Lipid Droplet Phase Transition in Freezing Cat Embryos and Oocytes Probed by Raman Spectroscopy

Abstract

Embryo and oocyte cryopreservation is a widely used technology for cryopreservation of genetic resources. One limitation of cryopreservation is the low tolerance to freezing observed for oocytes and embryos rich in lipid droplets. We apply Raman spectroscopy to investigate freezing of lipid droplets inside cumulus-oocyte complexes, mature oocytes, and early embryos of a domestic cat. Raman spectroscopy allows one to characterize the degree of lipid unsaturation, the lipid phase transition from the liquid-like disordered to solid-like ordered state, and the triglyceride polymorphic state. For all cells examined, the average degree of lipid unsaturation is estimated as ∼1.3 (with ±20% deviation) double bonds per acyl chain. The onset of the lipid phase transition occurs in a temperature range from −10 to +4°C and does not depend on the cell type. Lipid droplets in cumulus-oocyte complexes are found to undergo abrupt lipid crystallization shifted in temperature from the ordering of the lipid conformational state. In the case of mature oocytes and early embryos obtained in vitro, the lipid crystallization is broadened. In the frozen state, lipid droplets inside cumulus-oocyte complexes have a higher content of triglyceride polymorphic β and β′ phases than estimated for mature oocytes and early embryos. For the first time, to our knowledge, the temperature evolution of the phase state of lipid droplets is examined. Raman spectroscopy is proved to be a promising tool for in situ monitoring of the lipid phase state in a single embryo/oocyte during its freezing.

Introduction

Cryopreservation of preimplantation embryos and gametes is a primary tool used to back up and exchange laboratory animal strains (1, 2) and farm animal breeds (3, 4). More recently, some successful examples of using these technologies within the Genome Resource Bank (GRB) concept for endangered animal species were reported (5, 6, 7). Despite significant development of cryopreservation techniques over the last decades, these approaches can be applied to few mammalian species (8, 9). The problems of oocyte and embryo cryopreservation concern virtually all the representatives of the Carnivora order (5), many of them being endangered species. To expand the application of the GRB concept for mammalian species preservation, further investigations of factors affecting cell survival during freezing/cryopreservation procedures are needed.

Critical factors of cell injuries during freezing are the mechanical damage by ice and the toxicity of cryoprotectant solutions (10). Freezing protocols avoiding these effects make possible the cryopreservation of oocytes and embryos of some species. However, existing protocols face difficulties in successful cryopreservation of embryos and oocytes with high lipid content (11). This problem refers to a large list of species, which include cattle (12), pigs (13), sheep (14), and a variety of Carnivora representatives (15), including the Felidae family. The high lipid content indicates that there are other damage mechanisms associated with the properties of lipid structures in cells. Although the mechanisms not fully understood, thermotropic lipid phase transitions (LPTs) are believed to be responsible for this damage.

Within a cell, lipids are distributed in membranes and cytoplasmic lipid droplets (LDs). Great attention is paid to the investigation of the low-temperature damage associated with membranes (16, 17, 18, 19, 20). Cell cooling leads to the LPT that changes the properties of membranes: the functioning of membrane proteins (21), spatial lipid distribution (16), and membrane transport (22, 23). It was reported that ice formation could influence the phase state of membrane lipids via dehydration of the membrane surface (18, 19, 20). The changes in membrane properties are used to explain the cell injuries at “chilling” temperatures (0–10°C) in semen (17) and oocytes (24, 25, 26, 27). However, problems with cryopreservation are inherent in oocytes and embryos rich in LDs. The importance of LDs is supported by the observation that removal of LDs increases the survival ratio after cryopreservation for pig embryos (13). Damaging mechanisms related to LDs are unknown, but the knowledge about the details of the LPT in real cells is necessary for our understanding of LD-related injuries. Therefore, investigation of the LPT and description of the lipid phase states of LDs appears to be an actual task.

To describe the lipid phase state, two different order parameters are used. The first one is the degree of acyl chain ordering, which can refer to the number of gauche conformations in the hydrocarbon chain (28). The second parameter is related to the translational order in lipid molecule arrangement. In a single-component lipid system, these order parameters vary in a consistent manner. In the case of lipid mixtures, they might be independent, and intermediate phase states, such as the liquid-ordered state, become possible. In the liquid-ordered state, the lipid acyl chains are in an ordered conformational state, but the molecules are arranged randomly, as in the liquid state (29, 30). Biological lipid structures consist of multicomponent lipid mixtures. As a result, the description of the phase transition becomes complicated. Simultaneous coexistence of several phases can be observed (31), and the phase transition may occur via intermediate states (32).

Nowadays, the injury mechanisms associated with the lipid phase state remain obscure, not least because of the lack of experimental data on the lipid state in freezing cells. Arav et al. applied infrared spectroscopy to investigate LPT temperatures in bovine, ovine, and human oocytes (24, 25, 26, 27). It was found that the LPT for these mammalian species occurs at temperatures above 0°C and depends on lipid composition. Cells with the phase transition at lower temperatures are assumed to be more tolerant to chilling.

