Proteoglycans as Target for an Innovative Therapeutic Approach in Chondrosarcoma: Preclinical Proof of Concept

This article is featured in Highlights of This Issue, p. 2551

One of the key problems in chondrosarcoma, the second most common type of skeletal malignancy after osteosarcoma, is the lack of response to both chemotherapy and radiotherapy, giving rise to poor patient outcomes and leaving surgery as the only effective curative treatment. The type of required surgical procedure may vary according to the grade, the initial location, and the extent of the disease. Large en bloc dramatic resection or amputation, often necessary to allow the local control of the tumor and to avoid metastases, are at the origin of strong disability and morbidity (1). Nevertheless, wide resection is not always possible in large tumors and critical anatomic locations such as pelvis, axial skeleton, and also in many cases of local recurrence. The clinical behavior and the prognosis for patients with chondrosarcoma vary widely with a 10-year survival rate ranging from 29% to 83% depending on the tumor grades (2, 3).

For chondrosarcoma, the current clinical challenge is to prevent recurrences and to find better treatment options for patients with inoperable primary or recurrent disease, and metastases. Published data regarding the sensitivity of chondrosarcoma to cytotoxic agents are scarce in both preclinical and clinical reports. A recent study has shown that some chondrosarcoma cell lines may display moderate sensitivity to doxorubicin and cisplatin, the most commonly used chemotherapeutic agents in sarcomas (4). In 2012, in a large retrospective and multicenter series, chemotherapy using anthracycline was associated with a median progression-free survival of 4.7 months in the first-line (2, 3). The study of Van Maldegen and colleagues also demonstrated that the survival rate in patients with unresectable chondrosarcoma was also significantly higher in the case of systemic treatments (doxorubicin-based chemotherapy, and nonchemotherapy-based agents such as imatinib and sirolimus) compared with patients managed with best supportive care only (overall survival at 3 years: 26% vs. 8% respectively, P < 0.05; ref. 5). According to such poor outcome and no global improvement in the survival of patients with chondrosarcoma over the past 30 years, personalized medicine including a combination of targeted therapies will most likely be the best approach to tackle the resistance of human chondrosarcoma.

Treatments targeting tumor microenvironment have therefore been suggested as complements to surgical excision (6, 7). Nevertheless, due to the large amount of extracellular matrix (ECM) and the poor vascularity of chondrosarcoma, the anticancer agents have to diffuse over a relatively long distance to reach tumor cells and be effective (5).

The best approach for a personalized medicine of chondrosarcoma might be to take advantage of such phenotypic microenvironmental features, by developing an attractive approach of targeting the proteoglycan (PG)-rich cartilaginous tissues using a quaternary ammonium (QA) function as a carrier for drugs or radionuclides. The positively charged QA interacts with the sulfate and carboxylate groups present in PGs highly expressed in the chondrogenic ECM (8). With this strategy, we successfully delivered imaging agents as well as therapeutic drugs (such as anti-inflammatory drugs, metalloproteinase inhibitors, cytotoxics, and more recently radiosensitizing nanoparticles; refs. 8–13) selectively to PG-rich tissues such as cartilage and chondrosarcoma. In particular, the suitability and high sensitivity of the radiotracer N-[triethylammonium]-3-propyl-1,4,7,10,13-pentaazacyclopentadecane radiolabeled with ^99mTc (^99mTc-NTP 15-5) was validated for in vivo diagnosis and follow up of chondrosarcoma via scintigraphic imaging (14–16). This radiotracer is currently transferred into clinic for a first into human study. In parallel, we developed a therapeutic approach with a QA derivative of melphalan (Mel-QA), an alkylating agent, that showed promising results from in vitro and in vivo preclinical experiments (13). The aim of the current study was to characterize this therapeutic strategy.

