3D Chemical Imaging by Fluorescence-detected Mid-Infrared Photothermal Fourier Light Field Microscopy

Danchen Jia, Yi Zhang, Qianwan Yang, Yujia Xue, Yuying Tan, Zhongyue Guo, Meng Zhang, Lei Tian, Ji‐Xin Cheng

TL;DR

This paper introduces a novel microscopy technique, FMIP-FLF, that combines fluorescence and mid-infrared photothermal imaging to achieve high-speed, 3D chemical imaging of living cells. By leveraging thermosensitive fluorescent dyes to detect subtle temperature changes induced by mid-IR absorption, the system can visualize molecular composition with submicron resolution at 8 volumes per second, opening new avenues for studying dynamic biological processes.

What Problem Does This Paper Address?

Current 3D imaging techniques in biology, while excellent at visualizing structure, often lack detailed chemical information. Fluorescence microscopy, a workhorse for live-cell imaging, relies on fluorescent labels that can alter cellular behavior and don't inherently provide chemical identity. While vibrational spectroscopy methods like Raman microscopy offer chemical specificity, they are typically slow and have limited penetration depth or sensitivity. Mid-infrared (MIR) photothermal (MIP) microscopy offers a way to get chemical information from MIR absorption, which is orders of magnitude stronger than Raman scattering. However, traditional MIP microscopy is often slow, requiring scanning, which is incompatible with observing fast biological dynamics. This paper aims to bridge this gap by developing a technique that combines the speed and 3D capabilities of Fourier Light Field (FLF) microscopy with the chemical specificity of MIR photothermal imaging.

Methods

The core innovation is the Fluorescence-detected Mid-Infrared Photothermal Fourier Light Field (FMIP-FLF) microscopy. This technique builds upon Fourier Light Field (FLF) microscopy, which uses a lenslet array to capture multiple perspectives of emitted fluorescence in a single snapshot, enabling rapid 3D reconstruction. The novelty lies in how chemical information is encoded:

  • MIR Pump Beam: A pulsed mid-infrared laser is used to selectively excite specific molecular vibrations.
  • Thermosensitive Fluorescent Dyes: These dyes are chosen such that their fluorescence intensity is sensitive to small temperature changes (on the order of 1% per Kelvin).
  • Photothermal Effect: When the MIR laser is absorbed by molecules in the sample, it causes localized heating. This heat then slightly alters the fluorescence intensity of the nearby thermosensitive dyes.
  • FMIP Signal Extraction: By comparing fluorescence images taken with the MIR laser "on" and "off" (IR-on vs. IR-off), the subtle modulation in fluorescence due to the photothermal effect can be isolated. This difference signal directly reflects the MIR absorption at specific wavelengths.
  • 3D Reconstruction: The FLF system captures the light field information, allowing for 3D reconstruction of the fluorescence signal. This 3D reconstruction is then combined with the extracted FMIP signal to generate a 3D chemical image.
  • Deconvolution: A 3D deconvolution algorithm, using experimentally calibrated point spread functions (PSFs), is employed to reconstruct the final 3D chemical image from the raw FLF data.

The system achieves high speed by capturing the entire 3D light field in a single snapshot and then performing a rapid subtraction and deconvolution.

Datasets

The paper does not discuss training or evaluation on large datasets in the machine learning sense. Instead, it uses specific biological samples and phantoms for system characterization and demonstration:

  • 175 nm fluorescent beads: Used to characterize the spatial resolution (lateral and axial) of the FLF system.
  • Rhodamine 6G (R6G) stained Staphylococcus aureus (S. aureus) bacteria: Used to demonstrate the FMIP-FLF imaging of biological samples and its ability to map chemical composition (Amide-I band).
  • MIA Paca-2 cells (pancreatic cancer cells):
    • Stained with Lipi-red for visualizing lipid droplets.
  • Treated with 13C isotopically labeled fatty acids to study lipid metabolism.
  • Compared with gemcitabine-sensitive MIA Paca-2 cells and gemcitabine-resistant G3K cells to investigate altered lipid metabolism in drug resistance.

