Keywords
Endometriosis, GEO, Diagnostic marker, Nomogram, Immune cells inltration
Posted Date: May 3rd, 2023
DOI: https://doi.org/10.21203/rs.3.rs-2830815/v1
Page 2/28
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Page 3/28
Abstract
Endometriosis (EMS) is a common gynecological disease leading to chronic pelvic pain and infertility in
women of reproductive age, but its underlying pathogenic genes and effective treatment are still unclear.
To date, abnormal expression of NLRP3 activation-related genes has been identied in EMS patients and
mouse models. Therefore, this study sought to identify the key genes that could affect the diagnosis and
treatment of EMS. The GSE7307 dataset was downloaded from the Gene Expression Omnibus (GEO)
database, including 18 EMS samples and 23 control samples. 14 differential genes related to NLRP3
activation and EMS were obtained from the endometrial samples of GSE7307 by differential analysis. GO
and KEGG analysis showed that these genes were mainly involved in the production and regulation of the
cytokine IL-1 β , and the NOD-like receptor signaling pathway. Random Forest (RF) and support vector
machine recursive feature elimination (SVM-RFE) algorithms were used to select four diagnostic markers
related to NLRP3 activation (NLRP3, IL-1 β , LY96 and PDIA3) to construct the EMS diagnostic model. The
four diagnostic markers were veried using western blotting and validated in the GSE7305 and
GSE23339 datasets. The AUC values showed that the model had a good diagnostic performance. In
addition, the inltration of immune cells in the samples and the correlation between different immune
factors and diagnostic markers were further discussed. These results suggest that four diagnostic
markers may also play an important role in the immunity of EMS. Finally, 10 drugs targeting to four
diagnostic markers were retrieved from the DrugBank database, of which niclosamide proved useful for
treating EMS. Overall, we identied four key diagnostic genes for EMS. In addition, large-scale and
multicenter prospective cohort studies are necessary to conrm whether these four genes also have valid
diagnostic value in blood samples from EMS patients.
Introduction
Endometriosis (EMS) is a chronic inammatory disease dened as the presence of endometrial tissue
outside the uterus, causing pelvic pain and infertility, affecting 5–10% of women of reproductive age
worldwide[1, 2]. The most prevalent theory of the origin of endometriosis, advanced by Sampson (1928),
holds that during menstruation, endometrial cells and tissue fragments return through the fallopian tubes,
survive, attach and invade pelvic structures[3]. Depending on the location of the disease, EMS is
classied as supercial (or peritoneal), ovarian (endometrioma), and deep (dened as invasive lesions
greater than 5 mm in depth). The diagnosis of EMS is made primarily through laparoscopy, based on an
empirically guided clinical decision based on history, symptoms, clinical examination, and imaging,
whereas biochemical markers lack sensitivity and specicity[4, 5]. Therefore, it is necessary to explore the
underlying molecular mechanism and nd new markers to provide more meaningful and reliable
diagnostic information for EMS treatment, so as to promote early diagnosis and individualized treatment.
Nod-like receptor protein 3 (NLRP3) is a component of the inammasome that activates caspase1,
promoting the processing of IL-1 β and IL-18 into biologically active mature forms and inducing
pyroptosis[6, 7]. Furthermore, NLRP3 inammasome-mediated pyroptosis has been reported to play a
crucial role in the pathogenesis of inammatory diseases[8, 9]. EMS is also a chronic inammatory
Page 4/28
disease, and previous studies have also reported an association between NLRP3 inammasome and the
progression of EMS[10, 11]. A study has reported that targeted inhibition of NLRP3 can inhibit the
production of IL-1 β and cell proliferation in EMS, thereby improving ovarian function in EMS[12]. Estrogen
receptor(ER) β has been reported to interact with components of the inammasome in endometriotic cells,
increasing interleukin-1 β to enhance their cell adhesion and proliferative properties, thereby increasing the
invasive activity of endometrial tissues to establish ectopic lesions[13]. Due to the non-negligible
relationship between NLRP3 activation and EMS, it is necessary to explore the specic role of NLRP3
activation-related features in EMS and identify more diagnostic biomarkers.
In this study, we conducted a systematic study of NLRP3 activation-related genes to explore the
differential expression of these genes in EMS patients and normal controls and to construct an effective
diagnostic nomogram model for clinical application. Inammation is a vital pathological feature of EMS,
which is closely related to immune function. Therefore, we further explored the immune penetration of
these samples and the correlation between different immune factors and these diagnostic biomarker
candidates. In addition, ten drugs targeting these diagnostic biomarker candidates were retrieved from
the DSigDB database, which has implications for the pharmacotherapy of EMS. Therefore, this study
provides potential targets for the diagnosis and treatment of EMS. The workow of this study is shown in
Fig. 1.
Materials and methods
Data acquisition
First, the three datasets, GSE7307 (including 18 EMS samples and 23 normal samples) , GSE7305
(including 10 EMS samples and 10 normal samples), and GSE23339 (including 10 EMS samples and 9
normal samples ) were downloaded from Gene Expression Omnibus (GEO) database. The data
processing steps are as follows: (1) match multiple probes to the same gene, (2) remove the null probes,
(3) rst match probes to genes, and (4) log2 transformation of the obtained data set. The number of bits
used to normalize signal strength.
DEGs in EMS
Gene expression difference data between EMS samples and normal samples were examined by using a
limma package in R and using a threshold with a modied P value of 0.05 or less.
Functional enrichment analysis of the DEGs
To explore the biological mechanism of differentially expressed genes (DEGs) affecting EMS, a
functional enrichment analysis was carried out. First, the “clusterProler” and “ggplot2” packages were
employed for Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) was used
Page 5/28
for enrichment analyses of DEGs[14]. Adjusted P < 0.05 had signicance in statistics. Subsequently, Gene
Set Enrichment Analysis (GSEA) that used the ‘Clusterproler’ R package to perform revealed the
respective function of each key gene. The samples were divided into low-expression groups and high-
expression groups according to median values of the expression levels in key genes. The genes were
sorted from highest to lowest by logFC, and the rst 5 and last 5 GSEA results were taken after sorting.
