Fractional laser-assisted topical delivery of bleomycin quantified by LC-MS and visualized by MALDI mass spectrometry imaging

Abstract Bleomycin exhibits antiproliferative effects desirable for use in dermato-oncology but topical use is limited by its 1415 Da molar mass. Ablative fractional laser (AFL)-assisted drug delivery has been shown to enhance drug uptake in skin. The aim of this study was with AFL to deliver bleomycin into skin, quantify uptake, and visualize biodistribution with mass spectrometry. In a Franz diffusion cell study, pig skin samples (n = 66) were treated with AFL (λ = 10,600 nm), 5% density, and 0, 5, 20, or 80 mJ/microbeam (mb) pulse energies before exposure to bleomycin for 0.5, 4, or 24 h. Bleomycin was quantified in biopsy cryosections at depths of 100, 500, and 1500 µm using high-performance liquid chromatography-mass spectrometry (LC-MS), and drug biodistribution was visualized for 80 mJ/mb samples by matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI). The pulse energies 5, 20, and 80 mJ/mb resulted in microscopic ablation zones (MAZs) reaching superficial, mid, and deep dermis respectively. Bleomycin was successfully delivered into the skin and deeper MAZs and longer exposure time resulted in higher skin concentrations. After 24 h, AFL exposure resulted in significant amounts of bleomycin throughout all skin layers (≥510 µg/cm3, p ≤ .002). In comparison, concentrations in intact skin exposed to bleomycin remained below limit of quantification. MALDI-MSI supported the quantitative LC-MS results by visualizing bleomycin biodistribution and revealing high uptake around MAZs with delivery into surrounding skin tissue. In conclusion, topical drug delivery of the large and hydrophilic molecule bleomycin is feasible, promising, and should be explored in an in vivo setting.

Bleomycin is a cytotoxic glycopeptide used in the treatment of a variety of malignant diseases. The drug's mechanism of action includes intercalation in DNA and chelation with transition metals to induce reactive oxygen species (Saitta et al., 2008;Povirk et al., 1979). With chemical properties including molar masses of 1415-1426 Da and a predicted logP of À7.5 (Kim et al., 2016), bleomycin cannot cross the skin barrier unassisted to any significant degree.
Bleomycin is approved for systemic use in the treatment of squamous cell carcinoma, lymphoma, and testicular cancer (Saitta et al., 2008), and has been used off-label with intralesional administration for a wide variety of malignant and nonmalignant conditions in dermatology, e.g. basal-cell carcinoma (Glass et al., 1997) and keloids (Berman et al., 2017). AFL-assisted drug delivery of bleomycin may be an alternative to intravenous and intralesional administration, to enable a homogenous biodistribution in skin, and minimize systemic effects.
We aim to explore topical delivery of the large and hydrophilic drug bleomycin after AFL delivery, quantify skin uptake with high-performance liquid chromatography-mass spectrometry (LC-MS), and visualize drug biodistribution in superficial (100 mm), mid (500 mm), and deep (1500 mm) dermis using matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI).

Study design
Delivery of bleomycin by AFL treatment, reaching three different laser channel depths, was examined in an in vitro Franz diffusion cell model. Pig skin samples were exposed to bleomycin for 0.5, 4, or 24 h. A total of eleven interventions were tested in six repetitions (n ¼ 66) including non-AFL bleomycin interventions and saline controls (Table 1). LC-MS was used to quantitate bleomycin in cryosections from superficial, mid, and deep dermis (100, 500, and 1500 mm depths) along with receiver and donor fluids. MALDI-MSI imaging visualized cutaneous bleomycin biodistribution (100, 500, and 1500 mm depths) in skin samples treated with the deepest laser channels.

Skin preparation
Full-thickness skin samples from the flanks of two pigs (female, 46 and 44 kg) were utilized in the study. Skin was excised immediately after euthanasia, trimmed for hair, and relieved of subcutaneous fat. The skin was cut into square pieces of approximately 3 Â 3 cm, stored for up to a month at À80 C and thawed to room temperature at study initiation.

