Linking the Circadian Rhythm Gene Arntl2 to Interleukin 21 Expression in Type 1 Diabetes
- 1Department of Developmental & Stem Cells Biology, Institut Pasteur, CNRS URA 2578, Laboratoire de Génétique Moléculaire Murine, Paris, France
- 2Université Pierre et Marie Curie, Cellule Pasteur UPMC, Paris, France
- Corresponding author: Ute C. Rogner, .
The circadian rhythm–related aryl hydrocarbon receptor nuclear translocator-like 2 (Arntl2) gene has been identified as a candidate gene for the murine type 1 diabetes locus Idd6.3. Previous studies suggested a role in expansion of CD4+CD25− T cells, and this then creates an imbalance in the ratio between T-effector and CD4+CD25+ T-regulator cells. Our transcriptome analyses identify the interleukin 21 (IL21) gene (Il21) as a direct target of ARNTL2. ARNTL2 binds in an allele-specific manner to the RNA polymerase binding site of the Il21 promoter and inhibits its expression in NOD.C3H congenic mice carrying C3H alleles at Idd6.3. IL21 is known to promote T-cell expansion, and in agreement with these findings, mice with C3H alleles at Idd6.3 produce lower numbers of CD4+IL21+ and CD4+ and CD8+ T cells compared with mice with NOD alleles at Idd6.3. Our results describe a novel and rather unexpected role for Arntl2 in the immune system that lies outside of its predicted function in circadian rhythm regulation.
The murine type 1 diabetes (T1D) locus Idd6 is 1 of ∼40 genetic loci identified in the NOD mouse (1). The Idd6 candidate region (2), showing resistance to the spontaneous development of diabetes, overlaps with the candidate region for the resistance of immature T cells to dexamethasone (3–5) and for the control of low rates of proliferation in immature NOD thymocytes (6). Idd6 controls the activity of regulatory CD4+ T cells and invariant natural killer T cells (7,8).
Previous research has also focused on the identification of candidate genes and the underlying molecular networks (9–13). The analysis of three NOD.C3H subcongenic strains (6.VIIIa, 6.VIIIb, and 6.VIIIc) derived from the original 6.VIII congenic strain showed the presence of at least three diabetes-related subloci contributing to the overall T1D resistance (Idd6.1, Idd6.2, Idd6.3), but Idd6.3 alone controls the suppressive activity of splenocytes (10).
Transcription and sequence analysis revealed aryl hydrocarbon receptor nuclear translocator-like 2 (Arntl2) as candidate gene within Idd6.3. Arntl2 encodes a basic helix-loop-helix/Per-Arnt-Sim transcription factor controlling the circadian rhythm. Arntl2 is upregulated in the spleen and thymus of mice carrying C3H alleles at Idd6.3 compared with mice carrying NOD alleles at the locus. In addition, several polymorphism and different splice forms were identified when comparing the gene and its transcripts in C3H and NOD strains (10). Diabetes incidence is increased by Arntl2 mRNA interference in Idd6 congenic mice concomitant with an increase in CD4+ T cells and a decrease in regulatory CD4+CD25+ T cells in the peripheral immune system (14). In addition, upregulation of cellular Arntl2 levels correlates with inhibited CD4+ T-cell proliferation and their decreased diabetogenic activity (15).
The current study addresses transcriptional changes related to Arntl2 expression in CD4+ T cells. We show that Arntl2 and Idd6.3 control the expression of the interleukin 21 (IL21) gene (Il21) and the number of IL21-producing cells. The promoter of Il21, a gene that codes for an important cytokine involved in the proliferation of T cells, is directly targeted by ARNTL2 in an allele-specific manner. We propose that the circadian rhythm–related Arntl2 gene has specific functions in regulating cytokines involved in T1D.
