3.1.

Tumor Cell and Xenograft Models

These organic dyes have been reported to be directly taken up and preferentially accumulate in the mitochondria of tumor without inference in normal tissues in preclinical models, including mice and dogs.^{5,6,15,16,20,22} For example, Pz 247, a chiral porphyazine NIRF agent without chemical conjugation, show tumor-targeting property and exhibited preferential retention in the tumor cells of human breast tumor xenografts subcutaneously implanted in mice.⁴⁶ Heptamethine cyanine dye-based NIRF imaging is becoming an attractive modality for cancer detection with the acquisition of real-time pathophysiologic information. At the cellular level, these NIRF agents also show preferential uptake in cancer cells but not normal cells, as demonstrated in many different types of cancer cells including cultured, circulating and disseminated tumor cells.^5,6,16

3.2.

Patient-Derived Tumor Xenograft Models

In addition to solid tumors, these NIRF dyes can detect tumor cells either in the interstitial fluid or in the form of circulating tumor cells with high sensitivity and specificity.¹⁹ The analysis of malignant tumor cells in splanchnocoele fluid by NIRF imaging has become a fast and reliable method to monitor disease progression in cancer patients.^6,47 In recent years, patient-derived tumor xenograft (PDX) models established by transplanting human fresh tumor tissues into immunodeficient mice have become another valuable tool for maintaining intact tumor profiles, such as tumor heterogeneity. They can be used to screen and test anticancer drug efficacy tailored for individualized treatment.⁴⁸ Conventional PDX models are visualized in vivo by luminescence-based optical imaging after labeling tumors with a luciferase reporter gene, which is time-consuming and complicated.⁴⁹ On the other hand, the alternative in vitro screening of PDX-derived tumor cells often results in a loss of tumor cell subsets and information on tumor heterogeneity.⁵⁰ With the introduction of heptamethine cyanine dyes, we were able to characterize the PDX models by NIRF imaging by readily capturing subrenally grafted deep tissues. A wide spectrum of PDX models harboring different types of tumor samples, including gastric cancer, liver cancer, bladder cancer and renal cancer, were identified by NIRF imaging by the exclusive exhibition of fluorescence signals in the renal area of mice (Fig. 2), which was further confirmed concurrently by tumor histology using H&E stain (data not shown).

Fig. 2

NIRF imaging of different PDX models with renal capsule xenografts in nude mice. Representative H&E images indicating tumor histopathology for individual PDX model are shown. Magnification 400×.

3.3.

Clinical Experience

NIRF dyes show tumor-specific targeting inhuman tumor samples.³⁰ The fact that these dyes possess dual tumor-imaging and targeting capabilities suggests possible applications for the real-time monitoring of tumor growth as well as the recognition of surgical margins during surgery, which would significantly advance the accurate localization of tumor margins and thus improve surgical effects by reducing the possibility of missing positive lesions.⁵¹ In agreement with this idea, recent evidence has demonstrated that surgical resection of hepatic metastases with the assistance of NIRF imaging accurately recognized benign and malignant lesions of the liver and, more importantly, facilitated tumor detection with a focus of less than 5 mm.⁵²

In another preclinical trial, MHI-148 dye was applied for direct imaging of clinical tumor specimens, where the entire kidney removed surgically from a renal cell carcinoma (RCC) patient was subjected to immediate perfusion with saline-dissolved MHI-148 dye followed by ex vivo NIRF imaging. In this first-in-man ex vivo test without a tumor host, the NIRF signals were clearly detected in tumor areas in sharp contrast to the slight amount of dye retained in normal counterparts. Further quantitative analysis indicated a sixfold increase of MHI-148 dye uptake intensity in tumors relative to normal tissues, with coregistered tumor regions and fluorescence signals confirmed by pathologic analyses.²⁰ This pioneering study strongly suggests the potential clinical utility of these attractive NIRF agents.

