ATR-FTIR spectroscopy combined with chemometrics reveals molecular alterations and anticancer effects of Nigella sativa extract in human colon cancer cells

Abstract

Colorectal cancer remains the second leading cause of cancer-related mortality worldwide, with current therapeutic approaches often demonstrating limited efficacy. Nigella sativa L. (NS) seeds, historically valued for their medicinal properties, exhibit promising anticancer potential. This study investigates the molecular effects of NS methanolic extract on CaCo-2 human colon cancer cells, focusing on cellular composition, dynamics, and segregation patterns. Unsupervised chemometric analyses, including principal component and hierarchical cluster analyses, demonstrated a complete separation between control and NS-treated cells, indicating significant molecular divergence, further validated by supervised classification methods. Spectral analysis revealed reductions in unsaturated lipids, proteins, glucose, and DNA levels, along with a shortening of fatty acid acyl chain length. In contrast, saturated lipid and triglyceride content increased, accompanied by enhanced membrane fluidity and lipid disorder, indicating substantial alterations in cellular lipid dynamics and acyl chain flexibility. Furthermore, oxidative stress markers were elevated, as evidenced by increased protein carbonylation, while protein phosphorylation levels declined. NS treatment also induced protein conformational changes, notably an increase in aggregated β-sheet structures, suggesting protein denaturation. These biochemical modifications were strongly associated with NS-enhanced reactive oxygen species (ROS) levels. Overall, this study elucidates the molecular mechanisms underlying the anticancer effects of NS, supporting its potential as an adjunctive therapeutic strategy for colorectal cancer.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-34994-9.

Introduction

Colorectal cancer (CRC) is the third most common cancer and the second leading cause of malignant tumour-related mortalities worldwide. GLOBOCAN estimated that colorectal cancer accounted for ~ 1.9 million new cases and ~ 930,000 deaths in 2020¹.In line with WHO projections, these numbers are expected to rise sharply by 2040, reaching about 3.2 million new cases and 1.6 million deaths each year. representing an increase of 63% and 73%, respectively². The development of CRC is influenced by multiple factors, including age, family history, gender, geographic region, personal medical history, and lifestyle factors².

Current clinical approaches for CRC treatment include surgery, radiotherapy, chemotherapy, targeted therapy, and immunotherapy, all of which are cornerstone strategies critically endorsed by clinicians to improve patient survival and quality of life. Despite their efficacy, these methods have inherent limitations; for example, systemic treatments like chemotherapy are often associated with challenges such as adverse reactions and drug resistance, complicating therapeutic outcomes^3,4. Therefore, there is a growing need for novel therapeutic agents that provide greater efficacy and fewer side effects as complements to existing treatments to help reduce patient pain and improve survival rates.

Recent studies on drug discovery have increasingly focused on natural products due to their availability, cost-effectiveness, diverse biologically active components, and generally favourable safety profiles. Many natural compounds with chemotherapeutic potential for CRC treatment have been reported^5–7. Among these, Nigella sativa L. (Ranunculaceae family), commonly known as black cumin seed, has been used for thousands of years as a spice, food preservative, and treatment for various diseases^8–10. Over the past two decades, there have been many studies about the effect of N. sativa seeds on the different physiological systems, both in vitro and in vivo. These studies have demonstrated a wide range of biological activities of N. sativa seed or seed extracts¹¹. Most of these biological activities can be due to the major components of Nigella sativa L. seeds such as fixed oils, proteins, alkaloids, saponins, and essential oils¹². Among these components, thymoquinone (TQ) and p-cymene have been reported as the main compounds of N. sativa seed essential oil. Furthermore, TQ is one of the most active compounds and exhibits various biological activities such as anticonvulsant¹³, antioxidant¹⁴, anti-inflammatory¹⁵, antibacterial¹⁶, and antifungal properties¹⁷. In addition, these pharmacological activities, thymoquinone (TQ) has been extensively researched for its anticancer properties. For instance, the anti-neoplastic and pro-apoptotic activities of TQ, have been reported in the colon cancer cell line HCT116¹⁸. This study indicated that TQ’s pro-apoptotic activity was modulated by Bcl-2 protein and dependent on p53. The chemotherapeutic potential of TQ was also indicated in SW-626 colon cancer cells, showing similar efficacy to 5-fluorouracil¹⁹. Although these in vitro studies elucidate the mechanistic foundation of TQ’s anticancer properties, supplementary in vivo research has further validated its effectiveness in animal models. Gali-Muhtasib and colleagues evaluated the TQ’s therapeutic potential in two mouse colon cancer models. It was reported that TQ significantly reduced the number and size of ACF by 86% at week ten, with increased apoptosis²⁰. On the contrary, Rooney and Ryan have reported that TQ had no anticancer effect against HT-29 (colon adenocarcinoma) cells²¹.

Even though TQ has been thoroughly investigated for its anticancer properties, recent findings indicate that both whole-seed and seed extract forms of NS demonstrate considerable anticancer efficacy. Since these extracts contain diverse phytochemicals—fixed oils, saponins, alkaloids—that may act synergistically on multiple cellular targets²². Using the crude extract therefore enables detection of cumulative molecular alterations rather than isolating single-compound effects²³. Especially, crude extracts have several advantages over isolated compounds. First, they can synergistically interact with bioactive components. Second, they can improve pharmacokinetics of the main compounds. Third, they can buffer or lower toxic effects of the major actives. These mechanisms lead to greater efficiency than single bioactive components. Moreover, crude extracts have multi-target effects since many diseases involve multiple pathways. In addition, these extracts can be obtained in a cheap, easy and safety manner^23,24. Due to these advantages, this approach is particularly suitable for phenotypic assays and holistic evaluations of anticancer activity and is well established in anticancer screening studies employing plant-derived extracts as reported previously^11,25–34.

Several studies regarding the anticancer potential of NS seed confirmed that crude extracts and oils of NS seed have greater anticancer activity since they can capture multicomponent synergy and multi-target engagement^11,22,35. For instance, Agbaria et al. (2015) demonstrated that N. sativa seed oil/extracts prepared under varying thermal conditions inhibit the proliferation of mouse colon carcinoma (MC38) cells, exhibiting the strongest activity at moderate processing temperatures²². Moreover, the methanol extract of Saudi origin of NS has been demonstrated the most pronounced cytotoxic effect in HCT-116 cells³⁶. In another study, the hexane fraction of the methanol extract of N. sativa was found to be active against A-549 lung carcinoma cells and DLD-1 colon carcinoma, with IC₅₀ values of 31.0 and 63.0 mg/mL, respectively³³.These findings suggests that the full or crude extract, rather than solely isolated TQ, possesses biological relevance. Therefore, we used the crude extract of NS seed over TQ in the current study.