First of all, the binding ability of Mel-QA to aggrecan (major PG in chondrosarcoma) was demonstrated by surface plasmon resonance (SPR) respectively to its nontargeted equivalent (Mel). Quantitative whole-body autoradiography showed rapid accumulation of radioactivity in cartilaginous tissue after injection of the tritium-radiolabeled Mel-QA, as compared with Mel. The main interest of the QA vector was to strongly attenuate side effects, monitored by hematologic analyses and body weight, while maintaining a tumor growth inhibition assessed by in vivo^99mTc-NTP 15-5 scintigraphic imaging, ¹H-HRMAS NMR spectroscopy, and histologic analyses of biopsies.

The work reported here clearly validates the relevance of PGs as target for an innovative therapeutic approach in chondrosarcoma.

Mel was used as its hydrochloride form after treatment of Mel (Interchim) with a solution of 2 N hydrochloric acid in diethyl ether. The QA conjugate of Mel (Mel-QA) was prepared as already published (13, 17) and fully characterized (Supplementary Fig. S1).

The NTP 15-5 and its nontargeted equivalent (i.e., 15-5) were synthesized and radiolabeled with ^99mTc as described previously (18).

To demonstrate the binding affinity between the QA entity and aggrecan, we used the SPR method with aggrecan immobilized onto the surface of the sensor chip. When studied compounds interact with aggrecan, they form a complex that modifies the refractive index. This change is measured in real-time and the signal obtained is plotted in resonance unit (RU) versus time (19, 20). SPR assays were carried out on a Biacore T200 instrument with CM4 sensor chips (GE Healthcare). Sensor chip CM4 has a dextran matrix with a low degree of carboxylation which may be of value for reducing nonspecific binding with highly positive charged analytes and may be useful for kinetic analysis. To immobilize the ligand, sensor chip was firstly activated using a 1:1, v:v, mixture of a 0.2 mol/L aqueous 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride salt solution and a 0.5 mol/L aqueous N-hydroxysuccinimide (Amine Coupling Kit, GE Healthcare) solution at a flow rate of 5 μL/minute for 10 minutes. After that, aggrecan from bovine articular cartilage (Sigma Aldrich) was coated to the sensor chips at 400 μg/mL in running buffer HBS-P⁺ 1× (0.01 mol/L HEPES, 0.15 mol/L NaCl, 0.05%, v/v, surfactant p20, pH 7.4) containing 6 mmol/L of hexadecyltrimethylammoniumbromide (CTAB) with a flow rate of 5 μL/minute to a level of approximately 500 RU. Unoccupied binding sites were blocked using an aqueous 1 mol/L ethanolamine solution (pH 8.5; Amine Coupling Kit, GE Healthcare) with a flow rate of 5 μL/minute for 10 minutes. Mel and Mel-QA were injected with a flow rate of 30 μL/minute in increasing concentrations (0–2 mmol/L) using five 600-second injections and a 200-second dissociation time. After each experiment, flow cells were regenerated twice with NaCl 2 mol/L for 150 seconds. The interactions were recorded as the difference in response between the immobilized aggrecan flow cell and a corresponding control flow cell (activated and blocked, but without immobilized aggrecan). Kinetic and statistic parameters (K_d and χ²) were determined by Biacore T200 Evaluation software by fitting the steady-state values at equilibrium (Req). Experiments were performed in triplicate.

The Swarm rat chondrosarcoma (SRC) line (Dr. Patrick A. Guerne, Geneva, Switzerland) was delivered as tissue fragments which were frozen until use. Protocols were led under the authorization of the French Directorate of Veterinary Services and in accordance with the 2010/63/UE Directive and were conducted as described previously (13) using male Sprague–Dawley rats (Charles River Laboratories). All experiments were conducted after ethics committee approval (C2E2A, n°CE 82-12). Allograft transplantation of a 10-mm³ SRC fragment was performed on the right paw, the other paw being used as the contralateral control. These SRC fragments were collected from well-developed tumors on the paratibial area of donor Sprague–Dawley rats.