ICG is the only FDA-approved NIRF agent for clinical use. One of ICG’s applications in the clinic is to serve as a preoperative evaluator of liver function by intravenous injection prior to hepatectomy, with the dye being accumulated in hepatocellular carcinoma (HCC) tissue.⁵³ The ICG in combination with NIFR imaging of HCC has been developed for intraoperative navigation during open surgery and laparoscopic surgery.^54,55 But it is known that ICG can make normal tissues emit fluorescence signals with high intensity during some surgical procedures, such as partial nephrectomy,⁵⁶ which is similar to the observations made in ICG-receiving (i.v. route) mouse models for monitoring renal perfusion.⁵⁷ This disadvantage largely compromises the potential use of ICG for tumor visualization but provides opportunities for developing heptamethine cyanine dye-based NIRF imaging approaches in the near future to meet growing clinical needs for better cancer imaging and detection.

4.1.

Tumor Xenograft Mouse Models

NIRF dyes have been modified by conjugation with a radionuclide and tested for their imaging and targeting potential in xenograft tumor models, showing significant improvements in sensitivity, which suggest the preclinical utility of NIRF dyes for detection of deep-tissue tumors. In a RANKL-overexpressing LNCaP metastatic prostate tumor xenograft model, NIRF imaging detected two superficial tumors in the mouse after PC-1001 injection, while PC-1001-coupled PET revealed an extra tumor in a deep location in the same mouse,³⁰ indicating the enhanced sensitivity of NIRF nuclear imaging. To date, several PET tracers have been applied for NIRF nuclear imaging of experimental tumor models, including ${}^{18}\mathrm{F}$, ${}^{11}\mathrm{C}$, ${}^{99\mathrm{m}}\mathrm{Tc}$, ${}^{64}\mathrm{Cu}$, and ${}^{111}\mathrm{In}$.^31,62–64

Recent studies have reported on a PET/NIRF probe, $\mathrm{PC}\text{-}1001/{}^{64}\mathrm{Cu}$, conjugating NIRF dye PC-1001 with ${}^{64}\mathrm{Cu}$ for breast cancer PET and fluorescence imaging. This $\mathrm{PC}\text{-}1001/{}^{64}\mathrm{Cu}$ compound has been validated in both breast cancer cells and a breast cancer xenograft model.³¹ It showed a much higher rate of uptake and accumulation in tumors compared to normal organs, such as the heart, liver, lung, and spleen, which further exhibited a decrease in the uptake rate with time. However, developing the PET application with this probe in both the research and clinical settings was costly, due to the limited supply of cyclotron-produced ${}^{64}\mathrm{Cu}$ radioisotope. By contrast, the cancer-targeting SPECT/NIRF dual-modality imaging probe $\mathrm{PC}\text{-}1007/{}^{99\mathrm{m}}\mathrm{Tc}$ has been successfully synthesized in a low-cost scheme and also exhibited cancer-specific targeting and accumulation properties in cancer cells and tumor xenograft models.³¹ The multiple imaging agents were incorporated with Gd(III) chelates and conjugated to IR-783 with a polyethylene glycol (PEG) linker or a short alkyl linker. Both agents achieved outstanding tumor cell labeling capability and were detectable in vivo using both MR and optical imaging.⁶⁵ Accumulation of these agents in a nude mouse MCF7 tumor xenograft model was detected with fluorescence imaging, and renal clearance by MR imaging.²⁴

4.2.

Spontaneous Canine Tumor Models

To evaluate the tumor-specific targeting ability of NIRF dyes in large live animals, domestic dogs that grew spontaneous tumors were chosen as models to determine the uptake and accumulation of NIRF dyes by both optical imaging and PET/CT scan.²² Dogs that carried different spontaneous tumors were injected with MHI-148 dye via femoral vein. After removing the tumors, strong NIR fluorescence signals were detected from primary and metastatic canine tumor tissues, including breast cancer, liver cancer, lung cancer, duodenal cancer, and colon cancer.²² It has been demonstrated that these dyes could detect spontaneous tumors in large animals similar to the findings with mice. Since dogs develop spontaneous primary and metastatic tumors in a manner similar to humans,⁶⁶ the specificity and bioavailability of NIRF dyes proven in canine cancer imaging provides new insights into developing NIRF dyes for the clinical detection of human cancers. To further evaluate the tumor-specific targeting ability of NIRF dyes in large live animals, a $\mathrm{PC}\text{-}1001/{}^{68}\mathrm{Ga}$ probe was synthesized by conjugating PC-1001 with ${}^{68}\mathrm{Ga}$ for independent PET and fluorescence imaging in canine tumor models. PET/CT images clearly displayed the tumor locations by the preferential accumulation of $\mathrm{PC}\text{-}1001/{}^{68}\mathrm{Ga}$ in the canine tumor tissue versus the normal counterpart.²²

