In Vivo Assessment of Thermosensitive Liposomes for the Treatment of Port Wine Stains by Antifibrinolytic Site-Specific Pharmaco-Laser Therapy

Antifibrinolytic site-specific pharmaco-laser therapy (SSPLT) is an experimental treatment modality for refractory port wine stains (PWS). Conceptually, antifibrinolytic drugs encapsulated in thermosensitive liposomes are delivered to thrombi that form in semi-photocoagulated PWS blood vessels after conventional laser treatment. Local release of antifibrinolytics is induced by mild hyperthermia, resulting in hyperthrombosis and complete occlusion of the target blood vessel (clinical endpoint). In this study, 20 thermosensitive liposomal formulations containing tranexamic acid (TA) were assayed for physicochemical properties, TA:lipid ratio, encapsulation efficiency, and endovesicular TA concentration. Two candidate formulations (DPPC:DSPE-PEG, DPPC:MPPC:DSPE-PEG) were selected based on optimal properties and analyzed for heat-induced TA release at body temperature (T), phase transition temperature (Tm), and at T > Tm. The effect of plasma on liposomal stability at 37 °C was determined, and the association of liposomes with platelets was examined by flow cytometry. The accumulation of PEGylated phosphocholine liposomes in laser-induced thrombi was investigated in a hamster dorsal skinfold model and intravital fluorescence microscopy. Both formulations did not release TA at 37 °C. Near-complete TA release was achieved at Tm within 2.0–2.5 min of heating, which was accelerated at T > Tm. Plasma exerted a stabilizing effect on both formulations. Liposomes showed mild association with platelets. Despite positive in vitro results, fluorescently labeled liposomes did not sufficiently accumulate in laser-induced thrombi in hamsters to warrant their use in antifibrinolytic SSPLT, which can be solved by coupling thrombus-targeting ligands to the liposomes.

