Expression of the inflammatory cytokine TNF is tightly controlled. During endotoxin tolerance, transcription of TNF mRNA is repressed, although not entirely eliminated. Production of TNF cytokine, however, is further controlled by posttranscriptional regulation. In this study, we detail a mechanism of posttranscriptional repression of TNF mRNA by GAPDH binding to the TNF 3′ untranslated region. Using RNA immunoprecipitation, we demonstrate that GAPDH–TNF mRNA binding increases when THP-1 monocytes are in a low glycolysis state, and that this binding can be reversed by knocking down GAPDH expression or by increasing glycolysis. We show that reducing glycolysis decreases TNF mRNA association with polysomes. We demonstrate that GAPDH–TNF mRNA binding results in posttranscriptional repression of TNF and that the TNF mRNA 3′ untranslated region is sufficient for repression. Finally, after exploring this model in THP-1 cells, we demonstrate this mechanism affects TNF expression in primary human monocytes and macrophages. We conclude that GAPDH–TNF mRNA binding regulates expression of TNF based on cellular metabolic state. We think this mechanism has potentially significant implications for treatment of various immunometabolic conditions, including immune paralysis during septic shock.
The link between glycolysis and inflammation is well established. Many innate immune cell types specifically require glycolysis to perform their effector functions. When glycolysis is inhibited, leukocytes show decreased adhesion, mobility, and bacterial clearance (1–4). Monocytes produce less TNF cytokine when treated with the glycolysis inhibitor 2-deoxyglucose (2-DG), but not when treated with the mitochondrial inhibitor rotenone (4). Additionally, macrophages express greater levels of proinflammatory cytokines when forced to rely on glycolysis, but they express much lower levels when fatty acid oxidation is upregulated (5). This relationship between inflammation and glycolysis appears in certain disease states as well. As the endotoxin response proceeds to tolerance, monocytes downregulate glycolysis and upregulate fatty acid oxidation (6–8). This shift in metabolism occurs simultaneously with the onset of immunosuppression.
Recent findings indicate that glycolysis and inflammation communicate in ways not previously appreciated. One of the key enzymes in glycolysis is GAPDH, which converts glyceraldehyde-3-phosphate (G3P) into 1,3-bisphosphoglycerate in the sixth step of the glycolysis pathway (9). GAPDH also has a lesser known capacity as an RNA-binding protein (10). Specifically, GAPDH binds to AU-rich elements (ARE) found in the 3′ untranslated region (UTR) of many mRNAs. ARE are present in many inflammatory genes, including cytokines such as IFN-γ and TNF (11–13). GAPDH binding to a generic ARE is inhibited by G3P (14) and NAD+, a necessary cofactor for its enzymatic activity (10). Recently, it was shown GAPDH–ARE binding is responsible for posttranscriptional regulation of IFN-γ expression in T cells (15). This binding is disrupted by the metabolite G3P, making this mechanism sensitive to cellular metabolism. Some argue that these types of RNA–enzyme–metabolite interactions broadly affect gene expression (16); however, these mechanisms remain largely unexplored.
Expression of TNF is tightly regulated in immune cells. During endotoxin tolerance, much of this regulation occurs at the level of chromatin (17–24). Tolerant monocytes and other immune cells fail to generate TNF mRNA in response to an additional stimulus while they are in the immunosuppressed state. This repression of TNF expression also occurs at the posttranscriptional level (25–27). Even when transcription of TNF mRNA is restored to tolerant monocytes, they continue to show deficiencies in TNF cytokine production. Posttranscriptional repression mediated by microRNA accounts for part of this deficiency (25, 26). A number of reports describe other posttranscriptional mechanisms that regulate TNF expression (28–32); however, none of these mechanisms suggests that cellular metabolic state informs the regulation process. In this study, we propose a mechanism where glycolysis directly affects TNF expression through posttranscriptional regulation.
