Assessment of a novel nanoparticle hyperthermia therapy in a murine model of osteosarcoma

Abstract

Objective:

To evaluate the effects of nanoparticle hyperthermia therapy on monocyte function and tumor-derived factors associated with macrophage polarization in a murine osteosarcoma model.

Study design:

Experimental study.

Animals:

Female C3H mice.

Methods:

Peripheral blood monocyte cell surface phenotype, monocyte chemotaxis, tumor messenger RNA expression, and survival were compared among osteosarcoma (OS)-bearing mice treated with nanoparticle hyperthermia therapy, OS-bearing mice with osteomyelitis, OS-bearing mice, vehicle control mice, and normal control mice.

Results:

OS-bearing mice with osteomyelitis had a higher proportion of “nonclassical” monocytes (Ly6C^lo) compared with all other experimental groups. There were alterations in monocyte expression of multiple chemokine receptors among experimental groups including CXCR2, CCR2, and CXCR4. Monocytes from OS-bearing mice treated with hyperthermia therapy exhibited greater chemotaxis compared with monocytes from OS-bearing mice with osteomyelitis.

Conclusion:

OS likely induced alterations in monocyte phenotype and function. Nanoparticle hyperthermia therapy increased in vitro monocyte chemotaxis.

Clinical impact:

Enhancing monocyte/macrophage function in dogs with OS may enhance antitumor immunity.

1 |. INTRODUCTION

Osteosarcoma (OS), a malignant primary tumor of bone, is a devastating disease for both human and canine patients. It is the most common primary bone tumor in the dog as well as in children and adolescents and arises from mesenchymal cells that produce osteoid.^1,2 Despite successful removal of the primary tumor and administration of neoadjuvant/adjuvant chemotherapy, metastatic disease continues to be the primary cause of death for dogs and man, with stagnation in survival times for both species during the past 30 years. The 5-year disease-free interval for human patients with nonmetastatic OS remains about 70%, with only 20%−30% long-term survival in patients with metastases, and the median survival in dogs with OS remains between 10 and 12 months despite various permutations of chemotherapeutic regimens.^2–5 Advancements in OS therapy are clearly required to improve survival.

The improved survival associated with surgical site bacterial infections is an interesting phenomenon, observed in both canine and human OS.^6–8 A retrospective study reported that dogs with infections of the surgical limb-salvage site were half as likely to die and half as likely to develop metastasis, and these survival effects were due to a delay in metastasis.⁶ Studies using a murine model of chronic bacterial osteomyelitis have confirmed these observations; mice with OS and bacterial osteomyelitis experienced improved survival and had increased numbers of inflammatory monocytes and depletion of monocytes/macrophages in these mice negated the survival benefit.⁹ Macrophages exhibit a high degree of plasticity, with M1 (proinflammatory)/M2 (anti-inflammatory) macrophages being the most commonly recognized phenotypes, representing both ends of the polarization spectrum.¹⁰ M1 macrophages are induced by bacterial products such as lipopolysaccharide and cytokines such as interferon γ, whereas M2 macrophages are induced by cytokines such as interleukin (IL)-10, IL-4, and transforming growth factor-β (TGFβ). M1 macrophages exhibit heightened tumor antigen processing and presentation and efficiently produce inflammatory cytokines and cytotoxic mediators. M2 macrophages display increased expression of markers such as mannose receptor (CD206) and secrete anti-inflammatory molecules such as TGFβ.¹¹ M2 tumor associated macrophages (TAM) are protumorigenic in many cancers, whereas M1 TAM are associated with heightened tumor immunity.¹² Taken together, these findings suggest that the antitumor effects of infection may be modulated via upregulation of the inflammatory response and activation of monocytes and macrophages. Finding a way to recapitulate the antitumor effects associated with infection-induced inflammation holds promise as a novel treatment modality for OS.

Hyperthermia therapy has long been evaluated as an antitumor treatment, especially as a method of debulking primary tumors via irreversible thermal damage.^13–15 Nanoparticle-assisted hyperthermia therapy involves localized administration of nanoparticles within a tumor, followed by application of an external energy source such as near-infrared lasers and alternating magnetic fields to induce tumor hyperthermia. The tumor hyperthermia induces localized necrosis and apoptotic signals that result in upregulation of tumor specific antigens (TSA) and heat shock proteins (HSP).¹⁶ TSA and HSP released from the tumor in response to heat stress are subsequently recognized by antigen presenting cells, which in turn activate T cells, promoting a vigorous antitumor response.¹⁶ Nanoparticle-assisted hyperthermia therapy thus has the potential to eliminate the primary tumor and to target metastatic disease by enhancing the antitumor immune response.

