Immune cell type, cell activation, and single cell heterogeneity revealed by label-free optical methods

Perspectives on label-free microscopy of heterogeneous and dynamic biological systems

Significance

Advancements in label-free microscopy could provide real-time, non-invasive imaging with unique sources of contrast and automated standardized analysis to characterize heterogeneous and dynamic biological processes. These tools would overcome challenges with widely used methods that are destructive (e.g., histology, flow cytometry) or lack cellular resolution (e.g., plate-based assays, whole animal bioluminescence imaging).

Aim

This perspective aims to (1) justify the need for label-free microscopy to track heterogeneous cellular functions over time and space within unperturbed systems and (2) recommend improvements regarding instrumentation, image analysis, and image interpretation to address these needs.

Approach

Three key research areas (cancer research, autoimmune disease, and tissue and cell engineering) are considered to support the need for label-free microscopy to characterize heterogeneity and dynamics within biological systems. Based on the strengths (e.g., multiple sources of molecular contrast, non-invasive monitoring) and weaknesses (e.g., imaging depth, image interpretation) of several label-free microscopy modalities, improvements for future imaging systems are recommended.

Conclusion

Improvements in instrumentation including strategies that increase resolution and imaging speed, standardization and centralization of image analysis tools, and robust data validation and interpretation will expand the applications of label-free microscopy to study heterogeneous and dynamic biological systems.

Non-invasive detection of regulatory T cells with Raman spectroscopy

Regulatory T cells (Tregs) are a type of lymphocyte that is key to maintaining immunological self-tolerance, with great potential for therapeutic applications. A long-standing challenge in the study of Tregs is that the only way they can be unambiguously identified is by using invasive intracellular markers. Practically, the purification of live Tregs is often compromised by other cell types since only surrogate surface markers can be used. We present here a non-invasive method based on Raman spectroscopy that can detect live unaltered Tregs by coupling optical detection with machine learning implemented with regularized logistic regression. We demonstrate the validity of this approach first on murine cells expressing a surface Foxp3 reporter, and then on peripheral blood human T cells. By including methods to account for sample purity, we could generate reliable models that can identify Tregs with an accuracy higher than 80%, which is already comparable with typical sorting purities achievable with standard methods that use proxy surface markers. We could also demonstrate that it is possible to reliably detect Tregs in fully independent donors that are not part of the model training, a key milestone for practical applications.

Single-cell Raman microscopy with machine learning highlights distinct biochemical features of neutrophil extracellular traps and necrosis

The defining biology that distinguishes neutrophil extracellular traps (NETs) from other forms of cell death is unresolved, and techniques which unambiguously identify NETs remain elusive. Raman scattering measurement provides a holistic overview of cell molecular composition based on characteristic bond vibrations in components such as lipids and proteins. We collected Raman spectra from NETs and freeze/thaw necrotic cells using a custom built high-throughput platform which is able to rapidly measure spectra from single cells. Principal component analysis of Raman spectra from NETs clearly distinguished them from necrotic cells despite their similar morphology, demonstrating their fundamental molecular differences. In contrast, classical techniques used for NET analysis, immunofluorescence microscopy, extracellular DNA, and ELISA, could not differentiate these cells. Additionally, machine learning analysis of Raman spectra indicated subtle differences in lipopolysaccharide (LPS)-induced as opposed to phorbol myristate acetate (PMA)-induced NETs, demonstrating the molecular composition of NETs varies depending on the stimulant used. This study demonstrates the benefits of Raman microscopy in discriminating NETs from other types of cell death and by their pathway of induction.

Non-invasive monitoring of T cell differentiation through Raman spectroscopy

The monitoring of dynamic cellular behaviors remains a technical challenge for most established techniques used nowadays for single-cell analysis, as most of them are either destructive, or rely on labels that can affect the long-term functions of cells. We employ here label-free optical techniques to non-invasively monitor the changes that occur in murine naive T cells upon activation and subsequent differentiation into effector cells. Based on spontaneous Raman single-cell spectra, we develop statistical models that allow the detection of activation, and employ non-linear projection methods to delineate the changes occurring over a several day period spanning early differentiation. We show that these label-free results have very high correlation with known surface markers of activation and differentiation, while also providing spectral models that allow the identification of the underlying molecular species that are representative of the biological process under study.

High-sensitivity hyperspectral vibrational imaging of heart tissues by mid-infrared photothermal microscopy

Visualizing the spatial distribution of chemical compositions in biological tissues is of great importance to study fundamental biological processes and origin of diseases. Raman microscopy, one of the label-free vibrational imaging techniques, has been employed for chemical characterization of tissues. However, the low sensitivity of Raman spectroscopy often requires a long acquisition time of Raman measurement or a high laser power, or both, which prevents one from investigating large-area tissues in a nondestructive manner. In this work, we demonstrated chemical imaging of heart tissues using mid-infrared photothermal (MIP) microscopy that simultaneously achieves the high sensitivity benefited from IR absorption of molecules and the high spatial resolution down to a few micrometers. We successfully visualized the distributions of different biomolecules, including proteins, phosphate-including proteins, and lipids/carbohydrates/amino acids. Further, we experimentally compared MIP microscopy with Raman microscopy to evaluate the sensitivity and photodamage to tissues. We proved that MIP microscopy is a highly sensitive technique for obtaining vibrational information of molecules in a broad fingerprint region, thereby it could be employed for biological and diagnostic applications, such as live-tissue imaging.

