Hydrophobic cyanine dye-doped micelles for optical in vivo imaging of plasma leakage and vascular disruption

Abstract. Vascular leakage is an important feature of various disease conditions. In vivo optical imaging provides a great opportunity for the evaluation of this phenomenon. In the present study, we focus on the development and validation of a near-infrared (NIR) imaging formula to allow reliable, cost-efficient evaluation of vascular leakage in diverse species using the existing small-animal fluorescence imaging technology. IR-676, a moderately hydrophobic NIR cyanine dye, was doped into self-assembling aqueous micelles using a widely employed and safe nonionic emulsifier (Kolliphor HS 15), and was tested in several acute and chronic inflammatory disease models in both mice and rats. The imaging formula is stable and exerts no acute toxic effects in vitro. It accumulated specifically in the inflamed regions in all models, which could be demonstrated by both conventional epifluorescence imaging, and fluorescence tomography both as a standalone technique and also by merging it with computed tomography scans. Ex vivo verification of dye accumulation by confocal fluorescence microscopy was also possible. The present formula allows sensitive and specific detection of inflammatory plasma leakage in diverse models. Its potential for imaging larger animals was also demonstrated. IR-676-doped micelles offer an excellent opportunity to image inflammatory vascular leakage in various models and species.

Hydrophobic cyanine dye-doped micelles for optical in vivo imaging of plasma leakage and vascular disruption 1 Introduction The interest in near-infrared (NIR) optical in vivo imaging is constantly growing since its debut in the early 2000s, 1 and the sensitivity and versatility of the instruments have increased considerably since then. However, there is only a narrow range of reliable contrast agents currently available, and this is especially true for fluorescence markers dedicated to image plasma leakage from diseased vasculature. Most of these probes either suffer from low contrast and signal-to-background ratio, or have tremendous operational costs, thereby limiting large-scale imaging, and in general, do not enable studies involving research animals larger than mice.
The dilation and increased permeability of the microvessels caused by the contraction of the endothelial cells in the area of inflammation and subsequent formation of intercellular gaps is an important adaptive change during acute vascular hyperpermeability, as this helps the transmigration of immune cells (e.g., leukocytes, T, and B cells) as well as the transport of antibodies and inflammatory mediators to the site of the disease. 2 Prolonged tissue ischemia and the following reperfusion leads to a similarly increased perfusion and permeability of the vasculature within the affected area. This process leads to the formation of a plasma protein-rich fluid exudate as well, thereby contributing to edema formation, a characteristic feature of acute inflammation. This takes place primarily in the postcapillary venules. 3 Chronic vascular hyperpermeability (CVH) is mainly observed in tumors and chronic inflammatory conditions such as rheumatoid arthritis or immune-mediated dermatitis. The main difference of CVH is that the plasma leakage occurs primarily from newly formed pathological, "leaky" vessels with weakened endothelial structures. 3 It is not surprising that noninvasive, longitudinal, real-time evaluation of these processes provides a great opportunity to estimate the extent and characteristics of these conditions, for which NIR optical imaging is an excellent alternative in preclinical settings compared to other modalities (computed tomography etc.) due to the availability of tabletop instruments and lack of ionizing radiation. 4 The underlying principle of optical in vivo imaging of this phenomenon is based on employing NIR fluorophores (emission maximum between 650 and 900 nm), whose particle size is great enough to limit their escaping from healthy vasculature under normal conditions; however, they retain their ability to leave leaky vessels of inflamed or tumor tissues. The earliest example was the Evans-blue method (Miles-assay), which relied on the iv. administration of the azo dye with high affinity for serum albumin, for which the degree of extravasation could be traditionally determined by spectrophotometry. 3 It was recently demonstrated though that ex vivo NIR imaging of the affected organs is superior in terms of both specificity and sensitivity, 5 and in vivo fluorescence imaging of this dye is also feasible. 6 Unfortunately, this technique gives meaningful results only with a very high amount of concentrated dye, which necessitates the sacrificing of the subject, and can be easily biased if the residual dye in the vasculature is not properly removed. Consequently several authors employed other, mainly cyanine fluorophores such as indocyanine green (ICG), Cy5.5, or IR-820 which also have nonspecific binding ability to albumin and other plasma proteins. [7][8][9][10] The signal strength and contrast of these techniques remained relatively modest despite the high doses administered, presumably due to aspecific binding, small size, and the short plasma half-life (typically only minutes) of these compounds regardless of plasma protein binding (as the parenchymal cells of the liver rapidly clear these substances from the circulation). However, these studies substantiated that (1) cyanine dyes are, in general safe, and can be used for noninvasive tracking of plasma extravasation and (2) the signal intensity of extravasated dye is relatively stable, mainly due to nonspecific uptake by phagocytes. Among the dyes currently used, ICG received by far the most attention despite its significant disadvantages compared to novel, advanced cyanine fluorophores (low stability in aqeous media and low quantum yield, small Stokes-shift, autoaggregation, as well as poor solubility in physiologic buffers) as it is currently the only FDA-approved NIR-dye due to historical reasons. 11 Additionally, its cytotoxic effects have also been reported. 12,13 An advanced ICG-derivative SIDAG demonstrated somewhat better sensitivity and lower toxicity than ICG; nevertheless, it still required relatively large doses to be administered and contrast remained negligible at early imaging time points. 14 To overcome the difficulties outlined above, and to get better results, several investigators prepared preformulated NIR contrast agents that do not rely on plasma protein binding either by (1) performing a controlled labeling of a macromolecule (albumin, dextranes, etc.) in vitro, prior to use [15][16][17][18] or (2) synthesizing fluorescently labeled microparticles whose extravasating capabilities are limited by their size. [19][20][21] The latter method could especially provide additional benefits, as the incorporation of the dye into a nanoprobe minimizes the potential cytotoxicity and delays systemic elimination, thereby leading to prolonged plasma half-life. Here, we have employed Kolliphor HS 15 for this purpose, a nonionic solubilizer that forms stable micelles in water and physiologic buffers with an average size of 12 nm, and has a negligible toxicity on its own. 22 Based on these findings there is a particular need in preclinical inflammation and vascular research to have an NIR contrast agent with the following characteristics: (1) absolute compatibility with both the existing fluorescence molecular tomographic technology and confocal fluorescence microscopy to enable ex vivo verification, (2) imaging capability of both superficial and deep-tissue plasma leakage with a broad spectrum of applicability, (3) simple and cost-efficient synthesis requiring only standard life sciences laboratory equipment, and downto-earth steps during the process, and (4) potential for imaging larger animals (rats etc.). Therefore, we aimed at synthesizing a novel formula to fulfill these criteria and validate it in well established and routinely used rodent inflammation models of different organs.

