Precocious expression of Blimp1 in B cells causes autoimmune disease with increased self‐reactive plasma cells

Abstract The transcription factor Blimp1 is not only an essential regulator of plasma cells, but also a risk factor for the development of autoimmune disease in humans. Here, we demonstrate in the mouse that the Prdm1 (Blimp1) gene was partially activated at the chromatin and transcription level in early B cell development, although mature Prdm1 mRNA did not accumulate due to posttranscriptional regulation. By analyzing a mouse model that facilitated ectopic Blimp1 protein expression throughout B lymphopoiesis, we could demonstrate that Blimp1 impaired B cell development by interfering with the B cell gene expression program, while leading to an increased abundance of plasma cells by promoting premature plasmablast differentiation of immature and mature B cells. With progressing age, these mice developed an autoimmune disease characterized by the presence of autoantibodies and glomerulonephritis. Hence, these data identified ectopic Blimp1 expression as a novel mechanism, through which Blimp1 can act as a risk factor in the development of autoimmune disease.


Apoptosis assays
Ex vivo apoptosis of pro-B and pre-B cells was assessed by flow cytometry by using the Violet Ratiometric Membrane Asymmetry Probe/Dead Cell Apoptosis Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Alternatively, ex vivo apoptosis was assessed by intracellular staining with an anti-cleaved Caspase 3 antibody (5A1E; Cell Signaling Technology).

BrdU labeling of B cells Prdm1
ihCd2/+ and Prdm1 +/+ mice at the age of 3 months were intraperitoneally injected with 100 µl of 10 mg/ml BrdU (in PBS) at day 0. At the same time, BrdU was added at a concentration of 1 mg/ml to the drinking water, and the BrdU-containing drinking water, which was protected from light, was exchanged every day. At day 10, the mice were either sacrificed or received normal drinking water (without BrdU) for the next 15 days. At day 10 or 25, the BrdU incorporation into immature B and FO B cells of the spleen was analyzed by flow cytometry. Incorporated BrdU was detected by intracellular staining with an anti-BrdU antibody using the APC BrdU Flow kit (BD Pharmingen).

In vitro culture of pro-B cells Prdm1
ihCd2/+ and wild-type pro-B cells were cultured on OP9 stromal cells in IMDM medium containing IL-7 as described (Nutt et al., 1997).

In vitro B cell stimulation experiments
Immature B cells (CD19 + B220 + IgM + IgD -) were sorted by flow cytometry from the bone marrow, and mature FO B cells were isolated from the spleen by immunomagnetic depletion of CD43expressing cells using MACS cell separation (Miltenyi Biotec). Mature cells were resuspended in stimulation medium (IMDM medium supplemented with 10% heat-inactivated FCS (GE Healthcare; A15-101), 1 mM glutamine and 50 μM β-mercaptoethanol) and were seeded at a density of 2 × 10 6 cells in 4 ml of stimulation medium into one well of a 6-well plate. Immature B cells were seeded at a density of 1.5 × 10 5 cells in 750 µl of stimulation medium (RPMI 1640 instead of IMDM plus further addition of 1 mM sodium pyruvate and 10 mM Hepes) into one well of a 24-well plate. The stimulation medium additionally contained the following reagents: 25 μg/ml LPS (L4130; Sigma-Aldrich) for LPS stimulation; 3 μM CpG (ODN 1826, InvivoGen) for CpG stimulation; 20 ng/ml IL-4 (made in-house), 10 ng/ml IL-5 (405-ML, R&D Systems) and 2 μg/ml anti-CD40 antibody (HM40-3, eBioscience) for IL-4, IL-5 plus anti-CD40 stimulation. At 60 hours (immature B cells) or 4 days (mature B cells) of stimulation, the relative frequency of CD138 + CD22 lo plasmablasts was determined by flow cytometric analysis.

Analysis of somatic hypermutation
Following isolation of genomic DNA from sorted splenic plasma cells of non-immunized 4month-old mice, the intronic region downstream of the J H 4 segment of the Igh locus was PCRamplified with the PfuTurbo DNA polymerase (Agilent) using the PfuUltra II hotstart master mix (Agilent) and primers described in Table EV3. The 564-bp PCR fragments were A-tailed by Taq polymerase (made in-house) and directly cloned using the pGEM-T Easy Vector System (Promega), followed by transformation of the E. coli strain DH5α. The inserted DNA of at least 85 clones per genotype was analyzed by Sanger sequencing, and mutations were identified by comparison with the wild-type downstream J H 4 sequence.

