Microbial diversity investigation using 16S metagenomics in Tunisian patients with systemic lupus erythematosus

Abstract

Systemic lupus erythematosus (SLE) is a multisystemic autoimmune disease associated with significant morbidity and mortality. It is characterized by a loss of self-immune tolerance and autoantibody production, leading to multiple organ damage. Emerging investigations have confirmed the role of gut microbiota dysbiosis in patients with SLE, although the underlying mechanisms remain unclear to date. In this study, we aim to investigate the bacterial profile of SLE including phylum/class/genus relative abundance and diversity, to compare them with healthy controls and to study the correlation of relative abundance of different patterns with clinical/biological parameters. In this case–control study, the bacterial profile was investigated in 7 SLE patients and 7 healthy controls using 16S metagenomics clustering. The present study reported a low abundance of the class Bacilli (0.58% in SLE vs 1.26% in the controls), the genus Lactobacillus (0.43% vs 0.74%), as well as a higher abundance of the genera Gammaproteobacteria (2.37% vs 0.77%) and Escherichia–Shigella (2.04% vs 0.51%) in SLE samples compared to the controls (p < 0.05). We also found an association between the class Betaproteobacteria (4.42% vs 1.57%) and the genus Faecalibacterium (11.34% vs 3.35%) and renal manifestations (p < 0.05). The phylum Actinobacteria (0.21% vs 3.8%, p = 0.036) and the genus Bifidobacterium levels were lower in active SLE compared to the healthy controls. This study is the first report on the gut microbiota of SLE and the first case–control study in Tunisia and North Africa. We obtained a particular profile of bacterial gut microbiota for the SLE group. We found a specific clustering when compared to the healthy controls.

Resumen

El lupus eritematoso sistémico (LES) es una enfermedad autoinmune multisistémica que ocasiona elevada mortalidad y morbilidad. Se caracteriza por la pérdida de la tolerancia autoinmune y la producción de autoanticuerpos, que dañan múltiples órganos. En este trabajo, investigamos el perfil bacteriano del LES determinando las abundancias relativas a nivel de filo/clase/género y la diversidad en 7 pacientes con LES y 7 controles sanos, empleando agrupamiento metagenómico del gen 16S. Además, estudiamos la correlación de la abundancia relativa de diferentes clusters con parámetros clínicos/biológicos. En los pacientes con LES, encontramos baja abundancia relativa de la clase Bacilli (0,58% vs. 1,26% en controles) y del género Lactobacillus (0,43% vs. 0,74%), así como una mayor abundancia relativa de la clase Gammaproteobacteria (2,37% vs. 0,77%) y de los géneros Escherichia-Shigella (2,04% vs. 0,51%), en comparación con los controles (p < 0,05). También encontramos en el LES asociación de la clase Betaproteobacteria (4,42% vs. 1,57%) y el género Faecalibacterium (11,34% vs. 3,35%) con la manifestación renal (p < 0,05). Los niveles del filo Actinobacteria y del género Bifidobacterium disminuyeron en el LES activo frente a los controles sanos (0,21% vs. 3,8% y 0,15% vs. 3,7%, p = 0.036, ambos). Este es el primer informe sobre la microbiota intestinal en personas con LES y el primer estudio de casos y controles en Túnez y el norte de África. Observamos un perfil particular de microbiota intestinal bacteriana en el grupo con LES y detectamos un agrupamiento específico al compararlo con el perfil de los controles sanos.

Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by immune system deregulation, leading to the over-activation of immune T and B cells, and the production of antibodies against nuclear and cytoplasmic antigens. SLE manifestations are associated with multiple autoantibodies, including anti-nuclear antibodies (ANA) and anti-double-stranded DNA (anti-dsDNA) antibodies, causing immune complex deposition leading to tissue lesions in multiple organs (skin, kidneys, joints, muscles, heart, among others)^44,48. Lower levels of serum complement components C3 and C4 were also described.²⁴ Anti-Sm, anti-SSA and anti-SSB antibodies are specific to Sjögren's disease, which can be associated with SLE in some patients.

SLE mainly affects individuals between 15 and 40 years old and especially young women. There are no studies concerning the prevalence of SLE in North Africa, except in Tunisia. In Tunisia, the hospital prevalence varies from one study to another and is estimated between 0.1 and 0.3%^19,22,29. Patients suffer from variable clinical manifestations, ranging from mild joint and skin involvement to life-threatening features. They mainly suffer from renal, rheumatological and cutaneous mucosal features. The clinical complexity of SLE and the lack of pathological features or tests pose a diagnostic challenge for the clinician.

