Non-invasive transdermal delivery of chemotherapeutic molecules in vivo using superparamagnetic iron oxide nanoparticles

The skin is both a target and a potential conduit for the delivery of drugs, but its cornified cell layer resists penetration by most molecules. This study investigated the potential of superparamagnetic iron oxide nanoparticles to facilitate the transdermal delivery of anticancer agents. Chemotherapeutic cancer drugs were applied with or without nanoparticles to the skin of hairless mice, and their ability to penetrate the skin was assessed using fluorescence microscopy and tumor growth. Nanoparticles enhanced the penetration of the skin by doxorubicin and 5-fluorouracil as determined by fluorescence microscopy and growth retardation of experimental melanoma in immunocompetent, syngeneic mice. This drug enhancement did not require conjugation or encapsulation of the drugs by the nanoparticles—simple co-administration sufficed. Nanoparticles applied topically to melanomas increased the cytotoxicity and immune cell infiltration induced by co-administered 5-fluorouracil, and also reduced vascularization of the tumors independently of 5-fluorouracil. Correctly formulated superparamagnetic iron oxide nanoparticles can facilitate the chemotherapeutic effectiveness of cytotoxic drugs on skin tumors by both increasing their transdermal penetration and ameliorating host–tumor interactions. This enhancement of skin penetration occurs without the need for conjugation or encapsulation of the co-administered drugs, and so will likely be applicable to other drugs, also.

in the liver (Prausnitz and Langer 2008;Alkilani et al. 2015;Palmer and DeLouise 2016). In the case of skin diseases, transdermal delivery can enable local concentrations of drugs to achieve therapeutic levels while keeping the systemic levels very low, thereby reducing off-target toxicity. However, delivery of therapeutic molecules through the skin poses particular challenges. In particular, the stratum corneum (SC) is impermeable to almost all compounds larger than 500 Da (Matsui and Amagai 2015).
Nanoparticles have the potential to significantly improve the penetration of large molecules across the SC, thereby improving bioavailability without compromising the skin barrier to pathogenic microbes. Nanoparticles can improve penetration of drugs by acting either as carriers, in which the hydrophobicity or stability characteristics of a pre-formed nanoparticle-drug complex is improved relative to the drug alone, or as co-administered adjuvants, which change the behavior of the tissues to enhance their penetrability. They can offer stability, biodegradability, low toxicity, and biocompatibility (Dianzani et al. 2014;Wang et al. 2016;Pelaz et al. 2017;Carter et al. 2019). However, their size can be an impediment to transdermal applications. Generally, larger nanoparticles (20-30 nm) can penetrate only through hair follicles (Zvyagin et al. 2008), limiting their usefulness for transdermal applications. In contrast, nanoparticles smaller than 10 nm can permeate through both the SC and hair follicles (Hansen and Lehr 2012). However, aggregation of small nanoparticles in vivo can limit their tissue penetration and increase toxicity (Pelaz et al. 2017). The surface charge of nanoparticles also affects their ability to penetrate the skin.
Melanoma is the most aggressive common skin cancer with a 5-year survival rate of 10% among metastatic melanoma patients (Garbe et al. 2011;Mundra et al. 2015). In recent years, immunotherapies and chemotherapies targeted to specific mutant proteins have revolutionized the therapy of melanoma, but lack of response and relapse through resistance are common and so conventional cytotoxic chemotherapies still play a major role in its management (Tang et al. 2017;Vanella et al. 2019). Several chemotherapeutic drugs are available for melanoma treatment, including dacarbazine, temozolomide, 5-fluorouracil (5-FU), paclitaxel and platinum compounds. However, the overall response rate is less than 20% and each of these agents produces unwanted systemic side effects. Applications of nanoparticles to the treatment of melanoma is of growing interest (Tang et al. 2017), despite the aforementioned drawbacks of size, charge and aggregation inherent in many nanoparticles. Carbon nanotubes loaded with doxorubicin (DOX) induced melanoma cell death in vitro and abrogated tumor growth in xenograft mouse models (Chaudhuri et al. 2010). Inorganic nanoparticles based on silica and aluminum (Oh et al. 2012;Yu et al. 2013), protein-based nanoparticles (Vannucci et al. 2012) and DOX-loaded liposomes (Lohade et al. 2016) have also been investigated for melanoma treatment.
Superparamagnetic iron oxide nanoparticles (SPIONs) have been successfully used in various biomedical applications that benefit from their magnetic properties, biocompatibility, low toxicity, easy separation methodology and flexibility of surface modification (Shubayev et al. 2009;Veiseh et al. 2010;Kievit et al. 2011;Hasany et al. 2012;Ali et al. 2016;Palanisamy and Wang 2019). However, to date, no SPIONs have been commercialized successfully for drug delivery purposes. We have developed a unique coating technology for SPIONs that produces a highly stable dispersion for biomedical applications (Jain et al. 2010;Eamegdool et al. 2014;Pham et al. 2015Pham et al. , 2017. These SPIONs are nontoxic in vitro and in vivo and can be readily tailored to have different coating compositions and functional groups at the end of stabilizers (Bryce et al. 2013;Eamegdool et al. 2014;Pham et al. 2018). We have shown that co-administration of our nanoparticles with a cytotoxic drug can enhance the penetration of the drug in an in vitro spheroid model (Bryce et al. 2013).
In this study, we investigated the potential of our sterically stabilized SPIONs as chemotherapy adjuvants in transdermal drug delivery in vivo. We found that topical co-administration of these SPIONs improved transdermal delivery and the anti-tumor responses of mouse melanomas to chemotherapeutic drugs.

