The Histone Acetyltransferase HBO1 Functions As a Novel Oncogenic Gene in Osteosarcoma

1 The histone acetyltransferase HBO1 functions as a

2 novel oncogenic gene in osteosarcoma

3 4 Yan-yang Gao 1#, Zhuo-yan Ling 2#, Yun-Rong Zhu 3#, Ce Shi 4#, Yin Wang 5, Xiang-yang 5 Zhang 2, Zhi-qing Zhang 5, Qin Jiang 6*, Min-Bin Chen 7* Shuofei Yang 8*, and Cong Cao 5, 9* 6 7 1 Department of Pediatrics, Xinyi people's Hospital, Xinyi, China 8 2 Department of Orthopedics, the Second Affiliated Hospital of Soochow University, Suzhou, 9 China 10 3 Department of Orthopedics, The Affiliated Jiangyin Hospital of Medical College of Southeast 11 University, Jiangyin, China 12 4 Department of Orthopedics, The Affiliated Suqian Hospital of Xuzhou Medical University, 13 Suqian, China 14 5 Jiangsu Key Laboratory of Neuropsychiatric Diseases and Institute of Neuroscience, Soochow 15 University, Suzhou, China. 16 6 The Fourth School of Clinical Medicine, Nanjing Medical University, Nanjing, China. 17 7 Department of Radiotherapy and Oncology, Affiliated Kunshan Hospital of Jiangsu 18 University, Kunshan, China. 19 8 Department of Vascular Surgery, Renji Hospital, School of Medicine, Shanghai Jiaotong 20 University, Shanghai, China. 21 9 North District, The Affiliated Suzhou Hospital of Nanjing Medical University, Suzhou 22 Municipal Hospital, Suzhou, China. 23

24 #Co-first authors.

25 * To whom correspondence should be addressed: 26 Prof. Qin Jiang, The Fourth School of Clinical Medicine, Nanjing Medical University. 138 27 Han-zhong Road, Nanjing, China, 210029. Tel/Fax: +86-025-86677699. E-mail: 28 [email protected]. 29 Dr. Min-Bin Chen, M.D. Ph.D. Department of Radiotherapy and Oncology, Affiliated 30 Kunshan Hospital of Jiangsu University, 91 Qianjin Road, Kunshan, 215300, China. Tel/Fax: 31 +86-512-57559009. Email: [email protected].

32 Dr. Shuofei Yang, M.D., Ph.D., Department of Vascular Surgery, Renji Hospital, School of 1

33 Medicine, Shanghai Jiaotong University, Shanghai 200127, China, Email: 34 [email protected] 35 Prof. Cong Cao, Jiangsu Key Laboratory of Neuropsychiatric Diseases and Institute of 36 Neuroscience, Soochow University, 199 Ren-ai Road, Suzhou, Jiangsu 215123, China. Tel/Fax: 37 +86-0512-65883658. E-mail: [email protected]. 38

39 Abstract. 40 HBO1 (KAT7 or MYST2) is a histone acetyltransferase that acetylates H3 and H4 histones.

41 Methods: HBO1 expression was tested in human OS tissues and cells. Genetic strategies, 42 including shRNA, CRISPR/Cas9 and overexpression constructs, were applied to exogenously 43 alter HBO1 expression in OS cells. The HBO1 inhibitor WM-3835 was utilized to block HBO1 44 activation. Results: HBO1 mRNA and protein expression is significantly elevated in OS tissues

45 and cells. In established (MG63/U2OS lines) and primary human OS cells, shRNA-mediated 46 HBO1 silencing and CRISPR/Cas9-induced HBO1 knockout were able to potently inhibit cell 47 viability, growth, proliferation, as well as cell migration and invasion. Significant increase of 48 apoptosis was detected in HBO1-silenced/knockout OS cells. Conversely, ectopic HBO1 49 overexpression promoted OS cell proliferation and migration. We identified ZNF384 (zinc 50 finger protein 384) as a potential transcription factor of HBO1. Increased binding between 51 ZNF384 and HBO1 promoter was detected in OS cell and tissues, whereas ZNF384 silencing 52 via shRNA downregulated HBO1 and produced significant anti-OS cell activity. In vivo, 53 intratumoral injection of HBO1 shRNA lentivirus silenced HBO1 and inhibited OS xenograft 54 growth in mice. Furthermore, growth of HBO1-knockout OS xenografts was significantly 55 slower than the control xenografts. WM-3835, a novel and high-specific small molecule HBO1 56 inhibitor, was able to potently suppressed OS cell proliferation and migration, and led to 57 apoptosis activation. Furthermore, intraperitoneal injection of a single dose of WM-3835

58 potently inhibited OS xenograft growth in SCID mice. Conclusion. HBO1 overexpression 59 promotes OS cell growth in vitro and in vivo. 60 61 Running title: HBO1 overexpression in osteosarcoma

62 Keywords: Osteosarcoma; HBO1; Histone acetylation; WM-3835 63

64 Graphic Abstract

66 Introduction 67 68 Osteosarcoma (OS) is an aggressive malignant bone tumor arising from the primitive 69 transformed cells with mesenchymal origin [1, 2]. It is the most common histological form of 70 primary bone cancer, and it is mainly detected in teenagers and young adults [3]. It comprises 71 2.4% of all malignancies in pediatric patients, and about 20% of all primary bone tumors [3-5]. 72 The long-term overall survival rate of OS has improved significantly since the late 20th century 73 (65-70%)[6], and the current clinical OS treatments include combined chemotherapy, radiation, 74 and tumor resection [7, 8]. For the patients with advanced OS, including those with metastatic 75 or recurrence OS, the prognosis is far from satisfactory [6]. As a result, novel molecular 76 pathogenesis and targeted therapeutics for OS have been the focuses of recent studies [9-14]. 77 78 The epigenetic mechanism has been suggested to play a role in the cause of OS 79 pathogenesis and progression [15]. Among the epigenetic alternations, histone modification 80 through reversible acetylation is crucial for gene expression regulation [16-18]. Two enzymes, 81 histone acetyltransferase (HAT) and histone deacetylase (HDAC), are responsible for histone 82 acetylation [18, 19]. Dysregulation of histone acetylation is associated with OS tumorigenesis 83 and progression [15, 20, 21]. 84 85 HBO1 (KAT7 or MYST2) is a HAT that acetylates H4 and H3 histones at different lysine 86 residues [22]. It consists of a N-terminal domain and a C-terminal MYST domain, which are 87 required for acetyl-CoA binding and acetylation [23]. HBO1 is critical in regulating a number of 88 key cellular behaviors and physiological functions, including gene transcription, protein

89 ubiquitination, immune regulation, stem cell pluripotent, self-renewal maintenance, and 90 embryonic development [23-28], and it also forms complexes with native subunits and cofactors 91 [22, 23]. Complexes formed by HBO1 and BRPF scaffold (BRPF1, BRD1/BRPF2 or BRPF3) 92 are capable of directing HBO1 specificity toward histone H3K14 acetylation (H3K14ac). 93 Moreover, HBO1 complexes containing JADE (JADE1, JADE2 and JADE3) can also direct 94 HBO1 specificity toward histone H4 acetylation [24, 25, 29]. H3K14ac by HBO1 facilitates the 95 activation of replication origins, and histone H4 acetylation facilitates chromatin loading of 96 minichromosome maintenance protein complex (MCM) to promote DNA replication licensing 97 [24-28]. 98 99 HBO1 binds directly to cdt1 (Chromatin licensing and DNA replication Factor 1), a protein 100 required for G1 phase cell cycle progression [22, 30], to promote histone acetylation in cell cycle 101 progression and cell proliferation. Additionally, HBO1 acts as a positive regulator of 102 centromeric CENPA (Centromere Protein A) by preventing SUV39H1-mediated centromere 103 inactivation [31]. Niida et al., found that ATR-dependent HBO1 phosphorylation induced by 104 ultraviolet irradiation promoted its localization to DNA damage sites, which is essential for 105 XPC recruitment and nucleotide excision repair [32]. 106 107 Recent studies have proposed that HBO1 plays an oncogenic role in human cancers [22, 23, 108 33]. Wang et al., demonstrated that HBO1 overexpression in gastric cancer is negatively 109 correlated with patients’ survival [34]. In bladder cancer cells, HBO1 is involved in 110 Wnt/beta-catenin signaling and cancer cell proliferation [35]. Another study by MacPherson et 111 al., has shown that HBO1 HAT domain is essential for the acetylation of H3K14 (H3K14ac). 112 The latter promoted the processivity of RNA polymerase II to maintain high expression of key 113 genes (MYLK , HOXA9 and others) in leukemia stem cells [33]. While genetic silencing or 114 pharmacological inhibition of HBO1 potently inhibited leukemia stem cell progression[33]. 115 Kueh and colleagues, however, reported that HBO1 does not have an essential role in cell 116 proliferation and DNA replication in HEK293T, MCF7,or HeLa [36]. In this study, we tested 117 the expression and potential function of HBO1 in human OS [36], and we show that 118 overexpressed HBO1 acts as a novel oncogenic gene essential for OS cell tumorigenesis and 119 progression.

120

121 Methods

122 123 Chemicals and reagents. Cell Counting Kit-8 (CCK-8) was provided by Dojindo Co. 124 (Kumamoto, Japan). Puromycin and polybrene were provided by Sigma-Aldrich Chemicals (St. 125 Louis, Mo).Antibodies for HBO1 (#58418), cleaved caspase-3 (#9664), acetyl-Histone H3 at 126 Lys14 (H3K14ac, #7627), Histone H3 (#4499), acetyl-Histone H4 at Lys12 (H4K12ac, #13944), 127 Histone H4 (#2935),acetyl-Histone H4 at Lys5 (H4K5ac, #8674),cleaved-poly (ADP-ribose) 128 polymerase (PARP) (#5625), cleaved-caspase-9 (#20750), Caspase-3 (#9668), Caspase-9 129 (#9508), PARP(#9532) and Tubulin (#2125) were obtained from Cell Signaling Tech China 130 (Shanghai, China). A Zinc finger protein 384 (ZNF 384) antibody was provided by Abcam 131 (ab176689, Shanghai, China). Caspase inhibitors, z-DEVD-fmk and z-VAD-fmk, were provided 132 by Sigma-Aldrich Chemicals. The HBO1 inhibitor WM-3835, 133 N'-(4-fluoro-5-methyl-[1,1'-biphenyl]-3-carbonyl)-3-hydroxybenzenesulfonohydrazide was 134 synthesized by Min-de Biotech (Suzhou, China) based on the described protocol [33]. RNA 135 reagents, Lipofectamine 2000, and other transfection reagents were provided by Thermo-Fisher 136 Invitrogen (Shanghai, China). All primers, sequences, constructs, and vectors were designed and 137 provided by Genechem Co. (Shanghai, China), unless mentioned otherwise.

138 139 Cell culture. Primary human OS cells from Dr. Ji at Nanjing Medical University [37, 38] 140 were derived from two written-informed consent OS patients. Patients received no 141 chemotherapy and radiotherapy before surgery. OS tumor tissues were washed, minced, and 142 incubated in collagenase I solution (Gibco, Boston, MA). Cell pellets were isolated and cultured 143 in described medium [39]. Primary OS cells were named as pOS-1 and pOS-2. The established 144 human OS cell lines, U2OS and MG-63, were purchased from Cell Bank of Shanghai Institute 145 of Biological Science (Shanghai, China). The primary human osteoblasts were provided by Dr. 146 Ji as well [40, 41]. Human osteoblasts were differentiated and cultured as described [42, 43]. The 147 protocols of using primary human cells were approved by IACUC and Ethics committee of 148 Nanjing Medical University. 149 150 Human tissues. OS tissues and the surrounding normal bone tissues from a set of ten (10) 151 written-informed primary OS patients were provided by authors institutions. These patients 152 received no chemotherapy and radiotherapy before surgery. Tissues were incubated with the 153 described lysis buffer [37] and stored in liquid nitrogen. The written-informed consent was

154 obtained from each participant. The protocols of this study were approved by the Ethic 155 Committee of Soochow University according to Declaration of Helsinki. 156 157 Western blotting. The detailed protocols of Western blotting analyses were described in 158 details in our previous studies [44-46]. When testing different proteins, the exact same set of 159 lysates were run in parallel ―sister‖ gels. 160 161 Quantitative real time-PCR (qPCR). Total RNA was extracted by TRIzol reagents and 162 was reversely transcripted to cDNA [47]. qPCR was performed by an ABI Prism 7500 system 163 using a SYBR GREEN PCR Master Mix (Applied Biosystems, Shanghai, China). The product 164 melting temperature was always calculated. Glyceraldehyde-3-phosphatedehydrogenase(GAPDH) was 165 tested as the internal control and reference gene. The quantification of targeted mRNA was 166 done by 2−∆∆Ct method. The mRNA primers were listed in Table-S1. Other mRNA primers were

167 verified and provided by Genechem (Shanghai, China).

