Hepatic ZIP14-mediated zinc transport is required foradaptation to endoplasmic reticulum stressMin-Hyun Kima, Tolunay B. Aydemira, Jinhee Kima, and Robert J. Cousinsa,b,1

aFood Science and Human Nutrition Department, Center for Nutritional Sciences, College of Agricultural and Life Sciences, University of Florida, Gainesville,FL 32611; and bDepartment of Biochemistry and Molecular Biology, College of Medicine, University of Florida, Gainesville, FL 32611

Edited by Patrick J. Stover, Cornell University, Ithaca, NY, and approved June 13, 2017 (received for review March 9, 2017)

Extensive endoplasmic reticulum (ER) stress damages the liver,causing apoptosis and steatosis despite the activation of the un-folded protein response (UPR). Restriction of zinc from cells caninduce ER stress, indicating that zinc is essential to maintain normalER function. However, a role for zinc during hepatic ER stress islargely unknown despite important roles in metabolic disorders,including obesity and nonalcoholic liver disease. We have explored arole for the metal transporter ZIP14 during pharmacologically andhigh-fat diet–induced ER stress using Zip14−/− (KO) mice, which ex-hibit impaired hepatic zinc uptake. Here, we report that ZIP14-mediated hepatic zinc uptake is critical for adaptation to ER stress,preventing sustained apoptosis and steatosis. Impaired hepatic zincuptake in Zip14 KO mice during ER stress coincides with greaterexpression of proapoptotic proteins. ER stress-induced Zip14 KOmiceshow greater levels of hepatic steatosis due to higher expression ofgenes involved in de novo fatty acid synthesis, which are suppressedin ER stress-induced WT mice. During ER stress, the UPR-activatedtranscription factors ATF4 and ATF6α transcriptionally up-regulateZip14 expression. We propose ZIP14 mediates zinc transport intohepatocytes to inhibit protein-tyrosine phosphatase 1B (PTP1B) ac-tivity, which acts to suppress apoptosis and steatosis associated withhepatic ER stress. Zip14 KO mice showed greater hepatic PTP1B ac-tivity during ER stress. These results show the importance of zinctrafficking and functional ZIP14 transporter activity for adaptationto ER stress associated with chronic metabolic disorders.

unfolded protein response | p-eIF2α/ATF4/CHOP pathway | apoptosis |steatosis | zinc metabolism

The endoplasmic reticulum (ER) is a cellular organelle whereappropriate folding, assembly, and modification of proteins

occur (1). Normal ER function can be compromised by pharma-cological stimuli, such as tunicamycin (TM) or thapsigargin (2, 3),and by physiological stimuli, such as a high-fat diet (HFD), viralinfection, oxidative stress, or chronic alcohol consumption (4–8).Perturbed ER function is collectively defined as ER stress. WhenER stress occurs, mammalian cells activate the unfolded proteinresponse (UPR), which comprises three discrete signaling path-ways: the activating transcription factor 6 (ATF6) branch, theinositol-requiring enzyme 1 (IRE1) branch, and the dsRNA-activated protein kinase R-like ER kinase (PERK) branch (9).These pathways act to relieve the protein burden of the ER byreducing translation through mRNA decay (10) and by enhancingprotein-folding capacity through the expression of ER chaperones,including 78-kDa glucose-regulated protein (GRP78/BiP) and 94-kDa glucose-regulated protein (GRP94) (11). Cells survive if ERfunction is recovered through the activation of these adaptationpathways. However, if components of the UPR are compromisedor ER stress is too severe, the UPR leads cells to apoptotic celldeath via PERK-mediated phosphorylation of eukaryotic trans-lation initiation factor 2α (eIF2α) (12). Phosphorylated eIF2α(p-eIF2α) selectively enhances translation of activating transcrip-tion factor 4 (ATF4) mRNA, which leads to up-regulation of C/EBP-homologous protein (CHOP), a transcription factor that inducesexpression of apoptosis-associated components. Thus, enhancedactivation of the p-eIF2α/ATF4/CHOP pathway is a hallmark of

maladaptation against ER stress. Sustained apoptosis from ERstress influences numerous pathological conditions, includingdiabetes and obesity (8, 13). In the liver, apoptosis disrupts ho-meostasis of lipid regulation, causing hepatic steatosis (14).Zinc is an essential mineral required for normal cellular func-

tions playing catalytic, structural, and regulatory roles (15). Tomaintain zinc homeostasis, mammalian cells use 24 known zinctransporters that tightly control the trafficking of zinc in and out ofcells and subcellular organelles. These transporters are within twofamilies: ZnT (Zinc Transporter; SLC30) and ZIP (Zrt-, Irt-likeprotein; SLC39) (16–18). Physiological stimuli have been shown toregulate the expression and function of some of these transporters.Ablation of some of these transporters results in zinc dyshomeo-stasis and metabolic defects. Ablation of Zip14 (Slc39a14) in miceleads to a phenotype that includes endotoxemia, hyperinsulinemia,increased body fat, and impaired hepatic insulin receptor traf-ficking (19–21). Zinc and zinc transporters have been implicatedin ER stress and the UPR. Zinc deficiency may induce or exac-erbate ER stress and apoptosis. In yeast and some mammaliancells, the UPR was activated by zinc restriction (22, 23). A ratmodel of alcoholic liver disease created by zinc deficiency wasshown to trigger ER stress-induced apoptosis (24). We reportedthat consumption of a zinc-deficient diet by mice exacerbated ERstress-induced apoptosis and hepatic steatosis during TM-inducedER stress (25). Those recent studies showed that zinc can mod-ulate the proapoptotic p-eIF2α/ATF4/CHOP pathway during ERstress. A number of zinc transporters have also been associatedwith ER stress and the UPR. Administration of TM altered ex-pression of numerous zinc transporter genes in mouse liver, in-cluding ZnT3, ZnT5, ZnT7, Zip13, and Zip14 (23). In othervertebrate cells, knockdown (KD) of ZnT3, ZnT5, ZnT7, and

Significance

Unresolved endoplasmic reticulum (ER) stress corresponds withvarious chronic diseases, such as hepatic steatosis and diabetes.Although cellular zinc deficiency has been implicated in causingER stress, the effect of disturbed zinc homeostasis on hepaticER stress and a role for zinc during stress are unclear. This studyreveals that ER stress increases hepatic zinc accumulation viaenhanced expression of metal transporter ZIP14. Unfoldedprotein response-activated transcription factors ATF4 andATF6α regulate Zip14 expression in hepatocytes. During ERstress, ZIP14-mediated zinc transport is critical for preventingprolonged apoptotic cell death and steatosis, thus leading tohepatic cellular adaptation to ER stress. These results highlightthe importance of normal zinc transport for adaptation to ERstress and to reduce disease risk.

