Introduction

Lipopeptide biosurfactants produced by members of the genera Bacillus, Pseudomonas, Streptomyces, and Arthrobacter as secondary metabolites [1] are amphiphilic molecules with excellent surface activity and antifungal or antibacterial activities. Bacillus sp., such as Bacillus subtilis [2, 3], Bacillus cereus [4], and Bacillus amyloliquefaciens [5], have been reported as major producers of lipopeptide biosurfactants. Several Bacillus sp. strains have been screened to study their lipopeptide products for potential applications in the last few decades [6–8].

The three most well-known families of lipopeptides are the iturins [9, 10], the fengycins [11, 12], and the surfactins [13]. In addition, some other molecules [14–17], such as kurstarkin, mycosubtilin, bacillomycin, and lichenysin, with similar structures have also been identified. Surfactin contains a cyclic lactone ring consisting of a C12–C16 β-hydroxy fatty acid and a heptapeptide with a variable amino acid at positions 2, 4, and 7 [18]. Iturin is a lactam containing a C14–C17 β-amino fatty acid attached to a heptapeptide with Asp or Asn at position 1 [19]. Fengycin is composed of a β-hydroxy fatty acid chain attached to a decapeptide, which also forms a cyclic lactone ring. Two variants, fengycin A and fengycin B, have been reported with Val or Ala at position 6 [20, 21]. These lipopeptides are synthesized by nonribosomal peptide synthetases [22, 23], which can generate many variants that differ in the fatty acid chain length and amino acid sequence. Some microorganisms can simultaneously produce different families of lipopeptides. These two factors result in the production of complex mixtures of microbial lipopeptide products containing isoforms of iturins, fengycins, and surfactins, among others.

Acid precipitation is the most widely used method to primarily separate lipopeptides from cell-free fermentation broth on a laboratory scale [24]. The long fatty acid chains contribute to high solubility in organic solvents. Therefore, acid precipitation is usually followed by extraction using methanol, dichloromethane, chloroform, or ethyl acetate. Both chloroform–methanol in ratio of 2:1 [25, 26] and methanol alone [20] are commonly used for the extraction of lipopeptides. In consideration of the high toxicity of chloroform, we chose methanol alone in the extraction.

Reverse-phase (RP) high-performance liquid chromatography (HPLC) is highly suitable for the purification of molecules with different hydrophilicity and has been applied to discriminate between homologs of lipopeptides. The different amino acids with diverse hydrophilic characteristics will interact with the mobile phase, and the fraction with higher hydrophilicity in the peptide ring will be eluted at a lower proportion of methanol/acetonitrile. On the other hand, the hydrophobic fatty acid chains will combine with the modified filler of a C18 column; thus, the fraction containing the longer fatty acid will be harder to elute from the column and a higher proportion of methanol/acetonitrile is required. Iturins, fengycins, and surfactins can be eluted by 40–50 % [27], 50–70 % [12], and 85–100 % [1] acetonitrile in water, respectively, on a C18 column in an RP-HPLC system.

Precolumn derivatization of amino acids with phenyl isothiocyanate (PITC) was developed in the 1980s, and proved to have high sensitivity, excellent derivative stability, and high-resolution separation of amino acids [28]. For example, the derivative loss was less than 5 % in 2 days at 4 °C and the response was linear over the range 10–2,000 pmol [29]. The regression equations with a correlation coefficient above 0.999 were obtained from chromatograms of PITC derivatives of 18 standard amino acids [28, 30].

Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) with its high sensitivity and high sample throughput is well suited for the rapid measurement of the molecular masses of intact bioactive peptides. Glycopeptides and lipopeptides (mass below 5,000 Da) containing free amino acid groups can be protonated efficiently using α-cyano-4-hydroxycinnamic acid as the matrix, and these compounds frequently give strong signals in spectra [31]. Surfactins and iturins in the m/z range 1,000–1,100 and fengycins in the m/z range 1,400–1,600 [3, 12, 13, 21, 32] can be detected by MALDI-TOF-MS in single protonated forms or as Na+ or K+ adducts in the m/z range 800–2,000. There is no report about multiple adducts (dimers or trimers) of iturin, fengycin, and surfactin families. Further determination of the amino acid sequence in the peptide relies on collision-induced dissociation (CID) in MS/MS experiments.

