A fluorescent reporter assay of differential gene expression response to insulin in hepatocytes

Abstract

Insulin regulates multiple hepatic metabolic pathways in a seemingly heterogeneous manner. To understand this heterogeneity, we hypothesized that different subpopulations of hepatocytes have different sensitivity to insulin. To test this hypothesis, we developed a fluorescent reporter in which the insulin-responsive fatty acid synthase (FAS) promoter drove expression of a time-dependent fluorescent protein (“timer”) and characterized timer expression in flow-sorted cell populations. In Hepa1c1c7 and AML12 hepatocytes, we found that different cell populations express distinct timer fluorescence following insulin treatment, consistent with cellular heterogeneity in the response to insulin. RNA measurements indicated an enrichment of forkhead box O transcription factors in cells with a greater response to insulin. Moreover, we found evidence of increased Akt activation. These data are consistent with a heterogeneous cellular response to insulin and raise the possibility that these different subpopulations underlie the peculiar pathophysiology of hepatic insulin resistance.

INTRODUCTION

Hepatic insulin action is important in normal and disease states (18, 22). Impaired insulin action, especially in liver, adipose, and skeletal muscle, leads to defects in intracellular signaling pathways needed to maintain proper glucose homeostasis (6).

In a healthy liver, insulin promotes glucose utilization to generate energy, and its storage as glycogen. The latter is mediated through insulin-stimulated phosphorylation and inhibition of glycogen synthase kinase 3 (GSK3). Inactivation of GSK3 also allows another substrate, eukaryotic initiation factor 2B, to promote protein synthesis and storage of amino acids (18). In addition, insulin inhibits gluconeogenesis, while increasing the synthesis of triglycerides. Forkhead box O (Foxo) 1 is a key regulator of insulin-dependent glucose utilization vs. release, acting to promote gluconeogenesis through glucose-6-phosphatase and inhibit glucose utilization by suppressing glucokinase (11, 17, 28, 40).

Insulin resistance disrupts these processes through a variety of mechanisms, including inhibitory phosphorylation of insulin-signaling molecules (1), inflammatory pathways (4, 15, 25), lipotoxicity (32), and endoplasmic reticulum (ER) stress (31, 33). Hepatic insulin resistance has hepatocyte cell-autonomous and nonautonomous causes. For example, nitric oxide generated in endothelial sinusoids can affect glucose production (37). In addition, there is evidence of cellular heterogeneity of hepatocyte metabolism (23). For example, the periportal zone is thought to preferentially perform glycolysis and lipo- and ketogenesis, whereas the perivenous zone performs gluconeogenesis and fatty acid oxidation (20). It has been suggested that there is differential distribution of insulin receptor substrates 1 and 2 within the periportal and perivenous zones (24). However, new methods can shed light on molecular events underlying such heterogeneity in this organ.

Here, we developed a fluorescent reporter in which the insulin-responsive fatty acid synthase (FAS) promoter is used to drive expression of a time-dependent fluorescent protein. This reporter can discriminate between subpopulations of hepatocytes with variable sensitivity to insulin. Using this construct, we demonstrate a unique population of hepatocytes that is highly sensitive to insulin.

MATERIALS AND METHODS

Cloning constructs.

TimerS (pSlow-FT-N1; Addgene plasmid no. 31912) was a gift from Vladislav Verkhusha (36), pClover (pLenti-Foxo1-Clover; Addgene plasmid no. 67759) was a gift from Peter Rotwein (16), Foxo1 fused to a red fluorescent protein (FoxRFP) plasmid was previously generated in laboratory (Addgene plasmid no. 34678), and superfolder green fluorescent protein (GFP) was a gift from Erik Snapp (3). To create the FAStimerS plasmid, recognition sequences for the restriction enzymes AseI (5′-end) and XhoI (3′-end) were first added to a 257-bp fragment of the fatty acid synthase promoter (FASN) using primer extension PCR. The pGL2-FAS-luciferase (21) was used as a template for FASN. With the use of these restriction sites, the FASN DNA fragment was then cloned (using AseI and XhoI restriction enzymes and T4 DNA ligase purchased from New England Biolabs) into a precut TimerS vector backbone where the CMV promoter had already been removed and DNA purified from an agarose gel using the QIAquick gel extraction kit (QIAGEN). The sequences of constructs were validated before experiments.

