Introduction

Iron is an essential trace element for fish and other living organisms, vital for important biological processes (e.g., oxygen and electron transport, cellular respiration, synthesis of DNA and neurotransmitters) [1,2,3]. Biological activity of Fe, as of many other trace elements, is primarily associated with their binding to various proteins, thus forming metalloproteins, in which metals are essential components [2]. Iron deficiency can cause a number of disorders of different degrees of severity [2]. However, in excess, even essential trace metal like Fe can be toxic, causing the reactive oxygen species formation, and consequently leading to lipid peroxidation, DNA damage, altered calcium homeostasis, and eventually cell death [2,3,4,5].

Since there is a fine line between metal essentiallity and toxicity, organisms have developed complex systems for maintenance of trace element homeostasis, which are composed chiefly of transport and storage proteins, and certain hormones [2]. Following the uptake in the organism, Fe is transported in a bloodstream as a complex with transferrin, being finally delivered to cells of target organs. In the cytoplasm of those cells, Fe is present within the labile Fe pool, whereas excess Fe is stored in ferritin [2, 3].

Ferritins play a key role in Fe metabolism by sequestering potentially toxic free Fe, and maintaining Fe in a soluble, nontoxic, bioavailable form [6, 7]. Ferritin has a mass of ~450 kDa and consists of a central hydrous ferric oxide phosphate core surrounded by an outer protective hollow protein shell called apoferritin [3, 6]. The apoferritin has 24 subunits (a combination of heavy (H) and light (L) chains; molecular masses in a range from 18 to 28 kDa) [6,7,8]. The H-chains are associated with active roles in Fe metabolism, including ferroxidase activity, whereas L-chains are associated with a long-term storage function [7,8,9]. Thus, ferritins rich with L-chains are typical for organs with high Fe storage, such as the liver [3]. However, the native ferritins from different fish livers contain only one (middle) type of subunit instead of two, which possesses the characteristics of both H and L subunits [3, 6, 7] (in liver of smooth hammerhead (Sphyrna zygaena), 21 kDa [7]; in liver of the northern pike (Esox lucius), 20.4 kDa [10]).

Moreover, hemoglobin (Hb), which is contained in erythrocytes, contains Fe which reversibly binds oxygen and functions as the final acceptor of electrons originating from oxidative catabolic reactions [11,12,13]. The Hb molecule in vertebrates contains four globin chains (two α and two β) in a stable tetramer, each of them containing a prosthetic heme group which participates in oxygen binding through an iron atom in the ferrous form (Fe2+) [13]. Heme group is apparently identical in every studied fish species, whereas the globins differ from species to species and among Hb isoforms [12, 13]. The great majority of fish species have symmetric Hbs, i.e., two pairs of identical globin chains [13]. In the studies of fish liver metalloproteins, it is expected to detect Hb, as liver is perfused by blood [14].

Iron metalloproteins in freshwater fish were studied in various species (e.g., belonging to families Channidae, Cyprinidae, Esocidae, Leuciscidae, Salmonidae) by many different methods (e.g., PCR analysis and immunohistochemistry, electrophoresis, various immunological, chromatography, and mass spectrometry methods), with the special emphasis on Hb and ferritin studies [3, 6, 10, 14,15,16,17,18,19].

The studies on Fe-metalloproteins in the northern pike are fairly scarce. In adition to our studies on hepatic Fe-metalloproteins, by one-dimensional size exclusion or ion exchange chromatography coupled with inductively coupled plasma mass spectrometry (ICP-MS) [18, 19], only study that we are aware of is the one on hepatic ferritin as a possible natural iron supplement [10].

The current study aims to supplement our two previous studies [18, 19] by identification of the main Fe-binding biomolecules in the northern pike liver using two-dimensional chromatographic separation and subsequent mass spectrometry detection. Additional aim is to associate our results with those contained in databases (applying MASCOT search), and compare it subsequently with our corresponding study on the Vardar chub (Squalius vardarensis) [14], to establish relationships among various fish species with the regard to intracellular Fe-handling strategies.

Materials and methods

Fish sampling and the dissection of liver

The sampling of the northern pike (E. lucius Linnaeus, 1758) was carried out on 22 April and 1 May 2021 from the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town (N 45° 27′ 4.38″ E 15° 30′ 18.96″) using an electrofishing device (Hans Grassl, EL63 II GI, 5.0 KW, 137 Honda GX270, 300/600V max., 27/15A max.), in accordance with the Croatian standard HRN EN 14011 [20]. Following the requirements of the Ordinance on the protection of animals used for scientific purposes [21], the fish were euthanized at the location of the sampling, immediately after being caught, using unbuffered MS 222 (tricaine methane sulphonate, Sigma-Aldrich) in concentration of ~50 mg L−1 and in duration under 10 min [17]. Fish length and mass were measured, age was determined by counting the number of annuli (rings) on scales using an Olympus BH2 microscope (magnification 30×), whereas sex was determined by gonad examination at macroscopic level. The liver was taken from each fish and then stored, first in liquid nitrogen and thereafter in the freezer at −80°C, until further processing. Seventeen fish were caught in total at this location at the said dates, and livers of three of them were selected for this particular investigation (the same three northern pike specimens as used and presented in our previous paper) [19]. Three studied northern pike specimens were of the following characteristics: total mass of 345–1100 g; total length of 36.5–54.0 cm; they were all female with the age span from 2+ to 4+.

