Human Saos-2 and U-2OS OS cells were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The cell lines were routinely cultured in DMEM (#8119283, Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, #10270-106, Gibco), 100 μg/mL streptomycin, and 100 U/mL penicillin at 37°C in an incubator with a constant airflow of 5% CO₂ and 95% O₂. As for reverse test, the above-mentioned medium was supplemented with extra 400 μM serine (#GA10157, GlpBio Technology, CA, USA) in the serine supplementation group. AIL (#GC31742, molecular weight, 376.4, purity > 98.00%, GlpBio Technology) was dissolved to 10mM in dimethyl sulfoxide (DMSO, #D4540, Sigma-Aldrich, St. Louis, MO, USA) and stored at -80°C. The final concentrations were 0.039-10 μM. Cell proliferation assay was performed using Cell Counting Kit-8 (CCK-8, #96992, Dojindo Molecular Technologies, Dojindo, Japan). The absorbance value was measured at a wavelength of 450 nm, with the reference wavelength set at 630 nm. Data were collected from three independent experiments, and the AIL-induced cell growth inhibition rate was calculated by comparison to untreated control cells.

The migration ability of OS cells was detected using a wound healing assay as outlined previously (23). Briefly, OS cells in a 6-well plate were carefully scratched using 10 μL sterile pipette tips. Subsequently, the cells were washed twice and were cultured in DMEM with different concentrations (0, 0.01, 0.05 and 0.1 μM) of AIL for 24 h. Images were then taken to determine the width of the wound using a microscope (10× objective) (1×71, Olympus).

Transwell assays were performed using Boyden chambers (Corning Inc., Corning, NY, USA) as previously described (24). Briefly, OS cells were treated with AIL at various concentrations for 24h. Next, 1×10⁵ cells viable cells (confirmed by trypan blue exclusion) from each well were seeded in serum-free medium onto the Matrigel-coated upper chamber, while complete medium was added in the lower compartment as a chemoattractant. After incubation of 24 h, the cells that migrated through the filter were fixed and stained (23). The invading cells were evaluated in five random fields under the microscope, and images were obtained.

OS cells were seeded in 6-cm plates and incubated with different concentrations of AIL. After 24 h, all the cells including the floating and adherent ones were harvested and stained with Annexin V-fluorescein isothiocyanate (FITC) and propidium iodide (PI) apoptosis detection kit (#556547, BD Biosciences Pharmingen, San Diego, CA, USA) according to the manufacturer’s protocols. The resulting fluorescence was assessed by fluorescence-activated cell sorting scan (FACS) flow cytometry (Becton Dickinson, Mountain View, CA, USA) as reported previously (25).

After being treated with different concentrations of AIL for 24 h, the OS cells were harvested and fixed with 70% ethanol overnight at -20°C and stained with PI. The DNA content was detected using a FACS flow cytometry (Becton Dickinson) according to the manufacturer’s protocols. The cell cycle distribution was determined using FlowJo software 8.7.1 (Tree Star, Inc., Ashland, OR, USA).

Tandem mass tag (TMT)-based quantitative proteomic analysis was carried out on Saos-2 cells treated with IC50 (0.20 μM) of AIL or DMSO for 24 h (3 biological replicates each group). A schematic description of the experimental design and data process strategy is presented in Figure 5A . SDT (4%SDS, 100mM Tris-HCl, 1mM DTT, pH7.6) buffer was used for cell lysis and protein extraction. Quantitative proteomic analysis was performed at Shanghai Applied Protein Technology Co., Ltd as described previously (26). Briefly, 100 μg peptide of each sample was labeled using TMT-6-plex Isobaric Label Reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Then, the labeled peptides were separated into 10 fractions using the High pH Reversed-Phase Peptide Fractionation Kit (Thermo Fisher Scientific). Each fraction was separated with EASY-nLC 1000 Liquid Chromatograph (Thermo Fisher Scientific) and then subjected to LC-MS/MS analysis on a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific) for 60 min. The MS raw data for each sample were searched using the MASCOT engine (version 2.2, Matrix Science, London, UK) embedded into Proteome Discoverer 1.4 software (Thermo Fisher Scientific) against the SwissProt human proteome database (20395 items, released on 2021.01.06) concatenated with reverse decoy database for identification and quantitation analysis. All identified proteins were determined using a false discovery rate (FDR) threshold of ≤ 0.01. The methods for protein quantification were as followed: first, the ratio of label signals was output and calculated as a ratio to the reference sample; secondly, the necessary normalization on protein median was performed to eliminate experimental error utilizing Correcting Experimental Bias function with built-in settings in Proteome Discoverer 1.4 software (Thermo Fisher Scientific); finally, the protein is quantified based on the median of the relative intensity of the unique peptide of each protein.

