Abstract

   Chordoma is a special malignant tumor that lacks effective therapeutic
   targets, which can lead to incomplete treatment and metastasis.
   Inflammation plays an important role in chordoma progression and
   malignant phenotype. Inflammatory factors such as NF-kappaB and STAT3
   are continuously activated in many tumors and contribute to the
   malignant phenotype of tumors and are potential therapeutic targets.
   This study suggest TNF-alpha and NF-kappaB signaling pathways were
   consistently activated in chordomas. Long-term TNF-alpha treatment
   induces chordoma resistance to EGFR family inhibitors. The underlying
   mechanism is realized by the key molecules HS3ST3A and HS3ST3B1. These
   two enzymes are potential targets for chordoma treatment, as well as
   for combination drugs treatment. It should be emphasized that the above
   analysis lacks experimental verification.
     __________________________________________________________________

   Chordoma is a special malignant tumor that lacks effective therapeutic
   targets, which can lead to incomplete treatment and metastasis. graphic
   file with name d5md00258c-ga.jpg

Introduction

   Chordomas, derived from axial skeletons, are rare primary malignant
   tumors, which have surgical resection as their main treatment.^[38]1,2
   However, due to the special location of the dissection and the local
   destructive behavior of the malignant tumor itself, chordoma is
   difficult to treat and has a high recurrence rate (>50% of
   patients).^[39]3,4 In addition, chordoma is resistant to conventional
   chemotherapy and tends to metastasize during the course of the disease,
   resulting in poor prognosis.^[40]5,6 Therefore, it is necessary to
   clarify the underlying mechanism of chordoma tumorigenesis and find
   valuable therapeutic targets for chordoma treatment.

   Inflammation plays an important role in the occurrence and progression
   of tumors. NF-kappaB and STAT3, key modulators of inflammation, are
   constitutively active in most cancers. The activation of NF-kappaB can
   lead to chemoresistance and radio-resistance of tumors.^[41]7 Previous
   studies have shown that the effect of NF-kappaB on chordoma growth and
   the prognosis of patients with chordoma is significant.^[42]8 Tumor
   necrosis factor-alpha (TNF-alpha) is one of the classical prominent
   activators of NF-kappaB, and may be involved in the maintenance of
   tumor inflammatory microenvironment.^[43]9 Although TNF-alpha was
   originally thought to have a tumor-killing effect, it has also been
   shown to promote tumor survival and growth.^[44]10 In addition, breast
   cancer cells with long-term TNF-alpha treatment exhibited increased
   chemo-resistance.^[45]11 Thus, elucidating the effect of the
   inflammatory factor TNF-alpha on the malignant phenotype of chordoma
   can promote the understanding of tumor pathogenesis and the discovery
   of new therapeutic targets.

   Proteoglycans are composed of the covalent attachment of
   mucopolysaccharide chains to core proteins and play a significant role
   in the progression of cancer. The function of proteoglycans is mainly
   determined by their mucopolysaccharide chains. Enzymes responsible for
   the synthesis and modification of mucopolysaccharide chains can
   regulate the function of glycoproteins, thereby influencing the
   progression of cancer.^[46]12,13 Heparin sulfate 3-O-sulfotransferase
   (HS3ST) is a family of enzymes that catalyzes the 3-O-sulfidation
   modification of mucopolysaccharide chains.^[47]14,15 The 3-O-sulfated
   epitope is elevated in central nervous system tumors.^[48]16

   In this study, we analyzed the signaling pathway activation of
   NF-kappaB and TNF-alpha in chordoma, screened for malignant changes in
   chordoma cells caused by long-term TNF-alpha treatment, and finally
   obtained a key target, HS3ST3A1. Further functional analysis showed
   that HS3ST3A1 and its functionally related enzyme HS3ST3B1 promotes
   chordomas resistant to drugs that target the EGFR family. It has also
   been reported that the EGFR inhibitors were identified as potential
   treatment for chordoma. In addition, O[3] sulfuration of heparan
   sulfate proteoglycans (HSPGs) is more prevalent in gliomas of the
   nervous system.^[49]16,17 Therefore, HS3ST3A1 and HS3ST3B1 are the
   potential therapeutic targets for chordoma treatment and drug
   resistance.