Nuclear extract preparation and immunoblot analysis
Nuclear extracts of short-term cultured pro-B cells were prepared as described (Minnich et al., 2016). The protein concentration of the nuclear extract was determined by Bradford assay (BioRad). The proteins of the nuclear extract were denatured in 2× SDS sample buffer, boiled, separated by SDS-PAGE and analyzed by immunoblot analysis.

Immunization, ELISPOT and ELISA analyses
The immune response to a T cell-dependent antigen was studied by intraperitoneal injection of 100 µg of 4-hydroxy-3-nitrophenylacetyl-conjugated keyhole limpet hemocyanin (NP-KLH; Biosearch Technologies) in alum. The frequencies of NP-specific IgM antibody-secreting cells (ASCs) were determined in the spleen by enzyme-linked immunospot (ELISPOT) assay, as described (Smith et al., 1997). NP 24 -BSA-coated plates were used for capturing total anti-NP-IgM antibodies secreted by individual cells. ASCs were incubated for 6 h at 37 ºC and 5% CO 2 , and the resulting spots were visualized with a goat anti-mouse IgM antibody conjugated to alkaline phosphatase (SouthernBiotech), and color was developed by the addition of BCIP/NBT Plus solution (SouthernBiotech). After extensive washing, the spots were counted with an AID ELISPOT reader system (Autoimmun Diagnostika). The serum titer of NP-specific IgM, IgG1 and IgG2b antibodies was determined by enzymelinked immunosorbant assay (ELISA) (Smith et al., 1997) by using ELISA plates (Sigma-Aldrich), which were coated with 25 µg/ml of NP 7 -BSA or NP 24 -BSA to capture high-affinity IgG1 or total NP-specific IgM, IgG1 and IgG2b antibodies, respectively. The serum concentration of NP-specific IgG1 was determined relative to that of a standard anti-NP IgG1 antibody (hybridoma SSX2.1).

ELISA measurements of autoantibodies
ELISA plates (Sigma-Aldrich), which were coated with mouse liver DNA and then blocked with 1% BSA, were incubated with mouse serum for 2 h at 22 ºC. Anti-DNA-specific antibodies were detected by incubation with horseradish peroxidase-conjugated goat anti-mouse IgG or goat anti-mouse IgM antibodies (both from SouthernBiotech) in the presence of the TMB substrate (Biolegend). The absorbance was measured at 650 nm using an Epoch Microplate Spectrophotometer (BioTek Instruments). Antibodies against cardiolipin were measured as described (Wermeling et al., 2010). Briefly, ELISA plates were coated with cardiolipin (Sigma-Aldrich) overnight. Serum was added after blocking, and antigen-reactive IgG and IgM were measured with alkaline phosphate-or horseradish peroxidase-conjugated anti-mouse antibodies (SouthernBiotech). Antibodies against SSA (Ro-52) and SSB (La) were measured using commercial kits (all from Signosis Inc) following the manufacturer's instruction. All samples were corrected for background binding.

Indirect immunofluorescence assay using HEp-2 slides
Diluted mouse serum (1:100 in PBS) was incubated on HEp-2 slides (Orgentech) for 30 min in the dark at 22 ºC using a humidity chamber. Subsequently, the slides were rinsed once with a squirt bottle and washed twice for 5 min with PBS. For detection of mouse IgM or IgG, the slides were incubated for 30 min in the dark at 22 ºC in a humidity chamber with an Alexa488-conjugated goat anti-mouse IgM antibody or an Alexa488-conjugated goat anti-mouse IgG (H+L) antibody (both from Thermo Fisher Scientific and diluted in PBS) as a secondary antibody. Following two washing steps, DAPI-containing mounting medium (Life Technologies) was added, and images were acquired with a Zeiss Axio Imager 2 microscope and were analyzed with the Fiji software.

