The structural characterization and bioactivity assessment of nonspecific lipid transfer protein 1 (nsLTP1) from caraway (Carum carvi) seeds

Caraway seeds were morphologically authenticated by Dr. Muneeba Khan, Taxonomist at the Center for Plant Conservation, Karachi University Herbarium, and Botanical Garden, University of Karachi, Pakistan. As per institutional policy, seeds are not accepted as a voucher specimen, and no deposition number is issued required for the whole plant.

Ground caraway seeds immersed in n-hexane for 24 h defatted and dried in a fume hood. Defatted caraway seeds were continually stirred in Phosphate-Buffered Saline (PBS), pH 7.4, for 24 h at 4 °C. Extracted samples were filtered through mesh filter fabric and centrifuged at 14,000 rpm for 30 min. Then, the supernatant was precipitated using 80% ammonium sulfate and stirred for 24 h at 4 °C. Precipitated proteins were recovered by centrifugation at 14,000 rpm for 30 min. The pelleted proteins were dialyzed in water at 4 °C and lyophilized.

Tris/tricine Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) with 10% resolving gel and 4% stacking gel was performed for the crude protein, as well as the collected chromatography fractions during each step of the purification process. The resulting bands were stained by Coomassie blue dye followed by water destain.

Using FPLC AKTA pure (GE Healthcare), the gel filtration column HiLoad 26/600 Superdex 200 pg separated proteins as the first stage of the two-dimensional separation. The dried protein mixture dissolved in PBS was centrifuged, and the supernatant was injected into FPLC. The running buffer was PBS, pH 7.4, at 2.6 ml/min, and the absorbance of the proteins was detected at 280 nm.

The RP-HPLC was used to purify the proteins using Aeris 3.6 µm WIDEPORE C4 250 × 4.6 mm (Phenomenex). The running solvents consist of water containing 0.1% TFA (solvent A) and acetonitrile containing 0.1% TFA (solvent B). Gradient programming of the solvent system was carried out at 0–60% solvent B for 55 min. The absorbance was monitored at 280 nm. The column Nucleodur C18 (MACHEREY–NAGEL) was used for peptide purification experiments, and the absorbance was adjusted to 214 nm.

The matrix 3,5-Dimethoxy-4-hydroxycinnamic acid (SPA) was prepared in 50% acetonitrile containing 0.1% Trifluoroacetic acid (TFA). 1 μl of SPA and 1 μl of nsLTP1 (dissolved in 0.1% TFA) were mixed, spotted on the MALDI plate, and placed in Autoflex MALDI-TOF (Bruker) to determine the protein's molecular weight (m/z).

The purified proteins were dissolved in reduction and alkylation buffer composed of Guanidine/HCl 6 M, tris base 0.2 M, di-sodium EDTA, 2 mM, and 5 μl of 2-mercaptoethanol (βME) in a ratio of 1:1. Under nitrogen blowing, βME was added (1:10) and incubated at 50 ℃ for 4 h. Then, 4-vinylpyridine was added to the solution (1:10) and incubated for 1.5 h at 37 °C. The reaction was quenched by βME containing 5% acetic acid (1:1).

The pyridylethylated protein was digested by Trypsin, TPCK-treated (Sigma-Aldrich) to a final enzyme to protein ratio of 1:20 (w/w) in 50 mM Tris/HCl, pH 8.3 for 4 h at RT. Then, the digestion was quenched with 1 M acetic acid to pH 4, and RP-HPLC separated the resulting peptides.

The purified protein (1 mg) was dissolved in 2 ml of 70% formic acid. Then, 4 mg of Cyanogen Bromide (CNBr) was added to the solution at 25 °C and incubated in the dark for 24 h. A vacuum concentrator dried the cleaved methionyl bond peptides to isolate them by RP-HPLC.

The N-terminal amino acid sequence of the intact protein and the purified peptides were determined by automatic Edman degradation using a PPSQ-51A/53A gas-phase protein sequencer (Shimadzu). The retention time of the PTH-amino acid residues was compared to a PTH-amino acid standard for identification. Approximately 20 pmol of the protein/peptide of interest was spotted on a PVDF membrane, and the first 30 amino acid sequence was performed.