Author contributions: M.-H.K. and R.J.C. designed research;M.-H.K., T.B.A., and J.K. performedresearch; M.-H.K., T.B.A., and R.J.C. analyzed data; and M.-H.K. and R.J.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: cousins@ufl.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1704012114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1704012114 PNAS | Published online July 3, 2017 | E5805–E5814

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

Zip13 induced or potentiated ER stress (26–28). The role forZIP14 in ER stress has not been defined.In the present study, we explored a role of ZIP14 during

pharmacologically, HFD-induced ER stress using a Zip14−/− (KO)mouse model, which exhibits impaired hepatic zinc uptake. Wereport that during UPR activation, ATF4 and ATF6α transcrip-tionally regulate Zip14 expression and that ZIP14-mediated hepaticzinc accumulation is critical for adaptation to ER stress by pre-venting sustained apoptosis and steatosis. We propose that ZIP14-mediated zinc transport suppresses protein-tyrosine phosphatase 1B(PTP1B) activity, which may contribute to overcoming ER stress.

ResultsTM Administration Alters Hepatic Zinc Homeostasis and Zinc TransporterExpression, Particularly ZIP14. First, to define the effect of ER stresson zinc homeostasis, TM, a potent ER stress inducer, was in-jected i.p. into mice to induce systemic ER stress. TM is acompound that blocks N-glycosylation of newly synthesizedprotein. As measured by atomic absorption spectrophotome-try, TM-injected mice showed significantly higher hepaticzinc concentrations, along with hypozincemia, compared withvehicle-injected mice (Fig. 1 A and C). At 12 h after administra-tion, liver zinc concentrations of the TM group were ∼15%higher than liver zinc concentrations of the control group. Ad-ministration of 65Zn by gavage confirmed markedly increased zincuptake by the liver after administration of TM (∼1.6-fold) (Fig.1B). No difference in zinc uptake was observed 12 h after ad-ministration of TM in tissues that are known to be highly affected

by ER stress: the pancreas, kidney, and white adipose tissue (Fig.S1 B–D; WT). Therefore, the liver was our focus of further ex-periments. Because zinc homeostasis is maintained by zinc trans-porter activity, we examined expression of 24 known zinctransporter genes in the liver. The primer/probe sequences of genesare provided in Table S1. Among ZIP family transporters, Zip14expression was most highly up-regulated by TM (∼8.2-fold) (Fig.1D). Among ZnT family transporters, ZnT3mRNA expression wasmarkedly increased after administration of TM (∼15-fold) (Fig.1E). However, ZnT3 has a low hepatic abundance (16), and wasnot further evaluated. ZIP14 mRNA and protein peaked at 12 hafter administration of TM (Fig. 1 F and G), which coincided withmaximum zinc accumulation (Fig. 1A). The increased hepatic zincconcentration and ZIP14 expression were also observed in miceinjected with another ER stress inducer, thapsigargin, an inhibitorof ER Ca2+-ATPase (Fig. S2 A and B). Because liver tissue iscomposed of multiple cell types, we used human hepatomaHepG2 cells to support the in vivo studies. To measure total cel-lular zinc level in HepG2 hepatocytes, cell lysates were incubatedwith the zinc fluorophore FluoZin3-AM. Intensity of fluorescencerepresents the total cellular labile zinc concentration. Consistentwith data from the liver of TM-treated mice, TM treatment sig-nificantly increased fluorescence (∼2.1-fold after 12 h) (Fig. 1H),which coincided with increased ZIP14 expression (Fig. 1I).

Zip14 KO Mice Display Impaired Hepatic Zinc Uptake and HigherApoptosis During TM-Induced ER Stress. Next, we examined a po-tential role of ZIP14 during ER stress because increased hepatic

Fig. 1. TM-mediated ER stress increases hepatic zinc uptake and ZIP14. Hepatic Zn concentration (A) and 65Zn uptake (B) in mice (n = 3–4) are shown afteradministration of TM (2 mg/kg) or vehicle. (C) Serum Zn concentration in mice (n = 3–4) 18 h after administration of TM (2 mg/kg) or vehicle. Relative ex-pression of members of the ZIP family (D) and ZnT family transporter (E) genes in the liver of mice (n = 3–4) are shown after administration of TM (2 mg/kg) orvehicle for 12 h. Time-dependent expression of Zip14 mRNA (n = 3–4) (F) and immunoblot analysis of ZIP14 and markers of ER stress (G) in liver lysates (n = 3–4,pooled samples) are shown after administration of TM or vehicle for the indicated times. (H) Total cellular Zn concentrations in HepG2 cells were determinedby measurement of fluorescence after incubation with FluoZin3-AM (5 μM) following treatment with TM (1 μg/mL) or vehicle (n = 5). RFU, relative fluorescent unit.(I) Immunoblot analysis of ZIP14 and markers of ER stress in lysates of HepG2 cells after TM (1 μg/mL) or vehicle treatment. All data are represented as mean ± SD.*P < 0.05, **P < 0.01, ***P < 0.001. Cont, control.

E5806 | www.pnas.org/cgi/doi/10.1073/pnas.1704012114 Kim et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

ZIP14 expression and zinc accumulation after TM coincided witha marked reduction of CHOP expression (Fig. 1G). To test thehypothesis that ZIP14-mediated zinc uptake is critical to suppressapoptosis, we used conventional Zip14−/− (KO) mice (Fig. 2A).During steady-state conditions, the liver of Zip14 KOmice did notshow any indices of UPR activation (Fig. S3). Following TM,Zip14 KO mice exhibited less significant change in hepatic zincuptake, whereas WT mice displayed significant zinc accumulation(Fig. 2B). Measurement of radioactivity after 65Zn administrationby gavage also showed impaired zinc uptake in Zip14 KO miceduring ER stress (Fig. 2C). Expression of UPR pathway compo-nents, including proapoptotic pathway proteins (p-eIF2α, ATF4,and CHOP) and adaptation pathway proteins (BiP and GRP94),was examined. WT mice showed a marked reduction of proa-poptotic protein expression 24 h after administration of TM,whereas Zip14 KO mice displayed sustained expression (Fig. 2D).TUNEL assays of liver sections showed a significantly greaternumber of TUNEL-positive cells in Zip14 KO mice (∼2.3-fold),indicating that the KO mice had a greater level of apoptosis afterTM administration (Fig. 2E). Serum alanine aminotransferaseactivity, a common marker for liver damage, was also greaterin Zip14 KO mice after administration of TM (∼2.5-fold) (Fig.2F). Furthermore, Zip14 KO mice expressed significantly less

GRP94 than WT mice (∼0.6-fold) (Fig. 2D), which may indicatean impaired protein folding function.To test if there is a direct effect of zinc in the HepG2 hepa-

tocytes, we knocked down Zip14 with siRNA and then supple-mented zinc acetate along with pyrithione, a zinc ionophore.Pyrithione was added to improve zinc access under Zip14 KDconditions. The efficiency of Zip14 KD using siRNA transfectionwas ∼90% (Fig. S4A). TM-treated Zip14 KD cells showed an∼24% lower cellular zinc level compared with TM-treated con-trol cells, as measured by FluoZin3-AM (Fig. 3A, first four bars).To establish an optimal level of zinc supplementation, we treatedcells with zinc acetate, ranging from 2.5 to 20 μM. A maximallevel of cellular zinc was obtained with TM-treated Zip14 KDcells when 5 μM zinc acetate was added along with pyrithione(Fig. 3A). Thus, 5 μM zinc acetate was used subsequently to testthe effects of zinc supplementation. Higher expression of ATF4(∼2.6-fold) and CHOP (∼1.8-fold) and lower expression ofGRP94 (∼0.4-fold) were observed in Zip14 KD cells after TMtreatment (Fig. 3B, lanes 3 and 4). In response to TM, zinc-supplemented Zip14 KD cells expressed markedly reducedATF4 (∼34%) and CHOP (∼50%) proteins compared with non–zinc-supplemented Zip14 KD cells (Fig. 3B, lanes 4 and 6). Inaddition, expression of GRP94 was increased (∼1.6-fold) afterzinc supplementation of the cells. TM treatment increased p-eIF2α expression, which was reduced after zinc supplementationin Zip14 KD cells (Fig. 3B). The lower cell viability shown inTM-treated Zip14 KD cells (Fig. 3C) was eliminated by zincsupplementation (Fig. 3D).

