1.

Introduction

Understanding the fluid mechanics of the interstitium is indispensable for a variety of clinical situations and research areas, including immunology, pharmacokinetics, and oncology. However, even such fundamental issues as the regulatory mechanism and route of solute movement between microvessels, the interstitium, and the lymphatics have yet to be elucidated.¹

In the past decade, optical imaging technology, such as molecular imaging of living tissues with fluorescence emission, has been rapidly developed for the noninvasive evaluation of disease progression or drug effects. When visualizing fluorescent contrast agents existing in deep tissues, absorption and scattering of the excitation and fluorescence emission light by many biocomponents can be serious technical issues. In addition, autofluorescence, i.e., the natural emission of light by certain biological molecules, interferes with and obscures the detection of extrinsic fluorescent contrast agents, causing structures other than those of interest to become visible and manifest as a high background signal, which reduces the image contrast and clarity.² Near-infrared (NIR) light in the wavelength range of 700 to 900 nm can be used to overcome these issues. The absorption coefficient of physiological tissues is relatively low in the NIR region than in the visible light region.^3–5 Furthermore, using the excitation light in the NIR region can prevent tissue autofluorescence.^2,6 As a result, NIR light can penetrate into deep tissues and visualize the fluorescent contrast agents existing in those tissues, with a high contrast between the optical signals and the background fluorescence. Organic dyes and inorganic quantum dots (QDs) are two representative NIR fluorescent contrast agents that are currently available in the biomedical fields for both animal testing and clinical applications.⁷ Indocyanine green (ICG), a member of the NIR heptamethine indocyanine fluorochromes, is the US Food and Drug Administration (FDA)-approved and can be used safely in humans. Thus, it has been widely employed for in vivo optical imaging, such as fluorescent angiography, sentinel lymph nodes (SLN) mapping, and lymphangiography.^8–13 The molecular weight of ICG is 775 Da. ICG is water soluble because of its sulfonyl groups, but can form dimers and oligomers in aqueous solution. These aggregates cause self-quenching of ICG, which reduces its fluorescence in water. When administered to living tissues, ICG rapidly binds to plasma proteins such as albumin, globulins, or lipoproteins^14–16 and then becomes fluorescent under NIR light. The in vivo dynamics of ICG, therefore, are defined by those plasma proteins. When considering lymphatic and venous uptake from the interstitial fluid, the behavior of nanoparticles is greatly dependent on the particle size.^17–21 The particle size of plasma protein-bound ICG becomes $\sim 7$ to 22 nm in diameter;¹⁶ thus, fluorescent imaging using ICG can only partially reveal the overall movement of materials because it cannot demonstrate the movement of small materials (i.e., less than a few nanometers). Inorganic QDs, which are composed of a semiconductor core commonly made from CdSe or CdTe coated with an inorganic shell, are nanometer-sized fluorescent probes with unique optical and electronic properties. QDs have high quantum yields, high chemical and optical stabilities,^22–25 and tunable characteristics depending on the size and doping characteristics of the nanocrystals.²⁵ Thus, QDs represent a promising and powerful approach for in vivo tumor imaging or targeted drug delivery.^22,26–28 However, the safety of QDs remains a serious concern because toxic elements are present in their cores.^29,30

To address these problems, we developed an NIR fluorochrome that is a di-$\beta$-cyclodextrin-binding indocyanine derivative (named here as TK-1), which has two hydrophilic cyclic sugar chain cyclodextrins (Fig. 1). Here, we determined the optical properties of TK-1 in vitro in comparison with ICG. In addition, we report that the lymphatic and venous transport of fluorescent contrast agents, TK-1 and ICG, was differentially regulated depending on their particle sizes, based on real-time fluorescence imaging results in rats.

Fig. 1

Structural formulas of ICG and TK-1. (a) ICG is an amphiphilic molecule; two polycyclic groups interconnected by a carbon chain impart lipophilicity to the ICG molecule, and the sulfate groups are responsible for the hydrophilic nature. (b) TK-1 has two hydrophilic cyclic sugar chain $\beta$-cyclodextrins.

2.

Materials and Methods

2.5.

Animal Studies

2.5.1.

