The Cancer Genome Atlas dataset-based analysis of aberrantly expressed genes by GeneAnalytics in thymoma associated myasthenia gravis: focusing on T cells

Abstract

Background

Myasthenia gravis (MG) is a group of autoimmune disease which could be accompanied by thymoma. Many differences have been observed between thymoma-associated MG (TAMG) and non-MG thymoma (NMG). However, the molecular difference between them remained unknown. This study aimed to explore the differentially expressed genes (DEGs) between the two categories and to elucidate the possible pathogenesis of TAMG further.

Methods

DEGs were calculated using the RNA-Sequencing data from 11 TAMG and 10 NMG in The Cancer Genome Atlas (TCGA) database. GeneAnalytics was performed to characterize the associations between DEGs and tissues and cells, diseases, gene ontology (GO) terms, pathways, phenotypes, and drug/compounds, respectively. Genes related to T cells were sorted out using LifeMapDiscovery Cells and Tissues Database. Genes directly related to the phenotype of autoimmune diseases that were identified by VarElect were validated by reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

Results

The expression level of 169 genes showed a significant difference between the two groups, with 94 up-regulated and 75 down-regulated. Overexpression of six genes (ATM, SFTPB, ANKRD55, BTLA, CCR7, TNFRSF25), which are expressed in T cells and directly related to autoimmune disease through VarElect, was identified. The overexpression of soluble BTLA (sBTLA) (P=0.027), CCR7 (P=0.0018), TNFRSF25 (P=0.0013) and ANKRD55 (P=0.0026) was validated by RT-qPCR in thymoma tissues from our center.

Conclusions

Overexpression of sBTLA, CCR7, TNFRSF25 and ANKRD55 was identified and validated by RT-qPCR, which could partly explain the underlying pathogenesis in TAMG.

Introduction

Myasthenia gravis (MG), characterized by T cell involvement and complement dependence, is a group of autoimmune disease with antibodies targeting neuromuscular junctions (1). It presents as fluctuating muscular weakness, mainly affecting extraocular muscles, bulbar muscles and proximal limb muscles. Based on the clinical features and serum antibodies, MG can be classified as early/late-onset, thymoma/nonthymoma, acetylcholine receptor (AChR) antibody/muscle-specific tyrosine kinase (MuSK)/low density lipoprotein receptor related protein 4 (LRP4) antibody positive/antibody-negative and generalized/ocular subgroups (1). In 75–90% cases, MG is combined with abnormalities of thymus, including thymus hyperplasia with germinal centers (85%) and thymoma (15%) (2). Interestingly, thymoma is not the exclusive risk factor of MG, only 15–33% of thymoma patients may develop MG (3,4). With reference to WHO histological classification, type B thymoma is shown to be more frequently associated with MG than type A and AB. Among type B thymoma, type B2 is the most frequently associated with the disease (5).

Compared with early-onset and late-onset MG without thymoma, TAMG usually occurs after 50 years old and a small proportion are diagnosed in their adolescence. The symptoms are usually more severe and with bulbar muscles involved. The gender bias was not significant and its correlation with human leukocyte antigen (HLA) was not strong. Besides anti-AChR antibody, TAMG can also be accompanied by anti-Titin, anti-ryanodine antibodies and other autoimmune diseases (6). Some recent studies indicated that T cells may play a potentially key role in TAMG pathogenesis. Compared with non-MG thymoma (NMG), thymoma in TAMG can generate more mature CD4⁺CD45RA⁺T cells exporting to the peripheral while the peripheral AChR-reactive T lymphocytes can be recirculated to the thymus to be activated (2). A higher frequency of circulating IL-17 producing CD4⁺ T-cell subsets (Th17) and a lower proportion of CD4⁺CD25^highFoxp3⁺ regulatory T (Treg) cells were also observed (7,8). Besides, our study group verified a higher percentage of T follicular helper (Tfh) cells in MG patients with thymoma (9), which may also demonstrate a difference in pathogenesis between TAMG and NMG. However, the molecular mechanisms behind TAMG still need further investigations.

On a genome-wide scale, gene expression profiling is increasingly used to explore pathogenesis and to find out possible biomarkers in various human disorders. Technological advances in DNA microarrays and subsequent RNA-sequencing (RNA-Seq) are employed to analyze gene expression in a comprehensive and unbiased method (10). Previously, the overexpression of Interferon type I (IFN-I) with abnormal regulation of double-stranded RNA (dsRNA) signaling molecules was demonstrated in TAMG but not in NMG using microarrays (11).

To further explore the pathogenesis of TAMG, we calculated the differentially expressed genes (DEGs) using RNA-seq data from 11 TAMG and 10 NMG patients in The Cancer Genome Atlas (TCGA) database. Based on these genes, we performed gene ontology (GO) classification and functional enrichment analysis using GeneAnalytics (12). We further screened out genes related to T cells and verified these DEGs by reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

Methods

TCGA dataset of thymoma

High throughput RNA-Seq data of the thymoma tissues, including B2 thymoma tissues from 11 TAMG and 10 NMG were downloaded from TCGA database on December 31^st, 2017. The summary of the clinical information of these patients from TCGA was shown in Table S1. Considering TCGA data a community resource project, additional approval from the Medical Ethics Committee in our hospital was not necessary. TCGA data access policies and publication guidelines were also obeyed in the present study.

Table S1. Summary of the clinical information of patients from TCGA.

No.	Case ID	Gender	Age at diagnosis (year)	Pathology	Myasthenia gravis	Race	Case UUID
1	TCGA-X7-A8DB	Female	71	Thymoma; type B2	No	White	4165391B-6122-4E87-BA91-CE935E0B3CF4
2	TCGA-XU-A92U	Male	67	Thymoma; type B2	No	White	FAC44B50-AA8C-498E-88AF-4EB5C2F5F67A
3	TCGA-ZC-AAAH	Male	52	Thymoma; type B2	No	White	3CC2D49B-A3A6-4F55-A355-68CBE84F4056
4	TCGA-YT-A95E	Male	52	Thymoma; type B2	No	White	D3189C8E-FFE0-41FF-9C7B-64DA4C83B991
5	TCGA-X7-A8M4	Male	41	Thymoma; type B2	No	White	23CC4CD3-87DC-4342-9072-BE3695E393FA
6	TCGA-X7-A8DE	Female	35	Thymoma; type B2	No	White	15A2BAAE-D15E-4349-9358-74356A2D2AD0
7	TCGA-X7-A8M6	Male	43	Thymoma; type B2	No	White	A6885883-F1A2-48DD-8D72-0038319069A1
8	TCGA-XM-A8RC	Male	67	Thymoma; type B2	No	White	B8348ADB-17DD-4849-BD4F-C2E7E15DB1F4
9	TCGA-4X-A9FB	Male	44	Thymoma; type B2	No	White	2C2E169E-9F93-41A7-BFC4-ADE18BB5D7AE
10	TCGA-XM-A8RD	Female	69	Thymoma; type B2	No	White	68D2AB61-50FC-41E8-B3EE-75C70F0F4369
11	TCGA-4X-A9FD	Female	43	Thymoma; type B2	Yes	White	0A312064-878F-4B69-8BB7-E7F14E6CD7FE
12	TCGA-XU-AAXX	Female	44	Thymoma; type B2	Yes	White	5B7B3E7C-3230-4675-8FDF-3E5119B02CE1
13	TCGA-XU-AAXV	Female	45	Thymoma; type B2	Yes	White	80606B1C-E8C8-4C3D-8475-3D846A84CA98
14	TCGA-X7-A8M7	Male	26	Thymoma; type B2	Yes	White	56C4F5E8-4E89-4BA0-984D-E17F11A401F5
15	TCGA-X7-A8DG	Male	47	Thymoma; type B2	Yes	White	A653B0BB-AACF-4B11-BB11-42E043E49BF6
16	TCGA-XU-AAXY	Male	43	Thymoma; type B2	Yes	White	7933B320-AE8C-4D3D-9468-8ABB135E93D5
17	TCGA-4V-A9QS	Female	53	Thymoma; type B2	Yes	White	88C0AF76-B1BA-478F-A42B-8A50C0568D5E
18	TCGA-4V-A9QI	Male	68	Thymoma; type B2	Yes	White	EEAA0FA3-408C-4E0F-A18A-095514173FF1
19	TCGA-ZC-AAAA	Male	39	Thymoma; type B2	Yes	White	67FF20BA-1B56-45AE-9310-19DD74401E67
20	TCGA-YT-A95G	Female	61	Thymoma; type B2	Yes	White	7459BEC0-1B54-4FD7-90DB-8375168E6B1A
21	TCGA-X7-A8D6	Female	48	Thymoma; type B2	Yes	White	66C564BA-B42D-4BE9-9E5F-BF93984569EB