To identify different lipopeptide isoforms, effective separation and purification by RP-HPLC in a short time is important. Most studies have focused only on the separation of homologs of a certain family of lipopeptides with a property that meets their requirements. A few studies have detected two or three different lipopeptide families using a linear gradient [3, 12], but the elution time was too long and the isoform fractions were not thoroughly separated.

In this work, we optimized an efficient three-stage RP-HPLC strategy allowing simultaneous good separation of several iturin, fengycinand surfactin isoforms, which may differ in composition by only a single amino acid and/or the fatty acid residues. Subsequently, the identification of the amino acid sequence of peptides from their HPLC fractions was performed using MALDI-TOF-MS/MS assisted by amino acid analysis based on PITC derivatives. This systematic approach facilitates the simultaneous rapid discovery, high-quality purification, and molecular structure identification of isoforms of the iturin, fengycinand surfactin families of molecules from lipopeptide mixtures.

Materials and methods

Identification of the isolated strain

The strain used in this work, B. subtilis THY-7, was previously isolated from the sludge of a lotus pond at Tsinghua University. A 16S ribosomal RNA (rRNA) sequence (1,465 bp) was amplified from the strain by the primer pairs [33] 5′-AGAGTTTGATCCTGGTCAGAACGCT-3′ and 5′-TACGGCTACCTTGTTACGACTTCACCCC-3′. PCR amplification was performed in a total volume of 50 μL containing 0.5 μL template DNA, 5 μL Ex Taq reaction buffer, 4 μL dNTP (each at 2.5 mM), 1.5 U Ex Taq DNA polymerase and 1 μL of each primer (20 μM). The amplified 16S rRNA gene sequence was compared with ribosomal DNA sequences obtained from GenBank (http://www.ncbi.nlm.nih.gov/BLAST/).

Preparation of lipopeptide mixture from B. subtilis

A 5 % inoculum of B. subtilis THY-7 in Luria–Bertani medium was added to the fermentation medium (1 L: 60 g glucose, 1.0 g yeast extract, 25 g NaNO3, 0.333 g KH2PO4, 1.0 g Na2HPO4 · 12H2O, 0.15 g MgSO4 · 7H2O, 0.0075 g CaCl2, 0.006 g MnSO4 · H2O, 0.006 g FeSO4 · 7H2O, pH 7.0), and the culture was incubated for 48 h at 37 °C with shaking at 200 rpm. The fermentation broth was centrifuged at 12,000 g for 20 min. The pH of the supernatant was adjusted to 2.0 with 6 M HCl and was then stored overnight at 4 °C. The sample was centrifuged again at 12,000 g for 20 min to harvest the solid crude lipopeptides and then extracted with methanol. Insoluble impurities were removed by filtration, and the methanol was evaporated under a vacuum at 50 °C using a rotary evaporator to obtain the crude lipopeptides.

RP-HPLC

The lipopeptide mixture was dissolved in an aqueous solution of NaHCO3 at 1 g/L (the lipopeptide mixture at 2 g/L) and filtered through a 0.2 μm membrane filter. A 20 μL aliquot was injected into an InertSustain C18 column (5 μm, 250 mm × 4.6 mm) in the HPLC system to separate the isoforms. The 500 mg/L standard surfactin and iturin (purity of 98 % or greater, Sigma-Aldrich, Shanghai, China) were used to confirm the RP-HPLC fraction groups of three lipopeptide families. The mobile phases were water (A), methanol (B1), and acetonitrile (B2). All of the mobile phases contained 0.1 % trifluoroacetic acid (TFA). Optimization of the RP-HPLC strategy was done stepwise by adjusting the gradient program for the mobile phases, which focused on the concentration range of methanol/acetonitrile and the time of the gradient. The detailed gradient strategies used in the optimization are given in the figure captions. The final gradient strategy for the methanol–water mobile phase system was as follows: 0–3 min, 70 % methanol to 75 % methanol; 3–8 min, 75 % methanol to 85 % methanol; and 8–30 min, 85 % methanol to 95 % methanol. The final optimized gradient strategy for the acetonitrile–water mobile phase system was as follows: 0–3 min, 45 % acetonitrile to 50 % acetonitrile; 3–8 min, 50 % acetonitrile to 80 % acetonitrile; 8–25 min, 80 % acetonitrile to 100 % acetonitrile. The total flow rate of the mobile phases was kept at 0.8 mL/min, and the chromatograms were obtained at 205 nm. The fractions were harvested manually for amino acid analysis and MALDI-TOF mass spectra analysis.