NIR imaging

Real-time animal imaging was performed using a fluorescence stereomicroscope (SZX-RFL2, Olympus, Tokyo, Japan) equipped with a single-band filter set optimized for ICG (ICG-B-000; excitation wavelength $769\pm 20.5\mathrm{nm}$, emission wavelength $832\pm 18.5\mathrm{nm}$; Semrock Inc., Rochester, New York). All bright-field and fluorescence images were taken by an electron-multiplying, cooled charge-coupled device (CCD) camera (QuantEM 512SC, Photometrics, Tucson, Arizona), which was mounted on the top of the stereomicroscope. The fluorescence intensity of each image was analyzed by Image J software (National Institutes of Health, Bethesda, Maryland).

2.5.2.

Experimental animal model and surgical procedure

The animal study protocol was approved by the local animal ethics committee, in accordance with the Regulations for Animal Experiments at Nagoya University and the Japanese Government Animal Protection and Management Law (Act No. 105 of October 1, 1973). Twenty 10-week-old, male Sprague–Dawley rats (Charles River Japan Inc., Kanagawa, Japan) were used in this study, with a mean weight of 350 g. All animals were housed under specific pathogen-free conditions up to the time of the experiments.

All surgical procedures were performed under isoflurane anesthesia using a vaporizer for small animals (SF-B01; DS Pharma Biomedical. Co., Ltd., Osaka, Japan). After removing the body hair covering the lower extremity and groin of each rat, a longitudinal incision was made from the medial thigh to the groin, and the femoral vein and accompanying lymphatics were exposed by retracting the external oblique muscle (Fig. 2). Rats were then placed on a fluorescence stereomicroscope in the supine position. An aqueous solution of TK-1 was prepared by dissolving the green amorphous powder of TK-1 in distilled water. One hundred microliters of a 1 mM solution of TK-1 was injected subcutaneously into the hind footpad at a constant rate ($20\mu \mathrm{L}/\sec$) using a 27-gauge needle and a syringe pump (YSP-101, YMC Co., Ltd., Kyoto, Japan). Fluorescence images at the exposed groin site were observed in real time using a fluorescence stereomicroscope and were taken every 5 sec for up to 2 h with a CCD camera. For comparison purposes, we performed the same examination using $100\mu \mathrm{L}$ of a 1 mM ICG solution ($n=5$ per group). The pharmacokinetics of TK-1 and ICG were analyzed by measuring fluorescence intensity in the femoral vein or accompanying lymphatics, using Image J software. A small rectangle for the region of interest (ROI) was selected in the area of the femoral vein or one of accompanying lymphatics that exhibited peak fluorescence, and the average fluorescence intensity of each ROI was plotted as a function of time.

Fig. 2

Surgical field at the groin site of rats. The femoral vein (arrow) and an accompanying lymphatic (arrow-head) were exposed by retracting the external oblique muscle.

In addition, to confirm the presence of fluorescent agents (TK-1 or ICG) within the venous flow, venous blood was collected from the femoral vein at 10 min after injection. After acquiring a drop of venous blood, its fluorescence intensity was calculated using Image J software ($n=5$ per group). Data were analyzed by performing an unpaired Student $t$ test using SPSS version 19 software (SPSS, Chicago, Illinois); a $p$-value of $<0.05$ was considered statistically significant.

3.

Results

3.1.

Synthesis of TK-1

Synthesis of TK-1 was performed as outlined in Fig. 3. Synthesis of intermediate 3, which contains two carboxylic acid groups that bind with two cyclodextrin molecules, was performed as described previously.^35,36 However, the product yield was not suitable for effective TK-1 preparation. Thus, we developed a convenient and practical high-yield method (83% yield) for preparing compound 3 from 1.³¹ Compound 3 was treated with potassium methoxide prepared from methanol and $t$-BuOK in anhydrous methanol to afford 4 in a quantitative yield. Condensation of 4 with 2A-$O$-(3-Aminopropyl)-$\beta$-cyclodextrin (5)³² was successfully achieved in DMF/pyridine. The final product (TK-1) was purified using a C-18 open column chromatography, using 1 mM aqueous HCl and 1 mM HCl/methanol, and obtained as a green amorphous powder with a 69% yield. The three-dimensional particle size of TK-1 was calculated by computer-aided analysis. The structure of TK-1 was first drawn in two dimensions with ChemDraw software (PerkinElmer Inc., Waltham, Massachusetts) and then transformed to a three-dimensional structure using ChemBio3D Ultra 13.0 (PerkinElmer Inc., Waltham, Massachusetts). Based on results using these software packages, the largest diameter of TK-1 was calculated to be $<3\mathrm{nm}$. The effective molecular weight of TK-1 was 3441 Da. TK-1 has two hydrophilic cyclodextrin moieties, which enable water solubility.

