Extending optical chemical tools and technologies to mice by shifting to the shortwave infrared region

Abstract

Fluorescence imaging is an indispensable method for studying biological processes non-invasively in cells and transparent organisms. Extension into the shortwave infrared (SWIR, 1000–2000 nm) region of the electromagnetic spectrum has allowed for imaging in mammals with unprecedented depth and resolution for optical imaging. In this review, we summarize recent advances in imaging technologies, dye scaffold modifications, and incorporation of these dyes into probes for SWIR imaging in mice. Finally, we offer an outlook on the future of SWIR detection in the field of chemical biology.

Introduction

The ability to monitor dynamic processes without perturbing living systems is a hallmark of chemical biology. A common approach toward visualizing biological processes in real time is fluorescence imaging. While many fluorophores and technologies are available for imaging cells and transparent organisms [1–3], the use of optical imaging in mammals is limited by background autofluorescence and poor light penetration through tissue.

Recent developments in small animal imaging employ detection in the “biologically transparent” shortwave infrared region of the electromagnetic spectrum (SWIR or NIR-II, 1000–2000 nm). This region provides in vivo optical imaging with sub-millimeter resolution [4]. The benefits of the SWIR region were initially showcased with nanomaterials [5–7], which have since been optimized for use as bright biosensors [8,9] and contrast agents [10,11]. However, a recent explosion of interest in SWIR-emissive small molecule fluorophores has paralleled SWIR nanoparticle development and facilitated the creation of minimally perturbing, non-toxic SWIR probes. These optimized SWIR fluorophores, coupled with instrumentation advancements, will render ubiquitous chemical biology techniques more effective in animals.

SWIR imaging technologies and fluorophores

Traditional fluorescence imaging utilizes the visible (VIS, 400–700 nm) and near infrared (NIR, 700–1000 nm) regions of the electromagnetic spectrum (Figure 1a). The transition from these regions to the SWIR region results in favorable imaging conditions including reduced fluorescence from endogenous chromophores, decreased light scattering by tissue [4,12,13], and increased image contrast caused by water absorption [14]. These properties result in greater imaging depths and/or resolutions compared to VIS/NIR fluorescence imaging and improve in vivo imaging in mammals. We highlight the main advances in technology and fluorophore development necessary to leverage the SWIR region for chemical biology.

Figure 1.

Established technologies for SWIR bioimaging. a. Electromagnetic spectrum with defined visible (VIS), near infrared (NIR) and shortwave infrared (SWIR) regions. Spectral representations of b. Traditional single-color imaging, c. Fluorescence tail imaging, d. Emission-multiplexed imaging, and e. Excitation-multiplexed imaging. Absorption and emission traces are indicated by shaded and dotted curves, respectively. f. Generalized whole-animal imaging setup. Schematized g. confocal and h. light-sheet microscopy setups for three-dimensional imaging.

Technologies for SWIR bioimaging

Over the past decade, many imaging technologies have been adapted for the SWIR region. The major advance in SWIR detection has been the availability of indium gallium arsenide (InGaAs) detectors which are sensitive to 950–1700 nm light [15]. Challenges of this region such as the low fluorescence quantum yield (Φ_F) of SWIR contrast agents, and the differing depths or resolutions that can be achieved depending on detection wavelengths, provide additional considerations for technology development [16,17].

Single-color SWIR imaging

Simple single-color whole mouse imaging in the SWIR was first reported in 2009 by Dai and coworkers [5]. This report showcased the advantageous properties of the SWIR region; however, imaging rates were limited by the brightness of the carbon nanotube contrast agents. The brightness of a contrast agent in an imaging experiment is quantified as the product of the absorption coefficient (ε) at the wavelength of excitation and the Φ_F at the wavelengths collected. Due to the inherently low Φ_F values in SWIR dyes [17], efficient excitation is crucial for maximized detection. Optimal single-color imaging is achieved via excitation near the wavelength of absorbance where the ε is largest (λ_max,abs). This fact has prompted the development of fluorophores with λ_max,abs near the readily available 1064 nm laser line (see Dye scaffolds for SWIR bioimaging).