Research Design and Methods
The NOD.C3H 6.VIIIa and 6.VIIIc congenic strains (10), NOD/Lt, and NOD/severe combined immunodeficiency (SCID) mice were maintained by brother-to-sister mating at the Pasteur Institut animal facility. NOD/SCID.C3H congenic mice were established from the congenic strains (7,10) by crossing them to the NOD/SCID strain. F1 generation mice were intercrossed, and mice homozygous for both the C3H-derived intervals and the SCID mutation were selected. All animal studies were approved by the relevant institutional review boards (Comité d’éthique en expérimentation animale CEEA 59, IDF, Paris) under the protocol number 2009-0015.
CD4+ T cells were separated from splenocytes of three 7-week-old 6.VIIIa and 6.VIIIc female mice using the untouched isolation Kit II (Miltenyi Biotec SAS, Paris, France). One million cells were stimulated in a 24-well tissue culture plate with 25 µL anti-CD3/CD28 activator beads (Invitrogen, Carlsbad, CA) and 30 units (25 µg/mL) of hIL2 (R&D Systems Europe, Lille, France). The cells were collected 24 h later, and the DNA-free RNAs were isolated with the RNeasy Mini Kit (Qiagen, Hilden, Germany).
Splenocytes from two 8-week-old female NOD mice were used to separate the CD4+ T cells (untouched Isolation Kit II, Miltenyi Biotech). Five million CD4+ T cells were transfected with 10 µg plasmid pcDNA3.1-Arntl2 (15) or control pcDNA3.1/HisC (Life Technologies) by nucleofection using Program X-001 (Amaxa, Cologne, Germany). After culture for 3 h, the transfected cells were stimulated for 24 h and RNA was extracted.
The RNA quality was verified using a Bioanalyser 2100 (Agilent, Santa Clara, CA). Fluorochrome labeling and hybridization on Mouse Gene 1.0 ST Array were performed by the Pasteur Institute platform according to the manufacturers’ instructions (Affymetrix, Santa Clara, CA). The microarray data were deposited in the ArrayExpress database under the accession number E-MTAB-1957.
Diabetes Transfer Assays
CD25− splenocytes were prepared using the mouse CD4+CD25+ Regulatory T Cell Isolation Kit (Miltenyi Biotec), and 5.5 × 106 cells in PBS were injected intravenously into recipient mice. Diabetes development was monitored twice weekly using the Diabur 5000 test (Roche, Mannheim, Germany). Animals were considered diabetic when their urine glucose level exceeded 250 mg/dL. Time-to-event distributions were calculated by Kaplan-Meier estimation and compared by log-rank tests during the period of observation.
Five million CD4+ T cells were purified using the untouched isolation kit (Miltenyi Biotech) from spleens of 6- to 8-week-old 6.VIIIa and 6.VIIIc female mice and used for total RNA isolation using the RNeasy Mini Kit (Qiagen). For circadian analyses, spleen samples were collected and purified every 4 h during a 24-h period. Amounts of 200 ng to 10 µg of the total RNA were used for quantitative (Q)-RT-PCR, as previously described (12). The amplification quantification in arbitrary units was performed by the ΔCt method using three replicates and Arpo as an endogenous control to normalize mRNA levels. The values are presented as mean ± SD. Combined data of the experiments are presented as mean values ± SE. The differences between two groups were analyzed using the Mann-Whitney method.
Flow Cytometry Analysis
Expression of T-cell–specific proteins was analyzed by fluorescence-activated cell sorting after surface staining and/or intracellular staining using anti-mouse CD4 (Alexa Fluor 488 or Pacific Blue; BD Biosciences, Le Pont de Claix, France), anti-mouse T-cell receptor (PerCP Cy5.5, BD Biosciences), anti-mouse B220 (PE Texas Red, BD Biosciences), anti-mouse CD8 (PECy7, BD Biosciences), anti-mouse IL17 (PE, BD Biosciences; or Alexa Fluor 488; eBiosciences, Paris, France), phycoerythrin-conjugated anti-mouse IL21 (PE, eBiosciences), anti-mouse interferon-γ (IFN-γ) (AF700, BD Biosciences), anti-mouse IL4 (APC, BD Biosciences), anti-mouse CD11c (APC, BD Biosciences), and anti-mouse FoxP3 (APC, eBiosciences) antibodies at optimal concentration. Briefly, the cells were stimulated for 4 h with phorbol myristate acetate (25 ng/mL; Sigma-Aldrich, St. Louis, MO) and ionomycin (10 μg/mL; Sigma-Aldrich) in the presence of Brefeldin A (10 µg/mL; BD Biosciences). The cells were washed with PBS 0.5% BSA, incubated in the eBioscience fixation/permeabilization buffers according to the manufacturer’s protocol, and stained with fluorescent antibodies. Fluorescence-activated cell sorting was performed on a Cyan flowcytometer (Beckman Coulter, Villepinte, France) and analyzed using Flowjo software. Statistical analyses were performed using the Mann-Whitney method.