For the NaCl in MilliQ group, MilliQ water was adjusted to pH = 7.4 with 0.01 M NaOH, yielding an osmolarity of 0.0061 osmol/kg as a result of the pH adjustment. A starting solution of 0.9% (w/v) NaCl (154.0 mM) was prepared with the pH-adjusted MilliQ water. The starting solution was diluted to 115.5 mM, 77 mM, 38.5 mM, and 19.25 mM NaCl with the pH-adjusted MilliQ water.
For the NaCl in 10 mM HEPES buffer group, HEPES was dissolved in MilliQ water at a 10-mM concentration and the solution was adjusted to pH = 7.4 with 10 M NaOH in MilliQ, accounting for an osmolarity of 0.0244 osmol/kg. Next, a 5% (w/v) NaCl starting solution (855.5 mM) was prepared in the pH-adjusted HEPES buffer. The pH-adjusted HEPES buffer was used to dilute the 5% NaCl starting solution to 4% NaCl (684.4 mM), 3% NaCl (513.3 mM), 2% NaCl (342.2 mM), and 1% NaCl (171.1 mM) solutions. All listed concentrations were measured, but the osmolarity is only reported for the 0-4% NaCl concentration range.
For the calcein in MilliQ group, a 50-mM calcein solution was prepared by adding 350 mg of calcein to 10 mL of MilliQ water containing 50 mM NaOH. The pH was raised further by drop-wise addition of 10 M NaOH in MilliQ during continuous stirring with a magnetic stirrer and occasional heating until the calcein was completely dissolved. Next, the solution was allowed to cool down and thereafter gradually acidified by dropwise addition of 3.7% HCl until pH = 7.4 was reached during continuous stirring and real-time pH measurement. The final solution was serially diluted to a concentration of 0.8 mM with MilliQ that had been adjusted to pH = 7.4 with 0.01 M NaOH (yielding 0.0059 osmol/kg). It should be noted that the 50-mM calcein solution is an approximate concentration based on the additional volume that was added to the solution for solubilization and pH adjustments. For these experiments it was more important to achieve a self-quenching calcein solution with known osmolarity that could be adjusted to iso-osmolar levels using e.g., NaCl in pH-neutral MilliQ water. The osmolarity of different concentrations calcein and NaCl dissolved in pH-adjusted MilliQ (pH = 7.4) was also measured ( Figure S1) for heat-induced release assays in aqueous buffer and in human plasma-containing samples. Accordingly, iso-osmolar liposomes containing 52.8 mM calcein, 44.5 mM (0.26% w/v) NaCl, pH = 7.4, 0.292 osmol·kg -1 were prepared. To determine the effect of an osmotic gradient on leakage from liposomes at different plasma concentrations, lipid films were hydrated with a hypo-osmolar solution comprising 52.8 mM calcein in MilliQ, pH = 7.4, 0.215 osmol·kg -1 .
In the electrophoretic mobility determinations, the Zetasizer sample chamber was thoroughly rinsed with a solution containing 0.1% Triton X-100 (20 mL), ethanol (20 mL), and MilliQ water (20 mL) before each measurement. Gelfiltered LUVETs were diluted 5 × with physiological buffer (RT) to a 1-mM final phospholipid concentration in a 2.5-mL sample volume. The sample and physiological buffer were analyzed at RT. The zeta potential of LUVETs was corrected for the measured zeta potential of physiological buffer. Settings used for differential scanning calorimetry: scan rate = 55 °C/h, filter period = 2 s, resting temperature = 15 °C, cell concentration = 3.0 mM, starting temperature check enabled, desiccation while thermostat, real-time baseline subtraction, and pre-scan waiting period = 300 s.   Figure S3. Key procedural steps in intravenously injecting solutions, affixing the optical chamber to the animal's dorsal skin. (A) The thoracic skin overlaying the jugular and subclavian vein was excised to enable the injection of solutions via the subclavian vein. The needle, bent to a ~60 degree angle, was inserted at a sharp angle from the flank and guided to the subclavian/jugular vein junction, where the solution was gently infused. The needle was retracted slowly along the thoracic surface such that the fat and fascia covering the subclavian vein would deter excessive bleeding. (B) The dorsal skin was shaved and, in this case, depilated to reveal the blood vessels as seen through the annular window of the optical chamber. The skin was loosely 'sandwiched' between two aluminum frames and secured with sutures looped through the small holes in the frames and through the skin. (C) The skin overlaying the blood vessel of interest was removed with microsurgical scissors. (D) The animal was secured to a microscope stage through pins that interlocked with the optical chamber. The open wound was kept moist with PBS (37 °C) to maintain quasi-physiological conditions for the target blood vessel (insert) and prevent muscle twitching due to desiccation. A robustly secured optical chamber, albeit ensuring that the interpositioned skin is not squeezed so as to not hamper blood flow, also reduces motion of the region of interest during intravital microscopic imaging, shown in (E). Some features in the images deviate from the actual protocol for purposes of illustration (anesthesia, syringes, microscope stage, microscope, and laser system). The protocol and images were adapted from [1]. Figure S4. (A) Spectrum of emitted light by the Leica EL6000 light source. (B) The optical properties of the filter set used for intravital fluorescence microscopy. The transmission range of the excitation (ex) light filter is shown in blue (spectrum, black line), whereas the transmission range of the emission (em) filter is shown in green (spectrum, red line). The wavelengths at which NBD-PC, 5(6)carboxyfluorescein, calcein, and FITC exhibit their absorption maximum and fluorescence emission maximum are demarcated to show that the filter set was appropriate to image all fluorophores. Al spectra were normalized to maximum intensity.  9

8
The components of the custum-built laser system are shown in Figure S5. The laser beam was reflected by 2 adjustable 100% reflective mirrors (Thorlabs) into a light guide that could be rotated along its longitudinal axis to radially position the laser beam. A high-speed shutter (model Lambda SC SmartShutter, Sutter Instrument, Novato, CA) was placed between the reflective mirrors to control the laser pulse width (via a model LB-SC controller and Sutter software). The terminal aperture of the laser guide contained a plano-convex lens (Thorlabs). A reflective mirror was fixed at an angle (135 degrees relative to the light path) into a linearly adjustable probe slid over the light guide to focus the laser onto the target blood vessel at approximately 45 degrees relative to the plane of the optical chamber. The focal point of the laser was situated at approximately 10 cm from the reflective mirror in the light guide tip, producing a slightly oval spot size of 2 × 10 -3 mm 2 .

