Inhibitory Effects of Doxycycline on Tumor Progression In vitro and on Metastases in Early-Stage Osteosarcoma Xenografts Mice.

Osteosarcoma (OS) is the commonest primary osseous malignant tumor with a high propensity to metastasize in lungs. Pulmonary widespread micrometastatic lesions are present in up to 80% of patients at initial diagnosis and they are associated with signicantly worse prognosis. Doxycycline (Dox) is a synthetic tetracycline that has been shown to have anti-cancer properties in vitro and in vivo, and inhibit angiogenesis, effects that may prove benecial for several types of cancer. The aim of the present work was to study how Dox affects OS cells’ growth in vitro and in vivo and OS-driven pulmonary metastasis in vivo.


Background
Osteosarcoma (OS) is the most common primary malignant bone tumor with an estimated incidence of 1.7 -4.4 per million people being diagnosed every year [1]. While OS frequently affects people between the age of 5 and young adulthood, it can also occur in people over 65 [2]. The distal femoral metaphysis and the proximal tibia are the most usual sites for primary OS to develop [3]. Usually, OS presents as a high-grade intramedullary bone lesion with malignant mesenchymal cells, which invade surrounding tissues and produce immature bone, known as osteoid [3]. Approximately 80% of newly diagnosed OS cases have subclinical pulmonary micro metastases without radiologic evidence [4], leading to a signi cant decrease in the overall 5-year survival compared to the metastases-free OS patients (overall 5-year survival rate 30% compared to 70% for those free of distant metastases) [5]. Currently, the treatment protocol for OS includes neo-adjuvant chemotherapy (cisplatin, doxorubicin, ifosfamide, methotrexate) followed by wide surgical resection of the primary tumor and adjuvant chemotherapy [6]. Although the suggested OS treatment protocol presents the practice of choice for the treatment of OS, it does not seem to affect the metastatic rate of this very aggressive disease [7].
Dox is a synthetic third-generation tetracycline with a broad antibiotic spectrum. Its anti-cancer effects in several types of cancer have been previously examined with promising results [8,9,10,11]. There is also limited in vitro evidence obtained from OS cell lines suggesting that Dox may be a potential OS anticancer agent [12,13,14]. One of the mechanisms through which Dox may exert its anticancer effects has been suggested to be its ability to inhibit matrix metalloproteinases (MMPs) due to calcium chelation [15]. MMPs are calcium-dependent, zinc-containing endopeptidases secreted by tumor cells with fundamental involvement in the degradation of the extracellular matrix (ECM), cancer cell invasion, metastasis and angiogenesis [16,17]. MMPs may be also important for OS progression and targeting MMPs has been suggested as a potential therapeutic approach [18]. Experimental trials have suggested that MMPs interfere with the secretion of vascular endothelial growth factor A (VEGFA) from cancer cells [19]. MMPs and VEGFA help OS micro metastatic disease to stimulate local angiogenesis and remodel the microenvironment to support tumor neovascularization. In end-stage disease, micro metastases expand to become clinically detectable and give rise to lethal macro metastases [20].
The aim of the present study was to determine the in vitro and in vivo antitumor effect of Dox in OS and its potential for the prevention or limitation of pulmonary metastases in early-stage OS xenografts.

Cell viability Assay (Trypan blue exclusion)
OS cells were plated at a cell density of 3.5 × 10 4 cells/well in 6-well plates in EMEM supplemented with 10% FBS. Twenty-four and 48 h after seeding cells, trypan blue exclusion assay was used to determine the number of living MG-63 and 143B cells exposed to 0, 5, 10 and 20 µg/ml of Dox [21].
2.3 Cell migration assay (Wound healing assay) OS cells cultured in EMEM supplemented with 10% FBS were seeded into 24-well tissue culture plate wells at a density of 2×10 6 cells/well so that after 24 h of growth, they should reach 90-95% con uency as a monolayer. The monolayers were scratched vertically with a sterilized 200 µl pipette tip across the center of the well. After scratching, the wells were washed twice with 1x phosphate-buffered saline (PBS) to remove the detached cells and replenished with fresh medium containing Dox (20 µg/ml) or PBS (control). Cells were photographed at 0, 8, 16, and 24 h after scratching, using an Olympus Bx40 microscope. TScratch software version 7.8 (Computational Science and Engineering Laboratory, Swiss Federal Institute of Technology, Zurich, Switzerland) was then used to perform image analysis and measure the gap areas [21,22].

