Reagents and materials

Acetonitrile (ACN) of LC-MS grade, 2,5-dihydroxybenzoic acid (DHB), formic acid (FA) and trifluoroacetic acid (TFA, suitable for HPLC-MS) were purchased from Sigma-Aldrich, Merck (Burlington, MA, USA). Ammonium hydrogen carbonate (AHC, suitable for HPLC-MS) was obtained from Lachema (Brno, Czech Republic). Peptide calibration standard II was purchased from Bruker Daltonics (Bremen, Germany). Pierce trypsin protease MS grade was obtained from Thermo Fisher Scientific (Waltham, MA, USA). The commercially available reverse phase ZipTip C₁₈ pipette tips were purchased from Millipore Corporation (Burlington, MA, USA). The water was purified with a Milli-Q water purification system from Millipore Corporation.

Reference samples of fish meat

The meat from six selected fish species was analysed. Among them were three freshwater species: common carp (Cyprinus carpio), northern pike (Esox lucius), rainbow trout (Oncorhynchus mykiss) and three marine species: Atlantic mackerel (Scomber scombrus), Alaska pollock (Theragra chalcogramma) and Atlantic salmon (Salmo salar). Freshwater fish species were purchased from the local fish store named Štičí líheň ESOX spol. s. r. o. in Tábor (Czech Republic), marine species were from wholesale chains Albert and Lidl. Three individuals of each species (biological replicates) were analysed in this work.

Two types of samples were prepared for each fish: raw and cooked meat. The heat-treated samples were prepared by boiling 1–2 g of cut fish meat in boiling water for 10 min. Both types of samples were then stored in a freezer (-80 °C). These primary samples were later used for the analysis. Three samples (technical replicates) were weighed (approx. 1 mg) for each type of fish. Nine samples of raw meat and nine samples of cooked meat from each fish species were prepared for the analyses.

Sample preparation

A mass of 1 mg of each sample was digested in 20 µL of 50 mM AHC containing 0.02 mg/mL of trypsin at 37 °C with constant shaking for two hours. After two hours, the cleavage was terminated by adding 1 μL of 10 % TFA solution to a final amount of 0.5 % TFA. After the trypsin digestion, the samples were purified and concentrated on reverse phase ZipTip C₁₈. After purification, 10 μL of each purified sample were obtained.

MALDI-TOF MS measurements and data acquisition

A volume of 2 µL of purified peptide sample was mixed with 7.5 µL of DHB matrix solution (8.5 mg of DHB in 0.5 mL of a mixed solvent: water, acetonitrile and 0.1 % TFA). A volume of 1.3 μL of the resulting mixture was spotted thrice on the stainless steel MALDI target and air-dried. Mass spectra were acquired using MALDI-TOF Autoflex Speed mass spectrometer (Bruker Daltonics) equipped with an Nd:YAG laser (355 nm) in positive reflector mode. The obtained spectra contained peaks in the 900–4500 m/z interval. The corresponding spectrum was obtained from a total of 7000 shots for each spot.

LC-ESI-Q-TOF MS conditions and data acquisition

Measurements were carried out using UHPLC Dionex Ultimate3000 RSLC nano (Dionex, Bremen, Germany) connected to ESI-Q-TOF Maxis Impact mass spectrometer (Bruker Daltonics). Purified and air-dried samples (after trypsin digestion, see Sample preparation) were dissolved in 10 μL of a mixture of 3 % acetonitrile and 0.1 % formic acid. A volume of 3 µL of the solution was loaded into an Acclaim PepMap 100 C18 trap column (100 µm×2 cm, size of reverse phase particles 5 µm; Dionex) and 3 % of mobile phase B with a flow rate of 5 µL/min was run for 7 min. The peptides were then eluted from the trap column into Acclaim PepMap RSLC C18 analytical column (75 µm×150 mm, size of reverse phase particles 2 µm; Dionex) using the following gradient: 0–5 min 3 % B, 5–35 min 3–35 % B, 37 min 90 % B, 37–50 min 90 % B, 51 min 3 % B, 51–60 min 3 % B. The mobile phase A consisted of 0.1 % formic acid in water and mobile phase B of 0.1 % formic acid in acetonitrile. The flow rate during gradient separation was set at 0.3 µL/min. The peptides were eluted directly to an ESI source – Captive spray (Bruker Daltonics). Measurements were carried out in positive ion mode with a precursor selection in the range of 400–1400 Da; up to 10 precursors were selected for fragmentation from each MS spectrum.

