Metformin improves FOXP3 mRNA expression through suppression of interferon gamma levels in pristane-induced murine models of lupus

Background: A recent study has indicated the potential of metformin therapy for lupus in animal models, but there has been no study evaluating the effect on pristane-induced lupus. This study aims to evaluate the effect of intraperitoneal versus oral metformin on interferon (IFN)-γ levels and FOXP3 mRNA expression on pristane-induced female BALB/c mice. Methods: In total, 31 female BALB/c mice, aged 6 weeks, were intraperitoneally induced with 0.5 ml of pristane (2,6,10,14-tetramethylpentadecane). After 120 days, the mice were grouped and treated with various treatments: normal saline 100 MCL, oral metformin 100mg/kg-BW, or intraperitoneal metformin 100mg/kg-BW. After 60 days of treatment, all treatment groups were sacrificed, and kidney specimens prepared and stained using hematoxylin and eosin. Results: IFNγ levels of saline controls vs. oral metformin group was 309.39 vs. 292.83 pg/mL (mean difference 16.56 pg/mL; 95% CI 0.74-32.37; p=0.042), and saline control vs. intraperitoneal metformin group was 309.39 vs. 266.90 pg/mL (mean difference 42.49 pg/mL; 95% CI 29.24-55.73 pg/mL; p<0.001). FOXP3 mRNA expression changes in saline controls vs. oral metformin group was 6.90 vs. 7.79-fold change (mean difference -0.89-fold change; 95% CI -1.68-(-0.11); p=0.03) and in saline controls vs. intraperitoneal metformin group was 6.90 vs. 9.02-fold change (mean difference -2.12-fold change; 95% CI -2.99-(-1.25); p=<0.001). Correlation analysis of FOXP3 mRNA expression and IFNγ level changes revealed a Pearson correlation of -0.785 (p=0.001) and R2 value of 0.616 (p=0.001). Conclusion: Metformin is a potential new therapy to reduce the levels of IFNγ and increase FOXP3 mRNA expression in mice models of systemic lupus erythematosus.


Introduction
Systemic lupus erythematosus (SLE) is a complex systemic disease, which is defined by multiple organ damage and dysfunction resulting from auto-antibody generation and inherited immune system dysregulation 1 . The complicated pathophysiology and clinical manifestations result in difficulty in reaching an effective and comprehensive management of this condition. Lupus treatment currently relies on immunosuppressants and corticosteroids to suppress the immune system and reduce disease activity. This strategy is not ideal; there are several types of patients who do not respond well to immunosuppression and this therapy also produces side effects, such as recurrent infection, bone density loss, sarcopenia and psychological disturbances. This has led to infection and cardiovascular comorbidity becoming the major cause of mortality related to SLE, and not the disease itself 2 .
Recent studies on experimental models has shown that the key to effective SLE management doesn't rely on suppression of immune system, but how to manage and balance the activity of several key players, such as T regulator (Treg), T autoreactive (Th17), B autoreactive, and B regulator lymphocytes 3 . Inflammatory cytokines and cellular components, such as tumour necrosis factor (TNF)-α, type 1 and 2 interferons, B-lymphocyte stimulator and interleukin-10 has also been known to contribute to the development of auto-antibodies and immune complexes that destroy tissues, especially in the kidneys [4][5][6] . Recently the activity of interferon (IFN)-γ and Th1 cells has been the focus of several experimental and clinical studies, especially its relationship to the development of lupus nephritis and its effect on downstream T-helper cells, such as Treg and Th17 [7][8][9] .
Several studies has also shown the influence of oxidative stress from environmental exposure to the imbalance of Treg and Th1 cells, with subsequent effects on the elevation of IFNγ levels and the development of SLE in exposed murine models 10 . Exposure to reactive oxygen species is known to disrupt mitochondrial potential balance and activate the mTOR (mammalian Target of Rapamycin) metabolism regulation pathway by suppressing AMPK (Adenosine Monophosphate Kinase). This in turn results in a preference of Th1 pathway activation rather than Treg 11-13 .
Metformin, an old anti-diabetic drug with a reputable safety profile, recently has been known to be able to regulate the AMPK/mTOR pathway and from several studies has been shown to be able to regulate several autoimmune-, inflammation-, malignant-and aging-related conditions 14,15 . Studies on mice models of rheumatoid arthritis, autoimmune encephalitis and lupus nephritis has also shown this drug's ability as a potential therapy of autoimmune disease 11,16 . This study aims to evaluate the effect of intraperitoneal versus oral metformin in decreasing IFNγ and increasing FOXP3 mRNA expression levels on pristane-induced female BALB/c mice, as there no studies that have evaluated the route of metformin delivery, especially on an environmentally induced model of lupus nephritis.