Preparation of the IR-676-Doped Micelles
The moderately hydrophobic pentamethine cyanine fluorescent dye IR-676 (1,1′,3,3,3′,3′-hexamethyl-4,5,4′,5′-dibenzoindodicarbocyanine)-iodide [ Fig. 1(a)] was selected as a promising candidate, as (1) it has low water solubility (but is still not insoluble), and is moderately soluble in other polar solvents (e.g., lower alcohols) (2) its excitation/emission wavelength maxima differ only marginally from some other widely used "gold standard" NIR fluorophores (Cy5.5, Alexa Fluor 680, etc.), thereby ensuring compatibility with the existing imaging technology and wavelength filters. It has also been previously demonstrated that this dye is an excellent calibration standard for fluorescence tomography systems. 23 IR-676 was obtained in powder form (MW: 610.58 g∕mol, excitation maximum: 676 nm, emission maximum: 700 nm, Spectrum-Info Ltd., Kyiv, Ukraine). The dye can be easily dissolved in dichloromethane or alcohols. 23,24 Taking into consideration the incompatibility of dichloromethane with most plastic materials, the dye was initially dissolved in methanol at room temperature (RT), and was ultrasonicated then stirred until it formed a clear, deep blue solution with a concentration of 2 × 10 −4 M. This stable stock solution was either immediately used, or stored at −20°C protected from light. 100 μl batches were pipetted into microtubes which were covered with a parafilm membrane pierced by a 27G needle three to four times to allow evaporation of the solvent at a moderate rate. The vials were then centrifuged for at least 60 min in a Speed Vac centrifuge (Savant RVT4104 refrigerated vapor trap, Savant Instruments Inc, Farmingdale, New York), leaving behind a thin film of dried dye at the bottom of the vial, which was stored light protected at 4°C. For the reconstitution, a 5 w/v% solution of the nonionic surfactant Kolliphor HS 15 (Polyethylene glycol-15-hydroxystearate, Sigma-Aldrich) was prepared in sterile deionised water. Kolliphor HS 15 was selected based on its outstanding in vivo safety and previously demonstrated effectiveness for imaging contrast agent formulas such as ICG. 22,[25][26] 500 μl batches of this medium were directly pipetted into the microtubes containing the dye at RT. Complete dissolution of IR-676 could be achieved within 10 to 20 min by vortexing and ultrasonication. The resulting stable solution (4 × 10 −5 M) was either immediately used, or stored at 4°C for a maximum of 48 h.