Magnetic cationic liposomes (MCL) are an efficient method of inducing intracellular hyperthermia because MCL have been shown to accumulate with a 10-fold higher affinity for tumor cells compared with neutrally charged magnetic liposomes because of the electrostatic attraction between MCL and negatively charged cell membranes.¹⁷ After deposition of MCL into the tumor, an alternating magnetic field (AMF) is applied, which causes the MCL to generate heat through mechanisms including Brownian motion and particle-to-particle interaction.¹⁸ Several reports have documented the efficacy of MCL-hyperthermia therapy in rodent models against various tumors, including OS.^19–22 MCL-hyperthermia therapy thus warrants investigation as a novel therapeutic option for OS.

To this end, we sought to evaluate the efficacy of MCL-hyperthermia therapy using a syngeneic murine model of OS and to compare the effects of the therapy with effects of bacterial osteomyelitis on OS, which was previously reported to affect survival outcomes in OS positively.⁹ Normal healthy mice, mice with untreated OS, and mice with OS injected with MCL without receiving hyperthermia therapy served as controls. Tumor burden, monocyte cell surface receptor expression and chemotaxis, and tumor messenger RNA (mRNA) signatures were used for comparisons among experimental groups. Our previous work demonstrated that cell surface chemokine receptors and chemotaxis were downregulated in monocytes from dogs with OS compared with normal controls.²³ We posited that these findings represent ways in which OS causes dysregulation of the immune response.²³ We also sought to determine whether the previously observed effects of OS on canine monocyte receptor expression and chemotaxis occur in murine OS. The objectives of our study were (1) to compare survival between experimental groups, (2) to compare monocyte surface receptor expression and monocyte chemotaxis between experimental groups, and (3) to compare tumor mRNA signatures among experimental groups. Our hypotheses were that MCL-hyperthermia therapy or bacterial osteomyelitis leads to longer survival in OS-bearing mice and that MCL-hyperthermia therapy and bacterial osteomyelitis will reverse the effects of OS on monocyte receptor expression and chemotaxis and induce an inflammatory tumor mRNA signature.

2 |. MATERIALS AND METHODS

2.1 |. Ethics statement

All murine experiments were approved by the North Carolina State University, Institutional Animal Care and Use Committee, and humane killing of study animals was performed in accordance with current American Veterinary Medical Association Guidelines for the Euthanasia of Animals. All procedures were performed in an Association for Assessment and Accreditation of Laboratory Animal Care-approved facility.

2.2 |. Animals

Female C3H-HeN inbred mice were used for all experiments. In total, 46 mice aged 8–10 weeks were purchased from Charles River Laboratories (Kingston, Pennsylvania). All mice were randomly assigned to 1 of the following experimental groups: mice with OS that received MCL-hyperthermia treatment (n = 9), mice with OS inoculated with bacterial osteomyelitis (n = 13), mice with OS (n = 11), mice with OS injected with MCL without hyperthermia therapy (vehicle control, n = 6), normal (healthy) control mice (n = 7).

2.3 |. Cell line

The DLM8 murine OSA cell line was provided by Douglas Thamm VMD, DACVIM (Oncology), (Flint Animal Cancer Center, Colorado State University, Fort Collins, Colorado).

2.4 |. Bacteria

Staphylococcus aureus (XEN36) with stable expression of the luciferin and luciferase genes (Perkin Elmer, Waltham, Massachusetts) was used for inducing bacterial osteomyelitis.

2.5 |. Magnetite cationic liposomes

Magnetite particles were synthesized by Kaio Therapy (Raleigh, North Carolina; by CSP) by using a proprietary process, and MCL were prepared by using a previously described protocol.¹⁷

2.6 |. Syngeneic OS model

A subcutaneous syngeneic OS model was established in the mice by using a previously reported technique.⁹ Three days after suture implantation to induce an osteomyelitis, a subcutaneous injection of 2 × 10⁸ DLM8 murine OS cells suspended in 30 μL Hank’s balanced buffer solution (Corning Life Sciences, Tewksbury, Massachusetts) was administered into the right flank region. Tumor growth at the widest diameter was measured 2–3 times weekly with calipers, and mice were killed when their individual tumors measured 18 mm in diameter.

2.7 |. MCL-hyperthermia therapy

Access to the AMF device was provided by Kaio Therapy (Raleigh, North Carolina). MCL-hyperthermia therapy was administered to mice after their tumors had reached 8–12 mm in diameter. On day 1 of treatment, 300 μL of MCL was injected intratumorally for a duration of 30 minutes with a 25 gauge needle and an infusion pump. Hyperthermia therapy induced by an AMF was administered once daily for 30 minutes from days 2 to 4, for a total of 3 treatments. The AMF with a frequency of 100 kHz was created with a magnetic field generator. The mice were anesthetized by an intraperitoneal injection of ketamine (100 mg/ kg) and xylazine (10 mg/kg) and placed with the tumor centered under the AMF coil. Anesthetic variables were monitored every 5 minutes; skin temperatures overlying the tumor and 10 mm away from the tumor were measured with an Optocon (Dresden, Germany) FTC optical sensor, with a goal of reaching 43 °C over the tumor and maintaining normal surface temperature (21 °C-23 °C) 10 mm away from the tumor. Twenty hours after MCL injection, the tumors were heated by exposure to the AMF, whereas adjacent tissue and core body temperature remained constant.

