Cell Cycle-Related FAM64A Could be Activated by TGF-β Signaling to Promote Glioma Progression

Gliomas are aggressive brain tumors characterized by uncontrolled cell proliferation. FAM64A, a cell cycle-related gene, has been found to promote cell proliferation in various tumors, including gliomas. However, the regulatory mechanism and clinical significance of FAM64A in gliomas remain unclear. In this study, we investigated FAM64A expression in gliomas with different grades and constructed FAM64A silenced cell lines to study its functions. Our results demonstrated that FAM64A was highly expressed in glioblastoma (P < 0.001) and associated with a poor prognosis (P < 0.001). Expression profiles at the single-cell resolution indicated FAM64A could play a role in a cell-cycle-dependent way to promote glioma cell proliferation. We further observed that FAM64A silencing in glioma cells resulted in disrupted proliferation and migration ability, and increased cell accumulation in the G2/M phase (P = 0.034). Additionally, TGF-β signaling upregulates FAM64A expression, and SMAD4 and FAM64A co-localize in high-grade glioma tissues. We found FAM64A knockdown inhibited TGF-β-induced epithelial-mesenchymal transition in glioma. Our findings suggest that FAM64A could serve as a diagnostic and therapeutic target in gliomas.


Introduction
Glioma is the most prevalent and malignant brain tumor, and the prognosis remains poor due to its rapid progression (Wen and Packer 2021;Ostrom et al. 2016). The genetic and epigenetic markers are given increasing priority since the update of the 2021 WHO CNS5 tumor classification (Louis et al. 2016). IDH mutation, H3K27M mutation, cell cyclerelated CDKN2A, and other molecular events all contributed to the profiling of gliomas, which might indicate totally different therapeutic responses and clinical outcomes (Ross et al. 2021;Vitucci et al. 2011).
Given self-sufficiency in growth signals was one of the hallmarks of cancer (Hanahan 2022), cell-cycle-related Minjie Fu, Jingwen Zhang and Licheng Zhang have contributed equally to this study. genes like CDK4 and CDK6 have attracted attention and clinical trials are launched about their inhibitors (Sepulveda-Sanchez et al. 2020;DeWire et al. 2020). Family with sequence similarity 64 member A (FAM64A, aka PIMREG, RCS1) is another cell cycle-related protein (Jiang et al. 2020), which could control the metaphase-to-anaphase transition during cell division (Hashimoto et al. 2017). Abnormal FAM64A expression could promote tumorigenesis in multiple tumors including breast cancer and prostate cancer (Yao et al. 2019;Zhang et al. 2014;Zhou et al. 2021;Mizuno et al. 2020;Qiu and Qin 2021). In addition to its cell cycle functions, FAM64A could also enhance epithelial-tomesenchymal transition (EMT) and cell stemness Wang et al. 2021). Knockdown of FAM64A would suppress proliferation and migration of breast cancer (Yao et al. 2019).
EMT would allow cancer cells to acquire invasive properties and penetrate the neighboring stroma (Pastushenko and Blanpain 2019). Recent studies reveal that epithelial and mesenchymal characteristics are very plastic in GBM cells, and these tumors may transit in a broad phenotype spectrum during tumorigenesis and progression (Schichor et al. 2012;Zhang et al. 1999;Hitomi et al. 2015). limited amount of information is available about the differences between the classical EMT occurring in epithelial tumors and the glial-mesenchymal changes occurring in glioma (Iser et al. 2017). A better understanding of the regulatory mechanisms of EMT process during glioma progression would help to provide potential therapeutic interventions for glioma.
In the current study, we evaluated the impact of FAM64A on glioma proliferation and found High FAM64A expression in gliomas is indicative of poor prognosis. In addition to its cell cycle related functions, FAM64A was found essential to mediate TGF-β-induced EMT. Given the great abundance of FAM64A and its diverse functions, FAM64A might be a promising prognostic biomarker and potential tumor therapeutic target.

Data Collection and Processing
All clinical information and bulk RNA-seq and array expression data were downloaded from Gliovis (http:// gliov is. bioin fo. cnio. es/). All the data were preprocessed by normalizing. Single-cell expression profiles were obtained from a previous study published in the GEO database (GEO Accession No. GSE131928, GSE125116). Data analysis was conducted using R 4.0 (R Core Team, 2020).