The etiology of SLE is still unclear. Several recent studies have suggested that the modification of the gut microbial composition may be related to SLE manifestations^{14–16,28,36,41,47}. Increased evidence indicates that intestinal dysbiosis is related to several autoimmune diseases such as type 1 diabetes^3,12, multiple sclerosis^9,18,34, rheumatoid arthritis⁴², Crohn's disease/ulcerative colitis^20,45 and obesity².

The gut microbiota is a complex ecosystem that includes all unicellular organisms hosted in the gastrointestinal tract (GI), mainly bacteria but also viruses, fungi and archaea⁴³. It is composed of 10¹⁴ microorganisms and includes about 2000 different bacterial species²⁶. The composition of the intestinal microbiota is generally inferred from fecal samples and is characterized by the presence of five phyla: Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria and Fusobacteria. Firmicutes and Bacteroidetes represent the main dominant phyla and are faced with modifications due to factors such as environment, age, genetics, diet and infection¹³.

The precise mechanisms underlying gut microbiota dysbiosis in SLE remain elusive to date. Potential factors include molecular mimicry, compromised intestinal barrier function disrupting gut homeostasis, thereby facilitating the translocation of gut microbiota and their products (such as LPS from Gram-positive bacteria) into the systemic circulation. This process can trigger inflammation and tissue damage. Furthermore, it may involve the leakage of pathogens, toxins, and bacteria from the gut lumen into other organs, a phenomenon known as “leaky gut.” Other potential contributors include bacterial biofilms, specific intestinal pathogen infections, gender disparities related to sex hormones, autophagy of intestinal epithelial cells, extracellular vesicles, and microRNAs³⁷.

Dysbiosis in SLE involves Treg-Th17 trans-differentiation and lymphocyte over-activation occurring within the gut mucosa. This process initiates immune responses against commensal microbiota, where antigens from translocated bacteria may resemble host structures, triggering cross-reactivity that leads to autoantibody production and organ damage in SLE patients – a phenomenon known as molecular mimicry^27,40.

Recent investigations have increasingly highlighted the significance of gut microbiota dysbiosis in SLE, underscored by multiple studies^{14–16,28,46,51}. These studies predominantly focused on American and Asian populations, revealing geographic variations in the composition of fecal microbial communities.

The aim of our study was to identify the gut bacterial profile of Tunisian subjects suffering from SLE, by evaluating the relative abundance and diversity of different clusters, to compare them with the healthy controls and to study the correlation between the relative abundance of different clusters with clinical/biological parameters. The investigation of the gut microbiota correlation with clinical/biological parameters of SLE may help clinicians to improve diagnosis and prescribe the right treatment of lupus and its complications. This study is particularly original because it is the first to analyze gut microbiota in Tunisian SLE patients as well as across Tunisia, North Africa, and the MENA region.

Materials and methods

Ethics statement

All subjects included in this experiment provided their written informed consent, and the protocol of this study was approved by the Ethics Committee of the Military Hospital of Tunis.

Study subjects

Seven women with SLE were recruited from the Internal Medicine Department of the Military Hospital of Tunis between 2020 and 2021 (mean age 48 ± 13.8 years). All of them met at least four of the American College of Rheumatology criteria for the diagnosis of SLE⁵. Patients were excluded from further study based on the following criteria: (1) Pregnancy or breast-feeding; (2) Current inflammatory or autoimmune disease; (3) Current malignancy other than skin; (4) Serious infection within 3 months with hospitalization; (5) Antibiotic treatment within the preceding three months.

The activity levels have been categorized based on SLE activity index (SLEDAI) scores: no activity (SLEDAI = 0), remission (SLEDAI = 1–5), moderate to very high activity (6–20). A score greater than 5 often leads to initiating or modifying therapy in over half of the cases (Aringer 2019). SLEDAI values were obtained from only six patients, ranging from 3 to 22, indicating remission or active disease (Table 1).

Table 1.

Demographic and clinical parameters in the two groups.

	Patients	Healthy controls	p-Value
Population	SLE patients	Healthy controls
Place	Military Hospital of Tunis
Period	2020–2021
Age (years)*	48 ± 13.8	43.2 ± 12.2	0.51
BMI (kg/m^2)*	25.3 ± 3.8	23.8 ± 2.4	0.40
SLEDAI*	11 ± 7.1	–
Disease duration (years)*	15.4 ± 9.1	–
Clinical manifestations, n (%)		–
Rheumatological	3 (42.8)
Renal	3 (42.8)
Others**	3 (42.8)
Treatment, n (%)		–
Immunosuppressive	6 (85.7)
Immunomodulator	5 (71.4)
Immunotherapy	1 (14.3)
Drug discontinuation	1 (14.3)
Dietary intake, n (%)
Fiber diet***	67.8%	75%	>0.05
Tea consumer	6 (85.7)	3 (42.8)	>0.05
Dairy products	4 (57.1)	7 (100)	>0.05

SLEDAI: SLE disease activity index, SLE: systemic lupus erythematous, BMI: bone mass index.

