A transcriptional signature associated with non-Hodgkin lymphoma in the blood of patients with Q fever.

Coxiella burnetii, the agent causing Q fever, has been associated with B-cell non-Hodgkin lymphoma (NHL). To better clarify this link, we analysed the genetic transcriptomic profile of peripheral blood leukocytes from patients with C. burnetii infection to identify possible links to lymphoma. Microarray analyses revealed that 1189 genes were expressed differently (p <.001 and fold change ≥4) in whole blood of patients with C. burnetii infection compared to controls. In addition, 95 genes expressed in patients with non-Hodgkin lymphoma (NHL) and in patients with C. burnetii persistent infection have allowed us to establish the 'C. burnetii-associated NHL signature'. Among these, 33 genes previously found modulated in C. burnetii-associated -NHL by the microarray analysis were selected and their mRNA expression levels were measured in distinct C. burnetii-induced pathologies, namely, acute Q fever, focalized persistent infection, lymphadenitis and C.burnetii-associated NHL. Specific genes involved in anti-apoptotic process were found highly expressed in leukocytes from patients with C. burnetii associated-NHL: MIR17HG, REL and SP100. This signature differed from that found for NHL-control group. Patients with C. burnetii lymphadenitis presented significant elevated levels of BCL2 and ETS1 mRNAs. Altogether, we identified a specific transcriptionnal signature for NHL during C. burnetii infection reflecting the up-regulation of anti-apoptotic processes and the fact that lymphadenitis might constitute a critical step towards lymphomagenesis.

Introduction

NHL is the most common type (~90%) of lymphoma in humans. Many risk factors have been identified for malignant transformation: immune disorders, infection, lifestyle, genetics, family history and profession [1]. The most common type of lymphoma is diffuse large B-cell lymphoma (DLBCL) which manifests as fast-growing nodal or extranodal tumors [2]. Neoplasic transformation from germinal center B cells results from the accumulation of genetic modifications. Some indolent follicular lymphomas (FL) can evolve into more aggressive cancers, such as DLBCL or Burkitt lymphoma (BL) [2]. Gene expression profiling using microrrays or RNAseq has revealed the huge diversity of mechanisms accounting for neoplasic transformation [3]. This technology has identified two molecularly distinct forms of DLBCL, one characteristic of germinal center B-cells (GCB) and one of activated-B cells (ABC) [3].

It is estimated that one fifth of all human cancers are linked to viral and bacterial infections, and this also applies to NHL [4]. Helicobacter pylori causes most gastric mucosa-associated lymphoid tissues (MALT) lymphomas and is the only bacterium to be classified by the WHO as a class I carcinogen [4]. In addition, Epstein Barr virus (EBV) is associated with Burkitt and nasal NK T-cell lymphoma, hepatitis C virus with splenic marginal zone lymphoma and DLBCL, Borrelia burgdorferi and Chlamydia psittacii with marginal zone lymphoma [4-6].

Coxiella burnetii is an intracellular bacterium causing Q fever, a symptomatic disease in 40% of the cases. The clinical manifestations of acute Q fever are pneumonia and hepatitis, while persistent focalized C. burnetii infection is dominated by cardiovascular infections and is more rare (5%) [7]. C. burnetii infection has recently been described as being involved in lymphomagenesis. To date, 45 cases of C. burnetii associated NHL have been reported in the literature [8-11]. C. burnetii persistent focalized infection was associated with a 5 and 14 fold increased risk of NHL in the Netherlands and in France respectively. In situ, C. burnetii was detected in the tumoral microenvironment within macrophages and plasmacytoid dendritic cells, but not in B-cell. Interleukin 10 was described as an immunomodulatory cytokine favoring B-cell proliferation [12]. Lymphadenitis is considered a critical step to lymphomagenesis associated with C. burnetii [13]. Of note, C. burnetii replicates in a parasitophorous vacuole within human cells and must repress apoptosis of its host to optimally accomplish its life cycle [14]. This is achieved by increasing the expression of the anti-apoptotic BCL2 protein family member Bcl-2-related protein A1 (BCL2A1) and by reducing the expression of the pro-apoptotic BCL2 protein family members BAX and BAK [14,15].

Here, we evaluated the role of C. burnetii infection in B-cell NHL by analysing the whole blood transcriptomic profile of patients with C. burnetii infection. As NHL is a multifactorial disease, we wanted to know whether the expression of genes involved in DLBCL and FL suggested by Flowers et al. was involved in cases of C. burnetii associated NHL (EXOC2, MYC, NCOA1, PTV1, CXCR5, ETS1, LPP and BCL2) [1]. Our data strongly suggest that persistent infection with C. burnetii causes alterations in the transcription of genes involved in anti apoptotic and proliferative process.