Exploration of DEGs and GeneAnalytics

A simple set of transcript quantification was obtained from the RNA-Seq data of patients with B2 thymoma. RNA-seq relative read counts were measured by counts per million (CPM) to calculate DEGs using edgeR package (13). The value of fold change (FC) was log₂ transformed while false discovery rate (FDR) was −log₁₀ transformed for downstream analysis. We further performed differential analysis (|log₂FC| >1, log₂CPM >1, P value <0.05 and FDR <0.05) with Limma package (14).

Next, we used GeneAnalytics (http://geneanalytics.genecards.org/), a tool empowered by the GeneCards suite, including GeneCards human gene database, MalaCards human disease database, PathCards human biological pathways database and LifeMapDiscovery Cells and Tissues Database to analyze the associations between DEGs and tissues and cells, diseases, GO terms, pathways, phenotypes, and drug/compounds, respectively. Tissues and cells were scored with a matching algorithm that weighs tissue specificity, gene abundance and function. Disease matching scores were originated from the degree of gene-disease relations. Superpathways included the related pathways with high matched genes to reduce redundancy, strengthen inferences and pathway enrichment. Genes related to the T cells were sorted out according to LifeMapDiscovery Cells and Tissues Database (12). Among them, genes directly related to the phenotype of autoimmune diseases were identified by Varlect (15).

Human thymic samples

Eighteen thymoma samples were obtained from patients who underwent thymectomy at the Department of Thoracic Surgery, Huashan and Zhongshan Hospital, Fudan University in 2016–2018. Of them, 9 was with MG and 9 was not. The study was approved by Fudan University Institutional Review Board and written informed consents were granted.

The TAMG group was defined using following criteria: (I) age ≥18 years; (II) definitive diagnosis of generalized MG according to clinical, electrophysiological and antibody status; (III) histopathologic diagnosis of thymoma. The patients were classified by the Myasthenia Gravis Foundation of America (MGFA) Classification (16). Patients who had received corticosteroid or immunosuppressant treatment were excluded.

Validation of RT-qPCR

Total RNA was extracted from the thymoma tissues using Trizol^® (Invitrogen, USA) reagent based on the manufacturer’s instructions. The reverse transcription (RT) reactions were conducted with a cDNA synthesis kit (Takara, China). Quantitative polymerase chain reaction (qPCR) was performed with QuantStudio (Applied Biosystems, USA). The results were quantified with the 2^−ΔΔCT method against GAPDH for normalization. The data represents the mean value of three experiments. Real-time PCR primers were listed in Table S2.

Table S2. Primers used in RT-PCR.

Gene	Primers	Direction
TNFRSF25	CTTGTGTGTCCCCAAGACAC	F
	AGTCTAGGCATGGTTGGCAG	R
CCR7	TCTGCCTGGACTAGAGGGAC	F
	TATCTTCTGGAGCAGGGGCT	R
ANKRD55	GGCCGAATGTGTCCAGTCAC	F
	GTTGAGCACGTTGAACCGTC	R
SFTPB	CTACTTCCAGAACCAGACTGAC	F
	GCTCGGAGAGATCCTGTGTG	R
BTLA	CATCTTAGCAGGAGATCCCTTTG	F
	TGACCCATTGTCATTAGGAAGC	R
sBTLA	TGTGACAGGAAAGCAAAATGAAC	F
	CAGACCCTTCCTGCATCCTG	R
ATM	TGATCTTGTGCCTTGGCTAC	F
	TATGGTGTACGTTCCCCATG	R
GAPDH	GTGAAGGTCGGAGTCAACG	F
	GTTCTCAGCCTTGACGGTG	R

RT-PCR, reverse transcription-polymerase chain reaction.

Statistical analysis

For DEGs analysis, the values of log₂CPM were calculated and normalized by quantile normalization. Then the values of FC were calculated with the mean values of log₂CPM between the two groups. All data analysis of DEGs was conducted and related figures were generated with R 3.5.1 (www.r-project.org). The results of qPCR were analyzed by Student’s t-test or Mann-Whitney U test using GraphPad software. A P value <0.05 was regarded as statistically significant.

Results

GeneAnalytics

DEGs and GeneAnalytics

A total of 24,991 transcripts were included and transcripts with CPM value of 0 were excluded. Finally, we screened 16,477 transcripts for analysis. The expression level of 169 genes showed significant difference between the two groups (|log₂FC| >1, log₂CPM >1, P value <0.05 and FDR <0.05), with 94 up-regulated and 75 down-regulated. The final data was listed as the matrix in Table S3. DEGs calculated by edgeR were shown with volcano plot (Figure S1) and heat map (Figure S2).

Table S3. List of differentially expressed genes derived from edgeR.

