Up‐regulated SAMD9L modulated by TLR2 and HIF‐1α as a promising biomarker in tuberculosis

Abstract The aim of this study was to identify potential biomarkers of TB in blood and determine their function in Mtb‐infected macrophages. First of all, WGCNA was used to analyse 9451 genes with significant changes in TB patients’ whole blood. The 220 interferon‐γ‐related genes were identified, and then 30 key genes were screened using Cytoscape. Then, the AUC values of key genes were calculated to further narrow the gene range. Finally, we identified 9 genes from GSE19444. ROC analysis showed that SAMD9L, among 9 genes, had a high diagnostic value (AUC = 0.925) and a differential diagnostic value (AUC>0.865). To further narrow down the range of DEGs, the top 10 hub‐connecting genes were screened from monocytes (GSE19443). Finally, we obtained 4 genes (SAMD9L, GBP1, GBP5 and STAT1) by intersections of genes from monocytes and whole blood. Among them, it was found that the function of SAMD9L was unknown after data review, so this paper studied this gene. Our results showed that SAMD9L is up‐regulated and suppresses cell necrosis, and might be regulated by TLR2 and HIF‐1α during Mtb infection. In addition, miR‐181b‐5p is significantly up‐regulated in the peripheral blood plasma of tuberculosis patients, which has a high diagnostic value (AUC = 0.969).

Toll-like receptors (TLR) as pattern recognition receptors (PRR) can recognize multiple PAMP on the surface of Mtb. There is much literature showing that a variety of TLRs are involved in the entry process of Mtb and are closely related to the outcome of Mtb after entry. [5][6][7] Among them, TLR2 and TLR4 deserve special attention. 5,8,9 The aim of this study was to identify potential biomarkers of tuberculosis in blood and determine their function in Mtb infected macrophages. Bioinformatics methods were used to screen out genes that played a significant role in Mtb infection from the whole blood sample. In order to further narrow the range, four core genes were obtained by intersection with genes that played a significant role in monocytes. Subsequently, relevant experiments were designed to explore the function and upstream molecules of key genes. This will provide new ideas for the diagnosis and treatment of TB.

| Hub gene screening cohort
To select potential biomarkers, we first analysed whole blood samples (GSE19444, select 12 HC and 21 ATB) to obtain gene modules that were significantly changed in TB by WGCNA. 10 We then construct the protein-protein interaction (PPI) network of key genes in this module. Then Cytohubba plug-in in Cytoscape software was used to narrow the gene range further. AUC values of key genes were calculated to further screen key genes with diagnostic values.
In order to further narrow the range of key genes, and obtain differentially expressed genes (DEGs) with the value of mechanism research. We used GEO 2R and Cytohubba plug-in to screen out key genes from the monocyte database (GSE19443, select 4 HC and 7 ATB).
Taking the intersection of key genes from whole blood and monocytes, we end up with genes that have both diagnostic value and mechanistic value.

| Hub gene verification cohort
Whole blood samples are easy to obtain clinically, so they are suitable for screening diagnostic markers. In order to further explore the differential diagnostic value of key genes screened from whole blood, we extracted the expression data of key genes from GSE 42826 database, drew scatter plots and ROC curves, and calculated their AUC values.
In clinical treatment, how to evaluate the effectiveness of treatment measures, so appropriate monitoring indicators are needed. To this end, we further explore the changes of these genes during treatment, hoping to monitor their expression levels to evaluate whether treatment measures are appropriate and prevent the occurrence of undertreatment and overtreatment. The changes of expression levels of key genes in the database during treatment (GSE19435) were extracted, and the scatter diagram and broken line diagram were drawn.  After transfection about 24 h, Mtb infection was carried out at a ratio of 1:10. Twenty-four hours after infection, we discarded the supernatant and 200 μLPBS was added, followed by 5 μL AO and 5 μL EB staining solution. After mixing, the cells were incubated for 5 min and kept in dark place. And then, the supernatant was discarded and 500 μL PBS which containing 25 μL Hoechst and 25 μL PI staining solution was added. After mixing, the cells were incubated for 30 min, and then washed twice with PBS. Next, fluorescence microscope was used to observe. Finally, acid-fast staining was performed on the cell slide, and the number of acid-fast bacilli in 100 cells was counted under the oil microscope.