Chromatin Immunoprecipitation Analyses
Splenocytes (3 × 107) or CD4+ T cells (4 × 106) of female mice at 6–8 weeks of age were used for chromatin immunoprecipitation (ChIP) analyses using a previously described protocol (16) with the following modifications for chromatin digestion: 150 units MNase (Thermo Scientific, Villebon sur Yvette, France) were added for 10 min at 37°C, and the reaction was stopped with 4 µL EDTA (500 mmol/L) and protease inhibitors. Extracts were sonicated to obtain DNA fragments between 160 and 320 bp. Magnetic beads for immunoprecipitation (Dynabeads Protein G; Life Technologies, St. Aubin, France) were prepared according to the manufacturer’s indications and resuspended in TSE150 buffer (0.1% SDS, 1% Triton X-100, 2 mmol/L EDTA, 20 mmol/L Tris-HCl, 150 mmol/L NaCl). Chromatin was precleared by incubation with 50 µL magnetic beads for 1.5 h at 4°C on a rotating wheel. Precleared chromatin was then incubated with 4 µg anti-ARNTL2 (SC-376287; Santa Cruz Biotechnology, Dallas, TX), anti-ARNTL1 (ab3350; Abcam, Paris, France), anti-RNA Pol II (PB-7c2; Euromedex, Strasbourg, France), anti-CLOCK (circadian locomotor output cycles kaput) (ab3517; Abcam), anti-histone H3 (ab6002; Abcam), or anti-histone H3 acetyl (ab12179; Abcam) antibodies overnight at 4°C on the rotating wheel. Negative control precipitations used 10 µL total rabbit serum. Incubation was continued with 50 µL magnetic beads for 2 h at 4°C on the rotating wheel. Beads were then washed and eluted as described previously (16).
All quantifications were performed by real-time PCR in the presence of SYBR Green (Roche, Boulogne-Billancourt, France) on three replicates using 50 ng DNA. The results were calculated and normalized as (Ct of IP/Ct of input)/(Ct of control-IP/Ct of input). Data for single experiments are presented as means ± SE of experiments and as means ± SD of three technical replicates. We have verified that the anti-ARNTL2 antibodies were equally efficient for both the NOD and the C3H allelic version of ARNTL2 in IP experiments, followed by Western blotting using an anti-ARNTL2 antibody (ab86530; Abcam). For screening of Il21 by ChIP we designed primers from 2,500 bp 5′ of the start codon to 2,000 bp 3′ of the stop codon. For Arntl2, we designed primers every 250 bp from 600 bp 5′ of the start codon to 4,500 bp 3′ of it (Supplementary Table 1).
Identification of Genes Regulated by Idd6.3 and Arntl2 in CD4+ T Cells
We performed two complementary transcriptome analyses on CD4+ T lymphocytes to identify genes with a variation of expression linked to Arntl2. In the first experiment, we compared the transcriptomes of CD4+ T cells isolated from spleens of the NOD.C3H 6.VIIIa and 6.VIIIc strains containing, respectively, NOD and C3H alleles at the Idd6.3 locus, which spans Arntl2. In the second experiment, we compared the transcriptomes of NOD CD4+ T cells transfected with pcDNA3.1/HisC plasmids that allowed overexpression of Arntl2 or not (15). Both experiments were done in triplicate, and genes with P ≤ 0.05 in the Bonferroni test were considered for further analysis. We identified 44 genes differentially expressed in CD4+ T cells of the two congenic strains and 223 genes in the overexpression test. Although about half of the genes in the first experiment were up- or downregulated in strain 6.VIIIc compared with strain 6.VIIIa, the Arntl2 overexpression resulted in upregulation of more than 80% of the genes. Using medium stringency in the DAVID (Database for Annotation Visualization and Integrated Discovery) functional annotation, we identified 36 clusters among the 223 genes that were mostly associated with general cellular function, in particular, with ribosome complexes. This may reflect the increased protein production after overexpression. The 44 differential genes identified in the two strains clustered into only four clusters, including ribosomal proteins (enrichment score, 2.12), cytokine activity (score, 1.37), membrane proteins (score, 0.02), and ion binding (score, 0.01). Within the cytokine cluster we identified, for example, IFN-γ, IL3, IL21, IL22, and the high-mobility group box protein 1.