Amino acid analysis

Each fraction was harvested from its HPLC fraction using the vacuum freeze-drying method. Each fraction (3.0 mg) was hydrolyzed with 1 mL of 6 M HCl at 110 °C for 24 h in a sealed tube. The fatty acid residue was extracted three times using 1 mL ether, and the aqueous phase of the hydrolysis solution was dried at 50 °C under a vacuum to remove residual ether and HCl. Then, the dried sample was redissolved in 10 mL double-distilled water.

The analysis of amino acids was based on a precolumn derivatization with PITC [30, 34, 35]. Eighteen amino acids (Sigma-Aldrich, Shanghai, China) were used for standard reference: Asp, Glu, Ser, Gly, His, Arg, Thr, Ala, Pro, Tyr, Val, Met, Cys, Ile, Leu, Phe, Trp, and Lys. A 200 μL aliquot of the 18 standard amino acids (each at 30 mg/L) or the redissolved hydrolysis solution was mixed with 100 μL of 0.1 M PITC and 1 M triethylamine in acetonitrile after 10 μL of a 0.5 g/L solution of norleucine had been added as an internal standard for quantitative calculation. Then, the mixture was vortexed for 1 min and incubated for 30 min at room temperature. After derivatization, 400 μL n-hexane was added and the mixture was vortexed for 1 min and allowed to stand for 5 min. The bottom acetonitrile phase was filtered through a 0.22-μm polytetrafluoroethylene membrane filter before 6 μL (2 μL for the standard amino acid mixture reference) was injected onto an InertSustain C18 column (5 μm, 4.6 mm × 250 mm) at 38 °C. The derivative was stored at 4 °C if it was not analyzed within 2 h when a large number of samples were derived at a time, because the derivative stability is excellent, with less than 5 % loss in 2 days at 4 °C [28]. The mobile phases were as follows: 0.1 M sodium acetate (pH 6.5, adjusted with acetic acid)–acetonitrile (A; 93:7, v/v), and acetonitrile–water (B; 4:1, v/v). The proportion of the mobile phases was controlled using the following gradient program: 0 min, 0 % mobile phase B; 13 min, 7 % mobile phase B; 23 min, 23 % mobile phase B; 29 min, 35 % mobile phase B; 35 min, 40 % mobile phase B; 40 min, 100 % mobile phase B; 45 min, 100 % mobile phase B; 50 min, 0 % mobile phase B; and 60 min, 0 % mobile phase B. The mobile phase was kept at a flow rate of 1.0 mL/min, and the chromatograms were obtained at 254 nm.

MALDI-TOF-MS and MALDI-TOF-MS/MS analysis

MALDI-TOF analysis was used to elucidate the structure of HPLC-purified lipopeptides using a SCIEX 4800 Plus analyzer (Applied Biosystems, Foster City, CA, USA) with a 337-nm nitrogen laser for desorption and ionization. An equal volume of 0.1 % α-cyano-4-hydroxycinnamic acid was used as the matrix in a matrix-to-analyte ratio between 1:1 and 10:1. Mass spectra were accumulated over 50 individual laser shots and were obtained in reflector mode (mass accuracy 0.2 Da or less, sensitivity: 100 amol neurotensin, signal-to-noise ratio of 10:1 or greater) at an initial accelerating voltage of 20 kV [31, 36]. The m/z values were measured in the range from 800 to 2,000.

MALDI-TOF-MS/MS coupled with CID with the same spectrometer was used to analyze the fragment ions of peptides for further characterization of the amino acid sequence. The precursor ions submitted to MS/MS experiments were selected by the MS1 set and then focused in a collision cell. The collision cell was floated at 2 kV to attain a collision energy of 2 keV. Air used as the collision gas was introduced at a pressure leading to an attenuation of the precursor ion beam by almost 70 %.