MS spectrum was recorded every 3 s, MS/MS spectra were collected at 4–16 Hz depending on precursor intensity. Dynamic precursor exclusion was set to 1 min, preferred number of precursor charges was 2–5. Singly charged precursors were excluded from fragmentation. Collision-induced MS/MS spectra were recorded in the range of 50–2200 m/z. Mass spectra were extracted by DataAnalysis 4.1 (Bruker Daltonics) and loaded into Proteinscape 4.2 (Bruker Daltonics) and later into Mascot 2.4.1 (Matrix Science, Boston, MA, USA), which was used for protein identification. The identification was carried out against a single-species database containing the proteome of the investigated species (Cyprinus carpio and Salmo salar were downloaded from the UniProt website on 4 April 2022) (37), which was supplemented with common laboratory contaminants. The following identification parameters were used: enzyme trypsin (one missed cleavage site was allowed), oxidation of methionines as a variable modification, accuracy of assigning precursors of 10 ppm and fragments of 0.05 Da. Identified peptides and proteins were filtered to maintain a false positive identification rate of 1 %. All samples were analysed by LC-ESI-Q-TOF MS in three repetitions to obtain characteristic peptide profiles.

Searching for species-specific markers

To distinguish individual types of fish meat, species-specific markers were identified as m/z or peptides that occurred with certain frequency in the spectra obtained from one species but were absent in the spectra of other fish species. The frequency represents a number of spectra in which the particular m/z values or amino acid sequences occurred. Then these peptides or m/z values can be considered species-specific markers within the selected group of fish. Mass spectra were processed using two complementary methods: MALDI-TOF and LC-ESI-Q-TOF.

For MALDI-TOF data processing, the mMass software, v. 5.5.0 was used (38). The involved spectrum smoothing, baseline correction and manual peak selection, where 80–110 peaks were selected for each MALDI-TOF spectrum. The m/z values from the spectra obtained for a single species were then recorded in Microsoft® Excel® for further analysis.

To manage and analyse the extensive data, we used the PostgreSQL object-relational database system (v. 2022.4.4) with pgAdmin 4 (v. 6.21), an open-source graphical administration tool for PostgreSQL (36). The analysis focused on identifying m/z values that appeared consistently across spectra for each species. This process was adjusted to include only peaks present in a specific frequency, set in this study to a minimum of 23 out of 27 MALDI spectra (three individuals, three technical replicates and three spots per technical repetition).

Similarly, Excel data files containing the results (identified peptides and their corresponding proteins) from LC-ESI-Q-TOF MS were processed using comparable steps. Species-specific peptides were identified as those consistently present in all nine Excel data files corresponding to a single species (three individuals and three technical replicates, where each technical replicate was injected once).

Results of MALDI-TOF mass spectrometry

MALDI-TOF MS measurements were conducted on all six fish species, both raw and cooked samples. This analysis identified m/z values that serve as species-specific markers and characterise each fish species.

Searching for species-specific markers by MALDI-TOF MS

The list of peptides (m/z values) was obtained for each fish species using the data evaluation described in Searching for species-specific markers. The m/z values were determined separately for raw samples across all species and in a separate analysis for cooked samples. In the next step, values were compared between individual species to find characteristic values for one specific species, i.e. a given m/z value that occurred in one species, but not in any other. Table 1 lists the characteristic values for raw and cooked meat samples. Fig. 1 shows the MALDI-TOF MS spectra for carp, pollock and salmon, highlighting the differences in their characteristic m/z values. These spectra serve as a visual representation of some species-specific markers identified in our analysis.

Fig. 1

MALDI-TOF MS spectra of: a) salmon (blue), b) carp (red) and c) pollock (green). Characteristic m/z values for raw meat are indicated with asterisk (*)

The differentiation of fish species by MALDI-TOF MS using species-specific protein patterns has been shown previously (39–42). These studies used a simpler approach without trypsin digestion and analysed extracted proteins directly. Enzymatic cleavage was only used in a subsequent step for the identification of protein biomarker. This method has proven effective for distinguishing species, even without the need for tryptic digestion, particularly for species that are not closely related, such as the six species in our study.

Our approach differs fundamentally in the use of trypsin digestion prior to MALDI-TOF MS measurements. This technique, known as in-sample digestion, fragments proteins into peptides, resulting in spectra with a broader range of information and a higher resolution. While the approach of Mazzeo et al. (39) identified markers in the >11 000 m/z range, corresponding to small proteins like parvalbumins, or Stahl and Schroeder (40), who collected spectra in the mass range 2–20 kDa, our analysis focuses on the 900–4500 Da range to capture differences in the whole proteome rather than targeting specific proteins.

