The OS dataset GSE126209 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE126209) was retrieved from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) and includes 12 tumor samples along with 11 matched adjacent non-tumor tissues. Expression data for 40 kinesin family genes were extracted from this dataset. Differential gene expression between tumor and adjacent tissues was analyzed using R software (v4.3.2). Data visualization was performed with the aid of the ggplot2 (v3.5.2) and ggpubr (v0.6.0) packages in R. Clinical and gene expression data from 89 OS patients were retrieved from the TARGET database (https://www.cancer.gov/ccg/research/genome-sequencing/target), among which survival records were accessible for 86 individuals. Data integration was performed using customized computational scripts developed by the authors. Genes with an average expression level below 0.1 in the integrated transcriptomic dataset were excluded due to their limited biological relevance. The classification of mRNA and lncRNA was conducted using the human genome reference files downloaded from the ENSEMBL database (v113, https://www.ensembl.org). The interaction information between different proteins is directly analyzed online from the STRING database (v12.0, http://www.string-db.org) and visualized using Cytoscape software (v3.10.2, https://cytoscape.org/) [32].

Gene Set Enrichment Analysis (GSEA) is an algorithm to test if pre-defined gene sets are differentially and co-expressed significantly in two different phenotypes or conditions [33]. This method orders genes across a dataset by their relevance to a particular phenotype or state, and then it evaluates how genes from specific predefined sets are distributed within this ordered list to pinpoint significant overrepresentation. In our research, we utilized GSEA to conduct our biological enrichment analysis, leveraging gene sets from both the MSigDB hallmark gene set collection (v2024.1.Hs, https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) and Gene Ontology (GO) database (v2024-01-17, https://geneontology.org/). Patients with OS were divided into high- and low-KIF18A groups, differential expression analysis was performed and ranked gene list was selected as an input for GSEA.

The CIBERSORT (https://cibersortx.stanford.edu/) algorithm employs a computational technique to deduce the cellular composition within complex tissues through analysis of their gene expression profiles [34]. By using a distinct collection of gene expression markers unique to various cell types, it precisely assesses their relative abundances in a given sample. For our research, we utilized CIBERSORT to perform the immune microenvironment analysis, which allowed us to unravel the detailed constitution of immune cells present in our samples, offering a deeper understanding of the immune dynamics within the disease context examined.

Single-cell analysis was conducted using data from the Tumor Immune Single-cell Hub (TISCH2, http://tisch.comp-genomics.org/), a publicly available resource designed for the study of tissue-specific cellular heterogeneity [35]. The comprehensive data are given on cell identification, characterization, interaction network that facilitate the systematic study of cellular heterogeneity and how it is regulated in cellular processes. A drug sensitivity analysis was done on the data provided from the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org/) [36]. The Tumor Immune Dysfunction and Exclusion (TIDE) database (http://tide.dfci.harvard.edu/) was used for the quantification of the response rate of OS patients on the immunotherapy [37].

In total, 79 OS tissues and their matched non-cancerous adjacent tissues were collected from the patients undergoing surgical resection of Ningbo No. 6 Hospital within the years ranging from 2014 to 2023. After sectioning, tissue samples were snap-frozen in liquid nitrogen on site, stored at –80 °C and used thereafter. The participants that we included in the study have not yet undergone preoperative radiotherapy or chemotherapy. The written informed consents were provided from all the participants before collection. The utilization of human tissues and related procedures were reviewed and approved by the Ethics Committee of Ningbo No. 6 Hospital (Approval No.: 2025003). The hFOB, MG63, U2OS, and Saos-2 cell lines are obtained from the Cell Bank of Shanghai Branch of Chinese Academy of Sciences, which are cultured in Dulbecco’s modified Eagle medium (DMEM; Gibco BRL, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% antibiotics (Gibco) to support their growth and viability. Cells are grown in 37 °C, 5% CO₂ and 95% relative humidity controlled environment. All cell lines used here are identified by short tandem repeat (STR) profiling and tested mycoplasma negative.