Animals
For the experiments, 10 to 12-week-old C57Bl/6 mice (weighing 20 to 25 g) and 5-week-old Wistar rats (weighing 125-150 g) were used, for which the original breeding pairs were obtained from Charles-River Hungary Ltd.

Imaging Protocol
The IR-676-based contrast agent was injected iv. in a dosage of 2 × 10 −7 mol∕kg body weight, corresponding to a convenient injectate volume of 100 to 150 μl for mice or 0.6 to 0.8 ml for rats. The method of delivery was either lateral tail vein (oxazolone-and carrageenan-model) or retro-orbital injection (in the other models). Imaging was performed at several selected time-points with the in vivo imaging system (IVIS) Lumina II (PerkinElmer Ltd., Waltham, Massachusetts) with a 640-nm excitation (535, 570, and 605 for spectral unmixing) and ∼700-nm emission filters, auto exposure time, and a binning factor of 2. Fluorescence was expressed as the total radiant efficiency [ðphotons∕s∕cm 2 ∕steradianÞ∕ðμW∕cm 2 Þ]. The fluorescence molecular tomography (FMT) 2000 system (PerkinElmer Ltd.) was calibrated using the instrument's 680-nm laser channel, and a sample solution of 1 × 10 −6 M following the instructions of the instrument's manufacturer. Imaging with the FMT system was performed uniformly 3 h postinjection due to the lower throughput of the three-dimensional (3-D) imaging system using high-resolution acquisition parameters. Fluorescence was expressed as either the calculated amount of fluorophore (pmol) or as surface fluorescence (counts/energy). In the airway inflammation model, the extent of the inflamed area was also expressed in μm 3 . Animals were anesthetized for the injection and imaging with ketamine-xylazine ip. (dosage: 100∶5 mg∕kg). During anesthesia, body temperature was maintained by a heating lamp and vital signs were monitored by observation until recovery. Home cage activity after recovery was also observed.

Interaction with Serum Proteins
To determine whether IR-676 binds to serum proteins in its freely dissolved or encapsulated state, ethanol-based (20 v/v %) or micellar (5 w/v%) solutions (10 μl, 4 × 10 −5 M) of the dye were incubated with bovine serum albumin (BSA) (0.5 mg∕ml), fetal bovine serum (FBS), and horse serum (HS) (10x dilution) samples (100 μl) at RT. Additionally, the encapsulated dye was also incubated in an equal amount of deionised water. 10 μl batches of the samples were mixed with an equal amount of loading dye (bromophenol-blue solution) and 10 μl of the mixtures were pipetted into the gel. To preserve the physiologic protein structure and potential nonspecific binding, samples were separated by nondenaturing polyacrylamide gel electrophoresis (PAGE) (100 V, 1.5 h, Tris/glycine buffer). 27 Gels were scanned using the IVIS Lumina II, then stained with Coomassie brilliant blue and imaged again using the same settings (Note: the fluorescence spectra of the two dyes overlap, and the methanol-based Coomassie-staining procedure removes IR-676 from the gel).