| To explore the effect of upstream genes on key genes
In order to explore the regulatory role of upstream genes on key genes, we interfered with TLR2 of RAW264.7 cells, then infected with Mtb, and subsequently measured the expression level of SAMD9L. TLR4 in the same way. In addition, to investigate whether there is a regulatory relationship between HIF-1α and key genes, we added HIF-1α inhibitor into the medium, co-incubated with RAW264.7 cells for 24 h, and then infected Mtb. Subsequently, the expression level of SAMD9L was determined.

| miRNAs prediction of key hub gene and validation
miRWalk, TargetScan and StarBase were used to predict the miRNAs which may target key hub genes. Then, we take the intersection of the prediction results of the three databases and draw a Venn diagram. We selected the database GSE13 1708 (PBMC, 4 TB meningitis patients, 4 HC) to verify the reliability of the predicted TB miRNAs. Tuberculosis patients who did not receive anti-tuberculosis treatment at initial diagnosis. Patients with co-existing cancer, diabetes, autoimmune disease, hypertension, chronic inflammation and pregnant or lactating women were excluded from the active TB group.

| Clinical samples were collected and quantified by quantitative reverse transcriptionpolymerase chain reaction (qRT-PCR)
Informed consent was obtained from all patients prior to beginning the study.
Total RNA was extracted from each sample using Trizol-LS (Life Technologies, USA) according to the manufacturer's instruction.
Purified RNA was reverse transcribed to cDNA using Prime-ScriptH RT reagent Kit (TaKaRa) according to the manufacturer's protocol.
qRT-PCR was then performed using SYBRTM Green PCR Master Mix (Takara) following standard conditions on LightCycler 480 Ⅱ real-time PCR system (Roche). The relative amount of expressed miRNA was calculated by comparison between case and control using the 2 −ΔCt method.

| Dataset preprocessing and key gene screening
3.1.1 | The preliminary screening (GSE19444, whole blood samples) To screen out the key genes and the main biological processes that changed during TB, we downloaded gene expression matrix from GSE19444, selected 9451 genes with the highest expression variance in the top 25% after probe annotation (indicating high variability among different groups, including up-and down-regulated genes), the gene expression was filtered with mean FPKM = 0.5 as the standard and 9439 genes remained. Subsequently, in order to build a stable co-expression network, outlier sample GSM484628 was removed ( Figure 1 and Figure 2A). Taking R 2 = 0.85 as the criterion to meet the scale-free condition, we chose the optimal weighting parameter β = 9 (R 2 = 0.856). Not only ensured that the network was close to the scale-free network and the minimum threshold, but also ensured that the average connection degree of the network was not too small, which was conducive to the network to contain enough information. The inclusion criteria for important genes in brown module was set as follows: GS>0.30 and MM>0.80. Then we got 220 important genes and uploaded it to DAVID. We can see type Ⅰ interferon play a significant role in whole blood (Table 1). Simultaneously, 220 genes were uploaded to STRING and the results were imported into Cytoscape. According to node degree, the top 30 hub genes were identified by CytoHubba ( Figure 2C).
To further screen the essential genes that can distinguish TB patients from HC, expression levels of 30 hub genes were extracted from GSE19444 (Table 2). SPSS was used to calculate the AUC, and we set up AUC>0.9, log 2 FC>1.0 as filter criteria to narrow the range of hub genes down, finally obtained 9 essential  To screen out the essential up-regulated DEGs, we uploaded 206 up-regulated DEGs to STRING for further analysis. The PPI network was analysed by Cytoscape. According to node degree, the top 10 hub genes were identified by CytoHubba ( Figure 3C).

| Evaluation of differential diagnostic ability
Whole blood samples are easy to obtain, easy to handle and more suitable for diagnosis. So, to further evaluate differential diagnosis of these 9 gene in TB, sarcoidosis, pneumonia and lung cancer, we extracted their expression data from GSE42826 for comparison and drew ROC curves. The relative expression levels of these 9 genes in TB were higher than in another group ( Figure 2D).
Moreover, we drew ROC curves and calculated AUC of the 9 genes. The AUC of 9 genes were above 0.7, respectively, showing good diagnostic performance in differentiating TB patients from HC and another disease ( Figure 2E, Table 3).
In addition, to further evaluate the monitoring ability of 4 genes in drug therapy for TB patients, we extracted the expression data of 4 genes in GSE19435, and drew scatter plots and broken line plots ( Figure 3D,E). Blood was taken from the control, pre-treatment (PTB TA B L E 1 Top 10 of the BP analysis of 220 important genes in GSE19444