With our previous work in the background in regard to posttranscriptional repression of TNF mRNA and immunometabolic shifts in monocytes during the endotoxin response, we speculated that GAPDH–ARE binding might contribute to regulation of TNF expression in monocytes. We hypothesized that if glycolysis was limited, GAPDH would bind the ARE of TNF mRNA, thereby limiting its translation. To test this, we first cultured our THP-1 cells in media where glucose was replaced by galactose. Because galactose is metabolized more slowly than glucose (33), these cells adopted a less glycolytic, more oxidative metabolism. We not only found GAPDH binding to TNF mRNA in galactose-fed monocytic cells, but that this binding also occurs in endotoxin-tolerant cells following the natural downregulation of glycolysis that monocytes exhibit during tolerance. Furthermore, we found that GAPDH–TNF mRNA binding is affected by pharmacological manipulation of glycolysis. Our results indicate that this mechanism allows leukocyte cell metabolism to fine-tune TNF gene expression. These findings have potential implications for any number of disease states involving inflammation and metabolism, such as immunoparalysis during septic shock.
Materials and Methods
THP-1 cells were grown in RPMI 1640 with 10% FBS, l-glutamine, and penicillin/streptomycin media. Cells were kept in a 5% CO2 incubator at 37°C and subcultured every 1–3 d to maintain a density of 20–80 × 104 cells/ml (34). THP-1 cells were maintained in an undifferentiated state. Galactose-fed cells were taken from standard glucose-fed cultures, spun down, washed with PBS, and grown in RPMI 1640 (no glucose, 2 g/l galactose) for ≥5 d before use in any experiments.
THP-1 cells were tolerized with addition of 1 μg/ml LPS for 24 h. For experiments involving second-dose exposure of LPS, cells were spun down and resuspended in fresh media for 1 h before proceeding with second doses of LPS, also at 1 μg/ml.
Preparation of human primary monocytes/macrophages
Primary monocytes/macrophages were collected from heparinized venous blood samples donated by healthy adult volunteers according to the Institutional Review Board protocol approved by Wake Forest University (35). RBCs, platelets, and polymorphonuclear neutrophils were removed through Isolymph (Gallard-Schlesinger Industries) centrifugation of whole blood. Monocytes were then enriched through a 2-h adherence step, after which nonadherent cells were removed. Cells were then cultured overnight in fresh RPMI 1640 containing 10% FBS and either glucose or galactose, with or without 100 ng/ml LPS to induce ex vivo endotoxin tolerance. Brightfield analysis of morphology showed resulting cultures had >90% monocytes and macrophages.
Assessment of oxygen consumption rate and extracellular acidification rates (ECAR) were made using the Seahorse XF24 extracellular flux analyzer (Seahorse Bioscience) (36). Plates were coated with Cell-Tak (BD Biosciences) (37) and dried overnight before addition of 25 × 104 cells/well in unbuffered DMEM (10% FBS, 2 g/l glucose or galactose) and 1 h incubation in a CO2-free 37°C incubator. Plates were assayed according to the manufacturer’s instructions.
Lactate assays were performed using an L-Lactate assay kit (Eton Bioscience) according to the manufacturer’s instructions (38). Cells were kept in phenol red–free DMEM with 2 g/l glucose or galactose during the assay.
39). Cells were washed twice with PBS and resuspended to a density of 80 × 104 cells/ml in appropriate media before incubation with or without LPS. Supernatant of resulting cultures was collected when indicated and used for assay.
Real-time quantitative PCR
RNA was isolated using STAT60 (Tel-Test) when isolation was required outside the context of RNA immunoprecipitation (40). RNA quality was measured on a NanoDrop 1000 (Thermo Scientific) before reverse transcription using the qScript cDNA synthesis system (Quanta BioSciences) (41
For RNA stability assay, cells were stimulated with LPS for 1 h and then given 5 μg/ml actinomycin D for indicated time. Cells were then pelleted and RNA isolated as described above (23).
RNA immunoprecipitation was performed using the Magna RIP kit (Millipore) according to manufacturer’s instructions (42). Briefly, cultures of 10 × 106 cells were prepared as described above, spun down, washed, and lysed with −80°C freezing. Lysates were then spun down and supernatants transferred to tubes with magnetic beads that were previously treated with 5 μg anti-GAPDH Ab (Sigma-Aldrich) or nonspecific IgG. Lysates were rotated with beads overnight, washed the next day, eluted (alongside input RNA), isolated with phenol-chloroform-isoamyl alcohol, ethanol precipitated, and resuspended in RNase-free water. Quality of input RNA was assessed and all samples were measured through quantitative real-time PCR (RT-qPCR) as described above.