4.3.

Near-Infrared Fluorescence Imaging-Guided Clinical Surgery

The application of multimodality optical techniques has been recently brought into focus for better cancer imaging with improved specificity, sensitivity, and reliability as compared to a single modality. Following this trend, bimodal tracers for sentinel lymph node (SLND) by radiolabeled mannosylated dextran derivatives bearing a NIRF imager allowed a clear visualization of the popliteal node by SPECT/CT scan and also enabled the guided identification of SLND during surgery.⁶⁷ In a study of 32 breast cancer patients, the SLNDs of all patients could be identified through radionuclide-combined NIRF imaging in a simple, convenient and high-through putway.⁶⁸ This dual NIRF and radioactive imaging approach further simplifies the early-stage development and validation of in vitro and in vivo optimization of parameters prior to the final PET imaging in live animals and humans.

5.1.

Delivery of Conjugated Chemotherapy Drugs

Theranostic prodrugs equipped with fluorophores as optical reporters have become useful to monitor the drug delivery and release process. Heptamethine cyanine dyes labeled with different antitumor drugs to achieve targeted transport of drugs through the dye-guided specific uptake in tumors allowed both real-time monitoring of therapeutic effect and lower drug dosages to avoid potential side effects (Fig. 3).⁶⁹ Zhang et al. constructed IR-780NM, a tumor-targeting agent by introducing the antitumor drug nitrogen mustard into the parent structure of IR-780 iodide, and demonstrated the utility of this compound for tumor imaging and targeting.²⁸ Mostly recently, Wu et al. synthesized the NIR dye—MAOA inhibitor (NMI) compound by conjugating MHI-148 dye with the moiety derived from clorgyline, a small molecule inhibitor of monoamine oxidase A that has emerged as a novel therapeutic target for human prostate cancer (Fig. 3). NMI showed tumor-specific targeting properties by systemic circulation in prostate tumor xenograft mice and dramatically restricted tumor growth. This dye-drug conjugate possesses diagnostic potential and can become an important platform for a future generation of anticancer therapeutics.²¹ In addition, mice treated with IR-783-docetaxel conjugate showed significant shrinkage of prostate tumors grown in tibia, with no obvious systemic toxicity or reduced body weight observed in mice during treatment.³⁰ These results in aggregate suggest the feasible development of these dyes as a new platform for chemotherapy drug delivery with improved outcomes.

5.2.

Brain Drug Delivery

Chemotherapy is indispensable for brain tumor treatment after surgery. The major challenge of brain tumor therapy is how to deliver therapeutic agents effectively into the tumor core from the systemic circulation and reduce the side effects of nonspecific biodistribution, which is largely due to poor penetration of drugs through the blood-brain barrier (BBB) and blood-tumor barrier (BTB).^70,71 After conjugating IR-783 dye to the chemotherapy drug gemcitabine to synthesize a dye-drug conjugate (NIRG) (Fig. 3), we recently showed that the NIRF dye as a drug carrier was able to deliver the drug to brain tumors by penetrating the BBB/BTB in mice (Fig. 4). NIRG treatment also significantly inhibited the growth of intracranial glioma and prostate tumor brain metastases, prolonging survival in mice.²⁵ These results demonstrate the great potential of these NIRF dyes and derived dye-drug conjugates for future development and application in the clinic as effective brain tumor theranostics.

Fig. 3

Chemical structures of NIRF heptamethine cyanine dye-conjugated chemotherapy drugs. The structures of NIR dye and drug in conjugates are shadowed by red and blue colors, respectively.

