Zinc transporter Slc39a14 regulates inflammatory signaling associated with hypertrophic adiposity

Abstract

Zinc is a signaling molecule in numerous metabolic pathways, the coordination of which occurs through activity of zinc transporters. The expression of zinc transporter Zip14 (Slc39a14), a zinc importer of the solute carrier 39 family, is stimulated under proinflammatory conditions. Adipose tissue upregulates Zip14 during lipopolysaccharide-induced endotoxemia. A null mutation of Zip14 (KO) revealed that phenotypic changes in adipose include increased cytokine production, increased plasma leptin, hypertrophied adipocytes, and dampened insulin signaling. Adipose tissue from KO mice had increased levels of preadipocyte markers, lower expression of the differentiation marker (PPARγ), and activation of NF-κB and STAT3 pathways. Our overall hypothesis was that ZIP14 would play a role in adipocyte differentiation and inflammatory obesity. Global Zip14 KO causes systemic endotoxemia. The observed metabolic changes in adipose metabolism were reversed when oral antibiotics were administrated, indicating that circulating levels of endotoxin were in part responsible for the adipose phenotype. To evaluate a mechanism, 3T3-L1 cells were differentiated into adipocytes and treated with siRNA to knock down Zip14. These cells had an impaired ability to mobilize zinc, which caused dysregulation of inflammatory pathways (JAK2/STAT3 and NF-κB). The Zip14 deletion may limit the availability of intracellular zinc, yielding the unique phenotype of inflammation coupled with hypertrophy. Taken together, these results suggest that aberrant zinc distribution observed with Zip14 ablation impacts adipose cytokine production and metabolism, ultimately increasing fat deposition when exposed to endotoxin. To our knowledge, this is the first investigation into the mechanistic role of ZIP14 in adipose tissue regulation and metabolism.

white adipose tissue (WAT) contains a wide array of cell types that are characterized as connective tissue, nervous tissue, stromovascular cells, and immune cells. Together, these cells produce unique substances, i.e., leptin, adiponectin, and resistin, which contribute to the para- and autoendocrine regulation of lipid metabolism (21, 28). Limited epidemiological studies have linked aberrant trace mineral metabolism with adipose pathology; e.g., obesity is clinically correlated to deficiencies in iron, calcium, and zinc (18, 19, 49, 50). Zinc was first thought to be involved in metabolic activity of adipocytes through insulin-like effects (36) and through its inherent antioxidant properties (16). Human adipose tissue expresses many zinc transporters that may respond to the metabolic status of the patients, e.g., lean vs. obese (47). Despite the apparent correlation between zinc and adipose metabolism, a mechanism of action has yet to be defined.

Zinc partitioning within mammalian cells is due to ZIP and ZnT transporter activity, which control influx and efflux, respectively, of cytosolic zinc (27, 31). ZIP14 is a zinc influx transporter that is known to be upregulated during inflammation (3–5, 34, 40). We have focused on ZIP14 since it was found in our initial experiments to be the most responsive zinc transporter in mouse liver post-lipopolysaccharide administration (34). Knockout (KO) of ZIP14 results in a variety of unique phenotypes. We have reported previously that mice lacking ZIP14 have impaired liver zinc uptake (3) and elevated levels of serum endotoxin (25). Previously, Aydemir et al. (3) reported an increase in fat/lean ratios in Zip14-KO mice, a finding that led to our further investigation of ZIP14 in adipose function. The elevated level of serum endotoxin is particularly relevant to adipose in that chronic exposure to endotoxin can initiate obesity and insulin resistance (10). Inflammatory cytokines have a proliferative effect on adipocytes, leading to expansion of cell mass through both hypertrophy and hyperplasia (14, 20). Collectively, the marked induction of ZIP14 in WAT during inflammation along with the KO phenotype of increased adiposity, and metabolic endotoxemia suggests that this transporter alters zinc trafficking in adipocytes with functional outcomes.

Based on these previous findings, we hypothesized that ZIP14 would be critical to the inflammatory response and ultimately metabolic activity of WAT. Here, we report that WAT from KO mice appears to be insulin insensitive with hypertrophied adipocytes and dampened insulin signaling. Insulin resistance was linked to chronic inflammation within KO adipose tissue through upregulated cytokine expression and the Toll-like receptor 4 (TLR4) accessory protein myeloid differentiation primary response gene 88 (MyD88). Finally, we show that aberrant zinc signaling within the KO adipocyte is linked to enhanced JAK/STAT and NF-κB signaling, leading to impaired differentiation. Both findings tie directly to dyslipidemia and hypertrophy. The involvement of ZIP14 in major inflammatory pathways impacts adipocyte development and makes it a potential therapeutic target for inflammatory disorders in adipose, e.g., obesity and insulin resistance.

MATERIALS AND METHODS

Animals and diets.

Design and genomic validation of the Zip14-KO (Zip14^−/−) mice have been described previously (3, 5, 33). Briefly, Zip14 heterozygous mice were used to establish a breeding colony to generate wild-type (WT; Zip14^+/+) and KO mice. For the experiments described herein, female KO and WT mice were used between 8 and 16 wk of age. Mice were provided with ad libitum access to a commercial chow rodent diet (7912, with 60 mg zinc/kg provided by ZnO; Harlan-Teklad, Indianapolis, IN) and tap water.

