Antimicrobial Activity of Bacteriophage Endolysin Produced in Nicotiana benthamiana Plants

Abstract

The increasing spread of antibiotic-resistant pathogens has raised the interest in alternative antimicrobial treatments. In our study, the functionally active gram-negative bacterium bacteriophage CP933 endolysin was produced in Nicotiana benthamiana plants by a combination of transient expression and vacuole targeting strategies, and its antimicrobial activity was investigated. Expression of the cp933 gene in E. coli led to growth inhibition and lysis of the host cells or production of trace amounts of CP933. Cytoplasmic expression of the cp933 gene in plants using Potato virus X-based transient expression vectors (pP2C2S and pGR107) resulted in death of the apical portion of experimental plants. To protect plants against the toxic effects of the CP933 protein, the cp933 coding region was fused at its Nterminus to an N-terminal signal peptide from the potato proteinase inhibitor I to direct CP933 to the delta-type vacuoles. Plants producing the CP933 fusion protein did not exhibit the severe toxic effects seen with the unfused protein and the level of expression was 0.16 mg/g of plant tissue. Antimicrobial assays revealed that, in contrast to gram-negative bacterium E. coli (BL21(DE3)), the gram-positive plant pathogenic bacterium Clavibacter michiganensis was more susceptible to the plant-produced CP933, showing 18% growth inhibition. The results of our experiments demonstrate that the combination of transient expression and protein targeting to the delta vacuoles is a promising approach to produce functionally active proteins that exhibit toxicity when expressed in plant cells.

Introduction

The increasing spread of antibiotic-resistant microorganisms is a growing concern for modern animal production, agriculture, and medicine, and also has a significant negative impact on the US economy [11,14]. In this regard, phage-encoded endolysins that degrade peptidoglycan (murein) in the cell walls of bacteria, resulting in cell lysis, have acquired significant attention as antimicrobials owing to their high efficiency and mechanism of action [2,7,13,24,37].

Endolysins are bacteriophage-encoded peptidoglycan hydrolases, which are synthesized in phage-infected cells at the end of the phage lytic cycle [36]. During the process of phage maturation, lysins accumulate in the cytoplasm of infected bacterial cells. Most of the tailed phages possess a holin-endolysin system to achieve the release of their progeny through bacterial lysis. Holin is a small hydrophobic protein produced by double-stranded DNA phages at the end of the lytic cycle and is incorporated into and disrupts the bacterial cytoplasmic membrane, causing the exposure of the cell wall to the peptidoglycan hydrolyses [40,41]. Alternatively, the lysins containing an N-terminal signal sequence may be exported through the plasma membrane by the host sec system [26,35,42] or signal-anchor-release system (SAR) [27,29]. SAR endolysins that possess a non-cleavable N-terminal type II signal anchor require pinholins (a class of holins) that, at a genetically determined time, cause membrane depolarization and release of the SAR endolysin from the inner membrane into the periplasm, where it refolds and becomes muralytically active [27,29-31]. Lysins can also interfere with biosynthesis of peptidoglycan, leading to misassembled bacterial cell wall (single-stranded RNA and DNA phages) [12].

Externally applied lysins preferably lyse gram-positive bacteria owing to specific cell organization [13] and therefore could serve as topical antimicrobials or alternatives to antibiotics. The main advantages of lysins versus antibiotics are (i) exogenous lysin application leads to rapid lysis of the bacterial cell wall, avoiding such intracellular resistance mechanisms as efflux pumps, thereby making microbial resistance development rare; and (ii) endolysins possess narrow species specificity without affecting aboriginal (normal) microflora. Numerous studies have shown wide perspectives of lysin usage in human and veterinary medicine as well as in agriculture to control and treat bacterial pathogens, including antibiotic-resistant bacteria [8,25,33,36]. Although the efficiency of lysin application in the treatment and prevention of infectious diseases under laboratory conditions is well established, there is still a need for further investigations in this field before lysin preparations can be approved for human therapy.

