Quantitative Proteomic Analysis of Venoms from Russian Vipers of Pelias Group: Phospholipases A2 are the Main Venom Components

Venoms of most Russian viper species are poorly characterized. Here, by quantitative chromato-mass-spectrometry, we analyzed protein and peptide compositions of venoms from four Vipera species (V. kaznakovi, V. renardi, V. orlovi and V. nikolskii) inhabiting different regions of Russia. In all these species, the main components were phospholipases A2, their content ranging from 24% in V. orlovi to 65% in V. nikolskii. Altogether, enzyme content in venom of V. nikolskii reached ~85%. Among the non-enzymatic proteins, the most abundant were disintegrins (14%) in the V. renardi venom, C-type lectin like (12.5%) in V. kaznakovi, cysteine-rich venom proteins (12%) in V. orlovi and venom endothelial growth factors (8%) in V. nikolskii. In total, 210 proteins and 512 endogenous peptides were identified in the four viper venoms. They represented 14 snake venom protein families, most of which were found in the venoms of Vipera snakes previously. However, phospholipase B and nucleotide degrading enzymes were reported here for the first time. Compositions of V. kaznakovi and V. orlovi venoms were described for the first time and showed the greatest similarity among the four venoms studied, which probably reflected close relationship between these species within the “kaznakovi” complex.


Introduction
Venomous snakes inhabit all continents of the globe except Antarctica. They are particularly abundant in tropical areas of Asia, Africa, South America and Australia. Russia, despite its large territory, is inhabited by only a small number of poisonous snake species, which belong to three genera: Gloydius, Macrovipera and Vipera. The Vipera genus is the most speciose in Russia and includes more than ten species, the systematics within this genus being constantly updated [1,2]. The most abundant species is common (or European) adder Vipera berus, which has a very large habitat in Russia, ranging from its western borders to Sakhalin and the Ussuri region. V. berus is also spread throughout Europe-between 68 and 45 degrees north latitude. The venom of this species is fairly well studied. Biological activities of this venom were characterized and proteolytic, fibrinolytic, anticoagulant, and phospholipolytic ones were demonstrated by in vitro experiments [3]. Several toxic proteins were isolated from V. berus venom, including phospholipase A 2 (PLA2) [4], metalloproteinase (SVMP) [5], L-amino acid oxidase (LAAO) [6] and several others. Recently, we have partially characterized the steppe viper V. renardi venom, the PLA2s and Kunitz type protease inhibitors were isolated from this venom and sequenced [7]. The isolated PLA2s were studied in more details and found to exert their 9 Natriuretic peptide Pseudonaja textilis B-NAP 14

Composition of Russian Viper Venoms
As a result of venom protein quantification, it was found that the main venom components were PLA2s; their content ranged from about 24% in V. orlovi venom to more than 60% in V. nikolskii (Table 2, Figure 2). The overwhelming majority of PLA2s belonged to D49 subgroup of group IIA as it might be expected for the snakes from Viperidae family. The venom of V. nikolskii contained PLA2s only from this group. One PLA2 of S49 subgroup was highly represented in V. renardi. One PLA2 of group IA was observed in small amounts in three venoms and a low quantity of group IIE PLA2 was detected in V. renardi venom.

