chaperone properties of pdia3 participate in rapid membrane actions of 1α,25-dihydroxyvitamin d ...

13
Chaperone properties of Pdia3 participate in rapid membrane actions of 1,25-dihydroxyvitamin D3 Jiaxuan Chen 1 , Kirill S. Lobachev 2 , Brian J. Grindel 3 , Mary C. Farach-Carson 3 , Sharon L. Hyzy 1,4 , Khairat B. El-Baradie 2 , Rene Olivares-Navarrete 1,4 , Maryam Doroudi 2 , Barbara D. Boyan 1,2,4 , and Zvi Schwartz 1,4 1 Wallace H. Coulter Department of Biomedical Engineering and 2 School of Biology, GA Institute of Technology, 315 Ferst Dr NW, Atlanta, GA 30332-0363; 3 Department of Biochemistry and Cell Biology, Rice University, MS-140, Houston, TX 77251-1892; and 4 Department of Biomedical Engineering, VA Commonwealth University, 601 W Main St, Richmond, VA 23284 Journal: Molecular Endocrinology Protein disulfide isomerase family A, member 3 (Pdia3) mediates many of the plasma membrane (PM) associated rapid responses to 1,25-dihydroxyvitamin D3 (1,25(OH) 2 D 3 ). It is not well un- derstood how Pdia3, which is an endoplasmic reticulum (ER) chaperone, functions as a PM recep- tor for 1,25(OH) 2 D 3 . We mutated three amino acids (K214 and R282 in the calreticulin interaction site and C406 in the isomerase catalytic site), which are important for Pdia3’s ER chaperone function, and examined their role in responses to 1,25(OH) 2 D 3 . Pdia3 constructs with and with- out the ER retention signal KDEL were used to investigate the PM requirement for Pdia3. Finally, we determined if palmitoylation and/or myristoylation were required for Pdia3-mediated re- sponses to 1,25(OH) 2 D 3 . Overexpressing the Pdia3 R282A mutant in MC3T3-E1 cells increased PM phospholipase A2 activating protein (PLAA), c-Src, and caveolin-1, but blocked increases in 1,25(OH) 2 D 3 –stimulated protein kinase C (PKC) seen in cells overexpressing wild-type Pdia3 (Pdia3 Ovr cells). Cells overexpressing Pdia3 with K214A and C406S mutations had PKC activity comparable to untreated controls indicating that the native response to 1,25(OH) 2 D 3 also was blocked. Overexpressing Pdia3 [-KDEL] increased PM localization and augmented baseline PKC but the stimulatory effect of 1,25(OH) 2 D 3 was comparable to that seen in wild-type cultures. In contrast, 1,25(OH) 2 D 3 increased PGE2 in Pdia3 [KDEL] cells. While neither palmitoylation nor myristoylation was required for PM association of Pdia3, myristoylation was needed for PKC activation. These data indicate that both the chaperone functional domains and the subcellular location of Pdia3 control rapid membrane responses to 1,25(OH) 2 D 3 . O ver the past two decades, the steroid hormone, 1,25-dihydroxyvitamin D 3 (1,25(OH) 2 D 3 ) has drawn increasing attention due to its newly discovered functions in addition to calcium/phosphate homeostasis. These include regulation of mineralization by osteoblasts (1), matrix production and remodeling by chondrocytes (2), and contraction of cardiomyocytes (3). In patholog- ical conditions, 1,25(OH) 2 D 3 and its analogues have beneficial effects in treatment of multiple sclerosis (MS), diabetes and various types of cancer (4 – 8). While many of the effects of 1,25(OH) 2 D 3 occur through classic nu- clear vitamin D receptor (VDR) mediated gene expres- sion, receptor-mediated activation of membrane associ- ated signaling pathways also plays an important role. A number of rapid responses have been reported in 1,25(OH) 2 D 3 -responsive cells. In chondrocytes and os- teoblasts, 1,25(OH) 2 D 3 activates phospholipase A2 (PLA2) via phospholipase A2 activating protein (PLAA), resulting in release of arachidonic acid within seconds and subsequent production of prostaglandin E2 (PGE2) (9 – 13). In addition, phosphatidylinositol-dependent phos- pholipase C (PLC), protein kinase C (PKC) and the extra- cellular signal-regulated kinases 1 and 2 (ERK1/2) are rapidly increased downstream of PLA2 activation (9, 14, ISSN Print 0888-8809 ISSN Online 1944-9917 Printed in U.S.A. Copyright © 2013 by The Endocrine Society Received August 21, 2012. Accepted May 1, 2013. Abbreviations: ORIGINAL RESEARCH doi: 10.1210/me.2012-1277 Mol Endocrinol mend.endojournals.org 1 Molecular Endocrinology. First published ahead of print May 9, 2013 as doi:10.1210/me.2012-1277 Copyright (C) 2013 by The Endocrine Society

Upload: zvi

Post on 11-Dec-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

Chaperone properties of Pdia3 participate in rapidmembrane actions of 1�,25-dihydroxyvitamin D3

Jiaxuan Chen1, Kirill S. Lobachev2, Brian J. Grindel3, Mary C. Farach-Carson3,Sharon L. Hyzy1,4, Khairat B. El-Baradie2, Rene Olivares-Navarrete1,4,Maryam Doroudi2, Barbara D. Boyan1,2,4, and Zvi Schwartz1,4

1Wallace H. Coulter Department of Biomedical Engineering and 2School of Biology, GA Institute ofTechnology, 315 Ferst Dr NW, Atlanta, GA 30332-0363; 3Department of Biochemistry and Cell Biology,Rice University, MS-140, Houston, TX 77251-1892; and 4Department of Biomedical Engineering, VACommonwealth University, 601 W Main St, Richmond, VA 23284 Journal: Molecular Endocrinology

Protein disulfide isomerase family A, member 3 (Pdia3) mediates many of the plasma membrane(PM) associated rapid responses to 1�,25-dihydroxyvitamin D3 (1�,25(OH)2D3). It is not well un-derstood how Pdia3, which is an endoplasmic reticulum (ER) chaperone, functions as a PM recep-tor for 1�,25(OH)2D3. We mutated three amino acids (K214 and R282 in the calreticulin interactionsite and C406 in the isomerase catalytic site), which are important for Pdia3’s ER chaperonefunction, and examined their role in responses to 1�,25(OH)2D3. Pdia3 constructs with and with-out the ER retention signal KDEL were used to investigate the PM requirement for Pdia3. Finally,we determined if palmitoylation and/or myristoylation were required for Pdia3-mediated re-sponses to 1�,25(OH)2D3. Overexpressing the Pdia3 R282A mutant in MC3T3-E1 cells increased PMphospholipase A2 activating protein (PLAA), c-Src, and caveolin-1, but blocked increases in1�,25(OH)2D3–stimulated protein kinase C (PKC) seen in cells overexpressing wild-type Pdia3(Pdia3Ovr cells). Cells overexpressing Pdia3 with K214A and C406S mutations had PKC activitycomparable to untreated controls indicating that the native response to 1�,25(OH)2D3 also wasblocked. Overexpressing Pdia3[-KDEL] increased PM localization and augmented baseline PKC butthe stimulatory effect of 1�,25(OH)2D3 was comparable to that seen in wild-type cultures. Incontrast, 1�,25(OH)2D3 increased PGE2 in Pdia3[�KDEL] cells. While neither palmitoylation normyristoylation was required for PM association of Pdia3, myristoylation was needed for PKCactivation. These data indicate that both the chaperone functional domains and the subcellularlocation of Pdia3 control rapid membrane responses to 1�,25(OH)2D3.

Over the past two decades, the steroid hormone,1�,25-dihydroxyvitamin D3 (1�,25(OH)2D3) has

drawn increasing attention due to its newly discoveredfunctions in addition to calcium/phosphate homeostasis.These include regulation of mineralization by osteoblasts(1), matrix production and remodeling by chondrocytes(2), and contraction of cardiomyocytes (3). In patholog-ical conditions, 1�,25(OH)2D3 and its analogues havebeneficial effects in treatment of multiple sclerosis (MS),diabetes and various types of cancer (4–8). While manyof the effects of 1�,25(OH)2D3 occur through classic nu-clear vitamin D receptor (VDR) mediated gene expres-

sion, receptor-mediated activation of membrane associ-ated signaling pathways also plays an important role.

A number of rapid responses have been reported in1�,25(OH)2D3-responsive cells. In chondrocytes and os-teoblasts, 1�,25(OH)2D3 activates phospholipase A2(PLA2) via phospholipase A2 activating protein (PLAA),resulting in release of arachidonic acid within seconds andsubsequent production of prostaglandin E2 (PGE2) (9–13). In addition, phosphatidylinositol-dependent phos-pholipase C (PLC), protein kinase C (PKC) and the extra-cellular signal-regulated kinases 1 and 2 (ERK1/2) arerapidly increased downstream of PLA2 activation (9, 14,

ISSN Print 0888-8809 ISSN Online 1944-9917Printed in U.S.A.Copyright © 2013 by The Endocrine SocietyReceived August 21, 2012. Accepted May 1, 2013.

Abbreviations:

O R I G I N A L R E S E A R C H

doi: 10.1210/me.2012-1277 Mol Endocrinol mend.endojournals.org 1

Molecular Endocrinology. First published ahead of print May 9, 2013 as doi:10.1210/me.2012-1277

Copyright (C) 2013 by The Endocrine Society

15). Moreover, in skeletal muscle cells, c-Src was found tobe rapidly activated by 1�,25(OH)2D3 (16–18) and rapidmovement of Ca2� across cell membranes was shown in anumber of cells to follow 1�,25(OH)2D3 addition (19,20).