Cell culture, transfections, and treatments.

Hepa1c1c7 cells were a generous gift (Utpal Pajvani, Columbia University) and cultured in DMEM with 4.5 g/l glucose and l-glutamine without sodium pyruvate (Corning) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (P/S). AML-12 cells were a generous gift (Lale Ozcan, Columbia University) and cultured in DMEM-F-12 with GlutaMAX supplemented with 10% FBS and 1% P/S. Cells were transfected using Lipofectamine 3000 Reagent (ThermoFisher Scientific) according to the manufacturer’s instructions. To promote uptake of DNA to cells, cells were cultured in serum-free DMEM 2 h before transfections. In addition, fresh medium was given to the cells 3 h following transfection. Before insulin treatment, cells were cultured in serum-free DMEM for 4 h. Upon addition of insulin (Humalog) to the cells, unless otherwise indicated, medium was replaced to medium containing DMEM + 10% FBS.

Immunoblotting.

Cells were collected in single detergent lysis buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, 1% Triton X-100, and protease and phosphatase inhibitor cocktails (Cell Signaling) at 1× dilution. Cells were then lysed by sonication. Immunoblotting was performed with primary antibodies generated in rabbit directed toward phospho-Akt (S473) [Cell Signaling (RRID: AB_2315049), 1:2,000], phospho-GSK-3β (S9) [Cell Signaling (RRID: AB_10013750), 1:1,000], and RFP [which recognizes Timer; Rockland Immunochemicals (RRID: AB_2209751), 1:1,000] or primary antibodies made in mouse targeted to Akt [Cell Signaling (RRID: AB_1147620); 1:2,000], GSK-3β [Cell Signaling (RRID: AB_10839406), 1:1,000], and α-tubulin [Sigma-Aldrich (RRID: AB_477593), 1:20,000] diluted in Odyssey Blocking Buffer (PBS) (LI-COR). IRDye 800 and 680 secondary antibodies [LI-COR (RRID: AB_621847, AB_10954442, AB_621848, and AB_10953628), 1:10,000] of corresponding species were used and signal detected using LI-COR Odyssey. ImageJ software (National Institutes of Health) was used to quantify protein levels.

Fluorescence-activated cell sorting.

For flow cytometry analysis, cells were collected and fixed for 5 min in 2% paraformaldehyde. Cells were resuspended in 1× PBS supplemented with 0.5% FBS and analyzed using a BD LSRII. Red timer protein was detected with an orange (594-nm) laser and 620/20 band pass filter and the blue timer protein with a violet (405-nm) laser and 450/50 band pass filter. For cell sorting, cells remained unfixed and resuspended in ALDEFLUOR Assay Buffer (Stem Cell Technologies) or cell culture media. Cells were sorted using a BD Influx with a green (561-nm) laser and 610/20 band pass filter to detect red timer and a violet (405-nm) laser and 460/50 band pass filter to detect blue timer.

Imaging and immunofluorescence.

Cells were cultured in chamber slides (CORNING) and fixed with 4% paraformaldehyde for 10 min at room temperature. After permeabilization, cells were blocked in 10% donkey serum (Jackson Immunoresearch Laboratories) for 1 h at room temperature. Primary antibody was rabbit anti-phospho-Akt (S473) [Cell Signaling (RRID: AB_2315049), 1:25], and donkey anti-rabbit Alexa Fluor 488 [Molecular Probes (RRID: AB_141708), 1:1,000] was used for detection of primary antibody. All samples were covered with a coverslip with a VECTASHIELD HardSet Antifade Mounting Medium with DAPI (Vector Laboratories). Images were taken with a Zeiss LSM 510/710 confocal microscope (Zeiss) using a EC-Plan-Neofluor ×10 or Plan-Apochromat ×20 objective.

qPCR analysis.

