Transgenic SHH MB mice (MAP mice) were generated as described previously (39). PNA is produced by the reaction of 4-(2-bromoacetamido)-2,2,6,6-tetramethyl-1-piperidinyloxy (BrAcTPO) (Sigma-Aldrich, St. Louis, MO) and human serum albumin, and manufactured as a 10% solution. PNA (Lot 032597, AntiRadical Therapeutics LLC, Sioux Falls, SD) or 25% USP grade human serum albumin (Baxter, Deerfield, IL) vehicle was injected intraperitoneally three times a week (n = 6 for PNA treatment; n = 3 for vehicle control), starting at six weeks of age. The final dosage in all animals was 10 ml kg-1. This dosing was based on that used previously in breast cancer mouse studies (17) and in other systems. The PNA dose administered was near maximum due to injection volume considerations and the size of mice.

Animals were sacrificed and perfused at end-point with cold phosphate buffered saline (PBS) followed by cold 4% paraformaldehyde (PFA). Whole brain and spinal column were dissected and fixed in 4% PFA overnight at 4°C and processed for paraffin embedding and sectioning. Tissue sections were deparaffinized with CitriSolv (Decon Labs, King of Prussia, PA) and rehydrated through graded ethanol solutions. For frozen tissues, PFA fixed samples were further equilibrated in 20% sucrose for 24-48 hours at 4°C, embedded in TissueTek-Optimal Cutting Temperature (O.C.T.) compound (Sakura Finetek, Torrance, CA), and stored at -80°C. Frozen tissue blocks were sectioned at 15-20 µm thickness on a cryostat.

Immunostaining was carried out as described previously (39). Heat-induced epitope retrieval was performed for paraffin-embedded tissue sections using Rodent Decloaker (RD913 L, Biocare Medical, Concord, CA). Endogenous peroxidase activities were quenched with 3% H₂O₂ for 10 minutes. Tissue sections were treated with 10% normal serum in PBS-0.1% Triton X-100 for 1 hour and incubated with primary antibodies for 1 hour at room temperature. After three washes of 5 minutes each with PBS, biotinylated secondary antibodies were applied for 1 hour at room temperature, after which avidin enzyme complex and substrate/chromogen were used for color development (Vector Laboratories, Burlingame, CA). Stained tissue sections were counterstained with hematoxylin and eosin (H&E). If suboptimal staining occurred due to suboptimal fixation or the staining was uninterpretable, the slides were not included in the analysis and sample size was reduced.

For cell proliferation analysis, the number of Ki-67⁺ cells in 300 tumor cells was assessed from five distinct fields of each sample. For cell differentiation analysis, the number of NeuN⁺ cells in 300 tumor cells was assessed from five distinct fields of sample. The percentage of Ki-67⁺ or NeuN⁺ cells in total tumor cell population was calculated by averaging the number of positively-stained cells per 100 tumor cells of all samples for each treatment.

To quantify spinal tumors, high magnification images were taken of cross sections at different spinal cord levels. Boxes of 600 µm in perimeter were placed in the tumor region and the number of tumor cells within each box was counted. The area encompassed by tumor mass was calculated to determine total number and density of tumor cells on each section using StereoInvestigator software (MBF Bioscience, Williston, VT). Total tumor cell number and density was determined by averaging results from four representative sections at each spinal cord level for each treated animal. If no tumor cells were found at a certain level, the density was scored as zero to demonstrate the effects of PNA.

Light microscopic images were obtained using a Nikon SMZ1000 Stereomicroscope or a Nikon Eclipse 90i microscope system (Nikon Instruments, Melville, NY).

The technician performing cell scoring and quantitation was blinded to the treatment condition. Other investigators were not blinded during experiments and outcome assessment. Statistical analyses were performed with GraphPad Prism 6.0 (GraphPad Software Inc., La Jolla, CA). All pooled data were expressed as the mean ± standard error of the mean (SEM). Differences between groups were compared using an unpaired two-tailed t test. Results were considered significant at *P < 0.05; **P <0.01; ***P <0.001; ****P <0.0001. The significance of intergroup differences in the incidence of primary tumor, leptomeningeal disease, and metastasis was assessed using Fisher’s exact test.

Animals were housed in the AAALAC-accredited Laboratory Animal Research Facility at Sanford Research in accordance with NIH guidelines. All animal experimental procedures were approved by Sanford Research Institutional Animal Care and Use Committee (IACUC) and performed following national regulatory standards.

SHH MB mice used in this study were generated as previously reported and are referred to as MAP mice (39). MAP mice were treated with PNA or albumin vehicle starting at six weeks of age, a time point at which primary tumors in cerebellum are present and small clumps of tumor cells are present in CSF, indicative of leptomeningeal disease (39). Animals received 3 treatments per week until endpoint (end of life). At endpoint, all vehicle control mice showed leptomeningeal dissemination and metastasis to both the forebrain and to spinal cord. In PNA treated mice only 1 of 6 mice exhibited leptomeningeal dissemination and metastasis to the forebrain and 4 of 6 exhibited metastasis to the spinal cord ( Table 1 , P<0.05). Thus, PNA prevented metastasis to the forebrain in ~83% of animals and prevented metastasis to the spinal cord in ~33% of animals.

