Near-infrared fluorescence imaging in immunotherapy

Abstract

Near-infrared (NIR) light possesses many suitable optophysical properties for medical imaging including low autofluorescence, deep tissue penetration, and minimal light scattering, which together allow for high-resolution imaging of biological tissue. NIR imaging has proven to be a noninvasive and effective real-time imaging methodology that provides a high signal-to-background ratio compared to other potential optical imaging modalities. In response to this, the use of NIR imaging has been extensively explored in the field of immunotherapy. To date, NIR fluorescence imaging has successfully offered reliable monitoring of the localization, dynamics, and function of immune responses, which are vital in assessing not only the efficacy but also the safety of treatments to design immunotherapies optimally. This review aims to provide an overview of the current research on NIR imaging of the immune response. We expect that the use of NIR imaging will expand further in response to the recent success in cancer immunotherapy. We will also offer our insights on how this technology will meet rapidly growing expectations in the future.

Graphical Abstract

1. Introduction

The optophysical parameters of near-infrared (NIR; 650-1,700 nm) light have been used extensively in the field of medicine for diagnostic and therapeutic purposes [1-4]. Amongst all the imaging modalities, NIR light is of particular importance due to its unique physical properties suitable for optical imaging of biological tissues [2,5]. As such, NIR fluorescence imaging in the clinical setting has improved therapeutic outcomes for patients [6,7]. NIR imaging could overcome the past challenges of conventional imaging methodologies including cytotoxicity to host cells, interference from background radiation, and high cost. It has been employed for real-time in vivo monitoring of the localization, dynamics, and function of immune responses, whereas conventional imaging methods fall short in these abilities. Since then, this imaging modality has provided valuable pieces of information on not only the efficacy, but also the safety of immunotherapy. In this review, we intend to provide an overview of the current research on the use of NIR imaging of the immune response.

2. Photophysical properties of NIR light

NIR light offers several clear advantages over other wavelengths for use in imaging of biological tissues. NIR light is nonionizing radiation and poses no risk of tissue damage or genotoxicity [6-9]. In addition, the preparation of molecular probes and detection hardware for NIR radiation is far less expensive than those for ionizing radiation [8,9]. The imager should be not only cost-efficient but also relatively simple to operate. NIR laser technology has been used in the field of medicine for the past three decades, and its safety and methodology are well established [10]. Imaging using NIR light is therefore low-cost and simple, making it readily applicable in clinics.

NIR-I (650-900 nm) (Fig. 1) has relatively low light scattering and absorption by biomolecules with deep tissue penetration compared to visible light (400-650 nm) (Fig. 2) [11-15]. In addition, due to the low autofluorescence of biologic tissues at these wavelengths, NIR-I imaging produces a high signal-to-background ratio (SBR) in living organisms compared to imaging using visible light. These characteristics together enable high sensitivity and resolution imaging in the NIR-I region. NIR imaging is therefore an attractive modality among potential optical imaging technologies. The second window of NIR (NIR-II), in the range of 1,000-1,700 nm (Fig. 1), shows similar characteristics to NIR-I but with less scattering in the neighboring tissue [11]. Use of this window has several advantageous properties, including extremely low autofluorescence, minimal tissue absorption and scattering allowing for deeper tissue penetration (Fig. 2), and high resolution and contrast in images compared to visible or NIR-I light [16-18]. Indeed, Antaris et al. synthesized small molecule fluorophores that emit photons in the NIR-II window and demonstrated them for molecular cancer imaging after conjugating with anti-EGFR affibody [19].

Fig. 1. NIR optical window.

The absorption spectrum of human skin showing the first (NIR-I) and second (NIR-II) biological windows. Note that NIR imaging allows low light absorption and scattering into tissue, favoring its application in biological tissues. Modified from Son et al. [5] with permission from Elsevier.

Fig. 2. NIR fluorescence imaging of biological tissue.

NIR light has deeper tissue penetration and lower background fluorescence than visible light. The detection depths with the concurrent optical imaging equipment typically range from millimeters with NIR fluorescence to micrometers with visible-range fluorescence. Modified from Owens et al. [11] with permission from John-Wiley.

3. Labeling tools for NIR imaging

To study the effect of immunotherapy, noninvasive and real-time visualization and subsequent analysis of responses from critical players in the immune system are often crucial. To this end, a variety of fluorescent probes from small molecules to nanoparticles have been developed for use in NIR imaging of cancer and immunotherapy (Fig. 3). The basic principle of molecular imaging probe development is to make them biocompatible and nonimmunogenic for use in biomedical applications suitable for clinical imaging of immune cells in order to emerge to meet this ever-increasing demand.

Fig. 3. NIR imaging in cancer immunotherapy.

Small-molecule fluorophores, nanoparticles, and targeted and activatable probes as labeling tools have been developed for the NIR window, which can directly label immune cells including T cells, B cells, macrophages, dendritic cells (DC), and natural killer (NK) cells, subsequently targeting the cancer cell and ultimately playing a critical role in cancer immunotherapy.

3.1. Small-molecule fluorophores

Small molecule fluorophores have several advantages over other forms of fluorescent probes. These agents have well-established routes of synthesis, well-defined molecular structures, high batch-to-batch consistency, and an easy means of maintaining purity that can be difficult to achieve in other probes [19-21].

