A Brief Review of FT-IR Spectroscopy Studies of Sphingolipids in Human Cells

: In recent years, sphingolipids have attracted signiﬁcant attention due to their pivotal role in cellular functions and physiological diseases. A valuable tool for investigating the characteristics of sphingolipids can be represented via FT-IR spectroscopy, generally recognized as a very powerful technique that provides detailed biochemical information on the examined sample with the unique properties of sensitivity and accuracy. In the present paper, some fundamental aspects of sphingolipid components of human cells are summarized, and the most relevant articles devoted to the FT-IR spectroscopic studies of sphingolipids are revised. A short description of different FT-IR experimental approaches adopted for investigating sphingolipids is also given, with details about the most commonly used data analysis procedures. The present overview of FT-IR investigations, although not exhaustive, attests to the relevant role this vibrational technique has played in giving signiﬁcant insight into many aspects of this fascinating class of lipids.


Introduction
Lipids are a heterogeneous organic group of biomolecules that are insoluble in water due to the hydrocarbon nature of a large part of their structure. Lipids have several essential biological functions: they represent structural components of membranes, they serve as storage and transport for energy-rich molecules, they act as a protective coating on the surface of many organisms, and they also participate in recognition of specific characters. Some of them have significant biological activity, usually typified by vitamins and hormones. Although lipids are a distinct class of biomolecules, they are often bound with components of other biomolecules to form hybrid molecules, such as glycolipids containing carbohydrates, lipids, and lipoproteins, which contain lipids and proteins. In these biomolecules, their components' characteristic chemical and physical properties mix to perform specialized biological functions [1][2][3].
Several studies about SLs used methodologies based on NMR and vibrational techniques, such as Raman and Fourier-transform infrared (FT-IR) spectroscopies [22]. This last technique represents a unique tool thanks to its sensitivity in acquiring information on biological samples [27][28][29][30][31][32][33] and the effects of their interactions with external agents [34][35][36][37]. In addition, FT-IR spectroscopy has been also adopted for differentiating between benign and malignant tumors in the colon [38,39], breast tissues [40,41], and prostate [42,43]. The analysis of biofluids (saliva, serum, urine, and blood) represent another interesting field of application of this technique [44,45], even though particular attention should be paid to the contribution of water content that can interfere with the signal coming from the examined samples [46,47].
As far as concerns lipids, great attention has been devoted to the acquisition of infrared spectra of lipids. as is evident from [22,[48][49][50][51][52] and the references therein. The huge amount of work performed in this field has made available a certain number of databases dedicated to the FT-IR spectroscopy of lipids and similar compounds. In Table 1, the most cited ones are reported. Table 1. Online databases with information on lipids and lipidomics.

Source Url Notes
Lipid Bank-Japanese Conference on the Biochemistry of Lipids (JCBL) https://lipidbank.jp (accessed on 1 February 2023) LipidBank is a free database of natural lipids including fatty acids, glycerolipids, SLs, steroids, and various vitamins.The database contains more than 6000 unique molecular structures, their lipid names, and spectral and literature information.
NIST Chemistry WebBook https://webbook.nist.gov (accessed on 1 February 2023) The NIST Chemistry WebBook provides access to: thermochemical data; IR spectra, mass spectra, UV/Vis spectra, and gas chromatography data. It is possible to search for data on specific compounds based on name, chemical formula, CAS registry number, molecular weight, chemical structure, or selected ion energetics and spectral properties.
Spectral Database for Organic Compounds (SDBS) https://sdbs.db.aist.go.jp (accessed on 1 February 2023) SDBS is an integrated spectral database system for organic compounds, which includes 6 different types of spectra: an electron impact mass spectrum (EI-MS), a Fourier-transform infrared spectrum (FT-IR), a 1H nuclear magnetic resonance (NMR) spectrum, a 13C NMR spectrum, a laser Raman spectrum, and an electron spin resonance (ESR) spectrum.
Given the potentialities of FT-IR spectroscopy in lipid characterization and quantification, many researchers have adopted FT-IR for investigating particular aspects of SLs and their role in different biological frameworks.
In the present paper, we aim to revise the general characteristics of SLs in human cells and present the main findings of FT-IR spectroscopic studies obtained on these compounds. A short description of different FT-IR experimental approaches adopted for investigated sphingolipids is given, together with some information about the data analysis procedures commonly used in SL studies.