A novel approach to single-cell analysis reveals intrinsic differences in immune marker expression in unstimulated BALB/c and C57BL/6 macrophages

The origin of functional heterogeneity among macrophages, key innate immune system components, is still debated. While mouse strains differ in their immune responses, the range of gene expression variation among their pre-stimulation macrophages is unknown. With a novel approach to scRNA-seq analysis, we reveal the gene expression variation in unstimulated macrophage populations from BALB/c and C57BL/6 mice. We show that intrinsic strain-to-strain differences are detectable before stimulation and we place the unstimulated single cells within the gene expression landscape of stimulated macrophages. C57BL/6 mice show stronger evidence of macrophage polarization than BALB/c mice, which may contribute to their relative resistance to pathogens. Our computational methods can be generally adopted to uncover biological variation between cell populations.

Deriving accurate molecular indicators of protein synthesis through Raman-based sparse classification

Raman spectroscopy has the ability to retrieve molecular information from live biological samples non-invasively through optical means. Coupled with machine learning, it is possible to use this large amount of information to create models that can predict the state of new samples. We study here linear models, whose separation coefficients can be used to interpret which bands are contributing to the discrimination, and compare the performance of principal component analysis coupled with linear discriminant analysis (PCA/LDA), with regularized logistic regression (Lasso). By applying these methods to single-cell measurements for the detection of macrophage activation, we found that PCA/LDA yields poorer performance in classification compared to Lasso, and underestimates the required sample size to reach stable models. Direct use of Lasso (without PCA) also yields more stable models, and provides sparse separation vectors that directly contain the Raman bands most relevant to classification. To further evaluate these sparse vectors, we apply Lasso to a well-defined case where protein synthesis is inhibited, and show that the separating features are consistent with RNA accumulation and protein levels depletion. Surprisingly, when features are selected purely in terms of their classification power (Lasso), they consist mostly of side bands, while typical strong Raman peaks are not present in the discrimination vector. We propose that this occurs because large Raman bands are representative of a wide variety of intracellular molecules and are therefore less suited for accurate classification.

Non-invasive cell classification using the Paint Raman Express Spectroscopy System (PRESS)

Raman scattering represents the distribution and abundance of intracellular molecules, including proteins and lipids, facilitating distinction between cellular states non-invasively and without staining. However, the scattered light obtained from cells is faint and cells have complex structures, making it difficult to obtain a Raman spectrum covering the entire cell in a short time using conventional methods. This also prevents efficient label-free cell classification. In the present study, we developed the Paint Raman Express Spectroscopy System, which uses two fast-rotating galvano mirrors to obtain spectra from a wide area of a cell. By using this system and applying machine learning, we were able to acquire broad spectra of a variety of human and mouse cell types, including pluripotent stem cells and confirmed that each cell type can be classified with high accuracy. Moreover, we classified different activation states of human T cells, despite their similar morphology. This system could be used for rapid and low-cost drug evaluation and quality management for drug screening in cell-based assays.

A Computational Model of the Endothelial to Mesenchymal Transition

Endothelial cells (ECs) form the lining of lymph and blood vessels. Changes in tissue requirements or wounds may cause ECs to behave as tip or stalk cells. Alternatively, they may differentiate into mesenchymal cells (MCs). These processes are known as EC activation and endothelial-to-mesenchymal transition (EndMT), respectively. EndMT, Tip, and Stalk EC behaviors all require SNAI1, SNAI2, and Matrix metallopeptidase (MMP) function. However, only EndMT inhibits the expression of VE-cadherin, PECAM1, and VEGFR2, and also leads to EC detachment. Physiologically, EndMT is involved in heart valve development, while a defective EndMT regulation is involved in the physiopathology of cardiovascular malformations, congenital heart disease, systemic and organ fibrosis, pulmonary arterial hypertension, and atherosclerosis. Therefore, the control of EndMT has many promising potential applications in regenerative medicine. Despite the fact that many molecular components involved in EC activation and EndMT have been characterized, the system-level molecular mechanisms involved in this process have not been elucidated. Toward this end, hereby we present Boolean network model of the molecular involved in the regulation of EC activation and EndMT. The simulated dynamic behavior of our model reaches fixed and cyclic patterns of activation that correspond to the expected EC and MC cell types and behaviors, recovering most of the specific effects of simple gain and loss-of-function mutations as well as the conditions associated with the progression of several diseases. Therefore, our model constitutes a theoretical framework that can be used to generate hypotheses and guide experimental inquiry to comprehend the regulatory mechanisms behind EndMT. Our main findings include that both the extracellular microevironment and the pattern of molecular activity within the cell regulate EndMT. EndMT requires a lack of VEGFA and sufficient oxygen in the extracellular microenvironment as well as no FLI1 and GATA2 activity within the cell. Additionally Tip cells cannot undergo EndMT directly. Furthermore, the specific conditions that are sufficient to trigger EndMT depend on the specific pattern of molecular activation within the cell.