Results

The NF-kappaB and TNF-alpha signaling pathway are continuously activated in
chordoma tissue

   The transcriptome data of chordoma tissue and normal notochord tissue
   in [50]GSE224776 the dataset was downloaded for evaluation of the
   inflammation level in chordoma tissue. The three notochord samples and
   four chordoma samples were chosen for differential expression analysis.
   Gene set enrichment analysis (GSEA) was performed based on differential
   gene arrangement. The results showed that NF-kappa and TNF-alpha
   signaling pathways were continuously activated in chordoma compared to
   normal notochord tissue ([51]Fig. 1A and B). Then, we used the
   CIBERSORT algorithm to analyze the immune infiltration of different
   tissues. The results showed that in chordoma, M0 macrophages
   differentiated into M1 and M2 macrophages, which were the main sources
   of TNF-alpha ([52]Fig. 1C and D). Heat maps of the expression levels of
   TNF-alpha family genes in different samples showed that TNF-alpha
   family genes were mostly activated in chordomas, suggesting that the
   TNF-alpha signaling pathway plays an important role in the malignant
   phenotype of chordomas ([53]Fig. 1E). Further, correlation analysis of
   TNF-alpha family gene expression level and immune infiltration level
   showed that M1 and M2 macrophages contributed the most to the TNF-alpha
   pathway in chordoma ([54]Fig. 1F). The possibility that the chordoma
   itself secretes TNF-alpha cannot be ruled out, and there is a lack of
   published data to verify this.

Fig. 1. The inflammatory cytokines TNF-alpha and NF-kappaB are continuously
activated in chordoma tissue. (A) The CSEA results of the NF-kappaB signaling
pathway of differential expressional genes between chordoma and normal
notorchord tissue. (B) The CSEA results of TNF-alpha signaling pathway of
differential expressional genes between chordoma and normal notorchord
tissue. (C and D) Classification of immune cell subtypes in different tissue
samples. (E) Expression of genes in the TNF-alpha signaling pathway in
different tissue samples. (F) Correlation analysis between the TNF-alpha
signaling pathway related genes and immunoinfiltrating subtypes.

   [55]Fig. 1
   [56]Open in a new tab

Long time TNF-alpha treatment induces EGFR tyrosine kinase inhibitor
resistance in chordoma cells

   [57]GSE101866 and [58]GSE101867 dataset were downloaded for the
   analysis of the role of long-term TNF-alpha-induced inflammation in
   chordoma. The differential expression genes of chordoma cells treated
   with short-term TNF-alpha and long-term TNF-alpha were screened from
   the control group ([59]Fig. 2A and B). Venn diagram was used to show
   the different differential gene divisions ([60]Fig. 2C). There were 19
   common core genes that were consistently expressed during TNF-alpha
   treatment, and 83 specific targets that changed only after long-term
   TNF-alpha treatment. Biological process (BP) enrichment analysis of
   gene ontology (GO) was performed for these two subsets ([61]Fig. 3D and
   E). The three most significant enrichment results of core genes are the
   cellular response to peptide, response to alcohol, and response to
   gravity. The three most significant enrichment results of specific
   targets are the urogenital system development, renal system
   development, and regulation of chemotaxis. These biological processes
   do not appear to be associated with the malignant phenotype of
   chordoma. So KEGG pathway enrichment was performed to these two
   clusters for further analysis ([62]Fig. 2F and G). The KEGG enrichment
   of core genes showed that the TNF-alpha signaling pathway is included,
   suggesting that TNF-alpha keeps the TNF signaling pathway active in
   chordoma, both in the short and long term. The KEGG enrichment of core
   genes showed that long-term TNF-alpha treatment can induce EGFR
   tyrosine kinase inhibitor resistance.

Fig. 2. Analysis of the effect of TNF-alpha on chordoma cells. (A)
Differentially expressed genes in CH1 and MUG cell lines treated with short
term TNF-alpha compared with the control group. (B) Differentially expressed
genes in CH1 and MUG cell lines treated with long term TNF-alpha compared
with the control group. (C) Intersecting the results of screening for key
targets, red circles represent genes that are altered in both short- and
long-term TNF-alpha treatment, and green circles represent targets that are
specifically altered in long-term TNF-alpha treatment. (D) GO analysis result
of key genes altered in both short- and long-term TNF-alpha treatment. (E) GO
analysis result of specific targets specifically altered in long-term
TNF-alpha treatment. (F) KEGG analysis result of key genes altered in both
short- and long-term TNF-alpha treatment. (G) KEGG analysis result of
specific targets specifically altered in long-term TNF-alpha treatment.