THP-1 cells (5 × 106) were transfected with 1 μM small interfering RNA (siRNA) targeting either GAPDH mRNA or not targeting any mRNA (control). Transfection was done using the Amaxa Nucleofector II according to the manufacturer’s instructions (23). Cells were cultured in appropriate media for 3 d following transfection before use in Western blot or ELISA experiments.
Polysome fractionation profiling
Polysome fractionation analysis was performed as previously described (44). Briefly, 10 × 106 THP-1 cells were incubated with 100 μg/ml cyclohexamide before lysis in hypotonic buffer. Lysates were pelleted, and the supernatant was placed on top of a 10–45% continuous sucrose gradient. Samples were then centrifuged at 222,228 × g for 2 h at 4°C. After ultracentrifugation, tubes were pierced at the bottom and fractions were collected. UV absorbance was measured for each fraction using a NanoDrop 1000 (Thermo Scientific). RNA was extracted from each fraction using STAT50 (Tel-Test). RT-qPCR analysis was then performed as previously described.
THP-1 cells were plated in white 96-well plates in phenol red–free DMEM (5% FBS, 2 g/l glucose or galactose). Cells were then transfected with FuGENE transfection reagent and GoClone plasmids (SwitchGear Genomics) encoding Renilla luciferase with 3′UTR regions indicated in the figure legends. Transfections included Cypridina TK loading control plasmid. Transfection procedure followed the manufacturer’s instructions. Assay of luciferase activity was done 24 h after transfection using LightSwitch dual assay reagents (Active Motif) and the MicroLumat Plus LB96V (Berthold Technologies) plate luminometer. Relative luciferase units were calculated by subtracting background signal and normalizing Renilla signal to loading plasmid.
Statistical analysis and graphical presentations were performed using Microsoft Excel 2010. Significance was calculated using an unpaired Student t test. All data shown represent results from three or more independent observations, expressed as mean ± SEM.
Tolerance and galactose both affect metabolism and TNF-α expression
As our laboratory has previously reported (17, 22), endotoxin tolerance includes two distinct phenotypic characteristics in THP-1 monocytic cells. One characteristic of tolerance is an inability to produce TNF-α mRNA or protein in response to LPS restimulation. The other characteristic is a preference for fatty acid oxidation over glycolysis (8). To test our hypothesis that the latter influences the former, we compared responsive and tolerant cells to those grown in galactose-based media. The literature suggests that when glucose is replaced by galactose in cell culture media, cells use more mitochondrial oxidation and less glycolysis (15, 45, 46). Thus, this model allowed us to separate the metabolic impact of tolerance from its other effects on gene expression.
We first measured expression of TNF in three different culturing conditions: responsive (glucose-based media), tolerant (glucose-based media, prior overnight exposure to 1 μg/ml LPS), and galactose fed (galactose-based, glucose-free media). At the RNA level, we observed no significant difference between responsive versus galactose-fed cultures, with or without addition of LPS (Fig. 1A). TNF mRNA levels were significantly different in tolerant cultures, in line with previous reports (17). Despite showing no difference in TNF mRNA, however, galactose-fed cultures did show a significant reduction in TNF protein expression, as measured by ELISA (Fig. 1B). Culturing conditions did not appear to significantly impact stability of TNF transcript (Fig. 1C).
We next compared the differences in glycolysis between cells grown in responsive, tolerant, or galactose-fed culturing conditions. This was done in two ways. Lactate concentration following addition of LPS was measured using a commercial biochemical lactate assay (Fig. 2A). Responsive cells showed the highest concentration of lactate, followed by tolerant and galactose-fed cells, respectively. We also measured the ECAR of responsive, tolerant, and galactose-fed cells using the Seahorse XF24 (Fig. 2B). As a measurement of the rate of proton output by live cells, ECAR serves as an indicator of lactic acid production and glycolysis (36). Basal ECAR was the highest in responsive cells, followed respectively by tolerant and galactose-fed cells. Interestingly, responsive cells showed a sharp increase in ECAR after an injection of LPS into the assay wells, whereas neither tolerant nor galactose cells showed any significant change in ECAR in response to LPS. These differences in lactate (Fig. 2A) and ECAR (Fig. 2B) both indicate that galactose-fed THP-1 cells have a lower rate of glycolysis than do their glucose-fed counterparts.