Results

Optimization of RP-HPLC separation of isoforms from a mixture of three lipopeptide families

The 16S rRNA sequence of B. subtilis THY-7 was submitted and deposited in GeneBank (accession no. KJ777688.1). Comparison of the sequence in GenBank revealed 100 % identity with B. subtilis H12 (accession no. KC441785.1). After fermentation for 48 h, we harvested lipopeptides by acid precipitation, methanol extraction, and rotary evaporation for the further RP-HPLC separation.

Optimization of the RP-HPLC strategy was done stepwise by adjusting the gradient program for the mobile phases (A, water; B1, methanol; B2, acetonitrile; all containing 0.1 % TFA). The flow rate of the mobile phases was kept at 0.8 mL/min.

Methanol and acetonitrile are widely used in RP-HPLC mobile phase systems. In regard to the most important property of polarity, methanol is similar to acetonitrile. In consideration of the toxicity, we firstly chose the methanol–water mobile phase system to separate and purify the lipopeptide isoforms. By adjusting the ratio of methanol in the mobile phase and the gradient program, we were able to control the retention times of the isoforms with different hydrophilic peptide rings and lipophilic fatty acid chains, as displayed in Fig. 1. The optimal chromatogram, displayed in Fig. 1d, demonstrated that the use of a methanol–water system with an appropriate strategy could realize relatively good separation of the iturin and surfactin isoforms, but was inadequate for the fengycin family.

Fig. 1
figure 1

Reverse-phase (RP) high-performance liquid chromatography (HPLC) chromatograms with different elution strategies in the methanol–water system. a 0–3 min, 50 % methanol to 55 % methanol; 3–13 min, 55 % methanol to 85 % methanol; and 13–28 min, 85 % methanol to 100 % methanol. b 0–3 min, 65 % methanol to 75 % methanol; 3–13 min, 75 % methanol to 85 % methanol; and 13–30 min, 85 % methanol to 100 % methanol. c 0–3 min, 75 % methanol to 80 % methanol; 3–10 min, 80 % methanol to 85 % methanol; and 10–30 min, 85 % methanol to 95 % methanol. d 0–3 min, 70 % methanol to 75 % methanol; 3–8 min, 75 % methanol to 85 % methanol; and 8–30 min, 85 % methanol to 95 % methanol

Full size image

In most cases, the eluting capacity of acetonitrile is better than that of methanol. Therefore, we further explored the acetonitrile–water system as the mobile phase. The initial strategy was a linear gradient that increased from 30 to 100 % acetonitrile in water containing 0.1 % TFA for 60 min at a flow rate of 0.8 mL/min, similar to what was reported in the literature for lipopeptide antibiotics [3, 12]. Although the isoforms were separated (as shown in Fig. 2a), the elution time was too long (about 50 min). There were large time intervals between the groups, especially before the first lipopeptide component of itruin was eluted. Additionally, the second group of fengycin was too flat, indicating insufficient separation. Thus, it was necessary to reduce the retention time and adjust the gradient program systematically.

Fig. 2
figure 2

RP-HPLC chromatograms using different elution strategies in the acetonitrile–water system. a The initial strategy, eluting with a linear gradient increasing from 30 % acetonitrile to 100 % acetonitrile for 60 min. b Strategy I, the gradient program was 50 % acetonitrile to 75 % acetonitrile from 0 to 10 min and then 75 % acetonitrile to 100 % acetonitrile from 10 to 20 min. c Strategy II, the gradient program was as follows: 40 % acetonitrile from 0 to 3 min; 40 % acetonitrile to 50 % acetonitrile from 3 to 8 min; 50 % acetonitrile to 80 % acetonitrile from 8 to 13 min; and 80 % acetonitrile to 100 % acetonitrile from 13 to30 min. d The final optimized efficient three-stage strategy was as follows: 45 % acetonitrile to 50 % acetonitrile, from 0 to 3 min; 50 % acetonitrile to 80 % acetonitrile from 3 to 8 min; and 80 % acetonitrile to 100 % acetonitrile from 8 to 25 min

Full size image

Strategy I significantly reduced the initial run time by 30 min, and all of the active fractions in the initial chromatogram were observed, as shown in Fig. 2b. However, the first fraction was eluted too early, and thus homologs with shorter fatty acid chains would be eluted earlier and mixed with pigments before 5 min. Furthermore, the second fraction group was still too flat.

The result of strategy II indicated that the elution ability of 40 % acetonitrile during the first 3 min was not enough. However, the second group of fractions were concentrated (Fig. 2c).