Clinical Samples
Glioma tissues were resected in 2020 from patients in the Department of Neurosurgery, Huashan Hospital of Fudan University. Normal brain tissues were gathered from traumatic brain injury patients. The experiments were approved by the Human Ethics Committee of Huashan Hospital, and informed consent was signed by all patients.
U87 cells, U251 cells, and HA-MG cells from National Collection of Authenticated Cell Cultures were transfected with lentivirus according to the manufacturers' instructions. 2 μg/mL puromycin was used to select stable clonal cell lines.
Before TGF-β stimulation, U87 and U251 cells were split at 30% confluency and inspected daily. Cells were cultured in media supplemented with 5 ng/ml recombinant TGF-β1 (Peprotech, China) for TGF-β stimulation. Cells were collected for Western Blot and CCK8 assays at 0, 24 h, 48 h, and 72 h.

Western Blotting
RIPA lysis buffer was used for both tissues and cell protein extraction. After centrifugation, the supernatant was gathered, then the 5 × loading buffer was added and boiled for 5 min. Western blot was conducted as previously described (He et al. 2020). The data analysis and statistics were performed through ImageJ as reported previously (Stanic et al. 2015).

In vitro cell Growth and Migration Assays
In vitro cell growth assays were performed by cell-counting kit-8 (CCK8) assays. U87 cells, U251 cells, and HA-MG cells were seeded in 96-well plates and were incubated with 10% CCK8 solution (Meilunbio, MA0218) at 37 °C for 2 h, and then the absorbance at 450 nm was measured. In vitro migration assays were performed by wound healing assay. shFAM64A and shScramble U87 cells, U251 cells, and HA-MG cells were plated in 24-well culture plates. Cell monolayers were allowed to rest for 12 h and a wound was made by scraping the middle of the cell monolayer with a P10 pipette tip. Cells were washed with PBS twice and photographed. Migration was monitored and photographed 12 h later in DMEM supplemented with 10% FBS. Quantitation was performed on images in at least three independent experiments for each experimental condition.

Differential Gene Expression and Pathway Enrichment Analysis
In the gliovis database (http:// gliov is. bioin fo. cnio. es/), TCGA-GBM patients were divided into two groups (FAM64A-high and FAM64A low) depending on the expression of FAM64A mRNA expression level. The differential expressed genes (DEG) analysis were conducted using limma package. The differential expressed genes were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG, http:// www. genome. jp/ kegg/) functional annotation pipeline for gene function enrichment analysis using clusterProfiler package . Gene set enrichment analysis (GSEA) was carried out using the GSEA preranked tool (http:// softw are. broad insti tute. org/ gsea/ index. jsp).

Flow Cytometry Analysis of Cell-Cycle Progression
Cell cycle analysis was conducted with propidium iodide (Biosharp, China). 5 × 10 6 cells were harvested after digestion and fixed with ice-cold 70% ethanol at 4 °C for 1 h. After being washed with PBS, 100 µL RNase A and 400 µL PI were mixed well and added to the suspensions in a dark place for 30 min. Finally, a flow cytometer and data fitting software were employed for the cell cycle analysis.

Single-Cell Transcriptome Data Analysis
R package Seurat was applied to aggregate all samples (Butler et al. 2018), and perform quality control on the raw gene expression matrix obtained from the Cell Ranger pipeline. The filtering criteria for each cell were: (1) the number of genes expressed ranges from 200 to 4000; (2) the percentage of the mitochondrial genes is less than 25%. Cells that did not satisfy the above criteria were discarded. Gene expressions were normalized by log normalization with a factor 10,000, and the top 2000 highly variable genes were selected for PCA. Harmony was utilized for batch-effect correction based on the top 50 principal components of PCA (Korsunsky et al. 2019). The corresponding cell types of the clusters were annotated using SingleR with the reference datasets of human immune cells. The annotation was further verified manually following known gene markers.

Transcription Factor Analysis
CistromeDB toolkit (http:// dbtoo lkit. cistr ome. org/) was used to predict the function of this TF ). FAM64A (NM_001195228) was used as input. Transcription factors and chromatin regulators with a 10 kb half-decay distance to the transcription start site were predicted. The cutoff value of a relatively high regulatory potential (RP) score was set as 0.3. Visualization of normalized Chip-seq data was performed on Integrative Genomics Viewer (IGV) 2.12 (http:// www. broad insti tute. org/ igv).