Composition of Russian Viper Venoms
As a result of venom protein quantification, it was found that the main venom components were PLA2s; their content ranged from about 24% in V. orlovi venom to more than 60% in V. nikolskii (Table 2, Figure 2). The overwhelming majority of PLA2s belonged to D49 subgroup of group IIA as it might be expected for the snakes from Viperidae family. The venom of V. nikolskii contained PLA2s only from this group. One PLA2 of S49 subgroup was highly represented in V. renardi. One PLA2 of group IA was observed in small amounts in three venoms and a low quantity of group IIE PLA2 was detected in V. renardi venom. Altogether, the enzyme content in venom of V. nikolskii reached about 85%, however this venom was characterized by a very low content of SVMPs (less than 1%) and LAAO (less than 0.1%). PLA2s accounted for more than 40% in V. kaznakovi and V. renardi venoms. While the content of SVMPs was less than 1% in the V. nikolskii venom, they comprised 12%-16% in V. kaznakovi, V. orlovi and V. renardi. The highest content of SPs was in V. orlovi venom (24%) and the lowest in V. renardi (8%). LAAO was at the level of 4%-5% in all the analyzed venoms with the exception of V. nikolskii. Nucleic acid degrading enzymes (Nuc) represented about 2% in V. orlovi venom and less that 1% in all the others. Altogether, the enzyme content in venom of V. nikolskii reached about 85%, however this venom was characterized by a very low content of SVMPs (less than 1%) and LAAO (less than 0.1%). PLA2s accounted for more than 40% in V. kaznakovi and V. renardi venoms. While the content of SVMPs was less than 1% in the V. nikolskii venom, they comprised 12%-16% in V. kaznakovi, V. orlovi and V. renardi. The highest content of SPs was in V. orlovi venom (24%) and the lowest in V. renardi (8%). LAAO was at the level of 4%-5% in all the analyzed venoms with the exception of V. nikolskii. Nucleic acid degrading enzymes (Nuc) represented about 2% in V. orlovi venom and less that 1% in all the others. Phospholipase B (PLB) was found in all venoms (less than 1%) and very low amount of Hya (0.01%) was detected in three venoms. Among the non-enzymatic proteins, Dis (13%) in the V. renardi venom, CTL (12%) in V. kaznakovi, CRISPs (12%) in V. orlovi and vascular endothelial growth factors (VEGF, 8%) in V. nikolskii were the most abundant ones in the venoms studied. The total amount of non-enzymatic proteins was about 13% in the V. nikolskii venom and about 27%-28% in all the others. In addition to the proteins mentioned above, nerve growth factor (NGF) and Kunitz were present in all the venoms (less than 1%) with exception of V. kaznakovi, where Kunitz was absent.
Interestingly, comparison of both the nature of the identified proteins and their abundance showed very close venom compositions for the species V. kaznakovii and V. orlovii. The Pearson correlation coefficient for individual protein abundance LFQ was 0.83, while for the rest of pairs the correlation coefficient varied from 0.16 to 0.34 ( Figure 3). Phospholipase B (PLB) was found in all venoms (less than 1%) and very low amount of Hya (0.01%) was detected in three venoms. Among the non-enzymatic proteins, Dis (13%) in the V. renardi venom, CTL (12%) in V. kaznakovi, CRISPs (12%) in V. orlovi and vascular endothelial growth factors (VEGF, 8%) in V. nikolskii were the most abundant ones in the venoms studied. The total amount of nonenzymatic proteins was about 13% in the V. nikolskii venom and about 27%-28% in all the others. In addition to the proteins mentioned above, nerve growth factor (NGF) and Kunitz were present in all the venoms (less than 1%) with exception of V. kaznakovi, where Kunitz was absent.
Interestingly, comparison of both the nature of the identified proteins and their abundance showed very close venom compositions for the species V. kaznakovii and V. orlovii. The Pearson correlation coefficient for individual protein abundance LFQ was 0.83, while for the rest of pairs the correlation coefficient varied from 0.16 to 0.34 ( Figure 3).

Identification of Endogenous Peptides in the Venoms
It is well known that snake venoms may contain various peptides: several peptide families including bradykinin-potentiating peptides, natriuretic peptides, sarafotoxin, etc. were identified [17]. Moreover, the venoms studied in this work contain proteases, therefore their proteins may undergo proteolysis leading to generation of peptides. To study endogenously generated peptides in the venoms of interest, high molecular weight (MW) proteins were separated by ultrafiltration (10 KDa cut-off). The peptide fractions obtained were analyzed by LC-MS/MS in the same fashion as proteins, but without preliminary proteolysis. Peptides were searched at first against a full NCBI Serpentes database by Mascot search engine with 10% protein FDR (False Discovery Rate). A fusion database containing the peptidogenic proteins from Mascot search and the SwissProt Serpentes database was used for the final search in MaxQuant. Full NCBI database search with unspecific

Identification of Endogenous Peptides in the Venoms
It is well known that snake venoms may contain various peptides: several peptide families including bradykinin-potentiating peptides, natriuretic peptides, sarafotoxin, etc. were identified [17]. Moreover, the venoms studied in this work contain proteases, therefore their proteins may undergo proteolysis leading to generation of peptides. To study endogenously generated peptides in the venoms of interest, high molecular weight (MW) proteins were separated by ultrafiltration (10 KDa cut-off). The peptide fractions obtained were analyzed by LC-MS/MS in the same fashion as proteins, but without preliminary proteolysis. Peptides were searched at first against a full NCBI Serpentes database by Mascot search engine with 10% protein FDR (False Discovery Rate). A fusion database containing the peptidogenic proteins from Mascot search and the SwissProt Serpentes database was used for the final search in MaxQuant. Full NCBI database search with unspecific digestion failed in MaxQuant due to internal software limitations. In summary, 512 endogenous peptides from 80 proteins belonging to 13 protein families were found (Table S3). As expected, most of the peptides (462 peptides) belonged to proteins (48 proteins) which were earlier found in proteome, thus most likely representing venom protein degradation in vivo as a result of proteases and peptidases activity. At the same time, 50 peptides (Table S4) belonged to 32 unique proteins from nine protein families (Table 3).