An excess of N-t-Boc-4-amino-L-phenylalanine ethyl ester 1 was alkylated by reaction with [³H]-2-bromoethanol ([³H]-3), previously obtained in situ from ethyl bromoacetate 2 (2 mmol) and [³H]-NaBH₄ (3.7 GBq, diluted with 2.2 mmol of unlabeled NaBH₄). Then the crude reaction mixture containing amine 1 and N-t-Boc-4-(2-hydroxy-2-[³H]ethyl)amino-L-phenylalanine ethyl ester ([³H]-4) reacted with ethylene oxide (approximately 3 mL) in acetic acid to provide the corresponding 4-[bis(2-hydroxy-2-[³H]ethyl)]amino-phenyl derivative ([³H]-5) in 50% radiochemical yield (2.15 GBq, 2 mmol). Appel chlorination with triphenylphosphine (5.9 mmol) and carbon tetrachloride (2 mL) in dichloromethane (20 mL) gave, after silica gel chromatographic purification, the corresponding tritiated dichloro derivative [³H]-6 (0.96 mmol, 962 MBq) in 48% chemical yield. Then, the ester and the t-Boc protecting group of compound [³H]-6 were hydrolyzed with an aqueous 6 N hydrochloric acid solution (5 mL) to provide the hydrochloride salt of tritiated Mel ([³H]-Mel, 0.96 mmol, 740 MBq) in quantitative chemical yield. On the basis of this radiolabeling sequence, [³H]-Mel was isolated with a specific activity of 777 MBq/mmol and a radiochemical purity > 98%, as determined by HPLC analyses (Supplementary Fig. S2). The overall yield of this radiochemical procedure was 20%.

For the preparation of tritiated Mel-QA, a five-step radiosynthesis starting from hydrochloride salt of [³H]-Mel was developed. Thus, after Boc protection of the amino group of [³H]-Mel (0.69 mmol, 573 MBq), N-protected Mel was immediately coupled with 3-(dimethylamino)propylamine (0.83 mmol) in the presence of dicyclohexylcarbodiimide (DCC, 1.03 mmol) and 1-hydroxybenzotriazole (HOBt, 0.74 mmol) to give the corresponding amide intermediate [³H]-7 in 75% chemical yield after purification by column chromatography. Treatment with methyl iodide (2.3 mmol) in ethanol gave the QA compound [³H]-8 in quantitative yield. Finally, the t-Boc–protecting group was removed using a 2 N ethanolic hydrochloric acid solution (12 mL). A chromatography over a chloride ion-exchange resin (Dowex 1 × 8–200 mesh, Aldrich) provided the desired QA derivative of Mel ([³H]-Mel-QA, 0.46 mmol, 407 MBq) in 75% overall chemical yield from [³H]-Mel.

Finally, [³H]-Mel-QA was isolated with a specific activity of 814 MBq/mmol and a radiochemical purity > 97%, as determined by HPLC analyses (Supplementary Fig. S2). The overall yield of this radiosynthesis was 15% based on [³H]-NaBH₄.

This study was performed on thirty SRC-bearing rats at stage day 20 postimplantation. Animals were randomized into two groups (n = 15 per group) and received intravenous injection of 2.5 MBq (in 200 μL of physiologic saline) of [³H]-Mel or [³H]-Mel-QA.

At several time points after injection (5 and 15 minutes, then 1, 6, and 24 hours), intracardiac blood collection were performed (3 rats from each group per time point) under anesthesia by inhalation of isoflurane (CSP) in air (1.5%, 1 L/minute). Animals were then immediately sacrificed by carbon dioxide inhalation, frozen in liquid nitrogen, and embedded in blocks of carboxymethyl cellulose 2%, that were sagittally sectioned into 40-μm slices at −22°C with a cryomicro-tome (Reichert-Jung Cryopolycut). Each slice was dehydrated for 48 hours in a cryochamber. Whole-body slices (25 slices per animal and per time point) were then exposed for 1,000 minutes in a digital autoradiographic analyzer (βImager, Biospace Measures). Surfacic activity (counts in cpm/mm²) was quantified in regions of interest (ROI) delineated over organs of interest including tumors, corrected for radioactive decay, and expressed as % of injected dose (ID)/g of tissue, after calibration with tritium microscale standards [Autoradiographic [³H] Microscale, Amersham Biosciences]. Tumor-to-muscle ratio was calculated as follows: T/M = (counts per minute/mm² in tumor)/(counts per minute/mm² in muscle).