The final optimized three-stage elution strategy resulted in a significant reduction of run time, from 60 to 28 min, with good separation of all of the isoforms in the three families. The retention times were in the ranges of 6–10 min, 10–16 min, and 16–25 min, corresponding to the iturin, fengycin, and surfactin families of molecules, respectively, as shown in Fig. 2d. Twelve major fractions, which were eluted at retention times of 7.52, 8.89, 12.40, 12.92, 13.34, 13.84, 18.65, 19.84, 20.12, 21.06, 21.40, and 22.46 min, were marked as fractions 1–12 and were collected for MALDI-TOF-MS and amino acid analysis.

MALDI-TOF-MS of iturin, fengycin, and surfactin isoforms

Intense signals in the m/z ranges 1,000–1,100 and 1,400–1,600 were obtained in the MALDI-TOF-MS spectra of the lipopeptide mixture. In detail, the ions at m/z 1,044 and 1,058 and at m/z 1,065 and 1,079 were hypothesized to be surfactins and iturins, respectively, whereas the ions at m/z 1,485, 1,499, and 1,505 belonged to the fengycin family, as illustrated in Fig. 3a. MALDI-TOF-MS of the RP-HPLC fractions revealed a series of ions in the m/z range from 1,000 to 1,600. Peaks 1 and 2 in the first HPLC group were standard iturins containing C14 and C15 fatty acid chains, with series of H+, Na+, and K+ adduct ions at m/z 1,043, 1,065, and 1,081 (Fig. 3b) and at m/z 1,057, 1,079, and 1,095 (Fig. 3c). The mass spectra of HPLC peaks 3–6 in the subsequent group mainly exhibited the [fengycin + H]+ ions containing C16 and C 17 fatty acid chains at m/z 1,495, 1,463, 1,477 (Fig. 3d), and 1,505 (Fig. 3e). The mass spectra of HPLC peaks 7 and 9 showed the [surfactin + Na]+ ions at m/z 1,030 (Fig. 3f), the mass spectra of HPLC peak 8 and 11 showed the [surfactin + Na]+ ions at m/z 1,044, and the mass spectra of HPLC peaks 10 and 12 showed the [surfactin + Na]+ ions at m/z 1,058 (Fig. 3g) and 1,072, respectively. The MALDI-TOF-MS spectra of the other fractions are presented in Figs. S1–S6.

Fig. 3
figure 3

Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) spectra of the lipopeptide mixture and isoforms in HPLC peaks. a High mass intensity in the m/z range 1,000–1,100 proposed to contain surfactin and iturin ions, and in the m/z range 1,400–1,600 proposed to contain fengycin ions. b, c Iturins related to fractions 1 and 2 displayed a high mass intensity at m/z 1,043 and 1,057 in their protonated forms and at m/z 1,065 and 1,079 as the sodium adducts. d, e Protonated fengycins containing a C17 fatty acid chain with Ala or Val at position 6 related to fractions 5 and 6, displaying a high mass intensity at m/z 1,477 and 1,505. f, g Surfactin isoforms containing C14 and C15 fatty acid chains related to fractions 9 and 10, at m/z 1,030 and 1,058, respectively, as [M + Na]+

Full size image

MALDI-TOF-MS/MS analysis of the HPLC fraction structures

Some of the isoforms in different HPLC fractions detected by MALDI-TOF-MS had the same m/z. Thus, it is essential to precisely identify their structure by MS/MS techniques, especially for the amino acid sequence of the peptide portion of the molecules.

Figure 4 displays the MALDI-TOF-MS/MS spectrum of the precursor ion at m/z 1,043 of HPLC fraction 1. Nucleophilic attack by the amide oxygen of the Gln side chain on the carbon center of the protonated amide bond leads to the formation of cyclic derivatives [37, 38]; thus, the C-terminal fragment of Gln leaves the parent ion. The MSn experiments studying the His effect by Wysocki et al. [39] demonstrated that the peptide tended to break at the N terminus of Pro. Therefore, the peptide ring mainly cleaved at Gln-Pro, between the C terminus of Gln and the N terminus of Pro. The series of b+ ions at m/z 1,043 → 915 → 801 → 638 → 524 → 299 → 212 represented cleavage along the peptide bonds by the loss of Gln, Asn, Tyr, Asn, C14 β-OH fatty acid, and Ser from the C terminus of Gln, whereas the y+ fragment ions at m/z 832 → 745 and 520 → 406 showed the loss of Ser and Asn. The MS/MS spectrum exhibiting b+ and y+ fragment ions confirmed the sequence of Pro-Asn-Ser–C14 β-OH fatty acid–Asn-Tyr-Asn-Gln.