Nanoparticles
Maghemite nanoparticles (γ-Fe 2 O 3 , average 25 nm core diameter, unless stated otherwise) were synthesized by the coprecipitation method (Massart et al. 1995) and sterically stabilized with a mixture of 90% poly(ethylene glycol) methyl ether (MPEG)-end and 10% NH 2 -end macro-RAFT copolymers, prepared using RAFT polymerization (Pham et al. 2018). The polymers each contain a block of 10 units of MAEP to anchor MPEG or polyacrylamide to the SPIONs' surface, and either a block of 40 units of acrylamide and 17 units of poly(ethylene glycol) (PEG) (in MPEG-end-copolymer) or a block of 60 units of acrylamide (in NH 2 -end copolymer) for steric stabilization. We refer to these copolymer-coated nanoparticles as "10% NH 2 -SPIONs". Transmission electron microscopy images were obtained using a JEOL 1400 microscope. Hydrodynamic diameter was measured by dynamic light scattering (DLS) using Malvern's Zetasizer nano-ZS. The polymer content was 13% (w/w) of the dried solid as analyzed by thermogravimetric analysis (Additional file 1: Figure S1d). Iron concentration of the SPIONs was determined by flame atomic absorption spectroscopy (Varian AA800 spectrometer). For in vivo studies, 10% NH 2 -SPIONs were dispersed in PBS to the final concentration of 5 mg/mL as Fe and then filtered through a 0.22 µm sterile syringe filter.
Rho-SPIONs were made by labeling 10% NH 2 -SPIONs with rhodamine B isothiocyanate in PBS at room temperature in the dark for 4 h. Free rhodamine was removed using Amicon Ultra MWCO 3000 centrifugation filter units. Fluorescence intensity of Rho-SPIONs was measured with a RF-5301 PC spectrofluorophotometer (Shimadzu, Kyoto, Japan).

Cell culture
Amelanotic tyrosinase-knockout B16-F10 melanoma cells were kindly provided by Dr Shweta Tikoo from the Centenary Institute, Sydney, Australia, and maintained in Advanced Dulbecco's modified Eagle's medium and 5% fetal bovine serum (Thermo Fisher Scientific, Waltham, USA). They were stably transfected with CEFLP-tdTomato-H2B eukaryotic expression plasmids using GenePORTER ® 3000 (Genlantis, San Diego, CA, USA). Transfected cells were selected with puromycin (10 µg/mL) and tdTomato-expressing cells were purified by flow cytometry on a BD Influx Cell Sorter.

