Replacement of Trifluoroacetic Acid with HCl in the Hydrophobic Purification Steps of Pediocin PA-1: a Structural Effect

Abstract

Trifluoroacetic acid (TFA) is a purification contaminant associated with pediocin PA-1 that interferes with Fourier transform infrared spectroscopy structural analysis. As revealed by circular dichroism, its presence affects the structural folding of pediocin. Consequently, we propose a new pediocin PA-1 purification procedure using HCl instead of TFA in all of the hydrophobic steps. This procedural change does not affect the purification yield or the amount of pediocin PA-1 purified. Furthermore, removing HCl, as opposed to TFA, after purification is an easier procedure to carry out. In fact, the removal of TFA requires more experimentation and results in protein loss. Thus, HCl is a good alternative to TFA in pediocin PA-1 purification and can be extended to the purification of other proteins. We also show that TFA-induced structural modifications do not significantly affect the antimicrobial activity of pediocin PA-1.

For many years now, lactic acid bacteria (LAB) have been used in fermented foods in order to improve their organoleptic characteristics. Moreover, LAB are interesting because of their ability to improve food shelf life by inhibiting the growth of gram-positive bacteria and pathogenic species such as Listeria monocytogenes. This capacity is related to numerous metabolic end products (i.e., organic acids, hydrogen peroxide, and carbon dioxide) nutrient depletion, decreased redox potential, and bacteriocin production (24). Bacteriocins are bacterial proteins or peptides that inhibit strains and species that are generally closely related to the producing bacteria (38). Their antimicrobial activity seems to be related to their ability to disrupt the cytoplasmic membranes of bacterial cells. Four main classes of bacteriocins can be produced by LAB (25). Classes I and II are well characterized. They consist of small membrane-active, heat-stable peptides. The class I bacteriocins are called lantibiotics because of the presence of the thioether amino acids lanthionine and 3-methyl-lanthionine, which result from posttranslational modification of the propeptide (12). Class II corresponds to small (<10-kDa), heat-stable (at the pH of production) peptides without atypical amino acids, exhibiting both hydrophobic and cationic characteristics. This class is made up of three subclasses. Subclass IIa is characterized by the pediocin-like or antilisterial consensus sequence YGNGVXC in the N-terminal part. Class IIb bacteriocins need two polypeptide chains in order to be biologically active. Class IIc bacteriocins are dependent on the general secretory pathway of the cell and are thiol activated. Class III bacteriocins are high-molecular-mass (>30-kDa), thermolabile proteins (25). Class IV has not been well characterized at the biochemical level.

Because they can kill food pathogens and they are found in the food environment, these bacteriocins have promising potential as food grade preservatives. The lantibiotic nisin is the most studied bacteriocin and the only one presently being used. Pediocin PA-1 is a class IIa bacteriocin. Its antimicrobial activity was shown to be effective against 48 food-borne pathogens and 28 LAB (22). Pediocin PA-1 is particularly active against the psychrotrophic pathogen L. monocytogenes, which is killed in laboratory broth cultures and in model food systems, especially in meat products where nisin fails to work (30, 33). Its food preservative abilities are interesting for industrial application. Its primary structure consists of 44 amino acids (molecular mass, 4,629 Da; pI, 9.6) (17, 22). This bacteriocin is produced by Pediococcus acidilactici PAC 1.0, which is naturally found in fermented sausages (22) and other meat and vegetable fermentations (6). In order to obtain required Food and Drug Administration approval, complete chemical, physical, and genetic characterization of potentially new antimicrobials is necessary (22). Consequently, it is very important to obtain pure bacteriocin to test it at the biochemical level before its use in food preservation.