Materials and methods

Patients and case definition

Q fever definition

In the French National Reference Center (NRC) for Q fever, patients with positive serology consistent with C. burnetii infection according to the recently updated criteria were screened from 1991 to 2016. Acute Q fever infection was defined by the association of clinical symptoms (fever and/or hepatitis and/or pneumonia) with serologic criteria for phase II IgG levels ≥ 200 and phase II IgM levels ≥ 50 or seroconversion. Coxiella burnetii persistent infections, such as persistent hepatitis, endocarditis and vascular infection were determined as the recent update criteria [16][7]. In the French National reference center for Q fever, patients screened for C. burnetii serology are also screened for other intracellular bacteria such as Chlamydia, Bartonella and Rickettsia. Moreover, patients included in the group of C. burnetii associated with NHL were screened for EBV reactivation. All patients included in our study were found negative for these pathogen.

Lymphoma diagnosis

Diagnosis of lymphoma based on cytologic analysis was confirmed by a centralized reviewing by an expert haematologist as previoulsy described, and was typed according to criteria updated by the World health organization [17].

Study population

Among patients followed in the French NRC for Q fever, we selected 61 patients with acute Q fever, C. burnetii persistent infection, C. burnetii lymphadenitis or C. burnetii with lymphoma (Tables 1 & 2). For each patients, clinical information including medical past history, clinical symptoms, immunodepression as well as antibiotic and antitumoral treatment were collected. Healthy donor were used as control (Etablissement Français du sang). Twenty-one samples (5 controls, 8 acute Q fever and 8 C. burnetii persistent infection) were investigated with microarrays and 49 samples (10 controls, 11 acute Q fever, 11 C. burnetii persistent infection, 4 C. burnetii lymphadenitis, 4 C. burnetii associated-NHL and a control group of 10 patients with NHL without C. burnetii infection) were analyzed by real time quantitative reverse transcription-polymerase chain reaction (q-RTPCR). Among the 4 patients with C. burnetii associated NHL included in this study, one had follicular lymphoma, one had diffuse and large B-cell lymphoma, one a marginal zone lymphoma and one had a mucosa associated Gastric lymphoma [7]. Characteristics of patients selected for microarray and q-RTPCR experiments are summarized in Tables 1 and 2 respectively.

Table 1

Patients with C. burnetii infection characteristic for microarray analysis.

Patients characteristics	Acute Q fever	C. burnetii persistent infection
N =	8	8
Sex (men/women)	6/2	6/2
Age (mean±SD)	45±15	56±24
C. burnetii infectious focus
Endocarditis	0	6
Vascular infection	0	1
Hepatitis	5	1
Pneumonia	2	0
Flu-like syndrome	1	0
IgG I mediane [IQR]	75 [12.5–300]	6400 [25600–24800]
IgG II mediane [IQR]	800 [400–2000]	18 000 [6400–19200]
IgM II mediane [IQR]	300 [150–1200]	25 [0–425]
Outcome
Doxycycline alone	8	-
Doxycycline-Hydroxychloroquine	-	8
Surgery	0	2
Remission	8	8

IQR: interquartile range

Table 2

Patients characteristics for qRT-PCR analysis.

Patients characteristics	Acute Q fever	C. burnetii persistent infection	C. burnetii lymphadenitis	C. burnetii and lymphoma	ControlsNHL without C. burnetii infection
N =	11	11	4	4	10
Sex (men/women)	7/4	8/3	4/0	2/2	3/7
Age (mean±SD)	43±16	58±13	56±16	73±11	63±18
C. burnetii infectious focus					No C. burnetii infectious focus
Endocarditis	0	10	1	2
Vascular infection	0	1	-	1
Hepatitis	9	-	1	-
Pneumonia	1	-	1	1
Flu-like syndrome	1	-	-	-
IgG I mediane [IQR]	100 [0–200]	1600 [800–12800]	3400 [250–9600]	500 [150–800]	Negative
IgG II mediane [IQR]	1600 [800–2400]	3200 [1600–9400]	850 [1600–4200]	1000 [400–1600]
IgM II mediane [IQR]	400 [100–800]	0 [0–100]	50 [0–100]	0 [0–100]
Outcome					NA
Doxycycline alone	11	-	-	-
Doxycycline-Hydroxychloroquine	-	11	4	1
Surgery	0	0	-	-
Remission	11	11	4	2

IQR: interquartile range, NA: not applicable

An informed written consent was obtained from each subject and blood test was performed with the approbation of the ethical comitee of “Aix-Marseille University”. The study was conducted according to the principle of the Helsinki’s declaration.

Samples collection and ribonucleic acid (RNA) preparation

RNAs were extracted from whole blood (PAXgene tubes, Qiagen, Courtaboeuf, France) or PBMCs as previously isolated from EDTA tubes (Sigma Aldrich, Saint-Quentin Fallavier, France) using Ficoll (Eurobio, Lees Ulis, France) gradient and centrifugation as previously described [18] with RNeasy Mini Kit with a DNase I treatment to eliminate DNA contaminants and according to the manufacturer instructions (Qiagen, Courtaboeuf, France).