168 169 shRNA-mediating gene silencing. Two shRNAs targeting non-overlapping sequences of

170 human HBO1, shHBO1-a and shHBO1-b (sequences listed in Table-S1), were designed and 171 verified by Genechem Co. The shRNAs were sub-cloned into GV248 172 (hU6-MCS-Ubiquitin-IRES-puromycin) vector (Genechem), and the two constructs were 173 individually transfected to HEK-293T cells together with the lentivirus package plasmid mix 174 (Genechem). OS cells were seeded into six-well plates (0.8 × 105 cells per well)in 175 polybrene-containing complete medium. HBO1 shRNA lentivirus was added to cultured OS 176 cells. To select stable cells, puromycin (5.0 μg/mL) was added in complete medium, and cells 177 were cultured for five passages (10-12 days). HBO1 silencing in stable cells was assessed by 178 qPCR and Western blotting assays. Control cells were transfected with lentivirus encoding 179 scramble control shRNA (―shC‖). ZNF384 shRNAs (sequences listed in Table-S1) and stable 180 cell selection were carried out similarly.

181 182 HBO1 knockout. The small guide RNA (sgRNA) targeting human HBO1 (Target DNA 183 sequence: GATGAACGAGTCTGCCGAAG, PAM sequence PAM Sequence: AGG) was 184 inserted into the lenti-CRISPR-GFP-puro plasmid [48]. OS cells were seeded into six-well plates 185 (0.8 × 105 cells per well) in polybrene-containing complete medium. The construct was 186 transfected to OS cells by Lipofectamine 2000 (Thermo-Fisher Invitrogen, Shanghai, China).

187 GFP-positive OS cells were sorted by FACS, and then distributed into 192-well plates to achieve 188 single cells. Cells were further subjected to HBO1 knockout (KO) screening. HBO1 KO stable 189 cells were achieved via the selection by puromycin (5 μg/mL),and then verified by qPCR and 190 Western blotting analyses.

191

192 Forced HBO1 overexpression. The full-length HBO1 cDNA sequence was synthesized by 193 Genechem Co. (Shanghai, China), and it was sub-cloned into a GV369 gene expression 194 lentiviral vector (Genechem). The construct and the lentivirus package plasmid mix were 195 co-transfected into HEK-293T cells to generate HBO1-expression lentivirus. OS cells were 196 seeded into six-well plates (0.8 × 105 cells per well) in polybrene-containing complete medium. 197 The viruses were filtered, enriched, and added to cultured OS cells. To select stable cells, 198 puromycin was added in the culture medium. HBO1 overexpression in the stable cells was then 199 verified by qPCR and Western blotting analyses.

200 201 Cell viability and death assays. Briefly, cells with applied genetic modifications or 202 treatments were seeded into 96-well plates at 3 × 103 cells per well. After 96h incubation. In 203 each well 10 μL CCK-8 reagent was added for 3h. Later, CCK-8’s optical density (OD) value 204 was recorded at 450 nm. Cell death was tested by Trypan blue staining using an automatic cell 205 counter. 206 207 Cell migration and invasion. OS cells with the applied genetic modifications or treatments 208 were seeded at 5 × 104 cells per chamber (in serum-free medium) into the upper surfaces of 209 ―Transwell‖ chambers (BD Biosciences, Shanghai, China). The complete medium with12% 210 FBS was added into the lower compartments of the chambers. Following 24h incubation, OS 211 cells that migrated through the membrane were fixed by using 4% paraformaldehyde and 212 co-stained with crystal violet. Cells were then observed and counted under a microscope 213 (magnification: × 100). For cell invasion assays, the chamber surface was coated with Matrigel 214 (Sigma). 215 216 EdU staining. OS cells were seeded into six-well plates at 1.2 × 105 cells per well and were 217 cultured for 48h. An EdU (5-ethynyl-20-deoxyuridine) Apollo-567 Kit (RiboBio, Guangzhou, 218 China) was applied to quantify cell proliferation according to the attached protocols. EdU and 219 DAPI fluorescent dyes were added to OS cells, and the cells were visualized under a fluorescent

220 microscope (Leica, Shanghai, China). Five random views (n = 5) with a total of 1 000 cells per 221 each condition were used to calculate the average EdU ratio (% vs. DAPI). 222 223 TUNEL staining. OS cells with the applied genetic modifications or treatments were 224 seeded into six-well plates at 1.2 × 105 cells per well and cultured for 72h. Cells were then 225 stained with TUNEL (Invitrogen) for 4h under the dark. Nuclei were co-stained with DAPI and 226 visualized under a fluorescent microscope. Five random views (n = 5) contained a total of 1, 227 000 cells from each condition were used to calculate the average TUNEL ratio (% vs. DAPI). 228

229 Single stranded DNA (ssDNA) ELISA. In apoptotic cells, DNA breaks would lead to the 230 accumulation of ssDNA. Briefly, 48h after applied treatment, 30 μg of cell lysates per treatment 231 were analyzed by ssDNA ELISA kit (Roche Diagnostics, Shanghai, China) based on the 232 attached protocol. The ssDNA ELISA absorbance in each well was tested at 450 nm. 233 234 FACS.OS cells with applied genetic modifications were seeded into six-well plates at a 235 density of 1.2 × 105 cells per well. After applied treatments, cells were harvested, washed, and 236 resuspended in the binding buffer with Annexin V-FITC and propidium iodide (PI) (both at 5 237 μg/mL). Cells were then analyzed by a flow cytometer (Becton Dickinson, Shanghai, China) 238 using BD FACSDivaTM software. Annexin V-positive cells were gated as apoptotic cells, and 239 the ratio was recorded. To test cell cycle progression, cells were stained only with PI.

240

241 Caspase-3 activity. OS cells with the applied genetic modifications or treatments were 242 seeded into six-well plates at a density of 1.2 × 105 cells per well. Cells were collected and lysed 243 with cell lysis buffer. Caspase-3 activation was measured fluorometrically through a caspase-3 244 fluorescent assay kit (Sigma-Aldrich) with active caspase-3 substrate peptides Ac-DEVD-AMC 245 (7-amido-4-methyl coumarin) fluorogenic substrate. It was tested under a spectrofluorometer at 246 an excitation wavelength of 380 nm and an emission wavelength of 450 nm 247 248 Mitochondrial depolarization. With mitochondrial membrane potential (MMP) reduction 249 and mitochondrial depolarization, the fluorescent probe JC-1 aggregates are decomposed into a 250 monomer form, and the emitted fluorescence changes from red to green. Cells with applied 251 genetic modifications or treatments were seeded into six-well plates at a density of 1.2 × 105 252 cells per well and incubated with JC-1 at room temperature for 20 min under the dark. The JC-1

253 green monomers intensity (488 nm) was tested with a flow cytometer (Becton Dickinson). The 254 representative JC-1 images integrating both green and red fluorescence channels were presented. 255 256 Chromatin immunoprecipitation (ChIP) assay. As reported [49], cells or fresh human 257 tissues were first cross-linked with 1% formaldehyde, and glycine solution is added for 258 quenching. Afterwards, cells or tissues were washed and lysed using the lysis buffer. Lysate 259 samples were treated with a MisonixSonicator 3000 Homogenizer [49] to fragment genomic 260 DNA (200–800 bp). After centrifugation, the supernatant was removed and ChIP Dilution 261 Buffer was added. Lysates were then pre-cleared, and an anti-ZNF384 antibody was added. 262 Detailed protocols for ChIP were reported previously [49]. At last, ZNF384-bound DNA was 263 eluted from the protein A/G agarose with elution buffer, and NaCl was added to eliminate 264 cross-linking between proteins and genomic DNA. DNA containing HBO1 promoter was 265 analyzed via quantitative PCR with the primers at the promoter region of HBO1 (Genechem, 266 Shanghai, China). For all ChIP assay experiments, negative control regions were tested to 267 exclude the possible non-specific binding. 268

269 Small interfering RNA (siRNA). The siRNAs targeting SOX10, PITX3, ZNF384, and SP2 270 were designed, synthesized, and verified by Genechem (Shanghai, China). The sequences were 271 listed in Table-S1. OS cells were seeded into six-well tissue culture plate at 0.8×105 cells per well 272 (50-60% confluence). Cells were transfected with 200 nM of targeted siRNA or the scramble 273 control siRNA (―siR-C‖) by Lipofectamine 3000 (Thermo-Fisher Invitrogen, Shanghai, China) 274 for 48h. Over 80% knockdown efficiency of targeted mRNA was achieved(tested by qPCR 275 assays). 276

277 Xenograft assay. The severe combined immunodeficient (SCID) mice (18-19g, female) 278 were obtained from Experimental Animal Center of Soochow University (Suzhou, China). 279 pOS1 cells (5×106 cells per mouse, in Matrigel-containing serum free medium) were 280 subcutaneously (s.c.) injected to the right flanks. Tumor-bearing SICD mice were randomly 281 assigned into different groups (7-10 mice per group). Tumor volumes and the mice body weights 282 were recorded every seven days. The animal studies were approved by Institutional Animal Care 283 and Use Committee (IACUC) and Ethics Committee of Soochow University. 284

285 Statistical analyses. In vitro experiments were repeated at least three times, and similar

286 results were obtained. The quantitative data of normal distribution were presented as mean ± 287 standard deviation (S.D.). Statistical analyses were carried out by one-way ANOVA with the 288 Scheffe' and Tukey Test (SPSS 23.0, SPSS Co. Chicago, IL). When comparing the significance 289 between two treatment groups, a two-tailed unpaired T test was utilized (Excel 2007, Microsoft 290 Co. Seattle, Washington). P-values < 0.05 were considered as statistically significant differences.

291

292 Results

293

294 HBO1 upregulation in human OS 295 296 Over 60% of the common histological subtypes of bone sarcoma are OS [3].We first 297 searched the TARGET Pan-Cancer (PANCAN) database and examined HBO1 expression in 298 children’s sarcoma tissues using the UCSC Xena project, and then we screened RNASeq data of 299 HBO1 in sarcoma tissues. As demonstrated, HBO1 mRNA expression in children’s sarcoma 300 tissues (n = 166) is significantly upregulated [P < 0.05 vs. surrounding normal tissues (n = 70)]

301 (Figure 1A). Importantly, high expression of HBO1 mRNA in human sarcoma tissues is

302 associated with low overall survival (Figure 1B). 303 304 To confirm the bioinformatics results, we tested HBO1 expression in human OS tissues. 305 Tissue specimens from a set of ten OS patients were obtained. To test mRNA expression, qPCR 306 was performed. As shown in Figure 1C, HBO1 mRNA expression in OS tumor tissues (―T‖) was 307 over six-fold than that in normal tissues adjacent to tumor (―N‖) (P < 0.05 vs. ―N‖ tissues).

308 After testing HBO1 protein expression by Western blotting analyses, we confirmed the dramatic 309 HBO1 protein upregulation in OS tumor tissues from four representative patients (Patient

310 #1/#3/#6/#10) (Figure 1D). When combining all human tissues’ blotting data, we found that 311 HBO1 protein level in OS tumor tissues is significantly upregulated (P < 0.05 vs. ―N‖ tissues)

312 (Figure 1E). 313 314 HBO1 expression in OS cells was tested as well. In established OS cell lines, U2OS and 315 MG-63, as well as in the primary human OS cells, pOS-1, and pOS-2 (derived from two OS 316 patients), HBO1 mRNA expression was significantly higher than that in the primary human

317 osteoblasts (―Osteoblasts‖ [40, 41]) (P < 0.05, Figure 1F). Furthermore, HBO1 protein

318 upregulation was detected in established OS cells and primary human OS cells (Figure 1G)

319 when compared to the low expression in primary human osteoblasts (Figure 1G). These results 320 demonstrated that HBO1 is upregulated in OS.