Franz diffusion cell model
Immediately after AFL treatment, full-thickness skin samples were mounted on a Franz cell model (PermeGear Inc., Hellertown, PA, USA) consisting of a donor and receiver compartment. Skin sample stratum corneum (0.64 cm 2 ) faced donor compartments, while receiver compartments contained 5.5-5.8 ml phosphate buffered saline (pH 7.4, 37 C) and a magnetic stir bar. At baseline, 750 mL of bleomycin sulfate (Baxter, Deerfield, Chicago, USA) (15,000 IU/mL = 7.06 mg/mL) was topically applied to skin samples via donor compartments.
At 0.5, 4, or 24 h diffusion time, skin samples were dismounted, padded dry, and biopsied using an 8-mm punch. A biopsy from each sample was then mounted on Tissue-Tek (Sakura, Alphen, Netherlands) for LC-MS quantitation and laser channel morphology microscopy. The biopsies for MALDI-MSI were mounted on Milli Q water (ELGA CENTRA-R200, ELGA LabWater, Buckinghamshire, United Kingdom). Biopsies were sectioned at skin depths of 100-, 500-and 1500 lm to produce samples of 30 lm thickness for LC-MS and 10 lm for MALDI-MSI. Samples were stored at À80 C prior to analysis. For laser channel morphology, cryosections were stained with hematoxylin-eosin and photo documented with ProgRes Capture Pro v. 2.8.8 (Jenoptik, Jena, Germany). Laser channel and coagulation-zone dimensions were digitally measured with ImageJ software v. 1.52c (National Institute of Health, Maryland, USA).

LC-MS
Before LC-MS quantitation, donor and receiver fluids were collected directly from the Franz cells. For skin samples, bleomycin was initially extracted from cryosections for 2 h in 1 mL PBS, rotating vials every 30 min. All samples (250 lL) were then precipitated with 500 mL 2% ZnSO 4 · 7H 2 O in 25% methanol, centrifuged at 20,000 rcf for 5 min at 4 C and the supernatant was analyzed by LC-MS. Standards underwent equivalent analysis.
Quantitative analysis of bleomycin in skin sections, donor, and receiver compartments was performed using an Agilent LC system (Agilent Technologies, 1100 series, Santa Clara, CA, USA) with a binary pump (Agilent 1100 G1312A, Santa Clara, CA, USA) with a MS Single Quadrupole detector, multimode interface. The LC was equipped with a binary solvent delivery system and an autosampler. The injection volume was 10 lL. Separation was achieved at 40 C using a 10-cm Kinetex XB-C18 column (Phenomenex, Vaerlse, Denmark) with 1.7 mm particle size and 2.1 mm internal diameter. Bleomycin was analyzed under electrospray ionization in positive ion mode with detection in selection ion monitoring mode (SIM) at m/z 707.9 and m/z 713.3. The mobile phases were (A) 22% acetonitrile in milli Q water þ 0.05% heptafluorobutyric acid and (B) acetonitrile. The gradient started at 0% B at 4 min, increased to 90% B at 4-6 min, and returned to 0% B from 6-7 min, sustained until 16 min. The flow rate was 0.25 ml/min, the spray voltage 4000 V, and the drying gas flow rate 9 L/min. The limit of quantification (LOQ) was 75 ng/ml.

Units and calculations
Skin bleomycin concentrations (mg/cm 3 and IU/cm 3 ) were calculated based on cryosection dimensions of a halved biopsy with a surface area of 0.25 cm 2 and thickness of 0.003 cm. Receiver concentrations (mg/cm 2 ) were calculated based on the exposed skin area of the Franz cell setup (0.64 cm 2 ). Percentage of applied concentration present in a cryosection was based on the donor chamber concentration of bleomycin (7.06 mg/mL). In the result section, all concentrations mentioned are median values of 6 repetitions and pvalues represent the difference between the intervention and control (non-AFL bleomycin treated skin for 24 h).

Bleomycin imaging in skin
Qualitative determination of the distribution of bleomycin was performed on a Thermo QExactive Orbitrap mass spectrometer equipped with an AP-SMALDI-10 ion source (TransMIT, Giessen, Germany). MALDI imaging was performed on 12 samples and 4 control tissue sections. Prior to analysis, 10 mm sections of frozen tissue, both horizontal at 100, 500 and 1500 mm depth and selected vertical sections, were thawed in a vacuum desiccator and spray-coated with a 30 mg/mL solution of 2,5-dihydroxybenzoic acid in 50:50 methanol/water with 1% trifluoroacetic acid (Hsieh et al., 2007). 300 mL of matrix solution was pneumatically sprayed at a pressure of 2.5 bar, a flow rate of 30 mL/min and a distance between the sample and spray tip of 110 mm to form a homogenous layer of matrix crystals covering an area of approximately 18 mm in diameter. While applying the matrix solution, the sample was rotated at 600 rpm. The mass spectrometer was operated in positive ion mode (scan range m/z 400-1600) at mass resolving power of 14,0000@m/z 200. A mass accuracy of 1 ppm was ensured by using a matrix peak as lock-mass. The samples were imaged with pixel sizes of 40 mm for horizontal tissue sections and 35 mm for vertical tissue sections. The control tissue sections were analyzed using a pixel size from 100-120 mm, depending on the size of the section. Raw data was converted to imzML (Imaging mass spectrometry markup language) using an imzML converter (Schramm et al., 2012). Images were generated in MSiReader 0.06 using a bin width of 0.02 Da (Robichaud et al., 2013). The image analysis was performed using accurate masses of the two forms of bleomycin in positive ion mode (m/z 1415.52609 and 1425.56323) and the tissue was localized on basis of the localization of the endogenous phospholipid PC(34:1) (m/z 798.54092). 2,5-dihydroxybenzoic acid and trifluoroacetic acid were purchased from Sigma-Aldrich. Methanol was from ChemSolute/TH Geyer. Milli Q water was produced on a Millipore Direct-Q3 (Merck, Darmstadt, Germany).