Apoptosis Assay (Fluorescence-activated cell sorting/FACS)
The effect of Dox in OS cell lines apoptosis was carried out using ow cytometry. Brie y, 143B and MG-63 cells were xed overnight at 4 0 C in 70% ethanol. The xed cells were stained with RNase-containing propidium iodide (PI) and Annexin V FITC solution (TACS Annexin V FITC, Apoptosis Detection Kit, Gaithersburg, MD, USA). Cells were then separated as early apoptotic (Annexin V FITC-stained), late apoptotic (Annexin V FITC and PI-stained), necrotic (PI-stained) or non-apoptotic/live cells (no staining). DNA content was analyzed using a FACS caliber ow cytometer (Becton Dickinson, San Jose, CA) and MoD Fit software (Verity Software House, Topsham, ME) [23].

In vivo orthotopic implantation of 143B human OS cells in mice xenografts
143B is a highly tumorigenic OS cell line with high metastatic rate [24,25,26]. The injection of 143B cells in mice tibias generates either spontaneous (tumor cells interact with their native microenvironment, invade local vessels, and move to distant sites) or experimental (direct seeding of tumor cells in lungs during the injection procedure) pulmonary metastases [27,28,29].
Thirty-two 6 to 8-week-old female severe combined immunode cient (SCID) mice were obtained from the National Center for Scienti c Research (Demokritos, Greece). Mice were acclimatized for 10 days without any interventions after transportation to the Laboratory for Experimental Surgery and Surgical Research "N.S. Christeas", Medical School, University of Athens, Greece, where the in vivo study was conducted. The animals were housed individually in clean metabolic cages placed in a well-ventilated house with optimum conditions (temperature 23 ± 1°C; photoperiod 12 h natural light and 12 h dark; humidity 45-50%) with access to food and water ad libitum during the entire study period. On day 0, the animals were anesthetized using a ketamine (100 mg/kg, Imalgene 1000; Merial, France) -xylazine (10 mg/kg Rompun; Bayer Animal Health GmbH, Germany) mixture injected intraperitoneally (IP) [30]. 143B OS cells (1 x 10 6 in 100 µl of PBS) were injected into the left proximal tibia of SCID mice using a 25-G needle [31]. All procedures were approved by the Regional Veterinary Service (no:366107/July 9, 2019) the Ethics Committee of the NKUA/Medical School (no:163/September 18, 2019) and were in accordance with National Legislation and European Directive 63/2010.
Four weeks after tumor inoculation (Day 28), all animals were anesthetized, blood was drawn from the orbital sinus and mice were euthanized via cervical dislocation [34,35]. During the inoculation process 4/32 mice died after the acute onset of tachypnea, possibly due to pulmonary embolism. The left tibias (primary tumor sites) were evaluated with x-rays (Chirana type RK 75-10 EKv 2mm AL, Film Agfa 100NIF 25x30, Cassette Agfa CR MD 4.0 General) three weeks after engraftment of 143B cells to assess tumor formation at primary sites.
Resected xenograft tibias and tumor dimensions were measured using a digital caliper according to the formula: volume = (L + W) (L) (W) (0.2618). The value assigned as width (W) was the average between the anterior -posterior and medial -lateral planes of the proximal tibia. The value assigned as length (L) was the distance between the most proximal and the most distal tumor margin [28]. The lung volumes were calculated with a digital weight scale, and lung tissues were investigated for macroscopically detectable metastases using a stereoscope.

Surgical technique
Mice were anesthetized using a ketamine -xylazine mixture, as described above, injected IP [30]. A small animal heating pad was used during surgery to maintain normothermia. While under anesthesia, the left tibia of the mouse was shaved, cleaned with povidone iodine, and then rinsed with alcohol. The knees were exed in 90 0 and tibias were removed after midshaft femur cut to achieve tumor-free margins. The skin was closed with size 4-0 mono lament nylon sutures [33]. During this procedure, a mouse from group C died due to excessive bleeding (group C, new n=6). Buprenorphine (0.1 mg/kg, IP every 8 h; Bupaq® 0.3, Neocell Ltd.) was used for pain control over the rst 24 h after amputation (Day 5) [36].