321 322 HBO1 shRNA inhibits OS cell progression in vitro

323 324 To study the potential role of HBO1 in human OS cells, shRNA strategy was applied. Two 325 lentiviral shRNAs targeting non-overlapping sequences of human HBO1, shHBO1-a and 326 shHBO1-b, were individually transduced to pOS-1 cells. Stable cells were established by the

327 selection of puromycin containing medium. qPCR assay results in Figure 2A confirmed that 328 HBO1 mRNA decreased over 90% in stable pOS-1 cells containing HBO1 shRNA. HBO1

329 protein levels were significantly downregulated as well (Figure 2B), leading to dramatic

330 inhibition of H4 acetylation (H4K5ac and H4K12ac) and H3K14ac [33] (Figure 2B). 331 Expressions of HBO1-dependent mRNAs, MYLK and HOXA9 were downregulated in HBO1

332 shRNA-expressing OS cells (Figure S1A). As compared to the parental control cells, the growth

333 of HBO1 shRNA-expressing pOS-1 cells was significantly inhibited (Figure 2C). CCK-8 assay

334 results in Figure 2D demonstrated that cell viability was robustly decreased in HBO1-silenced 335 pOS-1 cells. 336 337 Miotto et al., have shown that HBO1 directly interacts with the replication licensing factor 338 Cdt1 [50] and its histone acetylase activity is required for DNA replication licensing [51]. 339 Conversely, Cdt1 phosphorylation by JNK1 inhibited Cdt1-HBO1 association and blocked 340 replication licensing [26]. Evidenced by significantly decreased percentage of EdU-positive 341 nuclei, pOS-1 cell proliferation was inhibited by shRNA-mediated knockdown of HBO1 (Figure 342 2E). The PI-FACS assay demonstrated that HBO1 knockdown induced reduction of S-phase

343 cells (Figure 2F). 344 345 Next, ―Transwell‖ and ―Matrigel Transwell‖ assays were performed to test cell migration 346 and invasion, respectively. We found that HBO1 shRNA significantly inhibited migration 347 (Figure 2G) and invasion (Figure 2H) of pOS-1 cell. For the ―Transwell‖ assays, cells were 348 incubated for only 24h to exclude the possible influence of cell growth/proliferation change. 349 Together, we reported here HBO1 silencing potently inhibited pOS-1 cell viability, growth, 350 proliferation, as well as cycle cycle progression, cell migration and invasion. The scramble

351 control shRNA (―shC‖), as expected, did not affect HBO1 expression (Figure 2A and B) and

352 pOS-1 cell functions (Figure 2C-H). 353 354 In the established human OS cell lines U2OS and MG63, and in primary OS cells-derived 355 another patient (pOS-2), transfection of HBO1 shRNA lentivirus (shHBO1-a) led to dramatic 356 HBO1 mRNA (Figure 2I) and protein (Figure S1B) downregulation. Functional studies

357 demonstrated that HBO1 shRNA largely inhibited cell viability (Figure 2J), proliferation (Figure

358 2K), and migration (Figure 2L). These results indicated that HBO1 silencing inhibits OS cell 359 progression in vitro. 360

361 HBO1 shRNA provokes OS cell apoptosis

362 363 The potential effect of HBO1 shRNA in cell apoptosis was studied. In pOS-1 primary cells, 364 as HBO1 was silenced by shHBO1-a and shHBO1-b (see Figure 2),we discovered significantly

365 enhanced activity of caspase-3 (Figure 3A) and increased cleavages of caspase-3, caspase-9, and

366 PARP (poly ADP-ribose polymerase) (Figure 3B). Increased single stand DNA (ssDNA) 367 contents, which indicates DNA breaks, were also detected in HBO1-silenced pOS-1 cells (Figure

368 3C). Evidenced by mitochondrial JC-1 green monomers accumulation, we demonstrated that

369 mitochondrial depolarization in pOS-1 cells with HBO1 shRNAs (Figure 3D). These results 370 suggested that mitochondrial apoptosis cascade was activated in HBO1-silenced OS cells [52, 371 53]. 372 373 We further showed that TUNEL-positive nuclei ratio was significantly increased in pOS-1 374 cells with HBO1 shRNAs, suggesting the upregulated apoptosis level (Figure 3E). Additionally, 375 the number of pOS-1 cells with positive Annexin V staining was largely increased following

376 HBO1 silencing (Figure 3F), further confirming apoptosis activation. The scramble control

377 shRNA (―shC‖) failed to provoke significant apoptosis activation in pOS-1 cells (Figure 3A-F). 378 In pOS-1 cells, two caspase inhibitors, the caspase-3 inhibitor z-DEVD-fmk and the pan caspase 379 inhibitor z-VAD-fmk, largely attenuatedshHBO1-a-induced viability (CCK-8 OD) reduction 380 (Figure 3G) and cell death (by recording Trypan blue-positive cells, Figure 3H). 381 382 In pOS-2 primary cells and established OS cell lines (U2OS/MG63), HBO1 silencing by 383 shHBO1-a (see Figure 2 and Figure S1B) induced mitochondrial depolarization (Figure 3I),

384 which also activated caspase-3 (Figure 3J) and increased TUNEL-positive nuclei ratio (Figure

385 3K), indicating apoptosis activation. These results indicated that HBO1 silencing would provoke 386 OS cell apoptosis. 387

388 HBO1 knockout induces potent anti-OS cell activity

389 390 To exclude possible off-target effect of the applied HBO1 shRNAs, the CRISPR/Cas9 gene 391 editing method was utilized to stably knockout (KO) HBO1. A lentiviral 392 CRISPR/Cas9-HBO1-KO construct (see Methods) was transduced in pOS-1 primary cells. 393 After KO screening and puromycin selection, two stable cell lines, koHBO1-1 and koHBO1-2,

394 were established. As shown, HBO1 mRNA (Figure S2A) and protein (Figure S2B) expressions

395 were depleted in koHBO1pOS-1 cells. H4 and H3 de-acetylation (Figure S2B), as well as

396 MYLK-HOXA9 downregulation (Figure S1C), were detected in HBO1 KO cells. 397 CRISPR/Cas9-induced HBO1 KO in pOS-1 cells potently inhibited cell viability (CCK-8 OD, 398 Figure S2C) and proliferation (by recording EdU-positive nuclei ratio, Figure S2D). ―Transwell‖

399 assay results in Figure S2E also demonstrated that cell migration was largely inhibited in 400 koHBO1 cells. 401 402 Further studies showed that the caspase-3 activity was robustly increased in koHBO1-1 403 cells and koHBO1-2 cells (Figure S2F). HBO1 KO also induced cleavages of caspase-3,

404 caspase-9, and PARP (Figure S2G) in pOS-1 cells, and accumulation of ssDNA that indicated

405 DNA damage (Figure S2H). Additionally, the increased TUNEL-positive nuclei ratio confirmed

406 the activation of apoptosis in koHBO1 pOS-1 cells (Figure S2I). 407 408 Similar results were observed in pOS-2 primary cells and established OS cell lines 409 (U2OS/MG63). Stable introduction of the CRISPR/Cas9-HBO1-KO construct (―koHBO1‖) 410 almost depleted HBO1 mRNA and protein (Figure S2J and Figure S1D). HBO1 KO largely

411 inhibited cell viability (Figure S2K), proliferation (Figure S2L), and migration (Figure S2M). 412 Therefore, CRISPR/Cas9-induced HBO1 KO produced significant anti-OS cell activity. 413

414 Ectopic HBO1 overexpression promotes OS cell growth

415 416 Because HBO1 shRNA or KO inhibited OS cell growth and induced cell apoptosis, we 417 hypothesized that ectopic overexpression of HBO1 should exert opposite effects. Therefore, a

418 lentiviral construct encoding full-length HBO1 cDNA was transduced to pOS-1 primary cells, and 419 we established stable OE-HBO1 cells through selection. Compared to the control cells with the 420 empty vector (―Vec‖), HBO1 mRNA levels were significantly increased (over nine folds) in

421 OE-HBO1 cells (Figure 4A). Consequently, HBO1 protein expression was also upregulated

422 (Figure 4B), with increased MYLK-HOXA9 mRNA expression detected as well (Figure S1E). 423 Through cell viability testing with CCK-8 assays, we confirmed that ectopic overexpression of

424 HBO1 increased the viability of pOS-1 cells (Figure 4C). EdU-positive nuclei ratio was also 425 significantly increased in the OE-HBO1cells, indicating that HBO1 overexpression promoted

426 pOS-1 cell proliferation (Figure 4D, results were quantified). Ectopic overexpression of HBO1

427 promoted pOS-1 cell migration (Figure 4E) and invasion (Figure 4F, results were quantified). As

428 expected, the vector control failed to affect HBO1 expression (Figure 4A and B) and the

429 functions of pOS-1 cells (Figure 4C-F). 430 431 In pOS-2 primary cells and established OS cell lines (U2OS/MG63), stable transfection of 432 the HBO1 construct increased HBO1 mRNA (Figure 4G) and protein (Figure S1F) expression.

433 We also found that Ectopic overexpression of HBO1 augmented cell proliferation (EdU

434 incorporation, Figure 4H) and migration (Figure 4I) in the OS cells. 435

436 Zinc finger protein 384 (ZNF 384) is a possible HBO1 transcription factor 437 438 To study the underlying mechanism of HBO1 upregulation in human OS, we focused on 439 possible transcription factors of HBO1 by searching in the JASPAR database [54]. Through 440 bioinformatical studies, we identified four possible transcription factors with the highest binding 441 affinity to HBO1’s promoter (sequences listed in Table-S1): SP2, ZNF384 (zinc finger protein

442 384), PITX3, and SOX10 (Figure 5A). Next, siRNA strategy was utilized to knockdown each of 443 the predicted transcription factors. In pOS-1 cells, the applied siRNA resulted in dramatic 444 knockdown of targeted transcription factor (over 80% mRNA knockdown efficiency). 445 Importantly, only SP2 siRNA and ZNF384 siRNA resulted in significant downregulation of 446 HBO1 mRNA in pOS-1 cells (Figure 5B), while the other two failed to do so (Figure 5B). Also,

447 ZNF384 siRNA-induced HBO1 mRNA reduction was significantly more potent than that of

448 SP2 siRNA(Figure 5B). These results suggested that ZNF384 could be an important 449 transcription factor of HBO1 in OS cells. 450

451 Chromatin immunoprecipitation (ChIP) assay results in Figure 5C confirmed that ZNF384 452 directly binds HBO1 promoter in human osteoblasts and in established (U2OS and MG63 lines) 453 and primary OS cells (pOS-1 and pOS-2). More importantly, there was a significant increase of 454 ZNF384-HBO1 promoter binding in OS cells (P < 0.05 vs. osteoblasts) (Figure 5C). Also,

455 ZNF384-HBO1 promoter binding was over six-time higher in OS tumor tissues (Figure 1) than

456 that in normal tissues(P < 0.05, Figure 5D). These results implied that increased binding 457 between HBO1 promoter and its proposed transcription factor ZNF384 could be one of the 458 primary mechanisms of HBO1 upregulation in OS. 459 460 To further test our hypothesis, lentiviral constructs, encoding three different ZNF384 461 shRNAs (shZNF384-s1/shZNF384-s2/shZNF384-s3, targeting non-overlapping 462 sequences),were individually transduced to pOS-1 cells. Stable cells were established via 463 puromycin selection. Each of the applied shRNA led to significant ZNF384 protein 464 downregulation (Figure 5E). Consequently, HBO1 protein expression was dramatically reduced

465 (Figure 5E). Protein expression of a known ZNF384-dependent gene, Cyclin D1 [49], was

466 significantly downregulated in pOS-1 cells with ZNF384 shRNA (Figure 5E). Of the applied 467 shRNAs, shZNF384-s3, led to most dramatic ZNF384-HBO1 downregulation, and this shRNA 468 was selected for further studies. As shown, ZNF384 shRNA (shZNF384-s3) induced robust 469 downregulation of HBO1 mRNA and HBO1-dependent mRNAs, MYLK and HOXA9, in pOS-1

470 cells (Figure 5F). Functional studies in pOS-1 cells showed that ZNF384 silencing resulted in

471 inhibition of proliferation (decreased EdU-positive nuclei ratio, Figure 5G) and migration

472 (Figure 5H), mitochondrial depolarization (JC-1 green monomers accumulation, Figure 5I),

473 and cell apoptosis (TUNEL ratio increase, Figure 5J), which mimicked the actions by HBO1 474 shRNA/KO. Importantly, ZNF384 shRNA-induced anti-OS cell activity was largely attenuated

475 by ectopic HBO1 expression using the HBO1-expressing construct (OE-HBO1, see Figure 4) as

476 shown in Figure S3A-E. These results implied that HBO1 downregulation should be the major 477 reason of ZNF384 shRNA-induced anti-cancer actions in OS cells. 478 479 Similarly, in pOS-2 primary cells and established OS cells (U2OS and MG63 lines), 480 transfection of the lentiviral ZNF384 shRNA construct (shZNF384) resulted in ZNF384 mRNA

481 (Figure 5K) and protein (Figure S1G) silencing, causing HBO1 mRNA reduction (Figure 5L).