Homogenization and centrifugation of the northern pike liver

The livers were cut into pieces in glass containers placed on ice, and then into each container, we have added cold homogenization buffer (100 mM Tris-HCl/base (Sigma-Aldrich, USA; pH 8.1 at 4 °C) supplemented with reducing agent (1 mM dithiothreitol (DTT), Sigma-Aldrich, USA)) (ratio of mliver and vbuffer was 1+5). Hepatic tissue suspensions were homogenized in an ice-cooled tube by 10 strokes of a Potter-Elvehjem homogenizer (Glas-Col, USA) at 6000 rpm. The homogenates were thereafter centrifuged for 2 h at 50,000×g and +4 °C (Avanti J-E centrifuge, Beckman Coulter, USA). The resulting supernatants (S50) represented soluble hepatic fractions, which mainly refer to cytosol, and further contain only microsomes [22] and extracellular fluids (including traces of blood). Thus obtained hepatic cytosols were immediately stored at −80 °C and kept there until further analyses [18].

High-resolution (HR) ICP-MS measurement of Fe and other element concentrations in the northern pike hepatic cytosols

The volumes of 300 μL of hepatic cytosols were first digested using the digestion mixture containing nitric acid (HNO3; Normatom® 67–69% for trace element analysis, VWR Chemicals, UK) and hydrogen peroxide (H2O2; Suprapur® 30%, Merck, Germany) in volume ratio 3+1. Volume ratio of cytosol to digestion mixture was 1+1. For each of the three samples, the digestion was carried out in duplicate in the laboratory dry oven (at 85 °C; 3.5 h) [18]. In thus obtained digested samples, diluted five times with Milli-Q water, Fe concentrations were measured using HR ICP-MS (Element 2, Thermo Finnigan, Germany). Measurements of 82Se, 95Mo, 109Ag, 111Cd, 133Cs, 205Tl, 208Pb, and 209Bi were performed in low-resolution mode; of 55Mn, 56Fe, 59Co, 63Cu, and 66Zn in medium-resolution mode; and of 75As in high-resolution mode. External calibrations were applied, which were based on adequate dilutions of multielement standard solution for trace elements (Analytika, Czech Republic) prepared in 2% (vol.) HNO3 (Normatom® 67-69% for trace element analysis, VWR Chemicals, UK) supplemented with adequately diluted standard solutions of Cs (1 g L−1; Fluka, Germany). The elemental concentrations used for creation of calibration curves were 0, 1, and 10 μg L−1 for all studied elements, and additionally 100 μg L−1 for Cu, 500 μg L−1 for Cu, Fe, and Zn, and 1000 μg L−1 for Fe and Zn, with respect to their actual concentrations in the analyzed samples. To all the samples and calibration standards, In (Fluka, Germany) was added as an internal standard (1 μg L−1). The cytosolic concentrations of all 14 analyzed elements in the northern pike liver are presented in Table 1, and, additionally, Fe concentrations in the selected northern pike hepatic cytosols are presented in Fig. 1.

Table 1 The cytosolic concentrations of Fe and 13 other trace elements in the liver of the three specimens of the northern pike used in this study, and caught from the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town in April/May 2021
Full size table
Fig. 1
figure 1

The Fe distribution profiles established by online coupled HPLC-ICP-MS with anion-exchange (AEX) column for the hepatic cytosols of the three northern pike specimens (samples 1–3), caught in the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town in April/May 2021. The results are presented as Fe intensities (y-axis) versus elution time (Et, min; x-axis). The maxima for two major Fe-AEX-peaks of each sample are annotated within the figure, as well as the cytosolic Fe concentrations

Full size image

Limits of detection (LODs) for HR ICP-MS measurement of Fe and other studied elements in digested and diluted hepatic cytosols were calculated based on the three standard deviations of ten successively measured metal/metalloid concentrations in the blank sample (comprised of homogenization buffer, HNO3 and H2O2, and processed using the same procedure as used for the digestion of hepatic cytosols). The obtained LODs (ng of element per g of hepatic tissue) were the following: Ag, 0.37; As, 1.40; Bi, 0.24; Cd, 0.08; Co, 0.22; Cs, 0.10; Cu, 16.3; Fe, 75; Mn, 1.7; Mo, 0.21; Pb, 0.49; Se, 1.5; Tl, 0.001; and Zn, 23. The HR ICP-MS measurement of several elements with certified concentrations in quality control sample (UNEP GEMS, Canada) on two separate occasions resulted with the following recoveries (average ± standard deviation): As, 100.7±0.6%; Cd, 102.6±1.1%; Co, 101.9±5.6%; Cu, 102.2±0.5%; Fe, 95.8±0.1%; Mn, 103.0±1.0%; Mo, 101.6±2.5%; Pb, 99.4±1.5%; and Zn, 107.1±2.4%.