Raman spectroscopy is a perspective approach for contactless in situ studies of frozen cells with high spatial resolution. In the last decade, this approach was used to investigate the distributions of ice, cryoprotectant, and eutectic crystallization products in freezing samples (33, 34, 35, 36). The capability of the Raman approach to study lipid phase state is proven by studies of frozen cells (2, 37) and model lipid systems (38, 39). Raman spectroscopy was applied in investigations of lipid content in LDs of algae (40, 41). Rinia et al. (42) demonstrated the applicability of the Raman approaches to examine lipid ordering and heterogeneity of LDs by using coherent anti-Stokes Raman spectroscopy. However, to our knowledge, the thermotropic LPT in LDs of living cells has not been studied yet.

This study was aimed at identifying the phase states and transitions of triglycerides in LDs of lipid-rich domestic cat embryos and oocytes during freezing. We investigated stretching CH, C=O, and CC Raman bands in the spectra measured from LDs in a wide temperature range to extract the degree of lipid unsaturation, the LPT parameters, and polymorphic phase content in a frozen state. The LPTs in cumulus oocyte complexes, mature oocytes, and early embryos were compared.

Materials and Methods

Sample preparation

Ovaries and epididymises from domestic cats were obtained after routine ovariohysterectomy and orchiectomy from local veterinary clinics and were transported to the laboratory within 3–4 hr at +4°С in HEPES buffered tissue culture medium (TCM)-199 (Thermo Fisher Scientific, Waltham, MA) supplemented with streptomycin (100 μg/mL) and penicillin (100 international unit (IU)/mL).

The ovaries were minced, and the cumulus-oocyte complexes (COCs) were collected into TCM-199 (Thermo Fisher Scientific) supplemented with 5.67 mM HEPES, 25 mM NaHCO₃, 2.2 mM pyruvate, 2.2 mM sodium lactate, 100 μg/mL streptomycin, 100 IU/mL penicillin, and 3 mg/mL bovine serum albumin at 38°C. Oocytes with uniformly dark ooplasm surrounded by several layers of cumulus cells were rinsed three times in HEPES-buffered TCM-199 and cultured in 50 μL of TCM-199 (Thermo Fisher Scientific) containing 5 IU/mL human chorionic gonadotropin (Chorulon; Intervet International B.V., Boxmeer, the Netherlands), 1 IU/mL equine chorionic gonadotropin (Follimag; Mosagrogen, Moscow, Russia), and supplemented with 2.2 mM sodium lactate, 2.2 mM pyruvate, 25 mM NaHCO₃, 100 μg/mL streptomycin, 100 IU/mL penicillin, and 3 mg/mL bovine serum albumin under mineral oil at 38°C, 5 v/v % CO₂ in 24 hr until the metaphase II stage was reached (in vitro maturation).

For in vitro fertilization (IVF), metaphase II oocytes were rinsed three times in Ham’s F-10 (Sigma Aldrich, St. Louis, MO) supplemented with 5 v/v % fetal calf serum, 1 mM L-glutamine, 10 μg/mL heparin, 100 μg/mL streptomycin, and 100 IU/mL penicillin; then, they were coincubated with 10⁶ motile epididimal spermatozoa/mL in 50 μL droplets of IVF medium under mineral oil in 5 v/v % CO₂ at 38°С.

Embryos were cultured in Ham’s F-10 (Sigma Aldrich) supplemented with 5 v/v % fetal calf serum, 1 mM L-glutamine, 100 μg/mL streptomycin, and 100 IU/mL penicillin at 38°C, 5 v/v % CO₂ under oil for up to 2 days, when the two- to four-cell stage was reached.

To perform the Raman study, one to three cells (COCs, mature oocytes, or embryos) were transported in plastic straws filled with Ham’s F-10 solution. Before freezing, oocytes/embryos were transferred to a cryoprotectant solution of Dulbecco’s phosphate-buffered saline (DPBS) and 10 v/v % glycerol. Equilibration with the cryoprotectant solution was performed in several steps: at the first step, oocytes/embryos were transferred into thrice-diluted DPBS/glycerol solution for 5 min; then, they were placed into a 10 μL drop of the twice-diluted DPBS/glycerol solution. Finally, the cells were transported into the undiluted DPBS/glycerol solution and placed on the glass with a cavity. The prepared sample was covered with a mica slice and sealed with paraffin.