In the present study, the antimicrobial properties of the CP933 endolysin, encoded by the cryptic bacteriophage of the gram-negative bacterium Escherichia coli O157:H7 str. EDL933 [32], were examined. The Potato virus X (PVX)-based vectors pP2C2S and pGD107 [3,22] were used for transient expression of CP933 in Nicotiana benthamiana plants. Although there are distinct advantages of widely used transient-expression systems in plants [6,23,34], additional approaches, such as targeting proteins to subcellular compartments [5,39] or secretory pathways [1,4,9,19], may be required if the target protein is toxic to the host plant and/or for improving recombinant protein accumulation in plants. Murray et al. [26] and Jackson et al. [15] reported successful targeting of a plant-toxic protein, avidin, to the vacuole in the cells of transgenic tobacco, apple, and sugarcane plants using an N-terminal signal peptide from the potato proteinase inhibitor I (PPI-I) protein. The PPI-I presequence is known to target PPI-I protein constitutively to delta-type vacuoles in potato tubers, and tomato leaves following wound induction [26]. The delta-type vacuoles are distinct from protein storage and lytic vacuoles and harbor pigments and vegetative storage proteins during plant development and in response to environmental triggers [16,17]. Results of these experiments suggested that the vacuole is the preferable organelle for stable storage of plant transgene proteins that exhibit toxicity to a host plant [26].

In our work, as a result of the toxicity of CP933 to plant cells being revealed during our experiments, the cp933 coding region was fused at its N-terminus to an N-terminal signal peptide from the PPI-I protein to direct CP933 to delta-type vacuoles. The antimicrobial activity of plantproduced CP933 was evaluated in vitro against Clavibacter ichiganensis and E. coli.

 

Materials and Methods

Plasmid Constructions

Cloning of the cp933 gene into the plasmid vector pET21a(+). The complete list of primers used for cloning is shown in Table 1. The genomic sequence encoding the CP933 endolysin originated from the cryptic prophage CP-933P (E. coli O157:H7 str. EDL933) and was predicted using BLAST at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/blast) (GenBank Accession No. NP_287988) [32]. The cp933 gene was synthesized by Genscript (Piscataway, NJ, USA) and cloned into the pUC57 vector at the NdeI/XhoI restriction sites. The coding region of the cp933 gene was amplified with the primer pair CP933NdeIF and CP933XhoIR and, after standard cloning procedures, was incorporated into pET21a(+), giving rise to pET21a(+)/cp933 with the addition of a C-terminal 6×His-tag to facilitate purification of CP933 using Ni-NTA resin.

Table 1.aRestriction sites are underlined; nucleotides in italics encode the 6×His-tag; overlapping regions are shown in bold.

Cloning of the ppi-I, ppi-I_cp933, and ppi-I_gfp genes into the plasmid vector pET28a+. To direct CP933 to the plant delta vacuole, the coding region of the cp933 gene was fused at its Nterminus to an N-terminal signal peptide from the PPI-I protein. The overlapping primers NcoIppiF and BamHppiIR were designed based on the sequence of PPI-I [26]. The amplified product was incorporated into the pET28a+ vector at the NcoI/BamHI restriction sites, to produce pET28a+/ppi-I. Because of problems with some restriction sites encountered during further cloning procedures, the ppi-I gene was amplified from pET28a+/ppi-I by PCR using the primer pair NcoIppiF+/EcoRIppiR and incorporated into the pET26b+ plasmid vector, giving rise to pET26b+/ppi-I. The resulting plasmid DNAs were digested with BamI/XhoI (plasmid pET28a+/ppi-I) and EcoRI/HindIII (plasmid pET26b+/ppi-I). The coding regions of the cp933 and gfp genes were amplified from the plasmids pET21a(+)/cp933 and pGDsmGFP using primer pairs EcoRICP933F/NotHisCP933R2 and BamHgfpF/XhoIgfpR, respectively, and incorporated into the corresponding plasmids, giving rise to pET26b+/ppi-I_cp933 and pET28a+/ppi-I_gfp. The ppi-I_cp933 gene was amplified from the plasmid pET26b+/ppi-I_cp933 using primer pairs NcoIpiF1/HindSHR1 and, after standard cloning procedures, was incorporated into the plasmid vector pET28a+ at the NcoI/HindIII sites, to produce pET28a+/ppi-I_cp933.