Total RNA was isolated from the tissues and cultured cells with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and oncolumn DNase I treatment was performed to eliminate genomic DNA contamination. The isolated RNA was following reverse transcribed into complementary DNA (cDNA) with the PrimeScript® RT Reagent Kit (TaKaRa, Dalian, China) following the manufactures recommended protocol. Quantitative real-time PCR (qRT-PCR) assays were conducted using Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) on real-time PCR detection system. Primer sequences are listed as follows: KIF18A, 5^′-TGGACTTACTTTACACCAGCCC-3^′ (forward) and 5^′-GCTGTTTTGTCTTGTTGTCGC-3^′ (reverse); GAPDH, 5^′-GGAGCGAGATCCCTCCAAAAT-3^′ (forward) and 5^′-GGCTGTTGTCATACTTCTCATGG-3^′ (reverse). The mRNA abundance was normalized to GAPDH and was determined through 2^{-Δ⁢Δ⁢Ct} method.

Two effective short hairpin RNAs (shRNAs) to specifically knock down KIF18A (sh-KIF18A#1 and #2) were constructed by shRNAs from Beijing Tsingke Biotech Co., Ltd. (Beiing, China) and the non-targeting shRNA (sh-NC) as negative control. Lentiviral particles were produced in 293T cells using a three-plasmid packaging construct, the shRNA-expressing plasmids being co-transfected with lentiviral packaging plasmids by Lipofectamine™ 3000 Transfection Reagent (Invitrogen, Waltham, MA, USA). Viral supernatants were harvested 48 h later. The lentiviral particles were used to infect the OS cells in 4 µg/mL polybrene (Sigma-Aldrich, St. Louis, MO, USA) for 72 h, and were selected with 1 µg/mL puromycin for 1 week to obtained the stable KIF18A-knockdown cell lines. KIF18A overexpression was transfected with pcDNA3.1-KIF18A expression plasmid and empty pcDNA3.1 vector (Hanbio Biotechnology, Shanghai, China), respectively with Lipofectamine^TM 3000. We confirm the efficiency of KIF18A knockdown and overexpression by quantitative real-time PCR (qRT-PCR) and Western blot analysis. Sequences of the target shRNA was described below:

sh-KIF18A#1: CCCGATTTGTAGAAGGCACAA

sh-KIF18A#2: CGCTTGTTAAAGGATTCTCTT

Cell proliferation was assessed by Cell Counting Kit-8 (CCK-8) assay, colony formation assay and 5-ethynyl-2^′-deoxyuridine (EdU) incorporation assay. In the CCK-8 assay, transfected OS cells (2 $\times$ 10³ cells per well) were seeded in 96 well microplates (Greiner, Frickenhausen, Germany) and then incubated with the CCK-8 reagent (Dojindo, Shanghai, China) at 10 µL per well for 24, 48 and 72 hrs. The colorimetric changes were measured using a microplate reader (Tecan, Sunrise, Austria) at 450 nm, with higher absorbance values reflecting increased cell proliferation. For the colony formation assay, 200 transfected OS cells were seeded per well into 6-well plates (Greiner) and cultured for 14 days to enable single-cell expansion and colony development. Following incubation, the colonies were fixed using 4% paraformaldehyde and stained with 0.1% crystal violet for visualization. The EdU assay was performed using the Cell-Light EdU Apollo567 In Vitro Kit (C10310-1, RiboBio, Guangzhou, China) following the manufacturer’s protocol. Briefly, 1 $\times$ 10⁵ transfected OS cells were plated in 24-well plates (Greiner) and incubated with EdU solution for 2 hours. Afterwards, cells were fixed with 4% paraformaldehyde for 15 minutes, permeabilized using 0.5% Triton X-100, and then labeled with the Apollo567 reaction mixture (RiboBio) for 30 minutes. Nuclei were counterstained with Hoechst 33342 (RiboBio) for 30 minutes as a positive control. EdU-positive cells were visualized and quantified under an inverted fluorescence microscope (Olympus Corporation, Tokyo, Japan) with five random fields analyzed per sample.