For NIR-I imaging, small molecule fluorophores including polymethine cyanines, phthalocyanines, porphyrin derivatives, squaraine derivatives, BODIPY analogs, benzo[c]heterocycles, and xanthenes are commonly used as labeling probes [22]. The most representative NIR fluorophore is indocyanine green (ICG), despite of its undesirable optical properties for in vivo imaging including short blood circulation, serum instability, high liver uptake, and nontargetability [1]. Since ICG is the only NIR fluorophore approved by the US Food and Drug Administration (FDA) for clinical NIR imaging, it has been extensively used for monitoring cardiac output, hepatic function, and retinal angiography without appreciable adverse effects [23]. Tetrapyrrole-based compounds, such as porphyrins, chlorins, benzochlorins, phthalocyanines, and expanded porphyrins, have been widely used as photosensitizers for photodynamic therapy (PDT) [24]. These PDT agents are also used for diagnostic purposes in humans because of their fluorescent optical properties in both the visible and NIR regions [25].

The development of small organic molecule fluorophores for NIR-II imaging is an emerging technology [26]. Several types of fluorophores with relatively favorable toxicity profiles and excretion pharmacokinetics have been reported, but polymethine cyanines and donor-acceptor–donor (D-A-D) structures with a benzothiadiazole (BBTD) core have been mainly explored to date [26,27]. Additionally, long non-negligible emission tails of some NIR-I fluorophores reaching past 1,000 nm have been characterized [28], allowing the existing clinically approved fluorophore ICG to be suitable for NIR-II imaging of superficial organs including vessels and lymphatics [29]. Recently, the Chen group at Stanford tested ICG-based NIR-II imaging in the clinic and demonstrated its advantages, inferring that combining the NIR-I/II spectral windows could improve image-guided cancer therapy with higher accuracy and sensitivity of cancerous tissues [3]. This is the first demonstration of the integrated NIR-I/II multispectral imaging system in the clinical setting, which opens a new door for its broad future applications to many clinical disciplines in dermatology, ophthalmology, neurology, and gynecology, and oncology, along with its important impact to molecular imaging-based therapy [1].

3.2. Nanoprobes

Nanomaterials are another option for NIR fluorescence imaging [30]. There are a wide variety of nanoprobes and substances tuned to emit NIR fluorescence. These materials include, but are not limited to, semiconductor nanocrystals, quantum dots (QDs), metal nanoshells, and rare-earth-doped nanoparticles (RENPs) [31]. Small organic fluorophores can form a nanoprobe when they are self-assembled into a nanoparticle through nanoprecipitation [32]. Although this approach has several notable disadvantages including complexity of the design, high cost, difficulty in large-scale production, and, most importantly, the unknown long-term toxicity to biological systems [33], nanoparticles are generally more amenable to broader approaches for bioimaging and tissue targetability compared to small molecules.

Metal nanoshells including gold possess unique optophysical properties to produce NIR fluorescence depending on their geometry, a phenomenon known as surface plasmon resonance (SPR) [34]. However, there is a significant concern about their long-term toxicity which may be caused by inefficient clearance and accumulation in non-targeted organs. QDs, small semiconductor particles, show superior optical performance with exceptional stability [31]. However, the heavy metal components with a high propensity for toxicity present a large obstacle to the clinical application of this technology. RENPs consist of a core structure comprised of the host substance and dopants surrounded by a shell with the undoped host material [35]. RENPs show the outstanding optical properties of lanthanide ions, favorable biocompatibility, photostability and low cytotoxicity, and they are therefore promising candidates for NIR bioimaging [36]. Lanthanide nanostructures, such as Gd₂O₃: Er³⁺, and Yb³⁺, can serve as optomagnetic markers allowing the combination of optical imaging and magnetic resonance imaging (MRI).

New classes of emitters in the NIR-II range have been explored to improve SBR upon imaging. Carbon nanotubes, RENPs, and QDs have been developed with excellent photostability and relatively rapid clearance [37,38]. Semiconducting single-walled carbon nanotubes (SWNTs) are quasi-one-dimensional materials exhibiting photoluminescence in the NIR range [38,39]. SWNTs generally show poor solubility and biocompatibility, but surface modification has been explored to resolve these issues with some success toward clinical applications [38]. QDs are optically versatile and can be tuned to emit a wide range of fluorescence, which can extend into the NIR-II window [31,38]. RENPs show multicolor NIR-II emissions [36,38], offering a new promising tool for NIR-II bioimaging.

3.3. Targeted probes for NIR imaging

Targeted NIR imaging agents can generally be divided into the following categories: small molecule, antibodies and peptides, nanoparticles, and protein complex probes (Fig. 4) [40]. Antibodies and peptides are useful in designing targeted imaging probes due to their high selectivity for a specific target [41]. This characteristic is utilized to construct NIR fluorescent probes targeted to the desired biomarker by conjugating a fluorescent molecule to antibody fragments or small peptides [40,42]. This strategy gives NIR fluorescent molecules the ability to bind to a specific target. A small-molecule fluorophore is a popular choice for bioconjugation, but nanoparticles can also be attached to antibodies and peptides (Fig. 4).

Fig. 4. Building blocks for targeted NIR imaging probes.

Targeted fluorophores currently tested in clinical trials include small molecules, peptides, activatable probes, antibodies, and multimodal bioconjugates. Activatable fluorophores are initially quenched, but upon cleavage by enzymes, emit fluorescence. The relative sizes of these targeted fluorophores are displayed in the bottom panel. Reprinted from Zhang et al. [40] with permission from Springer-Nature.

Nanoparticle-based multifunctional agents encapsulating NIR fluorophores and ligands have been used for a theranostic approach in the context of cancer imaging and therapy [43,44]. Nanoparticles themselves can be delivered to tumor tissue by enhanced permeability and retention (EPR) [45], which is defined as passive targeting. Alternatively, they can be prepared to achieve active targeting to a specific target by coating them with antibodies or ligands such as poly(ethylene glycol) (PEG)-functionalized with cyclic RGD peptides (cRGD) [46].