The two experiments had 24 genes in common (Table 1). We next asked if genes identified in the microrarray experiments follow the same circadian rhythm as Arntl2. We analyzed the expression in the 6.VIIIa and 6.VIIIc strains, and to this end, we isolated the spleens of 5- to 7-week-old female mice at different times of the day, selected the CD4+ T lymphocytes, and extracted their RNA to test gene expression by Q-RT-PCR. The level of Arntl2 expression in the 6.VIIIa strain was relatively low, but interestingly, 9 of the 24 genes—Ccl19, Cox7c, Hmgb1, Il21, Rpl26, Rpl35a, Rps8, Serpinb9, and Tmed2—found in the transcriptome analysis had a pattern very close to Arntl2 in the 6.VIIIa strain (Fig. 1). Gag and H2t22 showed this expression variation in strain 6.VIIIa, but an alteration in strain 6.VIIIc, which did not correspond to the expression variation of Arntl2. AW112010, Car2, Eraf, Il22, Rnu2, Rpl17, and Rpl29 had a very different expression variation. For most of the genes, we found expression differences at limited time points (Fig. 1 and Table 1). Interestingly, only Il21 showed differences between 6.VIIIa and 6.VIIIc mice during all the time points from 9:00 a.m. to 5:00 p.m. (1700 h). IL21 has an important role in the expansion of the T cells and on the immune response that may correlate with previous observation of the function of Arntl2 in T-cell proliferation. We therefore decided to analyze further the expression of this IL in our congenic mice.
IL21-Producing CD4+ T Cells Are Increased by NOD Alleles at Idd6.3
The transcriptome and the Q-RT-PCR results show a higher level of Il21 transcripts in the CD4+ T cells of the 6.VIIIa strain compared with the 6.VIIIc strain. This difference could be due to an increased number of cells expressing IL21. We addressed this question by flow cytometric analysis using anti-IL21 staining on T-cell receptor and CD4-labeled splenocytes (Fig. 2A). In agreement with the transcriptional results, the 6.VIIIc strain with C3H alleles at the Idd6.3 locus and expressing higher levels of Arntl2 exhibits less CD4+IL21+ T cells than the 6.VIIIa strain with NOD alleles at Idd6.3 and expressing lower levels of Arntl2. The test showed also that the CD4+IL21+ T-cell difference between the two strains decreased with age and that 10- to 13-week-old 6.VIIIc mice exhibit CD4+IL21+ T-cell numbers similar to 6.VIIIa mice (Fig. 2B). The age-dependent effect was also found in pancreatic lymph nodes. Significant differences were observed at 3 weeks of age (6.VIIIa, 4.24%; 6.VIIIc, 2.51% [n = 4]; P = 0.029) but not in older mice.