Quantitative measurement of blood biomarkers MMP9 and VEGFA
On day 28, blood samples were aspirated from mice under anesthesia with retro-orbital technique using ne-walled glass Pasteur pipettes (diameter:150mm). The blood samples were collected in heparin coated microhematocrit tubes and stored on crushed ice for no longer than 30 min before being centrifuged at 8000 rpm at 4°C for 10 min. Mouse serum was examined for MMP9 and VEGFA protein levels by ELISA (#ab100610 and #ab100662 respectively, Abcam, Cambridge, UK), according to manufacturer's instructions. The cell line we used to generate the in vivo OS model was human and we used human anti-VEGFA and anti-MMP9 ELISA kits to avoid cross-reaction with mouse VEGFA and MMP9 [37].

Histological Examination And Immunohistochemistry (Ihc)
Primary tumors and lungs were collected from mice, xed in 4% paraformaldehyde in Tris-buffered saline (TBS) at 4°C for 18 h. Para n embedded tissue specimens were sectioned at a thickness of 3.0 µm, followed by depara nization in xylene and dehydration in a graded series of ethanol solutions

Histological examination
Each section was photographed in its entirety and subsequently, digitally processed utilizing the IpWin6 program. Tumorous areas were manually demarcated, and the program proceeded to assess the percentage of tumor areas to overall tissue surface. The same procedure was conducted to assess the percentage of necrotic tumor areas to the overall tumor surface in primary sites.
The apoptotic effect of Dox on the same cells was examined by Annexin V-FITC and PI staining following a 24 h exposure of cells to either 20 µg/ml Dox or solute (PBS). Dox signi cantly enhanced late apoptosis in both types of OS cells, but also induced necrosis of these cells (Fig. 1B), in line with its inhibitory effect on the number of viable cells shown in Fig. 1A.

Doxycycline inhibits migration of MG-63 and 143B human OS cells
The in vitro effect of Dox in the migration capacity of OS cells was examined by a wound healing assay. Both Dox treated OS cell lines (MG-63 and 143B) presented a signi cant reduction in their mobility even at 8 h after the onset of treatment (p < 0.0001 in both cases) (Fig. 2).

Endpoint metrics in 143B OS xenografts
The volume (cm 3 ) and weight (g) of primary tumors in left tibias were measured in the Dox − /Amp − and Dox + /Amp − groups. No differences in primary tumor volume and weight were found between Dox treated and non-treated xenografts (Fig. 3A). The percentage of primary tumor necrosis in the Dox + /Amp − group was signi cantly increased (P < 0.0001) compared to the Dox − /Amp − group (Fig. 3B). Histopathologic evaluation of primary tumors with hematoxylin and eosin (H&E) staining revealed malignant tumor and stroma cells with destruction of normal bone (Fig. 3C).
The Ki67 nuclear protein is a well-established prognostic and predictive indicator in OS biopsies [42,43]. We found a statistically signi cant decrease (P <0.041) in the expression of Ki67 in primary tumors of the  (Fig. 4).

Doxycycline prevents the formation of pulmonary macro metastatic disease
In all groups, abnormal lesions were recognized on the lung surfaces of numerous mice and they were recorded as metastatic sites. We used H&E staining (Fig. 5A) to evaluate the percent metastatic surface / total lung surface ratio for each xenograft from our groups. As shown in Fig. 5B, the number of metastases was signi cantly lower in the Dox + /Amp − compared to the Dox − /Amp − group. Similarly, the number of metastases was signi cantly lower in the Dox − /Amp + compared to the Dox − /Amp − group. The decrease observed in the Dox + /Amp + group compared to the Dox − /Amp + group did not reach statistical signi cance. Early amputation also resulted in a signi cantly lower lung weight independently of Dox treatment (Fig. 3C).