Key Insights and Ideas

  • Bridging the Chemical-Structural Divide: The FMIP-FLF approach elegantly merges structural information from fluorescence with chemical information from MIR absorption, overcoming a major limitation of purely fluorescence-based 3D imaging.
  • Leveraging Thermosensitive Dyes: The use of thermosensitive fluorescent dyes is a clever way to amplify the weak photothermal effect into a detectable fluorescence modulation, significantly improving sensitivity.
  • Video-Rate 3D Chemical Imaging: Achieving 8 volumes per second for 3D chemical imaging is a significant leap in speed, enabling the study of dynamic cellular processes that were previously inaccessible.
  • Labeling Strategy for Metabolism Studies: The use of isotopically labeled fatty acids (13C) is crucial for studying lipid metabolism. Unlike fluorescent labels, isotopic labels minimally alter the chemical properties of the molecules, making the metabolic tracing more accurate and reliable.
  • Biomarker for Drug Resistance: The study suggests that differences in lipid droplet composition and uptake patterns in drug-resistant cancer cells (G3K vs. MIA Paca-2) could serve as potential biomarkers for drug resistance.

Key Results

  • Spatial Resolution: Achieved a lateral resolution of 0.5-0.9 µm and an axial resolution of 0.8-1.1 µm.
  • Imaging Speed: Demonstrated volumetric imaging at a rate of 8 volumes per second.
  • Imaging Volume: The captured image volume was approximately 60 µm × 60 µm × 5.6 µm.
  • Visualization of Biological Samples: Successfully visualized protein content in bacteria and lipid droplets in living pancreatic cancer cells with chemical specificity.
  • Lipid Metabolism Analysis: Demonstrated the ability to distinguish between endogenous lipids and exogenously supplied fatty acids in cancer cells using 13C labeling and FMIP-FLF.
  • Drug Resistance Correlation: Showed that gemcitabine-resistant G3K cells exhibit altered lipid metabolism, relying more on exogenous fatty acids compared to sensitive MIA Paca-2 cells.

Limitations and Future Work

  • Fluorescence Labeling Dependence: The current FMIP-FLF method relies on thermosensitive fluorescent dyes. While these dyes can be targeted to specific organelles, they can potentially alter cellular behavior and are subject to photobleaching.
  • Reconstruction Time: While the imaging is fast, the reconstruction process, though improved by deconvolution, can still be a bottleneck for true real-time video-rate applications.
  • Field of View (FOV) and Depth of Field (DOF): The current FOV and DOF are limited by the high magnification and NA of the objective, which is suitable for subcellular imaging but may need expansion for multicellular studies.

Future Work:

  • Label-Free Imaging: The authors express interest in developing label-free high-speed volumetric chemical imaging techniques to overcome the limitations of fluorescent labeling.
  • Expanded FOV/DOF: Exploring optical designs to achieve larger imaging volumes (e.g., 400 × 400 × 20 µm³ with 1 µm lateral resolution) for applications like multicellular systems.
  • Deep Learning for Reconstruction: Investigating deep learning frameworks to accelerate the reconstruction process and potentially achieve true video-rate reconstruction.
  • Further Calibration: Emphasizing the need for further calibration with established quantitative chemical imaging techniques (like coherent Raman microscopy) to accurately correlate FMIP signals with molecular concentrations.

Summary updated

Abstract

Three-dimensional molecular imaging of living organisms and cells plays a significant role in modern biology. Yet, current volumetric imaging modalities are largely fluorescence-based and thus lack chemical content information. Mid-infrared photothermal microscopy as a chemical imaging technology provides infrared spectroscopic information at submicrometer spatial resolution. Here, by harnessing thermosensitive fluorescent dyes to sense the mid-infrared photothermal effect, we demonstrate 3D fluorescence-detected mid-infrared photothermal Fourier light field (FMIP-FLF) microscopy at the speed of 8 volumes per second and submicron spatial resolution. Protein contents in bacteria and lipid droplets in living pancreatic cancer cells are visualized. Altered lipid metabolism in drug-resistant pancreatic cancer cells is observed with the FMIP-FLF microscope.
BibTeX
@inproceedings{jia2023d,
  author = {Danchen Jia and Yi Zhang and Qianwan Yang and Yujia Xue and Yuying Tan and Zhongyue Guo and Meng Zhang and Lei Tian and Ji‐Xin Cheng},
  title = {{3D Chemical Imaging by Fluorescence-detected Mid-Infrared Photothermal Fourier Light Field Microscopy}},
  booktitle = {Chemical},
  year = {2023},
  doi = {10.1021/cbmi.3c00022},
}
Figure