Skip to main content
NCBI home page
RSC Advances logoLink to RSC Advances
. 2026 Jul 6;16(35):36883–36893. doi: 10.1039/d6ra03939a

Integration of Raman spectroscopy and chemometric modeling to study lactose effects on human glioblastoma brain cells

Monika Anna Kopeć a,✉, Michał Sylwester Kopeć b, Jakub Maciej Surmacki a
PMCID: PMC13334372  PMID: 42440827

Abstract

Lactose, a disaccharide composed of glucose and galactose, serves not only as a dietary carbohydrate but also as a versatile metabolic substrate in mammalian cells. In this study, we examined the role of lactose in human glioblastoma cells (U-87 MG) using Raman spectroscopy supported by chemometric analysis. In addition, we compared the cellular responses of glioblastoma cells and normal astrocytes to identify potential biochemical differences between malignant and non-malignant cells. Our results indicate that Raman bands associated with cytochromes, proteins, carbohydrates, and nucleic acid-related vibrations may serve as potential spectroscopic markers of biochemical alterations in human brain cells following lactose supplementation. The observed spectral changes suggest that lactose exposure may be associated with modifications in cellular metabolism and bioenergetic homeostasis; however, direct effects on protein post-translational modifications or epigenetic regulation were not experimentally validated in the present study. Furthermore, our results indicate that Raman spectroscopy combined with PLS-DA analysis may provide additional insight into the biochemical responses of brain cells following lactose supplementation, potentially contributing to a better understanding of metabolic alterations relevant to neuronal physiology and disease-associated cellular remodeling.

Raman spectroscopy supported by chemometric analysis showed that lactose supplementation modulates the biochemical profile of U-87 MG glioblastoma cells. Changes in cytochrome-, protein-, carbohydrate-, and nucleic acid-related bands may serve as biomarkers.graphic file with name d6ra03939a-ga.jpg

Introduction

Lactose is a disaccharide carbohydrate composed of two monosaccharide units. Its molecular structure consists of one d-glucose and one d-galactose molecule, connected by a β-1,4-glycosidic bond, which defines its chemical structure and properties.1–3

Lactose is present in human breast milk as well as in the milk of domesticated animals, including cows, goats, and sheep. The concentration of lactose varies among species and is influenced by genetic, physiological, and environmental factors. Consequently, lactose is widely found in dairy products such as milk, yogurt, kefir, cream, and certain types of cheese, where it plays an important role in human nutrition, particularly during infancy and early childhood.4–7 In the digestive tract, lactose is hydrolyzed by the enzyme lactase into its monosaccharide components, glucose and galactose.8,9 These monosaccharides are absorbed in the small intestine and transported into the bloodstream.10 Glucose is the primary energy substrate for the central nervous system and is essential for maintaining proper neuronal activity and brain homeostasis. Neurons depend almost exclusively on glucose metabolism to generate ATP required for synaptic transmission, maintenance of membrane potentials, and neurotransmitter synthesis. During brain development, glucose supports neuronal proliferation, differentiation, axonal growth, and synaptogenesis, thereby contributing to the formation and maturation of neural networks. Impaired glucose metabolism may lead to disturbances in neuronal communication, cognitive dysfunction, and the development of various neurological disorders.11–14 Galactose is involved in the synthesis of glycoproteins and glycolipids, which are crucial for brain development and proper functioning of the nervous system.15,16 Lactose consumption has also been shown to enhance the intestinal absorption of minerals, including calcium, magnesium, and phosphorus. Additionally, lactose supports the growth of beneficial intestinal microbiota, particularly lactic acid bacteria. This effect contributes to maintaining gut health and supporting immune function. In infancy, lactose is especially important due to the high energy demands associated with rapid growth and development. Insufficient lactase activity leads to lactose intolerance, resulting in gastrointestinal symptoms caused by incomplete digestion of lactose.17,18

It is worth emphasizing that lactose isn't naturally present in the human brain, but lactate, its metabolic product, plays a crucial role in brain tumor biology.19 Neurons and astrocytes depend on a continuous supply of glucose and oxygen to sustain normal brain function. Neurons represent the primary consumers of energy within the brain, accounting for approximately 70–80% of total cerebral energy demand, while the remaining fraction is utilized by glial cells, including astrocytes, oligodendrocytes, and microglia. During periods of increased neural activity, lactate oxidation is markedly enhanced. Accordingly, the uptake of glutamate by astrocytes stimulates glucose metabolism, leading to increased lactate production and its subsequent release to support neuronal energy requirements during sustained activation. In addition in the brain, glucose is broken down through glycolysis, the tricarboxylic acid (TCA) cycle, and oxidative phosphorylation, where it is almost completely converted into carbon dioxide and water. This process produces about 30–36 molecules of ATP from each molecule of glucose, depending on how efficiently oxidative phosphorylation operates.20–25 This metabolic process in neuron and astrocyte cells is presented in Fig. 1.

Fig. 1. The metabolism process of glucose in neurons and astrocytes.

