The study group included 71 individuals with a diagnosis of RA and 69 healthy controls (HCs). The RA patients fulfilled the American College of Rheumatology (ACR) criteria (24) or the ACR/European League Against Rheumatism (EULAR) 2010 classification criteria for RA (25), and all RA patients were positive for rheumatoid factor and/or anticyclic citrullinated peptide antibody. The control subjects were matched with age, gender and ethnicity. A history of cancer, uncontrolled medical disease or any other inflammatory syndrome were excluded. Healthy individuals didn’t have any personal or family history of autoimmune disease. The demographic characteristics of the RA patients and HC donors are summarized in Table 1 . The study was approved by the Medical Ethics Committee of Hospital and all subjects provided appropriate informed consent.

RA disease activity was assessed using the Disease Activity Score in 28 joints with 4 variables, including C-reactive protein level. Radiographs of bilateral hands, wrists, and feet (anteroposterior view) were performed on all RA patients and assessed with the modified Sharp/van der Heijde score. Erosive disease was defined according to the 2013 EULAR definition when a cortical break was detected by radiography (6, 26).

Human Peripheral blood mononuclear cells (PBMCs) were collected using Ficoll-Paque™ PLUS (Mediatech, Inc., Herndon, VA, USA) gradient centrifugation. To sort CD14⁺ cells, PBMCs were positively selected with CD14 microbeads using autoMACS (Miltenyi Biotec Inc., Auburn, CA, USA).

For osteoclast differentiation, 1×10⁵ purified CD14⁺ monocytes were seeded in 96-well plates or 2 × 10⁵ in a 24-well plate (Corning, USA) per well. Change the fresh medium in the presence of 100 ng/ml RANKL and 50 ng/ml M-CSF every four days, finally terminate the differentiation in 21st day. Then the cells were stained by Tartrate Resistant Acid Phosphatase Staining (TRAP) staining kit and FITC-labeled phalloidin (Merck, Darmstadt, Germany). Cells including at least 3 nuclei and bands of F-actin containing podosomes were defined as mature osteoclasts, and finally mature osteoclasts were counted including the number of per well, the nuclei of each mature osteoclast, and the sizes of the osteoclasts were obtained by measuring the diameters of multinucleated TRAP-positive cells using Olympus IX71 inverted wide-field fluorescence microscope.

For bone resorption assay, bovine cortical bone slices were layered at the bottom of 96-well culture plates, and 1 × 10⁵ peripheral blood CD14⁺ monocytes were seeded onto the slices and cultured in the presence of 100 ng/ml of recombinant human RANKL and 50 ng/ml of M-CSF. Resorption pits on the slices were shown by toluidine blue staining and measured using an ImageJ 1.47 analysis system (NIH).

To test for the role of FAO in mediating osteogenesis and cell–cell fusion, etomoxir (MedChemExpress, Monmouth Junction, NJ, USA), which inhibiting the CPT1A enzyme activity results in a reduction in fatty acid oxidation, were used at concentrations of 25 μM. Etomoxir was added to each exchange of the osteoclast differentiation medium.

Total RNA was extracted from 1 × 10⁵ cells, and cDNA was synthesized with the AMV-reverse transcriptase and random hexamer primers (Roche Diagnostic Corp., Indianapolis, USA). Reverse transcription was performed using a standard procedure (Super Script First-Strand Synthesis System; Invitrogen) using 1 μg of total RNA. QPCR was performed using the iQ SYBR Green Supermix on the iCycler Real-Time Detection system (Bio-Rad, Hercules, CA, USA) on an StepOne Plus Real-Time Detection system (Applied Biosystems, USA) (27). The gene expression was normalized to GAPDH and the relative amount of mRNA was calculated using the 2^−ΔΔCt method. Primer sequences were listed in Table 2 . The relative amount of mRNA was normalized with a housekeeping GAPDH as indicated and was quantified.

Whole cell lysates were prepared in RIPA buffer (Cell signaling Technology Danvers, MA, USA) plus phenylmethylsulphonyl fluoride (PMSF) and a protease inhibitor cocktail (Merck, Germany) (28). For each sample, equal amounts of total protein were electrophoresed, transferred to a nitrocellulose membrane and blocked with 5% non-fat milk. Antibodies specific for CPT1A, CPT1B, CLTA, CLTC, C/EBPβ, and Actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were added 12 h at 4°C, followed by washing with washing buffer. Membranes were subsequently incubated with secondary antibodies (1:5000; Santa Cruz) for 1 h at room temperature and developed with a chemiluminescent detection system (Beyotime, Shanghai, China).