Sample freezing

We carried out experiments with COCs and mature oocytes (three experiments per group) and four experiments with preimplantation embryos (see the photos in Fig. 1, a–d). Samples with cells were placed into a FTIR600 cryostat (Linkam, Epsom, UK) cooled by a liquid nitrogen vapor flow. The freezing protocol was chosen to be close to the standard slow program freezing protocol conventionally used for mammalian embryos (2, 43, 44). The sample was cooled to the ice nucleation temperature T_n = −7°C at a cooling rate of 1°C/min. Ice nucleation was induced by touching the sample with a copper wire precooled in liquid nitrogen. After ice formation, the sample was kept at T_n from 10 to 30 min to ensure ice recrystallization. The sample was cooled to −40°C with a cooling rate of 0.3°C/min, to −70°C at a rate of 1–2°C/min, and then to −180°C at a rate of 5–10°C/min. Slow cooling results in strong dehydration of embryos and oocytes, which prevents the cells from damage by dendritic ice crystals. Sample cooling was paused at specified temperatures to obtain Raman spectra. The local temperature near the freezing cell was verified by the Raman spectrum of ice (see Fig. S1).

Figure 1.

Representative raw Raman spectra from LDs (left). The spectra are shifted vertically for illustrative purposes and are aligned in the following order (from bottom to top): COC at +20°C, mature oocyte at +20°C, and early embryo at +20, 0, −7, −32, −58, −115, and −183°C. Bright-field microscopy photos (right) are shown for (a) COC at +20°C, (b) mature oocyte at +20°C, (c) and (d) embryo at +20 and −50°C, (e) and magnified region with a large number of LDs. To see this figure in color, go online.

Raman experiment

Raman measurements were carried out using a laboratory-built experimental setup (37, 45). A solid-state laser (Millennia II; Spectra Physics, Santa Clara, CA) at a wavelength of 532.1 nm was used for Raman scattering excitation. A 100× objective (PL Fluotar L; Leica Microsystems, Wetzlar, Germany) with NA = 0.75 and a working distance of 4.6 mm was used to focus the laser beam into an ∼1 μm diameter spot (Fig. S2). The irradiation power after objective was 6.5 mW. Scattered radiation was collected using the same objective, and Raman spectra were measured using a monochromator (SP2500i; Princeton Instruments, Trenton, NJ) equipped with a charge-coupled device detector (Spec-10:256E/LN; Princeton Instruments). The spectral resolution of the experimental setup is 2.5 cm⁻¹ (see Fig. S3). The wavelengths for all measured spectra were calibrated using a neon-discharge lamp.

We measured Raman spectra from several substances with known numbers of double bonds. Palmitoleic, linoleic, and linolenic acids; triolein; and trillinolein were taken from Sigma Aldrich. Lyophilized phospholipids 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2-dilinoleoyl-sn-glycero-3-phosphocholine were taken from Avanti Polar Lipids (Alabaster, AL). Raman spectra were measured from fatty acids and triglycerides in the liquid phase state. To measure Raman spectra from phospholipids in the liquid-like disordered phase, suspensions of multilamellar lipid vesicles were prepared using the protocol described previously (39).

Cat oocytes and embryos contain LDs of different sizes from the submicron scale to several microns (see Fig. 1 e). The high spatial resolution allows collecting scattered radiation from a volume comparable to the size of a single big LD (5 μm in diameter). Thus, we tried to measure Raman spectra from the same LD during the experiment. However, it was not always possible because of LD movements and hiding during cell freezing. In this case, we were compelled to change the studied LD one or two times per experiment. Raman spectra were collected from the neighboring LDs from the same area inside the cell. For each experimental point, two spectral ranges were sequentially measured to provide the overall spectral range from 300 to 4000 cm⁻¹. For both spectral ranges, several spectra were acquired at each experimental position, followed by spectral averaging. The acquisition time for a single spectrum was 1 min, and the overall measurement time for one experimental point was 15–20 min.

Results

Domestic cat embryos and oocytes are rich in lipids, mainly found in LDs. Fig. 1 shows representative Raman spectra from LDs in COCs, in-vitro-matured oocytes, and in-vitro-derived early embryos measured at different temperatures. The Raman spectra of all cell types contain a similar set of Raman bands. The lipid contribution is manifested by lines of CC stretching vibrations at 1062, ∼1100, and 1130 cm^–1; twisting (1300 cm⁻¹) and scissoring (1440 cm⁻¹) deformational CH modes; and double bonded C=C (1660 cm⁻¹) and C=O (1745 cm⁻¹) bands. The Raman peaks at 2850 and 2882 cm⁻¹ correspond to symmetric CH₂ (sCH) and antisymmetric CH₂ (aCH) stretching vibrations, respectively. The absence of the CN mode at 720 cm⁻¹, which is typical for phospholipids (39), shows evidence that the lipid contribution comes mainly from triglycerides and free fatty acids.

