Identification of RIFL, a novel adipocyte-enriched insulin target gene with a role in lipid metabolism

Abstract

To identify new genes that are important in fat metabolism, we utilized the Lexicon-Genentech knockout database of genes encoding transmembrane and secreted factors and whole murine genome transcriptional profiling data that we generated for 3T3-L1 in vitro adipogenesis. Cross-referencing null models evidencing metabolic phenotypes with genes induced in adipogenesis led to identification of a new gene, which we named RIFL (refeeding induced fat and liver). RIFL-null mice have serum triglyceride levels approximately one-third of wild type. RIFL transcript is induced >100-fold during 3T3-L1 adipogenesis and is also increased markedly during adipogenesis of murine and human primary preadipocytes. siRNA-mediated knockdown of RIFL during 3T3-L1 adipogenesis results in an ~35% decrease in adipocyte triglyceride content. Murine RIFL transcript is highly enriched in white and brown adipose tissue and liver. Fractionation of WAT reveals that RIFL transcript is exclusive to adipocytes with a lack of expression in stromal-vascular cells. Nutritional and hormonal studies are consistent with a prolipogenic function for RIFL. There is evidence of an approximately eightfold increase in RIFL transcript level in WAT in ob/ob mice compared with wild-type mice. RIFL transcript level in WAT and liver is increased ~80- and 12-fold, respectively, following refeeding of fasted mice. Treatment of 3T3-L1 adipocytes with insulin increases RIFL transcript ≤35-fold, whereas agents that stimulate lipolysis downregulate RIFL. Interestingly, the 198-amino acid RIFL protein is predicted to be secreted and shows ~30% overall conservation with the NH(2)-terminal half of angiopoietin-like 3, a liver-secreted protein that impacts lipid metabolism. In summary, our data suggest that RIFL is an important new regulator of lipid metabolism.

Fig. 1.

Phenotype of RIFL (refeeding-induced fat and liver)-null mice and sequence of murine RIFL gene and protein. A: reduced serum triglyceride phenotype. Data taken from the Lexicon-Genentech null mouse phenotyping project was plotted to demonstrate levels of serum triglyceride (mg/dl) in wild-type (WT) littermates [n = 2 females (F) and 2 males (M)] and in RIFL-null mice (n = 4 F and 4 M). Data is shown as means ± SE. *P < 0.005 vs. WT mice. B: nucleic acid and protein sequence for murine RIFL. The italicized and underlined region indicates the predicted signal peptide. The 2 gray downward arrows demarcate the region of homology with murine angiopoietin-like 3 (ANGPTL3). Nos. at left indicate amino acid positions. C: alignment of murine RIFL protein sequence with that of murine ANGPTL3 and ANGPTL4. Nos. at left indicate amino acid positions, and dashes indicate discontinuity of sequence alignment. +Amino acid similarity. D: Western blot analysis of ectopic RIFL protein expression in 3T3-L1 adipocytes. Adipocytes were electroporated with either pcDNA3.1 [empty vector (EV)] or the RIFL-HA expression construct. Cell lysate was collected at 48 h posttransfection and analyzed using an HA primary antibody (top) or a peptidylpropyl isomerase A (PPIA) antibody for loading control (bottom). Digital images of enhanced chemiluminesence data are shown. Nos. on right are protein mass markers in kDa. The arrow shows the major species of RIFL-HA protein. *A slightly larger RIFL-HA protein species.

Fig. 2.

Dramatic increase in RIFL transcript with in vitro adipogenesis of 3T3-L1 and ScAP-23 preadipocytes. A: increase in RIFL transcript expression during 3T3-L1 adipogenesis. Real-time PCR was carried out on cDNA prepared from RNA collected from day 0 (D0) and daily through day 7 (D7). Top: RIFL transcript. Bottom: transcripts for 2 adipogenesis marker genes. The bracketed area marked as NS (not significant) indicates lack of significant change in RIFL transcript level between D0 and D1 or D2. One of 2 representative analyses is shown for the D0–D7 time course. B: increase in RIFL transcript expression during ScAP-23 adipogenesis. Top: real-time PCR data for ScAP-23 in vitro adipogenesis (D indicates days postinduction of adipogenesis). Bottom: adipogenesis marker transcripts. For A and B the respective values in each graph are normalized to the D0 value, which is set to 1. Bars represent means of measurement triplicates ± SD. *P < 0.001 and #P < 0.05 with respect to the value on D0. PPARγ, peroxisome proliferator-activated receptor-γ; FABP4, fatty acid-binding protein 4.