These gene sets were downloaded from the Molecular Signature Database (MSigDB, https://www.gsea-
msigdb.org/gsea/msigdb/ index.jsp). The nal result is displayed in the "enrichplot" package of the R
software. P-values < 0.05 was considered to be statistically signicant.
Screening for candidate diagnostic biomarkers
Diagnostic markers of EMS were classied using random forest (RF) and support vector machine
recursive feature elimination (SVM-RFE) arithmetic methods. The RF algorithm randomizes a single
decision tree overtting reduction algorithm and improves model accuracy based on many related
decision trees in the training set[15]. SVM-RFE as a support vector monitoring machine learning method
identies optimal variables via the deletion of the SVM-produced eigenvectors[16]. Markers selected for
EMS diagnosis were classied and analyzed using the SVM classier in the R package e1071.
Verication of the diagnostic biomarker results
Endometrium from endometriosis patients or matched control patients came from the Aliated Hospital
of Youjiang Medical College for Nationalities, Baise, China. We collected 12 cases diagnosed as ovarian
endometriosis cysts (n=6) and non-endometriosis (n=6) with inclusion conditions: ageg between 26 and
46 years, BMI of 21-27 kg/m2, non-vegetarian patients, no contraindications for surgery; The exclusion
criteria were endocrine diseases such as diabetes, and severe gastrointestinal, heart, lung, and liver
diseases. All the donors did not take drugs and hormones before surgery, and underwent combined
uterine and abdominal surgery at the same time due to abnormal uterine bleeding. All tissue samples
were obtained from tissue resected during surgery and were approved and informed by the patient and
the ethics review committee of our hospital. Protein preparation and western blotting were performed as
described previously[17]. Endometrial tissue lysates (500 μ L in RIPA lysis buffer PMSF= 94:6) were
processed for western blotting, where protein concentrations were determined with BCA Protein Assay Kit.
Then, proteins were denatured by boiling at 100 °C for 10 min. SDS-PAGE was conducted with denatured
protein samples of approximately 20 μ g loaded onto a 10% Tris–acetate gel, run at 120 V for 2 h.
Proteins in the gel were transferred onto polyvinylidene uoride membranes at 300 mA for 50 min, and
blocked in PBST containing 5% skim milk powder for 2 h. After washing with PBST, the membranes were
respectively incubated with antibodies against NLRP3 (1:1000 dilution, bs-10021R; Bioss, Beijing, China),
LY96 (1:1000 dilution, 11784-1-AP; proteintech, Wuhan, China), IL-1 β (1:3000 dilution, 26048-1-
AP; proteintech), PDIA3 (1:8000 dilution, 15967-1-AP; proteintech), and TUBULIN (1:20000 dilution, 66031-
1-Ig; proteintech) overnight at 4 °C, and washed with PBST three times for 10 min each. It was then
incubated for 2 h at room temperature with a goat anti-rabbit IgG (1:5000 dilution, SA00001-
Page 6/28
2; proteintech) and a goat anti-mouse IgG (1:5000 dilution, SA00001-1; proteintech) horseradish
peroxidase (HRP)-conjugated secondary antibody. After three 10-min washes of the membrane in PBST,
eECL was added, and the Tanon-5500 Chemiluminescence Imaging System was used to detect the
chemiluminescence of protein bands.
Diagnostic value of the key genes in EMS
The generalized linear model (GLM), support vector machine model (SVM), extreme gradient boosting
model (XGB), and random forest model (RF) were performed based on the combined training dataset.
The four models were analyzed using the interpretation function of DALEX package in R, and the residual
distribution was performed to obtain the optimal model. We constructed ROC curves to examine the
predictive utility of the proposed biomarkers by using mRNA expression data from 23 normal samples
and 18 EMS samples. EMS and control samples were separated by ROC analysis and further validated in
GSE7305 and GSE23339 datasets.
Distribution of the immune cells
The CIBERSORT (http://cibersortx.stanford.edu) algorithm was used to calculate the relative proportions
of the 22 immune cell types in the training dataset samples.
Analysis of genes identied and immune cells inltrated
Spearman rank correlation analysis was used to determine the correlation between the four key
diagnostic genes and the number of inltrating immune cells. The graph method of the ggplot2 package
is used to show the correlation of results.
Construction of nomogram and diagnostic score model
First, a nomogram model for predicting the occurrence of EMS was constructed using the rms package
(version 5.1-2)[18]. Then, the calibration curve was constructed to assess the diagnostic power of
the nomogram model. Finally, the clinical value of the nomogram model was evaluated by using decision
curve analysis (DCA).
Prediction of potential drug compounds
The Drug Signature Database (DSigDB) was used to screen potential drug molecules with gene-based
interactions. As a public web-based database derived from an enrichment web server, DSigDB provides
information about drugs and their target genes for GSEA[19]. Based on a statistical threshold of adjusted
Page 7/28
p-value < 0.05, we identied protein-drug interactions in the EMS-pyroptosis interaction that may affect
important central genes.
Statistical analysis.
All analyses were performed in R program 4.0.2. The t-test and the Mann-Whitney U-test were chosen
based on whether the data t a normal distribution. The Spearman correlation method was used to study
the relationship between the expressions of diagnostic genes and inltration immunocytes. P values <
0.05 was considered statistically signicant.
Results
Determination of DEGs in EMS
To investigate DEGs of EMS, we examined the GSE7307 dataset and identied 14 DEGs in EMS, including
11 up-regulated genes (CASP1, IL10, IL17RA, IL1B, IL1R1, IL6, LY96, NLRP3, P2RX7, PIK3CA and TLR4)
and 3 down-regulated genes (EIF2AK3, HMGB1 and PDIA3) (Fig. 2A and 2B).