482 ZNF384 shRNA induced significant viability (CCK-8 OD) reduction (Figure 5M) and

483 proliferation inhibition (EdU staining, Figure 5N) in the primary and established OS cells.

484 485 We also tested ZNF384 expression in OS cells and tissues. ZNF384 mRNA and protein 486 expressions were elevated in U2OS, MG63, pOS-1 and pOS-2 cells, when compared to its levels 487 in osteoblasts (Figure S1H). Furthermore, as shown in Figure 5O, ZNF384 mRNA levels in OS

488 tumor tissues (―T‖, see Figure 1) were significantly higher than those in normal tissues (―N‖, see

489 Figure 1) (P < 0.05 vs. ―N‖ tissues). ZNF384 protein upregulation was also detected in OS

490 tumor tissues from four representative patients (Patient #1/#3/#6/#10) (Figure 5P). 491 Furthermore, TCGA database showed ZNF384 transcript is upregulated in human sarcoma

492 tissues (Figure 5Q). Therefore, HBO1 expression is correlated with ZNF384 expression in 493 human OS tissues. ZNF384 upregulation, as well as the increased binding between ZNF384 494 andHBO1 promoter, could be the primary mechanisms of HBO1 upregulation in human OS. 495

496 HBO1 is required for OS xenograft growth in mice

497 498 To study the potential effect of HBO1 on OS cell growth in vivo, a mice xenograft model 499 was utilized. As described, pOS-1 primary cells were s.c. injected to the flanks of SCID mice. 500 Xenograft tumors were established within three weeks, with tumor volume close to 100 mm3 501 (―Day-0‖). The pOS-1 xenografts-bearing mice were randomly assigned into three groups (10 502 mice per group) receiving intratumoral injection of shHBO1-a, shHBO1-b (HBO1 shRNA 503 lentivirus) or scramble control shRNA lentivirus (―shC‖). Tumor growth curve shown in Figure

504 6A demonstrated that pOS-1 xenografts with shHBO1 injection grew significantly slower than 505 xenografts with shC lentivirus. By calculating estimated daily tumor growth using the formula: 506 (Tumor volume at Day-42—Tumor volume at Day-0)/42, we found that the growth of HBO1 507 shRNA-injected pOS-1 xenografts was significantly inhibited (Figure 6B). At experimental 508 Day-42, tumors were all separated and individually weighted. Xenografts receiving HBO1 509 shRNA injection were much lighter than those with control shRNA injection (Figure 6C). There

510 was no significant difference in the animal body weights within the three groups (Figure 6D). 511 We also failed to detect any apparent toxicities. 512 513 To test signaling changes in vivo, one tumor of each group was separated at experimental 514 Day-7 and Day-14. A total of six xenografts were obtained and tumor tissues were tested by 515 qPCR and Western blotting. As shown, HBO1 mRNA levels decreased over 90% in HBO1

516 shRNA-injected pOS-1 xenografts (Figure 6E). HBO1 protein expression and H4 acetylation

517 decreased as well in the pOS-1 xenografts with HBO1 shRNA (Figure 6F), at which the

518 cleavages of caspase-3 and PARP were detected (indicating apoptosis activation, Figure 6F).

519 Levels of total H3 and H4 histones were unchanged (Figure 6F). Therefore, HBO1 shRNA 520 lentivirus injection silenced HBO1 and inhibited pOS-1 xenograft growth in SCID mice. 521 522 Next, stable pOS-1 cells expressing the CRISPR/Cas9-HBO1-KO construct (ko-HBO1 cells, 523 see Figure S2) and cells with the CRISPR/Cas9 control construct (Cas9-C, see Figure S2) were 524 s.c. injected to the flanks of SCID mice. Forty days after cells inoculation, xenograft tumors 525 were separated and measured. As shown, the volumes of xenografts of ko-HBO1cells were 526 dramatically lower than xenografts of Cas9-C control cells (Figure 6G). Again, no difference

527 was found in mice body weights between the two groups (Figure 6H). After analyzing the tumor

528 tissues, we showed that HBO1 mRNA (Figure 6I) and protein (Figure 6J) expressions were

529 almost completely depleted in ko-HBO1 xenografts, and we detected H4 acetylation inhibition 530 as well as cleavages of caspase-3 and PARP (Figure 6J). H3 and H4 histones were unchanged

531 (Figure 6J). These results further confirmed that HBO1 is required for OS xenografts growth in 532 SCID mice. 533

534 The anti-OS activity by a HBO1 inhibitor WM-3835 535 536 WM-3835, a small-molecular HBO1 inhibitor, was synthesized based on the structure in a 537 early study[33], and its anti-OS cell activity was tested. CCK-8 viability assay results showed

538 that WM-3835 inhibited pOS-1 cell viability in a concentration-dependent manner (Figure 7A). 539 Furthermore, the HBO1 inhibitor required at least 48h to exert a significant anti-survival activity, 540 displaying a time-dependent manner as well (Figure 7A). Western blotting studies demonstrated 541 that WM-3835 did not alter HBO1 protein expression, but suppressed H4K12ac-H3K14ac in a 542 dose-dependent manner (Figure 7B). Expressions of total H3 and H4 histones were unchanged

543 (Figure 7B). Furthermore, WM-3835 inhibited pOS-1 cell proliferation dose-dependently

544 (EdU-positive nuclei ratio, Figure 7C). Titration experiments showed that 5 μM of WM-3835

545 resulted in robust viability reduction of pOS-1 cells (Figure 7A), while H3-H4 de-acetylation

546 (Figure 7B) and proliferation inhibition (Figure 7C) were observed. Thus this concentration

547 (close to IC-50, Figure 7A) was selected for following studies. 548 549 We found that WM-3835 (5 μM) downregulated MYLK-HOXA9 mRNA expression (Figure

550 S1I) in pOS-1 cells. It also potently inhibited pOS-1 cell migration (Figure 7D) and invasion

551 (Figure 7E). Additionally, caspase activation (Figure 7F) and cleavages of caspase-3, caspase-9,

552 and PARP (Figure 7G) were detected in WM-3835-treated pOS-1 cells. Further confirming cell 553 apoptosis activation by WM-3835, TUNEL-positive nuclei were significantly increased 554 following WM-3835 treatment in pOS-1 cells (Figure 7H). Additionally, Annexin V-positive cell

555 number was increased as well (Figure 7H). These results confirmed the significant anti-OS 556 activity of HBO1 inhibitor WM-3835. 557

558 Importantly, in HBO1-KO pOS-1 cells, koHBO1-1 and koHBO1-2 (see Figure S2), 559 treatment with WM-3835 (5 μM) failed to induce apoptosis and reduction of viability (CCK-8

560 and TUNEL assays, Figure 7I). Similarly, in HBO1-low human osteoblasts (see Figure 1), 561 treatment with WM-3835(5 μM) failed to induce significant cytotoxicity and apoptosis (Figure

562 7J). These results suggest that WM-3835 induced anti-OS cell activity through HBO1 inhibition. 563 In pOS-2 primary cells and established OS cell lines (U2OS/MG63), WM-3835 (5 μM) potently

564 inhibited cell viability (CCK-8 OD, Figure 7K) and proliferation (EdU ratio reduction, Figure

565 7L), while provoking apoptosis (TUNEL-positive nuclei ratio increase, Figure 7M). 566 567 The activity of WM-3835 in vivo was tested as well. The pOS-1 cells were injected s.c. to the 568 flanks of SCID mice to establish OS xenografts. As demonstrated, daily i.p. injection of 569 WM-3835 (at 10 mg/kg body weights for21 days) potently inhibited pOS-1 xenograft growth in 570 SCID mice (Figure 7N). WM-3835 did not provoke apparent toxicities to the experimental 571 animals, and no significant difference was found in mice body weights between two groups 572 (Figure 7O). Therefore, i.p. injection of WM-3835 successfully inhibited OS xenografts growth 573 in mice. 574

575 Discussion

576 577 Histone acetylation is a covalent modification of lysine residues in histone tails [19]. This 578 process is vital for the regulation of gene expression [19]. HBO1 (KAT7/MYST2) belongs to the 579 MYST family of acetyltransferases, and is responsible for the majority of H4 acetylation [23]. 580 HBO1 is evolutionarily conserved, and functions in the context of a multi-protein HAT complex 581 [23, 55]. HAT complex is composed of JADE1/2/3, HBO1, ING4/5, BRPF and Eaf6, and it 582 regulates DNA replication, gene transcription, expression, cell cycle progression, and other

583 important cellular functions [23]. 584 585 HBO1 plays an essential role in gene transcription, DNA replication, and is hence 586 implicated in cancer initiation and progression [22, 23, 33-35]. Quintela et al., reported that in 587 ovarian cancer cells, HBO1 acetylated histone H4 through the co-regulator JADE2 to regulate 588 mechano-transduction pathways and cell elasticity [30]. In embryonic stem cells, association of 589 HBO1 and Niam (Nuclear Interactor of ARF and Mdm2) is essential in early development, cell 590 survival, as well as cancer progression [56]. Recently, Taniue and colleagues reported that 591 TUSC3 expression induced by HBO1 via histone acetylation is critical for colon cancer cell 592 proliferation [57]. Bai et al., also reported that microRNA-639 silenced HBO1 via blocking 593 Wnt/β-catenin pathway to inhibit human hepatocellular carcinoma cell proliferation and 594 migration [58]. 595 596 The results of this study suggested that HBO1 could be an important novel oncogene for 597 OS cell progression. HBO1 is associated with poor overall survival as it was overexpressed in OS 598 tissues, whereas low expression was detected in parecancer normal tissues and cultured human 599 osteoblasts. In established and primary human OS cells, HBO1 silencing induced by shRNA 600 potently inhibited cell viability, proliferation, migration and invasion, but promoted significant 601 apoptosis activation. Furthermore, CRISPR/Cas9-induced HBO1 KO exerted robust anti-OS 602 cell activity. Conversely, ectopic overexpression of HBO1 augmented OS cell proliferation and 603 migration. In vivo, HBO1 silencing by intratumoral injection of HBO1 shRNA lentivirus 604 potently inhibited OS xenograft growth in SCID mice. Furthermore, growth of HBO1-KO 605 pOS-1 xenografts was largely inhibited in SCID mice. Therefore, HBO1 presented vital 606 oncogenic activities required for OS cell progression in vitro and in vivo, and it should be

607 considered as a novel therapeutic target for OS. 608 609 ZNF384, a C2H2-type zinc finger protein, is a transcription factor that regulates the 610 transcription of extracellular matrix genes [59]. Studies have proposed that ZNF384 could be a 611 potential oncogene and it is overexpressed in various human cancers [49, 60-62]. Zhong et al., 612 showed that ZNF384 is important for acute leukemia (ALL) progression and it fused with the 613 TET family genes including Ewing sarcoma breakpoint region 1 (EWSR1), TATA box binding 614 protein-associated factor (TAF15), and transcription factor 3 (TCF3) [62]. ZNF384 615 overexpression promoted melanoma cell migration [61], while in cervical cancer, ZNF384

616 bound to APOBEC3B’s promoter [60]. He et al., found that ZNF384 is overexpressed in 617 hepatocellular carcinoma (HCC) and it promoted HCC cell growth by upregulating cyclin D1 618 expression [49]. 619 620 In the present study, our results proposed that ZNF384 could be an important transcription 621 factor that directly binds to HBO1 promoter. In OS, increased binding of ZNF384-HBO1 622 promoter could be the primary mechanism of HBO1 upregulation. In support of our conclusion, 623 we found that shRNA-induced silencing of ZNF384 downregulated HBO1 expression and the 624 target mRNAs in OS cells. Moreover, ZNF384 shRNA produced significant anti-OS cell 625 activity, which mimics the activity induced by HBO1 silencing/KO. Importantly, ectopic 626 overexpression of HBO1 significantly attenuated ZNF384 shRNA-induced anti-OS cell activity. 627 628 WM-3835 is a cell permeable small molecule HBO1 inhibitor [33]. WM-3835 binds directly 629 to the acetyl-CoA binding site of HBO1 [33]. It is a highly specific inhibitor of HBO1, as it 630 makes additional interactions with HBO1 protein surface. Specifically, WM-3835 phenol forms 631 a hydrogen-bonding network with HBO1’s Glu525 and Lys488. The two amino acids are 632 specific for HBO1, but not conserved throughout the MYST family proteins. Therefore, 633 WM-3835 can exert robust and selective HBO1 inhibition [33]. 634 635 We found that WM-3835 activated apoptosis while inhibited OS cell proliferation, 636 migration and invasion. Notably, the activity of WM-3835 was nullified in HBO1-KO OS cells. 637 Also, the HBO1 inhibitor was non-cytotoxic in HBO1-low human osteoblasts, which indicates 638 that the anti-OS cell efficacy of WM-3835 is primarily through HBO1 inhibition. Importantly, 639 i.p. injection of a single dose of WM-3835 potently inhibited pOS-1 xenograft growth in SCID

640 mice, highlighting the potential translational value for this compound. Therefore, targeting 641 HBO1 by WM-3835 could be a novel therapeutic strategy against human OS.