Animals
A colony of hairless albino C57BL/6 mice was established from a breeding pair obtained from the Jackson Laboratory (strain 017840) and maintained by breeding homozygous males (Hr −/−) with heterozygous (Hr ±) females. Mice were housed with dust-free bedding (Able Scientific, Australia), regular chow, water, environmental enrichment and 12/12 h light/dark cycles.

Transdermal delivery of SPIONs and chemotherapeutic drugs
Mice were anesthetized by intraperitoneal injection of ketamine (75 mg/kg) and medetomidine (1 mg/kg), placed on a warming pad and the back skin of the mice was cleaned with a 70% isopropyl alcohol swab (Livingstone International, Australia). Rho-SPIONs (20 µl of 2 mg/mL in PBS) were applied to the skin and left for 1 h, after which excess liquid was wiped off with a wet cotton bud. PBS only (20 µL) was used for the control mouse. For chemotherapeutic drug delivery, a 30 µL mixture in PBS containing either both 10% NH 2 -SPIONs (1.3 mg/mL as Fe) and DOX (0.17 mM), 10% NH 2 -SPIONs only (1.3 mg/mL as Fe), or DOX only (0.17 mM), was applied to the mouse skin and then left for 1-2 h, at which time unabsorbed material was removed by wiping. Anesthesia was reversed with atipamezole (0.2 mg/kg).
The mice were imaged under anesthesia on a Carestream In Vivo FX PRO (Carestream Health, Woodbridge, USA) using 550 nm excitation and 600 nm emission wavelength filters before application and 1, 2, 24 and 48 h after application. They were then euthanized, and treated skins were fixed in 4% paraformaldehyde (PFA) overnight at 4 °C for fluorescent imaging.

In vivo melanoma model
Eight-week-old hairless albino C57BL/6 mice were injected subcutaneously with 1 × 10 4 B16F10-tdTomato amelanotic melanoma cells. Three days post-injection, the mice were randomized into different treatment groups. Each group was topically treated at the tumor site 3 times/week with 60 µL of either vehicle (0.9% saline), 5-FU (1 mM) alone, 10% NH 2 -SPIONs (1 mg/mL) alone, or a mixture of 5-FU (1 mM) and 10% NH 2 -SPIONs (1 mg/mL). Tumor growth was measured using calipers and micrographs taken with a Dino-Lite AM4515ZT Edge Handheld Microscope. Tumor volume was calculated by the modified ellipsoidal formula (Tomayko and Reynolds 1989). Mice were euthanized when tumors reached 1 cm 3 , or skin ulceration appeared to be imminent.

Image analysis
Cell morphology of H&E stained sections were characterized by the pathologist (LFK) to assess tumor borders and areas of tumor tissue necrosis. The pathologist was blinded to the treatments the specimens had received. Three mice per group were used for analysis and 4-6 regions were selected randomly in each tumor. Necrosis and whole tumor areas were measured using ImageJ software, and the percentage of the necrotic region was calculated by the total tumor area. Tumor-infiltrating lymphocytes were identified by a small cytoplasm:nucleus ratio, an oval, dark stained nucleus and a cell diameter of 7-10 µm and were counted manually using ImageJ.
The vascular area was quantified from CD31-stained images in which the diaminobenzidine signal had been segmented by spectral deconvolution (Ruifrok and Johnston 2001) and thresholded using Fiji image analysis software. The image representing CD31 staining was then binarized and holes (vascular lumens) were filled. The CD31-positive stained area and total tumor area were measured in each tumor section, from which the percentage of CD31-stained area was calculated.
Statistical analyses were done using GraphPad Prism v. 8.4.