Fig. 3

Synthesis of the di-$\beta$-cyclodextrin-bound indocyanine derivative, TK-1, in three steps. Intermediate 4 contains two carboxylic acid groups that bind with two $\beta$-cyclodextrin molecules (5).

3.2.

Optical Properties of TK-1

The absorption and fluorescence spectra of TK-1 in PBS at pH 7.4 are shown in Fig. 4. The absorbance of TK-1 was found to center at 800 nm ($ϵ=\mathrm{200,000}{\mathrm{M}}^{-1}{\mathrm{cm}}^{-1}$). As shown in Fig. 4, the shapes of the TK-1 absorption spectra were nearly identical at several concentrations (0.01 to 0.0001 mM), indicating that no or low self-aggregation occurred between the TK-1 molecules. In contrast, the ICG absorbance at 700 nm increased with the ICG concentration, which corresponded to H-aggregation.^6–8 The low self-aggregation property of TK-1 in PBS can be explained by the presence of the two hydrophilic cyclodextrin moieties. The TK-1 fluorescence in PBS at 25°C centered at 812 nm and the TK-1 fluorescence quantum yield (${\mathrm{\Phi }}_{\mathrm{F}}$) at 25°C was 0.048, which was calculated based on that of ICG in DMSO.³³ This TK-1 quantum yield was slightly higher than that of ICG in PBS at 25°C (${\mathrm{\Phi }}_{\mathrm{F}}=0.041$).

Fig. 4

Absorbance (solid line) and fluorescence (dotted line) spectra of (a) TK-1 and (b) ICG in PBS at pH 7.4 and 25°C. The TK-1 absorbance centered at 800 nm. The shapes of the TK-1 absorption spectra were nearly identical at several concentrations (0.01 to 0.0001 mM), indicating that no or low self-aggregation occurred between the TK-1 molecules. In contrast with TK-1, the shapes of the ICG absorption spectra were not identical at concentrations ranging from 0.01 to 0.0001 mM, indicating that ICG self-aggregates.

The three-dimensional fluorescence spectra of (a) TK-1 and (b) ICG in human venous whole blood at 25°C are shown in Fig. 5 (concentration of TK-1 and ICG: $20\mu \mathrm{M}$). TK-1 showed robust fluorescence excitation over excitation wavelengths ranging from 760 to 780 nm, as did ICG. Figure 6 shows fluorescence-quenching curves for TK-1 and ICG in human venous whole blood at 25°C. In the concentration range of 1 to $200\mu \mathrm{M}$, the TK-1 fluorescence intensity increased in a dose-dependent manner, up to the maximum of $200\mu \mathrm{M}$. At higher concentrations, the TK-1 fluorescence was slightly quenched, although the ICG fluorescence was quenched at concentrations exceeding $\sim 40\mu \mathrm{M}$. In addition, the maximum fluorescence peak gradually shifted from 813 to 825 nm with increasing TK-1 concentrations, compared to the marked increase in the wavelength for the maximum fluorescence peak of ICG (Fig. 6). These results indicated that TK-1 showed a lower tendency for aggregation in whole blood than did ICG. A high degree of ICG aggregation decreases its fluorescence intensity, resulting in inaccuracies in fluorescence imaging, while the use of TK-1 can prevent this because it does not aggregate appreciably.

Fig. 5

Three-dimensional fluorescence spectra of (a) TK-1 and (b) ICG in human venous whole blood at 25°C (concentration of TK-1 and ICG: $20\mu \mathrm{M}$). Both TK-1 and ICG showed robust fluorescence excitation at excitation wavelengths 760 to 780 nm.

Fig. 6

Relationships between (a) the fluorescence intensity at the maximum fluorescence peak (excitation wavelength: 760 nm), (b) the wavelength of maximum fluorescence peak (excitation wavelength: 760 nm), and the TK-1 and ICG concentrations in human venous whole blood at 25°C. TK-1 showed a tendency for decreased aggregation and low self-quenching in whole blood, compared to that observed with ICG.

A photostability study of TK-1 and ICG was performed by taking fluorescence measurements ($\text{excitation}=760\mathrm{nm}$, $\text{emission}=810\mathrm{nm}$) at a probe concentration of $0.1\mu \mathrm{M}$ in PBS at pH 7.4 at 25°C (Fig. 7). When the TK-1 solution was continuously exposed to light at 760 nm, the fluorescence intensity showed a more gradual decrease over time than did ICG. When not exposed to light, the fluorescence of TK-1 and ICG remained nearly constant.