Ideally, a detection window is utilized which includes the emission peak of the fluorophore (λ_max,em) (Figure 1b); however, some NIR dyes with broad or tailing fluorescence profiles can be used as SWIR contrast agents by detecting the tail region of their emission spectra (Figure 1c) [18–20]. A notable example is indocyanine green (ICG), a clinically approved NIR fluorophore with a SWIR tail [21] which has facilitated in vivo SWIR imaging [18,19] and image-guided surgery [22].

Multicolor SWIR imaging

Multicolor imaging is important for studies involving ratiometric detection and (co)localization. A major asset of the SWIR window is that it extends the range of wavelengths available for fluorescence imaging; thus, employing this region benefits studies where the detection of multiple contrast agents is necessary. Multicolor imaging often utilizes one excitation wavelength for multiple contrast agents which are detected in separate emission windows (Figure 1d). The first multicolor SWIR work detected three contrast agents in mouse brain tissues across different NIR and SWIR windows [23]. Other early works reported in vivo multicolor imaging of ovaries [24] and xenograft tumors [25–27] using different classes of contrast agents and/or image processing procedures.

Alternatively, multicolor imaging is achieved through orthogonal excitation of dyes with spectrally separated absorbance profiles (Figure 1e). Cosco et al. reported excitation-based multiplexed imaging where dyes are excited using distinct laser lines and emission is detected in one SWIR window, removing differences in depth/resolution between channels. This setup maximizes signal through excitation near λ_max,abs and produced video-rate images in mice in three [28] and four colors [29]. Another advantage of excitation multiplexing is that crosstalk between fluorophores only occurs in one direction. In the initial report [28], this crosstalk was removed via subtraction; in the subsequent report [29], optimization of the dosing or laser power reduced the need for unmixing. Zhou and coworkers reported a similar excitation-based approach for multicolor imaging of lanthanide-doped nanoparticles [30].

2D and 3D SWIR imaging

Whole animal imaging (Figure 1f) allows for real-time micron resolution images but does not provide detailed tomographic or cellular information. For these applications, confocal and light-sheet microscopy (LSM) are necessary. Early reports of home-built SWIR confocal microscopes (Figure 1g) incorporated SWIR-compatible detectors to achieve 3D renderings of mouse brain tissue [23], ovaries [24], chloroplasts [31], tissues ex vivo [26], and cerebrovasculature in animals [26,32–34].

Complementary to point-by-point confocal imaging, LSM functions by illuminating samples with planar sections of light (Figure 1h). However, the use of widefield detection limits the resolution of LSM due to light scattering. Dai and coworkers overcame this limitation by constructing a SWIR LSM system [35], which was further optimized to perform structured illumination microscopy on xenograft tumors with high contrast and resolution [36].

Dye scaffolds for SWIR bioimaging

A major limitation of early SWIR imaging efforts was the dearth of small molecule fluorophores for the SWIR region. Indeed, many of the technological advances highlighted above were first demonstrated with nanomaterials. The ideal SWIR fluorophore has λ_max,em>1000 nm, a narrow NIR or SWIR absorption band with high ε, and a large Stokes shift with a Φ_F that is optimized for the region of the SWIR in which the fluorophore emits. It is difficult to pinpoint an exact Φ_F threshold for SWIR fluorophores due to the significant differences in Φ_F predicted by the energy gap laws [17] and the large role that solvent plays [37,38]. Note that Φ_F values are often increased in organic solvents where non-radiative pathways facilitated by H-bonding and O–H vibrational modes are minimized. In this section, we highlight the three key scaffolds that have produced the majority of SWIR fluorophores to date.

Donor-Acceptor-Donor dyes

Donor-Acceptor-Donor (D-A-D) fluorophores are composed of two electron donating moieties flanking an electron-poor core. The first small molecule for bioimaging beyond 1000 nm was reported by Dai and coworkers who PEGylated D-A-D chromophore M1 (1, Figure 2a) previously prepared for organic lightemitting diodes [39] to yield CH1055-PEG (2, Figure 2a). CH1055-PEG has a broad NIR absorbance and large Stokes shift (305 nm) and yielded high resolution single-color lymphatic and cerebrovascular imaging with 808 nm excitation [40]. While CH1055-PEG was a significant milestone for the field, major improvements to the D-A-D scaffold have occurred. Yang and coworkers replaced the classic sulfur-containing benzobisthiadiazole acceptor with its selenium congener to access longer wavelengths (λ_max,abs=820 nm, λ_max,em=1080 nm) [41]. Donor moieties can also modulate D-A-D performance through backbone distortion [41–44]. Significant improvements in Φ_F were obtained upon inclusion of shielding (S) groups to afford S-D-A-D-S fluorophores (Figure 2a) [42–44]. Although much work has been performed to optimize D-A-D dyes for SWIR bioimaging, the brightness of these fluorophores remains restricted by low ε values [19].