For cell fusion analysis, CD14⁺ monocytes were pretreated with 100 ng/ml of RANKL and 50 ng/ml of M-CSF for 3 days. Then cells were surface stained with APC-CD14, FITC-CD36, FITC-DC-STAMP, and FITC-CD47 mouse polyclonal antibodies (BioLegend, San Diego, CA, USA).

Peripheral blood CD14⁺ monocytes were plated on 24-well culture plates with coverslips. After incubation, the medium was aspirated and cells were washed twice in PBS, fixed in 4% paraformaldehyde for 15-30 minutes, permeabilized in 0.2% Triton X-100 for 10 minutes at room temperature for the exposure of intracellular antigen, and blocked in PBS containing 3% bovine serum albumin (BSA; Affymetrix) for 30 minutes. Cells were stained with a mouse anti-human CD14 monoclonal antibody, rabbit anti-human DC-STAMP, rabbit anti-human CD36, mouse anti-human clathrin, and FITC-labeled phalloidin overnight at 4°C. Alexa Fluor 594–conjugated goat anti-rabbit IgG and Alexa Fluor 488–conjugated goat anti-mouse IgG (1:1,000; concentrations of stock solutions 2 mg/ml) (life science) were added and incubated for 1 hour at 37°C. After washing in PBS, the nucleus was stained with 4’6-diamindino-2-phenylindole (DAPI, Sigma-Aldrich) and coverslips were mounted with ProLong^® Gold Antifade Reagent (Life Technologies). The images were analyzed using a Zeiss LSM 880 Confocal Imaging System (Zeiss, Jena, Germany) (29).

To knockdown CPT1A expression, monocytes were pretreated with 100 ng/ml of RANKL and 50 ng/ml of M‐CSF for 4 days, 6 ug siRNA CPT1A and control vector were transfected into cells by oligofectamine (Life Technologies, Inc.; Invitrogen), following protocols provided by the manufacturer. The sequences of CPT1A siRNA were 5′-GCACCGUCAAUGCCUACAA dTdT-3′. For osteoclast differentiation, peripheral blood CD14⁺ monocytes were transfected with CPT1A sequence–specific short hairpin RNA (shRNA) expression lentivirus (pLV-mCherry-U6>CPT1A_shRNA) (VectorBuilder, Guangzhou, China) with (pLV-mCherry-U6>Scramble_shRNA) as a control. Stably transduced cells were selected using puromycin.

The ChIP assay was performed using ChIP kit (Merck, Germany) with C/EBPβ antibody (Santa Cruz Biotechnology, USA) (30). Cells in a 10-cm culture plate were crosslinked with 1% formaldehyde for 10 minutes. Crosslinking was neutralized with 0.2M glycine. Cells were collected and suspended in lysis buffer. Genomic DNA was released by using SDS Lysis Buffer, and sonicated to 200-1000 bps. Protein–DNA complexes were precipitated with C/EBPβ antibody (Cell Signaling Technology, USA). After purification, the precipitated DNA and input were de-crosslinked at 65°C and then purified. The capacity of C/EBPβ binding to CLTA and CLTC promoter were quantified by qPCR. Primer sequences used in ChIP-qPCR were as follows: CLTA sense 5’-ATGGCCCAGATGGAGAAAGC-3’ and antisense 5’-GGGAGGTGTTGGATGTGAGG-3’; CLTC sense 5’-ATGGCCCAGATGGAGAAAGC-3’ and antisense 5’-TGTTTCGACTGAGCCCCT-3’.

The liquid chromatography method was established on the QExactive LC (Thermo Fisher, Waltham, MA, USA) system, where Ultimate 3000 LC (Thermo Fisher) was used for the liquid phase portion. The mass spectrometry method was established in QExactive (Thermo Fisher) system. The lipidomics mass spectrometry method was performed in full scan/data-depend ms² Top10 analysis mode, and positive/negative spectrum analysis in ESI mode. Spray voltage was +3500V/-3000V, the sheath gas was 40arb, the auxiliary gas was 10arb, the ion transfer tube temperature was 320°C, the auxiliary gas heating temperature was 350°C. The first-level full scan resolution was 70000, and the mass range was 100-1500m/z. Ms² had a secondary resolution of 17500, top 10, impact energy of 20, 40, and mass range of 70-1500m/z.