In addition to these in vitro findings, several in vivo studies have further confirmed the anticancer potential of NS seed extracts. For instance, Salim and Fukushima (2003) demonstrated that the volatile oil of N. sativa markedly diminished aberrant crypt foci and inhibited cell proliferation (as indicated by BrdU labeling) in a rat colon carcinogenesis model, underscoring the chemopreventive properties of whole-seed-derived oil³⁵. Similarly, Ait Mbarek et al. (2007) found that different crude seed extracts (ethanolic, aqueous, essential oil) of N. sativa exerted both in vitro cytotoxicity and in vivo tumour size reduction in murine models¹¹. Although, many studies confirmed the anticancer effects of NS seed extract, Asfour et al. indicated that the ethanolic extract of NS seeds lacked chemopreventive efficacy in a rat model during the post-initiation phase of colon cancer³⁴. This discrepancy was also reported in TQ for HT-29 (colon carcinoma) cells²¹. These contradictions may be arisen from the differences in extraction methods, experimental context—e.g., chemoprevention versus treatment, solvent polarity, exposure window, and in vivo versus in vitro models.

While the anticancer activities of NS seed extracts and thymoquinone have been extensively reported^20,22, most studies have focused primarily on cell viability measurements, which provide limited insight into the global biochemical alterations underlying the cellular response. As a result, the holistic biomolecular changes and coordinated molecular remodeling induced by NS extract across lipids, proteins, and nucleic acids, remain poorly understood, especially in colon cancers. This implies a substantial research gap in understanding the impact of natural crude extracts on cellular metabolism and signaling pathways in colon cancer.

Therefore, rather than re-evaluating cytotoxicity alone and the effects of isolated individual compounds, the present study aimed to characterize the broader molecular response of CaCo-2 human colon cancer cells to crude methanolic NS seed extract using a holistic, chemometric and label-free approach. To achieve this, we combined XTT cytotoxicity assays with ATR-FTIR spectroscopy and multivariate chemometric modelling (PCA, HCA, LDA, SIMCA, SVM). To the best of our knowledge, this is the first integrative study to extract system-level spectral biomarkers induced by NS treatment. By mapping these spectral signatures, our novel study provides mechanistic insight into how NS extract affects membrane organization, protein structure, and nucleic acid content, thereby offering a comprehensive molecular perspective that extends beyond viability measurements and contributes new and holistic understanding to the anticancer potential of Nigella sativa.

Results

NS seed methanolic extract reduced cellular viability

The cytotoxic effect of NS seed extract, ranging from 0 µl/mL to 20 µl/mL on CaCo-2 was determined using XTT assay. A concentration-dependent reduction was acquired in cell viability with 24 h treatment with NS seed extract. Figure 1 shows that the NS seed extract reduces cell viability in a dose-dependent manner. In the figure, the first bar represents the control group, where cell viability is set to 100%. This serves as the baseline for comparing the effects of treatment on cell viability and cell viability at increasing concentrations of the extract. For example,10 µl/mL concentration of NS seed extract did not affect the cell viability since % 95 cell viability was obtained in 10 µl/mL treatment compared to the control group. 70% of the cells were non-viable at the highest concentration tested (20 µl/mL, ****p < 0.0001). The IC₅₀ (half-maximal inhibitory concentration) value of the NS seed extract was found to be 15 µl/mL for CaCo-2 cells.

Fig. 1.

Cytotoxicity assessments by XTT assay in CaCo-2 cells following the exposure of different concentrations of methanolic extract of Nigella sativa for 24 h. Values are mean ± SD of three independent experiments. (***p < 0.001, ****p < 0.0001).

Segregation of NS-treated and untreated colon cancer cells

Figure 2 shows the Control, NS-treated and their difference spectra of CaCo-2 cells in different spectral regions, and the assignments of the bands are presented in Table 1. As depicted in this figure, NS treatment induced variations in the spectral bands, particularly in two key regions: C-H stretching (3035–2800 cm⁻¹) and the fingerprint region (1800–900 cm⁻¹). Basen on these variations, we performed Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) on their spectra to determine the segregation patterns between control and NS treated colon cancer groups. Figure 3 shows unsupervised chemometric analysis of IR spectra using PCA and HCA, for NS-treated and control colon cancer cells. As seen from the Fig. 3a, the control and NS-treated samples form distinct clusters, highlighting the clear separation between the two groups along PC1 (93% of the total variance) and PC2 (5% of the variance). These variance values were also demonstrated in Fig. 3b. Figure 3e further supports this segregation through HCA dendrogram with 100% accuracy and a high relative distance. The high relative distance between the clusters suggests strong differentiation between the two groups, further confirming the effects of NS treatment on cellular molecular composition. In both analyses, the consistent grouping patterns between replicates confirmed that the clustering results originated from treatment-induced biochemical alterations instead of random fluctuations. Furthermore, PCA and HCA successful clusterings proved the reproducibility and robustness of the spectral data.

Fig. 2.

Unsupervised chemometric analysis of IR spectra of NS-treated and control colon cancer cells. PCA score (a), PCA explained variance (b), PCA loading PC1 (c), PC2 (d), and HCA dendrogram plots (e).

Table 1.

IR band assignment of the spectral bands in the colon cancer cells.

