The NCR peptides and their derivatives (Table 1 and Supplementary Table S1) were chemically synthesized with–COOH or–CONH₂ C-terminus. Peptides with–COOH -terminus were purchased from ProteoGenix (France), GenicBio Limited (People’s Republic of China), BBI Life Sciences Corporation (People’s Republic of China). Peptides with C-terminal–CONH₂ group were synthesized using an automated peptide synthesizer (CEM Liberty Blue) and TentaGel S RAM resin (loading of amino groups 0.23 mmol/g) according to the standard protocol of solid-phase peptide synthesis (SPPS) and as described before (Jenei et al., 2020). The isoelectric point (pI) and net charge (NC) of the peptides were calculated with https://pepcalc.com/. The calculation of other physicochemical properties were done with DBAASP v3 (Pirtskhalava et al., 2021) and AMP predictions were carried out with CS-AMPPred (Porto et al., 2012), CAMP_R3 (Waghu et al., 2016; Waghu and Idicula-Thomas, 2020), Antimicrobial Peptide Scanner vr.2 (Veltri et al., 2018), DBAASP v3 (Pirtskhalava et al., 2021) and Sense the moment (Porto et al., 2022).

The following pathogenic bacterial strains were used: The Gram-positive Enterococcus faecalis (ATCC 29212), Staphylococcus aureus (HNCMO112011), Listeria monocytogenes (ATCC 19111) and the Gram-negative Pseudomonas aeruginosa (ATCC 27853), Escherichia coli (ATCC 8739), Salmonella enterica (ATCC 13076), Klebsiella pneumoniae (NCTC 13440), Acinetobacter baumannii (ATCC 17978) obtained from the ATCC (United States) and NCTC (National Collection of Type Cultures–England). For the antifungal experiment the Candida albicans W01 strain (Ördögh et al., 2014) was used. Antimicrobial activities of peptides against ∼10⁷ log phase bacteria were tested in Potassium-Phosphate Buffer (PPB, pH 7.4) as described by Jenei et al. (2020) while anti-Candida assays were done according to Szerencsés et al. (2021). The two-fold dilution series of peptides ranged from 25 μM to 0.125 μM, while that of the antibiotics ampicillin (Merck) and miconazole (Duchefa Biochemie) from 10.24 mM to 0.1 μM. The lowest concentration of the antimicrobial agents, which completely eliminated viable bacteria or C. albicans were considered as the minimal bactericidal concentration (MBC) and minimal fungicidal concentration (MFC).

To test the possible effect of NCRs on the already formed biofilms, A. baumannii and C. albicans (at starting OD₆₀₀ = 0.1) were first incubated in 96 well plates containing 200 μl LB broth or YPD broth, respectively at 37°C for 48 h to allow biofilm formation (Kanugala et al., 2019). After formation of the biofilm 160 µl were gently discarded from the wells and then NCR solutions were added at 25, 50 and 100 µM final concentrations in the volume of 200 µl for overnight treatment (Montiel et al., 2017). Then, the unattached suspending cells were discarded and the remaining biofilms were washed twice with PBS. The bound biofilms were stained with 35 μl of 0.1% methanolic crystal violet for an hour at room temperature. Excess crystal violet was removed by washing the wells twice with distilled water. The stained biofilms were dried overnight at room temperature and then solubilized with 200 µl of 95% ethanol and mixing. The quantity of biofilms was measured by the absorbance of the samples at 570 nm. Untreated biofilms of A. baumannii and C. albicans were taken as controls corresponding to 100% of biofilms. All experiments were performed in duplicate and mean values and standard deviations (SD) were calculated.

Microbial cell membrane damage provoked by the peptides was assayed with Live/Dead staining. A. baumannii and C. albicans were grown as for the antimicrobial activity assay and washed gently with PPB. Cells were resuspended in PPB, and diluted to OD₆₀₀ = 0.1 and treated with peptides at various concentrations (as indicated in the Figures 3, 4 legends) at room temperature for 30 min. Untreated cells served as negative control. Cells were subsequently stained with 5 μM SYTO and 5 μM propidium iodide (PI). After 10 min incubation in the dark, 5 μl of the cell suspensions were spotted on microscope slide and covered with 2% (w/v) thin agar slices, and then observed with Olympus Fluoview FV 1000 confocal laser microscope at ×60 magnification using for SYTO9 488 nm laser for excitation, and 500–530 nm for emission. Excitation and emission wavelengths were 543 and 555–655 nm for PI, respectively. Sequential scanning was used to avoid crosstalk of the fluorescent dyes.

