The different biological effects of TMPyP4 and cisplatin in the inflammatory microenvironment of osteosarcoma are attributed to G‐quadruplex

Abstract Objective Osteosarcoma (OS) is characterized by high levels of the tumour‐associated inflammatory microenvironment. Moreover, in approximately 60% of OS, telomere length is maintained by alternative lengthening of telomeres (ALT) pathway. Whether the ALT pathway can be exploited for OS therapeutic treatment and how the OS inflammatory microenvironment influences the anti‐cancer drug effect remains unknown. Here, we examined the biological effects of TMPyP4 and cisplatin in the inflammatory microenvironment of OS cells. Materials and methods Immunofluorescence in situ hybridization (IF‐FISH) and C‐circle experiments were used to detect the G‐quadruplex and ALT activity. The redox potential of single guanine, G‐quadruplex and G‐quadruplex/TMPyP4 was evaluated by the lowest unoccupied molecular orbital energy (LUMO), zeta potential and cyclic voltammetry. Cell viability, flow cytometry and apoptosis, Western blot, comet assay, adhesion, transwell and scratch experiments were performed to compare the anti‐tumour proliferation and migration effects of TMPyP4 and cisplatin in the inflammatory microenvironment. Results This study indicated that compared with cisplatin, TMPyP4 could induce the formation of human telomeres and FAK G‐quadruplex in vitro and in vivo, and TMPyP4‐treated OS cells showed fewer extrachromosomal C‐circles and fewer ALT‐associated promyelocytic leukaemia bodies. Consequently, the ALT activity and FAK‐related cell migration were suppressed by TMPyP4. Mechanistically, the formation of G‐quadruplex resulted in both lower redox potential than G within the genome and FAK transcription inhibition, and TMPyP4 could enhance this phenomenon, especially in the inflammatory microenvironment. Conclusions Our results reveal that TMPyP4 is more suitable for OS treatment than cisplatin.


| INTRODUC TI ON
Currently, osteosarcoma (OS) is a highly aggressive bone tumour that most commonly affects children and adolescents. 1,2 Surgical resection of tumours followed by chemotherapy constitutes the current standard procedure for clinical OS therapy due to its relative resistance to radiotherapy. 3 However, OS is often refractory to standardized chemotherapy regimens, and the application of tumour chemotherapy drugs has many adverse effects. 4,5 In recent decades, several efforts have been made to develop OS therapy, including new pharmacological findings for novel drugs such as sorafenib (Nexavar) 6 and nanomedicines that aim to release chemotherapeutic drugs in local sites. 7,8 Despite all medical advances, treatment and outcomes for OS have remained unchanged over the past 30 years, with a 5-year survival rate below 30%. 9,10 Thus, uncovering the factors that reduce the therapeutic effects of chemotherapy on OS has important significance.
Local tumours in OS patients are often accompanied by excessive inflammation characterized as 'red and swollen as well as hot and pain', and this tumour-associated inflammatory microenvironment is closely related to the high morbidity, poor outcomes and mortality of OS in the clinic. 11 This local severe inflammation of OS infiltrated by macrophages is also considered one of the markers of OS. 12,13 Although the mechanisms are unclear, clinicians and scholars have found that the levels of inflammation are correlated with tumour resistance to treatment. 14,15 Additionally, chemotherapy further promotes inflammatory events 16 and induced inflammation seems to play a role in the proliferation, angiogenesis and metastasis of OS. 17 Consequently, the inflammatory microenvironment of OS together with chemotherapy-induced inflammation could minimize the clinical efficacy of chemotherapy, even causing its failure and enhancing the invasion and migration of OS cells, leading to death. 17,18 Therefore, finding a new treatment strategy is vitally important for the treatment of OS.
Additionally, OS is distinct from most cancers in that the majority of OS lack telomerase activity and use the alternative lengthening of telomeres (ALT) mechanism to maintain telomeres. 19 However, anticancer drugs targeting ALT are still unavailable, 20 which is another challenge of OS treatment in addition to the severe inflammatory microenvironment and represents a large obstacle in OS treatment.
Currently, mechanistic evidence suggests a model in which ALT is mediated by endogenous homologous recombination (HR) machinery. 21 Our previous study suggested that the formation of G-quadruplex in the 3′-terminus of single-stranded telomeric DNA could prevent the invasion/annealing of telomeric ssDNA and, therefore, have potential value as an anti-ALT cancer therapeutic. 22 In this regard, a chemotherapy drug endowed with both potent anti-tumour effects and specifically targeting telomeric G-quadruplex might have more advantages in OS treatment.
TMPyP4, which possesses a strong electron-hole transfer capability, is a novel type of synthetic water-soluble photosensitizer in photodynamic therapy (PDT). 23 Compared with normal tissues, TMPyP4 is easily enriched in tumour tissues and, therefore, has a stronger tumour targeting ability. 24 Recently, TMPyP4 was reported to stabilize G-quadruplex both in vitro and in vivo. [25][26][27] Since G has the lowest redox potential within the genome, 28 the formation of G-quadruplex might constitute a dominant site for electrons and/or redox, 29 and TMPyP4 could enhance this phenomenon. 27 Redox is the essential controlling factor of inflammation because activating adequate immune response cells requires sufficient and rapidly available energy resources, which are intrinsically linked with the redox state. 30 It is a reasonable assumption that TMPyP4 would be more suitable for OS treatment than most clinical chemotherapy drugs, such as cisplatin.
In this report, we provide evidence that TMPyP4 has more advantages than most clinical chemotherapy drugs in OS treatment, especially in ALT-positive OS cells, in the inflammatory microenvironment. The results showed that the inflammatory microenvironment can enhance the sensitivity of ALT-positive OS U2OS and SAOS-2 cells to TMPyP4 while reducing the anti-cancer effect of cisplatin and promoting OS cell migration. The implications of these results are discussed.