In Vitro Toxicity Screening
The biological safety of the formula was evaluated on the Chinese Hamster Ovary (CHO-K1) cell line. Cells were trypsinized, washed, and resuspended in PBS at a density of 2 × 10 5 ∕ml. 50 or 100 μl batches of the imaging formula or the vehicle alone were added to 1 ml amounts of cells; thereafter, the cells were incubated for 5 min at 37°C, pelleted by centrifugation at 1000 rpm for 5 min, after which they were resuspended in 1 ml PBS. Cell viability was evaluated by the fluorescein diacetate-propidium iodide (FDA/PI) assay, 28 as 500 μl batches of each sample were stained with either FDA (1 μl of an 500-mM stock solution in dry DMSO) or PI (5 μl of 0.75-mM stock solution) and incubated for 10 min. Viable cells fluoresce in green as they enyzmatically convert FDA to fluorescin, whereas dead cells fluoresce in red as PI stains the nuclei after passing through the damaged cell membranes. 2 × 10 4 cells∕cohort were evaluated by a CyFlow flow cytometer (Partec Ltd., Münster, Germany). Untreated and 6% formalin fixed cells were used to set the gates for living and dead cells, respectively. The percentages of viable and nonviable cells were calculated by identical gating for each sample.

Preparation of the Frozen Ear Sections for Microscopy
The samples were embedded in a frozen section compound (FSC 22, Surgipath Ltd., Richmond, Illinois), immediately frozen in liquid nitrogen, and stored at −80°C. The samples were cryosectioned (8 μm), and imaged within 24 h by an Olympus FluoView 1000 confocal laser scanning fluorescence microscope (Olympus, Center Valley, Pennsylvania). Single optical sections were taken from areas of interest using a 40× objective.

Oxazolone-Induced Contact Dermatitis Mouse Model
The ear model of oxazolone-induced allergic contact dermatitis was previously validated as a robust, reliable disease model, leading to localized edema and sensitization, as well as inflammatory cell activation primarily mediated by Th1 cell response. 29 Moreover, it has been recently demonstrated as a good proof-of-concept model to test potential NIR tracers of vascular leakage. 30 Anesthetized mice were sensitized on two consecutive days by applying 50 μl 2% oxazolone (Sigma-Aldrich) onto the shaved abdomen. Six days later, 15 − 15 μl of 2% oxazolone was applied on both the inner and outer surfaces of the right ear, whereas the left ear was treated with the vehicle only (96% ethanol) to serve as control. The oxazolone treatment induces a marked hyperemia and edema within hours, accompanied by leukocyte influx and T cell infiltration. Subjects were injected with the contrast agent 24 h later, and were repeatedly imaged between 5 to 180 min postinjection with the IVIS Lumina II. Thereafter, the mice were sacrificed, and ears were collected for cryosection histology.

Lipopolysaccharide-Induced Acute Airway Inflammation Mouse Model
The LPS model is a reliable tool to investigate acute inflammatory changes of the airways and lungs, and has been validated and optimized by our workgroup before. 31 In the present series of experiments, a modified, direct intratracheal delivery was used, since our preliminary experiments demonstrated that the conventionally used intranasal route produces far too high intra-group variations for in vivo imaging, Mice were anesthetized, and a PE-10 cannula was carefully introduced into the trachea via the mouth by a novel technique allowing simple visualization of the vocal cords. 32 The airway inflammation was induced by intratracheal instillation of 60 μl PBS in which 100 μg dry, purified Escherichia coli (serotype: 083) endotoxin was dissolved while controls received PBS. 21 h after LPS treatment, the mice were injected with the contrast agent, and they were imaged 3 h later by the FMT 2000 imaging system.

K/BxN Serum-Transfer Arthritis Mouse Model
The K/BxN serum-transfer model is a reliable immune-mediated model of rheumatoid arthritis, which is based on the transfer of autoantibody-rich sera from the spontaneously arthritic K/BxN mice into healthy recipients, thus eliciting a transient polyarthritis. Sera were obtained from the K/BxN mice, pooled, and stored at −80°C as described earlier. 33 To induce a systemic polyarthritis, mice were injected ip. with 300 μl serum diluted to 1 ml with sterile PBS. Subjects were injected 7 days after arthritis induction, and were imaged 3 h postinjection using the FMT 2000 system.

Carrageenan-Induced Acute Paw Inflammation Model of the Rat
This model was selected to evaluate the feasibility of in vivo imaging of larger species with this contrast agent. Carrageenan is a complex polysaccharide which elicits a massive edema formation, local hyperemia, and consequent leukocyte accumulation, thus serving as a reliable model of acute inflammation. 34 Wistar rats under anesthesia received an intraplantar injection of 100 μl 1% carrageenan (Sigma-Aldrich) in the left hindpaw, while the right paw was injected with an equal volume of saline. Two hours later, the animals were injected with the contrast agent, and were imaged 60 and 120 min postinjection with the IVIS Lumina II.