2.8 |. Bacterial osteomyelitis induction technique

Bacterial osteomyelitis was induced in the distal femur of the mice by using a technique modified from a previously reported technique.⁹ The XEN36 S aureus was cultured to achieve log growth. One 12-mm segment of 6–0 silk suture was placed in each well and incubated in a shaking incubator for 36 hours at 37 °C and 100 rpm, with media changes performed every 12 hours. The suture was then imaged by using an IVIS Lumina imaging system in Living Image version 4.2 software (Xenogen; Caliper Life Sciences, Hopkinton, Massachusetts) with a 30-second exposure time and medium sensitivity binning to confirm bacterial adherence to the suture. Mice were anesthetized with an intraperitoneal injection of ketamine (150 mg/kg) and xylazine (10 mg/kg) for suture implantation, and the suture was implanted by using the technique previously described by Sottnik et al.⁹ The mice were administered a subcutaneous dose of buprenorphine (0.05 mg/kg) for pain control during recovery from anesthesia.

2.9 |. In vivo luciferase imaging

The distal femora of all mice were imaged in the immediate postoperative period to confirm positive emission of luminescence from the implanted suture. The IVIS Lumina imaging system and the Living Image version 4.2 software with a 60-second exposure time as well as high sensitivity binning were used for in vivo imaging. Subsequently, the mice were anesthetized with isoflurane for imaging 1–2 times weekly to assess the presence of luminescence in the distal femora.

2.10 |. Femoral culture

When mice were killed, the left distal femur was aseptically harvested, placed in 25 mL of lysogeny broth (LB) broth, and cultured at 37 °C. When the LB exhibited gross evidence of organism growth, the broth was plated onto a blood agar plate. The resultant bacterial colonies were visually examined to determine whether they were pure cultures. The bacterial colonies were imaged by using the IVIS Lumina imaging system in Living Image version 4.2 software with a 5-second exposure time and medium sensitivity binning to confirm an osteomyelitis with luminescent XEN36 S aureus.

2.11 |. Histopathologic evaluation

When mice were killed, tumor samples and any suspected metastatic or grossly abnormal lesions in internal organs observed at necropsy were placed in 10% neutral buffered formalin (Fisher Scientific, Pittsburgh, Pennsylvania). After fixation and paraffin embedding, 5-μm histologic sections were made and stained with hematoxylin and eosin by using standard techniques. Sections from presumed metastatic lesions were reviewed by a board-certified veterinary pathologist (LBB) to confirm that the lesions were metastases from the primary injection site. The injection site masses were also assessed for the presence of black granular material (MCL), areas of necrosis, and local tissue reaction to heat treatment.

2.12 |. Flow cytometry

Surface receptor expression on monocytes were evaluated and quantified with flow cytometry. Blood was collected when mice were killed by using cardiac puncture. The blood was lysed with a standard whole blood lysis protocol, and leukocytes were stained the antibodies listed in Table 1. Normal donkey serum was added to the FACS buffer (phosphate buffered saline [PBS] with 2% fetal bovine serum [FBS]) for blocking against nonspecific staining. Monocytes were gated on the basis of forward and side scatter and positive expression of CD11b and Ly6C, with negative expression of Ly6G. Monocytes were further gated into subsets on the basis of intensity of Ly6C expression, namely classical (Ly6C^hi) and nonclassical (Ly6C^lo) monocytes. Samples were analyzed by using an LSRII flow cytometer (BD Biosciences, San Jose, California), and data analysis was performed in Kaluza version 1.5a (Beckman-Coulter, Indianapolis, Indiana).

TABLE 1.

List of antibodies used for murine flow cytometry

Antibodies	Clone	Fluorescence labeling	Manufacturer
CD11b	M1/70	BV421	BioLegenda
Ly6C	HK1.4	BV570	BioLegend
Ly6G	1A8	FITC	eBiosciencea
CCR2	475301	AF-700	R&Db
CXCR2	SA044G4	PE	BioLegend
CCR7	4B12	PE	eBioscience
CXCR4	2B11	AF700	BD Biosciencesc
CX3CR1	SA011F11	PE	Biolegend
CD62L	MEL-14	AF700	eBioscience

^aSan Diego, California.

^bMinneapolis, Minnesota.

^cSan Jose, California.