Animal treatments.

In experiments modeling acute inflammation, lipopolysaccharide (LPS; E. coli serotype 055:B5; Sigma, St. Louis, MO) in phosphate-buffered saline (PBS) was administered at 2.0 mg/kg via intraperitoneal (ip) injection. Mice received LPS injections up to 18 h prior to euthanasia. Control animals received ip injections of PBS. In other experiments, fasted mice were gavaged with ⁶⁵Zn (2 μCi/mouse in 250 μl of saline) and euthanized 3 h postgavage to determine the tissue distribution. Specific activity of the ⁶⁵Zn (Perkin-Elmer, Waltham, MA) when used was 4.4 mCi/mg. Tissue accumulation of ⁶⁵Zn was measured via γ-scintillation spectrometry. For the antibiotic experiment, ultrapure (MiliQ, Billerica, MA) drinking water was supplemented with neomycin (0.5 mg/ml) and ampicillin (1 mg/ml) for 4 wk prior to tissue collection. Mice were anesthetized using isoflurane (Baxter, Deerfield, IL) prior to injection or gavage. Euthanasia was conducted via cardiac puncture. Blood from cardiac puncture was collected into a clot activator microgel barrier collection tube (Capiject; Terumo Medical, Somerset, NJ). Serum was separated via centrifugation and stored at −80°C prior to further analysis. All harvested tissues (intestine, muscle, liver, and adipose) were snap-frozen in liquid nitrogen and stored at −80°C prior to further processing. All studies described herein used intra-abdominal white fat pads (parameterial fat pads) that lay along the uterine horn. Unless otherwise specified, tissues were homogenized in assay-specific buffers using a Bullet blender with zirconium oxide beads (Next Advance, Averill Park, NY). All animal protocols were approved by the University of Florida Institutional Animal Care and Use Committee.

Cell culture.

A well-established embryonic mouse fibroblast cell line (3T3-L1; ATCC, Manassas, VA) was used to model adipocyte growth and differentiation. Cells were cultured in growth medium (DMEM;4.5 g/l glucose, 15% FBS, and 1% pencillin-streptomycin; all from Corning, Manassas, VA). After two passages, subconfluent primary cultures were trypsinized and plated into 12-well plates for experimentation. Transient knockdown of ZIP14 was carried out with HiPerfect (Qiagen, Valencia, CA). 3T3-L1 cells were grown to 90% confluence, and cells were treated with 25 nM siRNA (Darmacon, Pittsburgh, PA) for 48 h prior to incubation with differentiation medium with or without LPS (10 ng/ml). Thereafter, cells were exposed to differentiation medium (DMEM; 4.5 g/l glucose, 10% FBS, and 1% pencillin-streptomycin supplemented with 1.7 μM insulin, 1 μM dexamethasone, and 0.5 mM isobutylmethylxanthine). Cells were exposed to differentiation medium for three days prior to postdifferentiation medium (DMEM; 4.5 g/l glucose, 10% FBS, and 1% pencillin-streptomycin). The entire experimental period from differentiation to collection was 8 days. For experiments which involved LPS, cells were plated into their experimental dishes (passage 3), and 10 ng/ml LPS was supplemented to the medium the next day. Thereafter, cells were exposed to LPS throughout the entire experimental course through differentiation; medium was changed every 2 days.

To confirm our findings in 3T3L1 cells, ear mesenchymal stem cells (EMSC) were collected from WT and KO mice, as described previously by Rim et al. (42). Briefly, ears were minced and digested in medium containing collagenase type I (Worthington, Lakewood, NJ). Prior to the cell suspension being filtered (100 μm; Fisher Scientific, Suwanee, GA) minced tissue and collagenase medium were placed in a 37°C shaking water bath for 1 h. Filtered cells were pelleted through centrifugation (360 g, 5 min, room temperature), and red blood cells were lysed. Isolated cells were plated in 100-mm petri dishes in the previously described growth medium. Cells were allowed to expand for 3–5 days, and then these subconfluent primary cultures were trypsinized and plated into 24-well plates for experimentation. At visually confirmed confluence the EMSC were stimulated into adipocytes with previously described differentiation medium. After 3 days in differentiation medium, cells were maintained in postdifferentiation media. The entire experimental period from differentiation to collection was 10 days.

Analytical procedures.

Serum was diluted 1:3 using ultrapure water, and zinc levels were determined using flame atomic absorption spectrophotometry (AAS). Adipose tissue was weighed prior to HNO₃ digestion (90°C for 3 h) and diluted 1:1 with ultrapure water prior to AAS analysis. In experiments where total zinc was determined in cells, 3T3-L1 cells were first cultured in 150-mm plates (Corning). Cells were collected in ice-cold PBS, and an aliquot was obtained for total protein determination (Pierce BCA assay; Thermo Fisher Scientific, Waltham, MA). The remaining cell pellet was digested in HNO₃ (80°C for 3 h) and diluted 1:1 with ultrapure water. Total zinc was measured by AAS, and values were normalized to total protein, as described above. Serum endotoxin was measured as reported previously (25) using the LAL Chromogenic Endotoxin Quantitation Kit (Thermo Fisher Scientific), with absorbance read at 407 nm. Serum and tissue leptin were measured using a mouse/rat-specific ELISA kit (Alpco, Salem, NH). Tissue was first homogenized in cell extraction buffer (20 mM Tris·HCl, 1 mM EDTA, and 254 mM sucrose, pH 7.4), and absorbance was measured at 450 nm (1). Leptin values were normalized to total protein, as described above. Adipose glucose content was determined enzymatically with a total glucose assay kit (Sigma-Aldridge, St. Louis, MO). Prior to analysis, tissues were homogenized in ultrapure water and diluted 1:2. Absorbance was measured at 540 nm.