Gene	Log2FC	Log2CPM	P value	FDR
SOHLH1	−8.748095249	4.030701899	1.36E−06	0.001177393
NEUROD1	−8.29403346	1.206443756	9.61E−06	0.003837044
SLC30A8	−7.394604588	2.46893017	5.70E−07	0.000758208
KRT14	−5.961451624	8.850437314	1.12E−09	1.11E−05
CYP19A1	−5.531242351	3.688722857	1.51E−05	0.004857241
PENK	−5.522444584	1.896122199	7.44E−05	0.013764626
CACNA1G	−4.800224071	1.502703602	8.66E−08	0.000208153
KRT33B	−4.678978991	3.516999508	1.10E−06	0.001051576
SEZ6L2	−4.394822626	2.815492394	9.38E−08	0.000208153
CHRNG	−4.233245516	1.055180283	5.61E−06	0.002730959
TBX1	−3.997357832	4.27653279	0.000715258	0.04638443
MYH6	−3.803434257	2.338179784	0.000116152	0.016448228
SCUBE1	−3.576521905	2.719921949	0.000125778	0.017320051
TRIM55	−3.458383411	1.634996091	0.000109941	0.016260644
TFDP3	−3.348329361	3.072405038	3.20E−05	0.007924085
CRYAB	−3.316888439	5.004937173	7.62E−06	0.003381488
DPEP1	−3.304235712	2.760709455	7.47E−06	0.003381488
NGEF	−3.259681002	1.090315496	0.000493254	0.038774813
TREM2	−3.206474566	1.610887941	5.00E−07	0.000712547
TWIST1	−3.197009539	2.286019868	0.000357804	0.031999183
KIF1A	−3.138683446	2.52383897	0.000224218	0.025293513
EPB41L4B	−3.114798511	1.811491458	1.51E−05	0.004857241
TNFRSF17	−3.052709294	3.52099794	9.02E−05	0.014635034
ANKFN1	−3.025104688	1.324634834	0.00016265	0.020685527
MYBPC2	−2.921742889	3.150154799	0.000176369	0.021342792
SMIM1	−2.911370466	2.541786224	1.69E−06	0.001347512
KCNQ4	−2.831610853	2.167085633	0.000127572	0.017411471
C1QTNF12	−2.825191103	3.222153422	0.000234636	0.02623382
DTNA	−2.811280027	1.326718473	8.55E−05	0.014253189
KRT1	−2.797637848	2.613754179	6.92E−05	0.013165864
RTN4RL2	−2.797383855	3.565203693	3.56E−06	0.002151673
NMB	−2.785464392	4.769423606	0.000144544	0.018741018
NKAIN4	−2.725835329	3.32315051	0.000175082	0.021316194
DUSP15	−2.712099618	1.402249593	7.45E−05	0.013764626
IGFBP6	−2.641385346	4.705648348	0.000357201	0.031999183
IRX1	−2.641371905	1.725281185	1.65E−06	0.001347512
KIF19	−2.599860383	2.634519615	0.000756833	0.047520993
LAMP5	−2.521963474	3.634943772	0.000114638	0.016448228
HEY1	−2.482922047	4.131715584	0.000115877	0.016448228
IL17RD	−2.482847407	1.563467996	0.00074938	0.047350848
NPTX2	−2.474916128	1.944105866	0.000114859	0.016448228
PRRX1	−2.418850644	5.086227953	0.000511045	0.039550531
RAMP1	−2.418519882	4.304251887	0.000157601	0.020171961
SPINK5	−2.401970407	1.419776083	0.000392375	0.033673167
PRDM6	−2.376809933	2.107794342	0.000258927	0.027114371
COL26A1	−2.360436459	4.132183181	0.000365692	0.03234187
PDZD7	−2.340246855	1.531649414	0.000262304	0.027136948
KRT17	−2.338793792	5.931256513	4.11E−05	0.009125326
S100A1	−2.335433447	2.256553923	8.31E−05	0.014253189
TSPEAR	−2.196880683	1.24295098	0.000641817	0.044285573
NNAT	−2.163234419	2.316903698	0.000406619	0.034257196
CPAMD8	−2.135732976	3.744044433	0.000728403	0.04638443
BCAM	−2.119331059	6.855164256	9.10E−05	0.014649376
KCNJ5	−2.074088058	2.820417316	4.19E−05	0.009192939
MAST1	−2.063241259	1.728215681	0.000334117	0.031503459
RARRES2	−2.050658472	4.309808115	0.00024593	0.026833276
MYL9	−2.037376153	6.269218628	3.35E−05	0.007924085
RADIL	−2.017851476	2.206839739	0.000617665	0.04373377
MAP1B	−1.983435991	3.693569233	0.000593846	0.042806191
ADPRHL1	−1.983165118	2.371933935	0.000411465	0.034519841
ADM	−1.908421627	2.164461803	0.000166279	0.02082272
KIF12	−1.839197112	4.042257933	6.80E−05	0.013162866
A4GALT	−1.746454553	3.356362608	0.000628883	0.044214475
AFAP1L2	−1.708529952	5.629806254	0.000699888	0.046273742
EPHX3	−1.704391406	1.161569209	0.000131591	0.017532068
RYR1	−1.690586943	2.119175707	0.000461127	0.036829271
KCNIP3	−1.591561724	6.27756293	0.000235181	0.02623382
RBP1	−1.396617446	4.330992858	0.000774057	0.048298738
RGS9	−1.386737859	2.212607056	0.000317832	0.03054823
IGFBP4	−1.379489998	6.898794948	0.000192914	0.022658341
RAB3IL1	−1.325468721	5.364223263	0.000459158	0.036827158
COPZ2	−1.319279321	4.042731219	0.000626393	0.044194982
PXDC1	−1.284119427	3.656120313	0.000660007	0.044907927
CTSF	−1.231637947	4.859352108	0.000638457	0.044264104
KIFC3	−1.14093435	3.501584415	0.000754573	0.047520993
CSRP1	1.031898606	7.236290106	0.000196095	0.022764126
C1QTNF6	1.048423636	4.215725309	0.000547396	0.04108972
PTPN3	1.064663452	4.79852239	0.00063754	0.044264104
PTPRS	1.084992749	6.321535934	4.01E−05	0.008992162
ATM	1.092312455	6.065074206	0.000166857	0.02082272
CAB39L	1.114221581	4.635029486	0.000459256	0.036827158
EOGT	1.13860395	5.023332339	2.33E−05	0.006539235
SYNE1	1.15698623	6.739375363	6.11E−05	0.012327227
ST3GAL1	1.160699732	5.608007883	6.86E−06	0.003183191