Zip14 KO Mice Show a Greater Level of Hepatic Steatosis During TM-Induced ER Stress. Prolonged ER stress in the liver has beenlinked to the occurrence of hepatic steatosis. Greater levels oflipid droplet accumulation in Zip14 KO mice were observed withH&E staining of the liver sections after administration of TM(∼1.9-fold) (Fig. 4A). Quantification of triglyceride (TG) accu-mulation in the liver supported this observation (Fig. 4B). Be-cause hepatic lipid homeostasis is mainly maintained by fourmechanisms, namely, de novo fatty acid (FA) synthesis, FA ox-idation, FA uptake into the liver, and lipoprotein secretion, wemeasured expression of genes involved in these pathways in theZip14 KO mice to elucidate which mechanism(s) led to severesteatosis. With TM administration, genes involved in de novo FAsynthesis, such as Srebp1c, Acc, Fasn, and Scd1, were significantlysuppressed in WT mice by ∼60%, ∼85%, and ∼74%, respectively(Fig. 4C). In contrast, expression of Srebp1c, Acc, Fasn, and Scd1in TM-treated Zip14 KO mice was ∼2.5-fold, ∼2.2-fold, ∼3.2-fold, and ∼fourfold greater, respectively, than in TM-treated WTmice, indicating that FA synthesis in Zip14 KO mice was en-hanced during the stress. In the same setting, there were nosignificant differences in expression of genes involved in FAoxidation, FA uptake, and lipoprotein secretion (Fig. 4D).

HFD-Mediated ER Stress Is Exacerbated in Zip14 KO Mice. Feeding anHFD has been used to trigger ER stress in rodents and provides amodel with high physiological relevance (8). Thus, we analyzed in-dices of ER stress after feeding mice with an HFD (60 kcal% fat) orchow diet (12 kcal% fat). After 16 wk, body weights of HFD-fedmice increased 10.2 ± 2.6 g, whereas chow-fed mice gained 2.8 ±0.2 g. Both genotypes gained weight at the same rate during HFDfeeding. WT mice showed enhanced ZIP14 expression after theHFD (Fig. 5 A and B). Hepatic zinc concentrations in HFD-fedWTmice were ∼17% higher compared with chow-fed WT mice, in-dicating that an HFD increases zinc uptake (Fig. 5C). However, zinclevels in HFD-fed Zip14 KO mice were unchanged, indicating thatZIP14 facilitates hepatic zinc accumulation. Although Zip9 mRNAexpression was also increased by the HFD (Fig. 5A), its hepaticabundance is known to be low (16); thus, it is unlikely thatZIP9 contributes markedly to hepatic zinc uptake. This conclusion

Fig. 2. ZIP14 KO mice show greater ER stress-induced apoptosis. (A) Immu-noblot analysis of ZIP14 from liver lysates of WT and Zip14−/− (KO) mice 12 hafter administration of TM (2 mg/kg) or vehicle. Hepatic Zn concentration(B) and 65Zn uptake (C) in WT and Zip14 KO mice (n = 3–4) are shown afteradministration of TM (2 mg/kg) or vehicle. (D) Immunoblot analysis of ER stressmarkers from liver lysates of WT and Zip14 KO mice (n = 3–4, pooled samples)after administration of TM (2 mg/kg). Individual blots (24 h after TM, n = 4)were quantified using digital densitometry to determine relative proteinabundance. (E) Representative images of TUNEL assays of liver sections of WTand Zip14 KO mice 24 h after administration of TM (2 mg/kg) or vehicle.TUNEL-positive cells in fields were quantified. (Magnification: 40×.) (Scale bars:25 μm.) (F) Serum alanine aminotransferase (ALT) activity of WT and Zip14 KOmice 24 h after administration of TM (2 mg/kg) or vehicle (n = 3–4). n.d., notdetected. All data are represented as mean ± SD. *P < 0.05. Labeled meanswithout a common letter differ significantly (P < 0.05).

Kim et al. PNAS | Published online July 3, 2017 | E5807

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

is also supported by showing that hepatic zinc does not increasedue to HFD consumption in Zip14 KO mice (Fig. 5C). Both gen-otypes showed elevated expression of UPR components, includingp-eIF2α, ATF4, CHOP, BiP, and GPR94, indicating UPR activation

in response to the HFD (Fig. 5D). However, HFD-fed Zip14 KOmice expressed greater levels of proapoptotic signatures, includingp-eIF2α, ATF4, and CHOP, compared with HFD-fed WT mice,indicating that ablation of Zip14 worsens ER stress-associated

Fig. 3. Supplementation with zinc rescues ER stress-induced apoptosis in Zip14 KD hepatocytes. (A) Total cellular Zn concentrations were determined by mea-surement of fluorescence after incubation with FluoZin3-AM (5 μM). HepG2 cells were treated with the indicated dose of zinc acetate and/or pyrithione (50 μM) for30 min, which was followed by TM (1 μg/mL) or vehicle treatment for 12 h. (B) Immunoblot analysis of ZIP14 and ER stress markers from cell lysates. Zip14 siRNA-transfected or control siRNA-transfected HepG2 cells were incubated for 30 min with zinc acetate (5 μM) and pyrithione (50 μM), which was followed by a 24-hincubation with TM (1 μg/mL) or vehicle. Individual blots (lanes 3–6, n = 3) were quantified using digital densitometry to determine relative protein abundance. (C andD) Cell viability was measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. (C) Zip14 siRNA-transfected or control siRNA-transfectedHepG2 cells were treated with TM (1 μg/mL) for 1 d or 2 d. (D) Cells were pretreated with zinc acetate (5 μM) and pyrithione (50 μM) for 30 min before TM (1 μg/mL)treatment for 18 h. All data are represented as mean ± SD. *P < 0.05. Labeled means without a common letter differ significantly (P < 0.05). Csi, control siRNA.

Fig. 4. Zip14 KO mice exhibit a greater level of hepatic TG accumulation after TM-induced ER stress. (A) Representative images of H&E-stained liver sections ofWT and Zip14 KO mice 24 h after administration of TM (2 mg/kg) or vehicle. The lipid droplet area in the field was measured. (Magnification: 10×.) (Scale bars:100 μm.) (B) Liver TG levels of WT and Zip14 KO mice were measured 24 h after administration of TM (2 mg/kg) or vehicle (n = 3–4). Relative expression of genesthat regulate FA synthesis (C) and FA β-oxidation, FA uptake, and lipoprotein secretion (D) were measured in livers of WT and Zip14 KO mice 12 h after ad-ministration of TM (2 mg/kg) or vehicle (n = 3–4). All data are represented as mean ± SD. Labeled means without a common letter differ significantly (P < 0.05).