Animal models
In total, 30 female BALB/c mice, aged 6 weeks and weighing approximately 25 grams, were purchased from Universitas Hasanuddin Makassar (Indonesia) and then maintained at the Animal Unit of the Molecular Biology Laboratory of Universitas Hasanuddin Makassar from January 2018. The mice was kept in a temperature controlled housing according to their study group, food and water was provided freely. The number of mice for intervention study was determined using Federer formula for 5 groups. Efforts was made to minimize suffering, such as minimal handling, less frequency of venous sampling, adequate space for living and no other experimentation or pain inducing procedures.
After 2 weeks of acclimatization the mice were then randomly divided into five groups (6 mice/group): group 1, normal control; group 2, SLE model; group 3, normal saline; group 4, oral metformin; and group 5, intraperitoneal metformin groups. Four of the groups (all apart from normal control) were induced with 0.5 ml of pristane (2,6,10,14-tetramethylpentadecane; Sigma Aldrich) intraperitoneally. The normal control group was injected with normal saline 0.5 ml intraperitoneally as a control group. After 120 days, groups 1 and 2 were sacrificed using chloroform euthanasia methods (dose of 20 g/m3 in a closed chamber system). Kidney specimen was then fixed with neutral buffered formalin (NBF) 10%, prepared in paraffin block, sliced to 2μM thickness and then stained with haematoxylin and eosin (H&E).
All intervention, analysis and reporting conducted in this study follows the ARRIVE guidelines for animal studies. Ethical approval for animal studies was obtained from Universitas Hasanuddin's Health Studies Ethical Committee, with protocol number UH17030037. Care and intervention conducted in the research animal, refers to Indonesian National Guidelines on Health Research Ethics and Indonesian Food and Medicine Regulatory Body Guidelines on Good Clinical Practice 17 .

Intervention
After 120 days, the intervention groups (groups 3-5) were given therapy every morning in the animal laboratory according to their designation: group 3, given 100 mcl normal saline via oral gavage once daily; group 4, given 100 mg/kg body weight of metformin diluted in 100 mcl normal saline via oral gavage once daily; group 5, given 100 mg/kg body weight of metformin diluted in normal saline via intraperitoneal injection once daily. Treatment lasted for 60 days and at the end of the period

Amendments from Version 1
We have updated the statistical analysis to one-way ANOVA with Tukey's posthoc analysis. Conclusion in the abstract has also been revised to reflect the limitation of this study. Several minor revisions have been done to miss-spellings and addition to legends in the figures.
Any further responses from the reviewers can be found at the end of the article REVISED all three groups were sacrificed, kidney specimen fixated with NBF 10%, prepared in paraffin block and then stained with H&E.
Cytokine and mRNA expression measurement Samples for IFNγ and FOXP3 mRNA measurements was collected from tail vein sampling (0.1-0.2 ml for each sample). IFNγ was measured using the murine IFNγ ELISA kit from Abcam (ab100689) and read using ELISA Reader 270 with 450 nm wavelength (Biomerieux, France).
Total RNA was isolated from using the Qiagen RNeasy Micro Kit (DNA Genotek, Qiagen) according to the manufacturer's instructions. Complementary DNA synthesis was performed by using iScript™ Advanced cDNA Synthesis Kit for RT-qPCR (Bio-Rad). Using cDNA synthesized from 150 ng of total RNA as a template for one amplification, real-time reverse transcriptase (RT)-PCR (CFX Connect system; Bio-Rad) was performed using SYBR ® Green PCR Master Mix one step according to the instructions provided (Bio-Rad). Final reaction volume was 20 μL, and included 2 μL cDNA, 10 μL SYBR Green Master Mix, 0.5 μL each of the forward and reverse primers (10 pmol), and 7 μL nuclease-free water. Amplification conditions used for qPCR were: 95°C for 2 minutes, followed by 40 cycles of denaturation and annealing/extension cycles at 95°C for 5 seconds and 60°C for 30 seconds.
The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was used as an internal control for normalization, GAPDH primer, forward (5'GAAGGTGAAGGTCGGAGT-3') and reverse (5'-GAAGATGGTGATGGGATTTC-3'). Fold change was determined by the ΔΔCt method. All measurements were conducted as per manufacturer's instructions and repeated three times to ensure the validity. Target protein concentration was read as picogram/millilitre and mRNA expression as fold change.
Histopathological analysis Kidney specimens were examined by two independent and blinded histopathologists experienced in evaluating murine renal samples. Glomerular scores were evaluated by measuring the level of destruction on 50 glomerular units in each mouse and scored as 0 = normal, 1 = mesangial expansion, 2 = endocapillary proliferation, 3 = capillaritis or necrotic changes and 4 = crescents. Interstitial scoring was measured by evaluating 50 high power fields and scored as 0 = no interstitial involvement, 1 = <25% interstitial involvement, 2 = 25-50% involvement and 3 = >50% involvement 18 .

