Infrared Fluorescent Proteins (IRFPs)

• Infrared fluorescent proteins (IRFPs) are genetically encodable molecular tags that utilize near-infrared light and biliverdin-based chromophores for enhanced cellular imaging.
• They are engineered from bacterial phytochromes using structure-guided mutagenesis to achieve red-shifted absorption, improved quantum yield, and optimized photostability.
• Recent advances, exemplified by mRhubarb720, demonstrate IRFPs’ capability to achieve high signal-to-background ratios and minimal photobleaching in electronically pre-resonant SRS microscopy.

Infrared fluorescent proteins (IRFPs) constitute a class of genetically encodable molecular tags that utilize the near-infrared spectral region, primarily for advanced imaging modalities such as stimulated Raman scattering (SRS) microscopy. Characterized by their unique protein scaffold and biliverdin-based chromophore, IRFPs are distinguished from classical fluorescent proteins (FPs) derived from jellyfish or coral, offering advantageous optical and vibrational properties for cellular and molecular imaging, particularly in contexts requiring deep tissue penetration and super-multiplex capability. Recent research demonstrates IRFPs, including mRhubarb720, emiRFP670, and miRFP670nano3, as the first genetically encodable probes for electronically pre-resonant SRS (epr-SRS), with performance metrics comparable to state-of-the-art synthetic near-IR dyes (Regan et al., 11 Nov 2025).

1. Molecular Architecture and Chromophore Integration

IRFPs are engineered from bacterial phytochromes and utilize the linear tetrapyrrole biliverdin IXα (BV) as their fluorogenic co-factor. The protein structure comprises a two-domain PAS–GAF fold, distinct from the β-barrel motif of classical FPs such as mCherry. In IRFPs, the GAF domain accommodates BV, with a thioether linkage formed between the C3 carbon of ring A and a cysteine residue (Cys15) (Fig. 1e, Fig. S2a).

Engineering strategies for spectral tuning and enhanced photostability involve structure-guided mutagenesis of residues within the chromophore pocket. For example, the development of mRhubarb720 from iRFP713 entailed key substitutions: L196Q (environment polarization), T202D (solubility improvement), and V203I (chromophore interaction reinforcement) (Fig. S2b–c). These modifications collectively yield red-shifted absorption, improved quantum yield, and optimized photostability.

2. Spectral Characteristics

IRFPs exhibit far-red to near-infrared absorption maxima, high extinction coefficients, and low quantum yields. Table S1 condenses these properties for principal IRFP variants:

Protein	λ_abs (nm)	ε (mM⁻¹ cm⁻¹)
mRhubarb720	703	72.5
emiRFP670	642	23.2
miRFP670nano3	645	31.9

All IRFPs are characterized by φ < 0.15. Emission maxima are typically 20–30 nm red-shifted relative to absorption (e.g., mRhubarb720 emits near 720–740 nm). Measured absorption spectra (Fig. 1c) show these proteins’ distinct spectral profiles compared with mCherry and free biliverdin, confirming the impact of chromophore integration and protein context.

3. Electronic Pre-Resonant SRS Enhancement

In SRS microscopy, coherent excitation of molecular vibrations is accomplished by overlapping “pump” (ω_p) and “Stokes” (ω_S) pulsed laser beams, driving vibrational transitions at Ω_v = ω_p – ω_S. The measured stimulated Raman loss (SRL) in the pump beam is defined by

$$I_{SRS} \propto N \cdot \sigma_{Raman} \cdot I_{pump} \cdot I_{Stokes}$$

where $N$ is the scatterer density and $\sigma_{Raman}$ the scattering cross-section. Proximity of the pump wavelength to the protein’s electronic absorption ($\omega_e$) invokes “electronic pre-resonance” (epr), dramatically enhancing $\sigma_{Raman}$ per Albrecht A-term approximation:

$$\sigma_{Raman}(\omega_p;\Omega_v) \propto \left[ F_A(\omega_p,\Omega_v) \right]^2$$

$$F_A(\omega_p,\Omega_v) = \frac{(\omega_p - \Omega_v)^2 (\omega_e^2 + \omega_p^2)}{(\omega_e^2 - \omega_p^2)^2}$$

Empirically, optimal epr-SRS signal is obtained for chromophores with the pump detuned by 2–6 homogeneous linewidths from $\omega_e$. A simplified scaling is $\sigma_{Raman} \propto (\omega_e - \omega_p)^{-4}$, favoring minimal detuning for higher cross-sectional enhancement.