Transwell migration and invasion assays were carried out using Corning Transwell-24-well plates with 8.0 µm Pore Polyester Membrane Insert (Corning, NY, USA). Transfected OS cells (1 $\times$ 10⁵ cells in 200 µL serum-free medium) were seeded in the upper chamber with Matrigel (BD Biosciences, Franklin Lakes, NJ, USA) for invasion assay, or without Matrigel for migration assay. The lower chamber was supplemented with 500 µL of DMEM (Gibco BRL, Grand Island, NY, USA) containing 20% FBS (Gibco). Following 24 hours of incubation, cells that had not migrated or invaded through the membrane were gently wiped away from the upper chamber using a cotton swab. Migrated or invaded cells located on the bottom side of the membrane were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet, and analyzed under an inverted microscope. Five randomly selected high-power fields were acquired from each filter and processed using ImageJ (v1.54g, National Institutes of Health, Bethesda, MD USA) for quantitative analysis.

Transfected OS cells (1 $\times$ 10⁵) were plated into individual wells of 6-well plates (Greiner, Frickenhausen, Germany) and grown to complete confluence. A scratch was created in the cell monolayer using a sterile 200 µL pipette tip, followed by three washes with PBS. The cells were then incubated for 24 hours in Dulbecco’s modified Eagle medium (DMEM; Gibco BRL, Grand Island, NY, USA) containing 1% fetal bovine serum (FBS; Gibco). Images were acquired at 0, 12, and 24 h using an inverted microscope. The width of the wound was measured using ImageJ software.

The Annexin V-FITC/PI Apoptosis Detection Kit (556547, BD Biosciences, San Jose, CA, USA) was used following the manufacturer’s protocol. Briefly, transfected OS cells (1 $\times$ 10⁵) were harvested and washed twice with cold PBS before being resuspended in a 1$\times$ binding buffer. The cells were next incubated with 5 µL of Annexin V-FITC and 10 µL of propidium iodide staining solution for 15 minutes at room temperature under light-protected conditions. Following incubation, samples were immediately analyzed by a BD LSRFortessa™ flow cytometer (BD Biosciences, San Jose, USA). All experiments were conducted in three independent replicates.

Equal quantities of extracted proteins were resolved by SDS-PAGE using a 10% gel and subsequently transferred onto polyvinylidene difluoride (PVDF) membranes. The membranes were incubated with a 5% (w/v) solution of nonfat dried milk for blocking, followed by overnight incubation at 4 °C with primary antibodies. Following this, the membranes were probed with horseradish peroxidase-conjugated secondary antibody (1:5000, sc-2357, Santa Cruz Biotechnology, Santa Cruz, USA). Protein signals were visualized with an enhanced chemiluminescence (ECL) kit (180-5001, Tanon, Shanghai, China), and the intensity of the bands was quantified using ImageJ software. Primary antibodies were: anti-KIF18A (1:1000, ab251863, Abcam, Cambridge, UK), anti-Bax (1:1000, ab32503, Abcam, Cambridge, UK), anti-Bcl-2 (1:1000, ab59348, Abcam, Cambridge, UK), anti-Cleaved Caspase-3 (1:1000, ab2302, Abcam, Cambridge, UK), anti-PARP1 (1:1000, ab32138, Abcam, Cambridge, UK), anti-Cleaved PARP1 (1:1000, ab32064, Abcam, Cambridge, UK), as well as the internal loading controls anti-$\beta$-Actin (1:2500, ab179467, Abcam, Cambridge, UK) and anti-GAPDH (1:2500, ab9485, Abcam, Cambridge, UK). The experiment was conducted a minimum of three times.