Microarray and data analysis

The quantity and the quality of extracted RNAs were assessed using a Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific, Eclubens, Switzerland) and 2100 Bioanalyzer (Agilent Technologies, Montpellier, France). Microarray experiments used chips containing 45.000 probes (4x44K Whole Human Genome) and One-Color Microarray-Based Gene Expression Analysis provided by Agilent Technologies. Four hundred ng of extracted RNAs were labeled using Cyanin-3 CTP using a low input Quick Amp Labeling kit, one color (Agilent technologies). This step lead to amplify and label target RNA to generate cRNA (Cy3-) for further oligo microarrays used in gene expression profiling. Labelled RNAs were hybridized for 17 hours at 65°C using QIAmp labeling kit according to the manufacturer’s recommendations. Slides were washed and scanned with a pixel size of 5 μm using a DNA Microarray scanner G2505C. The raw data were extracted using feature Extraction Software 10.5.1. Data were processed using GeneSpring GX 14.9 software (Agilent Technologies, Montpellier, France). Modulated genes were analyzed using Ingenuity Pathway Analysis (IPA, Qiagen, Courtaboeuf, France) or ClustVis software. The data have been deposited in NCBI’s Gene Expression Omnibus and are accessible via GEO series accession number GSE (GSE112086).

Reverse transcription-polymerase chain reaction

RT-PCR was performed in order to obtain cDNA using the Moloney murine leukemia virus-reverse transcriptase (MMLV-RT) kit (Life Technologies. Marseille. France) as previously described [18]. Specific genes found modulated by microarray or strongly involved in lymphoma according to prior work2 were then selected and validated by q-PCR using specific primers listed in the S1 Table as well as the SYBR Green Fast Master Mix (Roche Diagnostics, Meylan, France). The results were normalized to housekeeping gene β-actin and expressed as fold change (FC) = 2−ΔΔCt, which ΔΔCt = (CtTarget—CtActin)assay—(CtTarget—CtActin)control. The threshold cycle (Ct) was defined as the number of cycles required to detect the fluorescent signal. The expression of genes was considered as modulated when FC ≥ 1.5.

Statistical analysis

One-way ANOVA test was used to compare several groups of normally distributed variables. A post hoc Tukey’s honestly significant difference test was subsequently applied when ANOVA showed a p-value <0.05. A Kruskal-Wallis test by ranks was performed to compare non-normally distributed variables. A two-by-two comparison of non-parametric data was performed using a two-tailed non parametric Mann-Withney test. All tests were 2-sided and p <0.05 was considered significant.

Results

Relevant shift in gene transcription associated with Q fever

We used whole genome microarrays to study the whole blood transcriptional profile of Q fever patients (including patients with acute Q fever and patients with persistent C. burnetii infection) versus healthy donors. We identified 5588 genes differentially expressed between these two groups with a statistical significance (p ≤ 0.01) with 2189 up- and 3399 down-modulated genes (Fig 1). Because of the relatively large number of differentially expressed genes, we focused on genes that exhibit a >4 fold change between both groups and found 476 modulated genes (317 up- and 159 down-modulated genes. Non-supervised hierarchical clustering separated the transcriptional profiles of patients and controls (Fig 1A and 1B). Gene Ontology (GO) analysis of the transcriptional profiles underscored an association of Q fever-associated gene expression alterations with metabolic processes (18%), cardiovascular diseases (16%), cancers (10%), immune disorders (8%), infections (6%) and renal disorders (6%) (Fig 1C).

Fig 1

Transcritpional profile of patients with C. burnetii infection.

Microarray transcriptional profile using whole blood from patients with C. burnetii infection and healthy donors. Differential gene expression contrasting patients with Q fever and healthy controls was analysed by (A) hierarchical clustering with samples in row and genes in column. Gene expression was colored from green (down-regulated) to red (up-regulated). (B) Graphic representation of the principal component analysis indicating patients with Q fever (in blue) and controls (in red). (C) Differentially expressed genes in Q fever patients were subjected to gene ontology (GO) annotation for biological processes. The percentages of GO terms are classified by groups of diseases and concern all patients with Q fever.

Persistent C. burnetii infection induces transcriptional changes involved in non-Hodgkin lympoma

We next compared the blood transcriptome of patients with persistent C. burnetii infection with that of controls by microarray. Hierarchical clustering (Fig 2A) and principal component analysis (Fig 2B) clearly discriminated the gene expression profiles of patients and controls. 1687 genes were differentially expressed (p ≤0.01 and fold change ≥1.5). Among them, we found 739 up- and 948 down-modulated genes. Then, we used IPA software to inspect diseases or functions linked to persistent C. burnetii infection. The investigation of identifiers (ID) associated network functions (Fig 2C) revealed that “cancer and hematological disease” gene categories was at the top of this list (z-score = 46). The IPA “disease or function” ontology revealed 21 significant terms associated with lymphoma (Fig 2D). 76% of these terms was associated with NHL and included expression changes affecting 95 genes. Given the specific objective of this study, we decided to focus on this group of genes that we referred to as the C. burnetii-NHL (Cb-NHL) signature (Table 3).

Fig 2

Transcriptional profile of patients with C. burneti persistent infection links to lymphoma.

