Identification of a Novel Role for Foxo3 Isoform2 in Osteoclastic Inhibition

Cheng Xu, Gregory J. Vitone, Kazuki Inoue, Courtney Ng and Baohong Zhao
J Immunol October 15, 2019, 203 (8) 2141-2149; DOI: https://doi.org/10.4049/jimmunol.1900707

Key Points

  • This study reveals the first (to our knowledge) known biological function of Foxo3 isoform2 in bone.

  • Foxo3 isoform2 is an activator for IFN-I response and an osteoclastic inhibitor.

  • Foxo3 isoform2 mice exclusively expressing isoform2 show osteopetrosis phenotype.

Abstract

Foxo3 acts as an important central regulator that integrates signaling pathways and coordinates cellular responses to environmental changes. Recent studies show the involvement of Foxo3 in osteoclastogenesis and rheumatoid arthritis, which prompted us to further investigate the FOXO3 locus. Several databases document FOXO3 isoform2, an N-terminal truncated mutation of the full-length FOXO3. However, the biological function of FOXO3 isoform2 is unclear. In this study, we established a conditional allele of Foxo3 in mice that deletes the full-length Foxo3 except isoform2, a close ortholog of the human FOXO3 isoform2. Expression of Foxo3 isoform2 specifically in macrophage/osteoclast lineage suppresses osteoclastogenesis and leads to the osteopetrotic phenotype in mice. Mechanistically, Foxo3 isoform2 enhances the expression of type I IFN response genes to RANKL stimulation and thus inhibits osteoclastogenesis via endogenous IFN-β–mediated feedback inhibition. Our findings identify, to our knowledge, the first known biological function of Foxo3 isoform2 that acts as a novel osteoclastic inhibitor in bone remodeling.

This article is featured in In This Issue, p.2027

Introduction

Osteoclasts, derived from monocyte/macrophage precursors, are the exclusive cell type responsible for bone resorption in both bone homeostasis and pathological bone destruction. Bone loss is a major cause of morbidity and disability in many skeletal diseases, such as rheumatoid arthritis (RA), psoriatic arthritis, periodontitis, and periprosthetic loosening (1–3). Osteoclastogenesis is induced by the major osteoclastogenic cytokine receptor activator of NF-κB ligand (RANKL). Binding of RANKL to RANK receptors activates a broad range of signaling cascades, including canonical and noncanonical NF-κB pathways, MAPK pathways, and calcium signaling, which lead to the activation of an osteoclastic transcriptional network. The positive regulators in this transcriptional network, such as the transcription factors NFATc1, c-Fos, and Blimp1, drive osteoclast differentiation (1–4). In contrast, the process of osteoclast differentiation is delicately controlled by a “braking system,” in which negative regulators, such as IFN regulatory factor (Irf) 8, recombination signal binding protein for Ig κ J region (RBP-J), and differentially expressed in FDCP 6 homolog (Def6), restrain osteoclastogenesis to prevent excessive bone resorption (5–9). Thus, the extent of osteoclastogenesis is delicately modulated and determined by the balance between these osteoclastogenic and antiosteoclastogenic mechanisms.

Forkhead box class O (Foxo) proteins are a family of evolutionarily conserved transcription factors, which include Foxo1, 3, 4, and 6 in mammals. Foxo proteins consist of four conserved regions: a forkhead DNA-binding domain at the N terminus followed by a nuclear localization signal, a nuclear export signal, and a transactivation domain at the C terminus (10–12). Foxo proteins play important roles in diverse biological processes, such as metabolism, oxidative stress, cell cycle regulation, apoptosis, immunity, and inflammation. Foxo proteins are well known for their cell type– and context-specific effects on cellular processes because of their variable posttranslational modifications, subcellular localization, and binding cofactors in different scenarios (10–15). Foxo1, 3, and 4 were reported to regulate RANKL-induced osteoclast differentiation (16, 17). However, Foxo proteins seem to exhibit different functions in osteoclastogenesis. For example, some studies show that Foxo1, 3, and 4 proteins as a group are inhibitors of osteoclastogenesis (16), whereas others found that Foxo1 is a positive regulator (17). These results indicate that Foxo family plays an important but complex role in osteoclastogenesis. In disease settings, FOXO3 activity is correlated with outcomes in infectious and inflammatory diseases, such as RA. Increased expression of FOXO3 in monocytes due to a single-nucleotide polymorphism (FOXO3 [rs12212067: T>G]) is associated with reduced severity of RA (18, 19). Recently, we uncovered that Foxo3 is a target of miR-182 and plays an inhibitory role in inflammatory cytokine TNF-α–induced osteoclastogenesis and bone resorption (20). Thus, FOXO3 is closely involved in osteoclastogenesis and bone erosion in human RA. We further investigated the FOXO3 locus and found that there exists a short isoform of human FOXO3, named as isoform2 in contrast to the full-length isoform1. The current database supports the presence of FOXO3 isoform2 in human cells and tissues, such as fibroblasts and skeletal muscles, in physiological conditions (https://gtexportal.org/home/transcriptPage). The biological function of this FOXO3 isoform2 is unclear.

We primarily wished to take advantage of the conditional Foxo3 knockout (KO) mice (Foxo3f/f; LysMcre) to provide genetic evidence for the function of Foxo3 in vivo in osteoclastogenesis. However, to our surprise, these conditional KO mice express a truncated Foxo3 protein in addition to the lack of full-length protein. Sequence analysis demonstrated that this truncated Foxo3 is an ortholog of the human FOXO3 isoform2. Given this similarity, we named this truncated Foxo3 as mouse Foxo3 isoform2.

Over 90% of human genes are alternatively spliced to produce mRNA and protein isoforms, which may have shared, related, distinct, or even antagonistic functions. Alternative splicing is an essential biological process driving evolution and development. The isoforms resulting from alternative splicing contribute to transcriptomic and proteomic diversity and complexity in physiological conditions (21, 22). Aberrant splicing or deregulated isoform expression/function can lead to diseases, such as cancer and cardiovascular and metabolic diseases (21–23). Recent efforts have been made to investigate deregulated alternative splicing that could be used as diagnostic markers or therapeutic targets for diseases.