Besides the intensive lipid contribution, the Raman spectra also contain lines indicating the presence of proteins and glycerol. The low-intensity line at 1004 cm⁻¹ is assigned to the phenylalanine contribution. The peaks at 603, 750, and 1586 cm⁻¹ correspond to resonance Raman scattering of cytochromes. Other well-known protein lines, such as the cytochrome peak at 1130 cm⁻¹ or the amide I mode at ∼1655 cm⁻¹, overlap with the lipid lines. The existence of cytochrome Raman lines points to an external contribution from the cytoplasm and the nearest organelles. The protein contribution does not exceed 10% from the scissoring CH₂ band or the C=C mode of lipids; thus, it was neglected. The glycerol contribution from LD surroundings depends on cell dehydration and has to be taken into account. We used the glycerol peaks at 420 and 483 cm⁻¹ to evaluate the intensity of glycerol peaks (46) and subtracted the glycerol contribution from the raw spectra of the cells. The glycerol spectrum is temperature dependent (37); therefore, the subtraction procedure was applied to the Raman spectra of the cell and the glycerol solution measured at the same temperatures (see details in Supporting Materials and Methods).

We used Raman spectra to estimate the degree of lipid unsaturation and to investigate the LPT. To evaluate the degree of unsaturation, Raman intensities of the CH deformation mode (CH₂ scissoring and CH₃ antisymmetric bending vibrations) and the C=C peak were studied. The lipid lines demonstrate a pronounced temperature dependence (see Fig. 1). The C=O, CC, and CH stretching vibrations are sensitive to the LPT and reflect different aspects of the lipid structure. The quality of the measured spectra is sufficient for a comprehensive analysis of the LPT in all these three spectral regions.

Analysis of the lipid unsaturation degree

Lipid unsaturation decreases the LPT temperature and enhances cell survival after cryopreservation. To characterize the degree of unsaturation, we studied the ratio between the number of C=C bonds and the number of CH₂+CH₃ groups (N_C=C/N_CH2+CH3). The intensity ratio between the C=C peak (I_C=C) and the deformational mode at 1440 cm⁻¹ (I_δCH) is proportional to this ratio (40, 41, 42). Thus, we constructed a calibration curve (Fig. 2 a) based on Raman spectra of several triglycerides, phospholipids, and free fatty acids with the known number of double bonds per acyl chain (see Fig. 2 b). For the reference lipids, the N_C=C/N_CH2+CH3 ratio was calculated by taking into account C=C, CH₂, and CH₃ groups from acyl chains only. Reference Raman spectra were collected from samples in the liquid-disordered phase state at room temperature (+25°C), which is essential because the I_C=C/I_δCH intensity ratio depends on the temperature and phase state. The calibration curve is shown in Fig. 2 a.

Figure 2.

(a) Calibration curve for evaluation of the degree of unsaturation (N_C=C/N_CH2+CH3) from the intensity ratio Raman data I_C=C/I_δCH. The gray line is the linear fit (R² = 0.99). (b) Raman spectra from unsaturated lipids are used to calibrate the unsaturation degree in LDs. The spectra are shifted vertically for illustrative purposes. DLPC, 1,2-dilinoleoyl-sn-glycero-3-phosphocholine. To see this figure in color, go online.

We used spectra of LDs measured at +20°C to estimate the degree of unsaturation. The Raman experiment does not reveal any significant difference in the degree of lipid unsaturation for different cell types (see Fig. 2; Table 1). The average ratio I_C=C/I_δCH is ∼1.125 with an SD of 20%. This ratio corresponds to N_C=C/N_CH2+CH3 = 0.0925, equivalent to ∼1.3 double bonds per typical C18 acyl chain on average. The deviation in I_C=C/I_δCH can result from the diversity of the lipid content in LDs within the same cell and between different cells. However, this spread can also be explained by systematic experimental errors, such as variations in polarization conditions of the Raman experiment or an unaccounted protein contribution to the measured Raman spectra.

Table 1.

Summary of Raman Study Results

#	NC=C/NCH2+CH3	T∗, °C	TC, °C
COC #1	0.089	−4	−29 to −24
COC #2	0.084	−2	−7 to −4
COC #3	0.082	+1	−20 to −13
Oocyte #1	0.126	−2	−30 to −25
Oocyte #2	0.075	−2	−29 to −8
Oocyte #3	0.111	−1	−62 to −8
Embryo #1	0.076	−2	−33 to −10
Embryo #2	0.114	+4	−34 to −14
Embryo #3	0.088	−10	−20 to −10
Embryo #4	0.082	−5	−51 to −30

CH stretching band

Fig. 3 shows the temperature evolution of the CH band sensitive to acyl chain conformations and intermolecular interactions. The most striking effect associated with the changes in the lipid molecule state caused by the temperature decrease is the intensity increase of aCH mode at 2882 cm⁻¹. To investigate the changes in the aCH intensity, we examined the intensity ratio between the aCH and sCH modes (I_aCH/I_sCH). Evaluation of this ratio is described in (Fig. S4). At high temperatures (T > 0°C), the I_aCH/I_sCH ratio is low because of inhomogeneous broadening of the aCH peak caused by the high variance of the conformational states of lipid molecules. A decrease in temperature leads to freezing of the lipid conformational states, resulting in narrowing of the aCH peak and an increase in I_aCH/I_sCH. The abrupt increase in the I_aCH/I_sCH ratio with decreasing temperature reflects the transition of lipid acyl chains to the ordered state, which can be considered as evidence of the LPT occurrence.