   [63]Fig. 2
   [64]Open in a new tab

Fig. 3. Screening for targets of malignant phenotypes induced by long-term
TNF-alpha treatment in chordoma. (A) Differential expression analysis of
chordoma and normal chordal tissue. (B) KEGG enrichment analysis of
differentially expressed genes. (C) Results of screening key targets.

   [65]Fig. 3
   [66]Open in a new tab

Screening the key target of the malignant phenotype of chordoma induced by
long-term TNF-alpha treatment

   In order to further screen out the key target responsible for the
   malignant phenotype of chordoma under long-term TNF-alpha treatment,
   the specific targets induced by long-term TNF-alpha treatment were
   intersected with the malignant targets of glioma tissue compared with
   normal tissue, and 13 core targets were obtained ([67]Fig. 3C). In
   addition, KEGG enrichment analysis was performed on the differential
   genes of chordoma relative to normal tissues, and the five most
   significant results were the cytoskeleton in muscle cells, motor
   proteins, protein digestion and absorption, ECM–receptor interaction
   and dilated cardiomyopathy ([68]Fig. 3B). The malignant phenotypic
   enrichment of the chordoma itself did not overlap with the function
   induced by the long-term action of TNF-alpha treatment.

   Our scientific hypothesis is that TNF-alpha maintained inflammation
   promotes the malignant phenotype of chordomas, and therefore, we
   screened for indicators with consistent trends. The results shown in
   [69]Fig. 4A and B revealed that the co-upregulated target was HS3ST3A1,
   and the co-down-regulated targets were CNTN1 and NR3C2.

Fig. 4. Key targets HST3ST3A1 and HS3ST3B1 are correlated with the level of
TNF-alpha in chordoma cells. (A) Volcano plot of differential expressional
genes. (B) Differential genes in long time TNF-alpha treatment. (C)
Correlation screening of functional genes related with HS3ST3A1 in chordoma.
(D) KEGG enrichment analysis of HS3ST3A1 and its functional related genes.
(E) Results show the transcriptional levels of TNF-alpha, HS3ST3A1, and
HS3ST3B1 in different chordoma cell lines. (F) Results show the
transcriptional level changes of HS3ST3A1 and HS3ST3B1 after TNF-alpha
treatment.

   [70]Fig. 4
   [71]Open in a new tab

Key targets HST3ST3A1 and HS3ST3B1 are correlated with the level of TNF-alpha
in chordoma cells

   Finally, HS3ST3A1 was selected as the core target for further analysis.
   The [72]GSE239531 dataset was downloaded for further functional
   analysis of HS3ST3A1. Correlation analysis of HS3ST3A1 in chordoma
   tissue was performed, and indicators with absolute value of correlation
   coefficients greater than 0.7 and P value less than 0.05 were chosen
   ([73]Fig. 4C). The KEGG enrichment analysis was performed for HS3ST3A1
   related functional genes ([74]Fig. 4D). Among them, glycosaminoglycan
   biosynthesis-heparan sulfate/heparin got our interest. Proteoglycans in
   cancers were also found in the KEGG enrichment of special targets under
   long-term TNF-alpha treatment ([75]Fig. 2G). The genes enriched in this
   pathway are HS3ST3A1 and HS3ST3B1, both enzymes that are involved in
   protein glycosylation modification.

   The results of qRT-PCR showed that the transcriptional levels of
   TNF-alpha in the chordoma cells U-CH2 and MUG-Chor1 were lower than in
   the U-CH1 and JHC7 cells ([76]Fig. 4E). The relative transcriptional
   levels of HS3ST3A1 and HS3ST3B1 in these cell lines were consistent
   with the level of TNF-alpha ([77]Fig. 4E). Furthermore, we treated
   U-CH1 cells with TNF-alpha, and found that the transcription levels of
   HS3ST3A1 and HS3ST3B1 increased ([78]Fig. 4F). These results suggest
   that HS3ST3A1 and HS3ST3B1 are regulated by TNF-alpha.