Fig. 3.

Upregulation of RIFL transcript in a brown adipogenesis model. Top: real-time PCR data for RIFL transcript level during in vitro adipocyte conversion of a brown preadipocyte cell line, WT-BAT (brown adipose tissue). Bottom: marker genes for brown adipogenesis [uncoupling protein 1 (UCP1) and cell death-inducing DNA fragmentation factor-α-like effector A (CIDEA)] and for general adipogenesis (PPARγ and FABP4). D indicates days postinduction of adipocyte differentiation. Bars represent means of measurement triplicates ± SD. For PCR data the respective values in each graph are normalized to the D0 value, which is set to 1. *P < 0.001 compared with respective value for D0.

Fig. 4.

Upregulation of RIFL transcript level in adipogenesis of murine and human primary preadipocytes. A: increase in RIFL transcript level in adipogenesis of murine primary preadipocyte cultures. Left: real-time PCR assessment for transcript expression for murine RIFL. Left middle, right middle, and right: real-time PCR assessment for adipogenesis marker transcripts. B: increase in RIFL transcript level in adipogenesis of human primary preadipocyte cultures. Left: real-time PCR assessment for human RIFL transcript. Left middle, right middle, and right: real-time PCR assessment for 3 adipogenesis marker genes. One of 3 representative sets of preadipocyte (P) and in vitro-differentiated adipocyte (A) are shown. Bars indicate means of measurement triplicates ± SD. *P < 0.001 vs. value for preadipocytes, set to a value of 1 in the respective graphs. SCD1, stearoyl-CoA desaturase 1; PNPLA2, patatin-like phospholipase domain containing 2.

Fig. 5.

Downregulation of RIFL transcript in 3T3-L1 adipocytes with TNFα treatment and by PPARγ knockdown. A: decreased level of RIFL transcript in 3T3-L1 adipocytes with 24-h treatment with TNFα. 3T3-L1 adipocytes were treated for 24 h with either vehicle [control (C)] or 10 ng/ml TNFα, and transcript levels were assessed by real-time PCR for RIFL transcript and for 4 previously reported TNFα target genes in adipocytes. Bars represent means of measurement triplicates ± SD. Level in control was set to a value of 1; *P < 0.01. One of 2 representative experiments is shown. B: decreased level of RIFL transcript in human adipocytes with TNFα treatment. In vitro-differentiated human adipocytes were untreated (C) or treated for 36 h with 10 ng/ml TNFα, and RIFL transcript was assessed by real-time PCR. Bars indicate means ± SD; n = 2 for control and n = 2 for treated. Control was set to a value of 1; *P < 0.01 with respect to the control value. C: effective siRNA-mediated knockdown of PPARγ siRNA for PPARγ or nontargeting control (Con) was introduced into 3T3-L1 adipocytes, as described in materials and methods. Shown is Western blot analysis for PPARγ and for loading control (PPIA). Duplicate knockdown experiments are shown for each of control and PPARγ. D: effect of adipocyte PPARγ knockdown on RIFL transcript level. Real-time PCR analysis for indicated transcripts in 3T3-L1 adipocytes electroporated with nontargeting control siRNA (Con) or siRNA for PPARγ. Bars represent means ± SD; n = 2 for control and n = 2 for PPARγ knockdown. Control mean was set to a value of 1; *P < 0.02.