With mRhubarb720 (λ_abs ≅ 703 nm, ω_e), SRS was performed using λ_pump = 820 nm, establishing a detuning Δω ≈ 1790 cm⁻¹. This configuration suppresses one-photon absorption and photobleaching rates by ∼10⁴, attributed to the Boltzmann thermal tail exp(–ħΔω/k_BT) (Fig. S7). Two-photon absorption (TPA) artifacts are subtracted from SRL spectra (Fig. 2, Fig. S8–S10).

4. Genetic Encoding and Cellular Imaging Methodologies

IRFPs enable targeted genetic labeling for SRS imaging via fusion constructs. The protocol for nuclear targeting employed mRhubarb720 cloned downstream of human histone H2B, expressed in HeLa cell culture (Fugene6 transfection, paraformaldehyde fixation). Imaging utilized 820 nm pump and tunable Stokes pulses (centered at 948 nm for a 1640 cm⁻¹ vibrational resonance), raster-scanned through a 60×1.27 NA water-immersion objective.

SRL images at Ω_v = 1620 cm⁻¹ (specific to mRhubarb720) and 1660 cm⁻¹ (lipid C=C background) were acquired with 0.01 ms pixel dwell time over 50×50 μm (463×463 pixels, ~2 s per frame). Simultaneous epi- and forward fluorescence (600–700 nm pass) allowed direct comparison and localization. SRL and fluorescence confirmed subcellular targeting, with H2B-mRhubarb720 defining nuclear topology and lipid droplets visualized by vibrational contrast (Fig. 3b–g).

5. Performance Metrics and Comparative Evaluation

Quantitative performance assessments are as follows:

• Detection limit (solution): mRhubarb720 at 1 mM yields ΔI/I = 2.6×10⁻⁵ (pump 6 mW, Stokes 15 mW). Shot-noise limited sensitivity (~1 ms pixel) is ΔI/I ≅ 1.9×10⁻⁷, with a concentration detection threshold ≃ 7.2 μM. At elevated powers (pump 24 mW, Stokes 120 mW), ≃ 452 nM, closely matching ATTO740 (250 nM).
• Signal-to-background ratio: SRL S/B > 100:1, parallel to synthetic dyes (Fig. 2a, Fig. S11b).
• Spatial resolution: Lateral ≃ 0.61λ_pump/NA ≃ 400 nm; axial ≃ 2λ/NA² ≃ 1.2 μm.
• Photostability: Both SRS and fluorescence signals bleach over ~10 frames (1 ms exposure per nucleus pixel), dominated by TPA and excited-state absorption (Fig. 3h, Fig. S13). Triplet state formation accelerates bleaching under femtosecond pulses.

Comparison between mRhubarb720 and mCherry revealed a ~10× higher SRS cross-section in the former, corresponding to its more favorable pre-resonant detuning (Δω = 2030 cm⁻¹ vs 4754 cm⁻¹ for mCherry; Fig. 2c).

6. Challenges and Future Perspectives

Several technical challenges and avenues for advancement remain:

• Photobleaching: Rates are elevated under pre-resonant excitation. Mitigation strategies include reduced laser repetition rates, decreased pulse counts per pixel, expedited scanning, and protein scaffold modifications to lower TPA cross-sections.
• Multiplexed vibrational palette: Expansion to multiplex imaging necessitates genetically encoded tags with distinct vibrational markers (e.g., nitriles, alkynes via unnatural amino acids).
• Spectral coverage extension: SRS pump wavelengths in the visible region could facilitate epr-SRS of blue-, green-, and red-absorbing FPs, broadening vibrational imaging capabilities.
• Probe optimization: Future engineering efforts will focus on heightened quantum yield, improved extinction coefficients, and robust photostability during femtosecond-pulse excitation, while maintaining narrow vibrational linewidths essential for super-multiplex imaging.

A plausible implication is that IRFPs, in particular mRhubarb720, constitute the first genetically encodable epr-SRS probes capable of yielding signals on par with the best near-IR synthetic dyes. This marks a pivotal advance toward genetically targetable, vibrationally multiplexed cell imaging (Regan et al., 11 Nov 2025).