Cloning of the cp933, ppi-I_cp933, gfp, and ppi-I_gfp genes into the PVX-based vector pP2C2S. The coding regions of the cp933, ppi-I_cp933, gfp, and ppi-I_gfp genes were amplified from the plasmids pET21a(+)/cp933, pET28a+/ppi-I_cp933, pGDsmGFP, and pET28a+/ppi-I_gfp, using primer pairs CP933F/EcoRVStopHisR, EcoRVppiF/EcoRVStopHisR, EcoRVgfpF/EcoRVgfpR, and EcoRVppiF/EcoRVgfpR, respectively. The PCR products were incorporated into the pP2C2S vector at the EcoRV site, giving rise to pP2C2S/cp933, pP2C2S/ppi-I_cp933, pP2C2S/gfp, and pP2C2S/ppi-I_gfp, respectively.

Cloning of the cp933 and ppi-I_cp933 genes into the binary PVXbased vector pGR107. The multiple cloning site (MCS) of the vector pGR107 (a gift of Dr. David Baulcombe, Sainsbury Centre, UK [22]) was modified to incorporate additional restriction sites (not shown). The available restriction sites in the expanded MCS included MluI, CalI, SmalI, SalI, and NotI. The cp933 and ppi-I_cp933 genes were amplified from pET21a(+)/cp933 and pET28a+/ppi-I_cp933 using primer pairs MluKozCP933F/NotHisCP933R2 and MluKozppiIF/NotHisCP933R2, respectively, and were cloned into plasmid vector pGR107 at the MluI/NotI sites, giving rise to pGR107/cp933 and pGR107/ppi-I_cp933.

All constructs were verified by direct DNA sequencing.

Preparation and Delivery of Infectious Transcripts

Capped transcripts from SpeI-linearized pP2C2S/cp933, pP2C2S/ppi-I_cp933, pP2C2S/gfp, and pP2C2S/ppi-I_gfp plasmids weregenerated with the mMessage mMachine T7 kit (Ambion, Austin, TX, USA), diluted twice with 20 mM sodium phosphate buffer (pH 7.0), and rubbed onto three top carborundum-dusted leaves of N. benthamiana plants. Inoculated plants were grown in a greenhouse at 27 ± 2℃ under 18 h light and 6 h dark photoperiods until protein extraction was performed.

Transformation of Agrobacterium tumefaciens and Agroinfiltration

Introduction of pGR107/cp933 and pGR107/ppi-I_cp933 into A. tumefaciens and agroinfiltration of N. benthamiana plants were performed using procedures previously described [21].

RT-PCR for Detection of cp933 and ppi-I_cp933 Genes in Plants

Total cellular RNA was extracted by TRI Reagent (Molecular Research Center, Inc., Cincinnati, OH, USA) from systemically infected N. benthamiana leaves 1 week after inoculation with PVX RNA transcripts or agroinfiltration. RT-PCR analysis was carried out using the Titan One Tube RT-PCR System (Roche Molecular Biochemicals, Chicago, IL, USA) as described in the manufacturer’s instruction and using the primer pairs CP933F/EcoRVStopHisR, EcoRVppiF/EcoRVStopHisR, MluKozCP933F/NotHisCP933R2, and MluKozppiIF/NotHisCP933R2 (0.4 pmol final concentration) for pP2C2S/cp933, pP2C2S/ppi-I_cp933, pGR107/cp933, and pGR107/ppi-I_cp933, respectively. For RT-PCR, 35 cycles were conducted in a GeneAmp System 9700 (Applied Biosystems, Foster City, CA, USA) with AMV reverse transcriptase for the first-strand cDNA synthesis and the Expand High Fidelity enzyme blend (Roche) consisting of Taq DNA polymerase and Tgo DNA polymerase for amplification of cDNA by PCR. The PCR fragments were fractionated on a 1.0% agarose gel.