Fig. 4

Preferential uptake and retention of IR-783 dye and derived dye-drug conjugate in intracranial human brain tumor xenograft mouse models. A-B, IR-783 dye; C-D, IR-783-drug conjugate. Representative H&E images indicating tumor histopathology for individual xenograft are shown. Magnification 400×.

8.1.

Tumor Hypoxia and $\mathrm{HIF}1\alpha /\text{Organic}$ Anion-Transporting Polypeptides Signaling Axis

Tumor hypoxia is an important mechanism underlying tumor detection by heptamethine cyanine dye-assisted optical imaging,^20,22 which is different from ICG that functions primarily through binding to plasma proteins to enable tumor diagnosis by fluorescence imaging.⁸⁹ This is further supported by the lack of evidence indicating a direct association of ICG with tumor hypoxia. In our previous parallel comparison, ICG showed relatively low values of tumor-to-background ratio at 1.4 to 1.7 at 24 h in tumor-bearing mice by optical imaging,⁹⁰ while MHI-148 had a ratio of 9.1, which could be further increased by twofold in the presence of hypoxia-inducible factor $1\alpha$ ($\mathrm{HIF}1\alpha$) protein overexpression, a key mediator of hypoxic signaling.²⁰ These findings clearly demonstrated the advantage of heptamethine cyanine dyes over other NIRF agents for cancer imaging. Previous studies have demonstrated that OATP1B3, a representative OATP member, dominantly controls the transport of heptamethine cyanine dyes, including IR-780, IR-783, and MHI-148, into cancer cells, which could be antagonized by bromsulphthalein, an OATP competitive inhibitor.^19,20,22,23 The mechanisms underlying the uptake of NIRF dyes in cancer cells have been shown to be mediated by concerted actions exerted by hypoxia and activation of the $\mathrm{HIF}1\alpha /\mathrm{OATPs}$ signaling axis, providing a functional link between tumor hypoxia and OATPs to enhance dye uptake.^20,22 As shown in Fig. 5, tumor hypoxia, a frequent phenomenon in a wide range of cancers and typically associated with changes in metabolism, neovascularization, invasion, metastasis, drug resistance, and ultimately poor clinical outcomes,^91,92 mediates dye uptake mainly through $\mathrm{HIF}1\alpha$, which is capable of directly upregulating OATPs.²⁰ Under hypoxic conditions, active nuclear $\mathrm{HIF}1\alpha$ protein interacts with a hypoxia response element (HRE) in the OATP promoter, resulting in increased transcription of OATP genes. By contrast, $\mathrm{HIF}1\alpha$ is destabilized under normoxia due to rapid degradation initiated by the PHD/VHL pathway,²⁹ which is likely the mechanism responsible for the relatively low levels of OATPs generally observed in normal tissues.⁹³ Thus, inactivation of $\mathrm{HIF}1\alpha /\mathrm{OATPs}$ signaling in normal cells and tissues blunts their sensitivity to recruit NIRF dyes for the long-lasting retention and preferential organelle-specific cellular localization seen in cancer cells. The proposed regulatory relationship between $\text{hypoxia}/\mathrm{HIF}1\alpha$ and OATPs was supported by observations that treatment of cancer cells with $\mathrm{HIF}1\alpha$ stabilizers, such as cobalt chloride and DMOG, induced the transcription levels of select isoforms of OATPs in addition to known $\mathrm{HIF}1\alpha$-target genes, which was correlated with increased uptake of NIRF dye by cancer cells. Alternatively, genetic knockdown of $\mathrm{HIF}1\alpha$ in cancer cells reduced the expression of select OATPs, such as OATP1B3 and OATP2B1, along with $\mathrm{HIF}1\alpha$-responsive genes, corresponding with reduced uptake of NIRF dye by cancer cells.^20,22 These mechanisms have been well demonstrated in prostate cancer and brain cancer so far,^20,22,25 and are expected to work in other cancers as well given the prevailing activation of tumor hypoxia and OATP expression in most types of cancers.^94,95 In addition, we have recently established a NIRF imaging method to analyze clinical tumor samples ex vivo, where the activation of $\mathrm{HIF}1\alpha /\mathrm{OATPs}$ signaling could be recapitulated, evidenced by intense widespread nuclear $\mathrm{HIF}1\alpha$ and cytoplasmic OATP1B3 protein expression in RCC samples but not adjacent normal tissues.²⁰

