Enoxacin and bis-enoxacin stimulate 4T1 murine breast cancer cells to release extracellular vesicles that inhibit osteoclastogenesis

Enoxacin and its bone-seeking bisphosphonate derivative, bis-enoxacin, have recently captured attention as potential therapeutic agents for the treatment of cancer and bone disease. No differences in growth or survival of 4T1 murine breast cancer cells were detected at a concentration of 50 µM of enoxacin or bis-enoxacin. Growth was perturbed at higher concentrations. Both 50 µM enoxacin and bis-enoxacin stimulated increases in the number of GW/Processing bodies, but there were minimal changes in microRNA levels. Extracellular vesicles (EVs) released from 4T1 cells treated with 50 µM enoxacin or 50 µM bis-enoxacin stimulated proliferation of RAW 264.7 cells, and both significantly inhibited osteoclastogenesis in calcitriol-stimulated mouse marrow. EVs from 4T1 cells treated with enoxacin and bis-enoxacin displayed small reductions in the amount of microRNA (miR)-146a-5p and let-7b-5p. In marked contrast, miR-214-3p, which has been shown to regulate bone remodeling, was increased 22-fold and 30-fold respectively. We conclude that enoxacin and bis-enoxacin trigger the release of EVs from 4T1 cancer cells that inhibit osteoclastogenesis.

Enoxacin is a fluoroquinolone antibiotic first introduced in the 1980s 1 . Removed from the market in the United States by its manufacturer, it is still used in many nations for the treatment of gastroenteritis, respiratory infections, gonorrhea and urinary tract infections 2 . Enoxacin emerged from two independent screens for bioactive agents as a possible therapeutic agent for cancer and bone disease. Shan et al. identified enoxacin as a stimulator of RNA interference and microRNA activity 3 . This result was soon confirmed by a second group 4 . General reduction in microRNAs in cancer cells may allow the expression of oncogenic proteins 5 . The ability of enoxacin to stimulate microRNA activity prompted studies testing it as an anti-cancer agent. Enoxacin inhibited cancer cell growth in vitro 6,7 , and in vivo 6 . A more recent study expanded on these results, demonstrating that enoxacin inhibited both proliferating cancer cells and cancer stem cells in Ewing's Sarcoma, and that enoxacin acted synergistically with doxorubicin, a traditional anti-cancer agent, in the treatment of Ewing's sarcoma in an animal model 8 . Tests of enoxacin, which crosses the blood-brain barrier, for the treatment for amyotrophic lateral sclerosis (ALS) have also provided promising results 9 .
Separately, Ostrov et al. identified enoxacin in a "virtual screen" for small molecules predicted to block binding between the B-subunit of vacuolar H + -ATPase (V-ATPase) and microfilaments 10 . This protein-protein interaction is important for bone resorption by osteoclasts 11 . Tests of enoxacin in biochemical assays and cell culture confirmed its ability to block binding between V-ATPase and microfilaments and to inhibit osteoclast formation and bone resorption 10,12,13 . Enoxacin was used to block titanium particle-induced osteolysis 14 , but its short half-life in circulation, the fact that it is an antibiotic, and its side effects, made it impractical to consider enoxacin as an agent to treat osteoporosis. To overcome these problems, a bisphosphonate ester of enoxacin (bis-enoxacin) that directed enoxacin to bone was synthesized 15,16 . Like enoxacin, bis-enoxacin blocked V-ATPase binding to microfilaments and bone resorption in vitro and orthodontic tooth movement and periodontal bone loss when delivered systemically in rats 15,17 . Surprisingly, it also reduced systemic oxidative stress induced by periodontal infections 18 . Very recent studies showed that bis-enoxacin protects bone strength in rats after overiectomy more effectively than zoledronate 19,20 . In addition to blocking bone mineral loss, it also altered the glycoprotein composition of bone, making it more resistant to fractures 19 . Taken together, these data suggest that enoxacin and bis-enoxacin are a new type of therapeutic agent that may have distinct advantages over current therapeutics for the treatment of bone disease (bis-enoxacin) and cancer (enoxacin). Understanding the mechanisms that underlie the therapeutic effects is vital for the eventual use of these agents in the clinic.