Fig. 1

Open in a new tab

In gliomas and other brain cancers, tumour cells undergo metabolic reprogramming, relying heavily on glycolysis even under aerobic conditions (the Warburg effect), which leads to high lactate production.26 Accumulated lactate serves not only as an energy substrate but also as a signalling molecule, influencing pathways that promote tumour cell proliferation, migration, and survival via transporters such as MCT1 and MCT4 and receptors like HCAR1/GPR81.20,27,28 Additionally, lactate contributes to epigenetic regulation through histone lactylation, altering gene expression in a way that supports malignant phenotypes, therapy resistance, and stem-like properties in cancer cells.29 High lactate levels also modulate the tumour immune microenvironment by polarizing macrophages toward immunosuppressive phenotypes and inhibiting T-cell function, further facilitating tumour progression.30 These effects collectively underscore the importance of lactate metabolism in brain tumours and have spurred interest in targeting lactate production, transport, and lactylation pathways as potential therapeutic strategies to inhibit tumour progression, restore immune function, and enhance treatment responses.31

Recent Raman spectroscopy studies have demonstrated the growing importance of vibrational imaging and machine learning approaches in glioblastoma research.32,33 Ember et al.34 reported the application of Raman spectroscopy combined with machine learning for real-time intraoperative discrimination of brain tumors, including glioblastoma, highlighting the diagnostic potential of Raman-based biochemical profiling in clinical neurosurgery. In turn, Wurm et al.35 demonstrated that deep learning-enhanced Raman spectroscopy enables discrimination between pediatric and adult glioblastoma stem cells based on subtle differences in their biochemical fingerprints, further emphasizing the sensitivity of Raman methods to tumor-associated metabolic heterogeneity. These studies collectively confirm the growing role of Raman spectroscopy as a label-free analytical platform for investigating biochemical diversity and metabolic alterations in glioblastoma-related systems. The effect of glucose on human brain cell lines has been previously reported.11 The present study extends this line of research by examining the influence of lactose supplementation on human brain cells, with particular emphasis on biochemical and metabolic alterations occurring in brain cancer cells. Although Raman spectroscopy has previously been applied to the investigation of glioblastoma and normal brain cell biochemistry, the novelty of the present work lies in the analysis of organelle-resolved biochemical changes associated with lactose exposure using Raman imaging combined with multivariate statistical analysis, including Partial Least Squares Discriminant Analysis (PLS-DA). This approach enables characterization of subtle metabolic responses and intracellular biochemical heterogeneity at the single-cell level.

Materials and methods

Cell culturing and preparation for measurements

Experiments were performed using normal human astrocytes (NHA) and the human glioblastoma cell line (U-87 MG). The NHA cells were purchased from Lonza and grown in Astrocyte Medium Bulletkit Clonetics and ReagentPack without antibiotics in a humidified incubator at 37 °C and 5% CO2 atmosphere. The U-87 MG cells were purchased from ATCC and grown in Eagle's Minimal Essential Medium with l-glutamine supplemented with 10% fetal bovine serum without antibiotics in a humidified incubator at 37 °C and 5% CO2 atmosphere.

Twenty-four hours before Raman examination, cells (both without supplementation and supplemented) were seeded at a density of 5 × 104 cells per 35 mm Petri dish onto CaF2 Raman-grade windows (Crystran Ltd.).

After 24 h of growth, U-87 MG cells were supplemented with lactose at final concentrations of 5.5 and 50 mM in cell line-dedicated culture medium. Lactose was first dissolved in sterile water, and the appropriate working solutions were subsequently prepared directly in the culture medium immediately prior to cell treatment. Cells were then incubated with lactose for 24 h before Raman analysis.

Before measurements, all cells were fixed in 4% neutrally buffered formalin and subsequently immersed in phosphate-buffered saline (PBS) throughout the experiment.

Chemical compounds

Lactose 237696-50G (CAS Nr 64044-51-5) was purchased from PPH Galfarm Sp. z o.o.

Raman spectroscopy

All Raman measurements described in this manuscript were conducted using a confocal Raman microscope WITec alpha300 RSA+ (Ulm, Germany) equipped with 532 nm excitation wavelength (laser diode (SHG of the Nd:YAG laser)), UHTS (Ultra High Throughput Spectrometer) spectrometer and a CCD Camera (Andor Newton DU970N-UVB-353) operating in standard mode with 1600 × 200 pixels at –60 °C with full vertical binning. Measurements for individual cells were performed using 40× Nikon water dipping objective (NA = 1.0) and the power of the laser 10 mW. Raman maps of normal and cancer brain cells were recorded in the fingerprint region with an integration time of 0.5 s and in the high-frequency region with an integration time of 0.3 s. Acquisition and preprocessing of the data (cosmic rays removing (model: filter size: 2, dynamic factor: 10), smoothing (Savitzky and Golay procedure (model: order: 4, derivative: 0)), and removing background (model: shape, shape size: 1000)) were performed with WITec Project Plus software. The distribution of main cellular organelles was determined by using the Cluster Analysis (CA) method. A detailed description of the Cluster Analysis method is presented in our previous papers.36–38 Cluster Analysis is a method used to group Raman spectra into clusters based on their spectral similarity, where spectra within the same cluster are more similar to each other than to spectra belonging to different clusters. The separation of n observations (x) into k (k ≤ n) clusters S should be done to minimize the variance (Var) according to the formula:graphic file with name d6ra03939a-t1.jpgwhere µi is the mean of experimental points. In our study, the number of clusters was 7. Each cluster is characterized by specific average Raman spectra, which reflect the inhomogeneous distribution of the sample.

Raman images were recorded with a spatial resolution of 1 × 1 µm. A typical Raman map of a single cell consists of 1200 Raman spectra (map size 30 × 40 µm). Total of cells subjected to full Raman imaging: n(NHA) = 5, n(U-87 MG) = 7, n(U-87 MG with 50 mM lactose) = 7, n(U-87 MG with 5.5 mM lactose) = 7. The numbers of Raman spectra used for averaging were 6000, 18 125, 11 288, and 8273, respectively. Given the proof-of-concept nature of this study and the limited number of cells analyzed, it is important to note that the high sensitivity of Raman spectroscopy to molecular heterogeneity enables reliable biochemical characterization even without large sample sizes, with line scans across cells providing representative molecular information. To PLS-DA analysis we used Raman spectra from line scans. The line scans were acquired across entire cells, intentionally traversing different cellular regions in order to account for intracellular heterogeneity and to include contributions from various organelles, including both nuclear and cytoplasmic compartments. The positioning of the lines was therefore designed to provide a representative biochemical profile of the whole cell rather than of a single subcellular structure.