Analysis of the effects of HST3ST3A1 and HS3ST3B1 on the resistance of
chordoma to EGFR tyrosine kinase inhibitors

   Previous studies have reported that protein glycosylation can promote
   tumor drug resistance.^[79]18 Combined with KEGG enrichment results
   shown in [80]Fig. 2G, we speculated that the protein glycosylation
   modification mediated by HS3ST3A1 and HS3ST3B1 induces the resistance
   of chordoma cells to EGFR inhibitors. Therefore, we downloaded from the
   CCLE database the drug susceptibility data of EGFR family inhibitors,
   as well as the expression data of HS3ST3A1 and HS3ST3B1 in different
   cells to verify this speculation. The cells were divided into the high
   expression level group and low expression level group according to the
   median of the transcriptional level of HS3ST3A1 and HS3ST3B1,
   respectively. Results show that in the HS3ST3A1 group and HS3ST3B1 high
   expression group, the AUC values of afatinib, erlotinib, gefitinib and
   neratinib are higher, suggesting that the cells have a high level of
   HS3ST3A1 and HS3ST3B1 resistance to EGFR inhibitors ([81]Fig. 5). We
   also compared inhibitors targeting other members of the tyrosine kinase
   receptor family, such as dasatinib, imatinib, lapatinib, and nilotinib,
   and the results were not significant ([82]Fig. 6).

Fig. 5. Drug resistance analysis of cells with different expression levels of
HS3ST3A1 and HS3ST3B1 in afatinib, erlotinib, gefitinib and neratinib. (A, C,
E and G) Distribution of drug AUC in high and low expression groups of
HS3ST3A1. (B, D, F and H) Distribution of drug AUC in high and low expression
groups of HS3ST3B1.

   [83]Fig. 5
   [84]Open in a new tab

Fig. 6. Drug resistance analysis of cells with different expression levels of
HS3ST3A1 and HS3ST3B1 in dasatinib, imatinib, lapatinib and nilotinib. (A, C,
E and G) Distribution of drug AUC in high and low expression groups of
HS3ST3A1. (B, D, F and H) Distribution of drug AUC in high and low expression
groups of HS3ST3B1.

   [85]Fig. 6
   [86]Open in a new tab

   To further validate this conclusion, the cells in the CCLE database
   were divided into EGFR inhibitor-resistant and EGFR inhibitor-sensitive
   cells according to the median AUC values of afatinib, erlotinib,
   gefitinib and neratinib. Cells with larger AUC values than all four
   median AUC values are resistant cells, while cells with AUC values less
   than all median AUC values are sensitive cells ([87]Fig. 7A). For cells
   that cannot be clustered by this criterion, semi-supervised k-means
   clustering is used for classification. The semi-supervised k-means
   clustering of all cells was carried out with the clearly classified
   cells as reference, and all cells were successfully divided into
   drug-resistant cells and sensitive cells ([88]Fig. 7B). The expression
   levels of HS3ST3A1 and HS3ST3B1 in sensitive and drug-resistant cells
   were analyzed, and it was found that the expression levels of both
   genes were higher in drug-resistant cell lines ([89]Fig. 7C). This
   further supports the effect of these two enzymes on EGFR inhibitor
   resistance.

Fig. 7. Validation of HS3ST3A1 and HS3ST3B1 induced EGFR tyrosine kinase
inhibitors. (A) Schematic diagram of semi-supervised k-means classification.
(B) Presentation of classification results. The red dots represent resistant
cell lines and the green dots represent sensitive cell lines. (C) Results of
the expression levels of HS3ST3A1 and HS3ST3B1 in different classification
cells.

   [90]Fig. 7
   [91]Open in a new tab

   Furthermore, the genes enriched in the KEGG pathway EGFR tyrosine
   kinase inhibitor resistance were selected. As shown in [92]Fig. 8A,
   these genes are mainly distributed in the bypass pathway of the EGFR
   signaling pathway. Based on these genes, GSVA analysis was conducted on
   the transcriptional data of the CCLE database. The results showed that
   these genes were relatively activated in the resistant group ([93]Fig.
   8B and C). Moreover, the GSVA scores of the core genes were positively
   correlated with the transcriptional levels of HS3ST3A1 and HS3ST3B1
   ([94]Fig. 8D). These results are helpful in identifying the substrates
   of HS3ST3A1 and HS3ST3B1, but further experiments are needed for
   verification in the future.

Fig. 8. Verification of the changes of the EGFR TKI resistance gene set at
different cellular levels. (A) The distribution plot of gene sets in
pathways. (B) Heat map shows the relative expression of each gene in the gene
set. (C) The results of the differences in GSVA scores among different
groups. (D) Scatter plot showing the correlation between the GSVA score and
HS3ST3A1 and HS3ST3B1.

   [95]Fig. 8
   [96]Open in a new tab

Methods

Cell culture and treatment

   Human chordoma cells were cultured in basal media (U-CH1 and JHC7 in
   DMEM/F12, U-CH2 in 1640, and MUG-Chor1 in IMDM) supplemented with 10%
   FBS and 1% penicillin–streptomycin (Solarbio P1400) and grown at 37 °C
   in a humidified incubator with 5% CO[2].