3.4. Activatable or multimodal probes

In order to improve the specificity or broaden their applications, activatable or multimodal probes have been also explored (Fig. 5). For example, upon interaction with their intended target, imaging agents undergo a chemical reaction including enzymatic cleavage and oxidation which switches the probe from a non- or weakly fluorescent form to a strongly fluorescent form [47]. Other NIR fluorescent constructs can be designed with quenched dyes and certain peptide sequences to image cells expressing specific proteases. The enzymes cleave the peptide linker, ceasing the quenching on the fluorescent label, and produce an NIR signal within the associated cells [48-50]. These strategies are proven to be highly sensitive due to the catalytic nature of signal generation upon encountering the target enzyme and especially effective for the development of NIR fluorophore-based multifunctional agents in areas of cancer targeting and imaging [47,51].

Fig. 5. Design of smart NIR fluorescence probes for improved specificity and SBR.

Antibody-fluorophore conjugates are designed for activatable fluorescence imaging of cancer cells. NIR fluorescence signals stay “off” at normal tissues but turn “on” after binding to the target antigens overexpressed on cancer cell. Shown are an example of an Ab-dye conjugate trastuzumab-ATTO680. Reprinted from Kim et al. [52] with permission from Ivyspring.

4. Imaging of Immune Components

There are two broad categories of fluorescent probes: genetic probes (fluorescent proteins) and chemical probes (fluorescent molecules). In this section, we focus on translatable chemical probes.

4.1. Antibodies

Antibodies are the proteins produced by B cells that show specific binding to antigens and subsequently activate other arms of the immune system to fight pathogen threats [53]. The biodistribution of antibodies usually depends on their targets. Antibodies are therefore frequently used to determine the biodistribution of biomarkers in vivo. NIR fluorophore-labeled antibodies are therefore most frequently used for NIR imaging of biological tissue (Table 1). Antibodies labeled with NIR small-molecule fluorophores or nanoparticles are commonly used for NIR imaging of biomarkers. In light of the rise of therapeutic approaches using monoclonal antibodies, it is critical to determine the biodistribution of therapeutic antibodies at tissue and cellular levels to understand their mode of action. Conventional bioanalytical techniques including ELISA and immunohistochemistry lack the direct detection and high spatial and temporal resolution of NIR imaging. As such, radiolabeling is still the standard method for bulk organ and tissue distributions [54]. NIR imaging overcame these challenges and enabled real-time high-resolution imaging. However, there is still an outstanding issue with the possible alteration of the binding ability as a result of the labeling procedure [55].

Table 1.

NIR fluorescence imaging of antibodies.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Trastuzumab, bevacizumab, Trastuzumab emtansine	IRDye800CW, Alexa Fluor 680	4-17 d	Xenograft tumor mice	[56]
Cetuximab	IRDye800CW	1 d	Xenograft tumor mice	[57]
Cetuximab, bevacizumab	IRDye800CW	3 d	Xenograft tumor mice	[58]
Bevacizumab, trastuzumab	IRDye800CW	4 d	Xenograft tumor mice	[59]
MPDL3280A (PD-L1)	IRDye800CW	3-5 d	Xenograft tumor mice	[60]
Cys-diabody A2 (PSCA)	sCy5	2-4 h	Xenograft tumor mice	[61]
Atezolizumab (PD-L1)	ErNPs	3 d	Syngeneic and Xenograft tumor mice	[62]

Cilliers et al. successfully labeled two clinical antibodies, Herceptin (trastuzumab) and Avastin (bevacizumab) with NIR fluorophores IRDye800CW or Alexa Fluor 680 at an optimum degree of labeling without altering antibody disposition for 4-17 days. This method was further proved to be useful for ex vivo determination of the biodistribution of Kadcyla (a conjugate of trastuzumab to a cytotoxic agent DM1) up to 7 days in a xenograft tumor model in mice [56]. Olivier et al. labeled a chimeric (mouse/human) monoclonal antibody against epidermal growth factor receptor (EGFR) cetuximab with IRDye800CW and ⁸⁹Zr [57] The incremental uptake of fluorescence signals in xenograft tumor mice was observed 24 h post-injection of the NIR-antibody under NIR imaging. Both optical and positron emission tomography (PET) imaging gave consistent results in terms of tissue deposition of the antibody. However, Connor et al. found that IRDye800CW labeling via NHS chemistry significantly modified plasma pharmacokinetics and tissue deposition of the NIR fluorophore-labeled monoclonal antibody compared to the conventional radiolabeling [55]. Cohen et al. examined different labeling ratios of clinical monoclonal antibodies with IRDye800CW and concluded that alteration of biodistribution was observed when more than 1 equiv. of fluorophores were conjugated to antibodies [58]. Choi et al. confirmed this surface modulation using zwitterionic fluorophores conjugated on small peptides, proteins, and antibodies [63,64]. Considering the charge-to-mass ratio, when more than 2 small molecules are conjugated on the surface of a protein or an antibody, the fate of biodistribution and clearance could be changed significantly. Importantly, this method has been used to identify critical biomarkers in the field of cancer immunotherapy lately. Despite an improved overall survival with immune checkpoint inhibitors, the efficacy remains limited [65]. Programmed cell death-1 ligand-1 (PD-L1) expression by tumors has been shown to enrich for responses to anti-PD-L1 therapy for non-small cell lung carcinoma (NSCLC) and bladder cancer [66]. Chatterjee et al. conjugated IRDye800CW on a humanized, mouse and human cross-reactive PD-L1 antibody to detect the graded levels of PD-L1 expression in human tumor xenografts up to 120 h [60]. Zhong et al. further advanced this approach to conjugate anti-PD-L1 antibody with biocompatible rare-earth erbium-based nanoparticles (ErNPs) exhibiting downconversion luminescence at ~1,600 nm to realize NIR-IIb imaging with much improved SBR [62]. This group visualized T cells with lead sulfide (PbS) QDs to achieve noninvasive in vivo visualization of a critical biomarker and T cell population in the same NIR-IIb window in mouse models of breast and colon cancers at 3 d after injection. Remarkably, Zettlitz et al. established an universal dual-modality linker (DML) for conjugation of a fluorescent dye and an ¹⁸F-radioisotope to realize noninvasive, antibody-based dual-modality (PET/optical) imaging for noninvasive whole-body scanning and intraoperative fluorescence imaging [61]. Sequential immuno-PET and optical imaging 2-4 h after injection as a proof-of-principle experiment in prostate cancer xenograft model successfully showed feasibility of this strategy.