| Drug administration
TMPyP4 (Item Number: 323497) and LPS (Item Number: L6386) were purchased from Sigma-Aldrich with purity ≥97%. Cisplatin was from Selleck (Item number: S1166). TNFα was from Peprotech (Item number: 315-01A). TMPyP4 was dissolved in water, and cisplatin was dissolved in dimethyl sulfoxide (DMSO) for storage and further diluted to final concentrations.
The reaction product formazan was dissolved in 100 μL of DMSO after discarding the culture medium. Cell viability was determined by reading the absorbance at 490 nm with an American thermoelectric Thermo Fisher Multiskan FC automatic microplate reader. The result is expressed as the mean ±standard deviation of three measurements (n = 3).

| Immunofluorescence-FISH assay
Immunofluorescence (IF) was performed as previously described. 32 Cells on coverslips were fixed with 4% paraformaldehyde for 15 minutes, washed with PBS three times, permeabilized in 0.5% Triton ×-100 in PBS for 30 minutes and then incubated with blocking solution (5% goat serum in 1 × PBS) for 1.5 hours at room temperature. Cells were loaded with primary antibodies in PBST against BG4 (Sigma), 53BP1 (CST), FAK (CST) or phalloidin (CST) overnight at 4°C. Cells were washed with PBST three times and then incubated with DyLight 488-conjugated anti-rabbit or DyLight549conjugated anti-rabbit secondary antibodies for 1.5 hours at room temperature. The coverslip was washed with PBST six times, fixed with 4% paraformaldehyde for 30 minutes, washed with PBS, dehydrated with graded ethanol, incubated with PNA probe, denatured at 85°C for 5 minutes and hybridized overnight at 37°C. The cells were washed and mounted with DAPI and imaged using a Nikon Ti microscope.

| C-circle assay
A C-circle assay was performed as described previously. 33 Briefly, U2OS and SAOS-2 cell samples were harvested, and genomic DNA was extracted according to the instructions in the AxyPrep Blood Genomic DNA Miniprep Kit. The concentration was detected by a NanoDrop. Next, 1 μg of genomic DNA was mixed with Rsa I, HinfI (4 U/μg) and RNase A (25 ng/μg) to configure a 20 μL digestion system and digested at 37°C overnight. Then, 10 μL of digestion product diluted with TE to 50 μL, and 1 μL of the diluted sample were mixed with 10 μL of the C-Circle reaction system (0.1% Tween, 0.2 mg/mL BSA, 1 mmol L -1 dTTP, dGTP, dATP each, 0.5 U Φ29 DNA polymerase and 1×Φ29 buffer) and water, which was amplified at 30°C for 8 hours followed by 65°C for 30 minutes. The product was diluted to 60 μL with 2 × SSC solution, blotted onto an NC membrane and subjected to UV cross-linking. Subsequently, the membrane was exposed to a phosphor screen and scanned. The results were quantified using Image Q software.

| The calculation of LUMO energy
First, PDB (https://www.rcsb.org) was used to establish the initial geometric structure of the single-stranded nucleic acid sequence (TTAGGG) 4 and G-quadruplex. Second, quantum chemistry calculations at a density functional theory of ωB97X-D/6-31G(d) were used to fully optimize the simplified models. The lowest unoccupied molecular orbital (LUMO) energies of a single G, G4 and TMPyP4-G4 were calculated by Gaussian09.