Enrichment analysis of biological process and pathway
GO analysis results showed that DEGs were mainly rich in positive regulation of cytokine production,
cellular response to biotic stimulus, cellular response to lipopolysaccharide, cellular response to molecule
of bacterial origin, response to lipopolysaccharide, interleukin-1 production, regulation of interleukin-1
production, positive regulation of interleukin-1 beta production, positive regulation of interleukin-1
production, and regulation of cytokine production involved in the immune response (Fig. 3A). KEGG
assays revealed that DEGs were mainly enriched in the NOD-like receptor signaling pathway (Fig. 3B).
Detection and verication of diagnostic genes in EMS
Two machine learning methods were applied in selecting potential markers of EMS: SVM-RFE and RF.
We rst screened 14 DEGs with SVM-RFE analysis and identied 10 diagnostic biomarkers (Fig. 4A). RF
analysis identied 4 diagnostic biomarkers from 14 DEGs (Fig. 4B and 4C). After integrating the
characteristic genes from SVM-RFE and RF, four characteristic genes closely related to pyroptosis were
obtained: NLRP3, IL-1 β , LY96, and PDIA3 (Fig. 4D). The expression of NLRP3, IL-1 Β , and LY96 was
distinctly upregulated in EMS, and the expression of PDIA3 was distinctly upregulated in EMS (Fig. 5A).
The diagnostic value of four overlapping genes was shown by the use of ROC assays (Fig. 5B). We
veried the results of the screened differential genes at the protein level. The verication results revealed
that the protein (Fig. 6A and 6B) levels of NLRP3 (P<0.05), LY96 (P<0.01) and IL-1 β (P0.05) were the opposite, consistent
Page 8/28
with the Gene Chip data with multiple exposures. Moreover, we further determined the diagnostic value of
NLRP3, IL-1 β , LY96 and PDIA3 in the GSE7305 and GSE23339 datasets, and the results are shown in Fig.
7. We established the RF, SVM, XGB, and GLM models to choose NLRP3 activation-related genes to
predict the incidence of EMS. Both “Reverse Cumulative Distribution of Residual” (Fig. 8A) and “Boxplots
of Residual” (Fig. 8B) indicate that the GLM model has minimal residuals. The ROC curve was drawn to
evaluate the model, and the AUC value of the ROC curve also indicated that the GLM model had higher
accuracy compared with the other three models (Fig. 8C). Finally, the ROC curve also demonstrated the
good diagnostic performance of the four genetic biomarker combinations based on the four
classication models in both validation datasets (GSE7305 and GSE23339) for normal women and EMS
patients. (Fig. 8D and 8E).
Analysis of functional correlations
Gene set enrichment analysis (GSEA) was performed for the four key genes based on the median
expression prole. GSEA expression analysis of four genes in NLRP3 showed that the gene expressions
of malaria, pertussis, systemic lupus erythematosus immune-related diseases and biological processes
of complement and coagulation cascades and taurine and hypotaurine metabolism were signicantly
increased in the high expression group, while base excision repair, DNA replication, Fanconi anemia
pathway, homologous recombination, and mismatch repair in the low expression group (Fig. 9A). The IL-
1 β in the high expression group were signicantly enriched in the legionellosis, malaria, systemic lupus
erythematosus, complement and coagulation cascades, and viral protein interaction with cytokine and
cytokine receptor, whereas those in the low-expression groups were signicantly enriched in
aminoacyl−tRNA biosynthesis, base excision repair, DNA replication, homologous recombination, and
mismatch repair (Fig. 9B). The LY96 in the high-expression group was signicantly enriched in asthma,
legionellosis, malaria, systemic lupus erythematosus, and complement and coagulation cascades. On the
other hand, LY96 in the low expression groups was signicantly enriched in base excision repair, DNA
replication, Fanconi anemia pathway, homologous recombination, and mismatch repair (Fig. 9C).
Ultimately, the PDIA3 in the high expression group was signicantly enriched in DNA replication, Fanconi
anemia pathway, homologous recombination, mismatch repair, and protein export, nevertheless, those in
the low expression group were signicantly enriched in asthma, malaria, pertussis, systemic lupus
erythematosus, and taurine and hypotaurine metabolism (Fig. 9D).
Correlation between four diagnostic genes and EMS
immune permeation level
In this study, the proportion of 21 immune cells in 18 EMS samples and 23 Normal samples was
estimated with the CIBERSORT algorithm, which can be seen in the histograms. The abundance of B cells
memory, Plasma cells, T cells CD4 memory resting, Macrophages M2, Mast cells activated, and
Eosinophils were higher in EMS samples than that in normal samples, while the inltration abundance of
Page 9/28
B cells naïve, NK cells activated, and Mast cells resting in EMS samples were lower (Fig. 10A and 10B).
Four biomarkers (NLRP3, IL-1 β , LY96 and PDIA3) were associated with 6 signicantly different immune
cells in 21 immune cell types, including Dendritic cells activated, Dendritic cells resting, Eosinophils,
Macrophages M1, Monocytes, and Plasma cells (Fig. 10C and 10D). As shown in Figure 9D, NLRP3 was
negatively correlated with B cells naïve only. IL-1 β was positively correlated with Plasma cells but
negatively correlated with Mast cells resting and T cells CD4 naïve. LY96 was related to Plasma cells,
Macrophages M2, B cells mEMSory, T cells gamma delta, and Eosinophils in a positive way, while
negatively related with B cells naïve, NK cells activated, Dendritic cells activated, Mast cells resting, T cells
CD8, and T cells follicular helper. PDIA3 has positively correlated with T cells CD4 memory activated
when negatively correlated with B cells memory and Macrophages M2.