Statistical analysis

All presented Raman spectra were normalized (model: divided by norm) using Origin software. Detailed descriptions of the normalization process are described extensively in our previous papers.39,40

PLS-DA analysis was performed using Matlab and PLS_Toolbox Version 4.0. As an input matrix for the PLS-DA models, we used Raman spectra collected along a line-scan across individual cells in the fingerprint region. Data were preprocessed using mean-centering and smoothing (0th-order, 15-points window). Cross-validation was conducted using the Venetian blinds approach with 10 data splits. The model was constructed using seven latent variables (LVs).

The Matthews correlation coefficient (MCC) was calculated according to the formula:graphic file with name d6ra03939a-t2.jpg

The Youden's index (J) was calculated according to the formula:graphic file with name d6ra03939a-t3.jpgwhere: TN- the true negative result. FP- the false positive result. FN- the false negative result.

Results and discussion

This study aimed to characterize biochemical alterations induced by lactose supplementation in human brain cells using vibrational spectroscopy coupled with chemometric analysis. Raman spectroscopy, together with Raman imaging, was employed to obtain detailed biochemical information and to characterize molecular alterations occurring in both normal and cancerous human brain cell lines. Fig. 2 shows the Raman spectra and images for a typical normal human brain cell (NHA) and a cancer human brain cell (U-87 MG). Based on Raman maps obtained using Cluster Analysis, the spatial distribution of the nucleus (red color), mitochondria (magenta color), lipid droplets (orange color), endoplasmic reticulum (blue color), cytoplasm (green color) and cell membrane (gray color) can be resolved, and their biochemical composition can be inferred from the corresponding Raman spectra.

Fig. 2. The typical microscope images, Raman maps (from the area marked with a red frame), Cluster Analysis images and Raman spectra of cell organelles of a normal astrocyte (NHA) and glioblastoma (U-87 MG) cells.

Fig. 2

Open in a new tab

To elucidate the cellular physiological response to lactose in brain cancer cells, Raman spectroscopy and Raman imaging were employed. Fig. 3 presents the Raman images and corresponding spectra acquired from U-87 MG cells following lactose supplementation.

Fig. 3. The typical microscope images, Raman maps (from the area marked with a red frame), Cluster Analysis images and Raman spectra of cell organelles of a glioblastoma (U-87 MG) cells supplemented with 5.5 mM and 50 mM lactose.

Fig. 3

Open in a new tab

Results presented in Fig. 1 and 2 demonstrate that Raman spectra exhibit characteristic features enabling the identification of the major biochemical components of single human brain cells. The fingerprint region of the Raman spectra is especially informative, as it reflects the complex biochemical composition of the investigated samples. In the fingerprint region, the band at 750 cm−1 is characteristic of cytochrome c and b,41 while a band at 1004 cm−1 is characteristic of phenylalanine.42 Another vibration characteristic for cytochrome c and b is observed at 1126 cm−1.41 The band at 1245 cm−1 corresponds to Amide III.43 The band at 1305 cm−1 corresponds to CH2 deformation (lipid), adenine, cytosine.44 The band at 1444 cm−1 corresponds to scissoring vibration observed in lipids, fatty acids and cholesterol.45–48 The band at 1583 cm−1 is associated with cytochrome c.49,50 Band characteristic of Amide I is observed at 1656 cm−1.51 The high-frequency spectral region reveals characteristic Raman bands attributed to the symmetric stretching mode of CH2 groups in lipids at 2845 cm−1 and the asymmetric stretching mode of CH2 groups in proteins at 2930 cm−1.42,52

To better visualize the effects of lactose on brain cell lines, chemometric approaches were applied. PLS-DA analysis was employed to discriminate between normal and cancer cell lines, as well as between cancer cell line without supplementation and lactose-supplemented cancer cell line. Fig. 4 shows the score plots and ROC curves obtained from PLS-DA analysis for NHA and U-87 MG cell lines. Raman spectra characteristic for astrocytes are marked by blue color, for glioblastoma by red color, for glioblastoma supplemented with 50 mM of lactose by green color and for glioblastoma supplemented by 5.5 mM of lactose by black color.

Fig. 4. PLS-DA score plot on Raman spectra acquired from a line scan across individual cell for NHA (blue circle), U-87 MG (red circle), U-87 MG cells supplemented with 50 mM lactose (green circle) and 5.5 mM lactose (black circle) (panel A); the ROC curve graphs for plotting specificity versus sensitivity for the PLS-DA model based on Raman data for normal and cancer brain cell lines and cancer brain cell lines supplemented with 50 mM lactose and 5.5 mM lactose (panel B).