Given that pure bacteriocin must be active to be tested, purification procedure should be performed under gentle conditions and maintain the structural integrity of the protein. Indeed, protein activity is a complex phenomenon, dependent on the correct structural folding of the protein. The folded conformation can be disrupted by environmental changes that do not involve variations in the covalent structure (10). These environmental changes may be caused by pH, ionic strength, and temperature. Some drastic variations may induce irreversible changes in protein conformation. It is therefore very important to use a purification procedure that does not affect the native structural folding of the protein. Given that trifluoroacetic acid (TFA) was detected in apparently pure bacteriocin preparations and that its presence could influence protein structure, the purpose of this study was to find a new pediocin PA-1 purification procedure using HCl rather than TFA in all of the hydrophobic step. Our results indicate that HCl is a good alternative to TFA, because it does not affect the purification yield, it is easily removed, and it has no effect on the structural characterization of the protein, as determined by circular dichroism (CD) and Fourier transform infrared spectroscopy (FTIR).

MATERIALS AND METHODS

Pediocin PA-1 purification.

The pediocin PA-1-producing strain P. acidilactici PAC 1.0 (a kind gift of Quest International, Sarasota, Fla.) was grown in lactobacillus MRS broth (Rosell Laboratories, Montréal, Québec, Canada) at 37°C for 20 h, and the sensitive strain Listeria innocua ATCC 33090 was grown in Trypticase soy broth (Difco) enriched with 0.3% yeast extract for 18 h at 37°C. Culture supernatant was obtained by centrifugation (8,000 × g, 15 min) to remove cells.

Pediocin PA-1 was purified by two methods. The first method is a procedure adapted from that described by Guyonnet et al. (16). Briefly, the cell-free supernatant was directly deposited on a sulfopropyl (SP)-Sepharose column (Econosystem Bio-Rad, Hercules, Calif.). The column was washed with 5 mM ammonium acetate (pH 5) supplemented with 0.25 M NaCl. The elution was performed with a linear gradient from 0.25 to 0.5 M NaCl in the same buffer. The absorbance was monitored at 280 nm. Active fractions were collected and desalted onto a Sep-pack C₁₈ column (Cartridge 10 mg; Waters), where they were eluted with 20 to 50% acetonitrile-0.11% TFA-high-pressure liquid chromatography (HPLC) water. After removal of the acetonitrile by evaporation, the preparation was loaded on a reverse-phase HPLC (Beckman Instruments Inc., Mississauga, Canada) by using a C₁₈ hydrophobic affinity column. The column was maintained at 39°C with a column heater. After equilibration of the column with solvent A (0.11% TFA), the peptides were eluted by increasing the concentration of solvent B (60% acetonitrile, 0.11% TFA). Fractions were directly collected in Centricon tubes (Millipore Corporation, Bedford, Mass.; molecular weight, 3,000) in order to concentrate the pure pediocin. Retentates were dried in a vacuum centrifuge (Savant Instrument Inc., Hicksville, N.Y.) and stored at −80°C under an N₂ atmosphere.

The second method was essentially the same except that TFA was replaced by 5 mM HCl in all hydrophobic steps.

FTIR analyses.

Samples were prepared by dissolving 1% pediocin PA-1 in deuterated phosphate buffer (0.1 M) at pD 6 (pD = pH + 0.4). Infrared spectra at 2-cm⁻¹ resolution were recorded with a Magna 560 spectrometer (Nicolet Instrument Corporation, Madison, Wis.) equipped with a mercury-cadmium-telluride detector. The sample chamber was continuously purged with dry carbon dioxide-free air. Pediocin PA-1 spectra were recorded in a cell with CaF₂ windows separated by 23-μm polyethylene terephtalate film spacers. The spectrum of pediocin PA-1 is the result of the average of 128 scans and is apodized with a Happ-Genzel function. For the study of the amide I region of the protein, subtraction of the buffer spectrum contribution and Fourier self-deconvolution with a resolution enhancement of 2 and a bandwidth of 18 cm⁻¹ was performed by the software provided with the spectrophotometer (Omnic software; Nicolet Instrument Corporation). Water vapor subtraction was carried out when necessary.

CD analyses.