This approach can be used not only for diagnostic purposes but also as a therapeutic strategy. One of the remarkable advances in the use of antibodies with NIR fluorophores is photoimmunotherapy (PIT) (Fig. 6). PIT combines the cytotoxic effects of photodynamic therapy using a photo-sensitizer with antibodies specific to tumor antigens [67,68], which also allows for monitoring therapeutic effects under NIR imaging [69]. PIT which uses a photosensitizer activated in the NIR range (NIR-PIT) represents a new modality and has been shown to not only cause immediate cell death resulting in tumor shrinkage [70], but also induce host immune responses via the rapid release of immunogenic signals [71]. This approach has been tested in many preclinical settings and recently has been evaluated in clinical settings with promising results. In particular, monoclonal antibodies (mAb) targeting cancer antigen EGFRs conjugated to a photosensitizer such as IR700 or ICG successfully delayed tumor growth of breast [72], bladder [73], lung [74], pancreatic [75], prostate [76], and ovarian [77] cancer, as well as B-cell lymphoma [78,79] models in mice when activated with NIR light.

Fig. 6. Monitoring the effect of NIR-PIT.

Serial histology pictures show microdistribution of IR700 fluorescence in the tumor after repeated PIT. A) Only a few cells are fluorescent in the diffuse necrotic cells 1 h after the first PIT (top right). Scattered fluorescent micro-clusters of survived tumor cells (bottom; arrows) indicating proliferated cancer cells 1 d post PIT. B) Schematic drawing of a humanized antibody conjugate with IR700. Reprinted from Mitsunaga et al. [80] with permission from ACS Publications.

NIR-PIT is a highly flexible theranostic platform, as any antibody specific to a choice tumor antigen can be conjugated to an NIR photosensitizer. In addition, the fluorescence signal from the conjugate can be used for noninvasive imaging to monitor the effect of NIR-PIT. This approach is highly specific because antibody-targeted photosensitizers activate and induce cell death only in target cells, reducing some side effects of cancer treatment [80]. Phase 1 and 2 trials of NIR-PIT using the anti-EGFR antibody cetuximab, and the photo-absorber IR700, are ongoing in patients with recurrent head and neck cancer (https://clinicaltrials.gov/ct2/show/NCT02422979).

Targeting immune cells with PIT is another strategy to augment the anti-tumor immune response. It is well established that regulatory CD4⁺CD25⁺FOXP3⁺ regulatory T cells (Tregs) play a critical role in tumor immuno-evasion [81]. CD25-targeted NIR-PIT induced depletion of Tregs in tumors and enhanced anti-tumor response, resulting in tumor regression in a mouse model [82]. The same research group further demonstrated that combined CD44- and CD25-targeted NIR-PIT induced long-term antitumor immunity and prolonged survival compared with CD25-targeted NIR-PIT alone in both colon cancer and lung carcinoma models [83].

4.2. Vaccines

Vaccination is one of the most successful medical interventions to decrease morbidity and mortality caused by infectious diseases [84]. This modality administers bioactive molecules to challenge and enhance the immune response. To this end, efficient and timely delivery of the vaccine to the secondary lymphoid tissue, where the immune response is coordinated, is critical for optimal efficacy [85-87]. However, despite the clear advantages of NIR imaging in tracking biological molecules, there is a lack of methodology and a lack of tools to determine the biodistribution of the injected vaccines using NIR imaging (Table 2).

Table 2.

NIR fluorescence imaging of vaccines.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Yellow fever prME mRNA model vaccine	64Cu-DyLight 680	1 d	Rhesus macaque	[88]
Model vaccine	ZW800-1C	3 d	Mice	[89]
Mucosal model vaccine	PLGA-QD	2 d	Mice	[90]

Lindsay et al. demonstrated that a dual radionuclide-NIR probe allows for quantitative, longitudinal, and noninvasive monitoring of the biodistribution of a model yellow fever prME mRNA vaccine labeled with ⁶⁴Cu-DyLight 680 in the secondary lymphoid tissue by PET-CT and NIR imaging in monkeys [88]. In this method, the model vaccine was monitored up to 28 h by PET-CT while NIR imaging was used to extract relevant portions of organs that contained the RNA of interest. Future work could be done on whether NIR imaging could provide a noninvasive, low-cost, and high-resolution means of assessing translocation of the vaccine into lymph nodes. Katagiri et al. labeled model vaccines with a zwitterionic NIR fluorophore ZW800-1C, which showed minimal interaction with biological tissue and size-dependent uptake in the lymph nodes up to 72 h [89]. Panthani et al. used FDA-approved biocompatible poly(lactic-co-glycolic acid) (PLGA) microparticle-encapsulated QDs as a model mucosal vaccine. The QD vaccines were gavaged orally and imaged noninvasively in the gastrointestinal tract up to 48 h [90].

