Ultra-bright Raman dots for multiplexed optical imaging

Multiplexed imaging of many molecular targets remains a grand challenge in optical microscopy. While immunofluorescence microscopy has achieved nanometer resolution via super-resolution methods, simultaneous channels are limited (typically 4–5) by fluorescence spectral linewidths (~50 nm). Raman scattering circumvents spectral crowding due to much narrower vibrational linewidths (enabling >20 channels), but Raman cross-sections are 10^10–10^14 smaller than fluorescence, limiting sensitivity for immunostaining. SRS can enhance detection by up to 10^8 but is still insufficient using common probes. SERS provides strong enhancement but typically requires large particles (>50 nm), suffers from low imaging speed, resolution, and hotspot variability; practical quantitative immunostaining is difficult. Raman-active polymers/nanoparticles can increase signal but often require complex synthesis, provide limited color palette, and are large (50–100 nm), impeding intracellular access. The authors identify the need for new Raman probes combining high brightness and small size to enable multiplexed imaging and propose compact, ultra-bright Raman nanoparticles (Rdots) prepared by a simple swelling–diffusion approach.

Prior work established super-resolution IFM (MINFLUX and others) and multiplexed IFM workflows that rely on iterative staining cycles, facing registration and epitope degradation issues. Fluorescence is fundamentally limited by broad spectra. Raman approaches allow dense spectral packing due to narrow vibrational lines. SRS improves Raman sensitivity but not enough for immunostaining with typical small molecules. SERS can yield up to ~10^11 enhancement yet requires large metallic nanoparticles, has long acquisition times, low spatial resolution, dependence on hotspots, and challenges in quantitative staining. Polymer-based Raman nanoparticles incorporate many vibrational tags (alkyne, nitrile, C–D) but involve complex chemistry, limited spectral diversity, and large sizes restricting intracellular diffusion. Recent engineered polyynes (Carbow dyes) and MARS dyes improved organic Raman probe performance, yet ultra-bright, compact, and versatile probes for practical multiplexed immunostaining remained lacking.

Probe design and fabrication: Rdots were produced by a swelling–diffusion method. Hydrophobic small-molecule Raman probes (Carbow dyes and other alkyne/nitrile-containing molecules) dissolved in THF/DMSO were added dropwise to aqueous suspensions of 20 nm carboxylated polystyrene (PS) nanoparticles to swell the polymer and allow probe diffusion into the matrix. Addition of aqueous buffer shrank the particles, entrapping probes. Colloids were clarified by brief centrifugation and extensively washed using 30 kDa MWCO filters to remove solvents. Particle concentrations were quantified against the 3054 cm−1 PS Raman peak. Spontaneous Raman spectra were acquired on calibrated microspectrometers; size and morphology were measured by DLS and SEM. Probe panel: Six Carbow probes were demonstrated initially; in total 19 common alkyne/nitrile probes were tested and successfully incorporated without spectral broadening or significant peak shifts. Ten resolvable Rdots were generated, with potential to expand to >20 using the full Carbow panel. Biofunctionalization: Rdots surfaces (abundant –COOH) were PEGylated by EDC/sulfo-NHS coupling with a mixture of amine-PEG8-OH and amine-PEG16-COOH to reduce non-specific binding and maintain colloidal stability. In a second step, IgG, protein A, or other amine-bearing biomolecules were conjugated to PEG-COOH via EDC/NHS. Excess reagents were removed by filtration or SEC; final buffers were borate or PBS with Tween-20 as needed. PEGylation increased hydrodynamic diameter modestly to ~23 nm. SRS microscopy: An integrated picosecond SRS system (picoEMERALD) on an inverted scanning microscope was used. Stokes: 1064 nm, 6 ps, modulated at 8 MHz; Pump: tunable 720–990 nm, 5–6 ps. Objective: 25× water, 1.05 NA. Detection used lock-in demodulation at 8 MHz. Titrations used 150 mW pump and Stokes with 1 ms time constant. Cell/tissue imaging used ~100 mW pump and 150 mW Stokes with 10–30 μs time constant and matching pixel dwell time. For single-particle validation, Rdots (~2 nM) were embedded in agarose gels prepared in MES pH 6.0 or borate pH 8.2 and imaged; Raman spectra were acquired by tuning the pump wavelength. Immunostaining protocols: Cells (Cos-7, HeLa, SKBR3) were fixed (4% PFA/0.1% GA), permeabilized (0.5% Triton X-100), blocked (BSA/Triton), incubated with primary antibodies, then with secondary antibody–conjugated Rdots (typically 30 nM). For primary-only staining, protein A–conjugated Rdots were preloaded with primary antibodies or Rdots were directly conjugated to primary antibodies (e.g., anti-CD44). Mouse colon frozen sections were fixed, blocked, incubated with primary antibodies, then secondary Rdots. Washes were performed between steps. Imaging used narrowband SRS at specific Raman shifts corresponding to Rdots; fluorescence channels (e.g., Alexa647 phalloidin, NucGreen) were acquired separately. Photostability was assessed by repeated SRS scanning.