Among these, proteins in six families mostly had one peptide per protein, which can explain their identification only in the peptidome analysis as a result of a very low concentration of original proteins before degradation, so they were missed in the shotgun MS/MS selection in proteome analysis (or were beyond the taken FDR cut off). The largest number of unique peptides was found in proteins belonging to Dis and SVMP/Dis families: 25 peptides from 14 proteins were found. The peptides identified were mainly fragments of larger venom proteins. However, we found 12 peptides from proteins belonging to B-NAP family (Figure 4). These peptides may represent real endogenous peptides and possess their own biological activity. This is the first indication for the presence of bradykinin-potentiating and natriuretic peptides in venoms of vipers from the Pelias group. found 12 peptides from proteins belonging to B-NAP family (Figure 4). These peptides may represent real endogenous peptides and possess their own biological activity. This is the first indication for the presence of bradykinin-potentiating and natriuretic peptides in venoms of vipers from the Pelias group.

Discussion
We have analyzed venom proteomes and peptidomes for four species of Vipera, for which there is no genomic or transcriptomic data published. For each species, the venoms of at least 15 individual animals were pooled for the analysis. Protein identification for such "non-sequenced" species is problematic for inherently database oriented bottom-up LC-MS/MS-based proteomics [18]. A possible solution is to use the protein sequences of closely related species, based on the assumption of their high homology level [18]. Thus, when the exact protein sequence is missing in the database, the protein might still be identified by partial/full homology with a known protein of another species. Here, we searched LC-MS/MS data against the database containing all the proteins from the taxon Serpentes in the NCBI database on the date of the experiment (the results are given in Table 1).
In the bottom-up proteomics, there are two major approaches for the quantitative analysis: (a) relative quantification of a single protein across samples; and (b) comparison of different proteins within a single sample. Principle (a) is based on the measurement of all the peptides belonging to a protein in several samples (and pair-wise peptide Fold Changes estimation) followed by protein Fold Change calculation as, e.g., mean or median value of the peptide fold changes. Principle (b) is based on the assumption that the sum of peptide peak areas (either all or just some of them, like in the top 3 theory [19,20]) for a given protein is proportional to its absolute abundance. Thus, comparison of these sums for two proteins is supposed to give the difference in their content within one sample. What is the most important, when making a comparison between several samples, the two approaches (a) and (b) are supposed to give consistent results.
In case of protein analysis of "non-sequenced" species, both these approaches encounter significant albeit different problems arising from the incomplete peptide identification due to the lack of adequate protein amino acid sequences in the search database. Principle (a) works best when as many as possible shared peptides per protein are identified and quantified for a pair of samples, since individual peptide measurements are prone to err due to possible post-translational modifications or isoforms. When it comes to different species, the number of shared peptides between samples goes down just because of different protein sequences. Besides, this approach works only when there are shared peptide sequences identified and quantified in both samples (recommended number of shared peptides for a reliable quantitation is 2). Thus, if a protein is unique for a sample, it cannot be quantified this way at all. Besides, it provides no data for concentration comparison between different proteins within a single sample.
Principle (b) was developed and verified for systems (artificial protein mixtures) where all the best flyer peptides for a protein (the peptides which have the best proportion between peptide concentration and intensity and thus have the maximum impact on the summed protein intensity) can be easily identified and quantified [19]. For "non-sequenced" species it would mess the final results through protein abundance underestimation if the missed peptides were among the best flyers for the given protein of some particular species but were overlooked because their amino acid sequence was missing in the database. For that, peptide MS/MS sequencing de novo might help a bit, but many peptides would still be missed for the reasons that are not clarified.