Construction and assessment of the diagnostic nomogram
model
A nomogram was created to validate the diagnostic power of the four EMS biomarkers (Fig. 11A). The
calibration curve showed a small error between the actual EMS risk and the predicted risk, indicating that
the nomogram model had a high accuracy in predicting EMS (Fig. 11B). DCA showed that the nomogram
curve was higher than the gray critical line and the four biomarker curves, indicating that the nomogram
model had the best clinical benet (Fig. 11C).
Identication of candidate pharmacologic agents
Four well-validated diagnostic genes (NLRP3, IL-1 β , LY96, and PDIA3) were used as genetically linked
targets to facilitate further studies in therapeutic development by a drug target enrichment analysis was
performed (Table 1). We have listed the top 10 candidate enrichment agents targeting gene links, which
may be targeted for the development of co-onset therapeutic strategies. Studies have shown that
niclosamide can target and inhibit macrophage-induced endometriosis inammatory microenvironment,
thereby improving reproductive function[20–22]. In addition, our analysis of the results based on Table 1
indicates that niclosamide is considered a candidate molecule for the treatment of EMS because of its
high odds ratio and high composite score.
Discussion
Endometriosis, a gynecological disease characterized by the implantation of endometrial tissue outside
the uterus, has attracted increasing attention worldwide recently. However, despite its prevalence,
diagnosis is often delayed by years, misdiagnosis is common, clinical presentations are varied, and the
presence of pelvic lesions is heterogeneous[1]. Currently, EMS is usually treated with a combination of
hormone therapy and surgery to remove the lesion[23]. Despite recent advances in the medical treatment
of endometriosis-related pelvic pain, there are still no drugs available to treat the condition[24]. Therefore,
Page 10/28
the identication of novel molecular factors and the unraveling of the underlying mechanisms of EMS
will facilitate early diagnosis and the development of effective therapies.
In the present study, we identied DEGs across the GSE7307 dataset including 18 EMS samples and 23
normal samples. These DEGs included 11 upregulated and 3 downregulated genes, respectively. GO and
KEGG analyses revealed that these DEGs were involved in regulating immune-related signaling pathways,
such as the production and regulation of the cytokine IL-1 β , as well as the NOD-like receptor signaling
pathway. We applied the machine learning methods of SVM-RFE and RF to integrate the feature genes
and obtained four feature genes: NLRP3, IL-1 β , LY96 and PDIA3. Then, we further conrmed their
diagnostic using GSE7305 and GSE23339 datasets, and further demonstrated NLRP3, IL-1 β , LY96 and
PDIA3 as critical biomarkers for EMS based on the results of ROC assays.
The inammasome is newly discovered and plays an important role in innate immunity. The most typical
is the NLRP3 inammasome, the so-called NLRP3 protein because the NLRP3 protein in this complex
belongs to the family of nucleotide-binding and oligomerization domain-like receptors (NLRs), also
known as "pyrin domain-containing protein 3"[25]. The expression of NLRP3 in EMS tissues and serum of
EMS patients was higher than that of normal controls, which contributed to the invasion of EMS[26].
Since EMS is a gynecologic disorder associated with estrogen, higher estrogen receptor (ER)- β levels and
enhanced ER- β activity are detected in endometriotic tissues. ER- β is known to interact with NALP3, and
the activation of NALP3 increases interleukin IL-1 β and IL-18, thereby enhancing cell adhesion and
proliferation[27]. The NLRP3 inhibitor (MCC950) has recently been reported to signicantly reduce the co-
localization of NLRP3 and IL-1 β in cyst-derived stromal cells (CSCs) and IL-1 β concentration in the
supernatant of CSCs. Thereby contributing to Ovarian endometriosis (OE) suppression and improving
ovarian function in endometriosis[12]. However, the specic mechanism of NLRP3 in EMS is still unclear.
IL-1 β plays a considerable role in inammatory diseases and the main sources of its secretion are
macrophages and monocytes, dendritic cells (DC), B lymphocytes, neutrophils and natural killer (NK) cells
and non-immune cells such as keratinocytes[28]. The maturation of IL-1 β and IL-18 induces pyroptosis, a
form of cell death[29]. The inammatory factor IL-1 β has been reported to be signicantly increased in
ectopic endometrial tissue of patients with EMS[30], which is consistent with our results. Chronic pelvic
pain is one of the main clinical symptoms of female endometriosis, but its specic mechanism is unclear.
IL-1 β is closely related to inammatory pain by sending inammatory pain signals to the hypothalamus
and participating in afferent pain response[31]. A retrospective study of peritoneal uid samples obtained
by laparoscopy found an increase in IL-1 β concentration compared with normal controls that were
statistically signicant only in women with endometriosis who reported pelvic pain[32]. In addition, some
studies have revealed the mechanism of deep dyspareunia in endometriosis, that is, IL-1 β stimulates the
expression of nerve growth factor (NGF), promotes local neurogenesis around endometriosis, and then
leads to tenderness in the pelvic anatomy, leading to deep dyspareunia[33]. Because IL-1 β can induce
NGF expression and increase NGF bioactivity, dienogest, a selective P-receptor (PR) agonist, Inhibition of
NGF expression by PR-A and PR-B in human endometrial epithelial cells (hEECs) may contribute to the
relief of endometriosis pain[34]. Sulforaphane alleviates pain caused by sciatic nerve endometriosis,
Page 11/28
which is mediated by the inhibition of inammatory cytokines IL6, IL-1 β , and TNF- α [35]. Overall,
pharmacological targeting of IL-1 β /NGF may also have the potential for the treatment of endometriosis-
related pain.
LY96 plays a crucial role in inammation-related and immune-related diseases, such as Crohn's disease,
rheumatoid arthritis, and inammatory diabetic cardiomyopathy[36]. Key protein disulde isomerase 3
(PDIA3) is a kind of chaperone, that can adjust the folding of the new synthesis glycoprotein, the
heterogeneous and REDOX enzyme activity, and immune activation marks, cancer immune cells and
immune modulators signicantly related, is a powerful cancer prognosis biomarkers, which can
effectively predict the immune response to treatment[37, 38]. Similarly, the expression and function of
LY96 and PDIA3 in EMS remain are largely unknown.