Protein disulfide isomerase family A, member 3 (Pdia3,also called ERp57, ERp60, Grp58, and 1,25-MARRS)has been proposed to mediate many of these rapid re-sponses to 1�,25(OH)2D3. Pdia3 was initially isolatedfrom the basal lateral membranes of chicken intestinalepithelial cells based on its saturable binding to1�,25(OH)2D3 (21). Antibodies to the N-terminal pep-tide of the protein block Ca2� and phosphate transportacross the membrane in response to 1�,25(OH)2D3 (22)and interfere with rapid activation of PKC� in chondro-cytes and osteoblasts (23–25). Similarly, epithelial cellsisolated from Pdia3-conditional knockout mice lack sur-face binding of 1�,25(OH)2D3 and 1�,25(OH)2D3-stim-ulated calcium uptake (19). The stimulatory effect of1�,25(OH)2D3 is stereospecific, indicating a receptor-mediated mechanism (26). Moreover, mice lacking afunctional VDR possess Pdia3 and cells isolated fromthese mice respond to 1�,25(OH)2D3 with an increase inPKC activity (27). Recently, we showed that embryonicstem cells possess Pdia3 and respond to 1�,25(OH)2D3

with an increase in PKC activity (28).These observations support the hypothesis that Pdia3

is a receptor for the secosteroid. Pdia3 can be located incaveolae where it physically interacts with the scaffoldingprotein caveolin-1 and with PLAA (1, 29). Disruption ofcaveolae with �-cyclodextrin prevents 1�,25(OH)2D3-dependent PKC activation (13, 30). In addition, cells frommice lacking functional caveolin-1 (Cav1-/-) fail to in-crease enzyme activity in response to the secosteroid (30),demonstrating the importance of this specialized plasmamembrane domain to the function of the receptor. Whilethese data demonstrate the role of Pdia3 in the rapidresponse to 1�,25(OH)2D3, global knockout of Pdia3 isembryologically lethal (31), suggesting it also plays addi-tional critical roles along with its function as a receptorfor 1�,25(OH)2D3.

Outside the field of vitamin D, Pdia3 can act as a chap-erone protein in the endoplasmic reticulum (ER) where itpromotes formation of disulfide bonds in its N-glycosy-lated protein substrate through interaction with the ERlectin chaperones, calreticulin and calnexin (32–34).Pdia3 is also intensively studied for its role in assisting theformation of the major histocompatibility complex(MHC) class I peptide-loading complex, which is essen-tial for formation of the final antigen conformation andexport from the ER to the cell surface (31, 35, 36). Howthese chaperone properties might impact Pdia3’s role as a

membrane receptor for 1�,25(OH)2D3 is not known.However, a number of studies suggest that Pdia3 does actin the assembly of proteins involved in membrane signal-ing by the hormone. In addition to the association ofcalreticulin and calnexin with Pdia3 as a chaperone pro-tein (32–34), calreticulin was also shown to participate inthe action of 1�,25(OH)2D3 (37). PLAA and caveolin-1can be immunoprecipitated with Pdia3 in MC3T3-E1preosteoblasts (29), and c-Src was found to be immuno-precipitated with VDR and to be involved in rapid re-sponses to 1�,25(OH)2D3 in muscle cells (16).

Previous mutagenesis studies have found certain aminoacids are important to the chaperone function of Pdia3.Pdia3 is composed of four thioredoxin-like domains a, a�,b, and b� with two catalytic CGHC motifs in the a and a�domains (38). Amino acid C406 is located in the a� do-main and functions as one of the two catalytic sites; mu-tating C406 (C406S) significantly decreases oxidoreduc-tase activity and chaperone function (36, 39, 40). The cliffbetween b and b� interacts with the proline-rich P-domainof calnexin and calreticulin (38, 41, 42). This interactionis required for the proper folding of the substrate and iscompletely blocked by mutating arginine 282 to alanine(R282A) (38). Another important amino acid at this in-terface is K214, which when mutated (K214A), reducesthe interaction by eight fold (38). Interestingly, calreticu-lin has been shown to interact with the ligand-VDR com-plex in ROS 2.8 osteoblastic cells (37).

Pdia3 has broad subcellular distribution. It has beenfound in the ER, plasma membrane, extracellular matrix,cytoplasm, and nucleus (1, 40, 43, 44). This is partiallyattributable to its compromised and inefficient ER reten-tion signal. Pdia3 has a C-terminal QEDL retention sig-nal, rather than the classic KDEL retention signal. As aresult, Pdia3 can escape from ER retrieval, go through theconstitutive secretary pathway, and eventually be se-creted into the extracellular matrix (40); or as is the casein cartilage, Pdia3 is released into the matrix as a compo-nent of extracellular matrix vesicles (23). We have shownthat 1�,25(OH)2D3 directly activates Pdia3-dependentPKC and PLA2 in isolated plasma membranes (23, 29, 33,45), indicating that the apparatus needed is present, butwhether Pdia3 can function as a mediator of1�,25(OH)2D3 rapid actions in other subcellular com-partments is less clear.

The mechanism of how Pdia3 associates with theplasma membrane is not known. The protein has myris-toylation sites, which may participate in its retention as aplasma membrane associated protein during membranebiosynthesis. The recent observation that steroid hor-mone receptors associate with the plasma membrane viapalmitoylation (46–48), suggests the possibility that

2 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

Pdia3 uses this mechanism as well. However, no palmi-toylation sites have been identified in Pdia3, but its func-tion may require other components that use this mem-brane-association strategy.

The purpose of this study was to determine if the sameamino acids that are critical for the chaperone function ofPdia3 in the endoplasmic reticulum are also critical for itsfunction in mediating rapid signaling by 1�,25(OH)2D3

at the plasma membrane. While there are considerabledata indicating that Pdia3 is present in caveolae and thatcaveolae are required for its receptor function, it is notknown if some of its 1�,25(OH)2D3-dependent effectsoccur in other cellular compartments, including the endo-plasmic reticulum, where it works as a chaperone. Toaddress the first question, we used site-directed mutagen-esis of three important amino acids: C406, K214 andR282, which are required for the chaperone role of Pdia3,to dissect these individual functions. To answer the sec-ond question, we changed the original QEDL ER-reten-tion signal to either a classic KDEL signal to trap Pdia3 inER or removed QEDL completely to prevent ER retrievaland allow Pdia3 to move to other cellular locations in-cluding the plasma membrane. We also inhibited palmi-toylation and myristoylation to investigate if they are alsoimportant to the plasma membrane association of Pdia3and its function in rapid responses.

Materials and Methods

Plasmid Construction

Chaperone FunctionTo study whether the amino acids involved in Pdia3’s chap-

erone function in the ER are also important to its receptor func-tion, we mutated lysine 214 to alanine (K214A), arginine 282 toalanine (R282A) and cystine 406 to serine (C406S) on a nativemouse Pdia3 overexpression vector (Pdia3Ovr, OriGene, Rock-ville, MD) (Figure 1A) using a QuikChange II XL Site-DirectedMutagenesis Kit (Agilent Technologies, Santa Clara, CA). In thekit, a polymerase chain reaction (PCR) was used to generate themutation. Briefly, the original plasmid was denatured and an-nealed with the mutagenic primers containing the desired mu-tation. Then the primers were extended by pfuUltra DNA poly-merase followed by Dpn I enzyme digestion to break down theparental methylated and hemi-methylated DNA. The mutatedplasmid was further transformed into XL10-Gold ultracompe-tent cells for nick repair. Plasmids were isolated and purifiedwith EndoFree plasmid maxi kit (Qiagen, Valencia, CA) forsequencing (Eurofins MWG Operon, Huntsville, AL) or latertransfection. The whole open reading frames of Pdia3Ovr,K214A, R282A and C406S were sequenced to ensure no ran-dom mutations had occurred (data not shown). Sequence resultsshowed that the desired site-directed mutations were achieved inall constructs (Figure 1B).

FIGURE 1. Diagram showing strategy for site-directed mutagenesis. (A): Four different plasmid constructs. (B): 3D structure of Pdia3 showing thelocation of the three mutated amino acids. Purple: K214; green: R282; yellow: C406. Blue: the cliff between b and b� domain. (C): Sequencingresult of mutated 214, 282 and 406. For 214, AAG (lysine) was mutated to GCG (alanine); for 282, AGG (arginine) was mutated to GCG (alanine);for 406; TGT (cystine) was mutated to TCC (serine).

doi: 10.1210/me.2012-1277 mend.endojournals.org 3

Endoplasmic Reticulum v. Plasma MembraneLocation

To study the effect of subcellular location of Pdia3 on rapidresponses, we used a Pdia3[�KDEL] construct generated as de-scribed previously (44). Briefly, the DNA plasmid was based ona mammalian expression vector from Clontech (MountainView, CA). This plasmid contains the full length human Pdia3gene with a deletion of the ER retention signal QEDL on theC-terminal. This Pdia3 was fused with green fluorescent protein(GFP) containing an A206K mutation on the C-terminal fol-lowed by a classic ER retention signal KDEL (Figure 2A). ThisGFP fusion was previously found not to change the function ofPdia3 in response to a TNF-� treatment (44). In order to deleteKDEL, the last amino acid codon of GFP on the C-terminal wasmutated to a stop codon (Pdia3[-KDEL]) using the same strategydescribed above for the site directed mutagenesis of the chaper-one catalytic sites. Plasmids containing Pdia3[�KDEL] or Pdia3[-

KDEL] were prepared as above. The entire open reading framewas sequenced; no random mutations were observed (Figure2B). Sequences of the primers used in site-directed mutagenesisare available upon request.