RNA was isolated from cells sorted by fluorescence-activated cell sorting (FACS) using an Arcturus PicoPure RNA Isolation Kit (Applied Biosystems) according to the manufacturer’s protocol. cDNA was generated using qScript cDNA SuperMix (Quanta Biosciences) as suggested by the manufacturer and qPCR reactions completed using GoTaq qPCR Master Mix (Promega). Primers used are described in Table 1. Measurements were made using a Bio-Rad CFX96 real-time PCR system, and relative gene expression levels were determined by the ΔΔC_t method with β-actin as the reference gene.

Table 1.

Primers used for qPCR

Gene	Sequence
Timer	AAGCTGACCCTGAAGTTCATCTGC
	CTTGTAGTTGCCGTCGTCCTTGAA
FAS	TTGGCCCAGAACTCCTGTAG
	CTCGCTTGTCGTCTGCCT
Srebp-1c	CGACTACATCCGCTTCTTGCAG
	CCTCCATAGACACATCTGTGCC
IR	GAGAGGATGTGAGACGACG
	CAGGTTGTTCCGGATGTCC
IRS1	CCCGTTCGGTGCCAAATAGC
	GCCACTGGTGAGGTATCCACATAGC
IRS2	ACTTCCCAGGGTCCCACTGCTG
	GGCTTTGGAGGTGCCACGATAG
Foxo3	TTGTCCCAGATCTACGAGTGGA
	CGTGCCTTCATTCTGAACGCGCA
Foxo4	TGAACTCCTTGCGTCAGTCACC
	CGGTGCTAGCCTGAGACATCAA
Foxo1	TCCAGTTCCTTCATTCTGCACT
	GCGTGCCCTACTTCAAGGATAA
β-Actin	TACCACAGGCATTGTGATGG
	TTTGATGTCACGCACGATTT

FAS, fatty acid synthase; Srebp-1c, sterol regulatory element-binding protein-1c; IR, insulin receptor; IRS1, insulin receptor substrate-1; Foxo, forkhead box O.

Statistical analysis.

Data are presented in graphs in Figs. 1–7 as means ± SE. Comparisons between two groups were conducted using a two-tailed Student’s t-test. For comparisons between multiple groups, analysis was performed using analysis of variance followed by Tukey’s post hoc test. Statistical tests were completed using GraphPad Prism 7 software (La Jolla, CA), and P < 0.05 was deemed significant.

Fig. 1.

Detection of fluorescent timers by fluorescence, immunohistochemistry, and FACS. A: representative fluorescent images of reporter vectors in Hepa1c1c7 cells 48 h after transfection with pTimerSlow (pTimerS), pTimerMedium (pTimerM), p-superfolder (sf) green fluorescent protein (GFP) (control for GFP), and pFoxo1 fused to red fluorescent protein (RFP, control for RFP). Images were taken with a ×10 objective, and the scale bar is 200 μM. B and C: representative immunoblots for the indicated proteins in Hepa1c1c7 cells transfected with vectors for 48 h demonstrating Timer is recognized with an antibody directed toward RFP but not GFP. Arrowhead indicates a nonspecific band. D: FACS dot plot of Hepa1c1c7 cells expressing pTimerSlow 24 h after transfection. Cells were first gated by forward (FSC) and side (SSC) scatter (top) and separated by blue and red fluorescence (bottom).

RESULTS

Using fluorescent timers to identify cell populations.