642 643 Despite significant progresses that have been made in surgical techniques and new adjuvant 644 chemotherapies, over 30% of OS patients still die from lung metastases within five years of 645 diagnosis [3, 8], and the prognosis of advanced and recurrent OS patients remains extremely 646 poor [3, 8]. The results of this study indicated that targeting HBO1 via pharmacological 647 (WM-3835 or others) or genetic methods could be a valuable novel therapeutic approach against 648 human OS . Moreover, we here identified ZNF384 as a potential transcription factor of HBO1.

649 Increased binding between ZNF384 and HBO1 promoter could be a key mechanism of HBO1 650 upregulation in OS. Considering that HBO1 is upregulated in several other human cancers 651 (gastric cancer, bladder cancer, etc) [22, 23, 33] and functions as a potential oncogene [22, 23, 652 33], it certainly will be interesting to further test the potential anti-cancer activity of HBO1 653 genetic silencing or pharmacological inhibition in other cancer models. 654 655 Competing interests. The authors declare no conflict of interest. 656

657 Funding. This work was generously supported by grants from the National Natural Science 658 Foundation of China (81922025, 81974388, 81870679, 81101676, 81472786, 81302195, 659 82072712, 31371139, 81571282, 81771457, 81371055, 81773192, 81502162 and 81570859) and 660 Suqian Sci&Tech Program (Grant No.K202016), a Project Funded by the Priority Academic 661 Program Development of Jiangsu Higher Education Institutions. The funders had no role in the 662 study design, data collection and analysis, decision to publish, or preparation of the manuscript.

663

664 References 665

666 1. Isakoff MS, Bielack SS, Meltzer P, Gorlick R. Osteosarcoma: Current Treatment and a Collaborative Pathway to 667 Success. J Clin Oncol. 2015; 33: 3029-35. 668 2. Kansara M, Teng MW, Smyth MJ, Thomas DM. Translational biology of osteosarcoma. Nat Rev Cancer. 2014; 14: 669 722-35. 670 3. Ottaviani G, Jaffe N. The epidemiology of osteosarcoma. Cancer Treat Res. 2009; 152: 3-13. 671 4. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020; 70: 7-30. 672 5. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019; 69: 7-34. 673 6. Anderson ME. Update on Survival in Osteosarcoma. Orthop Clin North Am. 2016; 47: 283-92. 674 7. Clark JC, Dass CR, Choong PF. A review of clinical and molecular prognostic factors in osteosarcoma. J Cancer Res 675 Clin Oncol. 2008; 134: 281-97. 676 8. Carrle D, Bielack SS. Current strategies of chemotherapy in osteosarcoma. Int Orthop. 2006; 30: 445-51. 677 9. Zhao J, Dean DC, Hornicek FJ, Yu X, Duan Z. Emerging next-generation sequencing-based discoveries for targeted 678 osteosarcoma therapy. Cancer Lett. 2020; 474: 158-67. 679 10. Bishop MW, Janeway KA, Gorlick R. Future directions in the treatment of osteosarcoma. Curr Opin Pediatr. 2016; 680 28: 26-33. 681 11. Denduluri SK, Wang Z, Yan Z, Wang J, Wei Q, Mohammed MK, et al. Molecular pathogenesis and therapeutic 682 strategies of human osteosarcoma. J Biomed Res. 2015; 30. 683 12. Liang X, Wang X, He Y, Wu Y, Zhong L, Liu W, et al. Acetylation dependent functions of Rab22a-NeoF1 Fusion 684 Protein in Osteosarcoma. Theranostics. 2020; 10: 7747-57. 685 13. Hu XK, Rao SS, Tan YJ, Yin H, Luo MJ, Wang ZX, et al. Fructose-coated Angstrom silver inhibits osteosarcoma 686 growth and metastasis via promoting ROS-dependent apoptosis through the alteration of glucose metabolism by 687 inhibiting PDK. Theranostics. 2020; 10: 7710-29. 21

688 14. Wang S, Zhong L, Li Y, Xiao D, Zhang R, Liao D, et al. Up-regulation of PCOLCE by TWIST1 promotes metastasis in 689 Osteosarcoma. Theranostics. 2019; 9: 4342-53. 690 15. Cui J, Wang W, Li Z, Zhang Z, Wu B, Zeng L. Epigenetic changes in osteosarcoma. Bull Cancer. 2011; 98: E62-8. 691 16. Jenuwein T, Allis CD. Translating the histone code. Science. 2001; 293: 1074-80. 692 17. Schuldner M, Dorsam B, Shatnyeva O, Reiners KS, Kubarenko A, Hansen HP, et al. Exosome-dependent immune 693 surveillance at the metastatic niche requires BAG6 and CBP/p300-dependent acetylation of p53. Theranostics. 2019; 9: 694 6047-62. 695 18. Yang G, Feng J, Liu Y, Zhao M, Yuan Y, Yuan H, et al. HAT1 signaling confers to assembly and epigenetic regulation 696 of HBV cccDNA minichromosome. Theranostics. 2019; 9: 7345-58. 697 19. Marmorstein R, Zhou MM. Writers and readers of histone acetylation: structure, mechanism, and inhibition. Cold 698 Spring Harb Perspect Biol. 2014; 6: a018762. 699 20. Wittenburg LA, Bisson L, Rose BJ, Korch C, Thamm DH. The histone deacetylase inhibitor valproic acid sensitizes 700 human and canine osteosarcoma to doxorubicin. Cancer Chemother Pharmacol. 2011; 67: 83-92. 701 21. Roh MS, Kim CW, Park BS, Kim GC, Jeong JH, Kwon HC, et al. Mechanism of histone deacetylase inhibitor 702 Trichostatin A induced apoptosis in human osteosarcoma cells. Apoptosis. 2004; 9: 583-9. 703 22. Iizuka M, Takahashi Y, Mizzen CA, Cook RG, Fujita M, Allis CD, et al. Histone acetyltransferase Hbo1: catalytic 704 activity, cellular abundance, and links to primary cancers. Gene. 2009; 436: 108-14. 705 23. Lan R, Wang Q. Deciphering structure, function and mechanism of lysine acetyltransferase HBO1 in protein 706 acetylation, transcription regulation, DNA replication and its oncogenic properties in cancer. Cell Mol Life Sci. 2020; 77: 707 637-49. 708 24. Feng Y, Vlassis A, Roques C, Lalonde ME, Gonzalez-Aguilera C, Lambert JP, et al. BRPF3-HBO1 regulates replication 709 origin activation and histone H3K14 acetylation. EMBO J. 2016; 35: 176-92. 710 25. Lalonde ME, Avvakumov N, Glass KC, Joncas FH, Saksouk N, Holliday M, et al. Exchange of associated factors 711 directs a switch in HBO1 acetyltransferase histone tail specificity. Genes Dev. 2013; 27: 2009-24. 712 26. Miotto B, Struhl K. JNK1 phosphorylation of Cdt1 inhibits recruitment of HBO1 histone acetylase and blocks 713 replication licensing in response to stress. Mol Cell. 2011; 44: 62-71. 714 27. Burke TW, Cook JG, Asano M, Nevins JR. Replication factors MCM2 and ORC1 interact with the histone 715 acetyltransferase HBO1. J Biol Chem. 2001; 276: 15397-408. 716 28. Iizuka M, Stillman B. Histone acetyltransferase HBO1 interacts with the ORC1 subunit of the human initiator 717 protein. J Biol Chem. 1999; 274: 23027-34. 718 29. Saksouk N, Avvakumov N, Champagne KS, Hung T, Doyon Y, Cayrou C, et al. HBO1 HAT complexes target chromatin 719 throughout gene coding regions via multiple PHD finger interactions with histone H3 tail. Mol Cell. 2009; 33: 257-65. 720 30. Quintela M, Sieglaff DH, Gazze AS, Zhang A, Gonzalez D, Francis L, et al. HBO1 directs histone H4 specific 721 acetylation, potentiating mechano-transduction pathways and membrane elasticity in ovarian cancer cells. 722 Nanomedicine. 2019; 17: 254-65. 723 31. Ohzeki J, Shono N, Otake K, Martins NM, Kugou K, Kimura H, et al. KAT7/HBO1/MYST2 Regulates CENP-A 724 Chromatin Assembly by Antagonizing Suv39h1-Mediated Centromere Inactivation. Dev Cell. 2016; 37: 413-27. 725 32. Niida H, Matsunuma R, Horiguchi R, Uchida C, Nakazawa Y, Motegi A, et al. Phosphorylated HBO1 at UV irradiated 726 sites is essential for nucleotide excision repair. Nat Commun. 2017; 8: 16102. 727 33. MacPherson L, Anokye J, Yeung MM, Lam EYN, Chan YC, Weng CF, et al. HBO1 is required for the maintenance of 728 leukaemia stem cells. Nature. 2020; 577: 266-70. 729 34. Wang Y, Chen S, Tian W, Zhang Q, Jiang C, Qian L, et al. High-Expression HBO1 Predicts Poor Prognosis in Gastric 730 Cancer. Am J Clin Pathol. 2019; 152: 517-26. 731 35. Chen Z, Zhou L, Wang L, Kazobinka G, Zhang X, Han X, et al. HBO1 promotes cell proliferation in bladder cancer via