Microarray analyses were performed on whole blood from patients with persistent C. burnetii infection and healthy controls. Transcriptional profiles of patients with persistent C. burnetii infection were analyzed by (A) hierarchical clustering, (B) principal component analysis, (C) using IPA software to identify associated networks, and (D) according to the “disease or function” ontology of this software. (E) Comparison of the transcripitional effects of C. burnetii and S. aureus on genes linked to NHL or other cancers. The numbers at the top of each column indicates the number of genes incriminated.

Table 3

C. burnetii—Non-Hodgkin lymphoma signature.

Gene	Fold change	Gene description
ADAMTSL3	5.880324	ADAMTS like 3
AIF1	-2.9805992	Allograft inflammatory factor 1
ALOX5	-2.834	Arachidonate 5-lipoxygenase
ANKIB1	4.8102517	Ankyrin repeat and IBR domain containing 1
ANKRD36	2.2068887	Ankyrin repeat domain 36
ASAP2	-4.2549253	ArfGAP with SH3 domainAnkyrin repeat and PH domain 2
ATP10A	-4.458795	ATPase phospholipid transporting 10A (putative)
B2M	-3.521	Beta-2-microglobulin
BAG1	-2.899064	BCL2 associated athanogene 1
BCL2L11	2.6857266	BCL2 like 11
CASP10	-1.7338927	Caspase 10
CASP2	3.9026406	Caspase 2
CASP4	-2.6195037	Caspase 4
CASP8	3.4592438	Caspase 8
CCDC97	-2.008394	Coiled-coil domain containing 97
CD44	-2.1155155	CD44 molecule
CD63	-2.6959426	CD63 molecule
CEBPB	-1.9058212	CCAAT/enhancer binding protein beta
CNTRL	2.5811584	Centriolin
CREBBP	3.0448112	CREB binding protein
CTAGE1	3.4368377	Cutaneous T-cell lymphoma-associated antigen 1
CTAGE5	3.5890672	CTAGE family member 5, export factor
CTNNB1	4.6916003	Catenin beta 1
CUL9	-2.9435835	Cullin 9
DCLRE1C	1.8426613	DNA cross-link repair 1C
DNMT3B	-3.6707778	DNA methyltransferase 3 beta
DYNLL2	-2.3400176	Dynein light chain LC8-type 2
EGR1	-3.795406	Early growth response 1
FBXO11	2.1503875	F-box protein 11
FYN	3.284	FYN proto-oncogene, src family tyrosine kinase
G6PD	4.5535316	Glucose-6-phosphate dehydrogenase
GNB1	1.9694024	G protein subunit beta 1
GP1BA	-3.7912498	Glycoprotein Ib platelet alpha subunit
GRK3	3.358	G protein-coupled receptor kinase 3
HIC1	3.1705003	Hypermethylated in cancer 1
HLA-G	5.7384872	Major histocompatibility complex, class I, G
HUWE1	2.3448784	HECT, UBA and WWE domain containing 1
IKZF1	2.0111005	IKAROS family zinc finger 1
IL13RA1	-2.7759755	Interleukin 13 receptor subunit alpha 1
IL2RG	3.2423832	Interleukin 2 receptor subunit gamma
ITGAX	-2.8220086	Integrin subunit alpha X
JAK3	-3.003575	Janus kinase 3
KIAA1551	3.162272	KIAA1551
KLHDC3	2.2549937	Kelch domain containing 3
KLHL6	-3.3247852	Kelch like family member 6
KMT2C	2.083985	Lysine methyltransferase 2C
LIG3	-2.9250553	DNA ligase 3
MAP3K8	2.3423563	Mitogen-activated protein kinase kinase kinase 8
MAP4K4	1.5909171	Mitogen-activated protein kinase 4
MED1	-4.125631	Mediator complex subunit 1
MICALCL	-2.5829322	MICAL C-terminal like
MIR17HG	3.2123563	miR-17-92a-1 cluster host gene
MKNK2	-2.5849397	MAP kinase interacting serine/threonine kinase 2
NF1	2.524433	Neurofibromin 1
NFKBIA	-2.8016303	Nuclear factor of Kappa light polypeptide gene enhancer in B cells inhibitor alpha
NIPBL	3.1714787	NIPBL, cohesin loading factor
NOTCH2	-2.1726353	Notch 2
OGG1	2.5774076	8-oxoguanine DNA glycosylase
PBX1	-3.845063	PBX homeobox 1
PLCG1	1.9097688	Phospholipase C gamma 1
PLEKHG3	2.318101	Pleckstrin homology and RhoGEF domain containing G3
PPP6R3	3.2296784	Protein phosphatase 6 regulatory subunit 3
PTPN11	-3.1850703	Protein tyrosine phosphatase, non-receptor type 11
QARS	3.6468294	Glutaminyl-tRNA synthetase
RAPGEF1	3.936472	Rap guanine nucleotide exchange factor 1
RARA	-2.5248845	Retinoic acid receptor alpha
REL	1.7198911	REL proto-oncogene, NF-kB subunit
RELB	-2.859732	RELB proto-oncogene, NF-kB subunit
RPS2	-3.7340784	Ribosomal protein S2
RPSA	-3.64	Ribosomal protein SA
SERPINA1	-2.598	Serpin family A member 1
SLC16A7	2.801318	Solute carrier family 16 member 7
SMARCA4	4.9222918	SWI/SNF related actin dependent regulator of chromatin, subfamily a, member 4
SMARCB1	3.59208	SWI/SNF related actin dependent regulator of chromatin, subfamily b, member 1
SP100	2.505	SP100 nuclear antigen
SPECC1	-2.0302236	Sperm antigen with calponin homology and coiled-coil domains 1
TDRD1	3.4261622	Tudor domain containing 1
TET2	5.2559843	Tet methylcytosine dioxygenase 2
TGFB1	2.3384285	Transforming growth factor beta 1
TLR7	4.3670783	Toll like receptor 7
TNFRSF8	-1.8059382	TNF receptor superfamily member 8
TNIK	-2.174313	TRAF2 and NCK interacting kinase
TPM1	-4.584908	Tropomyosin 1 (alpha)
TRAF1	1.6773132	TNF receptor associated factor 1
TRIM38	2.384246	Tripartite motif containing 38
TRIP12	2.4232147	Thyroid hormone receptor interactor 12
TTC21B	2.1786764	Tetratricopeptide repeat domain 21B
TUBA1C	-3.2958088	Tubulin alpha 1c
TUBA3C	-2.4673393	Tubulin alpha 3c
TUBB3	-8.211237	Tubulin beta 3 class III
TUBB4B	-2.5735104	Tubulin beta 4B class IVb
UBE2F	-2.2094207	Ubiquitin conjugating enzyme E2 F (putative)
UIMC1	-2.096621	Ubiquitin interaction motif containing 1
WWC3	-1.7466848	WWC family member 3
XRCC5	-2.593	X-ray repair cross complementing 5