| Zeta potential measurements
The hTel oligomer was resuspended in 10 mmol L -1 Tris-HCl (pH 7.4) that contained 100 mmol L -1 KCl or no metal cations. The concentration of the oligomer was 3 μM, and TMPyP4 was dissolved at 10 mmol L -1 in water for later use. Next, the oligomers were heated at 90°C for 5 minutes, slowly cooled to RT (this process is best to perform overnight in a heat preservation device to prevent toorapid cooling) and then incubated at 4°C before the experiment. The zeta potential was determined using a Zetasizer Nano ZS apparatus (Malvern Instruments). All measurements were tested in triplicate (n = 3).
Conventional three-electrode tests were performed at RT with a glassy carbon electrode (GCE) used as the working electrode, a saturated calomel electrode (reference electrode) and a platinum electrode serving as the auxiliary electrode. Among them, the working electrode was polished with metallographic sandpaper before each measurement, polished with an Al 2 O 3 suspension and then washed with ethanol and pure water ultrasonically for use. The electrochemical test electrolyte solution was 10 mmol L -1 Tris-HCl (pH 7.4) or 10 mmol L -1 Tris-HCl (pH 7.4) that contained 100 mmol L -1 KCl buffer solution, and the potential scanning range was from −2.5 V to 1 V at a rate of 5 mV S -1 . It is worth noting that nitrogen was passed through to remove oxygen for more than 1 hour before the experiment, and the experimental process was carried out under nitrogen.

| Circular dichroism measurements
Circular Dichroism (CD) spectra were recorded using a spectropolarimeter (Applied Photophysics Ltd., UK) with a 1 cm long quartz cell, the wavelength range of 200-390 nm, and 200 nm min −1 scan speed with three acquisitions at room temperature. The oligomer of FAK was resuspended in 10 mmol L -1 Tris-HCl (pH 7.4) that contained 100 mmol L -1 KCl or no metal cations. The concentration of the oligomer was 10 μmol L -1 , and TMPyP4 was dissolved in 10 mmol L -1 water for later use. Subsequently, the oligomers were heated at 90°C for 5 minutes, slowly cooled to RT (this process is best to stay overnight in a heat preservation device to prevent too fast cooling) and then incubated at 4°C before the experiment. The buffer baseline correction is measured in the same quartz cell. During the titration experiment, the oligomer was fixed to 10 μmol L -1 , and different concentrations of TMPyP4 were added and equilibrated for at least 10 minutes (until there is no change in the CD signal) before performing spectral scanning. Data analysis was carried out by using GraphPad Prism 5.0.

| Cell adhesion assay
The cell adhesion assay was performed as described previously. 34 At room temperature, we coated a 96-well plate with 2.5 μg/mL human fibronectin in PBS (Millipore). Then, the treated cells were seeded into serum-free medium at a density of 4 × 10 4 cells/well and cultured at 37°C for 30 minutes under 5% CO2. Cells treated with vehicle (0.1% DMSO) were used as controls. The medium was gently removed, and then, the cells were fixed with 4% paraformaldehyde and stained with crystal violet at room temperature for 5 minutes.
After dissolving the crystal violet with 100 μL DMSO, the absorbance was measured at 560 nm. The following formula was used to calculate the relative number of cells attached to the extracellular matrix: average of treated cells OD/average OD control unit. The relative number of cells attached to the extracellular matrix was calculated using the following equation: (mean OD of treated cells/ mean OD of control cells) ×100%.

| Scratch-wound assay
Cell migration was detected by the scratch-wound assay. In short, U2OS and SAOS-2 cells were treated with or without drug, seeded in 6-well plates and grown to confluence in the growth medium.

| Comet assay
A comet assay was used to detect DNA damage. 35 Briefly, U2OS cells were mixed with 0.5% low-melting-temperature agarose before being transferred onto slides, which were coated with 1.5% normal agarose. For the alkaline assay, the slides were lysed in 1% Triton ×-100, 10 mmol L -1 Tris (pH 10.0), 100 mmol L -1 EDTA, 2.5 M NaCl overnight at 4°C and then electrophoresed in 1 mmol L -1 EDTA and 300 mM NaOH at 2 V/cm for 15 minutes.

| Statistical analysis
The student's two-tailed unpaired t test was used to determine the statistical significance and the resulting P-values are indicated in figures (*p < .05; **p < .01; ***p < .001).