Transfection and OverexpressionThe respective Pdia3 native or mutant proteins were overex-

pressed in wild type (WT) mouse MC3T3-E1 preosteoblast-likecells (CRL-2593, ATCC, Manassas, VA). As a result, normalPdia3 was also present. Unfortunately, we were not able togenerate MC3T3-E1 cells that were both silenced for Pdia3 andwere viable after transfection with any of the overexpressionplasmids. The MC3T3-E1 cells were plated at a density of20,000 cells/cm2 in 6-well plates and cultured in �-MEM sup-plied with 10% (v/v) fetal bovine serum (FBS). After 24 h, 500�lof Opti-MEM (Invitrogen, Carlsbad, CA) containing 5�l lipo-fectamine LTX (Invitrogen), 2.5�l plus reagent (Invitrogen) and2.5 �g plasmids were added into each well. After 48 h, stabletransfected cell lines were created by culturing the cells in selec-tion medium (�-MEM containing 10% FBS, 1% (v/v) penicillin/streptomycin (P/S) and 550 �g/ml G418 (Cellgro, Manassas,VA)) for two weeks according to the company’s directions (Ori-Gene). Overexpression rates were monitored to be stable acrossthe passages used (data not shown).

Successful overexpression of Pdia3[�KDEL] and Pdia3[-KDEL]

(Pdia3[�KDEL]) was determined by realtime PCR measurementof Pdia3GFP fused mRNA in Pdia3[�KDEL] and Pdia3[-KDEL]

cells. Ten days after plating, cell layers were dissolved in TRIzol(Invitrogen) and mRNA extracted and reverse-transcribed intocDNA using the high-capacity cDNA reverse transcription kit(Applied Biosystems, Carlsbad, CA) according to the manufac-turer’s directions. Real-time PCR was performed using SYBRGreen Master Mix (Applied Biosystems) for Pdia3GFP and glyc-eraldehyde 3-phosphate dehydrogenase (Gapdh) encoding tran-scripts. Oligonucleotide primers were designed using BeaconDesigner 7.0 software. For Pdia3GFP transcripts, the forwardprimer (ACCATATACTTCTCTCCAGCCAAC) targeted the Cterminal of human Pdia3 (NM 005313.4) and the reverseprimer (TCCTCGCCCTTGCTCACC) targeted the N terminalof GFP (FM177581.1). The primers were designed using BeaconDesigner 7.0 software (PREMIER Biosoft International, PaloAlto, CA) and synthesized by Eurofins MWG Operon (Hunts-ville, AL). Real-time PCR was performed using the Veriti 96well Thermal Cycler (Applied Biosystems, Carlsbad, CA) withStep One software (Applied Biosystems). Data were normalizedto the endogenous reference gene Gapdh.

Cell CultureAll cell lines were plated at 10,000 cells/cm2 in T75 or 24-

well plates with full medium (�-MEM containing 10% FBS, and1% P/S) with or without 550 �g/ml G418 to select for success-fully transfected cells. After 48 h, full medium was changed tofull medium containing 1% (w/v) vitamin C to enable cross-linking of type I collagen in the extracellular matrix (49). Otherthan for experiments involving confocal microscopy, all cellswere cultured for 10 d after plating and treated with full mediumcontaining either the ethanol vehicle alone or with the appro-priate dose of 1�,25(OH)2D3.

Effect of Mutant Pdia3 Overexpression

Plasma Membrane IsolationIn order to study the changes in plasma membrane associa-

tion of Pdia3 and of proteins potentially involved in its mecha-nism of action, a detergent-free method of plasma membraneisolation was used as described previously (50). Ten days afterplating, cell layers were scraped in isolation buffer (0.25M su-

FIGURE 2. Diagram showing strategy for changing ER retention signal. (A): Pdia3[-KDEL] and Pdia3[�KDEL] constructs. (B): The sequencing result ofthe deletion of ER retention signal KDEL. Lower panel is the sequence result of Pdia3[�KDEL] showing that the KDEL retention signal is present onthe c-terminal of green fluorescence protein. Upper panel is the sequence result of Pdia3[-KDEL] showing that the last amino acid (nucleotides TAC)of green fluorescent protein was mutated to a stop codon (TAA).

4 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

crose, 1 mM EDTA, 20 mM tricine, pH 7.8). Samples werehomogenized using a tissue grinder for twenty strokes. Homog-enates were centrifuged at 20,000g for 10 min to pellet celldebris including nuclei, mitochondria, and ER. The supernatantwas collected, placed on top of 30% (v/v) Percoll (GE Health-care, Piscataway, NJ) in isolation buffer, and then centrifugedfor 30 min at 84,000g. The plasma membranes formed a visibleband and were collected by aspiration. This method was re-ported to have 5.2 fold enrichment of plasma membrane (51)and we showed this fraction is free of nucleus and mitochondriacontamination by western blot against TATA box binding pro-tein and COXVI (data not shown). The levels of each protein ofinterest in the plasma membrane fraction were examined bywestern blots.

Western BlotsWestern blots were performed using whole cell lysates and

isolated plasma membranes to examine the protein level ofPdia3, caveolin-1, Pdia3-interacting proteins calreticulin andcalnexin, and the signaling mediators c-Src and PLAA. The sam-ples were mixed with loading buffer, boiled and followed by gelelectrohoresis using NuSep 4%–20% LongLife Gels (NuSep,Lawrenceville, GA). Proteins were transferred to nitrocellulosemembrane by iBlot Dry Blotting System (Invitrogen). The mem-brane was subsequently blotted in 1% (w/v) bovine serum al-bumin (Sigma-Aldrich, St. Louis, MO) in PBS with primaryantibodies against Pdia3 (Alpha Diagnostic International, Inc.),Gapdh (MAB374, Millipore, Billerica, MA), calreticulin (TO-5,Santa Cruz Biotechnology), calnexin (A-9, Santa Cruz Biotech-nology), caveolin-1 (N-20, Santa Cruz Biotechnology), c-Src(B-12, Santa Cruz Biotechnology), and PLAA (custom antibody,Strategic Diagnostics, Inc., Newark, DE). After washing withPBS containing 0.05% (v/v) Tween-20, the membrane was in-cubated with goat antirabbit, goat antimouse or donkey anti-goat horseradish peroxidase conjugated secondary antibodies(Bio-Rad, Hercules, CA) in PBS containing 5% (w/v) dry milkand 0.05% Tween-20. After washes, the membrane was devel-oped using SuperSignal West Pico Chemiluminescent System(Thermo Fisher Scientific, Rockford, IL) and imaged with theVersaDoc imaging system (Bio-Rad, Hercules, CA).

In order to compare the effect of each Pdia3 mutation onassociation of proteins with the plasma membrane, we quanti-fied the data by normalizing the intensity of targeted proteins tothe corresponding Pdia3 intensity and the ratio in the mutant-overexpressing cells was further divided by the ratio for cellsoverexpressing the wild type Pdia3 (Pdia3Ovr).

Subcellular Location of Pdia3To determine if the presence of the ER-retention signal

KDEL altered the amount of Pdia3 in the plasma membrane,plasma membranes were isolated from wild type, Pdia3[�KDEL]

and Pdia3[-KDEL] cells as described before. A Synergy H4 mul-timode plate reader (Biotek, Winooski, VT) was used to detectthe green fluorescence in the samples. The relative fluorescenceunits were further normalized by total protein levels of the iso-lated membranes.

To visualize the difference in subcellular location, confocalfluorescence microscopy was performed to detect the green flu-orescence signal of the Pdia3GFP fused protein. Pdia3 was pre-viously shown to colocalize with lipid rafts in the plasma mem-

brane (1), wild type, Pdia3[�KDEL] and Pdia3[-KDEL] cells werecultured in chamber slides for 24 h and stained with Vybrant®

Alexa Fluor® 594 Lipid Raft Labeling Kit (Invitrogen). Cellswere further stained with Hoechst 33342 (Invitrogen) to labelthe nucleus. After washing, the cells were fixed with FLURO-GEL mounting medium (Electron Microscopy Sciences, Hat-field, PA) and visualized using a Zeiss LSM 510 confocal micro-scope (Carl Zeiss MicroImaging, Thornwood, NY). Imageswere obtained at room temperature with 40 � 1.3 objective lensand 10X ocular lens. No digital enlargement was applied. Toquantify the pixel distribution of Pdia3, the intensities of thegreen Pdia3 signal and the red lipid raft signal were calculatedalong a cross section, and plotted against the distance by ZeissLSM Image software (Carl Zeiss MicroImaging).

Signaling by 1�,25(OH)2D3

To study the effect of genetically modifying Pdia3 on1�,25(OH)2D3 induced rapid responses, 10 d after plating, allcell lines were treated with vehicle (ethanol) or 10-7 M1�,25(OH)2D3. To measure PKC activity, treatment wasstopped at 15 min and cell layers were washed twice with PBSand lysed in RIPA buffer (20 mM Tris-HCl, 150 mM NaCl, 5mM disodium EDTA, 1% (v/v) NP-40). PKC activity was mea-sured using a commercial kit (RPN77, GE Healthcare, Piscat-away, NJ) and normalized by total protein of the cell layer.