CCDC88C	1.175757769	6.84430386	2.17E−05	0.006279865
MCOLN3	1.176635199	4.748109433	0.000175053	0.021316194
AMOTL1	1.224644988	7.927186252	0.000573374	0.042245614
CD226	1.251137605	2.717868443	0.000356512	0.031999183
PAQR6	1.278244251	2.951705012	0.000813914	0.049547031
C11orf21	1.302413542	2.972934344	0.000586808	0.042551936
ADAMTS15	1.346155525	3.887698093	0.000550948	0.041201438
COL4A5	1.376611342	5.435165895	0.000658274	0.044907927
TSPAN32	1.415877755	3.247474464	0.000352642	0.031999183
PCED1B	1.447718826	3.95648763	0.000580164	0.042551936
PEX11G	1.450854953	3.682162391	0.000358722	0.031999183
BNC1	1.463281485	4.041859479	0.000164864	0.02082272
CABLES1	1.466009022	5.524096702	2.80E−05	0.007261196
BTLA	1.493058148	2.296589938	0.000661238	0.044907927
ZNF831	1.521658299	1.931796715	0.000521141	0.040021588
POU5F1B	1.568868299	1.274084132	0.000509381	0.039550531
AMDHD1	1.573918649	1.388503093	0.000806365	0.049388597
PABPC4L	1.585601776	3.026351018	0.000604812	0.043284181
CCR7	1.613695957	5.827382462	0.000334489	0.031503459
SUSD4	1.62114103	5.625537896	9.50E−05	0.015059557
CHRM3−AS2	1.687459799	2.668354407	0.000305892	0.029793877
DSC2	1.713299303	5.811930077	0.000128245	0.017411471
BTNL9	1.718941315	6.493067076	0.000171341	0.021118357
OGFRL1	1.724670542	6.3800536	1.27E−07	0.000253168
KIAA1614	1.724837466	4.390292809	8.45E−05	0.014253189
TMEM221	1.747277292	3.258767333	0.000793345	0.048829234
SULT1B1	1.751701245	1.066856666	0.000253265	0.027108324
TNFRSF25	1.760189383	4.648424342	1.57E−07	0.000285793
SYBU	1.795524926	5.823073573	0.000209855	0.023943912
S1PR5	1.818352037	1.726708795	0.000513812	0.039611173
TLR5	1.874988549	3.928291175	1.61E−05	0.005102521
ABCD2	1.912033944	1.47977496	8.64E−05	0.014253189
RGPD1	1.962107442	1.447866581	0.000240462	0.026426386
VIPR1	1.990615221	2.346466784	4.73E−05	0.010050012
DSC3	2.060586958	6.23696081	2.64E−06	0.001797271
PITPNM3	2.066183086	3.19477902	0.000279793	0.028215335
CADPS2	2.072964696	5.37328009	0.000237964	0.026396822
CNKSR2	2.118672531	3.24874403	1.34E−05	0.004690262
LRRN3	2.156274312	4.605360993	1.75E−05	0.005305741
SCN2B	2.185969726	3.524182011	0.000259038	0.027114371
TMEM163	2.222626461	5.546379096	0.000443157	0.03611641
ADGRB1	2.268231032	4.599635445	0.000354712	0.031999183
TRMT9B	2.270450369	4.289344976	0.000546679	0.04108972
RBFOX3	2.281853526	1.166636114	0.000730959	0.04638443
LRRC4	2.310897045	5.022877443	3.72E−06	0.002186462
VSIG1	2.324453749	2.360575948	0.000350873	0.031999183
SGIP1	2.364819235	3.463574899	8.82E−06	0.003621892
ANKRD55	2.385252316	2.396103806	0.000343918	0.031939584
HR	2.392649271	5.773120702	5.40E−06	0.002695495
CYP21A2	2.396917721	2.816431582	0.000123312	0.017098387
PLEKHD1	2.427148309	2.545280033	0.000722988	0.04638443
DCST2	2.469699059	1.817009447	0.000108736	0.016202501
CACNA1H	2.554062761	2.753556499	0.000397447	0.033771062
MAP2	2.595734303	5.832143722	1.22E−05	0.004425089
C2CD4A	2.621694006	3.171878901	0.000440538	0.036050107
NAPSA	2.648866778	2.812059645	0.000111476	0.016366482
DYSF	2.654315409	6.120007721	4.50E−06	0.002430803
BBOX1	2.707435961	2.624542486	2.94E−05	0.00744023
MIR452	2.726183047	4.100742925	9.79E−05	0.015278367
TRIM7	2.749984944	3.966026316	3.02E−08	0.000120607
MEGF11	2.768338744	3.901573036	0.000510894	0.039550531
ANAPC1P1	2.791896532	1.466174301	0.000240878	0.026426386
CDH26	2.817492308	3.004824858	2.05E−06	0.001512784
B3GAT1	2.893594108	3.129109812	7.51E−05	0.013764626
NGB	2.951037162	2.930569607	0.000684676	0.045695125
TCEAL2	3.097773295	2.284127821	0.000392941	0.033673167
DAB1	3.238208706	1.459708099	0.000669547	0.045110642
CADPS	3.292163784	1.366927466	0.000501109	0.039237812
VWC2	3.392777854	3.600284028	0.000248537	0.026970359
HAS2	3.449016578	1.627563507	5.77E−05	0.011869445
PLXNB3	3.475439297	3.383279965	7.05E−07	0.000880073
CYP4F3	3.787744948	1.921251767	6.75E−05	0.013162866
SFTPD	3.953404307	2.313645037	0.000457005	0.036827158
MIR614	4.203046253	1.614918477	0.000108146	0.016202501
LHFPL4	4.340877226	1.0676253	0.000438816	0.036050107
VEGFD	5.089525645	1.953670015	7.86E−13	1.57E−08
OLFM3	5.49125031	2.657082163	1.86E−08	9.29E−05
SALL3	5.858632642	1.337707021	2.36E−05	0.006539235
SCGB1A1	6.134443217	1.230518068	0.000617646	0.04373377
SLITRK6	6.433479385	4.049025947	1.70E−08	9.29E−05
SFTPA2	7.881535806	5.599887412	1.45E−05	0.004857241
OLIG2	7.883686911	2.267494657	0.000286309	0.028543087
SFTPA1	8.457267837	5.468790881	8.88E−06	0.003621892
SFTPC	9.764859884	6.800645422	5.66E−08	0.00018848
SFTPB	10.40211808	5.172167416	3.64E−07	0.000605947