Figure 2.

Tuning of pre-existing a. donor-acceptor-donor, b. polymethine, and c. aza-BODIPY scaffolds into SWIR-emissive dyes for improved imaging in vivo. All values were calculated relative to IR-26 (Φ_F = 0.05% in dichloroethane) as a reference unless otherwise noted. ^aIR-26 (Φ_F = 0.11% in dichloroethane) used as a reference. ^bMagnesium tetra-tert-butyl phthalocyanine (Φ_F = 84% in chloroform) used as a reference. ^cIR-125 (Φ_F = 23% in DMSO) used as a reference.

Polymethine dyes

Polymethine dyes consist of two heterocycles connected by a namesake methine linker. Polymethine fluorophores often have decreased Stokes shifts, photo-stability, and Φ_F values compared to D-A-D dyes. However, their high ε, narrow absorption bands, scaffold tunability, and extended emission tails have made them excellent SWIR emitters [45].

The exceptional absorption properties of polymethine dyes are evident in the success of ICG (3, Figure 2b) as a SWIR contrast agent even when only 5% of its emission is collected [19,21]. Strategies to red-shift 3 into the SWIR through both heterocycle [46–48] and polymethine chain modification [49,50] have been performed. Notable heterocycle advances include: 1) the benzo[c,d]indolium heterocycle, first reported by Henary et al. [46] and named/applied to in vivo imaging by Zhang and coworkers (4, FD-1080, Figure 2b) [47]; 2) appending extra indolium heterocycles to produce tetra-indolium dye LZ-1105 [48]. Polymethine dye extension has resulted in variants of 3 (ICG-C11) [49] and aryl-fused nonamethine dyes (FNIR-1072, 5, Figure 2b) [50]. Fluorophores 4, 5, and LZ-1105 can all be imaged with excitation at 1064 nm. Schnermann and coworkers demonstrated three-color excitation-multiplexed imaging using 3, 5, and nonamethine FNIR-872 [50].

The use of heteroatoms other than nitrogen in polymethine dye heterocycles has also been fruitful towards the creation of SWIR fluorophores. This work builds off thiaflavylium dye IR-26 (6, Figure 2b) and thiapyrylium dye IR-1061. Optimization of sulfur-containing dyes has yielded pentamethine dyes BTC1070 [51] and 5H5 [52]. While sulfur provides impressive red-shifts, non-radiative pathways may be facilitated by heavy atoms. Consequently, oxygen-containing heterocycles have been explored. The first polymethine designed for SWIR imaging, Flav7 (7, Figure 2b), was inspired by 6 with replacement of sulfur with oxygen and the addition of an electron donating dimethylamino group. Flav7 exhibited a Φ_F ten-fold greater than 6 while maintaining λ_max,abs>1000 nm [53]. Further optimization has yielded fluorophores that can be excited at 1064 nm (JuloFlav7, λ_max,abs=1061 nm) [28] and fluorophores with increased brightness (Chrom7, Φ_F=1.7%) [29] (Figure 2b). Zhang and coworkers have also developed a series of chromenylium dyes coined CX dyes with impressive chemo- and photostability [54].