*

Mean ± standard deviation.

**

Others: cutaneous/mucosal, neurologic/neuropsychiatric and gastrointestinal symptoms with 14.3% for each manifestation.

***

Consumer proportion of high-fiber foods (fruits, vegetables, cereals and/or dried vegetables).

SLEDAI is a validated model of disease activity in lupus. Four of 7 patients had active disease at the time of sampling (SLEDAI > 5) and only 2 were in remission (SLEDAI ≤ 5).

The control group was composed of 7 healthy women recruited from the Military Hospital of Tunis (mean age 43.2 ± 12.2 years); they were age and sex-matched with the SLE patients, with no history of inflammatory/autoimmune diseases, nor any gastrointestinal tract infection, and had not received antibiotics within the preceding 3 months. Clinical laboratory tests were conducted as part of routine care. Patients and healthy controls were requested to provide fecal samples. Samples were conserved at −80 °C at the Military Hospital of Tunis and then transferred to the African Biotechnology Society (ABS) laboratory for the next steps.

• (i)

  Anthropometric measurements. Body mass index (BMI) was calculated using the following formula: weight/(height)² (in kg/m²). Individuals with a BMI higher than 30 were excluded from the study given that the gut microbiota is influenced by obesity statuses². Mean BMI value in SLE patients was 25.3 ± 3.8 and 23.8 ± 2.4 in the healthy controls (Table 1).

• (ii)

  Nutritional evaluation. Participants were surveyed about their food consumption and evaluated as “regular consumers” or “medium frequency consumers” by the nutritionist. The food survey focused on beverage consumption, high-fiber foods such as fruits, vegetables, cereals, and dried fruits. Participants were asked about their intake of meats, oils and fat foods.

• (iii)

  Lifestyle-related factors. Various lifestyle factors of the subjects were recorded. These factors encompassed smoking habits, physical activity, alcohol consumption and the use of supplements such as vitamins and minerals within the last month.

Immunologic assays

C3 and C4 levels and the presence or absence of autoantibodies including anti-nuclear antibodies (ANA), anti-dsDNA, anti-SSA, anti-SSB, anti-histone, anti-Sm, and anti-nucleosome, were recorded in SLE patients (Table 2). Normal values of C3 and C4 complement components were between 0.97–1.576 and 0.162–0.445, respectively.

Table 2.

Presence of autoantibodies, C3/C4 and treatment in SLE patients.

Patients	Autoantibodies	C3	C4	Treatment	Immunosuppressive therapy
P1	P1. ANA (+), anti-dsDNA (−), anti-histone (−), anti-nucleosome (−)	1.550	0.161	Hydroxychloroquine, methotrexate, steroids	+
P2	P2. ANA (−)	0.745	0.217	Steroids, rituximab	+
P3	P3. ANA (+)	0.172	0.064	Hydroxychloroquine + steroids + MMF	+
P4	P4. ANA (+), anti-dsDNA (+), anti-histone (+), anti-nucelosome (+), anti-SSA (−), anti-SSB (−), anti-Sm (−)	–	–	Drug discontinuation	−
P5	ANA (+), anti-histone (−), anti-nucleosome (−), anti-SSA (+), anti-SSB (+), anti-Sm (−)	–	–	Hydroxychloroquine + steroids + MMF	+
P6	ANA (+), anti-histone (+), anti-nucleosome (+), anti-SSA (+), anti-SSB (+), anti-Sm (+)	1.550	0.217	Hydroxychloroquine + steroids + methotrexate	+
P7	ANA (+), anti-dsDNA (+), anti-histone (+), anti-nucleosome (+), anti-SSA (+), anti-SSB (−), anti-Sm (−)	1.120	0.110	Hydroxychloroquine + steroids	+

(+) positive, (−) negative, ANA: anti-nuclear antibody, MMF: mycophenolate mofetil, +: immunosuppressive therapy, values in bold: diminished levels of C3 or C4 complement component, anti-SSA: anti–Sjögren's-syndrome-related antigen A autoantibodies, anti-SSB: anti- Sjögren syndrome type B antigen autoantibodies.