FC, fold change; CPM, counts per million; FDR, false discovery rate.

Figure S1.

Differentially expressed genes by edgeR. Volcano plot illustrating the differential expression levels of genes of TAMG and NMG group. Genes that are significantly up- and down-regulated in the TAMG group compared to the NMG group are shown in red and green dots, respectively. TAMG, thymoma-associated myasthenia gravis; NMG, non-myasthenia gravis.

Figure S2.

Heat map of the hierarchical clustering based on DEGs (fold change =1, P value <0.05). DEG, differentially expressed gene.

Analysis of tissues and organs (Table S4), diseases (Table S5), biological superpathways (Table S6), GO pathways (Table S7), phenotypes (Table S8) and compounds (Table S9) using GeneAnalytics were listed in Tables S4-S9, respectively. However, we did not see a clearly biological meaning from these results.

Table S4. GeneAnalytics program mapping of systems with 10 highest match scores for 169 DEGs in TAMG and NMG.

System	Genes matched to system	No. of genes in systems	Score
Nervous system	ABCD2, ADGRB1, ADM, AFAP1L2, AMDHD1, ANKRD55, ATM, B3GAT1, BBOX1, CACNA1G, CACNA1H, CADPS, CCR7, CD226, CNKSR2, COL26A1, COL4A5, COPZ2, CRYAB, CSRP1, CTSF, CYP19A1, CYP4F3, DAB1, DTNA, DUSP15, DYSF, EOGT, EPB41L4B, HAS2, HEY1, IGFBP4, IGFBP6, IL17RD, IRX1, KCNIP3, KCNJ5, KIF1A, KRT14, KRT17, LAMP5, LRRN3, MAP1B, MAP2, MAST1, MCOLN3, MEGF11, MYL9, NAPSA, NEUROD1, NGB, NGEF, NKAIN4, NMB, NNAT, NPTX2, OGFRL1, OLFM3, OLIG2, PABPC4L, PAQR6, PENK, PLEKHD1, PLXNB3, PRRX1, PTPN3, PTPRS, PXDC1, RAB3IL1, RAMP1, RARRES2, RBFOX3, RBP1, RGPD1, RGS9, S100A1, SALL3, SCN2B, SCUBE1, SEZ6L2, SGIP1, SLITRK6, ST3GAL1, SULT1B1, SUSD4, SYBU, SYNE1, TBX1, TCEAL2, TMEM163, TNFRSF25, TRMT9B	92	11.62
Muscoskeletal system	ABCD2, ADAMTS15, ADPRHL1, AMOTL1, ANKFN1, ATM, B3GAT1, BCAM, BNC1, BTLA, BTNL9, C11orf21, CABLES1, CACNA1G, CACNA1H, CADPS, CCR7, CHRNG, COL26A1, COL4A5, COPZ2, CPAMD8, CRYAB, CSRP1, CTSF, CYP4F3, DAB1, DSC2, DTNA, DYSF, EOGT, EPB41L4B, HAS2, HEY1, HR, IGFBP4, IGFBP6, IL17RD, IRX1, KCNIP3, KIF1A, KRT14, KRT17, LAMP5, LRRN3, MAP2, MCOLN3, MIR452, MYBPC2, MYH6, MYL9, NEUROD1, NGEF, NKAIN4, NMB, NNAT, NPTX2, OGFRL1, OLIG2, PENK, PITPNM3, PRRX1, RAB3IL1, RAMP1, RARRES2, RBFOX3, RBP1, RGPD1, RYR1, S100A1, SALL3, SEZ6L2, SFTPA1, SLITRK6, ST3GAL1, SULT1B1, SYBU, SYNE1, TBX1, TCEAL2, TMEM163, TREM2, TRIM55, TRIM7, TSPAN32, TWIST1, VEGFD, VIPR1	88	10.03
Reproductive system	ADAMTS15, ADM, ANKRD55, ATM, B3GAT1, BNC1, CAB39L, CABLES1, CADPS, CADPS2, CCR7, CD226, COL26A1, COL4A5, COPZ2, CRYAB, CSRP1, CTSF, CYP19A1, CYP4F3, DAB1, DPEP1, DSC2, DSC3, DTNA, DUSP15, DYSF, HAS2, HEY1, HR, IGFBP4, IGFBP6, IL17RD, IRX1, KCNIP3, KCNJ5, KIF1A, KIFC3, KRT1, KRT17, MAP1B, MAP2, MCOLN3, MEGF11, MYL9, NEUROD1, NNAT, NPTX2, PAQR6, PENK, PEX11G, PRRX1, PTPRS, RAB3IL1, RADIL, RAMP1, RARRES2, RBP1, RGPD1, RGS9, S100A1, SOHLH1, SPINK5, ST3GAL1, SUSD4, SYNE1, TCEAL2, TFDP3, TMEM163, TWIST1, VEGFD, VSIG1	72	7.19
Cardiovascular system	ADAMTS15, ADM, ADPRHL1, ANKRD55, ATM, B3GAT1, BCAM, BNC1, C1QTNF6, CACNA1H, CD226, COPZ2, CRYAB, CSRP1, CTSF, DAB1, DSC2, DSC3, DTNA, DYSF, EOGT, HAS2, HEY1, IGFBP4, IGFBP6, IRX1, KCNIP3, KCNJ5, KIF1A, KRT1, KRT14, LRRN3, MAP1B, MCOLN3, MEGF11, MIR452, MYH6, MYL9, NEUROD1, NKAIN4, OGFRL1, PAQR6, PENK, PLEKHD1, PRRX1, PXDC1, S100A1, SALL3, SPINK5, ST3GAL1, SULT1B1, TBX1, TMEM163, TRIM55, TRIM7, TWIST1	56	6.39
Hematopoietic system	A4GALT, ADGRB1, ADM, AFAP1L2, AMOTL1, ANKRD55, ATM, BTLA, BTNL9, CABLES1, CACNA1G, CADPS2, CCDC88C, CCR7, CD226, CSRP1, CYP4F3, DSC2, DYSF, EPB41L4B, HAS2, HEY1, KCNIP3, KCNJ5, KRT1, KRT14, KRT17, LAMP5, LRRN3, MYL9, NAPSA, NEUROD1, OGFRL1, OLIG2, PABPC4L, PCED1B, PENK, PITPNM3, PRRX1, RAB3IL1, RAMP1, RBP1, RGPD1, S100A1, SFTPB, SFTPC, SGIP1, SLITRK6, SMIM1, SPINK5, ST3GAL1, SULT1B1, SYNE1, TBX1, TLR5, TNFRSF17, TNFRSF25, TREM2, TSPAN32, VEGFD, VSIG1, ZNF831	62	6.32
Respiratory system	AMOTL1, ANKFN1, ATM, BCAM, BNC1, BTNL9, CCR7, CD226, CDH26, COL4A5, CRYAB, CSRP1, CYP4F3, DAB1, DSC2, DSC3, EOGT, HR, IRX1, KCNIP3, KCNJ5, KIF1A, KIFC3, KRT14, KRT17, LAMP5, MAP2, MYH6, MYL9, NAPSA, NKAIN4, NMB, NNAT, NPTX2, PAQR6, PENK, PLXNB3, PRRX1, RADIL, RARRES2, RBP1, RYR1, S1PR5, SCGB1A1, SFTPA1, SFTPA2, SFTPB, SFTPC, SFTPD, SLITRK6, SULT1B1, SYNE1, TBX1, TMEM163, TREM2, VEGFD, VIPR1	57	6.26
Sensory organs	ABCD2, ANKRD55, C1QTNF6, CADPS, CADPS2, CCR7, CHRNG, COL4A5, COPZ2, CPAMD8, CRYAB, CSRP1, CYP4F3, DAB1, DSC2, DTNA, EPHX3, IGFBP6, IRX1, KCNIP3, KCNJ5, KIAA1614, KIF1A, KRT14, LAMP5, MAP1B, MAP2, MEGF11, MYL9, NEUROD1, NKAIN4, NNAT, PABPC4L, PENK, PRRX1, PTPN3, PTPRS, RAB3IL1, RADIL, RARRES2, RBP1, RGPD1, S1PR5, SALL3, SCUBE1, SEZ6L2, SFTPC, SGIP1, SLITRK6, ST3GAL1, SYNE1, TBX1, TNFRSF17, TRMT9B, VWC2	55	6.05
Integumentary system	A4GALT, ADGRB1, ADM, ATM, BBOX1, BNC1, CABLES1, CACNA1G, CADPS, CCR7, COL4A5, COPZ2, CRYAB, CSRP1, CYP4F3, DAB1, DPEP1, DSC2, DSC3, DYSF, EPHX3, HAS2, HR, IGFBP4, IRX1, KCNIP3, KCNJ5, KIF1A, KRT1, KRT14, KRT17, LAMP5, LRRC4, LRRN3, MAP2, MCOLN3, MYL9, NEUROD1, NKAIN4, NPTX2, PENK, PRRX1, PTPRS, RAMP1, RYR1, S100A1, S1PR5, SCUBE1, SPINK5, ST3GAL1, SYBU, SYNE1, TBX1, TCEAL2, TWIST1	55	5.81
Endocrine system	AFAP1L2, ATM, B3GAT1, C2CD4A, CAB39L, CACNA1G, CACNA1H, CADPS, CADPS2, CRYAB, CSRP1, CTSF, CYP21A2, CYP4F3, DPEP1, DSC2, EPB41L4B, HEY1, IGFBP6, IRX1, KCNIP3, KCNJ5, KIF1A, KRT14, KRT17, MAP1B, MAP2, MCOLN3, MYH6, MYL9, NEUROD1, NGEF, NNAT, NPTX2, PENK, PRRX1, RAMP1, RARRES2, RBFOX3, RBP1, RGS9, S100A1, SCGB1A1, SEZ6L2, SLC30A8, SLITRK6, SPINK5, ST3GAL1, SYBU, SYNE1, TRMT9B	51	5.12
Gastrointestinal tract	ABCD2, ADM, AFAP1L2, ATM, B3GAT1, BCAM, C2CD4A, CABLES1, CCR7, CD226, CHRNG, COL4A5, CRYAB, CSRP1, CTSF, CYP4F3, DAB1, DPEP1, DSC2, DSC3, DYSF, EPB41L4B, HAS2, HEY1, IGFBP4, IGFBP6, IRX1, KCNIP3, KCNJ5, KIF1A, KRT1, KRT14, KRT17, LAMP5, MAP1B, MAP2, MCOLN3, MYBPC2, MYH6, MYL9, NNAT, PENK, PRRX1, RARRES2, RBP1, S100A1, SEZ6L2, SFTPD, SLITRK6, SPINK5, ST3GAL1, SULT1B1, TBX1, TNFRSF17, TRIM7, VIPR1, VSIG1, ZNF831	58	4.68

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Table S5. GeneAnalytics program mapping of diseases with 10 highest match scores for 169 DEGs in TAMG and NMG.