| SAMD9L can inhibit cell necrosis, which
is not conducive to Mtb survival during Mtb infection siRNA with the most significant interference effect was screened ( Figure 4C). After cell transfection, the cells were infected for 24 h and stained for observation. Next, AO/EB and Hoechst/PI staining were used to test the effects of SAMD9L in RAW264.7. We found that compared with NC-siRNA, necrosis was significantly increased after SAMD9L inhibition. This suggests that SAMD9L may not conducive to the intracellular survival of Mtb.

| The effect of TLR2, TLR4 and HIF-1α
TLRs and HIF-1α play an important role in Mtb infection. To further prove whether there is a regulatory relationship between TLR2, TLR4 and HIF-1α, we interfered with TLR2 and TLR4 and inhibited HIF-1α, respectively, followed by Mtb infection. We found that the expression of SAMD9L was significantly decreased after interference with TLR2 and inhibition of HIF-1α. This implies that TLR2 and HIF-1α regulate SAMD9L. However, after TLR4 interference, the expression of SAMD9L only decreased. ( Figure 5A).

| miRNAs prediction and validation
miRNA is non-coding small molecule that can play an essential role by suppressing gene expression. We tried to find miRNA that could interact with SAMD9L. To make the prediction more credible, we used StarBase, Targetscan and miRWalk to predict it. We got 25 miRNAs after intersecting of the results ( Figure 5B). We selected GSE13 1708 dataset to verify the reliability of the above miRNAs in TB. Finally, we found that compared with HC, miR-146a-5p and miR-181b-5p were significantly reduced in TB.

| miRNA revalidation in Mtb-infected macrophage and plasma from TB patients
Based on previous research by others, miR-146a-5p was proved to increase significantly in Mtb-infected RAW264.7 and exosomes from it. 11 Therefore, we chose to measure the expression of miR-181b-5p. Although the expression of miR-181b-5p was not statistically significant, it showed an upward trend after infection in RAW264.7 and tissue ( Figure 5C).
What is the expression level of miR-181b-5p in the plasma of TB patients? To explore this question, we collected plasma from 8 HC and 8 TB patients, then has-miR-181b-5p was quantitatively analysed by qRT-PCR. The results showed that the expression of miR-181b-5p was a significant increase in the plasma from TB patients ( Figure 5D).

Count p value Genes
interferon-gamma-mediated signalling pathway TB screened in the original text. 17 We intersected critical genes from the whole blood and monocyte and found that the SAMD9L may play a role in TB. SAMD9L is a type Ⅰ interferon-induced gene that has important roles during virus infection and innate immunity. Its role in the control of influenza virus, poxvirus infection, and also pathogenesis has been proposed. [18][19][20] We searched for SAMD9L related pathways through KEGG but found no pathway information. Therefore, the function of SAMD9L in tuberculosis was preliminarily explored in this paper. Therefore, in order to explore whether there is a regulatory relation-  between miR-181b-5p and SAMD9L. However, miR-181b-5p can be a good diagnostic marker for TB.
In conclusion, we found that SAMD9L was significantly upregulated in whole blood of TB patients, monocytes in whole blood, Mtb infected RAW264.7 cells, and BMDM, and showed an upward trend in lung and spleen of Mtb infected mice. SAMD9L has good diagnostic value and differential diagnostic value. SAMD9L inhibits cell necrosis in Mtb infection. In Mtb infection, TLR2 and HIF-1α regulate the expression of SAMD9L. In addition, although the experiment verified that there was no interaction between miR-181b-5p and SAMD9L, miR-181b-5p can be used as a good diagnostic marker for TB.

ACK N OWLED G EM ENTS
The current study was supported by the Program of Shandong Province Natural Science Foundation of China (No. ZR2021MH401).
The funder had no role in the design of the study and collection, analysis or preparation of the manuscript.

CO N FLI C T S O F I NTE R E S T S
All the authors declare that there are no conflicts of interest relevant to this article.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used for the current study are available from the corresponding author upon reasonable request.