To create a cell-based technique to assess heterogeneity, we took advantage of the unique properties of fluorescent timer proteins that change their emission from the blue to red wavelength as a function of time (36). For example, timer slow (timerS) has a blue excitation/emission peak of 402/465 nm with a red excitation/emission peak of 583/604. This shift in wavelength permits individual detection of each timer form by fluorescence (Fig. 1, A and D). The half-life of timerS blue fluorescence under culture conditions is 9.8 h. Accordingly, cells expressing timerS have a mixture of blue and red fluorescence when analyzed by FACS 24 h after transfection (Fig. 1D). It is also possible to measure timer abundance by immunoblot with an antibody directed toward RFP, but not GFP (Fig. 1, B and C).

To establish a cell system, we examined insulin signaling in murine Hepa1c1c7 hepatocytes exposed to different concentrations of insulin overnight. We observed that phosphorylation of the protein kinases Akt at S473 (Fig. 2, A and B) and GSK-3β at S9 (Fig. 2, C and D) increased in an insulin dose-dependent manner.

Fig. 2.

Insulin-induced signaling in hepatoma cells. A and C: representative immunoblots for the indicated proteins. Hepa1c1c7 cells transfected with FAStimerS for 24 h followed by overnight (O/N) treatment with insulin (ins) at the indicated dose. B and D: quantification of Akt (B) or glycogen synthase kinase (GSK)-3β (D) phosphorylation. All graphs represent the average of at least 3 independent experiments. U, untransfected cells; C, vehicle control. *P < 0.05, **P < 0.01, and ****P < 0.0001 versus untransfected cells. #P < 0.05 and ####P < 0.0001 versus control cells.

A fluorescent reporter driven by FAS promoter.

Next, we established a reporter system to monitor the transcriptional response to insulin. To this end, we cloned timerS downstream of the insulin-responsive fatty acid synthase promoter (FAStimerS). Following transfection of the FAStimerS reporter into Hepa1c1c7 cells, levels of the timer reporter increased 50–100% upon stimulation by insulin in a time- and dose-dependent manner (Fig. 3, A–C). This was accompanied by an increase in the number of blue fluorescent cells from 4 to 12% (Fig. 3D).

Fig. 3.

FAStimer induction by insulin. A: representative immunoblot for induction of timer reporter protein [recognized by a red fluorescent protein (RFP) antibody] in Hepa1c1c7 cells after overnight (O/N) insulin (ins) treatment. B: quantification of A. Graph represents average of 3 independent experiments. C: immunoblot of Hepa1c1c7 cells transfected with FAStimerS and cultured in serum-containing (+serum) or serum-free (–serum) medium. D: representative fluorescent images (left) and quantification (right) demonstrating induction of timer blue in Hepa1c1c7 cells after insulin treatment. Images were taken with a ×10 objective, and the scale bar is 200 μM. U, untransfected cells; C, vehicle control. *P < 0.05 and ****P < 0.0001 versus control. Arrowhead indicates a nonspecific band.

Using the shift in fluorescence of the FAStimerS reporter, we compared the response of the FAS promoter to overnight stimulation with 10 nM insulin, a condition known to reduce insulin signaling (14). The nonfluorescent population (~70% of total) did not change (Fig. 4, A and B), indicating that insulin treatment did not result in an increased transfection efficiency. The size of this nonfluorescent population is likely because of a combination of transfection efficiency and deletion of the CMV promoter from the reporter vector backbone. Cells expressing the FAStimerS could be divided into three subsets based on their fluorescence in response to insulin. Fifteen percent of cells showed only red fluorescence [timer late (T_L), Fig. 4D], i.e., they did not translate new FAStimer protein, consistent with an early insulin response. In contrast, the populations of cells expressing red and blue, i.e., with ongoing FAStimer synthesis indicative of continued responsiveness to insulin, increased from 1.2 to 2% [timer middle (T_M)] and from 4 to almost 6% [timer early (T_E), Fig. 4C]. These data indicate that there are different populations of cells with regard to regulation of the fluorescent FAStimerS reporter, consistent with the possibility that cells differ in their response to insulin treatment.