   U-CH1 and MUG-Chor1 cells were obtained from Meisen (Zhejian, China).
   U-CH2 and JHC7 cells were obtained from ATCC (USA). Fetal bovine serum
   was purchased from Gibco.

   TNF-alpha agent was purchased from Proteintech (no. Ag11413). The
   treatment conditions are 20 ng ml^−1 and two weeks.

RNA extraction and qRT-PCR

   TRIzol reagent (15596026, Invitrogen, Waltham, MA, USA) was used to
   extract total RNA from cells. After mixing with chloroform, it was
   centrifuged at 12000 rpm for 4 °C for 10 minutes. The upper phase was
   extracted and mixed with an equal volume of isopropanol. Then the
   mixture was centrifuged at 12 000 rpm for 10 minutes at 4 °C, then the
   isopropanol was discarded. 1 ml of 75% ethanol was added, and the
   mixture was centrifuged at 7500 rpm for 5 minutes at 4 °C, the ethanol
   was discarded, and the total RNA of the cells was collected.

   The cDNA reverse transcription kit (KR116-02, Tiangen, China) is used
   to convert total RNA into cDNA.

   qRT-PCR analyses were performed using the SYBR Premix Ex Taq kit
   (AQ601, TransGen Biotech, Beijing, China), employing beta-actin as an
   internal reference.

   The primer sequences:

   Beta-actin-F: TGGCACCCAGCACAATGAA,

   Beta-actin-R: CTAAGTCATAGTCCGCCTAGAAGCA;

   Human TNF-alpha-F: CAGAGGGCTGATTAGAGAGAGGT,

   Human TNF-alpha-R: TGCTTGTTCCTCAGCCTCTT;

   HS3ST3A1-F: CATGGAGAAGACGCCCAGTT,

   HS3ST3A1-R: CACGATGAGCTTGGTGTCCT;

   HS3ST3B1-F: TCTTCGATCGCAGCTACGAC,

   HS3ST3B1-R: AACTGGGCGTCTTCTCCATG.

   The primers were synthesized by Sangon (shanghai, China).

Data source

   Chordoma data used for signaling pathway difference analysis and
   expression difference analysis were derived from the GEO database
   ([97]GSE224776), including three notochord samples and four chordoma
   samples. The [98]GSE101866 and [99]GSE101867 dataset were downloaded
   from the GEO database for analysis of the effect of long term TNF-alpha
   on chordoma cells. The [100]GSE239531 dataset was downloaded from the
   GEO database for the function analysis of the key target HS3ST3A1. Drug
   susceptibility data and gene expression data of cells were obtained
   from the CCLE database ([101]https://sites.broadinstitute.org/ccle/).

Immune cell subtype proportion analysis

   The cell-type identification by estimating relative subsets of RNA
   transcripts (CIBERSORT) algorithm used for immune cell subtype
   proportion analysis is provided by the IOBR package (v0.99.9) of R
   software (v4.3.1).^[102]19,20

GSEA and KEGG pathway enrichment analysis

   Gene set enrichment analysis (GSEA) was used for the analysis of
   signaling pathway difference between tumor tissue and normal tissue.
   The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analysis was used for the analysis of downstream of differential genes.
   Both methods are implemented by the clusterProfiler package (v4.8.3) of
   R software.^[103]21

Differential gene analysis and correlation analysis

   Differential analysis between tumor tissues and normal tissues were
   performed by the limma package (v3.56.3) of R software. The absolute
   value of logarithm of foldchange was greater than 2, and the P value
   was less than 0.05 (|log FC| > 2, P value < 0.05). Correlation analysis
   was performed by the corrplot package (v0.92) of R software. Genes with
   absolute values of correlation coefficients greater than 0.7 and
   P-values less than 0.05 were selected (|cor| > 0.7, P value < 0.05).

Semi-supervised k-means algorithm

   The semi-supervised k-means algorithm uses labeled data to get the
   initial cluster center, and then trains the unlabeled data according to
   the initial cluster center until the model converges. The purpose of
   classifying unlabeled data according to labeled data is also achieved.
   The above operations are based on the Scikit-learn library (v1.6.0) of
   Python (v3.13.1).

Statistical analysis

   All statistical analyses in the present research were implemented using
   R software (version 4.3.1) and python (v3.13.1). p < 0.05 was used as
   the threshold for statistical significance.