Fig. 7

Photostability of TK-1 and ICG in PBS exposed to 760 nm light at 25°C. $F/F_0$ is the fluorescence-intensity ratio at 810 nm ($\text{excitation wavelength}=760\mathrm{nm}$) at the indicated time points ($F$) to the initial fluorescence intensity ($F_0$). When the TK-1 solution was continuously exposed to light at 760 nm, the fluorescence intensity decreased more gradually over time than did ICG.

3.3.

NIR Imaging in Rats

3.3.1.

Real-time fluorescence imaging

We obtained continuous fluorescence images at the exposed groin site of rats and analyzed the fluorescence intensity of a femoral vein and accompanying lymphatics, in which TK-1 or ICG was transported. During real-time fluorescence imaging of the TK-1 injected group, the femoral vein was visualized first and successively accompanying lymphatics were then detected, as shown in Fig. 8(a). In the ICG-injected group, most ICG was transported via the lymphatics, as shown in Fig. 8(b). ICG in the femoral vein was not detected. Subsequently, we plotted the time-dependent changes of the average fluorescence intensities of the ROIs defined on the femoral vein and an accompanying lymphatic, which exhibited the peak fluorescence. We found that the transport of TK-1 via lymphatics showed a steep peak at 10 to 15 min after injection, which then decreased immediately to almost the same level as that of the femoral vein [Fig. 9(a)]. In contrast, ICG was transported only through the lymphatics at a nearly constant rate [Fig. 9(b)]. In addition, we performed statistical analysis of the time-dependent data. First, we calculated the average time required for TK-1 and ICG to travel from the footpad to the groin lymphatics. The average time required before detecting the maximum peak of the lymphatic fluorescence was $10.7\pm 3.3\min$ in the TK-1 group and $24.8\pm 11.6\min$ in the ICG group ($p<0.05$). Second, we also calculated the average residence time of TK-1 and ICG in the lymphatics and femoral vein. The average residence time until the lymphatic fluorescence decreased to 20% of the maximum intensity was $67.6\pm 25.7\min$ in the TK-1 group and $120.0\pm 0.0\min$ in the ICG group (i.e., ICG remained in the lymphatics for $>2\mathrm{h}$) ($p<0.01$). On average, TK-1 remained in the femoral vein for $>2\mathrm{h}$, whereas ICG was not detected from the beginning.

Fig. 8

Real-time fluorescence imaging; bright field images versus fluorescence images at 0 to 120 min postinjection. TK-1 and ICG were injected into the hind footpads of rats at 0 min. (a) In the TK-1-injected group, the femoral vein was visualized first (arrow, 2-min image) and successively accompanying lymphatics were detected next (arrow head, 10-min image). (b) In contrast, in the ICG-injected group, most ICG was transported via the lymphatics (arrowhead, 20-min image), and ICG in the femoral vein was not detected.

Fig. 9

Time-dependent changes of fluorescence intensities detected in the lymphatic and venous flow, with two representative cases shown for each group. The upper-left image in each group shows a bright field image, and the upper-right image shows the fluorescence. The small rectangles for the ROIs were defined, based on one of the accompanying lymphatics (green rectangles) and femoral veins (blue rectangles). The lower images show the measured fluorescence intensities of lymphatics (green lines) and femoral veins (blue lines). TK-1 and ICG were injected into the hind footpads of rats at 0 min. The transport of TK-1 via a lymphatic showed a steep peak at 10 to 15 min postinjection, which then decreased immediately to nearly the same level as observed with the femoral vein. In contrast, ICG was only transported through the lymphatics at a nearly constant rate.

3.3.2.

Fluorescence intensity of venous blood

We calculated the intensity of single drops of venous blood from rats in the TK-1-injected and ICG-injected groups, using Image J software. The mean fluorescence intensities of venous blood collected at 10 min were $13824.7\pm 2146.2$ and $133.6\pm 11.3$ arbitrary units ($p<0.001$; Fig. 10). These results indicated that TK-1 actually infiltrated into the venous flow, whereas ICG did not.

Fig. 10

Fluorescence intensity of venous blood collected at 10 min from rats in the TK-1-injected group (a) and the ICG-injected group (b). The mean fluorescence intensity in the TK-1 group was significantly higher than that measured in the ICG-injected group (***$p<0.001$).