General Characteristics of Sphingolipids in Human Cells
SLs are a group of lipids that present a sphingosine (SP) backbone to which a fatty acid and a polar group are bonded. The SP (2-amino-4-trans-octadecene-1,3-diol) is an unsaturated amino alcohol made of three parts: a three-carbon chain containing two alcohol groups and one amino group linked to a long hydrocarbon chain ( Table 1). The SP is synthesized from serine and a long-chain fatty acyl CoA and then transformed into ceramides (Cers), phosphosphingolipids, glycosphingolipids, and other species [11]. Cers are the most basic form of SL and are in high concentrations in cell membranes. They are made up of fatty acid chain lengths ranging from 14 to 26 carbon atoms connected to SP via an amide linkage. When the Cers is esterified at the terminal OH group with a monosaccharide or an oligosaccharide, it is a glycolipid, respectively known as cerebroside or ganglioside. If the Cer is instead linked to phosphorylcholine (choline linked to a phosphate group), we have sphingomyelin (SM), the only membrane phospholipid free of glycerol. SM is the most abundant class of SLs in human cells; it is an essential component of the cell membranes, especially in the myelin sheath surrounding neuronal axons.
SLs are common in neural tissue and have an essential role in signal transmission and cell recognition; they also are known as bioactive signaling molecules in physiological reactions and stresses [4]. Some of them, such as Cer, SP, and S1P, in addition to their structural role in cell membranes, play a crucial role in regulating cell growth, differentiation, aging, and apoptosis [12][13][14]. Cers have been proposed to function as pro-apoptotic messengers because they increase the permeability of the mitochondrial outer membrane and contribute to release cytochrome c, which acts as a mediator of caspase activation downstream. SP functions as an extracellular ligand for G protein receptors and a second intracellular messenger to facilitate cell survival [17,18]. A phosphorylated version of SP, sphingosine 1-phosphate (S1P), is a modulator of calcium homeostasis and a regulator of cellular survival, proliferation, and differentiation.
Many investigations have also been carried out on investigating the role of SLs, particularly Cer, S1P, and SM, in regulating tumor growth and response to chemotherapy, radiotherapy, and immunotherapy [53][54][55][56][57]. Recent research has revealed that cancer alters the lipid metabolism in cells, including endogenous lipid synthesis, storage, exogenous biogenesis uptake, and cell membrane function, involving changes in SL structure and properties such as fluidity, stiffness, and membrane dynamics [58]. The increased rate of lipid synthesis in cancer tissues, their role in cancer cell migration and invasion, and their influence in tumor angiogenesis for tumor cell proliferation and metastasis have been considered in [59].
In the presence of many apoptotic stimuli, the SM pathway is activated, causing the accumulation of Cer and SP. Cer has been considered an apoptotic second messenger, while SP has been neglected as a mediator of apoptosis. However, recent investigations report significant hints about the role of the second messenger for sphingosine and the identification of potential factors influencing the mechanism of sphingosine-mediated cell death [60]. Lipid metabolism significantly affects primary tumors and shapes the tumor microenvironment to boost malignancy and dissemination [61]. Cer accumulation dysregulation is a common resistance mechanism to standard cancer treatments [62]. Because they cause cell death via Cer production, Cer synthase and sphingomyelinase are essential targets for γ-radiation and chemotherapy drugs [63][64][65][66]. One example of the oxidative stress induced by ionizing radiation is the transfer of the enzyme acid sphingomyelinase from lysosomes to the outer layer of the cell membrane, which causes SM hydrolysis and Cer formation. Ceramide-cholesterol fusion forms lipid microspheres in the plasma membrane and increases signal receptor clustering (death receptors such as FAS, TNF, CD40, TRAIL, or G protein-coupled receptors) [67,68].