Detection of IgG immune complexes
IgG immune complexes were detected on paraformaldehyde-fixed cryosections of kidneys by staining with an Alexa488-conjugared goat anti-mouse IgG (H+L) antibody (Thermo Fisher Scientific) followed by addition of DAPI-containing mounting medium (Life Technologies). The abundance of IgG immune complexes was quantified by determining the mean fluorescence intensity (MFI) of at least 12 individual glomeruli for each kidney. Image acquisition was performed with a LSM710 (Zeiss) confocal microscope, and images were analyzed with the Fiji software.

Histopathological analysis
For histopathological analyses, one kidney from each mouse was isolated, fixed in 4% paraformaldehyde, trimmed, dehydrated and processed with the Logos Tissue processor. Processed kidneys were embedded in paraffin, sectioned at a thickness of 2 µm and stained either with haematoxylin and eosin (H&E) or periodic acid-Schiff (PAS) stains. Stained slides were reviewed by a board-certified pathologist with a Zeiss Axioskop 2 MOT microscope (Carl Zeiss Microscopy) and representative microscopic images were acquired with a SPOT Insight digital camera (SPOT Imaging). From each kidney, 40 glomeruli were individually evaluated based on criteria adapted from the following references (Alperovich et al., 2007;Chowdhary et al., 2015;Weening et al., 2004). Each glomerulus was evaluated for the presence of the following microscopic lesions: active lesions -mesangial hypercellularity, fibrinoid necrosis; endocapillary hypercellularity, capillary basement membrane lesions (including wireloops, reduplication and focal granular sub-or supra-basement membrane deposits) and intracapillary hyaline thrombi as well as chronic lesions -mesangial sclerosis, mesangial crescents and capsular fibrous adhesions. Light microscopic lesions that were notably evident in kidneys from female Prdm1 ihCd2/+ mice included capillary basement membrane lesions (focal granular deposits, focal or segmental basement membrane thickening or reduplication) and mesangial sclerosis (expansion of the mesangium with PAS-positive matrix with narrowing of capillary lumina). Intracapillary hyaline thrombi were evident in a few mice as were obsolescent glomeruli. Tubular and interstitial lesions were not assessed, as they were not a prominent feature in any of the kidneys. Histopathologic scores were assigned to each glomerulus, based on the extent of involvement by one or more of the above lesions as follows: score 0 -within normal limits; score 1 -minimal (less than 10%); score 2 -mild, segmental (11% to 30%); score 3 -moderate, segmental (31% to 60%); score 4severe, segmental to global (greater than 60%), as documented in Table EV2.

RT-qPCR analysis of nascent transcripts and spliced mRNA
Total RNA was isolated from sorted pro-B cells, pre-B cells and in vitro LPS-stimulated plasmablasts by using the RNeasy Mini kit (Qiagen). Genomic DNA was eliminated by using an eliminator spin column (Qiagen). Reverse transcription was performed by using randon hexamer or oligo-dT primers (New England Biolabs) and SuperScript II reverse transcriptase (Life Technologies). Prdm1, Atg5 and Tbp nascent transcripts were analyzed by PCR amplification with primers located in intronic sequences (Table EV3), and the data were normalized to those obtained for nascent Tbp transcripts. The Prdm1 and Tbp mRNAs were analyzed by PCR amplification with primers located in different exons (Table EV3), and the data were normalized to the Tbp mRNA.
GRO-seq analysis CD19 + pro-B cells from the bone marrow of Rag2 -/mice and CD23 + FO B cell from the spleen of Cd23-Cre Ebf1 fl/+ or Cd23-Cre Ebf1 fl/mice were isolated by immunomagnetic enrichment using MACS cell separation (Miltenyi Biotec). The nuclei were prepared from approximately 10 million cells by incubation with nuclear preparation buffer (0.30 M sucrose, 10 mM Tris, pH 7.5, 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 0.1 mM EGTA, 0.1% NP40, 0.15 mM spermine, 0.5 mM spermidine and 2 mM 6AA) for 3 min and were then subjected to nuclear run-on for 5 min at 30 ˚C using BrUTP-containing NTPs, as described (Core et al., 2008). The reaction was stopped by the addition of TRIzol reagent. The RNA was isolated, fragmented and the nascent transcripts were isolated using anti-BrdU antibody-conjugated agarose beads (Santa Cruz Biotech, sc-32323ac). The purified nascent RNA was subjected to the small RNA library preparation procedure (Reimão-Pinto et al., 2015) with anti-BrdU antibody-mediated purification following each ligation step.