A. baumannii and C. albicans were grown and treated the same way as for the membrane permeability assay. Cells were then fixed with 2.5% (v/v) glutaraldehyde in phosphate buffered saline (PBS, pH 7.4). 5 μl of the above bacterial and yeast suspensions were spotted on a silicon disk coated with 0.01% Poly-L-Lysine. The filters were washed twice with PBS and gradually dehydrated with ethanol series (30, 50, 70, 80%, and three times 100% ethanol, each for minimum 30 min). The samples were dried with a critical point dryer (K850: Quorum Technologies Ltd.), followed by 12 nm gold coating and observed under a JEOL JSM-7100F/LV scanning electron microscope.

RTS 100 E. coli HY Kit (biotechrabbit Gmbh, Germany, catalog number: BR1400102) was used for the in vitro transcription/translation assay, according to the protocol. pIVEX2.3 control vector was used to produce GFP protein and to measure the translational efficiency. The amount of GFP produced was measured using a Hidex Plate Reader (Hidex Sense Microplate Reader with Plate Reader Software version 5064). The translational inhibitors, streptomycin and MtNCR247 as well as MtNCR001 without effect on translation (Farkas et al., 2014) were used as controls. All peptides and streptomycin were used at 100 µM concentration.

Interactions of peptides with DNA was visualized with gel retardation assay (Hsu et al., 2005). 100 ng Lambda DNA/HindIII marker (ThermoFisher Scientific) was incubated with 0, 5 or 10 µM synthetic NCRs or NCR derivatives in water for 30 min at room temperature and then loaded into 1% agarose gel. Electrophoresis of DNA fragments separated the mobile HindIII-digested DNA fragments while the aggregated DNA remained in the loading wells.

Human blood was purchased from the Regional Blood Centre in Szeged. The use of human blood for the hemolysis assay has been authorized by the Regional Hungarian Ethics Committee and approved by the Ethics Review Sector of DG RTD (European Commission) in connection with EK’s ERC AdG SymBiotics. The cells from 10 ml of EDTA-blood were centrifuged at 1,500 × g for 1 min and washed several times in TBS buffer (10 mM Tris, pH = 7.2, 150 mM NaCl) until the supernatant became colorless. The cells were then resuspended in 12 ml TBS buffer and 100 μl of this cell suspension was incubated for 1 h at 37°C in the presence of NCR peptides (0.4–100 μM). The cells were centrifuged at 1,500 g for 1 min and the supernatants were transferred into sterile 96-well plates and the hemoglobin release was measured at OD₅₆₀ (Hidex Sense Microplate Reader with Plate Reader Software version 5064). Cells in TBS buffer were used to determine baseline OD values, while 0.5% Triton X-100 (Serva) added at the same time as NCRs represented 100% of hemolysis. The hemolytic activity was calculated as % of red blood cell disruption relative to the positive control sample lysed with detergent Triton X-100.

Peptides were incubated in 10% Fetal Bovine Serum at 37°C for 18 h. Full length NCRs were used at 25 μM, while short derivatives at 50 μM. The serum proteins and peptides were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, 12% bis-tris gel, MES buffer pH 7.3) and visualized by Coomassie blue staining. NCRs with StrepII tag at their C-terminus were also detected by Western blot analysis using Precision Protein StrepTactin-HRP conjugate antibody (BioRad) and Clarity Western ECL (BioRad) for chemiluminescence detection according to the supplier.

We investigated the antimicrobial activity of 78 NCR peptides from M. truncatula and 12 NCRs from other IRLC legumes as well as their shorter derivatives against E. coli, S. enteritidis, P. aeruginosa, L. monocytogenes, S. aureus, E. faecalis, K. pneumoniae and A. baumannii (Supplementary Table S1). In the selection of the M. truncatula peptides, we considered their symbiotic expression levels but also their physicochemical properties. The antimicrobial activity of NCRs has so far been attributed to cationic peptides, with the exception of one anionic peptide. To investigate whether other anionic NCRs are antimicrobial, in addition to the cationic peptides, neutral and anionic NCRs were also included covering a broad range of peptides with different net charges (NC: from 7.8 to −10.2) and isoelectric points (pI: from 10.69 to 2.95). 50 peptides had positive net charge and pI > 7.0, while 28 peptides had negative net charge and pI < 7.0 (Supplementary Table S1A). Of the other legumes, only cationic NCR peptides were studied (Supplementary Table S1B).