4.

Discussion

NIR-excitable fluorescent contrast agents facilitate a broad range of applications for noninvasive in vivo imaging, due to their low tissue autofluorescence and deep tissue penetration in this spectral range.^2–6 Regarding NIR fluorochromes that are effective in vivo optical imaging, several general properties are desirable, including a high quantum yield, high optical and chemical stabilities, high water solubility, low nonspecific binding, and low toxicity. ICG is an FDA-approved NIR fluorescence-contrast agent that has been widely used for many therapeutic and diagnostic applications, such as cardiac output measurement, liver functional studies, and fluorescence angiography, because it has low toxicity and the capacity to absorb and emit in the NIR spectral region.^8,9,37–40 Recently, ICG fluorescence imaging was also developed for SLN mapping and lymphangiography.^{10–13,41–47} However, conventional ICG has several physicochemical disadvantages. ICG is water-soluble because sulfonyl groups are bonded thereto, whereas it has lipophilicity because of the two hydrophobic polycyclic groups.^48,49 Due to its amphiphilic character, ICG has a propensity for adsorbing to lipids. The adhesion or leakage of ICG to undesired tissues, therefore, may interfere with surgical operations or medical diagnoses because ICG cannot be easily removed from the biological tissues to which ICG can adhere. Second, ICG can self-aggregate in water. The optical properties of ICG strongly depend on dye concentration. In aqueous solution, ICG starts to form dimers and oligomers, even at relatively low concentrations.^3,50–52 The presence of aggregates causes self-quenching of the dye, which reduces its fluorescence. In addition, ICG undergoes degradation in aqueous media resulting in the simultaneous loss of absorption and fluorescence.⁵ Finally, ICG administered into living tissues immediately binds to plasma proteins, such as albumin, high-density lipoprotein, and low-density lipoprotein.^14–16 The particle size of plasma protein-bound ICG becomes $\sim 7$ to 22 nm in diameter,¹⁶ and the vascular or tissue permeability of ICG is defined by those proteins.^16,53 Most ICG fluorescence observed in living tissues, therefore, demonstrates the in vivo dynamics of plasma lipoproteins, especially that of HDL alone, or in combination with LDL.¹⁶ Thus, it is possible to study material movement only partially when using ICG.

In contrast, inorganic QDs, which are composed of a semiconductor core commonly made from CdSe or CdTe coated with an inorganic shell, have several desirable characteristics, such as nano-size particles, high quantum yields, and high chemical and optical stabilities.^22–25 Moreover, QDs are tunable, based on their size and structure, so that a narrow-width emission light can be generated across the visible and NIR spectrum, depending on the size and doping characteristics of the nanocrystals.²⁵ Nonetheless, issues remain that need to be addressed before these nanocrystal probes can be broadly used in biomedical applications. Importantly, the long-term biocompatibility and the potential toxicity of QDs need to be improved.^29,30,54

The di-$\beta$-cyclodextrin-bound indocyanine derivative, TK-1, can overcome these problems. TK-1 was originally designed to mimic the absorption and fluorescence of ICG. A noteworthy feature of TK-1 is that it possesses two water-soluble cyclodextrin molecules. Cyclodextrins are cyclic oligosaccharides produced by the enzymatic digestion of starch and are named $\alpha$-, $\beta$-, and $\gamma$-cyclodextrin depending on the number of glucose units included (6, 7, or 8 glucose units, respectively). Cyclodextrins have a toroidal shape with a hydrophobic surface inside and form inclusion compounds with smaller hydrophobic compounds of appropriate size, which fit into their 0.5 to 0.8 nm cavity. These inclusion complexes have many advantages, such as increased solubility, enhanced bioavailability, and improved stability.^55–57 In this study, we expected that TK-1 would have high chemical stability, low self-aggregation, low self-quenching, and low binding to tissues because of its two $\beta$-cyclodextrin molecules. To take advantage of the cyclodextrin properties, two short carbon-chain linkers were placed between the indocyanine fluorophore and cyclodextrin. The results of our in vitro optical studies indicated that both hydrophilic cyclodextrin moieties of TK-1, which potentially enclose the hydrophobic indocyanine moiety, enabled TK-1 to exist as a monomer in aqueous solution (with no or low self-aggregation) and consequently to achieve high water solubility and high optical and chemical stabilities. In particular, the results from our photostability study indicated that the cyclodextrin moiety can protect the indocyanine fluorophore from photodegradation and suggested that TK-1 would be effective during lengthy in vivo optical imaging because of its low photobleaching property. Furthermore, we expect that TK-1 will show low nonspecific binding in vivo, owing to its hydrophilic cyclodextrin moieties and reduced tendency for self-aggregation. The Stoke’s shift of TK-1 is much smaller than that of ICG. However, in general, when the Stoke’s shift is small, the excitation and emission wavelengths will be set at slightly shorter and slightly longer wavelengths, respectively. In NIR fluorescence imaging using TK-1, a filter set optimized for ICG (ICG-B-000) can efficiently detect TK-1 fluorescence (Fig. 5). Regarding the toxicity of TK-1, we suggest that TK-1 would be as safe as ICG because it has no toxic elements in its core, in contrast to inorganic QDs. In fact, our preliminary data indicated that tail vein injection with $0.75\mathrm{mg}/\mathrm{kg}$ TK-1 had no adverse effects in rats. Furthermore, another water-soluble indocyanine compound, which we prepared by a method similar to that used to synthesize TK-1, was nontoxic in animal experiments based on good laboratory practice (unpublished data).