4.3. Immune cells

The recent clinical success of cancer immunotherapy highlights the importance of T cell imaging, in particular, noninvasive methods to assess the density and functions of tumor-infiltrating lymphocytes (TIL) to predict its therapeutic responses [91,92]. To date, ex vivo direct staining of isolated cells is a popular method for labeling lymphocytes (e.g., CAR-T cells) for NIR imaging [93]. However, this method has significant limitations due to unfavorable effects on lymphocytes during the isolation and labeling procedures. Therefore, a new methodology is highly desired to indirectly label T and B cells in vivo with an injectable probe.

4.3.1. T cells

T cells are major effector cells for combating foreign pathogens and cancer, as well as being responsible for peripheral tolerance to self- and non-harmful antigens [94]. T cell immunology gains more significant interest lately due to the success of cancer immunotherapy [91,92]. Tumor infiltrating T lymphocytes (TIL) density is a strong positive prognostic indicator for many types of tumors, and responses to immunotherapy preferentially occur in tumors with a preexisting antitumor T-cell response [91,92]. It is therefore critical to describe the in vivo behavior of T cells to understand the pathologic basis of diseases. NIR imaging has been extensively used for T cell imaging (Table 3).

Table 3.

NIR fluorescence imaging of T cells.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Tumor infiltrating T lymphocytes	DiR	3 wk	Xenograft tumor mice	[95]
Model antigen vaccination	CIR38M	7 d	C57BL/6 mice	[96]
Tumor infiltrating T lymphocytes	VivoTag 680	3 d	Xenograft tumor mice	[97]
Tumor infiltrating T lymphocytes	IRDye800CW	3 d	Xenograft tumor mice	[98]
Antigen-specific T cells	NIR-797-isothiocyanate	3 d	Syngeneic tumor model and bacterial infection in mice	[99]
Tumor infiltrating T lymphocytes	ErNP	3 d	Xenograft tumor mice	[62]

A cell-permeable lipophilic NIR fluorophore 1,1-dioctadecyltetramethyl indotricarbocyanine iodide (DiR) is frequently used for this purpose. Younis et al. used this approach and observed the fluorescence signal from ex vivo-stained T cells injected into a syngeneic mouse breast cancer model for 3 wk under a multi-spectral fluorescent imaging system [95]. No toxicity was detectable for up to 10 d after staining in vitro, suggesting that this approach could be durable and safe for the purpose of NIR imaging. Mellanby et al. used a tricarbocyanine N-triazole CIR38M with improved brightness and photostability compared to ICG or DiR to track post-transferred T cells responding to model vaccination in the secondary lymphoid tissues of mice for 7 d [96]. This method did not cause any significant functional alterations on T cells. CIR38M was also able to stain human T cells ex vivo without affecting their function for 24 h.

When diffused into cytotoxic T cells, an NHS ester of a (benz)indolium-derived far-red fluorescent probe VivoTag 680 (VT680) covalently bonded to cellular components which enabled longitudinal observation of T cells in xenografts of colon cancer for 3 d after ex vivo labeling with fluorescence-mediated molecular tomography (FMT) [97]. This labeling had no apparent toxicity or other adverse effects and elicited the same tumor growth delay as non-labeled control T cells with adoptive transfer therapy. Foster et al. used IRDye800CW for ex vivo T cell labeling, which also bears an NHS reactive group to couple with amino groups on proteins [98]. Human EBV-specific cytotoxic T cells stained with IRDye800CW showed normal viability, cytokine production and migration. The stained T cells also accumulated in a xenograft tumor model in mice that were imaged using in vivo NIR imaging up to 72 h. In order to compensate the limited penetration depth in NIR imaging, Zheng et al. applied a photoacoustic approach in bioimaging. The Nie group was able to observe gradual uptake of model cancer antigen (ovalbumin, OVA)-specific T cells labeled with NIR-797-isothiocyanate ex vivo up to 72 h into OVA-expressing breast cancer in mice with a peak signal at 12 h [99]. In fact, NIR imaging can be applied to deeply and noninvasively monitor the dynamic changes of immune cells in the lymph node and tumor.

For indirect labeling, the antibody-based approach has been extensively explored in the field of immuno-PET [100]. Similarly, Zhong et al. used anti-CD8α antibody targeting T cells conjugated with PbS QDs, which emits NIR-IIb fluorescence in murine breast and colon cancer models [62]. In this approach, the labeled antibody was intravenously injected and successfully detected tumor-infiltrating T cells by NIR-IIb imaging with high resolution and SBR at 3 d post-intravenous injection.

Recently, NIR imaging of T cell function has been increasingly studied. NIR macromolecular reporters, composed of a peptide-caged NIR signaling moiety linked with a hydrophilic PEG passivation chain, become fluorescent by specific granzyme B activation and can be used for real-time in vivo evaluation of the efficacy of immunotherapy. Especially, He et al. demonstrated that these probes passively targeted tumor and detected granzyme B with the NIR fluorescence at 707 nm in immunotherapeutics-treated breast tumor in mice [101]. Notably, the probes show high renal clearance which allowed noninvasive optical urinalysis and evaluations of immunoactivation with immunotherapy.

4.3.2. B cells

B cells and their progeny plasma cells produce antibodies responsible for humoral responses. These antibodies protect against infection, but also contribute to pathological tissue injury in autoimmunity [102]. B cell depletion has demonstrated successful treatment of non-Hodgkin's lymphoma and is beginning to see use in other B cell pathologies. Because of this, the ability to monitor B cells as they respond to an immune event is important for determining the efficacy and safety of B cells affecting drugs. Importantly, depletion of B cells is a clinically approved approach for the treatment of lymphoma [103]. However, the exploration of NIR imaging of B cells is relatively scarce compared to T cells (Table 4).