Basic Principles of FT-IR Spectroscopy
FT-IR spectroscopy analyzes the spectral infrared radiation absorbed by a sample. The energy of the absorbed infrared radiation equates the energy needed to transition between the vibrational states of functional groups. The radiation intensity absorbed by a sample produces the infrared absorption spectrum, in which there are peaks or bands related to a particular mode of the vibrations that is characteristic of functional groups in the sample [27,37]. The main component of a modern FT-IR spectroscopic apparatus is a Michelson interferometer (see Figure 1 and [37] for further details), which substitutes the monochromator that was present in the old infrared spectrometers. FT-IR spectrometers enable the collection of absorption spectra quickly and precisely by obtaining qualitative and quantitative information on the structure of the compounds analyzed using modern software algorithms. the sample [27,37]. The main component of a modern FT-IR spectroscopic apparatus is a Michelson interferometer (see Figure 1 and [37] for further details), which substitutes the monochromator that was present in the old infrared spectrometers. FT-IR spectrometers enable the collection of absorption spectra quickly and precisely by obtaining qualitative and quantitative information on the structure of the compounds analyzed using modern software algorithms. The sample under investigation is positioned in one of the two arms of the interferometer, the frequencies that are related to the excitation of vibrational states of the functional groups of the molecules of the sample are attenuated, and the obtained interferogram is Fourier transformed into an absorption spectrum in terms of the wavenumber, usually given in cm −1 . The sample under investigation is positioned in one of the two arms of the interferometer, the frequencies that are related to the excitation of vibrational states of the functional groups of the molecules of the sample are attenuated, and the obtained interferogram is Fourier transformed into an absorption spectrum in terms of the wavenumber, usually given in cm −1 .
FT-IR spectrometers offer many advantages compared to traditional infrared ones [28][29][30]. As previously said, the most relevant benefit is the ability to collect spectra in a short time. In fact, the detector can simultaneously analyze and record all frequencies and wavelengths. FT-IR spectroscopy also offers high precision and accuracy in the determination of wavenumbers thanks to the use of the monochromatic radiation of a laser source as an internal standard. Unwanted sample heating effects are excluded since the infrared source is sufficiently far from the sample. Then FT-IR spectra can be obtained from samples at controlled temperatures using proper accessories [69]. The use of suitable cells also allows the acquisition of spectra from samples under different pressures [70]. Moreover, an FT-IR spectrometer can be equipped with a microscope to allow the analysis of samples or parts of samples at the microscopic level [71].

Experimental Aspects of FT-IR Spectroscopy
There are three main infrared spectroscopy sampling modes: transmission, reflectance, and attenuated total reflection (ATR), each with advantages and disadvantages [31,72]. For lipid analysis, the most largely used approaches are transmission and ATR modes. For transmission measurements, the samples are placed between two IR transparent windows. Generally, CaF 2 , BaF 2 , and KBr windows are used together with a proper sample holder in which the Teflon spacer can obtain the appropriate thickness, typically around 6-10 µm [73].
As an alternative, the samples can be included within a KBr matrix and pressed to form a pellet that can be subsequently analyzed after being dried [74].
The ATR approach is based on the presence of an evanescent wave at the reflecting interface between the sample and the crystal [75]. When a radiation beam penetrates inside a crystal, and the angle of incidence at the interface between the sample and crystal is greater than the critical angle, which is given by the refractive indices of the two surfaces, the beam penetrates a fraction of a wavelength beyond the reflecting surface, and when it comes into close contact with a material that selectively absorbs radiation, the beam loses energy at the wavelength where the material absorbs. The spectrometer measures and plots the attenuated radiation as a function of the wavelength, yielding the sample's absorption spectral characteristics. ATR employs crystal materials, such as diamond, ZnSe, and ZnS, in contact with the sample under investigation. A typical configuration for ATR equipment is reported in Figure 2. A pressure plate is needed to ensure the connection between the samples and the crystal. Except for diamond objectives, excessive pressure can harm some crystals [76,77].
There are three main infrared spectroscopy sampling modes: transmission, reflectance, and attenuated total reflection (ATR), each with advantages and disadvantages [31,72]. For lipid analysis, the most largely used approaches are transmission and ATR modes. For transmission measurements, the samples are placed between two IR transparent windows. Generally, CaF2, BaF2, and KBr windows are used together with a proper sample holder in which the Teflon spacer can obtain the appropriate thickness, typically around 6-10 µm [73]. As an alternative, the samples can be included within a KBr matrix and pressed to form a pellet that can be subsequently analyzed after being dried [74].
The ATR approach is based on the presence of an evanescent wave at the reflecting interface between the sample and the crystal [75]. When a radiation beam penetrates inside a crystal, and the angle of incidence at the interface between the sample and crystal is greater than the critical angle, which is given by the refractive indices of the two surfaces, the beam penetrates a fraction of a wavelength beyond the reflecting surface, and when it comes into close contact with a material that selectively absorbs radiation, the beam loses energy at the wavelength where the material absorbs. The spectrometer measures and plots the attenuated radiation as a function of the wavelength, yielding the sample's absorption spectral characteristics. ATR employs crystal materials, such as diamond, ZnSe, and ZnS, in contact with the sample under investigation. A typical configuration for ATR equipment is reported in Figure 2. A pressure plate is needed to ensure the connection between the samples and the crystal. Except for diamond objectives, excessive pressure can harm some crystals [76,77].  Fringeli and Gunthard, in their pivotal paper about infrared spectroscopy, demonstrated that the ATR approach offers many advantages in studying lipids [79]. For this acquisition geometry, a drop of a few microliters of lipid extract is deposited on the crystal and evaporated under a gaseous N 2 flow to form a thin film. This way to acquire measurements is fast and able to give a strong signal with a small sample quantity. In addition, it enables for obtaining information about the orientation of various parts of the molecule under investigation in an oriented system [75,80].