732 activation of Wnt/beta-catenin signaling. Mol Carcinog. 2018; 57: 12-21. 733 36. Kueh AJ, Eccles S, Tang L, Garnham AL, May RE, Herold MJ, et al. HBO1 (KAT7) Does Not Have an Essential Role in 734 Cell Proliferation, DNA Replication, or Histone 4 Acetylation in Human Cells. Mol Cell Biol. 2020; 40. 735 37. Chen W, Chen M, Xu Y, Chen X, Zhou P, Zhao X, et al. Long non-coding RNA THOR promotes human osteosarcoma 736 cell growth in vitro and in vivo. Biochem Biophys Res Commun. 2018; 499: 913-9. 737 38. Fei HD, Yuan Q, Mao L, Chen FL, Cui ZH, Tao S, et al. Assessment of GSK1904529A as a promising 738 anti-osteosarcoma agent. Oncotarget. 2017; 8: 49646-54. 739 39. Chaiyawat P, Pruksakorn D, Phanphaisarn A, Teeyakasem P, Klangjorhor J, Settakorn J. Expression patterns of class I 740 histone deacetylases in osteosarcoma: a novel prognostic marker with potential therapeutic implications. Mod Pathol. 741 2018; 31: 264-74. 742 40. Liu G, Chen FL, Ji F, Fei HD, Xie Y, Wang SG. microRNA-19a protects osteoblasts from dexamethasone via targeting 743 TSC1. Oncotarget. 2018; 9: 2017-27. 744 41. Zhao S, Mao L, Wang SG, Chen FL, Ji F, Fei HD. MicroRNA-200a activates Nrf2 signaling to protect osteoblasts from 745 dexamethasone. Oncotarget. 2017; 8: 104867-76. 746 42. Zhu CY, Yao C, Zhu LQ, She C, Zhou XZ. Dexamethasone-induced cytotoxicity in human osteoblasts is associated 747 with circular RNA HIPK3 downregulation. Biochem Biophys Res Commun. 2019; 516: 645-52. 748 43. Fan JB, Liu W, Zhu XH, Cui SY, Cui ZM, Zhao JN. microRNA-7 inhibition protects human osteoblasts from 749 dexamethasone via activation of epidermal growth factor receptor signaling. Mol Cell Biochem. 2019; 460: 113-21. 750 44. Wang SS, Lv Y, Xu XC, Zuo Y, Song Y, Wu GP, et al. Triptonide inhibits human nasopharyngeal carcinoma cell 751 growth via disrupting Lnc-RNA THOR-IGF2BP1 signaling. Cancer Lett. 2019; 443: 13-24. 752 45. Liu YY, Chen MB, Cheng L, Zhang ZQ, Yu ZQ, Jiang Q, et al. microRNA-200a downregulation in human glioma leads 753 to Galphai1 over-expression, Akt activation, and cell proliferation. Oncogene. 2018; 37: 2890-902. 754 46. Cao C, Huang X, Han Y, Wan Y, Birnbaumer L, Feng GS, et al. Galpha(i1) and Galpha(i3) are required for epidermal 755 growth factor-mediated activation of the Akt-mTORC1 pathway. Sci Signal. 2009; 2: ra17. 756 47. Tang XF, Liu HY, Wu L, Li MH, Li SP, Xu HB. Ginseng Rh2 protects endometrial cells from oxygen glucose 757 deprivation/re-oxygenation. Oncotarget. 2017; 8: 105703-13. 758 48. Wu F, Liu F, Dong L, Yang H, He X, Li L, et al. miR-1273g silences MAGEA3/6 to inhibit human colorectal cancer cell 759 growth via activation of AMPK signaling. Cancer Lett. 2018; 435: 1-9. 760 49. He L, Fan X, Li Y, Chen M, Cui B, Chen G, et al. Overexpression of zinc finger protein 384 (ZNF 384), a poor 761 prognostic predictor, promotes cell growth by upregulating the expression of Cyclin D1 in Hepatocellular carcinoma. 762 Cell Death Dis. 2019; 10: 444. 763 50. Miotto B, Struhl K. HBO1 histone acetylase is a coactivator of the replication licensing factor Cdt1. Genes Dev. 764 2008; 22: 2633-8. 765 51. Miotto B, Struhl K. HBO1 histone acetylase activity is essential for DNA replication licensing and inhibited by 766 Geminin. Mol Cell. 2010; 37: 57-66. 767 52. Riedl SJ, Shi Y. Molecular mechanisms of caspase regulation during apoptosis. Nat Rev Mol Cell Biol. 2004; 5: 768 897-907. 769 53. Chen M, Wang J. Initiator caspases in apoptosis signaling pathways. Apoptosis. 2002; 7: 313-9. 770 54. Khan A, Fornes O, Stigliani A, Gheorghe M, Castro-Mondragon JA, van der Lee R, et al. JASPAR 2018: update of the 771 open-access database of transcription factor binding profiles and its web framework. Nucleic Acids Res. 2018; 46: 772 D260-D6. 773 55. Sapountzi V, Cote J. MYST-family histone acetyltransferases: beyond chromatin. Cell Mol Life Sci. 2011; 68: 774 1147-56. 775 56. Pardo M, Yu L, Shen S, Tate P, Bode D, Letney BL, et al. Myst2/Kat7 histone acetyltransferase interaction

776 proteomics reveals tumour-suppressor Niam as a novel binding partner in embryonic stem cells. Sci Rep. 2017; 7: 777 8157. 778 57. Taniue K, Hayashi T, Kamoshida Y, Kurimoto A, Takeda Y, Negishi L, et al. UHRF1-KAT7-mediated regulation of 779 TUSC3 expression via histone methylation/acetylation is critical for the proliferation of colon cancer cells. Oncogene. 780 2020; 39: 1018-30. 781 58. Bai Z, Xia X, Lu J. MicroRNA-639 is Down-Regulated in Hepatocellular Carcinoma Tumor Tissue and Inhibits 782 Proliferation and Migration of Human Hepatocellular Carcinoma Cells Through the KAT7/Wnt/beta-Catenin Pathway. 783 Med Sci Monit. 2020; 26: e919241. 784 59. Nakamoto T, Yamagata T, Sakai R, Ogawa S, Honda H, Ueno H, et al. CIZ, a zinc finger protein that interacts with 785 p130(cas) and activates the expression of matrix metalloproteinases. Mol Cell Biol. 2000; 20: 1649-58. 786 60. Mori S, Takeuchi T, Ishii Y, Kukimoto I. Identification of APOBEC3B promoter elements responsible for activation 787 by human papillomavirus type 16 E6. Biochem Biophys Res Commun. 2015; 460: 555-60. 788 61. Sakuma T, Nakamoto T, Hemmi H, Kitazawa S, Kitazawa R, Notomi T, et al. CIZ/NMP4 is expressed in B16 789 melanoma and forms a positive feedback loop with RANKL to promote migration of the melanoma cells. J Cell Physiol. 790 2012; 227: 2807-12. 791 62. Zhong CH, Prima V, Liang X, Frye C, McGavran L, Meltesen L, et al. E2A-ZNF384 and NOL1-E2A fusion created by a 792 cryptic t(12;19)(p13.3; p13.3) in acute leukemia. Leukemia. 2008; 22: 723-9. 793 794 Figure legends

795 796 Figure 1. HBO1 upregulation in human OS. The TARGET Pan-Cancer (PANCAN) database 797 shows HBO1 expression (RNASeq-TOIL RSEM) in 166 cases of children's sarcoma tissues and

798 70 cases of normal adjacent tissues (A). Kaplan Meier Survival analyses of HBO1-low (n = 53)

799 and HBO1-high (n = 113) children's sarcoma patients (B). HBO1 mRNA and protein expression

800 in OS tumor tissues (―T‖) and in normal tissues adjacent to tumor (―N‖) of ten (n = 10) primary 801 human OS patients was shown (C-E). HBO1 mRNA and protein expression in established OS 802 cell lines (U2OS and MG63), primary human OS cells (pOS-1 and pOS-2), as well as in primary

803 human osteoblasts (―Osteoblasts‖) were shown (F and G). Data were presented as mean ±

804 standard deviation (SD). * P < 0.05 vs. ―N‖ tissues/Osteoblasts (C-G).

805

806 Figure 2. HBO1 silencing inhibits OS cell progression in vitro. Stable primary OS cells (pOS-1 807 and pOS-2) and established OS cell lines (U2OS and MG63) that expressed the HBO1 shRNA 808 (shHBO1-a/shHBO1-b) or the scramble control shRNA (―shC‖) were established. After 809 cultured for applied time periods, expression of HBO1 mRNA and listed proteins was shown (A,

810 B, and I); Cell growth (C), viability (D and J), proliferation (E and K), S-phase cell percentage

811 (F), cell migration (G and L) and invasion (H) were tested by the listed assays, with results 812 quantified. For nuclear EdU staining, five random views (n = 5) of total 1, 000 cell nuclei per 813 each condition were used to calculate the average EdU ratio (% vs. DAPI, same for all EdU 24

814 studies). For ―Transwell‖ and ―Matrigel Transwell‖ assays, five random microscopy views were 815 included to calculate the average number of migrated or invaded cells in each condition (same 816 for all ―Transwell‖ assays). For the in vitro cellular functional studies, the exact same number of 817 viable cells with the applied genetic modifications was initially seeded into each well/dish 818 (―Day-0‖/―0h‖, same for all figures), and cells were cultured for indicated time periods. ―Pare‖ 819 stands for the parental control OS cells. The data were presented as mean ± standard deviation 820 (SD, n = 5). * P < 0.05 vs. ―sh-C‖ cells. The experiments were repeated five times with similar

821 results obtained. Scale bar = 100 μm (E, G and H).

822 823 Figure 3. HBO1 shRNA provokes OS cell apoptosis. Stable primary OS cells (pOS-1 and 824 pOS-2), as well as the established OS cell lines (U2OS and MG63) that expressed HBO1 shRNA 825 (shHBO1-a/shHBO1-b) or the scramble control shRNA (―shC‖) were established and cultured 826 for applied time periods, caspase activation (A, B and J), single strand DNA (ssDNA) contents

827 (C), and mitochondrial depolarization (by recording JC-1 green monomers, D and I) were

828 tested; Cell apoptosis was tested by nuclear TUNEL staining (E and K) and Annexin V FACS

829 (F) assays. Stable pOS-1 cells with shHBO1-a or shC were treated with z-DEVD-fmk (50 μM), 830 z-VAD-fmk (50 μM) or vehicle control (0.1% of DMSO), and cultured for 96h. Cell viability

831 and death were tested by CCK-8 (G) and Trypan blue staining (H) assays, respectively. For 832 nuclear TUNEL staining assays, five random views of a total 1, 000 cells per each condition 833 were included to calculate the average TUNEL ratio (% vs. DAPI, same for all TUNEL assays). 834 ―Pare‖ stands for the parental control OS cells. The data were presented as mean ± standard 835 deviation (SD, n = 5). * P < 0.05 vs. ―sh-C‖ cells. #P < 0.05 vs.―DMSO‖ treatment (G and H).

836 The experiments were repeated five times with similar results obtained. Scale bar = 100 μm (D,

837 E, I and K). 838 839 Figure 4. Ectopic HBO1 overexpression promotes OS cell growth. Stable primary OS cells 840 (pOS-1 and pOS-2) and established OS cell lines (U2OS and MG63) with lentiviral construct 841 encoding full-length HBO1 cDNA (―OE-HBO1‖) or the empty vector (―Vec‖), were established 842 and cultured for applied time periods; Expression of HBO1 mRNA and listed proteins was

843 shown (A, B and G); Cell viability (CCK-8 OD, C), proliferation (by recording EdU-positive

844 nuclei ratio, D and H), migration and invasion (―Transwell‖ assays, E, F and I) were tested by 845 the listed assays, and results were quantified. ―Pare‖ stands for the parental control OS cells. 846 The data were presented as mean ± standard deviation (SD, n = 5). * P < 0.05 vs. ―Vec‖ cells.

847 The experiments were repeated five times with similar results obtained. Scale bar = 100 μm (E).

848

849 Figure 5. ZNF384 is a transcription factor of HBO1 in OS cells. JASPAR database predicted

850 the potential transcription factors of HBO1 (A). pOS-1 cells were individually transfected with 851 200 nM of applied siRNAs or scramble control siRNA (―siC‖) for 48h, and HBO1 mRNA

852 expression was tested by qPCR (B). ChIP assay results demonstrated the direct binding between 853 ZNF384 protein and HBO1 promoter in listed human cells and tissues, with results quantified 854 (C and D). Stable OS cells, including pOS-1 cells, pOS-2 primary cells, as well as the established 855 OS cell lines (U2OS and MG63) that expressed listed ZNF384 shRNAs (―shZNF384‖) or the 856 scramble control shRNA (―shC‖), were established and expression of listed genes was shown (E,

857 F, K, and L); Cells were cultured for applied time periods; cell proliferation (EdU staining assay,

858 G and N), viability (CCK-8 OD, M), migration (H), mitochondrial depolarization (by recording

859 JC-1 green monomers intensity, I), and apoptosis (nuclear TUNEL staining assay, J) were tested. 860 ZNF384 mRNA and protein expressions in OS tumor tissues (―T‖) and in normal tissues

861 adjacent to tumor (―N‖) were shown (O and P). TCGA database showed relative ZNF384

862 transcript expression in human sarcoma tissues and normal tissues (Q). The data were presented

863 as mean ± standard deviation (SD, n = 5). * P < 0.05 vs. ―siC‖/―shC‖ cells (B, E, G-N).* P <

864 0.05 vs. ―Osteoblasts‖ (C). * P < 0.05 vs. ―N‖ tissues (D, O and P). The experiments were

865 repeated five times with similar results obtained. Scale bar = 100 μm (G-J).

866 867 Figure 6. HBO1 is required for OS xenograft growth in mice. The pOS-1 xenografts-bearing 868 SCID mice were subjected to intratumoral injection with HBO1 shRNA lentivirus

869 (shHBO1-a/shHBO1-b) or scramble control shRNA lentivirus (―shC‖); tumor volumes (A) and

870 mice body weights (D) were recorded every seven days, and estimated daily tumor growth was

871 calculated by using the described formula (B). At experimental Day-42, all tumors were

872 separated through surgery and weighted individually (C). Expression of HBO1 mRNA (E) and

873 listed proteins (F) in indicated OS xenograft tissues was shown. The stable pOS-1 cells 874 expressing the CRISPR/Cas9-HBO1-KO construct (ko-HBO1 cells) or CRISPR/Cas9 control 875 construct (―Cas9-C‖) were s.c. injected to the flanks of SCID mice. Forty days after cell

876 inoculation, tumors were separated and measured (G); Mice body weights were recorded (H);

877 Expressions of HBO1 mRNA (I) and listed proteins (J) were shown. Data were presented as

878 mean ± standard deviation (SD). * P < 0.05 vs. ―shC‖/―Cas9-C‖ tumors.