Our lab has previously shown the effect of 1�,25(OH)2D3 onPKC is via a PLA2-dependent pathway (11). PGE2 as an indirectproduct of PLA2 action has been used as an outcome measure-ment of rapid responses (1). To measure PGE2 release into themedia, after 30 min of treatment, the conditioned media werecollected and acidified by adding hydrochloride to a final con-centration of 0.1M. PGE2in the conditioned media was mea-sured using a commercial PGE2 radioimmunoassay (RIA) kit(Perkin Elmer, Waltham, MA) and normalized by total DNA ofthe cell layer.

Palmitoylation and MyristoylationTo determine if palmitoylation or myristoylation is required

for the membrane association of Pdia3 or its function in medi-ating 1�,25(OH)2D3 dependent PKC activation, we depletedthe palmitoylated protein or myristoylated protein from theplasma membrane by pretreating the cells with tunicamycin or2-hydroxymyristic acid (HMA) prior to treatment with1�,25(OH)2D3 (52). Tunicamycin inhibits the N-linked glyco-sylation of proteins and has been demonstrated to inhibit pal-mitoylation of membrane associated receptors (46, 52). HMA isalso used to inhibit myristoylation of membrane associated pro-tein (53). Ten days after plating, wild type MC3T3-E1 cells weretreated with vehicle (DMSO) or 1 �g/ml tunicamycin (Sigma-Aldrich) or 1 mM HMA (Cayman Chemical, Ann Arbor, MI)for 48 h. To detect the changes in plasma membrane associationof Pdia3, at the end of 48 h, plasma membrane isolation wasperformed followed by western blots against pan-cadherin(CDH, Abcam, Cambridge, MA), caveolin-1 and Pdia3 as de-scribed previously. To study the effect of these inhibitors on1�,25(OH)2D3 induced rapid responses, at the end of 48 h, cellswere treated with vehicle (ethanol) or 10-7M 1�,25(OH)2D3 for15 min and the PKC activity was measured as describedpreviously.

doi: 10.1210/me.2012-1277 mend.endojournals.org 5

Statistical AnalysisEach experiment was repeated at least once to ensure the

validity of the data. The data presented are from a single repre-sentative experiment. Each data point represents the means �standard error for six independent cell cultures. Significancewas determined by one-way analysis of variance (ANOVA) andpost hoc testing performed using Bonferroni’s modification ofStudent’s t test for multiple comparisons. P � .05 was consid-ered significant.

For experiments using separation of functional alleles, be-cause 5 different cell lines were compared, treatment/controlratios were calculated to show the effect of 1�,25(OH)2D3. Thevalue for each sample in the treated group was divided by themean of the control group. Each data point represents themeans � standard error for six normalized values and a dashedline with value of 1 represents the control. Due to the non-normal distribution, significance was determined using theMann-Whitney test. P � .05 was considered significant.

For quantification of western blots, the values of experimen-tal samples were divided by the value of control sample for eachexperiment. Four independent experiments were performed.Each data point represents the mean � standard error of mean(SEM) of the 4 independent experiments with the control rep-resented by a dashed line with value of one. Because the data areboth non-normal distributed and paired, significance was deter-mined by Wilcoxon matched pair test between the mutants andthe Pdia3Ovr control and Wilcoxon rank-sum test between mu-tants; P � .05 was considered to be significant.

Results

Overexpression of Pdia3 mutants changed theplasma membrane presence of signaling molecules

Western blots showed that all cells containing the over-expression plasmid, including the wild type control plas-mid Pdia3Ovr and the plasmids for overexpressing themutant proteins K214A, R282A and C406S had a darkerimmunoreactive Pdia3 band compared to wild type cellswithout a plasmid, indicating a possible feedback loop forsynthesis (Supplemental Figure 1). Bands for the internalloading control Gapdh were comparable among cell lines.Plasma membranes isolated from the overexpressing celllines also had more immunoreactive Pdia3 compared towild type plasma membranes (Figure 3A). Calreticulinwas also increased in the plasma membranes isolatedfrom the Pdia3Ovr cells and to an even greater extent inplasma membranes isolated from each of the mutant celllines. PLAA was reduced in Pdia3Ovr and K214A cellmembranes compared to plasma membranes from non-transfected cells, whereas there appeared to be more im-munoreactive PLAA in plasma membranes from R282Acells. Similarly, c-Src band intensity was increased in theR282A membranes but it was reduced in the Pdia3Ovr

membranes. Caveolin-1 band intensity was comparable

FIGURE 3. Effect of overexpressing site directed mutants of Pdia3 on plasma membrane association of signaling molecules. Ten days afterplating, wild type, Pdia3Ovr, K214A, R282A, and C406S MC3T3–E1 cells were harvested for plasma membrane isolation. (A): Western blots ofPdia3, calreticulin, caveolin-1, PLAA, and c-Src in plasma membranes. (B to E): Image quantification of (A). The pixel intensity of the targetedproteins was first normalized by its corresponding Pdia3 intensity, then the ratio in the mutant cells was further divided by the value in Pdia3Ovr

blots. & P � .05, R282A and C406S vs. K214A; % P � .05, C406S vs. R282A.

6 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

in all plasma membranes. Immunoreactive calnexinbands were not seen in any of the plasma membranesexamined.

Normalizing individual protein bands to their corre-sponding Pdia3 and then comparing the ratio to that ofthe Pdia3Ovr plasma membranes demonstrated that mu-tations in the chaperone interaction sites altered the pres-ence of proteins in the plasma membrane fraction. Theratio between calreticulin and Pdia3 was unchanged inany of the mutants (Figure 3B), but R282A cells exhibitedincreased caveolin-1 per Pdia3 than K214A cells (Figure3C), greater PLAA per Pdia3 than K214A cells (Figure3D) and greater Src/Pdia3 than either K214A or C406Scells (Figure 3E). In contrast, no differences were ob-served in western blots of the whole cell lysates for any ofthese proteins (data no shown). These data suggest that bychanging specific amino acids in Pdia3 we could changethe plasma membrane association of downstream signal-ing molecules.

Mutating Pdia3 alters 1�,25(OH)2D3-stimulatedrapid responses

1�,25(OH)2D3 stimulated PKC specific activity inwild type MC3T3-E1 cells, causing a 30% increase after15 min (Figure 4A). The effect of 1�,25(OH)2D3 wasgreater in Pdia3Ovr cells overexpressing the native Pdia3.The stimulatory effect of the secosteroid was abolished incells overexpressing Pdia3 with the K214A or C406S mu-tation. The effect of 1�,25(OH)2D3 on PKC in cells over-expressing the R282A mutation was comparable to thatof wild type cells.

PGE2 release was affected in a similar manner (Figure4B). 1�,25(OH)2D3 increased PGE2 in the conditionedmedia at 30 min in wild type cells and this response wasenhanced in cells transfected with the normal Pdia3Ovr

plasmid. Mutating C406S blocked PGE2 release whereas

mutating R282A or K214A resultedin PGE2 release comparable to thatof wild type cells, preventing the in-crease observed in the Pdia3Ovr cells.

The subcellular location ofPdia3 was sensitive to thepresence of the retention signal

Realtime PCR of the fusedPdia3GFP mRNA showed a similaramount of overexpression in bothPdia3[�KDEL] cells and Pdia3[-KDEL]

cells (Figure 5A). Pdia3[�KDEL] cellshad greater green fluorescence in-tensity in the isolated plasma mem-branes than wild type cells and thefluorescence signal was markedly

greater in Pdia3[-KDEL] cells (Figure 5B). Confocal micro-scope images of wild type cells exhibited lipid raft stainingdiffusely throughout the cell surface and no green signalwas observed in the background (Figure 5C). The GFPsignal of Pdia3[�KDEL] was present in the peri-nuclearregion and nucleus but little fluorescence was observed inthe cytoplasm and plasma membrane. In contrast, theGFP signal of Pdia3[-KDEL] was observed in the perinu-clear region, nucleus, cytoplasm, and plasma membraneas indicated by the white arrows. The histogram of pixelintensity across the cell confirmed these observations. Thegreen Pdia3[�KDEL] signal peaked near the blue nuclearsignal (Figure 5D) whereas the red plasma membrane sig-nal distributed evenly across the cell. In contrast, Pdia3[-

KDEL] fluorescence intensity was high in the plasma mem-brane and nuclear regions while fluctuating at a low levelin cytosol (Figure 5E). These data suggest that the ERretention signal is an important switch to control the pres-ence of Pdia3 on the plasma membrane.

Rapid responses to 1�,25(OH)2D3 are sensitive tothe presence of the ER retention signal

The presence of KDEL did not alter the level of PKCactivity in MC3T3-E1 cells in comparison to wild typecultures, which contained endogenous Pdia3 (Figure 6A).However, removal of the ER retention signal resulted in a100% increase in PKC compared to the wild type cells.Despite this increase in enzyme activity due to overexpres-sion of Pdia3[-KDEL] in these cells, the stimulatory effect of1�,25(OH)2D3 was unchanged. In contrast, PGE2 releasewas not affected by the presence or absence of the KDELmotif (Figure 6B). 1�,25(OH)2D3 increased PGE2 by50% in the conditioned media of wild type cells but over-expression of either Pdia3[�KDEL] or Pdia3[-KDEL] aug-mented the increase to more than 100%.