IL21 is produced by several CD4+ T-cell subpopulations, but mostly by Th17+ cells. We analyzed different CD4+ T-cell subpopulations in our strains to test if Idd6.3 influences their number. We used antibodies against IL17, a specific IL of Th17 cells, antibodies against IL4, specific for Th2 cells, and antibodies against IFN-γ, specific for Th1 cells. Th1, Th2, and Th17 are effector cells involved in regulating the immune balance together with Treg cells. We also tested the proportion of CD4+FoxP3+ T cells. FoxP3 is a specific factor of regulatory T cells (Fig. 3A). We found an effect of the locus Idd6.3 on the proportion of the Th17 and Th2 cells but not on the Th1 and the Treg cells. There is a strong variation between the strains 6.VIIIa and 6.VIIIc for IL17, indicating that the 6.VIIIa strain splenocytes contain more Th17 CD4+ T lymphocytes than those of the 6.VIIIc strain. A smaller but significantly different effect was found for IL4-producing CD4+ T cells. This result indicates that the difference in CD4+IL21+ T cells correlates with the Th17 and Th2 cells producing this cytokine. We also tested the proportions of CD4+IL22+ T cells, a cytokine previously identified in the transcriptome analyses but not further confirmed by Q-PCR. IL22 is also produced by Th17 cells. There was no significant difference for CD4+IL22+ T cells at 3 weeks (6.VIIIa, 0.49%; 6.VIIIc, 0.38% [n = 9]; P > 0.298), 6 weeks (6.VIIIa, 0.42%; 6.VIIIc, 0.40% [n = 9]; P > 0.124), or 12 weeks (6.VIIIa, 1.06%; 6.VIIIc, 0.88% [n = 9]; P > 0.169).
We also tested whether the congenic mice exhibit differences in major immune cell populations (Fig. 3A). We found significant differences for CD4+ T cells and CD8+ T cells that were increased in 6.VIIIa splenocytes. B cells and CD11c+ cells were increased in 6.VIIIc splenocytes. The relative expansion of T cells in 6.VIIIa splenocytes may be the direct result of increased IL21 levels (17). This result correlates with the previous findings of inhibition of CD4+ T-cell proliferation by Arntl2 (14,15). Interestingly, the differences in T-cell numbers were not yet significant at 3 weeks of age for CD4+ T cells (6.VIIIa, 37.24%; 6.VIIIc, 29.37% [n = 4]; P > 0.112) or CD8+ T cells (6.VIIIa, 14.10%; 6.VIIIc, 19.14% [n = 4]; P = 0.665), although differences in the IL21-producing CD4+ T cells were already present. This would suggest that the differences in IL21-producing cells precede the expansion of T cells. At 12 weeks of age, the significant T-cell differences were no longer present for CD4+ T cells (6.VIIIa, 27.09%; 6.VIIIc, 23.87% [n = 4]; P = 0.061) or CD8+ T cells (6.VIIIa, 17.60%; 6.VIIIc, 16.77% [n = 4]; P = 0.47).
Because T-cell development and selection takes place in the thymus, we performed flow cytometric analyses using thymic cells of the strains 6.VIIIa and 6.VIIIc (Fig. 3B). The number of CD4+ T cells and CD8+ T cells were similar, and no difference was found for double-positive or double-negative T cells. Only among the CD4+ T cells did we detect differences in IL21+, IL17+, and IL4+ cells. This result indicates that the observed differences between the 6.VIIIa and 6.VIIIc strains may be established during the negative selection of CD4+ T cells in the thymus.
NOD Alleles at Idd6.3 Increase the Diabetogenic Activity of Splenocytes
Our previous studies showed that the Idd6.3 interval, containing the gene Arntl2, influences protection in diabetes cotransfer assays. Splenocytes from the congenic strain NOD.C3H 6.VIIIc (carrying C3H alleles at Idd6.3) confer more protection than those of the congenic strain NOD.C3H 6.VIIIa (carrying NOD alleles at Idd6.3) (10). Our results on increased IL21 production in 6.VIIIa splenic CD4+ T cells and the changes in the balance between T cells and B cells and macrophages indicate that splenocytes from 6.VIIIa mice are likely to be more diabetogenic than those from 6.VIIIc mice. To validate this assumption, we injected CD25− splenocytes obtained from 10-week-old female NOD.C3H 6.VIIIa, NOD.C3H 6.VIIIc, and NOD mice to NOD/SCID mice. Cells from NOD.C3H 6.VIIIc induced significantly less diabetes than cells from NOD.C3H 6.VIIIa and NOD mice (Fig. 4A). These results confirm that C3H alleles at Idd6.3 decrease the diabetogenic activity of splenocytes.