Figure 3.

Temperature evolution of the Raman CH band from an LD in vivo (embryo #4 in Table 1). The inset shows the temperature dependence of the I_aCH/I_sCH ratio for the LD (filled circles) and DOPC vesicle data (empty circles) taken from (39). The arrow marks the temperature corresponding to the onset of the LPT. To see this figure in color, go online.

The inset in Fig. 3 shows the effect of temperature on I_aCH/I_sCH ratio for synthetic DOPC vesicles and LD inside the cat embryo. DOPC has one double bond per C18 acyl chain. This is comparable to the degree of lipid unsaturation in LDs inside cat oocytes and embryos. Above 0°C, the I_aCH/I_sCH ratio for both samples can be described with a temperature-independent constant. The temperature dependence of the I_aCH/I_sCH ratio for temperature dependence of DOPC has a sharp gap corresponding to the LPT at −17°C. Triolein, which contains three C18 acyl chains with one double bond each, also exhibits an abrupt increase in I_aCH/I_sCH at −5°C (see Fig. S9). However, in LDs, the LPT is broadened, and the increase in I_aCH/I_sCH occurs without abrupt changes. In this case, the I_aCH/I_sCH ratio deviation from the high temperature constant can be considered as the onset of the LPT. The LPT can also be detected by the peculiarity in the temperature behavior of the sCH mode (Figs. S6–S8).

The temperature of the LPT onset (T^∗) demarcates the disordered liquid state and intermediate states with a higher degree of ordering (Fig. 3, inset). For all cells studied, the I_aCH/I_sCH ratio increase begins in the temperature range from −10 to +4°C, with an average value of −2°C. The estimated value of T^∗ does not correlate with the cell type or degree of lipid unsaturation (see Table 1). However, the maximal spread in T^∗ was observed at the early embryo stage. At the low-temperature limit, the I_aCH/I_sCH temperature dependence of the LDs is similar to that of the synthetic phospholipid samples (45). This is in agreement with the earlier reported data for preimplantation mouse embryos (37).

It can be noted that T^∗ is close to the ice formation temperature (T_n). To verify the effect of ice formation on the state of lipids, we carried out Raman measurements at T_n before and after ice nucleation. The values of this ratio were the same (Figs. S6–S8). Therefore, we concluded that the phase state of LDs does not depend on the ice formation, and the proximity of T^∗ and T_n is a coincidence.

C=O stretching band

In Raman spectra, the ester carbonyl stretching region can provide an insight into the lipid phase organization (47, 48). Triglycerides have three polymorphic forms in the solid-like ordered phase: α, β′, and β (49). The Raman C=O band can distinguish the liquid-like disordered state and the three polymorphic forms of the ordered phase (50). The C=O band corresponding to the liquid state demonstrates no pronounced spectral features and can be described by using a Gaussian function centered at ∼1750 cm⁻¹. For the polymorphic α-form, the C=O band also has a Gaussian-like shape. However, the band position is shifted to lower frequencies as compared to the spectrum of the liquid state. Other polymorphic forms show more complex shapes of the C=O band. For example, the β-phase of triolein demonstrates two sharp peaks in the Raman spectra at 1727 and 1744 cm⁻¹; the spectra of the β′-phase have peaks at 1730 and 1741 cm⁻¹ (50). In the Raman spectra obtained in our experiments, three peaks at 1727.5, 1734, and 1740.5 cm⁻¹ can be discerned. The full set of these lines does not match any known lipid polymorphic forms. The Raman spectra of the different phases depend on the particular triglyceride studied, and the identification of particular β- and β′-phases only by Raman spectra seems to be an incorrect task for such a complex object as a natural LD. A frozen LD can be formed by a mixture of β- and β′-phases of different triglycerides. Therefore, for simplicity, further in the text we will use the term “β-phases,” implying β, β′, or a mixture of these two phases.

To reveal lipid crystallization (transition from the liquid to solid state, related to ordering in molecule arrangement), we traced the position of the C=O band evaluated with a Gaussian fit. Fig. 4 shows the temperature dependence of the C=O band position (see also Figs. S6–S8). It can be seen that the temperature dependence of COCs has a discontinuity that is absent in the temperature dependence of mature oocytes and early embryos. The detected gap was associated with the transition to the solid ordered states and was used to determine the lipid crystallization temperature (T_C). For three COCs, we obtained T_C ≈ −5, −17, and −27°C, i.e., it varies significantly from cell to cell. The temperature dependence of mature oocytes and embryos demonstrates the broadened lipid crystallization occurring in the temperature range from −10 to −50°C. The temperature ranges of the phase transformation are shown in Table 1. In some cases, the temperature dependence demonstrates both a gradual change and a short gap in the C=O band position (oocyte #1 and embryo #4 in Fig. 4; Table 1).