RNA quantification.

For RNA isolation, a portion of frozen fat pad was placed into TRIzol reagent (Life Technologies, Thermo Fisher Scientific). Early experiments (Fig. 1, A–C) utilized a Polytron blender to lyse tissue. Subsequent tissues were homogenized as described previously using the Bullet Blender. cDNA was generated using the iScript reagents (Bio-Rad, Hercules, CA). Quantitative real-time PCR was performed using the Real-time PCR Fast SYBR Green Master Mix and a StepOnePlus Fast Thermocycler (Applied Biosystems, Thermo Fisher Scientific). Primers to detect specific mRNAs were designed to span intron/exon boundaries: peroxisome proliferator-activated receptor-γ (Pparγ), 5′-GGAAGACCACTCGCATTCCTT-3′ and 5′-GTAATCAGCAACCATTGGGTCA-3′; plasminogen activator inhibitor-1 (Pai-1), 5′-AGGGCTTCATGCCCCACTTCTTCA-3′ and 5′-GTAGAGGGCATTCACCAGCACCA-3′. All other primer sequences were selected using the qPrimerDepot database, http://mouseprimerdepot.nci.nih.gov/, as described in Cui et al. (17). TBP mRNA was the normalizer for relative expression, as described previously (3, 5).

Fig. 1.

Expression of Zip14 in white adipose tissue (WAT) is induced by lipopolysaccharide (LPS). A: tissue expression profile of Zip14 transcripts. Transcript abundance was determined by quantitative PCR, starting with an equal amount of total RNA from each tissue. B: fold change in Zip14 tissue expression 18 h post-intraperitoneal (ip) injection of LPS (2 mg/kg). C: of the WAT zinc transporters significantly (P ≤ 0.05) impacted by LPS, Zip14 had the highest induction. For all graphs, n = 5 mice/genotype ± SE. DUO, duodenum.

Western blotting.

Western analysis was performed using polyclonal rabbit antibodies against ZIP14 developed and affinity purified (Life Technologies, Thermo Fisher Scientific) as described previously (3, 5). Akt, phosphorylated Akt (p-Ser⁴⁷³), hormone-sensitive lipase (HSL), phosphorylated HSL (Ser⁶⁶⁰), IκB, phosphorylated IκB (p-Ser^32/36), IR1β, phosphorylated IR1β (p-Y^1150/1151), mammalian target of rapamycin (mTOR), phosphorylated mTOR (Ser²⁴⁴⁸), MyD88, NF-κB, phosphorylated NF-κB (p65), PPARγ, preadipocyte factor-1 (PREF-1/DLK-1), SOCS3, STAT3, and phosphorylated STAT3 (p-Y⁷⁰⁵) antibodies were purchased from Cell Signaling Technology (Boston, MA). F4/80 and zinc α2-glycoprotein (ZAG) antibody were purchased from Santa Cruz Biotechnology (Dallas, TX). Frozen tissues were processed over liquid nitrogen to prevent freeze thaw and protein degradation. Tissues were homogenized in RIPA lysis buffer (Santa Cruz Biotechnology) with 100× protease and phosphatase inhibitors (Thermo Fisher Scientific), along with phenylmethanesulfonyl fluoride (Sigma-Aldrich). Proteins were separated using a 10% acrylamide gel for SDS-PAGE and transferred to nitrocellulose membranes. Transfer was verified through Ponceau Red staining, and proteins were visualized through chemiluminescence (SuperSignal West Pico, Thermo Fisher Scientific) and digital imaging (Protein Simple, San Jose, CA). Tubulin (Abcam, Cambridge, MA) abundance was used as the loading control.

Adipose histology.

Collected fat pads (parametrial fat pads) were fixed in 10% neutral buffered formalin for 24 h prior to paraffin embedding and sectioning. Slides were hematoxylin and eosin stained prior to microscopy using a Zeiss Axiovert 100 microscope at ×10 magnification. Adipocyte area was measured using ImageJ software (National Institute of Mental Health, Bethesda, MD). Cell areas were determined on cells with contiguous borders using the Adiposoft open-source ImageJ plugin http://fiji.sc/Adiposoft (22). Between 600 and 1,500 cells/image were used. Ten full images per genotype (5 mice, 2 images/mouse) were used to generate cell area images. In cell experiments where Oil Red O was used, cells were first fixed in 3.3% formaldehyde and stained in 3 μg/ml Oil Red O. Images were collected using the previously described Zeiss Axiovert 100 at ×10 magnification. To quantify staining, Oil Red O was extracted with isopropanol containing 4% NP-40, and absorbance was measured at 520 nm (62).