Fig. 4

Open in a new tab

The score plots (Fig. 4A) demonstrate clear separation between the spectra of NHA cells (blue circle) and those of U-87 MG cells (red, green and black circles), indicating distinct biochemical profiles between normal and cancer cells. The most visible separation is observed along the first latent variable (LV1), which represent 30.10% of the data variability. All of the spectra of astrocyte cells are positioned on LV1(–) and most of the glioblastoma cells on LV1(+). The spectra of U-87 MG cells supplemented with 50 mM lactose (green circle) are separated along the second latent variable (LV2). All spectra representing this type of supplementation are located on LV2(–), whereas the spectra of U-87 MG cells supplemented with 5.5 mM (black circle) are separated along LV4. To verify the accuracy of the test, we calculated the values of sensitivity and specificity. Fig. 4B presents the ROC curve graphs with the values of sensitivity and specificity. One can see that the classification accuracy parameters are characterized by high values. The sensitivities (calibration) are equal 100%, 97.1%, 97.1%, and 97.1% for NHA, U-87 MG, U-87 MG with 50 mM lactose, and U-87 MG with 5.5 mM lactose, respectively. While the specificities (calibration) for NHA, U-87 MG, U-87 MG with 50 mM lactose, and U-87 MG with 5.5 mM lactose are equal 100%, 99% 99% and 100%, respectively. The values of sensitivity for cross-validation equal 100% (NHA), 82.9% (U-87 MG), 97.1% (U-87 MG with 50 mM lactose), 85.7% (U-87 MG with 5.5 mM lactose). The specificity (cross-validation) is equal to 100% (NHA), 95% (U-87 MG), 97% (U-87 MG with 50 mM lactose) and 91% (U-87 MG with 5.5 mM lactose).

To evaluate the classification potential of Raman spectra for the investigated normal and cancer human brain cells, as well as cancer brain cells without supplementation and lactose-supplemented cancer cells, the PLS-DA was applied, including analysis of latent variables (LVs) and variable importance in projection (VIP) scores. The spectral features most relevant to cell differentiation were identified using VIP scores derived from the PLS-DA model. The obtained results are presented in Fig. 5.

Fig. 5. Loading plots for PLSDA, VIP scores for the PLS-DA model based on Raman data for NHA cells (blue line), U-87 MG cells (red line), U-87 MG cells supplemented with 50 mM lactose (green line) and 5.5 mM lactose (black line) and the confusion matrix and confusion table from the PLS-DA analysis.

Fig. 5

Open in a new tab

It is visible from Fig. 5 that LV1, LV2 and LV3 give the dominant account for the maximum variance of the data. It is 30.10%, 5.36% and 10.16%, respectively. Fig. 5 also illustrates the VIP scores obtained from the PLS-DA model, highlighting the Raman spectral features that contribute most to sample differentiation. Peaks with VIP values above 1 are emphasized as the most influential variables. This findings indicate that peaks at 553 cm−1, 624 cm−1, 783 cm−1, 1043 cm−1, 1078 cm−1, 1224 cm−1, 1444 cm−1 and 1656 cm−1 were identified as the most influential variables. These important features reflect underlying biochemical differences in the samples. Overall, VIP-based interpretation provides a clear way to connect spectral data with relevant biological information. The Raman band observed at 1043 cm−1 can be attributed to vibrational modes characteristic of carbohydrates, primarily C–O and C–C stretching as well as glycosidic (C–O–C) bonds.42,53,54 In contrast, the band located at 1483 cm−1 is typically associated with CH2 and CH3 bending (scissoring) vibrations, commonly observed in lipids and proteins, as well as with aromatic ring stretching modes. Literature reports indicate that in biological samples, this band is often linked to aliphatic hydrocarbon chains and may reflect contributions from membrane lipids or protein side chains.53

To assess the classification performance of the PLS-DA model, a confusion matrix was constructed to evaluate class assignment accuracy.

To evaluate the quality of the classification model applied to Raman spectra, we have calculated the Matthews correlation coefficient. Additionally, to provide a single, prevalence-independent measure of the diagnostic test's overall discriminatory ability and to identify the optimal cutoff value that maximizes the combined sensitivity and specificity, we have calculated the Youden's index. The obtained values of MCC and J are presented in Table 1.

Table 1. The MCC and J values for NHA, U-87 MG, U-87 MG with 5.5 mM lactose and U-87 MG with 50 mM lactose cells.

Class MCC Youden's index
NHA 1 1
U-87 MG 0.799891 0.779
U-87 MG with 5.5 mM lactose 0.832459 0.767
U-87 MG with 50 mM lactose 0.913196 0.941
Open in a new tab

One can see from Table 1 that the values of MCC and J are high for all analyzed classes, indicating a strong agreement between predicted and true class labels.

Now, let us focus on the interpretation of the PLS-DA loadings plots (Fig. 5). The results confirm that Raman bands associated with cytochromes (750 cm−1, 1126 cm−1, 1583 cm−1), lipids (1444 cm−1), DNA (1305 cm−1), carbohydrates (1043 cm−1 and 1483 cm−1) and proteins (1004 cm−1, 1245 cm−1, 1656 cm−1) serve as key spectral biomarkers for monitoring biochemical changes in brain cells following lactose supplementation. The changes observed in Raman bands characteristic of cytochromes may reflect alterations in mitochondrial redox balance and respiratory activity. As we mentioned earlier, lactose is not directly metabolized by brain cells or mitochondria; its effects are likely mediated indirectly through increased availability of glucose and galactose following extracellular or systemic hydrolysis. Enhanced monosaccharide supply can stimulate glycolysis and increase pyruvate flux into mitochondria, leading to elevated electron transfer through the respiratory chain and a shift in cytochrome redox state toward the reduced form.37,55–59 What is interesting in the loading plot, we observed only one strong band characteristic for lipid and derivatives (1444 cm−1). The lack of Raman markers of lipid oxidation can suggest that lactose supplementation promotes metabolic activation rather than mitochondrial dysfunction or oxidative damage.60–62 In brain cells, such effects may be amplified by neuron–astrocyte metabolic coupling, where increased substrate availability enhances mitochondrial oxidative phosphorylation in neurons.60,63,64 Therefore, the Raman spectral changes observed after lactose supplementation could be associated with adaptive bioenergetic responses of mitochondria linked to altered cellular metabolism.