Furthermore, we also evaluated the degree of inltration of 22 immune cells in EMS and normal samples.
Compared with normal samples, EMS samples had distinctly higher levels of B cells memory, Plasma
cells, T cells CD4 memory resting, Macrophages M2, Mast cells activated, and Eosinophils. It is well
known that endometriosis lesions are characterized by the presence of numerous plasma cells and
activated macrophages[39]. One study found that the relative abundance of resting CD4 + memory T cells
and even less abundant memory B cells were signicantly higher in patients with stage III/IV
endometriosis, which was strongly associated with endometrial receptivity[40]. Activated macrophages
are generally divided into two groups, M1-like macrophages and M2-like macrophages, both of which are
closely related to the inammatory response[41]. Another study found a signicant increase in total
macrophages in ectopic endometrial tissue (EC) and a syngeneic mouse model of endometriosis, and M2
macrophages were the predominant macrophages in EC[42, 43]. Of note, M2 macrophages have been
found to increase gradually from stage I to stage IV in ectopic endometrial tissues of EMS patients,
whereas M1 macrophages have been found to increase in contrast, which may contribute to the
proinammatory microenvironment in the early stage of the disease, as well as to the probrotic activity
in the late stage[44]. Diffuse inltration of large numbers of mast cells (MCs) is observed in the
interstitium of ectopic endometrial tissues of EMS. MCs promote the development of endometriosis
through NLRP3 inammasome activation mediated by nuclear-initiated estrogen signaling pathway and
the production of mature IL-1 β [45, 46]. In addition, the preoperative pain score of EMS is closely related to
the density and number of mast cells, and the use of MCs stabilizers and inhibitors can relieve
endometriosis and its associated pain[47, 48]. Eosinophils can regulate broblasts and stimulate
collagen synthesis. Eosinophils are abundant in brotic areas and peritoneal uid of endometriosis,
which is associated with brosis of endometriosis lesions[49, 50]. Moreover, we found that the
expressions of NLRP3, IL-1 β , LY96, and PDIA3 were related to the levels of many immune cells,
highlighting their potential used as therapeutic targets for EMS. Hence, The differential expression of
NLRP3, IL-1 β , LY96, and PDIA3 between EMS and normal groups pointed out they might be crucial
immune-related biomarkers to diagnose EMS. A diagnostic nomogram model based on four core genes
with excellent predictive power was established, and patients would benet from this model.
Page 12/28
Finally, based on four diagnostic genes as gene junction points, we identied niclosamide as a candidate
molecule for EMS treatment due to its high odds ratio and high comprehensive score among the top 10
candidate enrichment agents by drug target enrichment analysis. Niclosamide was discovered in 1953 at
the Bayer Chemotherapy Research Laboratory to treat tapeworm infections in humans and is listed on the
World Health Organization's Essential Medicines List[51]. However, accumulating evidence suggests that
niclosamide is a multifunctional drug capable of inhibiting or modulating multiple signaling pathways
and biological processes, which has been demonstrated to treat endometriosis in vitro and in vivo
experiments. Recent niclosamide was reported can effectively reduce the endometriosis of endometriosis
lesions in mice (ELL) induced by peritoneal macrophages (LPM) and LPM the expression of related
genes, and by reducing ELL and pelvic organs may stimulate the peripheral nerve as a medium of
inammatory factor to reduce the abnormal inammation, may eventually reduce the pain
associated[52]. In the endometriosis model in mice, chlorine nitramine can not only through targeted
STAT3 and NFkB signaling pathways, reduce the growth and progress of the sample endometriosis
lesions, but also can inhibit the inammatory mechanisms, thereby reducing the macrophages induced
cell activity and the secretion of cytokine/chemokine based on not affect reproductive function[20–22].
At present, although niclosamide has not been used in the clinical treatment of endometriosis, the results
of previous studies and our results suggest that niclosamide may be a novel strategy for the targeted
treatment of inammatory dysfunction in endometriosis.
Conclusion
Four diagnostic genes related to NLRP3 activation and EMS were identied by bioinformatics. Although
we have explored the biological processes and pathways involved in them, further experimental
verication is needed to verify these functions. The nomogram based on these four genes has a good
diagnostic effect on EMS, which provides a theoretical basis for clinical diagnosis. In addition, these
diagnostic genes were found to be associated with different immune factors, suggesting that they may
also have important roles in the immune microenvironment. Currently, drugs targeting these pyroptosis
genes are predicted to alleviate EMS, but only niclosamide has the results of basic studies to support its
use in the treatment of EMS, and further studies are needed to explore its specic mechanism of action.
Firstly, the GSE7307 dataset was downloaded from the GEO database, including 18 EMS samples and 23
control samples. The nal 14 DEGs related to pyroptosis and EMS were obtained from the endometrial
samples of GSE7307 by differential analysis. GO and KEGG analysis showed that these genes were
mainly involved in two major biological processes. Random Forest (RF) and support vector machine
recursive feature elimination (SVM-RFE) algorithms were used to select four genes to construct the EMS
diagnostic model. The model was validated in endometrial samples of GSE7305 and GSE23339, and the
AUC values showed that the model had a good diagnostic performance. In addition, the inltration of
immune cells in the samples and the correlation between different immune factors and pyroptosis genes
were further discussed. These results suggest that these diagnostic genes may also play an important
role in the immunity of EMS. Finally, 10 drugs targeting these four diagnostic genes were retrieved from
the DrugBank database, of which niclosamide proved useful for treating EMS. Overall, we identied four
Page 13/28
key diagnostic genes for EMS. In addition, large-scale and multicenter prospective cohort studies are
necessary to conrm whether these four genes also have valid diagnostic value in blood samples from
EMS patients.