Cell culture and AIL treatment were the same as those in quantitative proteomic analysis. RNA samples were prepared using TRIzol (#15596026, Invitrogen) and subjected to RNA-seq analysis by Genome Center, WuXi App Tec as reported previously (27). Generally, 1 μg of qualified RNA (RIN > 8.0) was used as input for library construction following the Illumina TruSeq RNA Sample Preparation protocol. Then, RNA libraries were sequenced on an Illumina NovaSeq 6000 platform, PE 2×150bp. The average data yield for each sample was 20M PE reads with % of Q30 bases > 90. Subsequently, RNA-seq reads of each sample were aligned to the human genome (build 38) using short reads aligner STAR (version 2.5.1b). Gene annotation and quantification were then performed using RSEM24 with GENCODE annotation (release 24: http://www.gencodegenes.org).

Untargeted metabolomics profiling of Saos-2 cells treated with AIL or DMSO was performed using ultra-high-performance liquid chromatography (1290 Infinity LC, Agilent Technologies, USA) coupled with a quadrupole time-of-flight system (AB Sciex TripleTOF 6600, AB SCIEX, USA) at Shanghai Applied Protein Technology Co., Ltd. Briefly, the cell samples were collected in 15-mL Vacutainer tubes and then centrifuged for 15 minutes. Aliquots of the samples were stored at −80°C for use. Then the samples were thawed at 4°C, and 100-μL aliquots were mixed with 400 μL of cold methanol/acetonitrile (1:1, v/v) to remove the protein. The mixture was centrifuged for 20 minutes. The supernatants were dried in a vacuum centrifuge. Detailed experimental procedures for mass spectrometry and data process methods are described previously (28). After normalization to the total peak intensity, the data were exported to the SIMCA-P software for multivariate data analysis to screen for differential metabolites.

The mRNA levels of PHGDH, phosphoserine aminotransferase 1 (PSAT1) and phosphoserine phosphatase (PSPH) were detected using a PrimeScript RNA RT-PCR Kit (#Q711-03, Vazyme Biotech co., ltd, Nanjing, China) according to the manufacturer’s protocol. The primer pairs are shown in Table 1 . The relative mRNA levels were calculated using the 2^−ΔΔCt method.

Expression of proteins in OS cells was quantified using a BCA Protein Assay Reagent Kit (#23228, Thermo Fisher Scientific) as described previously (23). The rabbit polyclonal cleaved-Caspase-3 (#9664), BCL-2 (#3498), BAX (#5023), cyclinD1 (#55506) and CDK4 (#12790) antibodies were purchased from Cell Signaling Technology (Boston, USA). The PHGDH (#14719-1-AP), PSAT1 (#10501-1-AP) and PSPH (#14513-1-AP) antibodies were purchased from Proteintech Group, Inc. (Chicago, IL, USA), and the band density was quantified using ImageJ software 1.8.0 (National Institutes of Health, Bethesda, MD, USA).

The PHGDH overexpression pcDNA3.1 plasmid was constructed by Genechem Co., Ltd. (Shanghai, China). The plasmid was transfected into Saos-2 and U-2OS cells using Lipofectamine 3000 (#L3000015, Invitrogen) following the manufacturer’s protocol.

The animal study was performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals approved by the Scientific Investigation Board of Shanghai Jiao Tong University Affiliated Sixth People’s Hospital. 5×10⁶ Saos-2 cells were suspended in 75 μL serum-free DMEM and combined with 75 μL of Matrigel (#356235, Corning, NY, USA), and then the mixture was injected into the right hind thigh of 6-8 weeks NOD-Prkdc^scid Il2rg^em1/Smoc (NSG) mice (Model organisms center, Shanghai, China). Cell line-derived xenografts were allowed to grow to an average volume of 100 mm³, then mice were intraperitoneally injected with AIL (2mg kg^-1) or DMSO (as controls) every 3 days. Length and width of resulting tumors as well as weight of mice were measured every 5 days, and tumor volumes were calculated using the formula (length×width²)/2. Mice were sacrificed and tumors were harvested 30 days after treatment initiation.