Nomogram and Chemotherapeutic Responses Evaluation
A nomogram was built on the predictive model as a graphical presentation in the software package R using the nomogram function from the R package "rms" (https:// cran.r-proje ct. org/ web/ packa ges/ rms/). Chemotherapeutic responses of glioma were assessed by R package "pRRophetic" based on gene expression profiles (Geeleher et al. 2014). Cgp2016 dataset was used as a reference dataset.

Statistical Analysis
Comparisons between groups were done using either Student's t-test or a one-way ANOVA test on GraphPad Prism 8. P-value for multiple comparisons was adjusted by the Tukey-Kramer post hoc test. Kaplan-Meier analysis was used to analyze the data and used the log-rank p-value to compare the two groups. The difference between groups is considered significant at P-values < 0.05.
A nomogram plot including FAM64A, WHO grade, age, and IDH mutation, was successfully constructed to predict glioma prognosis (Fig. 1F). Meanwhile, FAM64A expression could also predict the sensitivity to chemotherapy based on Cancer Genome Project (GGP). Patients with low FAM64A might be more responsive to temozolomide (TMZ), the most commonly used chemotherapy drug for gliomas (P = 2.2e-16, Pearson correlation r = − 0.48; Fig. 1G).

FAM64A Could Promote the Proliferation of Glioma Cells
To define the cell types FAM64A play roles in, a scRNAseq dataset (GSE131928) was analyzed to specify FAM64A expression profiles in different cell types of glioma tissues. After removing the batch effect (Suppl. Fig. 2A), the infiltration of T cells, monocytes/macrophage/microglia in glioma was verified by classic markers (Suppl. Fig. 2B). It was clear that FAM64A was mainly expressed in glioma and glial cells instead of the immune cells ( Fig. 2A).
FAM64A-knockdown U87 cells (shFAM64A) and their scramble controls (shScramble) were successfully established (Suppl. Fig. 2C). shFAM64A U87 cells exhibited significant lower cell viability after 48 h (Unpaired t-test, P = 0.001) and 72 h (Unpaired t-test, P = 0.011) with CCK8 assay (Fig. 2B). A similar phenomenon was observed in U251 cells. FAM64A-knockdown U251 cells exhibited significant lower cell viability after 24 h (Unpaired t-test, P < 0.0001), 48 h (Unpaired t-test, P < 0.0001), 72 h (Unpaired t-test, P = 0.0001 and P < 0.0001, respectively), and 96 h (Unpaired t-test, P < 0.0001) with CCK8 assay (Suppl. Fig. 2D). The immunofluorescent assay indicated that FAM64A-knockdown U87 cells exhibit lower Ki-67 expression (Unpaired t-test, P = 0.017, Fig. 2C). With differential expression analysis according to the TCGA-GBM, pathways associated with cell division (such as cell cycle and oocyte meiosis) were significantly enriched in FAM64A highly expressed samples by KEGG analysis (Fig. 2D). Taken together, FAM64A was mainly expressed in glioma cells and could promote glioma proliferation, probably by affecting cell-cycle. A FAM64A mRNA expression of gliomas in different WHO grades. B FAM64A expression in glioblastoma is higher than in other histological types (astrocytes, oligodendrocytes, oligoastrocytoma). C Kaplan-Meier survival curves of TCGA-GBMLGG cohorts of FAM64A in high expression and low expression, with 95% confidence bands. D FAM64A was detected by immunoblotting in normal brain tissues and glioma samples. E FAM64A protein expression in glioma samples with different grades was assessed by quantitative analysis. F Nomogram plot to predict patients' prognosis based on the FAM64A expression, Grade, age, and IDH status. For example, a patient who was diagnosed with gliomas at the age of 50 years (50 points), with IDH mutation (0 points), grade 2 (0 points), and the normalized FAM64A expression of 4 (20 points), has a total of 70 points. The cumulative points correspond to a survival probability of 87% and 77% at 3 and 5 years respectively. G The sensitivity of temozolomide is inversely correlated with FAM64A expression. Data represents the mean ± SD of triplicate samples, *P < .05, **P < .01, ***P < .001, t test was used to compare two individual groups