Disease	Genes matched to disease type	No. of genes in disease type	Score
Lung cancer	KRT14, KRT17, SFTPC, SFTPD, VEGFD, ATM, NAPSA, NMB, SCGB1A1, DSC3, SFTPA2, BNC1, BTNL9, CAB39L, LRRN3, SFTPB	16	9.81
Pulmonary fibrosis, idiopathic	SFTPA2, SFTPA1, SFTPC, CCR7, SCGB1A1, SFTPB, SFTPD	7	8.80
Breast cancer	ATM, KRT14, KRT17, CYP19A1, TWIST1, VEGFD, AMDHD1, BBOX1, DSC3, IRX1, KCNIP3, LAMP5, MAP2, SCUBE1, TCEAL2	15	8.33
Polycystic kidney disease	BBOX1, CYP4F3, DPEP1, IGFBP6, PRRX1, ADAMTS15, AMDHD1, CNKSR2, LAMP5, MAP1B, MYL9, NAPSA, TWIST1	13	7.22
Neuroblastoma	CYP21A2, KRT1, CADPS, MAP1B, MAP2, ADM, CADPS2, SNKSR2, EPB41L4B, KIF1A, MYL9, NNAT, RAB3IL1, TCEAL2	14	7.03
Respiratory distress syndrome in premature infants	SFTPB, SFTPC, SFTPA1	3	6.83
Dilated cardiomyopathy	MYH6, CRYAB, DSC2, DTNA, MYBPC2, SYNE1, TWIST1	7	6.50
Lung cancer susceptibility 3	SCGB1A1, SFTPD, VEGFD, NAPSA, SFTPB, SFTPC, NPTX2, PLXNB3, SFTPA2, VIPR1	10	6.48
Colorectal cancer	DPEP1, ATM, CACNA1G, SYNE1, VEGFD, NEUROD1, CAB39L, CD226, LRRN3, MYL9, TFDP3	11	6.02
Tuberous sclerosis complex	CADPS, CNKSR2, DTNA, PRRX1, CYP4F3, EPB41L4B, LAMP5, NEUROD1, NGEF, OLFM3, SCN2B	11	5.97

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Table S6. GeneAnalytics program mapping of superpathways with 10 highest match scores for 169 DEGs in TAMG and NMG.

Superpathways	Genes matched to superpathways	No. of genes in superpathways	Score
Defective CSF2RB causes pulmonary surfactant metabolism dysfunction 5	SFTPA2, SFTPB, SFTPC, SFTPA1, SFTPD	5	26.70
Surfactant metabolism	SFTPA2, SFTPB, SFTPC, SFTPA1, SFTPD, NAPSA	6	22.03
Diseases of metabolism	SFTPA2, SFTPB, SFTPC, SFTPA1, SFTPD	5	14.75
FOXA1 transcription factor network	SCGB1A1, SFTPA2, SFTPA1, SFTPD	4	11.03
Antiarrhythmic pathway, pharmacodynamics	CACNA1G, CACNA1H, SCN2B, KCNJ5	4	9.72
Circadian entrainment	KRT17, KRT14, KRT33B, COL4A5, VEGFD, CACNA1G, CACNA1H, RYR1, KCNJ5, KCNQ4	10	9.70
Keratinization	KRT17, KRT14, KRT1, KRT33B, SPINK5, DSC3, DSC2	7	8.89
Myometrial relaxation and contraction pathways	ADM, RAMP1, RGS9, IGFBP6, IGFBP4, RYR1, KCNJ5	7	8.74
Cardiac conduction	DYSF, RYR1, SCN2B, KCNIP3, MYBPC2, MYH6, MYL9	7	8.28
Developmental biology	KRT17, KRT14, KRT1, CNKSR2, KRT33B, COL4A5, NEUROD1, NGEF, PLXNB3, SPINK5, DSC3, DSC2, CACNA1G, CACNA1H, IL17RD, SCN2B, MYL9, TREM2	18	8.14

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Table S7. GeneAnalytics program mapping of gene ontology (GO) molecular function and biological processes with 5 highest match scores for 169 DEGs in TAMG and NMG.

Variable	Genes matched	No. of genes	Score
A. GO-molecular function
Lipopolysaccharide binding	ADGRB1, SFTPA2, SFTPA1, SRTPD, TREM2	5	16.93
Microtubule binding	CRYAB, MAP1B, MAP2, SYBU, CCDC88C, KIF12, KIF19, KIF1A, MYH6, KIFC3, SGIP1	11	15.87
Monosaccharide binding	SFTPA2, SFTPA1, SFTPD	3	12.74
Low voltage-gated calcium channel activity	CACNA1G, CACNA1H	2	10.89
Signaling receptor activity	KRT1, OGFRL1, RAMP1, TNFRSF25, RTN4RL2, TNFRSF17, TREM2	7	10.49
B. GO-biological process
Surfactant homeostasis	SFTPA2, SFTPA1, SFTPD, NAPSA	4	19.15
Muscle contraction	CRYAB, DYSF, CACNA1H, RYR1, MYBPC2, MYH6, MYL9, CHRNG	8	17.62
Respiratory gaseous exchange	SFTPA2, SFTPB, SFTPC, SFTPA1, SFTPD	5	15.93
Cornification	KRT17, KRT14, KRT1, KRT33B, SPINK5, DSC3, DSC2	7	14.34
Cellular protein metabolic process	PENK, IGFBP6, IGFBP4, SFTPA2, SFTPB, SFTPC, SFTPA1, SFTPD, NAPSA	9	13.35

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Table S8. GeneAnalytics program mapping of phenotypes with 10 highest match scores for 169 DEGs in TAMG and NMG.

Phenotypes	Genes matched to phenotypes	Genes	Score
Abnormal surfactant physiology	SFTPB, SFTPC, SFTPA1, SFTPD	4	18.61
Dehydration	KRT1, NEUROD1, MAP1B, SPINK5, SALL3, CHRNG	6	14.98
Impaired glucose tolerance	COL4A5, SLC30A8, CYP19A1, VIPR1, RARRES2, RGPD1, IGFBP4, CADPS2	8	11.57
Increased or absent threshold for auditory brainstem response	ADGRB1, SLITRK6, MCOLN3, PDZD7, TBX1, IL17RD, KCNQ4	7	10.3
Unresponsive to tactile stimuli	MAP1B, RYR1, KIF1A	3	10.2
Blistering	KRT14, KRT1, SPINK5	3	10.2
No spontaneous movement	OLIG2, RYR1, CHRNG	3	10.02
Abnormal cerebellum morphology	NEUROD1, MAP1B, DAB1, CADPS2, IL17RD	5	9.82
Centrally nucleated skeletal muscle fibers	CRYAB, DYSF, SYNE1, RYR1	4	9.72
Dystrophic muscle	DTNA, DYSF, SYNE1	3	9.52

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Table S9. GeneAnalytics program mapping of compounds with 10 highest match scores for 169 DEGs in TAMG and NMG.