The changes observed in DNA-related Raman signatures within brain cells following lactose supplementation may be associated with metabolic alterations and could potentially reflect indirect epigenetic-related processes. Lactose, after hydrolysis into glucose and galactose, provides additional substrates for central carbon metabolism, increasing ATP production and the availability of key metabolites such as acetyl-CoA, S-adenosylmethionine (SAM), and nucleotide sugars.65,66 These metabolites serve as cofactors for chromatin-modifying enzymes, including histone acetyltransferases, DNA methyltransferases, and enzymes involved in nucleotide biosynthesis, thereby influencing DNA methylation patterns, chromatin accessibility, and overall transcriptional activity. Enhanced energy flux and redox balance can further modulate DNA repair processes and replication fidelity, while changes in osmotic pressure and macromolecular crowding may subtly affect DNA packaging and higher-order chromatin structure.67–69

We now turn to the analysis of Raman bands attributed to proteins. The alterations observed in the Raman band characteristic for protein structure and composition (1004 cm−1, 1245 cm−1, 1656 cm−1) in brain cells following lactose supplementation can be explained by a combination of metabolic and structural mechanisms. Lactose, after hydrolysis into glucose and galactose, provides additional energy substrates that enhance glycolytic flux and mitochondrial ATP production, thereby increasing the availability of biosynthetic precursors required for protein synthesis.65,66 This metabolic shift can modulate the expression of cytoskeletal proteins, enzymes, and membrane-associated proteins, while also influencing post-translational modifications such as O-GlcNAcylation and glycosylation, which affect protein folding, stability, and intermolecular interactions.70,71 In parallel, changes in osmotic balance, macromolecular crowding, and hydration dynamics may subtly alter hydrogen bonding networks within protein backbones, leading to detectable shifts in secondary structure.72,73

Taken together, Raman spectroscopy combined with chemometric analysis provides robust discrimination between different stages of brain cancer and enables accurate identification of lactose-supplemented cells.

While the present study provides new insights into organelle-specific biochemical alterations in NHA and U-87 MG cells following lactose supplementation, the findings should be interpreted within the context of several important limitations. The observed spectroscopic changes are based exclusively on Raman imaging data and therefore remain primarily correlative in nature. Consequently, although the results suggest the presence of metabolic and bioenergetic alterations associated with lactose exposure, direct mechanistic conclusions cannot be drawn at this stage. Future studies integrating Raman spectroscopy with complementary functional and biochemical assays, including viability assessment, ATP and LDH measurements, ROS analysis, mitochondrial membrane potential evaluation, lactate quantification, and western blot validation, will be important to further characterize the underlying cellular mechanisms and validate the observed metabolic alterations.

In addition, the relatively limited number of independent biological replicates analyzed in each experimental group should be considered when interpreting the data. Although thousands of Raman spectra were acquired, these measurements originated from a restricted number of biological samples and therefore should not be regarded as fully independent observations. Nevertheless, the consistency of the spectral trends observed across analyzed cells supports the overall reproducibility of the detected biochemical patterns.

An additional aspect requiring further investigation concerns the precise mechanism underlying the observed lactose-associated cellular responses. Since brain cells are not generally considered to directly metabolize lactose, it remains unclear whether lactose hydrolysis occurred in the culture medium, whether the cells utilized lactose itself or its hydrolysis products, or whether some of the observed biochemical alterations were partially influenced by osmotic effects in the extracellular environment. Therefore, future studies incorporating osmolarity controls, measurements of lactose degradation, glucose and galactose utilization, and complementary metabolic analyses will be necessary to better distinguish specific metabolic responses from nonspecific osmotic effects.

Conclusions

Our study demonstrates the effectiveness of Raman spectroscopy coupled with chemometric methods as a powerful approach for differentiating normal and cancer brain cells and cancer brain cells supplemented with lactose.

The Raman spectroscopic and PLS-DA analyses clearly demonstrate that supplementation with lactose induces profound biochemical and metabolic changes in brain cancer cells. These alterations are reflected in the spectral regions associated with cytochromes, proteins, carbohydrates and nucleic acids. Conducted results suggest that bands at 728 cm−1, 750 cm−1, 785 cm−1, 1004 cm−1, 1043 cm−1, 1095 cm−1, 1126 cm−1, 1245 cm−1, 1305 cm−1, 1375 cm−1, 1483 cm−1, 1583 cm−1 and 1656 cm−1 are the candidates for Raman biomarkers to track metabolism changes in human brain cells supplemented with lactose.

Chemometric analysis was performed to assess the similarities between the Raman spectra of brain cells without supplementation and those supplemented with lactose. The values of sensitivity (cross-validation) obtained for analyzed cell types equal 100% (NHA), 82.9% (U-87 MG), 97.1% (U-87 MG with 50 mM lactose), 85.7% (U-87 MG with 5.5 mM lactose). The values of specificity (cross-validation) equal 100% (NHA), 95% (U-87 MG), 97% (U-87 MG with 50 mM lactose), 91% (U-87 MG with 5.5 mM lactose).

By integrating Raman spectroscopy with chemometric analysis, our approach demonstrates strong potential for rapid and reliable diagnostic applications based on vibrational spectral biomarkers. These findings advance the understanding of lactose-associated metabolic pathways in cancer progression and may contribute to the development of objective tools for clinical diagnosis.

Author contributions

Conceptualization: M. A. K.; investigation: M. A. K.; methodology: M. A. K., J. M. S.; sample preparation: J. M. S.; chemometric analysis: M. A. K.; M. S. K.; writing – original draft: M. A. K.; manuscript editing: M. A. K., J. M. S., M. S. K.; all authors reviewed and provided feedback on the manuscripts. All authors have read and agreed to the published version of the manuscript.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgments

The research was supported by the Statutory Funding from Lodz University of Technology, Poland.