In this study, we identified Foxo3 isoform2 as a novel osteoclastogenic suppressor that leads to an osteopetrotic phenotype in mice. Foxo3 inhibits osteoclast differentiation through type I IFN–mediated feedback inhibition. These Foxo3f/f mice could be used as a model to investigate the function of human FOXO3 isoform2 because of the high protein sequence homology between human and mouse Foxo3. To the best of our knowledge, this is the first report unveiling the biological function of Foxo3 isoform2, which provides novel knowledge and research tools and opens new avenues for studying the function of Foxo3 isoform2 in different scenarios and areas.

Materials and Methods

Mice and analysis of bone phenotype

We generated mice with myeloid-specific expression of mouse Foxo3 isoform2 (full-length Foxo3 is replaced by the isoform2) by crossing Foxo3flox/flox mice (stock no. 024668; The Jackson Laboratory) with a lysozyme M promoter-driven Cre transgene on the C57BL/6 background (known as LysMcre; The Jackson Laboratory). Gender- and age-matched Foxo3flox/floxLysMcre+ mice (referred to as Foxo3isoform2) and their littermates with Foxo3+/+LysMcre+ genotype as wild-type controls (WT) were used for experiments. Global Foxo3−/− were purchased from The Jackson Laboratory (stock no. 022097). We maintained all mice under standard 12 h light/dark cycles with ad libitum access to regular food and water. All animal studies were approved by the Hospital for Special Surgery Institutional Animal Care and Use Committee and Weill Cornell Medical College Institutional Animal Care and Use Committee.

Microcomputed tomography analysis was conducted to evaluate bone volume and three-dimensional bone architecture using a SCANCO μCT-35 scanner (SCANCO Medical) as described (24). Twelve-week-old male mouse femora were fixed in 10% buffered formalin and scanned at 6 μm resolution. Proximal femoral trabecular bone parameters were analyzed using SCANCO software, according to the manufacturer’s instructions and the American Society of Bone and Mineral Research guidelines.

Cell culture

Mouse bone marrow cells were harvested from tibiae and femora of age- and gender-matched mutant and control mice and cultured for 3 d in Petri dishes (Corning Falcon, no. 351029) in α-MEM (Thermo Fisher Scientific) with 10% FBS (S11550; Atlanta Biologicals), glutamine (2.4 mM, no. 25030164; Thermo Fisher Scientific), 1% penicillin/streptomycin (no. 15070063; Thermo Fisher Scientific), and L929 cell supernatant (conditioned medium [CM]), which contained the equivalent of 20 ng/ml of murine rM-CSF and was used as a source of M-CSF. The attached bone marrow–derived macrophages (BMMs) were scraped, seeded at a density of 4.5 × 104 cells per cm2, and cultured in α-MEM with 10% FBS, 1% penicillin/streptomycin, 1% glutamine, and CM for overnight. The cells were then treated without or with an optimized concentration of RANKL (40 ng/ml, no. 310-01; PeproTech) in the presence of CM for times indicated in the figure legends. Culture media were exchanged after 3 d. For RAW264.7 cell culture, 5 × 102 cells were seeded per well in 96-well plates in α-MEM with 10% FBS, 1% penicillin/streptomycin, and 1% glutamine overnight. The cells were then treated without or with RANKL (100 ng/ml, no. 310-01; PeproTech). Culture media were exchanged every 2 d. Tartrate-resistant acid phosphatase (TRAP) staining was performed with an acid phosphatase leukocyte diagnostic kit (Sigma-Aldrich), in accordance with the manufacturer’s instructions.

Plasmids, cloning, and sequencing

cDNA fragments encoding mouse full-length Foxo3 protein or exon 2 fused with FLAG tag at the C terminus was amplified by PCR using the cDNA templates from WT BMMs and then subcloned into the XbalI/BamHI sites of pcDNA3.1+ vector to construct the pcDNA3.1+ full-length Foxo3-Flag plasmid or pcDNA3.1+-Foxo3 exon 2-Flag plasmid, respectively. Furthermore, cDNA fragment encoding mouse Foxo3 isoform2 fused with FLAG tag at the C terminus was amplified by PCR using the cDNA templates from Foxo3isoform2 BMMs, followed by subcloning into the XbalI/BamHI sites of pcDNA3.1+ vector to construct the pcDNA3.1+-Foxo3 isoform2-Flag plasmid. The following primers were used for cloning: for Foxo3 full-length fragment, forward 5′-ATTCTAGAGCCACCATGGCAGAGGCACCAGCC-3′, reverse 5′-ATGGATCCTCACTTGTCGTCATCGTCTTTGTAGTCGCCTGGTACCCAGCTTTGA-3′; for exon 2 of Foxo3 fragment, forward 5′-ATTCTAGAGCCACCATGGCAGAGGCACCAGCC-3′, reverse 5′-ATGGATCCTCACTTGTCGTCATCGTCTTTGTAGTCCTTCCAGCCCGCAGAGCT-3′; and for Foxo3 isoform2 fragment, forward 5′-ATTCTAGAGCCACCATGCGCGTTCAGAATGAAGG-3′, reverse 5′-ATGGATCCTCACTTGTCGTCATCGTCTTTGTAGTCGCCTGGTACCCAGCTTTGA-3′. The sequence integrity of the inserted fragments in each expression plasmid was verified by restriction enzyme digestion and DNA sequencing at Cornell University Genomics Facility.

Transfection of human embryonic kidney 293 cells and RAW264.7 cells

Lipofectamine 3000 reagent (L3000015; Thermo Fisher Scientific) was used for the transfection of the human embryonic kidney (HEK) 293 cells or RAW264.7 cells. Briefly, the cells were seeded (2.5 × 105 HEK cells/well and 1.2 × 105 RAW264.7 cells/well) and cultured with DMEM for HEK293 cells or α-MEM for RAW264.7 cells supplemented with 10% FBS and 1% penicillin/streptomycin in a 24-well plate at 37°C in a humidified atmosphere containing 5% CO2 overnight. The cells were then transfected with 500 ng plasmid DNAs using Lipofectamine 3000 reagent, according to the manufacturer’s instructions. After 24 h, the medium was replaced with fresh completed DMEM for HEK293 cells or α-MEM for RAW264.7 cells. The protein lysates from cell cultures were collected after 48 h to assess plasmid expression.

In vitro gene silencing by small interfering RNAs

In vitro gene silencing by small interfering RNAs (siRNAs) was performed as previously described (20). Briefly, siRNAs targeting Foxo3 or their corresponding control oligos (80 nM) were transfected into murine BMMs using TransIT-TKO transfection reagent (Mirus Bio), in accordance with the manufacturer’s instructions.