Statistical Analysis
Results are expressed as mean±standard error of the mean. Statistical evaluation was performed by Graphpad Prism®. The imaging results in the oxazolone model were evaluated by two-way ANOVA with Sidak's multiple comparisons test, while other imaging results were achieved by Student's twotailed t-test. At least p < 0.05 was considered significant.

The IR-676-Based Formula Demonstrates Negligible Binding to Serum Proteins
The nondenaturing-PAGE demonstrated that IR-676 binds weakly if at all to serum proteins regardless of being present as freely dissolved (ethanol) or encapsulated (Kolliphor HS 15). As strong fluorescence was exclusively observed in the wells where micellar IR-676-containing samples were pipetted, which did not propagate in the gel, we can conclude that the formula did not disintegrate even when exposed to strong electric current, and its size prevented it from entering into the gel [Figs. 1(b) and 1(c)].

The Dye Formula Has No Impact on Cell Viability in the Fluorescein Diacetate-Propidium Iodide Acute Toxicity Assay
Exposing CHO-K1-cells to two concentrations of the formula or the vehicle alone exerted no apparent acute toxic effects. The percentage of viable and dead cells showed minimal fluctuations between study groups ( Table 1). As the higher concentration employed (100 μl formula∕1 ml cell suspension) exceeded even the expected peak plasma concentration upon intravenous injection, we assume that the present formula poses minimal if any risks of toxicity within its suggested operational range. Although no formal in vivo toxicity tests were performed, we carefully observed IR-676-treated animals to assess any major untoward effect. No lethality occurred throughout the study and no observable pathological sign was ever found upon close inspection of the animals during and after anesthesia. Heart rate, as well as respiratory frequency and depth remained regular, and no changes in skin color and visible mucous membranes were seen. Dye-injected animals regained consciousness normally and afterward showed normal home cage activity (locomotion, eating/drinking, climbing, and grooming). Based on these observations and the in vitro cytotoxicity data, we can conclude that the potential in vivo acute toxicity of the dye is of minor significance.

IR-676-Doped Micelles Selectively Accumulate in Areas Affected by Chemically Induced Contact Dermatitis in Mouse Ear
Immediately 10 min postinjection, the contrast agent yielded a more than 260% greater signal in the inflamed ears than on the contralateral noninflamed side. Moreover, the signal strength showed minimal variations in the vehicle-treated ears, demonstrating the importance of body weight-based dosage. In the following 3 h, the signal strength slowly decreased in the control ears indicating the gradual clearance of the dye-loaded micelles from the circulation. At 3 h, the fluorescence intensity in the affected ears was more than 650% higher than in the controls, pointing out the prolonged presence of the extravasated dye molecules in diseased tissues [ Figs. 2(a) and (b)]. The confocal fluorescence microscopic examination supported these findings, showing a patchy accumulation of the fluorophore in the perivascular regions within oxazolone-treated ears [ Fig. 2(c)].   ]. IR-676 showed a strong accumulation in the liver, similarly to other cyanine dyes. This, however, could be easily demarcated from the specific signal (despite being at least a magnitude greater) in the 3-D tomographic scans. In contrast, it was not possible to separate the specific signal from the liver signal in the 2-D epifluorescence scans [ Fig. 3(d)] due to the scattering of the latter.

Visualization of Immune-Mediated Polyarthritis of the Mouse Ankle Joints
In the control mice, the fluorescence in the hindlimbs was negligible 3 h after the injection (less than 1 pmol or not quantifiable) and was not localized in the ankle joints. In comparison, the K/BxN serum-treated mice displayed a circumscribed,