In a parallel experiment, groups of 6 rats received [³H]-Mel or [³H]-Mel-QA, and were housed in metabolic cages for urine collection from 6 to 24 hours after injection.

Radioactivity of urine or blood samples was directly measured after addition of Packard Ultima Gold cocktail (PerkinElmer) in a Wallac Winspectral 1414 Liquid Scintillation Spectrometer (PerkinElmer).

Experiments were conducted on SRC-bearing rats being randomized into three groups (control, Mel-QA, and Mel), with a number of 6–8 animals/group. Treated rats received three doses of Mel or Mel-QA (16 μmol/kg per injection) by intravenous route at 4-day intervals (q4d × 3 schedule), beginning on day 8 postimplantation. Controls received the excipient by intravenous route according to the same schedule. Animal weight and tumor volume were recorded twice a week. The tumor volume (TV) of each tumor was estimated using the formula: TV (mm³) = (L × W²)/2 where L is the length in mm, and W the width in mm.

At the end of the study, tumors were removed, fixed in 10% buffered formalin, and then embedded in paraffin. Slices (5 μm) were then stained with hematoxylin–eosin.

Antitumor efficacy was monitored by scintigraphic imaging, using the radiotracer ^99m Tc-NTP 15-5 developed by our group at day 7 (one day before treatment) and day 30 (2 weeks after treatment; ref. 15). Acquisitions were performed 30 minutes after intravenous tracer administration (25 MBq/animal), with a 10-minute planar acquisition for each posterior paw positioned over the parallel-hole collimator of a small-animal γ camera (GammaImager, Biospace) and with a 15% window centered on the 140-keV photopeak of ^99mTc. All the scans were evaluated by the same experienced investigator, using fixed-size ROIs delineated over tumor and muscle patterns. For each ROI, total activity, count in cpm (counts per minutes) per mm² were obtained. Tumor-to-muscle ratio was calculated: T/M = (counts per minute/mm² in tumor)/(counts per minute/mm² in muscle).

To assess the in vivo specificity of ^99mTc-NTP 15-5 imaging, additional SRC animals (n = 10) were injected with ^99mTc-NTP 15-5 or its nontargeted equivalent (^99mTc-15-5) at day 20 postimplantation when tumors exhibited a mean volume of 949 ± 223 mm³. Dynamic planar imaging (DPI) was performed to determine the kinetics of the tracer distribution in tumor. SRC animals were intravenously injected via the tail vein with 25 MBq of ^99m Tc-NTP 15-5 or ^99mTc-15-5 simultaneously to the starting of a 120-minute duration list mode acquisition using a 5-minute sampling time. Time–activity curves were obtained from ROIs drawn around tumor and muscle with all measured activities corrected for radioactive decay and T/M ratio calculated at each time point.

¹H HRMAS spectroscopy profile analyses were performed in a parallel study following the same protocol with additional animals (4 animals per group). Tumors were excised one week after the final injection, rapidly frozen in liquid nitrogen and stored at −80°C until analysis.

All experiments were performed on frozen tumor samples flushed for a few seconds in cooled D₂O. ¹H NMR spectroscopy was performed on a small-bore Bruker DRX 500 magnet (Bruker) equipped with an HRMAS probe. Samples were then set into 4-mm diameter, 50-μL free volume ZrO₂ rotor tubes without upper spacer. A total of 3 μL of D₂O containing 0.75% 3-(trimethylsilyl)propionic-2,2,3,3-d₄ acid (TSP; Sigma), used for peak referencing, was added to the rotor tubes to lock the spectrometer. Rotors were spun at 4 kHz to keep the rotation sidebands out of the acquisition window. A Bruker cooling unit was used to maintain the sample temperature at 277 K to minimize tissue degradation. One-dimensional ¹H NMR sequence was a saturation recovery sequence. The resonances were assigned on the basis of the known chemical shifts of the major structural groups using tissue data (21). The relative content of metabolites was estimated by peak area integration. To compare the metabolic profiles between experimental groups, each peak was normalized with the respect of the total integral of the spectrum in the 0.5 to 4.7 ppm region (21–23). Sample stability inside the rotor was assessed at a temperature of 277 K over several hours by a series of analyses. During the 48-minute overall acquisition period, no significant changes in NMR signals of biological material were observed. Each experiment was performed in triplicate.