RNA sequencing and bioinformatics analysis

RNA sequencing (RNA-seq) and bioinformatics analysis were performed as previously described (24). Briefly, total RNA was extracted using RNeasy Mini Kit (QIAGEN) following the manufacturer’s instructions. TruSeq RNA Library preparation kits (Illumina) were used to purify poly-A+ transcripts and generate libraries with multiplexed barcode adaptors, following the manufacturer’s instructions. All samples passed quality control analysis using a Bioanalyzer 2100 (Agilent Technologies). RNA-seq libraries were constructed per the Illumina TruSeq RNA sample preparation kit. High-throughput sequencing was performed using the Illumina HiSeq 4000 in the Weill Cornell Medical College Genomics Resources Core Facility. RNA-seq reads were aligned to the mouse genome (mm10) using TopHat (25). Cufflinks (26) was subsequently used to assemble the aligned reads into transcripts and then estimate the transcript abundances as reads per kilo base per million values. HTseq (27) was used to calculate raw reads counts, and edgeR (28) was used to calculate normalized counts as counts per million. Heatmaps were generated by pheatmap package in R. RNA-seq data (accession no. GSE 135479) have been deposited in National Center for Biotechnology Information’s Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE 135479).

Reverse transcription and real-time PCR

Reverse transcription and real-time PCR were performed as previously described (24). DNA-free RNA was obtained with the RNeasy Mini Kit (no. 74106; QIAGEN, Valencia, CA) with DNase treatment, and 1 μg of total RNA was reverse transcribed using a First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA), according to the manufacturer’s instructions. Real-time PCR was done in triplicate with the QuantStudio 5 Real-time PCR System and Fast SYBR Green Master Mix (Thermo Fisher Scientific). Gene expression was normalized relative to GAPDH. The primers for real-time PCR were as follows: Acp5: 5′-ACGGCTACTTGCGGTTTC-3′ and 5′-TCCTTGGGAGGCTGGTC-3′; Dcstamp: 5′-TTTGCCGCTGTGGACTATCTGC-3′ and 5′-AGACGTGGTTTAGGAATGCAGCTC-3′; Ctsk: 5′-AAGATATTGGTGGCTTTGG-3′ and 5′-ATCGCTGCGTCCCTCT-3′; Itgb3: 5′-CCGGGGGACTTAATGAGACCACTT-3′ and 5′-ACGCCCCAAATCCCACCCATACA-3′; Calcr: 5′-ACATGATCCAGTTCACCAGGCAGA-3′ and 5′-AGGTTCTTGGTGACCTCCCAACTT-3′; Foxo3-F3R3: 5′-CTGTCCTATGCCGACCTGAT-3′ and 5′-CTGTCGCCCTTATCCTTGAA-3′; Foxo3-F4R4: 5′-ATGGGAGCTTGGAATGTGAC-3′ and 5′-TTAAAATCCAACCCGTCAGC-3′; Foxo3-F5R5: 5′-AGGAGGAGGAATGTGGAAGG-3′ and 5′-CCGTGCCTTCATTCTGAAC-3′; Ifnb1: 5′-TTACACTGCCTTTGCCATCC-3′ and 5′-AGAAACACTGTCTGCTGGTG-3′; Mx1: 5′-GGCAGACACCACATACAACC-3′ and 5′-CCTCAGGCTAGATGGCAAG-3′; Ifit1: 5′-CTCCACTTTCAGAGCCTTCG-3′ and 5′-TGCTGAGATGGACTGTGAGG-3′; Irf7: 5′-GTCTCGGCTTGTGCTTGTCT-3′ and 5′-CCAGGTCCATGAGGAAGTGT-3′; Ifit2: 5′-AAATGTCATGGGTACTGGAGTT-3′ and 5′-ATGGCAATTATCAAGTTTGTGG-3′; Stat1: 5′-CAGATATTATTCGCAACTACAA-3′ and 5′-TGGGGTACAGATACTTCAGG-3′; and Gapdh: 5′-ATCAAGAAGGTGGTGAAGCA-3′ and 5′-AGACAACCTGGTCCTCAGTGT-3′.

Immunoblot analysis

Total cell extracts were obtained using lysis buffer containing 150 mM Tris-HCl (pH 6.8), 6% SDS, 30% glycerol, and 0.03% bromophenol blue; 10% 2-ME was added immediately before harvesting cells. Cell lysates were fractionated on SDS-PAGE, transferred to Immobilon-P membranes (MilliporeSigma), and incubated with specific Abs. Western Lightning Plus-ECL (PerkinElmer) was used for detection. Foxo3 N-terminal (no. 2497, specifically recognizing the residues surrounding Glu50 in exon 2 of Foxo3) and C-terminal (no. 12829S, specifically recognizing the C terminus of Foxo3) Abs were purchased from Cell Signaling Technology. Anti-Flag tag Ab (no. 637301) was purchased from BioLegend. p38α (sc-535) Ab was from Santa Cruz Biotechnology.

Statistical analysis

Statistical analysis was performed using GraphPad Prism software. Two-tailed Student t test was applied if there were only two groups of samples. In the case of more than two groups of samples, one-way ANOVA was used with one condition, and two-way ANOVA was used with more than two conditions. ANOVA analysis was followed by post hoc Bonferroni correction for multiple comparisons. A p value <0.05 was taken as statistically significant: *p < 0.05 and **p < 0.01. Data are presented as the mean ± SD, as indicated in the figure legends.

Results

Absence of Foxo3 enhances osteoclastogenesis

To provide genetic evidence for the role of Foxo3 in osteoclasts, we first took advantage of Foxo3 global KO mice, in which the Foxo3 protein is completely deleted. We first used BMMs as osteoclast precursors to examine in vitro osteoclast differentiation in response to RANKL, the master osteoclastogenic inducer. We found that Foxo3 KO–derived BMMs showed an increased responsiveness to RANKL, determined by more TRAP-positive multinucleated osteoclasts (Fig. 1A, 1B). Furthermore, we performed an RNA-seq experiment using WT and Foxo3 KO BMMs to examine gene expression in response to RANKL. In parallel with increased osteoclast formation, the expression of osteoclastic genes, such as Nfatc1 (encoding NFATc1), Prdm1 (encoding Blimp1), Acp (encoding TRAP), Oscar (encoding OSCAR), and Ctsk (encoding cathepsin K), was significantly enhanced by RANKL in Foxo3 KO BMM cultures compared with the BMMs cultured from WT controls (Fig. 1C). These results indicate that Foxo3 functions as a negative regulator in RANKL-induced osteoclast differentiation.