Discussion
The primary outcome of our study is that we successfully compounded a novel IR-676 dye-containing NIR imaging formula for both in and ex vivo imaging. It can be rapidly and economically established and easily mass produced for routine experimental applications using only the standard equipment and common reagents of a life sciences laboratory, with the sole exception of the fluorophore. We have validated the adaptability of our contrast agent in a range of different inflammation models in distinct organs displaying different characteristics regarding the underlying pathophysiological changes, and it proved to be a highly sensitive and perfectly reproducible tool to detect vascular leakage in mouse ear, joint and lung, as well as the rat paw. The observed superior contrast and signal-to-noise ratio can be attributed to several different factors: (1) employing a hydrophobic dye ensures complete encapsulation of the fluorophore and makes the amount of free dye in the aqueous phase negligible. This results in easier preparation (since it is not neccessary to test the encapsulation-efficiency); moreover, it reduces the unspecific signal from healthy tissues, which is produced at least partially by the free dye.
(2) Rigorous observance of body weight-based dosage. According to our results, this alone can greatly enhance specificity mainly by reducing the variation of signal strength in controls (as observed, e.g., in the oxazolone-experiment). However, fluctuation within the diseased groups will nevertheless be present due to the intrinsic biological variations of the models. Fluorophores encapsulated into self-assembling, microparticles using biologically inert surfactants represent a simple, yet sensitive method to evaluate a wide variety of inflammatory diseases and to follow tumor progression in vivo. A similar concept has recently been described for two-photon imaging microscopy using a di-stryl benzene derivative. 35 Kolliphor HS 15 and other nontoxic nonionic surfactants (Pluronic, Cremophor etc.) have been validated for the encapsulation of ICG within self-assembling micelles and provide a good solution to countervent the rapid clearance of the dye. 25,36 Nevertheless, some intrinsic problems of ICG still remain: its low quantum yield (necessitating at least a magnitude higher doses), its tendency to autoaggregate, and its broad excitation and emission maxima, making specific filtering of its fluorescence difficult. Additionally, its good water solubility poses the risk of high unspecific signal due to the unpredictable kinetics and biodistribution of the free dye. IR-676, in contrast, has a quantum yield comparable to most other advanced fluorophores (lower amounts are needed for sensitive detection), 24 and has definitive peak excitation and emission maxima, which match the parameters of the current NIR lasers and filters, as these are almost identical to the respective parameters of the most widely used NIR dyes (Cy5.5, Alexa Fluor 680 etc.). Moreover, the poor water solubility of IR-676 minimizes the presence of freely dissolved dye in the solution. We observed that IR-676 displays a robust liver accumulation, therefore, imaging disease conditions within the liver or in its close proximity are not technically possible with the present formula. We have to emphasize, however, that this is a common feature of all the other cyanine fluorophores, as well. 4,19 Additionally, it is important to note that the emission maximum of this dye lies very close to that of chlorophyll (680 to 740 nm) which is abundantly present in most standard rodent food pellets. Therefore, for imaging in the abdominal region mice should be previously fed with alfalfa-free food to reduce abdominal autofluorescence. 37 Traditionally, good water solubility was deemed as one of the essential pharmacological characteristics of fluorophores employed in in vivo imaging. 11 While this is generally beneficial for fluorophores intended to be used for labeling, moderate hydrophobicity increases the efficiency of micellar encapsulation and reduces unspecific background fluorescence. However, it is also important to keep in mind that lipophilic substances can have an unspecific cytotoxic effect by influencing the plasma membrane fluidity. 38 However, our in vitro toxicity screening test suggests that the present formula is relatively inert within its operational range, as we observed no toxic effects on CHO-K1 cells. As IR-676 is (1) not entirely hydrophobic, (2) as it was coated by the surfactant, and (3) as the employed dosage was very low compared to other similar studies, its possible in vivo effect (if present) was marginal at best, and caused no observable behavioral or physiologic alterations in the animals similarly to other closely related cyanine dyes. 13 As a conclusion, the novel in vivo imaging formula presented here provides excellent sensitivity and high specificity in a wide range of disease models, both in superficial and deep-tissues, and it can be more easily and economically prepared using standard laboratory equipment without any complicated steps in the process as compared to the existing state-of-the-art NIR plasma extravasation contrast agents. Additionally, this formula can be seamlessly incorporated into most optical imaging instruments, even the fluorescent optical tomography system. IR-676doped micelles present an excellent opportunity for the noninvasive optical imaging of plasma extravasation, and also make in vivo imaging of these disease conditions in larger species realistically feasible (as demonstrated on rats) due to the easy and economical large-scale synthesis of the formula.