Band No	Band location	Band assignment	References
1	3002	Olefinic C = C-H stretching: unsaturated lipids	38
2	2958	CH3 antisymmetric stretching: lipids, protein side chains	39
3	2920	CH2 antisymmetric stretching: lipids	39
4	2870	CH3 symmetric stretching: mainly proteins	39
5	2850	CH2 symmetric stretching: mainly lipids	39
6	1739	C= O stretching vibrations of lipids (triglycerides and cholesterol esters)	39
7	1654	Amide I: C= O stretch: proteins α-helix	39
8	1624	Intermolecular β-sheets	40
9	1534	Amide II: (protein N–H bend, C–N stretch)	39
10	1468	CH2 bending vibrations: lipids and proteins	39
11	1435	δ (CH2) polysaccharides	41
12	1404	COO symmetric stretching: fatty acids	41
13	1347	CH2 wagging	41
14	1314	Amide III of proteins	41
15	1240	(νasPO2−) asymmetric phosphodiester vibrations of nucleic acids	42
16	1150	Glycogen	43
17	1080	(νasPO2−) asymmetric phosphodiester vibrations of nucleic acids	39
18	1015	ν(C–O) deoxyribose: DNA	42
19	951	Z-DNA	44
20	934	Z-DNA	42
21	903	Phosphodiester stretching: DNA	45

Fig. 3.

(a) LDA discrimination plot for the training sets of the NS-treated and control colon cancer cells, (b) Distance between PCA models of the NS-treated and control colon cancer cells, (c) Cooman’s plot of the NS-treated (red) and control (blue) colon cancer cells with test samples (green) and (d) SVM classification plot for the training sets for IR spectra of the NS-treated and control colon cancer cells.

To validate these findings, linear discriminant analysis (LDA) was performed on the training data set, as shown in Fig. 4a. The NS samples were successfully differentiated from the control samples with 100% discrimination accuracy. The NS-treated samples are clearly differentiated from the control samples. This complete segregation is further supported by the high discrimination accuracy of the model. Figure 4b displays the model distance graph obtained from Soft Independent Modelling of Class Analogies (SIMCA) analysis. A high distance value, approximately 50, between the control and NS-treated groups further validates the separation. This graph highlights the significant molecular differences between the groups, confirming the distinct effects of NS treatment on the cells.

Fig. 4.

Control, NS-treated, and their difference spectra of colon cancer cells in the (a) 3035–2800 cm⁻¹ and (b) 1800–800 cm⁻¹ spectral regions.

Similar results were acquired in the SIMCA and Support Vector Machine (SVM) plots (Fig. 4c and d). The high classification accuracy in SVM further confirms the robustness of the molecular segregation between the treated and untreated samples. All supervised analyses were performed on the test data sets to test the classification efficiency. The prediction tables of LDA and SVM for training and test data sets are given in Table S1 and Table S2, respectively. All these data indicated a classification of the test samples with 100% accuracy. The classification analysis of test samples was repeated five times with different test data sets, consistently achieving 100% discrimination accuracy. The obtained high classification accuracy in training and test sets at more than 90% from models LDA, SIMCA, SVM demonstrated excellent predictive power along with reliable and reproducible spectral biomarkers.

NS seed extract- induced compositional and structural changes in colon cancer cell molecules

The loading spectra were evaluated to elucidate NS-induced changes in the molecular composition of colon cancer cells. PC1 and PC2 loading spectra were similar to those of the control and NS-treated cells (Fig. 3c and d). In loading plots, the Y-axis denotes the eigenvalue of each spectral band. Higher eigenvalues of a spectral band indicate a higher variation and stronger contribution of the discrimination for this band³⁷. The visual analysis of PC1 and PC2 loading plots revealed that higher eigenvalues were found in the 1800–750 cm⁻¹ spectral region and profound differences were observed between control and NS-treated colon cancer cells regarding protein, lipid, triglyceride, carbohydrate, and nucleic acid contents.

To support the loading plot findings, the spectra of the NS-treated and control colon cancer cells and their difference spectrum were quantitatively analysed in detailed (Fig. 2). As depicted in this figure, there are quantitative differences in the biomolecular contents between control and NS treated colon cancer cells, particularly in two key regions: C-H stretching (3035–2800 cm⁻¹) and the fingerprint region (1800–900 cm⁻¹). Band area, band area ratio, band intensity, band location, and bandwidth values of the spectral bands were measured to quantify these changes. The performed quantitative spectral analysis and their indications are given in Table 2. Furthermore, the detailed numerical results obtained from these analyses are demonstrated in Table S3. As seen from this table, there are significant variations in the analyzed parameters due to NS treatment, which are also demonstrated in Figs. 5 and 6.

Table 2.

Quantitative spectral analysis and their indications.

Band area/ area ratio	Indications
A3002	Unsaturated lipid
A2850 /A2920 + 2850	Saturated lipid concentration
A1739 /A2920 + 2850	Triglycerides-cholesterol ester concentration
A2920 /A2958	Acyl chain length of fatty acids
A1654 + A1624 / A1654 + A1624 + A1534	Protein concentration
(A2950 + A2920) / (A1654 + A1624 + A1534)	Total lipid/Total protein
A1654 + A1624 /A1534	Protein structural and conformational changes
A1240/A2958	Protein phosphorylation
A1739/A1534	Protein carbonylation
A1150/(A1654 + A1624 + A1534)	Glucose/Protein
Band intensity
I1624	Aggregated β-sheets concentration
I1654	α-helix concentration
I1692	Antiparallel β-sheets concentration
A1015/(A1240 + A1080)	DNA concentration
Wavenumber (W)	Indications
W2920	Membrane lipid order
Bandwidth (BW)	Indications
BW2850	Membrane fluidity

Fig. 5.

The quantitative changes in lipids in the control and NS-treated (a) unsaturated lipid, (b) saturated lipid, (c) triglyceride-cholesterol ester, (d) acyl chain length of fatty acids, (e) membrane lipid order, (f) membrane lipid fluidity (*p < 0.05, **p < 0.01, ****p < 0.0001).

Fig. 6.

The quantitative changes in lipid and protein in the control and NS-treated colon cancer cells. (a) protein concentration, (b) lipid/protein, (c) protein conformation, (d) aggregated β-sheets, (e) antiparallel β-sheets, (f) α-helix, (g) protein carbonylation, (h) protein phosphorylation, (j) glucose/protein and (k) DNA concentration.*p < 0.05, ** p < 0.01,***p < 0.001,**** p < 0.0001.

Figure 5 shows the quantitative spectral analyses for the lipid-associated spectral bands. The band at 3002 cm⁻¹, the olefinic band, originates from the C-H stretching of HC = CH groups and is generally used to acquire unsaturated lipid content⁴⁶. NS treatment resulted in a significant decrease (****p < 0.0001) in the unsaturated lipid content (Fig. 5a). In addition to unsaturated lipids, the amount of saturated lipids and triglyceride-cholesterol esters were determined by analyzing the band area ratios of the corresponding spectral bands^47,48. NS treatment resulted in a significant increase (**p < 0.01) in the band area ratio for the saturated lipids and in the ratio for triglyceride-associated spectral bands (****p < 0.0001) (Figs. 5b and c).