As a query sequence, representative nsLTP1 sequences from different genera were obtained from NCBI blastp against caraway nsLTP1. The alignment of multiple sequences retrieved from NCBI was carried out using the Muscle program [17]. The sequence alignment data was analyzed by the Jalview tool. MEGA 11 software [18] was used to perform the phylogenetic analysis. The neighbor-joining method was used to construct the phylogenetic tree. The parameters were set as default according to the selected method, which uses Jones–Thornton–Taylor (JTT) model for pairwise distance matrix. Bootstrapping was performed to calculate the percentage of taxa clusters with a bootstrap value of 1000 (1000 replicates).

The homology modeling approach was used to predict the three-dimensional structure of caraway nsLTP1. The amino acid sequence was subjected to blastp [19] against the PDB database (Protein Data Bank) to look for the closest similarity of protein structures. 5TVI was the best template based on identity, similarity, and other statistical parameters. Using 5TVI as the reference structure, the model of the amino acid sequence was built by MODELLER v9.23 [20]. Modeller constructs the structure of a protein target by a comparative modeling approach, which uses the structure of a known protein. The software generated five default models. The best homology model was chosen based on the least negative DOPE score. The model was superimposed on the template structure to see the difference by RMSD calculation through visualizing software UCSF Chimera [21].

The modeled structure of caraway nsLTP1 protein was analyzed through SAVES v6.0 structure validation server, and the Ramachandran plot was assessed using PROCHECK [22]. The Ramachandran plot was used to examine the stereochemical quality by calculating phi and psi angles of amino acid residues. Additionally, the modeled protein was also evaluated through the ProSA web server [23], which indicates potential structure errors. It calculates the residues' z-score and energy plot, thus considering the model's overall quality.

Structures of fatty acids, i.e., linolenic, palmitic, and linoleic acid, were obtained from PubChem Database [24] and stearic acid from ChemSpider [25]. Both databases provide a vast collection of chemical molecules for biological assays. All the structures were converted into the required format of the molecular docking tool, prepared by adding hydrogen atoms, and saved in PDBQT format through the AutoDock tool (ADT). Likewise, the modeled protein was prepared in ADT by adding Kollman charges and polar hydrogen for utilization in the docking program. Further, the binding pocket of modeled structure was predicted by the DoGSite Scorer server [26]. This server generates a grid around protein, applies Gaussian filter difference, and identifies favorable locations by making spheres on the grid positions. The potential pocket is selected based on the maximum density threshold of each pocket cluster.

Molecular docking is a structure-based approach employed to explore the binding orientation attained by ligands and the interaction pattern between receptor and ligand. In this regard, AutoDock Vina [27] software was run for docking fatty acids against the modeled structure. The grid box was set around the residues of modeled protein defined by the DoGSite server. The x, y, and z dimensions of the grid box were adjusted by 38, 38, and 34, and the box was centered by 86.954, 35.12, and -10.554 x, y, and z coordinates, respectively. As a result, nine default conformers were obtained from the docking of each ligand. Discovery Studio Visualizer [28] was used for the docking analysis. The analysis was carried out considering the binding affinities and non-covalent interactions in a protein–ligand complex, including hydrogen bonding, hydrophobic, and van der Waal interactions.

Triple-negative breast cancer (MDA-MB-231) and human breast cancer (MCF-7) cell lines were purchased from ATCC (Manassas, USA), cultured, and maintained in DMEM medium supplemented with 10% (v/v) of FBS and 1% penicillin–streptomycin. Cells were incubated at 37 ºC humidified at a 5% CO₂ incubator.