Fig. 4.

Different populations of insulin-responsive FAStimer-expressing cells. A: FACS dot plot for blue and red fluorescence of Hepa1c1c7 cells expressing FAStimerS and treated overnight (O/N) with insulin (Ins). Shaded regions represent cell populations timer early (T_E), timer middle (T_M), and timer late (T_L). B–D: quantification of cells in A having no fluorescence (B), those induced upon insulin treatment (C), and a cell population that is unchanged after long-term insulin treatment (D). C, control. Each bar represents the average of at least 3 independent replicates. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Increased Foxo mRNA levels in cells with elevated FAS reporter activity.

We compared gene expression in the four different cell populations (Fig. 5A). We chose transcripts encoding insulin-signaling components or insulin-responsive genes. We saw the largest increases in Timer and FAS mRNA in the T_E population, validating the assumption that these cells have active transcription of the FAStimerS construct (Fig. 5, B and C). Both T_M and T_E populations also had increased expression of sterol regulatory element-binding protein-1c (SREBP1-c), a FAS transcription factor, consistent with the elevated levels of reporter expression (Fig. 5D). Although there were no significant changes in insulin receptor or insulin receptor substrate-1 mRNA between cell populations, the T_E group had a threefold increase in insulin receptor substrate 2 when compared with the nonfluorescent (N) cells (Fig. 5, E–G). Interestingly, mRNAs encoding Foxo1, Foxo3, and Foxo4 were significantly increased in the T_E population (Fig. 5H). Specifically, Foxo1 mRNA was ≥ 50-fold higher in the highly sensitive T_E cells compared with nonfluorescent N cells and ≥ 20-fold higher in T_E when compared with nonresponder T_L cells (Fig. 5H).

Fig. 5.

Insulin-responsive liver cells have increased forkhead box O (Foxo) expression. A: FACS dot plot of Hepa1c1c7 control cells (left) or cells expressing FAStimerS after overnight insulin (Ins) treatment (right) separated by blue and red fluorescence. Cells were sorted into indicated cell populations: timer early (T_E), timer middle (T_M), timer late (T_L), and nonfluorescent (N). B–H: quantitative PCR analysis of cell populations shown in A for expression of the indicated genes. mRNA levels are normalized to β-actin. All bars represent average values of at least 5 independent replicates. *P < 0.05, **P < 0.01, and ***P < 0.001 versus N. #P < 0.05, ##P < 0.01, and ####P < 0.0001 versus T_L.

Because Hepa1c1c7 cells are derived from a mouse hepatoma, we sought to validate the findings in a second cell line, AML12 mouse hepatocytes. FAStimerS was significantly induced in AML12 cells upon treatment with insulin (Fig. 6, A and B, G and H). Analysis of these cells by FACS revealed three separate populations of FAStimerS-expressing AML12 cells (Fig. 6C). Similar to the results in Hepa1c1c7 cells, the number of T_M and T_E cells increased with insulin treatment, indicating that these cells maintain active FAStimerS transcription in response to insulin (Fig. 6, D and E). On the other hand, there was no significant change in the number of T_L cells (Fig. 6F). Furthermore, consistent with Hepa1c1c7 cells, the responder T_E population showed increased levels of Foxo1, Foxo3, and Foxo4 mRNAs (Fig. 6I). In summary, elevated Foxo1 expression marks cells that preserve their response to insulin during an overnight incubation in two independent mouse hepatocyte lines.

Fig. 6.