SPIONs penetrate mouse skin
SPIONs were synthesized from maghemite particles by coating them with polyacrylamide and PEG copolymers, 10% of which were NH 2 -end derivatized for steric stabilization (Additional file 1: Figure S1a). The particles had a core diameter of 20-30 nm, a hydrodynamic diameter of 72 nm, a polymer content of 13% w/w, a polydispersity index of 0.11 and a zeta potential of − 40 mV (Additional file 1: Figure S1b-e). To evaluate the penetration of SPIONs into mouse skin, fluorescent rhodamine-labeled SPIONs (Rho-SPIONs) (Additional file 1: Figure S1f ) were topically applied to hairless albino C57BL/6 mice. Live-mouse images (Fig. 1a, Additional file 1: Figures S2, S3) show a strong rhodamine fluorescent signal on the mouse's skin after one hour of application. After 24 h, the signal intensity of the Rho-SPIONs was lower and dispersed over a larger area. No fluorescent signal above autofluorescence was detected on the control mouse. Fluorescence microscopy of fixed skin from the application areas of Rho-SPION-treated mice show that SPION particles penetrated through the SC within 24 h of application, accumulating most strongly in the epidermis, but also penetrating and being retained throughout the dermis (Fig. 1b, Additional file 1: Figure S3).

SPIONs improve the transdermal penetration of doxorubicin
To test whether topical application of SPIONs enhances the penetration of chemotherapy drugs into the skin, we topically applied the chemotherapeutic DNA-intercalating drug, doxorubicin (DOX, molecular mass 544), with and without SPIONs to the skin of the dorsal back and ear. The fluorescent properties of DOX enabled microscopic and macroscopic visualization of drug penetration into the mouse skin. Live-mouse imaging 2 h after topical application showed strong DOX fluorescence in the application area for DOX applied with or without SPIONs, which decreased after 24 and 48 h (Fig. 2a). Macroscopically, the DOX signal was higher in the area treated with DOX alone compared with the area treated with both DOX and SPIONs at 2, 24 and 48 h after application. Confocal microscopy of cryosections of fixed tissue was used to determine the extent of penetration of DOX into the skin (Fig. 2b, Additional file 1: Figure S4). Mice treated with DOX alone showed poor penetration of DOX into the skin whereas mice treated with both DOX and SPIONs showed a strong DOX signal within the epidermis that extended into the dermis. It should be noted that DOX remaining on the surface of the epidermis would not have intercalated into cell DNA and therefore may have been lost during the fixation and counterstaining process; this would explain why there is no DOX signal detected in the DOX + PBS cryosections despite a strong signal being observed at the application site in the whole animal images.