Bacterial DNA extraction

DNA extraction and sequencing steps were performed in the ABS laboratory under sterilized and adequate conditions. A quantity of 180–220 mg of fecal samples were collected to perform bacterial DNA extraction from the study subjects using the QiAmp Fast DNA Stool Mini Kit (Qiagen) according to the manufacturer's guidelines. QiaXpert Spectrophotometer (Qiagen) was used to quantify DNA and 260/230 and 260/280 ratios were measured to check purity. DNA solutions were stored at −20 °C until being analyzed (https://support.illumina.com/documents/documentation/chemistry_documentation/16s/16s-metagenomic-library-prep-guide-15044223-b.pdf, downloaded on 21-07-2024).

Preparation of the QIAseq 16S V3-V4 region panel PCR reaction

In this step, we mixed 1 μl of extracted bacterial DNA, 2.5 μl of UCP master mix, 1 μl of region panel and UCP PCR water for a volume of 10 μl per sample. The setup of QIAseq 16S Region PCR Reaction included a hold step at 95 °C for 2 min, 12 cycles with a 3-step cycling consisting of denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s and extension at 72 °C for 2 min, and a final step at 72 °C for 7 min. The product was maintained at 4 °C. Then, we purified the amplified fragments using a magnetic reaction.

Preparation of the QIAseq 16S V3-V4 region panel sample index PCR reaction

To 32.5 μl of the QIAseq 16S region panel PCR product, we added 12.5 μl of UCP master mix, 2.5 μl of each primer (4 μM) and UCP PCR water to complete a volume of 50 μl per sample. The setup of QIAseq 16S Region PCR Reaction included a hold step at 95 °C for 2 min, 19 cycles with a 3-step cycling consisting of denaturation at 95 °C for 30 s, annealing at 60 °C for 30 s and extension at 72 °C for 2 min, and a final step at 72 °C for 7 min. The product was maintained at 4 °C. We performed the same protocol of purification as the previous step and we stored the completed QIAseq 16S Region Panel Sequencing library at −20 °C.

Illumina MiSeq sequencing of 16S rRNA gene-based amplicons

To prepare for MiSeq sequencing, pooled libraries were denatured with NaOH, and then heat-denatured following a dilution in the hybridization buffer. For each sequencing run, at least 5% PhiX was included as an internal control. Sequencing was conducted on Illumina MiSeq technology using paired 300-bp reads and MiSeq v3 reagents. The overlapping ends of each read produce full-length reads of the V3 and V4 regions.

Bioinformatic and statistical analysis

The bioinformatic and statistical analysis was performed at the Pasteur Institute of Tunis and Military Hospital of Tunis. The FASTQ files were analyzed using mothur v.1.48.0 following the MiSeq Standard Operating Process pipeline (https://mothur.org/wiki/miseq_sop/). Operational Taxonomic Units (OTUs) were calculated at phylum, class and genus levels. Only taxa with relative abundance higher than 1% were considered for statistical analysis. Contigs were checked to remove sequences that did not align with the SILVA V4 database³⁸. For statistical analysis and data visualization, several R packages including phyloseq, ggplot2 and dplyr were used. Alpha-diversity estimators, corresponding to variables describing diversity within the samples (“Observed”, “Chao1”, “ACE”, “Shannon”, “Simpson”, “InvSimpson” diversity indexes), were computed. An ordination technique based on non-metric multidimensional scaling (NMDS) was used to explore microbiome data in the two groups (cases and controls). Beta-diversity analysis, including Jaccard and Bray–Curtis distance metrics^7,38, was then performed to check whether bacterial communities in the two groups were significantly different from each other. Statistical analysis was performed using SPSS 22.0 software. The normality of quantitative variables was evaluated and adequate parametric or non-parametric tests were used. A T-test was used to compare independent variables with a normal distribution and the Mann–Whitney test was used to compare independent variables to non-normally distributed ones. The chi-square test was performed to investigate the relationship between clinical manifestations and the relative abundance of different bacterial clusters. p-Values <0.05 were considered statistically significant.

Results

The demographic parameters of the participants and their clinical data such as age, BMI, disease duration, SLEDAI, clinical manifestations, treatment and dietary intake are shown in Table 1. The presence of autoantibodies, C3/C4 levels and treatment data of each patient are detailed in Table 2.