Ten BALB/c nude male mice (4–6 weeks old; body weight 18–24 g) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China). They were equally classified into two groups (n = 5 per group), injected subcutaneously with U2OS cells stably transfected with either sh-KIF18A or sh-NC. 1 $\times$ 10⁶ U2OS cells in the logarithmic growth phase were suspended in 100 µL of phosphate-buffered saline (PBS) and subsequently injected subcutaneously into the right axilla of the mice. Tumor size was monitored and recorded every four days. After 20 days, the mice were euthanized via CO₂ asphyxiation at a flow rate between 30% and 70% of the chamber volume per minute, with cervical dislocation carried out afterward to confirm euthanasia. Tumor volume was calculated using the formula (mm³): tumor volume = length $\times$ width² $\times$ 0.5. The animal use protocol has been reviewed and approved by the Animal Ethics and Welfare Committee (AEWC) of Ningbo University (Approval No.: AEWC-2024-0002).

IHC was performed as previously described [38]. In brief, xenograft sections from paraffin were incubated at 4 °C in a humid chamber with primary antibodies against Ki-67 (1:500, ab15580, Abcam, Cambridge, UK) and KIF18A (1:500, ab251863, Abcam, Cambridge, UK). The proteins were visualized in situ using the BioGenex Super Sensitive LinkLabel IHC Detection System (BioGenex, Fremont, CA, USA).

R and GraphPad Prism 10 (San Diego, CA, USA) were applied to statistical analyses of this study. Differences between two groups were analyzed by Student’s t test and those between multi-groups were analyzed by one-way ANOVA, and the chi-square test was done to analyze the comparison of the rate of the category. Spearman’s correlation analysis was done to analyze the correlation of expression levels. Survival curves were depicted with Kaplan-Meier and differences were analyzed by Log-rank test. The data are expressed as mean $\pm$ standard deviation (SD). The p-value 0.05 was recognized as significant.

Firstly, we identified the KIFs family in OS patients of the TARGET database and analyzed their underlying protein interaction based on the STRING database (Fig. 1A). Using the cytohubba plugin in Cytoscape software, we noticed that the KIF20A, KIF15, KIF6, KIF13B, KIF16B, KIF23, KIF11, KIF22, KIF18A and KIF2A were the key nodes of the protein-protein interaction network (Fig. 1B). We examined the expression levels of 40 kinesin family genes in the GSE126209 OS dataset by comparing tumor tissues with adjacent non-cancerous tissues (Fig. 1C,D). Analysis revealed that KIF18A was significantly overexpressed in cancerous tissues (p = 0.0009) (Fig. 1D). Then, we evaluated the expression level of the KIFs family in metastatic (n = 22) and non-metastatic (n = 64) OS patients in TARGET-OS database. We found that all members of the KIF family are not statistically significant in terms of survival analysis (Fig. 1E,F). However, considering the relatively small sample size of the TARGET-OS, we believe this limitation can be relaxed. Therefore, due to the higher expression of KIF18A in OS, we chose KIF18A for further study. Not surprisingly, we identified that KIF18A was upregulated in OS tissues compared with its paired non-tumorous tissues (Fig. 1G) and showed significantly greater expression in metastatic tissues as well (Fig. 1H). To examine the potential correlation between the expression of KIF18A and clinical features, we performed an analysis of KIF18A expression and some clinical features of 79 patients with OS. According to the mean expression level of KIF18A, the samples were divided into 2 groups, the KIF18A high expression group and the KIF18A low expression group. Table 1 showed that KIF18A expression was significantly correlated with different clinical parameters, such as tumor size (p = 0.011), distant metastasis (p = 0.041), TNM stage (p = 0.022), and degree of differentiation (p = 0.002). We also evaluated the prognostic meaning of KIF18A. Kaplan-Meier analysis revealed that OS with high expression of KIF18A showed poor overall survival (Fig. 1I). Thus, the present results imply that KIF18A is over-expressed in OS and has a prognostic value for patients with OS.