For proteomics analysis, Perseus software (v1.6.5.0) was used. Differentially ex-pressed proteins (DEPs) were filtered at P < 0.05 and with a fold change of ≥ 1.2 or ≤ 0.83. Gene Ontology (GO) annotation for DEPs was performed using the DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/). Moreover, proteome and transcriptome data were subjected to Gene Set Enrichment Analysis (GSEA) separately based on the whole gene set with the rank list of all the available expression values. C2.cp.kegg from the Molecular Signatures Database (MSigDB, v7.4) were selected as gene sets for the KEGG pathway enrichment analysis. Significantly enriched pathways were selected at a level of 0.05 norminal P-value and 0.25 FDR q-value. Subsequently, enriched pathways from proteome and transcriptome data were combined to determine the pathway that was most significantly affected by AIL treatment.

For metabolomics data, differentially expressed metabolites were determined using an unpaired Student’s t-test. The variable importance in the projection (VIP) value >1 and P < 0.05 was considered as statistically significant. The above-obtained metabolites were blasted against the online KEGG database to retrieve their Compounds and were subsequently subjected to enrichment analysis. Only pathways with P-values under a threshold of 0.05 were considered as significantly changed pathways.

The non-omics data analysis was performed using GraphPad Prism 7 (GraphPad Software Inc., CA, USA). P values were two-tailed, and P < 0.05 was considered statistically significant. Continuous variables are presented as the mean ± SD.

The proliferation assay revealed that AIL significantly inhibited the proliferation of Saos-2 and U-2OS cells in a dose and time-dependent manner ( Figure 1A ). The IC50 at 24 h was determined to be 0.20 and 0.13 μM for the Saos-2 and U-2OS cells, respectively. Wound healing assays and Transwell assays were performed to determine the migratory and invasive abilities of OS cells. As shown in Figures 1B, D , compared with the migration rates of the control group, the migration rates of the Saos-2 and U-2OS cells treated with AIL were decreased dose-dependently. In the Transwell assay, AIL dose-dependently suppressed OS cell invasion ( Figures 1C, D ).

To evaluated the efficacy of AIL in vivo, we treated Saos-2 xenografts in nude mice with AIL every 3 days. As a result, AIL significantly inhibited the increase of tumor volume in Saos-2 xenografts ( Figure 2A ). Mice were sacrificed after treatment, the mean tumor volume in AIL-treated mice was significantly lower than that in DMSO-treated mice ( Figures 2B, C ). In addition, there was no significant difference in mice weight between the two groups ( Figure 2D ), indicating that the dose of AIL was well tolerated.

Since AIL showed a significant inhibitory effect on OS cell growth both in vitro and in vivo, we next explored the mechanism of action. As a result, AIL induced apoptosis in Saos-2 and U-2OS cells dose-dependently, which was supported by the downregulation of BCL-2 protein and upregulation of cleaved-Caspase-3 and BAX protein by Western blot assay ( Figure 3 ). Cell cycle analysis showed that the S phase was significantly decreased, while the G1 phase was significantly increased in Saos-2 and U-2OS cells after AIL treatment ( Figures 4A, B ). G1-S transition related protein cyclinD1 and CDK4 were downregulated dose-dependently ( Figures 4C, D ). The results indicated that AIL blocked G1-S transition in OS cells in a dose-dependent manner.