Compounds	Genes matched to compounds	Genes	Score
Calcium	KRT14, KRT1, NEUROD1, NMB, CRYAB, PITPNM3, MAP1B, MAP2, NPTX2, CYP21A2, DAB1, VIPR1, B3GAT1, DSC3, DSC2, PTPN3, DUSP15, DYSF, RARRES2, PENK, IGFBP6, IGFBP4, CACNA1G, CADPS2, CACNA1H, CADPS, TNFRSF17, RYR1, S100A1, CCR7, CDH26, SCGB1A1, KCNIP3, KCNJ5, KCNQ4, SFTPA2, MYH6, SFTPB, SFTPC, SFTPA1, SFTPD, MYL9	42	19.79
Progesterone	KRT14, KRT1, ADM, CRYAB, MAP2, CYP21A2, CYP19A1, VEGFD, B3GAT1, RBP1, PENK, IGFBP6, IGFBP4, S100A1, SCGB1A1, SFTPB	16	15.17
Estrogen	KRT14, KRT1, ADM, TWIST1, MAP2, CYP21A2, CYP19A1, VEGFD, ATM, B3GAT1, PTPN3, SULT1B1, RBP1, PENK, IGFBP6, IGFBP4, CACNA1G, S100A1, SCGB1A1, KCNJ5, MYH6	21	14.51
Infasurf	SFTPB, SFTPC	2	12.88
Calcium citrate	CACNA1G, CADPS2, CACNA1H, CADPS, RYR1	5	12.78
Calcium phosphate	CACNA1G, CADPS2, CACNA1H, CADPS, RYR1	5	12.34
Retinoic acid	KRT17, KRT14, KRT1, NEUROD1, ADM, MAP2, OLIG2, VIPR1, HAS2, RARRES2, RBP1, PENK, IGFBP6, IGFBP4, CACNA1G, S100A1, MYH6, SFTPB, SFTPC	19	11.82
Paraffin	KRT17, KRT14, KRT1, ADM, CRYAB, MAP1B, MAP2, ATM, RBP1, S100A1, SFTPB	11	11.76
POPG	SFTPB, SFTPC	2	11.72
ML218	CACNA1G, CACNA1H	2	11.72

DEG, differentially expressed gene; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

DEGs related to T cells

Since thymus is the place where T cells develop, we further selected DEGs in T cells between thymoma tissues of TAMG and NMG by LifeMapDiscovery Cells and Tissues Database. Among 169 DEGs, there were 6 genes mainly expressing in T helper cells (ANKRD55, BTLA, CCR7, LRRN3, TNFRSF25, VSIG1), 3 in T cytotoxic cells (CCR7, LRRN3, TNFRSF25) and 4 in double positive (DP) thymocytes (ATM, CCR7, SFTPB, TNFRSF25) with all 8 genes up-regulated. The genes were exhibited with their log₂FC, log₂CPM, P value and FDR in Table 1. Further screened by VarElect, 6 genes (ATM, SFTPB, ANKRD55, BTLA, CCR7, TNFRSF25) were directly related while 2 genes (LRRN3, VSIG1) were indirectly related to the phenotype of autoimmune diseases.

Table 1. Eight T-cell associated DEGs screened by LifeMapDiscovery Cells and Tissues Database.

Gene	Log2FC	Log2CPM	P value	FDR
ATM	1.09	6.07	0.000167	0.0208
SFTPB	10.4	5.17	3.64E−07	0.000606
ANKRD55	2.39	2.40	0.000344	0.0319
BTLA	1.49	2.30	0.000661	0.0450
CCR7	1.61	5.83	0.000334	0.0315
TNFRSF25	1.76	4.65	1.57E−07	0.000286
LRRN3	2.16	4.61	1.75E−05	0.00531
VSIG1	2.32	2.36	0.000351	0.0320

DEG, differentially expressed gene; FC, fold change; CPM, counts per million; FDR, false discovery rate.

Validation by RT-qPCR

Patients and samples

Demographic and clinical characteristics were summarized in Table 2. Disease severity was evaluated using the MGFA Clinical Classification (16). There was no statistical difference in sex, age, WHO histologic classification of thymoma and MGFA classification between TAMG and NMG group.

Table 2. Demographic and clinical characteristics of the subjects included in the study.

Classification/group	NMG	TAMG
Gender
Female	5	2
Male	4	7
Age	18–62 (mean 50.2, median 55)	30–62 (mean 46, median 42)
WHO histologic classification
A	1	1
AB	2	–
B1	2	–
B2	3	6
B3	1	2
MGFA classification
1	–	1
2a	–	2
2b	–	3
3a	–	3
AChR antibody
Positive	–	9
Negative	–	0

TAMG, thymoma-associated MG; NMG, non-MG thymoma; WHO, World Health Organization; MGFA, Myasthenia Gravis Foundation of America; AChR, acetylcholine receptor; MG, myasthenia gravis.

RT-qPCR

The relative expression level of mRNA in 6 genes (ATM, SFTPB, ANKRD55, BTLA, CCR7, TNFRSF25) that are direct related to autoimmune disease were compared between the thymoma tissues of the TAMG and NMG group. Compared with NMG, significantly higher expression of BTLA (P=0.022), CCR7 (P=0.0018), TNFRSF25 (P=0.013) and ANKRD55 (P=0.0026) were observed in the TAMG group (Figure 1A,B,C,D). However, we did not observe a difference in expression of ATM (P=0.082) and SFTPB (P=0.11) between the two groups (Figure 1E,F). These results may suggest a specific association between the overexpression of BTLA, TNFRSF25, CCR7 or ANKRD55 and the mechanism of TAMG. Since BTLA negatively regulated the immune response, the elevated expression of BTLA could not be explained in the TAMG group. However, soluble BTLA (sBTLA), the isoform of BTLA from alternative splicing, could exhibit a counteracted function from BTLA (17). We further performed RT-qPCR of sBTLA using its specific primers and it was revealed that there was a significant difference (P=0.027) between the two groups (Figure 2).

Figure 1.

Validation of the differences of six directly immune-related DEGs using RT-qPCR. (A-D) Compared with NMG, significantly higher expression of BTLA (P=0.022), CCR7 (P=0.0018), TNFRSF25 (P=0.013) and ANKRD55 (P=0.0026) was observed in TAMG. (E,F) The difference in expression of ATM (P=0.082) and SFTPB (P=0.11) was not significant between the two groups. *, P<0.05; **, P<0.01. DEG, differentially expressed gene; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; NMG, non-myasthenia gravis thymoma; TAMG, thymoma-associated myasthenia gravis.

Figure 2.

Validation of the difference of soluble BTLA using RT-qPCR. A significant difference (P=0.027) was observed between the two groups. *, P<0.05. RT-qPCR, reverse transcription-quantitative polymerase chain reaction.