Data Analysis Procedures
Data preprocessing has been identified as essential in biomedical spectroscopy's effective data analysis workflow. It includes background subtraction, scattering effect removal, and normalization procedures. Regarding background subtraction, a background spectrum must be acquired without a sample to consider the effects due to the instrument and the environment. As said before, particular attention should be given to water vapor, which is responsible for the characteristic bands around 3500 cm −1 and 1630 cm −1 , and carbon dioxide, which is responsible for the bands at 2350 cm −1 and 667 cm −1 [46,47,81]. Another critical step in the preprocessing procedures of FT-IR spectra is removing scattering effects that can occur during the interaction between infrared radiation and samples. When the scattering causes the loss of a significant portion of the light, data interpretation does become problematic because it is arduous to determine whether the signal at a given wavelength decreases due to scattering or chemical absorption. When the wavelength of the electromagnetic wave has a value similar to the size of the scatterer, the amount of radiation power lost due to scattering is very large. The scattering is known as Mie-type scattering when the scatterer has spherical symmetry. An intense Mie scattering alters the spectra of single cells and tissues in infrared spectroscopy. These artifacts can be removed using good algorithms, such as the ones proposed in [82,83]. Normalization procedures represent a class of data treatments for scaling spectra in a similar range. This method is commonly used to reduce the effects of different optical paths on data or compensate for source intensity changes while indicating one of the possible causes [84]. After data preprocessing, different approaches can be used to analyze the FT-IR spectra in detail. Procedures for spectra deconvolution are primarily used and allow for the individuation of the main vibrational modes that are present in the spectrum. Typically, Gaussian and mixed Gaussian-Lorentzian functions are adopted in the case of FT-IR spectra. The components' spectral positions, intensities, and widths are evaluated via a fitting procedure [34][35][36].
In some cases, the second derivative of the spectrum is also performed to preliminarily evaluate the local minima corresponding to the vibrational modes [85]. In addition, univariate and multivariate data analysis procedures are primarily used for FT-IR spectra analysis [35,86,87]. In the investigation of FT-IR lipids, the evaluation of the ratios between the intensity or the band area values of chosen spectral components has been largely used for investigating the effects of interactions with external agents [36,[88][89][90][91].

FT-IR Characterization of SLs
Before discussing the principal results reported in the literature about FT-IR studies on SLs, it is worthwhile to mention the fundamental paper of Fringeli and Gunthard [79] that provides the basis for all subsequent studies about lipids components in biological samples. Some other authors give relevant contributions in this field [75,80,[92][93][94]. In Table 2, an almost exhaustive list of the peaks present in lipid samples is reported together with their assignments. Table 2. Main sphingolipids present in human cells and their structures [21].