Fig. 4
figure 4

MALDI-TOF-MS/MS spectrum of the iturin precursor ion at m/z 1,043. The spectrum exhibits b+ fragments of the precursor ion at m/z 212, 299, 524, 638, 801, and 915 and y+ fragments of the precursor ion at m/z 406, 520, 745, and 832. The precursor ion at m/z 1,043 was confirmed as an iturin consisting of Asn-Tyr-Asn-Gln-Pro-Asn-Ser and a C14 fatty acid chain

Full size image

Figure 5c displays the MS/MS spectrum of the precursor ion at m/z 1,058 in HPLC fraction 10. The series of b+ fragment ions at m/z 1,058 → 945 → 832(−H2O, 814) → 717 revealed the loss of Leu-Leu-Asp from the C terminus. Furthermore, another typical set of y+ fragment ions at m/z 707 → 594 → 481 → 382 → 267 suggested the loss of the Leu-Leu-Val-Asp in the middle of the peptide chain. The peak at m/z 832 in Fig. 5c was formed by loss of the Leu-Leu from molecule, and the peak of m/z 814 resulted from the loss of Leu-Leu–OH2, which confirmed that the ester structure has been formed by the carboxyl group of Leu and the hydroxyl group of the aliphatic part in the surfactin molecule. The precursor ion at m/z 1,058 was demonstrated to be the sodium adduct of a surfactin containing a C15 β-OH fatty acid, whose peptide sequence was Glu-Leu-Leu-Val-Asp-Leu-Leu. Additionally, the MS/MS spectra of sodium adducted ions found at m/z 1,030 (Fig. 5a), 1,044 (Fig. 5b), and 1,072 (Fig. 5d) with a difference of 14 Da (−CH2–) proved to be homologs possessing the same peptide sequence but different C13, C14, and C16 β-OH fatty acids, respectively.

Fig. 5
figure 5

MALDI-TOF-MS/MS spectra of surfactin precursors containing C13–C16 fatty acid chains. a Surfactin precursor ion [M + Na]+ at m/z 1,030, containing a Glu-Leu-Leu-Val-Asp-Leu-Leu peptide and a C13 β-hydroxy fatty acid chain. b Surfactin precursor ion [M + Na]+ at m/z 1,044, containing a C14 β-hydroxy fatty acid chain. c Surfactin precursor ion [M + Na]+ at m/z 1,058, containing a C15 β-hydroxy fatty acid chain. d Surfactin precursor ion [M + Na]+ at m/z 1,072, containing a C16 β-hydroxy fatty acid chain

Full size image

Another peptide sequence slightly different from the typical surfactin was discovered in fraction 9 (Fig. 6a) and fraction 11 (Fig. 6b). For the precursor ion at m/z 1,030 in HPLC fraction 9 (Fig. 6a), the significant ion series at m/z 1,030 → 931 → 818(−H2O,800) → 703 → 590 and 693 → 594 → 481 → 368 → 253 represented the sequence of Val-Leu-Asp-Leu from the C terminus and Val-Leu-Leu-Asp in the middle, suggesting the precursor ion possessed a peptide sequence of Glu-Val-Leu-Leu-Asp-Leu-Val and a C14 fatty acid.

Fig. 6
figure 6

Isoforms of surfactin containing a Glu-Val-Leu-Leu-Asp-Leu-Val peptide. a Surfactin precursor ion [M + Na]+ at m/z 1,030, containing a C14 β-hydroxy fatty acid chain. b Surfactin precursor ion [M + Na]+ at m/z 1,044, containing a C15 β-hydroxy fatty acid chain