S2.6. Image analysis
The isolated frames depicting the largest thrombus (group 1; section 2.9) or the most fluorescent thrombus (groups 2-6; section 2.9) were loaded into Photoshop and duplicated. The duplicated images were subjected to a green filter in the hue/saturation control panel as shown in Figure S6. The standard color range for the red, yellow, green, cyan, blue, and magenta hues spans 30° of the core hue + 30° into the neighboring hues on each end ( Figure S6G). The lightness of all hues except for the green was completely dimmed to a value of -100. In order to retain the full bandwidth of the green hue, the long periphery of the yellow was cut off at 75° ( Figure S6B) while the short periphery of the cyan hue was cut off at 165° ( Figure S6D). The 165° and 75° values correspond to R/G/B coordinates 0/255/191 and 191/255/0, respectively. These RGB coordinates in turn translate to 496 nm and 559 nm, respectively (https://academo.org/demos/wavelength-to-colour-relationship/). Accordingly, only pixels that match these hue values were retained in the filtered duplicates, and hence show whether any of the fluorophores had been incorporated into the laser-induced thrombus ( Figure S4). , and magenta (F). The hue range of the yellow and cyan was cut off at 75° and at 165°, respectively (bottom spectral sliders in B and D, respectively), so as to allow both 30° peripherals of the green hue (G). The legend in (G) shows the corresponding RGB coordinates of the green hue boundaries as well as the corresponding wavelength. Consequently, the filtered duplicate fames showed only colors in the 496-559 nm wavelength range -i.e., the fluorescence emission range of calcein, CF, NBD, and FITC ( Figure S4).

15
The low TA:lipid ratio in our formulations, compared to the Eeffs reported for other drugs in Table S3, stems from the fact that we deliberately chose to add 316 mmoles of TA to 5 mmoles of lipid, and is further accounted for by the liposome preparation method. The high-drug-low-lipid proportions were deliberate so that 1) the intraliposomal environment would be iso-osmolar relative to physiological conditions, and 2) only a small number of liposomes would have to be targeted to laser-induced thrombi to account for a pharmacologically relevant anti-fibinolytic milieu in laserilluminated PWS vessels. These rationales were explained and demonstrated in [3]. When one adds > 60 × more hydrophilic drug molecules than lipids to a solution where only a tiny fraction of the volume will be occupied by the aqueous compartment of the formed liposomes, most of the waterdissolved TA will reside in the extraliposomal fraction. This TA-containing fraction is cleared in the size exclusion chromatography step(s), whereby only intraliposomal TA is retained in the eluate, and replaced by TA-lacking physiological buffer that the Sephadex G50 matrix was equilibrated in. Consequently, the ultimate TA content is profoundly reduced by the Sephadex matrix exchange while the lipid content stays more or less the same. Figure S7. Differential scanning calorimetry thermograms of TA-encapsulating LUVETs of different phospholipid compositions (from bottom to top, formulations 3, 6, 14, and 19) in physiological buffer compared to empty DPPC and DSPC liposomes in water. The phase transition temperature (Tm) is indicated in the warm temperature shoulder of each thermogram. The orange region designates temperatures that the LUVETs will be exposed after intravenous administration (up to the normal body temperature; 36.5-37.5 °C). The red temperature zone represents the range corresponding to the onset of fever (> 37.5 °C or 38.3 °C).  LUVETs, prepared with a hypo-osmolar calcein solution relative to physiological buffer and plasma, had a mean ± SD diameter of 130.4 ± 1.6 nm and 162.8 ± 3.5 nm, respectively, and a mean ± SD polydispersity index of 0.270 ± 0.067 and 0.077 ± 0.022, respectively. The passive leakage rates and kinetics at 37 °C are plotted in Figure S9. DPPC:DSPE-PEG LUVETs exhibited comparable calcein leakage rates irrespective of the PPP concentration, which were significantly lower than for DPPC:MPPC:DSPE-PEG LUVETs at 20%, 40%, 60%, and 80% PPP. When DPPC:MPPC:DSPE-PEG LUVETs were added to equilibration buffer, 100% calcein leakage occurred within 1 min. The addition of PPP stabilized passive calcein leakage to some extent. The high relative fluorescence of the LUVETs following addition to PPP-containing solution suggests a biphasic process of initially rapid leakage (few seconds) followed by slower leakage. It should be noted that the plotted leakage rates do not reflect the rapid leakage phase since the leakage rates were measured between roughly 3-5 min. Given the fact that DPPC:MPPC:DSPE-PEG LUVETs containing an iso-osmolar calcein concentration had similar physicochemical properties but lower fluorescence following addition to the PPP-containing solutions, it is proposed that the initial rapid leakage of calcein from hypo-osmolar DPPC:MPPC:DSPE-PEG LUVETs was due to the osmolarity gradient.