Fig. 5

A schematic outlining the molecular mechanisms underlying the specific uptake of NIRF dyes by cancer cells but not normal cells, which is coordinately mediated by tumor hypoxia, a $\mathrm{HIF}1\alpha /\mathrm{OATP}$ signaling axis and mitochondrial membrane potential.

8.2.

Enhanced Permeability and Retention Effect

In addition to the well-elucidated $\text{hypoxia}/\mathrm{HIF}1\alpha$- and OATPs-mediated mechanisms, alternative mechanisms may also be used by some NIRF dyes, such as IR-780 and Pz 247.^28,96 The uptake of IR-780 iodide by tumor cells has been shown to be dependent on cellular energy metabolism, plasma/mitochondrial membrane potential, and membrane-bound OATPs.²⁸ Both OATP1B3, an influx transporter and multidrug resistance p-glycoprotein-3 (MDR3), an efflux transporter, as main ICG-related transporters play pivotal roles in the uptake of ICG in HCC. The uptake of ICG mediated by OATP1B3 and impaired biliary excretion of ICG caused by MDR3 contribute to the accumulation of ICG in HCC tissues.⁵³ In addition, Ishizawa et al. found that the portal uptake of ICG in HCC cells is mediated by ${\mathrm{Na}}^{+}/\text{taurocholate}$ transport protein and OATP1B3 after preoperative intravenous administration.⁵⁴

Serum components, such as albumin and low-density lipoprotein (LDL), play important roles in the delivery of hydrophobic NIRF dye to cells. For example, Pz 247 is taken up by cells into lysosomes through LDL receptor-mediated endocytosis, leading to accumulation in lysosomes and accompanied NIR emission.⁷⁷ Addition of heparinto at a dose of $25\mathrm{mg}/\mathrm{mL}$ inhibited LDL binding to LDL receptor and completely eliminated the lysosomal localization of Pz 247. Along with the elevated LDL-bound cholesterol in highly proliferative tumor cells relative to normal cells, Pz 247 has been suggested for a potential application for targeting tumor cells.⁹⁶ In short, heptamethine cyanine dyes can achieve organelle-preferential accumulation in cancer cells by enhanced permeability and retention (EPR) effect, which facilitates the entrance of dyes to the intracellular compartment with higher retention in the mitochondria and lysosomes through transmembrane transport proteins or endocytosis.^19,77

8.3.

Reactive Sulfur Species

The targeting of mitochondria by NIRF dyes may be closely related to intracellular reactive sulfur species. Endogenous hydrogen sulfide is an important gasotransmitter involved in critical physiological and pathological processes.⁹⁷ A NIRF multiresponse probe, Cy–NO2, was designed for hydrogen sulfide detection and imaging in live cells.⁹⁸ Further study found that sulfane sulfur instead of ${\mathrm{H}}_2\mathrm{S}$ is the actual signaling molecule, which can be generated as a result of the reaction between ${\mathrm{O}}_2^{-}$ and ${\mathrm{H}}_2\mathrm{S}$ in the mitochondria.⁹⁹ A dual response NIRF probe was employed to image ${\mathrm{O}}_2^{-}/{\mathrm{H}}_2\mathrm{S}$ in cells and in vivo. This probe was derived from the NIRF heptamethine cyanine dye Cy.7.Cl.¹⁰⁰ Because mitochondria are the main production source of ROS in cells, and the mitochondrial fraction contains $\sim 60\%$ of bound sulfane sulfur,¹⁰¹ this probe specifically targets and localizes in mitochondria and can be employed to directly detect changes in mitochondrial ${\mathrm{O}}_2^{-}$ and ${\mathrm{H}}_2\mathrm{S}$.⁹⁹

9.1.