The minimal bactericidal concentrations (MBCs) were determined after 3 h of treatment of ∼10⁷ log phase bacteria with 2-fold serial dilution of peptides from 25 µM to 0.8 µM. Only 25 of the 50 cationic M. truncatula peptides had activity at ≤25 µM (Supplementary Table S1). NCR073, NCR336 and NCR358 were the most active peptides capable of killing all tested bacteria. Of the 3 peptides, the less cationic peptides, NCR073 and NCR336 were somewhat more effective than the highly cationic NCR358. Interestingly, peptides with almost identical net charge and pI to the fully active NCR073 and NCR336 were inactive or only partially active, confirming high importance of the amino acid sequence. NCR377 and NCR645 could also kill all bacteria except E. faecalis. NCR280 and NCR700 were active against all strains except S. aureus and E. faecalis. NCR055 was also effective against six bacterial strains but not against S. aureus and K. pneumonia. Likewise, NCR384 was able to kill all tested bacteria except E. faecalis and K. pneumoniae. NCR183, NCR299, NCR471 and NCR520 were effective against five or four strains, while NCR135 against three strains. Three peptides were only active against two bacterial strains; NCR649 against L. monocytogenes and A. baumannii, NCR281 against L. monocytogenes and P. aeruginosa and NCR350 against A. baumannii and P. aeruginosa. NCR035 was only effective against E. coli, while NCR030 and NCR032 only against A. baumannii. None of the anionic NCR peptides affected the survival of the tested bacterial strains. However, the neutral peptide, NCR147 showed activity against four strains: E. coli, P. aeruginosa, L. monocytogenes and A. baumannii.

Of the other legumes, CaNCR13, CaNCR63, MsNCR443, MsNCR463, PsNCR349 and PsNCR352 had broad-spectrum activity, with the exception of MsNCR463, which was only effective against P. aeruginosa at 25 µM (Supplementary Table S1B). In MsNCR463 the positively charged amino acid residues accumulated at the C-terminal half of the molecule and we assumed that the relative inefficiency of the full length MsNCR463 might be caused by the N-terminal region. Therefore, we synthetized and tested two C-terminal derivatives of MsNCR463; MsNCR463_16-35 and MsNCR463_17-30. Likewise, nine lysine residues (K) were present in the C-terminal part of PsNCR349, that could be responsible for the antimicrobial activity, therefore, PsNCR349_26-50 and PsNCR3493_31-50 were also synthetized. On the contrary, the positively charged residues spread over the whole sequence in CaNCR63, indicating that likely the full-length sequence is necessary for the activity, though it could not be excluded that shorter derivatives of CaNCR63 were also active. From CaNCR63 we synthetized both the N-terminal sequence, CaNCR63_1-20 and a C-terminal fragment, CaNCR63_15-34. The MBC and the minimal fungicide concentration (MFC) of these peptides were determined against A. baumannii and C. albicans as examples of top bacterial and fungal pathogens on the WHO priority list (Table 1). All peptides were effective against both A. baumannii (MBCs: 1.6–6.25 µM) and C. albicans (MFCs: 1.6–25 µM), but the most active ones were the C-terminal derivatives of MsNCR463; MsNCR463_16-35 and MsNCR463_17-30 (MBC: 1.6 and 3.1 µM, MFC: 3.1 µM), which have the highest pI (11.79 and 12.44) of the tested NCRs. PsNCR349 and its C terminal derivative PsNCR349_26-50 were similarly effective against A. baumannii (MBC: 3.125 µM) and C. albicans (MFC: 3.125 µM). The even shorter PsNCR349_31-50, had slightly reduced activity against A. baumannii (MBC: 6.25 µM), but showed the best activity against C. albicans (MFC: 1.6 µM). The MBC value of the other peptides (CaNCR13, CaNCR63, CaNCR63_1-20, CaNCR63_15-34, MsNCR443, PsNCR349_31-50, PsNCR352) was 6.25 µM.

In terms of pathogens, E. faecalis was the most resistant to NCRs, as only nine peptides (NCR055, NCR073, NCR183, NCR299, NCR336, NCR358, NCR520, GoNCR308, PsNCR351) were able to kill this Gram-positive bacterium. K. pneumoniae was sensitive to 13 NCRs, S. aureus to 18 NCRs, S. enterica to 21 NCRs, L. monocytogenes to 24 NCRs, A. baumannii to 27 NCRs, E. coli to 33 NCRs and P. aeruginosa to 36 NCRs (Supplementary Table S1).