Confocal laser-scanning microscopy.

Imaging was done with a Leica TCS SP5 laser-scanning confocal microscope with LAS-AF imaging software, using a ×40 oil objective. For detection of labile zinc, cells were incubated with FluoZin-3 AM (Invitrogen, Waltham, MA), as described previously (4). Briefly, cells were incubated in 5 μM FluoZin-3 for 30 min, followed by a 30-min incubation in Locke's buffer (154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO₃, 1.2 mM MgSO₄, 5.6 mM glucose, 2.5 mM CaCl₂, and 10 mM HEPES). Cells were then stimulated with 40 μM ZnCl₂, and fluorescence was measured at 516 nm with excitation at 494 nm. A similar method was used to quantify FluoZin-3 AM in 12-well plates, with fluorescence being determined as above. Immediately after fluorescence was read, the cells were trypsanized and diluted 1:1 with trypan blue to determine cell number. FluoZin-3 AM relative fluorescent units were normalized to cell number. To generate confocal images, 3T3-L1 cells were passaged into two-well chambered coverglass (Thermo Fisher Scientific) prior to siRNA treatment and a subsequent 8-day differentiation period. Imaging was performed at the University of Florida Cell & Tissue Analysis Core with a Leica TCS SP5 laser-scanning confocal microscope, using a ×40 oil objective. LAS-AF imaging software was used for image analysis.

Statistics.

Statistical analyses was performed using SAS verison 9.2 (SAS Institute, Cary, NC). The effects of WT vs. KO genotype were compared using Student's t-tests. The independent effects of genotype and antibiotic were analyzed using the Proc Mixed procedure (SAS) with mouse within treatment as the random effect. Multiple comparison significance was analyzed using the Tukey adjustment. Reported values represent means ± SE.

RESULTS

Zip14 is highly expressed in WAT during acute inflammation.

Relative Zip14 mRNA abundance in WAT of WT mice is low compared with that found in duodenum or liver but greater than that found in muscle (Fig. 1A). Aydemir et al. (3) demonstrated that the highest induction of liver Zip14 mRNA occurred 18 h post-ip injection of LPS. Therefore, the 18-h time point was used to examine the comparative expression of Zip14 across tissues during LPS challenge (Fig. 1B). Expression of Zip14 mRNA increased two- and 32-fold in liver and WAT, respectively, with LPS administration. A transcript screening of 24 zinc transporters was performed. The transporter mRNAs that were significantly (P ≤ 0.05) altered by LPS are shown (Fig. 1C). With the exception of ZnT3 and ZnT8, all zinc transporter genes were expressed in WAT. Zip14 had the greatest LPS induction of the zinc transporters.

LPS-induced Zip14 expression is correlated with upregulated cytokine expression.

To compare WAT with our previous liver findings (3), we next sought to evaluate the induction of WAT ZIP14 during acute LPS challenge. The endotoxin-induced acute phase response was shown through increased expression of Il-6, Tnfα, and Il-1β mRNA (Fig. 2A). The induction of Zip14 expression peaked at 6 h post-ip injection and preceded the 9-h peaks of cytokine transcripts (Fig. 2A). Serum hypozincemia occurred as expected (3, 5); however, adipose levels of zinc fluctuated over the course of the LPS challenge (Fig. 2B). This finding suggests that a redistribution of zinc occurs within adipose during inflammation. The upregulation of cytokine mRNA transcripts coincided with increased ZIP14 protein levels, as shown in the Western blots from adipose lysates (Fig. 2C). In adipose, the acute-phase response is associated with increased markers of lipolysis and decreased markers of differentiation (14, 23, 64). Our data confirmed findings of others in that the phosphorylation of HSL (HSL660) was upregulated during endotoxemia. In contrast, phosphorylated insulin receptor and PPARγ were downregulated, which suggested a loss of adipocyte differentiation.

Fig. 2.

LPS-induced Zip14 expression is correlated with cytokine expression. A: mice received LPS (2 mg/kg ip) or the same volume (250 μl) of saline [control (CTRL)] 1–18 h before being euthanized. Relative transcript levels of Zip14 and specific cytokines; n = 3 mice/treatment ± SE. B: serum hypozincemia confirms the acute-phase response. Zinc concentrations in serum and WAT were measured by atomic absorption spectrophotometry (AAS). Adipose zinc values were normalized to wet tissue weight; n = 3 mice/treatment ± SE. C: Western analysis of ZIP14, lipolytic marker hormone-sensitive lipase (HSL), and differentiation marker PPARγ (peroxisome proliferator-activated receptor-γ) during the acute-phase response in WAT. Each lane is a pooled lysate of 3 mice/treatment. *P < 0.05; **P < 0.01; ***P < 0.0001. P-IR, phosphorylated insulin receptor.

Zip14 KO mice have hypertrophy of adipocytes and greater leptin production.