We then analyzed genes of Cb-NHL signature by means of GO annotation. The GO term analysis revealed a regulation of apoptotic process, a positive regulation of NF-κB signaling, a cellular response to mechanical stimulus, an innate immune response, cell adhesion, cell migration, extrinsic apoptotic signaling pathway and an inflammatory response and apoptotic process (S1 Fig).

To determine whether acute Q fever might be a risk factor for lymphoma, we used IPA software to explore categories of network functions in microarray data (S2A and S2B Fig). This analysis led to a more diffuse association with multiple disease categories, including multiple different types of cancer, not just lymphoma (gastrointestinal disease, organismal injury and abnormalities, neurological and dermatological disease, S2C and S2D Fig). Four genes overexpressed in Q fever were previously implicated in NHL, namely, BCL2, BCL7A, BCL9 and BCL11 (S2 Table). Altogether, these results confirm the suspicion that acute Q fever is much less associated with NHL than persistent C. burnetii infection.

Finally, to determine whether the correlation found with NHL was specific to C. burnetii infection, we compared the C. burnetii–related dataset to published [19] microarray analyses of patients with S. aureus and persistent endocarditis previously performed from whole blood (S3 Fig). In patients with S. aureus endocarditis, only 15% terms of “disease or function” ontology, including 6 genes, were associated with NHL (Fig 2E). This comparison confirms the existence of a specific link between persistent C. burnetii infection and lymphoma/NHL, which is not observed in case of S. aureus endocarditis.

The Cb-NHL signature in patients with different types of C. burnetii infection

We used qRT-PCR to investigate the Cb-NHL signature in a cohort of C. burnetii-infected patients including 11 with acute Q fever, 11 patients with persistent infection, 4 patients with lymphadenitis and 4 patients with C. burnetii infection and lymphoma and 10 with NHL without C. burnetii infection included in the NHL-control group (Table 2). Among the large number of genes (95) listed in the Cb-NHL signature, we selected 24 genes based on their functional links to cancer and lymphoma [1]. In addition, we included 9 genes specifically associated with lymphoma based on Flowers’ signature [1]. This type of PCA revealed particular Cb-NHL signature in Q fever patients (log2FC) normalized to 10 healthy donors (Fig 3A). In this PCA, the group of C. burnetii-infected patients with lymphoma, and that with C. burnetii lymphadenitis largely overlapped. In contrast, no such kind of superposition was observed when the Flowers’ signature is applied (S4 Fig). Both PCA and further analysis of Cb-NHL signature by unsupervised hierarchical clustering (Fig 3B) suggest that patients with persistent C. burnetii infection can be separated from patients with C. burnetii lymphadenitis or C. burnetii associated NHL. Again, these distinctions were not observed using Flowers’s signature (S4 Fig). Three genes of the Cb-NHL signature were strongly up-regulated (fold change between 100 and 1000) in patients with C. burnetii and lymphoma and significantly increased when compared to C. burnetii persistent infection and to NHL-control: MIR17HG (p = 0.0026), SP100 (p = 0.026) and REL (p = 0.0024) (Fig 3C). In the group of patients with C. burnetii infection and lymphoma, BAG1, OGG1 and NCOA1-v1 were also significantly up-regulated, as compared to other groups, namely, acute Q fever and persistent C. burnetii infection (Table 4).