Side effects of treatments with Mel or Mel-QA were assessed by both animal's weight and hematologic parameter measurements.

• (i) Weight variation was calculated according to the formula:

  % = [(weight at day x − weight at day 4)/weight at day 4) × 100.

• (ii) Hematologic parameters were assessed at day 18, that is, two days after the end of treatments. The blood (50 μL by retro-orbital puncture) of each animal was removed and processed on a Melet Schloesing MS9-5 Hematology Analyzer (Diamond Diagnostics).

Statistical analyses were conducted using the GraphPad Prism 5 software. Data are reported as mean ± SD. Results were analyzed by ANOVA followed by Tukey post test. We considered P values < 0.05 to indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Studies were performed for Mel-QA, respectively, to Mel injected in increasing concentrations (0–2 mmol/L). The sensorgrams obtained after signal subtraction of control was displayed versus time. As illustrated in Fig. 1 and regards to K_d values, Mel-QA exhibited affinity to aggrecan (K_d = 2.25 ± 0.92 mmol/L) in contrast to Mel, demonstrating that QA vector allows aggrecan binding.

The radiosyntheses of [³H]-Mel and [³H]-Mel-QA are shown in Fig. 2A. Both Mel-QA and Mel were radioisotopically labeled on the alkylating moiety, also ³H radionuclide was introduced in the chloroethyl groups. [³H]-Mel was synthesized from N-t-Boc-4-amino-L-phenylalanine ethyl ester according to a slightly modified sequence previously reported in the literature (24). For the preparation of [³H]-labeled Mel-QA, the radiosynthetic sequence starting from [³H]-Mel was adapted from the five-step synthesis developed for the corresponding nonradioactive derivative and ¹⁴C-labeled Mel-QA (17, 25). Both [³H]-Mel and [³H]-Mel-QA were successfully obtained with a specific activity of 814 MBq/mmol and a radiochemical purity > 97 %, as determined by radio-RP-HPLC analyses.

The distribution of radioactivity in several tissues after intravenous injection of 2.5 MBq of [³H]-Mel or [³H]-Mel-QA in SRC animals is expressed as the percentage of injected dose per gram of tissue (% ID/g; Table 1). After injection of [³H]-Mel-QA, radioactivity rapidly accumulates in PG-rich tissues such as cartilages and tumor (Fig. 2C). Five minutes postinjection (p.i.) of [³H]-Mel-QA, the T/M value was significantly higher (P < 0.05) compared with [³H]-Mel (T/M values 5 minutes postinjection: 5.77 ± 1.11 for [³H]-Mel-QA versus 2.22 ± 1.20 for [³H]-Mel; Fig. 2B).

Furthermore, high levels of each drug were found in the metabolizing and elimination organs, such as liver and kidneys. As expected for small hydrophilic compounds, Mel and Mel-QA are cleared in the kidney (i.e., about 16% of cumulative renal excretion) and no significant difference was observed between the two compounds at 24 hours postinjetcion (Table 1). However, radioactivity was observed in the liver after [³H]-Mel-QA injection as compared with [³H]-Mel.

Finally, radioactivity in the blood was higher after injection of [³H]-Mel than [³H]-Mel-QA (Table 1).

Mel-QA and Mel were given at 16 μmol/kg according to a q4d × 3 schedule beginning on day 8 after tumor implantation. The control group received intravenous saline injection.

On the basis of the tumor volume monitoring, a significant inhibition of tumor growth (P < 0.01) was observed for both Mel-QA- and Mel-treated rats compared with the control group, but no significant difference was observed between the two treated groups (Fig. 3A).