Fig. 1.

Identification of KIFs family in the OS patients. (A) The protein-protein interaction network of KIFs family in OS. (B) The key nodes identified by cytoHubba plug-in in the cytoscape software. (C,D) Expression profiling of KIFs family in tumor tissues compared to paired adjacent normal tissues. (E,F) The expression level of KIFs family in metastatic and non-metastatic OS patients. (G) qRT-PCR was used to detect the expression level of KIF18A in OS tissues (n = 79) and corresponding adjacent normal tissues (n = 79). (H) qRT-PCR was used to evaluate the expression level of KIF18A in OS tissues with (n = 38) or without metastasis (n = 41). (I) Kaplan–Meier analysis was conducted on OS patients categorized by low (n = 35) and high (n = 44) expression levels of KIF18A. Statistical analysis for (C,D,G) was conducted using Paired Samples t-test, whereas Independent Samples t-test was utilized for (E,F,H). Data are shown as mean $\pm$ SD. The Kaplan-Meier survival analysis shown in (I) was conducted using the log-rank test. *p 0.05, ***p 0.001. KIFs, kinesin superfamily proteins; OS, Osteosarcoma; KIF18A, Kinesin family member 18A; SD, standard deviation. The red boxes illustrated in figures D and E highlight the statistical results associated with KIF18A.

Moreover, we explored the biological function of KIF18A in OS patients. GSEA based on the Hallmark gene set revealed that in OS patients exhibiting elevated KIF18A expression, several pathways were significantly upregulated, including KRAS signaling, E2F targets, G2M checkpoint, mitotic spindle formation, cholesterol homeostasis, androgen response, and Wnt/$\beta$-catenin signaling. Conversely, pathways associated with NOTCH signaling, transforming growth factor-beta (TGF-$\beta$) signaling, and interleukin-6 - janus kinase - signal transducer and activator of transcription 3 (IL6-JAK-STAT3) signaling demonstrated downregulation (Fig. 2A). GSEA analysis utilizing the GO gene set suggested that OS patients with high levels of KIF18A might experience increased activity related to GABAergic synapses, negative regulation of myotube differentiation, keratinization processes, cardiac right ventricle morphogenesis, and positive regulation of cyclic AMP responsive element-binding protein (CREB) transcription factor activity. In contrast, there was a notable reduction in activities involving CCR chemokine receptor binding, chemokine activity, eosinophil chemotaxis migration patterns pertaining to MHC class II protein complexes (Fig. 2B,C). Collectively these findings underscore a significant involvement of immune response-associated pathways alongside tumor growth-related mechanisms.

Fig. 2.

Biological role of KIF18A in OS. (A) GSEA analysis of KIF18A based on Hallmark gene set. The color scheme represents the Normalized Enrichment Score (NES), while circle size corresponds to -log₁₀(adjusted p-value), reflecting the level of statistical significance for each enriched pathway. (B,C) GSEA analysis of KIF18A based on GO gene set. GSEA, Gene Set Enrichment Analysis.

Since the function of KIF18A in OS has not been fully elucidated, we first assessed its expression levels in OS cell lines and clinical tissue specimens. Subsequently, we explored the impact of KIF18A on the biological behaviors of OS cells. KIF18A expression was assessed in OS cell lines in comparison to normal hFOB cells. Both mRNA and protein levels of KIF18A were markedly increased in OS cell lines relative to normal controls, with protein expression levels showing a strong correlation to the corresponding mRNA expression trends (Fig. 3A,B). Then, we tried to knock down the KIF18A expression and found that the sh-KIF18A#2 exhibited the highest knockdown efficiency (Fig. 3C,D). Colony formation and CCK8 assay indicated that the inhibition of KIF18A can significantly suppress the proliferation ability of OS cells (Fig. 3E–G). Also, the EdU assay indicated a significantly reduced ratio of EdU-positive cells in the KIF18A knockdown group compared with the normal group (Fig. 3H). Transwell assay results demonstrated that silencing KIF18A significantly suppressed the invasive and migratory capacities of OS cells (Fig. 4A). The wound-healing assay also indicated the same trend (Fig. 4B). Furthermore, we constructed a plasmid for the overexpression of KIF18A and observed a significant increase in KIF18A levels in OS cells following plasmid transfection (Supplementary Fig. 1A–C). KIF18A overexpression enhances the proliferation (Supplementary Fig. 1D–G), invasion (Supplementary Fig. 2A), and migration (Supplementary Fig. 2B) of OS cells.