Co-application with SPIONs improves the response of melanoma tumors to 5-FU
5-FU is a hydrophilic, negatively charged molecule that cannot easily penetrate through the hydrophobic, negatively charged SC (Fang et al. 2004). To determine whether SPIONs could improve the response to transdermal treatment by 5-FU, we Fig. 1 Rhodamine-labeled SPIONs penetrate mouse skin. To examine the penetration of topically applied SPIONs into the skin, rhodamine-labeled SPIONs (2 mg/mL in 20 µL PBS, mouse on the right) or vehicle (20 µL, mouse on the left) were applied to the back skin of mice. One hour after application, unabsorbed solution was removed by wiping. a Mouse whole-body fluorescence images were acquired at the time points indicated. b Fluorescence micrographs of frozen fixed sections of the Rho-SPION-treated area of skin, 24 h after application, showing brightfield, nuclear counterstain (DAPI) and rhodamine signals. Representative images of 1-2 mice and experiment repeated 3 times topically treated subcutaneous B16-F10 melanoma tumors in the skin of mice with 5-FU and/or SPIONs, starting three days post-tumor cell inoculation, then 3 times per week until mice were euthanized. Comparing the growth curves up to day 17, the latest point at which all mice survived, 5-FU on its own appeared to slightly accelerate tumor growth compared with saline (Fig. 3a), but this effect did not reach statistical significance (p = 0.52 by repeated measures ANOVA). Topical application of 5-FU with SPIONs significantly reduced tumor growth compared with topical application of 5-FU alone (p = 0.0072), but SPIONs alone did not (p = 0.96). Furthermore, topical co-application of 5-FU with SPIONs significantly improved the survival of mice (i.e., the time for the tumor to reach 1 cm diameter, the ethical endpoint; Fig. 3b) when compared with 5-FU alone (p = 0.012, log-rank test).
To determine whether the reduced growth of melanoma tumors in the SPION/5-FU co-administration group was due to an enhanced cytotoxicity, we examined tumors histologically for necrosis. Photomicrographs showed more pronounced necrosis in tumors treated topically with 5-FU + SPIONs than in those treated with 5-FU alone, SPIONs alone or saline vehicle control (Fig. 4a). Quantification by image analysis confirmed that this pro-necrotic effect was statistically significant (p = 0.040 and p = 0.026, respectively, by one-way ANOVA with Sidak's multiple comparisons correction) (Fig. 4b). Moreover, combined 5-FU + SPION topical treatment increased the infiltration of tumors by both lymphocytes (Fig. 5a) and macrophages (Fig. 5b), and reduced their vascularization (Fig. 5c) compared with tumors treated topically with 5-FU alone or saline control. Fig. 3 Topical co-application of 5-FU and SPIONs reduces mouse melanoma tumor growth. Melanoma tumors in 8 week old hairless mice were initiated by subcutaneous injection of 1 × 10 4 B16F10-tdTomato mouse melanoma cells on the back of the mouse. Three days post cell injection, mice were treated with topical application at the tumor site of 60 µL of either saline alone, SPIONs alone, 5-FU (1 mM) alone or 5-FU plus SPIONs (1 mg/mL) three times a week (n = 5-8 mice per group). a Tumor size. **p < 0.01, repeated measures ANOVA. b Survival, in terms of the time taken for the tumor to reach 10 mm diameter. *p < 0.05, log-rank (Mantel-Cox) test Interestingly, SPIONs alone were able to increase lymphocyte and macrophage infiltration and reduce vascularization (Fig. 5a-c). Thus, similar to the DOX studies, co-administration of SPIONs facilitated the transdermal effects of 5-FU.