In vivo antitumor effects were also assessed by ^99mTc-NTP 15-5 scintigraphic imaging, as this radiotracer was previously demonstrated to provide suitable criteria for monitoring chondrosarcoma (14–16). As illustrated in Fig. 3B, quantitative analysis of in vivo ^99mTc-NTP 15–5 imaging of tumors evidenced significant changes in radiotracer uptake in treated animals versus controls. A significant increase of 84% of T/M ratio (P < 0.05) was observed in nontreated rats between day 7 (1 day before treatment) and day 30 (14 days after the last drug injection), whereas this ratio remained overall constant in Mel- or Mel-QA–treated rats. Furthermore, to assess the in vivo specificity of ^99mTc-NTP 15-5 imaging and to exclude the possibility of unspecific uptake in tumoral tissue, ^99mTc-NTP 15-5 and its nontargeted equivalent (^99mTc-15-5) were injected by intravenous route in additional SRC-bearing rats (n = 10). Time–activity curves are shown in Fig. 3C in term of T/M ratio. From 5 minutes postinjection of ^99mTc-NTP 15-5, T/M ratio gradually increased and reached a plateau as soon as 30 minutes which was maintained at least for 1 hour (T/M values 1 hour postinjection: 2.25 ± 0.11), allowing tumor visualization with an excellent contrast (Fig. 3D). On the contrary, T/M ratio for ^99mTc-15-5 was around 1 throughout the course of the imaging. According to these results, we demonstrated the potential of ^99m Tc-NTP 15-5 imaging as a “methodology companion” of targeted therapies of chondrosarcoma.

To characterize the antiproliferative effects, tumors were removed one day after the end of the treatments and anatomopathologic analyses were performed. For both treated groups, histopathologic analyses showed signs of tumor regression, such as necrotic cells, fibroinflammatory component, and increased vascularization. These signs were observed for 57% and 40% of tumors in the Mel-QA and Mel groups, respectively (Fig. 4A). Furthemore, mitotic activity index values showed a significant decrease (P < 0.01) in the proliferating activity in the two treated groups (Fig. 4B). In contrast, the tumors of nontreated rats were characterized by hypercellularity with a high mitotic activity index.

Finally, in a parallel study, SRC-bearing rats were treated following the same protocol and tumors were excised one week after the final injection for ¹H-HRMAS NMR spectroscopy analyses (Fig. 4C) which confirmed necrosis and apoptosis. Results revealed a significant increase (P < 0.01) in the spectral intensity ratio of methylene (CH₂) resonance (at 1.28 ppm) to methyl (CH₃) resonance (at 0.9 ppm) in the Mel-QA–treated group as compared with controls (Fig. 4D). This ratio, which is considered as an apoptosis marker (26, 27), was also significantly higher (P < 0.01) in the Mel-QA–treated group than in the Mel-treated group (Mel-QA: 2.34 ± 0.22 versus Mel: 1.47 ± 0.08). In the same way, a significant increase in CH₂(lip)/total creatine(tCr) ratio, which is considered as a necrosis marker (28), was observed in the Mel-QA group (Mel-QA: 11.5 ± 2.07 vs. Mel: 5.0 ± 1.26).

The body weight of each rat was regularly recorded and hematologic parameters were assessed 48 hours after the last dose of both compounds (Fig. 5). Mel-treated rats exhibited significant weight loss starting from the second injection which reached 15% at day 23 postimplantation (Fig. 5A). Furthermore, the clinical score including diarrhea, rough coat, closed eyes, lethargy, and bleeding were determined for each animal throughout the time course of the study: we assigned a score of 1 (presence) or 0 (absence) for each of these criteria. One day after the last dose of treatment, no clinical side effects were observed for Mel-QA–treated rats and controls, in contrast to the Mel-treated rats with an average clinical score of 2.7 ± 2.1. For Mel-treated rats, these symptoms were associated with a severe leucopenia, a common side effect of alkylating agents, which was marked by a significant decrease in white blood cells and lymphocytes numbers (Fig. 5B). Furthermore, in comparison with Mel-treated animals, an improvement in hematologic parameters was observed for Mel-QA treatment, that is, attenuation of leucopenia with, more specifically, a significant attenuation of lymphopenia (P < 0.01).