FIGURE 4. The effect of overexpressing site directed mutants of Pdia3 on rapid responses to1�,25(OH)2D3. 10 d after plating, wild type, Pdia3Ovr, K214A, R282A, and C406S MC3T3–E1cells were treated with or without 10-7 M 1�,25(OH)2D3. (A): PKC activity was measured at 15min. (B): PGE2 in conditioned media was measured at 30 min. PKC activity was normalized tototal protein and PGE2 was normalized to total DNA. Data were normalized to the vehicletreated group (dashed line � 1). * P � .05, 1�,25(OH)2D3 vs. control; # P � .05, mutations vs.WT; ˆ P � .05, mutations vs. Pdia3Ovr.

doi: 10.1210/me.2012-1277 mend.endojournals.org 7

Palmitoylation and myristoylation are not requiredfor PM association of Pdia3

Treatment with tunicamycin significantly decreasedthe plasma membrane association of N-glycosylated pan-

cadherin and slightly decreasedS-palmitoylated caveolin-1 but didnot affect Pdia3 (Figure 7A). More-over, the rapid 30% increase in PKCactivity in response to1�,25(OH)2D3 was not affected inthe tunicamycin treated cells (Figure7B). HMA treatment also did not af-fect the membrane association ofPdia3; neither did it affect the non-myristoylated caveolin-1 or pan-cadherin (Figure 7C). Interestingly,48 h of pretreatment with HMA didattenuate 1�,25(OH)2D3-inducedPKC activation, indicating myris-toylation was needed for the overall

function of the rapid response (Figure 7D).

FIGURE 5. Subcellular location of Pdia3[�KDEL] and Pdia3[-KDEL] proteins. (A): mRNA for Pdia3GFP in wild type, Pdia3[�KDEL] and Pdia3[-KDEL] cells.The realtime PCR was performed with forward primer targeting on Pdia3 and reverse primer targeting on GFP. (B): Green fluorescence intensity inthe plasma membranes of wild type, Pdia3[�KDEL] and Pdia3[-KDEL] cells. Cells were harvested for plasma membrane isolation. The greenfluorescence intensity in the plasma membranes was measured by microplate reader and normalized by total protein. (C): Confocal microscopyimage of wild type, Pdia3[�KDEL], and Pdia3[-KDEL] cells. At 80% confluence, lipid raft staining (red) and Hoechst staining (blue) of the nucleus wereperformed. Green fluorescence (green) of Pdia3GFP fused protein was detected. Yellow represents the merged signal of Pdia3 and lipid rafts.White arrows show green merges with red on the cell boundary. (D, E): Fluorescence intensity histogram of confocal microscopy image. Pixelintensity was calculated along the blue lines indicated Pdia3[�KDEL] and Pdia3[-KDEL] images. Green: Pdia3; Blue: nucleus; Red: lipid raft. *: P � .05Pdia3[�KDEL] vs. WT.

FIGURE 6. The effect of changing subcellular location of Pdia3 on rapid responses to1�,25(OH)2D3. Cells were treated with vehicle or 10-7M 1�,25(OH)2D3. PKC activity wasmeasured at 15 min (A). PGE2 in conditioned media was measured at 30 min (B). PKC wasnormalized to total protein and PGE2 was normalized to DNA. Then the 1�,25(OH)2D3 treatedgroup was normalized to the vehicle treated group. The dashed line with value of 1 representsvehicle treated group. *: P � .05 vehicle vs. treatment; #: P � .05 Pdia3[�KDEL] vs. WT.

8 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

Discussion

This study shows that the chaperone scaffolding propertyof Pdia3 plays an important role in mediating the rapideffects of 1�,25(OH)2D3 on two signal transductionpathways, PKC and PLA2, in MC3T3-E1 preosteoblasts.Site directed mutations of three amino acids involved inPdia3’s chaperone function impacted the presence ofdownstream mediators associated with the plasma mem-brane, the activity of PKC in response to 1�,25(OH)2D3

and the release of PGE2 into the conditioned medium.The subcellular location of Pdia3 was influenced by thepresence of the ER retention signal KDEL. Interestingly,baseline PKC activity was sensitive to the retention signal,but the net increase in enzyme activity due to1�,25(OH)2D3 was not impacted by this variable. In con-trast, PLA2 activity was unaffected by the retention ofPdia3 in the ER or its elevated levels in the cytoplasm inthe absence of KDEL, but the stimulatory effect of theseco-steroid was increased in cells overexpressing Pdia3with or without ER retention signal. Finally, our resultsshow that the association of Pdia3 with the plasma mem-brane does not require palmitoylation, nor does the rapidactivation of PKC by 1�,25(OH)2D3.

Our results support the hypothe-sis that Pdia3 serves a scaffoldingfunction in assembling proteins in-volved in the mechanism of1�,25(OH)2D3 action at the plasmamembrane and that the same sitesthat are required for its role as achaperone in the ER are required forits role in mediating 1�,25(OH)2D3

signaling. Overexpression of nativePdia3 increased the presence of cal-reticulin in the plasma membranecompared to wild type MC3T3-E1cells, as was noted by others study-ing the interaction of Pdia3 and cal-reticulin in cancer cells (54). Calre-ticulin was also increased in theplasma membranes from cells over-expressing all three mutant Pdia3transcripts, but there was no changein its ratio to Pdia3 regardless of themutation. This indicates that theamino acids required for the interac-tion mechanism between Pdia3 andcalreticulin in the ER are not impor-tant to their plasma membrane asso-ciation. Calreticulin can mediatecalcium signaling and interact withPKC (55, 56), both of which are also

important in rapid responses to 1�,25(OH)2D3 (33, 57).Hruska and colleagues (37) reported that calreticulin wasinvolved in intracellular transport of 1�,25(OH)2D3-VDR in ROS 17/2.8 osteoblastic cells, but whether this isthe case in the MC3T3-E1 cells was not addressed in thepresent study.

The ratios of Pdia3 to caveolin-1, PLAA and c-Src var-ied among the cell lines suggesting that specific sites wereresponsible for assembling each protein. We previouslyreported that Pdia3 forms a complex with PLAA andcaveolin-1 (29) and c-Src also interacts with caveolin-1 inrapid response to 1�,25(OH)2D3 (58). In the presentstudy we found that by changing single amino acids inPdia3 we could alter the complex formation. There wasmore caveolin-1, PLAA and c-Src per Pdia3 in R282Acells compared to Pdia3ovr, K214A or C406S cells. As amultifunctional protein, Pdia3 was also reported to inter-act with nuclear factor kappa B in NB4 promyelocyticleukemia cells and STAT3 in HepG2 cells in response to1�,25(OH)2D3/phorbol ester cotreatment or IL-6 re-spectively (43, 59). Which domain of Pdia3 is involved inthese interactions is not known. It will be interesting toinvestigate whether R282, K214 and C406 are involved

FIGURE 7. The effect of tunicamycin and HMA on the membrane association of Pdia3 andrapid responses to 1�,25(OH)2D3. Wild type MC3T3–E1 cells were treated with vehicle (DMSO)or 1 �g/ml tunicamycin (A, B) or 1 mM HMA (C, D) for 48 h. For A and C, plasma membraneswere isolated and western blots of pan-cadherin, caveolin-1 and Pdia3 were performed. For Band D, after 48 h, media were replaced by media containing vehicle or 10-7M 1�,25(OH)2D3.

PKC activity was measured at 15 min and normalized to total protein. * P � .05, 1�,25(OH)2D3

vs. control. # P � .05, HMA vs. control.

doi: 10.1210/me.2012-1277 mend.endojournals.org 9

in a common mechanism mediating the protein-proteininteraction of Pdia3. However, this question is out of thescope of this study.

These data support the hypothesis that changes in thechaperone sites can alter formation of the complex re-quired for 1�,25(OH)2D3 signaling at the plasma mem-brane and suggest that the protein assembly in R282Acells may be more active than in the other cell line exam-ined. Indeed, R282A cells were the only mutant cell linethat responded to 1�,25(OH)2D3 treatment by increasingboth PKC activity and PGE2 release. However, the effectof 1�,25(OH)2D3 in the R282A cells was comparable towild type cells, indicating that the mutation blocked theeffect of overexpression seen in Pdia3Ovr cells but did notimpact the native Pdia3 function. In contrast, K214A andC406S cells had levels of caveolin-1, c-Src and PLAA sim-ilar to the Pdia3Ovr cells. K214A cells exhibited no changein PKC and increased PGE2 to levels comparable to wildtype cells. PKC activity was reduced in C406S cells com-pared to wild type and there was no effect of1�,25(OH)2D3 on PGE2 activity. Thus, even singleamino acid changes to Pdia3 to the same sites that mediateits role as a chaperone protein in the ER, can have pro-found consequences for its scaffolding function at theplasma membrane, particularly with respect to1�,25(OH)2D3 signaling.