Full size image

Figure 7 displays the MALDI-TOF-MS/MS spectra of the precursor ions at m/z 1,463 in fraction 4, m/z 1,477 in fraction 5, and m/z 1,505 in fraction 6. The γ-OH fatty acyl moiety had the effect of stabilizing the positive charge on the N terminus. It encouraged proton migration to the amide bonds in the first two positions of the peptide chain, making these two positions more labile to bond cleavage. Thus, the precursor ions mainly cleaved at the Glu–Orn and Orn–Tyr bonds, resulting in the specific octapeptide ring ions at m/z 966 and 1,080 (Fig. 7a, b) and at m/z 994 and 1,108 (Fig. 7c), respectively [40]. The extra stability of the charged octapeptide ring system suppressed further fragmentation along the peptide backbone and thus minimized the amount of sequence information. We analyzed some other ionic species for further analytical information and found some ions derived from the octapeptide ring. Significant ion series at m/z 665 → 520(+H2O) → 389 → 226 → 115(+H2O) in Fig. 7b represented the sequence of Pro-Gln-Tyr-Ile-Tyr from the N terminus after the first cleavage of bonds between Ala/Val and Pro in the specific octapeptide ring ions. Besides, we also found z8 and z9 ions corresponding to the y8 and y9 ions with a loss of NH (15 Da). The ions in Fig. 7a and b agreed with fengycin containing the same peptide ring, but with a different length of fatty acid chain, C16 or C17. The ions in Fig. 7b and c agreed with fengycin containing a C17 fatty acid chain, with the only difference being Ala or Val at position 6. The amino acid sequence of the lactone ring could not be perfectly identified. In this situation, amino acid analysis based on PITC derivatives verified the identities of the amino acids and their molar ratios.

Fig. 7
figure 7

MALDI-TOF-MS/MS spectra of the fengycin precursor ions. a Precursor ion at m/z 1,463 with Ala at position 6, containing a C16 γ-hydroxy fatty acid chain. b Precursor ion at m/z 1,477 with Ala at position 6, containing a C17 γ-hydroxy fatty acid chain. c Precursor ion at m/z 1,505 with Val at position 6, containing a C17 γ-hydroxy fatty acid chain

Full size image

Amino acid analysis of isoforms from the HPLC fractions

Fractions harvested from the HPLC fractions were hydrolyzed with 6 M HCl and derivatized with PITC after norleucine had been added as an internal standard. Figure 8 displays the chromatograms of the PITC derivatives of 18 standard amino acids (Fig. 8a) and amino acid residues from samples in fractions 5 and 6, which showed m/z values of 1,477 and 1,505, respectively, in their protonated forms. Figure 8b demonstrates that the hydrolyzed fraction of m/z 1,477 contained Glu (Gln), Thr, Ala, Pro, Tyr, Ile, and Orn in a molar ratio of 2.90:1.00:1.01:1.10:2.04:1.10:1.03, and Fig. 8c demonstrates that the hydrolyzed fraction of m/z 1,505 contained Glu (Gln), Thr, Pro, Tyr, Val, Ile, and Orn in a molar ratio of 3.06:1.00:1.16:2.17:1.10:1.15:1.07, which just differed in the substitution of Val for Ala. The amino acid analysis results were in perfect agreement with the amino acid sequence of the fengycin molecules, as illustrated in Fig. 7d. The amino acids of the major fraction at m/z 1,058 in fraction 10 identified by MALDI-TOF-MS/MS were also verified by amino acid analysis, as shown in Fig. S10.

Fig. 8
figure 8

Chromatograms of phenyl isothiocyanate derivatives of amino acids obtained by C18-RP-HPLC. a Standard amino acids: Asp, Glu, Ser, Gly, His, Arg, Thr, Ala, Pro, Tyr, Val, Met, Cys, Ile, Leu, Phe, Trp, and Lys. b The hydrolyzed sample of m/z 1,477 in fraction 5, with Ala at position 6. c The hydrolyzed sample of m/z 1,505 in fraction 6, with Val at position 6

Full size image

Finally, the structures of 12 lipopeptide isoforms detected in 12 fractions from the RP-HPLC were all identified and are summarized in Table 1.

Table 1 Assignment of the structures of lipopeptides in RP-HPLC fractions in this study
Full size table

Discussion

Lipopeptide biosurfactants have the advantages of low toxicity, high efficiency, high biodegradability, and high sustainability compared with chemical surfactants. There are a wide range of potential applications in agriculture and industry [41–46], such as microbial-enhanced oil recovery and cosmetics. The iturin and fengycin families display strong antifungal and antibacterial activity [47, 48], and the surfactin family also displays hemolytic, antiviral, antimycoplasma, and antibacterial activities [49, 50].