Table 4.

NIR fluorescence imaging of B cells.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
B cell depletion immunotherapy	CellVue-NIR815	14 d	Anti-CD79 antibody in mice	[104]
B cells	SWNTs-rituximab	–	In vitro study only	[105]

Thorek et al. used an approach combining NIR imaging with superparamagnetic iron oxide (SPIO) nanoparticles to provide a means for whole body, longitudinal, and noninvasive imaging of B cells [104]. NIR fluorophore CellVue NIR815 (a lipophilic membrane dye) provided detectable signals from ex vivo-stained B cells injected into mice for up to 14 d using the whole body in vivo NIR imaging. In this study, a decrease of fluorescence signal was consistently observed following a model B cell depletion immunotherapy using an anti-CD79 antibody, confirming the specificity of this methodology. This method did not induce any detectable functional changes when tested in vitro.

Biologically inert semiconducting single-walled carbon nanotubes (SWNTs) with PEG functionalization conjugated to Rituxan (anti-CD20 antibody) have also been used for B cell imaging. Welsher et al. demonstrated that this method showed minimal non-specific binding and staining of non-targeted cells in vitro, but no in vivo imaging nor toxicity test was performed in this study [105].

4.3.3. Macrophages

Macrophages are immune cells that reside in local tissues and fulfill various functions in host defense, tissue homeostasis, and inflammatory response through phagocytosis and production of inflammatory cytokines [106]. Similar to T cells, tumor-associated macrophages (TAMs) have been gaining considerable attentions because of their predominant abundance in solid tumors and key roles in tumor progression [107,108]. Due to their significant roles in pathophysiology, macrophages are a frequent target to study.

NIR imaging has been extensively used to monitor the dynamics and function of macrophages (Table 5). A membrane-selective lipophilic carbocyanine dye (DiR) has been used for labeling macrophages ex vivo. Eisenblatter et al. demonstrated that the migration of ex vivo-labeled macrophages into inflammation induced by polyacrylamide gel pellets could be tracked up to 7 d after the injection of labeled cells using FMT [109]. Labeling did not induce appreciable changes in viability or the function of macrophages.

Table 5.

NIR fluorescence imaging of macrophages.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Migration into inflammation	DiR	7 d	Polyacrylamide gel-induced inflammation	[109]
Brain cancer imaging	Silica iron oxide NPs + Cy5.5	1 d	Xenograft tumor mice	[110]
Accumulation in cardiac infarction	CLIO + VT750	1 d	Myocardial infarction or ischemia in mice	[111]
Migration into inflammation	F4/80 Ab + Cy5.5	3 d	Arthritis mouse model	[112]
Liver tumor imaging	TLR4 ab + ZW800-1C	3 d	Syngeneic tumor mice	[113]
Mer imaging in colorectal cancer	UNC2025-SiR-COOH	1 d	model of metastatic tumor in mice	[114]
Migration into inflammation	PNIPAM + IR750 + folate NPs	1 d	LPS-induced tissue inflammation in mice	[115]
Migration into inflammation	CDnir7	18 h	LPS-induced tissue inflammation in mice	[116]

Macrophages show strong phagocytic activities toward small particles; engulfing and digesting them in their phagosomes [117]. Taking advantage of this phenotype, macrophages have been labeled and visualized with fluorescent nanoparticles indirectly. Lee et al. synthesized NIR fluorescent (Cy5.5) silica-coated iron oxide nanoparticles (NF-SIONs) for this purpose [110]. NF-SIONs were about 37 nm in size and efficiently taken up by macrophages without apparent toxicity. Tumor-associated macrophages (TAMs) may occupy 30-50% of the whole cell population of a brain tumor [118,119], and can be used to identify the location of glioblastomas (GBM). Interestingly, NF-SIONs penetrated the blood-brain barrier after intravenous injection and specifically delineated an orthotopic xenograft model of GBM in mice up to 24 h under in vivo NIR imaging. Nahrendorf et al. intravenously injected a magneto-fluorescent iron oxide nanoparticle CLIO-VT750 for dual imaging in mice with cardiac infarction. This method detected an accumulation of inflammatory cells including macrophages and neutrophils in the infarction zone under FMT and MRI up to 24 h. The signal from simultaneously injected prosense-680, an activatable fluorescence sensor reporting on cathepsin activity colocalized with inflammatory cells, detected function of these inflammatory cells simultaneously in vivo under dual channel FMT [111].

Other indirect labeling methods have also been explored. NIR labeled fluorophore-antibody conjugates have been used for the monitoring of macrophage migration. Hansch et al. used a pan-macrophage marker F4/80 to visualize macrophages. Injection of anti-F4/80 monoclonal antibodies (mAb) labeled with Cy5.5 fluorophores revealed macrophage accumulation in the inflamed knee joints of a murine model of antigen- induced arthritis using in vivo NIR imaging up to 72 h [112]. Toll-like receptor 4 (TLR4) plays an important role in innate immunity and is highly expressed in macrophages [120]. Ji et al. labeled anti-TLR4 antibody with a zwitterionic NIR fluorophore with minimal interaction with biological tissue to avoid any possible interference from the fluorophore with the biodistribution of the mAb. Hepatocellular carcinoma is characterized by its inflammatory background and macrophages are closely related to its carcinogenesis and progression [121,122]. Upon intravenous injection of the conjugate, a strong signal was observed in TAM-enriched liver cancer grafts in a mouse model up to 72 h after injection (Fig. 7) [113].

Fig. 7. NIR fluorescence imaging of tumor-associated macrophages in hepatic tumor.