BODIPY and aza-BODIPY dyes

Boron-dipyrromethene (BODIPY) difluoride dyes and their derivatives have recently gained interest as SWIR emitters. BODIPYs display excellent Φ_F values and stability and can be red-shifted through nitrogen incorporation [55]. First synthesized in 2002 as promising phototherapeutic agents [56] (8, Figure 2c), aza-BODIPYs have high ε, large Stokes shifts, and λ_max,abs values spanning the NIR region [55–57]. Two teams concurrently reported the first SWIR aza-BODIPY contrast agents in 2019. Huang and coworkers appended electron donating moieties to the aza-BODIPY core to produce D-A-D-like dyes NJ1030 and NJ1060 (λ_max,em=1030 and 1060 nm) [58]. In contrast, Sancey et al. synthesized blue-shifted aza-SWIR-01 (λ_max,em=980 nm, 9, Figure 2c). Ammonium groups were appended to 10 to produce water-soluble SWIR-WAZABY-01 (10, Figure 2c), which was employed for single-color emission tail imaging in mice [59].

Small molecule SWIR probes

A prominent avenue of chemical biology is modifying fluorophores such that they provide a functional readout on a biological system as fluorescent “probes”. Initial reports toward SWIR probes are highlighted below.

Targeted and molecular imaging probes

Targeted probes consist of fluorophores conjugated to agents which label specific biomolecules, metabolites, or tissues. The first reported small molecule SWIR molecular imaging probe consisted of D-A-D dye CH1055 conjugated to an anti-EGFR affibody [40]. S-D-A-D-S probes have also been conjugated to chorionic gonadotropin and poly-l-lysine for hormone receptor imaging [24] and dendritic cell tracking (Table 1, entry 1) [60], respectively. Additionally, Tan and coworkers used a tetraazido D-A-D dye [23] to label D-propargyl glycine in gut microbiota via ex vivo click chemistry (11, Figure 3a; Table 1, entry 2). The labeled microbes were administered back to mice via gavage and imaged [61].

Table 1.

Characteristics of recently reported D-A-D, polymethine, and BODIPY-based SWIR probes.

Entry	Probe	Mechanism	Target/trigger	λmax,abs/λmax,em (nm)	ΦF (%)	Ref.
D-A-D
1	TTQ-PLL@OVA	Molecular imaging	Dendritic cells	740/1050	1.24a (water)	[60]
2	IR-FGN	Molecular imaging	D-propargyl glycine	745/1050	1.9b (water)	[23,61]
3	NIR-II-CLS	Ratiometric	Peroxynitrite	540/680; 700/985	N.A.	[66]
Polymethine
4	ICG-C11-Herceptin	Molecular imaging	HER2	1021/1058	1.3c (DMSO)	[49]
5	FNIR-1072-Dex	Molecular imaging	Lymphatic system	1072/1103	0.055d (10% FBS/PBS)	[50]
6	5H5-PEG8-cRGDfk	Molecular imaging	αvβ3 integrin	1069/1125	0.26e (acetonitrile)	[52]
7	QT-RGD	Molecular imaging	αvβ3 integrin	1042/1068	0.12e (DMSO)	[62]
8	CEAF-RGD	Molecular imaging	αvβ3 integrin	980/1017	0.20 (dichloromethane)	[63]
9	IRBTP-B	Fluorogenic	Peroxynitrite	785/970	0.10 (methanol)	[67]
10	CEAF-OMe	Fluorogenic	Endocytosis, pH	980/1017	0.20 (dichloromethane)	[63]
11	PN1100	Ratiometric	Peroxynitrite	883f/920; 1089/1140	0.66f; 0.091 (chloroform)	[54]
12	DSNP@MY-1057	Ratiometric	Peroxynitrite	N.A./1060; 1057/1081	N.A.; 0.71 (DMSO)	[68]
BODIPY
13	BTC1070	Ratiometric	pH	~600–900/1015; 980/1,065g	N.A.; 0.016g (PBS)	[51]
14	NTR/NQO/ALP-InD	Fluorogenic	NTR, NQO, ALP	730/900	0.048h (DMSO/TRIS-HCl)	[69]
15	BOD-M-βGal	Fluorogenic	β-galactosidase	723/853	N.A.	[70]
16	WD-NO2	Fluorogenic	Viscosity	818/982	0.3128c (glycerol)	[71]

Photophysical values reflect dyes prior to nanoparticle formulation or conjugation with targeting agents unless otherwise stated. Values for fluorogenic probes reflect the emissive state. For ratiometric probes, values for each fluorescent component are listed in order of increasing λ_max,abs. To facilitate probe comparison, all Φ_F values were calculated to be referenced to IR-26 Φ_F = 0.05% in dichloroethane or IR-1061 Φ_F = 0.32% in dichloromethane. All instances where a recalculation was performed due to a different reference in the original report are noted. Otherwise reported values were calculated relative to IR-26 Φ_F = 0.05% in dichloroethane).