Although enoxacin and bis-enoxacin completely blocked osteoclast formation in vitro at a concentration of 50 µM, over 100 µM enoxacin was required to inhibit cancer cell growth by 50% and 100 µM was used to demonstrate stimulation of microRNAs 6,7 . We hypothesized that these agents might have effects on cancer cells at lower concentrations that had not been detected, which might help account for their anti-cancer effects in vivo. In this report, we sought effects of low dose (50 µM) enoxacin and bis-enoxacin on 4T1 murine breast cancer cells, which have often been used to study the ability of breast cancer cells to invade bone [21][22][23] . Our data show that enoxacin and bis-enoxacin inhibit proliferation of 4T1 cells in a manner similar to various other cancer cells that have been tested in vitro 6,7 . Extracellular vesicles (EVs), which include exosomes and microvesicles, have emerged as important intercellular regulators in cancer invasion and in regulating bone remodeling 24 . We find that both enoxacin and bis-enoxacin alter the regulatory activity and composition of EVs released by 4T1 breast cancer cells at a lower concentration than that required to inhibit their proliferation.

Enoxacin and Bis-enoxacin inhibit growth of 4T1 cancer cells only at high concentrations.
Enoxacin inhibits the growth of various types of cancer cells in vitro, but requires concentrations of 100 µM or greater 6,7 . Enoxacin had not been tested on 4T1 cells. Enoxacin selectively inhibited osteoclast formation and bone resorption at much lower concentrations without inducing apoptosis 10,12 . Complete inhibition of osteoclast differentiation was achieved by 50 µM enoxacin or 50 µM bis-enoxacin and the IC 50 for each was about 10 µM 12,15 . Suspecting that enoxacin and bis-enoxacin may have undetected effects on cancer cells at lower concentrations, we first examined concentration dependence. Consistent with reports on enoxacin's effects on other types of cancer cells [6][7][8] , no changes in growth parameters were detected in the presence of 50 µM of either enoxacin or bis-enoxacin. At a concentration of 100 µM, 4T1 growth was inhibited by about 40% after 4 days (Fig. 1A). This level of inhibition was very similar to previous studies of the effects on enoxacin on other types of cancer cells [6][7][8] .
To further explore the effects of enoxacin, we used the MTT assay. As in cell counts, we detected no difference in proliferation at concentrations of 50 µM enoxacin and bis-enoxacin (Fig. 1B). We tested for apoptosis using a caspase-3 assay; there was no increase in caspase-3 activity at enoxacin and bis-enoxacin concentrations under 150 µM and apoptosis levels were modest even at high concentrations (Fig. 1C) Enoxacin and bis-enoxacin at 50 µM stimulated formation of GW/Processing (P) bodies but little increase in cellular levels of selected microRNAs was detected. As a first test of whether low concentrations of enoxacin and bis-enoxacin affect cancer cells, we examined GW/P bodies, which are considered surrogate markers for microRNA-mediated repression of translation 25,26 . Enoxacin and bis-enoxacin, at 50 µM, stimulated significant increases in GW/P bodies ( Fig. 2A-D).
To test whether microRNA levels increased, qPCR was performed to examine the relative levels of a panel of microRNAs that were selected based on published data. MiR-146a-5p, let-7b-5p and miR-214-3p were upregulated during osteoclast formation and were very abundant in osteoclasts. MiR-290 and miR-689 were abundant in precursors and downregulated as osteoclasts form [27][28][29] . For this reason, we considered them to likely be important regulators in osteoclasts. MiR-146a-5p and miR-214-3p have also been implicated in regulating bone remodeling [30][31][32][33][34][35] . Because enoxacin has been reported to generally stimulate microRNA levels and GW/P bodies are thought to be surrogate markers for microRNA increases, it was surprising that of the microRNAs examined, only miR-214-3p was significantly stimulated by enoxacin. Bis-enoxacin stimulated small, but significant increases, in cellular levels of miR-214-3p, miR-689 and miR-290 (Fig. 2E).

Enoxacin and bis-enoxacin induced the release of EVs by 4T1 cells that inhibit calcitriol-stimulated osteoclast formation.
In addition to being sites of microRNA-mediated translation repression, GW/P bodies have been proposed as sites where microRNAs are packaged into EVs 36 . EVs from 4T1 cells treated with vehicle, 50 µM enoxacin or 50 µM bis-enoxacin were examined by nanoparticle tracking. EVs from 4T1 cells treated with enoxacin and bis-enoxacin were significantly smaller (Fig. 3A). Positive/negative stain transmission electron microscopy confirmed that nanoparticles were EVs (Fig. 3B). An immunoblot was also performed showing that consistent with EVs from breast cancer cells 37 , EpCAM (CD326) were abundant in the isolated EVs (Fig. 3C).
