2.1. Development of the Vector
The purpose of this study was to create a bacterial vector for the production and secretion of a variety of AMPs from a single probiotic organism. We remodeled a MccV production plasmid, pHK22, in order to create a vector that contained the entire MccV secretion machinery, as well as a strong DNA promoter system and an AMP cloning site.
We refer to this final vector as pMPES, for Modular Peptide Expression System. Figure 1
shows a diagram of this vector.
The MccV secretion system was selected for this application because it is compatible with E. coli
, because it is among the most well-characterized AMP secretion systems and because it recognizes a glycine-glycine type signal peptide, a common class of signal peptide used by many AMPs [22
The MccV secretion pathway relies on the ABC transporter CvaB and the accessory protein CvaA [18
]. The outer membrane protein, TolC, is encoded in the E. coli
chromosome and is also required for transport [27
Vector pHK22 contains a 9.1-kb fragment originally isolated from the native MccV production plasmid, pColV-K30. This region, located between the HindIII and SalI restriction enzyme cut sites shown in Figure 1
, was previously found to contain all of the necessary components for native MccV production, immunity and secretion [18
]. The sequence of this region is reported under GenBank Accession Number X57524.1 [27
For the development of pMPES, we made three primary modifications to pHK22: (1) we mutated the MccV structural gene to remove native AMP activity; (2) we added a well-characterized synthetic DNA promoter; (3) we added a multiple cloning site and a terminator site to facilitate insertion and removal of AMPs.
The mutation of the native mccV
gene was essential for testing of pMPES, since MccV production would interfere with the characterization of other peptides. This mutation was introduced into the start codon of the native mccV
, to encode a stop codon as shown in Figure 1
. This new vector is referred to as pHK22Δ. Mutation of the stop codon was selected over complete removal of mccV
in order to minimize the risk of disrupting any unknown secretory components. To verify that MccV or any potentially active truncated peptides were no longer being produced, E. coli
MC1061 F’ containing pHK22Δ was tested against the indicator strain, E. coli
DH5α, and shown to have no activity (Figure S1
Next, a strong DNA promoter was incorporated into the vector to enable high-level protein expression. We have previously developed a synthetic DNA promoter for E. coli,
which relies on a synthetic hybrid activator protein, ProTeOn [31
]. ProTeOn was constructed by physically linking the reverse tetracycline repressor protein to the activating domain of the Vibrio fischeri
transcription factor, LuxR. ProTeOn makes strong contacts within the engineered DNA promoter site (Pon
), which contains optimally-spaced tetracycline and LuxR operator binding regions. This system recruits RNA polymerase and strengthens the holoenzyme-DNA interactions to up-regulate gene expression.
ProTeOn has been further modified in this study to include a positive feedback loop by inserting the gene encoding ProTeOn downstream of Pon. By using this feedback loop, we are able to amplify promoter expression and obtain high levels of the proteins of interest compared to the original ProTeOn promoter. Herein, we will refer to this new expression construct as ProTeOn+.
A multiple cloning site (MCS) containing the five restriction sites SacI, ApaI, AvrII, NotI and PciI and a rho-independent terminator sequence was then inserted downstream of ProTeOn+.
2.2. Activity Tests
In this study, we evaluated the production of seven different peptides from the MccV secretion machinery; MccV, Microcin L (MccL), Microcin N (McnN), Enterocin A (EntA), HirJM79, Enterocin P (EntP) and Enterocin B (EntB). These peptides were selected to represent a wide range in percent sequence identity to MccV and its production system. We note that these tests were not intended to be comprehensive, but only indicative of combinatorial possibilities.
MccV, MccL and McnN are natively produced by E. coli
while the remaining peptides are produced by Enterococcus
, a Gram-positive genus of bacteria [17
]. Like MccV, the studied enterocins are considered class II bacteriocins, which generally lack major post-translational modifications and are commonly secreted by ABC-transporters using N-terminal secretion tags [34
With the exception of EntP and HirJM79, all of the peptides tested herein are naturally encoded with glycine-glycine leader peptides [17
]. EntP and HirJM79 are believed to be secreted via the general Sec-type secretion pathways of their native producers [3
shows the percent identity and similarity of the signal peptides, the mature peptides and the two primary secretion genes associated with MccL, McnN and EntA compared to MccV. The EntB, HirJM79 and EntP transporter genes are not compared because they are either unknown as in the case of EntB or they belong to a different class of transporters [3
]. Identity and similarity values were calculated using the EMBOSS Needle (European Molecular Biology Open Software Suite version 188.8.131.52, Needleman-Wunsch global alignment application) global sequence alignment program with default parameters [38
Based on the alignment results, one can see that the signal peptides of all three AMPs share significant similarity to MccV’s signal peptide. These similarities are further discussed below. Importantly, the transporters for MccL and MccV are nearly identical despite the differences in the mature peptides.