E5808 | www.pnas.org/cgi/doi/10.1073/pnas.1704012114 Kim et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

apoptosis in this setting. Expression of ATF4 and CHOP was∼3.7-fold and ∼3.2-fold greater, respectively, in Zip14 KO mice.Of particular note is the similarity of these results to the resultsattained with TM-induced ER stress. Specifically, the HFD-fedZip14 KO mice expressed a lower level of GRP94 than HFD-fed WT mice (∼0.6-fold). Hepatic TG accumulation after theHFD tended to be higher in Zip14 KO mice (Fig. 5E). Moreover,this observation coincided with significantly higher mRNA ex-pression of Srebp1c (∼1.7-fold), Fasn (∼1.8-fold), and Scd1 (∼1.9-fold) (Fig. 5F), suggesting a greater level of FA synthesis led tomore hepatic TG accumulation in the Zip14 KO mice.

Increased PTP1B Activity Is Observed in Zip14 KO Mice During ERStress. As a possible mechanism underlying the increased ERstress in Zip14 KO mice, we focused on PTP1B, an enzyme thathas been implicated in ER stress. PTP1B expression was in-creased in ER stress induced by TM and the HFD, and deletionof the protein in vivo and in vitro significantly reduced ER stress-associated apoptosis and steatosis (29–31). Because zinc is aknown inhibitor of PTP1B activity (32, 33), we hypothesized thatZip14 KO would exhibit increased PTP1B activity during ERstress due to impaired zinc uptake, and thus disrupt adaptationto ER stress. Consistent with previous reports (30, 31), expres-sion of proapoptotic proteins was significantly decreased in TM-treated Ptp1b KD cells compared with TM-treated control cells(Fig. 6A). ATF4 and CHOP expression was reduced by ∼42%and ∼30%, respectively, in the Ptp1b KD condition. This findingwas supported by markedly higher cell viability in Ptp1b KD cellsafter TM challenge (Fig. 6B). At steady state, Zip14 KO miceexpressed less PTP1B protein than WT mice (∼0.6-fold), butPTP1B activity was comparable (Fig. 6 C and D). Following TMadministration, hepatic PTP1B expression was increased in bothgenotypes (Fig. 6C, TM). However, PTP1B activity was signifi-

cantly greater in Zip14 KO mouse liver (∼1.7-fold) (Fig. 6D). Thesame pattern was observed in HFD-fed mice. PTP1B activity wassignificantly greater in the liver of Zip14 KO mice compared withWT mice (∼1.5-fold), although the amount of protein expressionwas not different (Fig. 6 E and F). In HepG2 hepatocytes, Zip14siRNA KD did not alter PTP1B protein levels (Fig. 6G). Incontrast, PTP1B activity was increased in the hepatocytes follow-ing Zip14 siRNA KD and TM treatment, but it was significantlyreduced with zinc supplementation (Fig. 6H). These data dem-onstrate a direct effect of zinc on PTP1B activity in this setting.Collectively, these data suggest that normal hepatocytes suppressPTP1B activity during ER stress by regulating cellular zinc avail-ability via ZIP14 induction. Impaired zinc uptake in the KO miceor by siRNA KD disturbs these events. A model underlyingZIP14-mediated adaptation and defense against ER stress ispresented in Fig. 6I.

Zip14 Is Transcriptionally Regulated by ATF4 and ATF6α During TMTreatment. Because we observed an induction of Zip14 duringTM- and HFD-induced ER stress, our next focus was to identifytranscription factor(s) that regulate Zip14. For efficient genemanipulation, we performed in vitro experiments using HepG2hepatocytes. During TM treatment, Zip14 mRNA was increasedin a dose-dependent manner, and treatment with actinomycin D,a transcription inhibitor, suppressed induction, indicatingthat Zip14 induction by TM is regulated at the transcrip-tional level (Fig. 7A). Transcriptional regulation was sup-ported by the time-dependent increases in Zip14 mRNA andheterogeneous nuclear RNA. Both exhibited similar expressionpatterns following TM treatment (Fig. 7B). Global gene expressionscreening data provided clues regarding the potential Zip14-regu-lating transcription factors ATF4 and ATF6α (34, 35). ATF4 andATF6α have been shown to have a strong binding affinity to the

Fig. 5. HFD-fed Zip14 KO mice show greater hepatic ER stress-induced apoptosis and TG accumulation. Mice were fed the HFD or a chow diet for 16 wk.(A) Relative gene expression of members of the ZIP family transporter in WT mice (n = 4). (B) Immunoblot analysis of ZIP14 from liver lysates of WT and KOmice. (C) Hepatic Zn concentration of WT and Zip14 KO mice (n = 4). (D) Immunoblot analysis of ER stress markers from liver lysates of WT and KO mice (n = 4,pooled samples used). Individual blots (HFD, n = 4) were quantified using digital densitometry to determine relative protein abundance. (E) Liver TG levelswere quantified in WT and Zip14 KOmice (n = 4). (F) Relative expression of genes that regulate FA synthesis were measured in livers of WT and Zip14 KOmice(n = 4). All data are represented as mean ± SD. *P < 0.05. Labeled means without a common letter differ significantly (P < 0.05).

Kim et al. PNAS | Published online July 3, 2017 | E5809

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

cAMP-response element (CRE) sequence [TGACGT(C/A)(G/A)](Fig. 7C). Matinspector software analysis revealed that the Zip14promoter has a potential binding site for both ATF4 and ATF6αat −94 to −89, which matches with a core motif of CRE, and wasconserved in mice and humans (Fig. 7D). To test the expression ofZip14 under ATF4-KD or ATF6α-KD conditions, HepG2 hepato-cytes were transfected with either Atf4 or Atf6α siRNA (Fig. S4 B andC). KD of Atf4 resulted in significantly reduced Zip14 induction atboth 6 h and 24 h after TM treatment (Fig. 7E), whereas KD ofAtf6αreduced induction only at 24 h after TM treatment (Fig. 7G). Toensure actual binding of transcription factors to the potential bindingsite of the Zip14 promoter, we conducted ChIP-PCR. Detection ofDNA enrichment revealed high ATF4 DNA binding 6 h after TMtreatment. Binding was reduced thereafter (Fig. 7F). However, en-hanced binding of ATF6α was only detected 24 h after TM treatment(Fig. 7H). Western blotting showed that ATF4 expression was in-duced until 12 h after TM treatment, and was then decreased at 24 h(Fig. 7I). This finding suggests a time-dependent regulation byATF4 and ATF6α in which ATF4 first binds to the Zip14 promoter

due to increased expression and higher binding affinity and ATF6αthen binds to the promoter after ATF4 expression is decreased.