| TMPyP4 induces the formation of Gquadruplex and inhibits the ALT activity of OS cells
It is well known that TMPyP4 stabilizes G-quadruplex in vitro 26 and may promote the formation of G-quadruplex in vivo. To test this hypothesis, immunofluorescence using antibodies to G-quadruplex (BG4) 36  The formation of G-quadruplex could result in a variety of cellular consequences, including altered gene expression, impaired DNA replication and/or cell cycle arrest. 26 Additionally, our previous report demonstrated that the formation of telomeric G-quadruplex inhibits ALT activity. 22 ALT was characterized by the presence of extrachromosomal C-circle DNA and the formation of ALT-associated PML bodies (APBs) and a high frequency of HR at telomeres. 20 Using a previously characterized assay for C-circle DNA, 22 we observed less C-circle DNA in TMPyP4-treated U2OS cells and SAOS-2 cells ( Figure 1D-1E). The number of APBs also decreased in TMPyP4treated U2OS and SAOS-2 cells but did not change after cisplatin treatment ( Figure 1F-1H, Figure S1D-1F). These results indicated that TMPyP4 treatment suppresses ALT activity by inducing the formation of telomeric G-quadruplex.

| The G-quadruplex has a lower redox potential than a single guanine base, and TMPyP4 can enhance this phenomenon
As mentioned before, a single G has the lowest redox potential within the genome, and the formation of a G-quadruplex might be a dominant site for electrons and/or redox; therefore, we first cal- Moreover, cyclic voltammetry (CV) assays further determined that compared with G-oligo, G-quadruplex and/or G-quadruplex/ TMPyP4 was confirmed to pass the current more effectively and observed a more obvious oxidation peak ( Figure 2C). This is consistent with the findings of quantum chemistry calculations and zeta potential tests.

| TMPyP4 was endowed with stronger anticancer activity in the inflammatory microenvironment
The formation of G-quadruplex triggers intense DNA damage and provokes a strong DNA damage response. 26 Considering the high

| TMPyP4 reduces the pseudopodia area of OS, while cisplatin increases it in the inflammatory microenvironment
During the experiments, we observed that compared with the TMPyP4-treated alone group, cells in the TNFα or LPS and TMPyP4 combination group became shrunken and decreased round and smooth ( Figure 5). Conversely, cisplatin-treated cells cultured in a TNFα-or LPS-induced inflammatory microenvironment had a more elongated structure, and cell-cell junctions were more obvious than those cultured in a normal culture environment ( Figure 5). The formation of G-quadruplex would affect the related gene expression. 26 We speculated that  Figure 6E-6F). This phenomenon was also observed in LPS pre-conditioning-treated U2OS and SAOS-2 cells ( Figure S4).

| TMPyP4 induced the formation of Gquadruplex in the FAK promoter and inhibited the transcription of FAK
To explore whether TMPyP4 inhibited the expression of FAK by inducing the formation of G-quadruplex in FAK-transcribed genes, we adopted the 'https://genome.ucsc.edu/' database to obtain the promoter of PTK2 encoding FAK. Then, the website 'https://bioin forma tics.ramapo.edu/QGRS/index.php' was used to preliminarily predict the G-quadruplex formation ability of the PTK2 promoter region.
Surprisingly, no less than 11 oligo sequences had G-quadruplex formation potency (Table S1). Then, a circular dichroism (CD) assay was performed to determine the FAK G-quadruplex induced by TMPyP4.
As expected, we found that the selected FAK oligo sequences would form a parallel G-quadruplex (a negative peak at 241 nm and a positive peak at 264 nm on CD spectrometry, Figure 7A  Cell adhesion is often associated with cell migration. 43 Next, we performed a scratch-wound healing assay to determine the migration rate of TMPyP4-or cisplatin-treated U2OS and SAOS-2 cells in the presence or absence of TNFα or LPS. Our results showed that the TNFα-or LPS-induced inflammatory microenvironment sensitizes the anti-migration ability of TMPyP4 but, to some extent, restores the migration ability of cisplatin-treated cells ( Figure 8E, Figure S5E).