shows an alignment of the secretion signal peptides and the first ten amino acids of the mature peptides. The first ten amino acids were included in the alignment because it has been previously hypothesized that the amino acids adjacent to the signal peptide may significantly impact secretion and processing [39
]. The multiple sequence alignment program Clustal Omega was used for alignment results [40
]. Amino acid conservation scores are based on the Gonnet PAM250 substitution matrix. Residues with a score >0.5 are considered highly conserved in similarity, and those with scores <0.5 are considered to have low conservation.
In addition to testing the capacity of the MccV secretion system with seven different AMPs, another objective of this study was to compare the effect of different signal peptides on active peptide secretion. Consequently, in addition to testing peptide production using the MccV signal peptide (denoted Vsp), we also tested the production of MccL, McnN and EntA using their own, native signal peptides. For EntA, the impact of fusing the MccL signal peptide to EntA (denoted LspA) was also tested because of the high level of similarity between MccV and MccL secretion tags and transporters.
Ultimately, the combination of distinct AMPs and of separate secretion signal tags resulted in eleven systems, listed in Table 2
In order to reduce the toxicity of the constructs to the producer strain, the immunity genes MccV, MccL and McnN were included in constructs encoding these peptides. Immunity genes were not included for HirJM79, EntP or EntB because these peptides were previously found to be inactive against the producer strain, E. coli MC1061 F’.
Agar diffusion assays with appropriate indicator strains were used to screen for peptide activity from the different constructs. Figure 3
shows the agar diffusion assay results of E. coli
Mc1061 F’ containing the nine AMP constructs listed in Table 2
, as well as the empty control, pMPES.
For these tests, 3 μL of overnight producer strain culture were spotted on agar containing the indicator strain and then incubated overnight. The white spots are the producer strain, and the dark regions surrounding the producer are zones of inhibition.
Inhibition is likely the result of expressed and secreted AMPs. We acknowledge that in this study, despite our efforts, no peptides were isolated and quantitatively measured as direct proof that inhibition was due to their production and secretion. This may be due to the small culture sizes used (40 mL), which may not produce sufficient peptide amounts for detection using canonical protein isolation methods. It may be useful to scale up culture sizes to one liter or more and to test alternative concentrating, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and purification methods for each of the tested systems, but this was beyond the scope of this study.
Nevertheless, all negative controls, which are included for all studies for activity comparison, suggest that antimicrobial activity is indeed the result of secreted AMPs. This conclusion that AMPs are expressed and secreted is further supported by the fact that in all cases, activity was observed only against the expected indicator strain. For example, MccL constructs were inactive against the Gram-positive E. faecium, whereas EntA constructs were inactive against the Gram-negative E. coli DH5α.
For these assays, MccV, MccL and McnN were tested against E. coli
DH5α (Figure 3
a) because it has previously been used as the indicator strain for MccV and McnN [13
]. EntA, HirJM79, EntP and EntB were tested against E. faecium
8E9 (Figure 3
b) because these peptides were previously shown to have activity against this strain [1
]. Note, E. faecium
8E9 is a multidrug-resistant pathogenic isolate.
MccL and MccV producers were also tested against foodborne pathogen E. coli
O157:H7 (Figure 3
c, top), and McnN producers were tested against foodborne pathogens Salmonella enterica
serovar Enteritidis and Salmonella enterica
serovar 4,,12:i:- (Figure 3
c, bottom) [41
In Figure 3
a, one can see clear zones of inhibition around E. coli
MC1061 F’, pMPES:V, L, VspL and VspN. These results strongly suggest that all of these strains are producing and secreting inhibitory levels of their respective peptides.