K214A and R282A are both located at the b and b� cliffof Pdia3. This region interacts with calreticulin and cal-nexin, and when mutated, the interaction between Pdia3and calreticulin or calnexin is abolished (38). In ourstudy, calreticulin was present in the plasma membrane ofall cell lines whereas calnexin was not. This suggests thatPdia3 recruits only calreticulin to the plasma membrane.However, in the mutant K214A cells, association betweenPdia3 and calreticulin did not support downstream sig-naling even of native Pdia3. This indicates that there is aspecific conformational arrangement that mediates thesubsequent recruitment of other components of the sig-naling complex, and that there may be novel interactionsthrough the same b and b� cliff of Pdia3 to other proteinsthan the well-established interaction with calreticulin,which are important to rapid responses. Moreover, theR282A mutation prevented the stimulatory effect of1�,25(OH)2D3 on PKC and PGE2 seen in Pdia3Ovr cells,indicating that calreticulin may not be important for me-diating the action of the secosteroid on these two signal-ing pathway, but instead the role of calreticulin may be asa chaperone for 1�,25(OH)2D3-VDR after activation ofthe membrane receptor complex (37).

Amino acid C406 is the second cystine in CGHC cat-alytic motif of the thioredoxin-like a� domain of Pdia3and is important for the isomerase activity of Pdia3. Our

mutagenesis data show C406 is also important to rapidresponse to 1�,25(OH)2D3. Pdia3 forms transient disul-fide bonds with all of its N-glycosylated protein sub-strates, many of which are plasma membrane proteins(60), and forms stable disulfide bonds with some of its ERpartners (35). Mutation of C406 in our study resulted inloss of downstream signaling in response to1�,25(OH)2D3 compared to wild type and Pdia3Ovr cellsbut it did not alter the relationship between Pdia3 andcaveolin-1, PLAA or c-Src. This suggests that C406 isresponsible for catalyzing formation of disulfide bondswith other proteins required for PKC activation andPGE2 production.

Our data showed the C-terminal ER retention signal isimportant to the subcellular location of Pdia3. Pdia3[-

KDEL] was present in the cytoplasm as well as the plasmamembrane, indicating transport in the form of membranevesicles. Pdia3[-KDEL] and Pdia3[�KDEL] were both presentin the nucleus as well. Some of our previous work hassuggested the inefficient N-terminal signal peptide ofPdia3 will cause a portion of Pdia3 to escape the ER intothe cytosol and later translocate into the nucleus via thenuclear location sequence (44), which could explain ourobservation. The significance of nuclear Pdia3 is not yetknown but PDI family members have been speculatedthat it can alter the redox state of transcription factorssuch as NF�B and AP1 and alter their binding with DNA(61, 62).

Our results show that palmitoylation is not requiredfor plasma membrane localization of Pdia3 nor for itsability to mediate rapid actions of 1�,25(OH)2D3. Theglycosylation and palmitoylation inhibitor tunicamycinreduced the membrane association of cadherin and caveo-lin-1, but did not affect Pdia3. Cadherin is N-glycosylated(63) and caveolin-1 is palmitoylated (64), explaining theirreduced association with the membrane. Caveolin-1 isimportant to the function of Pdia3 and 75% of the plasmamembrane protein was still present after tunicamycintreatment.

Myristate is another lipid that can increase plasmamembrane affinity and sequence analysis shows thatPdia3 possesses myristoylation sites (65). In contrast toreversible palmitoylation, through which estrogen recep-tors increase membrane affinity upon estradiol treatment(48), myristoylation is irreversible. Our data showed thatthe myristoylation inhibitor-HMA did not affect the PMassociation of Pdia3, indicating other mechanisms wereinvolved. Interestingly, HMA was able to block PKC ac-tivation, possibly by inhibiting other mediators in therapid response to 1a,25(OH)2D3 such as G� (66) andc-Src (58), which were previously reported to be myris-toylated (67, 68). A detailed mechanism of how myris-

10 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

toylation may affect these mediators is of interest but notthe focus of this Pdia3 sequence-function centered study.

Surprisingly, we found that the stimulating effect of1�,25(OH)2D3 on PKC activity was comparable amongwild type, Pdia3[�KDEL] and Pdia3[-KDEL] cells. Instead,there was a marked increase in baseline PKC activity inPdia3[-KDEL] cells compared to wild type control cultures.One possibility is that the effect of 1�,25(OH)2D3 wasmaximized in all cultures and did not depend on the in-creased amount of enzyme but on other factors. The factthat our Pdia3[�KDEL] cells were transfected, cultured,and treated in the same way as Pdia3[-KDEL] cells, but hadbaseline PKC activity similar to wild type cells, supportsthe role of plasma membrane associated Pdia3 in mediat-ing PKC activity.

Both Pdia3[�KDEL] and Pdia3[-KDEL] responded to1�,25(OH)2D3 with an increase in PGE2 production. Inchondrocytes, 1�,25(OH)2D3 activates PLA2through aPdia3-dependent pathway resulting in arachidonic acidrelease, which is further processed to PGE2 through cy-clooxygenase-1 (COX-I) (9–11, 13). Cox-1 is found pre-dominantly in endoplasm reticulum and peri-nuclear re-gion (69) and cytosolic PLA2 translocates to the peri-nuclear region after activation (70). OverexpressedPdia3[�KDEL] and Pdia3[-KDEL] were present in this regionas well, providing spatial proximity to c-PLA2 andCOX-1, the data may suggest a role for ER-associatedPdia3 in mediating the production of PGE2. Antibodies toPdia3 block the stimulatory effect of 1�,25(OH)2D3 onPKC supporting its plasma membrane location (23, 24),but it is not known if anti-Pdia3 antibody could block1�,25(OH)2D3 -stimulated PGE2 release. We measuredPGE2 production at 30 mins, whereas PKC activationoccurs as quickly as 3 min (71) and arachidonic acidrelease as early as 15 s in chondrocytes (72). It is possiblethere is a secondary role of Pdia3 in mediating the PGE2pathway in the peri-nuclear region downstream of its ini-tial action on PLA2 and PKC. Alternatively, Pdia3 maymediate effects in PGE2 production in the ER itself. Anovel estrogen receptor GPR30 was found to be present inER where it binds estradiol and initiates rapid responses(73).

In conclusion, this study shows that amino acids K214and R282 in the calreticulin interaction site and C406 incatalytic site are important to Pdia3 dependent rapid re-sponses to 1�,25(OH)2D3. Moreover, K214, R282 andC406 are important for the membrane association ofcaveolin-1, c-Src and PLAA. Deletion of the ER retentionsignal can increase plasma membrane associated Pdia3but its effects on rapid responses are complex implyingadditional levels of regulation occur. Palmitoylation isnot required for the plasma membrane association of

Pdia3 and rapid increases in PKC and PLA2 signaling by1�,25(OH)2D3.

Acknowledgments

We would like thank Po-Yi Ho for her help and advice.

Received August 21, 2012. Accepted May 1, 2013.Address all correspondence and requests for reprints to: Bar-

bara D. Boyan, Ph.D., Department of Biomedical Engineering,VA Commonwealth University, 601 West Main Street, Rich-mond, VA 23284, Phone: 804-828-0190, Email:[email protected]

Disclosure Summary: The authors have nothing to disclose.This work was supported by the Price Gilbert, Jr. Foundation

and Children’s Healthcare of Atlanta.

References

1. Chen J, Olivares-Navarrete R, Wang Y, Herman TR, Boyan BD,Schwartz Z. Protein-disulfide isomerase-associated 3 (Pdia3) medi-ates the membrane response to 1,25-dihydroxyvitamin D3 in osteo-blasts. J Biol Chem. 2010;285:37041–37050.

2. Schwartz Z, Schlader DL, Ramirez V, Kennedy MB, Boyan BD.Effects of vitamin D metabolites on collagen production and cellproliferation of growth zone and resting zone cartilage cells in vitro.J Bone Miner Res. 1989;4:199–207.

3. Tishkoff DX, Nibbelink KA, Holmberg KH, Dandu L, SimpsonRU. Functional vitamin D receptor (VDR) in the t-tubules of car-diac myocytes: VDR knockout cardiomyocyte contractility. Endo-crinology. 2008;149:558–564.

4. Becklund BR, Hansen DW, Jr., Deluca HF. Enhancement of 1,25-dihydroxyvitamin D3-mediated suppression of experimental auto-immune encephalomyelitis by calcitonin. Proc Natl Acad Sci U S A.2009;106:5276–5281.

5. Mathieu C, Waer M, Casteels K, Laureys J, Bouillon R. Preventionof type I diabetes in NOD mice by nonhypercalcemic doses of a newstructural analog of 1,25-dihydroxyvitamin D3, KH1060. Endocri-nology. 1995;136:866–872.

6. Yudoh K, Matsuno H, Kimura T. 1alpha,25-dihydroxyvitamin D3inhibits in vitro invasiveness through the extracellular matrix and invivo pulmonary metastasis of B16 mouse melanoma. The Journal oflaboratory and clinical medicine. 1999;133:120–128.

7. Lee HJ, Paul S, Atalla N, Thomas PE, Lin X, Yang I, Buckley B, LuG, Zheng X, Lou YR, Conney AH, Maehr H, Adorini L, UskokovicM, Suh N. Gemini vitamin D analogues inhibit estrogen receptor-positive and estrogen receptor-negative mammary tumorigenesiswithout hypercalcemic toxicity. Cancer Prev Res (Phila). 2008;1:476–484.

8. Wali RK, Bissonnette M, Khare S, Hart J, Sitrin MD, Brasitus TA.1 alpha,25-Dihydroxy-16-ene-23-yne-26,27-hexafluorocholecal-ciferol, a noncalcemic analogue of 1 alpha,25-dihydroxyvitaminD3, inhibits azoxymethane-induced colonic tumorigenesis. CancerRes. 1995;55:3050–3054.