Statistical analyses
Descriptive statistics are presented as medians with interquartile ranges. Testing for statistical significance (5% level), unpaired samples were compared using the non-parametric Kruskal-Wallis and Mann-Whitney U tests. If concentrations were below LOQ (75 ng/ml) the LOQ value was used in the calculations. All p-values are exact values and two-tailed. Statistical tests were computed in SPSS 24 (IBM Corporation, Armonk, NY, USA). Graphical presentation of data was plotted using GraphPad Prism 7 (GraphPad Software Inc., San Diego, CA, USA).

LC-MS quantified bleomycin concentrations
Overall, drug uptake depended on MAZ depth and exposure time, resulting in higher bleomycin concentrations using deeper MAZs and longer exposure times ( Figure 2 and Table 2).
Bleomycin concentrations in the transdermal receiver chambers were below LOQ or negligible after 0.5 and 4 h of drug exposure with all MAZs. After 24 h, bleomycin reached quantifiable concentrations (Table 2). For all interventions, donor compartments contained more than 75% of the originally applied bleomycin concentration at examined time points (data not shown). Normal saline controls were negative for bleomycin by LC-MS analysis.

Enhancement ratios
In Figure 3, enhancement ratios depending on MAZ depth and exposure time are shown as a heatmap of investigated skin layers. As illustrated, skin pretreatment with AFL enabled topical delivery of bleomycin with deeper MAZs resulting in enhanced delivery. Longer drug exposure times further enhanced concentrations. Four hours drug exposure time using MAZ-Deep thus offered the same enhancement ratio in superficial skin as 24 h exposure with MAZ-Supf.

MALDI mass spectrometry imaging
Qualitative imaging revealed bleomycin concentrating primarily in CZs, though also present in surrounding tissue (Figure 4(A)). Supporting quantitative LC-MS data, a trend of higher concentrations with increased drug exposure time was seen. The highest concentrations of bleomycin were noted in the superficial dermis, with decreasing intensity down through the skin layers.
Drug imaging of the superficial dermis (Figure 4(Ba-Bc)) indicated increasing concentrations of bleomycin over time.  Despite poorly defined cryosection outlines, imaging also revealed visible laser channels in analyzed samples. Mid dermis imaging (Figure 4(Bd-Bf)) showed well-defined skin sample outlines with an organized grid of MAZs. Peak bleomycin concentrations were detected corresponding to CZs and a trend for higher concentrations is seen with increasing drug exposure time.
Deep dermis imaging (Figure 4(Bg-Bi)) showed no visible bleomycin in the center of the 0.5-hour image but the drug is present peripherally (Figure 4(Bg)). This corresponds to the quantitative analysis by LC-MS being below LOQ. The 4 and 24 h images (Figure 4(Bh,Bi)) show well-defined outlines and grid-organized laser channels with a trend of higher bleomycin concentrations with increasing drug exposure time.
No bleomycin was detected in negative control samples with normal saline (data not shown).