2.13 |. Chemotaxis

An in vitro transwell chemotaxis assay was used to assess monocyte migration. Single cell suspensions of the spleen were obtained at humane killing. Transwell inserts with a 3-μm membrane pore size (BD Biosciences, Bedford, Massachusetts) were loaded with 2 × 10⁶ splenic cells suspended in 300 μL of Roswell Park Memorial Institute medium + 10% fetal bovine serum. The lower chamber (ie, the 24-well cell culture plate) was loaded with the following: 100 ng/mL monocyte chemoattractant protein-1 (MCP-1; Peprotech, Rocky Hill, New Jersey), 100 ng/mL stromal cell-derived factor-1 (SDF-1; Peprotech) as positive chemoattractants, and phosphate-buffered saline + 1% bovine serum albumin (Sigma-Aldrich, St Louis, Missouri) as a negative control. The cells were incubated at 37 °C for 4 hours, after which the nonmigrated cells remaining on the top of the membrane were removed with cotton swabs. The transmigrated cells on the underside of the membrane were fixed with ice-cold methanol before staining with 1% crystal violet aqueous solution. The membranes were air dried overnight and then removed and mounted on glass microscope slides for counting. The membranes were counted by an independent clinical pathologist blinded to the study (KM). For each mouse, the ability of monocytes to move to the chemoattractants was measured in duplicate by using aggregate counts from 5 microscope fields.

2.14 |. Quantitative real-time polymerase chain reaction

Tumor samples were collected at humane killing, placed in RNALater (Qiagen, Germantown, Maryland), and stored at −80 °C for subsequent polymerase chain reaction (PCR) analysis. The Direct Zol RNA Miniprep kit (Zymo Research, Irvine, California) was used to extract total RNA from tumor samples, according to the manufacturer’s protocol. The concentration and quality of the extracted RNA was determined by using a NanoDrop 2000c spectrophotometer (Thermo Scientific, Wilmington, Delaware) as well as the Agilent Bioanalyzer (Agilent Technologies, Santa Clara, California). RNA samples were stored at −80 °C until processing. Complementary DNA (cDNA) was synthesized from 1000 ng RNA per sample by using the Quantabio qScript (Quantabio, Beverly, Massachusetts) cDNA synthesis kit, according to manufacturer’s directions. mRNA gene expression was determined with SYBR Green quantitative real-time PCR (qPCR) by using the QuantaBio PerfeCTa SYBR Green FastMix (Quantabio), per manufacturer’s directions, and reactions were cycled in a Roche Lightcycler 480 (Roche Diagnostics, Indianapolis, Indiana). All samples were run in triplicate. Primers for detection of mRNA expression levels of arginase, Fizz, IL-6, CD206, tumor necrosis factor (TNF)-α, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were selected on the basis of previous reports and purchased from Life Technologies (Gaithersburg, Maryland).²⁴ The qPCR cycling conditions were denaturation at 95 °C for 1 minute, followed by 95 °C for 5 seconds, 60 °C for 15 seconds, 72 °C for 10 seconds, for a total of 45 cycles. The PCR product was sequenced by GENEWIZ DNA sequencing services (South Plainfield, New Jersey), and the sequences were subjected to alignment searches in the National Center for Biotechnology Information database to confirm matches with the gene of interest. Expression of the target genes was normalized to expression of the housekeeping genes GAPDH. The 2^−ΔΔCT method was used to calculate the normalized relative mRNA expression of the target genes.

2.15 |. Statistical analysis

All statistical analyses were performed in SAS 9.4 (SAS, Cary, North Carolina) or Prism 7 (GraphPad Software, La Jolla, CA). For flow cytometric analyses, summary statistics were first calculated for all monocyte receptor values. Two-sample Wilcoxon rank-sum tests were used to test for differences among the nanoparticle groups when collected at humane killing compared with before humane killing. One-factor ANOVA was used to test for differences in the percentage of Ly6C^hi and Ly6C^lo monocytes among experimental groups as well as to test for differences between marker expressions within monocyte receptor subgroups for Ly6C^hi, Ly6C^lo, and CD11b^hi. When statistical significance was indicated, post hoc pairwise comparisons of means were performed. Pearson’s correlation coefficient was calculated to examine the relationship between Ly6C^hi and Ly6C^lo monocytes within each treatment group. Analysis of covariance (ANCOVA) was used to test for the relationship between Ly6C^hi and Ly6C^lo and its interaction with treatment groups. Differences between marker expressions within each experimental group and monocyte expression were tested by using a series of Wilcoxon signed-rank tests comparing these pairwise differences to zero. Differences in survival times among treatment groups were also tested by using 1-way ANOVA. To examine the relationship between monocyte markers (within Ly6C^hi and Ly6C^lo) and survival, the markers were included as a covariate in a series of 1-way ANCOVA. For each hypothesis of interest, the Holm-Bonferroni correction was applied to adjust for multiple testing and to control the type I error rate. For survival analysis, which was defined as the number of days from tumor injection to time of sacrifice when tumors reached 18 mm in diameter, the Kaplan-Meier log-rank analysis was used. For chemotaxis experiments, a linear mixed effects model (ie, a completely randomized, split-plot design with subsampling) was used to analyze the chemotaxis data. This model includes random effects for mouse nested within treatment and chemoattractant-by-mouse combination also nested within treatment because each such combination was subsampled in duplicate. The model also includes fixed effects for treatment and chemoattractant and their interaction. Residual diagnostics indicated that variability in monocyte movement increased with the mean, as is the case with counting variables. Therefore, the square root transformation was applied to the counts, resulting in better homogeneity of variance. The cutoff for statistical significance before adjusting was set at P < .05 throughout.