Based on our experiments, the net charge and pI were not the only determinants of the antimicrobial activity. AMP predictions for all peptides by eight different tools (Supplementary Table S1) gave different and only partially overlapping results, and none were fully consistent with our experimental data, as several NCRs, with high pI and net charge predicted as AMPs, were inactive. Therefore, other physicochemical properties of the peptides may play a role in the antimicrobial property. We used DBAASP v3 database that provides tools for multifactor analysis of physicochemical properties of peptides (Pirtskhalava et al., 2021). The normalized hydrophobic moment, normalized hydrophobicity, net charge, pI, penetration depth, tilt angle, disordered conformation propensity, linear moment, propensity to in vitro aggregation, angle subtended by the hydrophobic residues, amphiphilicity index and propensity to PPII coil of all peptides is shown in Supplementary Table S1. Comparing these properties of the active and inactive peptides revealed significant differences in normalized hydrophobicity, net charge, pI, disordered conformation propensity, amphiphilicity index and to a lesser extent in linear moment and angle subtended by the hydrophobic residues (Supplementary Figure S1). The active NCRs are positively charged, cationic and disordered peptides, but less hydrophobic and more amphiphilic than the inactive peptides. The plot of pI and net charge and the amphiphilicity index and disordered conformation propensity of each peptide together with the activity shown in Figure 1. As it was evident from the antimicrobial tests, high pI and net charge were required for antimicrobial activity but in some cases they alone were insufficient (Figure 1A). NCRs with net charge higher than 2 can be antimicrobial, but NCRs with net charge higher than 5 were all active, with the exception of NCR192. Moreover, the disordered nature and higher amphiphilicity index of active NCRs (Figure 1B) indicate higher possibility for antimicrobial activity likely due to adoption of secondary structures at membrane-water interphases (Vishnepolsky and Pirtskhalava, 2014; Pirtskhalava et al., 2021).

Due to the protective effect of the matrix, pathogens in the biofilm are less accessible to antimicrobials and are therefore more resistant. As the biofilms are the main sources of persistent or recurrent infections, we tested whether NCRs could degrade the already formed biofilms. Biofilms of A. baumannii and C. albicans were treated overnight with 25, 50 and 100 µM concentrations of the peptides listed in Table 1. The quantity of biofilms remaining after the peptide treatments were compared to the control untreated biofilms (Figure 2). The peptides had only a mild effect on biofilms of A. baumannii while they were more effective in destroying the C. albicans biofilms reducing the biofilm quantity by 50–60%.

Live/Dead staining of microbes with the membrane-permeable SYTO9 dye, which emits green fluorescence and propidium iodide (PI), which only enters the damaged dead cells and emits red fluorescence, allows investigation of membrane damage caused by the NCR peptides. In addition, morphological changes were also observed by scanning electron microscopy (SEM). A. baumannii (Figure 3) and C. albicans (Figure 4) cells were treated for 30 min with different concentrations of NCRs as indicated in the figure legends and then were either co-stained with SYTO9/PI dyes and observed under confocal microscopy, or the cells were fixed and observed with SEM. The untreated A. baumannii showed green fluorescence after the SYTO9/PI staining, while the treatment of A. baumannii with any of the peptides resulted in varying degrees of red fluorescence (Figure 3). Only red dead cells were observed after the treatment with CaNCR13, MsNCR443, PsNCR349_26-50 and PsNCR349_31-50 (Figures 3B,F,J,K). The few green, live cells observed among the red, dead cells might indicate a slower killing action of peptides CaNCR63, MsNCR463_16-35, MsNCR463_17-30 and PsNCR349 (Figures 3C,G–I), which can be even more pronounced in the case of PsNCR352 (Figure 3L) based on the generally yellowish staining of cells.

SEM revealed uniform size, morphology and smooth cell envelop of A. baumannii control cells (Figure 3M). This cell form was preserved in the case of PsNCR352 (Figure 3X) in line with the slower death indicated by the SYTO9/PI staining. The treatment by MsNCR463_16-35 and MsNCR463_17-30 caused budding of vesicles (Figures 3S,T). The treatment with CaNCR63, CaNCR63_1-20 and CaNCR63_15-34 caused rough cell surface, lysis and flattening of the cells (Figures 3O–Q). Only dead cells were visible after treatment with CaNCR13, PsNCR349 and PsNCR349_31-50 (Figures 3N,U,W). Filament-like structures, observed also by confocal microscopy by attachment of cells at their poles, were formed in the presence of CaNCR63_15-34, MsNCR443, MsNCR463_16-35, MsNCR463_17-30 and PsNCR349_26-50 (Figures 3Q–V).