Sphingolipid Compound Structure
Ceramide (Cer) The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by The spectra of different membrane lipids include two different regions; the high wavenumber part of the spectrum (3100-2800 cm −1 ) that only contains C-H stretching vibrations and originates mainly from hydrocarbon chains and the low wavenumber region of the spectrum (below 1800 cm −1 ) that is mainly associated to the polar headgroups of lipids. In Figure 3 this region is reported for different membrane lipids.
The C-H stretching bands in the lipid carbon chain region are related to different vibrational modes (see Table 3 for details). This region is interesting to study because there is no overlap with other vibrational modes; even in complex systems such as the biological samples, it is independent of the polar headgroups, and can give information about the structure of the hydrocarbon chain. In fact, lipid phase transition can be investigated by monitoring the features (position, intensity, and width) of the bands present in this region [95]. The range associated to the lipid polar headgroup allows for the identification of the different lipids, as evident in Figure 3, in which FT-IR spectra of commercial samples of different cell membrane lipids are reported. Among these lipid spectra, there is the one of SM. The most relevant contributions in this region are related to the C=O stretching mode positioned around 1740 cm −1 and the phosphate group near 1240 and 1080 cm −1 for the antisymmetric and symmetric modes, respectively. It is worth noting that below 1800 cm −1 each lipid molecule has a distinctive FT-IR spectrum. This characteristic has been exploited to identify the various components in lipid mixtures [96]. different lipids, as evident in Figure 3, in which FT-IR spectra of commercial samples of different cell membrane lipids are reported. Among these lipid spectra, there is the one of SM. The most relevant contributions in this region are related to the C=O stretching mode positioned around 1740 cm −1 and the phosphate group near 1240 and 1080 cm −1 for the antisymmetric and symmetric modes, respectively. It is worth noting that below 1800 cm −1 each lipid molecule has a distinctive FT-IR spectrum. This characteristic has been exploited to identify the various components in lipid mixtures [96]. FT-IR spectroscopy has been largely used for investigating several aspects of SL characteristics and functions; it would be very hard work to try to summarize all the number of contributions that can be found in the literature. For this reason, we prefer to present a brief synopsis of some papers concerning SM, Cer, SP, and S1P that can contribute to evidence of the pivotal role of FT-IR spectroscopy in this field. In Table 4, some significant papers on sphingomyelin are briefly summarized in chronological order. As far as concerns the class of investigated samples, commercial ones have been used in many cases. For some studies, lipids extracted from cells and tissues have been adopted. In these cases, Bligh and Dyer as well as Folch methods have been used [97,98]. Further details on the lipid extraction methods are reported in Appendix A. Regarding the FT-IR geometry adopted for spectra collection, it is possible to notice that both transmission and ATR approaches have been used. For transmission measurements, in a large number of cases, CaF 2 windows have been used even though KBr pellets have shown also to be useful.
The paper of Villalain et al., cited in Table 4, is particularly interesting due to the joint use of differential scanning calorimetry and FT-IR spectroscopy to investigate the molecular interactions between PC and SM. The authors observed that no significant change in the wavenumber position for the antisymmetric and symmetric CH 2 stretching is present when SM is added to PC at temperatures above or below the phase transition and then they demonstrated that the gauche ratio to all-trans is not affected by SM and PC interaction. The authors also examined the characteristics of the C=O stretching band of PC located at 1735 cm −1 , evidencing that the phosphate group of SM takes part in hydrogen bonding between the molecules of SM and possibly PC. Particular attention was also given to the Amide I contribution at 1635 cm −1 and to bands related to the phosphate group and located at 1220 and 1080 cm −1 [99]. To examine the spatial distribution of molecular changes associated with C6 glioma progression.
The concentrations of SM, nucleic acids, PS, and glucocerebroside are significantly affected during C6 glioma development. [103] Lipids extracted from brain tissues using Folch and Bligh and Dyer methods.

Transmission KBr pellets
To analyze the lipid extracts from the brain to identify their composition.
Lipid content can be evaluated via FT-IR spectroscopy, which may improve the differential diagnosis of brain cancers. [104,105] Commercial samples and lipids extracted from PC-3 cells using Bligh and Dyer method.

ATR
To analyze the changes in the lipidome of prostate cancer PC-3 cells after exposure to sub-lethal ouabain levels.
Lipid alterations induced by ouabain can be identified by variations in the ester/choline/phosphate ratios in FT-IR spectra. [96] Commercial samples and lipids extracted from PC-3 cells using Bligh and Dyer method.