There were no significant differences between the patients and the healthy controls in terms of age, BMI, smoking history, and alcohol or dietary intake; none of them were smokers or alcohol consumers. The group of SLE patients included individuals with a wide variety of symptoms, mainly rheumatological (42.8%) or renal (42.8%). Of seven patients, five (71.4%) were treated with hydroxychloroquine as an immunomodulator, five received immunosuppressives such as steroids, methotrexate, or mycophenolate mofetil (MMF), one (14.3%) was treated with immunotherapy (Rituzimab) and one received no treatment (Tables 1 and 2).

Furthermore, the dietary intake proportions of fiber compounds, tea consumption, dairy products, proteins and lipids were recorded. There were no significant differences between the patients and the healthy controls (Table 1).

Six patients (85.7%) were positive for anti-nuclear antibodies (ANA) while only two (28.5%) patients were positive for anti-dsDNA antibodies. Two patients exhibited a decrease in C3 levels whereas three showed decreased levels of the C4 complement component (Table 2).

The metagenomics analysis of the samples showed a significant variation in the number of reads, ranging from 65 reads to 73 354 reads across the samples.

Gut microbiota profile in SLE patients and comparison with healthy controls:

The gut bacterial distribution of SLE patients and controls is illustrated in Figure 1. Healthy controls were numbered from 049 to 055, while SLE patients from 056 to 062 SLE patients. Bacteroidetes and Firmicutes were found to be the dominant phyla in all samples from both groups. These results are in line with previous studies, indicating that a healthy human gut microbiota is primarily composed of Bacteroidetes and Firmicutes⁴⁶. The Firmicutes/Bacteroidetes ratio was 1.29 in SLE cases and 0.81 in the healthy controls. Moreover, Bacteroidetes and Firmicutes, Proteobacteria and Actinobacteria were present with different abundance levels across samples. However, no significant differences were observed at the phylum level. At the class level, we observed that Bacilli, belonging to the Firmicutes phylum, were less abundant in the SLE cases than in the healthy controls (0.58% vs 1.26%, p < 0.05). Furthermore, Gammaproteobacteria, belonging to the Proteobacteria phylum (2.37% vs 0.77%, p < 0.05) seemed to be more abundant in SLE patients compared to the controls. Finally, abundance at the genus level confirmed the microbial distribution observed at the class level, showing an increase in the abundance of Enterobacteriaceae, especially the genera Escherichia and Shigella (2.04% vs 0.51%, p < 0.05) in SLE samples compared to the controls. A decrease in the abundance of Lactobacillaceae, particularly Lactobacillus (0.43% vs 0.74%, p < 0.05) was observed within SLE samples compared to controls.

Figure 1.

Relative abundance of the gut microbiota identified in our study within each control and SLE sample (049–055 numbers represent controls and 056–062 represent SLE patients). Relative abundance bar plots represent the bacterial composition in the human gut microbiota at the phylum, class, and genus levels, identified in each sample. Each legend box shows the top classified bacterial taxa across the samples.

We estimated alpha diversity using the Observed_OTUs, Chao1, ACE, Shannon, Simpson, and InvSimpson alpha-diversity indices to compare the richness and evenness of the human gut microbiota between the two studied groups. The results of the alpha-diversity measures are highlighted in Figure 2. We observed that there was a difference in diversity between the case and the control samples.

Figure 2.

Comparison of average alpha-diversity scores in our study between cases and controls. Box plots show the average alpha-diversity scores within the two groups, which were measured by using Observed_OTUs, Chao 1, ACE, Shannon, Simpson, and InvSimpson.

Figure 3 displays the abundances of the top 100 OTUs for the case and control samples, ranked on the abundance for each group. The number of detected OTUs is generally significantly higher for the case samples compared to the control samples. In addition, it seemed that the evenness (Shannon) in the control and case samples were similar, indicating that the OTUs may be different in terms of abundance.

Figure 3.

The abundances of the top 100 OTUs for the cases and controls communities in our study were ranked by abundance for each group.

The beta-diversity analysis between the two groups of samples showed that the Jaccard distance did not provide a clear separation between cases and controls. Conversely, the Bray–Curtis distance provided a slightly better separation, showing that the two groups shared their abundant taxa but differed in their rare taxa. This suggests that although the groups had taxa in common, their specific taxa were different (Fig. 4).

Figure 4.

Beta-diversity analysis between the two groups (cases and controls) in our study based on Jaccard and Bray–Curtis distance calculation.

Relationship between the relative abundance of bacterial clusters and clinical/biological parameters

No differences or correlations with the clinical features (SLEDAI and clinical manifestations), complement components or any antibodies were detected for phylum clustering (p > 0.05) [data not shown].