Data availability

Data for this article are available at Repozytorium Otwartych Danych Badawczych at https://doi.org/10.34658/RDB.OIUID8.

References

  1. Odell O. J. Wallis G. A. Int. Dairy J. 2021;116:104970. doi: 10.1016/j.idairyj.2020.104970. [DOI] [Google Scholar]
  2. Dominici S. Marescotti F. Sanmartin C. Macaluso M. Taglieri I. Venturi F. Zinnai A. Facioni M. S. Foods. 2022;11:1486. doi: 10.3390/foods11101486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Fassio F. Facioni M. S. Guagnini F. Nutrients. 2018;10(11):1599. doi: 10.3390/nu10111599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Portnoy M. J. Dairy Sci. 2021;104:8314–8325. doi: 10.3168/jds.2020-18706. [DOI] [PubMed] [Google Scholar]
  5. Chengolova Z. Shentov P. Ivanova R. Syarova R. Dairy. 2024;5:702–726. doi: 10.3390/dairy5040052. [DOI] [Google Scholar]
  6. Anguita-ruiz A. Vatanparast H. Walsh C. Natoli S. Eisenhauer B. Ramirez-mayans J. Anderson H. Guerville M. Ligneul A. Gil A. Vatanparast H. Walsh C. Natoli S. Eisenhauer B. Ramirez-mayans J. Anderson G. H. Guerville M. Ligneul A. Gil A. Alternative F. CRC Crit. Rev. Food Sci. Nutr. 2025:1–14. doi: 10.1080/10408398.2025.2470394. [DOI] [PubMed] [Google Scholar]
  7. Lee B. and Szilagyi A., in Dietary Sugars: Chemistry, Analysis, Function and Effects, eds. J. R. Alonso-Fernandez, D. J. Timson, A. Szilagyi, R. L. Obendorf, M. F. Abdelmalek, M. McGrath, L. Boukraâ, M. D. Gastrich, A. Haukioja, J. Jensen, M. Erkkola, K. A. Amin, M. F. Varela, F. Vieira dos Santos, L. M. Moreira, L. Hansen, M. B. Hall, T. Fujisawa, D. S. Millington, E. Denova-Gutierrez, M. Benaissa, D. Centonze, D. A. L. Aarts, J. J. Gagliardino, M. L. Sanz, L. A. Terry, A. Abdul-Aziz, M. M. F. Choi, E. M. S. M. Gaspar, J. L. Todoli, K. Sugimoto, W. Surareungchai, L. M. Moreira, J. N. Keith, N. Silanikove, D. S. Newburg, O. I. Bermudez, A. L. Deierlein, K. McCowen, M. J. Rodriguez-Yoldi, J. G. Jones, C. A. Pinkert, N. T. Tamaki, S. Navarro, C. Cheeseman, M. Soupioni, J. M. Pingarron, L. Valgimigli and V. R. Preedy, The Royal Society of Chemistry, 2012, p. 0 [Google Scholar]
  8. Vesa T. H. Marteau P. Korpela R. J. Am. Coll. Nutr. 2000;19:165S–175S. doi: 10.1080/07315724.2000.10718086. [DOI] [PubMed] [Google Scholar]
  9. Mattar R. de Campos Mazo D. F. Carrilho F. J. Clin. Exp. Gastroenterol. 2012;5:113–121. doi: 10.2147/CEG.S32368. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Grant J. D. Bezerra J. A. Thompson S. H. Lemen R. J. Koldovskk O. Udall J. N. Gastroenterology. 1989;97:895–899. doi: 10.1016/0016-5085(89)91494-7. [DOI] [PubMed] [Google Scholar]
  11. Kopec M. Beton-Mysur K. Surmacki J. Abramczyk H. Sci. Rep. 2024;14:16626. doi: 10.1038/s41598-024-67718-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Magistretti P. J. Allaman I. Neuron. 2015;86:883–901. doi: 10.1016/j.neuron.2015.03.035. [DOI] [PubMed] [Google Scholar]
  13. Bélanger M. Allaman I. Magistretti P. J. Cell Metab. 2011;14:724–738. doi: 10.1016/j.cmet.2011.08.016. [DOI] [PubMed] [Google Scholar]
  14. Mergenthaler P. Lindauer U. Dienel G. A. Meisel A. Trends Neurosci. 2013;36:587–597. doi: 10.1016/j.tins.2013.07.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Conte F. van Buuringen N. Voermans N. C. Lefeber D. J. Biochim. Biophys. Acta, Gen. Subj. 2021;1865:129898. doi: 10.1016/j.bbagen.2021.129898. [DOI] [PubMed] [Google Scholar]
  16. Shwe T. Pratchayasakul W. Chattipakorn N. Chattipakorn S. C. Exp. Gerontol. 2018;101:13–36. doi: 10.1016/j.exger.2017.10.029. [DOI] [PubMed] [Google Scholar]
  17. Robles L. Priefer R. Diagnostics. 2020;10:412. doi: 10.3390/diagnostics10060412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lisker R. Aguilar L. Zavala C. Am. J. Clin. Nutr. 1978;31:1499–1503. doi: 10.1093/ajcn/31.9.1499. [DOI] [PubMed] [Google Scholar]
  19. Han M. He W. Zhu W. Guo L. Cell death Discov. 2025;11:213. doi: 10.1038/s41420-025-02408-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Yildiz C. Medina I. Nutrients. 2024;16:631. doi: 10.3390/nu16050631. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Genc S. Kurnaz I. A. Ozilgen M. BMC Syst. Biol. 2011;5:162. doi: 10.1186/1752-0509-5-162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kim Y. Dube S. E. Park C. B. Korean J. Physiol. Pharmacol. 2025;29:1–8. doi: 10.4196/kjpp.24.388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Alberini C. M. Cruz E. Descalzi G. Bessières B. Gao V. Glia. 2018;66:1244–1262. doi: 10.1002/glia.23250. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Suzuki A. Stern S. A. Bozdagi O. Huntley G. W. Walker R. H. Magistretti P. J. Alberini C. M. Cell. 2011;144:810–823. doi: 10.1016/j.cell.2011.02.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kimelberg H. K. J. Cereb. Blood Flow Metab. 2004;24(11):1235–1239. doi: 10.1097/01.WCB.0000138668.10058.8C. [DOI] [PubMed] [Google Scholar]