Discussion
There has been a limited number of reports investigating the permeability of inorganic nanoparticles through the skin, with or without conjugated drugs (Baroli et al. 2007;Nose et al. 2012;Anandhakumar and Raichur 2013;Fernandes et al. 2015;Santini et al. 2015;Hsiao et al. 2016;Ozcan et al. 2020). Most previous transdermal nanoparticle studies focused on gold nanoparticles, with studies investigating Fe 3 O 4 magnetic nanoparticles for topical application rare (Baroli et al. 2007;Rao et al. 2015;Santini et al. 2015). In this study, the excellent penetration through the skin by our SPIONs, despite their relatively large size, might be explained by their steric stabilizers, which prevent the particles from aggregating and accumulating on the skin surface. In addition, NH 2 end groups on the coating layer are protonated at low pH, which might facilitate their penetration of the SC.
DOX has been used on its own or in combination with other agents against many cancers, including melanoma, lung, breast and ovarian cancer (Weiss 1992;Pan et al. 2019). However, there are two main concerns of DOX: cardiotoxicity and poor Fig. 4 Co-treatment of 5-FU with SPIONs increases necrosis and tumor-infiltrating lymphocytes in melanoma tumors. a Representative images of hematoxylin and eosin-stained sections of B16F10-tdTomato melanoma tumors that had been treated topically with either vehicle (saline), 5-FU alone, SPIONs alone or both 5-FU and SPIONs. b The extent of necrosis in the tumors. c The number of infiltrating lymphocytes per 1.3 mm 2 field in the tumors. The data are shown as the mean ± standard deviation of n = 3 mice per group. One-way ANOVA with Sidak's multiple comparisons adjustment indicating significant difference at *p < 0.05 and **p < 0.01 penetration into a solid tumor. To overcome these limitations, the first nano-based, PEGylated liposomal DOX was approved for the treatment of ovarian cancer and Kaposi's sarcoma (Barenholz 2012). Since then, several nanoparticle formulations conjugated or loaded with DOX have been developed for use in cancers Chen et al. 2015;Lv et al. 2017;Sheth et al. 2020). However, nanoparticles conjugated or loaded with chemotherapy drugs may be susceptible to inefficient release of drugs at the target site, which is avoided by our adjuvant approach (Bryce et al. 2013). We previously showed that co-administration of 10% NH 2 -SPIONs with DOX increased drug penetration into cancer cell spheroids compared with DOX alone (Bryce et al. 2013). The results in this study confirm that co-administration of SPIONs Co-administration of SPIONs with 5-FU increases macrophage infiltration and inhibits blood vessel formation in tumors. a Representative images of B16F10-tdTomato melanoma tumors treated topically with either vehicle (Saline), 5-FU alone, SPIONs alone or both 5-FU and SPIONs. Sections were stained for macrophages with anti-F4/80 antibody (brown). b Quantification of tumor-infiltrating macrophages from F4/80 immunostained sections. c Representative images of melanoma tumors treated topically with vehicle (Saline) 5-FU, SPIONs or both 5-FU and SPIONs. Sections were stained for endothelial cells with anti-CD31 antibody. The data are shown as the mean ± SEM of n = 5 measurements from 2 mice per group. d Quantification of vascularization from CD31-immunostained sections. One-way ANOVA followed by Sidak's multiple comparisons adjustment indicating significant difference at *p < 0.05, **p < 0.01 and ***p < 0.001 additionally enhances the penetration and bioactivity of DOX and 5-FU through the SC and epidermal layers into the dermal layer of the skin of a living animal, without the need for covalently binding DOX to the nanoparticles. The range of molecules whose transdermal penetration is enhanced by these SPIONs without covalent attachment remains to be determined, but conceivably could include other anticancer and immunomodulatory drugs.
For many years, the infiltration of tumors by lymphocytes has been used as a measure of treatment success in clinical trials for melanoma treatment (Rosenberg et al. 1994) and numerous reports indicate that tumor-infiltrating lymphocytes play an essential role in mediating response to chemotherapy in many cancers (Galluzzi et al. 2015). In this study, we showed that SPIONs applied topically to a skin melanoma can enhance the cytotoxicity of 5-FU, increasing necrosis and slowing tumor growth. The increased necrosis may be a combination of enhancing the penetration of 5-FU, increasing leukocyte infiltration and reducing the vascularity of the tumors. The effects of SPIONs on tumor vascularization and immune cell infiltration are independent of 5-FU. A recent study on magnetically localized polyethyleneimine-coated SPIONs also found an enhancement of macrophage infiltration and reduced vascularization of tumors in the absence of cytotoxic drugs (Mulens-Arias et al. 2019), but did not examine the effect of SPIONs on drug delivery.
Chemotherapy of cutaneous melanomas is usually restricted to patients with metastatic disease, in part because of the toxicity of systemically delivered drugs. Further investigation needs to be undertaken to determine the suitability of topical SPION/ drug co-administration for treating premalignant lesions without surgery, as a neoadjuvant treatment for the surgical removal of primary melanomas, which sometimes invade beyond excision margins, and of cutaneous metastases, which can be too numerous to be removed surgically. In particular, the range of molecules whose penetration can be enhanced by the co-administration of specifically stabilized SPIONs needs to be determined.

Conclusion
Appropriately designed SPIONs show promise for improving the treatment of skinlocalized tumors through a 3-pronged effect: enhancing the immune response by increasing leukocyte infiltration, reducing the nutrient supply by decreasing its vascularity, and enhancing the penetration of chemotherapeutic drugs such as DOX and 5-FU through the skin. The application of such drugs transdermally offers a non-invasive treatment that ensures that the drugs reach their targets while reducing systemic toxicity. We anticipate that these SPIONs could improve the performance of other drugs in current use and enable drugs that had been previously ruled out due to systemic toxicity issues to be revisited.