The expression of markers in pulmonary macrometastatic lesions
The expression of Ki67, MMP2, MMP9, VEGFA and Ezrin in metastatic lesions was high in all xenografts and no evidence of statistically signi cant difference between the Dox-treated and untreated groups was found ( Fig. 6 and Table 1). The metastatic lesions were vimentin positive and E-cadherin negative, similarly to the primary tumors shown in Fig. 4. Table 1 The intensity of antibodies staining in pictures like those presented in Fig. 6  The hypothesis that blood levels of VEGFA could have a predictive and prognostic value in OS patients has been previously assessed, showing that VEGF blood levels are signi cantly higher in OS patients compared with healthy counterparts and correlate with response to chemotherapy or metastasis [45]. The predictive role of MMP9 blood levels has been discussed for several types of tumors [46], but not yet in OS patients. In the present work, we evaluated the correlation between VEGFA and MMP9 blood levels and response to Dox. Both VEGFA and MMP9 blood levels were high in the non-treated Dox − /Amp − control group and they were signi cantly decreased in all other groups (P < 0.0001, one way ANOVA). The decrease observed in the Dox + /Amp + group compared to the Dox − /Amp + group did not reach statistical signi cance (Fig. 7), similarly to what has been observed in the formation of pulmonary metastases shown in Fig. 5B.