The data obtained from the above lipidomics method were imported into Lipid Search 4.1 software for Lipid identification and queue analysis. The full scan/data-depend ms² data collected by QExactive instrument was selected to use the daughter ion mode for lipid identification. The accurate mass deviation threshold of parent ion and daughter ion was set to 5ppm, and isotope matching was performed to display the ions with fragment matching score above 2. The retention time window threshold was set to 1min. Positive mode selects +H peak, +Na peak, +NH4 peak addition and negative mode selects -H peak, +HCOO peak addition and sum for database retrieval. The retrieval results select and identify the ions with grade A, B, and C for queue analysis to obtain the lipid profile information and the relevant data results could be exported.

Data were analyzed using SPSS version 13.0 software. Groups were compared using unpaired Student’s t-test as appropriate. The results were expressed as the mean ± standard error of the mean (SEM). The correlation of parametric data was assessed using Pearson’s correlation test. P values of less than 0.05 were considered significant.

Lipids are major molecular constituents of cells that have important functions in membrane structure, energy production, and signal transduction. To study alterations in the lipid metabolic pathway of OCPs in RA, the lipidome of peripheral CD14⁺ monocytes isolated from RA patients and HCs was comprehensively examined by liquid chromatography coupled with dynamic quantitative mass spectrometry. Overall, the lipid profiles of RA patients contained markedly reduced amounts of diacylglycerol (DAG), triacylglycerol (TAG), phosphatidylcholine (PC), phophatidylethanolamine (PE), phosphatidylnositol (PI), phosphatidylserine (PS), lyso - PC, lyso - PE, lyso - PI, and lyso - PS than those of HCs ( Figure 1A ). To further characterize the metabolic state of lipids in RA monocytes, CD14⁺ monocytes from RA patients and age-matched HCs were subjected to Nile red staining to examine the accumulation of intracellular lipid droplets. Deletion of cytoplasmic neutral lipids in RA-monocytes resulted in a red intensity that was 1.74-fold lower in RA monocytes than in those from HCs, suggesting either impaired lipid synthesis or enhanced FAO in RA monocytes ( Figures 1B, C ).

Next, the gene expression of CIDEC (FSP27) and PLIN1, which is critical for the process of droplet formation, was examined (31). Expression of both genes was comparable to that of HCs ( Figure 1D ). Similarly, no significant differences in the key genes involved in lipid biosynthesis (FASN and SCD) were observed ( Figure 1E ) (31). However, the genes that are relevant for mitochondrial FAO pathway (ACADS, ACADM, ACADL, HADHB, and ACOX) were expressed at significantly higher levels in RA monocytes than HCs ( Figure 1F ), suggesting the reduced content of lipids in RA monocytes induced by metabolic activation of the FAO pathway.

CPT1 is the rate-limiting enzyme for β-oxidation of long- and medium-chain lipids in the mitochondria (16). To assess which subtype of CPT1 is upregulated in the CD14⁺ monocytes of patients with RA, the mRNA levels of CPT1A and CPT1B were determined in the CD14⁺ monocytes from age-matched HCs and RA patients. A significantly higher expression of CPT1A, but not CPT1B, was observed in the CD14⁺ monocytes of patients with RA than in those from the HCs (CPT1A, HCs:1.430 ± 0.081 vs. RA: 2.594 ± 0.112, P < 0.001; CPT1B, HCs: 0.949 ± 0.081 vs. RA: 0.878 ± 0.059, P = n.s.) ( Figure 2A ). To assess the impact of immunosuppressive therapy and the disease state on CPT1A induction, CPT1A expression was examined in patients with newly diagnosed RA who were not treated with disease-modifying anti-rheumatic drugs or underwent corticosteroid treatment within 3 months before recruitment to the study. There was no significant difference between CPT1A levels in the treated and untreated patients ( Figure 2B ), suggesting that drugs are not the causative factors of CPT1A upregulation. The increased expression of CPT1A protein, but not CPT1B in RA patients was also validated by western blotting ( Figures 2C, D ). Furthermore, there was a positive correlation observed between CPT1A gene expression in peripheral blood CD14⁺ monocytes and total modified Sharp/van der Heijde score (r = 0.787, P = 0.001) ( Figure 2E ) and erosion subscore (r = 0.719, P = 0.001) ( Figure 2F ), suggesting that high CPT1A expression in circulating osteoclast precursors may be associated with bone erosion in RA.