DiscussionIn this report, we demonstrate that TM- and HFD-induced ERstress triggers ZIP14-mediated zinc accumulation in mouse liver.Both routes to ER stress yield remarkably comparable results(Figs. 2 and 5). This event is critical for adaptation to ER stressby preventing prolonged apoptosis and steatosis, possibly viainhibition of PTP1B activity. During TM-induced ER stress,Zip14 up-regulation in HepG2 hepatocytes is modulated at thetranscriptional level by the UPR-activated transcription factorsATF4 and ATF6α. These findings are summarized in Fig. 8.During ER stress, cellular metabolism is largely altered by

activation of UPR pathways in an effort to restore ER homeo-stasis. This activation includes enhancing expression of proteinsthat help the ER folding function, such as ER chaperones, andreducing translation of mRNA to reduce cellular protein burden(11). Several lines of evidence indicate that ER stress also altersthe metabolism of some metals. For example, mice injected with

Fig. 6. ZIP14 is required to suppress hepatic PTP1B activity after TM administration and the HFD. Immunoblot analysis of PTP1B and ER stress markers (A) andmeasurement of cell viability using the MTT assay (B) in HepG2 hepatocytes transfected with Ptp1b siRNA or control siRNA are shown. Cells were treated withTM (1 μg/mL) or vehicle for 24 h. In A, individual blots (TM, n = 3) were quantified using digital densitometry to determine relative protein abundance.Immunoblot analysis of PTP1B protein (C) and measurement of PTP1B activity (D) in livers of WT and Zip14 KOmice are shown 12 h after administration of TM(2 mg/kg) or vehicle (n = 3–4, pooled samples used for C). Analysis of PTP1B protein (E) and measurement of PTP1B activity (F) in livers of WT and Zip14 KOmice fed with HFD or chow for 16 wk (n = 4, pooled samples used for E) are shown. Immunoblot analysis of PTP1B protein (G) and measurement of PTP1Bactivity (H) in HepG2 hepatocytes transfected with Zip14 siRNA or control siRNA are shown. Cells were pretreated with zinc acetate (5 μM) and pyrithione (50 μM)for 30 min before TM (1 μg/mL) treatment for 12 h. (I) Proposed model for ZIP14-mediated zinc transport and inhibition of PTP1B activity. All data are representedas mean ± SD. Labeled means without a common letter differ significantly (P < 0.05).

E5810 | www.pnas.org/cgi/doi/10.1073/pnas.1704012114 Kim et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

TM have been shown to exhibit hypoferremia and iron seques-tration in the spleen and liver (36). Regarding zinc metabolismand ER stress adaptation, it has been shown recently in mice thatTM administration alters hepatic gene expression of multiplezinc transporters (23). That observation is supported by ourpresent study, in which ZIP14-mediated hepatic zinc uptake wasobserved, along with hypozincemia, after TM and thapsigargintreatments (Fig. 1 A–C and Fig. S2A), demonstrating that acuteER stress induction alters zinc metabolism. We focused on theliver because no difference in zinc uptake was observed in thepancreas, kidney or white adipose tissue, which are known to behighly affected by ER stress (Fig. S1 B–D).We also examined ER stress with an HFD model, which is

more physiologically relevant. It was previously demonstratedthat 16 wk of HFD feeding to mice induces ER stress in the liver(8). The HFD also produced ZIP14-mediated hepatic zinc ac-cumulation in WT mice, whereas Zip14 KO mice did not accu-mulate zinc during that time period (Fig. 5C). The impairedhepatic zinc uptake of Zip14 KO mice possibly produced agreater signature of ER stress-induced apoptosis and steatosis,which is consistent with the TM-induced ER stress model (Fig. 5

D–F). Phenotypic changes in Zip14 KO mice, such as hepaticzinc accumulation and steatosis, are remarkably similar in thosetwo models. However, the magnitude of change is relativelymarginal in the HFD model compared with TM model. Thisdifference may be caused by the chronic vs. acute nature of thesetwo models of ER stress. The HFD is a well-known model of themetabolic syndrome, causing insulin resistance or glucose in-sensitivity (37). Our recent report has shown that HFD-fed(6 wk) Zip14 KO mice do not develop a disturbed metabolicphenotype, such as hepatic insulin resistance or glucose in-sensitivity (20), indicating that the impaired stress adaptation ofZip14 KO mice is probably not influenced by glucose stress,which is a known cause of ER stress (38).Our first question regarding the zinc, ZIP14, and ER stress

relationship was whether the increased hepatic zinc uptake afterER stress would be a factor in stress adaptation. In the sense thatER function is dependent on zinc availability to provide zinc forincorporation into and folding of zinc metalloproteins (39), in-creased zinc may play a role in the restoration of ER homeo-stasis. In support of that notion, it has been shown that a supplyof zinc to the early secretory pathway via ZnT5, ZnT6, and

Fig. 7. Zip14 is transcriptionally regulated by ATF4 and ATF6α in a time-dependent manner during TM treatment. (A) Relative expression of Zip14 mRNA inHepG2 cells treated with TM (1 μg/mL) and/or actinomycin D (Act D; 2 μg/mL) for 12 h. (B) Relative expression of Zip14 mRNA and heterogeneous nuclear RNA(hnRNA) in HepG2 cells treated with TM (1 μg/mL) or vehicle. (C) Consensus motif of CRE and binding motifs of ATF4 and ATF6. (D) Sequence of mouse andhuman Zip14 promoter regions (from −120 to +1). Identical nucleotides are indicated by an asterisk. The TGACG sequence (from −94 to −89) is marked by abox. Relative expression of Zip14mRNA in TM-treated HepG2 cells (1 μg/mL) after transfection with control siRNA, Atf4 siRNA (E), or Atf6α siRNA (G) is shown.Enrichment of DNA bound to ATF4 antibody (F) or ATF6α antibody (H) was measured by quantitative real-time PCR after ChIP assays in TM-treatedHepG2 cells (1 μg/mL). Nonspecific rabbit IgG antibody was used as a negative control. (I) Immunoblot analysis of ATF4 and full-length and cleaved ATF6αin TM-treated HepG2 cells (1 μg/mL). All data are represented as mean ± SD. *P < 0.05, **P < 0.01.

Kim et al. PNAS | Published online July 3, 2017 | E5811

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

ZnT7 is essential for helping protein folding via zinc incorporationinto alkaline phosphatases, which are zinc metalloproteins (40, 41).The importance of an adequate zinc supply for protein folding wasillustrated when TM-induced ER stress was exacerbated duringzinc-deficient conditions in ZnT5−/−/ZnT7−/− cells (26). In contrast,some metals may exhibit toxicity when their cellular concentrationsare elevated and can aggravate processes such as ER stress. Forexample, excessive accumulation of manganese (42–44), cadmium(45–47), and fluoride (48) induces ER stress and UPR activation.Here, we found that ZIP14-mediated zinc uptake was critical foradaptation to ER stress, because Zip14 KO mice displayed greaterapoptosis (Figs. 2 D and E and 5D). The demonstration of im-paired GRP94 expression in ER stress-induced Zip14 KO mice(Figs. 2D and 5D) supports the notion that GRP94 is a criticalfactor that supports recovery of ER homeostasis. AlthoughZIP14 can contribute to uptake of manganese and non–transferrin-bound iron under certain circumstances (49, 50), the hepaticconcentration of these metals after TM administration wascomparable between WT and Zip14 KO mice (Fig. S1 E and F).In support, in vitro supplementation with zinc rescued Zip14 KDHepG2 hepatocytes from ER stress-induced apoptosis (Fig. 3 Band D), demonstrating a direct effect of zinc. Because prolongedapoptosis causes many pathogenic disorders, suppression of thep-eIF2α/ATF4/CHOP pathway can be a therapeutic target. Ourdata collectively demonstrate that ZIP14-mediated hepatic zincuptake also influences this pathway and provides a mechanismby which apoptosis is suppressed.Our next focus was to elucidate a direct target for zinc. Here,