Micro-ATR
To develop PLS models based on FT-IR spectra to determine the changes in the amounts of different lipids in extracts from PC-3 cells treated with four antitumor drugs.
After treatments with anticancer drugs, the spectral region of the polar headgroups of samples did not show any noticeable alterations. However, the developed PLS models can be used for high-throughput measurements. [106] Commercial samples ATR To investigate the changes occurring in detergent-resistant membranes (DRM) extracted from human breast cancer cells when treated with the omega 3 fatty acid docosahexaenoic acid.
FT-IR spectroscopy and multivariate analysis enables to monitor the changes in the composition of DRMs.
This approach can be useful for label-free characterization of lipid component in cells.
[107] Commercial samples Transmission geometry using CaF 2 windows To examine the interaction profile of carboplatin at varying concentrations with SM multilamellar vesicles.
Carboplatin affects the phase transition, enthalpy, the cooperativity parameter, the phase transition temperature, the lipid order, the lipid fluidity, and the hydrogen state of specific groups in hydrophilic parts of the examined samples.
Dreissig et al. investigated the different SLs and phospholipids that are present in lipids extracted from the porcine brain. They presented the FT-IR spectra of some SLs, phospholipids, and neutral lipids (see Figure 4). If SM spectrum is considered in detail, it is possible to recognize the contribution of different vibration modes reported in Table 3 and in Table 2 of [103]. In the high wavenumber region related to hydrocarbon chains, peaks related to CH 2 asymmetric and symmetric stretching are positioned at 2924 and 2852 cm −1 , respectively. In the spectral region related to the lipid polar headgroup, contributions attributed to C=O stretching (1647 cm −1 ), N-H bending (1545 cm −1 ), CH 2 bending (1466 cm −1 ), CH 3 symmetric bending (1378 cm −1 ), PO -2 asymmetric stretching (1240 cm −1 ), PO -2 symmetric stretching (1090 cm −1 ), and C-O-C stretching (1055 cm −1 ) can be noticed. In [103], the authors examined the difference in band position and intensities in the spectra of the different commercial samples. They also prepared lipid mixtures to train and validate a quantification model for determining the composition of brain lipid extracts. The results of the PLS regression of FT-IR spectra were successfully related to the lipid quantification obtained by using TLC. Even though the FT-IR spectroscopic approach presented in this paper cannot be generally used for the quantitative analysis of all lipids, it might contribute to the diagnosis of brain tumors by evaluating the changes occurring in the lipid composition of tumor cells [103].  The papers of Gasper et al. [104,105] represent a typical example of the use of FT-IR spectroscopy for studying the effects of drugs on SLs, a topic largely investigated by using this vibrational technique. In the present case, the authors aim to examine the number of changes in lipids in PC-3 cells exposed to sub-lethal levels of ouabain, a well-known cardiotonic steroid that demonstrates an anti-cancer activity in vitro and in vivo that indicates the possibility of adopting these compounds as chemotherapeutic agents in The papers of Gasper et al. [104,105] represent a typical example of the use of FT-IR spectroscopy for studying the effects of drugs on SLs, a topic largely investigated by using this vibrational technique. In the present case, the authors aim to examine the number of changes in lipids in PC-3 cells exposed to sub-lethal levels of ouabain, a wellknown cardiotonic steroid that demonstrates an anti-cancer activity in vitro and in vivo that indicates the possibility of adopting these compounds as chemotherapeutic agents in oncology. The authors examined the difference spectra obtained between the average spectra for lipid extracts related to different experimental conditions. In Figure 4 of [104], three curves are reported: the first is related to the changes occurring in the first 6 h after the treatment, the second is related to the changes between 24 and 6 h of treatment and the last one concerns the changes between 36 and 24 h of treatment. The most relevant changes for lipids are present between 6 and 24 h. Principal component analysis (PCA) was also performed for evaluating the differences among the FT-IR spectra related to different experimental conditions. According to the findings of [104,105], FT-IR spectroscopy not only detects cellular changes caused by ouabain but also qualitatively assesses the evolution of these alterations during therapy. Ouabain's overall biological effect on PC-3 cell lipids increased SM and decreased PC [104,105].
A particularly interesting aspect of FT-IR spectroscopy is related to the possibility of its use for the quantitative determination of a single component in a mixture. This possibility has been evidenced by the paper of Derenne et al. [96], which developed PLS algorithms for quantifying lipids in complex mixtures of lipids. The lipid composition obtained via FT-IR spectroscopy was first validated by using HPLC and used for determining the lipid composition of lipids extracted from cells exposed to four different drugs. These treatments did not cause significant changes in the polar headgroups' spectral region. However, the authors suggested that their PLS models appear to be reliable and can be helpful for routine analysis [96].