| D ISCUSS I ON
Locally higher inflammatory responses, which would limit the effect of chemotherapy, 13,14 are typical clinical symptoms of OS.
Chemotherapy tends to aggravate local inflammatory symptoms, finally resulting in chemotherapy failure. 44 Furthermore, nearly 40% of OS cell lines are ALT-positive, 19 which are resistant to most chemotherapy, and anti-cancer therapeutics targeting ALT are not yet available, adding extra challenges to OS treatment in the clinic. 9 The aim of this study was to explore why chemotherapy is stagnant in OS treatment and how to further improve the curative effect. Our results reveal that TMPyP4, a G-quadruplex stabilizer and photosensitizer, has more advantages in OS treatment than cisplatin in an inflammatory microenvironment.
Cisplatin is a common chemotherapeutic drug that exerts its anticancer effect by producing cross-linking with guanines (G) in DNA. 45 Due to the lowest redox potential of G throughout the genome, it has become the preferred target of ROS, 46 which are produced abundantly during inflammation. 30 We speculated that cisplatin might not cross-link with G effectively because of the competitive effect of ROS in the inflammatory microenvironment of OS; thus, the anti-cancer activity was impaired. Our results revealed that the TNFα-or LPS-simulated inflammatory microenvironment increased DNA resistance to cisplatin (Figure 3, Figure S2) and reduced the sensitivity of OS cells to cisplatin, thereby reducing tumour apoptosis and promoting survival (Figures 3 and 4, Figure S3). Our discovery also revealed that cisplatin was ineffective against the migration of OS cells, especially in the TNFα-or LPS-simulated inflammatory microenvironment ( Figure 8, Figure S5). Mechanistically, cisplatin treatment increased the expression and distribution of FAK, which is attributed to cell migration, at the leading edge of OS cells and promoted the formation of pseudopodia ( Figure 6, Figure S4), leading to an increased ability to adhere to the extracellular matrix ( Figure 8A and 8B). Accordingly, increased transferability ( Figure 8C and 8D) and migration rate ( Figure 8E) were observed for cisplatin-treated cells. It has been widely accepted that increased cell migration and transferability are closely associated with cancer invasion and metastasis, a major cause of osteosarcoma patient death. 47 Thus, our study implies that cisplatin may increase the risk of OS metastasis.
It has been proposed that G-quadruplex formation in telomeres inhibits ALT activity and alters gene transcription and expression in the genome. Nearly 40% of OS cells are ALT-positive, 19 and G-quadruplex more easily attract electrons than a single G. 29 In this regard, a G-quadruplex inducer and/or stabilizer might have more advantages in OS treatment. TMPyP4 is a photosensitizer in PDT, 27 and an excellent G-quadruplex stabilizer 27 satisfies the above criteria. TMPyP4 induced the formation of telomeric Gquadruplexes, thereby attenuating C-circles/APBs, indicating the suppression of ALT (Figure 1, Figure S1). Persistent G-quadruplex activated stronger DNA damage, which might be exacerbated by ROS generated from inflammation. Consequently, compared with the conventional culture environment, TMPyP4 triggered more intense DNA damage but failed to activate comparable DNA damage responses in the inflammatory microenvironment ( Figure 3, Figure S2), leading to cell cycle arrest and apoptosis or senescence ( Figure 4, Figure S3).
Migration accounts for therapy failure and cancer-related death in OS. 2,8,9,43 In these scenarios, anti-migration is worthy of more attention in OS treatment. Because G-quadruplex are widely located at the promoter regions of migration-related genes, 26 such as VEGF 48 and WNT1, 49 TMPyP4 was expected to inhibit the migration and The cells were pre-treated with 10 μmol L -1 TMPyP4 or 5 μmol L -1 cisplatin for 24 h, incubated with 10 ng/mL TNFα for 48 h, and then collected for RT-PCR. The Values are represented as the mean ± SD of at least three independent experiments. The statistical significance was calculated using the unpaired Student's two-tailed t test (*p < .05, **p < .01, ***p < .001) and it could not completely simulate the local OS-associated microenvironment. Thus, the applicability of TMPyP4 from this study to mouse models, PDX models and patients may be questioned. In addition, TMPyP4-induced G-quadruplex are widely present in the genome and could result in a variety of cellular consequences, which might influence the function of normal cells. Our future studies will not only focus on the effect of TMPyP4 on solid tumours, including mouse models and PDX models but also explore stabilizers that specifically induce the formation of G-quadruplex in metastasisrelated genes and telomeres, which may reduce the risk of adverse off-target effects in normal human cells.

ACK N OWLED G EM ENTS
The present study was supported by the National Natural Science

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

AUTH O R CO NTR I B UTI O N S
GL, XHZ and JQC designed the research. JQC, XXJ, YNM, ZS, JFZ, HYS and MSW performed the research study and analysed the data.
XHZ, GL and JQC wrote the paper. All the authors approved the final manuscript and agreed for the publication.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.