Fig. 6.

siRNA-mediated knockdown of RIFL during 3T3-L1 adipogenesis decreases lipid content. A: assessment of effectiveness of siRNA-mediated knockdown of RIFL transcript and protein. Top: real-time PCR analysis of RNA prepared from 3T3-L1 adipocytes electroporated with either siRNA for nontargeting control siRNA (siCon) or siRNA for RIFL (siRIFL). Bars represent means of measurement triplicates ± SD. Level in control was set to a value of 1; *P < 0.001. One of 2 representative experiments is shown. Bottom: Western blot using HA-tag antibody conducted on lysates of 293T cells transiently transfected with either EV or the RIFL-HA expression construct. For RIFL-HA transfections, cells were cotransfected with either siCon or siRIFL. Western blot for tubulin is shown as loading control. Results of 2 independent siRNA knockdowns are shown each for siCon and siRIFL. B: effect of RIFL knockdown on intracellular lipid content; microscopic view. Microscopic view of day 7 adipocytes following electroporation of indicated siRNAs into 3T3-L1 preadipocytes, which were then subjected to an adipogenic differentiation protocol. Top: live, unstained cells. Middle and bottom: paraformaldehyde fixed, Oil Red O (ORO)-stained cultures. Relative magnifications are indicated in parentheses. C: effect of RIFL knockdown on intracellular lipid content; view of whole culture wells. Cells harboring indicated siRNA were fixed with paraformaldehyde and stained with ORO. Culture plate was scanned, and resultant digital image is shown. For B and C, images are representative of 3 independent experiments. D: quantification of culture neutral intracellular lipid content. Triplicate independent wells (n = 3) each were transfected and analyzed for siCon or siRIFL. Following differentiation, cultures were stained with ORO and extracted with isopropanol, and absorbance was measured at an optical density of 490 nm. Data are shown as means of triplicate wells ± SD; *P < 0.01. Results of 2 independent studies are shown (black bars, experiment 1; gray bars, experiment 2). For each experiment, level in the siCon was set to a value of 1. E: assessment of adipocyte marker transcripts. Real-time PCR was used to assess level of indicated mature adipocyte marker gene expression in control siRNA cultures (C; black bars) or in RIFL siRNA knocked-down cultures (R; gray bars). For each indicated transcript, the value of the respective control was set to 1. *Nonsignificant difference (P > 0.05) for control siRNA vs. RIFL siRNA. Bars represent means ± SD of n = 3 culture dishes for either control siRNA or RIFL siRNA. The study was carried out 2 independent times with essentially the same results; 1 of the 2 studies is shown.

Fig. 7.

Tissue and cell type enrichment of RIFL transcript expression. A: real-time PCR assessment of RIFL transcript expression in a panel of murine tissues. Level in heart was set to a value of 1. B: in white adipose tissue (WAT), RIFL transcript expression is restricted to adipocytes. Real-time PCR analysis was carried out on murine WAT tissue that had been subjected to cellular fraction into the preadipocyte-containing stromal vascular fraction (SVF) or the adipocyte fraction (AF). Signal in SVF was set to a value of 1. Left: data for the RIFL transcript; middle left and middle right: data for adipocyte marker transcripts; right: level of a preadipocyte marker transcript. C: relative expression level of RIFL transcript in human liver, adipocytes, and WAT. Human adipocytes obtained from in vitro differentiation of human primary preadipocytes and human WAT, liver, testis, and heart RNA were analyzed for relative transcript expression for human RIFL using real-time PCR. Level in human heart sample was set to a value of 1. For A –C, data shown are means of measurement triplicates ± SD; *P < 0.001 vs. unmarked samples (i.e., no asterisk) within the same graph.

Fig. 8.

Increased expression of RIFL transcript in tissues of genetically obese (ob/ob) mice. A: RIFL transcript level in WAT of ob/ob mice. Real-time PCR analysis was conducted to assess RIFL transcript levels in either WT or genetically obese (ob/ob) epididymal WAT of male mice. B: RIFL transcript level in liver of ob/ob mice. Real-time PCR analysis was conducted to assess RIFL transcript levels in either WT or genetically obese (ob/ob) liver tissue of male mice. A and B, left: RIFL transcript level; middle and right: levels of transcripts used as controls, as described in the text. Bars represent means ± SD of n = 5 WT and n = 5 ob/ob mice. *P < 0.05 for ob/ob vs. respective WT level, with the mean level of each WT group set to 1.

Fig. 9.