Here, we used two approaches to quantify proteins. First, we used MaxLFQ approach [21] which is basically principle (a), but it also uses absolute peptide intensities in addition to peptide FC comparison between samples (such results are labeled LFQ in Table 2). Second, we used direct comparison of sums of peptide intensities to make quantitation within each sample (such results are labeled INT in Table 2). The results of protein quantitation made by different methods gave quite similar results ( Table 2, Table S1), especially when potential errors in individual protein contents were compensated by consolidation of proteins into families.
There is also a question of which types of peptides should be used for protein quantitation. Protein identification process deals not with separate proteins, but with protein groups, which are sets of individual proteins (at least partially homologous) sharing a set of identified peptide sequences. In the absence of unique specific peptides, no distinction between these proteins within a group can be made. A standard approach is to take as a hit the protein from a protein group which has a maximum number of assigned identified peptides. Thus, there are three types of peptides for a single protein group in the identification list: unique, razor and other (shared) peptides (MaxQuant terminology) [22,23]. Usually, protein groups have some unique peptides to pinpoint them as "correct" hits, but it is also possible that the number of unique peptides for a protein is zero. Absence of unique peptides means that all the peptides from the current protein group are shared and can be just as successfully assigned to some other protein groups. In such situation, the final set of protein groups shown in the identification list is the minimal one sufficient to explain all the identified peptides (Occam's razor principle). Shared peptides are named "razor" when they belong to the protein group with the maximum total number of peptides among other possible protein groups. These razor peptides are used for quantitation (along with unique peptides), both LFQ and intensity based [24]. A shared peptide, which is "razor" for some particular group, is counted in "all peptides" in all the protein groups to which it can be potentially assigned, and "all peptides" list is used to calculate Sequence Coverage.
Importantly, any analytical method may prove only that the amount of the compound under investigation is below the method sensitivity, rather than show the absolute absence of the compound in the sample. This is specifically applicable for the LC-MS/MS-based shotgun identification principle which selects peptide ions pseudo-randomly, sometimes missing the peptides with very low intensities just because of a wrong choice. Thus, quantitation is much more reliable for showing the absence of the compound (or, more accurate, the concentration being lower than its Low Limit of Detection). MaxQuant features chromatogram alignment and the possibility to quantify peptides on the basis of similarity of their retention time and m/z in the sample, where they were identified by MS/MS and in another sample where this particular m/z signal got lost during the shot-gun selection for the MS/MS analysis (proteins with such peptides are marked "By matching" in "Identity Type" column in Tables S1 and S2). In this work the protein is considered to be identified (and considered as present) in the sample only if it has an MS/MS spectrum identified in this particular sample. However, for quantitation both MS/MS identified peptides and the peptides identified on the basis of the above described similarity were used. This might lead to apparent contradictions when there are no proteins identified, but the protein abundance is non-zero (like natriuretic peptides (B-NAP) in the V. nikolskii venom-0.01/0.01 (0) in Table 2).
At the present time, the genus Vipera includes 22 species, however it is not homogenous. Molecular phylogeny studies showed that this genus comprises the V. aspis group, the V. ammodytes complex, and the Pelias group as separate clades [25]. Of these clades, only snakes from the Pelias group inhabits Russia. The Pelias was further classified into two subgroups, one comprising V. dinniki, V. kasnakovi, and V. ursinii, and another including V. berus, V. barani, V. nikolskii, and V. seoanei [25]. The first subgroup was further subdivided into the "kaznakovi" complex, including V. kasnakovi, V. orlovi and some other closely related species, and the "ursinii" complex, in which V. renardi was included [26,27]. Earlier, for the vipers of the Pelias group, we have studied the venom toxicity towards crickets Gryllus assimilis [12] and found that it differed depending on feeding preferences. The snakes from the V. renardi, V. lotievi, V. kaznakovi, and V. orlovi species feed on a wide range of animals including insects, whereas the snakes from V. berus and V. nikolskii species do not include insects in their diet. The venom from vipers which hunt insects was found to possess a greater toxicity towards crickets. This suggests that the venom composition may greatly differ among these species. As concerns the toxicity to other animals, it was shown that the venom of V. nikolskii was more toxic than that of V. berus to frogs (9-11 µg/g vs. 30-52 µg/g) and mice (0.93 vs 1.58 µg/g) at intraperitoneal injection [28]. The venom of V. renardi was less toxic to mice (2.96 µg/g) than that of V. berus [28]. We were not able to find any data about toxicity of V. orlovi and V. kaznakovi venoms.