9. Boyan BD, Sylvia VL, Dean DD, Pedrozo H, Del Toro F, Nemere I,Posner GH, Schwartz Z. 1,25-(OH)2D3 modulates growth platechondrocytes via membrane receptor-mediated protein kinase C bya mechanism that involves changes in phospholipid metabolism andthe action of arachidonic acid and PGE2. Steroids. 1999;64:129–136.

10. Schwartz Z, Gilley RM, Sylvia VL, Dean DD, Boyan BD. Prosta-glandins mediate the effects of 1,25-(OH)2D3 and 24,25-(OH)2D3

doi: 10.1210/me.2012-1277 mend.endojournals.org 11

on growth plate chondrocytes in a metabolite-specific and cell mat-uration-dependent manner. Bone. 1999;24:475–484.

11. Schwartz Z, Graham EJ, Wang L, Lossdorfer S, Gay I, Johnson-PaisTL, Carnes DL, Sylvia VL, Boyan BD. Phospholipase A2 activatingprotein (PLAA) is required for 1alpha,25(OH)2D3 signaling ingrowth plate chondrocytes. J Cell Physiol. 2005;203:54–70.

12. Doroudi M, Schwartz Z, Boyan BD. Phospholipase A(2) activatingprotein is required for 1alpha,25-dihydroxyvitamin D(3) depen-dent rapid activation of protein kinase C via Pdia3. J SteroidBiochem Mol Biol. 2012;132:48–56.

13. Boyan BD, Wang L, Wong KL, Jo H, Schwartz Z. Plasma mem-brane requirements for 1alpha,25(OH)2D3 dependent PKC signal-ing in chondrocytes and osteoblasts. Steroids. 2006;71:286–290.

14. Schwartz Z, Sylvia VL, Luna MH, DeVeau P, Whetstone R, DeanDD, Boyan BD. The effect of 24R,25-(OH)(2)D(3) on protein ki-nase C activity in chondrocytes is mediated by phospholipase Dwhereas the effect of 1alpha,25-(OH)(2)D(3) is mediated by phos-pholipase C. Steroids. 2001;66:683–694.

15. Schwartz Z, Ehland H, Sylvia VL, Larsson D, Hardin RR, BinghamV, Lopez D, Dean DD, Boyan BD. 1alpha,25-dihydroxyvitaminD(3) and 24R,25-dihydroxyvitamin D(3) modulate growth platechondrocyte physiology via protein kinase C-dependent phosphor-ylation of extracellular signal-regulated kinase 1/2 mitogen-acti-vated protein kinase. Endocrinology. 2002;143:2775–2786.

16. Buitrago C, Vazquez G, De Boland AR, Boland RL. Activation ofSrc kinase in skeletal muscle cells by 1, 1,25-(OH(2))-vitamin D(3)correlates with tyrosine phosphorylation of the vitamin D receptor(VDR) and VDR-Src interaction. J Cell Biochem. 2000;79:274–281.

17. Boland R, De Boland AR, Buitrago C, Morelli S, Santillan G,Vazquez G, Capiati D, Baldi C. Non-genomic stimulation of ty-rosine phosphorylation cascades by 1,25(OH)(2)D(3) by VDR-de-pendent and -independent mechanisms in muscle cells. Steroids.2002;67:477–482.

18. Buitrago C, Vazquez G, De Boland AR, Boland R. The vitamin Dreceptor mediates rapid changes in muscle protein tyrosine phos-phorylation induced by 1,25(OH)(2)D(3). Biochem Biophys ResCommun. 2001;289:1150–1156.

19. Nemere I, Garbi N, Hammerling GJ, Khanal RC. Intestinal cellcalcium uptake and the targeted knockout of the 1,25D3-MARRS(membrane-associated, rapid response steroid-binding) receptor/PDIA3/Erp57. J Biol Chem. 2010;285:31859–31866.

20. Van Cromphaut SJ, Dewerchin M, Hoenderop JG, Stockmans I,Van Herck E, Kato S, Bindels RJ, Collen D, Carmeliet P, BouillonR, Carmeliet G. Duodenal calcium absorption in vitamin D recep-tor-knockout mice: functional and molecular aspects. Proc NatlAcad Sci U S A. 2001;98:13324–13329.

21. Nemere I, Dormanen MC, Hammond MW, Okamura WH, Nor-man AW. Identification of a specific binding protein for 1 alpha,25-dihydroxyvitamin D3 in basal-lateral membranes of chick intestinalepithelium and relationship to transcaltachia. J Biol Chem. 1994;269:23750–23756.

22. Nemere I, Farach-Carson MC, Rohe B, Sterling TM, Norman AW,Boyan BD, Safford SE. Ribozyme knockdown functionally links a1,25(OH)2D3 membrane binding protein (1,25D3-MARRS) andphosphate uptake in intestinal cells. Proc Natl Acad Sci U S A.2004;101:7392–7397.

23. Nemere I, Schwartz Z, Pedrozo H, Sylvia VL, Dean DD, Boyan BD.Identification of a membrane receptor for 1,25-dihydroxyvitaminD3 which mediates rapid activation of protein kinase C. J BoneMiner Res. 1998;13:1353–1359.

24. Boyan BD, Bonewald LF, Sylvia VL, Nemere I, Larsson D, NormanAW, Rosser J, Dean DD, Schwartz Z. Evidence for distinct mem-brane receptors for 1 alpha,25-(OH)(2)D(3) and 24R,25-(OH)(2)D(3) in osteoblasts. Steroids. 2002;67:235–246.

25. Pedrozo HA, Schwartz Z, Rimes S, Sylvia VL, Nemere I, PosnerGH, Dean DD, Boyan BD. Physiological importance of the

1,25(OH)2D3 membrane receptor and evidence for a membranereceptor specific for 24,25(OH)2D3. J Bone Miner Res. 1999;14:856–867.

26. Schwartz Z, Shaked D, Hardin RR, Gruwell S, Dean DD, Sylvia VL,Boyan BD. 1alpha,25(OH)2D3 causes a rapid increase in phos-phatidylinositol-specific PLC-beta activity via phospholipase A2-dependent production of lysophospholipid. Steroids. 2003;68:423–437.

27. Boyan BD, Sylvia VL, McKinney N, Schwartz Z. Membrane ac-tions of vitamin D metabolites 1alpha,25(OH)2D3 and24R,25(OH)2D3 are retained in growth plate cartilage cells fromvitamin D receptor knockout mice. J Cell Biochem. 2003;90:1207–1223.

28. Olivares-Navarrete R, Sutha K, Hyzy SL, Hutton DL, Schwartz Z,McDevitt T, Boyan BD. Osteogenic Differentiation of Stem CellsAlters Vitamin D Receptor Expression. Stem Cells Dev. 2012.

29. Doroudi M, Schwartz Z, Boyan BD. Phospholipase A(2) activatingprotein is required for 1alpha,25-dihydroxyvitamin D(3) depen-dent rapid activation of protein kinase C via Pdia3. J SteroidBiochem Mol Biol. 2012.

30. Boyan BD, Wong KL, Wang L, Yao H, Guldberg RE, Drab M, JoH, Schwartz Z. Regulation of growth plate chondrocytes by 1,25-dihydroxyvitamin D3 requires caveolae and caveolin-1. J BoneMiner Res. 2006;21:1637–1647.

31. Garbi N, Tanaka S, Momburg F, Hammerling GJ. Impaired assem-bly of the major histocompatibility complex class I peptide-loadingcomplex in mice deficient in the oxidoreductase ERp57. Nat Im-munol. 2006;7:93–102.

32. Costa JL, Eijk PP, van de Wiel MA, ten Berge D, Schmitt F, NarvaezCJ, Welsh J, Ylstra B. Anti-proliferative action of vitamin D inMCF7 is still active after siRNA-VDR knock-down. BMC genom-ics. 2009;10:499.

33. Sylvia VL, Schwartz Z, Ellis EB, Helm SH, Gomez R, Dean DD,Boyan BD. Nongenomic regulation of protein kinase C isoforms bythe vitamin D metabolites 1 alpha,25-(OH)2D3 and 24R,25-(OH)2D3. J Cell Physiol. 1996;167:380–393.

34. Lindquist JA, Jensen ON, Mann M, Hammerling GJ. ER-60, achaperone with thiol-dependent reductase activity involved inMHC class I assembly. EMBO J. 1998;17:2186–2195.

35. Peaper DR, Wearsch PA, Cresswell P. Tapasin and ERp57 form astable disulfide-linked dimer within the MHC class I peptide-load-ing complex. EMBO J. 2005;24:3613–3623.

36. Peaper DR, Cresswell P. The redox activity of ERp57 is not essentialfor its functions in MHC class I peptide loading. Proc Natl Acad SciU S A. 2008;105:10477–10482.

37. Kim YS, MacDonald PN, Dedhar S, Hruska KA. Association of 1alpha,25-dihydroxyvitamin D3-occupied vitamin D receptors withcellular membrane acceptance sites. Endocrinology. 1996;137:3649–3658.

38. Kozlov G, Maattanen P, Schrag JD, Pollock S, Cygler M, Nagar B,Thomas DY, Gehring K. Crystal structure of the bb’ domains of theprotein disulfide isomerase ERp57. Structure. 2006;14:1331–1339.