Two-dimensional high-performance liquid chromatography (HPLC) separations of Fe-binding biomolecules from the northern pike hepatic cytosol

During all chromatographic separations, we have undertook all the necessary measures, as mentioned by de la Calle Guntiñas et al. [23], which are required to avoid metal dissociation from metal-binding biomolecules and maintain the physiological surroundings, such as the use of ice-cold conditions during the sample handling, and the slightly alkaline pH of 7.4-8.0 throughout the HPLC separations.

Anion exchange (AEX)-HPLC-ICP-MS analysis

The first step of separation of Fe-binding biomolecules from the northern pike hepatic cytosol, with subsequent Fe detection, was based on AEX principle (i.e., separation based on differences in net charges) using hyphenated HPLC-ICP-MS system (HPLC, Agilent 1260 Infinity II with a diode array UV/VIS detector; ICP-MS, Agilent 7900; Agilent Technologies, USA). The analyzed sample was the soluble hepatic fraction of the northern pike. The protocol for AEX-HPLC separation was described in our previous paper [19], and adapted from the previously described procedures [14, 24]. The details on separation protocol are given in Table 2. In Fig. 1, we have presented only the obtained AEX-HPLC-profiles for Fe, since the comparison with several other metals was already discussed in details in our previous paper [19]. Selected Fe-containing AEX fractions were collected in plastic tubes at 30-s intervals using a fraction collector (Agilent InfinityLab LC Series 1260 Infinity II Bio-inert Fraction Collector). Fractions were supplemented with 1 mM DTT (Honeywell Fluka, Germany) and then pooled under ice-cold conditions and preconcentrated by Amicon Ultra centrifugal filters (cut-off at 3 kDa, Merck, Millipore, Ireland) for second step of chromatographic separation. Iron recovery from AEX column was calculated to be 85–97%, based on the analyses of samples 1 and 2. During recovery estimation, quality control indicated measurement accuracy in the range from 99 to 104%.

Table 2 Separation protocols for analyzing Fe-binding biomolecules from the soluble hepatic fraction of the northern pike by anion-exchange (AEX) and size-exclusion (SEC) HPLC
Full size table

Size exclusion (SEC200)-HPLC-ICP-MS analysis

The second step of chromatographic separation was based on size exclusion principle (i.e., separation based on differences in molecular masses), using the same hyphenated HPLC-ICP-MS system as described above. The details on separation protocol are given in Table 2. In order to identify potential overlapping peaks of the other metals/metalloids with Fe-peaks, in addition to Fe we have also measured 13 other elements (Ag, As, Bi, Cd, Co, Cs, Cu, Mn, Mo, Pb, Se, Tl, and Zn) in selected AEX-fractions separated by SEC200-column. The results are presented in Figs. 2, 3, and 4 and Table ESM 1 (see Electronic Supplementary Material). For the column calibration, six protein standards were run through the column under the same conditions as the samples and their elution times, void volume elution time, and calibration straight line equation are shown in Table 3. Metal-binding biomolecules were defined as belonging to the high molecular mass region (HMM, >100 kDa), the medium molecular mass region (MMM, 30–100 kDa), and the low molecular mass region (LMM, <30 kDa).

Fig. 2
figure 2

The Fe distribution profiles established by online coupled HPLC-ICP-MS with size-exclusion (SEC200) column in the four selected chromatographic fractions obtained by AEX-HPLC separation of the hepatic cytosols of the two northern pike specimens (samples 1–2), caught in the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town in April/May 2021: a AEX-fraction A, 8.5–9.5 min; b AEX-fraction B, 9.5–10.0 min; c AEX-fraction C, 10.0–10.5 min; d AEX-fraction D, 10.5–11.5 min. The results are presented as in Fig. 1. The maxima (Et, min; and estimated molecular masses according to SEC-column calibration, kDa) for the major Fe-SEC-peaks of each sample are annotated within the figures for each AEX-fraction

Full size image
Fig. 3
figure 3

The distribution profiles established by online coupled HPLC-ICP-MS with size-exclusion (SEC200) column for additional three metals in a single chromatographic fraction obtained by AEX-HPLC separation of the hepatic cytosols of the two northern pike specimens (samples 1–2), caught in the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town in April/May 2021 (namely, AEX-fraction B; 9.5–10.0 min): a Mn; b Pb; and c Se. The results are presented as in Fig. 2

Full size image
Fig. 4
figure 4

The distribution profiles established by online coupled HPLC-ICP-MS with size-exclusion (SEC200) column for additional two metals in a single chromatographic fraction obtained by AEX-HPLC separation of the hepatic cytosols of the two northern pike specimens (samples 1–2), caught in the Mrežnica River (Croatia) in front of the former cotton industry facility in Duga Resa town in April/May 2021 (namely, AEX-fraction C; 10.0–10.5 min): a Pb and b Zn. The results are presented as in Fig. 2

Full size image
Table 3 Concentrations, sources, elution times (Et), and molecular masses (MM) of blue dextran (for determination of void volume) and six standard proteins used for calibration of Superdex® 200 Increase 10/300 GL (Cytiva) size exclusion column. Equation of calibration straight line was: y = −0.2256x + 1.3165; (y = Kav; x = logMM)
Full size table

Selected SEC200-fractions were collected in plastic tubes at 1-min intervals, supplemented with 1 mM DTT (Honeywell Fluka, Germany), pooled under ice-cold conditions and preconcentrated by Amicon Ultra centrifugal filters (cut-off at 3 kDa, Merck, Millipore, Ireland) for further mass spectrometry analysis.