FIGURE 1.
FIGURE 1.

RANKL-induced osteoclast differentiation is enhanced by Foxo3 deficiency. BMMs derived from WT control and Foxo3 KO mice were stimulated with RANKL for 4 d. TRAP staining was performed (A), and the number of TRAP-positive multinucleated cells per well is shown in (B). TRAP-positive cells appear red in the photographs. Scale bar, 100 μm. Data are representative of three independent experiments. (C) Heat map of RANKL-induced osteoclastic gene expression enhanced by Foxo3 deficiency. Row z-scores of CPMs of osteoclast genes are shown in the heat map. **p < 0.01.

Foxo3f/f;LysMcre mice express a truncated Foxo3 protein that is an ortholog of human FOXO3 isoform2

We next wished to examine the role of Foxo3 in vivo using conditional Foxo3 KO mice. We deleted Foxo3 (encoding Foxo3) in myeloid lineage osteoclast precursors by crossing Foxo3flox/flox mice (The Jackson Laboratory) with LysMcre mice that express Cre under the control of the myeloid-specific lysozyme M promoter. We used Foxo3flox/flox; LysMcre+ mice and littermate controls with a Foxo3+/+; LysMcre+ genotype (hereafter referred to as WT) in the experiments. The mouse Foxo3 gene has four exons, and the coding region within exons 2 and 3 produces a full-length Foxo3 protein with 672 aa. The Foxo3flox/flox mice (The Jackson Laboratory) possess loxP sites flanking exon 2 of the Foxo3 gene (Fig. 2A). To verify Foxo3 deletion, we first designed a series of PCR primers that cover the coding region from exon 2 and exon 3 (Fig. 2B, Table I). As shown in Fig. 2C, PCR products were detected in WT BMM cDNAs using all primer sets. As expected, the exon 2–3 primer set did not produce any PCR bands using the Foxo3flox/flox; LysMcre+ BMM cDNAs. Surprisingly, other primer sets covering exon 3 or exon 3–4 generated the same PCR products using BMM cDNAs obtained from either Foxo3flox/flox; LysMcre+ or WT mice (Fig. 2C). We further designed quantitative PCR primer sets and found that the primers other than those located within exon 2 amplified the Foxo3 cDNAs in Foxo3flox/flox; LysMcre+ BMMs (Fig. 2D). These results indicate that there exists a truncated Foxo3 mRNA transcript in the Foxo3flox/flox; LysMcre+ mice. Interestingly, the primers located within exon 1 and exon 3 (F5 and R5 primers) were also able to generate PCR products shorter than 300 bp, strongly implying that this truncated Foxo3 mRNA is transcribed from exon 1, skips exon 2, and is elongated to exon 3. To directly demonstrate this, we cloned Foxo3 transcripts from WT or Foxo3flox/flox; LysMcre+ BMMs using a primer set (Exon 1F and Exon 3R in Fig. 2E, 2F) that covers WT Foxo3 mRNA starting from the transcription start site in exon 1 to the end of the coding sequence in exon 3. As shown in Fig. 2E, we detected the normal junction between exon 1 and exon 2 in WT BMMs (Fig. 2E). However, the entire exon 2 was absent, and a novel exon 1 to exon 3 junction was present in Foxo3flox/flox; LysMcre+ BMMs (Fig. 2F). These results confirm the presence of a novel Foxo3 mRNA with exon 2 truncated in the Foxo3flox/flox; LysMcre+ BMMs, resulting from an in-frame (nonframeshift) deletion by the cre-lox recombination in these mice. We next set off to detect the Foxo3 protein expression in the WT and Foxo3flox/flox; LysMcre+ BMMs. We used two Abs; one Ab recognizes the C-terminal region of Foxo3, whereas the other is an mAb that specifically targets the exon 2 of Foxo3. As shown in Fig. 2G, the full length of WT Foxo3 proteins were detected by both Abs in WT BMMs. In Foxo3flox/flox; LysMcre+ BMMs, the full length of Foxo3 proteins were deleted as expected. In contrast, a truncated Foxo3 protein (55 kDa) was detected by the C-terminal Ab in Foxo3flox/flox; LysMcre+ BMMs but not by the Ab specifically targeting exon 2. Furthermore, knockdown of Foxo3 by RNA interference completely deleted the truncated protein (55 kDa) in the Foxo3flox/flox; LysMcre+ BMMs (Fig. 2G, top panel). Taken together with the cloning data in Fig. 2F, these results demonstrate that the full-length Foxo3 protein is absent, but there exists an exon 2–truncated Foxo3 protein in Foxo3flox/flox; LysMcre+ BMMs.

FIGURE 2.
FIGURE 2.

Foxo3f/f;LysMcre (Foxo3isoform2) mice express a truncated Foxo3 protein that is an ortholog of human FOXO3 isoform2. (A) Molecular structure of mouse Foxo3 and Loxp sites. (B) PCR primer locations in Foxo3. (C) Foxo3 gene expression detected in WT and Foxo3f/f;LysMcre BMMs by PCR using the indicated primer sets whose locations are shown in (B). n = 5 per group. (D) Foxo3 gene expression detected in WT and Foxo3f/f;LysMcre BMMs by quantitative PCR using the indicated primer sets whose locations are shown in (A). (E and F) DNA sequences of the cloned Foxo3 transcripts from WT or Foxo3f/f;LysMcre BMMs using a primer set (Exon 1F and Exon 3R) that covers WT Foxo3 mRNA starting from the transcription start site in exon 1 to the end of the coding sequence in exon 3. (G) Foxo3 protein expression detected in WT and Foxo3f/f;LysMcre BMMs by Western blot using Abs recognizing C terminus or exon 2 of Foxo3, respectively. p38 was used as a loading control. All the primer sequences are shown in Table I.