To explore a possible connection between CPT1A expression and osteoclastogenesis, the RA patients were further divided into two groups based on the relative gene expression levels of CPT1A in CD14⁺ monocytes, CPT1A ^low group (CPT1A < 3) and CPT1A ^high group (CPT1A > 3). The gene expression levels of C/EBPβ, RUNX2, Cathepsin K, and NFATc1, which associated with osteoclast-prone phenotype and function, were compared in the fresh isolated CD14⁺ monocytes from HC, RA patients with lower expression level of CPT1A, and RA patients with higher CPT1A expression. The strongly enhanced expression levels of those genes were observed in CPT1A ^high group than those in CPT1A ^low and HC groups ( Figure 2G ). This result indicated that CPT1A expression may associated with the dysfunction of osteoclast differentiation in RA patients. The samples were then cultured in the presence of 100 ng/ml RANKL and 50 ng/ml M-CSF for 21 days. TRAP staining and FITC-labeled phalloidin were used to detect mature osteoclasts in 24-well plates. The results in Figure 2H show that RA monocytes have a greater ability to form osteoclasts, as the numbers of mature osteoclasts in both groups of RA monocytes were much higher than that in HCs (HCs: 53.4 ± 3.8; CPT1A ^low RA: 155.6 ± 19.6; CPT1A ^high RA: 387.6 ± 46.7) ( Figure 2I ). As the overall size of the osteoclasts and number of nuclei were positively correlated with their resorption capability (32), our results also indicate that osteoclasts in the CPT1A^low RA and CPT1A ^high RA samples have a larger diameter (HCs: 47.20 ± 5.6 μm; CPT1A^low RA: 79.60 ± 9.7 μm; CPT1A^high RA 203.0 ± 18.2 μm), more nuclei for TRAP and F-actin (TRAP, HCs: 6.1 ± 0.7, CPT1A^low RA: 10.9 ± 1.2, CPT1A^high RA: 18.9 ± 1.7; F-actin, HCs: 9.2 ± 1.5, CPT1A^low RA: 16.1 ± 2.4, CPT1A^high RA: 22.3 ± 2.6) staining and significant higher pit area of bone resorption lacunae than those in HCs (HCs: 144.5 ± 15.7; CPT1A^low RA: 223.2 ± 14.9; CPT1A^high RA: 433.8 ± 22.6) ( Figure 2J – M ). In particular, the number of mature osteoclasts was significantly higher in RA patients with higher expression levels of CPT1A compared to patients with lower CPT1A expression (P <0.01), as was also found in case of osteoclast diameter (P <0.01), number of nuclei per osteoclast (P <0.01), and the pit area of bone resorption lacunae on day 21 (P <0.01) ( Figure 2I – M ). Meanwhile, induction of the genes associated with osteoclast differentiation and activity, such as C/EBPβ, RUNX2, Cathepsin K, TRAP, MMP-9 and NFATc1 was also detected using qPCR. There was significantly higher expression of these genes in the monocytes isolated from RA patients with higher CPT1A than those of patients with lower CPT1A and HCs ( Figure 2N ). This suggests that upregulation of CPT1A expression is not only associated with osteoclastogenesis but also with OCP fusion, which contributes to the formation of giant osteoclasts in RA patients.

Osteoclast fusion and multinucleation are pivotal steps in osteoclastogenesis. To assess osteoclast fusion dynamics under RA inflammatory conditions, the mRNA expression of ADAM8, OC-STAMP, DC-STAMP, CD47, and CD36 in the CD14⁺ monocytes from RA patients and HCs was detected by qPCR. The abundance of ADAM8, OC-STAMP, DC-STAMP, CD47, and CD36 mRNA was significantly higher in patients with RA than that in the HCs ( Figure 3A ). In addition, flow cytometry confirmed an approximately two-fold increase in in the mean fluorescence intensity (MFI) of DC-STAMP, CD47, and CD36 in the OCPs from RA patients compared to those in the HCs ( Figures 3B, C ).