Fig 3

qRT-PCR analysis of disease-associated alterations in gene expression within the Cb-NHL signature.

PBMC samples from patients with acute Q fever, persistent C. burnetii infection, C. burnetii associated NHL, C. burnetii lymphadenitis and NHL-controls were compared to healthy controls by using qRT-PCR. The expression of genes from the Cb-NHL signature were evaluated as fold change of investigated gene/ β-actin mRNA. (A) Principal component analysis reveals the overlap between the modulated genes (log2 fold change) of the three groups of patients with persistent C. burnetii infection (blue), C. burnetii associated NHL (purple) or C. burnetii lymphadenitis (green) and NHL-control (red). (B) Modulated genes from Cb-NHL (log2 fold change) were represented as a heatmap with samples in columns and genes in rows. Gene expression was colored from blue (down-regulated) to red (up-regulated). (C) Comparison of patients affected by persistent C. burnetii infection to patients with lymphoma and lymphadenitis led to the identification of MIR17HG, REL and SP100 as significantly up-regulated in patients with lymphoma and lymphadenitis. (D) In patients with lymphadenitis, BCL2 and ETS1 were significantly up-regulated.

Table 4

Quantitative-RT-PCR analysis.

Gene expression profile	Gene symbol	Fold change (mean)				p-value	Group of patients
		NHL-control	Persistent C. burnetii infection	C. burnetii and lymphadenitis	C. burnetii and lymphoma
							NHL-control	PersistentC. burnetii infection	C. burnetiilymphadenitis	C. burnetii and lymphoma
Oncogene, proliferation&Anti-apoptotic function	MIR17HG	2.089	5.058	302.3	567.0	<0.0001	Up	Up	Up	Up
	REL	6.304	1.593	118.5	63.60	0.0001	Up	Up	Up	Up
	BAG1	2.709	1.414	3.277	5.161	0.0008	Up	-	Up	Up
	OGG1	1.629	0.7404	3.418	6.006	0.0012	Up	Down	Up	Up
	SP100	0.1206	1.455	1134	306.6	<0.0001	Down	-	Up	Up
	NCOA1-v1	0.311	1.140	2.835	2.591	<0.0001	Down	-	Up	Up
	CREBBP	0.7572	2.494	0.5745	0.7397	0.0001	Down	Up	Down	Down
	ETS1	2.683	0.8918	3.457	2.477	0.0034	Up	Down	Up	Up
	BCL2	5.885	1.650	4.961	3.038	0.0044	Up	Up	Up	Up
Pro-apoptotic properties	CASP4	1.317	1.078	1.798	2.952	0.0001	-	-	Up	Up
	CASP8	2.916	1.237	641.4	328.3	0.0018	Up	-	Up	Up
	CASP2	1.732	2.032	127.8	130.1	0.0084	Up	Up	Up	Up
Immunity	TLR7	2.214	2.460	408.4	2179	<0.0001	Up	Up	Up	Up
	JAK3	9.395	0.4601	12.46	29.80	0.0001	Up	Down	Up	Up

P value is calculated as a fold change in comparison to the control group (healthy donors)

Up = fold change>1.5; down = fold change<1; -: fold change between 1 and 1.5

The specificity of the C. burnetii associated NHL transcriptomic signature was supported by gene expression analysis of the NHL control group, in which MIR17HG, SP100, REL, and ETS1 were not upmodulated, whereas BCL2 was significantly highly expressed (Fig 3C and 3D). Interestingly, antibiotic treatment with doxycycline and hydroxychloroquine (the standard medication to treat C. burnetii infection) did not affect the gene expression profiles found in patients (S5 Fig).

Some similarities were observed between the group of patients carrying C. burnetii and lymphoma, and those with C. burnetii lymphadenitis. Both groups exhibited similar gene expression profiles, and the aforementioned lymphoma-associated genes (MIR17HG, REL, BAG1, OGG1, SP100 and NCOA1-v1) were up-regulated (Fig 3C and S6 Fig). It is interesting to note that in addition to the SP100 and REL genes, the lymphadenitis group has significantly up-regulated ETS1 (p = 0.0046) and BCL2 genes (p = 0.021), both genes involved in the inhibition of apoptosis, tumorigenesis and cell proliferation, while these patient groups had no known occurrence of lymphoma. (Fig 3D). This latter result supports the idea that lymphadenitis might constitute a step towards lymphomagenesis.

Discussion

The present study identifies a specific transcriptomic signature of Cb-NHL among patients with persistent C. burnetii infection strengthening the link between Q fever and the development of NHL.

Three genes, REL, MIR17HG and SP100, constitute the core of this transcriptomic signature. This is intriguing because these genes have already been involved in cell survival, B-cell proliferation and tumorigenesis via anti-apoptotic process. MIR17HG is involved in cell survival proliferation and differentiation both in solid tumor and in lymphoma, and is associated with poor prognosis in Burkitt lymphoma and DLBCL [20]. SP100 stimulates the transcriptional activity of ETS1, a protooncogene ubiquitously expressed in the lymph nodes and in bone marrow, which is involved in apoptosis, invasion and metastasis (Table 4) [21]. REL is a proto-oncogene involved in survival and proliferation of B-lymphocytes and its mutation is associated with B-cell lymphoma [22]. REL-BCL11A fusion or gain of REL, has been previously linked to the transformation of FL to DLBCL, and this may be a genetic marker for tumor progression [22]. In the lymphadenitis group, as in the lymphoma group, two genes of the core of the transcriptomic C. burnetii-NHL signature, REL and SP100, were significantly upmodulated when compared to the control group.