The in vitro antiproliferation effect of nsLTP1 against MDA-MB-231 and MCF-7 cells was determined using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) assay. In a 96-well plate, 10,000 cells/well were seeded and incubated for 24 h at 37 ºC humidified 5% CO₂ incubator. Then, a final concentration range (0–95 μM) of caraway nsLTP1 was added, and cells were incubated for 48 h. After that, the media was discarded, and fresh media was added containing 0.5 mg/mL MTS. After 4 h, the absorbance was recorded at 490 nm by a microplate reader. The percentage of cell inhibition was calculated as follows:

Three independent triplicate experiments were used to graph the dose–response curve over a range of concentrations by nonlinear regression analysis using Prism 9.5.0 (GraphPad).

The total antioxidant capacity (TAC) assay kit (MAK334, Sigma-Aldrich) was used to detect the antioxidant activity of nsLTP1 according to the manufacturer's instructions. Trolox standards were prepared from Trolox stock (50 mM) by combining 5 μL of standard with 245 μL of ultrapure water. Then, the standards were diluted from (0–1000 μM). The assay was started by transferring 20 μL of standards and caraway nsLTP1 into a flat bottom 96-well plate in triplicate. Then, 100 μL of the reaction mixture (100 μL of reagent A and 8 μL of reagent B) was added to all assay wells, and then the plate was incubated for 10 min at room temperature. The absorbance was measured at 570 nm (A₅₇₀) through a microplate reader, and the blank A₅₇₀ value was subtracted from standard and sample values. The standard curve was generated by plotting Trolox concentrations against their correspondence A₅₇₀. Total antioxidant capacity is expressed as μM equivalent of Trolox, calculated using the following equation.where: (A₅₇₀) sample = the absorbance of the sample

(A₅₇₀) blank = the absorbance of the medium blank.n = sample dilution factor.

Caraway nsLTP1 (15 μM), dissolved in ultrapure water, was used for CD spectroscopy analysis. The following parameters were applied using a J-1500 spectropolarimeter (Jasco) equipped with the Peltier PTC-510 temperature controller (Jasco) under the following conditions: Nitrogen flow 20 SCFH, PM-539 detector, data interval 0.1 nm, data pitch 0.1 nm, CD scale 200 mdeg/0.1 dOD, FL scale 200 mdeg/0.1 dOD, bandwidth 1 nm, cell length 1 mm, start mode immediately, scanning mode continuous, scanning speed 50 nm/min, and shutter control auto. An average of three readings between 190 and 260 nm was collected for each sample. The background of ultrapure water was subtracted. For the temperature-dependent CD spectroscopy, the sample was scanned at 20, 40, 60, 80, and 95 °C after 20 min of heating at the specified temperature. After collecting the data at 95 °C, the sample was kept at 20 °C for 20 min. Then, the sample was scanned again at 20 °C. Under the same conditions and parameters, lysozyme was used as a control to compare the conformational changes in the secondary structure after the thermal treatment. For pH-dependent CD spectroscopy, samples in artificial gastric pH 1.6 and intestinal pH 6.8 fluids were scanned at 37 °C. Analysis was carried out by Spectra Manager software (Jasco) using principal component regression (PCR), with a basis set containing 26 proteins reference set under the following conditions: standardization of result 100, replaced negative value to zero, and rejection percentage 1%.

Prism 9.5.0 (GraphPad) was used to analyze the experimental data. One-way ANOVA was used for group comparison, followed by Tukey’s post-hoc test to determine the statistical significance of the antioxidant assay. Results with P values of < 0.05 were considered statistically significant and presented as *P < 0.05, **P < 0.01, ***P < 0.001.

The crude proteins were extracted from defatted caraway seeds in PBS pH 7.4 at 4 °C, and protein precipitation was accomplished using 60% ammonium sulfate successfully. Gel filtration chromatography partially separated the protein mixture (Fig. 1a). 10%Tris/Tricine SDS-PAGE gel displayed a protein band around 9–10 kDa in the pooled fractions 33–41 (Fig. 1b), and the full-length gel figure is present in Supplementary Figure S1. The pooled fraction was further purified by RP-HPLC that yielded a highly purified protein eluted at 30 min (Fig. 2a). MALDI-TOF mass spectrometry determined the protein mass at m/z 9652 Da (Fig. 2b).