GAPDH binds to TNF mRNA in THP-1 cells with low glycolysis
Our observation that TNF protein but not mRNA was reduced in galactose-fed cells (Fig. 1A, 1B) suggests a mechanism of posttranscriptional repression. These data are consistent with our hypothesis that low glycolysis causes GAPDH to bind the ARE of TNF mRNA. To determine whether this was the case, we used RNA immunoprecipitation (RNA-IP) with an anti-GAPDH Ab to probe for an interaction between GAPDH protein and TNF-α mRNA.
Our initial RNA-IP experiments compared responsive, glucose-fed cells with responsive, galactose-fed cells. As shown in Figs. 1 and 2, these cultures differed in metabolism, but not TNF mRNA. After stimulation with LPS for 1 h, significantly more TNF mRNA was pulled down by the GAPDH Ab in galactose-fed cultures than in glucose-fed cultures (Fig. 3A). This indicates greater GAPDH protein–TNF mRNA binding occurs in galactose-fed cells.
Additionally, GAPDH showed no off-target binding to its own mRNA (Fig 3B). GAPDH mRNA is constitutively expressed and lacks an ARE, making it an unlikely target for GAPDH protein to bind. This made GAPDH mRNA a suitable negative indicator of nonspecific RNAs isolated by the RNA-IP. As shown in Fig. 3B, minimal GAPDH mRNA was pulled down during the RNA-IP. This indicates that there is specificity to the GAPDH protein–TNF-α mRNA interaction. To test whether the increase in GAPDH–TNF mRNA binding reflected an increase in total GAPDH protein, we measured GAPDH protein levels by Western blotting (Fig. 3C). We observed no significant change in GAPDH protein concentration in response to galactose-based media, or in response to stimulation with LPS.
Comparison of glucose-fed and galactose-fed cultures indicated that our hypothesized mechanism of metabolism-sensitive RNA binding took place in monocytes, but under idealized and artificial conditions. We next sought to investigate whether it also took place during endotoxin tolerance. Tolerant THP-1 cells show reduced glycolysis (Fig. 2) and serve as a model for septic shock (47–49).
To determine whether this mechanism participated in tolerance, we again used RNA-IP to probe for interactions between GAPDH protein and TNF-α mRNA. Tolerant cultures were stimulated with LPS for 24 h prior to assay, whereas responsive cultures were not exposed to any LPS prior to assay.
Real-time PCR analysis of the RNA pulled down by the GAPDH Ab shows that GAPDH binds to TNF mRNA in tolerant cells (Fig. 4A). The amount of TNF mRNA bound by GAPDH was significantly greater in tolerant cells than responsive cells, despite the repression of TNF mRNA in tolerant cells. As in the glucose versus galactose model, no significant off-target binding to GAPDH mRNA is observed (Fig. 4B). We also observed no significant change in total GAPDH protein level (Fig. 4C).
GAPDH is responsible for posttranscriptional repression
Our observation that GAPDH binds to TNF mRNA when TNF protein, but not mRNA, is reduced immediately suggests a mechanism of posttranscriptional repression. To test this potential mechanism, we used siRNA to knock down GAPDH expression in both glucose and galactose-fed cells (Fig. 5A). We observed that whereas the GAPDH knockdown had no significant effect on glucose-fed cells, the knockdown increased production of TNF protein in galactose-fed cells (Fig. 5B). This supports our hypothesis that GAPDH protein limits production of TNF protein in cells with reduced glycolysis.
We next sought to verify that posttranscriptional repression was, in fact, responsible for the loss of TNF protein. To test this, we used polysome fractioning to determine whether TNF mRNA ceased associating with polyribosomes when glycolysis was limited. Lysates from glucose and galactose-fed cells were separated over a 10–45% sucrose gradient and fractionated by density. Polysome fractions were determined by UV absorbance at 254 nm (Fig. 6A). We observed that in glucose-fed cells, a greater portion of the TNF mRNA was present in the polysome fractions (Fig. 6B, 6C). In galactose-fed cells, TNF mRNA was found in less dense fractions, indicating fewer associated ribosomes. Neither media affected the density of actin mRNA (Fig. 6D, 6E).
GAPDH binding to TNF mRNA is sensitive to changes in glycolysis
After demonstrating GAPDH binding to TNF mRNA in two conditions with low glycolysis, we sought to further establish that glycolysis regulated the level of this binding. We also sought to determine whether this binding was reversible. To test this, we treated tolerant THP-1 cells with different substances that alter glycolysis. We then used RNA-IP to study corresponding changes in GAPDH–TNF mRNA binding.