To test whether enoxacin or bis-enoxacin alter the regulatory activity of EVs from 4T1 cells, EVs were isolated from cells treated for 3 Days with vehicle, 50 µM of enoxacin or 50 µM bis-enoxacin. The EVs (5 × 10 7 based on nanoparticle tracking) were then added to RAW 264.7 cells, a model for osteoclast precursors, which had been plated at a low and high cell density, and an MTT proliferation assay was performed. EVs from vehicle -treated 4T1 cells did not affect proliferation of RAW 264.7 cells. However, EVs from enoxacin and bis-enoxacin treated cells triggered significant increases in proliferation (Fig. 4A).
Calcitriol-stimulated mouse marrow were treated with 5 × 10 7 control EVs or EVs from 4T1 cells treated with 50 µM enoxacin or bis-enoxacin on Days 1 and 4 of a 6-day culture period. The marrow cells were fixed and stained for tartrate-resistant acid phosphatase (TRAP) activity, a marker for osteoclasts, on day 6. EVs from vehicle treated 4T1 cells did not significantly alter osteoclast formation compared with no EV controls. EVs from enoxacin and bis-enoxacin-treated 4T1 cells significantly inhibited osteoclast formation compared with no EV controls as well as EVs from untreated 4T1 cells ( Fig. 4B-F).

Discussion
For the first time, we show that enoxacin and bis-enoxacin alter the regulatory activity of EVs released by a type of cancer cell at a relatively low concentration (50 µM). EVs from 4T1 murine breast cancer cells treated with enoxacin or bis-enoxacin, but not controls, increased proliferation by RAW 264.7 cells and strongly inhibited osteoclast formation in calcitriol-stimulated mouse marrow, which is a proven and widely-used cell culture model for the bone microenvironment. Enoxacin and bis-enoxacin treated 4T1 cells released EVs that were enriched 22-fold and 30-fold in miR-214-3p, while other microRNAs were reduced. Why would EVs from enoxacin and bis-enoxacin-treated 4T1 cells stimulate proliferation of RAW 264.7 cells (a model for osteoclast precursors), but inhibit osteoclast formation in mouse marrow? In mouse marrow hematopoietic osteoclast precursors and mesenchymal osteoblast precursors carry on an intercellular communication dialog in response to calcitriol that leads to osteoclast formation. For this to occur in cell culture, the cells must be at the correct density and stage of differentiation. Disruption of any aspect of this differentiation process, like stimulating excess proliferation of osteoclast progenitors, could disrupt the overall process. At this point, we consider the crucial result to be the clear demonstration that treatment of 4T1 cells with low concentrations of enoxacin or bis-enoxacin alters the regulatory activities of the EVs that they release.
We found that enoxacin and bis-enoxacin inhibited proliferation of 4T1 cells with a concentration-dependence that was very similar to previous studies, which had examined various types of human cancer. These included colorectal, gastric, lung, breast, liver and pancreatic cancer, as well as leukemia and lymphoma 6,7 . Despite failing to affect proliferation, 50 µM of both enoxacin and bis-enoxacin stimulated GW/P body formation in 4T1 cells. Enoxacin at 50 µM significantly increased only one of five cellular microRNAs tested, miR-214-3p, while bis-enoxacin stimulated modest increases in the levels of miR-214-3p, miR-290 and miR-689. Our data show that enoxacin stimulates GW/P body formation in 4T1 cancer cells at concentrations where it does not function as a general stimulator of microRNA activity. This aligns with a very recent report that showed that although mouse embryonic fibroblasts lacking TAR RNA-binding protein 2 (TARP2), the proposed target of enoxacin 6 , were more resistant to growth inhibition by enoxacin 39 , changes in microRNAs stimulated by enoxacin were not influenced by the lack of TARP2. Taken together, these results suggest that the mechanism of enoxacin's effects on cells may involve TARP2, but is likely more complex than general upregulation of cytosolic microRNAs. In addition to stimulating microRNAs, other proposed mechanisms for the observed therapeutic effects of enoxacin include, blocking interaction between V-ATPase B-subunit and microfilaments 12 , reduction in expression of lymphocyte cytosolic protein 1 40 , induction of alternative splicing of mdmX transcript and rescue of P53 activity 41 , ROS-activated DNA damage 42 , and suppression of the JNK-signaling pathway 14 .