FAM64A Expression Could Change Dynamically in a Cell Cycle-Dependent Manner
To demonstrate the details about how FAM64A affected cell-cycle, cell-cycle analysis was conducted on the scRNA-seq profile. Malignant cells were extracted according to their copy number variation (CNV) and cell-cycle analysis showed cells in the different cell cycles are characterized by distinct expression profiles (Suppl. Fig. 3A-B). FAM64A was highly expressed in the G2M phase cells (Fig. 3A). Pseudotime and RNA velocity analysis revealed the temporal trajectory of cells with different FAM64A expressions and showed cells with high FAM64A expression were in the late developmental stage. The primitive glioma cells (in a blue rectangle) had two developmental trajectories. Part of them would enter the stage with high FAM64A expression. This finding indicated that primitive cells might be activated by cell signaling which determines their fate (Fig. 3B).
To validate the effect of FAM64A on the cell cycle, flow cytometry was conducted Intriguingly, FAM64A silencing resulted in a higher proportion of cells in the G2/M phase (P = 0.034) and a lower proportion of cells in S phase (P = 0.007) compared to control samples, indicating that knockdown of FAM64A induces G2/M phase arrest (Fig. 3C). In addition, the expression of G1/S checkpoint proteins and G2/M checkpoint proteins was examined, indicating that FAM64A knockdown could disrupt the cell cycle. Specifically, FAM64A knockdown was found to affect the expression of cell cycle-related factors (E2F1, cycling D1, GADD45, and Phospho-Rb), which may lead to G2/M phase arrest. However, no differences in expression of CDK2, Rb and cyclin B1 were observed between FAM64A silencing cells and non-silencing cells, indicating that FAM64A does not regulate these proteins (Fig. 3D).

FAM64A Could be Induced by TGF-β Pathway Activation
To decipher the potential signaling pathway that regulate FAM64A expression, we intercellular crosstalk analysis of GSE131928. TGF-β signaling communication was found between tumor-associated macrophages and microglia (TAMs) and neoplasm cells, especially neoplasm cells with high FAM64A expression ( Fig. 4A and Suppl. Fig. 4A), suggesting that TGF-β might be responsible for the initiation of FAM64A expression. TGF-β1 with TGFBR1 and TGFBR2 pairs contributed most to the signaling communication (Suppl. Fig. 4B). CistromeDB toolkit further demonstrated that TGF-β downstream transcription factors (SMAD2/3, SMAD3, and SMAD4) had a regulatory potential higher than 0.3 on FAM64A, reaffirming that that FAM64A might be regulated by the TGF-β signaling (Fig. 4B). In addition, previous ChIP-seq datasets also showed TGF-β associated transcriptional factors could bind to the promoter of FAM64A (Suppl. Fig. 4C), while FAM64A expression showed a downregulating trend in the Smad3 knock out embryonic cells (Suppl. Fig. 4D). In vitro cultivation with exogenous TGF-β further confirmed that TGF-β stimulation could enhance glioma cell proliferation with upregulated FAM64A expression (Fig. 4C).
To further investigate the association between TGF-β and FAM64A, glioma samples of different WHO grades were collected and TGF-β, FAM64A, and TGF-β downstream transcription factors were detected. ELISA and immunoblotting assays showed the concentration of TGF-β had a positive correlation with FAM64A expression, while grade 4 gliomas have a higher TGF-β level compared with normal brain tissues, grade 2, and grade3 gliomas (one-way ANOVA, P = 0.004; Tukey-Kramer test, P = 0.011, 0.009, 0.026 respectively; Suppl. Figure 4E). IHC analysis also demonstrated similar results (Suppl. Fig. 4F). As expected, SMAD3 and SMAD4 were also higher expressed in highgrade glioma (Fig. 4D). These proteins showed a high correlation with each other (Fig. 4D). Immunofluorescence also showed increased number of FAM64A and SMAD4 silencing groups (shFAM64A#1 and shFAM64A#2). Control groups exhibited higher E2F1, cyclin D1, GADD45, and Phospho-Rb expression compared with FAM64A silencing groups. The protein level of E2F1, cyclin D1, cyclin B1, GADD45, CDK2, Phospho-Rb, and Rb in cells was determined by western blotting, normalized by β-TUBULIN colocalization in high grade gliomas (one-way ANOVA, FAM64A P = 0.002; SMAD4 P = 0.002; Fig. 4E). The proportions of cells expressing FAM64A and SMAD4 in grade 4 gliomas were higher than grade 2 or 3 gliomas (Tukey-Kramer test, FAM64A, P = 0.002 and 0.008; SMAD4, P = 0.002 and 0.029; Fig. 4F). Additionally, Fig. 4 FAM64A could be induced by TGF-β signaling. A Heatmap showing the TGF-β signaling pathway network among cells (T cells, TAMs, FAM64A highly expressed neoplasm cells, FAM64A lowly expressed neoplasm cells). TAMs are both the strongest TGFβ signaling senders and receptors. B Transcription factor prediction for FAM64A. TGF-β associated transcription factors have a high regulatory potential score. C Enhanced TGF-β signaling promoted proliferation of U87 cells and induced FAM64A expression. D The concentration of TGF-β and expression of FAM64A, SMAD3, and SMAD4 in glioma samples. E Immunofluorescence testing was performed with a Leica confocal immunofluorescence microscope. High-grade glioma has a higher expression of FAM64A and SMAD4. F FAM64A co-localizes with SMAD4, a transcription factor of the TGF-β signaling pathway FAM64A expression has distinct correlation with SMAD4 (Pearson r = 0.76, P = 0.017). Importantly, colocalization of FAM64A and SMAD4 could be observed in grade 4 gliomas (Fig. 4E). Together, these results suggest that FAM64A could be upregulated by TGF-β signaling in glioma.