^aIR-1061 used as a reference but no Φ_F or solvent reported for the reference.

^bHiPco carbon nanotubes used as a reference.

^cRecalculated with IR-1061 (Φ_F = 0.32% in dichloromethane) as a reference.

^dRecalculated with IR-26 (Φ_F = 0.05% in dichloroethane) as a reference.

^eListed value is for dye–peptide conjugate.

^f1% trifluoroacetic acid was added to obtain these values.

^gDetermined from dye loaded into phospholipid micelles. First and second set of values refer to probe below and above pH 4, respectively.

^hA reference fluorophore was not reported for Φ_F calculations.

Figure 3.

Targeting and activation mechanisms for select SWIR probes. a. Conjugation of IR-FGN to metabolically incorporated D-propargyl glycine facilitates in vivo tracking of gut microbiota. © 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. b. IRBTP-B fluorescence is activated through boronate ester cleavage by peroxynitrite and spontaneous elimination of p-quinone methide. This probe was used for detection in an acetaminophen-induced livery injury model. Reprinted (adapted) with permission from ref 67. Copyright 2019, American Chemical Society. c. Protonation of CEAF-OMe in lysosomes disrupts aggregation induced quenching, yielding bright SWIR fluorescence for image-guided surgery. Scale bar represents 2 cm. Reproduced from ref 63 with permission from the Royal Society of Chemistry. d. Reduction of the nitro group and spontaneous elimination of a 2,5-cyclohexadiene imine activates nitroreductase-responsive probe NTR-InD in a xenograft mouse model. Reproduced from ref 69 with permission from the Royal Society of Chemistry. e. Rigidification of a piperidinyl moiety favors fluorescence of probe WD-NO₂ in a high viscosity liver models induced by monensin (Mon), nystatin (Nys), and lipopolysaccharide (LPS). Reprinted (adapted) with permission from ref 71. Copyright 2020, American Chemical Society.

Polymethine dye conjugates have also been utilized as targeted imaging agents, with most relying on antibodies or cyclic RGD peptide (cRGD) for targeting. A IRDye 800CW-trastuzumab conjugate was used by Carr et al. for SWIR tail imaging of xenografts in 2018 [19]. Both ICG-C11 and the nonamethine FNIR-872 have been conjugated to antibodies for labeling overexpressed receptors in xenograft tumors (Table 1, entries 4, 5) [49,50]. The xenograft imaging with FNIR-872 was paired with a nonamethine-dextran conjugate (Table 1, entry 5) and 3 to simultaneously observe the surrounding lymphatics and vasculature [50]. α_vβ₃ integrin targeting has also been a popular approach for targeted imaging with polymethine dyes including thiapyrylium (Table 1, entry 6) [52], benzo[c,d]indolium (Table 1, entry 7) [62], and thiaflavylium (Table 1, entry 8) [63] dye-cRGD conjugates.

Fluorogenic and ratiometric probes

Compared to “always-on” dye conjugates for molecular imaging, responsive fluorescent probes display higher specificity. Fluorogenic or “turn-on” probes display enhanced Φ_F in response to a stimulus, while ratiometric probes can monitor subtle changes to stimuli by using ratios between multiple signals. In the past three years, SWIR dyes have been developed into a number of stimuli-responsive probes for in vivo detection.

Metabolite detection in the SWIR region has enabled diagnostics in animal disease models. Early reports of metabolite-responsive SWIR probes include BODIPY-based fluorogenic probes for H₂S detection [64] and a hydroxyl radical-reactive fluorogenic cyanine probe [65]. A recent development in ratiometric SWIR probes is a micellar hydrogen peroxide-activated probe by Yang et al. (Table 1, entry 3) which involves chemiluminescence energy transfer from H₂O₂-generated dioxetanedione to a D-A-D dye, followed by Förster resonance energy transfer (FRET) to a second D-A-D fluorophore [66]. Multiple SWIR probes have also been developed for peroxynitrite (ONOO⁻) detection. Lei et al. found that polymethine dyes CX-1 and CX-3 formed a highly efficient FRET pair when encapsulated in micelles, with the latter being susceptible to oxidative degradation by ONOO⁻ (Table 1, entry 11) [54]. Li et al. developed polymethine probe IRBTP-B (12, Figure 3b; Table 1, entry 9) by including a phenyl boronate ester which yields a non-emissive species until uncaged by ONOO⁻ [67]. Finally, a carboxylated LZ-1105 derivative (Table 1, entry 12) was paired with lanthanide-doped nanoparticles to detect ONOO⁻ levels [68].