ChIP-qPCR analysis of histone modifications
Short-term cultured pro-B cells were used for ChIP analysis with an anti-H3K4me1 antibody (rabbit polyclonal; Abcam; ab8895) or an anti-H3K27ac antibody (rabbit polyclonal; Abcam; ab4729), as described (Schebesta et al., 2007). Different regions of the Prdm1 locus were analyzed by ChIP-qPCR analysis with specific primers (shown in Table EV3), as described (Minnich et al., 2016).

ChIP-seq analysis of Blimp1 binding
Chromatin of 1 × 10 8 in vitro cultured pro-B cells from Prdm1 ihCd2/+ mice was prepared using a lysis buffer containing 0.25% SDS and was then subjected to ChIP with anti-V5 agarose beads (Sigma-Aldrich), as described (Schebesta et al., 2007). The quantification of precipitated DNA was performed using qPCR, and about 1-5 ng of ChIP-precipitated DNA was used for library preparation and subsequent Illumina deep sequencing (Table EV4).
cDNA preparation for RNA-seq Total RNA from ex vivo sorted pro-B and pre-B cells was isolated with the RNeasy Plus Mini Kit (Qiagen), and mRNA was purified by two rounds of poly(A) selection with the Dynabeads mRNA purification kit (Invitrogen). The mRNA was fragmented by heating at 94 °C for 3 min in fragmentation buffer. The fragmented mRNA was used as template for first-strand cDNA synthesis with random hexamers and the Superscript Vilo First-Strand Synthesis System (Invitrogen). The second-strand cDNA synthesis was performed with 100 mM dATP, dCTP, dGTP and dUTP in the presence of RNase H, E. coli DNA polymerase I and DNA ligase (Invitrogen). The incorporation of dUTP allowed for specific elimination of the second DNA strand during library preparation, thereby preserving strand specificity (Parkhomchuk et al., 2009).

Library preparation and Illumina Deep Sequencing
About 1-5 ng of cDNA or ChIP-precipitated DNA was used as starting material for the generation of sequencing libraries with the NEBNext Ultra Ligation Module and NEBNext End Repair/dA-Tailing module. DNA fragments of the following sizes were selected: 200-500 bp for ChIP-seq and 150-700 bp for RNA-seq with AMPure XP beads (Beckman Coulter). For strand-specific RNA-seq, the uridines present in one cDNA strand were digested with uracil-N-glycosylase (New England BioLabs) as described (Parkhomchuk et al., 2009), followed by PCR amplification with the KAPA Real Time Amplification kit (KAPA Biosystems). Completed libraries were quantified with the Bioanalyzer dsDNA 1000 assay kit (Agilent) and QPCR NGS Library Quantification kit (Agilent). Cluster generation and sequencing was carried out by using the Illumina HiSeq 2000 system with 50 nucleotides read length according to the manufacturer's guidelines.

Database of RefSeq-annotated genes
The database generation of RefSeq-annotated genes was performed as previously described (Wöhner et al., 2016). To refine the annotation of immunoglobulin genes, the immunoglobulin λ light-chain segments were replaced with their corresponding converted GRCm38.p3 annotations (Ensembl version 79; Yates et al., 2016). The resulting number of genes was 24,732.

Sequence alignment
In case of RNA-seq experiments, reads corresponding to mouse ribosomal RNAs (BK000964.1 and NR046144.1) were removed. The remaining reads were cut down to a read length of 44 nucleotides and aligned to the mouse transcriptome (genome assembly version of July 2007 NCBI37/mm9) using TopHat version 1.4.1 . In case of ChIP-seq, GRO-seq and ATAC-seq experiments, all sequence reads that passed the Illumina quality filtering were considered for alignment after adapter trimming. The remaining reads were aligned to the mouse genome assembly version of July 2007 (NCBI37/mm9), using the Bowtie program versions 0.12.1, 1.0.0 and 2.1.0, respectively (Langmead et al., 2009). For GRO-seq, additional four bases were eliminated after adapter trimming and filtered against the rDNA with Bowtie version 2.1.0 before mouse genome alignment. For ATAC-seq, additional alignment parameters were 'sensitive -X 5000'. Bio/Bio plasmablasts with a P value of < 10 -5 . These Blimp1 peaks were further filtered for a P value of < 10 -10 , which resulted in 762 peaks in pro-B cells and 9,320 peaks in plasmablasts.