Figure 4.

Temperature dependence of the C=O band position and band decomposition (at T < −60°C). The left panel shows the temperature dependence of the C=O band position for different cell types. The presented temperature dependence is shifted vertically by 5 cm⁻¹ for illustrative purposes; the major ticks denote the 10 cm⁻¹ frequency scale. The vertical dashed lines in the subpanel with the COC data mark the gaps corresponding to lipid crystallization. The right panel shows the decomposition of the experimental C=O spectra on three Lorezians and one Gaussian peak described in the text. The empty circles denote the experimental data, the red lines denote the applied fits, the green lines show the Gaussian contribution, and the navy lines show the contribution from the sum of the Lorentz peaks. The vertical dashed lines mark the positions of the Lorentz peaks shown by the blue lines. To see this figure in color, go online.

To study the phase content of frozen LDs, the spectral shape of the C=O band was investigated. We used spectra with a high signal/noise ratio obtained from averaging of the spectra from cells of the same type and all the temperatures below −60°C (Fig. S11). The average spectra were fitted with the sum of three Lorentz peaks and one Gaussian peak (see Fig. 4). The Lorentz peaks simulate the contribution from the β-phases, and the Gaussian peak models the contribution from the less-ordered α-phase. Fitting the peak included non-negativity constraints, whereas the positions and widths of the peaks were fixed (for details, see Supporting Materials and Methods). In the case of mature oocytes and embryos, the fits demonstrate similar values for the peak intensities (Fig. S12). For COCs, the Lorentz peaks at 1734 and 1740.5 cm⁻¹ appear to be more intensive, and the Gaussian component is reduced. The intensities of the Lorenz peaks indicate the same phase composition of LDs in mature oocytes and embryos, which is different from the composition of LDs in COCs. The low magnitude of the Gaussian contribution implies that LDs in frozen COCs have a higher fraction of β-phases (58% of the overall area of the C=O band) than in mature oocytes and early embryos (only 43%). The investigation of the C=O band shape proves that the crystallization of lipids in LDs of frozen COCs and later developmental stages is different.

C-C stretching region

The CC stretching region is widely used in investigations of acyl chain ordering in model lipid systems (28, 39, 51). Therefore, the possibility of investigating the CC region in Raman spectra from biological samples was examined. Fig. 5 a shows the effect of temperature on Raman spectra in the CC stretching region after subtraction of the baseline and glycerol contributions. The temperature decrease leads to enhancement of the peak intensity at 1062, ∼1100, and 1130 cm⁻¹. The mode at the highest frequency at ∼1130 cm⁻¹, also known as the “all-trans” mode, is considered as a reliable measure of all-trans conformations (52). The intensity measurement for the CC modes is problematic because of the ambiguity in baseline correction and the overlap with the cytochrome peak. Thus, we inspected how the position of the all-trans peak depends on temperature (see Fig. 5 b). At temperatures above T^∗, the precision of the parameter evaluation is low because of the low intensity of the all-trans peak. Below T^∗, the all-trans peak becomes sharper, and the frequency of the peak increases. A further temperature decrease leads to peak sharpening, and its position shifts toward higher frequencies. This temperature dependence is in qualitative agreement with the data obtained from synthetic lipids such as DOPC (Fig. 5 b). However, low precision of the all-trans peak position estimation at high temperatures makes it difficult to detect the LPT using this approach.

Figure 5.

Temperature evolution of the Raman spectrum in the CC stretching region. (a) shows a representative stretch CC region of the Raman spectra measured from the LD in vivo (embryo #4 in Table 1). The gray vertical line denotes the so-called “all-trans” peak. The spectra are shifted vertically for illustrative purposes. On the right, (b) shows the temperature dependence of the all-trans peak position (embryo #4). The filled circles represent the data obtained from the LD; the empty circles correspond to the DOPC vesicle data taken from (39). (c) and (d) demonstrate the temperature dependence of η in the case of embryo #4 and COC #1, respectively. The vertical dashed lines denote T^∗ determined from CH stretching region analyses. The vertical dashed-and-dotted line shows T_C evaluated from the C=O band analysis. To see this figure in color, go online.