To study if Idd6.3 alleles contribute to diabetes resistance factors outside of the immune system, we transferred diabetogenic CD25− splenocytes isolated from diabetic NOD mice to NOD/SCID.6.VIIIa and NOD/SCID.6.VIIIc mice. These strains did not significantly differ in their diabetes incidence from each other, neither from NOD/SCID nor the previously described Idd6 congenic strain NOD/SCID.C3H 6.VIII (Fig. 4B). We concluded that the Idd6.3 alleles, like the Idd6 alleles (7), do not provide major resistance factors in the absence of an intact immune system. We also tested the proportions of CD4+IL21+ T cells 1 and 3 days after transfer to NOD/SCID.C3H 6.VIIIa and NOD/SCID.C3H 6.VIIIc mice. We found no difference between strains or time points (day 1: 6.VIIIa, 1.22%; 6.VIIIc, 1.39%; day 3: 6.VIIIa, 1.37%, 6.VIIIc, 1.21% [n = 4]; P = 0.357 to P = 0.440 for all comparisons). Together, these tests indicate that an intact immune system may be required to establish the strain-dependent differences in this cell population.
Il21 Is a Target Gene of ARNTL2
The rapid induction of Il21 but not of the Il17 expression within 24 h after transfection with ARNTL2-expressing plasmids suggests that ARNTL2 acts on the transcriptional level of Il21. To test if Il21 could be a direct target of ARNTL2, we used ChIP to analyze the presence of the ARNTL2 protein on the Il21 gene. To this end, we designed and tested a set of primer pairs throughout the gene for use in ChIP-Q-PCR (Fig. 5D). Our results show that ARNTL2 binds with a strong peak of ∼1,000 bp upstream of the start codon of Il21 (Fig. 5A). The binding is found in the 6.VIIIc strain but absent in the 6.VIIIa strain. The use of anti-histone H3 and anti-acetylated histone H3 antibodies resulted in no differences in the Il21 region, suggesting that the binding pattern for ARNTL2 is not due to general differences in the chromatin accessibility in the strains (Supplementary Fig. 2B and C). Our results suggest that ARNTL2 may act as a transcriptional repressor in strain 6.VIIIc. This hypothesis is reinforced by our finding that RNA polymerase II binds to the same position as ARNTL2 not only in 6.VIIIa but also in strain 6.VIIIc (Fig. 5B).
Interestingly, we could not detect any specific binding or differential binding of CLOCK or ARNTL1 on Il21 (Supplementary Fig 2A and Fig. 5C), strongly suggesting that these factors have no direct influence on the expression of Il21.
The results presented here suggest that the Idd6.3 locus, containing the circadian rhythm–related Arntl2 gene, controls the development of IL21-producing CD4+ T cells in the thymus of the NOD mouse. This is underlined by the finding that mice carrying NOD alleles at Idd6.3 show a higher number of CD4+IL21+ T cells in the thymus and the spleen than mice carrying C3H alleles at Idd6.3. Mice carrying NOD alleles at Idd6.3 show increased number of splenic CD4+ and CD8+ T cells and reduced numbers of B cells and macrophages. This observation may be directly linked to the increased number of IL21-producing cells and their effect on the expansion of T cells (18). The differences in IL21 expression can also, at least partly, explain the differences in diabetogenic activity between 6.VIIIa- and 6.VIIIc-derived splenocytes. The effects on IL21 production and T-cell expansion occur in younger mice and disappear with age, and interestingly, diabetes incidence of 6.VIIIa and 6.VIIIc mice was found to be similar (10).
Our transcriptome studies suggest that ARNTL2 controls the expression of the Il21 gene and that high levels of ARNTL2 suppress Il21 expression. Knowing the role of IL21 in the immune system, the expansion of T cells, and in particular as being a major factor in T1D (19–22), our previous results become explainable (14,15). The decrease in diabetogenic activity of Arntl2-overexpressing splenocytes could be related to decreased IL21 production. Because Arntl2 is ubiquitous and IL21 is a pleiotropic cytokine, effects on other cell types in our mouse model must be expected and remain to be studied, and ARNTL2 may target other genes that have not yet been revealed by our analysis.