Mass spectrometry analysis of selected AEX-SEC200-HPLC-fractions of the northern pike hepatic cytosol

After two-dimensional chromatographic fractionation (AEX-SEC200-HPLC), selected Fe-containing fractions were further analyzed with two mass spectrometry techniques (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and liquid chromatography tandem mass spectrometry (LC-MS/MS)) in order to determine exact molecular masses and possibly identify the isolated northern pike hepatic biomolecules of interest.

MALDI-TOF-MS analysis

For the purpose of MALDI-TOF-MS analysis, we have used the bench-top Bruker Microflex LT mass spectrometer equipped with the Bruker flexControl v 3.4 and flexAnalysis v 3.4 software (Bruker Daltonics, Germany). The volumes of 1 µL of preconcentrated samples were spotted in two replicates onto a 96-spot steel target plate (Bruker Daltonik, Germany) and allowed to visibly dry at room temperature. Subsequently, 1 µL of HCCA matrix, containing α-cyano-4-hydroxycinnamic acid (10 mg mL−1; Bruker Daltonik, Germany) in 50% acetonitrile (Optima® LC/MS grade, Thermo Fisher Scientific, UK) and 2.5% trifluoroacetic acid (Sigma-Aldrich, USA), was added on top of the samples and allowed to dry at room temperature.

Modified protocol by Huang et al. [7] was used to analyze fractions containing HMM Fe-binding biomolecules. Volumes of 5 µL of samples were mixed with 5 µL of acetone (Thermo Fisher Scientific, UK) and 5 µL of the HCCA matrix. Volume of 5 µL of this mixed solution was loaded onto the target plate, and the solvent was removed by air drying.

The mass spectra were acquired manually. Ions were captured in the positive linear mode (mass range 5–100 kDa), and positive ions were extracted at accelerated voltage of 20 kV. Spectra with the sum of the ions were obtained by 1000 laser shots in different regions of every target plate spot. The results are presented in Fig. 5.

Fig. 5
figure 5

Mass spectra obtained by MALDI-TOF-MS for fish sample 2: a ~360 kDa Fe-binding biomolecule isolated by two-dimensional chromatographic separation (AEX-HPLC (fraction C, 10.0–10.5 min) followed by SEC-HPLC (peak 1, 17–19 min)); b ~50 kDa Fe-binding biomolecule isolated by two-dimensional chromatographic separation (AEX-HPLC (fraction B, 9.5–10.0 min) followed by SEC-HPLC (peak 2, 23–26 min)). The results are presented as intensities (y-axis) versus m/z values (x-axis). The maxima of decisive peaks are annotated within the figure (m/z)

Full size image

LC-MS/MS analysis with MASCOT database search

LC-MS/MS analyses with MASCOT database search were performed in accordance with the method described in our previously published paper [14], which will be, further on, presented in detail. In preconcentrated AEX-SEC200-fractions, DTT (Honeywell Fluka, Germany) was again added in a concentration of 1 mM in order to reduce possible disulfide bonds in the proteins. After 1 h at room temperature, 54 mM iodoacetamide (Sigma-Aldrich, USA) was added to the reaction mixture to a final conncentration of 5.5 mM and left in the dark at room temperature for another hour. Finally, 1 µL of trypsin (1 µg µL−1; Trypsin Gold, Promega, USA) was added and the mixture was incubated overnight at room temperature. The reaction was stopped by adding trifluoroacetic acid (Sigma-Aldrich, USA) to a final concentration of 1%.

Analyses of obtained samples of tryptic peptides were performed via low-resolution ion trap LC-MS/MS (LC, Dionex Ultimate 3000 RSLC; MS/MS, Amazon ETD ion trap; Bruker Daltonik, Germany). Peptides were loaded onto a trap column (C18 resin, Acclaim Pepmap, 100 Å, 5 mm, 1 mm, 5 mm) in 1 mL of solution containing 0.1% formic acid (Fisher Chemical, UK). Peptides were separated on a capillary column (C18 resin, Acclaim Pepmap, 100 Å, 2 µm, 0.3 mm × 150 mm) at a flow rate of 3.5 μL min−1. Mobile phase A consisted of 0.1% formic acid in solution of HPLC grade water (98%)/acetonitrile (2%), and mobile phase B consisted of 0.1% formic acid in solution of acetonitrile (98%)/HPLC grade water solution (2%). The 45-min multistep gradient consisted of mobile phase B: 1 min 5%, 30 min linear gradient to 45%, 1 min linear gradient to 90%, 4 min hold on 90%, 3 min linear gradient to 5%, 6 min hold on 5%. The MS/MS was operated at an ESI capillary voltage of 4500 V. High voltage end plate offset was 500 V. The nebulizer was set at 10 psi.