  26. Warburg O. Source Sci. New Ser. 1956;123:309–314. doi: 10.1126/science.123.3191.309. [DOI] [PubMed] [Google Scholar]
  27. Longhitano L. Vicario N. Tibullo D. Giallongo C. Broggi G. Caltabiano R. Barbagallo G. M. V. Altieri R. Baghini M. Di Rosa M. Parenti R. Giordano A. Mione M. C. Li Volti G. Front. Oncol. 2022;12:871798. doi: 10.3389/fonc.2022.871798. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Falkowska A. Gutowska I. Goschorska M. Nowacki P. Chlubek D. Baranowska-Bosiacka I. Int. J. Mol. Sci. 2015;16:25959–25981. doi: 10.3390/ijms161125939. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Fan Y. Chen Z. Li J. Ding L. Oncol. Lett. 2025;30:492. doi: 10.3892/ol.2025.15238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Wang Z.-H. Peng W.-B. Zhang P. Yang X.-P. Zhou Q. EBioMedicine. 2021;73:103627. doi: 10.1016/j.ebiom.2021.103627. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Luo T. Liu L. Wang H. Wen S. Int. J. Oncol. 2025;67(58):1–13. doi: 10.3892/ijo.2025.5764. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Imiela A. Surmacki J. Abramczyk H. J. Mol. Struct. 2020;1217:128381. doi: 10.1016/j.molstruc.2020.128381. [DOI] [Google Scholar]
  33. Abramczyk H. Surmacki J. M. Brozek-Pluska B. J. Mol. Struct. 2022;1252:132134. doi: 10.1016/j.molstruc.2021.132134. [DOI] [Google Scholar]
  34. Ember K. Dallaire F. Plante A. Sheehy G. Guiot M.-C. Agarwal R. Yadav R. Douet A. Selb J. Tremblay J. P. Dupuis A. Marple E. Urmey K. Rizea C. Harb A. McCarthy L. Schupper A. Umphlett M. Tsankova N. Leblond F. Hadjipanayis C. Petrecca K. Sci. Rep. 2024;14:13309. doi: 10.1038/s41598-024-62543-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Wurm L. M. Fischer B. Neuschmelting V. Goldbrunner R. Croner R. S. Jagielski M. W. Laue D. Ertel W. Hacker M. C. Dybas J. Kahlert U. D. Front. Oncol. 2026;16:1748133. doi: 10.3389/fonc.2026.1748133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kopeć M. Beton K. Jarczewska K. Abramczyk H. Sci. Rep. 2022;12:18561. doi: 10.1038/s41598-022-21483-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Abramczyk H. Surmacki J. M. Brozek-Pluska B. Kopec M. Cancers. 2021;13(11):2599. doi: 10.3390/cancers13112599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Abramczyk H. Brozek-Pluska B. Kopec M. Sci. Rep. 2022;12:2120. doi: 10.1038/s41598-022-04803-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kopec M. Abramczyk H. Spectrochim. Acta, Part A Mol. Biomol. Spectrosc. 2022;268:120667. doi: 10.1016/j.saa.2021.120667. [DOI] [PubMed] [Google Scholar]
  40. Kopec M. Beton-Mysur K. Chem. Phys. Lipids. 2024;259:105375. doi: 10.1016/j.chemphyslip.2023.105375. [DOI] [PubMed] [Google Scholar]
  41. Kopeć M. Borek-Dorosz A. Jarczewska K. Barańska M. Abramczyk H. Analyst. 2024;149:2697–2708. doi: 10.1039/D4AN00015C. [DOI] [PubMed] [Google Scholar]
  42. Movasaghi Z. Rehman S. Rehman I. U. Appl. Spectrosc. Rev. 2007;42:493–541. doi: 10.1080/05704920701551530. [DOI] [Google Scholar]
  43. Shetty G. Kendall C. Shepherd N. Stone N. Barr H. Br. J. Cancer. 2006;94:1460–1464. doi: 10.1038/sj.bjc.6603102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Stone N. Kendall C. Smith J. Crow P. Barr H. Faraday Discuss. 2004;126:141–157. doi: 10.1039/B304992B. [DOI] [PubMed] [Google Scholar]
  45. Pezzotti G. J. Raman Spectrosc. 2021;52:2348–2443. doi: 10.1002/jrs.6204. [DOI] [Google Scholar]
  46. Augustyniak K. Lesniak M. Golan M. P. Latka H. Wojtan K. Zdanowski R. Kubiak J. Z. Malek K. Int. J. Mol. Sci. 2024;25:1–20. doi: 10.3390/ijms252212274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Lo W.-L. Lai J.-Y. Feinberg S. E. Izumi K. Kao S.-Y. Chang C.-S. Lin A. Chiang H. K. J. Raman Spectrosc. 2011;42:174–178. doi: 10.1002/jrs.2688. [DOI] [Google Scholar]