3 |. RESULTS

Mice were divided into 5 experimental groups, healthy mice (normal control), OS-bearing mice (tumor control), OS-bearing mice infected with XEN36 S aureus (osteomyelitis), OS-bearing mice treated with MCL-hyperthermia therapy (hyperthermia), and OS-bearing mice injected with MCL without subsequent hyperthermia therapy (vehicle control). Anesthesia and procedures were well tolerated by all mice, and, apart from 1 mouse with a thermal wound as a result of the MCL hyperthermia therapy, none of the mice exhibited unusual signs of pain postprocedure.

3.1 |. Induction of bacterial osteomyelitis

Bacterial osteomyelitis was successfully induced in mice, as indicated by positive luciferase imaging of the distal femur and/or positive bacterial cultures of the bacterial implantation site in the distal femur after mice were killed. All bacterial cultures were positive for luminescence, confirming the identity of XEN36 S aureus (Figure 1).

FIGURE 1.

Induction of bacterial osteomyelitis. Mice were inoculated with luciferase-positive XEN36 S aureus. A, Osteomyelitis was confirmed after femoral inoculation by using luminescence live imaging in anesthetized mice (arrow). B, After mice were humanely killed, the femur was aseptically harvested and cultured. Luciferase-positive bacterial cultures confirmed the presence of pure luminescent XEN36 S aureus

3.2 |. MCL-hyperthermia therapy

The complete course of MCL-hyperthermia therapy was administered to 6 of 9 total mice. Three mice did not receive all 3 hyperthermia treatments because of the absence of gross tumor after 1–2 treatments. These 3 mice exhibited wounds at the tumor site that were suggestive of thermal necrotic injury; 1 mouse was killed immediately upon development of the wound 1 day posthyperthermia treatment because of the severity of the injury, and the other 2 mice were killed 12 days after the last hyperthermia treatment to standardize the time of killing in treated mice without gross tumors.

3.3 |. Histopathologic analysis

The morphologic features of masses that appeared at the injection site of all mice were consistent with primary OS. The masses were composed of bundles, streams, and sheets of neoplastic mesenchymal cells that expanded the subcutis and infiltrated and replaced dermis and underlying musculature. In both primary and metastatic lesions, the neoplastic masses often had curvilinear areas of necrosis. Tumor morphology from all presumed metastatic lesions was identical to that observed in the primary injection site. The most frequent sites of metastasis in these mice were liver (17.9%), ovary (7.7%), and lung (7.7%).

To assess the depth and delivery of the MCL, primary tumors were further assessed for presence of a dark brown-black granular material. In the hyperthermia and vehicle control groups, the MCL were easily visualized in large lakes that were distributed unevenly throughout the masses (Figure 2A–C). In mice that were treated with hyperthermia therapy, including the 3 mice that had absence of gross tumor, tumor lysis and changes in the epidermis and subcutis consistent with thermal injury were observed.

FIGURE 2.

Distribution of nanoparticles in tumors arising at injection sites and tumor histopathology. Photomicrographs at × 20 of hematoxylin and eosin-stained sections of injection site tumors. A, Neoplastic cells only. The subcutis contains a large expansile neoplasm that extends into the subjacent musculature. The overlying skin is intact. B, Neoplastic cells and magnetic cationic liposomes (MCL). The neoplasm arising from the subcutis has large areas of pale pink necrosis (asterisks) and intralesional black nanoparticle material (arrow). C, Neoplastic cells, MCL, and hyperthermia therapy. There is extensive coagulation necrosis (homogenous red staining), which is consistent with thermal injury, that extends from the epidermis into the panniculus and affects approximately 75% of the neoplasm. Note the uneven distribution of black nanoparticle material and areas of necrosis. Scale bars = 500 μm