Based on the literature and our past experience, we selected four substances, each with a distinct mechanism of affecting glycolysis (Fig. 7A). To block glycolysis, we used 2-DG, an inhibitor of hexokinase and phosphoglucose isomerase (50). To promote glycolysis, we used EX527, a sirtuin 1 inhibitor that limits the ability of cells to transition from glycolysis to fatty acid oxidation (8); human insulin, which increases glucose uptake and phosphorylation (51, 52); and oligomycin, an ATP synthase inhibitor that blocks mitochondrial ATP production (53) and causes an acute increase in glycolysis.
The effects of these four substances on glycolysis were verified by Seahorse XF analysis (Fig. 7B). Tolerant cell cultures were treated with 2-DG (5 mM, 1 h before assay), EX527 (5 μM, 18 h before assay), human insulin (100 nM, 18 h before assay), or oligomycin (10 μM, 15 min before assay) as indicated. Cultures were then lysed and analyzed by RNA-IP. Inhibition of glycolysis using 2-DG resulted in a greater level of TNF-α mRNA in the resulting GAPDH RNA-IP (Fig. 8A). Similarly, promotion of glycolysis with any of the other three treatments decreased the level of TNF mRNA isolated by RNA-IP. This indicates that lowering glycolysis increases GAPDH–TNF mRNA binding, whereas increasing glycolysis reduces that binding. This reciprocal relationship is predicted by our hypothesis. No significant binding to GAPDH mRNA was observed (Fig. 8B), again indicating that the GAPDH–TNF mRNA interaction is specific. Additionally, we saw no significant change in total GAPDH protein in response to the treatments (Fig. 8C).
We next explored whether these changes in glycolysis produced measurable changes in TNF protein. If GAPDH–TNF mRNA binding truly represents a mechanism of posttranscriptional repression, we would expect that treatments that increase glycolysis and decrease GAPDH–TNF binding would increase TNF protein production. To test this, we measured expression of TNF mRNA and protein in tolerant THP-1 cells treated with either EX527 or insulin versus untreated. We were unable to use 2-DG or oligomycin in this study due to higher toxicity and the longer incubation period required for ELISA.
TNF mRNA levels were not increased by addition of insulin or EX527 to tolerant cultures (Fig. 9A); however, we observed small but statistically significant increases in TNF protein levels following treatment with either substance (Fig. 9B). Because the increase in cytokine production cannot be explained by an increase in RNA, it follows that a greater amount of the transcript is translated. This supports our hypothesis that GAPDH binding represses translation of TNF mRNA.
Transcripts with the 3′UTR of TNF mRNA are repressed in a metabolism-sensitive manner
Our data indicate that GAPDH–TNF mRNA binding correlates with a decrease in TNF-α protein expression. To further demonstrate that this decrease in cytokine production is due to posttranscriptional repression, we used a luciferase reporter system (Fig. 10A). We used plasmids encoding a Renilla luciferase transcript, with or without the TNF 3′UTR present. Because the plasmids contained the same constitutive promoter, and because Renilla luciferase is not affected by ATP, changes in luminescence should be directly attributable to posttranscriptional regulation. We reasoned that if GAPDH–TNF mRNA binding results in posttranscriptional repression, altering glycolysis should alter luciferase signal in a consistent manner.
We observed a significant reduction in luciferase signal in tolerant cells transfected with the TNF 3′UTR reporter, compared with those with the control 3′UTR (Fig. 10B). This immediately demonstrated the importance of posttranscriptional repression of TNF, which has been previously shown (25, 26). When cells transfected with the TNF 3′UTR reporter were treated with substances that affected both glycolysis and GAPDH–TNF mRNA binding (Figs. 7B, 8A), luciferase signal was also affected (Fig. 10B). Addition of 2-DG caused a decrease in luciferase signal, whereas addition of insulin or oligomycin resulted in increased signal. These results match the RNA-IP data (Fig. 8A), which indicated the treatments respectively increased or decreased posttranscriptional repression of TNF mRNA.
GAPDH binds to TNF-α mRNA in primary cells
After characterizing this mechanism of posttranscriptional repression in THP-1 cells, we tested whether this mechanism was also present in primary human monocytes. Primary monocytes were isolated from whole blood samples collected from healthy donors. Donor monocytes were either cultured overnight in glucose-based media, tolerized ex vivo, or cultured overnight in galactose-based media. Examination of cell morphology the following day by brightfield staining showed >90% of isolated cells were monocyte/macrophage cell types (data not shown).