The temperature of dry gas was set at 300 °C with a flow rate of 8 L min−1. Helium was used as a collision gas. The fragmentation amplitude was set at 0.60 V and ramped between 30 and 300% of this value. Product ion spectra were sequentially recorded for each selected precursor. The acquisition software was set up in autoMS/MS mode using up to three precursor ions with enabled active exclusion (precursor exclusion after two MS/MS spectra for 2 min). MS and MS/MS spectra were acquired within a scan range from 300 to 1500 m/z using averages from five spectra and a scan rate of 8100 (m/z) s−1. DataAnalysis software 4.0 (Bruker Daltonik GmbH, Germany) was used to extract MS and MS/MS data to create a MASCOT file for database search.

Search parameters were enzyme trypsin cleavage, allowed two missed cleavages, fixed modification carbamidomethyl and variable modification methionine oxidation, mass tolerance 0.5 Da. The NCBI non-redundant database version 18122019 and UniProt/Swiss-Prot sprot_15042019 were used as protein sequence databases. The results are presented in Tables 4 and 5, and in Electronic Supplementary Material.

Table 4 The main hits obtained by the LC-MS/MS analyses with MASCOT search of two databases for the following chromatographically separated and collected fraction: AEX-HPLC fraction B (9.5–10.0 min; Figs. 1 and 2) further fractionated by SEC-column (peak 2; 23–26 min; HPLC-estimated molecular mass of intact protein (MM): ~30–80 kDa; Fig. 2b). The presented results refer to obtained hits for Fe-containing proteins of adequate molecular masses, with the highest scores and overlapping in both databases. The complete MASCOT search results are presented as Supplementary Information
Full size table
Table 5 The additional hits obtained by the LC-MS/MS analyses with MASCOT search of two databases for the following chromatographically separated and collected fraction: AEX-HPLC fraction B (9.5–10.0 min; Figs. 1 and 2) further fractionated by SEC-column (peak 2; 23–26 min; HPLC-estimated molecular mass of intact protein (MM): ~30–80 kDa; Fig. 2b). The complete MASCOT search results are presented as Supplementary Information
Full size table

Data processing and graphical data presentation

All calculations (e.g., SEC-HPLC calibration, calculations of molecular masses associated to individual metal SEC-HPLC peaks) were executed in Microsoft Office Excel (version 16). The graphs were created in statistical program SigmaPlot 11.0 for Windows.

Results and discussion

We have started our study on metalloproteins/peptides in the liver of the northern pike (including Fe-binding biomolecules) by establishing the distributions of several elements of interest among the biomolecules of different sizes (by use of SEC-HPLC) [18] and their distributions among the biomolecules of different total charges (by use of AEX- and CEX-HPLC) [19].

Our previously published one-dimensional chromatographic separation of the northern pike hepatic cytosol applying the SEC200-column [18] revealed predominant Fe binding to HMM biomolecules. Major obtained Fe-peak in that study had HPLC-estimated molecular mass maximum at ~400 kDa (range ~200 to 600 kDa), near the elution time of standard protein apoferritin (molecular mass 443 kDa), leading to a conclusion that it referred to Fe-storage protein ferritin [18]. Another prominent Fe-peak reported in that study was obtained with maximum at ~50 kDa (range ~30 to 80 kDa; Fe-MMM-peak), and based on the comparison with the previously published reports [14,15,16,17, 25], it was hypothesized to be Hb [18]. We have also reported that increase in Fe bioaccumulation in the liver of the northern pike resulted with the increased quantity of Fe associated with the SEC200-HMM-peak, i.e., presumably ferritin [18].

Moreover, our previously published one-dimensional chromatographic separation of the northern pike hepatic cytosol applying the AEX/CEX-columns [19] revealed that more than 95% of Fe-metalloproteins bind to the AEX-column at physiological pH (7.4), i.e., has negative total charge and acidic isoelectric point. Moreover, in that study, Fe was found in five peaks eluted from the AEX-column (Fig. 1) [19]: the first two were poorly resolved (max. at Et of 9.1 and 9.3 min) and, compared to standard proteins, had pIs close to neutral values, whereas the majority of eluted Fe was associated with the third AEX-peak (max. at Et of 10.6 min) which, compared to standard proteins, had pI probably within pH range of 4–6; the remaining two Fe-peaks were observed with maxima at Et of 14.7 and 18.8 min, indicating more acidic Fe-binding biomolecules. Based on the comparison with our previously published SEC200-analysis [18], we have presumed that the major Fe-AEX-peak referred to ferritin, whereas the first two peaks (containing second highest Fe-quantity) possibly referred to Hb. The plausibility of that conclusion was further corroborated by the known pIs of these two Fe-metalloproteins (pIferritin in fish liver, 4.1–7.0 [6]; pIHb in fish liver, 5.9–8.1 [26]) (see Electronic Supplementary Material, Table ESM 2). We have also reported that increase in Fe bioaccumulation in the liver of the northern pike resulted with the increased quantity of Fe associated with the probable AEX-ferritin peak [19].