TLR4 antibody-conjugated ZW800-C was injected intravenously into mice of liver cancer model. Intraoperative tumor targeting and imaging were performed after a single intravenous injection of ZW800-conjugated TLR4 antibody under the real-time NIR fluorescence imaging system. Reprinted from Ji et al. [113] with permission from AME Publishing Company.

Proto-oncogene tyrosine-protein kinase Mer (MerTK) is a receptor tyrosine kinase influencing cell survival, migration, differentiation, and phagocytic activity. Importantly, Mer signaling influences tumor-associated leukocytes, including macrophages, to promote tumor progression [123]. Millar et al. developed a small molecule probe for NIR imaging of Mer based on its kinase inhibitor UNC2025 linked to silicon rhodamine carboxylate (SiR-COOH) fluorochrome [114]. An accumulation of a selective fluorescent molecular probe (MERi-SiR) in Mer⁺ TAMs in a model of metastatic murine colorectal cancer was confirmed by microscopic imaging, but not noninvasive NIR imaging.

Macrophages express elevated levels of the folate receptor (FR) when activated in inflamed tissues [124]. Low et al. synthesized an NIR fluorophore (IR750)-trapped poly(N-isopropyl acrylamide-co-styrene) nanoparticles and conjugated them with folate to target macrophages. The FR-targeting nanoprobes did not induce any toxicity in vitro and distinctively showed the fluorescence signal under NIR imaging in the LPS-induced inflamed tissues 24 h after intravenous administration [115]. Kang et al. took a diversity-oriented fluorescence library approach (DOFLA) [125] to find a macrophage-targeted NIR fluorophore CDnir7 with a heptamethine structure [116]. CDnir7 did not show any apparent toxicity in vitro, and upon intravenous injection, it detected infiltrating macrophages in a murine tissue inflammation model under FMT. The authors also demonstrated that CDnir7 could be used for photoacoustic imaging. However, compared to other labeling modalities such as ⁸⁹Zr-labeled dextran nanoparticles [126], this method is rather short-lived (up to 18 h only).

4.3.4. Dendritic cells

Dendritic cells (DCs) play critical roles in both the innate and adaptive immune systems. In particular, they are known as the most versatile antigen-presenting cells, orchestrating immune responses against infectious diseases and cancer as well as maintaining tolerance to host and harmless antigens [127]. NIR imaging has been useful for analyzing the dynamics and function of DCs in vivo and informative for understanding DC biology in various settings including infection, allergy, autoimmune disease, transplantation, vaccination, and immunotherapy. To analyze DC migration, multiple imaging approaches have been used including MRI, PET and NIR imaging (Table 6).

Table 6.

NIR fluorescence imaging of dendritic cells.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Migration into the secondary lymphoid tissues	Tat-NIR polymersomes	33 d	Mice	[128]
	CdSeTe/ZnS QD800	3 d		[129]
	Gd + aza-BODIPY	2 d		[130]
	SPIO + NIR797	5 d		[131]
	PLGA + iron oxide + ICG/OVA	2 d		[132]
	PLGA + ICG/OVA	2 d		[133]

Christian et al. conjugated a cell-permeable peptide Tat to NIR fluorescent polymersomes (Tat-NIR-emissive polymersomes) [134]. Ex vivo labeling of bone-marrow-derived DCs did not significantly affect the viability and function of the cells. Fluorescence lifetime imaging (FLIM) allowed a longitudinal observation of the accumulation of subcutaneously injected DCs into the popliteal lymph node over 33 d after the initial injection [128]. This approach was used in another study and found to have no significant effect on DC behavior compared to other labeling methods, including genetic or small fluorescence molecule approaches [135]. Noh et al. used CdSeTe/ZnS QD800 for ex vivo labeling of bone-marrow-derived DCs and tracked DC migration into lymph nodes up to 72 h after injection under in vivo NIR imaging. They similarly found that NIR cell labeling did not change the viability or antigen presentation capacity of DCs [129].

NIR imaging can be combined with other modes of imaging to create multimodal probe molecules for NIR imaging of DCs. Kim et al. combined gadolinium (Gd)-chelating MR probe and an NIR fluorophore, aza-BODIPY (boron-dipyrromethene) (AB-BCA) and successfully imaged migration of ex vivo-labeled bone marrow-derived DCs through the lymphatic vessels to the lymph nodes via MR and NIR imaging [130]. This labeling method did not influence obvious toxicity by in vitro testing. The fluorescence signal was observed under in vivo NIR imaging for 48 h after labeled cell injection. Chen et al. developed a dual-modal nanoprobe composed of superparamagnetic iron oxide (SPIO) and NIR fluorophore NIR797 to label bone-marrow derived DCs ex vivo. Migration of labeled DCs into lymph nodes was detected by both MRI and NIR imaging simultaneously for 5 d after cell injection [131]. Noh et al. further advanced this approach using multifunctional PLGA nanoparticles containing the MRI contrast agent iron oxide, NIR fluorophores (ICG), and the model antigen OVA [132]. Migration of ex vivo-labeled bone-marrow-derived DCs through lymph draining could be observed through NIR imaging, and accumulation of the DCs in the lymph nodes was detected by MRI imaging in response to specific antigen stimulation with OVA for 48 h. T cells isolated from treated mice showed high cytotoxicity against OVA-expressing EG7, indicating that the DCs loaded with the nanoparticles efficiently cross-presented. Heo et al. used PLGA nanoparticles containing both small interfering RNA (siRNA) for the knock-down of immune-suppressor gene (signal transducer and activator of transcription-3, STAT3) and an immune response modifier (imiquimod, R837) for the activation of DCs through the TLR7 along with OVA and ICG. DCs treated with the nanoparticles were similarly confirmed to prime CD8+T cells through cross-presentation, and migration of DCs to lymph nodes was detected by in vivo NIR imaging for 48 h [133].