3.4 |. Monocyte receptor expression

On the basis of our previous finding that monocyte chemokine receptors were significantly downregulated in dogs with OS compared with normal controls, we compared the monocyte surface receptors among our experimental groups to determine whether a similar effect could be observed in mice and whether osteomyelitis or MCL-hyperthermia therapy affects receptor expression.²³ Monocytes were evaluated on the basis of Ly6C expression, with high Ly6C expression representing the inflammatory monocyte population and low or negative Ly6C expression representing the alternatively activated monocytes (Figure 3). Mice with osteomyelitis exhibited a significantly greater proportion of Ly6C^lo monocytes (median = 20%, min-max = 14%−38%) compared with normal control mice (median = 14%, min-max = 10%−19%, P = < .001), tumor control mice (median = 19%, min-max = 7%−34%, P = .0184), hyperthermia-treated mice (median = 11%, min-max = 5%−23%, P < .001), and vehicle control mice (median = 15%, min-max = 13%−24%, P = .0048; Table 2). Next, when surface receptor expression was evaluated for all monocytes, chemokine receptor CCR7 expression was significantly higher in monocytes of normal control mice (median = 37%, min-max = 4%−66%) compared with tumor control mice (median = 16%, min-max = 1%−37%, P = .0146), hyperthermia-treated mice (median = 3%, min-max = 2%−25%, P < .001), vehicle control mice (median = 13%, min-max = 6%−27%, P = .0282), and mice with osteomyelitis (median = 27%, min-max = 2%−98%, P = .0012; Table 3). When surface receptor expression was evaluated between the monocyte subsets, CCR7 expression was significantly higher in the Ly6C^lo monocyte subset of normal control mice (median = 41%, min-max = 6%−70%) compared with tumor control mice (median = 10%, min-max = 0%−29%, P < .001), hyperthermia-treated mice (median = 4%, min-max = 0%−29%, P < .001), vehicle control mice (median = 13%, minmax = 3%−22%, P = .0029), and mice with osteomyelitis (median = 32%, min-max = 0%−98%, P < .001; Table 3). Chemokine receptor CXCR4 expression was significantly higher in the Ly6C^lo monocyte subset of mice with osteomyelitis compared with normal control mice (P = .0082), tumor control mice (P = .0267), hyperthermia-treated mice (median = 23%, min-max = 15%−41%, P < .001), and vehicle control mice (median = 16%, min-max = 5%−32%, P = .0082; Table 2). In the Ly6C^hi monocyte subset, CCR2 expression was significantly higher in tumor control mice (median = 99%, min-max = 96%−100%) compared with hyperthermia-treated mice (median = 96%, min-max = 81%−100%, P = .001). Finally, in the Ly6C^hi monocyte subset, CXCR4 expression was significantly lower in tumor control mice (median = 5%, min-max = 2%−11%) compared with hyperthermia-treated mice (median = 8%, min-max = 7%−18%, P = .002).

FIGURE 3.

Monocyte identification and phenotypic analysis by flow cytometry. A, Flow cytometry gating illustrates typical forward light scatter (FSC) vs side light scatter (SSC; left panel) and lack of Ly6G expression (Ly6G⁻, circled, right panel). B, The Ly6G⁻ monocytes (from A, circled) were then gated on the basis of Ly6C and CD11b expression for identification of classical (CD11b⁺ Ly6C^hi, upper right) and nonclassical (CD11b⁺ Ly6C^lo, middle right) monocytes

TABLE 2.

Monocyte markers in mice with osteomyelitis (n = 13) compared with all other experimental groups^a

Monocyte markers	Osteomyelitis vs normal control (n = 7)	Osteomyelitis vs tumor control (n = 11)	Osteomyelitis vs vehicle control (n = 6)	Osteomyelitis vs hyperthermia (n = 9)
Ly6Clo subset	↑	↑	↑	↑
CXCR2 expression in Ly6Clo subset	↑	↑	↑	↑

^aArrows indicate an increase in mice with osteomyelitis compared with each treatment group (P < .05 for all arrows, see Results for specific P values)

TABLE 3.

Monocyte markers in normal control mice (n = 7) compared with all other experimental groups^a

Monocyte markers	Normal control vs osteomyelitis (n = 13)	Normal control vs tumor control (n = 11)	Normal control vs vehicle control (n = 6)	Normal control vs hyperthermia (n = 9)
CCR7 expression in all monocytes	↑	↑	↑	↑
CCR7 expression in Ly6Clo subset	↑	↑	↑	↑

^aArrows indicate an increase in normal control mice compared to each treatment group (P < .05 for all arrows, see Results for specific P values).

3.5 |. Monocyte chemotaxis

Because we had previously demonstrated decreased chemotaxis in monocytes from dogs with OS compared with monocytes from normal controls, we asked whether murine OS may have a similar effect on monocytes.²³ Our hypothesis was that osteomyelitis and/or MCL-hyperthermia therapy enhances monocyte chemotaxis in mice with OS. When MCP was used as a chemoattractant, monocytes from mice with osteomyelitis exhibited significantly decreased chemotaxis (median = 13, min-max = 0–63) compared with monocytes from tumor control mice (median = 50, min-max = 20–100, P = .0043) and hyperthermia-treated mice (mean = 37, min-max = 1–140, P = .0099; Figure 4A). With SDF as a chemoattractant, we observed that monocytes from hyperthermia-treated mice exhibited greater chemotaxis (median = 51, min-max = 19–162) compared with monocytes from mice with osteomyelitis (median = 25, min-max = 1–61, P < .001; Figure 4B).