For research in the fields of cosmetics and medicine, RP-HPLC is widely used for high-quality purification. Literature reports on lipopeptide separation usually focus on a certain family, such as the report by Yuan et al. [27], who eluted iturin using 40 % acetonitrile in water for 20 min. Additionally, Villegas-Escobar et al. [12] eluted fengycin between retention times of 26 and 50 min (48–60 % acetonitrile in water) and Liu et al. [35] eluted surfactin with 90 % methanol containing 0.05 % TFA for 24 min. Some studies have detected more than one family of molecules when the sample was eluted with a rather long gradient. Chen et al. [3] detected iturin, fengycin, and surfactin at the same time with acetonitrile increasing from 30 to 80 % for 60 min, but the fractions were not perfectly separated. With use of our three-stage gradient strategy, isoforms of iturin, fengycin, and surfactin were separated in groups at retention times in range of 6–10 min, 10–16 min, and 16–25 min, respectively. The purity of the RP-HPLC fractions was very high, and thus the MALDI-TOF-MS spectra exhibited only one series of [M + H]+, [M + Na]+, and [M + K]+ ions corresponding to a lipopeptide isoform in a fraction. Good purification of the isoforms is crucial for the structural identification and for some applications with strict demands of purity. An aim would be to optimize further the separation of molecules of the fengycin family.

MS/MS coupled with CID is a powerful tool for identifying the amino acid sequence of peptides. The mobile proton model provides a solid background for qualitatively understanding the fragmentation pathways of protonated peptides. The model states that protonated peptides activated under low-energy collision conditions fragment mainly by charge-directed reactions at the amide bonds along the backbone [38], forming structurally informative sequence ions, especially the b and y ions containing the N terminus and the C terminus, respectively. In double hydrogen transfer, the cleavage happens between alpha and beta atoms and two hydrogens are transferred to the alkoxy part including alpha atoms [51]. As a result, for the ester structure the H2O would be added to the C-terminal part and its mass would be increased by 18. MALDI-TOF-MS/MS affords high sensitivity and exhibits abundant singly charged sequence ions of the cyclic peptides without the need for hydrolytic treatment. However, partial base hydrolysis treatment (such as treatment with 1 M KOH at room temperature for 1 h [12, 14]) can contribute to produce adequate sequence information for the peptide, especially the fengycin family.

The precolumn derivatization method with PITC was proved to have enough derivative stability, high sensitivity, and high-resolution separation of 18 amino acids [28, 29]. The results of amino acid analysis further confirmed the amino acid species and molar ratio in a certain peptide, especially for the two isobaric amino acids Leu and Ile. Since the lipopeptide hydrolysate samples were treated with 6 M HCl, the Asn and Gln would become Asp and Glu. Thus, this analysis method could not distinguish Asn and Gln from Asp and Glu [52]. The peaks of Asp and Glu were considered as Asx and Glx, which could be verified in combination with the mass loss in the MS/MS spectra [14].

Conclusion

An efficient three-stage gradient RP-HPLC strategy was optimized for the rapid and high-quality purification of a lipopeptide mixture produced by B. subtilis. The methanol–water mobile phase system realized a relatively good separation of the iturin and surfactin isoforms. Further optimization in an acetonitrile–water mobile phase system resulted in a better simultaneous separation of the iturin, fengycin and surfactin families in groups with retention time ranges of 6–10 min, 10–16 min, and 16–25 min. The isoforms obtained using the optimized acetonitrile–water strategy showed a high level of purity in MALDI-TOF-MS. MALDI-TOF-MS/MS analysis precisely elucidated the amino acid sequence of the cyclic isoforms of iturin and surfactin harvested from RP-HPLC fractions without hydrolytic treatment. The typical cyclic fengycins mainly cleaved at Glu–Orn and Orn–Tyr bonds, and amino acid analysis based on PITC derivatization was used to verify the identities and molar ratios of the amino acids present. To sum up, we have presented a systematic approach for the simultaneous rapid discovery, high-quality purification, and characterization of lipopeptide isoforms from the iturin, fengycin, and surfactin families, which may differ in composition by only a single amino acid and/or the fatty acid residue.