CD spectra were recorded using a Jasco J-710 spectropolarimeter. Measurements were performed at 20°C, using a 0.05-cm-path-length quartz cell from 250 to 190 nm and at least 20 scan accumulations. A protein-free control spectrum was recorded for each condition and subtracted from the protein spectra. Pediocin PA-1 was dissolved at 0.5 mg/ml in TFA-water or HCl-water buffer at the same pH. The pH of a pure fraction of pediocin PA-1, purified in the presence of TFA and dissolved in water, is 2.2.

Bacteriocin activity and protein assay.

During purification, the bacteriocin activity was measured by an agar spot test on Trypticase soy agar-0.3% yeast extract medium (32). Serial twofold dilutions of each bacteriocin preparation were spotted onto fresh indicator lawns of L. innocua ATCC 33090. These lawns were prepared by propagating fresh culture to an optical density at 600 nm of 0.1 and adding 100 μl of the 100-fold-diluted cell suspension to 5 ml of overlay agar. The plates were then incubated for 12 h at 37°C. Activity was defined as the reciprocal of the highest dilution demonstrating complete inhibition of the indicator lawn and was expressed in activity units milliliter of culture medium⁻¹.

In order to measure bacteriocin activity in the presence of TFA, 50 μg of pure pediocin was dissolved in 50 μl of H₂O. For the sample without TFA, the pH was adjusted to 2.2 with HCl in order to be the same as that of 50 μg of pediocin PA-1 purified with TFA. Serial twofold dilutions of pediocin PA-1 preparations were performed in microtiter plates, and 50 μl of a 10⁻⁴ dilution of fresh culture (optical density of 0.1) of L. innocua ATCC 33090 was added. This preparation was incubated for 30 min at 37°C. After incubation, 50 μl from each well was plated on Trypticase soy agar-0.3% yeast extract medium and incubated for 18 h at 37°C. Surviving cells were counted, and the dilution corresponding to 50% lethality was used to determine the lethal units.

Protein analyses were performed with a bicinchoninic acid protein assay kit (Pierce, Rockford, Ill.) according to the directions provided with the kit. Bovine serum albumin (Pierce) was used as the standard.

RESULTS

Purification of pediocin PA-1.

Three peaks were observed in the HPLC chromatogram of the last purification step in the first procedure, i.e., TFA in the HPLC buffer (Fig. 1A). The major peak corresponds to protein with antimicrobial activity. The HPLC chromatogram of the same purification step obtained with the second procedure (i.e., HCl) is comparable (Fig. 1B). The same peaks are present in similar quantities and sizes. The major difference is in the retention times, which are shorter when HCl is used. This is due to the fact that HCl is less acidic than TFA. The purification results obtained for pediocin PA-1 by using the first (i.e., TFA) and the second (i.e., HCl) procedures are shown in Table 1. The data reveal that there are no significant differences between these procedures (i.e., TFA and HCl) in purification yield or in the amount of pediocin PA-1 purified. The amount of purified bacteriocin is 2 mg for 1 liter of supernatant. The purification yield is 38%.

FIG. 1.

Chromatograms of the final purification step. A semipreparative C₁₈ reverse-phase column was used to isolate pediocin PA-1 after cation-exchange chromatography in the presence of 0.1% TFA (A) and 5 mM HCl (B). OD, optical density.

TABLE 1.

Purification of pediocin PA-1 during TFA and HCl procedures^a

Sample	Activity recovery (%)b	Purification (fold)b
Supernatant	100	1
SP-Sepharose	68.9	542
Sep-pak	39.7, 39.67	2,358, 2,366
C18 column	38, 38.1	2,665, 2,660

^aProtein quantity was estimated by the bicinchoninic acid method, and activity was tested against L. innocua ATCC 33090.

^bWhere two numbers are given, the first is for TFA purification and the second is for HCl purification.

Structural results with FTIR.