FIGURE 4.

Comparison of monocyte chemotaxis among treatment groups. A transwell chemotaxis assay was used to assess monocyte migration. Visualization of monocytes on the transwell membrane (A, × 100 objective). Box and whisker plots compare the numbers of migrated monocytes counted on the transwell membrane in response to the chemoattractants MCP (B) and SDF (C). The horizontal line represents the median, the box represents the first and third quartiles, and the whiskers represent the minimum and maximum values. *P < .05. MCP, monocyte chemoattractant protein; SDF, stromal cell-derived factor

3.6 |. Relative tumor mRNA expression

Interleukin-6 and TNF-α are markers of inflammation, and, together with arginase, Fizz, and CD206, can be used to differentiate macrophage subsets in mice.²⁴ To investigate the effects of osteomyelitis and MCL-hyperthermia therapy on the expression of the above markers within tumor tissue, we compared the relative mRNA expression of IL-6, TNF-α, arginase, Fizz, and CD206 in tumor tissue from mice with osteomyelitis, MCL-hyperthermia mice, and vehicle control mice to tumor control mice. There were no significant differences in relative mRNA expression of all markers of interest in tumor tissue of experimental groups compared with tumor control tissue. However, there was a trend toward decreased CD206 expression in tumors from hyperthermia mice compared with tumor control mice (P = .0513)

3.7 |. Survival analysis

To evaluate the effects of MCL-hyperthermia therapy and of osteomyelitis on survival in OS, we performed a Kaplan-Meier survival analysis comparing survival between MCL-hyperthermia-treated mice, mice with osteomyelitis, tumor control mice, and vehicle control mice. Median survival for hyperthermia-treated mice was 28 days, for mice with osteomyelitis it was 24 days, for vehicle control mice it was 26 days, and for tumor control mice it was 25 days. There was no significant difference in median survival between experimental groups (P = .53).

4 |. DISCUSSION

The goal of this study was to evaluate the efficacy of MCL-hyperthermia therapy with a murine model of OS in comparison to an established osteomyelitis model and to characterize further the role of monocytes/macrophages in driving an antitumor response. The efficacy of MCL-hyperthermia therapy was previously demonstrated in a hamster model of OS in which all tumors in the nanoparticle treatment group reportedly resolved grossly by day 15 posttreatment, and no local recurrence of tumor was observed for 3 months.²² In contrast, we did not demonstrate a survival benefit afforded by either MCL-hyperthermia therapy or bacterial osteomyelitis. The MCL-hyperthermia mice exhibited signs of thermal injury at their treatment site as evidenced by the gross presence of an eschar and histologic presence of coagulative necrosis. One potential explanation for this observation is the administration of a single large volume (300 μL) of MCL into each tumor that prevented the MCL from distributing in a more diffuse pattern through the tumor (Figure 2). This single volume administration was based upon a pilot study in mice with mammary carcinoma (C. Petty, personal communication, June 2015) and a previous rodent study.¹⁹ Such an accumulation of all injected MCL within 1 region of the tumor may have resulted in excessive focal heat production during AMF application. We are currently investigating the effects of dividing up the MCL dose into 3 smaller volumes injected into 3 tangential sites within the tumor in an effort to achieve a more diffuse distribution of MCL throughout the tumor. We did not demonstrate a survival benefit in mice with OS and bacterial osteomyelitis even though all the post mortem femoral cultures were positive for XEN36 S aureus. However, greater than 50% of the mice stopped exhibiting in vivo luminescence in the last 5 days prior to killing, suggesting a relatively low bacterial burden. Sottnik et al⁹ reported that bacterial osteomyelitis inhibited OS growth and increased survival in mice with OS and that the osteomyelitis was durable, with mice exhibiting in vivo bacterial luminescence more than 50 days postinfection. Our findings suggest that infections were present in the mice but were not robust enough to incite a strong inflammatory response.