The N-terminal Edman sequencing of intact protein identified the amino acid residues up to 20 residues (DIDCKTVDSALSPCIPYLLG). The cysteine residues were successfully modified using 4-vinylpyridine and recognized as PE Cystein in Edman sequencing. The protein was subjected to TPCK-treated trypsin digestion to determine the primary sequence completely. The produced peptides were purified by RP-HPLC and sequenced. Based on the homology with other nsLTPs, the peptide linkage was assigned. The absence of arginine and lysine from one of the peptides is implicit as the C-terminal peptide.

Moreover, employing CNBr cleavage reaction yielded two fragmented peptides that validated the presence of a single Met residue at position 38. When aligned with Trypsin peptides, CNBr peptide-2 revealed the presence of -Lys-Arg- at residues 45–46. Combining the intact protein sequence data, trypsin digestion, and CNBr cleavage, peptides elucidate the complete primary structure of caraway seeds nsLTP1 consisting of 91 amino acids (Fig. 2c). The nsLTP1 contains eight conserved cysteine residues that form and retain its typical tertiary folding. The protein sequence data reported in this paper will appear in the UniProt Knowledgebase under the accession number C0HM61.

Global alignment of multiple nsLTP1 sequences was performed, as shown in Fig. 3. Cysteine is a highly conserved residue in all nsLTP1 homologs. Additionally, Val7, Leu18, Pro24, Gly32, Val33, Lys53, Arg58, Ala67, Leu70, and Ser83 are also conserved in caraway nsLTP1 and all analyzed nsLTP1 sequences. The sequence of caraway nsLTP1 showed a maximum identity of 77.78% with the sequence of ajwain nsLTP1 (Trachyspermum ammi) and the least identity of 50.5% with multiple sequences including Solanum stenotomum, Solanum verrucosum, and Gossypiuum hirsutum.

An unrooted phylogenetic tree was analyzed by aligning different nsLTP protein sequences using the neighbor-joining method. Homologous sequences were excluded from the result of the blastp homology search for alignment purposes. The nodes of tree were assessed by 1000 repetitions. Figure 4 shows that the tree contains several families of nsLTP sequences. From the top, a clade with bootstrap value 70 is divided into branches as well as its sub-branches consisting of nsLTPs of different taxonomy as well as Solanaceae family members, i.e., Nicotiana and Solanum. The other side of a clade is joined with Manihot esculenta, Gossypium, and members of the Apiaceae family. In Apiaceae, C. carvi (caraway), T. ammi (ajwain), Anethum foeniculum (fennel), and Daucus carota (carrot) are clustered together, in which C. carvi shares the most common ancestor with T. ammi and D. carota. Besides, the rest of the tree contains many other branches having species of different genera.

The three-dimensional structure of caraway nsLTP1 sequence was constructed with the help of searching known structures available in protein databases. BLAST (Basic Local Alignment Search Tool) was used to input the protein query. We chose nsLTP1 of Solanum melongena (PDB ID: 5TVI) as the modeling template of caraway nsLTP1 with the highest identity of 47.78%, similarity 67%, lowest E-value 5e⁻²⁹ and query coverage of 98% among all other homologous proteins (Supplementary Table S1). MODELLER program aligned target and template, i.e., caraway nsLTP1 and 5TVI, and built different models. Among the models obtained from basic modeling, one having the least DOPE score of -8445.872 was chosen as the final structure (Fig. 5a). The modeled structure was superimposed upon 5TVI (template), which shows a low of RMSD 0.6 Å, suggesting that the suitable model is constructed under the range of ≤ 2 Å structural variation (Fig. 5b). Upon evaluating the predicted model by PROCHECK, it was found that 94.9% residues lie in the most favored region making the structure acceptable (Supplementary Figure S2). The -6.37 Z-score by the ProSA tool also indicated the model's accuracy within the scores of X-ray/NMR structures available in the PDB database (Supplementary Figure S3). The modeled caraway nsLTP1 structure can be accessed in PMDB (protein model database) [29] with ID PM0084405.