we propose that zinc influences the p-eIF2α/ATF4/CHOPpathway through suppression of the activity of PTP1B. Variouspathways, including the pathways for insulin and leptin signaling,are regulated by PTP1B, and dysregulation of PTP1B activity wasdemonstrated to contribute to the pathogenesis of various dis-eases, such as cancer and diabetes (51). Therefore, there hasbeen a significant effort to develop a PTP1B inhibitor as adrug target (52). PTP1B has been implicated in ER stress (53).Liver-specific Ptp1b−/− mice showed decreased activation of thep-eIF2α/ATF4/CHOP pathway, along with reduced indices ofmetabolic syndrome, during TM- and HFD-induced ER stress(30, 31). Consequently, our data (Fig. 6 A and B) are consistentwith the data from TM-treated Ptp1b KD hepatocytes. Zinc isa known inhibitor of PTP1B through physical binding to theenzyme (32). We previously reported that overexpression ofZIP14 in AML12 hepatocytes suppressed PTP1B activity (33).

Based on these reports, we hypothesized that ER stress-inducedmice increase hepatic ZIP14 expression to enhance zinc uptakefor suppression of PTP1B activity. Zip14 KO mice showed sig-nificantly elevated PTP1B activity, whereas activity in WT miceremained unchanged after TM- and HFD-induced ER stress(Fig. 6 D and F), possibly due to impaired hepatic zinc uptake. Invitro, a direct effect of zinc on PTP1B activity inhibition wasdemonstrated using zinc supplementation, where the greaterPTP1B activity in TM-treated Zip14 KD cells was reversed byzinc supplementation (Fig. 6H). These data are consistent withour previous study, where increased PTP1B activity was observedin mice fed a zinc-deficient diet compared with mice fed a zinc-adequate diet during TM challenge (25).Disrupted hepatic lipid homeostasis is another hallmark of un-

resolved ER stress. This defect has been observed in cells withcompromised UPR function. Genetic ablations of UPR components,including ATF6α, IRE1α, and eIF2α, result in the development ofhepatic steatosis (14, 54, 55). Loss of ATF4 increased free cholesterolin rodent liver (35), and liver-specific deletion of XBP1 producedhypocholesterolemia and hypotriglyceridemia (56). Similar to theseobservations, the liver of Zip14 KO mice showed potentiated TGaccumulation after TM administration due to enhanced FA synthesis(Fig. 4). Zinc-mediated PTP1B activity suppression may mechanis-tically explain these data because liver-specific Ptp1b−/− mice showeddiminished hepatic FA synthesis and TG storage during HFD feed-ing (30), which is consistent with the ER stress-induced TG accu-mulation we showed in Zip14 KO mice.UPR activation induces a variety of genes involved in protein

folding, degradation, and trafficking to restore ER homeostasis viatranscription factors, such as ATF6α, ATF4, and XBP-1 (12). Wepropose that one arm of UPR pathway activation is to modulatezinc trafficking, based on our finding that Zip14 is a transcriptionaltarget of ATF4 and ATF6α (Fig. 7 E–H). ATF4 and ATF6α have asignificant binding affinity for the CRE consensus sequence (57–59).We found that the Zip14 promoter has a CRE-like sequence, anddemonstrated here that both ATF4 and ATF6α bound to the CRE-like sequence after TM treatment in a time-dependent manner,leading to Zip14 mRNA increases through increased transcription(Fig. 7). Our data agree with the observations from global tran-scriptional profiling, which showed significantly reduced Zip14mRNA levels after TM treatment in liver-specific Atf4−/− mice (35)and in Atf6α−/− fibroblasts (34). ATF4 and ATF6α regulation ofZip14 suggests that these events lead to increased cellular zincavailability through ZIP14-mediated zinc transport. Of note is thatthe Zip14 promoter lacks the sequence required for the liver-specific transcription factor CREBH, which is needed for in-duction of hepcidin by ER stress (36). Limited evidence indicatesthat UPR components regulate expression of other zinc transportergenes. XBP-1–mediated hZnT5 up-regulation during ER stressoccurs via direct binding to that promoter (26). In addition, ZnT5−/−/ZnT7−/− cells displayed an exacerbated ER stress response (26).Similarly, KD of Zip7, Zip13, and ZnT3 in cells induced ER stressand higher cell death, indicating the importance of normal zinchomeostasis for ER stress adaptation (27, 28, 60).In conclusion, we demonstrate that ZIP14-mediated hepatic

zinc uptake plays an important role in suppressing ER stress-induced apoptosis and hepatic steatosis (Fig. 8). ZIP14 displayed aprotective effect against ER stress by influencing the proapoptoticp-eIF2α/ATF4/CHOP pathway and de novo FA synthesis throughzinc-mediated inhibition of PTP1B activity. Deletion of Zip14produced a phenotype presenting with hepatic lipid accumulationduring ER stress. In addition, we demonstrated that Zip14 istranscriptionally regulated by ATF4 and ATF6α. The presentfindings document the importance of functional zinc transporterfor adaptation to ER stress.

Fig. 8. Role of ZIP14-mediated zinc transport in ER stress adaptation. Basedon the data in this report, we propose a cycle where ER stress sequentiallyincreases expression of ATF4 and ATF6α. The transcription factors increasetranscription of Zip14, leading to increased ZIP14 in hepatocytes. Enhancedtransporter activity increases intracellular zinc concentration, leading to in-hibition of PTP1B activity.

E5812 | www.pnas.org/cgi/doi/10.1073/pnas.1704012114 Kim et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

Materials and MethodsMice and Diets. Development and characterization of the murine Zip14−/− (KO)mice have been described previously (19). A colony of Zip14+/− heterozygotes onthe C57BL/6/129S5 background was used to produce KO and Zip14+/+ (WT) mice.The mice were backcrossed on a mixed 129S5 and C57BL/6 background. Mice hadfree access to a chow diet (Teklad 7912; Harlan) and tap water, and weremaintained on a 12-h light/dark cycle. The same response to treatments was seenin both male and female mice in pilot experiments. Young adult (8–16 wk of age)male WT and KOmice were used for TM experiments. To model HFD-induced ERstress, WT and KO mice were fed either the chow diet (17 kcal% fat) containing63 mg of zinc per kilogram as ZnO or an HFD (60 kcal% fat, D12492; ResearchDiets) containing 39 mg of zinc per kilogram as ZnCO3 for 16 wk.