39. Beynon-Jones SM, Antoniou AN, Powis SJ. Mutational analysis ofthe oxidoreductase ERp57 reveals the importance of the two centralresidues in the redox motif. FEBS letters. 2006;580:1897–1902.

40. Hirano N, Shibasaki F, Sakai R, Tanaka T, Nishida J, Yazaki Y,Takenawa T, Hirai H. Molecular cloning of the human glucose-regulated protein ERp57/GRP58, a thiol-dependent reductase.Identification of its secretory form and inducible expression by theoncogenic transformation. European journal of biochemistry /FEBS. 1995;234:336–342.

41. Frickel EM, Riek R, Jelesarov I, Helenius A, Wuthrich K, EllgaardL. TROSY-NMR reveals interaction between ERp57 and the tip ofthe calreticulin P-domain. Proc Natl Acad Sci U S A. 2002;99:1954–1959.

42. Leach MR, Cohen-Doyle MF, Thomas DY, Williams DB. Local-

12 Pdia3 as receptor for 1�,25(OH)2D3 Mol Endocrinol

ization of the lectin, ERp57 binding, and polypeptide binding sitesof calnexin and calreticulin. J Biol Chem. 2002;277:29686–29697.

43. Wu W, Beilhartz G, Roy Y, Richard CL, Curtin M, Brown L,Cadieux D, Coppolino M, Farach-Carson MC, Nemere I, MecklingKA. Nuclear translocation of the 1,25D3-MARRS (membrane as-sociated rapid response to steroids) receptor protein and NFkappaBin differentiating NB4 leukemia cells. Exp Cell Res. 2010;316:1101–1108.

44. Grindel BJ, Rohe B, Safford SE, Bennett JJ, Farach-Carson MC.Tumor necrosis factor alpha treatment of HepG2 cells mobilizes acytoplasmic pool of ERp57/1,25D(3) -MARRS to the nucleus.J Cell Biochem. 2011.

45. Schwartz Z, Schlader DL, Swain LD, Boyan BD. Direct effects of1,25-dihydroxyvitamin D3 and 24,25-dihydroxyvitamin D3 ongrowth zone and resting zone chondrocyte membrane alkalinephosphatase and phospholipase-A2 specific activities. Endocrinol-ogy. 1988;123:2878–2884.

46. Acconcia F, Ascenzi P, Bocedi A, Spisni E, Tomasi V, TrentalanceA, Visca P, Marino M. Palmitoylation-dependent estrogen receptoralpha membrane localization: regulation by 17beta-estradiol. Mo-lecular biology of the cell. 2005;16:231–237.

47. Marino M, Ascenzi P. Steroid hormone rapid signaling: the pivotalrole of S-palmitoylation. IUBMB Life. 2006;58:716–719.

48. Li L, Haynes MP, Bender JR. Plasma membrane localization andfunction of the estrogen receptor alpha variant (ER46) in humanendothelial cells. Proc Natl Acad Sci U S A. 2003;100:4807–4812.

49. Franceschi RT, Iyer BS, Cui Y. Effects of ascorbic acid on collagenmatrix formation and osteoblast differentiation in murineMC3T3–E1 cells. J Bone Miner Res. 1994;9:843–854.

50. Huhtakangas JA, Olivera CJ, Bishop JE, Zanello LP, Norman AW.The vitamin D receptor is present in caveolae-enriched plasmamembranes and binds 1 alpha,25(OH)2-vitamin D3 in vivo and invitro. Mol Endocrinol. 2004;18:2660–2671.

51. Perret B, Chap HJ, Douste-Blazy L. Asymmetric distribution ofarachidonic acid in the plasma membrane of human platelets. Adetermination using purified phospholipases and a rapid methodfor membrane isolation. Biochimica et biophysica acta. 1979;556:434–446.

52. Fan G, Goldsmith PK, Collins R, Dunn CK, Krapcho KJ, RogersKV, Spiegel AM. N-linked glycosylation of the human Ca2� recep-tor is essential for its expression at the cell surface. Endocrinology.1997;138:1916–1922.

53. Vaandrager AB, Ehlert EM, Jarchau T, Lohmann SM, de JongeHR. N-terminal myristoylation is required for membrane localiza-tion of cGMP-dependent protein kinase type II. J Biol Chem. 1996;271:7025–7029.

54. Obeid M. ERP57 membrane translocation dictates the immunoge-nicity of tumor cell death by controlling the membrane transloca-tion of calreticulin. J Immunol. 2008;181:2533–2543.

55. Coppolino MG, Woodside MJ, Demaurex N, Grinstein S, St-Ar-naud R, Dedhar S. Calreticulin is essential for integrin-mediatedcalcium signalling and cell adhesion. Nature. 1997;386:843–847.

56. Rendon-Huerta E, Mendoza-Hernandez G, Robles-Flores M.Characterization of calreticulin as a protein interacting with proteinkinase C. The Biochemical journal.1999; 344 Pt. 2:469–475.

57. Liu R, Li W, Karin NJ, Bergh JJ, Adler-Storthz K, Farach-CarsonMC. Ribozyme ablation demonstrates that the cardiac subtype ofthe voltage-sensitive calcium channel is the molecular transducer of1, 25-dihydroxyvitamin D(3)-stimulated calcium influx in osteo-blastic cells. J Biol Chem. 2000;275:8711–8718.

58. Buitrago C, Boland R. Caveolae and caveolin-1 are implicated in1alpha,25(OH)2-vitamin D3-dependent modulation of Src, MAPKcascades and VDR localization in skeletal muscle cells. J SteroidBiochem Mol Biol. 2010;121:169–175.

59. Eufemi M, Coppari S, Altieri F, Grillo C, Ferraro A, Turano C.ERp57 is present in STAT3-DNA complexes. Biochem Biophys ResCommun. 2004;323:1306–1312.

60. Elliott JG, Oliver JD, High S. The thiol-dependent reductase ERp57interacts specifically with N-glycosylated integral membrane pro-teins. J Biol Chem. 1997;272:13849–13855.

61. Clive DR, Greene JJ. Cooperation of protein disulfide isomeraseand redox environment in the regulation of NF-kappaB and AP1binding to DNA. Cell biochemistry and function. 1996;14:49–55.

62. Grillo C, D’Ambrosio C, Scaloni A, Maceroni M, Merluzzi S,Turano C, Altieri F. Cooperative activity of Ref-1/APE and ERp57in reductive activation of transcription factors. Free radical biology,medicine. 2006;41:1113–1123.

63. Langer MD, Guo H, Shashikanth N, Pierce JM, Leckband DE.N-Glycosylation Alters Cadherin-Mediated Intercellular BindingKinetics. Journal of cell science. 2012.

64. Uittenbogaard A, Smart EJ. Palmitoylation of caveolin-1 is requiredfor cholesterol binding, chaperone complex formation, and rapidtransport of cholesterol to caveolae. J Biol Chem. 2000;275:25595–25599.

65. Khanal RC, Nemere I. The ERp57/GRp58/1,25D3-MARRS recep-tor: multiple functional roles in diverse cell systems. Curr MedChem. 2007;14:1087–1093.

66. Schwartz Z, Sylvia VL, Larsson D, Nemere I, Casasola D, Dean DD,Boyan BD. 1alpha,25(OH)2D3 regulates chondrocyte matrix ves-icle protein kinase C (PKC) directly via G-protein-dependent mech-anisms and indirectly via incorporation of PKC during matrix ves-icle biogenesis. J Biol Chem. 2002;277:11828–11837.

67. Mumby SM, Heukeroth RO, Gordon JI, Gilman AG. G-proteinalpha-subunit expression, myristoylation, and membrane associa-tion in COS cells. Proc Natl Acad Sci U S A. 1990;87:728–732.

68. David-Pfeuty T, Bagrodia S, Shalloway D. Differential localizationpatterns of myristoylated and nonmyristoylated c-Src proteins ininterphase and mitotic c-Src overexpresser cells. Journal of cell sci-ence. 1993;105 (Pt 3):613–628.

69. Rollins TE, Smith WL. Subcellular localization of prostaglandin-forming cyclooxygenase in Swiss mouse 3T3 fibroblasts by electronmicroscopic immunocytochemistry. J Biol Chem. 1980;255:4872–4875.

70. Muthalif MM, Benter IF, Uddin MR, Malik KU. Calcium/calmod-ulin-dependent protein kinase IIalpha mediates activation of mito-gen-activated protein kinase and cytosolic phospholipase A2 in nor-epinephrine-induced arachidonic acid release in rabbit aorticsmooth muscle cells. J Biol Chem. 1996;271:30149–30157.

71. Sylvia VL, Schwartz Z, Schuman L, Morgan RT, Mackey S, GomezR, Boyan BD. Maturation-dependent regulation of protein kinaseC activity by vitamin D3 metabolites in chondrocyte cultures. J CellPhysiol. 1993;157:271–278.

72. Swain LD, Schwartz Z, Boyan BD. 1,25-(OH)2D3 and 24,25-(OH)2D3 regulation of arachidonic acid turnover in chondrocytecultures is cell maturation-specific and may involve direct effects onphospholipase A2. Biochimica et biophysica acta. 1992;1136:45–51.

73. Revankar CM, Cimino DF, Sklar LA, Arterburn JB, Prossnitz ER.A transmembrane intracellular estrogen receptor mediates rapidcell signaling. Science. 2005;307:1625–1630.

doi: 10.1210/me.2012-1277 mend.endojournals.org 13