(b) Ceramide (Cer)
In Table 5, some papers related to the study of the properties and functions of ceramide (Cer) have been reported. Additionally, for Cer, commercial samples have been used in the largest number of cases, and measurements have been performed using both ATR and transmission geometries.
Moore et al. [108] investigated the conformational order and phase behavior of hydrated Cers to provide the first complete investigation of the intermolecular and intramolecular chain and headgroup interactions in hydrated non-hydroxy fatty acid (NFA) and hydroxy fatty acid (HFA) Cers. In particular, the authors examined the temperature dependence of the methylene stretching, scissoring, and rocking mode frequencies. This study indicates the occurrence of two relevant phase transitions, one around 60 • C and the other around 80 • C. The behavior of the Amide I and II modes shows different contributions from NFA and HFA Cers. The results reported in this paper were also of relevance to the domain mosaic model of the stratum corneum lipid barrier that was proposed in the seventies of the last century. A thorough investigation of headgroup and intermolecular chain interactions in hydrated non-hydroxy fatty acid (NFA) and hydroxy fatty acid (HFA). Cers is obtained via FT-IR spectroscopy. In addition, information regarding the structure and conformational order of the NFA and HFA Cer chain subcells is derived from the temperature dependency of the methylene stretching, scissoring, and rocking mode frequencies.
[ To review some methods for analyzing skin lipids and SL structures.
FT-IR spectroscopy is used: to evaluate the state of the order of hydrocarbon chains in terms of population of trans and gauche conformers, the packing behavior, and phase transitions; to elucidate the polymorphism and the hydrogen-bonding network of the headgroups; to characterize the hydration properties of phytosphingosine and SP-Cer.  The study describes the peculiar properties of the SM and Cer sphingosine-based headgroups that are not found in glycerolipids.
Another paper addressing the analysis of the conformational order of Cer by using FT-IR spectroscopy is reported in [114]. In addition, these authors investigated the dependence on temperature of the conformational stability of Cers. They studied different Cers and mixtures and polar-nonpolar lipid interactions. Their study evidenced the impact of the polar headgroup on the conformation of the hydrocarbon chains when the temperature increases, with the role of the endogenous molecules also being more prominent.
More recently, de Arada et al. [118] used FT-IR spectroscopy to examine the interactions of polar headgroups from Cers and SM. These authors studied four regions of the FT-IR spectra of these compounds: C-H stretching and CH 2 scissoring vibrations, the Amide I region, and the phosphate vibration range. The study of the two latter areas was the novelty of the de Arata et al. paper. The temperature dependence of hydrated samples of pure SM and SM-Cer mixtures was examined and compared with the results of differential scanning calorimetry measurements. The data from the hydrocarbon chain region show a transition in agreement with previous observations. The results related to the Amide I region are more interesting since they evidence that SM and Cer carbonyl groups strongly interact, probably through H bonds. Furthermore, the results pertaining to the phosphate group suggest a relevant role of H bonds in the interaction between SM and Cer. The results of [118] evidence that SM and Cer can have an interaction through their polar headgroups that is differently from what occurs to other lipids.
(c) Sphingosine (SP) and sphingosine 1-phosphate (S1P) In Table 6, we summarized the main aspects of some papers devoted to applying FT-IR spectroscopy for studying SP and S1P. Commercial samples and lipids extracted from human cells, biofluids, and natural substances are examined in this case. The FT-IR investigations have been made using different acquisition geometries (transmission with CaF 2 windows and KBr pellets and ATR). A certain number of selected papers can be evidenced by the joint use of FT-IR spectroscopy with mass spectrometry for precisely identifying the SL involved in the examined process.
The contribution reported in [119] is related to the joint use of differential scanning thermometry and FT-IR spectroscopy for investigating the effect of SP and stearylamine on the interaction of SP with calcium. Additionally, in this case, spectroscopic observations confirm the phase transition observed by using differential scanning thermometry. The inspection of the ester C=O stretching mode appearing as a broad band at 1734 cm −1 suggests that the amino bases do not introduce hydrogen bonding with the C=O group and do not modify its hydration state. The same evidence is obtained by considering the band related to the phosphate group located at 1220 cm −1 . The degree of dehydration of the phosphate group can also be quantitatively determined by using a partial least-squares multivariate statistical analysis. A further study from the same research group [120] is devoted to the use of FT-IR for a quantitative analysis of the dehydration process of the phosphatidylserine phosphate group in the presence of Ca 2+ caused by various molecules, such as diacylglycerol, SP, and stearylarnine, by adopting a partial least-squares statistical method. FT-IR can also be used for estimating the apparent pKa of lipid carboxyl groups.
The absorbance signals given by the protonated and the unprotonated forms of the specific group under investigation can be determined. In so doing, it is possible to evidence that diacylglycerol increases the dehydration of the phosphate group due to Ca 2+ while the amino-bases SP and stearylamine avoid the dehydration of the phosphate group. The paper of Derenne et al. [96] that has been cited in the previous table also gives interesting results for SP and S1P.

(d) FT-IR Lipidomic studies involving sphingolipids
It is worth noting that, recently, great attention has been devoted to two aspects of FT-IR spectroscopy technology to broaden its field of application further. The former of these aspects is related to the development of experimental approaches allowing the characterization of a considerable number of samples by developing a high-throughput approach [127][128][129]. The second concerns the "omics" aspect. Usually, this term is used for indicating the analysis of genes (genomics), proteins (proteomics), and metabolites (metabolomics), but also the analysis of all lipids and the molecules with which they interact, and their function within biological systems assumes a particular relevance since lipidomics represent one of the most critical aspects of modern analytical biochemistry. The most general techniques used in lipidomics are the techniques of TLC, HPLC, GC, and CE mentioned in the Introduction of the present paper; however, also in this framework, FT-IR spectroscopy is increasingly appreciated [22,130,131]. Considering sphingolipids, Ramos-Garcia et al. [25] showed that FT-IR analysis could provide a qualitative and quantitative biochemical characterization of isolated exosomes, allowing for a fast and direct quantification of the total lipid content and evidencing sphingolipid contribution. Another contribution in this field has been given by Guleken et al. that individuated changes in sphingolipid metabolism in the blood serum of endometriosis-affected patients [24,26].