TGF-β Induced EMT is Dependent on FAM64A
TGF-β could induce EMT in multiple tumors (Xu et al. 2009;Soukupova et al. 2021;Hao et al. 2019). To further define the role of FAM64A in TGF-β-induced EMT, EMT related marker (N-cadherin) was detected via western blotting analysis. The expression level of the N-cadherin was lower in the shFAM64A groups compared with shScramble groups.
To further gain insights of regulatory effects by TGF-β on FAM64A knockdown cells, shFAM64A and shScramble cells were treated with various concentrations of TGF-β (0, 10 ng/mL, 50 ng/mL, 200 ng/mL) for 48 h and EMT markers expression was quantificated by western blotting. For shScramble cells, the expression levels of N-cadherin were higher in TGF-β treated samples compared with untreated ones. No such trends existed in FAM64A-knockdown cells, indicating that FAM64A plays an essential role in TGFβ-induced EMT process (Fig. 5A-B and Suppl. Fig. 4G). Scratch wound healing assay showed FAM64A knockdown could inhibit cell migration significantly as well (Unpaired t-test, P = 0.003, Fig. 5C), further supporting the notion that FAM64A could mediate TGF-β-induced EMT and promote glioma cells migration.

Discussion
The World Health Organization (WHO) classification of CNS tumors is periodically updated to reflect advances in our understanding of tumor biology. The current version, CNS5, incorporates molecular information to supplement traditional histological criteria for diagnosis and classification (Louis et al. 2021). Recent studies in molecular biology have further improved our understanding of glioma pathogenesis and the identification of potential therapeutic targets. Fu et al. 2021). Additionally, cell cyclerelated proteins have attracted attention in the molecular diagnosis and treatment of gliomas (Zhu et al. 2021;Wen et al. 2019;Bronner et al. 2019). Abnormal cell cycle progression is one of the mechanisms underlying tumorigenesis, making cell cycle regulators potential therapeutic targets (Liu et al. 2022).
In this study, we focused on FAM64A, a cell cycle-related protein upregulated in various human cancers. Our results demonstrated that knockdown of FAM64A inhibited glioma cell proliferation by inducing G2/M phase arrest and downregulating classical cell cycle-related proteins (E2F1, cycling D1, GADD45, and Phospho-Rb), which is in accordance with the previous study on other solid tumors (Jiao et al. 2019;Yao et al. 2019;Jiang et al. 2020). We also found that FAM64A could be stimulated by the TGF-β signaling pathway, which is widely expressed in gliomas and plays a crucial role in glioma pathogenesis (Kaminska et al. 2013;Kaminska and Cyranowski 2020). Moreover, we revealed the association between FAM64A and EMT in gliomas. Taken together, the TGF-β signaling activated downstream transcription factors (SMAD3, and SMAD4) and upregulated the expression of FAM64A. FAM64A mediated TGF-β-induced EMT process and confers poor prognosis of glioma (Fig. 6). Activation of EMT is a key process in cancer cell metastasis, during which epithelial cells acquire characteristics of mesenchymal cells, with enhanced cell motility and migration capabilities. In gliomas, EMT is not a metastatic process, but may result in a widespread dissemination (Bellail et al. 2004). This process is followed by the change of N-Cadherin and E-Cadherin and regulated by a complex network of signaling pathways and transcription factors . Our findings suggest that FAM64A confers a poor prognosis in gliomas and could serve as a potential therapeutic target.