Declarations
Acknowledgements
None.
Authors’ contributions
MD and PH directed the study; WN, HW and SD performed the experiments; LH, TL, LL and BW performed
the data collection and statistical analysis; SZ edited the language; all the authors approved the
manuscript for submission.
Funding
This research was supported by the National Natural Science Foundation of China (No. 81960274), the
2020 university-level Research Project of Youjiang Medical University for Nationalities (No.
yy2020gcky039), and the 2021 Scientic Research and Technology Development Program of Baise City
(No. 20212351).
Availability of data and materials
Publicly available datasets were analyzed in this study. This data can be found here: All the raw data
used in this study are derived from the public GEO data portal (https://www.ncbi.nlm.nih.gov/geo/;
Accession numbers: GSE7307, GSE7305 and GSE23339).
Ethics approval and consent to participate
All procedures were approved by the ethics review committee of Aliated Hospital of Youjiang Medical
College for Nationalities. All methods were carried out in accordance with relevant guidelines and
regulations of the ethics review committee of Aliated Hospital of Youjiang Medical College for
Nationalities, and informed consent was obtained from all subjects and/or their legal guardian(s).
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Page 14/28
References
1. Taylor HS, Kotlyar AM, Flores VA. Endometriosis is a chronic systemic disease: clinical challenges
and novel innovations. Lancet. 2021;397:839–52.
2. Chapron C, Marcellin L, Borghese B, Santulli P. Rethinking mechanisms, diagnosis and management
of endometriosis. Nat Rev Endocrinol. 2019;15:666–82.
3. Vallvé-Juanico J, Houshdaran S, Giudice LC. The endometrial immune environment of women with
endometriosis. Hum Reprod Update. 2019;25:564–91.
4. Koninckx PR, Fernandes R, Ussia A, Schindler L, Wattiez A, Al-Suwaidi S, et al. Pathogenesis Based
Diagnosis and Treatment of Endometriosis. Front Endocrinol (Lausanne). 2021;12:745548.
5. Kiesel L, Sourouni M. Diagnosis of endometriosis in the 21st century. Climacteric. 2019;22:296–302.
. Humphries F, Bergin R, Jackson R, Delagic N, Wang B, Yang S, et al. The E3 ubiquitin ligase Pellino2
mediates priming of the NLRP3 inammasome. Nat Commun. 2018;9:1560.
7. Beckerman P, Bi-Karchin J, Park ASD, Qiu C, Dummer PD, Soomro I, et al. Transgenic expression of
human APOL1 risk variants in podocytes induces kidney disease in mice. Nat Med. 2017;23:429–38.
. Wang L, Hauenstein AV. The NLRP3 inammasome: Mechanism of action, role in disease and
therapies. Mol Aspects Med. 2020;76:100889.
9. Zhao K, An R, Xiang Q, Li G, Wang K, Song Y, et al. Acid-sensing ion channels regulate nucleus
pulposus cell inammation and pyroptosis via the NLRP3 inammasome in intervertebral disc
degeneration. Cell Prolif. 2021;54:e12941.
10. Hang Y, Tan L, Chen Q, Liu Q, Jin Y. E3 ubiquitin ligase TRIM24 deciency promotes NLRP3/caspase ‐
1/IL ‐ 1 β ‐ mediated pyroptosis in endometriosis. Cell Biology International. 2021;45:1561–70.
11. Zhou F, Zhao F, Huang Q, Lin X, Zhang S, Dai Y. NLRP3 activated macrophages promote endometrial
stromal cells migration in endometriosis. Journal of Reproductive Immunology. 2022;152:103649.
12. Murakami M, Osuka S, Muraoka A, Hayashi S, Bayasula null, Kasahara Y, et al. Effectiveness of
NLRP3 Inhibitor as a Non-Hormonal Treatment for ovarian endometriosis. Reprod Biol Endocrinol.
2022;20:58.
13. Han SJ, Jung SY, Wu S-P, Hawkins SM, Park MJ, Kyo S, et al. Estrogen Receptor β Modulates
Apoptosis Complexes and the Inammasome to Drive the Pathogenesis of Endometriosis. Cell.
2015;163:960–74.
14. Yu G, Wang L-G, Han Y, He Q-Y. clusterProler: an R package for comparing biological themes among
gene clusters. OMICS. 2012;16:284–7.
15. Gers FA, Schmidhuber E. LSTM recurrent networks learn simple context-free and context-sensitive
languages. IEEE Trans Neural Netw. 2001;12:1333–40.
1. Huang S, Cai N, Pacheco PP, Narrandes S, Wang Y, Xu W. Applications of Support Vector Machine
(SVM) Learning in Cancer Genomics. Cancer Genomics Proteomics. 2018;15:41–51.
Page 15/28
17. Liang A, Huang L, Liu H, He W, Lei X, Li M, et al. Resveratrol Improves Follicular Development of
PCOS Rats by Regulating the Glycolytic Pathway. Mol Nutr Food Res. 2021;e2100457.
1. Tibshirani R. The lasso method for variable selection in the Cox model. Stat Med. 1997;16:385–95.
19. Yeoh YK, Zuo T, Lui GC-Y, Zhang F, Liu Q, Li AY, et al. Gut microbiota composition reects disease
severity and dysfunctional immune responses in patients with COVID-19. Gut. 2021;70:698–706.
20. Sekulovski N, Whorton AE, Shi M, MacLean JA, Hayashi K. Endometriotic inammatory
microenvironment induced by macrophages can be targeted by niclosamide † . Biol Reprod.
2019;100:398–408.
21. Prather GR, MacLean JA, Shi M, Boadu DK, Paquet M, Hayashi K. Niclosamide As a Potential
Nonsteroidal Therapy for Endometriosis That Preserves Reproductive Function in an Experimental
Mouse Model. Biol Reprod. 2016;95:76.