However, to be able to associate undoubtedly the Fe-SEC200-peaks of particular molecular masses with specific Fe-AEX-peaks, i.e., to define the relations between the main Fe-binding biomolecules regarding their molecular masses and total charges, two-dimensional chromatographic separation had to be performed, with either additonal AEX-HPLC separation of chosen SEC200-peaks, or additional SEC200-HPLC separation of chosen AEX-peaks. Two-dimensional HPLC-fractionation is further beneficial because it enables the purification of the biomolecules of interest and their more reliable identification by further analysis using mass spectrometry methods [14]. Namely, when SEC200-HPLC fractionation was applied, coelution of Fe with several other elements was observed (Bi, Co, Mn, Mo, and Zn coeluted with HMM-Fe-peak; Bi, Co, Cu, Mn, Se, and Zn coeluted with MMM-Fe-peak) indicating that each SEC200-peak contains a mixture of metalloproteins [18]. The same conclusion could be made for fractionation by AEX-column, where Fe-binding biomolecules were eluted together with biomolecules that bind Ag, Bi, Cd, Co, Cu, Mn, Pb, Tl, and Zn [19].

Thus, as a starting point in this study, we have chosen Fe-AEX-profile (Fig. 1), and for further SEC200-HPLC separation collected the AEX-fractions containing the first three major Fe-AEX-peaks (eluted from 8.5 to 11.5 min). Since the peaks are placed rather close to one another, and it is difficult to separate them clearly due to a possibility of minor variability of elution time between chromatographic runs, we have collected and pooled four fractions, as follows: AEX-fraction A, 8.5–9.5 min; AEX-fraction B, 9.5–10.0 min; AEX-fraction C, 10.0–10.5 min; and AEX-fraction D, 10.5–11.5 min.

SEC200-HPLC-ICP-MS analysis of Fe in selected AEX-HPLC fractions, as a second dimension of fractionation of Fe-binding biomolecules

The Fe-profiles obtained by SEC200-HPLC-ICP-MS for four above-mentioned AEX-fractions are presented in Fig. 2. As said above, the tails of AEX-Fe-peaks overlapped to a certain degree, and thus in each fraction, it was possible to observe appearance of one Fe-peak and/or disappearance of another. Accordingly, it can be observed that Fe-binding biomolecule of molecular mass of ~50 kDa appears already in the earliest AEX-fraction (A, 8.5–9.5 min; Fig. 2a), and has its maximum in the second one (B, 9.5–10.0 min; Fig. 2b), and then gradually disappears (Fig. 2c, d). On the other hand, Fe-binding biomolecule of molecular mass of ~360 kDa appears already in the second AEX-fraction (B, 9.5–10.0 min; Fig. 2b), but has its maximum in the last two fractions (C and D, 10.0–11.5 min; Fig. 2c, d). The results obtained by two-dimensional chromatographic separation have thus confirmed our hypothesis on the association between the Fe-SEC200-peaks of particular molecular masses with specific Fe-AEX-peaks. It was confirmed that presumable ferritin peak of higher molecular mass has earlier SEC200-elution time and later AEX-elution time (more acidic pI) compared to the other major Fe-binding biomolecule, presumably Hb, of lower molecular mass, i.e., later SEC200-elution time, and earlier AEX-elution time (pI closer to neutral values).

Since our aim was also to purify Fe-binding biomolecules, to be able to identify them more reliably, in addition to Fe we have measured 13 other elements in AEX-fractions separated by SEC200-column, and their peak maxima are presented in Table ESM 1 (see Electronic Supplementary Material). It can be seen that the majority of metals/metalloids were separated from Fe, and overlapping peaks with 50 kDa Fe-peak were found only for Mn, Pb, and Se (see Electronic Supplementary Material, Table ESM 1; Fig. 3), and with 360 kDa Fe-peak for Pb and Zn (see Electronic Supplementary Material, Table ESM 1; Fig. 4). This, however, had to be considered during the interpretation of mass spectrometry analysis, for which we have chosen the fractions marked in Fig. 2b (~50 kDa; collected from 23 to 26 min) and Fig. 2c (~360 kDa; collected from 17 to 19 min).