Conclusions
The relevance of SLs in an extremely large number of physiological processes and pathologies is nowadays clearly recognized, and all the experimental techniques that can contribute to shedding light on the mechanisms in which SLs are involved are under intense investigation. The large number of papers that are now present in the literature and the representative results discussed here evidence that FTI-R spectroscopy represents an excellent tool for investigating SLs simply and directly. In fact, the selected articles testify to the variety of contributions offered by FT-IR spectroscopy, from studies on the conformational order of different ceramides in stratum corneum to SL quantification in cancer cells, drug effects, and new biomarkers for tumor diagnosis. This review does not present an exhaustive collection of all the results made available via FT-IR spectroscopy. Still, it proves that it can be an essential research tool for studying SL roles in very different experimental conditions.

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

Appendix A. Sample Preparation
The extraction of lipids from biological matrices is a crucial step for preparing samples for FT-IR spectroscopy. In particular, it is necessary to reach two goals: to have a high extraction efficiency and to completely remove potentially interfering substances such as proteins, carbohydrates, and other metabolites. Depending on the nature and physical state of the sample (vegetable or animal, tissue or fluid), optimizing the procedure begins with the choosing of the appropriate solvent. Non-polar organic solvents solubilize neutral lipids, while polar ones are better for polar lipids, such as phospholipids and membrane lipoproteins. Therefore, the best choice may be to use solvents of different polarity [132,133]. To facilitate separation after extraction, other factors, such as volatility and the ability to form a two-phase aqueous system, must be considered, while the absence of toxic or reactive impurities prevents unwanted reactions with the lipids to be extracted [134]. Finally, the effects of solvents on health and the impact on the environment are also worthy of attention.
As regards the protocols for lipid extraction, the papers cited in the present review essentially adopted two liquid-liquid extraction (LLE) protocols based on a ternary mixture of chloroform, methanol, and water at different ratios by using the Folch method [135] and the Bligh and Dyer method [97].
Developed in the 1950s to extract fat from animal tissue, they are still used as reference extraction processes for different biological matrices, such as plasma, urine, seminal plasma, liver, saliva, and human milk (Saini et al., 2021). The Folch method is usually used to extract lipids from solid tissue, while for biological fluids, the Bligh and Dyer method is preferred [134].
The Folch and Bligh and Dyer protocols differ in the chloroform/methanol/water ratio: in the Folch method it is 2:1:0.75, while in the Bligh and Dyer method it is 1:1:0.9%. In addition, in the Folch method the added aqueous fraction contains salts. When comparing the total volume of solvents, it is twenty times the sample in Folch and only four times in the Bligh and Dyer method [136].
In any case, these protocols are equally efficient in extracting the total lipids. The efficiency relies on the amphipathic properties of lipids to obtain a differential partition between an aqueous phase and an immiscible organic phase at a given temperature, pH, and ionic strength. Hence, the lipids are collected in the lower organic chloroform/methanol phase while the upper aqueous phase contains more hydrophilic compounds and salts. The negative aspects of these protocols are essentially two: the toxicity of chloroform, a known carcinogen, and the extract contamination during the sampling of the lower organic phase. A pipette must cross the upper and middle aqueous phases, containing precipitate and insoluble material, and an inevitable contamination of lipid extract occurs.
In recent years, research activities have focused on the development of new extraction protocols characterized by simplicity, economy, automation, and the use of non-toxic solvents. Alternative methods, based on a two-phase mixture, such as tert-butyl methyl ether (MTBE)/methanol [137], or a single phase, such as 1-butanol/methanol [138], have gained popularity. In both methods the lipids are extracted in the upper phase, avoiding unwanted contamination, and saving time compared to the Folch and Bligh and Dyer methods. In addition, they are safer and more environmentally friendly as they do not use toxic chloroform. A single-phase method may even be more cost-effective since, by not involving solvent separation into a two-phase form, it potentially yields better lipid recoveries and automation opportunities.