Regarding the danger to humans, the data about bites by these snakes are sparse. Most of the documented cases refer to steppe viper V. renardi and report that it usually has calm and timid behavior, is reluctant to bite, and seeks to escape. This viper bites only when it is in danger, for example, if the snake is suddenly stepped on or picked up. V. renardi is considered less dangerous to humans than common adder. The human fatalities as a consequence of steppe viper bites are not reliably known [29], though there are some cases of the death of horses and small ruminants. A picture of human envenomation is characterized mainly by local signs which include severe pain at the site of the bite, redness, swelling that spreads far beyond the site of the biting. In severe cases, drowsiness, dizziness, nausea, increase of heart rate, and reduction in body temperature may be observed [30].
Records of the bites of humans by the Caucasian viper V. kaznakovi and Nikolsky's viper V. nikolskii are practically absent. However, V. kaznakovi may be dangerous. Solitary human deaths and livestock losses after Caucasian viper bites were mentioned [30]. We were able to find only one report about human fatalities after the Nikolsky's viper bites [31]. No information on the V. orlovi bites is available.
It should be noted that the venoms of not all Pelias species were studied equally well. The venom of V. berus is the best characterized. As mentioned earlier, the V. berus venom displayed in vitro proteolytic, fibrinolytic, anticoagulant, and phospholipolytic activities. In mice, significant local tissue-damaging effects, including edema, hemorrhage and myonecrosis were observed for this venom [3]. Several proteins involved in manifestation of those effects were isolated from V. berus venom. These proteins included basic PLA2 [4], SVMP [5], LAAO [6] and some others.
The V. nikolskii species is phylogenetically very close to V. berus and is included in the same subgroup within the Pelias group. It is regarded as a V. berus subspecies in some publications. However, the analysis of the V. nikolskii venom has shown it to differ greatly from that of V. berus. Thus, two heterodimeric PLA2s were isolated from the V. nikolskii venom [10], but similar proteins are absent in V. berus. The data obtained in the present work are in good agreement with the published results; the basic and acidic PLA2 subunits forming heterodimeric enzymes account for more than 50% of the V. nikolskii venom (Table 1). Earlier, cDNA encoding SP nikobin and Kunitz type inhibitor in the V. nikolskii venom gland was cloned and sequenced [11]. In this study we have found that nikobin is the main SP in the V. nikolskii venom (more than 12% of the total protein content, Table 1) and Kunitz-type serine protease inhibitor ki-VN was also the main protein of the Kunitz family in this venom (about 0.6%, Table 1). CRISP, the sequence of which was also deduced from cDNA analysis [32], was found in the venom in fairly low amount (0.66%, Figure 2). Interestingly, the content of CRISPs was much higher in other venoms studied and accounted for 8%, 10% and 12% in V. renardi, V. kaznakovi and V. orlovi venoms, respectively ( Figure 2).
The steppe viper V. renardi is included in the "ursinii" complex [33] while the other two vipers, V. kaznakovi and V. orlovi, belong to the "kaznakovi" complex. Among these vipers only the composition of the V. renardi venom was in some way studied [7]. The amino acid sequences for several PLA2s and Kunitz-type inhibitor were deduced from the cloned cDNA of venom gland. Some PLA2s and Kunitz protein were isolated from the venom. The most abundant was ammodytin I2d analogue. In this work we have found all the PLA2s described by Tsai et al. [7] in the V. renardi venom, Vur-PL2 having the highest content (Table 1). Interestingly, this viper venom has very high content of disintegrins which accounts for about 13% of total protein, while the V. kaznakovi and V. orlovi venoms contain less than 1% and in the V. nikolskii venom no disintegrins were detected.
There are no published data on the composition of the V. kaznakovi and V. orlovi venoms and they are characterized for the first time in this work. These two venoms have the highest similarity among the four ones studied (Figure 3) that confirms the inclusion of V. orlovi in the "kaznakovi" complex. They have a fairly high content of SVMPs (15%-16%), CTL (11%-12%) and CRISPs (11%-12%) ( Figure 3). The V. orlovi venom has the highest amount of SP (24%) among the four venoms studied ( Figure 3) and only V. kaznakovi contains a small quantity of hyaluronidase (Hya) at the level of 0.01%. However no Kunitz type proteins were detected in the latter venom.