MALDI-TOF-MS analysis of selected AEX-SEC200-HPLC fractions

The analysis of 360 kDa Fe-peak resulted with a MALDI-peak at m/z equal to 20.65 kDa (Fig. 5a). This finding closely corresponds to molecular mass of ferritin subunits previously reported for the northern pike liver (20.4 kDa) [10], thus indicating that the detected peak refers to single charged ferritin subunit. Similar molecular mass was reported for the liver of S. zygaena (~21 kDa) [7]. The finding of subunits instead of the intact ferritin molecule is very common since, as explained by Huang et al. [7], even high laser intensity in MALDI-TOF MS does not have capacity to make a molecular ion of a whole protein for mass determination of ferritin, because its molecular mass is too big (>400 kDa). However, high laser intensity can weaken binding energy of noncovalent bonds among ferritin subunits, creating their molecular ions detectable by mass spectrometry [7]. Huang et al. [7] managed to detect single charged ferritin subunit of 21.07 kDa in the liver of S. zygaena, together with double charged monomers, and single charged dimers and trimers, but only after they applied high laser intensity, whereas application of low laser intensity did not produce visible spectra. In our study, even laser intensity of 90% resulted with an m/z peak characteristic for ferritin subunit in only one sample and having very low intensity (~100; Fig. 5a), indicating the possibility that the bonds among ferritin subunits in the liver of the northern pike, without any pretreatment, were too strong for the potential of the applied instrument. However, there is another possible explanation for absence/low intensity of MALDI-peak of ferritin in the collected HMM-fraction of the northern pike liver. There is a possibility that the concentration of this Fe-binding biomolecule was too low in our collected HPLC-fractions. Although the intensity of Fe in this HMM-peak was high (Fig. 2), it has to be taken into consideration that one molecule of ferritin can store up to 4500 Fe3+ ions [27, 28], and thus high intensity of Fe does not necessarily reveal high protein concentration. Accordingly, our further research will be focused on the optimization of the pretreatment and conditions for mass spectrometry analysis of Fe-binding proteins of high molecular masses.

The analysis of 50 kDa Fe-peak obtained from two fish samples resulted with a detectable MALDI-spectra (Fig. 5b), containing several m/z peaks. The main peak was observed at 16.05 kDa, presumably indicating single-charged Hb-subunit, which is typical MALDI-spectra for Hb according to available literature [14, 29, 30]. For example, in hemocytes of whiteleg shrimp Litopenaeus vannamei, Hb was detected as a subunit (Hb β) of 16.12 kDa [29], and in the liver of Vardar chub (S. vardarensis) as a subunit (Hb β) of 15.4 kDa [14]. It is not unusual to detect subunits by MALDI-TOF MS instead of intact protein, because, as mentioned above, the laser in this method can decompose unstable multisubunit proteins into the free subunits [7]. In addition, we have observed much smaller peaks at 32.12 kDa and 48.19 kDa, as a possible sign of presence of dimers and trimers of Hb-subunits, whereas tetramer (i.e., intact protein) was not detected. A similar finding was reported for Vardar chub liver, with detected dimers (31.5 kDa) and trimers (46.9 kDa) [14]. A peak recorded at 7.95 kDa possibly indicated double-charged Hb-subunit (Fig. 5b). The same was also reported for human blood, where Hb was detected by MALDI-TOF-MS as a double charged subunit (Hb α-chain) with m/z equal to 7.56 kDa [30]. In the MALDI-TOF-MS spectra, several other small peaks were observed, which can be explained by the fact that the isolated protein was not completely purified, but contained the traces of metalloproteins that bind Mn, Pb, and/or Se (see Electronic Supplementary Material, Table ESM 1; Fig. 3).

LC-MS/MS analysis of selected AEX-SEC200-HPLC fraction with subsequent MASCOT search

Finally, only Fe-binding protein of HPLC-estimated molecular mass equal to ~50 kDa (AEX-Et, 9.5–10.0 min; SEC200-Et, 23–26 min), that was isolated from the liver of the northern pike from the Mrežnica River, was detected by LC-MS/MS analysis. Subsequent MASCOT search of two databases (UniProt/Swiss-Prot, NCBInr) recognized it as Hb-subunit β (Hb β), based on the similarities with Hb β sequences stored in the databases for northern pike and sockeye salmon (Oncorhynchus nerka) (Table 4). Our previous LC-MS/MS study combined with MASCOT search of corresponding Fe-binding protein from the liver of Vardar chub also indicated the presence of Hb β subunits, but the similarities were obtained with Hb β sequences stored in the databases for fishes Danio rerio, Carassius auratus, and Arctogadus glacialis [14].

Among the different fish species, the high degree of genetic variation was detected [13, 31]. Thus, the studies of metal-binding biomolecules conducted on various species enable the identification of similarities/differences in metal handling strategies among different fish species, families, and/or orders. Based on our two so far performed studies of this kind, we have identified the following grouping regarding the Fe-sequestering by Hb within the class Teleostei (teleosts) (Table 4; Krasnići et al., 2019): (1) [Esociformes (order, pikes and mudminnows) > Esocidae (family, pikes) > Esox lucius Linnaeus, 1758 (species, the northern pike)] was associated with [Salmoniformes (order, salmons) > Salmonidae (family, salmons) > Oncorhynchus nerka (Walbaum, 1792) (species, the sockeye salmon)]; (2) [Cypriniformes (order, carps) > Leuciscidae (family, minnows) > Squalius vardarensis Karaman, 1928 (species, the Vardar chub)] was associated with [Cypriniformes (order, carps) > Danionidae (family, danios) > Danio rerio (Hamilton, 1822) (species, the zebrafish)], [Cypriniformes (order, carps) > Cyprinidae (family, minnows or carps) > Carassius auratus (Linnaeus, 1758) (species, the goldfish)], and [Gadiformes (order, cods) > Gadidae (family, cods and haddocks) > Arctogadus glacialis (Peters, 1872) (species, the Arctic cod)]. Similarities in metal-handling strategy between esocids and salmonids, and between leuciscid and cyprinid fish, were already observed when Zn distribution among cytosolic biomolecules was studied using SEC-HPLC-ICP-MS [18].