Similarly, results in Figure 3
b suggest the production and secretion of EntA, HirJM79, EntP and EntB using all signal peptides tested. The activity depicted in Figure 3
c demonstrates that the V, L and VspN constructs are potent enough to exhibit activity against pathogenic E. coli
In order to further validate the hypothesis that the peptides were in fact being secreted using the MccV secretion machinery, negative controls were made for VspA, L and N. For these controls, cvaA
were removed from the vector to abolish MccV secretion by digesting pMPES with XmaI (see Figure 1
) then relegating the digestion. Using agar diffusion tests, no activity was detected from any of these negative controls.
In the future, we aim to use this type of AMP production system to deliver peptides using probiotic bacteria. We therefore sought to test pMPES’s compatibility with probiotic E. coli
Nissle 1917. The pMPES:V, L and VspA constructs were transformed into E. coli
Nissle 1917 and tested using agar diffusion assays (Figure S2
). Definitive activity could be detected from all three constructs compared to the negative control, suggesting that the pMPES expression and secretion systems are compatible with Nissle. Figure S3
shows an additional agar diffusion assay, the results of which are consistent with those shown in Figure S2
. Note that, unlike E. coli
MC1061 F’, E. coli
Nissle 1917 naturally produces AMPs, Microcin H47 and Microcin M, which could account for the activity observed in the negative controls against E. coli
]. In Figure S3
, it appears that VspN shows a more defined halo than pMPES, implying, but not definitively determining activity.
2.4. Supernatant Activity Assays
To quantitatively compare the activities of the different constructs, we performed liquid supernatant inhibition assays. Figure 5
shows the growth curves of the two indicator strains in the presence of 75% supernatant from the constructs discussed above. These growth curves are averaged over three biological replicates. On the left, E. coli
DH5α growth curves are shown in the presence of supernatant from E. coli
MC1061 F’ pMPES (negative control), the MccV, MccL and McnN constructs. On the right, E. faecium
8E9 growth is shown in the presence of supernatant from pMPES, VspA, LspA, A, VspH, VspP and VspB constructs.
Note that V, L, VspL, VspA, LspA and VspH curves remain at OD600 = 0 on their respective graphs. No regrowth was observed in any of these cultures after 48 h, indicating complete killing of the indicator strains, which were originally inoculated at ~5 × 103 CFU/mL. In this study, an OD600 between ~0.05 and 1 is approximately linearly correlated with CFU/mL.
To further quantify the effectiveness of the constructs, the activity of the supernatants containing the AMPs was evaluated based on liquid growth assays. We stress that without the specific activities of the individual peptides, peptide production cannot be quantitatively compared across the different AMPs. Unfortunately, we were unable to isolate the peptides in any of the supernatants after multiple SDS-PAGE and HPLC attempts (see Materials and Methods). Additionally, it is important to consider peptide stability, particularly when comparing supernatant activity in which the calculated activity depends on peptide accumulation over several hours. Nevertheless the tests herein are potentially useful in comparing the efficacy of different signal peptides for a given AMP. In the case of the enterocins, which were tested against a pathogenic strain, these tests provide insight into which constructs may be the most potent for future application in probiotics.
reports the inhibitory activities of the supernatants in terms of Bacteriocin Units (BUs) [23
]. One BU is defined as the reciprocal of the highest dilution of supernatant required to reduce the growth of the indicator strain, with p
< 0.05 compared to growth with pMPES supernatant. For these studies, supernatant was diluted in 2x dilutions from 0.75 down to 0.0059 (1.3–170 BUs). Reported values are the average of three biological replicates. Error represents the standard deviation of these replicates. Note, a higher number of supernatant BUs indicates greater potency against the indicator strain.
also contains previously-reported minimum inhibitory concentrations (MICs) of the peptides in nM. We acknowledge however that these values were obtained by a variety of methods using different indicator strains than those used here. These values only are provided to give some idea of the potential activity of the different peptides, but should not be directly compared to this study.
Inhibition with p < 0.05 was observed for all supernatants tested in 75% supernatant. Interestingly, pMPES:L was more potent than pMPES:VspL, implying that the naturally-encoded signal peptide was more effective than Vsp for this particular AMP. This is in contrast to what was observed with EntA and McnN (see the growth curves and Figure 3
It should be noted that in multiple trials, VspA slightly out-performed LspA, but the difference is not observable in the BU calculations provided here.