Spielmann et al. (42) explored the use of MALDI-TOF MS for processed meat and developed a database of species-specific fish proteins using the Biotyper tool. While effective, the Biotyper tool is proprietary and requires a paid license, making it less accessible. In contrast, the approach described in this paper utilises PostgreSQL, an open-source database system, which not only reduces costs but also provides broader accessibility to researchers and laboratories.

Although the six tested species could probably be differentiated using simpler methods, this method shows strong potential in the context of heat-treated fish meat, where the higher resolution and broader marker identification provide added value. By focusing on peptide-level markers and leveraging accessible tools, the shown approach offers a cost-effective and innovative alternative for fish meat authentication.

Markers for different types of cooked fish meat were then searched in the same way. In this analysis, at least two specific markers were found for each species. The list of characteristic values of the cooked meat samples is shown in Table 1. A slightly bigger number of characteristic values was found for cooked meat. This can be explained by the fact that proteins in their native structure are not as easily accessible to trypsin cleavage in some positions as the loosened, thermally denatured proteins.

Some m/z values are identical for raw and cooked fish meat. These values can be considered as characteristic markers for the respective species regardless of the type of meat processing (underlined values in Table 1). The other half of the m/z values, which are not the same for raw and cooked meat, proves that some markers differ depending on the type of processing to distinguish the respective species. The obtained results prove the feasibility of the method also for cooked meat and at the same time confirm the stability of some MALDI-TOF MS markers during heat treatment (markers found simultaneously for raw and cooked meat), as mentioned in the literature (39, 40).

The next step was to search for markers that could distinguish between raw and heat-treated meat of the same species. In this phase, we therefore focused on the m/z values that differentiate the raw and cooked form of a particular fish—for example, markers unique to raw carp versus cooked carp, and by analogy the other species analysed. The search for markers was carried out sequentially for all analysed species and the results are shown in Table 2. The results show that the raw fish meat samples contained more specific markers than their cooked counterparts. The underlined values represent peaks that are already included in Table 1 and serve as markers for distinguishing between the raw or cooked meat species. These markers have a high discriminatory value and when present in a sample, can determine not only the type of fish meat but also its type (raw or heat treated). Such markers have been found for raw meat of carp, mackerel and pike, in the case of cooked meat of carp and mackerel.

Results from LC-ESI-Q-TOF mass spectrometry

The LC-ESI-Q-TOF analyses were performed only on carp and salmon samples, both raw and cooked (a total of 36 samples were analysed – nine for each type of processing of these species). The reason for this was that in the used UniProt protein database (in which the peptides and proteins are searched) the fish species category is not complete. Therefore, only the two species whose protein sequences are available in the database were selected. This method can detect the amino acid sequences of peptide fragments and identify the protein from which the fragments originate. This provides more accurate data than the m/z values usually obtained by MALDI-TOF MS.

Comparison of markers found by both mass spectrometric methods

A possible similarity between markers found with MALDI-TOF and LC-ESI-Q-TOF mass spectrometry was investigated. The m/z markers obtained from MALDI-TOF MS (see Table 1 and Table 2) were compared to the list of the characteristic peptide sequences (their m/z) obtained by LC-ESI-Q-TOF.

Since MALDI produce ions of peptides and proteins with a uniform charge of +1 and the ions generated by electrospray ionization can have multiple charges, it was not possible to directly compare the m/z values obtained by these two methods. The molecular mass of the peptide fragments were compared after charge subtraction. For MALDI-TOF, these values were easily obtained by subtracting the mass of one proton (MH⁺=1 Da), for LC-ESI-Q-TOF the mass of the peptide fragment was obtained directly from the results exported by Mascot 2.4.1. The tolerance was set to ±0.3 Da. Seven and three characteristic MALDI-TOF MS markers were obtained for raw carp and salmon samples respectively (see Table 1). Of these, three values for carp and none for salmon were found in the LC-ESI-Q-TOF markers in the form of characteristic sequences within the required tolerance. An overview of these values can be found in Table 6.

In cooked samples of the same fish species, 10 markers for carp and two for salmon were obtained using the MALDI-TOF method (see Table 2). Within the same tolerance, four out of ten carp markers were also found in the LC-ESI-Q-TOF data. However, not a single match was found for salmon.