Sample classification revealed some specific clustering patterns in the patient samples, such as a significant difference between the relative abundance of class Betaproteobacteria (phylum Proteobacteria) (4.42% vs 1.57%) and genus Faecalibacterium (phylum Firmicutes, class Clostridia, family Ruminococcaceae) (11.34% vs 3.35%) and renal manifestations (p = 0.042, p = 0.048 respectively).

A negative correlation was detected between genus Clostridium_XVIII (phylum Firmicutes, class Erysipelotrichia, family Erysipelotrichaceae) and SLEDAI scores (r = −0.812, p = 0.050), whereas a positive correlation was found between genus Odoribacter genus (phylum Bacteroidetes, class Bacteroidia, family Porphyromonadaceae) and disease duration (r = 0.924, p = 0.003). The genus Prevotella (phylum Bacteroidetes, class Bacteroidia, family Prevotellaceae) seems to be negatively correlated with BMI index (r = −0.757, p = 0.014).

Moreover, we checked the relationship of gut microbiota abundance and diversity with the disease activity. All six patients with SLEDAI values were treated with immunosuppressive drugs.

We found lower levels of phylum/class Actinobacteria (0.21% vs 3.8%, p = 0.036) and genus Bifidobacterium (phylum Actinobacteria class Actinobacteria) (0.15% vs 3.7%, p = 0.036) in SLE patients with active disease compared to the healthy controls.

Discussion

SLE manifestations are linked to multiple autoantibodies, which facilitate immune complex formation and deposition, as well as other immune processes⁴⁴. This intricate clinical presentation and pathogenesis make SLE a challenging disease to understand and define.

This study is the first report on the gut microbiota of SLE and first case–control study in Tunisia and North Africa. In the present study we performed a 16S metagenomics analysis to investigate the gut microbiota of Tunisian subjects, with and without SLE, through the evaluation of the proportions of Firmicutes, Bacteroidetes and other bacterial species. Indeed, the healthy controls recruited in the present study had no gastrointestinal tract disorders and had not received antibiotics within 3 months prior to sample collection. All patients were women, six of the seven individuals recruited were in the active and remission form of the disease at the time of sampling (SLEDAI score ranging from 3 to 22) and had received immunotherapy, immunosuppressive or immunomodulatory treatment in the previous months. These treatments have an impact on the physiology of SLE.

We found an altered microbiome in human SLE patients compared to the healthy controls. These results are in line with previous studies, indicating that a healthy human gut microbiota is primarily composed of Bacteroidetes and Firmicutes⁴⁶. The present study reported a low abundance of Bacilli (phylum Firmicutes), Lactobacillaceae, particularly genus Lactobacillus as well as a higher abundance of Gammaproteobacteria, Enterobacteriaceae, especially Escherichia in SLE samples compared to the controls. The beta-diversity investigation based on the Bray–Curtis distance revealed that the two groups share their abundant taxa but not the rare ones, which are different.

Bacillus and Lactobacillus strains are enteric bacteria and are widely used as probiotics¹¹. Bacillus strains promote the growth of Lactobacillus to balance the microbiota and achieve therapeutic purposes³².

Proteobacteria are Gram-negative bacteria that include many pathogens such as Escherichia coli, Salmonella, Vibrio, Helicobacter, and Shigella. The gut microbiota dominated by the genera Escherichia/Shigella is associated with low short-chain fatty acid concentrations and an increase in metabolic pathways, creating a pathogenic and inflammatory environment in these patients⁶. The lipopolysaccharides of Gammaproteobacteria are more immunogenic compared to those of some commensal Gram-negative bacteria, such as Bacteroides species. Enterobacteriaceae has been previously linked to intestinal inflammation¹⁷, reflecting the common inflammatory response observed in SLE patients.

Similar results were published by He et al. (2020) in a study of 21 Chinese women with SLE and 10 healthy volunteers. They reported an increase in Proteobacteria (class Gammaproteobacteria, order Enterobacteriales, and family Enterobacteriaceae) in SLE patients and Escherichia/Shigella¹⁴.

Hevia et al. studied 20 Spanish women with SLE of Caucasian origin and 20 healthy controls. They found a significantly lower Firmicutes/Bacteroidetes ratio in SLE patients (median ratio, 1.97) compared to healthy controls. The normalized abundances of classes Bacteroidia (phylum Bacteroidetes), Clostridia (phylum Firmicutes), as well as orders Bacteroidales and Clostridiales, are significantly different between the SLE patients and healthy controls. Lachnospiraceae and Ruminococcaceae (family level) were positively associated with the healthy controls¹⁶.