View this table:
Table I. Sequences of regular PCR primers

When we investigated the human FOXO3 locus, we found annotations for a short isoform of FOXO3 (Supplemental Fig. 1A), which is named as isoform2 (RefSeq gene database, Ensembl genome database, and Uniprot Knowledgebase). The full length of FOXO3 is named as isoform1, which contains 673 aa. The human full-length FOXO3 isoform1 has two subisoforms (1a and 1b), which have an identical coding sequence with variable 5′ untranslated region. The isoform2, generated by alternative splicing with an alternate promoter, is a truncated FOXO3 protein with 453 aa that are encoded by exon 2 (Supplemental Fig. 1B, 1C). The coding sequences of the mouse and human FOXO3 are highly conserved, determined by 95% of identical amino acids (Supplemental Fig. 2). When comparing the coding and amino acid sequences of the human FOXO3 isoform2 with the mouse truncated Foxo3 in Foxo3flox/flox; LysMcre+ BMMs, we found that 96% of the amino acids are identical (Supplemental Fig. 3). These new findings indicate that the mouse truncated Foxo3 in Foxo3flox/flox; LysMcre+ BMMs is a mouse ortholog of human FOXO3 isoform2. We therefore name this novel Foxo3 isoform as mouse Foxo3 isoform2. The biological function of the human FOXO3 isoform2 is unclear. Because the Foxo3flox/flox; LysMcre+ mice express Foxo3 isoform2 instead of the full-length protein, we hereafter refer to these mice as Foxo3 isoform2 mice, which could be useful as a promising model for studying the function of the newly identified Foxo3 isoform2.

Mouse Foxo3 isoform2 suppresses osteoclastogenesis and leads to the osteopetrotic phenotype in mice

To investigate the role of Foxo3 isoform2 in osteoclastogenesis, we used BMMs as osteoclast precursors to examine osteoclast differentiation in response to RANKL. As shown in Fig. 3A, 3B, the osteoclast differentiation indicated by TRAP-positive multinucleated osteoclast formation induced by RANKL was significantly suppressed in Foxo3 isoform2 BMM cell cultures compared with the WT littermate control cell cultures (Fig. 3A, 3B).

FIGURE 3.
FIGURE 3.

Mouse Foxo3 isoform2 suppresses osteoclastogenesis and leads to the osteopetrotic phenotype in mice. (A) BMMs derived from WT control and Foxo3isoform2 mice were stimulated with RANKL for 4 d. TRAP staining was performed (A), and the number of TRAP-positive multinucleated cells (MNCs) per well is shown in (B). Scale bar, 100 μm. Data are representative of and statistical analysis was performed on three independent experiments. (C) μCT images and (D) bone morphometric analysis of trabecular bone of the distal femurs isolated from the WT and Foxo3isoform2 mice. n = 8 per group. (E) BMMs transfected with either control or Foxo3 siRNA (80 nM) were stimulated with RANKL for 5 d. The number of TRAP-positive MNCs (≥3 nuclei per cell) per well was calculated. *p < 0.05, **p < 0.01. BV/TV, bone volume per tissue volume; Tb.N, trabecular number; Tb.Sp, trabecular separation; Tb.Th, trabecular thickness.

We next performed microcomputed tomographic (μCT) analyses to examine the bone phenotype of Foxo3 isoform2 mice. The Foxo3 isoform2 mice and their littermate controls exhibit similar body weight and body length (data not shown). As shown in Fig. 3C, 3D, Foxo3 isoform2 mice show an osteopetrotic phenotype indicated by significantly increased trabecular bone volume and number but decreased trabecular bone spacing. Taken together with the suppressed osteoclast differentiation in Foxo3 isoform2 cells, these data demonstrate that expression of Foxo3 isoform2 in mice leads to an osteopetrotic bone phenotype.

Consistent with the Foxo3 global KO data (Fig. 1), knockdown of Foxo3 using RNA interference in WT BMMs enhanced osteoclast differentiation (Fig. 3E). Furthermore, knockdown of Foxo3 isoform2 in Foxo3 isoform2 BMMs significantly elevated osteoclastogenesis (Fig. 3E), supporting the inhibitory role of Foxo3 isoform2 in osteoclast differentiation.

We next performed a structure-functional analysis of Foxo3 protein in osteoclast differentiation. We cloned and generated a series of plasmids that express full-length WT Foxo3 or recombinant Foxo3 peptides encoded by the isoform2 or by exon 2 (hereafter referred to as Exon 2). We confirmed the protein expression of each plasmid in HEK293 cells (Fig. 4A) and RAW264.7 cells (Fig. 4B) after transfection. As shown in Fig. 4C, RANKL induced osteoclast differentiation in the RAW264.7 cells transfected with empty vector. Overexpression of WT full-length Foxo3 or isoform2 drastically inhibited osteoclast differentiation. The isoform2 seems to possess a stronger inhibitory effect on osteoclast differentiation than the full-length protein. Interestingly, expression of exon 2 significantly promoted osteoclast differentiation (Fig. 4C). These data were further corroborated by the corresponding changes in osteoclast marker gene expression, such as TRAP and cathepsin K (Fig. 4D). Because isoform2 is encoded by exon 3, these results argue that exon 3 is mainly responsible for osteoclastic inhibition, whereas exon 2 likely counteracts this effect.

FIGURE 4.
FIGURE 4.

Overexpression of Foxo3 isoform2 inhibits osteoclastogenesis. (A and B) Immunoblot analysis of the expression of full-length Foxo3, Foxo3 isoform2, and exon 2 in whole cell lysates of HEK293 cells (A) or RAW264.7 cells (B) transfected with corresponding pcDNA3.1+ plasmids containing specific Foxo3 fragments as indicated in the Materials and Methods. Anti-Flag Ab was used in (A). In (B), Foxo3 C-terminal Ab was used to detect full-length Foxo3 and Foxo3 isoform2. Foxo3 N-terminal exon 2 Ab was used to detect Foxo3 exon 2. (C) RAW264.7 cells transfected with the indicated plasmids were stimulated with RANKL for 6 d. TRAP staining was performed (left panel), and the number of TRAP-positive multinucleated cells per well is shown in the right panel. Scale bar, 100 μm. Data are representative of and statistical analysis was performed on three independent experiments. (D) Quantitaive PCR analysis of the relative expression of CtsK and Acp5 induced by RANKL for 6 d in the RAW264.7 cells transfected with the indicated plasmids. The induction folds of gene expression by RANKL relative to each basal condition was calculated and shown in the figure. Data are representative of three independent experiments. *p < 0.05, **p < 0.01.