FTIR was used to study the molecular structure of pediocin. Figure 2 shows the original spectrum of pediocin PA-1, purified in the presence of TFA, in the amide I′ band region. This band (called amide I′ in D₂O) corresponds mainly to the C=O stretching mode of the peptide backbone and is sensitive to changes in the secondary structures (4, 15, 21, 26). It is composed of several overlapping components attributed to specific secondary structures such as helices, β-sheets, turns, and nonordered structures. The original spectrum of pediocin (Fig. 2) has three components: a strong one located at 1,672 cm⁻¹, another one at 1,641 cm⁻¹, and a shoulder at 1656 cm⁻¹. The strong band, which has also been observed in other studies (21, 35, 37), arises from the counterion trifluoroacetate (Fig. 2). Its strong absorbance in the amide I region makes it impossible to carry out a more detailed analysis and to obtain any information about pediocin PA-1 secondary structure.

FIG. 2.

FTIR spectra of the amide I′ region of pediocin PA-1 purified with 0.1% TFA (thick line) and of TFA (thin line). a.u., arbitrary units.

Figure 3 displays the original and deconvoluted FTIR spectra of pediocin PA-1 purified in the presence of HCl. The deconvoluted spectrum, resulting from a mathematical treatment of the original spectrum, allows for a better separation of the amide I′ band into its components and a more detailed interpretation. The amide I′ band maximum occurs at 1,645 cm⁻¹, which is characteristic of proteins and peptides with little or no well-defined secondary structures (3, 37). The spectrum also has a shoulder around 1,673 cm⁻¹ due to β-turn structures and another one at 1,611 cm⁻¹, characteristic of amino acid residues (15). These results suggest that pediocin PA-1 is composed mainly of unordered structures with a small contribution of β-turns. The contribution of amino acid side chain vibrations, such as Asn located at 1,648 cm⁻¹ in D₂O medium, is negligible, since it represents only 5 to 15% of the amide I region (11, 20).

FIG. 3.

Native spectrum (thick line) and deconvoluted spectrum (thin line) of the amide I′ region of pediocin PA-1 purified with 5 mM HCl. a.u., arbitrary units.

The original spectrum of pediocin PA-1 purified in the presence of HCl (Fig. 3) is different from that obtained for the protein purified in the presence of TFA (Fig. 2), even though a detailed comparison cannot be done because of the strong interference of TFA.

Structural results obtained by using CD.

Figure 4 shows the far-UV CD spectra of pediocin PA-1 at pH 2.2 adjusted with TFA or HCl. CD spectra, based on the dependence of optical activity of protein between 190 and 240 nm (2, 23), are not affected by the presence of TFA, unlike FTIR data, which are based on chemical component vibrations. Thus, CD provides information on the secondary structure that is complementary to FTIR data. The two CD spectra display a well-defined minimum located at 198 nm for pediocin PA-1 in TFA and at 199 nm pediocin PA-1 in HCl, suggesting that in both cases, proteins adopt a random-coil conformation (2, 7). The spectra show a second minimum between 200 and 230 nm, centered at 218 nm, which is characteristic of a helical structure; this minimum was more pronounced when TFA was used. Moreover, there is a positive dichroic power at 232 nm for pediocin PA-1 in the presence of HCl and a negative one in the presence of TFA.

FIG. 4.

Effect of TFA and HCl on the CD spectra of pediocin PA-1 in aqueous solution. The spectra were recorded in the wavelength range of 190 to 300 nm.

These results suggest that, despite having a similar overall conformation composed of unordered structures, pediocin presents slightly different structures depending on whether TFA or HCl is used.

Activity results.

Since conformational differences could have functional implications, we analyzed the effect of TFA on pediocin PA-1 activity. To confirm that the effect observed is caused only by TFA and not by a pH effect against L. innocua ATCC 33090, the activity tests were done with the same samples as those used in the CD analyses. Although a slight decrease in activity was observed when TFA was used, it was not significant (data not shown).