Treatments. ER stress was induced by i.p. administration of TM (Sigma) orthapsigargin (Sigma) dissolved in 1% DMSO/150 mM glucose at 2 mg/kg of bodyweight (TM) or 1mg/kg of bodyweight (thapsigargin), respectively. To assess zincabsorption and tissue distribution, 65Zn (2 μCi; PerkinElmer) was given to mice byoral gavage 3 h before euthanasia. Accumulated 65Zn in tissue and plasma wasmeasured by gamma scintillation spectrometry. Mice were anesthetized by iso-flurane for injections and euthanasia by cardiac puncture. All mice were killedbetween 9:00 AM and 10:00 AM. Collected tissues were snap-frozen in liquidnitrogen and stored at −80 °C. All animal protocols were approved by theUniversity of Florida Institutional Animal Care and Use Committee.

Western Blot Analysis. Tissue samples or cells were homogenized in radio-immunoprecipitation assay lysis buffer (Santa Cruz Biotechnology) supple-mented with protease and phosphatase inhibitors (Thermo Fisher) usingBullet Blender (Next Advance). Proteins were then separated by SDS/PAGEand transferred to a nitrocellulose membrane. Polyclonal rabbit antibodyagainst ZIP14 was raised in-house as described previously (61). Purchasedantibodies were BiP, CHOP, and phosphorylated eIF2α (Ser52) (all from SantaCruz Biotechnology); ATF4, GRP94, and PTP1B (all from Cell Signaling); ATF6(Novus Bio); and tubulin (Abcam). Immunoreactivity was visualized by ECLreagents (Thermo Fisher). Immunoblots for ATF4, CHOP, and GRP94 werequantified using densitometry.

Statistical Analyses. Results are expressed as mean ± SD. Statistical signifi-cance was analyzed using GraphPad Prism 6 (GraphPad Software). Com-parisons between two groups were conducted by the Student’s t test, andcomparisons among multiple groups were conducted using ANOVA fol-lowed by a Tukey’s test. Statistical significance was set at P < 0.05.

ACKNOWLEDGMENTS. This research was supported by NIH Grant R01DK94244and the Boston Family Endowment of the University of Florida Foundation (toR.J.C.). M.-H.K. is an Alumni Graduate Fellow of the College of Agricultural andLife Sciences.

1. Ellgaard L, Helenius A (2003) Quality control in the endoplasmic reticulum. Nat RevMol Cell Biol 4:181–191.

2. Nair S, et al. (2007) Toxicogenomics of endoplasmic reticulum stress inducer tunica-mycin in the small intestine and liver of Nrf2 knockout and C57BL/6J mice. Toxicol Lett168:21–39.

3. Zhang X, et al. (2014) Endoplasmic reticulum stress induced by tunicamycin andthapsigargin protects against transient ischemic brain injury: Involvement of PARK2-dependent mitophagy. Autophagy 10:1801–1813.

4. Ji C, Kaplowitz N (2003) Betaine decreases hyperhomocysteinemia, endoplasmic re-ticulum stress, and liver injury in alcohol-fed mice. Gastroenterology 124:1488–1499.

5. Zhang K, Kaufman RJ (2008) From endoplasmic-reticulum stress to the inflammatoryresponse. Nature 454:455–462.

6. Li J, et al. (2012) Accumulation of endoplasmic reticulum stress and lipogenesis in theliver through generational effects of high fat diets. J Hepatol 56:900–907.

7. Zhang K (2010) Integration of ER stress, oxidative stress and the inflammatory re-sponse in health and disease. Int J Clin Exp Med 3:33–40.

8. Ozcan U, et al. (2004) Endoplasmic reticulum stress links obesity, insulin action, andtype 2 diabetes. Science 306:457–461.

9. Schröder M, Kaufman RJ (2005) ER stress and the unfolded protein response. MutatRes 569:29–63.

10. Baird TD, Wek RC (2012) Eukaryotic initiation factor 2 phosphorylation and trans-lational control in metabolism. Adv Nutr 3:307–321.

11. Walter P, Ron D (2011) The unfolded protein response: From stress pathway to ho-meostatic regulation. Science 334:1081–1086.

12. Rutkowski DT, et al. (2006) Adaptation to ER stress is mediated by differential sta-bilities of pro-survival and pro-apoptotic mRNAs and proteins. PLoS Biol 4:e374.

13. Yoshida H (2007) ER stress and diseases. FEBS J 274:630–658.14. Rutkowski DT, et al. (2008) UPR pathways combine to prevent hepatic steatosis

caused by ER stress-mediated suppression of transcriptional master regulators. DevCell 15:829–840.

15. King JC, Cousins RJ (2014) Zinc. Modern Nutrition in Health and Disease, ed Shilis ME(Lippincott Williams & Wilkins, Baltimore), 11th Ed, pp 189–205.

16. Kambe T, Tsuji T, Hashimoto A, Itsumura N (2015) The physiological, biochemical, andmolecular roles of zinc transporters in zinc homeostasis and metabolism. Physiol Rev95:749–784.

17. Jeong J, Eide DJ (2013) The SLC39 family of zinc transporters. Mol Aspects Med 34:612–619.

18. Lichten LA, Cousins RJ (2009) Mammalian zinc transporters: Nutritional and physio-logic regulation. Annu Rev Nutr 29:153–176.

19. Aydemir TB, et al. (2012) Zinc transporter ZIP14 functions in hepatic zinc, iron andglucose homeostasis during the innate immune response (endotoxemia). PLoS One 7:e48679.

20. Aydemir TB, Troche C, Kim MH, Cousins RJ (2016) Hepatic ZIP14-mediated zinctransport contributes to endosomal insulin receptor trafficking and glucose metab-olism. J Biol Chem 291:23939–23951.

21. Troche C, Aydemir TB, Cousins RJ (2016) Zinc transporter Slc39a14 regulates in-flammatory signaling associated with hypertrophic adiposity. Am J Physiol EndocrinolMetab 310:E258–E268.

22. Ellis CD, et al. (2004) Zinc and the Msc2 zinc transporter protein are required forendoplasmic reticulum function. J Cell Biol 166:325–335.

23. Homma K, et al. (2013) SOD1 as a molecular switch for initiating the homeostatic ERstress response under zinc deficiency. Mol Cell 52:75–86.

24. Sun Q, et al. (2015) Zinc deficiency mediates alcohol-induced apoptotic cell death inthe liver of rats through activating ER and mitochondrial cell death pathways. Am JPhysiol Gastrointest Liver Physiol 308:G757–G766.

25. Kim MH, Aydemir TB, Cousins RJ (2016) Dietary zinc regulates apoptosis through thephosphorylated eukaryotic initiation factor 2α/activating transcription factor-4/C/EBP-homologous protein pathway during pharmacologically induced endoplasmic re-ticulum stress in livers of mice. J Nutr 146:2180–2186.

26. Ishihara K, et al. (2006) Zinc transport complexes contribute to the homeostaticmaintenance of secretory pathway function in vertebrate cells. J Biol Chem 281:17743–17750.

27. Jeong J, et al. (2012) Promotion of vesicular zinc efflux by ZIP13 and its implicationsfor spondylocheiro dysplastic Ehlers-Danlos syndrome. Proc Natl Acad Sci USA 109:E3530–E3538.

28. Kurita H, Okuda R, Yokoo K, Inden M, Hozumi I (2016) Protective roles ofSLC30A3 against endoplasmic reticulum stress via ERK1/2 activation. Biochem BiophysRes Commun 479:853–859.