Peak overlap analysis
The peak overlap analysis was performed with the Multovl program (Aszódi, 2012) by using a minimal overlap length of one bp and allowing for all possible overlaps.

Motif discovery analysis
Sequences +/-150 bp around the most significant MACS2 summit of the top 300 P value-ranked Blimp1 peaks have been used as input for the MEME-ChIP suite version 4.9.1 (Machanick and Bailey, 2011). The most significant motif was the Bimp1 motif with an E-value of 1.3 x 10 -379 .

Read density analysis
Read density profiles were calculated using jnomics (I. Tamir, unpublished).

Peak-to-gene assignment
Common and unique Blimp1 targets in Prdm1 ihCd2/+ pro-B cells and Prdm1 Bio/Bio plasmablasts were identified by peak-to-gene assignment as described (Revilla-i-Domingo et al., 2012). Peaks were assigned to genes in a stepwise manner by prioritizing genes containing peaks in their promoter and/or gene body. For this, peaks overlapping with the promoter (-2.5 kb to +2.5 kb relative to TSS) or gene body (+2.5 kb to TES) were first assigned to the corresponding gene.
Other peaks within a specified region of 50 kb upstream of the TSS or downstream of the TES were assigned to the gene containing peaks in the promoter or gene body. All other peaks within the same specified region were assigned to the nearest gene, and all non-assigned peaks were classified as intergenic.

Analysis of RNA-seq data
The number of reads per gene was counted using featureCounts version 1.5.0 (Liao et al., 2014) with default settings. TPM (transcripts per million) values were calculated as described (Wagner et al., 2012). For analysis of differential gene expression of Prdm1 ihCd2/+ and wild-type pro-B and pre-B cells, the datasets were grouped according to cell type, genotype and replicate number and were analyzed using the R package DESeq2 version 1.8.2 (Love et al., 2014). Wald tests were performed with the model design formula "~ replicate + type" (type is a linear combination of cell type and genotype). Sample normalizations and dispersion estimations were conducted using the default DESeq2 settings. Regularized log transformations were computed with the blind option set to 'FALSE' and were transformed from log 2 to log 10 scale for the scatterplots shown in Fig 4A,B. Genes with an adjusted P value < 0.05 and an absolute fold change > 3 as well as a mean TPM (averaged within conditions) > 5 were called as significantly expressed. Immunoglobulin and T cell receptor genes were filtered from the list of significantly expressed genes, but were included in the TPM calculations.

Statistical analysis
Statistical analysis was performed with the GraphPad Prism 7 software. The two-tailed Student's t-test analysis was used to assess the statistical significance of differences between two experimental groups in all experiments, with the exception of those involving NGS-based approaches and ELISA measurement. The statistical evaluation of the RNA-seq data is described above (Analysis of RNA-seq data). The ELISA data were analyzed using the Mann-Whitney test.