To investigate the possible network related to the inhibitory effect of AIL on OS cells, we performed a comparative proteomic analysis in Saos-2 cells treated with AIL or DMSO. In total, 6860 unique proteins (5684 proteins identified with at least two peptide fragments) were identified with high confidence (FDR ≤ 0.01). Among them, 6831 proteins were quantifiable (5674 proteins were quantifiable with at least two peptide fragments). 899 DEPs (342 upregulated and 557 downregulated, Supplementary Table 1 ) were filtered at P < 0.05 and fold change ≥ 1.2 or ≤ 0.83. The classification of DEPs by GO database annotation showed that 617 (68.6%) proteins are involved in the metabolic process ( Figure 5B ), suggesting that AIL treatment leads to widespread changes in metabolic programs in OS cells. To identify the specific signaling pathways in detail, GSEA was performed. As a result, 28 KEGG pathways were significantly enriched ( Figure 5C ). Furthermore, we performed RNA-Seq analysis to interrogate changes at transcript level in OS cells exposed to AIL. GSEA analysis revealed that 30 KEGG pathways were significantly enriched ( Figure 5D ). Then we combined the significantly enriched pathways from proteome and transcriptome data, we found out that 7 KEGG pathways were common ( Figure 5E ). Among the 7 common pathways, the serine biosynthetic or metabolic pathway was the pathway most significantly changed ( Figures 5C–G ). Notably, the GSEA Blue-Pink O’gram ( Figure 5H ) revealed that PHGDH, PSPH and PSAT1, the three rate-limiting enzymes in SSP, are the top three downregulated proteins. These bioinformatic results suggested that the SSP is a key target of AIL in OS cells. Serine is a precursor to folate, which is the principal donor of one-carbon units (29, 30), it serves as a major source of metabolic intermediates that are required for nucleotide synthesis. Serine is also the precursor to sphingolipids (29, 31–33). Interestingly, GSEA results indicated that the one-carbon pool by folate, DNA replication and sphingolipid metabolism pathways were among the 7 common pathways downregulated by AIL treatment ( Figures 5E, I–K ). These data further support the connection between the effects of AIL and the SSP.

To validate the changes of SSP expression induced by AIL, we measured the expression of mRNA and protein of the SSP in Saos-2 and U-2OS cells using qRT-PCR and Western blot assay, respectively. In line with the RNA-Seq-based gene expression, all of the PHGDH, PSAT1 and PSPH mRNA was downregulated significantly in both of the two cell lines after AIL treatment ( Figure 6A ). Similarly, protein expression of each enzyme in the SSP was confirmed to be downregulated by AIL dose-dependently ( Figures 6B, C ).

Since the above-mentioned data suggested that the effects of AIL are associated with changes in metabolic programs, untargeted metabolomics with high resolution, high throughput, and highly sensitive technology was performed to execute a large‐scale detection of the metabolite features after AIL treatment ( Figure 7A ). As shown in Figure 7B , 21 KEGG pathways were significantly enriched, out of which, 13 were involved in metabolism and Supplementary Table 2 . To be noted, glycine serine and threonine metabolism was one of the pathways that are changed the most. And then, we asked about the change of intermediate metabolites of the SSP. As expected, the relative levels of 3-phosphohydroxypyruvate, 3-phosphoserine and serine in OS cells were reduced significantly after AIL treatment ( Figures 7A, C ). Serine is important in metabolism in that it participates in the biosynthesis of purines, pyrimidines and sphingolipids. Interestingly, the intermediate metabolites of the downstream pathways of the SSP including the purine metabolism, pyrimidine metabolism and sphingolipid metabolism were also downregulated significantly ( Figure 7C ). These results are in line with the data at mRNA and protein levels. Taken together, we proposed that AIL inhibited proliferation, migration and invasion of OS cells by downregulating the SSP, further downregulating the pathways of one-carbon pool by folate, purine metabolism, pyrimidine metabolism, DNA replication and sphingolipid metabolism ( Figure 7D ).

To further determine the functional importance of the SSP in AIL effects, the PHGDH in OS cells were overexpressed by the treatment with IC50 of AIL ( Figures 8A, B ). Besides, we supplemented exogenous serine in the media to explore if downregulation of serine biosynthesis by AIL can be compensated for with exogenous serine. As expected, the viability of Saos-2 and U-2OS cells significantly increased in the PHGDH overexpression group compared with that in the negative control group. However, the inhibitory effect of AIL on OS cell proliferation can’t be reversed by exogenous serine supplementation. Moreover, exogenous serine supplementation didn’t affect the reverse effect of PHGDH overexpression ( Figure 8C ). Similarly, the wound healing assay and Transwell assay revealed that overexpression of PHGDH but not exogenous serine supplementation attenuated the suppression of AIL on the migration and invasion of OS cells ( Figures 8D–F ). These data demonstrated that AIL suppresses OS cells through downregulating the serine de novo biosynthesis but not exogenous serine importation.