To further assess the cellular membrane accumulation of the fusogenic molecules, DC-STAMP and CD36 expression was visualized by co-immunostaining with CD14, which is a cellular surface marker of monocytes. Elevated DC-STAMP/CD14 and CD36/CD14 co-localization signals, shown in yellow, were observed on the cell surface of monocytes in patients with RA. In contrast, the expression levels of both DC-STAMP and CD36 in the surface membranes were markedly lower in the monocytes of HCs than in those of RA patients ( Figures 3D, E ).

To assess whether elevated CPT1A levels contribute to osteoclastogenesis and fusion of circulating OCPs in RA, CD14⁺ monocytes derived from RA patients were transfected with small interfering RNA or lentiviral vector with specific CPT1A sequence to knock down the CPT1A expression. The transfected cells were cultured with 100 ng/ml RANKL and 50 ng/ml M-CSF for 21 days. CPT1A sequence-specific knockdown markedly reduced CPT1A gene and protein expression ( Figures 4A, B and Supplementary Figure 1A ), and also decreased the numbers of mature osteoclasts (con shRNA: 389.8 ± 17.2 vs. CPT1A shRNA: 202.47 ± 31.1, P < 0.001), cell diameter (con shRNA: 271.4 ± 23.6 μm vs. CPT1A shRNA: 102.1 ± 15.6 μm, P < 0.001), nuclei per osteoclast (TRAP, con shRNA: 18.4 ± 2.1 vs. CPT1A shRNA: 7.8 ± 0.9, P < 0.001; F-actin, con shRNA: 20.1 ± 2.8 vs. CPT1A shRNA: 8.4 ± 1.0, P < 0.001), and pit area of bone resorption lacunae (con shRNA: 453.4 ± 11.5 vs. CPT1A shRNA: 222.6 ± 22.5, P < 0.001) ( Figures 4C – H ). In addition, CPT1A knockdown resulted in the decrease in the expression of osteoclastogenesis genes (C/EBPβ, RUNX2, Cathepsin K, TRAP, MMP9, and NFATc1) ( Figure 4I ) and fusogenic genes that participated in OCP fusion (ADAM8, OC-STAMP, DC-STAMP, CD47, and CD36) ( Figure 4J ). This result also was conformed through the flow cytometry analysis of the cellular membrane expression of DC-STAMP, CD47, and CD36 ( Figures 4K, L ).

Alternatively, RA-derived CD14⁺ monocytes were treated with etomoxir, a pharmacological inhibitor that specifically inhibits the CPT1A-mediated FAO metabolic pathway. We did not find any distinguishable difference in the transcript levels of CPT1A after etomoxir treatment compared to those in the vehicle group, which is consistent with the previous studies demonstrated that etomoxir binds irreversibly to the catalytic site of CPT-1 and etomoxir treatment does not change the mRNA level of CPT1 in rat tissues (33, 34). However, this inhibitor effectively suppressed osteoclastogenesis-associated gene expression (C/EBPβ, Cathepsin K, and NFATc1). Combining with the results from CPT1A gene specific knockdown which rule-out the possible off-target effects of the etomoxir, it may suggest that this inhibitor mainly work through the inhibition of CPT1A enzymatic function ( Supplementary Figure 1B ). Furthermore, treatment with etomoxir resulted in a significant decrease in the cell number (Vehicle: 277.4 ± 26.8 vs. Etomoxir: 76.7 ± 13.1, P < 0.001), size (Vehicle: 257.6 ± 57.2 μm vs. Etomoxir: 92.47 ± 15.3 μm, P < 0.01) and the number of nuclei per osteoclast (TRAP, Vehicle: 18.1 ± 2.0 vs. Etomoxir: 8.2 ± 1.1, P < 0.001; F-actin, Vehicle: 22.8 ± 2.7 vs. Etomoxir: 9.3 ± 1.9, P < 0.001) of mature osteoclasts, as well as and pit area of bone resorption lacunae in patients with RA (Vehicle: 437.2 ± 11.7 vs. Etomoxir: 142.13 ± 9.9, P < 0.001) ( Figures 4M-R ).

Consistently, reduced mRNA expression osteoclastogenesis genes and cellular membrane distribution of fusogenic molecules were found in the monocytes treated with etomoxir ( Figures 4S – V ). In summary, these results confirm that CPT1A directly regulates OCP fusion.