It is apparent that ZIP14 is highly induced by endotoxin in WAT. Previously, we established that global Zip14-KO impaired intestinal barrier function, which precipitates systemic endotoxemia (25). Therefore, we hypothesized that systemic endotoxemia would enhance the inflammatory status of adipose tissue. In confirmation of our previous findings, plasma endotoxin levels were greater in KO mice (Fig. 3A). We first sought to characterize the phenotypic profile of KO adipose. Enhanced expression of MyD88 (an adaptor protein that facilitates LPS induction of IL-6), NF-κB, and STAT3 is indicative of WAT inflammation in the KO mice (Fig. 3A, right). In agreement, the steady-state levels of Il-6, Tnfα, and Il-1β mRNAs were greater in WAT from the KO mice (Fig. 3B). Similarly, upregulation of Pref-1 (29, 58) and Pai-1 (35) mRNAs combined with downregulation of Pparγ in the WAT suggests a proinflammatory state, a characteristic of preadipocytic cells (Fig. 3B). Adiponectin (Adpn) expression, a hormone positively correlated with insulin sensitivity, was reduced with KO. Leptin protein levels were significantly (P < 0.05) greater in both KO serum and WAT (Fig. 3C). Moreover, cell areas of KO adipocytes were on average 40% larger than WT adipocytes (Fig. 3D). KO adipose had increased expression of key adipogenic enzyme transcripts and altered lipid homeostasis (Fig. 3E). Compared with WT mice, steady-state lipolysis markers, p-HSL (HSL660), and ZAG (39) were reduced in WAT from KO mice. The observed dyslipidemia was accompanied by reduced PPARγ levels (Fig. 3E). Overall, the KO mutation appeared to enhance adiposity and depress adipocyte differentiation. An evaluation of steady-state insulin signaling revealed that KO WAT had decreased phosphorylation of the insulin receptor (IR), protein kinase B (Akt), and mTOR (Fig. 3F) (9). These data show that KO adipose demonstrates a phenotype of dampened insulin signaling. Furthermore, data in Fig. 3F also show that the macrophage marker (F4/80) is not different between genotypes, suggesting minimal influence of macrophage infiltration (59) on our observed KO phenotype.

Fig. 3.

Zip14-knockout (KO) mice have enhanced levels of circulating endotoxin, predicating an inflammatory state characterized by adipose hypertrophy and dampened insulin signaling. A: colorimetrically determined that serum endotoxin was higher in Zip14-KO mice; n = 3/genotype ± SE (left). Proinflammatory signaling pathways in wild-type (WT) and KO mice are shown in WAT (right). Pooled lysates are depicted; lanes are triplicate repeats. B: expression of cytokines Il-6, Tnfα, Il-1β, and adiponectin (Adpn) in WT and KO mice. Preadipocyte marker [plasminogen activator inhibitor-1 (Pai-1) and preadipocyte factor-1 (Pref-1)] and differentiation marker (Pparγ) are also depicted; n = 28 mice/genotype ± SE. C: serum and WAT leptin concentrations in WT and KO mice; n = 3 mice/genotype ± SE. ELISA values from WAT were normalized to total protein. D: representative hematoxylin and eosin (H & E) images of WAT from WT and KO adipose. Cell areas are from n = 10 images/genotype ± SE. Bars, 150 μm. E: transcripts for key adipogenic enzymes (n = 28 mice/genotype ± SE) and Western analysis of lipolytic markers (n = 3 mice/genotype). F: Western analysis of insulin signaling (n = 3 mice/genotype). MyD88, myeloid differentiation primary response gene; mTOR, mammalian target of rapamycin; ZAG, zinc α2-glycoprotein.

The Zip14 phenotype is enhanced with LPS challenge and minimized with oral antibiotics.

Given the systemic endotoxemia in the KO, we hypothesized that the phenotype would be exacerbated by acute LPS and improved with antibiotics (AB). The highest induction of Zip14 mRNA during our LPS time course occurred 6 h post-ip injection (Fig. 1C). Therefore, we used the 6-h LPS response as the point of comparison between genotypes. The loss of Zip14 caused higher induction of WAT cytokine transcripts (Fig. 4A, left) along with serum IL-6 (Fig. 4A, right). These findings suggest that ZIP14 may serve as a negative regulator of cytokine induction. In contrast to the exacerbating effects of LPS, the impact of endotoxin-induced inflammation was reduced with AB. Expression of Il-6 and Tnfα mRNAs in KO WAT was reduced with AB (Fig. 4B). Furthermore, AB prevented the adipocyte hypertrophy seen in KO WAT (Fig. 4C). The AB treatment also normalized insulin signaling between the genotypes (Fig. 4D).

Fig. 4.

Zip14 KO enhances LPS-induced inflammation. The effects of the KO are reversed with oral antibiotics (AB). A: WT and KO mice were administered LPS (2 mg/kg ip) for 6 h. Zip14, Tnfα, and Il-1β mRNAs were measured by quantitative PCR. Serum and WAT IL-6 concentrations were measured by ELISA; n = 3 mice/treatment ± SE. In the absence of LPS, baseline IL-6 levels in either genotype were not detected (ND). B: WT and KO mice were provided AB in drinking water (0.5 mg/ml neomycin and 1 mg/ml ampicillin) for 4 wk. WAT expression of Il-6, Tnfα, and Il-1β mRNAs in WT and KO mice with AB treatment; n = 5 mice/treatment ± SE. C: H & E images of WAT were used to measure cell areas; n = 10 images/treatment ± SE. Bars, 150 μm. D: Western analysis of WAT insulin signaling in WT and KO mice with AB treatment; n = 3 mice/treatment.