Our flow cytometric analysis revealed a higher proportion of Ly6C^lo peripheral monocytes in mice with osteomyelitis compared with all other experimental groups, which is an unexpected finding. Monocytes are a heterogeneous population of mononuclear leukocytes primarily found in the peripheral circulation, making up approximately 2%−4% of murine leukocytes.²⁵ Murine monocytes are divided into inflammatory and alternative subpopulations, with the inflammatory monocytes exhibiting high Ly6C (Ly6C^hi) expression and alternative monocytes exhibiting low or negative Ly6C (Ly6C^lo) expression. Ly6C^hi inflammatory monocytes are recruited to inflammatory foci, including sites of bacterial infection where they then carry out defensive microbicidal functions, whereas Ly6C^lo monocytes typically mediate tissue repair.²⁶ Thus, in our study, we expected the mice with osteomyelitis to exhibit higher proportions of Ly6C^hi inflammatory monocytes, which was previously reported.⁹ The predominance of Ly6C^lo monocytes in the mice with osteomyelitis (Table 2) along with the loss of in vivo bacterial luminescence in this study suggests that these mice were likely resolving their infections. The increased expression of the chemokine receptor CXCR2 on the Ly6C^lo monocytes of mice with osteomyelitis (Table 2) suggests a mediation of chemotaxis as well as integrin-mediated arrest of these monocytes at sites of inflammation.²⁷ The higher CCR7 expression observed in monocytes from normal control mice (Table 3) parallels our previous finding of increased chemokine receptor expression, including CCR7, in monocytes from normal control dogs compared with dogs with untreated OS.²³ However, we did not observe a similar decrease in monocyte chemotaxis from tumor control compared with normal control mice in this study. We also observed higher CCR2 and lower CXCR4 expression in the Ly6C^hi monocytes of tumor control and hyperthermia mice compared with their Ly6C^lo monocytes. In tumor-bearing animals, the CCR2-CCL2 axis is important for Ly6C^hi monocyte recruitment to the tumor; Ly6C^hi monocytes are generally considered as precursors of TAM.²⁸ The higher CCR2 expression in the Ly6C^hi monocytes of tumor control mice and mice with hyperthermia may represent tumor-driven upregulation of CCR2 expression to facilitate Ly6C^hi monocyte migration toward the tumor. CXCR4 overexpression by tumors has been extensively studied, with CXCR4 implicated in the promotion of angiogenesis and metastasis.²⁹ Therefore, CXCR4 expression on the Ly6C^hi monocytes of tumor control and mice with hyperthermia could represent another mechanism by which the tumor recruits monocytes.

The hyperthermia-treated mice exhibited heightened monocyte chemotaxis compared with mice with osteomyelitis. This was due to a relative increase in chemotaxis in hyperthermia-treated mice and a relative decrease in chemotaxis in mice with osteomyelitis, in comparison to controls (Figure 4). The decreased monocyte chemotaxis exhibited by mice with osteomyelitis compared with hyperthermia-treated and tumor control mice suggests a low intensity of infection in these mice, resulting in decreased monocyte chemotaxis. The reduced monocyte migration in mice with osteomyelitis could alternatively be a result of an immune escape mechanism induced by S aureus, which has been reported in some cases to inhibit leukocyte migration toward sites of infection.³⁰

We did not find significant differences in relative mRNA expression of IL-6, TNF-α, arginase, Fizz, or CD206 in the tumor tissue from hyperthermia, osteomyelitis, and vehicle control mice compared with tumor tissue from tumor control mice. CD206, or mannose receptor, is a pattern recognition receptor expressed on macrophages, and has been used to differentiate between macrophage subsets, with M2 macrophages exhibiting higher expression of CD206.²⁴ There was a trend toward decreased CD206 expression in tumors from hyperthermia mice compared with tumor control mice (P = .0513), suggesting a shift away from M2 polarization. Additional studies with an optimized MCL-hyperthermia therapy protocol are required to evaluate the potential M1 macrophage polarizing effects.

This study had limitations. One potential issue was the use of small batch-manufactured MCL in our laboratory, so there were likely inconsistencies in the performance of the MCL. The large volume of MCL injected into a single site may have contributed to the thermal damage evident at the treatment sites. Current optimization studies using commercial pharmaceutical grade MCL and injection of smaller volumes of MCL into multiple sites within each tumor are yielding promising results, with more uniform tumor heating and the absence of gross thermal injury (personal communication, C. Petty, March 2018). Another limitation was the low intensity of osteomyelitis present in the mice. Even though all femora in the osteomyelitis group cultured positive for XEN36 S aureus at time mice were killed, greater than 50% of the mice stopped exhibiting in vivo luminescence in the last 5 days prior to being killed. This suggests that the intensity of the infection may not have been robust enough to elicit a strong inflammatory reaction against the tumor.

Even though in the current study we did not identify a survival benefit conferred by either MCL-hyperthermia therapy or osteomyelitis, we have identified features of the MCL-hyperthermia therapy that we can optimize for follow-up studies. We observed higher CCR7 monocyte expression in normal control mice compared with other experimental groups, similarly to our previous observations in dogs with OS and normal control dogs. Furthermore, we also identified monocyte surface expression of CXCR4 as a potential mechanism by which tumors recruit monocytes. The most important finding of our study was identification of heightened monocyte chemotaxis in hyperthermia-treated mice in relation to mice with osteomyelitis and a trend toward decreased M2 polarization in tumor tissue from hyperthermia-treated mice.