4.3.5. Natural Killer Cells

Natural killer (NK) cells have gained attention as a promising candidate for future immunotherapeutic approaches due to their critical role in immune surveillance against viral infection [136] and cancer [137]. In response to this, a reliable imaging methodology to analyze their dynamics and function in vivo is desirable, and NIR imaging has been explored for this type of cell (Table 7).

Table 7.

NIR fluorescence imaging of natural killer cells.

Application (Target)	Contrast agent	Tracking time	Animal model	Ref
Migration into breast cancer	ESNF13	7 d	Xenograft tumor mice	[138]
Migration into tumor	CD56 + QD705	1-7 d		[139]
Migration of CAR NK-92 cells	DiD	1 d		[140]

Similar to DCs, ex vivo staining is a dominant method for this type of cell. Uong et al. stained ex vivo expanded NK cells from healthy donors with a small NIR fluorophore ESNF13 [138]. These NK cells were injected intravenously in human breast cancer xenograft mice and imaged using mini-FLARE NIR fluorescence imaging. The high fluorescent signal from the ESNF13-stained NK cells allowed for tracking of cells migrating to the tumor sites for 7 d. Lim et al. were able to observe the signal from NK92MI cells labeled with anti-human CD56 antibody-coated QD705 after intratumoral injection in a xenograft melanoma in mice under in vivo NIR imaging at least 24 h after injection [139]. Tavri et al. stained NK-92 cells with an NIR fluorophore DiD, which were engineered with a chimeric antigen receptor (CAR) for the epithelial cell adhesion molecule (EpCAM) expressed on prostate cancer, and intravenously injected them into a xenograft prostate cancer model in rats [140]. Progressive accumulation of DiD-labeled cells was observed up to 24 h after injection by NIR imaging. In these studies, the labeling procedures minimally affected the function of NK cells.

5. Perspectives

NIR fluorescence imaging not only displays high-sensitivity and specific real-time imaging of the biological system due to minimal autofluorescence, low light scattering, and low absorption in neighboring tissues, but also offers multispectral and multiplexing capability, making this technology suitable for bioimaging the inherently complicated immune system [113]. Accordingly, this imaging technique has been successfully used to track immune cells in vivo and contributed to improving our understanding of immune responses. Thus, monitoring the behavior and function of injected DCs via NIR imaging is critical to optimizing this cell-based therapeutic modality [141]. Since the first approval of DC-based cancer vaccines (i.e., Provenge) by the US FDA [142], DCs have played a pivotal role in cancer therapy due to their safety and immunogenicity. Consequently, detailed analyses of the dynamics and functions of DCs in vivo are desirable for studying infections, allergies, autoimmune diseases, transplantation, vaccination and immunotherapies. In particular, it has been shown that macrophages are highly associated with the initiation, progression, and metastasis of cancer [143]. Macrophages have also been implicated in the onset, progression, and symptoms of multifarious pathologic processes, such as the formation of malignant tumors and inflammatory diseases [106,144]. Due to its importance in many pathological conditions, the development of NIR imaging for these types of cells has been expanding. In response to the rise of cancer immunotherapy, the importance of an imaging methodology is expected to increase further. Similarly, considering the important roles of NK cells in immunosurveillance against cancer, this type of cells also represents a potentially ideal target for cancer immunotherapy. Further development of NIR imaging of NK cells is therefore warranted.

Nevertheless, there are two unsolved questions remined in the field: 1) Clinical imaging systems that can provide clinicians fast answers in the OR with high temporal and spatial resolution noninvasively for image-guided interventions, 2) clinical contrast agents that can be used in the human body with specificity and safety in the clinical setting [145,146]. To this end, novel imaging modalities are increasingly emerging, such as NIR-II fluorescence imaging [1,3], photoacoustic imaging [49,99], and afterglow imaging [147,148] to attack the first unsolved question. As discussed above, these imaging modalities have great promise as an advanced noninvasive molecular imaging tool with high sensitivity levels in the body. Additionally, due to the limited penetration of optical imaging, combination approaches including MRI-NIR and PET-NIR using multimodal probes would be necessary for noninvasive deep tissue imaging [5].

However, the latter is more serious. Since ICG is currently the only FDA-approved fluorophore in the NIR window, novel fluorophores in the NIR-II window showing superior resolution and sensitivity over the visible-NIR-I window are desperately needed to provide deep anatomical and functional features [11]. Additionally, we expect that more classes of multifunctional NIR fluorophores will emerge for the target-specific treatment of human diseases. However, the feasibility of applying this approach in the clinic, e.g. biocompatibility, nonimmunogenicity, and safety, needs to be fully explored, as many nanoparticle-based imaging probes used in this technology raise concerns about unknown long-term toxicity. Most works carefully examined the toxicity of probes in vitro, but safety assessment on long-term accumulation and toxicity upon in vivo application of nanoprobes is difficult using the current system. In order to move these candidate nanoprobes into clinical practice, a standardized guideline and reliable methodology for safety evaluation should be established. Despite the clear advantages of NIR imaging in tracking biological molecules, the methodology to determine the biodistribution of vaccines based on NIR imaging is surprisingly limited. Since the current gold standard radiologic method cannot reveal the detail of immune responses and immune cell dynamics, further exploration of this technology in this field is warranted. Noninvasive, high-resolution, real-time NIR imaging of vaccines and immunotherapeutics would provide a critical piece of information on how the immune system is primed and ultimately develops protective immune responses from vaccination or induces immunological tolerance with immunotherapy.