Fig. 3.

KIF18A promotes the proliferation ability of OS cells. (A) The expression level of KIF18A in OS and normal cells. (B) Western blot analysis was performed to detect the expression of the indicated proteins in OS and normal control cell lines. (C) qRT-PCR was used to evaluate the knockdown efficiency of KIF18A. (D) Western blot analysis was conducted to evaluate the expression of the indicated proteins in OS cells 48 hours post-transfection with sh-KIF18A lentiviruses. (E–G) Colony formation and CCK-8 assays were conducted to compare KIF18A-knockdown cells with control cells. (H) EdU assays were conducted in both KIF18A-knockdown and control cells. Scale bar: 100 µm. Statistical analysis for (A–E) was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD, and all experiments were independently repeated at least three times. *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. CCK-8, Cell Counting Kit-8; EdU, 5-ethynyl-2^′-deoxyuridine.

Fig. 4.

KIF18A promote invasion and migration ability of OS cells. (A) Transwell assay was performed in control and KIF18A knockdown cells. Scale bar: 100 µm. (B) Wound-healing assay was conducted in control and KIF18A knockdown cells. Scale bar: 200 µm. Statistical analysis for (A,B) was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD, and all experiments were independently repeated at least three times. **p 0.01, ***p 0.001.

Then we transfected sh-KIF18A into the OS cells in order to study the effect of KIF18A on cellular apoptosis. By cell apoptosis analysis, we observed that the rate of cell apoptosis was increased remarkably in MG63 and U2OS cells treated with KIF18A knockdown (Fig. 5A). In accordance with these observations, the expressions of the apoptosis-related protein, Cleaved Caspase-3, Cleaved PARP, and Bax, was significantly upregulated in KIF18A knockdown cells (Fig. 5B). Conversely, the expression of Bcl-2 was notably downregulated (Fig. 5B). The data indicate that the downregulation of KIF18A expression in MG63 and U2OS cells triggers apoptosis.

Fig. 5.

Knockdown of KIF18A promotes apoptosis of OS cells. (A) The apoptosis of MG63 and U2OS cells was assessed through an apoptosis assay, conducted both with and without the silencing of KIF18A. (B) Western blotting was performed to evaluate the expression of apoptosis-associated proteins after transfection with sh-NC or sh-KIF18A. Statistical analysis for A and B was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD, and all experiments were independently repeated at least three times. *p 0.05, **p 0.01, ***p 0.001.

We subsequently examined the impact of KIF18A on the growth of OS cells in vivo. While all U2OS cells were capable of tumor formation when injected at a density of 1 $\times$ 10⁶ cells, the KIF18A-knockdown group exhibited a marked decrease in both tumor volume and weight compared to controls (Fig. 6A–C). Tumors derived from sh-KIF18A-transfected cells showed an almost twofold reduction in weight relative to the control group (Fig. 6C). Furthermore, the IHC assay indicated that KIF18A knockdown led to a decrease in Ki-67 positivity within in vivo tumors (Fig. 6D,E). Overall, these results indicate that KIF18A knockdown effectively suppresses the progression of human OS xenograft tumors in vivo.

Fig. 6.