Nutritional Regulation of RIFL in fasted/refed mice. A: marked increase in RIFL transcript expression level in WAT by refeeding of fasted mice. Mice were subjected to a fasting and refeeding dietary regimen, as described in materials and methods. Real-time PCR analysis was used to determine transcript level in WAT for RIFL (top) and for the control transcripts PNPLA3 and PNPLA2, which are known to be increased by refeeding and fasting, respectively. B: marked increase in RIFL transcript expression level in liver by refeeding of fasted mice. Mice were subjected to a fasting and refeeding dietary regimen, as described in materials and methods. Real-time PCR analysis was used to determine transcript level in liver for RIFL (top) and for the control transcripts PNPLA3 and PNPLA2, which are known to be increased by refeeding and fasting, respectively. For A and B, bars represent means ± SD; n = 2 fasted mice and n = 2 refed mice. *P < 0.001 for refed vs. fasted, with the mean level of the fasted group set to a value of 1.

Fig. 10.

RIFL is a new insulin-induced gene in adipocytes. A: upregulation of RIFL transcript level by insulin in 3T3-L1 adipocytes. 3T3-L1 adipocytes were untreated [control (C)] or treated with 100 nM insulin (Ins) for 48 h under serum-free conditions, as described in materials and methods. Levels for RIFL transcript and for transcripts of established insulin target genes (CIDEC, PNPLA3, and PNPLA2) were analyzed by real-time PCR. Three independent experiments were conducted, with 1 representative study shown. B: effect of insulin on RIFL transcript in human adipocytes. In vitro-differentiated human adipocytes were treated with insulin for 36 h, and RNA was assessed for RIFL transcript level with real-time PCR. Bars represent means of Ins-treated (n = 2) and C (n = 2) ± SD; *P < 0.005 for Ins vs. C. C: time course of RIFL transcript upregulation by Ins. 3T3-L1 adipocytes were exposed to 100 nM Ins starting at time 0 and RNA harvested at indicated hourly intervals thereafter through 48 h. #P < 0.05, ^P < 0.01, and *P < 0.001 vs. the 0 h, which was set to a value of 1. D: dose response of RIFL transcript regulation by insulin. 3T3-L1 adipocytes were exposed from 0 to 1,000 nM insulin, as indicated, and samples analyzed by real-time PCR 24 h later. Level on day 0 is set to a value of 1. For A, C, and D, bars represent means of measurement triplicates ± SD; *P < 0.001 vs. either control (for A) or day 0 (for C and D), which was set to a value of 1. E: regulation of RIFL transcript level in 3T3-L1 adipocytes by insulin and glucose singly or in combination. 3T3-L1 adipocytes were cultured as described in materials and methods; control (Con), glucose-treated only (Glu), insulin-treated only (Ins), and treatment with glucose and insulin combined (Glu + Ins). RNA was analyzed for either RIFL transcript level (left) or PNPLA3 transcript level (right) by real-time PCR. Bars represent means (n = 2) for each condition ± SD, with control set to a value of 1. For RIFL data, *P < 0.01 for Glu + Ins vs. Con, Glu, or Ins. For the PNPLA3 graph, *P < 0.01 for Glu or Glu + Ins vs. Con or Ins.

Fig. 11.

RIFL transcript in adipocytes is decreased by agents that stimulate lipolysis. A: effect of dibutyrl cAMP (db-cAMP), forskolin (Forsk), and isoproterenol (Iso) on RIFL transcript level in adipocytes. 3T3-L1 adipocytes were exposed for 24 h to vehicle control (Con), db-cAMP, Forsk, or Iso, and RIFL transcript level (top) or levels of fatty acid synthase (FASN) and glucose transporter 4 (GLUT4) transcript (middle and bottom) were measured by real-time PCR. Because Iso treatment was carried out in a separate study from that in A, it is shown in a separate graph with its respective control. B: dexamethasone (DEX) decreases RIFL transcript level in adipocytes. 3T3-L1 adipocytes were treated with 1 μM DEX for 24 h, and RIFL transcript was assessed by real-time PCR. For A and B, bars represent means (n = 2 for control and n = 2 for treated) ± SD, with level in control set to a value of 1. *P < 0.02 and #P < 0.05 for treatment with indicated agent vs. respective control.