Acyl chain length and membrane lipid order e.g., acyl chain flexibility are structural parameters, and membrane fluidity is a dynamic parameter. These parameters were calculated in control and NS-treated colon cancer cells to detect ROS-induced alterations in lipids. A 25% decline was observed in the acyl chain length with NS treatment (****p < 0.0001) (Fig. 5d). Changes in the wavenumber and bandwidth values of CH₂ stretching bands were calculated following black cumin seed extract treatment. Significant increases in these parameters were found in this group (Fig. 5e and f). A significant increase (****p < 0.0001) in the wavenumber value of this band indicates a decrease in membrane lipid order. In addition, a significant increase in the bandwidth (*p < 0.05) implies higher membrane fluidity^38,49–51.

In addition to lipids, the protein concentration was calculated by using the band area ratio of amide I/amide I + amide II⁴⁷. A significant decrease (**p < 0.01) in the protein concentration was observed with NS treatment, implying a protein breakdown or decreased protein synthesis (Fig. 6a). On the contrary, an increased lipid/protein ratio (***p < 0.001) was obtained in the NS-applied CaCo-2 cells (Fig. 6b).

Protein conformational changes were also detected from a significant decline (*p < 0.05) in the band area ratio of the amide I/amide II (A₁₆₅₄ + A₁₆₂₄)/A₁₅₃₄) (Fig. 6c). This ratio is generally used to determine conformational alterations of the proteins since these bands originate from the different functional groups of the protein^38,52. To elucidate NS-induced protein secondary structural changes, i.e., the alterations in alpha helix and beta sheets structures, the signal intensity values were calculated from the vector normalized second derivative amide I spectral band (Figure S1). The results reveal significant protein structural changes in colon cancer cells following NS treatment. Increased intensities at 1624 cm⁻¹, 1654 cm⁻¹, and 1690 cm⁻¹ correspond to aggregated β-sheets, α-helices, and antiparallel β-sheets, respectively (Fig. 6d and e, and 6f). It is well known that when proteins are denatured, the intensity of the aggregation band increases (Severcan and Haris 2003). Therefore, the results indicate NS-induced protein denaturation. To confirm protein denaturation, protein carbonylation was also measured. In IR spectroscopy, this carbonylation can be detected from the alteration of the band area ratio A₁₇₃₉/A₁₅₃₄⁵³. Protein carbonylation was detected following NS treatment, and a higher protein carbonyl level (****p < 0.0001) was measured in NS-treated cells (Fig. 6g). It was also observed that NS treatment caused a significant reduction (***p < 0.001) in protein phosphorylation (Fig. 6h). Glucose and DNA content were also evaluated, and their respective concentrations were found to decrease significantly following treatment with the black cumin seed methanolic extract. Specifically, the glucose/protein ratio, as shown in Fig. 6j, revealed a substantial reduction (****p < 0.0001). This indicates that while both glucose and protein concentrations decreased, the decrease in glucose content was more pronounced, resulting in the observed reduction in the ratio. Similarly, a dramatic decrease in DNA concentration (Fig. 6k) was also observed (****p < 0.0001), further supporting the extract’s potential impact on cellular components.

Discussion

We conducted this study to investigate, for the first time, the alterations in molecular composition, structure, and dynamics induced by the methanolic extract of NS seed L. (black cumin) seeds in colon cancer cells. These seeds, historically valued for their medicinal properties, exhibit promising anticancer potential. To determine the cytotoxic concentration, CaCo-2 cells were treated with NS seed extract at concentrations ranging from 0 to 20 µL/mL for 24 h, and the IC₅₀ (half-maximal /inhibitory concentration) value was found as 15 µl/mL. Consistent with our findings, several studies have demonstrated the anticancer/cytotoxic effects of NS seed extracts in various cancer cell lines, including prostate and cervical cancers. Hasan et al. (2013) reported that the methanolic extract of NS seed inhibited human cervical cancer cell proliferation at a concentration of 125 µL/mL after 24 h of treatment²⁹. The anti-proliferative activity of NS seed methanolic extract was also reported on human prostate cancer cell line PC-3⁵⁴. The anti-neoplastic and pro-apoptotic activities of TQ extracted from black cumin seed have been reported against the colon cancer cell line HCT116¹⁸. Moreover, TQ has been shown to increase tumor cell cytotoxicity and trigger cell cycle inhibition at various stages, including G2/M, G1/S, S, and G0/G1 check points while also stimulating apoptosis and necroptosis in cancer cells⁵⁵. However, conflicting results exist, as Rooney and Ryan (2005) reported that TQ did not exhibit anticancer activity against HT-29 (colon adenocarcinoma) cells²¹. It is important to note that most studies have investigated only the effects of TQ, while N. sativa seed extract comprises multiple bioactive compounds with diverse biological activities. These constituents may act synergistically, enhancing the overall pharmacological effects⁵⁶. Extensive work has documented the anticancer potential of Nigella sativa seed extracts and thymoquinone^20,22. However, most studies emphasize either cytotoxicity outcomes, single-biomarker assays, or isolated constituents rather than integrated molecular profiling. However, this study was designed to mainly examine the methanolic crude extract at cellular level, focusing holistic biological responses rather than the effects of isolated individual compounds. This approach was also used previously for other plant-derived extracts in anticancer screening studies^11,25–34. Different than the other studies on N. sativa, our approach links XTT-based viability measurements with ATR-FTIR chemometric analyses (PCA, HCA, LDA, SIMCA, SVM) to capture system-level molecular alterations across lipids, proteins, and nucleic acids after treatment. This combined methodology extends conventional cytotoxicity testing by delivering quantitative spectral biomarkers that illuminate treatment-induced biochemical remodelling beyond simple viability loss.