Statistical analysis
Statistical analysis (SPSS Statistics ver. 20, IBM) was done by measuring mean difference (one way ANOVA with Tukey's post test for multiple comparisons) to evaluate the difference in IFNγ levels, FOXP3 mRNA fold change and histopathological scores between the three intervention groups (comparing saline control, oral and intraperitoneal metformin groups). Mean difference analysis using t-test was done for the proof of concept group (comparing normal control and SLE model groups). Correlation analysis was also done to evaluate the relationship between IFNγ and FOXP3 mRNA changes to determine the strength of the causality. All tests were reported with 95% confidence interval, standard error and statistical significance score (p<0.05).

Results
Two female mice expired in the adaptation period; therefore, only 29 mice entered the intervention period and finished the experiments without problems, analysis was done with 5 mice from each group (n=25).
Histopathological analysis on kidney specimens resulted in a variable change in each intervention group (Figure 3). Histopathological scoring between intervention groups (normal saline, oral and intraperitoneal metformin) did not reveal significant differences, although qualitative analysis by blinded pathologists revealed difference in the degree of nephritis occurring in each group (Figure 4).

Discussion
SLE is a multifactorial inflammatory autoimmune disease, with clinical manifestations that involves various tissues and organs. The aetiology of this autoimmune condition is linked to dysfunctional B and T lymphocyte responses to environmental stimulus in a genetically susceptible individual, which in turn determines the immune response to various autoantigens and can cause tissue damage. The application of pristane, an aromatic hydrocarbon, has an advantage to genetically modified mice, because the ability of this model to mimic SLE in humans, which in a genetically susceptible individual is usually caused by environmental exposure. Pristane mouse models also enables researchers to evaluate pathophysiological changes in a timely manner, and to give a picture of the cellular mechanism involved in the development and progressivity of SLE 12 .
In this study, we showed that after induction with pristane, there was a significant difference in the level of IFNγ     Mardani et al. showed that an intervention with probiotics could reduce IFNγ and IL-17 levels in pristane induced murine lupus, followed by a reduction in autoantibodies such as ANA, anti-dsDNA and anti-RNP 9 . Reduction of IFNγ levels could improve the outcome of lupus nephritis by inactivating B7/CD28 signalling pathway, which results in a reduction in ANA autoantibodies, IL-4 and IFNγ levels. The inactivation of B7/CD28 pathways also caused anergy, tolerance and apoptosis of T-cells, which results in a decrease of urine protein and immune complex deposition in the kidneys of pristane induced C57BL/6J mice 22 .
This study revealed that the administration of oral and intraperitoneal metformin gave a significantly better suppression of IFNγ than placebo, in accordance to the study conducted by Yin et al. 23 . In that study, intervention with metformin and 2-DG (2-deoxy-glucose) in B6.Sle1Sle2.Sle3 mice resulted in a suppression of IL-17 and IFNγ levels through a blockade on the glucose oxidation pathway. This blockade on the glucose oxidation pathway also normalized T-cell metabolism, which in turn suppresses the activation of CD4+ T-cells and returns the balance of pro/anti-inflammatory cytokines in mice models 10 .
Regarding FOXP3 mRNA expression, this study showed that intervention with metformin via oral (6.90 vs. 7.79-fold change; mean difference -0.89; 95% CI -1.68 -(-0.11); p=0.03) or intraperitoneal route (6.90 vs. 9.02-fold change; 95% CI -2.99 -(-1.25); p<0.001) gave a significantly superior increase in FOXP3 mRNA expression compared to saline control. We also showed that there was a strong significant inverse correlation between the increase of FOXP3 mRNA expression and decrease of IFNγ resulting from metformin intervention (R=-0.785; p=0.001). Furthermore, it seems that the reduction of IFNγ explains the increase in FOXP3 mRNA expression rather strongly, as shown by the R 2 value of 0.616 (p=0.001). The above result was consistent with several studies in pristane induced models; a decrease in FOXP3+ T-cells and increase in CD4+CD69+ T-cells coincide with an increase in IFNγ levels in intraperitoneal fluid 20,23 .
To the best of our knowledge, this is the first study that evaluates the effect of metformin on the expression of FOXP3 mRNA in lupus, although it is also known that metformin has the ability to induce AMPK pathway activity and suppress mTOR signalling 24,25 . Metformin has also been known to be able to improve disease activity index, histological and inflammatory profiles in several other autoimmune models, such as inflammatory bowel disease 26 and autoimmune insulitis 14 through the modulation of AMPK-mTOR pathway and the resulting changes in IL-17, IFNγ, IL-10 and FOXP3 associated cytokines and cells.
Intraperitoneal route of metformin gave a superior effect on the suppression of IFNγ levels and increasing FOXP3 mRNA expression compared to the oral route, and to the best of our knowledge this was the first study that observed this effect in pristane induced murine model of lupus Although this study was able to prove that there was a characteristic change in accordance to lupus nephritis in pristane induced models compared to normal BALB/c, subsequent therapy with metformin failed to produce a statistically significant score change. However, qualitative analysis by blinded pathologists has confirmed that there was at least a difference in renal changes that showed better results in intraperitoneally treated mice compared with oral metformin and placebo control. This result could be caused by a short period of intervention; a longer treatment time could possibly result in a significant difference in renal scoring.