Two pH-sensitive thiapyrylium probes were recently designed by Zhang and coworkers. One probe exhibited ratiometric fluctuations in emission intensity based on protonation states of the dimethylamino substituents (Table 1, entry 13) [51]. Another employed PEGylated dye aggregate CEAF-OMe (13, Figure 3c; Table 1, entry 10), which exhibited aggregation-induced quenching when neutrally charged and displayed a 72-fold signal increase upon endocytosis [63].

Dye precursor-enzyme substrate conjugates have been reported as enzyme-responsive SWIR probes. Wang et al. created three indolium-substituted BODIPY probes (Table 1, entry 14) which were conjugated to substrates for nitroreductase (NTR), NAD(P)H quinone oxidoreductase isozyme 1 (NQO), or alkaline phosphatase (ALP) via a benzothioether linker. In a mechanism analogous to 12, substrate cleavage and linker elimination resulted in a dye with SWIR tail emission (14, Figure 3d) [69]. A similar β-galactosidase-responsive probe was also developed (Table 1, entry 15) [70].

Unlike aforementioned probes, fluorescence of BODIPY probe WD-NO₂ (15, Figure 3e; Table 1, entry 16) is activated by a physical stimulus. This probe by Liu and coworkers possesses a rotating piperidinyl “spin group” which rigidifies in viscous environments such as inflamed tissues (Figure 3e). Rigidification favors fluorophore backbone conjugation to yield a ~5-fold increase in brightness in vivo [71].

Conclusions and outlook

Expansion of optical imaging into the SWIR window has revolutionized non-invasive imaging and detection within mammals. Much work has been done to adapt pre-existing technologies for SWIR imaging. Concurrently, the development of bright, robust fluorophores is necessary for chemical biology studies in vivo. As highlighted here, significant work has been performed to create such contrast agents for the SWIR window. The tuning of D-A-D, polymethine, and aza-BODIPY scaffolds has resulted in SWIR dyes with enhanced brightness and utility in living systems. Recent work transforming these SWIR dyes into fluorogenic and ratiometric probes demonstrates great potential for the creation of chemical tools that can be visualized non-invasively in mice. This work aims to bridge the gap between fundamental studies in cells and translational research in mammals.

Still, there is ample space for the development of SWIR technologies and probes. Techniques such as multi-photon excitation [72] are now being adapted for the SWIR region. Opportunities remain for expansion and/or optimization of such advanced optical imaging technologies for this window. Further development and availability of SWIR-compatible microscopes, flow cytometers, and plate readers is also essential to fully integrate the SWIR region into standard chemical biology procedures. Other fluorophore scaffolds such as xanthene [73] and squaraine [74,75] are showing potential as SWIR contrast agents. We anticipate further advancements in SWIR probes including ion sensing, fluorogens, and turn-on FRET probes, which will be powerful additions to the optical imaging toolbox for non-invasive detection in mammals.

Abbreviations

SWIR

shortwave infrared

VIS

visible

NIR

near infrared

ICG

indocyanine green

InGaAs

indium gallium arsenide

LSM

light-sheet microscopy

D-A-D

donor-acceptor-donor

PEG

polyethylene glycol

EGFR

epidermal growth factor receptor

FRET

Fӧrster resonance energy transfer

NTR

nitroreductase

NQO

NAD(P)H:quinone oxidoreductase isozyme 1

ALP

alkaline phosphatase

Mon

monensin

Nys

nystatin

LPS

lipopolysaccharide

FBS

fetal bovine serum

PBS

phosphate buffered saline

DMSO

dimethyl sulfoxide