29. Gu F, et al. (2004) Protein-tyrosine phosphatase 1B potentiates IRE1 signaling duringendoplasmic reticulum stress. J Biol Chem 279:49689–49693.

30. Delibegovic M, et al. (2009) Liver-specific deletion of protein-tyrosine phosphatase 1B(PTP1B) improves metabolic syndrome and attenuates diet-induced endoplasmic re-ticulum stress. Diabetes 58:590–599.

31. Agouni A, et al. (2011) Liver-specific deletion of protein tyrosine phosphatase (PTP) 1Bimproves obesity- and pharmacologically induced endoplasmic reticulum stress.Biochem J 438:369–378.

32. Bellomo E, Massarotti A, Hogstrand C, Maret W (2014) Zinc ions modulate proteintyrosine phosphatase 1B activity. Metallomics 6:1229–1239.

33. Aydemir TB, Sitren HS, Cousins RJ (2012) The zinc transporter Zip14 influences c-Metphosphorylation and hepatocyte proliferation during liver regeneration in mice.Gastroenterology 142:1536–1546.e5.

34. Wu J, et al. (2007) ATF6alpha optimizes long-term endoplasmic reticulum function toprotect cells from chronic stress. Dev Cell 13:351–364.

35. Fusakio ME, et al. (2016) Transcription factor ATF4 directs basal and stress-inducedgene expression in the unfolded protein response and cholesterol metabolism in theliver. Mol Biol Cell 27:1536–1551.

36. Vecchi C, et al. (2009) ER stress controls iron metabolism through induction of hep-cidin. Science 325:877–880.

37. Buettner R, Schölmerich J, Bollheimer LC (2007) High-fat diets: Modeling the meta-bolic disorders of human obesity in rodents. Obesity (Silver Spring) 15:798–808.

38. Lee AS (2001) The glucose-regulated proteins: Stress induction and clinical applica-tions. Trends Biochem Sci 26:504–510.

39. Eide DJ (2006) Zinc transporters and the cellular trafficking of zinc. Biochim BiophysActa 1763:711–722.

40. Suzuki T, et al. (2005) Zinc transporters, ZnT5 and ZnT7, are required for the activation ofalkaline phosphatases, zinc-requiring enzymes that are glycosylphosphatidylinositol-anchored to the cytoplasmic membrane. J Biol Chem 280:637–643.

41. Suzuki T, et al. (2005) Two different zinc transport complexes of cation diffusionfacilitator proteins localized in the secretory pathway operate to activate alkalinephosphatases in vertebrate cells. J Biol Chem 280:30956–30962.

42. Chun HS, Lee H, Son JH (2001) Manganese induces endoplasmic reticulum (ER) stressand activates multiple caspases in nigral dopaminergic neuronal cells, SN4741.Neurosci Lett 316:5–8.

43. Oubrahim H, Chock PB, Stadtman ER (2002) Manganese(II) induces apoptotic celldeath in NIH3T3 cells via a caspase-12-dependent pathway. J Biol Chem 277:20135–20138.

44. Seo YA, Li Y, Wessling-Resnick M (2013) Iron depletion increases manganese uptakeand potentiates apoptosis through ER stress. Neurotoxicology 38:67–73.

45. Hiramatsu N, et al. (2007) Rapid, transient induction of ER stress in the liver andkidney after acute exposure to heavy metal: Evidence from transgenic sensor mice.FEBS Lett 581:2055–2059.

Kim et al. PNAS | Published online July 3, 2017 | E5813

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0

46. Biagioli M, et al. (2008) Endoplasmic reticulum stress and alteration in calcium ho-meostasis are involved in cadmium-induced apoptosis. Cell Calcium 43:184–195.

47. Kitamura M, Hiramatsu N (2010) The oxidative stress: Endoplasmic reticulum stressaxis in cadmium toxicity. Biometals 23:941–950.

48. Kubota K, et al. (2005) Fluoride induces endoplasmic reticulum stress in ameloblastsresponsible for dental enamel formation. J Biol Chem 280:23194–23202.

49. Liuzzi JP, Aydemir F, Nam H, Knutson MD, Cousins RJ (2006) Zip14 (Slc39a14) mediatesnon-transferrin-bound iron uptake into cells. Proc Natl Acad Sci USA 103:13612–13617.

50. Pinilla-Tenas JJ, et al. (2011) Zip14 is a complex broad-scope metal-ion transporterwhose functional properties support roles in the cellular uptake of zinc andnontransferrin-bound iron. Am J Physiol Cell Physiol 301:C862–C871.

51. Zabolotny JM, et al. (2002) PTP1B regulates leptin signal transduction in vivo. Dev Cell2:489–495.

52. Zhang S, Zhang ZY (2007) PTP1B as a drug target: Recent developments in PTP1Binhibitor discovery. Drug Discov Today 12:373–381.

53. Popov D (2012) Endoplasmic reticulum stress and the on site function of residentPTP1B. Biochem Biophys Res Commun 422:535–538.

54. Yamamoto K, et al. (2010) Induction of liver steatosis and lipid droplet formation inATF6alpha-knockout mice burdened with pharmacological endoplasmic reticulumstress. Mol Biol Cell 21:2975–2986.

55. Zhang K, et al. (2011) The unfolded protein response transducer IRE1α prevents ERstress-induced hepatic steatosis. EMBO J 30:1357–1375.

56. Lee AH, Scapa EF, Cohen DE, Glimcher LH (2008) Regulation of hepatic lipogenesis by

the transcription factor XBP1. Science 320:1492–1496.57. Wang Y, et al. (2000) Activation of ATF6 and an ATF6 DNA binding site by the en-

doplasmic reticulum stress response. J Biol Chem 275:27013–27020.58. Koyanagi S, et al. (2011) cAMP-response element (CRE)-mediated transcription by

activating transcription factor-4 (ATF4) is essential for circadian expression of the

Period2 gene. J Biol Chem 286:32416–32423.59. Luo S, Baumeister P, Yang S, Abcouwer SF, Lee AS (2003) Induction of Grp78/BiP by

translational block: Activation of the Grp78 promoter by ATF4 through and upstream

ATF/CRE site independent of the endoplasmic reticulum stress elements. J Biol Chem

278:37375–37385.60. Ohashi W, et al. (2016) Zinc transporter SLC39A7/ZIP7 promotes intestinal epithelial

self-renewal by resolving ER stress. PLoS Genet 12:e1006349.61. Liuzzi JP, et al. (2005) Interleukin-6 regulates the zinc transporter Zip14 in liver and

contributes to the hypozincemia of the acute-phase response. Proc Natl Acad Sci USA

102:6843–6848.62. Rebouche CJ, Wilcox CL, Widness JA (2004) Microanalysis of non-heme iron in animal

tissues. J Biochem Biophys Methods 58:239–251.63. Xie L, Collins JF (2013) Transcription factors Sp1 and Hif2α mediate induction of the

copper-transporting ATPase (Atp7a) gene in intestinal epithelial cells during hypoxia.

J Biol Chem 288:23943–23952.

E5814 | www.pnas.org/cgi/doi/10.1073/pnas.1704012114 Kim et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

19,

202

0