Foxo3 isoform2 represses osteoclast differentiation via endogenous type I IFN–mediated feedback inhibition

We next set off to explore the mechanisms by which Foxo3 isoform2 inhibits osteoclastogenesis. In parallel with the suppressed generation of TRAP-positive polykaryons, we found that the expression of osteoclast marker genes Acp5 (encoding TRAP), Ctsk (encoding cathepsin K), Itgb3 (encoding β3 integrin), Dcstamp (encoding Dc-Stamp), Calcr (encoding calcintonin receptor), and Atp6V0d2 (encoding ATPase H+ Transporting V0 Subunit D2) was drastically decreased in RANKL-treated Foxo3 isoform2 cells relative to the WT control cells (Fig. 5A). A previous study shows that Foxo3 targets catalase and Cyclin D1 to arrest cell cycle and promote apoptosis in RANKL-induced osteoclastogenesis (16). Such Foxo3-mediated changes, however, were not detected in the Foxo3 isoform2 osteoclastogenesis (data not shown). In contrast, we found that the expression of Irf7, an IFN-responsive gene, was markedly elevated in RANKL-treated Foxo3 isoform2 cells relative to WT control cells (Fig. 5B). IRF7 has been identified as a Foxo3 target (29). It is also well established that endogenous IFN-β produced by osteoclast precursors is a strong feedback mechanism that restrains osteoclastogenesis (5, 30, 31). We therefore asked whether the inhibitory effects of Foxo3 isoform2 involves type I IFN–mediated inhibition. Previous studies showed that RANKL treatment can induce a low level of IFN-β expression in macrophages/osteoclast precursors. Although the magnitude of type I IFN induction by RANKL is small (<10 pg/ml after 24 h stimulation) when compared with other stimuli such as TLR stimulation, the high potency of type I IFN effects allow these small concentrations to inhibit osteoclast differentiation (30, 31). Consistent with these observations (30, 31), we found that RANKL induced IFN-β expression in WT BMMs and Foxo3 isoform2 significantly increased IFN-β induction (Fig. 5B). The enhancement of IFN expression by Foxo3 isoform2 was further corroborated by the elevated expression of IFN-responsive genes, such as Mx1, Ifit1, Ifit2, Irf7, and Stat1 after RANKL treatment (Fig. 5B). These results clearly demonstrate enhanced Ifnb expression and response by Foxo3 isoform2 during osteoclastogenesis and indicate that Foxo3 isoform2 suppresses osteoclastogenesis via type I IFN–mediated feedback inhibition.

FIGURE 5.
FIGURE 5.

Mouse Foxo3 isoform2 suppresses osteoclastic gene expression but enhances type I IFN–responsive gene expression. BMMs derived from WT control and Foxo3isoform2 mice were stimulated with RANKL for 3 d. The expression of osteoclastic marker genes (A) and type I IFN response genes (B) was examined by quantitative PCR. Data are representative of three independent experiments. **p < 0.01.

Discussion

Similarly to the other Foxo proteins, the function of Foxo3 is largely regulated through posttranslational modifications, such as phosphorylation, acetylation, methylation, and ubiquitination. These posttranslational modifications are context dependent and create a complex set of codes, which affect the subcellular location of Foxo3 and give rise to the diverse functions of Foxo family proteins in response to different stimuli (10–14). For example, Foxo3 can be phosphorylated by various protein kinases at many phosphorylation sites from the N to C terminus of the protein. Phosphorylation of specific sites by kinases, such as AKT, SGK1, CDK2, ERK, and IKK, induces cytoplasmic translocation and/or degradation of Foxo3, leading to target gene inhibition. In contrast, phosphorylation of the activating sites by kinases MST1, JNK, and AMPK usually leads to nuclear localization of Foxo3 and the activation of its target genes (10–14, 32). Foxo3 isoform2 lacks most of the N-terminal DNA-binding domain while maintaining the nuclear localization signal, the C-terminal nuclear export signal, and the transactivation domain at C terminus. This molecular structure implies that Foxo3 isoform2 is likely to lose the direct transcriptional regulation of the genes targeted by the full-length Foxo3 because of the lack of DNA-binding domain. However, Foxo3 isoform2 holds several activating phosphorylation sites that usually contribute to gene activation. In addition to the direct DNA-binding transcriptional activity, Foxo transcription factors are able to regulate transcription in a DNA-binding-independent manner, often by interaction with other transcriptional activators or repressors. Hence, we cannot exclude the possibility that Foxo3 isoform2 regulates gene transcription in the nucleus together with other partners. In addition, Foxo3 isoform2 carries the nuclear localization signal as well as the nuclear export signal that allow it to shuttle between the nucleus and cytoplasm in response to environmental cues. The overall impact from these possibilities will determine the subcellular localization of Foxo3 isoform2 and the mechanisms by which it inhibits osteoclastogenesis. The exon 2 peptide is shown to promote osteoclast differentiation. With the consideration that exon 2 contains an N-terminal DNA-binding domain, the direct DNA binding presumably results in the osteoclastogenic activity of exon 2, which in turn attenuates the full-length Foxo3′s ability in osteoclast inhibition. Further experiments are needed to elucidate the shared or distinct mechanisms mediated by full-length Foxo3 and the isoform2.

Protein isoforms from the exon skipping mode of alternative splicing often end up with a lack of certain domains that distinguish the function of the isoforms from their original full-length proteins. For example, previous studies identify IRF7 as a critical direct target of FOXO3, and FOXO3 negatively regulates IRF7 transcription in the antiviral response (29). Our results show that Foxo3 isoform2 expression elevates Irf7 transcription and corresponding type I IFN response during osteoclastogenesis. Foxo3 isoform2 lacks the DNA-binding domain and thus may function as an activator to increase Irf7 expression in a DNA-binding independent manner. Although Irf7 is a common target by both full-length Foxo3 and the isoform2, they show distinct regulatory effects on Irf7 expression presumably because of their different DNA binding capacity.

Our results revealed that Foxo3flox/flox;LysMcre+ mice are not fully conditional KO mice because of the existence of the isoform2. The position of the loxp sites caused an in-frame deletion of exon 2 in this mouse line. This was not known at the time when previous loss-of-function studies used this Foxo3flox/flox line. The interpretation of the mutant phenotype in such studies might be now questionable, dependent on cell types. Therefore, future work should pay close attention to the verification of frameshift deletion by cre-loxp recombination as well as the presence of isoforms.