Discussion
The present study is the first assessment of AFL-assisted topical delivery of bleomycin quantified with LC-MS and visualized with MALDI-MSI. Bleomycin was successfully delivered into the deep dermis by AFL while drug detection in non-AFL exposed skin was negligible even at superficial depths (Table 2 and Figure  2(C)). Bleomycin concentrations depended on MAZ depth, and higher concentrations of drug in tissue were achieved using deeper MAZs and longer drug exposure times. As expected, a non-linear relationship between drug uptake and MAZ depth as well as exposure time was observed, indicating a saturation level of bleomycin in the skin similar to data published for other drugs previously delivered by AFL in vitro, e.g. 5-FU (Wenande et al., 2017) and methotrexate (Taudorf et al., 2015).
The quantitative LC-MS results are supported by the qualitative MALDI-MSI. Furthermore, the qualitative imaging data provided insight in biodistribution and confirms high uptake of bleomycin in CZs as well as substantial delivery in the surrounding skin tissue. It has been hypothesized that AFLinduced CZs function as a barrier against drug uptake . However, in the present study, we utilized MALDI imaging to demonstrate substantial biodistribution beyond CZs, similar to what has previously been shown with smaller drugs than bleomycin, e.g. 5-FU (Wenande et al., 2017). Thus, our data is in line with previous research which shows high uptake in CZs with other topically-applied compounds (Haak et al., 2017).
AFL-assisted delivery of bleomycin appears to be a promising candidate for a new topical treatment with uniform and reproducible biodistribution. The AFL modality is associated with less interphysician variance than might be expected with current local treatment methods, e.g. injection delivery. AFL-assisted drug delivery may also enable precision delivery to specific skin depths and might also limit transdermal access to the vascular system and systemic effects; an advantage of local therapy. Although cutaneous biodistribution is highly dependent on drug concentrations and characteristics, a recent in vivo study investigating AFL-assisted cisplatin (300 Da) and 5-FU (130 Da) delivery did not find any quantifiable transdermal uptake , which indicates that systemic access might be limited depending on the drug and treated area.
AFL-assisted drug delivery in a clinical setting can reduce the need for long drug exposure times. After just 30 min, deep laser channels yielded almost the same drug uptake enhancement ratio in superficial skin as superficial channels after 4 h drug exposure. Thus, increasing MAZ-depth may dramatically reduce drug exposure time.
Compared with reported IC50 test results, our data suggests that AFL-assisted drug delivery of bleomycin can provide clinically relevant concentrations with all tested MAZs. Two studies listed in the Genomics in Drug Sensitivity Cancer (GDSC) project database differ greatly in their IC50 test results for bleomycin on a wide range of cell lines. For the often-used SCC-25 cell line, reported values were 0.97 mg/cm 3 (0.69 mM) and 219 mg/cm 3 (155 mM), respectively (Yang et al., 2012).
Another study on the same cell line suggests 15.0 mg/cm 3 (10.7 mM) after 24 h (Olesen et al., 2017a). Importantly, all three values can easily be reached using AFL under in vitro conditions. We used the standard vehicle for bleomycin for our Franz diffusion cell study, an aqueous saline solution. Topical application in the clinic may benefit from a more viscous formulation in order to adhere securely on the skin for the duration of the treatment, e.g. a gel-based vehicle. In a recent study, we compared different types of vehicles for use with laser-assisted delivery and found that a gel-based vehicle enters the skin almost as effectively as a liquid solution (Olesen et al., 2017b). Future in vivo studies might investigate bleomycin's efficacy and compatibility with different vehicles compared with normal saline. Other factors including drug loading in skin (Karadzovska et al., 2012), closing of laser channels over time (Banzhaf et al., 2017), and vascular wash-out effects also represent areas of further investigation, though best studied in vivo (Oni et al., 2012;Ibrahim et al., 2018).
A limitation of this in vitro study represents the almost infinite supply of bleomycin dissolved in normal saline available to the skin during exposure. This does not necessarily translate to clinical settings, but may emulate continuous application under occlusion during a similar time-frame in the clinic. Furthermore, Figures 4(Bg-Bi) suggest that punch biopsy can cause depression of peripheral tissue from the skin surface downwards, resulting in cryosections from deep skin compartments displaying tissue from more superficial layers in their outer rims. Higher drug concentrations would in consequence be visualized in the periphery of the cryosections, a phenomenon similarly observable in earlier studies (Wenande et al., 2017). Finally, a few samples of non-AFL-exposed skin were shown to contain bleomycin. However, this finding most likely reflects inadvertent contamination during handling and these low concentrations near the LOQ had a limited impact on our findings.
In conclusion, we show for the first time that AFL-assisted topical drug delivery of the large hydrophilic molecule bleomycin is feasible. Cutaneous bleomycin concentrations can be increased by using deeper MAZs and longer exposure times. Imaging further reflected the quantitative LC-MS data by visualizing bleomycin distributed in CZs and surrounding tissue, as well as higher drug concentrations associated with increased exposure time. AFL-assisted bleomycin delivery appears to be a promising candidate for future in vivo studies.