ATR-FTIR spectroscopy is a well-established technique for assessing holistic biochemical alterations within cells in response to drug candidates or plant-derived compounds in a high- throughput manner^57,58. This analytical approach is particularly valuable for investigating drug-cell interactions by detecting chemical variations in treated cells over specific exposure periods⁵⁹. In this study, ATR-FTIR spectroscopy was utilized to evaluate biomolecular alterations in human colon cancer cells, focusing on changes in molecular content, structure and dynamics following NS treatment. To achieve this, CaCo-2 cells were first exposed to the IC₅₀ concentration of the methanolic NS seed extract for 24 h. Subsequently, IR spectra of both control and NS-treated groups were acquired using ATR-FTIR spectroscopy to identify molecular-level changes induced by the treatment. The comparative spectral analysis provided insights into alterations in cellular biochemical composition, molecular architecture and dynamics as a consequence of treatment. In the current study, 24-hour exposure at IC₅₀ concentration level was chosen since this IC₅₀ concentration level reduces cell viability by 50% and leads to detectable biochemical changes within 24 h while retaining viable cells. This allows researchers to observe specific treatment effects beyond nonspecific late-stage responses^60–62. Under these established experimental conditions, FTIR spectroscopy captures robust, treatment-associated biochemical alterations. For instance, Derenne et al., 2013 reported that IR spectral changes were successfully detected when T98G glioma cells were treated at their IC₅₀ with several polyphenols for 2, 6, and 24 h⁶⁰. Moreover, the robust biochemical alterations in SW620 colon cancer cells can be easily captured under IC₅₀ of 5-FU exposure via FTIR spectroscopy, validating IC₅₀ exposure⁶¹.In addition, 24 h treatment concept was confirmed by Altharawi et al., 2019, revealing better spectral discrimination at longer exposure time (24 h)⁶². These findings prove that our experimental design aligns with best practices in spectroscopic profiling.

An IR spectrum provides extensive biochemical information about a single cell, consisting of approximately a thousand spectral data points. Given the complexity of these datasets, multivariate data analyses is essential for efficient and accurate interpretation within a short timeframe³⁷. Among these methods, unsupervised pattern recognition approaches such as PCA and HCA are widely used to assess relationships and discrimination between groups based on their spectral variation³⁷. Therefore, PCA and HCA were performed to evaluate the segregation patterns between control and NS-treated colon cancer cells. Both analyses confirmed a clear separation between the two groups, indicating significant molecular divergence induced by NS treatment. To further validate these distinctions, supervised classification techniques, including LDA, SIMCA and SVM were applied to the control and NS-treated spectral data. Notably, all classification models achieved 100% accuracy, demonstrating the robustness of the molecular differences between the treated and untreated cells.

To characterize the biochemical alterations induced by NS treatment, PCA loading spectra, IR spectra of the NS-treated and control colon cancer cells, and their difference spectra were qualitatively examined. Significant changes were observed in two main spectral regions: The C-H stretching (3035–2800 cm⁻¹) and fingerprint (1800–900 cm⁻¹) region. The C-H stretching region primarily corresponds to unsaturated and saturated lipid bands, while the fingerprint region contains characteristic spectral bands associated with proteins, lipids, triglycerides, carbohydrates, and nucleic acids. According to Beer Lambert’s law, the band area/intensity of the spectral peak is proportional to the concentration of the corresponding molecules. Therefore, the band area and band area ratio of these bands were calculated to define these macromolecular alterations. A significant reduction in unsaturated lipid content was observed following NS seed extract treatment which may be attributed to the degradation of unsaturated lipids. This degradation is likely driven by lipid peroxidation, a process induced by elevated ROS, as excessive ROS levels are known to trigger lipid peroxidation⁶³. Moreover, the pro-oxidant property of TQ has been reported to generate ROS in a concentration-dependent manner⁶⁴, further supporting NS-induced lipid peroxidation. In addition to unsaturated lipids, an increased band area ratio was observed for saturated lipids and triglyceride-cholesterol esters. The significant increase in the ratio for triglyceride-associated spectral bands may be associated with increased lipid carbonylation due to oxidative damage, as this band originates from the C = O stretching vibration of triglycerides and serves as a key indicator of carbonylation levels³⁸.

Elevated ROS levels can lead to structural and functional alterations in lipids. To assess these changes, acyl chain length and membrane lipid order were analyzed. NS treatment resulted in a reduction in acyl chain length, which is linked to the increased breakdown of unsaturated lipids, as this process triggers membrane lipid peroxidation⁶⁵. Lipid peroxidation is well known to alter membrane lipid ratios, thereby affecting membrane fluidity and order⁶⁶. To quantify these alterations, membrane order and fluidity were assessed by analyzing the wavenumber and bandwidth values of CH₂ stretching bands, respectively. A significant decrease in membrane lipid order, coupled with an increase in membrane fluidity in colon cancer, was obtained in NS-treated cells. These spectral changes are consistent with the effects of ROS-driven lipid peroxidation, which destabilizes membranes, increases permeability, and promotes mitochondrial outer membrane permeabilization (MOMP) during early apoptotic signaling^67,68. Increased lipid disorder therefore represents a biophysical hallmark of apoptosis induction under oxidative stress conditions.

Our findings on NS-induced lipid disorder and enhanced membrane fluidity are align with previous studies conducted on colon cells^69,70. Lipid metabolic rearrangement, or deregulated lipid metabolism, is a well-known phenotypic hallmark of cancer cells. Alterations of lipidome profiles and lipid metabolic pathways have also been reported in colon cancer⁷⁰. In these pathways, lipid synthesis, desaturation, elongation, and mitochondrial β-oxidation of fatty acids are highly active in CRC cells. These metabolic alterations lead to membrane structural and functional changes, influencing key cellular processes such as apoptosis, proliferation, differentiation, and cancer cell growth. Therefore, the NS-induced reduction in cellular proliferation may be associated with altered saturated and unsaturated lipid metabolism, and subsequent changes in membrane order and dynamics⁶⁹. Supporting this finding, Luchi et al. (2019) reported that oxidized polyunsaturated fatty acids inhibited cell proliferation in THP-1 (a human monocytic leukemia cell line) and DLD-1 (a human colorectal cancer cell line) cells⁷¹.