22. Sekulovski N, Whorton AE, Tanaka T, Hirota Y, Shi M, MacLean JA, et al. Niclosamide suppresses
macrophage-induced inammation in endometriosis † . Biol Reprod. 2020;102:1011–9.
23. Suryawanshi S, Huang X, Elishaev E, Budiu RA, Zhang L, Kim S, et al. Complement pathway is
frequently altered in endometriosis and endometriosis-associated ovarian cancer. Clin Cancer Res.
2014;20:6163–74.
24. Moses AS, Taratula OR, Lee H, Luo F, Grenz T, Korzun T, et al. Nanoparticle-Based Platform for
Activatable Fluorescence Imaging and Photothermal Ablation of Endometriosis. Small.
2020;16:e1906936.
25. Shao B-Z, Xu Z-Q, Han B-Z, Su D-F, Liu C. NLRP3 inammasome and its inhibitors: a review. Front
Pharmacol. 2015;6:262.
2. Huang Y, Li R, Hu R, Yao J, Yang Y. PEG2-Induced Pyroptosis Regulates the Expression of HMGB1
and Promotes hEM15A Migration in Endometriosis. Int J Mol Sci. 2022;23:11707.
27. Di Nicuolo F, Castellani R, De Cicco Nardone A, Barbaro G, Paciullo C, Pontecorvi A, et al. Alpha-Lipoic
Acid Plays a Role in Endometriosis: New Evidence on Inammasome-Mediated Interleukin
Production, Cellular Adhesion and Invasion. Molecules. 2021;26:E288.
2. Wenjing F, Tingting T, Qian Z, Hengquan W, Simin Z, Agyare OK, et al. The role of IL-1 β in aortic
aneurysm. Clin Chim Acta. 2020;504:7–14.
29. He Y, Hara H, Núñez G. Mechanism and Regulation of NLRP3 Inammasome Activation. Trends
Biochem Sci. 2016;41:1012–21.
30. J H, X C, Y L. HMGB1 Mediated Inammation and Autophagy Contribute to Endometriosis. Frontiers
in endocrinology [Internet]. Front Endocrinol (Lausanne); 2021 [cited 2022 Oct 16];12.
31. Wang W, Li G, De Wu null, Luo Z, Pan P, Tian M, et al. Zika virus infection induces host inammatory
responses by facilitating NLRP3 inammasome assembly and interleukin-1 β secretion. Nat
Commun. 2018;9:106.
32. Akoum A, Al-Akoum M, Lemay A, Maheux R, Leboeuf M. Imbalance in the peritoneal levels of
interleukin 1 and its decoy inhibitory receptor type II in endometriosis women with infertility and
Page 16/28
pelvic pain. Fertil Steril. 2008;89:1618–24.
33. Peng B, Alotaibi FT, Sediqi S, Bedaiwy MA, Yong PJ. Role of interleukin-1 β in nerve growth factor
expression, neurogenesis and deep dyspareunia in endometriosis. Hum Reprod. 2020;35:901–12.
34. Mita S, Shimizu Y, Sato A, Notsu T, Imada K, Kyo S. Dienogest inhibits nerve growth factor expression
induced by tumor necrosis factor- α or interleukin-1 β . Fertil Steril. 2014;101:595–601.
35. Liu Y, Zhang Z, Lu X, Meng J, Qin X, Jiang J. Anti-nociceptive and anti-inammatory effects of
sulforaphane on sciatic endometriosis in a rat model. Neurosci Lett. 2020;723:134858.
3. Nie K, Li J, Peng L, Zhang M, Huang W. Pan-Cancer Analysis of the Characteristics of LY96 in
Prognosis and Immunotherapy Across Human Cancer. Front Mol Biosci. 2022;9:837393.
37. Mo H-Q, Tian F-J, Ma X-L, Zhang Y-C, Zhang C-X, Zeng W-H, et al. PDIA3 regulates trophoblast
apoptosis and proliferation in preeclampsia via the MDM2/p53 pathway. Reproduction.
2020;160:293–305.
3. Tu Z, Ouyang Q, Long X, Wu L, Li J, Zhu X, et al. Protein Disulde-Isomerase A3 Is a Robust
Prognostic Biomarker for Cancers and Predicts the Immunotherapy Response Effectively. Front
Immunol. 2022;13:837512.
39. Hever A, Roth RB, Hevezi P, Marin ME, Acosta JA, Acosta H, et al. Human endometriosis is associated
with plasma cells and overexpression of B lymphocyte stimulator. Proc Natl Acad Sci U S A.
2007;104:12451–6.
40. Xiang R, Chen P, Zeng Z, Liu H, Zhou J, Zhou C, et al. Transcriptomic analysis shows that surgical
treatment is likely to inuence the endometrial receptivity of patients with stage III/IV endometriosis.
Front Endocrinol (Lausanne). 2022;13:932339.
41. Yunna C, Mengru H, Lei W, Weidong C. Macrophage M1/M2 polarization. Eur J Pharmacol.
2020;877:173090.
42. Zhong Q, Yang F, Chen X, Li J, Zhong C, Chen S. Patterns of Immune Inltration in Endometriosis and
Their Relationship to r-AFS Stages. Front Genet. 2021;12:631715.
43. Je M, Sh A, Rm M, Sp M, At F, M K, et al. IL-17A Modulates Peritoneal Macrophage Recruitment and
M2 Polarization in Endometriosis. Frontiers in immunology [Internet]. Front Immunol; 2020 [cited
2022 Oct 19];11.
44. Laganà AS, Salmeri FM, Ban Frange ž H, Ghezzi F, Vrta č nik-Bokal E, Granese R. Evaluation of M1 and
M2 macrophages in ovarian endometriomas from women affected by endometriosis at different
stages of the disease. Gynecol Endocrinol. 2020;36:441–4.