Chondrosarcoma, with an estimated incidence of 0.2 in 100,000 patients per year, is a rare disease with most patients being cured by a surgical excision. However, in case of unresectable locally advanced, inoperable, or metastatic disease, chemotherapy and radiotherapy appear largely ineffective. The poor prognostic and lack of effective treatments markedly highlight a pressing need to develop new therapeutic approaches (1, 5).

As efficacy of antineoplastic treatments is mainly hampered by toxic side effects, improving their therapeutic index, by reducing adverse effects and increasing drug accumulation in tumor tissue, remains a challenge in drug development. Among the numerous strategies aimed at such objectives, “carrier” with preferential binding to one or more tumor targets may offer promising therapeutic opportunities. This was the strategy employed here for experimental chondrosarcoma: a QA function (i.e., a carrier exhibiting affinity for the high negative fixed charged density of PGs) was conjugated to Mel. In this way, we expected to improve its therapeutic index by reducing systemic toxicity while keeping an antitumor activity.

First of all, the role of QA function for PG binding was demonstrated in vitro by SPR with aggrecan, the major PG component of chondrosarcoma. As expected, binding of the QA-functionalized derivative of Mel to aggrecan was characterized with a K_d value in the millimolar range (2.25 ± 0.92 mmol/L) which is not observed for Mel. These results confirmed our hypothesis that the Mel–QA conjugate establishes ionic interactions with the high negative fixed-charge density of aggrecan. The next step was then to confirm in vivo the interaction of both Mel-QA and Mel with the ECM of chondrosarcoma by a biodistribution study using the SRC model. This preclinical model reproduces the histologic and clinical behaviors of the human grade II disease (29–32). For biodistribution study, ³H radionuclide was introduced with excellent radiochemical purities (>97%) on the common alkylating group to allow comparative biodistribution of both compounds. Quantitative whole-body autoradiography showed rapid distribution of radioactivity after [³H]-Mel-QA intravenous administration: radioactivity rapidly accumulated in healthy cartilaginous tissues and tumor. A high accumulation of radioactivity was also observed in the kidney and liver, as previously reported for Mel-QA and Mel in healthy rats (25). This accumulation of radioactivity in liver could be ascribed to metabolism but specific PG binding could not be excluded. As a parenchymal organ, liver contains mainly heparan sulfate-GAGs with additional minor amounts of dermatan sulfate-GAGs, hyaluronic acid (HA), and chondroitin sulfate-GAGs which are barely detectable (33, 34). According to us, as autoradiography showed radioactivity of all radiolabeled derivatives (unchanged compound as well as its metabolites), liver signal could also be mainly attributable to metabolism and bile excretion of Mel-QA since previous distribution, in healthy rats, of [¹⁴C]-Mel-QA evidenced fecal biliary excretion of around 30% (25). These results are also consistent with published data of Mel (35, 36).

In the SRC model, there is a tendency for radioactivity to accumulate at a higher level in the tumor after [³H]-Mel-QA injection as compared with [³H]-Mel but statistically non significant due to the variability between animals. To correct the inter-rat variability, T/M ratio was calculated for each animal. If radioactivity is expected to accumulate specifically in tumor tissue, and not only due to blood flow, T/M ratio should be expected to be higher than 1. The mean T/M ratio for Mel-QA at 5 minutes after injection (5.77 ± 1.11) was significantly higher than for [³H]-Mel (2.22 ± 1.20, P < 0.05). It could be due to an enhanced accumulation in tumor as well as a lower circulating activity of the radiolabeled Mel-QA.

Finally, we confirmed the interest of PG-targeting strategy with the QA function for chondrosarcoma therapy notably in term of therapeutic index, essential in oncology. Combining tumor volume assessment, in vivo ^99mTc-NTP 15-5 scintigraphic imaging of PGs, ¹H-HRMAS NMR spectroscopy, and histologic analyses of biopsies, this work demonstrates that the conjugation of a QA function to Mel does not hamper its in vivo efficiency. More importantly, a significant decrease of side effects in terms of body weight and haematologic profile was observed for animals treated with Mel-QA as compared with those receiving Mel. Thus, QA function leads to a significant improvement of the therapeutic index. The excellent tolerability of Mel-QA could enable repeated cycles for the therapeutic effects in patients with chondrosarcoma.