879 26

880 Figure 7. The anti-OS activity by a HBO1 inhibitor WM-3835. The pOS-1 cells were treated 881 with WM-3835 (at 5 μM, expect for A-C) or vehicle control (―Veh‖, 0.5% DMSO), and cells 882 were further cultured in complete medium for indicated time periods. Cell viability (CCK-8 OD, 883 A), expression of listed proteins (B), cell proliferation (by recording EdU-positive nuclei ratio,

884 C), migration (D), and invasion (E) were tested; Caspase activation (F and G) and cell apoptosis

885 (nuclear TUNEL staining and Annexin V FACS assays, H) were tested as well. The stable

886 HBO1-KO pOS-1 cells, koHBO1-1 and koHBO1-2 (I) or the human osteoblasts (―Osteoblasts‖,

887 J) were treated with WM-3835 (5 μM) or vehicle control for applied time periods, cell viability 888 and apoptosis were tested by CCK-8 and TUNEL staining assays, respectively. The pOS-2 889 primary cells, as well as the established OS cell lines, U2OS and MG63, were treated with 890 WM-3835 (5 μM) or vehicle control for applied time periods, cell viability, proliferation and

891 apoptosis were tested by CCK-8 (K), nuclear EdU staining (L) and TUNEL staining (M) assays, 892 respectively. The pOS-1 xenografts-bearing SCID mice were subjected to i.p. injection of

893 WM-3835 (10 mg/kg, daily for 21 days) or vehicle control (―Veh‖); tumor volumes (N) and

894 mice body weights (O) were recorded every seven days. The data were presented as mean ± 895 standard deviation (SD). * P < 0.05 vs. ―Veh‖ treatment. The in vitro experiments were repeated

896 five times with similar results obtained. ―n. s. ‖ stands for no statistical difference (I and J). 897 Scale bar = 100 μm (D and E).

Figure 1 A B Children’s sarcoma tissues 1.0

12 * 0.8

0.6 HBO1-high (n= 113) 11

Survival 0.4 Children’s sarcoma HBO1-low (n= 53) tissues (n = 166) 0.2 10 HBO1 gene expression (RNASeq-TOIL RSEM) Normal tissues P = 0.022 (n = 70) 0.0 0 50 100 150 200 (months) C D E 8 * 7 N T 6 1.2 * 5

4 0.8 Patient #1 Patient #3 Patient #6 Patient #10 Patient #1 Patient #3 Patient #6 Patient #10

3 80 kDa- HBO1 2 55 kDa- Tubulin 0.4 HBO1 (vs. Tubulin)

HBO1 mRNA (Folds vs. “N”) 1 n = 10 0 N: Normal tissues adjacent to tumor T: OS Tumor tissues n = 10 N T 0 N: Normal tissues adjacent to tumor N T T: OS Tumor tissues

F 12 G

10 * * 8

* Osteoblasts U2OS MG63 pOS-1 pOS-2 6 * 80 kDa- HBO1 4 55 kDa- Tubulin 2 Osteoblasts HBO1 mRNA (Folds vs. “Osteoblasts”) U2OS MG63 pOS-1 pOS-2 0 Figure 2

A 1.2 pOS-1 B shHBO1 C D 8 Pare shC -a -b 0.9 1 80 kDa- HBO1 Pare shC 0.8 55 kDa- Tubulin 6 shHBO1-a 0.6 shHBO1-b 0.6 11 kDa- H4K5ac 4 11 kDa- H4K12ac 0.4 * 0.3 * * 11 kDa- H4

0.2 2 Viability (CCK-8 OD) 17 kDa- H3K14ac * * * 96 h HBO1 mRNA (Fold vs. “Pare”) * Cell No. per dish (×10, 000) 0 17 kDa- H3 * 0 Pare shC -a -b 0 Pare shC -a -b shHBO1 0h 48 h 96 h shHBO1 E 40 F Cell proliferation 45 Pare shC shHBO1-a shHBO1-b 30 30 20 * *

* S-phase (%) 15 10 * EdU (% vs. DAPI) 72 h 48 h EdU/DAPI 100 μm 0 0 Pare shC -a -b Pare shC -a -b Cell migration shHBO1 shHBO1 G 120 Pare shC shHBO1-a shHBO1-b 100

60 * * 40

20 100 μm 24 h 0 Migrated cells (% vs. “Pare”) Pare shC -a -b H Cell invasion 120 shHBO1 Pare shC shHBO1-a shHBO1-b 100

60 40 * * 20 24 h 100 μm 0 Invaded cells (% vs. “Pare”) Pare shC -a -b shHBO1 120 I 1.2 J 120 K L 45 100 1 100 80 0.8 80 shHBO1 30 shC shHBO1 shHBO1 shC shC shHBO1 shHBO1 shHBO1 0.6 60 shC shC shC 60 shHBO1 shHBO1 shHBO1 * shC shC shC * 40 shHBO1 shHBO1 shHBO1 0.4 shC shC shC 40 * * * 15 *

* EdU (% vs. DAPI) * 0.2 20 * 20

* CCK-8 OD (% vs. “shC”) 96 h 72 h 24 h HBO1 mRNA (Fold vs. “shC”) 0 * * 0 0 Migrated cells (% vs. “shC”) 0 pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63 Figure 3 A B shHBO1 C Pare shC -a -b 48 h pOS-1 Clvd- 9 * 17 kDa- 0.6 * * Caspase-3 * 36 kDa- Caspase-3

115 kDa- Clvd- 0.4 6 PARP 89 kDa- PARP Clvd- 47 kDa- 3 Caspase-9 0.2 35 kDa- Caspase-9 48 h 48 h DNA breaks (ssDNA ELISA OD) 0 55 kDa- Tubulin 0 Pare shC -a -b Pare shC -a -b Caspase-3 activity (Fold vs. “Pare”) shHBO1 shHBO1 Mitochondrial depolarization D 0.6 Pare shC shHBO1-a shHBO1-b *

0.4 *

0.2

JC-1/JC-1 100 μm JC-1 intensity (488 nm) 48 h 0 Pare shC -a -b shHBO1 E Cell apoptosis F 20 20 * Pare shC shHBO1-a shHBO1-b * * *

10 10 Annexin V (%)

TUNEL (% vs. DAPI) 72 h 100 μm 72 h TUNEL/DAPI 0 0 Pare shC -a -b Pare shC -a -b shHBO1 shHBO1 G DMSO H I 0.6 z-DEVD-fmk Mitochondrial depolarization * 0.9 z-VAD-fmk 30 shC shHBO1 * *# DMSO * *# 0.4 z-DEVD-fmk * 0.6 20 z-VAD-fmk

* shC shC shC 0.2 0.3 10 shHBO1 shHBO1 shHBO1 *# *# 100 μm JC-1 intensity (488 nm) Viability (CCK-8 OD) JC-1/JC-1 48 h Cell death (Trypan blue%) 0 0 0 pOS-2 shC shHBO1-1, 72 h shC shHBO1-1, 72 h pOS-2 U2OS MG63 J K 12 Cell apoptosis 20 * * 10 shC shHBO1 15 * 8 * * 6 10

* shC shC shC 4

5 shHBO1 shHBO1 shHBO1 TUNEL (% vs. DAPI) 2 shC shC shC

shHBO1 shHBO1 shHBO1 100 μm 48 h TUNEL/DAPI 72 h 0 0 pOS-2 U2OS MG63 pOS-2 pOS-2 U2OS MG63 Caspase-3 activity (Fold vs. “shC”) Figure 4 A B C D 12 1.2 pOS-1 * 60 10 * 1 * 8 0.8 40 6 Pare Vec OE-HBO1 0.6 80 kDa- HBO1 4 0.4 20 55 kDa- Tubulin EdU (% vs. DAPI) Viability (CCK-8 OD) 2 Pare Vec OE-HBO1 0.2

Pare Vec OE-HBO1 96 h HBO1 mRNA (Fold vs. “Pare”) 0 0 Pare Vec OE-HBO1 72 h 0 E F Cell migration 1.6 * 1.8 Vec OE-HBO1 1.4 1.6 * 1.2 1.4 1 1.2 0.8 1 0.8 0.6 0.6 0.4 0.4 100 μm 0.2

Pare Vec OE-HBO1 0.2

24 h Pare Vec OE-HBO1 0 24 h Invaded cells (Folds vs. “Pare”)

Migrated cells (Folds vs. “Pare”) 0 G H I 2 9 * * 60 * * * * * * 6 40 * 1

3 20 EdU (% vs. DAPI) Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 Vec OE-HBO1 HBO1 mRNA (Folds vs. “Vec”)

72 h Migrated cells (Folds vs. “Vec”) 24 h 0 0 0 pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63 Figure 5 A

B 1.2 C D E F 9 ChIP-ZNF384 8 1.2 1 * * shZNF384 ZNF384 shC -s1 -s2 -s3 1 0.8 6 HBO1 siSP2 6 63 kDa- ZNF384 0.8 HOXA9 * 4 MYLK 0.6 * * 80 kDa- HBO1 0.6 2 n = 10 n = 10

0.4 * HBO1 promoter (Folds vs. “N”) 36 kDa- 0.4 siZNF384 3 Cyclin D1 0 0.2 * N T 0.2 * * mRNA (Fold vs. “shC”) Osteoblasts U2OS MG63 pOS-1 pOS-2 N: Normal tissues 55 kDa- Tubulin * siSOX10 siC siPITX3

HBO1 mRNA (Fold vs. “siC”) 0 *

HBO1 promoter (Folds vs. “Osteoblasts”) adjacent to tumor 0 shC shZNF384 0 T: OS Tumor tissues ChIP-ZNF384 G Cell proliferation H Cell migration shC shZNF384 40 shC shZNF384 120 100 30 80 20 60 * 40 * 10 20 24 h EdU (% vs. DAPI) 72 h 0 100 μm 0 100 μm EdU/DAPI shC shZNF384 shC shZNF384 I J Migrated cells (% vs. “shC”) Mitochondrial depolarization Cell apoptosis shC shZNF384 0.6 shC shZNF384 30 * * 0.4 20

0.2 10

48 h 72 h

TUNEL (% vs. DAPI) 0 100 μm 0 100 μm

JC-1/JC-1 JC-1 intensity (488 nm) shC shZNF384 TUNEL/DAPI shC shZNF384

K 1.2 L 1.2 M 120 N 45 1 1 100

0.8 0.8 80 shC shC shC shZNF384 shZNF384 shZNF384 30 0.6 0.6 60 shZNF384 shZNF384 shZNF384 * * shC shC shC shZNF384 shZNF384 shZNF384 0.4 0.4 shC shC shC 40 * * * shC shC shC shZNF384 shZNF384 shZNF384 * 15 0.2 0.2 * * 20 EdU (% vs. DAPI)

* CCK-8 OD (% vs. “shC”) * 96 h 72 h 0 * * HBO1 mRNA (Folds vs. “shC”) 0 0 0 ZNF384 mRNA (Folds vs. “shC”) pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63

ZNF384 expression O P Patient #1 Patient #3 Q 3 N T N T * 63 kD- ZNF384 2 37 kD- GAPDH

1 Patient #6 Patient #10 N T N T

ZNF384 mRNA (Folds vs. “N”) n = 10 0 63 kD- ZNF384 N T Transcript per million N: Normal tissues 37 kD- GAPDH adjacent to tumor Normal tissues Sarcoma tissues T: OS Tumor tissues N: Normal tissues adjacent to tumor (n = 2) (n = 260) T: OS Tumor tissues TCGA samples Figure 6 A B C D 30 1200 shC Day-42 21 3 shC

3 1200 shHBO1-a shHBO1-a shHBO1-b 20 20 800 shHBO1-b

800 * 19 10 * 400 * 400 * *

Tumor weights (mg) 18 Mice body weights (g) Tumor volume (mm ) *

n = 10 n = 10 n = 10 n = 10 0 0 0 17

Estimated tumor growth (mm /day) -a -b -a -b Day-0 7 14 21 28 35 42 shC shC Day-0 7 14 21 28 35 42 shHBO1 shHBO1 shHBO1 shHBO1 E F shC -a -b shC -a -b 1.2 Day-7 1.2 Day-14 80 kDa- HBO1 HBO1 1 1 11 kDa- H4K12ac H4K12ac