18
Finally, calcein release DPPC:DSPE-PEG (96:4) and DPPC:MPPC:DSPE-PEG (86:10:4) LUVETs containing a hypo-osmolar, self-quenching calcein concentration (49.6 mM calcein in MilliQ, pH = 7.4, 0.210 osmol·kg -1 ) was assessed during a 20 °C → 47.0 °C temperature ramp in a repurposed Roche LightCycler 480 II system. These experiments were performed to prove the importance of an osmotic balance between the intra-and extraliposomal space. As shown in Figure Figure S11. Representative flow cytograms of resting and convulxin-activated hamster platelets incubated with (A) 34 µM 5(6)-carboxyfluorescein (CF) and (B) 50 µM calcein (green traces). In (C) and (D), flow cytograms are shown of resting and convulxin-activated human platelets that had been incubated with the same concentration CF and calcein, respectively (blue traces). The purple (A, B) and red traces (C, D) represent fluorescence histograms of platelets that had not been incubated with fluorophores (control). The fold increase in fluorescence relative to non-labeled platelets in provided in (E) for CF and (F) for calcein (Hu, human; Ha, hamster; ∅, resting platelets; ⊕, activated platelets) (N = 2 per group). The extent of platelet staining with activation-dependent anti-CD62P-FITC antibodies and their respective anti-IgG1K-FITC isotype control antibodies is present in (G) for hamster platelets and in (H) for human platelets (N = 2 per group). Figure S12. Thrombus kinetics in the negative control group (no fluorescent thrombus labeling) following multiple laser pulses (Table S2) administered to the encircled region in the top left panel.

S3.5. LUVETs do not incorporate into laser-induced thrombi in vivo
The arrows indicate the direction of blood flow in the venule (V) and arteriole (A). After illumination, a thrombus began to form and expand in time, indicated as minutes:seconds relative to the last laser pulse in the bottom right corner. The thrombus was visible as a white patch (arrowheads), resulting from the absence of light absorbers in the clot compared to blood, which comprises erythrocytes that contain the strongly light-absorbing chromophore hemoglobin. Consequently, the blood vessels always appear as dark columns regardless of the type of microscope illumination (epi-and transillumination). The panels in the bottom row (15:06 -41:06) were duplicated and edited in Photoshop to reveal the laser-induced thrombus. The vessel segment was layered via cut after manual contouring and the background in the source image was blackened. The layered vessel segment was augmented in brightness (150%) and contrast (200%) to show the thrombus, which was visible due to reflection of incident light from the EL6000 light source and heating lamp.    Half of the blood smear in (A) was enhanced by augmenting the intensity by 375% to demonstrate the lack of NBD fluorescence in blood. The other half of the smear was contoured. Half of the urine smear in (K) was enhanced by augmenting the intensity by 375% to demonstrate NBD fluorescence in urine. Organs were harvested approximately 135 min after antibody infusion and imaged using an exposure time of 1.5 s and a gain of 5.0 ×.