As a further interesting result, our real-time in vivo fluorescence imaging data demonstrated that the in vivo dynamics of TK-1 differed markedly from those of ICG with respect to both lymphatic and venous transport. More specifically, TK-1 drained from hind footpad via the peripheral venous flow and the lymph flow, whereas ICG drained only through the lymphatics. Recently, some physiologists have concluded that the original Starling equation does not describe fluid transport adequately and that fluid reabsorption does not occur across blood vessels due to their steady-state properties; consequently, the lymphatic vasculature provides the only escape route for extravasated fluid and macromolecules.^1,58 In contrast, from a radiological point of view, it has been demonstrated that the behavior of nanoparticles injected subcutaneously is strongly dependent on their particle size.^17,18 Some studies regarding contrast agents used for magnetic resonance (MR) lymphangiography or lymphoscintigraphy have provided evidence that the particle size is the major determinant of lymphatic uptake from the interstitial fluid, and particles with small diameters (i.e., less than a few nanometers) are mostly exchanged through the blood capillaries.^19–21 In addition, regarding QDs, some authors have emphasized that the in vivo behavior of QDs was greatly dependent on their hydrodynamic diameters when considering renal clearance or retention in the SLN.^59–63 As noted above, the regulatory mechanism and route of solute movement are controversial. The particle size of TK-1 in aqueous solution is $<3\mathrm{nm}$. We suggest that TK-1 is probably absorbed into microvessels and lymphatics at the injection site because its particle size is smaller than that of plasma protein-bound ICG, $\sim 7$ to 22 nm. Furthermore, regarding the lymphatic transport, we found that time-dependent changes of TK-1, which showed a steep peak, were quite distinct from those of ICG. These results indicated that solute movement via the lymphatic and venous pathways can be differentially regulated, depending on the particle size. The advantage of NIR fluorescence imaging using TK-1 against MR lymphangiography or lymphoscintigraphy is that TK-1 provides real-time imaging in an inexpensive and convenient manner. In addition, the intravascular administration of TK-1 may also be valuable for NIR fluorescence imaging. In fact, we performed tail vein injection of TK-1 in rats and subsequently measured its fluorescence. The results demonstrated that TK-1 showed higher vascular permeability than ICG and could more sensitively detect mild states of localized inflammation or edema (unpublished data). We propose that TK-1 can be used to visualize the route of microcirculation, which cannot be shown by conventional ICG, and can enable real-time NIR fluorescence imaging of whole-fluid and solute movements via both the lymphatics and microvessels.

This study has some limitations. Regarding the characteristics of TK-1, we did not investigate some of its optical properties, such as the circulation time or the fluorescence lifetime. Furthermore, the following limitations should also be considered when interpreting our results. We analyzed the pharmacokinetics of TK-1 by observing the fluorescence intensities at the groin site, which is separated from the injected hind footpad; thus, we did not identify the exact peripheral route at the injected site through which TK-1 was absorbed. In addition, there is a possibility that factors other than the particle size, such as the surface charge or shape of a particle, affect the regulation of solute movement. Additional experimental methods should be devised that can solve these problems. Nevertheless, TK-1 is surely an NIR fluorescence dye, which can overcome the disadvantages of ICG, and we believe that the use of TK-1 can provide new opportunities for the visualization of microcirculation.