DISCUSSION

TFA is a common solvent in reverse-phase HPLC protein purification (6, 16, 17, 31) because of its effectiveness in solubilizing hydrophobic peptides (36). This effect is mainly due to its high acidity. However, although it is a volatile component, it remains in samples after drying, as revealed by infrared data. This suggests that it is strongly bound to the protein. Since it can kill eucaryotic cells (9), it must be eliminated from the sample, especially if pediocin is intended to be used as a food preservative. Many authors working on synthetic peptides produced in the presence of TFA remove it from their samples by washing proteins in dialysis membranes, by several steps of freeze-drying of the peptide in 0.1 M HCl, or by washing it away with trifluoroethanol (8, 18, 19, 27, 34). This strategy is not satisfactory, because it results in more experimental steps. Indeed, a protein purification procedure must be rapid and use as few steps as possible in order not to lose or affect the protein. Moreover, the use of dialysis membranes in bacteriocin purification reduces protein yield by coating the membrane with protein. In the light of these facts, another method is proposed. Since acidic conditions are needed in the hydrophobic steps of purification, we propose to replace TFA with HCl.

Our results demonstrate that the use of HCl as a replacement for TFA in the purification procedure does not affect pediocin PA-1 recovery, which was 38% (i.e., 2 mg of pediocin PA-1 for 1 liter of supernatant). Our findings agree with those of Guyonnet et al. (16), from whose method the technique was adapted. By using TFA in the HPLC buffer, they obtained a purification yield of 25%, which corresponds to 1.4 mg of purified pediocin PA-1 for 1 liter. Our slightly higher yield may result from the use of a stronger cationic exchanger (SP-Sepharose instead of carboxymethyl-Sepharose) and a more hydrophobic column (C₁₈ instead of C₈). Nevertheless, this result suggests that the same amount of pediocin PA-1 was present in the initial medium (i.e., around 5.4 mg). Several authors (5, 28, 29, 41) have reported intermediate or final yields of recovery activity of above 100% for bacteriocin purification. This odd result may be attributed to the presence of different reagents (sulfate ammonium salt or Tween 80, etc.) which modify the activity of bacteriocins, resulting in an increase in the recovery yield (16). The absence of these reagents in our procedure does not induce such an apparent increase of pediocin PA-1 activity. Thus, our purification procedure is gentle for the protein, and the use of HCl in all hydrophobic steps does not affect pediocin PA-1 recovery. Moreover, HCl can be removed easily from the sample by using a low-coating filtration membrane, which makes it possible to obtain contaminant-free pediocin.

A second major point, which underlies the necessity to remove TFA from protein purification, is that TFA is bound to the protein and thus modifies its conformation. CD data reveal that TFA induces an increase in α-helix structure, even though pediocin is composed mainly of unordered structures. These results can be interpreted in two different ways: either the conformational change is caused by TFA, or it is caused by a pH effect, as proposed recently (1). Consequently, pediocin was solubilized under the same acidic conditions in the presence or absence of TFA, and the CD spectra were recorded. This leads to the conclusion that TFA-protein linkage alone was responsible for these differences. However, it is interesting that these structural changes have no significant effect on pediocin PA-1 activity. This result is not surprising, because we recently showed (H. Gaussier and M. Subirade, unpublished data) that pediocin PA-1 needs to have a flexible structure in order to be active and because the CD data obtained in this study reveal that pediocin PA-1 is composed mainly of flexible, unordered polypeptide chains. The same behavior has been proposed for pediocin AcH (the same molecule as PA-1) on the basis of CD (40). Such a structural behavior seems to be characteristic of many class II bacteriocins, such as leucocin A (14), carnobacteriocin B2 (39), and mesentericin Y105 (13).

In conclusion, it is clear from ours results that TFA, a contaminant of protein purification, affects the secondary structure of pediocin. Its presence interferes with FTIR structural analysis, and CD spectral data indicate that it induces a small increase in helical structures. However, this small structural effect does not affect protein flexibility and thus protein activity. Moreover, its capacity to decrease pH, in any buffer, induced difficulties in a rigorous study on pH dependence of structure-function relationships. Its replacement with HCl in all of the hydrophobic steps of pediocin PA-1 purification seems to be a good alternative. Indeed, HCl does not affect the purification of pediocin PA-1 or its conformation, and it can be easily removed with a filtration membrane. This method can be extended to the purification of other proteins.