To study the temperature evolution of the CC stretching region and avoid problems with all-trans peak analysis, we used a simplified approach based on the CC spectrum linear decomposition into spectral components. This concept was already successfully tested on synthetic lipids (53, 54). At the low-temperature limit (below −100°C), acyl chains of lipid molecules are in the ordered all-trans conformation state, whereas at the high-temperature limit (above 10°C), acyl chains can be considered as completely disordered. Thus, we described the CC region as a combination of spectral components corresponding to ordered ($S_o\left(\omega \right)$) and disordered ($S_d\left(\omega \right)$) states and linear background. The Raman spectrum in the CC region, $I\left(\omega \right)$, was fitted with the following linear combination:

$$I\left(\omega \right)=a+b\cdot \omega +C_o\cdot S_o\left(\omega \right)+C_d\cdot S_d\left(\omega \right)$$ (1)

where a and b correspond to the parameters of the linear function, whereas $C_o$ and $C_d$ are the magnitudes of the ordered and disordered components, respectively. The details of data handling and examples of the CC region fits are presented in (Fig. S5). This approach can extract more reliable data because the analysis involves not only the all-trans peak but also other spectral features of the CC region.

The ratio η, defined as $C_o/\left(C_o+C_d\right)$, was used as the acyl chain ordering parameter. The value η = 0 corresponds to acyl chains in the completely disordered conformation state, whereas η = 1 corresponds to the ordered state. Fig. 5 c shows an example of η(T) for embryo #4. The parameter η begins to increase at approximately the same temperatures as the I_aCH/I_sCH ratio. In the case of COC #1, η(T) demonstrates an abrupt increase at a temperature corresponding to the peculiarity observed in the C=O band (Figs. 4 and 5 d). However, η(T) does not show pronounced changes at T_C for COC #3, which may result from the insufficient quality of the measured Raman spectra. For COC #2, T^∗ and T_C are too close to distinguish these two peculiarities in η(T) (Fig. S6). Thus, we conclude that the CC region appears to be sensitive to the onset of the LPT related to the ordering of acyl chain conformational states. Under certain conditions, it seems to be possible to detect the lipid order in the molecule arrangement.

Discussion

The GRB concept was successfully applied to some laboratory and farm animal species (2, 3, 4). Freezing of embryos and oocytes is still challenging for some more exotic mammalian species, especially for those whose oocytes/early embryos are rich in lipids (5, 11, 15). Thus, to achieve better survival rates and to avoid massive injuries, knowledge about the lipid content and the LPT in LDs during freezing is needed (24, 25, 26, 37). Here, we presented the most accurate and detailed data to date about the lipid state in LDs inside frozen mammalian oocytes and embryos using a contactless Raman approach.

The average unsaturation degree was estimated for triglycerides in domestic cat COCs, mature oocytes, and early embryos. LDs of all the cells demonstrate a similar degree of unsaturation, corresponding to 1.3 double bonds per C18 chain with 20% deviation (N_C=C/N_CH2+CH3 = 0.0925). In comparison, the average unsaturation degree for ovine oocytes, calculated from phospholipid chromatography data (26), is approximately three times lower (N_C=C/N_CH2+CH3 = 0.0313). The observed spread in our data may result from the diversity in LDs inside the cell or unspecified parameters, such as the cat breed or diet. Considering the data deviations, we conclude that no drastic changes in the lipid unsaturation degree occur during domestic cat COC development to mature oocytes and early embryos. At the same time, our results cannot exclude the possible changes in the average unsaturation degree at the level of ∼10%.

This study measuring Raman spectra evolution with temperature reveals two peculiarities in the lipid phase state in LDs of freezing oocytes and embryos. The first one (at T^∗) is observed in the temperature dependence of the Raman spectra in the CH₂ and CC stretch regions. Above T^∗ = −2 (−10 to +4) °C, the LDs are in a liquid disordered state. Upon cooling below T^∗, gradual ordering of the acyl chains of triglycerides begins. In earlier studies of LPTs in oocytes (24, 25, 26) and embryos (37), only CH₂ stretching modes were investigated. The averaged T^∗ agrees with the recently reported Raman study of LPTs in mouse embryos, in which the LPT was detected in the temperature range from −7 to 0°C (37). However, for other species, the LPT was reported at temperatures above 0°C (24, 25, 26). Bovine oocytes undergo the LPT in the temperature range from +13 to +20°C (25), and for ovine oocytes, the broadened LPT occurs at +16°C (26). According to the concept that chilling injury depends on the LPT temperature (17, 24, 26), embryos and oocytes of the domestic cat should have a higher chilling tolerance than ovine or bovine ones. This might be the reason why the embryos of the domestic cat were successfully cryopreserved in 1988 (55), well before the embryos of any other Carnivora species (5).

Below T^∗, the lipids turn to more ordered crystalline states (at T_C). From the temperature dependence of the C=O band, it follows that further cooling of oocytes and early embryos leads to a gradual increase of the translational order of the lipid molecule packing. For COCs demonstrating an additional peculiarity in the behavior of C=O mode, a sharp transition to the solid state was found at T_C ≈ −19 (−27 to −5.5) °C. Based on the Raman data (Fig. S10), we can assume that triglycerides in LDs of COCs are in the intermediate liquid-ordered state at the temperatures between T^∗ and T_C. For comparison, in a homogeneous single-component triolein system, the ordering of hydrocarbon chains and molecular arrangement occur abruptly at the same temperature (see example in Fig. S9).