Many recent studies suggest EVs released by cancer cells have the capacity to regulate cells in their microenvironment so that they accept the presence of the cancer cells and promote their survival [43][44][45][46] . For example, cancer-derived EVs containing transforming growth factor-β trigger cancer associated fibroblasts to differentiate into myofibroblasts and promote cancer growth and angiogenesis 47 . EVs from breast cancer cells have been shown to deliver microRNAs that alter glucose metabolism in the premetastatic niches to promote metastasis 48 . Integrins in EVs released by cancer cells have been implicated in their organotropism 49 . Transfer of integrins from tumor cells to non-tumor cells in EVs can trigger migration 50,51 . EVs have been reported to either promote immune response or be strongly immunosuppressive 52,53 . Pharmacological alterations of the regulatory activity of cancer EVs, as shown for enoxacin and bis-enoxacin, therefore may represent a novel type of anti-cancer therapeutic approach. Breast cancer cells are thought to stimulate the formation of osteoclasts and inappropriate bone resorption. This produces growth factors for the cancer cells and allows the cancer cells to have access to the bone 54 . The ability of enoxacin and bis-enoxacin to stimulate the release of EVs that inhibit osteoclast formation could be useful in the treatment of bone cancer. Bis-enoxacin is a bone-targeted molecule that is an attractive candidate to be an agent to treat bone cancer. In addition to its intrinsic anti-resorptive effects, it also influences the EVs released by cancer cells so that they inhibit osteoclast formation. Changes in EV composition may also contribute to the effects of enoxacin and bis-enoxacin on bone remodeling. Recent studies support the idea that EVs from both osteoclasts and osteoblasts have roles in intercellular communication between osteoclasts and osteoblasts 33,38,55,56 . Transfer of microRNA-214-3p was recently implicated as a mechanism by which osteoclast EVs regulate osteoblasts 33,38 . The work of Li and colleagues provided evidence both in vitro and in vivo for a role for the transfer of miR-214-3p from osteoclasts to osteoblasts in EVs in the development of osteoporosis. Their data also correlated with human studies which showed increased levels of circulating miR-214-3p to be linked to osteoporosis. Our finding that miR-214-3p is greatly increased in cancer EVs provides a plausible mechanism by which the observed changes in the regulatory activities of EVs from 4T1 cells could be produced, although additional studies are required to test this hypothesis. Increased miR-214-3p in osteoclasts was identified as a mediator of bone cancer metastasis 34 , which suggests increasing its levels in cancer EVs may not be helpful. However, others showed miR-214-3p inhibited hepatocellular carcinoma proliferation 33 and increased miR-214-3p enhanced the effects of cisplatin on squamous carcinoma cells 57 .  In summary, we find that enoxacin and bis-enoxacin alter 4T1 cancer cell-derived EVs so that they inhibit osteoclast formation in calcitriol-stimulated mouse marrow. Enoxacin and bis-enoxacin strongly enhanced the export of miR-214-3p from 4T1 cells in EVs while significantly reducing export of others. MiR-214-3p has been implicated as an important regulator of bone remodeling 33,35,58 . Finally, we suggest that bis-enoxacin represents a new potential agent for the treatment of bone cancer. By both inhibiting osteoclasts directly and triggering the release of EVs from cancer cells that inhibit osteoclast formation, it may prove useful for disrupting the ability of cancer cells to invade bone.
Cell culture. 4T1 murine breast cancer cells (kind gift of Gary Sahagian, Tufts University, Boston, MA) were grown in dMEM plus 10% fetal bovine serum (FBS) as described 59 . In some experiments the 4T1 cells were grown in dMEM plus 10% exosome-free fetal bovine serum (Systems Biosciences). Calcitriol-stimulated mouse marrow was prepared as described previously 60 . All protocols using of animals were approved by the University of Florida Institutional Animal Care and Usage Committee. All procedures were performed in accordance with guidelines of the National Institutes of Health of the United States. Cervical dislocation was performed to sacrifice Swiss-Webster mice. Femora and tibia were dissected from the mice, and marrow was removed by cutting both bone ends, inserting a syringe with a 25-gauge needle, and flushing the marrow using α-MEM plus 10% fetal bovine serum (α-MEM D10). Marrow was washed and plated at a density of 4 × 10 4 cells/cm 2 on 24-well plates for 5 d in α-MEM D10 with 10 −8 M calcitriol plus EVs as noted. Cultures were fed on day 3 by replacing half the media per plate and adding fresh calcitriol. After 6 days in culture, osteoclasts were present. These were detected as giant cells that stained positive for tartrate-resistant acid phosphatase activity (TRAP; a marker for mouse osteoclasts). RAW 264.7 cells (ATCC), a cell line that serves as a model for macrophages and osteoclast precursors 11 , were grown in α-MEM D10.
Growth Curves and Cell counting. Cells in three random locations (1000 µm × 1000 µm square) per well were counted by blinded counters. At least 3 wells per treatment group were counted at each time point.