Data availability
RNA-seq, ChIP-seq and GRO-seq data (Table EV4), which are first reported in this study, are available at the Gene Expression Omnibus (GEO) repository under the accession numbers GSE111692. Previously published ATAC-seq, ChIP-seq and RNA-seq datasets, which were used in this study, are available at the GEO repository under the accession numbers indicated in Table  EV4.
(A) Schematic diagram of the human ATG5-PRDM1 region. The exon-intron structures of both genes are shown together with the positions of the two SNPs (rs6568431 and rs548234) that have been associated with human SLE and RA (Gateva et al., 2009;Raychaudhuri et al., 2009;Zhou et al., 2011). Each dot corresponds to one mouse.
(A) Induction of active chromatin at the 3' end of the Prdm1 ihCd2 gene by the inserted MoMLV enhancer. ChIP analysis with H3K4me1-or H3K27ac-specific antibodies was used to determine the abundance of active chromatin (H3K4me1 and H3K27ac) at 7 different regions of the Prdm1 locus in short-term cultured wild-type (WT) and Prdm1 ihCd2/+ pro-B cells. Input and precipitated DNA were quantified by qPCR with primers amplifying the indicated regions, shown below a schematic diagram of the Prdm1 gene, or the promoter of the ubiquitously expressed control Tbp gene. The amount of precipitated DNA was determined as percentage relative to input DNA for each region analyzed and is shown as relative enrichment at the Prdm1 region compared to the Tbp promoter (set as 1). Average values with SEM are shown for two independent experiments. An inactive region downstream of the Cd19 gene (Cd19ds) was analyzed as negative control. Site A corresponds to the upstream region A shown in Fig  (white) mice. Bar graphs show absolute cell numbers for each cell type and indicated genotype. The different cell types were defined as described in detail in the Appendix Supplementary Methods. Statistical data (A-F) are shown as mean value with SEM and were analyzed by the Student's t-test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Each dot corresponds to one mouse.
Appendix Figure S4. Blimp1-dependent deregulation of the B cell gene expression program.
Multiple overlap of Blimp1-activated (left) or Blimp1-repessed (right) genes, which were identified in Prdm1 ihCd2/+ pro-B and pre-B cells (as described in Fig 4A,B) as well as in wild-type pre-plasmablasts (as described in Fig 1g; Minnich et al., 2016). and repressed Blimp1-bound (right) target genes, which were determined in Prdm1 ihCd2/+ pro-B and pre-B cells (as described in Fig 4F) as well as in wild-type pre-plasmablasts (as indicated in  (Fuxa and Busslinger, 2007). Different B cell subsets (gray) and T cells (black) of Pax5 iGfp/+ mice were analyzed as positive or negative control for Pax5 expression.
GFP expression is shown as a histogram for pre-B cells (left) or as bar graphs (right) for all B cell subsets relative to the GFP expression determined for the control Pax5 iGfp/+ genotype (set as 1).
(G) Expression of selected Blimp1-activated and Blimp1-repressed genes coding for cell surface receptors. Blimp1-bound genes are underlined. The mRNA expression of the indicated genes is shown as mean expression value (TPM) with SEM, based on two different RNA-seq experiments for the pre-B cells of each genotype.
Appendix Figure S5. Impaired GC B cell formation and increased plasmablast differentiation in Prdm1 ihCd2/+ mice.
(A) The number of total cells in the spleen (left) and bone marrow (bone marrow) of nonimmunized Prdm1 ihCd2/+ (red) and wild-type (WT, gray) mice at the age of 2, 4 and 12 months was determined by flow cytometry. These absolute cell numbers correspond to the data shown in Fig 5A (spleen) and Fig 5B (bone marrow). (B) Presence of IgM and IgG antibodies in the serum of non-immunized Prdm1 ihCd2/+ (red dots) and wild-type (gray dots) mice at the age of 2, 4 and 12 months. The titers (mg/ml) of total IgM and IgG antibodies were determined by ELISA and correspond to the mice analyzed in Fig 5A, control Eβ -/-Prdm1 +/+ (gray) mice. Flow cytometric analysis of plasma cells from the bone marrow (C) and spleen (D) of the indicated genotypes is shown to the left, and bar graphs indicate the absolute cell numbers of plasma cells in each organ to the right. Statistical data (A-D) are shown as mean value with SEM and were analyzed by the Student's t-test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Each dot corresponds to one mouse.
after the last injection. (B) Presence of IgM and IgG antibodies detecting dsDNA, cardiolipin, SSA (Ro-52) and SSB (La) in the serum of injected and non-injected mice at the age of 10 weeks. The titers of the different IgM and IgG antibodies in the serum of Prdm1 ihCd2/+ (red dots) and wild-type (gray dots) mice were determined by ELISA and are displayed as arbitrary units (AU). The serum of 6-month-old Fasl gld/gld mice was used as positive control. Statistical data are shown as mean value with SEM and were analyzed by the Mann-Whitney test; *P < 0.05. Each dot corresponds to one mouse. (C) Detection of anti-nuclear antibodies (ANA) of the IgM and IgG isotype in the serum of injected and non-injected Prdm1 ihCd2/+ or wild-type mice. ANA staining was performed with the serum of the indicated mice by indirect immunofluorescence assay on HEp-2 cells with an Alexa488-conjugated anti-mouse IgM or IgG antibody, respectively. The serum of Fasl gld/gld mice was used as positive control.