Water bend–libration as a cellular Raman imaging probe of hydration

Abstract

Water is a ubiquitous and vital component of living systems. Hydration, which is the interaction between water and intracellular biomolecules, plays an important role in cellular processes. However, it is technically challenging to study water structure within cells directly. Here, we demonstrate the utility and power of the water bend–libration combination band as a unique Raman spectral imaging probe of cellular hydration. Hydration maps reveal distinct water environments within subcellular compartments (e.g., nucleolus and lipid droplet) due to the spectral sensitivity of this coupled vibrational band. Spectroscopic studies using the water bend–libration are broadly applicable, offering the potential to capture the chemical complexity of hydration in numerous systems.

The importance of water in biology is universally recognized; yet, visualization of water and characterization of its molecular interactions with diverse biomolecules in cellulo are limited (1–5). Nevertheless, it is suggested that the chemical structure/environment of water molecules within a cell is unique because of crowding and confined compartments (6). Here, we report an intrinsic Raman imaging probe of cellular hydration, the water bend–libration combination band. This coupled vibration between the H–O–H bend and librational motion provides a signal where no other endogenous biomolecular vibrations exist (Fig. 1A), responsive to the chemical environment [e.g., hydrogen-bonding network (7–9)]. The coupling of the bend with low-frequency librations, which report on interfacial water (10), makes it possible to gain insights into the hydration shell(s) of intracellular biomolecules.

Fig. 1.

Raman spectral imaging of cellular hydration. (A) Average Raman spectra collected from the cytoplasm and nucleus of a SH-SY5Y cell shown in the bright-field image (B). Spectral assignments of nucleotides (pyrimidine ring breathing mode) and proteins (phenylalanine ring breathing mode) are denoted by arrows. The water bend–libration band is shaded gray (1,900 to 2,450 cm⁻¹) and magnified for clarity. The inset illustrates the combination of water bending and librational motions. (C) Raman maps for nucleotides (magenta) and proteins (green) were generated by integration over the specified spectral regions, and a composite image was generated by overlaying both maps, where gray indicates overlapping regions. (D and E) Raman maps of water generated by integration (I_a) and first moment (m₁) of the bend–libration band (1,900 to 2,450 cm⁻¹). (Top) SH-SY5Y cell (λ_ex = 514 nm, 500-ms accumulation time per pixel, 500-nm step per pixel), (Middle) SK-MEL28 cell (λ_ex = 488 nm, 250-ms accumulation time per pixel, 400-nm step per pixel), and (Bottom) HEK293T cell (λ_ex = 488 nm, 150-ms accumulation time per pixel, 300-nm step per pixel).

Results and Discussion

A total of 32 cells from three human cell lines (HEK293T, SH-SY5Y, and SK-MEL28) were imaged using a home-built Raman inverted microscope (11). Raman spectral imaging is a label-free method ideal for cellular imaging that allows for simultaneous detection (i.e., multiplexing) of various biomolecules based on intrinsic molecular vibrations with spatial resolution. Representative Raman spectra collected in the cytoplasm and nucleus are shown in Fig. 1A along with the associated bright-field image (Fig. 1B). By integrating the respective Raman peaks of nucleotides and proteins (Fig. 1A, arrows), the nucleus is differentiated from the cytoplasm (Fig. 1C).

Unlike the widely used O–H stretch (12), the bend–libration combination resides in a spectrally “quiet” region (1,900 to 2,450 cm⁻¹, Fig. 1A, gray area) without interference from endogenous biomolecules, thus simplifying data analysis. Hydration maps are generated from the integrated area (I_a) of the combination band (Fig. 1D). The difference between the cytoplasm and nucleus is clearly seen, consistent with other studies (2, 3, 5). Notably, nucleoli in the nucleus are also discerned (Fig. 1D, magenta arrows). Next, we took a model-independent approach to evaluate spectral changes by calculating the first moment (m₁), which defines the mean frequency of the bend–libration band (1,900 to 2,450 cm⁻¹). m₁ maps reveal greater molecular information (Fig. 1E), where intracellular water at all locations has a higher m₁ frequency than that of extracellular water. We interpret this higher m₁ frequency to indicate stronger H-bonding and/or more ordered-water environments (8, 13). Remarkably, this analysis pinpoints the localization of distinct subcellular water populations. The hydration map of a HEK293T cell (Fig. 1 E, Bottom) shows two very clear nucleoli, suggesting a more ordered-water network compared to the nucleus. Interestingly, the surrounding cellular peripheral is also differentiated from the extracellular water, signifying exquisite sensitivity of the bend–libration.