Insulin-sensitive FAStimer cell population in AML12 cells is also enriched for forkhead box O (Foxo) mRNA. A: representative immunoblot for induction of timer reporter protein in mouse liver AML12 cells after overnight (O/N) insulin (ins) treatment. Arrowhead is nonspecific band. B: quantification of A. C: FACS dot plot for blue and red fluorescence of AML12 cells expressing FAStimerS. Shaded regions represent cell populations timer early (T_E), timer middle (T_M), and timer late (T_L). D–F: quantification of cell populations in C induced upon insulin treatment (D and E) or unchanged after insulin treatment (F). C, control. *P < 0.05 and **P < 0.01. NS, not significant. G–I: quantitative PCR analysis of T_E, T_M, T_L, and nonfluorescent (N) cell populations for the indicated genes. mRNA levels are normalized to β-actin. All bars represent average values of 3 independent replicates. *P < 0.05 and **P < 0.01 versus N. #P < 0.05, ##P < 0.01, and ###P < 0.001 versus T_L. $P < 0.05 and $$P < 0.01 versus T_M.

Cellular heterogeneity of Foxo1 localization following insulin treatment.

We next analyzed the phosphorylation status of Akt in the different FAStimer-expressing cell populations after insulin treatment. Interestingly, the T_E population had more Akt phosphorylation than any of the other populations (Fig. 7A). Given the tight connection between insulin signaling and Foxo1 localization, we analyzed Foxo1 subcellular localization after insulin treatment using a similar Timer reporter approach. We used two constructs: a Foxo1 construct fused to the fast version of the Timer fluorescent protein (FoxTimer) and a GFP protein containing the Foxo1 nuclear localization signal (pClover) (16). The majority of Foxo1Timer and pClover translocated to the cytoplasm after insulin stimulation, whereas ~25% of cells continued to demonstrate Foxo1 nuclear localization (Fig. 7, B and C). This was further confirmed in Hepa1c1c7 cells expressing FoxRFP, where ~20% Foxo1 remained in the nucleus after insulin treatment (Fig. 7, D and E). Interestingly, cells with nuclear Foxo1 had a twofold increase in colocalization with phosphorylated Akt (Fig. 7, D and F).

Fig. 7.

Cells harboring nuclear forkhead box O (Foxo) 1 are more sensitive to stimulation by insulin. A: representative immunoblot (top) with quantification (bottom) of cultured FAStimerS Hepa1c1c7 cell populations sorted from FACS [timer early (T_E), timer middle (T_M), timer late (T_L), and nonfluorescent (N)] after insulin treatment. Graph represents average of 5 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 versus respective controls. Within insulin treatment: ###P < 0.001 versus N, $$$$P < 0.0001 versus T_L, and ^^P < 0.01 versus T_M. B: fluorescent images of Hepa1c1c7 cells cotransfected with Foxo1Timer (red) and pClover (green) 24 h before start of overnight (O/N) insulin treatment. DAPI counterstains nuclei. Arrows, cell coexpressing Foxo1 and pClover. C: quantification of Foxo1Timer subcellular localization. ****P < 0.0001 versus control. D: immunofluorescence for Akt phosphorylation (pAkt, green) after insulin treatment in Hepa1c1c7 cells transfected with Fox red fluorescent protein (RFP, red). DAPI counterstains nuclei. Arrow, cells coexpressing FoxRFP and pAkt. E: quantification of FoxRFP nuclear localization. All images taken with ×20 objective, and the scale bar is 100 μM. ****P < 0.0001 versus control. F: quantification of pAkt⁺ cells. ****P < 0.0001 versus respective controls. Within insulin treatment: ####P < 0.0001 versus FoxRFP⁻ and $$$$P < 0.0001 versus Foxo1 cyto⁺.

DISCUSSION

In this study we generated a new method to identify heterogeneous cell populations with a fluorescent timer. Using this tool, we characterized a population of hepatocytes highly sensitive to insulin. Although our experiments focused on conditions of prolonged insulin stimulation, our methods can easily be modified to address other biological questions. For example, insulin-signaling pathways are rapidly activated after treatment with this hormone. Heterogeneity of this response can readily be measured with the fast and medium versions of timers, which have a blue emission half-life of 0.25 and 1.2 h, respectively (36). In addition, while we demonstrated proof-of-principle in two hepatocyte cell lines, experiments can also be adapted for use in primary culture and mouse models.