Upregulated cytokine pathways due to the KO mutation lead to impaired capacity to differentiate.

In an effort to eliminate the effect of circulating endotoxin on KO adipose, mesenchymal stem cells were cultured in the absence of LPS. Stem cells were derived from WT and KO ears and differentiated into adipocytes. Zip14-KO was confirmed with quantitative PCR (Fig. 5A). We hypothesized that KO cells would exhibit enhanced steady-state cytokine signaling in culture. Cytokine mRNA expression (Il-6 and Il-1β) was higher in fully differentiated KO cells. Mt1 expression was not significantly different between genotype but tended to be lower in KO cells. Additionally, KO preadipocytic markers Pref-1 (29, 58) and Pai-1 (35) were upregulated (Fig. 5A). Western analysis of both PPARγ and PREF-1 (Fig. 5B) confirmed our KO tissue findings (Figs. 3, B and E, and 4D). Although the same antibody was used for cells and tissue, persistent bands appear in the Western blots of primary mesenchymal stem cells (Fig. 5B). However, this same persistent banding was again noted in siRNA-treated 3T3-L1 cells, which may serve as our negative control for cellular ZIP14 expression (Fig. 6F). Oil Red O staining also demonstrated that cultured KO cells are more preadipocytic and have a lower capacity to differentiate and accumulate lipids (Fig. 5C, top). It was not until cells were cultured in LPS that KO cells began to overproduce lipids (Fig. 5C, bottom). Quantification of Oil Red O confirmed our visual observations (Fig. 5D).

Fig. 5.

Stem cells from Zip14-KO mice are inflammatory and preadipocytic in culture. Primary stem cells from WT and KO mice were cultured in 3T3-L1 growth media. Upon confluence, cells were placed in differentiation medium and cultured for 10 days. A: transcript levels of Zip14, cytokine (Il-6, Tnfα, and Il-1β), Mt1, and preadipocytic markers (Pref-1 and Pai-1) were measured; n = 3 pooled wells ± SE. B: Western analysis of the preadipocytic markers in WT and KO cells; n = 3 pooled wells ± SE. C: accumulation of lipids in WT- and KO-derived cells, as measured by Oil Red O (± 10 ng/ml LPS for 10 days). D: Oil Red O was eluted from cell cultures and measured colorimetrically and normalized to protein; n = 3 wells/treatment ± SE.

Fig. 6.

Zip14-KO adipose and 3T3-L1 adipocytes treated with Zip14 siRNA display altered zinc homeostasis and enhanced proinflammatory signaling though the NF-κB and JAK2/STAT3 pathways. A: WT and KO mice received ⁶⁵Zn by gavage, and WAT zinc accumulation was calculated from ⁶⁵Zn-specific activity. Mt1 mRNA was measured as an index of available intracellular zinc; n = 10 mice/genotype ± SE. B: 3T3L1 cells were transfected with Zip14 siRNA. Total zinc (AAS) and Mt1 mRNA; n = 3 plates/group replicated 3 times. C: representative laser-scanning confocal images used to detect labile zinc pools using FluoZin-3 AM fluorescence. FluoZin-3 AM was quantified and normalized to cell count; n = 24 wells/group replicated twice. D: representative Western blots show the effect of Zip14 siRNA on ZIP14 and ZIP8 (n = 2 replicated 3 times). E: relative mRNA expression of Zip14, Il-6, Tnfα, and Il-1β in Zip14 siRNA-transfected 3T3-L1 cells ± 10 ng/ml LPS (n = 3 wells/group ± SE). F: representative Western blots show the effect of Zip14 siRNA on proinflammatory signaling pathways (n = 4 replicated 3 times). SCR, scrambled; RFU, relative fluorescence units.

ZIP14-KO alters intracellular zinc, which enhances key inflammatory pathways.

Since the focus of the molecular site responsible for the KO phenotype is likely related to metal transport, it was necessary to place those findings within a function of dyshomeostasis of cellular zinc metabolism. ⁶⁵Zn was administered orally to measure zinc uptake/retention. KO mice accumulated more zinc (⁶⁵Zn) in WAT. Figure 6A shows that Mt1 mRNA expression, a surrogate measure of cytosolic zinc, is reduced. Adipogenic 3T3-L1 cells transfected with siRNA had a pattern of zinc distribution similar to KO tissue. Specifically Zip14 knockdown increased total zinc in cells, whereas Mt1 mRNA was decreased (Fig. 6B). Intracellular labile zinc was visualized with a zinc probe, fluozin-3 AM. Fluorescence was detected using laser-scanning confocal microscopy (Fig. 6C). The silencing of Zip14 with siRNA clearly resulted in increased (P < 0.0001) vesicular zinc. Transfection with siRNA increased expression of ZIP8, a transporter that is a homologue of ZIP14 (Fig. 6D) (31). These in vitro data are compatible with the hypothesis that less functional cellular zinc is available in adipocytes of KO mice. This finding is indicative of the “zinc trap” hypothesis associated with Zip14 ablation that we advanced earlier (25) in mouse intestine. Zip14 siRNA clearly resulted in increased expression of Il-6 and Tnfα (P < 0.05) and to a lesser extent Il-1β (Fig. 6E).