45. Sugamata M, Ihara T, Uchiide I. Increase of activated mast cells in human endometriosis. Am J
Reprod Immunol. 2005;53:120–5.
4. Guo X, Xu X, Li T, Yu Q, Wang J, Chen Y, et al. NLRP3 Inammasome Activation of Mast Cells by
Estrogen via the Nuclear-Initiated Signaling Pathway Contributes to the Development of
Endometriosis. Front Immunol. 2021;12:749979.
Page 17/28
47. Anaf V, Chapron C, El Nakadi I, De Moor V, Simonart T, Noël J-C. Pain, mast cells, and nerves in
peritoneal, ovarian, and deep inltrating endometriosis. Fertil Steril. 2006;86:1336–43.
4. Binda MM, Donnez J, Dolmans M-M. Targeting mast cells: a new way to treat endometriosis. Expert
Opin Ther Targets. 2017;21:67–75.
49. Blumenthal RD, Samoszuk M, Taylor AP, Brown G, Alisauskas R, Goldenberg DM. Degranulating
eosinophils in human endometriosis. Am J Pathol. 2000;156:1581–8.
50. Eidukaite A, Tamosiunas V. Activity of eosinophils and immunoglobulin E concentration in the
peritoneal uid of women with endometriosis. Clin Chem Lab Med. 2004;42:590–4.
51. Chen W, Mook RA, Premont RT, Wang J. Niclosamide: Beyond an antihelminthic drug. Cell Signal.
2018;41:89–96.
52. Shi M, Sekulovski N, Whorton AE, MacLean JA, Greaves E, Hayashi K. Ecacy of niclosamide on the
intra ‐ abdominal inammatory environment in endometriosis. FASEB j [Internet]. 2021 [cited 2022 Oct
20];35.
Table 1
Table 1. Predictive top 10 pharmacologic agent candidates for EMS with NLRP3 activation.
Page 18/28
Drug name Padj OddsRatio CombinedScore Genes
Triclocarban CTD 00000497 0.00 713.14 8021.85910 IL1B;NLRP3
chlorhexidine CTD 00005633 0.00 498.90 5271.69962 IL1B;NLRP3
uric acid CTD 00006967 0.00 498.90 5271.69962 IL1B;NLRP3
GNF-Pf-4325 BOSS 0.00 486.71 5119.715195 IL1B;NLRP3
niclosamide BOSS 0.00 464.02 4838.479209 IL1B;NLRP3
9-Methoxyellipticine TTD 00001373 0.00 464.02 4838.479209 IL1B;NLRP3
niclosamide 0.00 415.58 4245.007066 IL1B;NLRP3
chlorhexidine 0.00 415.58 4245.007066 IL1B;NLRP3
5H-quinolino[8,7-c][1,2]benzothiazine 6,6-dioxideTTD 00001179 0.00 398.92 4043.246313 IL1B;NLRP3
thimerosal BOSS 0.00 332.27 3249.969098 IL1B;NLRP3
Figures
Page 19/28
Figure 1
The workow of the analysis
Figure 2
Page 20/28
The dysregulated genes were shown in the (A) boxplot and (B) heat map via analyzing GSE7307
datasets. 14 DEGs were identied between EMS samples and normal samples.
Figure 3
(A) GO and (B) KEGG pathway analyses of 14 DEGs in EMS.
Page 21/28
Figure 4
Selection procedure of diagnostic markers for EMS diagnoses. (A) An illustration of biological marker
screening through the SVM-RFE arithmetic. (B)Tuning feature selection in the random forest model. (C)
Analyze the meaning of NLRP3 activation-related genes using the scores returned by the random forest
model. (D) Venn graph presenting 4 diagnostic biomarkers shared by the random forest and SVM-RFE
arithmetic methods.
Page 22/28
Figure 5
The expression of four biomarkers in EMS and the diagnostic value of four biomarkers in GSE7307 were
studied by ROC analysis. (A) NLRP3, IL-1 Β , and LY96 were highly expressed in EMS. PDIA3 was lowly
expressed in EMS. (B)The diagnostic value of the 4 biomarkers was studied using ROC assays in
GSE7307.
Page 23/28
Figure 6
Results
of differential gene expression of NLRP3, LY96, IL-1 β and PDIA3 at the protein level between EMS
and control samples. (A) Western blotting results. (B) Densitometry of the western blot. N=6 each group.
*P<0.05 and **P<0.01.
Page 24/28
Figure 7
ROC curve analysis of the model EMS in the validation datasets (GSE7305 and GSE23339).
Page 25/28
Figure 8
Identication of NLRP3 activation-related genes. (A) Boxplots of the residuals of the sample. Red dot
stands for root mean square of residuals. (B) Cumulative residual distribution map of the sample. (C) The
AUC value of the ROC curve indicated that the accuracy of GLM model (0.933) was the highest among
the four models. (D) The ROC curve assesses the accuracy of the four models in the validation datasets
(GSE7305 and GSE23339).
Page 26/28
Figure 9
GSEA revealed the enriched pathways of four biomarkers. (A)NLRP3, (B) IL-1 Β , (C) LY96, and (D)PDIA3.
Page 27/28
Figure 10
Immune inltration and immune-related factors. (A)The number of immune cells in each EMS sample is
represented by a different color. (B) Differences of the inltrate immune cells between the normal (blue)
and EMS (red) groups. (C) Heat map of correlation of four biomarkers (NLRP3, IL-1B, LY96, and PDIA3)
with 22 immune cell subsets. (D) Correlation between NLRP3, IL-1B, LY96, and PDIA3 and inltrating
immune cells in EMS.
Page 28/28
Figure 11
Construction and assessment of a nomogram model for EMS diagnosis based on the training set. (A)
Nomogram to predict the occurrence of EMS. (B)Calibration curve to assess the predictive power of the
nomogram model. (C) Decision curve analysis (DCA) to evaluate the clinical value of the nomogram
model.
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.