He et al. (2016) showed that 45 Chinese female lupus patients presented an intestinal microbiota profile indicative of dysbiosis compared to 48 volunteers. The Firmicutes:Bacteroidetes ratio was decreased in patients compared to the healthy controls. Indeed, an abundance in the number of bacteria of the genera Rhodococcus, Eubacterium, Flavonifractor, Eggerthella, Klebsiella and Prevotella was significantly high in lupus patients, whereas the bacteria of the genera Pseudobutyrivibrio and Dialister exhibited low abundance. The intestinal microbiota (IM) in SLE patients includes a high abundance of Gram-negative bacteria and other types of bacteria, such as those of the genera Blautia, Odoribacter, and a genus from the family Rikenellaceae, with lower diversity than in controls¹⁵.

Li et al. (2019) investigated the intestinal microbiota in 40 Chinese female SLE patients and 22 healthy controls. They reported a lower Firmicutes/Bacteroidetes (F/B) ratio in the feces of remissive SLE patients compared to healthy controls. Among the altered microbiota, the genera Campylobacter, Streptococcus, Veillonella, and the species S. dispar and S. anginosus were positively correlated with lupus activity, whereas the genus Bifidobacterium was negatively associated with disease activity²⁴.

In another study, van der Meulen et al. examined the intestinal microbiota of 30 white SLE patients from the Netherlands with a European ethnic background and 965 controls from the general population. They also found a lower Firmicutes/Bacteroidetes ratio in SLE patients compared to controls. The relative abundance of the phylum Bacteroidetes and the genus Bacteroides was significantly higher in SLE patients compared to the general population controls. Additionally, the genus Alistipes (belonging to Bacteroidetes) and the phylum Proteobacteria showed a notably higher relative abundance in SLE patients compared to controls⁴⁹.

Luo et al. studied the composition of IM in 14 African American (not Caribbean) and Caucasian, non-Hispanic lupus patients and 17 healthy controls and showed a high abundance of the phylum Proteobacteria in patients. Other species were more abundant in lupus patients, mainly Odoribacter and Blautia (family Lachnospiraceae). The Firmicutes/Bacteroidetes ratios were not significantly different for SLE patients in remission versus non-SLE controls²⁸.

Bacteroidetes and Firmicutes are the two most abundant bacterial phyla in the human gut. Hevia et al. reported a decrease in the Firmicutes/Bacteroidetes (F/B) ratio in SLE patients¹⁶. In contrast, other research groups found no alteration in the F/B ratio of SLE patients^15,28. Our results, however, revealed a 1.29-fold increase in the F/B ratio in SLE patients (1.29 in cases compared to 0.81 in controls), along with a non-significant decrease in the abundance of Bacteroidetes.

Intestinal dysbiosis is also characterized by an increase in the F/B ratio in obesity². Conversely, a lower F/B ratio is evident in other autoimmune diseases^31,35. The F/B ratio decreases in patients with type 2 diabetes compared to controls²¹. Additionally, most studies on patients with Crohn's disease report a decrease in the abundance of Firmicutes and an increase in Bacteroidetes²¹. This imbalance in the F/B ratio appears to be a key feature of SLE dysbiosis, regardless of lifestyle, disease duration or stage, or diet.

Among the altered/disordered microbiota in our study, the genus Clostridium_XVIII (Firmicutes phylum) was negatively correlated with SLEDAI scores, while the class Betaproteobacteria (Proteobacteria phylum) and the genus Faecalibacterium (Firmicutes phylum) were positively associated with renal manifestations. The genus Odoribacter (Bacteroidetes phylum) was positively correlated with disease duration, while the genus Prevotella (Bacteroidetes phylum) was negatively correlated with BMI index. Phyla Firmicutes and Bacteroidetes were involved in host metabolism and immunity according to Round et al.³⁹.

Faecalibacterium is a key member of the core functional group of the gut microbiota, playing several crucial roles in maintaining human health. These roles include generating energy components and nourishing the intestinal epithelium, mitigating inflammation severity, preserving intestinal barrier functions, and enhancing colon motility¹⁰. Recent findings suggest that Faecalibacterium can also alleviate renal dysfunction in chronic kidney disease patients through butyrate-mediated GPR43 signaling in the kidneys²³. Elevated levels in SLE patients might be a compensatory response to improve renal symptoms.

Clostridium sp. can metabolize daidzein, an isoflavone shown to reduce inflammation in lupus-prone mice, potentially influencing SLE by decreasing daidzein levels⁸. Additionally, it catabolizes tryptophan, consumed by systemically activated T cells, via the kynurenine pathway, which has been found to be altered in SLE patients^1,33.