Zinc's ability to inhibit key cytokine pathways is one way in which zinc may regulate cytokine expression. We hypothesized that zinc trapped within vesicles, through either Zip14 ablation in mice or with siRNA transfection in cells, would limit the inhibition of key proinflammatory signaling cascades by zinc. With the exception of A20 (11), Zip14 knockdown enhanced activation of both the JAK2/STAT3 and NF-κB pathways in 3T3-L1 cells (Fig. 6F). Upregulation of these pathways depressed the ability of cells to differentiate (upregulation of PREF-1 expression) and increase cytokine expression. As shown in the model presented in Fig. 7, zinc transported by ZIP14 appears to impact the TLR4 pathway as early as MyD88 and also influences signaling further downstream, e.g., phosphorylated IκBα (p-IκBα) and NF-κB. Zip14 knockdown limits cytosolic availability of zinc, and inhibition of NF-κB is lifted, allowing enhanced induction of proinflammatory genes Il-6, Tnfα, and Il-1β (Fig. 6E). Similarly, Zip14 deficiency enhanced JAK2 and STAT3 activity (Fig. 3A, right) with a concomitant stimulation of the Il-6 transcript (Fig. 3B) that could in turn execute an autocrine response leading to increased leptin production and secretion (Fig. 3C).

Fig. 7.

Proposed model for increased IL-6 and leptin production caused by global Zip14 deletion. Increased cytosolic zinc due to ZIP14 inhibits cytokine signaling. Systemic endotoxemia and/or LPS administration leads to Toll-like receptor 4 (TLR4) activation and induction of the NF-κB pathway. Without surface expression of ZIP14, zinc is trapped in intracellular compartments, where it is unable to inhibit activities of the NF-κB and JAK2/STAT3 pathways, thereby increasing IL-6 and leptin production.

DISCUSSION

Here we show for the first time that deletion of Zip14 impacts adipose function. Although Zip14 was first cloned from adipogenic 3T3-L1 cells (53), little has been reported regarding its expression/function in adipose metabolism. In this report, it is shown that KO alters zinc signaling pathways, intracellular zinc trafficking, and ultimately adipocyte metabolism.

Inflammation, expansion of tissue mass, and recruitment of inflammatory cells are normal functions of healthy adipose (49, 60). However, a prolonged inflammatory response will eventually lead to metabolic alterations within WAT. Acute inflammation impacts ZIP14 expression in a tissue-specific manner (3, 25). From the experiments described to this point, it has been established that adipose ZIP14 is highly responsive to LPS. Zip14 transcript abundance was upregulated 30-fold by 18 h after LPS injection, suggesting a critical role for ZIP14 in adipose inflammatory response. These data suggest that upregulation of Zip14 expression in WAT is kinetically more rapid than liver expression (3). Furthermore, Zip14 ablation drastically influences cytokine and leptin production along with signaling pathway activity in WAT. Upregulated Il-6 production with KO may be responsible for the induction of leptin secretion (55). We interpret these responses as being indicative of a greater requirement for ZIP14 and/or zinc during inflammatory challenge (43). The reversal of the KO phenotype with antibiotics supports this hypothesis.

The findings with antibiotics are noteworthy in that the treatment was a cocktail of neomycin and ampicillin. Neomycin in particular, is poorly absorbed through the gastrointestinal tract (15). Therefore, these antibiotics target intestinal microflora populations directly, a finding that suggests that intestinal microflora (along with any treatment that alters the microfloral load, dietary or otherwise) impacts adipose development and function. In fact, the tie between microflora and adipose has been established previously by Ley et al. (30). The impact of KO on the intestinal microfloral load is certainly relevant considering the proposed links between the microbiome and metabolic diseases, a finding that we plan to evaluate in further studies. Relevant to this report is that the in vivo phenotype was successfully modeled in vitro using LPS alone, confirming that KO predisposes cells to an inflammatory phenotype that is exacerbated with LPS.

Increased intestinal permeability increased serum endotoxin in KO mice (25). This finding was of particular interest, as chronic exposure to endotoxin leads to metabolic dysfunction. Metabolic endotoxemia is characterized by atypically high levels of circulating endotoxin, which produces low-grade, systemic inflammation (10, 34, 46). This inflammation has far-reaching physiological implications, including increased weight gain, hepatic insulin resistance (12, 26, 37), and macrophage recruitment (52, 59), all of which precede the development of type 2 diabetes upon the consumption of a high-fat diet. Luche et al. (35) found that a 28-day LPS pretreatment caused mice consuming high-fat diets to gain more weight than cohorts who had not received the LPS conditioning. The reported increase in body weight was coupled with an increased fat/lean ratio, an influx of small inflammatory preadipocytes, and an overall higher glycemic index (35). Unique to our KO model, dampened insulin signaling and low-grade inflammation were coupled with hypertrophy. Enhanced markers of lipogenesis and decreased markers of lipolysis (8, 38) were also noted with KO. These markers are suggestive of a chronic (vs. acute; see Ref. 61) inflammatory state in KO WAT. Furthermore, these physiological differences occur in the absence of a dietary intervention such as the feeding of a high-fat diet.