Targeting of the promoter of Il21 by ARNTL2 at the RNA polymerase binding site appears to be allele-specific, and the targeted DNA sequence contains potential E-box sequences. Although we detected binding of proteins expressed from Arntl2 C3H alleles, no binding was found for NOD alleles. Our ChIP experiments indicate that ARNTL2 acts as a transcriptional repressor of Il21, but whether ARNTL2 itself, its protein partners, or other mechanisms are involved is not clear at this point. Our controls did not indicate that the antibodies for ChIP experiments discriminate between the allelic versions; however, we cannot exclude the existence of a NOD-specific isoform that is undetectable by our antibodies.
It is interesting to note that the circadian profile of Il21 resembles that of the NOD allelic version of Arntl2 in 6.VIIIa mice but not the C3H allelic version of Arntl2 in 6.VIIIc mice. Il21 was constantly downregulated during daytime in 6.VIIIc mice compared with 6.VIIIa mice. One interpretation of this observation would be that both genes are under a common control in NOD mice, either of the peripheral or central circadian regulatory system. Interestingly, CLOCK and ARNTL1 were not found to bind to the promoter of Il21, and this makes a direct regulation through these factors in CD4+ T cells rather unlikely.
We analyzed the expression of other circadian genes in splenic CD4+ T cells of both congenic strains. Several of them exhibited expression differences. At 1700 h (5:00 p.m.) period 1 (Per1) was found upregulated in 6.VIIIa and period 2 (Per2) in 6.VIIIc. Clock was found upregulated in 6.VIIIa at 1300 h (1:00 p.m.) and in 6.VIIIc at 1700 h (5:00 p.m.). No changes in the circadian rhythm were observed for Arntl1, Arnt, Per1, and Per2, suggesting that there is no general difference in the expression of circadian rhythm–associated genes (Supplementary Fig. 1). However, for the Clock gene we found a shift of the peak of expression in strain 6.VIIIc compared with strain 6.VIIIa. This may, in turn, again influence the circadian expression of Arntl2 because CLOCK binds to the promoter of the Arntl2 gene (Supplementary Fig. 3). It is interesting to note that Arntl1, Per1, and Per2 have a similar profile in expression in both strains as Arntl2 in strain 6.VIIIa. Changes in the circadian rhythm of Arntl2 may be due to its sequence alteration in 6.VIIIc mice. ARNTL1 and ARNTL2 bind to the Arntl2 gene but with a different profile between the two strains. This may not be surprising because a number of polymorphisms between the NOD and C3H sequences are detectable in the Arntl2 gene (13). These observations point to an interest of further detailed studies on the circadian circuit in immune cells in the future.
In summary, our results suggest a novel role for ARNTL2 in the control of the immune system and in the development of T1D, which is not necessarily related to its proposed action in the circadian regulation. These findings reinforce the evidence that components of the immune system undergo circadian rhythm control (23).
Acknowledgments. The authors thank Philip Avner (EMBL, Italy) and Christian Boitard (Institut Cochin, France) for continuous support during this project; Agnès Lehuen and Yannick Simoni (both at Institut Cochin, France) for helpful discussions; Guillaume Soubigou (Institut Pasteur, France) for microarray processing; and Chantal Becourt (Institut Cochin, France) and Corinne Veron and Pierre Alcan (Institut Pasteur, France) for technical assistance.
Funding. This work was supported by grants from the Agence Nationale de la Recherche, the Association pour la Recherche sur le Diabète, and the European Foundation for the Study of Diabetes/JDRF/Novo Nordisk Programme, Cardiovasculaire-Obésité-Rein-Diabète, and by recurrent funding from the Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale, and Institut Pasteur.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. B.L. and U.C.R. researched data and wrote the manuscript. C.H. researched data and reviewed the manuscript. U.C.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1702/-/DC1.
- Received November 6, 2013.
- Accepted February 3, 2014.
- © 2014 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.