Clathrin-mediated endocytosis plays a critical role in the cellular surface distribution of fusogenic molecules and OCP fusion (10). To evaluate whether clathrin-dominant endocytosis is augmented OCPs that are in the peripheral blood of RA patients, CD14⁺ monocytes were purified from PBMCs of both RA patients and HCs. There was a significant increase in the transcription levels of CLTA and CLTC in the monocytes of patients with RA compared to those in the HCs (CLTA: HCs: 2.5 ± 0.3 vs. RA: 4.7 ± 0.6, P < 0.001; CLTC: HCs: 2.8 ± 0.2 vs. RA: 7.6 ± 0.4, P < 0.001) ( Figure 5A ). Western blotting results also showed that both the CLTA and CLTC band intensities were significantly upregulated in RA-derived monocytes compared to those in HCs ( Figures 5B, C ).

To clarify whether clathrin expression is regulated by the CPT1A-mediated FAO metabolic pathway, CD14⁺ monocytes purified from RA patients were treated with etomoxir or CPT1A specific RNA interference ( Figures 5D – F ). Impaired CPT1A enzymatic activity and CPT1A gene expression significantly downregulated the mRNA and protein levels of CLTA and CLTC in the monocytes from RA patients ( Figures 5G – I ).

Podosomes or podosome-like structures are required for OCP fusion (35, 36). Next, the effect of clathrin upregulation on OCP fusion in RA was assessed. To do this, the formation of podosome structure was examined in RA patients and HCs at day 3 after the induction of osteoclast differentiation. In RA patients, markedly elongated F-actin–rich protruding structure were found between OCPs, which had clear DC-STAMP and clathrin co-localization with the F-actin ring. The number of podosome structures was significantly higher in the OCPs of RA patients than those of the HCs ( Figures 5J, K ). Furthermore, upon RANKL stimulation, DC-STAMP was internalized by clathrin-dependent mechanisms in RA OCPs with DC-STAMP co-localization with clathrin endocytic vesicles. However, this colocalization signal was much less in HCs. Both the protruding F-actin-rich structures and colocalization of DC-STAMP and clathrin endocytic vesicles were successfully disrupted by CPT1A enzymatic inhibition or CPT1A knockdown ( Figures 5J, K ), indicating that CPT1A promotes OCP fusion in RA patients via activating clathrin-dependent endocytic activity.

C/EBPβ is a master transcription factor involved in lipid metabolism and also involved in osteoclast differentiation (37–39). The results of transcript factor analysis also suggested that C/EBPβ is a potential transcription factor may bind to the promoters of CLTA and CLTC. Therefore, C/EBPβ expression in the monocytes was examined in patients with RA and HCs. The abundance of C/EBPβ mRNA and protein in the monocytes from RA patients was significantly higher than in those of HCs ( Figures 6A – C ). To clarify whether enhanced C/EBPβ expression is regulated by CPT1A, the transcript and protein expression levels of C/EBPβ were detected in RA monocytes treated with or without etomoxir. Both transcript and protein expression levels of C/EBPβ were dramatically reduced in the RA monocytes cultured with etomoxir ( Figures 6D – F ).

To further investigate whether C/EBPβ could bind to the promoters of CLTA and CLTC, C/EBPβ chromatin immunoprecipitation (ChIP) assays was performed in RA-derived monocytes treated with or without etomoxir. ChIP analysis results confirmed the immunoprecipitation of C/EBPβ and the CLTA and CLTC promoters ( Figure 6G ). Clear bands that decrease C/EBPβ bindings to the CLTA and CLTC promoters in RA monocytes supplied with etomoxir were observed ( Figures 6G, H ).

In an alternative approach, the function of CPT1A was inhibited by siRNA interference by transfecting RA monocytes with CPT1A specific-interfering sequences. Reduction in CPT1A gene expression significantly downregulated the transcript and protein expression levels of C/EBPβ ( Figures 6I – K ), and also strongly attenuated the binding of C/EBPβ to the CLTA and CLTC promoter regions, indicating that CPT1A-mediated FAO metabolism upregulates CLTA and CLTC expression by promoting C/EBPβ binding to the CLTA and CLTC promoters ( Figures 6L, M ).