In our construct design, we used an insulin-regulatable FAS promoter reporter gene to show that the ability of cells to respond to insulin by synthesizing more FAS protein differs. Although we cannot conclude from these data that similar differences occur in hepatocytes in vivo, one can speculate that a similar heterogeneity underlies the complex pathophysiology of liver insulin action. In these experiments, we found that cells that continued to respond to insulin had increased mRNA levels of Foxo1, as well as increased Akt phosphorylation. This latter feature is consistent with prior reports of a homeostatic loop between Foxo and Akt such that increased nuclear Foxo begets increased Akt, while decreased Foxo decreases Akt signaling (2, 13, 26).

Foxo1 localization is primarily controlled by its phosphorylation by Akt, which triggers its export from the nucleus and degradation by the ubiquitin-proteasome system (5, 7, 29, 39). It is unclear why cells differ with regard to nuclear exclusion of Foxo1 in the presence of insulin. It is possible that Foxo1 is being sequestered in the nucleus because of other posttranslation modifications. In fact, a second posttranslational modification, deacetylation, promotes nuclear retention of Foxo1 (8, 10, 27, 38). Also adding to the complexity of this regulation is that dephosphorylation and deacetylation of Foxo1 can occur independently of each other (35). Further work will be necessary to determine if Foxo1 protein is differentially modified in the population with increased FAStimer fluorescence.

While Akt phosphorylates Foxo1, nuclear-localized Foxo1 also promotes Akt phosphorylation (12, 30). While our interpretation of occurrence of these two events in a cell was that this cell demonstrated an increased sensitivity to insulin, it is also possible that this may merely indicate coregulation of these two markers. In support of the former, in our analysis, we discovered that increased Foxo expression marks cells that are more sensitive to stimulation by insulin as indicated by increased levels of our FAStimer reporter. Foxo factors may have an active role in maintaining insulin responsiveness in these cells. This is consistent with studies showing that Foxo1 regulates lipid and glucose metabolism in the liver (2, 9, 19, 26, 34). Furthermore, regulation of Foxo1 levels has already been linked to a FAS-dependent mechanism involving SREBP1c-targeted degradation (21). Further studies will be needed to clarify the role of Foxo1 in T_E cells.

Insulin has multiple effects on liver cells in addition to its effects on FAS gene expression. Consequently, our decision to use a reporter that monitors induction of the FAS promoter by insulin limits the scope of our analysis to effects in pathways downstream of this promoter. It will be informative to perform similar screens for markers of cellular heterogeneity using reporters that target other branches of the insulin-signaling pathway.

Alteration of intracellular signaling is pivotal to the development of insulin resistance and diabetes. However, much less is known about the contribution of cellular heterogeneity to the pathogenesis of this disease. In addition, many hepatic metabolic processes are demarcated by liver zonation, raising the question of how cell-intrinsic and -extrinsic mechanisms interact to cause heterogeneity. In support of this possibility is a recent study which demonstrated that selective insulin resistance is attributable to the liver zonation of insulin receptor substrates (24).

GRANTS

W. M. McKimpson is supported by NIH Postdoctoral Training Program in Arteriosclerosis Research (T32-HL-007343-38) and is a Berrie Fellow in Diabetes Research. This work was supported in part by NIH Grants S10 RR-027050, S10 OD-020056, 5P30 DK-063608, and DK-57539.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

W.M.M. and D.A. conceived and designed research; W.M.M. performed experiments; W.M.M. and D.A. analyzed data; W.M.M. and D.A. interpreted results of experiments; W.M.M. prepared figures; W.M.M. drafted manuscript; W.M.M. and D.A. edited and revised manuscript; W.M.M. and D.A. approved final version of manuscript.