Furthermore, MASCOT search resulted with several additional hits (Table 5). Some of them were immediately dismissed as improbable (DNA-directed RNA polymerase subunit β, caskin-2, pre-mRNA-splicing factor prp12, genome polyprotein, tRNA (Met) cytidine acetyltransferase; Table 5) due to much higher mass of the intact protein or a subunit (molecular mass, ~120 to 330 kDa; Table 5) compared to molecular mass estimated by SEC200-HPLC (~30 to 80 kDa; Fig. 2). In some cases, an additional reason for dismissal was protein intracellular location (e.g., nuclear placement of prp12 [32]) or species of origin (genome polyprotein inherent to viruses [33]). Five MASCOT hits seemed plausible according to molecular masses (18–46 kDa); however, some other features rendered most of them unlikely. Since we have isolated soluble, cytosolic, hepatic fraction, containing additionally only microsomes [22], we have concluded that most likely the protein characteristics for organelles and membrane were not contained within the studied fraction. Those included ribosomal proteins (ribosomal RNA large subunit methyltransferase, 30S ribosomal protein S3; Table 5) and transmembrane transporter (ATP-binding cassette domain-containing protein [34]). Which left us with only two additional options, i.e., two enzymes of acceptable molecular mass (fumarylacetoacetase (FAH), 46.5 kDa; and malate dehydrogenase (MDH), 38.4 kDa; Table 5), which have pI close to neutral pH values, and cytoplasmic placement in the cell (according to Malécot et al. [35] and Table 5, respectively). According to literature reports, FAH is a cytosolic homodimer with two 46-kDa subunits, where equilibrium always exists between dimer and monomer [36], and for which pI of 6.2 was reported in medaka fish (Oryzias latipes) [35] (see Electronic Supplementary Material, Table ESM 2). MDH, cytosolic isoform, is also a homodimer, and in teleost fish, the reported mass of its subunit was ~36 kDa [37] and a pI ~7.0 [38] (see Electronic Supplementary Material, Table ESM 2). Neither of these two enzymes is an Fe-containing protein, as will be demonstrated below. However, the cytosolic fraction isolated in our study from the northern pike liver could additionally contain the proteins that do not bind metals, as well as proteins that bind some other metals, both those that we measured and those that we did not. Moreover, as mentioned earlier, we have found traces of Mn, Pb, and Se in the analyzed fraction (see Electronic Supplementary Material, Table ESM 1; Fig. 3), revealing the possibility that our sample contained, in addition to Fe-binding biomolecules, the compounds that contain one of these three elements. The score for FAH was rather low (28–43), the species of its origin was very distant from the studied one (domestic cattle, Bos taurus) (Table 5), and the possible cofactors of this enzyme include Mg2+ and/or Ca2+ [39, 40], but we cannot exclude the possibility that our sample contained this enzyme or its monomer. On the other hand, malate dehydrogenase had considerably higher score (97), the species of origin was the same as the studied one (the northern pike) (Table 5), and the known cofactor of this enzyme is Mn2+ [41]. Detection of Mn in our sample (Fig. 3a), as well as a small, but clearly visible m/z MALDI-peak at approximately ~36 kDa (Fig. 5b), leads to a conclusion that our studied fraction, in all likelihood, contained, in addition to Fe-binding Hb, Mn-containing malate dehydrogenase or its subunit.

Conclusions

The analytical approach applied in this study, namely two-dimensional chromatographic separation, enables determination of relationship among various metalloproteins according to their molecular masses and net charges. Furthermore, it is also useful in partial purification of particular metalloproteins, in this case Fe-binding proteins, thus allowing more reliable identification with mass spectrometry techniques. Accordingly, the combination of chromatographic separation and mass spectrometry analysis in this study of Fe-binding biomolecules in the northern pike liver revealed ferritin and Hb as two main participants in Fe-metabolism in this organ, the latter one being the sign of hepatic blood perfusion. Apparent molecular masses of two proteins, estimated by SEC-HPLC, were ~360 kDa and ~50 kDa, respectively, with ferritin having more acidic pI values than Hb, according to AEX-chromatographic separation. The presence of ferritin in the hepatic samples was further demonstrated by detection of probable ferritin subunit at 20.65 kDa by MALDI-TOF-MS. The presence of Hb in the hepatic samples was, on the other hand, confirmed by two techniques of mass spectrometry. MALDI-TOF-MS detected typical m/z spectra with the predominant peak referring to single-charged Hb-monomer (16.1 kDa). LC-MS/MS with subsequent MASCOT database search recognized Hb-β subunits of the northern pike and the sockeye salmon, thus pointing to connection between esocid and salmonid fishes. Further recognition of Mn-binding malate dehydrogenase in our samples revealed that additional separation methods could be applied to obtain better degree of protein purification.