Limitations
We did not perform an evaluation of autoantibodies related to SLE, such as anti-dsDNA, anti-Sm and anti-RNP1. However, several murine studies has confirmed the ability of pristane induced BALB/c in producing related auto-antibodies 29,30 . Our research also did not evaluate the antibody response to metformin therapy; however several studies has shown the ability of metformin in reducing autoantibodies related to SLE 10,22 . We also did not evaluate the expression of mRNAs related to IFNγ, but several studies has shown that Th1 activity is closely related to IFNγ levels 8,31,32 . Furthermore it has been recently suggested that the cytokine balance could play an important role in determining active T-cell subsets, changing the phenotype of peripheral T-cells and contributes to the pathogenesis of lupus 13,33 .

Conclusions
A murine model of SLE by pristane induced female BALB/c mice could be used to represent a model of lupus similar to the human condition. The increased activity of Th1 and reduced activity of Treg, in this study represented by pro-inflammatory IFNγ levels and FOXP3 mRNA expression, has proven to be related to the development of lupus nephritis. Metformin is a potential new therapy to reduce the levels of IFNγ and increase FOXP3 mRNA expression in SLE and in turn inhibits the development of glomerulonephritis. Intraperitoneal metformin, intravenous in humans, could provide a novel route of administration to improve the effect of metformin on lupus patients.

Open Peer Review
Yes

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
In the manuscript by Sumantri et al., the authors evaluated metformin in the pristane-inducible model of SLE. The authors tested both oral gavage and intraperitoneal administration of metformin as a treatment. Lupus was induced by a single ip injection of pristane into female BALB/c mice, after which lupus was allowed to develop for 120 days. At this point, mice were given vehicle, oral metformin, or IP metformin daily for 60 days. The experimental endpoints were as follows: the expression of mRNA for the Treg-associated transcription factor FoxP3 and circulating IFN-ɣ, both of which were assessed using blood collected from the tail vein. In addition, and histological examination of H&E stained kidneys was performed. The administration of pristane resulted in an increase in circulating IFN-ɣ and a decrease in the expression FoxP3, and treatment with metformin decreased IFN-ɣ and increased FoxP3 expression. The authors also conclude that treatment with pristane leads to renal injury, including glomerular changes, but minimal mesangial expansion. Treatment with metformin did not lead to significant changes in renal injury. Overall, this manuscript is written clearly, and the methods would be easy to follow if one were to repeat these experiments.
One specific concern is the statistical analyses that are used. It is stated that t-tests were used; however, there are five experimental groups. I suggest that the statistical tests be redone using one-way ANOVA, followed by Tukey's post test for multiple comparisons. Another issue/point of confusion that I have is with the presentation of the data in the Figures and Tables. In both Figure  1 and Figure 2, I suggest adding asterisks or other markers to indicate statistically significant differences among the groups. While it is stated in the text, it would be clearer if it was also on the graphs. Also, are the marks in Figure 1B, 1C, and 2A statistical outliers? Please indicate in the figure legends. In addition, Table 2 states that it is composed of IFN-ɣ and FoxP3 expression after intervention with metformin. While I think these are the data that are presented, the table says "post induction" as in Figure 1, which was referring to post treatment with pristane. I think that Table 2 should be relabeled with "post treatment" or something similar. Finally, the kidney specimens were scored, but those scores were not presented in any format. I would suggest presenting the scoring data in a graphical format in the manuscript.
The pristane model of SLE mimics human disease in many ways, including increases in IFNproduction and dysregulation of T cell subsets. The data that the authors present suggest that metformin may improve the Th subset imbalances (increasing FoxP3, decreasing IFN-ɣ), but direct evidence of this is lacking. The findings could be strengthened by the addition of flow cytometry to assess Th subsets. In addition, the authors indicate that they did not measure immunoglobulins or autoantibodies in their animals. While data may be available in other studies, it would be a good additional piece of data for this study.

Is the work clearly and accurately presented and does it cite the current literature? Partly
Is the study design appropriate and is the work technically sound? Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? effect of metformin on other manifestations, i.e. serological, in this model. The conclusion of "intravenous administration in human, could provide a novel route of administration...." could not be drawn from the data provided in the manuscript.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Partly

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? Partly