The tumor growth inhibition observed in the SRC model treated by Mel-QA raises the question of the mechanism at the origin of this effect. Assessment of alkylating capacity using the 4-(4-nitrobenzyl)pyridine assay, as described previously (17), proved that an alkylating activity was maintained for Mel-QA with an alkylation rate of 52.5 ± 4.9 × 10⁻⁵ min⁻¹ compared with 20.9 ± 2.6 × 10⁻⁵ min⁻¹ for Mel (Supplementary Table S1). Nevertheless, is the biological activity observed only attributed to DNA alkylation, like for Mel (37), or to other events, such as the degratdation of PGs? Indeed, we think that the mechanism of action of Mel-QA is more complex than a simple alkylation and may be due to a remodeling of the PGs of ECM. We previously demonstrated that treatments with Mel-QA induce changes in the expression and degradation of the PGs (13). PGs are an important component of the tumor microenvironment, which represents a complex and highly dynamic media basically composed of nonmalignant cells as well as a ECM consisting of fibrous structural proteins, fibrous adhesive proteins, and PGs. Because of their complex structure, PGs play an important role in cell–cell and cell–ECM interactions and signaling in a variety of cellular functions (motility, adhesion, growth; refs. 38, 39). Modifications of PGs contribute to altered composition of ECM and also could explain the reduced tumor growth observed with Mel-QA in SRC-bearing rats. Concerning chondrosarcoma, the importance of PGs in SRC growth was firstly suggested by Oegema and colleagues who hypothesized that reducing the PG content was a way to decrease accessibility of growth factors and to increase the immunologic cell infiltration (40).

Recently, tumor microenvironment has been gradually recognized as a key contributor for cancer progression and drug resistance (41, 42). PGs appear as major partners for multiple organ integrity and function, and might represent interesting targets in other malignant pathologic processes, such as head and neck carcinomas (43, 44). As QA strategy has demonstrated its potential for chondrosarcoma management through ECM interaction, it will be of great interest to take advantage of another phenotypic feature of chondrosarcoma microenvironment that is hypoxia (7, 45, 46). Exploiting hypoxia is well-documented in cancer therapy through hypoxia-activated prodrugs and inhibitors of molecular targets upon which hypoxic cell survival depends. Therefore, combining PG targeting and hypoxia-activated prodrugs could serve as basis for innovative and selective, microenvironmentally targeted chondrosarcoma therapy.

No potential conflicts of interest were disclosed.

Conception and design: C. Peyrode, V. Weber, A. Vidal, P. Auzeloux, J.-M. Chezal, E. Miot-Noirault

Development of methodology: C. Peyrode, V. Weber, A. Voissière, A. Vidal, P. Auzeloux, M. Borel, E. Miot-Noirault

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Peyrode, V. Weber, A. Voissière, M. Borel, M.-M. Dauplat, F. Rédini, E. Miot-Noirault

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Peyrode, V. Weber, M. Borel, E. Miot-Noirault

Writing, review, and/or revision of the manuscript: C. Peyrode, V. Weber, A. Maisonial-Besset, P. Auzeloux, J.-M. Chezal, E. Miot-Noirault

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Gaumet, M. Quintana, F. Degoul

Study supervision: C. Peyrode, V. Weber, J.-M. Chezal, E. Miot-Noirault

Other (radiochemistry): A. Maisonial-Besset

The authors thank Delphine Skrzydelski and Romain Vives and especially Sophie Besse for her technical expertise and help for in vivo treatments and hematologic sampling.

Researchers assigned to our laboratory UMR 990 Inserm/UdA received financial support from Ligue contre le cancer Auvergne, CPER, and PRTK (INCa/DGOS 2015-051) for coordination of all the work presented here.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.