Discussion

   In this study, we verified that the TNF-alpha–NF-kappaB inflammatory
   pathway is continuously activated in chordoma. Long time TNF-alpha
   treatment can induce partial tyrosine kinase inhibitor (TKI) resistance
   of chordoma. Further analysis showed that HS3ST3A1 and HS3STSB1 are key
   molecules mediating the malignant phenotype of chordoma. Drug
   sensitivity data of CCLE were used to demonstrate that cells with a
   high expression of HS3ST3A1 and its related enzyme HS3ST3B1 were more
   resistant to EGFR-targeting TKIs. The cell lines were divided into TKI
   resistant and sensitive types according to whether they were resistant
   to afatinib, erlotinib, gefitinib and neratinib. The resistant cell
   lines have high expression level of these two enzymes.

   To date, the main treatment for chordoma is surgical resection, and
   there is lack of effective target drugs in clinics.^[104]1,2 In
   addition, chordomas have unexplained chemotherapy resistance and are
   prone to progression.^[105]5,6 Therefore, it is urgent to search for
   therapeutic targets for chordoma treatment. Inflammation plays an
   important role in tumor progression, and we demonstrated that the
   TNF-alpha–NF-kappaB inflammatory pathway is continuously activated in
   chordoma. According to the immune infiltration algorithm CIBERSORT,
   this malignant phenotype may be related to the immune microenvironment
   of chordoma. Based on data from the GEO database, we analyzed the
   transcriptional characteristics of chordoma cells under long-term
   TNF-alpha. The results showed that long-term TNF-alpha treatment
   induced EGFR inhibitor resistance and tumor protein glycosylation.
   Further, we screened out the target of the malignant phenotype of
   chordoma from the above genes. The HS3ST3A1 gene is highly expressed in
   both long-term TNF-alpha treatment and chordoma, suggesting that the
   HS3ST3A1 gene is the mediator of chordoma malignancies induced by the
   inflammatory factor TNF-alpha. Further screening for HS3ST3A1 related
   functional genes in chordoma was performed. The final results indicated
   that the glycosylation modification of HS3ST3A1 and HS3ST3B1 may be the
   underlying mechanism of the malignant phenotype induced by TNF-alpha in
   chordoma. The results of qRT-PCR verified that HS3ST3A1 and HS3ST3B1
   were related to the level of TNF-alpha in chordoma cells and were
   regulated by TNF-alpha.

   The CCLE database was used to verify whether HS3ST3A1 and HS3ST3B1 have
   an impact on EGFR inhibitor resistance. Both direct comparison and
   comprehensive analysis after semi-supervised k-means clustering showed
   that high expression of HS3ST3A1 and HS3ST3B1 resulted in EGFR
   inhibitor resistance.

   The enzymes of the HS3STs family are responsible for the 3-O-sulfated
   modification of heparin sulfate to form the sulfated
   epitope.^[106]14,22 The biological function of proteoglycans is largely
   determined by the sulfation site of the oligosaccharide, but not the
   core protein.^[107]15,23–26 Proteoglycans have been shown to be
   involved in the regulation of multiple signaling pathways, so their
   modification enzymes are important for the development and progression
   of cancers.^[108]12,13,27,28 However, previous studies have shown that
   glycoproteins regulate signaling pathways by interacting with cytokines
   and receptors,^[109]29 and this study found that HS3ST3A1 and HS3ST3B1
   are associated with EGFR inhibitor resistance, proving that the
   modification of glycosylation may be one of the causes of chordoma drug
   resistance. In addition, compared with the O6 sulfuration of HSPG,
   sulfuration of O[3] promotes the malignancy of central nervous system
   tumors.^[110]16

   In addition, precision treatment comes from the accurate classification
   of tumors.^[111]30 Assessing the TNF-alpha level of chordoma is
   conducive to identifying more suitable tumor subgroups for treatment.

   In summary, HS3ST3A1 and HS3ST3B1 are potential therapeutic and
   combination drug targets for chordoma. However, this study lacks
   experimental verification and needs further research.

Author contributions

   DW: formal analysis, investigation, methodology, project
   administration, software, supervision, and writing – original draft.
   HT: writing – original draft and validation. QZ: writing – review and
   editing and validation. Others do ancillary work.

Conflicts of interest

   The authors declare that the research was conducted in the absence of
   any commercial or financial relationships that could be construed as a
   potential conflict of interest.

Acknowledgments