In our study, Bifidobacterium is less abundant in SLE patients with active disease compared to healthy controls. Liu et al. showed that short-chain fatty acids generated by Bifidobacterium and Lactobacillus combine and activate receptors (FFAR2, FFAR3, or GPR109a) on enterocytes to inhibit inflammatory responses by blocking nuclear factor-κ-light chain enhancer of B cells activation pathway. This point may explain the inflammatory response in SLE patients²⁵.

In our study, we observed some similarities with previous studies in bacterial microbiota but also noted differences likely due to variations in inclusion criteria. We included only female patients with a SLEDAI score <22 and also investigated the bacterial gut in SLE patients with active disease (SLEDAI >5), while Luo et al. and van der Meulen et al. included both male and female SLE patients of different ethnicities and with various levels of disease severity. Hevia et al.¹⁶ included only Caucasian female patients with SLEDAI scores <8 considering the remission course. Moreover, we included SLE patients requiring immunosuppressive, immunomodulatory therapies, or who had discontinued treatment, whereas Hevia et al.¹⁶ excluded those receiving immunological or steroid treatments. Our exclusion/inclusion criteria likely selected subjects who better represent the SLE patient population. This highlights the importance of considering the country of origin in microbiota comparisons and exercising caution in cross-cohort analyses from different countries due to strong geographical influences, mainly attributed to associated eating habits and dietary diversity, which significantly influence bacterial composition.

In our study, five of seven patients were treated with hydroxychloroquine and six of seven patients were treated with steroids. Angelakis et al.⁴ and Pan et al.³⁷ showed that short-term/high-dose or long-term use of hydroxychloroquine can also alter the gut microbiota, resulting in a decreased level in the relative abundance of Firmicutes. We did not find a decreased in relative abundance of Firmicutes in SLE patients treated with hydroxychloroquine compared to those not receiving this treatment. However, some commonly used medicines to treat patients with SLE, such as steroids and methotrexate, can inhibit the growth of certain gut bacteria or have been previously associated with reduced diversity of the gut microbiota³⁰.

Wu et al. (2011) reported that dietary interventions, including whole grains, increase the Firmicutes/Bacteroidetes ratio and that Firmicutes levels were positively associated with a low-fat/high-fiber diet. In our study, there were no significant differences in age, dietary intake, and lifestyle factors (smoking, alcohol consumption, physical activity, and use of vitamin and mineral supplements [data not shown]) between SLE patients and healthy controls, reducing the likelihood that these factors influenced our gut microbial profile analysis⁵⁰.

SLE is characterized by abnormal antibody responses to nuclear and cytoplasmic antigens, including mainly anti-nuclear antibodies and anti-double-stranded DNA. Sample classification revealed no specific association between bacterial clustering and different clinical features or antibodies. The development of SLE is due to a dysbiotic gut microbiota, leading to compromised gut mucosal barrier control and facilitating antigen transit. Additionally, the production of autoantibodies can result from molecular mimicry with different bacterial populations in SLE. Bacteria have been found mimicking orthologous epitopes similar to host proteins, thereby activating autoimmune T and B cells⁴¹.

We identified a specific type of microbiota for the SLE group. There are a few limitations to this study, most notably the small sample size, which was influenced by challenges in sample collection and exclusion criteria. Therefore, this may be insufficient to generate reproducible statistics. However, the strengths of our study lie in the investigation of the gut microbiota in female SLE patients with active disease receiving immunosuppressive treatment. Furthermore, our population is considered homogenous. Increasing the number of SLE patients and investigating the species level and the crosstalk between commensal bacteria, metabolites and the immune system may be crucial for understanding the immunopathology of this disease.

CRediT authorship contribution statement

Amira Messadi: writing, statistical analysis

Sameh Sayhi: revision, clinical evaluation

Kais Ghedira: bioinformatic analysis

Chadia Zaouaoui: data analysis

Bilel Arfaoui: clinical evaluation

Soumaya Khouikhi: revision

Maha Rebai: clinical evaluation

Nour El Houda Guediche: clinical evaluation

Nadia Abdelhafidh: revision

Bassem Louzir: clinical evaluation

Ajili Faida: revision and validation

Ethical considerations

The protocol of this study was approved by the Ethics Committee of the Military Hospital of Tunis.

Funding

This study was funded by department resources.

Conflict of interest

There is no conflict of interest.

Acknowledgments

The authors would like to thank Dr Imen Nahdi, the CEO of African Biotechnology Society and her collaborators for making available their laboratory and required equipment to perform the study and for performing a part of bioinformatic analysis. This study was funded by department resources.