Although a limited number of B-NAP proteins (one in V. kaznakovi and one in V. orlovi, Table 2) were detected in the proteome analysis, several peptides derived from proteins of this family were found in the peptidomes of all the venoms studied ( Figure 4). The mature bradykinin-potentiating peptide QGGLPRPGPEIPP was observed in the V. nikolskii venom and several fragments of similar peptides were detected in the other analyzed venoms. Several fragments of C-type natriuretic peptides were found in all four venoms as well (Figure 4). It should be noted that no bradykinin-potentiating and C-type natriuretic peptides from the vipers of Pelias group were reported so far.
In total, 210 proteins (Table 1) and 512 endogenous peptides (Table S3) were identified in four viper venoms. The overwhelming majority of the proteins (98%-99% of the total protein content) and the peptides represented 14 snake venom protein families ( Table 2). The comparison of our results with those for other snakes of the Vipera genus shows higher representation of venom protein families in our data (Table 4). For example, while Nuc and PLB were found in all venoms studied in this work, no proteins of these families were reported for other venoms from the Vipera species (Table 4). Hya was observed in three of the studied venoms and this is also the first indication for the presence of this enzyme in the venoms of the Vipera species. We have found that the main components of all venom studied are PLA2s, while SVMPs were prevailing in venoms of V. anatolica [13] and V. raddei [14].

Conclusions
In this work, quantitative proteomic and peptidomic characterization of venoms from four vipers inhabiting Russia was done; the compositions of the venoms from V. kaznakovi and V. orlovi, which showed the highest similarity among the four studied species, were analyzed for the first time.
More than 200 proteins and over 500 peptides were detected in total in all four venoms. They represented 14 snake venom protein families. In all venoms studied, over 70% of the total proteins were enzymes, the highest enzyme content (85.7%) being in the V. nikolskii venom. The main components of the venoms were PLA2s, which accounted for 65% of total protein content in the V. nikolskii venom. For the first time, bradykinin-potentiating and C-type natriuretic peptides were reported for vipers of the Pelias group. Nucleic acid degrading enzymes and phospholipase B were found in the venoms of Vipera species for the first time.
Due to the low toxicity of the steppe viper, or a limited habitat of the Caucasian and Orlov's vipers, these snakes do not pose an epidemiological threat to Russian population. However, the envenomation by Nikolsky's viper, the venom of which was shown in this study to contain a considerable amount of neurotoxic phospholipase A 2 , may represent certain danger. An antiserum "Antigadyuka" ("Antiviper") produced by Russian company "Allergen" is based on the venom of the common viper and may not be effective against the Nikolsky's viper bites due to strong differences in the composition of the venoms. The need to consider the differences in the composition of the venoms in the antivenom production is discussed in recent publications [28,35] and should be taken into account by antiserum manufacturers.

Materials and Methods
The venoms of V. kaznakovi, V. nikolskii, V. orlovi and V. renardi vipers were obtained as described earlier [12]. The venoms from several individual animals were pooled as described in [12]. Snakes were captured in their natural habitat: V. kaznakovi in Krasnodar Territory near Adler, V. nikolskii in Penza region near Zubrilovo village, V. orlovi in Krasnodar Territory at Mikhaylovskiy mountain pass and V. renardi in Krasnodar Territory near Beysugskiy firth.

In-Solution Trypsin Digestion of Venom Samples
Lyophilized venom sample (100 µg each) was dissolved in 10 µL of a buffer containing 100 mM ammonium bicarbonate (ABC), 5% sodium deoxycholate (SDC) and 5 mM dithiothreitol (DTT) and incubated for 40 min at 60˝C to reduce cysteine residues. Then, 5 µL of 50 mM iodoacetamide (IAA) water solution was added and the mixture was incubated 30 min at RT, in the dark. Residual IAA was neutralized by 5 µL of 50 mM DTT and sample was diluted with 50 µL 50 mM ABC and trypsin was added in a 1:100 (enzyme/protein) ratio to the final volume 100 µL and the protein concentratioñ 1 mg/mL. Samples were incubated overnight at 37˝C. Trypsin was deactivated by addition of 5 µL of 10% TFA. Tryptic peptides were desalted using reverse-phase solid extraction cartridges Discovery DSC-18 (100 mg) (Supelco, Bellefonte, PA, USA) according to the manufacturer protocol. Final peptide solution was dried in vacuum and stored at´80˝C prior to LC-MS/MS analysis.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript: