DHM/SERS reveals cellular morphology and molecular changes during iPSCs-derived activation of astrocytes

The activation of astrocytes derived from induced pluripotent stem cells (iPSCs) is of great significance in neuroscience research, and it is crucial to obtain both cellular morphology and biomolecular information non-destructively in situ, which is still complicated by the traditional optical microscopy and biochemical methods such as immunofluorescence and western blot. In this study, we combined digital holographic microscopy (DHM) and surface-enhanced Raman scattering (SERS) to investigate the activation characteristics of iPSCs-derived astrocytes. It was found that the projected area of activated astrocytes decreased by 67%, while the cell dry mass increased by 23%, and the cells changed from a flat polygonal shape to an elongated star-shaped morphology. SERS analysis further revealed an increase in the intensities of protein spectral peaks (phenylalanine 1001 cm−1, proline 1043 cm−1, etc.) and lipid-related peaks (phosphatidylserine 524 cm−1, triglycerides 1264 cm−1, etc.) decreased in intensity. Principal component analysis-linear discriminant analysis (PCA-LDA) modeling based on spectral data distinguished resting and reactive astrocytes with a high accuracy of 96.5%. The increase in dry mass correlated with the increase in protein content, while the decrease in projected area indicated the adjustment of lipid composition and cell membrane remodeling. Importantly, the results not only reveal the cellular morphology and molecular changes during iPSCs-derived astrocytes activation but also reflect their mapping relationship, thereby providing new insights into diagnosing and treating neurodegenerative diseases.


Introduction
Astrocytes play a key role in the central nervous system, with their activation state directly influencing neural function and stability [1][2][3][4].The advancement of induced pluripotent stem cells (iPSCs) technology provides researchers with a model that closely resembles human physiological conditions for simulating and studying their activation in vitro [5,6].The establishment of this model not only enhances our understanding of the pathogenesis of neurological diseases but also opens up possibilities for screening potential therapeutic drugs [7][8][9][10].Investigating the activation process of iPSCs-derived astrocytes is crucial for an advanced comprehension of the mechanisms of protection and repair in the nervous system.However, the activation mechanisms of iPSCs-derived astrocytes are not fully elucidated, necessitating further research.The activation of astrocytes from a resting to a reactive state involves complex morphological changes and molecular interactions.Traditional biochemical techniques, such as western blot, immunofluorescence, and flow cytometry, have made significant progress in studying various aspects of astrocytes activation, including detailed analyses of cell morphology, protein expression profiles, and intracellular signaling pathways [11][12][13].However, these methods have inherent limitations, such as invasiveness, cumbersome sample preparation, and challenges in dynamic monitoring, which may restrict their effectiveness in capturing real-time, label-free, and comprehensive cellular response information.
In this study, label-free optical imaging and molecular spectroscopy techniques are emerging as powerful tools for studying astrocytes activation processes based on their morphological and biomolecular information.Specifically, Digital Holographic Microscopy (DHM) is a noninvasive, real-time, high-resolution imaging technique capable of dynamically monitoring the three-dimensional morphological changes of cells and extracting biophysical parameters at the single-cell level, such as cell projected area and dry mass [14][15][16][17][18], in which dry mass, defined as the mass of all cellular contents excluding water, is a crucial biomarker that reflects cellular metabolic activity, growth status, and health.Although suspended microchannel resonator has high sensitivity in measuring dry mass [19], but DHM offers higher throughput, elimination of complex sample preparation and labeling steps, and suitability for adherent cells.Dry mass, representing the mass of all cellular contents excluding water, is a crucial biomarker that reflects cellular metabolic activity, growth status, and health.By measuring dry mass, we can gain deeper insights into the regulatory mechanisms of physiological processes such as cellular responses to environmental stimuli, proliferation, and apoptosis [20][21][22].On the other hand, Surface-Enhanced Raman Scattering (SERS) technology, with its ultra-high sensitivity and resolution, provides cellular component and structural information such as proteins, nucleic acids, and lipids [23][24][25].Combined with multivariate statistical analysis methods, such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), it further enables the parsing of spectral data to reveal molecular differences underlying changes in cellular states, demonstrating promising applications in disease diagnosis and sample classification [26][27][28].
Here, we combined DHM and SERS to obtain quantitative morphological and molecular metabolic information of unlabeled cells.By investigating the changes at the morphological and molecular levels during the activation of iPSC-derived astrocytes, we hope to achieve some mapping relationship between morphological and molecular metabolic information.This study not only provides a multidimensional detection tool for cell analysis, but also enhances our understanding of the astrocyte activation process.

Cell culture and treatments
The iPSCs were purchased from iCell Bioscience Inc. (iCell, China).The iPSC complete medium (iCell, China) was prepared by adding the iPSC supplement (iCell, China) to the iPSC basal medium, iPSCs were cultured in dishes coated with Matrigel (iCell, China) at 37°C in a 5% CO2 environment.The protocol was established to differentiate iPSCs into neural progenitor cells (NPCs) and further into astrocytes.First, iPSCs were cultured in neurobasal medium supplemented with GlutaMAX, N2, and B27 for 72 hours.Next, the cells are treated with digestive enzymes and transferred to ultra-low attachment 6-well plates (Corning, USA) to promote the formation of embryoid bodies.Subsequently, cells were transferred to culture flasks and cultured in DMEM supplemented with GlutaMAX, N2, B27, and FGF2 until the appearance of neural rosettes.Neural Rosette Selection Reagent (Stem Cell, China) was used to purify the NPCs.Finally, astrocyte differentiation medium (DMEM supplemented with N2, B27, fetal bovine serum, and antibiotics) was used to obtain astrocytes.NPCs and astrocytes were cultured using STEMdiff Neural Progenitor Medium (Stem Cell, China) and DMEM medium supplemented with 10% fetal bovine serum (Gibco, USA).The iPSCs-derived astrocytes were seeded into petri dishes and cultured for 24 hours.Following this, 100 ng/mL of lipopolysaccharide (LPS) (Sigma-Aldrich, USA) was introduced, and the culture was sustained for an additional 24 hours in a cell culture incubator to induce astrocytes activation.

DHM system and calculation of biophysical parameter
Quantitative phase imaging was performed using a DHM system developed by our research group, and details of the system have been described in previous work [29].Briefly, as illustrated in Fig. 1, the system utilized a short-coherence laser source with a central wavelength of 639.67 nm and a bandwidth of 0.836 nm.The reference beam passed through the microscope objective Obj2 and the tube lens TL2 interfered with the object beam passed through Obj1 and TL1, digital holograms were recorded by Cam1 (BFS-U3-16S2M-CS, 1440 × 1080 pixels, pixel size of 3.45 µm).The Wollaston prism WP, converging lens L, polarizer P3, and camera Cam2 (BFS-U3-16S2M-CS, 1440 × 1080 pixels, pixel size of 3.45 µm) together comprise a synchronous phase shifts measurement assembly (SPSMA).The phase shifts measured by the SPSMA are exactly the same and synchronized with the phase shifts between phase measurement interferograms, so it can accurately and rapidly realize phase shift measurement online.Subsequently, the cellular phase was reconstructed using the proposed phase-shifting digital holography algorithm.Among them, the laser source was purchased from Changchun Leishi Photo-Electric Technology Ltd (China), model MW-RRL-638.The liquid crystal phase shifter used is the LCC1413-A from Thorlabs (USA), with an aperture size of 10 mm.
The phase images of cells reconstructed from holographic recordings are attributed to the phase differences generated by the varying refractive indices between cells and the culture medium, denoted as where λ represents the wavelength of the illuminating light wave, n m denotes the refractive index of the surrounding medium, while n c (x, y) and h(x, y) represent the refractive index and thickness of the cell at position (x, y).
The formula for calculating the projected area of cells is where N represents the number of pixels in the cell region, ps = 3.45 × 3.45µm 2 denotes the area of a single pixel on the CMOS sensor, and M = 20 is the magnification factor of the microscope.The total phase value of the entire cell is integrated and multiplied by a constant coefficient to obtain the dry mass of the cell: where s c represents the area of the cell region, α represents a specific refractive increment related to the protein content within the cell, varying within the range of 0.0018 ∼ 0.0021m 3 /kg, with an average value of approximately 0.0019m 3 /kg [20], and φ represents the average phase of the cell.

SERS measurement
Astrocytes were inoculated at a suitable density on slides with a gold film base and incubated for 24 hours with the addition of 45µg/mL AuNPs (NanoSeedz, China).Spectra ranging from 400 to 1800cm −1 were acquired using a confocal Raman spectrometer (Renishaw, UK) equipped with a 785 nm laser as the excitation source and an L50 × (NA = 0.6) objective lens, with a laser power of 25.93 mW, a spot size of 1 µm, and an exposure time of 10 seconds were used.40 cells were randomly selected, and more than 5 different regions were chosen of each cell.The obtained SERS spectra underwent preprocessing using WiRE 4.3 software, which included de-baselining, cosmic ray removal, and curve smoothing, followed by post-processing using Origin software.

Identification of the PCA-LDA model
The combination of PCA and LDA to process Raman spectral data sets to construct a PCA-LDA reference model has been widely used in the field of Raman spectroscopy-based cell detection [30].The unsupervised PCA method downscales the Raman spectral information of astrocytes before and after activation, which preserves the information of the original variables while making the metadata uncorrelated with each other [31].Next, the salient metadata are used as input to LDA, which enables the classification of cell states by maximizing the differences between categories to establish decision boundaries [32].Finally, the discriminative performance of the PCA-LDA model was evaluated using a confusion matrix and a tenfold cross-validation method [33].

Differentiation of iPSCs to NPCs
The protocol for obtaining NPCs from iPSCs through induced differentiation, followed by further differentiation into astrocytes, was utilized.Here, the success of induced differentiation of NPCs from iPSCs was demonstrated by optical morphological observation as well as immunofluorescence staining experiments.As shown in Fig. 2(a), at the early stage of differentiation, iPSCs were clustered and grew adherently to the wall with clear colony edges.When differentiated at day 4∼10, the cells were suspended and aggregated into clusters to form Embryoid Bodies (EBs) (Fig. 2(b)).As shown in Fig. 2(c), when differentiated to day 13, the EBs had already switched to adherent growth, the peripheral cells exhibited floral cluster growth and unfolded multiple structures, forming neural rosette junctions and enriched with a large number of NPCs. Figure 2(d) displays the purified NPCs, serving as intermediaries in the differentiation of iPSCs into neuroepithelial cells, which can subsequently differentiate into various neuroepithelial cells including astrocytes, neurons, and oligodendrocytes.To determine the differentiation status of the cells, we conducted immunofluorescence staining.As observed in Fig. 3, the iPSCs exhibited strong nuclear expression of Oct4 and SOX2, with almost no expression of Nestin, which is localized to the cytoskeleton.This aligns with the characteristic features of iPSCs, where SOX2 and Oct4, key regulators maintaining pluripotency, are localized within the nucleus [34].In contrast, in NPCs, SOX2 and Oct4 were not expressed at all, whereas the cytoskeletal proteins Nestin and the nuclear-localised PAX6 were strongly expressed.Nestin is highly expressed in proliferatively active cells and regulates cell proliferation through the key regulator PI3 K, which directly controls the proliferation of NPCs [35].These findings demonstrate the robust induction of NPCs from human iPSCs, exhibiting typical molecular and morphological characteristics of NPCs.

Model of iPSCs-derived astrocyte activation and molecular characterization
We derived astrocytes from NPCs and subsequently induced their activation using 100 ng/mL LPS, establishing an in vitro model of astrocytes inflammation.Fig 4 illustrates the immunostaining results of astrocytes before and after activation, transitioning from a resting state to a reactive state.Glial fibrillary acidic protein (GFAP) is mainly distributed in astrocytes in the central nervous system, it is involved in the composition of the cytoskeleton and maintains its tensile strength, serving as a marker for astrocytes activation.The results indicated positive GFAP expression, which showed obvious red fluorescence.Cell nuclei were stained with Hoechst,

Quantitative phase imaging of iPSCs-derived astrocyte activation
Conventional light microscopy can only qualitatively observe the differences in cell status, but cannot accurately quantify them with specific values.Here, the morphological changes of astrocytes were quantitatively evaluated by the DHM system.
Quantitative phase images of astrocytes by DHM are shown in Fig. 5(a).The resting astrocyte was polygonal in shape, with a maximum phase height of approximately 1.5 rad, and generally flatter.However, after stimulation with LPS, a significant morphological transformation occurs; the reactive astrocyte demonstrated a noticeable increase in phase height, reaching up to 3 rad, with the cell body becoming thicker and adopting a stellate shape.In addition, the high-resolution DHM system vividly captured the changes in cell fine structure, such as elongated processes in the reactive astrocyte.Based on the monitoring of more than 100 cells in the samples, the projected area and dry mass data of cells were counted to quantify the morphological changes during the activation of astrocytes.Figures 5(b) and (c) show the average changes in cell projected area and dry mass before and after activation.The measurements showed that the average projected area of resting astrocytes was 4426 µm 2 and reactive astrocytes was 1468 µm 2 , a decrease of about 67%, which is consistent with the observed phenomenon of cellular morphological contraction.However, after the activation, the average dry mass of astrocytes increased by about 23%, from 840 to 1037 pg, indicating significant change in the biomolecular content of astrocytes.

SERS spectra analysis of iPSCs-derived astrocyte activation
To achieve cellular molecular information such as proteins, lipids, nucleic acid, SERS with high-sensitivity fingerprinting was used to reveal the molecular events during LPS-induced astrocytes activation.Table 1 summarizes the assignments of the main changing spectral peaks.As shown in Fig. 6, the changes in several spectral peaks related to protein structure and function: reactive astrocytes exhibited higher intensities at 1001, 1043, 1053, 1083, 1169, 1200, and 1655 cm −1 compared to resting astrocytes.Among them, the peaks at 1001 and 1169 cm −1 were attributed to the ring breathing modes of phenylalanine and tyrosine, while the signal at 1043 cm −1 represented proline, the 1200 cm −1 peak is associated with amide III, and the peaks at 1655 and 1665 cm −1 correspond to amide I. Conversely, it can be observed that the peaks intensity decreased at 618, 755, 1142, and 1606 cm −1 , these changes were associated with the twisting of the C-C bond, symmetric breathing of tryptophan, cysteine, and the protein C = C bending.Lipid-related peaks showed a decreasing trend, including phosphatidylserine (524 cm −1 ), cholesterol ester (538 cm −1 ), phosphatidylinositol (776 cm −1 ), triglycerides (1264 cm −1 ), CH 3 symmetric stretch (1379 cm −1 ), phospholipids (1745cm −1 ), and the 1444 cm −1 band attributed to cholesterol, lipids and fatty acids.
In conclusion, the biological changes of proteins, DNA, and lipids during the activation process of astrocytes can be sensitively detected by SERS, providing insights into the molecular mechanisms associated with astrocytes activation in neurological diseases.These peaks can be considered key differences between resting and reactive astrocytes, playing an important role in distinguishing cells before and after activation.

Statistical analysis of SERS spectra
To effectively extract key information and features from complex SERS spectra, we employed the PCA-LDA method for a deeper statistical analysis of these spectral differences.This approach not only allows for more accurate differentiation of cellular activation states but also unveils the molecular mechanisms behind cell state transitions.We selected 200 spectra from astrocytes before and after activation as inputs for the PCA algorithm.Figure 7(a) displays the scoring plots for PC1 and PC2, where blue represents data from resting cells, and red indicates data from reactive cells.The spectral data are separated into two clusters, confirming their biochemical feature differences.The PC1 loading plot in Fig. 7(b), through its significant positive and negative coordinates, highlights the key molecular features distinguishing cell states.Positive contributions emphasize molecules closely associated with the reactive state, such as phenylalanine (1001 cm −1 ), proline (1043 cm −1 ), C-O stretching, C-N stretching (1053 cm −1 ), the O-P-O backbone of nucleic acids (1101 cm −1 ), guanine (1333 cm −1 ), and amide I (1655, 1670 cm −1 ).In contrast, negative contributions reveal molecules more prominent in the resting state, like cholesterol, cholesterol esters (538, 702 cm −1 ), symmetric breathing of tryptophan (755 cm −1 ), fatty acids, palmitic acid (1131 cm −1 ), cysteine (1142 cm −1 ), triglycerides (1264 cm −1 ), and phospholipids (1745cm −1 ).By inputting the scores of PC1 to PC4 into LDA and using Fischer's discriminant function to determine the decision boundary, we successfully established the PCA-LDA model.Table 2 illustrates the recognition performance of resting and reactive astrocytes using the PCA-LDA model.The model accurately classified 98 out of 100 resting astrocytes spectra and 97 out of 100 reactive astrocytes spectra, achieving an accuracy, sensitivity, and specificity of 97.5%, 98%, and 97%.Lastly, a tenfold cross-validation method yielded an approximate classification accuracy of 96.5%, further validating the model's capability to distinguish cellular activation states.This demonstrates its potential application in the detection of related diseases.

Discussion
This study comprehensively explored the morphological and molecular changes during iPSCsderived astrocytes activation using DHM and SERS techniques.DHM experiments result showed that LPS stimulation led to a transition of astrocytes from a flattened polygonal shape to a hypertrophic stellate shape, which is consistent with previously reported morphological changes in iPSCs-derived astrocytes upon activation [37].These significant morphological changes likely reflect cytoskeletal reorganization and intracellular molecular adjustments.Meanwhile, SERS accurately captured changes in the content and conformation of intracellular proteins, DNA, and lipids.Our study found that the decrease in cell projected area was closely related to lipid changes, while the increase in protein content led to an increase in dry mass, revealing a correlation between cell morphology and molecular changes.Upon activation by LPS stimulation, the cell projected area decreased by 67%, which may reflect cytoskeletal reorganization and intracellular molecular adjustments.Astrocytes activation is not controlled by a simple "on-off" pathway but is finely regulated by the complex interplay of intracellular and extracellular signaling mechanisms [38].The RhoA-ROCK signaling pathway is a key pathway that regulates the stellate morphology of flattened astrocytes and has an important impact on multiple biological processes, including cell skeleton formation, migration, and adhesion [39][40][41].The large reduction in the projected area of the cell may be related to changes in intracellular lipids, since lipids, as the main structural components of biological membranes, play an important role in a large number of biological processes, such as energy homeostasis, material transport, signal transduction, neurogenesis, and synaptogenesis [42].Phospholipids, as the most abundant category of cell membrane lipids, are crucial for the maintenance, repair, and function of neural cells [43].SERS detected a decrease in phospholipid-related spectral peaks after astrocyte activation, including phosphatidylserine at 524 cm −1 , phosphatidylinositol at 776 cm −1 , and phospholipids at 1745cm −1 .Phosphatidylserine, an aminophospholipid located in the inner leaflet of the membrane, is closely related to inflammatory responses, oxidative stress, and apoptosis [44].Previous reports have indicated that changes in phospholipids suggest alterations in the composition of glial cell membranes during neuroinflammation [45].Additionally, cell activation is usually accompanied by increased energy demands, with cells adjusting lipid metabolism to meet these needs.SERS analysis showed that the band at 1264 cm −1 was attributed to triglycerides, and that reactive astrocytes had significantly lower triglyceride levels than resting astrocytes.It has been reported in the literature that after establishing a rat brain injury model, pancreatic triglyceride lipase (PTL) is less expressed in resting astrocytes, while its expression is greatly increased in proliferating and activated astrocytes after brain injury [46,47].It is hypothesized that PTL may mediate the hydrolysis of triglycerides and that the free fatty acids produced play an important role in the synthesis of myelin sheaths in the peripheral nervous system [48,49].The peak at 1444 cm −1 was attributed to cholesterol, lipids, and fatty acids, which are essential for maintaining cell membrane integrity and function.The decrease in these lipid levels is closely related to membrane reconstruction, the initiation of stress responses, and the triggering of neuroprotective and repair mechanisms during astrocyte activation.These adjustments enable cells to maintain their functionality and stability in a new environment.
Reactive astrocytes had a 23% increase in dry mass compared to resting astrocytes, similar to previously reported increases in dry mass in activated T cells [50].At the molecular level, SERS revealed an increase in protein content and changes in protein conformation in activated astrocytes, particularly in spectral peaks at 1001, 1043, 1169, 1200 and 1655 cm −1 .The peaks at 1001 and 1169 cm −1 , corresponding to the breathing modes of phenylalanine and tyrosine, their increasing might be due to disturbances in monoamine neurotransmitter pathways, such as adrenaline and noradrenaline, where phenylalanine and tyrosine are precursors in their synthesis [51][52][53][54] Proline signaling at 1043 cm −1 was increased, and it was previously shown that the proline-rich tyrosine kinase 2(Pyk2) is involved in LPS-induced activation of rat brain astrocytes, and LPS mediates cellular release of proinflammatory proteins (including matrix metalloproteinase 9) and causes cell migration through the TLR4/c-Src/Pyk2 pathway [55].The increase in protein content is the primary cause of the increase in cell dry mass, as previous studies have shown that total cell dry mass is mainly regulated by protein synthesis and degradation [56].
In addition, the PCA-LDA model enabled us to distinguish between the resting and reactive states of astrocytes with 96.5% accuracy, paving a new path for personalized therapy based on cell state.More importantly, this study identified molecular features that are closely related to cell activation, such as phenylalanine, proline, and amide I.These molecular markers may become targets for future therapeutic strategies, bringing new hope for personalized treatment of neurological diseases.

Conclusion
This study demonstrates the potential of DHM and SERS technologies as label-free tools in cell biology research, providing new molecular and morphological evidence for the activation mechanisms of iPSCs-derived astrocytes.The mapping relationship between cellular morphology and biochemical information enriches our understanding of astrocyte biology and opens new perspectives for diagnosing and treating neurological disorders associated with astrocyte activation.

Disclosures.
The authors declare no conflicts of interest related to this work.

Fig. 5 .
Fig. 5. (a) Quantitative phase images of resting and reactive astrocytes.Boxplots showing changes in morphological parameters before and after astrocytes activation.(b) Projected area.(c) Dry mass.Within each boxplot, the lower boundary corresponds to the 25th percentile, the line within the box represents the median, and the upper boundary reflects the 75th percentile.Whiskers extending above and below the box encompass non-outlier data points.

Fig. 6 .
Fig. 6.Comparison of the mean ± standard deviation of SERS spectra for resting astrocytes (blue spectral line) and reactive astrocytes (red spectral line), with shaded areas representing the standard deviations.The orange spectral line represents the difference spectrum of the means between the two.The dominant Raman characteristic peaks of the resting and reactive states are labeled with blue and red bands.