Discussion

In the present study, with high throughput data from 11 TAMG and 10 NMG from TCGA database, we tried to find the possible molecular changes in the pathogenesis of TAMG. With the edgeR analysis tool, we found that among 24,991 transcripts, 169 genes showed a significant difference between the TAMG and NMG group. Further screening using LifeMapDiscovery Cells and Tissues Database and VarElect, overexpression of six genes (ATM, SFTPB, ANKRD55, BTLA, CCR7, TNFRSF25) directly related to autoimmune diseases expressed on T cells was found and four (ANKRD55, BTLA, CCR7, TNFRSF25) of them were validated in thymoma tissues in our center by RT-qPCR.

C-C chemokine receptor-7 (CCR7), a member of the C-C chemokine receptor subfamily, can combine with its ligand (CCL19 or CCL21). Mainly expressed on naïve and central memory CD4⁺ T cells, naïve B cells and mature dendritic cells (DCs), this receptor activates them to travel to their target tissues selectively and initiates the autoimmune response (18,19). It also mediates the settling of T-cell precursors in the thymus to initiate thymopoiesis (20). CCR7-deficient mice and mice lacking the expression of CCL19 and CCL21 exhibited defective migration of DCs and lymphocytes toward the T-cell zones (21). When under antigenic stimulation, some activated T cells lost their CCR7 expression and upregulated CXCR5, the receptor for CXCL13 expressed on B cells. CXCR5⁺ T cells then migrated towards the B cell-rich follicles to interact with B cells (22). Moschovakis et al. demonstrated that CCR7-deficient mice showed complete resistance to collagen-induced arthritis, and the antibody response to collagen II was severely impaired (23). Belikan et al. demonstrated that the induction of experimental autoimmune encephalomyelitis (EAE) was halted with the specific constitutive deletion of CCR7 on CD4⁺ T cells (24). When compared with normal subjects, the frequency of CD4⁺CCR7⁺ and CD8⁺CCR7⁺ T cells increased significantly in patients with primary Sjögren’s syndrome. The expression level of CCR7 on CD4⁺ T cells was associated with disease activity (25). In MG thymus, Li Qi et al. showed overexpression of CCR7 may block the differentiation of CD4⁺CD8⁺ DP thymocytes from the CD4^-CD8^- double negative (DN) stage, leading to the production of AChR autoreactive T cells (26). Our results also showed a significant higher expression of CCR7 in TAMG (P=0.0018), which further validated the positive correlation between CCR7 and TAMG.

B and T lymphocyte attenuator (BTLA), the ligand of herpesvirus entry mediator (HVEM), is expressed on DCs, natural killer (NK) cells, T and B lymphocytes (27). It was detected on double positive (DP) thymocytes, CD4⁺ or CD8⁺ single positive but not DN thymocytes (28). Belonging to the immunoglobulin superfamily and with similarities to cytotoxic T lymphocyte antigen 4 (CTLA-4) and programmed death 1 (PD-1), BTLA is another coinhibitory receptor that negatively regulate the immune response. It was demonstrated that BTLA-deficient mice have enhanced specific antibody reactions and improved sensitivity to EAE, with earlier disease onset, prolonged disease duration and increased clinical score (29). DCs require the functions of BTLA to actively adjust tolerizing T cell responses in CD5 dependent mechanisms (30). BTLA could also inhibit γδ T cell-dependent dermatitis through negatively regulating the production of IL-17 and tumor necrosis factor (TNF) (31). In multiple sclerosis, CD19⁺BTLA⁺ cells were significantly reduced compared with health controls while the disease remission was related to the increase in CD19⁺BTLA⁺ cells, suggesting a potentially pathogenic role (32). In our study, we first found the relationship between the higher expression of BTLA and TAMG. It could be explained by the possibility of reactive responses following antigen specific activation of T cells. Since we did not see such responses in other autoimmune diseases in literature, we further quantified and confirmed the elevation of sBTLA by RT-qPCR. In collagen-induced arthritis, soluble PD-1 aggravates the disease progression through Th1 and Th17 pathway (33). Like PD-1, the production of soluble form of BTLA may also interfere the normal regulatory function of HVEM/BTLA pathway and lead to the phenotype we observed.

Tumor necrosis factor receptor superfamily member 25 (TNFRSF25), also called death receptor 3 (DR3), is expressed on T cells and combines with the ligand of TNF-like protein 1A (TL1A) (34). In addition to mediating necroptotic cell death, this receptor has been shown to stimulate NF-κB activity through promoting the activation of phosphoinositide 3-kinase (PI3K) and Akt (protein kinase B, PKB), leading to the increased production of pro-inflammatory cytokines, enhanced differentiation and clonal expansion of T cells (35,36). It was demonstrated that the contact of TNFRSF25 and TL1A is required for the immunopathological process in EAE with IL-9 producing Th9 cells (37,38). This may explain the effect of elevated expression of TNFRSF25 in TAMG samples.

Although the precise function of ankyrin repeat domain containing protein 55 (ANKRD55) is still unknown, it functions almost exclusively to mediate protein-protein interactions (39). Single-nucleotide polymorphism (SNP) of this gene was reported to be related to several autoimmune diseases, such as rheumatoid arthritis (40) and multiple sclerosis (41). ANKRD55 is also overexpressed in EAE with neuroinflammation (42). These demonstrations in autoimmune diseases were in accordance with our results. However, the detailed mechanisms underlying TAMG need to be further clarified.

ATM, the ataxia telangiectasia-mutated gene, is required for DNA damage response and genome stability (43). The mutations in ATM have been reported in patients with brain and breast tumor (44). Surfactant Protein B (SFTPB) could influence lung function and innate immunity (45). Altered SFTPB is associated with several lung diseases including acute respiratory distress syndrome (46) and chronic obstructive pulmonary disease (47). In this study two genes mentioned above did not show significant difference between the two groups, and this may partly be due to the lower number of thymoma samples.

Combined with the findings above, the overexpression of 4 T-cell associated genes could be a characteristic of MG-associated thymoma with potential pathogenic importance. The induction and maintenance of MG may be related with the overexpression of some functional genes from T cells and other immune cells in the thymoma microenvironment. In other words, whether or not thymoma develops MG may depend on the heterogeneity of the thymoma microenvironment. As for how this heterogeneity was formed, further research should be conducted.

Of course, there are some limitations in this study. First, the conclusion may be limited by the small sample size. Second, the heterogeneity of individual patients including disease severity and WHO thymoma classification would also influence the reliability of the study. Third, tissue validation could not limit its results to T cells like bioinformatics analysis. Although aberrantly expressed genes were validated using RT-qPCR, they may originate in various cells from thymic tissues, so the results should be interpreted with caution. This was the first study to use TCGA database for investigating the possible molecular changes and pathogenesis in TAMG and in the study we find the correlation of TAMG with higher expression of CCR7, BTLA, TNFRSF25 as well as ANKRD55. The expression and SNPs of these genes may be promising biomarkers in TAMG and need to be further investigated.

Conclusions

In this study, we first identified 6 genes that were directly related to autoimmune diseases through GeneAnalytics from 169 DEGs calculated in TCGA. Then, overexpression of sBTLA, CCR7, TNFRSF25 and ANKRD55 was validated in thymoma tissues from TAMG in our center, which could partly explain the underlying pathogenesis of TAMG.

Ethical Statement: The study was approved by Fudan University Institutional Review Board and written informed consents were granted.