Inflammatory cytokines are well-known mediators of altered adipose function and development (12, 48). Furthermore, preadipocytes themselves are able to acquire phagocytic activity and express macrophage-specific antigens under acute inflammatory conditions (13). Using cultures at tiered stages of differentiation (0, 50, and 90%), Chung et al. (14) elegantly demonstrated that preadipocytes were the primary producers of the endogenous cytokines responsible for endotoxin-induced suppression of insulin-stimulated glucose uptake. It has been shown previously that Zip14 mRNA is induced during the early stages of adipocyte differentiation (53). Here, we show that KO increased preadipocytic markers and enhanced cytokine expression. The differentiation marker PPARγ was downregulated with KO. This finding is relevant, as PPARγ is zinc responsive and has the ability to inactivate STAT3 in myeloma cells (38, 56, 57). Zip14 KO limits the ability of mesenchymal stem cells to differentiate and accumulate lipids (Fig. 4). Although this may initially seem counterintuitive, limiting a preadipocyte's ability to differentiate precipitates hypertrophic obesity in humans (24, 49). Hypertrophic obesity is the likely cause of our previous finding where KO mice had enhanced liver lipids (3).

Aberrant zinc signaling with Zip14 KO impacts the growth and development of adipocytes. Here, we report a phenotype characterized by high tissue zinc but low Mt1 expression. Trayhurn et al. (54) found that adipose expression of Mt1 was not influenced by subcutaneous zinc injection. However, experiments conducted with metallothionein-KO (7) revealed that Mt^−/− mice had increased visceral WAT weight coupled with adipocyte hypertrophy and higher levels of circulating leptin (7). Furthermore, this phenotype was exacerbated by high-fat feeding (44). Previous literature has reported conflicting relationships between serum leptin and fat pad zinc content (2, 41, 51). In humans, obesity has long been associated with hypozincemia. In terms of zinc transporters, obesity is correlated with downregulated ZnT (−2, −3, −6, and −8) and ZIP (−1, −2, −3, −4, −5, −6, and −7) expression in subcutaneous adipose (47). Our data shows that Zip14 deletion causes adipose to act as if it were experiencing a zinc deficiency. Evidence for that includes the lower Mt1 mRNA levels in WAT of the KO mice (Fig. 6A). This zinc-deficient status apparently cannot be overcome by the concomitant upregulation of zinc transporters, most notably ZIP8 (Fig. 6D). The ability of adipocytes to regulate zinc trafficking may determine how adipose is remodeled during tissue expansion. Furthermore, ZIP14 appears to lie at the plastic interface of adipose metabolism and inflammation.

As shown in Fig. 7, our data show that ZIP14 influences adipocyte cytokine expression through two well-established signaling pathways, NF-κB and JAK2/STAT3 (6, 34, 55, 63). The adipose mass expresses TLR4 (45) and is a major secretory organ of both leptin and IL-6 (28). Hence, the demonstration here that both are overproduced with Zip14 deletion points to ZIP14 acting to provide control over the production of both mediators. Zip14 deletion enhances phosphorylation of these signaling pathways and is a likely cause for the upregulation of cytokines and leptin observed with KO. It should be noted that the ZIP14 homolog ZIP8 has been shown to inhibit the TLR4 pathway though inhibition of IκB kinase (IKK) (32). In this report, ZIP8 levels are greater with Zip14 ablation, demonstrating that ZIP8 is unable to compensate for the upregulation of the NF-κB pathway. Therefore, ZIP8-mediated inhibition of IKK may be of minor significance in our Zip14-KO adipose model.

In summary, adipocytes with a Zip14 deletion appear to have an impaired ability to mobilize intracellular zinc. In our model, limited availability of intracellular zinc disinhibits key cytokine pathways. Adipocytes have enhanced cytokine signaling and are more preadipocytic with KO. ZIP14 knockdown cells did not produce excess lipids until they were cultured in LPS media. These results in part model hypertrophic WAT observed with KO in vivo. Furthermore, the KO phenotype of hypertrophy was diminished with oral antibiotics, indicating that low-grade inflammation is necessary to induce enhanced adiposity. Our results demonstrate that ZIP14 is critical to cytokine production in adipose tissue and that targeted zinc transport is critical to adipocyte development.

GRANTS

This project was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant R01-DK-094244 and the Boston Family Endowment Funds of the University of Florida to R. J. Cousins.

DISCLOSURES

The authors report no conflicts of interest, financial or otherwise.

AUTHOR CONTRIBUTIONS

C.T., T.B.A., and R.J.C. conception and design of research; C.T. and T.B.A. performed experiments; C.T. analyzed data; C.T., T.B.A., and R.J.C. interpreted results of experiments; C.T. prepared figures; C.T., T.B.A., and R.J.C. drafted manuscript; C.T., T.B.A., and R.J.C. edited and revised manuscript; R.J.C. approved final version of manuscript.