Collectively, our findings in the current study identify the first, to our knowledge, known biological function of Foxo3 isoform2, which acts as an important suppressor of osteoclast differentiation via endogenous type I IFN–mediated feedback inhibition. The Foxo3 floxed allele mice (Foxo3flox/flox) could be used as a mouse resource in various areas to investigate the function of Foxo3 isoform2 that recapitulates human FOXO3 isoform2. Environmental cues often affect gene transcription and alternative splicing. For example, bone marrow macrophages/osteoclast precursors mainly express full-length Foxo3 with a trace amount of isoform2 in a physiological condition. Upon RANKL stimulation, Foxo3 isoform2 expression is increased (Fig. 2), which contributes to osteoclastic feedback inhibition. Thus, we speculate that the expression patterns and functions of Foxo3 isoform2 may be altered in response to different environmental settings. It will be of particular interest and clinical relevance to investigate the expression levels and functions of FOXO3 isoform2 in human cells, for instance, in human osteoclasts in healthy conditions versus disease settings, such as in osteoporosis and RA. Future studies are expected to uncover the expression profile of Foxo3 isoform2 in different cells and tissues, the regulation of its expression/function, related biological significance, and potential therapeutic implications.

Disclosures

The authors have no financial conflicts of interest.

Acknowledgments

We thank Dr. Matthew B. Greenblatt and the members of the Zhao laboratory for valuable discussion.

Footnotes

  • ↵1 Equal contribution.

  • This work was supported by Grants AR062047, AR071463, and AR068970 from the National Institutes of Health (to B.Z.) and by support for the David Z. Rosensweig Genomics Research Center from The Tow Foundation.

  • The sequences presented in this article have been submitted to the Gene Expression Omnibus under accession number GSE 135479.

  • The online version of this article contains supplemental material.

  • Abbreviations used in this article:

    BMM
    bone marrow–derived macrophage
    CM
    conditioned medium
    Foxo
    forkhead box class O
    HEK
    human embryonic kidney
    Irf
    IFN regulatory factor
    KO
    knockout
    RA
    rheumatoid arthritis
    RANKL
    receptor activator of NF-κB ligand
    RNA-seq
    RNA sequencing
    siRNA
    small interfering RNA
    TRAP
    tartrate-resistant acid phosphatase
    WT
    wild-type control.

  • Received June 25, 2019.
  • Accepted August 20, 2019.