By now, three different models describing the liquid-crystalline phase of triglycerides in the disordered state have been proposed (47): smectic (56), nematic (57), and discotic (58, 59). The last model assumes that triglyceride molecules in the liquid state have a splayed orientation of acyl chains, forming a discotic (Y-like) conformation state with disordered hydrocarbon chains. The first two models use the concept that triglycerides in the liquid state have an h-like orientation of acyl chains, resembling a tuning fork. Because the h-like conformation appears in crystalline phases, it can be considered as more predisposed to hydrocarbon chain ordering. The smectic phase seems to be the most ordered of the three phases mentioned. In this phase, the lipids form distinct lamellar structures with a translational disorder inside the layers. Therefore, it can be argued that triglycerides of freezing LDs undergo the transition to the smectic phase below T^∗ (32). In this phase, triglycerides turn to the h-like orientation suitable for the ordering of acyl chains. The complex lipid content of biological LDs results in suppression of further crystallization of triglycerides. Only at T_C does the phase separation and crystallization of the supercooled mixture take place.

The detected change in the C=O band position reflects triglyceride crystallization. The investigation of the band shape reveals the formation of different polymorphic forms. In frozen LDs, the fraction of triglycerides in β-phases is higher for COCs than for mature oocytes and embryos. COCs were taken directly from the ovarian tissue, whereas mature oocytes and early embryos were obtained after in vitro procedures (in vitro maturation/IVF). Thus, we suggest that incubation might be the source of the differences between COCs and the later development stages. Mass spectrometry studies show that the lipid content may differ in fresh and in-vitro-cultured oocytes and embryos (60, 61). These observations agree with different biological properties of in-vitro- and in-vivo-matured oocytes (62). COCs and the later stages show similar degrees of lipid unsaturation and values of T^∗ but different values of T_C and crystallized states. Probably this effect is associated with the changes in the composition of lipophilic admixtures such as cholesterol.

Although Raman spectroscopy was already introduced to investigate the LPT in early embryos (37), this study expands the use of this approach to reveal the details of the LPTs in single oocytes and embryos. Raman spectroscopy is considered as a method of choice for in situ LPT research comprising individual cell monitoring. The last advantage is especially critical for rare and endangered species. A single-cell investigation also can help to avoid the effect of LPT blurring, which inevitably happens when multiple cells are simultaneously studied. This is important in the case of sharp transitions (for example, see COC data in Fig. 4). A Raman experiment can be performed with a high spatial resolution corresponding to the resolution of a confocal microscope and does not suffer from water absorbance limitations. It is noteworthy that the same set of Raman spectra contains information about the degree of lipid unsaturation, the onset of the LPT, and triglyceride crystallization in cells. In perspective, a contactless, label-free Raman approach can be embedded into actual cryopreservation systems and protocols to monitor the lipid phase state of the cells at different stages of cryopreservation.

Conclusion

In this study, we investigated the phase transitions in the LDs within frozen COCs, mature oocytes, and early embryos of domestic cats using Raman spectroscopy. The specific results of the study can be summarized as follows:

• 1)The average degree of lipid unsaturation (N_C=C/N_CH2+CH3) was estimated to be ∼0.0925 (with 20% deviations). No significant differences in lipid unsaturation were found between COCs, matured oocytes, and preimplantation embryos.
• 2)The investigation of the temperature dependence of the CC and CH₂ Raman lines made it possible to detect the onset of the LPT, which occurs typically at −2°C. No significant differences were found between different developmental stages, from COCs to early embryos. The temperature behavior of the CH₂ modes in Raman spectra from LDs in all these samples appears to be close to the known temperature dependence of the CH₂ stretching modes in synthetic lipid systems. Above the LPT onset, lipids are in the liquid disordered state, in which the LDs can participate in cellular metabolism.
• 3)The C=O band was used to reveal triglyceride crystallization in the LDs during freezing. It was demonstrated that crystallization of LDs occurs differently for different cell types. COCs undergo a sharp transition, which occurs within the temperature range from −27 to −5°C. In the case of early embryos and mature oocytes, lipid crystallization occurs gradually during freezing. In our experiments, the composition of polymorphic forms of triglycerides in frozen LDs differs for complexes of COCs and other stages of development.

Finally, we demonstrated that single-cell Raman spectroscopy can provide in situ label-free characterization of LPTs in freezing oocytes and embryos. The proposed approach opens the prospects for monitoring of LPTs in various lipid-rich embryos and oocytes.

Author Contributions

K.A.O., S.Y.A., and N.V.S. designed the research. K.A.O. and V.I.M. performed the experiments. K.A.O. processed the raw data. K.A.O. and N.V.S. analyzed the data. K.A.O. wrote the article with contributions from all coauthors.

Editor: Arne Gericke.