Alongside lipids, ROS oxidize cellular proteins, leading to alterations in protein concentration, conformation, and function⁷². To evaluate these parameters, protein concentration was quantified by following NS treatment. A significant decrease in protein concentration was observed, implying protein degradation or decreased protein synthesis. In addition, NS treatment led to a protein conformational change, especially increasing the amount of aggregated β-sheets, α-helices, and antiparallel β-sheets. These structural modifications imply that NS treatment triggers oxidative stress, leading to protein misfolding and aggregation. The simultaneous increase of both α-helical and β-sheet structures highlights a complex cellular response: while α-helices may be preserved in apoptotic or regulatory proteins, β-sheet-rich aggregates likely contribute to cytotoxicity and apoptosis. Given that NS components, particularly TQ, are known to modulate redox balance and promote apoptosis in colorectal cancer cells, these findings further support its anticancer potential. Furthermore, the increased aggregated β-sheet content serves as a key indicator of protein denaturation, consistent with previous findings⁷³. These spectral shifts collectively suggest that NS treatment disrupts protein homeostasis, overwhelms cellular protein quality control mechanisms, and facilitates apoptosis—an essential process in cancer therapy. Protein denaturation is primarily driven by oxidative stress, which can be assessed through protein carbonylation, the most detrimental and irreversible oxidative alteration of proteins⁷⁴. A high protein carbonyl level in NS-treated cells confirms that NS treatment generates excessive ROS, reinforcing its role in oxidative stress-mediated cytotoxicity.

Protein phosphorylation plays a critical role in regulating key cellular processes, including cell division, protein degradation, signal transduction, gene expression regulation, and protein interactions⁷⁵. Consequently, any alteration in this mechanism leads to cancer pathogenesis by promoting cancer cell proliferation, invasion, and metastasis, while simultaneously suppressing apoptosis⁷⁶. Regarding phosphorylation, TQ treatment has been shown to significantly reduced phosphorylated p65 levels in the nucleus of colon cancer cells⁷⁷. Consistent with these findings, our study demonstrated a significant reduction in the protein phosphorylation following NS treatment. Furthermore, Kundu et al. (2014) reported that thymoquinone induces apoptosis in human colon cancer HCT116 cells through inactivation of STAT3 by blocking JAK2-and Src-mediated phosphorylation of EGF receptor tyrosine kinase⁷⁸. The observed decrease in cell viability following NS treatment may be linked to reduced protein phosphorylation, which could inhibit critical signaling pathways necessary for cancer cell survival and proliferation. To further investigate the cytotoxic effect of NS treatment on colon cancer cells, glucose and DNA content were evaluated. NS-treatment resulted in a significant decrease in glucose content, which may be attributed to the inhibition of glucose metabolism. Supporting this, Karim et al. (2022) demonstrated that TQ suppresses glycolytic metabolism in colon cancer cells⁷⁹. Similarly, the inhibitory effects of TQ on the glycolytic pathway have been reported in colon cancer cells⁸⁰. Wu et al. (2018) further demonstrated that elevated glucose levels promote CRC cell proliferation, migration, and invasion while simultaneously inhibiting apoptosis⁸¹. Therefore, the NS-induced reduction in cell proliferation may be linked to glucose depletion and impaired metabolic activity. In addition to glucose depletion, NS treatment led to a dramatic decrease in DNA concentration, supporting its impact on cellular components. This decline may be attributed to either increased DNA degradation or the inhibition of DNA synthesis. The observed DNA damage may be driven by NS-enhanced ROS levels, as radical species are known to induce oxidative damage DNA⁷². ROS can interact with genomic DNA, causing substantial damage through DNA strand breakage, thereby contributing to genomic instability and trigger apoptotic signalling pathways⁸². Consistent with this, TQ-induced DNA damage has been previously reported in colorectal cancer stem/progenitor cells⁸³, further corroborating our findings. In line with our results, earlier reports have reported that NS seed extracts and thymoquinone can induce oxidative stress, lipid peroxidation, and programmed cell death in colorectal cancer models^20,22. Likewise, the observed shift toward lipid-ester/triglyceride enrichment aligns with studies demonstrating that seed processing conditions influence extract composition and cytotoxic potency through multi-component synergistic activity²².

In summary, our findings demonstrate that this work provides the first multi-parameter, system-level molecular characterization of Nigella sativa seed extract activity in colon cancer cells using ATR-FTIR spectroscopy combined with advanced chemometric modelling. In contrast to previous NS or thymoquinone studies, which are typically limited to cytotoxicity assays or single biochemical endpoints, our analysis simultaneously quantified changes across a broad set of biomolecular parameters, each linked to a specific functional consequence. In addition, the membrane lipid order/fluidity parameters and protein secondary structural changes represent particularly novel aspects of this study, as neither has been previously characterized in the context of Nigella sativa treatment in colon cancer cells. These biophysical changes are functionally important because ROS-driven lipid peroxidation disrupts membrane integrity and promotes mitochondrial outer membrane permeabilization which is a hallmark of apoptosis. Likewise, alterations in protein secondary structure, including increased aggregated β-sheet content, indicate protein misfolding and denaturation, all of which are signature events in oxidative stress-mediated apoptotic signaling. Demonstrating these shifts for the first time under NS extract treatment provides new mechanistic evidence linking NS-induced oxidative stress to membrane destabilization and protein structural collapse. While NS-induced protein denaturation contributes to apoptosis in cancer cells, these findings also highlight the need for delivery strategies that selectively target malignant cells, ensuring that such potent oxidative and structural damage is restricted to tumor tissue while sparing healthy cells. The broader system-level signatures obtained here, including triglyceride enrichment, protein misfolding and aggregation, elevated protein carbonylation, suppressed phosphorylation, reduced glucose content, and diminished DNA integrity, collectively demonstrate a coordinated ROS-mediated biochemical remodelling of cancer cells.

Altogether, our study establishes a novel analytical framework and provides the most comprehensive biochemical map to date of how Nigella sativa seed extract changes colon cancer cell physiology, strengthening ATR-FTIR profiling as a powerful tool for elucidating the multi-target actions of complex natural products. Our findings also highlight the value of ATR-FTIR spectroscopy as a label-free, rapid, and comprehensive tool for monitoring treatment-induced biochemical remodelling, while also indicating that targeted follow-up approaches (e.g., proteomics, metabolomics, lipidomics) will be useful for identifying specific molecular drivers within the crude extract.