In addition to nucleoli, lipid droplets also exhibited different bend–libration signatures. For comparison, specific regions of interest (ROIs) for individual nucleoli (magenta) and lipid droplets (green) were isolated for each cell (Fig. 2 A, Left), and their hydration content and structure were visualized by I_a and m₁ maps, respectively (Fig. 2 A, Center and Right). The lipid droplets exhibit the highest I_a and m₁ values compared to the rest of the cellular regions. Comparing the average spectra of the ROIs reveals a striking difference in the bend–libration for nucleoli vs. lipid droplets (Fig. 2B). While both spectra have peak intensity maxima at ~2,100 cm⁻¹, the nucleoli spectrum has a shoulder at 2,180 cm⁻¹, whereas the lipid droplet spectrum shows a sharp peak in the same position. These spectral features (2,100 and 2,180 cm⁻¹) are tentatively attributed to water–water and water–solute contributions, respectively (7). Clearly, more detailed investigation is needed to elucidate the chemical nature of these bands as the water bend–libration is not well understood. Nonetheless, we can utilize these spectral variations to document aqueous environmental differences at these cellular regions (Fig. 2C).

Fig. 2.

Cellular water subpopulations are revealed by first-moment (m₁) analysis. (A) Bright-field image (Left) of SH-SY5Y cells overlaid with locations of lipid droplets (green, 1,720 to 1,770 cm⁻¹, lipid ester carbonyl stretching mode) and nucleoli (magenta, 775 to 790 cm⁻¹) mapped by Raman spectral imaging (λ_ex = 514 nm, 500-ms accumulation time per pixel, 500-nm steps per pixel). Corresponding water maps generated by integration (I_a) over the bend–libration (1,900 to 2,450 cm⁻¹, Center) and m₁ analysis (Right). (B) Corresponding Raman spectra (averaged over ROIs shown in A, Left) for lipid droplets (N = 153 spectra) and nucleoli (N = 156 spectra). Lines and shading represent mean and SDs, respectively. Spectra are offset for clarity. Reference guides are drawn at 2,100 and 2,180 cm⁻¹. (C) Comparison of m₁ values at various cellular regions for individual cells: HEK293T (N = 9), SH-SY5Y (N = 17), and SK-MEL28 (N = 6). Center lines and error bars represent the mean and SDs, respectively. P values were calculated using unpaired t tests.

The bend–libration of the cytoplasm has the lowest m₁ frequency, followed by the nucleus, nucleoli, and lipid droplets. We postulate that this observed increasing trend is related to the higher density of molecules (i.e., molecular crowding) and confinement, which would create greater water organization and structure in these subcellular compartments. This hypothesis is supported by the hydration intensity maps (Fig. 2 A, Center), where the intracellular water, in particular in the lipid droplets, appears to be at a higher level than that of the extracellular region. While seemingly counterintuitive, it can be explained because the bend–libration is reporting on a specific population of confined water within the droplets of highly concentrated lipids, which is absent in bulk water. This sensitivity to water associated with lipid droplets, in part, is due to the information accessible from the libration, which has been shown to detect the hydration shell of phospholipids (14). Altogether, this work establishes the bend–libration of water as a Raman spectral imaging probe of cellular hydration. Importantly, water within nucleoli and lipid droplets were visualized directly, a feat not previously achieved by vibrational imaging. This superb sensitivity to the molecular environment is due to coupling with low-frequency librations, which are responsive to the hydrogen-bonding network. We anticipate this ability to observe and quantify hydration states within cellular compartments to be broadly impactful. This work opens up possibilities for understanding cellular processes such as liquid–liquid phase separation, protein aggregation, and amyloid formation, paving the way for future developments in biomedical research and diagnostics.

Methods

HEK293T and SH-SY5Y cells (American Type Culture Collection (ATCC)) were grown in Dulbecco’s modified eagle medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37 °C in 5% CO₂ atmosphere. HEK293T cells (ATCC) were grown similarly using minimum essential medium. Prior to imaging, cells were fixed using 4% paraformaldehyde in phosphate-buffered saline, pH 7.4.

Raman spectral imaging was conducted as previously reported (15) with some modifications. An additional laser line at 488 nm (Coherent Sapphire-SF) was used for excitation (80 mW at the sample) along with a laser clean-up filter (LL01-488-25, Semrock), dichroic mirror (Di02-R488-25x36, Semrock), and long pass filter (BLP01-488R-25, Semrock). For collection, a 600 mm⁻¹ grating was used. The CCD image was binned from 124 to 132 pixels in the y-dimension.

Raman spectra were only processed by subtracting a constant offset, and maps were generated as specified (Matlab R2022B).

Data, Materials, and Software Availability

All study data are included in the article and have been deposited in Figshare at https://doi.org/10.25444/nhlbi.23802519 (16).