We first measured the effect of our responsive, tolerant, and galactose-fed culturing conditions on glycolysis. As in our THP-1 model, responsive cultures showed the highest level of glycolysis before and after the addition of LPS (Fig. 11A). Tolerant and galactose-fed cell cultures both showed reduced concentration of lactate, indicating a reduced rate of glycolysis.
We next determined whether culturing conditions affected production of TNF cytokine. ELISA analysis of cell supernatant revealed that cells in tolerant and galactose cultures produced less cytokine in response to LPS than did their responsive counterparts (Fig. 11B). These results are consistent with THP-1 results (Fig. 1B), supporting the hypothesis that a similar mechanism was responsible. When analyzed by RNA immunoprecipitation, GAPDH binding to TNF mRNA was confirmed (Fig. 11C). We found significantly greater GAPDH–TNF mRNA binding in tolerant and galactose-cultured cells than in responsive-cultured cells. This difference is particularly prominent when responsive and galactose cultures are compared.
In this study, we show that TNF mRNA is posttranscriptionally repressed by GAPDH binding to the 3′UTR. As summarized in Fig. 12, this mechanism of repression is sensitive to changes in cellular metabolism, specifically the rate of glycolysis. When the rate of glycolysis is high, GAPDH binds TNF mRNA at a relatively low level. When glycolysis is downregulated due to limited availability of glucose or endotoxin tolerance, GAPDH binds TNF mRNA to a greater degree. This binding inhibits translation of the transcript, thus limiting TNF cytokine production.
This study further demonstrates that GAPDH binding to TNF mRNA can be reversed by increasing glycolysis. Others have shown that GAPDH metabolic substrates G3P and NAD+ interfere with GAPDH binding to ARE (10, 14, 15). As neither G3P nor NAD+ is membrane permeable, however, these data were observed in ex vivo experiments or in saponin-permeabilized cells. We think our approach of reversing binding by increasing glycolysis better illustrates the central role of metabolism in regulating translation.
We propose that this mechanism of posttranslational repression through GAPDH–TNF mRNA binding represents a way of fine-tuning the inflammatory response. Our data indicate that glycolysis affects production of TNF cytokine, although only modestly (Fig. 7). When compared with mechanisms regulating production of TNF mRNA (54), the effects we observe are relatively small. Although this mechanism is not a primary determinant of TNF expression, we argue it makes a unique contribution.
We suggest that GAPDH–TNF mRNA binding refines expression of TNF depending on the metabolic environment. We imagine this mechanism of regulation is advantageous in a number of biological situations. For example, endothelial cell responses to TNF signaling allow for immune cell migration to a site of infection (55). Effector immune cells require glucose for effector functions such as phagocytosis and generating reactive oxygen species for the respiratory burst (56). Because GAPDH binding limits TNF mRNA translation when glycolysis decreases, we propose that this mechanism essentially helps keep the demand for glucose from exceeding the microenvironment supply.
In this study, we describe how glycolysis influences TNF protein expression through a mechanism not previously observed in monocytes. These findings may have implications for any number of immunometabolic conditions. One such condition of great clinical significance is sepsis. Endotoxin-tolerant mechanisms are closely aligned with the immunosuppressed state observed in septic shock (57). This state increases risk of secondary infection and overall patient mortality (58, 59). Following the failure of anti-TNF therapies to decrease patient mortality, there is increasing reason to explore stimulation of the immune system to improve survival in patients with severe sepsis or septic shock (60, 61). Our findings underscore the importance of approaching such efforts metabolically as well as immunologically.
The authors have no financial conflicts of interest.
We acknowledge David Long and Dr. Michael Seeds for technical assistance during this project as well as Dr. Martha Alexander-Miller and Dr. Anthony Molina for guidance during this project.
This work was supported by National Institutes of Health Grants R01AI079144, R01AI065791, R01GM099807, and R01GM102497.
Abbreviations used in this article:
- AU-rich element
- extracellular acidification rate
- RNA immunoprecipitation
- real-time quantitative PCR
- small interfering RNA
- untranslated region.
- Received June 15, 2015.
- Accepted January 3, 2016.
- Copyright © 2016 by The American Association of Immunologists, Inc.