Furthermore, we found that FAM64A could predict glioma sensitivity to DNA-methylating agents like TMZ. This is an important finding, as TMZ is a commonly used chemotherapeutic drug for glioma treatment. (Friedman et al. 2000;Stupp et al. 2005;Singh et al. 2021). Currently, MGMT promoter methylation is widely used for predicting TMZ sensitivity (Singh et al. 2021), but additional mechanisms and predictive factors are needed to improve treatment outcomes (Hegi et al. 2005). our study suggests the glioma with low expression of FAM64A may have a higher sensitivity to TMZ. This would be due to FAM64A and TMZ has the opposite effect on the cell cycle. We have uncovered that FAM64A knockdown led to G2/M phase arrest, while TMZ resulted in cell cycle arrest at G2/M phase ). Archangelo et al. also revealed that in glioblastoma cells, when the cells enter the S phase, FAM64A would be progressively upregulated as the cell cycle progresses, and the FAM64A biological protein level was the highest in the G2 phase (Archangelo et al. 2013), which hinted that FAM64A promoted the G2 to M phase transition. Higher expression of FAM64A might promoted the passage of cells through G2/M phase cell checkpoints and antagonizes the cytotoxic effects of TMZ.
Although there are some strengths to this study, there are also some limitations. Tumor progression is the result of a complex and multifaceted interaction in TME. Due to the blood-brain barrier (BBB), TME in brain tumors is distinguished from other solid tumors (Daneman and Prat 2015). In this study, only neoplasm cell lines were 1 3 investigated for FAM64A functions. There is a major difference between glioma cell lines and intracranial tumors. To bridge this gap, we analyzed published scRNA-seq datasets of GBM and performed cell crosstalk analysis, which indicated the TGF-β signaling pathway played important roles in the interactions between neoplasm cells and immune cells (mainly tumor-associated macrophages, and microglia in CNS) (Gratchev 2017).

Conclusions
In conclusion, our study sheds light on the role of FAM64A in glioma proliferation and its regulatory mechanism. Targeting FAM64A could be a promising strategy for glioma treatment, and FAM64A could serve as a predictive biomarker for TMZ sensitivity. Our findings A TGF-β signaling upregulated SMAD3 and SMAD4 in U87 scramble controls and FAM64A silencing groups (shFAM64A#1 and shFAM64A#2). TGF-β of low concentration (10 ng/mL) could promote N-Cadherin expression. N-Cadherin expression decresased with the elevated TGF-β concentrations (50 ng/mL and 200 ng/mL). Control groups exhibited higher N-cadherin expression compared with FAM64A silencing groups (Paired t-test, P = 0.0090, 0.0037, respectively). The protein level of N-Cadherin, SMAD3, and SMAD4 in cells was determined by western blotting, normalized by β-TUBULIN. B TGF-β signaling upregulated SMAD3 and SMAD4 in U251 scram-ble controls and FAM64A silencing groups (shFAM64A#1 and shFAM64A#2). TGF-β of low concentration (10 ng/mL) could promote N-Cadherin expression. However, N-Cadherin expression decresased with the elevated TGF-β concentrations (50 ng/mL and 100 ng/mL). Control groups exhibited higher N-cadherin expression compared with FAM64A silencing groups (Paired t-test, P = 0.0234, 0.0417, respectively). The protein level of N-Cadherin, SMAD3, and SMAD4 in cells was determined by western blotting, normalized by β-TUBULIN. C Cell migration ability was determined by cell scratch assay. FAM64A silencing U87 cells have a higher migration ability compared with scramble control groups 1 3 provide new insights into the molecular mechanisms underlying glioma pathogenesis and suggest potential avenues for developing new therapies.