0.8 0.8 11 kDa- H4 H4 0.6 0.6 17 kDa- H3 H3

89 kDa- Clvd- Clvd- 0.4 0.4 PARP PARP 17 kDa- Clvd- Clvd- 0.2 0.2 Caspase-3 Caspase-3

HBO1 mRNA (Folds vs. “shC”) 55 kDa- HBO1 mRNA (Folds vs. “shC”) * * Tubulin Tubulin 0 * 0 * shC -a -b shC -a -b Day-7 Day-14 shHBO1 shHBO1 G H I J 800 20 1.2 Cas9-C koHBO1 80 kDa- HBO1

3 1 600 15 11 kDa- H4K12ac 0.8 11 kDa- H4

400 10 0.6 17 kDa- H3 Clvd- 89 kDa- * 0.4 PARP

200 Mice body weights (g) 5

Tumor volume (mm ) Clvd- 0.2 17 kDa- n = 7 Caspase-3 n = 7 n = 7 55 kDa- Tubulin 0 0 HBO1 mRNA (Folds vs. “Cas9-C”) 0 * Cas9-C koHBO1 Cas9-C koHBO1 Cas9-C koHBO1 40 days after cell inoculation Figure 7

A 120 B C pOS-1 WM-3835 40 100 Veh 1 5 10 25 μM 80 kDa- HBO1 80 * 30 * * 11 kDa- H4K12ac 60 * * 11 kDa- H4 20 40 24 h * * 17 kDa- 48 h* * H3K14ac * 10 * 20 EdU (% vs. DAPI) 72 h 17 kDa- CCK-8 OD (% vs. “Veh”) * H3 96 h * * 72 h 0 * 55 kDa- Tubulin 0 Veh 1 5 10 25 μM Veh 1 5 10 25 μM WM-3835 WM-3835 120 120 D Cell migration E Cell invasion Veh WM-3835 100 Veh WM-3835 100 80 80 60 60 40 * 40 20 20 * 100 μm 24 h 100 μm 24 h 0 Invaded cells (% vs. “Veh”) 0 Migrated cells (% vs. “Veh”) Veh WM-3835 Veh WM-3835 F 9 G H Veh WM-3835 30 * Veh WM-3835, 48h * * 20 PARP 6 20 Clvd- PARP 10 3 Caspase-3 10 Annexin V (%) Caspase-3 activity (Folds vs. “Veh”)

Clvd- TUNEL (% vs. DAPI) Propidium Iodide (PI) 48 h Caspase-3 72 h 72 h 0 Tubulin 0 0 Veh WM-3835 Veh WM-3835 Annexin V Annexin V Veh WM-3835 I J Osteoblasts N n. s. 30 10 120 3 n. s. n. s. n. s. n. s. 1200 Veh 0.6 100 8 n. s. WM-3835 80 20 800 0.4 6 60 4 *

40 10 CCK-8 OD 400 0.2 2 TUNEL (% vs. DAPI) Tumor volume (mm )

20 TUNEL (% vs. DAPI) WM-3835 WM-3835 Veh Veh Veh Veh WM-3835 WM-3835 n = 7

CCK-8 OD (% vs. “Veh”) 72 h 96 h 72 h 0 96 h 0 0 0 0 koHBO1-1 koHBO1-2 koHBO1-1 koHBO1-2 Veh WM-3835 Veh WM-3835 Day-0 7 14 21 28 35 42 K L M O 120 45 30 21 100 * Veh 20 WM-3835 80 30 20 * *

60 Veh WM-3835 Veh WM-3835 Veh WM-3835 19 WM-3835 WM-3835 WM-3835 * Veh Veh Veh 40 15 * * 10 18

* Veh Veh Veh TUNEL (% vs. DAPI)

EdU (% vs. DAPI) * 20 * WM-3835 WM-3835 WM-3835 CCK-8 OD (% vs. “Veh”) Mice body weights (g) 96 h 72 h 72 h n = 7 0 0 0 17 pOS-2 U2OS MG63 pOS-2 U2OS MG63 pOS-2 U2OS MG63 Day-0 7 14 21 28 35 42 Table S1: sequences utilized in this study.

qPCR primers Forward5’-3’ Reverse 5’-3’ Hoxa9 AGAATGAGAGCGGCGGAGACAA CTCTTTCTCCAGTTCCAGGGTC HBO1 TTTTGGCCGCTATGAACTG GGAGGATTGTCTGGCTCTTC GAPDH GTCTCCTCTGACTTCAACAGCG ACCACCCTGTTGCTGTAGCCAA MYLK GAGGTGCTTCAGAATGAGGACG GCATCAGTGACACCTGGCAACT shRNA shHBO1-a CCGGGCCTTCTTCTGAGTCTGACATCTCGAGATGTCAGACTCAGAAGAAGGCTTTTT shHBO1-b CCGGCTCGTTCATCTGGTTCAGAAACTCGAGTTTCTGAACCAGATGAACGAGTTTTT shZNF384-s1 CCGGGCCTTCACACAACTCTCCAATCTCGAGATTGGAGAGTTGTGTGAAGGCTTTTTG shZNF384-s2 CCGGGATCGAGAACACAATGTTCATCTCGAGATGAACATTGTGTTCTCGATCTTTTT shZNF384-s3 CCGGCGGCAACACAACAAAGATAAACTCGAGTTTATCTTTGTTGTGTTGCCGTTTTTG siRNA SOX10 CCGUAUGCAGCACAAGAAA SP2 UGCAGACCAUCAACAUCAA ZNF 384 AAUUACUUUAUAACAACUGGU PITX3 UUUAUUUCAUUUAUCUUUGAA

HBO1 promoter region

GCAGGAGAATCGCCTGAACCCGGGAGGCGGAGGCTGCGGTGAGCCGAGCTCGTGCCATTGC CCTCCAGCCTGGCAACAAGAGCAAAACCCTGTCTCAGAAAAAAAAAAAAAAAAAGAAAGGC TGGTATGGCTGGGAGTGGTGGCTCACACCTTAATCCCAGCACTCTGGAAGGCCGAGACAGTA GTTCAGCTCAGTAGGTCGAGACCAGCCTGGGCAGCATGGTGAAACTCTGTCTCTACAAAAAAT ACAAAAATTAGCCAGGTGTGGTGGCACACGCCTGCAGTCCCAGCTACTCGGGAGGCTGAGGC AAGAGAATCCCTTGAGCCCACGAGGTGGAGCTTGCAGTGAACTGACATCCAGCCATTGCACT CCAGCCTGGGCAACACAGCAAGACTCTGTCTTGAAAGAAAAAAAAAGAAGAGAGAGAAAAA AAGTAGGATGGTATGAAATAACTGAGATATGCTCTTAATGTAAGAGATCCTTACCACCAGTTGG GGAGCCAGAACATTCAAATGTAACCTCAGGAACATATCCGTGAGTGCAAAATAACAGAGTAAT GTATGAAATTAAATGCTTAAGTGGTGAGGGGCTACATGAAATTAAATGCTTAAGTGGTGAGAA GTAATGTGAACCATTTTCAAATAAGTTCTTACGTGCAGTTCAACCTAAGTAGGTTATCAAAGGT ATTTATACTAAGCACTAAAAGCTAAGTATTAAATACTAAAAGCTGCCTGTGTCAACACTGGCCC AGGACATCCCCAAGGAGAGTGCTGGCTAAAGGTCTGGAACTTTAAATCTTTAGAGTTCTTTAC TTAGCACATCAATCAGATATTTACCTTCTGCTTCTTTGTTCTTTAGCCCTGCTCCTGGGGGAAAT CTCCATCTTAACGGGGCCTCAGTTTGAAATTGGGGTTCGGGATCCTAACTTAATGACTGAGCG TGCCTGGGTTCCACCCAGTCGCAATTGGTCAAGACTTGGTCCTCTCAAGGCCTTTCCTCCGGA GAGGTTAAATCCTCAACGCCTGCTCTCCACTTTCCAGCCTTTTCTTAAGGCTCTTCTTTCAGCCT AGTATCTGTGACCAACTTCTCCTTCCTAGGCCCTCTCATTATTATTTAAATGTCCATTCTTTTGGG GGCTTCAACTTCACTAAAAACATTCCTTCCTCTTCCTCCGCCTCGTAACTCTCTGCTCAAGCATC TGGATGCCAGAGGTCATTTGAGCTAAATGGGGTTCAAAGCTTTTCTGACTCTGGTACTGCATTC CCCACTTTCAAATCACATTTTTCACCCCAGGGCCCCCAGTGGTCTCAAACCCAAACACAGTGC GGTTAGGAAGTGGGCACTCGCCACAGAGCTAAATTGGGAATTATTTGACCTGCAGCTTCTGGA GAATGGTTTTGCCTCCTCCCCGCGACATCAACTAACTCACTTGCTGTTAGCTAAGGGCAAGGA CAATGTCTCCTCTACCAGACAGGAAACTTGATGGCGGACGCTATCCGTTAGTTCCTTTTTCAAT TCCCCATTCCCACCCCTTCGAGACCACGATGTTTCTTAAAATACAAAGTTTCTTTTACAAAACAT CCCTCCGCAACTTTGAGTTTATATAATAAACGGAAACGTCCTTTAAACAATCAGGCTACGAAG GGGTTAAATCTAATCTCAGCAACCTGATGGATTCTGGGGGGTTTCCTTCCTCGACTCTAGAAAT TAAAACGAAATTTTGCCCAAGGTAAAGATGAGAACCGTTTACATCAACTGTCAGGCTTCTGCC CCAAACAAGAATGGCTACCCACGCCCACAAACAGAGCCACTCACGTAACTAAAAATGGTTGG CTCCGGGTTCCACCACTGGAGTCACTTTCTGCCCAACTTCAAAATGGCGCCCACTAGCTTCACC AAAAGGATCGAACTTCCCAGCACCCTCTCTGGCCCGTAACGTCAGGTGACGCGAGACCCAGC CGGAAGTGAAGGAAAAAGCGCTTCAGCCCGCGGCGCCTGCGCAGAACGCTCCAGACGCTGA GAGGCAGGAGGCACTAGGGATCGTCCGCAGGATTGGGACTGATACAGAGGCCGCCACGGAG CCCGCCGGAGCCACCGTT

Figure S1. Expressions of MYLK-HOXA9 mRNA and listed proteins in primary OS cells (pOS-1 and pOS-2) and established OS cell lines (U2OS and MG63), with described genetic modifications, were tested by qPCR and Western blotting analyses (A-G). ZNF384 mRNA and protein expressions in established OS cell lines (U2OS and MG63), primary human OS cells (pOS-1 and pOS-2), as well as in primary human osteoblasts (“Osteoblasts”) were shown (H). The pOS-1 cells were treated with WM-3835 (5 μM) or vehicle control for 24h, MYLK-HOXA9 mRNA expression was tested (I). The data were presented as mean ± standard deviation (SD, n=5). * P< 0.05 vs. “Pare”/“Cas9-C” cells, “Osteoblasts”, or “Veh” treatment. The experiments were repeated five times with similar results obtained.

Figure S2. Stable primary OS cells (pOS-1 and pOS-2) and established OS cell lines (U2OS and MG63) with the lentiviral CRISPR/Cas9-HBO1-KO construct “koHBO1”, or the CRISPR/Cas9 control construct “Cas9-C”, were established. Expression of HBO1 mRNA and listed proteins was shown (A, B and J); Cells were cultured for applied time periods; cell viability (C and K), proliferation (by recording EdU-positive nuclei ratio, D and L), and migration (E and M) were tested by the assays mentioned, with data quantified. The caspase-PARP activation (F and G), single strand DNA (ssDNA) contents (H), and cell apoptosis (by recording nuclear TUNEL ratio, I) were tested as well. The data were presented as mean ± standard deviation (SD, n=5). * P< 0.05 vs. “Cas9-C” cells. The experiments were repeated five times with similar results obtained. Scale bar= 100 μm (E).

Figure S3. Stable pOS-1 cells expressing ZNF384 shRNA (“shZNF384”) were further transduced with or without the lentiviral construct encoding full-length HBO1 cDNA (“OE-HBO1”), and control cells were with the scramble control shRNA (“shC”). Expressions of listed proteins were shown (A); Cells were cultured for applied time periods; cell proliferation (EdU staining assay, B), migration (C), mitochondrial depolarization (by recording JC-1 green monomers intensity, D), and apoptosis (nuclear TUNEL staining assay, E) were tested. #P< 0.05. The experiments were repeated five times with similar results obtained.