MTT assay. MTT proliferation assays were performed using a kit from Cayman Chemical (Ann Arbor, MI) according to the manufacturer's directions. From 1 × 10 5 to 5 × 10 5 cells 4T1 or RAW 264.7 cells were seeded onto 96-well plates and cultured for 24 or 48 H. Absorbance was determined using a BioTek Synergy HT microplate reader (Winooski, VT) at 570 nm.
Caspase-3 assay. Caspase-3 activity was assayed as described previously 12   cells were then washed with PBS two times, for 10 minutes each, blocked in blocking buffer (PBS with a pH of 7.4 plus 1% bovine serum albumin (BSA)) for 60 minutes and probed with 200 μL of the human anti-GW bodies (GWBs) reference serum 18033, which reacts against GW182 and GE-1 at 4 °C overnight 61 . Then, the cells were washed two times in PBS for 10 minutes each, and secondary antibody, Alexa-Fluor-594 conjugated goat antibodies to human IgG was incubated for 60 minutes at room temperature in the dark. Then, the cells were rinsed two times in PBS for 10 minutes each. GW/P bodies were detected by epifluorescence microscopy. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). The number of GW/Ps per cell, as well as the number of multinuclear cells, was quantitatively assessed using the computer program, ImageJ. Results are expressed as mean plus/ minus standard error. Samples were compared by a Student's t test using GraphPad software. P values < 0.05 were considered significant.
Isolation of cancer EVs from cell culture and testing miRNA levels. Cells were grown in dMEM plus exosome-free media. The media was harvested and any debris and cells were removed by performing centrifugation at 5000 X g for 30 min. The cell-free culture media was then added to ExoQuick and incubated at 4 degrees Celsius overnight. The ExoQuick pellet was obtained by centrifugation at 5000 × g/30 min. EVs were then spun at 150,000 × g for 2 hours in an Airfuge (Beckman) in order to remove the solubilized ExoQuick. EVs were present in the pellet and then resuspended in sterile PBS. EVs were assessed by nanoparticle tracking. Quantitation of EVs was performed by using the Exocet kit (Systems Biosciences) following the manufacturer's instructions or by nanoparticle tracking using a Nanosight NS-300 (Malvern Size determination by Nanoparticle tracking. The diameter size and concentration of EV population was determined using a NanoSight NS-300 (Malvern). Samples were evaluated using different dilutions (1/10 to 1/50) in sterile-filtered PBS and the following parameters: camera at 30 frames per second, camera level at 16, temperature between 21-25 °C and video recording time at 60 s in order to estimate the concentration and size distribution of vesicles by light scattering and Brownian motion. Nanosight NTA Software analyzed raw data videos by triplicate.
Transmission electron microscopy. EVs were visualized as described previously 55 . EV pellets were resuspended in 2% paraformaldehyde, placed on Parafilm, and Formvar-carbon-coated electron microscope (EM) grids were floated atop the EVs for 20 min. Grids were washed in PBS, transferred to 1% glutaraldehyde for 5 min, and then washed in deionized water 8 times. The grids were contrasted and embedded by placing them on a drop of uranyl-oxalate solution and transferred to methyl cellulose-uranyl acetate on ice for 10 min. The grids were blotted dry with filter paper and observed under an electron microscope. Electron microscopy was performed at the University of Florida, College of Medicine, Electron Microscopy Core Facility using a Hitachi (Tokyo, Japan) 7600 transmission electron microscope operated at 80 kV.
Immunoblotting. EVs isolated from 4T1 cells treated with vehicle, enoxacin or bis-enoxacin were extracted by solubilizing the pellet in Laemmli buffer (BioRad, Hercules, CA) followed by boiling for 5 min. Samples were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis on 4-20% gradient gels and transferred to nitrocellulose. Blots were probed with Anti-EpCAM (D269-3) obtained from MBL International (Woburn, MA, USA).

Detection of osteoclasts.
Osteoclasts were detected by staining for tartrate resistant acid phosphatase (TRAP) activity as described previously 60 . Cells were counted by volunteers who were blinded to the treatment conditions and who had been calibrated. We documented mononuclear, multinuclear (2-10 nuclei) and Giant (>10 nuclei) cells that were positive for TRAP activity as described previously 60 Statistics. Results are expressed as mean plus/minus Standard Error. Samples were compared by One-Way ANOVA and Student's t test using the program GraphPad Prism 5 (GraphPad Software, La Jolla, CA). P values < 0.05 were considered significant. Nanoparticle tracking was analyzed by ANOVA with Tukey's modification.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.