Discussion
Concurring to the literature [12,13,14], our in vitro evidence suggests a dose-dependent effect of Dox in decreasing the number and the migration of OS cell lines. In addition, Dox effectively induced both apoptosis and necrosis of OS cell lines, in line with the decreased number of cells and the enhanced necrosis observed in vivo. The in vitro anti-tumor effect of Dox on MG-63 and 143B cells was studied at a dose between 5 to 20 µg/ml. To our knowledge, this in vitro treatment range has been proposed as the equivalent human dosage for Dox [11][12][13].
Since lung is the most common site of metastasis in OS patients and pulmonary metastasis is associated with a signi cant worse prognosis in OS patients, we also aimed to assess the use of Dox as a supportive therapy in OS patients for the prevention of pulmonary metastasis. In our in vivo human 143B OS xenograft model, doxycycline hyclate at 50 mg/kg/d was intraperitoneally administered for 28 days in dox-treated mice. A study by Luccheti J et al, revealed that the intraperitoneal injection of Dox in C57BL76 mice at a dose of 10 -100 mg/kg had led in a peak plasma concentration of 2-10 µg/mL which is superimposable to the established oral treatment of Dox at 100-200 mg per day in humans [47]. In addition, we designed our study in accordance with a previously conducted study which reported a great tolerance of Dox when administered at a dose of 50 mg/kg/d in breast BALB/c xenografts [32]. The only study which evaluated the suppressing effect of Dox in tumor progression of human OS-or rhabdomyosarcoma-implanted athymic mice had used Dox at ∼30 mg/kg/d in drinking water with promising results [48]. However, in our opinion the administration of Dox 'ad libitum' via drinking water is not well controlled and this consists a limitation of the experimental study by Dickens D et al. Other limitations of that study are that the effect of Dox in the prevention of pulmonary metastatic disease was not examined in lung tissue biopsies, and the design of that study had not mimicked the standard treatment strategy of human OS, which includes neoadjuvant and adjuvant administration of chemotherapeutic agents accompanied with surgical excision of primary site [6]. Our study is the rst experimental trial that has mimicked the clinical therapeutic strategy as in human OS. We administered Dox preoperatively, as well as postoperatively, after wide resection of the primary tumor in one of our study groups (Dox + /Amp + ) and the results were compared with control groups.
In our study, we observed that treatment with 50 mg/kg Dox resulted in a smaller number of metastases compared with the numerous metastatic lesions in the Dox − /Amp − control group. In addition, the mice treated with neo-adjuvant Dox, early amputation and adjuvant administration of Dox were almost free of metastases. Dox has been suggested to have anti-cancer properties through inhibition of MMPs, but the timing of an MMP inhibitor application in cancer is critical to achieve the desired therapeutic effect [49]. However, previously conducted phase I, II and III clinical trials that have evaluated the effect of MMPinhibitors (e.g., the broad-spectrum MMP inhibitor Marimastat and the chemically modi ed tetracycline Col-3) as chemotherapeutic agents in advanced end-stage metastatic cancers (e.g., breast, pancreatic, gastric) [50,51,52,53] were not encouraging. This may have resulted from the fact that these MMP inhibitors had not been used at early-stage cancers before the establishment of macro metastatic disease. A novel proposal of our in vivo study is the administration of Dox in early-stage OS that seems to prevent the progression of micro metastatic to macro metastatic lethal disease.
The present study demonstrated that circulating levels of human VEGFA and MMP9, secreted by the human OS cells, were decreased by Dox in line with the fewer macro metastases, resulting in better prognosis. The circulating levels of human VEGFA and MMP9 in the Dox − /Amp + and Dox + /Amp + groups were much lower compared to the groups that did not undergo early amputation. In these groups, both VEGFA and MMP9 were most likely secreted by the OS cells that formed pulmonary micro metastases and could be used as a biomarker even in cases when these metastases cannot be detected.
Similarly, to the in vitro model, the Dox-treated mice developed increased tumor necrosis in primary tumors, supporting the prognostic value of tumor necrosis in primary site for the development of systemic disease in OS [54]. Furthermore, Dox treated mice depicted fewer metastases compared to nontreated mice. On the contrary, the tumor size and the tumor volume were not affected by Dox administration and were not correlated with a higher or a lower metastatic surface / total lung surface ratio. This nding is not in accordance with bibliography which tends to recognize the signi cance of primary tumor size as a prognostic factor for subsequent lung metastases in OS [55]. The reason for this discrepancy was that some of our non-amputated mice developed limping and skin ulcers on primary sites and had to be euthanized early from the 28th day respectful to their welfare. Consequently, a signi cant difference in tumor size between Dox-/Amp-and Dox+/Amp-groups could not be entirely assessed.
The effect of Dox on EMT reversal in our in vivo model is veri ed by the immunohistochemical evaluation of E-cadherin (epithelial marker) and Vimentin (mesenchymal marker) [56]. In our study, primary tumors as well as pulmonary metastatic lesions were Vimentin positive and E-Cadherin negative. No signi cant difference in the intensity of the signal was recognized among Dox-treated and non -treated xenografts. Subsequently, we concluded that Dox prevents metastatic spreading to the lungs in OS, but this effect is not related to an EMT process, which has been also described in other tumors, such as lung cancer [57].
In Dox-treated mice, the expression of VEGFA, MMP2, MMP9, Ki67 and Ezrin in primary tumors were all downregulated and were strictly associated with a better prognosis. The association between VEGFA, MMP2, MMP9, Ki67 and prognosis is well known for many types of cancer [58,59]. Ezrin has been shown to positively regulate the expression of MMPs and VEGFA in tumor cells and promote the metastatic potential of tumor cells [60,61]. In pulmonary metastases, the expression of these markers was decreased in the amputated compared to the non-amputated groups but were not affected by Dox treatment. This may be due to the well described differences between metastatic cell clones and primary tumor cells that include their chemosensitivity to anticancer agents [62, 63].
Despite our effort to provide a close approximation of human OS using a humanized mouse model and provide therapeutic outcomes of Dox with clinical relevance, our study has two main limitations.  Figure 1 The  The effect of Dox in the migration of OS cell lines. Migration of MG63 and 143B cells was detected via wound healing assays (magni cation X20). At different time points after scratch, the gap areas were quanti ed using proper software as described. Solid lines mark the initially scratched area and dotted lines show the area that was not covered with cells at the time points shown. When the whole area was covered with cells, no lines are present. The area that remained uncovered with cells was quanti ed at Weight of the lungs at the end of the experiment expressed as mean ± SD. In all groups n=6 except Dox+/Amp-in which n=8.

Figure 6
Expression of markers in pulmonary macrometastatic lesions. Representative para n-embedded mouse lung sections stained using antibodies against vimentin and E-cadherin, Ki67, MMP2, MMP9, VEGFA and Ezrin. Positive staining is brown. Counterstain with hematoxylin is shown as blue. Quanti cation is presented in Table 1.