Regarding the suggested genetic risk factors proposed by Flowers et al., only NCOA1, was significantly up-regulated in two groups: lymphoma and lymphadenitis, whereas BCL2 was significantly expressed in cases of C. burnetii lymphadenitis [1]. NCOA1, nuclear receptor co-activator 1, acts as a transctiptional coactivator for steroid and nuclear hormones receptors and may increase the induction of the T-cell receptor endocytosis [23]. BCL2 gene expression is well known to block apoptosis and promote B-cell lymphoma and is even the target of antitumoral therapy [24]. Its high expression observed in case of C. burnetii lymphadenitis supports lymphadenitis as a critical step in lymphomagenesis, whereas its high expression in case of NHL control group is concordant with previous reports [25]. While Flowers et al. suggested that a preexisting genetic risk factor could play a role in the development of NHL in patients with C. burnetii infection, our results rather indicate that C. burnetii infection triggers the expression of genes implicated in anti apoptotic and proliferative mechanisms.

Regarding the Cb-NHL signature and using pubmed data base, we identified 42 genes associated with tumor development in patients carrying viral or bacterial pathogens. REL is among the common genes modulated in infection by EBV, HTLV, C. jejuni, H. pylori, B. burgdorferi or C. psittaci (Fig 4A). During EBV infection, REL and STAT3 have a potent effect on cell growth and in lymphomagenesis [26]. On the contrary, REL deficiency leads to decreased proliferation, decreased survival of EBV transformed cells and could even lead to their necrotic death [27]. In HTLV infection, REL proteins are involved in the canonical NF-κB pathway, which induces the transcription of anti-apoptotic BCL2A1 and BCL2L1 [28]. In Borrelia burgdorferi infection, REL is responsible for the ppGpp synthesis that facilitates the bacterium growth and virulence [29]. Nevertheless, the precise role of REL in infections by C. jejuni and C. psittacii has not been yet clarified.

Fig 4

(A) Link between genes contained in Cb-NHL signature and infection by cancer-associated microorganisms. Venn diagram showing overlap between genes contained in the Cb-NHL signature that are also deregulated in non-Hodgkin lymphoma associated with Epstein-Barr virus (EBV), human T lymphocyte virus (HTLV), Campylobacter jejuni, Helicobacter pylori, Borrelia burgdorferi or Chlamydia psittaci. (B) Pathophysiological mechanism. Fom persistent C. burnetii infection (endocarditis, vascular infection), to C. burnetii lymphadenitis and lymphoma. Peripheral blood monocyte cells genomic transcriptional profiles indicated an overexpression of BCL2 and ETS1 mRNAs in cases of C. burnetii lymphadenitis and an overexpression of MIR17HG, REL and SP100 genes in cases of C. burnetii associated lymphoma. All these gene are related to the inhibition of apoptosis and/or cellular proliferation.

There is a vast literature on the effects of EBV and H. pylori on gene expression. Epstein barr virus latency gene products drive viral persistence in memory B-cells and malignancy by controling cell growth, division, differentiation and apoptosis [30,31]. Latent transcipts, such as Epstein-Barr virus nuclear antigen 1 (EBNAs) and latent membrane proteins (LMPs) were detected in EBV-associated B-cell lymphoma [30,31]. EBV-positive and EBV-negative BL could even be differentiated by their apoptosis-related gene expression profile [31]. Interestingly, the transcriptomic profile of EBV-positive benign lymphadenophathy was described as closer to BL than to post-transplant lymphoproliferative disease [32], supporting EBV-positive lymphadenopathy as a critical step leading to lymphomagenesis, similar to what we suggest here for C. burnetii infection [32].

C. burnetii modulates the apoptotic pathway through Beclin 1/BCL2 to establish succesfull infection of the host cell [33]. Here, we report that BCL2 was significantly up-regulated in the blood of patients with C. burnetii lymphadenitis. The activity of PKB/AKT and MAPK1/ERK2 or MAPK3/ERK1 seems to be required for the apoptosis-suppressive effect of C. burnetii infection [34]. More recently, the anti-apoptotic protein myeloid cell leukemia-1 (MCL1) has been found to be increased in neutrophils infected by avirulent C. burnetii and this effect was linked to the inhibition of caspase-3 cleavage and the activation of the MAPK survival pathway [34]. Further in vitro experimentation is necessary to unravel the mechanims through which infection of certain leukocyte subtypes (such as monocytes/macrophages) may be coupled with the inhibition of apoptosis in B cells.