Figure 6 represents the docked complexes of linolenic, linoleic, stearic, and palmitic acid in the binding cavity of caraway nsLTP1. Linolenic acid with caraway protein showed a maximum binding score of -3.6 kcal/mol, whereas palmitic acid showed the least binding score of -3.0 kcal/mol. However, the other two fatty acid molecules, i.e., linoleic acid and stearic acid, exhibited -3.5 and-3.4 kcal/mol, respectively (Supplementary Figure S4). Further, the molecular interactions between caraway protein and ligands are tabulated in Supplementary table S2.

The antiproliferative activity of purified nsLTP1 from caraway seeds against MDA-MB-231 and MCF-7 cell lines was evaluated using the MTS assay. In a dose-dependent manner, the results showed inhibitory activity of nsLTP1 against MDA-MB231 and MCF-7 cells (Fig. 7). The IC₅₀ was calculated using nonlinear regression analysis; For MDA-MB-231 cells, the value was 52.93 μM, while for MCF-7 cells, it was 44.76 μM. At the lowest concentration (5 μM), caraway nsLTP1 exhibited more inhibitory effect in MCF-7 cells than in MDA-MB-231 cells. Besides, at the highest concentration (95 μM), proliferation was inhibited by 65% and 82% for MDA-MB-231 and MCF-7, respectively.

The antioxidant ability determination is inferred based on caraway nsLTP1 capability to reduce Cu2 + . The occurring Cu + primarily enhances the color of the solution by forming a colored complex in proportion to oxidation inhibition. Therefore, the sample displayed intense absorbance, confirming the protective effect of nsLTP1. The total antioxidant capacity (TAC) was dose-dependent, and the obtained value was 750.4 ± 6.9 μM Trolox equivalent at the highest concentration (41.4 μM). At the lowest concentration (10.4 μM), nsLTP1 showed a TAC value of 268.2 μM ± 0.29 μM Trolox equivalent (Fig. 8).

The stability of caraway nsLTP1 under different temperatures, artificial gastric fluid, and artificial intestinal fluid was investigated using CD spectroscopy. First, the spectra were captured for nsLTP1 and lysozyme (positive control) at temperatures between 20 to 95 °C. Then, nsLTP1 and lysozyme were cooled to 20 °C to collect the data again (Fig. 9a). Lysozyme is reported to be stable in thermal treatments [30]. At high temperatures, the spectra changes of nsLTP1 were minimal, and the negative maxima remained stable. In addition, nsLTP1 completely returned to its original secondary structure when cooled to 20 °C. In contrast, lysozyme minimum shifts from 208 to 203 nm at high temperatures. Overall, lysozyme refolds almost completely once the temperature is reduced (Fig. 9b). Therefore, we can suggest that the nsLTP1 is resistant to thermal denaturation up to 95 °C. In addition, The data illustrates nsLTP1 superior thermal stability compared with lysozyme. Based on PCR's estimated secondary structure contents, α-helix is the predominant structure in nsLTP1 with a value of 34.2%, which increased slightly to 45.6% as the temperature rose to 95 ºC. The turn and other secondary structures had minor changes in higher temperatures, while the β-sheet secondary structures dropped to 7.5% from 27.8% at 95 °C (Table 1). nsLTP1 secondary structure stability was assessed in artificial gastric and intestinal fluids without proteolytic enzymes. Both fluids had some effects on the secondary structure, but there were minimal changes (Supplementary Figure S5). The spectra illustrated an increase and decrease in Molar CD intensity. The PCR analysis indicated a slightly lower content of β-sheet (AIF 2.4% and AGF 18.4%), almost unchanged turn (AIF 15.3% and AGF 13.0%), higher α-helix (AIF 46.4% and AGF 37.3%) and almost unchanged other (AIF 35.9% and AGF 31.3%) in artificial gastric fluid compared to nsLTP1 dissolved in ultrapure water (Supplementary table S3). Like thermal treatment results, percentages of α-helix increase and β-sheet decrease in AGF and AIF environments. The data suggest that nsLTP1 has a high tolerance to keep its secondary structure in the gastrointestinal tract without proteolytic enzymes.