Protein Extraction and Characterization

Protein production in bacteria. To produce CP933 protein in bacteria, the construct pET21a(+)/cp933 was introduced into E. coli strains BL21(DE3) (Stratagene, La Jolla, CA, USA), C43(DE3) pLysS, C43(DE3), C41(DE3) pLysS, and C41(DE3) (Lucigen, Middleton, WI, USA) according to the manufacturer’s instructions. Protein induction, extraction, inclusion body (IB) purification, solubilization, and refolding were performed as previously described [20].

Protein extraction from N. benthamiana and purification. Protein extraction from plant tissues was performed using 1× PBS buffer (Bio-Rad, Hercules, CA, USA) or 10 mM Tris-HCl, pH 8.0, containing 8 M urea. Both buffers were supplemented with a 1:100 dilution of a plant protease inhibitor cocktail (Sigma Chemical Co., St. Louis, MO, USA). After two rounds of centrifugation (at 4,000 ×g for 20 min and at 10,000 ×g for 10 min) for extracts in 1×PBS buffer and six rounds of centrifugation (one round at 4,000 ×g for 20 min and five rounds at 10,000 ×g for 10 min) for extracts in 8 M urea buffer, the protein samples were analyzed by western blot assay or purified using Ni-NTA resin after filtration through 0.22 μm low protein binding membrane filter (Millipore Corporation, Bedford, MA, USA). For purification, filtered protein samples obtained with 1× PBS buffer were applied to a column with 1 ml of Ni-NTA resin pre-washed with 1× PBS containing 10 mM imidazole and incubated for 1 h at 4℃, with rotation. Columns were washed with 3 column volumes of 1× PBS containing 20 mM imidazole and bound protein was eluted twice with 500 μl and 300 μl of 1× PBS containing 500 mM imidazole. Eluted fractions were combined and dialyzed against 1× PBS overnight at 4℃. In the procedure where proteins were extracted using 8 M urea buffer, protein samples were applied to a column with 1 ml of Ni-NTA resin pre-washed with 7 M urea buffer, pH 8.0, and incubated for 1 h at 4℃, with rotation. Columns were washed with 3 column volumes of 8 M urea buffer, pH 6.3, and bound protein was eluted twice with 500 μl and 300 μl of 8 M urea buffer, pH 4.5. The eluted fractions were combined and dialyzed against 20 mM Tris-HCl, pH 8.0, overnight at 4℃. The resulting protein samples were analyzed by western blotting and used for antibacterial assays after filtration through a 0.22 μm low protein binding membrane filter (Millipore Corporation). To determine the concentration of purified protein samples, measurement of absorbance at 280 nm was performed (Thermo Scientific NanoDrop ND-8000 8-Sample Spectrophotometer).

Gel electrophoretic and western blot analyses of proteins. Aliquots of the CP933 protein produced in bacteria or plants were electrophorized on a Novex Tris-Glycine Gradient Gel (10% to20%; Invitrogen) under denaturing conditions as described in the manufacturer’s instructions. The protein was visualized by staining with SimplyBlue SafeStain (Invitrogen). Alternatively, proteins were transferred to a nitrocellulose membrane (following manufacturer’s instructions; Invitrogen) that was subsequently incubated with 1:1,000 dilution of Anti-His HRP Conjugate solution (Penta His HRP Conjugate Kit; 5 Prime Inc., Gaithersburg, MD, USA) according to the manufacturer’s instruction, followed by membrane development using the TMB Membrane Peroxidase Substrate System (KPL, Gaithersburg, MD, USA).

Confocal laser scanning microscopy (CLSM). Transverse sections (approx. 0.2 mm) and pieces of fresh leaves of N. benthamiana were excised and placed in cover glass-bottom Petri dishes (MatTeck Corp., Ashland, MA, USA) for observation. A Zeiss LSM710 CLSM system (Thornwood, NY, USA) was used to obtain images. The images were observed using a Zeiss Axio Observer inverted microscope with 40× 1.3 NA Plan-Apochromat water immersion objectives. A 488 nm argon laser with a pin hole of 30 μm passing through a MBS 488 beam splitter filter with limits set between 495 and 535 nm was used. The Zeiss Zen 2012 (Thornwood, NY, USA) 64 bit software was used to obtain images. Forty to seventy 30 μm per frame z-stack images were obtained to produce 3D renderings, which were used to develop the 2D maximum intensity projections for publication.

Microorganisms

Clavibacter michiganensis subsp. sepedonicus (hereinafter referred to as C. michiganensis), a gram-positive bacterium causing potato ring rot disease, was cultivated on nutrient-broth yeast extract agar (NBY) [21] at 25℃. For antibacterial assays, NBY broth was inoculated with the colonies from the plate and incubated at 25℃ with shaking at 250 rpm for 5 days.

E. coli BL21(DE3) (Stratagene), a gram-negative bacterium, was transformed with pET26b+ plasmid for growth on YT medium supplied with kanamycin. For antibacterial assays, the colonies from the plate were transferred into YT broth containing kanamycin and incubated overnight at 37℃ with shaking at 250 rpm.

Antibacterial Assays

The antibacterial activity of the plant-produced and purified CP933 protein was examined against C. michiganensis and E. coli as previously described [20], using a final protein concentration of 54 μg/ml in 500 μl of total reaction volume. The bacterial concentration before experiment was 5.6 × 104 colony-forming units (CFU)/ml for C. michiganensis and 3.6 × 105 CFU/ml for E. coli. The reaction mixtures were incubated at 25℃ and 37℃ with shaking at 250 rpm for 5 days and 17 h for C. michiganensis and E. coli, respectively. Following incubation, 100 μl aliquots of protein-treated bacterial cultures were serially diluted in sterile water (from 10-2 to 10-6) and 25 μl of diluted bacterial suspensions was plated onto the appropriate solid medium. The plates were incubated at 25℃ and 37℃ for 5 days and 17 h for C. michiganensis and E. coli, respectively, and the bacterial growth was examined by counting CFU. Each experiment has at least three repetitions. The data are expressed as the mean ± the standard error of the mean. Statistical significance of obtained data was analyzed by calculating the p-value (http://www.socscistatistics.com).

 

Results

In order to examine the antimicrobial properties of CP933, we attempted to produce the protein in a prokaryotic expression system. The cp933 gene was subcloned into the pET21a(+) plasmid vector under the transcriptional control of the bacteriophage T7 promoter (Fig. 1A) and introduced into E. coli strains BL21(DE3), C43(DE3) pLysS, C43(DE3), C41(DE3) pLysS, and C41(DE3), as described in Materials and Methods. Expression of the cp933 gene in bacterial cells led to growth inhibition and lysis of the host cells (strains BL21(DE3), C43(DE3) pLysS, and C41(DE3) pLysS) or production of trace amounts of the CP933 (strains C43(DE3) and C41(DE3)). Analysis of both total and soluble fractions of CP933 protein showed that it was localized in the total fraction containing insoluble inclusion bodies (Fig. 2 A). Expression of cp933 in E. coli was confirmed by western blot analysis (Fig. 2B). The attempt to purify the bacterially produced protein with Ni-NTA resin failed owing to the extremely low yield of protein.

Fig. 1.Schematic representation of vector constructions carrying cp933 and fusion genes ppi-I_cp933 and ppi-I_gfp.

Fig. 2.SDS-PAGE and Western blot analysis of E. coli-produced CP933.

To overcome the expression problems encountered in E. coli, we explored the possibility of producing the CP933 protein in N. benthamiana plant cells. The cp933 gene containing a 6× His-tag at its C-terminus was engineered into the PVX-based transient expression vector pP2C2S [3] and binary PVX-based vector pGR107 [22], under the transcriptional control of the bacteriophage T7 and Cauliflower mosaic virus 35S promoters, respectively, with the expectation that the CP933 protein would be expressed in the cytoplasm of plant cells. Expression of the cp933 gene from both of these vectors resulted in growth inhibition and death of the apical region of experimental plants after 10 dpi (Fig. 3C). Western blot analysis did not detect the presence of CP933 in plants producing cytoplasmic CP933. Moreover, the collection of plant samples was complicated by the low quality of plant material due to the negative impact of CP933 on plants. The lower, initially uninfected leaves remained uninfected during the period of experiment, so they could not be used for protein extraction. To protect plants against the detrimental impact of CP933, the cp933 gene was fused at its N-terminus to an N-terminal signal peptide from PPI-I (Figs. 1B and 1C) to direct CP933 to delta-type vacuoles. To verify the ability of the PPI-I signal peptide to target a protein of interest to a vacuole, we designed a construct containing the gfp reporter gene (pP2C2S/ppi-I_gfp) (Fig. 1D), introduced it into the N. benthamiana plants, and examined GFP localization in the cells of systemically infected leaves on the 10th dpi. Confocal microscopy showed expected autofluorescence in N. benthamiana cells (Figs. 4A and 4B) of the “control-mock” (plants treated with 20 mM sodium phosphate buffer) and “empty-pP2C2S” (plants infected with empty pP2C2S vector) experimental groups, and confirmed the vacuolar localization of GFP in infected N. benthamiana leaf mesophyll cells (Fig. 4D) compared with its primary localization in the nucleus and cytoplasm in the construct lacking the PPI-I signal peptide (Fig. 4C).

Fig. 3.Effect of CP933 on growth of N. benthamiana plants.

Fig. 4.GFP localization in systemically infected N. benthamiana leaves (10 dpi).

We next examined the expression of the ppi-I_cp933 gene in N. benthamiana as described in Materials and Methods. Plants producing the vacuolar-targeted PPI-I_CP933 fusion did not exhibit the severe phenotypic symptoms (Fig. 3D ) observed with cytoplasmic expression of cp933 (Fig. 3C). The fact that plants producing the fusion protein PPII_CP933 remained phenotypically normal indicates that targeting was successful.

The stability of the cp933 and ppi-I_cp933 genes within the PVX-based vector in N. benthamiana plants was confirmed by RT-PCR assays performed on RNA samples isolated from systemically infected leaves at 8 dpi (Fig. 5).

Fig. 5.RT-PCR analysis of total cellular RNA isolated from the systemically infected N. benthamiana leaves (8 dpi).

The plant-produced PPI-I_CP933 fusion protein was extracted and purified with Ni-NTA resin at 7-10 dpi. Western blot analysis using His-tag specific antibodies confirmed CP933 protein production in systemically infected N. benthamiana leaves (Fig. 6). A difference in migration of the CP933 protein on the gel was dependent on the buffer used for protein extraction from plant tissues. The molecular weight of the detected protein was consistent with the predicted molecular mass of the fusion PPI-I_CP933 after cleavage (21.6 kDa) (Fig. 6A) when the 1× PBS extraction method was used. However, when 8 Murea buffer, pH 8.0, was employed for protein extraction and purification, the estimated molecular weight of the CP933 product was twice that of the predicted molecular mass (Fig. 6B). The level of production of fusion CP933 in plants was 0.16 mg per gram of plant tissue.

Fig. 6.Western blot analysis of CP933 expressed in N. benthamiana.

To test the antimicrobial activity of the plant-produced endolysin, we performed growth inhibition assays using the PPI-I_CP933 protein, as described in Materials and Methods. The experiments demonstrated that, in contrastto E. coli, C. michiganensis was more susceptible to treatment with CP933, showing statistically significant 18% growth inhibition (p < 0.05 (0.006)) (Fig. 7).

Fig. 7.Antimicrobial activity of plant-produced CP933 protein (54 μg/ml) against C. michiganensis (Cms) and E. coli BL21(DE3).

 

Discussion

In this study, the CP933 endolysin encoded by a cryptic prophage CP-933P (Escherichia coli O157:H7 str. EDL933) was produced in N. benthamiana plants using a PVX-based transient expression vector, and its antimicrobial activity was evaluated in vitro against the gram-positive bacterium C. michiganensis and the gram-negative bacterium E. coli BL21(DE3).

Prior to producing CP933 in plants, we attempted to express the cp933 gene in a prokaryotic expression system. Several strains of E. coli were chosen (BL21(DE3), C43(DE3) pLysS, C43(DE3), C41(DE3) pLysS, and C41(DE3)), but only two that were specially designed for toxic protein production were able to produce CP933 (strains C43(DE3) and C41(DE3)), although at unsatisfactory levels. Strains C43(DE3) and C41(DE3) were derived from C41(DE3) and BL21(DE3), respectively, and are suitable for production of protein from the genes cloned into expression vectors under the control of the T7 promoter (http://www.lucigen.com).

The mode of action of CP933 is not completely known. Analysis of the CP933 amino acid sequence revealed that it is a positively charged protein, rich in hydrophobic amino acids. Bioinformatic comparison of the predicted CP933 amino acid sequence with the complete genome sequences of all endolysins found in double-stranded DNA and RNA phages [27] revealed that CP933 contains an N-terminal transmembrane domain with a catalytic residue (Fig. 1A), suggesting that CP933 belongs to the SAR family of endolysins. Moreover, according to the literature, SAR endolysins were encoded only by phages of gram-negative bacteria, mostly from the Enterobacteriaceae family, with few exceptions [27]. In the SAR system, the non-cleavable N-terminal transmembrane domain, being a part of the mature endolysin, stays embedded in the plasma membrane in an enzymatically inactive form during the latent period until membrane depolarization caused by pinholin activity [27,29,30,31]. However, the robust lytic activity of recombinant CP933 in E. coli, when pinholin is not available, is most likely explained by the impact of positively charged CP933 endolysin, primarily, on negatively charged bacterial membrane, leading to membrane disruption, followed by enzymatic degradation of the cell wall peptidoglycan in bacteria. This hypothesis leads us to speculate that CP933 may also affect the plant cell wall or membrane or both when expressed in the cytoplasm, leading to collapse of rapidly growing plant tissues. It was reported that some lysins, especially those originating from phages of gram-negative bacteria, are capable of affecting bacterial cells in an enzymatic-independent way. It was found that some endolysins contained a sequence in the C-terminus similar to a sequence of cationic antimicrobial peptides [10,28]. Düring et al. [10] showed that T4 lysozyme, derived from the phage of a gram-negative bacterium, contained in its sequence at least one positively charged α-helix domain that consisted of basic amino acid residues able to interact with the negatively charged bacterial membrane. It has also been shown that synthetic peptides with an amino acid sequence corresponding to the sequence of this helix possessed strong bactericidal and not enzymatic activity.

Furthermore, the impact of plant viral vector itself on plant tissues cannot be excluded. In our previous studies, we observed the synergistic effect of co-infection by the empty-PVX-based vector and the phytopathogenic fungus Colletotrichum coccoides that led to plant collapse [21]. However, in the present study, the absence of the toxic effect in plants expressing vacuolar-targeted CP933 using PVX-based vectors (pP2C2S/ppi-I_cp933 or pGR107/ppi-I_cp933) indicates that the vectors themselves do not play a crucial role in the severe symptom development (growth inhibition and death of the plant apical region) that was observed in the case of cytoplasmic protein expression (pP2C2S/cp933 or pGR107/cp933). Therefore, we concluded that the phenotypic changes in plants producing CP933 (pP2C2S/cp933 or pGR107/cp933) in the cytoplasm were primarily a result of CP933 activity.

To protect plants against the detrimental impact of CP933, the cp933 gene containing a 6×His-tag at its C-terminus was fused at its N-terminus to the PPI-I signal peptide to direct CP933 to delta-type vacuoles. The successful expression of plant-toxic proteins in transgenic tobacco leaves and sugarcane through the use of vacuolar targeting was shown for the biotin-binding proteins avidin and streptavidin [15,26]. Targeting to the delta vacuoles resulted in the high yields of avidin for both plants and streptavidin for tobacco plants, although in sugarcane a biotin-deficient phenotype was observed [15]. Tobacco plants appeared phenotypically indistinguishable from non-transgenic plants [26]. In our attempt to express the endolysin gene in N. benthamiana, a combination of transient expression and vacuole targeting strategies were employed. Plants producing PPI-I_CP933 fusion protein did not exhibit the severe toxic effects on apical shoot growth observed with cytoplasmic expression of CP933, indicating successful protein targeting. Since one of the aims of this work was to obtain functionally active CP933 protein, the fusion of CP933 (20.7 kDa) with GFP (26.3 kDa) for monitoring of CP933 localization in plant cells was avoided as this protein fusion could affect protein folding and consequently the activity of CP933.

Low protein expression in N. benthamiana may be due to the toxicity of CP933 itself, even after the vacuole targeting strategy was applied. Electrophoretic analysis revealed the difference in sizes between CP933 extracted with 1× PBS and 8 M urea buffer, where the protein migrated with the size of a dimer after treatment with 8 M urea. It has been reported that the Norovirus capsid P protein spontaneouslyformed dimers that were very stable over a broad range of pH (2 to 11) or under strong denaturing conditions (60 min of 8 M urea or 6 M guanidine treatment) [38]. Moreover, dimerization through the cysteine SH group of calreticulin (ubiquitous protein, located primarily in the endoplasmic reticulum) was induced by lowering the pH to 5-6, by heating, or in the presence of urea at a concentration above 2.6 M or an SDS concentration above 0.025% [18]. The same protein formed very stable oligomers through non-covalent interactions at concentrations of urea above 2.6 M (pH below 4.6 or above 10) at temperature above 40oC, or in the presence of organic solvents at high concentrations (25%) [18]. It may be that extraction of CP933 under denaturing conditions (using 8 M urea buffer) allows monomers to closely interact with each other, forming very stable dimers.

The antimicrobial activity of purified plant-produced PPI-I_CP933 protein, when added exogenously to the bacteria, was evaluated against C. michiganensis and E. coli BL21(DE3). C. michiganensis was more susceptible to CP933 treatment than E. coli. Current literature indicates that lysins, when added externally, preferably work against gram-positive bacteria, because of the ability of lysins to contact peptidoglycan of the cell wall directly, whereas the presence of the outer membrane in gram-negative bacteria makes lysin-peptidiglycan interaction difficult [13].

Our study showed that targeting of CP933 to the delta vacuoles using the signal peptide of the PPI-I protein protected plants against the toxic effect of the CP933 endolysin and allowed recovery of protein possessing antimicrobial activity. Thus, the combination of transient expression and protein targeting is a promising approach for the production of functionally active proteins that exhibit toxicity when expressed in E. coli and cytoplasmically in plant cells.