• Rdots synthesis: Successful incorporation of six Carbow probes and 19 common alkyne/nitrile probes into ~20 nm PS nanoparticles without spectral shift/broadening; generated at least 10 spectrally resolvable Rdots with potential to exceed 20.
• Brightness: Relative Raman intensity versus EdU (RIE) exceeded 10^4 for most Rdots, making them the brightest organic-based Raman probes reported; nearly 4× brighter than PDDA despite PDDA’s visible resonance enhancement.
• SRS cross-section: Estimated SRS cross-section for Rdots2220 up to 5 × 10^−16 cm^2, >40× larger than the best organic Raman dyes (MARS) and even higher than typical absorption cross-sections of fluorescent dyes.
• Probe loading: Independent estimates showed thousands of dye molecules per nanoparticle; e.g., ~2.7 × 10^3 Carbow 2-yne per 20 nm Rdots2220, yielding local concentrations ~1.2 M and intermolecular spacing <1 nm. Raman does not quench at this packing density.
• Detection sensitivity: SRS detection limit for Rdots in solution was 900 ± 50 pM (99% CI), with linear response (R^2 = 0.999), representing 2–3 orders of magnitude improved sensitivity over previously reported organic Raman probes.
• Single-particle evidence: Immobilized Rdots produced diffraction-limited spots with spectra matching bulk; intensity histograms at pH 8.5 and pH 6.0 showed quantized peaks corresponding to single-, double-, triple-, and quadruple-particle aggregates. Line profiles exhibited integer multiples of single-particle intensity. Estimated single-particle SNR ~8, consistent with theory.
• Size and stability: DLS showed no size change after doping (~21 nm); PEGylation increased size to ~23 nm. Rdots retained signal with no leakage for at least 5 months in aqueous storage. Photobleaching under SRS was minimal (<10% loss over ≥20 frames).
• Specificity and performance in cells/tissues: Minimal non-specific binding observed. High-contrast SRS images of α-tubulin and vimentin in cells; off-resonance background negligible. Correlative fluorescence/SRS imaging showed strong colocalization (Pearson’s r = 0.88 for microtubules; 0.82 for vimentin) with matching line profiles. In mouse colon sections, e-cadherin staining localized to epithelial cell membranes as expected, with no labeling in submucosa or muscularis externa.
• Sensitivity to low-abundance targets and primary-only staining: Rdots enabled immunostaining using primary antibodies alone (microtubules) and detection of low-abundance membrane protein CD44 with strong SRS signals; MARS probes were much weaker under identical targets.
• Multiplexing: Demonstrated three-color imaging of microtubules (Rdots2177), vimentin (Rdots2220), and actin (phalloidin-Alexa647) with minimal cross-talk, and four-color imaging including nuclei (NucGreen). Rdots channels were compatible with fluorescence imaging.

The study addresses the key bottleneck of Raman-based multiplexed imaging by creating probes that are both ultra-bright and compact, overcoming the trade-off between signal strength and size seen in SERS and polymer-based approaches. Dense volumetric packing of Raman-active small molecules within a hydrophobic polymer matrix provides exceptional SRS cross-sections and RIE values without fluorescence-like quenching or hotspot dependency, enabling sub-nM detection and evidence of single-particle SRS imaging. The compact ~20–23 nm size facilitates diffusion and access to intracellular targets, supporting high-contrast immunostaining of gold-standard cytoskeletal markers and tissue biomarkers with specificity comparable to conventional IFM. Strong colocalization with fluorescence validates staining fidelity, while primary-only staining and low-abundance target detection highlight improved sensitivity. Narrow spectral linewidths underpin multi-channel separation and compatibility with fluorescence channels, demonstrating practical multiplexed imaging. These results position Rdots as quantitative, scalable Raman probes suitable for a range of applications, including highly multiplexed imaging, barcoding, and integration with advanced SRS modalities.

A simple, general swelling–diffusion strategy was developed to produce compact, ultra-bright Raman nanoparticles (Rdots) with exceptional SRS sensitivity, stability, and biofunctionalization versatility. Rdots deliver superior brightness (RIE > 10^4; SRS cross-section up to 5 × 10^−16 cm^2), sub-nM detection, and evidence for single-particle SRS imaging. They enable specific, high-contrast immunostaining in cells and tissues, including primary-only approaches and low-abundance targets, and support multiplexed imaging alongside fluorescence channels. The method is broadly extendable to many small-molecule probes, allowing expansion to >20 resolvable Raman channels. Future work could focus on further size reduction to enhance accessibility, integrating with hyperspectral and high-speed SRS platforms, coupling to nucleic acid probes for in situ hybridization, applying to flow cytometry for multiplex profiling, and combining with expansion microscopy for super-resolution, thereby expanding the impact of Raman-based multiplexed bioimaging.

Although Rdots are relatively small (~20–23 nm), further size reduction would be desirable to improve diffusion and surface accessibility in crowded environments. Narrowband SRS imaging requires sequential acquisition at different Raman frequencies, which can extend imaging time in highly multiplexed experiments. The study focused on fixed cells and thin tissue sections; while diffusion barriers were not problematic under these conditions, performance in thicker or in vivo specimens was not evaluated. Despite minimal observed non-specific binding and aggregation due to PEGylation and surface charge management, such artifacts can be context-dependent and warrant application-specific assessment.