Chemical Modification

Additional chemical conjugations may be needed to enhance the EPR ability of currently available heptamethine cyanine dyes to adjust their lipid and water solubility, to facilitate their transportation through cell membranes and effective recognition of intracellular molecules, to weaken background interference by lowering noncovalent hydrophobic binding of probes to proteins in blood, and to reduce toxicity to normal cells.¹⁰² Moreover, a series of synthetic modifications should be addressed, such as solubility improvement to allow higher injection doses or an alternate route of administration. In addition, given that heptamethine cyanine dyes show preferential accumulation and retention mainly in the mitochondria, which contain most of the sulfane sulfur, we will also be able to design novel NIRF probes for hydrogen polysulfides to further improve mitochondrial targeting.⁷

The ideal fluorescence probe should display high sensitivity in living cells. Improvement of intracellular retention of NIRF dye is a new approach to increase the sensitivity of fluorescence.¹⁰³ Given that individual OATP transport substrates with compound specificity, a highly sensitive, high-throughput fluorescence-based OATP1B1 inhibition assay system was established to screen the most sensitive fluorescence.¹⁰⁴ Several fluorescein derivatives as substrates of OATPs were evaluated for their uptake via OATP1B1-expressing human HEK293 cells.¹⁰⁵ By determination of substrate-dependent $K_i$ variations, dichlorofluorescein showed the highest OATP1B1-mediated uptake among all fluorescent compounds examined. In line with the maintenance of OATP substrate specificity, several structure optimizations of NIRF heptamethine cyanine dyes can be further made by detection of substrate-dependent $K_i$ value.

Protein nanoparticles have been reported to protect NIRF dyes from being destroyed by ROS and reduce bleaching, thereby enhancing the fluorescence stability of the dyes.¹⁰⁶ Protein nanoparticles are more compatible than other forms of nanoparticles and thus can be used as transport carriers for a number of chemotherapy drugs in tumor-targeted therapy. For example, the NIR prodrug dicyanomethylene-disulfide-camptothecin and its derivative, PEG-polylactic acid, when loaded in nanoparticles, emitted NIRF signals.^107,108 These NIR-encapsulated nanoparticles showed higher antitumor activity and longer retention in the plasma than free camptothecin.¹⁰⁸ Additionally, the excellent fluorescence intensity and photostability of DCM chromophores with a large Stokes shift makes the in vivo and in situ tracking of drug release and therapeutic efficacy in live animals possible.¹⁰⁸

9.2.

Preclinical Experiments

The penetrability and sensitivity of NIRF dyes need to be improved before clinical use. For example, NIRF dyes can be labeled with radionuclide for combined PET/CT scanning for early diagnosis of microtumors in vivo.^32,109 Bright NIRF agents further conjugated to disease-targeting moieties promise molecular imaging and image-guided surgery. In addition to the mouse or rat tumor models usually preferred for preclinical testing, it would be worthwhile to verify the reliability and sensibility of NIRF imaging in large animal models, such as dog or pig models carrying different types of spontaneous tumors, to address essential issues in human cancers that are not reflected in small animal model systems.¹¹⁰ Finally, crucial evaluations of pharmacokinetics, pharmacodynamics, and toxicity of dyes and dye-drug conjugates are also needed in mouse and dog models before this group of novel compounds can be moved into the clinic for improved cancer diagnosis, prognosis and therapy.¹⁰⁸ Taking into account the complexity of targeted drug development, the development of diagnostic NIRF agents for tumor imaging should give the priority to the selection conducted in preclinical experimental models, where both the PDX mouse model and dog spontaneous tumor model are obviously convenient tools for preclinical tests.

In summary, a group of NIRF heptamethine cyanine dyes with dual imaging and targeting properties are promising novel anticancer theranostic agents. These NIRF dyes and their derivatives can be exploited for the detection and treatment of different types of tumors to meet preclinical and clinical needs without chemical conjugation. Although obstacles remain, valuable and exciting opportunities exist in this field.