References

  1. ↵
    1. Novack, D. V.,
    2. S. L. Teitelbaum
    . 2008. The osteoclast: friend or foe? Annu. Rev. Pathol. 3: 457–484.
    OpenUrlCrossRefPubMed
    1. Sato, K.,
    2. H. Takayanagi
    . 2006. Osteoclasts, rheumatoid arthritis, and osteoimmunology. Curr. Opin. Rheumatol. 18: 419–426.
    OpenUrlCrossRefPubMed
  3. ↵
    1. Schett, G.,
    2. E. Gravallese
    . 2012. Bone erosion in rheumatoid arthritis: mechanisms, diagnosis and treatment. Nat. Rev. Rheumatol. 8: 656–664.
    OpenUrlCrossRefPubMed
  4. ↵
    1. Asagiri, M.,
    2. H. Takayanagi
    . 2007. The molecular understanding of osteoclast differentiation. Bone 40: 251–264.
    OpenUrlCrossRefPubMed
  5. ↵
    1. Binder, N.,
    2. C. Miller,
    3. M. Yoshida,
    4. K. Inoue,
    5. S. Nakano,
    6. X. Hu,
    7. L. B. Ivashkiv,
    8. G. Schett,
    9. A. Pernis,
    10. S. R. Goldring, et al
    . 2017. Def6 restrains osteoclastogenesis and inflammatory bone resorption. J. Immunol. 198: 3436–3447.
    OpenUrlAbstract/FREE Full Text
    1. Li, S.,
    2. C. H. Miller,
    3. E. Giannopoulou,
    4. X. Hu,
    5. L. B. Ivashkiv,
    6. B. Zhao
    . 2014. RBP-J imposes a requirement for ITAM-mediated costimulation of osteoclastogenesis. J. Clin. Invest. 124: 5057–5073.
    OpenUrlCrossRefPubMed
    1. Zhao, B.,
    2. S. N. Grimes,
    3. S. Li,
    4. X. Hu,
    5. L. B. Ivashkiv
    . 2012. TNF-induced osteoclastogenesis and inflammatory bone resorption are inhibited by transcription factor RBP-J. J. Exp. Med. 209: 319–334.
    OpenUrlAbstract/FREE Full Text
    1. Zhao, B.,
    2. L. B. Ivashkiv
    . 2011. Negative regulation of osteoclastogenesis and bone resorption by cytokines and transcriptional repressors. Arthritis Res. Ther. 13: 234.
    OpenUrlCrossRefPubMed
  9. ↵
    1. Zhao, B.,
    2. M. Takami,
    3. A. Yamada,
    4. X. Wang,
    5. T. Koga,
    6. X. Hu,
    7. T. Tamura,
    8. K. Ozato,
    9. Y. Choi,
    10. L. B. Ivashkiv, et al
    . 2009. Interferon regulatory factor-8 regulates bone metabolism by suppressing osteoclastogenesis. Nat. Med. 15: 1066–1071.
    OpenUrlCrossRefPubMed
  10. ↵
    1. Hedrick, S. M.,
    2. R. Hess Michelini,
    3. A. L. Doedens,
    4. A. W. Goldrath,
    5. E. L. Stone
    . 2012. FOXO transcription factors throughout T cell biology. Nat. Rev. Immunol. 12: 649–661.
    OpenUrlCrossRefPubMed
    1. Tia, N.,
    2. A. K. Singh,
    3. P. Pandey,
    4. C. S. Azad,
    5. P. Chaudhary,
    6. I. S. Gambhir
    . 2018. Role of Forkhead Box O (FOXO) transcription factor in aging and diseases. Gene 648: 97–105.
    OpenUrl
  12. ↵
    1. Wang, X.,
    2. S. Hu,
    3. L. Liu
    . 2017. Phosphorylation and acetylation modifications of FOXO3a: independently or synergistically? Oncol. Lett. 13: 2867–2872.
    OpenUrl
    1. Salih, D. A.,
    2. A. Brunet
    . 2008. FoxO transcription factors in the maintenance of cellular homeostasis during aging. Curr. Opin. Cell Biol. 20: 126–136.
    OpenUrlCrossRefPubMed
  14. ↵
    1. van der Vos, K. E.,
    2. P. J. Coffer
    . 2008. FOXO-binding partners: it takes two to tango. Oncogene 27: 2289–2299.
    OpenUrlCrossRefPubMed
  15. ↵
    1. Morris, B. J.,
    2. D. C. Willcox,
    3. T. A. Donlon,
    4. B. J. Willcox
    . 2015. FOXO3: a major gene for human longevity--A mini-review. Gerontology 61: 515–525.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Bartell, S. M.,
    2. H. N. Kim,
    3. E. Ambrogini,
    4. L. Han,
    5. S. Iyer,
    6. S. Serra Ucer,
    7. P. Rabinovitch,
    8. R. L. Jilka,
    9. R. S. Weinstein,
    10. H. Zhao, et al
    . 2014. FoxO proteins restrain osteoclastogenesis and bone resorption by attenuating H2O2 accumulation. Nat. Commun. 5: 3773.
    OpenUrlCrossRefPubMed
  17. ↵
    1. Wang, Y.,
    2. G. Dong,
    3. H. H. Jeon,
    4. M. Elazizi,
    5. L. B. La,
    6. A. Hameedaldeen,
    7. E. Xiao,
    8. C. Tian,
    9. S. Alsadun,
    10. Y. Choi,
    11. D. T. Graves
    . 2015. FOXO1 mediates RANKL-induced osteoclast formation and activity. J. Immunol. 194: 2878–2887.
    OpenUrlAbstract/FREE Full Text
  18. ↵
    1. Gregersen, P. K.,
    2. N. Manjarrez-Orduño
    . 2013. FOXO in the hole: leveraging GWAS for outcome and function. Cell 155: 11–12.
    OpenUrlCrossRefPubMed
  19. ↵
    1. Lee, J. C.,
    2. M. Espéli,
    3. C. A. Anderson,
    4. M. A. Linterman,
    5. J. M. Pocock,
    6. N. J. Williams,
    7. R. Roberts,
    8. S. Viatte,
    9. B. Fu,
    10. N. Peshu, et al, UK IBD Genetics Consortium
    . 2013. Human SNP links differential outcomes in inflammatory and infectious disease to a FOXO3-regulated pathway. Cell 155: 57–69.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Miller, C. H.,
    2. S. M. Smith,
    3. M. Elguindy,
    4. T. Zhang,
    5. J. Z. Xiang,
    6. X. Hu,
    7. L. B. Ivashkiv,
    8. B. Zhao
    . 2016. RBP-J-regulated miR-182 promotes TNF-α-induced osteoclastogenesis. J. Immunol. 196: 4977–4986.
    OpenUrlAbstract/FREE Full Text
  21. ↵
    1. Vacik, T.,
    2. I. Raska
    . 2017. Alternative intronic promoters in development and disease. Protoplasma 254: 1201–1206.
    OpenUrlCrossRef
  22. ↵
    1. Kim, H. K.,
    2. M. H. C. Pham,
    3. K. S. Ko,
    4. B. D. Rhee,
    5. J. Han
    . 2018. Alternative splicing isoforms in health and disease. Pflugers Arch. 470: 995–1016.
    OpenUrl
  23. ↵
    1. Dlamini, Z.,
    2. F. Mokoena,
    3. R. Hull
    . 2017. Abnormalities in alternative splicing in diabetes: therapeutic targets. J. Mol. Endocrinol. 59: R93–R107.
    OpenUrlAbstract/FREE Full Text
  24. ↵
    1. Inoue, K.,
    2. Z. Deng,
    3. Y. Chen,
    4. E. Giannopoulou,
    5. R. Xu,
    6. S. Gong,
    7. M. B. Greenblatt,
    8. L. S. Mangala,
    9. G. Lopez-Berestein,
    10. D. G. Kirsch, et al
    . 2018. Bone protection by inhibition of microRNA-182. Nat. Commun. 9: 4108.
    OpenUrl
  25. ↵
    1. Trapnell, C.,
    2. L. Pachter,
    3. S. L. Salzberg
    . 2009. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25: 1105–1111.
    OpenUrlCrossRefPubMed
  26. ↵
    1. Trapnell, C.,
    2. B. A. Williams,
    3. G. Pertea,
    4. A. Mortazavi,
    5. G. Kwan,
    6. M. J. van Baren,
    7. S. L. Salzberg,
    8. B. J. Wold,
    9. L. Pachter
    . 2010. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat. Biotechnol. 28: 511–515.
    OpenUrlCrossRefPubMed
  27. ↵
    1. Anders, S.,
    2. P. T. Pyl,
    3. W. Huber
    . 2015. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics 31: 166–169.
    OpenUrlCrossRefPubMed
  28. ↵
    1. Robinson, M. D.,
    2. D. J. McCarthy,
    3. G. K. Smyth
    . 2010. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26: 139–140.
    OpenUrlCrossRefPubMed
  29. ↵
    1. Litvak, V.,
    2. A. V. Ratushny,
    3. A. E. Lampano,
    4. F. Schmitz,
    5. A. C. Huang,
    6. A. Raman,
    7. A. G. Rust,
    8. A. Bergthaler,
    9. J. D. Aitchison,
    10. A. Aderem
    . 2012. A FOXO3-IRF7 gene regulatory circuit limits inflammatory sequelae of antiviral responses. Nature 490: 421–425.
    OpenUrlCrossRefPubMed
  30. ↵
    1. Takayanagi, H.,
    2. S. Kim,
    3. K. Matsuo,
    4. H. Suzuki,
    5. T. Suzuki,
    6. K. Sato,
    7. T. Yokochi,
    8. H. Oda,
    9. K. Nakamura,
    10. N. Ida, et al
    . 2002. RANKL maintains bone homeostasis through c-Fos-dependent induction of interferon-beta. Nature 416: 744–749.
    OpenUrlCrossRefPubMed
  31. ↵
    1. Yarilina, A.,
    2. K. H. Park-Min,
    3. T. Antoniv,
    4. X. Hu,
    5. L. B. Ivashkiv
    . 2008. TNF activates an IRF1-dependent autocrine loop leading to sustained expression of chemokines and STAT1-dependent type I interferon-response genes. Nat. Immunol. 9: 378–387.
    OpenUrlCrossRefPubMed
  32. ↵
    1. Calnan, D. R.,
    2. A. Brunet
    . 2008. The FoxO code. Oncogene 27: 2276–2288.
    OpenUrlCrossRefPubMed
View Abstract
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools

Jump to section