Treatment of C. albicans with any of these peptides resulted in PI staining of the cells while in the untreated control culture, all cells were alive and showed green fluorescence (Figure 4A–L). A few viable green cells were, however, observed in the CaNCR63_1-20 sample (Figure 4D) in line with its highest MFC value. SEM observation revealed both yeast and hyphal forms in the control sample (Figure 4M). The CaNCR63_1-20 treated sample was the most similar to the control (Figure 4P) where a few living cells were observed with confocal microscopy (Figure 4D). Flattening of cells was visible after CaNCR63, MsNCR463_16-35, MsNCR463_17-30, PsNCR349 and PsNCR352 treatments (Figures 4O,S,T,U,X) and aggregation of shrunken cells in a spider net like matrix in the samples treated with CaNCR13, CaNCR63, MsNCR443 and PsNCR352 (Figures 4N,O,R,X).

There is growing evidence that AMPs in addition to their membrane interactions can also have intracellular bacterial targets and affect various cellular functions. Previously it was shown that MtNCR247 inhibits the bacterial translation, thus here we investigated if any of the peptides in Table 1 have similar functions. The effect of peptides was studied using an E. coli in vitro coupled transcription/translation assay, which measures the translational efficiency by GFP production (Figure 5). As controls, we included streptomycin and MtNCR247 as inhibitors of translation, as well as MtNCR001 without effect (Farkas et al., 2014). In agreement with previous data, streptomycin inhibited strongly and MtNCR247 to lesser extent the GFP production. The tested peptides resulted in GFP production comparable to the water and MtNCR001 controls indicating that these peptides have no significant effects on translation.

In the natural symbiotic system NCRs provoke amplification of bacterial DNA via endoreduplication cycles (Mergaert et al., 2006). While the functions of NCRs in this process is still unknown we investigated whether the peptides could interact with DNA using HindIII-digested λ DNA as test molecule (Figure 6). The peptides were added to the DNA at 5 μM and 10 μM concentrations for 30 min and then loaded into agarose gel. Separation of the λ HindIII-fragments by gel electrophoresis occurred in the control untreated samples, while the DNA remained in the wells in the presence of 10 μM CaNCR63, CaNCR63_15-34, MsNCR463_16-35, PsNCR349, PsNCR349_26-50 and PsNCR349_31-50. While these peptides aggregated the DNA fragments, CaNCR13, MsNCR443 and PsNCR352 did not. The N-terminal fragment of CaNCR63, CaNCR63_1-20 was less efficient than either the full-length peptide or the C-terminal CaNCR63_15-34, which retained completely the DNA-binding ability. CaNCR63_1-20 and CaNCR63_15-34 have similar pI and net charge values, indicating the contribution of the sequence downstream of the 15th amino acid to the DNA-binding. In PsNCR349 the positively charged amino acid residues are at the C-terminal part, accordingly the sequence from 31 to 50 was sufficient for aggregation of DNA. PsNCR352 was less active in DNA-binding than PsNCR349, which may be due to its lower net charge (6.80). MsNCR463_17-30 was less efficient than MsNCR463_16-35 indicating that the last five amino acids are necessary to the DNA-interaction.

While AMPs can be very effective against microbes, they should not be toxic for human cells. The potential hemolytic activity of the legume peptides listed in Table 1 was tested in hemolysis assay (Jenei et al., 2020). Human red blood cells were co-incubated with the peptides used in two-fold dilutions from 100 to 0.2 µM or with 0.5% Triton X-100 provoking complete hemolysis. Disruption of red blood cells was measured by spectrophotometry and compared to samples lysed with the detergent Triton X-100 (Supplementary Figure S2). None of the peptides caused hemolysis in the studied concentration range, except MsNCR465, provoking mild hemolysis at 50 and 100 μM concentrations, far above the MBCs of this peptide.

We also tested if NCRs remain stable in the presence of serum. 18 antimicrobial NCRs and NCR derivatives were incubated overnight at 37°C with Fetal Bovine Serum (FBS). The amount of peptides with or without FBS was comparable based on Coomassie Blue staining of SDS-PAGE gels (Supplementary Figure S3). To confirm the results, NCR183 and NCR055 carrying a StrepII tag at their C-terminus were also visualized by Western blot analysis using anti-StrepII-tag antibody and in line with the results of Coomassie Blue staining both of them were stable in the presence of FBS.