Kaplan-Meier survival analysis ( Figure 1 ) showed MAP mice treated with albumin vehicle had a mean survival of 71 days (n = 3) while PNA treated MAP mice had a mean survival of 95 days (n = 6, P <0.05, Figure 1 ). Thus, PNA increased lifespan by ~28%.

Cell proliferation and differentiation in primary and secondary tumors were evaluated using quantitative Ki-67⁺ and NeuN⁺ immunohistochemistry ( Figure 2 ) in animals at end point. Ki-67⁺ and NeuN⁺ are used as markers for mitotic cells and postmitotic cells, respectively (39). Ki-67⁺ expression was significantly reduced by PNA in primary brain tumors ( Figure 2B , **P<0.01) but not in spinal cord tumors ( Figure 2C ). NeuN⁺ expression as a measure of differentiation increased in primary tumors but was not significantly changed in secondary tumors ( Figures 2B, C ). Thus, PNA treatment significantly affected cell proliferation and differentiation only in primary tumors.

The effects of PNA on tumor burden and cell densities in spinal cord was evaluated in animals at endpoint using H&E staining ( Figure 3A ). Tumor burden (mean total number of cells) for the entire spinal cord was significantly reduced in mice treated with PNA compared to albumin vehicle ( Figure 3B , P<0.05). The density of cells (number of cells/µm2) in tumors at four levels of the spinal cord was quantified ( Figure 3C ). MAP mice treated with PNA were found to have significantly reduced metastatic cell density in the thoracic, lumbar and sacral regions (P<0.05), while cell density in the cervical region did not show significant change.

Numerous studies have shown that PNA is a vascular therapeutic in animal models of stroke, traumatic brain injury, sickle cell disease, myocardial infarction, and preventing lung capillary leak (27, 30–35, 37). The sustained therapeutic effects observed in all indications were thought to be derived from responses within the vasculature. In agreement, PNA’s demonstrated vascular effects to date have included reversal of hypoxia, blood flow enhancement, modulation of NO signaling, and alleviation of oxidative and nitrosative stress. It is unclear whether the effects of PNA on metastasis in MB requires crossing the BBB and entry into the CSF. The BBB is intact in SSH MB while in WNT MB it is leaky (40, 41). As a result, WNT MB is considered more treatable compared to SHH MB (40, 41). However, it is possible that highly local geographic variation in BBB leakiness could exist in SHH MB as suggested in glioma (43), despite overall integrity of the barrier in this MB subtype. Although data on the geographic variability of CNS drug penetration is quite sparse, we cannot rule out loss of BBB integrity in the specific vicinity of tumors in our model while the bulk of the BBB remains intact. On the other hand, given recent work by Taylor and colleagues on hematogenous dissemination of MB (9), it is also possible that PNA exerts its anti-tumor effects in MB without crossing the BBB.

Determining PNA’s concentration in CSF in treated animals would help in addressing whether PNA is acting within the CSF, yet is made difficult by the small volume of CSF in mice as well as the risk of contamination from blood. The distribution of PNA is thought to be similar to the well-studied distribution of serum albumin (44, 45). The BBB and the blood-spinal cord barrier allow only very low levels of albumin in CSF. In contrast, BBB disruption or leakiness associated with pathological conditions leads to increased albumin levels in CSF (2, 44). Entry of albumin into the CNS normally is an active process involving receptor-mediated internalization into endothelial cells and transcytosis dependent on the albumin-binding protein SPARC (45). SPARC has been found to be over-expressed in glioma and provides a mechanism for albumin-based drug delivery to the CNS (46, 47). The expression of SPARC and its potential role in PNA transport into the CSF in SHH MB will need investigation.

Assuming an intact BBB and lack of enhancement in SPARC-mediated transport, the concentration of albumin in CSF is normally ~3 µM (48), a >200 fold lower concentration than in plasma (~640 µM). In the elderly or in chronic pathological conditions where the BBB is thought to be leaky, the CSF albumin concentration increases only ~2 fold (48, 49). Although CSF albumin levels in brain cancers such as MB and glioblastoma have not been directly measured, mass spectrometry data supports slightly increased albumin levels in glioblastoma (50).