  48. Ma C. Zhang L. He T. Cao H. Ren X. Ma C. Yang J. Huang R. Pan G. Stem Cell Res. Ther. 2021;12:555. doi: 10.1186/s13287-021-02619-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Hu S. Spiro T. G. Morris I. K. Singh J. P. Smith K. M. J. Am. Chem. Soc. 1993;115:12446–12458. doi: 10.1021/ja00079a028. [DOI] [Google Scholar]
  50. Strekas T. C. Spiro T. G. Biochim. Biophys. Acta Protein Struct. 1972;278:188–192. doi: 10.1016/0005-2795(72)90121-3. [DOI] [PubMed] [Google Scholar]
  51. Notingher I. Green C. Dyer C. Perkins E. Hopkins N. Lindsay C. Hench L. L. J. R. Soc. Interface. 2004;1:79–90. doi: 10.1098/rsif.2004.0008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Koljenović S. Schut T. B. Vincent A. Kros J. M. Puppels G. J. Anal. Chem. 2005;77:7958–7965. doi: 10.1021/ac0512599. [DOI] [PubMed] [Google Scholar]
  53. De Gelder J. De Gussem K. Vandenabeele P. Moens L. J. Raman Spectrosc. 2007;38:1133–1147. doi: 10.1002/jrs.1734. [DOI] [Google Scholar]
  54. Notingher I. Sensors. 2007;7:1343–1358. doi: 10.3390/s7081343. [DOI] [Google Scholar]
  55. Abramczyk H. Brozek-Pluska B. Kopec M. Surmacki J. Błaszczyk M. Radek M. Cancers. 2021;13:960. doi: 10.3390/cancers13050960. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Abramczyk H. Kopeć M. Surmacki J. ACS Omega. 2025;10:27788–27805. doi: 10.1021/acsomega.5c00393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Surmacki J. M. Abramczyk H. Sci. Rep. 2023;13:1–11. doi: 10.1038/s41598-023-42301-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Alshehri B. Saudi Pharm. J. 2024;32:102194. doi: 10.1016/j.jsps.2024.102194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. LaManna J. C., in Oxygen Transport to Tissue XXIV, eds. J. F. Dunn and H. M. Swartz, Springer US, Boston, MA, 2003, pp. 535–546 [Google Scholar]
  60. Bay B.-H. Lee Y.-K. Tan B. K.-H. Ling E.-A. Neurosci. Lett. 1999;277:127–130. doi: 10.1016/S0304-3940(99)00872-1. [DOI] [PubMed] [Google Scholar]
  61. Escribano B. M. Muñoz-jurado A. Luque E. Conde C. Fei- M. Latorre M. Valdelvira M. E. Buendía P. Giraldo A. I. Ca-Cancer J. Clin. 1871:1–13. [Google Scholar]
  62. Li R. Yang Y. Wang H. Zhang T. Duan F. Wu K. Yang S. Xu K. Jiang X. Sun X. Cell. Mol. Neurobiol. 2023;43:2541–2555. doi: 10.1007/s10571-023-01335-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Magistretti P. J. Allaman I. Neuron. 2015;86:883–901. doi: 10.1016/j.neuron.2015.03.035. [DOI] [PubMed] [Google Scholar]
  64. Kann O. Kova R. Am. J. Physiol.: Cell Physiol. 2007;292(2):C641–C657. doi: 10.1152/ajpcell.00222.2006. [DOI] [PubMed] [Google Scholar]
  65. Lundgaard I. Li B. Xie L. Kang H. Sanggaard S. Haswell J. D. R. Sun W. Goldman S. Blekot S. Nielsen M. Takano T. Deane R. Nedergaard M. Nat. Commun. 2015;6:6807. doi: 10.1038/ncomms7807. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Li H. Guglielmetti C. Sei Y. J. Zilberter M. Le Page L. M. Shields L. Yang J. Nguyen K. Tiret B. Gao X. Bennett N. Lo I. Dayton T. L. Kampmann M. Huang Y. Rathmell J. C. Vander Heiden M. Chaumeil M. M. Nakamura K. Cell Rep. 2023;42:112335. doi: 10.1016/j.celrep.2023.112335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Xing S. Zhang L. Lin H. Mao Z. Bao K. Hao P. Pei Z. Li J. Hu Z. Int. J. Clin. Exp. Pathol. 2019;12:2034–2045. [PMC free article] [PubMed] [Google Scholar]
  68. Pantiya P. Thonusin C. Ongnok B. Chunchai T. Kongkaew A. Nawara W. Arunsak B. Chattipakorn N. Chattipakorn S. C. Toxicology. 2023;492:153553. doi: 10.1016/j.tox.2023.153553. [DOI] [PubMed] [Google Scholar]
  69. Wellen K. E. Hatzivassiliou G. Sachdeva U. M. V Bui T. Cross J. R. Thompson C. B. Science. 2009;324:1076–1080. doi: 10.1126/science.1164097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Holden H. M. Rayment I. Thoden J. B. J. Biol. Chem. 2003;278:43885–43888. doi: 10.1074/jbc.R300025200. [DOI] [PubMed] [Google Scholar]
  71. Slawson C. Copeland R. J. Hart G. W. Trends Biochem. Sci. 2010;35:547–555. doi: 10.1016/j.tibs.2010.04.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Back J. F. Oakenfull D. Smith M. B. Biochemistry. 1979;18(23):5191–5196. doi: 10.1021/bi00590a025. [DOI] [PubMed] [Google Scholar]
  73. Arakawa T. Timasheff S. N. Biochemistry. 1982;21(25):6545–6552. doi: 10.1021/bi00268a034. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data for this article are available at Repozytorium Otwartych Danych Badawczych at https://doi.org/10.34658/RDB.OIUID8.

Articles from RSC Advances are provided here courtesy of Royal Society of Chemistry

RESOURCES