In this study, the transcriptomic profile was obtained from the whole blood. In blood, BCL2 gene expression was proved to be significantly up-regulated in patients with chronic lymphoid leukemia [35]. In patients with NHL, a concordant expression between bone marrow and peripheral blood BCL2/JH expression was reported [25]. The identification of the high-level expression of BCL2 in the blood of patients with lymphadenitis might constitute a biomarker of a prelymphomatous grade requiring close monitoring.

The up-regulation of genes coding for caspases (CASP2, CASP4, CASP8) that we have found is not in clear contradiction with the anti-apoptotic mechanism involved in C. burnetii infection insofar, as the regulation of caspase activation essentially occurs at the post-translational level [36].

In conclusion, we have identified a transcriptomic signature from PBMC of patients with C. burnetii-associated lymphoma that consists of the overexpression of several genes including MIR17HG, REL and SP100. Patients with C. burnetii lymphadenitis presented high level of BCL2 mRNA in blood, shared a similar transcriptomic signature with patients infected with C. burnetii and were diagnosed with lymphoma. This argues in favor of the hypothesis that C. burnetii lymphadenitis may facilitate subsequent lymphomagenesis leading to NHL.

Cb-NHL signature by using the GO annotation for biological process.

GO terms related to regulation of apoptotic process, positive regulation of NF-κB signaling, cellular response to mechanical stimulus, innate immune response, cell adhesion, cell migration, extrinsic apoptotic signaling pathway, inflammatory response and apoptotic process.

(TIFF)

Click here for additional data file.

Gene expression signature between patients with acute Q fever (n = 8) and patients with C. burnetii persistent infection (n = 8).

Differential gene expression between acute Q fever and persistent C. burnetii infection was analysed by hierarchical clustering (A) and bioinformatic plots results (B). In panel B the bioinformatic plots results according to 3 axis shows a segregation gene modulation between patients with acute Q fever and patients with C. burnetii persistent infection as compared with the control group. Differentially expressed genes from Cb-NHL signature were subjected to GO annotation to identify the biological process. The histogram graphs (C) and (D) show the repartition of dieases associated with acute C. burnetii infection.

(TIFF)

Click here for additional data file.

Microarray investigation between C. burnetii and S. aureus induced endocarditis.

Differential gene expression between persistent C. burnetii infection and S. aureus endocarditis was analysed by (A) hierarchical clustering and (B) bioinformatic plots results. In panel B, the results of the bioinformatics plots along 3 axes show a modulation of the segregation gene between patients with persistent C. burnetii infection and S. aureus endocarditis compared to the control group.

(TIFF)

Click here for additional data file.

q-RTPCR analysis of genes modulation from Flowers’ signature.

PBMC samples of patients with acute Q fever, C. burnetii persistent infection, C. burnetii associated NHL, C. burnetii lymphadenitis and NHL-control, were investigated by q-RTPCR and normalized to controls for the expression of genes from Flowers’ signature. (A) Generated by ClustVis software, principal composant analysis reveals the overlaping of the modulated genes (log2FC) from the three groups C. burnetii persistent infection (blue), C. burnetii and lymphoma (purple) and C. burnetii lymphadenitis (green) and NHL-control (red). (B) Modulated gene from Flowers’signature were represented as heatmap with samples in column and gene in row. Gene expression was colored from blue (down-modulated) to red (up-modulated).

(TIFF)

Click here for additional data file.

Investigation of the NHL genes expression according to treatment.

Among the large number of genes (95) listed in the Cb-NHL signature, we selected 24 genes based on their functional links to cancer and lymphoma. In addition, we included 9 genes specifically associated with lymphoma based on Flowers’ signature. A total of 33 genes were analysed according to antibiotic treatment with doxycycline and hydroxyplaquenil (the standard medication to treat C. burnetii infection). 0. Samples from patients never treated (red). 1. Samples from patients with ongoing doxycycline and hydroxyplaquenil treatment (blue). 2. Samples from patients before receiving doxycycline and hydroxyplaquenil treatment (green). 3. Samples from patients after doxycycline and hydroxyplaquenil treatment. 4. Samples from patients after doxycycline treatment (purple). 5. Samples from patients with ongoing doxycycline treatment (brown). Using q-RTPCR, we investigated the Flowers’ signature (log2FC) which (A) principal composant analysis and (B) hierarchical clustering with samples from persistent Q fever (red), lymphoma (green) or remission (blue) periods from index case. Up- and down-modulated genes were represented in red and blue respectively. The analysis shows that gene expression was not affected by the antibiotic treatment received from patients.

(TIFF)

Click here for additional data file.

Gene expression analysis.

PBMC samples of patients with C. burnetii persistent infection, C. burnetii associated-NHL, C. burnetii lymphadenitis and NHL-control were investigated by q-RTPCR, normalized to controls to investigate the expression of genes from Cb-NHL and Flowers’ signatures: (A) OGG1, (B) BAG1 and (C) NCOA1-v1.

(TIFF)

Click here for additional data file.

Specific primers used to performed qRT-PCR.

(PDF)

Click here for additional data file.

List of genes modulated in acute Q fever.

(PDF)

Click here for additional data file.