In previous in vitro studies on DAOY medulloblastoma (21), U-87 glioblastoma cells (21) and in 4T1 breast cancer cells (17), the effective PNA concentration for inhibiting tumor cell proliferation was 30-120 µM. The amount of PNA administered in this study predicts a PNA concentration of ~49 µM in plasma (an ~8% increase in total serum albumin). If PNA distributes similarly to endogenous albumin and the BBB isn’t appreciably leaky, then PNA within the CSF would be ~0.25 µM in our experiments, 100-480 fold lower than the concentrations effective in suppressing tumor cell proliferation (17, 21). Yet, PNA exerted significant effects on MB at primary and metastatic sites in our model. These observations argue that PNA is acting from within the vasculature. PNA’s concentration in CSF would have to rise 100-480 fold to be effective. Although significant increases in CSF albumin levels have been seen to occur transiently in trauma situations (51), BBB disruption to the extent needed to reach effective PNA levels in CSF would have immediate life-threatening consequences and couldn’t be sustained for long. Therefore, our results suggest that tumor suppression from PNA results from PNA’s effects within the vasculature.

The anticancer effects of PNA mainly result from the covalently attached caged nitric oxide groups. While serum albumin on its own has anti-oxidative effects (52), total albumin in plasma (PNA + endogenous albumin) in our experiments increased only by ~8% after administration of PNA or control serum albumin (see above). Low molecular weight cyclic nitroxides are known to be superoxide dismutase (SOD)-mimetics, act as antioxidants by oxidizing cuprous and ferrous ions, and also react directly with a variety of free radicals to detoxify them e.g. lipid peroxidation (16). Numerous studies have shown the efficacy of low molecular nitroxides in animal models (13–15, 53). However, as they easily cross membranes, they are toxic and subject to rapid reduction intracellularly to the inactive hydroxylamine form. Covalent attachment of CNO groups to a macromolecule, as in PNA, addresses these issues by localizing activity extracellularly, prolonging pharmacological effects and preventing toxicity.

Imbalances in antioxidant defenses are one of the hallmarks of cancer (13, 15). There are 3 superoxide dismutases, two of which are intracellular (SOD1 and SOD3) and one is extracellular (SOD3/EcSOD) (15). Low molecular weight cyclic nitroxides on their own can act as SOD mimetics of all three enzymes (13, 15), but when coupled to PNA no longer cross membranes and can only act as an SOD3 mimetic in the vasculature. Without PNA, nitric oxide (NO) reaction with elevated superoxide in the vasculature leads to depletion of NO levels and production of peroxynitrite. Peroxynitrite is where the oxidative and nitrosative stress pathways merge, and peroxynitrite is the major free radical in the pathogenesis of multiple diseases. Lowering of vascular superoxide with PNA would reduce peoxynitrite production and oxidative/nitrosative stress.

Hypoxia is well known in solid tumors and has been shown in glioblastoma to limit the effectiveness of chemotherapeutics (54). The CSF in MB is a hypoxic environment (55) while a derangement in ROS homeostasis promotes metastasis of SHH MB through stabilization of hypoxia-inducible factor 1α (HIF1α) (56). It has been shown using live animal imaging that PNA can increase blood flow within minutes in peripheral solid tumors (17, 18). Thus, in addition to reducing oxidative/nitrosative stress, the lowering of superoxide and elevation of NO levels with PNA would also increase blood flow to tumors, resulting in decreased hypoxia within tumors.

Anti-hypoxic effects are also seen with low molecular weight cyclic nitroxides that can cross membranes and act intracellularly. TEMPOL is one such example and has been shown to inhibit both hypoxia-inducible factor 1α (HIF-1α) and hypoxia-inducible factor 2 α(HIF-2α) in glioblastoma (54). Compared with PNA, these nitroxides do not show the vascular and therapeutic effects observed with PNA in diverse animal disease models. Nevertheless, comparison of the effects of these two classes of drugs in inhibiting dissemination and metastasis is warranted.

If hematogenous dissemination occurs in our transgenic model, the main candidate locations where PNA may be acting are:

The TME of brain tumors is comprised of diverse cell types including malignant cells, astrocytes, macrophages/microglia, neurons, pericytes and endothelial cells (57–59). PNA may affect any of these cells as well as the dynamically changing extracellular matrix. The glycocolyx on the luminal site of endothelial cells may also be important owing to the central role of albumin in its properties (60). For peripheral solid tumors, it has been suggested that the vasculature can also adopt a hybrid structure that incorporates primary tumor cells (61), which if operative in MB would make tumor cells directly accessible to PNA from the luminal side of the vasculature.

The CCL2 pathway receptor is known to play an important role in the transit of leukocyte populations ito the CNS (62, 63). Recently, the CCL2 pathway has also been implicated to be involved in the reentry of CTCs into CSF in MB (9). In this respect, PNA might affect the CCL2 pathway in the entry of CTCs into the CSF.

Finally, all MB subtypes are characterized as having an immunosuppressive tumor environment and are considered not amenable to immunotherapy (42). As elevated NO levels after PNA treatment may help to create a favorable tumor environment for endogenous immune cell functions, this mechanism would not require the involvement of hematogenous dissemination of CTCs.