Knockdown of KIF18A inhibits OS cell growth in vivo. (A) Typical images of tumors harvested from the sh-KIF18A and sh-NC groups. (B,C) Tumor volume and weight were assessed in both the sh-KIF18A and the sh-NC group. (D,E) Representative images illustrate the staining of Ki-67 and KIF18A in xenograft sections, with the percentages of positively stained cells indicated. Scale bar: 200 µm. Statistical analysis for (B,C,E) was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD, and all experiments were independently repeated at least three times. ***p 0.001.

Immune microenvironment has an important impact on the disease progression. To assess the immune landscape of OS, we employed the CIBERSORT algorithm for quantification (Fig. 7A). Correlation analysis revealed that KIF18A was positively associated with CD4+ memory resting T cells, activated NK cells, and CD4+ naïve T cells, but inversely correlated with follicular helper T cells, M2 macrophage, and resting dendritic cells (Fig. 7B). Immune function assessment indicated that patients exhibiting high KIF18A expression may have an inhibited immune function, like APC_co_inhibition, CCR, Check_point, HLA, inflammation_promoting and T_cell_co_stimulation (Fig. 7C). Meanwhile, we found that the immune-related gene HMGB1 was upregulated (p = 0.00132) in the OS patients with high KIF18A expression (Fig. 7D). Single-cell analysis showed that in the OS microenvironment, KIF18A was mainly expressed in the malignant, fibroblasts, osteoblasts and also several immune cells (Fig. 8A,B). Cell interaction results indicated an obvious interaction between fibroblasts and malignant (Fig. 8C). Then, the TIDE algorithm was utilized to estimate the immunotherapy response in OS patients (Fig. 8D). We noticed a positive correlation between KIF18A and TIDE score, indicating that patients with high expression of KIF18A are more likely to develop immune therapy resistance compared to those with low expression (Fig. 8E). Also, we noticed that KIF18A was negatively correlated with immune dysfunction, but positively correlated with MDSC infiltration (Fig. 8F–H).

Fig. 7.

Immune microenvironment analysis of KIF18A.(A) The immune microenvironment of OS patients was quantified using the CIBERSORT algorithm. (B) Correlation between KIF18A and quantified immune cells. (C) The immune function terms in patients with high and low KIF18A expression. (D) The immune checkpoints level in patients with high and low KIF18A expression. Statistical analysis for (C,D) was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD, and all experiments were independently repeated at least three times. *p 0.05, **p 0.01. The red box in figure B highlights the most significant correlation.

Fig. 8.

Single-cell and immunotherapy analysis. (A,B) The single-cell expression of KIF18A in OS microenvironment. (C) Cell interaction at single-cell level in OS microenvironment. (D) TIDE analysis was used to quantify the immunotherapy response of OS patients. (E) Correlation between KIF18A and TIDE score. (F) Correlation between KIF18A and immune dysfunction. (G) Correlation between KIF18A and CAFs infiltration. (H) Correlation between KIF18A and MDSC infiltration.

Subsequently, we performed a drug sensitivity analysis of common drugs for OS patients. We found that the patients with low KIF18A expression might be more sensitive to lapatinib (Fig. 9A). Following this, we tried to identify the lncRNA associated with KIF18A with the threshold of $|$cor$|$ $>$0.2 and p 0.05 (Fig. 9B). Based on the univariate Cox regression analysis, we found that among these lncRNAs, AC124798.1, MEF2C-AS1, AC147067.1, AC083843.2, SNHG1, NCAM1-AS1, GSEC and AF241728.1 were risk factors, while AC091271.1, AC015819.1, AC005911.1 and AC104561.1 were protective factor (Fig. 9C).

Fig. 9.

Drug sensitivity analysis and lncRNA network of KIF18A. (A) The IC₅₀ difference of specific drugs in patients with high and low KIF18A level. Statistical analysis for Figures A was conducted using Independent Samples t-test. Data are shown as mean $\pm$ SD. *p 0.05. (B) The lncRNA regulatory network of KIF18A. (C) Among these regulatory lncRNAs, 12 lncRNAs were identified to have a significant prognosis effect.