Conclusion

Our study delivers a comprehensive characterization of the biochemical responses of colon cancer cells to Nigella sativa seed extract by coupling ATR-FTIR spectral profiling with machine learning–based chemometric tools. The research shows that NS seeds methanolic extract decreases cell survival rates and triggers major biochemical changes in CaCo-2 colon cancer cells. The XTT–ATR-FTIR holistic method revealed specific spectral indicators which showed how the extract damaged cellular homeostasis through lipid peroxidation and protein breakdown and metabolic disruption. The study achieves stable and reproducible molecular profiling results through its use of a standardized IC₅₀ concentration and 24-hour exposure period which enables future research including time-dependent studies and sub-toxic concentration tests and normal cell model evaluations. By simultaneously evaluating lipid composition, membrane order and fluidity, protein secondary structure, protein carbonylation, phosphorylation status, glucose metabolism, and DNA integrity, we demonstrate a coordinated ROS-driven biochemical remodelling that underlies the extract’s anticancer action. Importantly, membrane biophysical parameters and protein structural transitions, which are evaluated here for the first time, offer new structural-level evidence for NS-induced membrane destabilization, protein misfolding, and apoptosis.

Furthermore, machine learning-supported chemometric classifiers (LDA, SIMCA, SVM) achieved 100% discrimination accuracy, underscoring the strength and diagnostic potential of the spectral fingerprints obtained. Future research should validate these results through lipidomic and proteomic analysis to create nanotechnology-based delivery systems that will improve Nigella sativa bioactive selectivity and therapeutic outcomes while protecting normal cells from adverse effects. In this context, incorporating chromatographic based fractionation and isolation of single-compound together with machine learning–assisted ATR-FTIR profiling will help clarify which specific bioactive components drive the observed spectral patterns and support their translation into more targeted therapeutic application.

Materials and methods

Preparation of NS seed extract

Nigella sativa seeds were obtained from a local market and ground into a fine powder. A total of 200 g of powdered seeds were extracted using 500 mL of methanol (99.8% purity, Merck) via maceration. The extraction was performed three times for 24 h each, with an additional 1.5-hour sonication step to improve extraction efficiency. After maceration, the extracts were filtered through filter paper, and the methanol was evaporated under reduced pressure at 45 °C using a rotary evaporator (Buchi, Labortechnik, Flawil, Switzerland) to obtain the crude extract. The extract was further dried to ensure the removal of residual solvent. The final yield was 10 g of crude methanolic extract, consistent with the literature⁸⁴. The experimental setup used to obtain NS seed methanolic extract is shown in Figure S2A. In this study, rather than isolation of the single compound via chromatographic methods, the methanolic crude extract was used to preserve natural phytochemical complexity and to enable evaluation of its overall biological effects at the cellular level.

Methanol was selected as extraction solvent because its strong polarity enables the extraction of multiple bioactive compounds from NS seeds. The extraction process using methanol effectively retrieves phenolic acids and flavonoids and alkaloids and saponins and fatty acids which less polar solvents cannot extract efficiently. Studies indicated that N. sativa methanolic extracts include more diverse phenolic compounds than extracts obtained through water or non-polar solvent extraction methods^85,86. Similarly, Zribi et al., reported the highest levels of total phenolic content (TPC), total flavonoid content (TFC), and total flavonoid-like compounds (TFlC) in the methanolic extracts of N. sativa⁸⁷.The results show that methanol extracts contain a wide range of phenolic compounds which proves the effectiveness of this solvent.

Cell culture growth

The human colon cancer cell line Caco-2 (HTB-37) was obtained from ATCC (The American Type Culture Collection). Caco-2 cells were cultured using Eagle’s Minimum Essential Medium (EMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Cells were grown to 80–85% confluency and maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

NS seed extract treatment and cell proliferation assay

Caco-2 cells were seeded in a 96 well tissue culture plate (1 × 10⁴ cells/well) in a final volume of 100 µL in culture medium. After 24 h incubation, the cells were treated with a range of concentrations of N. sativa (NS) methanolic extract (10 µl/mL, 12 µl/mL, 15 µl/mL, 18 µl/mL, 20 µl/mL) for 24 h. Prior to cellular treatment, crude extract was first dissolved in 1% dimethyl sulfoxide (DMSO) and then diluted 50–100 times with complete cell culture medium. The final DMSO concentration was obtained ≤ 0.1% which is considered non-toxic for most mammalian cell culture systems. Concentrations were expressed as volume per culture medium (µL/mL) because the crude methanolic extract is a complex mixture without a defined molecular weight; therefore, conventional mass-based units (e.g., µg/mL) are not applicable unless the active constituents are fully purified and quantified This is standard practice for crude extracts which is consistent with the previous phytochemical screening studies^29,88.

Each concentration of the NS seed extract was tested using six technical replicates (six wells per concentration) in a 96-well tissue culture plate. The entire experiment was performed three independent times to ensure reproducibility. The XTT (XTT Cell Viability Assay, Cell Signaling) method was performed using the manufacturer’s protocol for the cell proliferation assay. At the end of the 24 h treatment of NS seed extract, XTT solution (50 µl/well) was directly added to each well and incubated at 37 °C for 4 h. The absorbance at 450 nm was measured with an ELISA plate reader with methanol as a blank control. The cell viability was plotted as a percentage of the value obtained for untreated control cells, and the IC50 value (or the concentration of the drug that caused a 50% inhibition of cell growth) was calculated.

IR spectral data collection and spectral analysis

Sample preparation and spectrum collection

For IR spectroscopy experiments, cells were grown independently three times, maintaining the same cell passage number. When the cells reached 70–80% confluency, 15 µl (IC50 value) NS seed extract was applied. Both control and 15 µl NS-treated cells were cultured for 24 h since the IC50 value this IC₅₀ concentration level reduces cell viability by 50% and leads to detectable biochemical changes within 24 h while retaining viable cells. Then, the cells were washed with PBS, trypsinized, and counted. Two million cells were resuspended in 10 µl PBS, and three different aliquots of 3 µl each were prepared from each application. After each application, the samples were dried by mild nitrogen flux for 5 min on an ATR crystal (Bruker Alpha II).

The samples were scanned over the 4000 –750 cm⁻¹ spectral range with 64 scans per sample at 4 cm⁻¹ resolution. The spectrum of air was used as a reference and automatically subtracted from the sample spectra. For both NS-treated and untreated controls, three independent biological replicates were examined, which were processed under identical experimental conditions. Each biological replicate represented a separate cell culture preparation and treatment. The three technical replicate spectra from each biological replicate sample were collected from separate aliquots of each application to measure intra-sample variability. In total, nine spectra were obtained for each experimental group. The reproducibility of these replicates was confirmed by consistent clustering patterns observed in PCA and HCA analyses. The experimental workflow for ATR-FTIR measurement of NS treated Caco-2 cells, from sample loading on the ATR crystal to spectral acquisition, is demonstrated in Figure S2B.

Chemometric analysis

All chemometric analyses were conducted on FTIR spectra obtained from experimentally treated biological samples rather than on in silico spectral data set. These analyses relied on computational post-processing of the collected sample spectra.

Prior to chemometric analyses, the spectra were baseline-corrected with the Rubber band correction method with 64 baseline points (Opus Software). Then, the second derivative of the spectra was acquired (polynomial order:2, smoothing points:13) using Unscrambler^® X 10.3 (CAMO Software AS, Norway) software. For this analysis, nine spectra were used for each group.

Two unsupervised chemometric analyses, including PCA and HCA, were performed on the whole range of the spectra as exploratory analyses. To reduce random variation effects, the singular value decomposition (SVD) algorithm with full cross-validation were used in the PCA. PCA results were given as scatter and loading plots. PCA loading plots were analyzed to identify the contribution of spectral bands for their discrimination as reported previously^51,89. HCA was performed using Euclidean distance as the similarity measure and Ward’s linkage method, which are widely recommended for spectral datasets because they minimize intra-cluster variance and produce compact, interpretable groups^51,90–92.The result was demonstrated in the HCA as a dendrogram.

Numerous studies in biomedical vibrational spectroscopy adopt that Principal component analysis (PCA) was used for unsupervised dimensionality reduction and exploratory visualization before applying supervised classifiers. For example, Baker et al. (2014) and Severcan et al., 2024 used PCA as a robust unsupervised tool to visualize clustering and identify discriminative spectral bands, often followed by supervised methods for confirmatory analysis^46,91. Our analytical workflow aligns with these best practices. Because our primary objective was exploratory to assess whether NS seed extract produced reproducible and statistically separable biochemical fingerprints, PCA was an appropriate first-line approach and then we applied supervised methods including LDA, SIMCA, and SVM.

Based on the PCA and HCA segregations, three supervised analyses were employed namely linear discriminant analyses (LDA), Soft Independent Modelling of Class Analogies (SIMCA), and support vector machine (SVM). To achieve these analyses, training (7 samples/group) and test (2 samples/group) data sets were formed^46,89. In the training/test set, a category variable was added. The PCA training data set was used as LDA model input for LDA. The linear method was used for the projection of the 2 PC components. The results were shown as a discrimination plot and prediction table for the training data set. In the case of SIMCA, PCA models were acquired for each group. These models were used as input for the SIMCA. For the SVM analysis, Nu-Support Vector Classification (nu-SVC) SVM was selected using a linear function (nu value: 0.826, Weights: 1.00, Cross-validation segments:10). In addition to the training data set, LDA, SIMCA, and SVM analyses were performed to test data set, and the prediction was given as a table.

Qualitative and quantitative spectral analyses

All qualitative and quantitative spectral analyses were carried out using computational post-processing and calculation approaches, performed exclusively on experimentally obtained, baseline-corrected FTIR spectra, rather than on in silico spectra.

To visually demonstrate the spectral differences among the studied groups, difference spectra were obtained by subtracting the mean absorbance spectra of the control group from those of the NS-treated group. To further characterize these biomolecular alterations, quantitative spectral analyses were performed. For this purpose, all the spectrum was first baseline-corrected to perform quantitative spectral analyses (Rubber band correction method with 64 baseline points). Then, quantitative spectral analyses were then performed. Firstly, band wavenumber (peak position) called the wavenumber corresponds to maximum absorbance within a given spectral region, was determined after second-derivative transformation for peak separation. Then band area was measured by integrating absorbance between pre-defined start and end wavenumbers of the peak and band area ratio were obtained by dividing the areas of two selected bands. Finally, the full width at half maximum, (FWHM) of band width was determined by measuring peak widths at 80% of the maximum absorbance height.

Membrane lipid order and fluidity were also evaluated using the CH₂ symmetric (~ 2850 cm⁻¹) and asymmetric (~ 2920 cm⁻¹) stretching bands of lipid acyl chains. The peak positions and full width at 80% maximum values were determined from second-derivative, baseline-corrected spectra. Shifts toward higher wavenumbers and broadening of band width indicate increased acyl chain disorder and membrane fluidity, respectively^46,93,94 .

The performed analyses were performed in the two main spectral regions: the 3035–2800 cm⁻¹ - C-H stretching and 1800–800 cm⁻¹ - fingerprint regions. The first region contains mainly lipid associated spectral bands whereas the second region includes protein, carbohydrate, nucleic acid and carbohydrate related spectral bands, which are the most diagnostically relevant vibrational modes of cellular macromolecules. Thus, they are widely used for distinguishing biochemical changes associated with cancer progression and treatment response^46,52,95,96. All qualitative and quantitative analyses were employed using OPUS 5.5 (Bruker, Germany) software.

Statistical analysis

Statistical analyses were performed via GraphPad Prism 6 software (La Jolla, CA, USA). All data were presented as mean + SEM. One-way ANOVA with Sidak’s multiple comparison test was used for cell viability results. For quantitative spectral analysis, the unpaired t-test was performed. The degree of significance was denoted as *p < 0.05; **p < 0.01; ***p < 0.001., ****< p.0001.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: F.S. Sample collection: H.S. Spectral collection: I.O. IR spectral data analysis: F.C., I.O., N.S.O. Multivariate data analysis: N.S.O. Interpretation: All authors Visualization: F.S., F.C., I.O., N.S.O. Supervision: F.S. Writing—original draft: All authors.

Funding

This work was funded by the Altinbas University Research Fund (PB2018-GÜZ-TIP-2).

Data availability

The datasets generated during the current study are available from the corresponding author and will be made publicly available via Zenodo upon manuscript acceptance (10.5281/zenodo.17950195).

Declarations

Competing interests

The authors declare no competing interests.