2 cell lines

ELSEVIER

BB. Biochi~mic~a et Biophysica A~ta Biochimica et Biophysica Acta 1297 (1996) 57-68

HSP binding and mitochondrial localization of p53 protein in human HT1080 and mouse C3H10T1/2 cell lines B. Alex Me,rrick a,*, Chaoying He a Lora L. Witcher a Rachel M. Patterson a JoAnne J. Reid b, p. Miki Pence-Pawlowski a, James K. Selkirk a a Laboratory of Molecular Carcinogenesis, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709, USA Laboratory of Experimental Pathology, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709, USA

Received 29 April 1996; accepted 20 May 1996

Abstract In normal cells, the tumor suppressor actions of p53 protein are mediated by specific DNA binding and protein-protein interactions within the nucleus. Mutant p53 proteins, however, often assume an aberrant conformation devoid of tumor suppressor activity and newly capable of binding to the cognate or inducible HSP70. Recent reports from our laboratory and others show that additional unknown proteins may also complex with mutant p53. In this study, we characterize p53:HSP complexes and their subcellular location in the transformed cell lines, human HT1080 and murine C3HI0T1/2, which both contain aberrant p53 conformers. Immunoprecipitation and SDS-PAGE of p53 from whole cell lysates revealed the additional presence of a broad 70 kDa band and a 90 kDa band in both lines, while p53 isolated from nuclear lysates was free from other proteins. 2D-PAGE was used to isolate and identify HSP members from cytoplasmic and nuclear lysates by immunoprecipitation, Western blotting and protein sequencing. Anti-p53 immune complexes from cytoplasmic lysates contained not only HSC70 but also GRP75, GRP78 and a weakly basic 90 kDa protein, which may be related to HSP90. The inducible form of HSP70 was not complexed to p53 protein, even though expressed in these cells. Analysis of anti-HSP70, anti-GRP75 and anti-HSP90 immune complexes suggests that HSP members exist as preformed complexes in the cytoplasm, but not the nucleus. The presence of the mitochondrial and endoplasmic reticular chaperones, GRP75 and GRP78, in p53:HSP complexes suggested that p53 might be found in these cytoplasmic organelles which was confirmed in mitochondria by biochemical and immunoelectron microscopic evidence. These studies suggest that newly identified members of p53:HSP complexes represent components of a chaperone program which affects the subcellular distribution of p53 protein in these transformed lines. Keywords: Chaperone; p53 protein; Heat shock protein; Binding; Mitochondrion; (Human); (Mouse)

1. Introduction Chaperones comprise a phylogenetically conserved class of proteins which assist polypeptides in folding, assembly, translocation, repair and degradation [1,2]. Three general families of chaperones are well known for their stressor responses particularly to 'heat shock' and accordingly

Abbreviations: HSP, heat shock protein; 2D, two-dimensional; Hepes, 4-(2-hydroxyethyl)-l-piperazineelhanesulfonicacid; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; PBS, phosphatebuffered saline; TBS, Tris-buffered saline. * Corresponding author. Fax: + 1 (919) 5414704; e-mail: [email protected].

these proteins have been grouped by their molecular masses as HSP70, HSP90 and HSP60. In mammalian cells, HSP70 members include the cognate form HSC70 (73 kDa) and heat-inducible HSP70 (72 kDa) localized in cytosol along with organelle-resident members, GRP78 in the ER lumen and GRP75 in the mitochondrial matrix [ 1-4]. GRP75 and GRP78 are induced by non-heat shock stressors such as accumulation of unfolded or misfolded proteins, disruption in calcium homeostasis and glucose deprivation [3]. HSP90 family members show a similar subcellular distribution and involve GRP94 in the ER and the two gene products for HSP90 in cytosol, termed HSP89-o~ and HSP89-[3 in humans or Hsp84 and Hsp86 in mouse [5-7]. The third chaperone family, mitochondrial HSP60, and its cytosolic counterpart, TCP-I, assemble into barrel-like structures

0167-4838/96/$15.00 Copyright: © 1996 Elsevier Science B.V. All rights reserved. PII S0167-4838(96)00089- 1

58

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

with low molecular mass cohort proteins to assist proper polypeptide folding [8]. A special type of stress requiring a cellular response is the accumulation of aberrant or malformed polypeptides. One such recognition system has been described as the 'unfolded protein response' (UPR) signaling pathway, triggered by various stimuli including glucose starvation, inhibitors of glycosylation or disulfide bonding, intracellular calcium depletion or overexpression of mutant secretory proteins [9]. The binding of HSPs to the tumor suppressor protein, p53, may play a role in malignant transformation. Inactivation of the p53 gene is the most common genetic abnormality in human cancer [10]. The p53 gene often undergoes point mutation, leading to abnormal conformations which fail to perform wild-type protein functions as a negative regulator in the cell cycle [10]. The topographical changes in mutant proteins are evidenced by changes that result in exposure of cryptic epitopes and loss of normal epitopes [11]. When wild-type and mutant p53 proteins coexist in the cell, they can form cytoplasmic oligomers and assume an aberrant conformation capable of binding HSP70 that sequester the growth inhibitory wild-type protein from the nucleus [12]. A similar sequestering phenomenon has been proposed for cognate or inducible forms of HSP70 that bind to mutant p53 protein in transformed cells [13,14]. High transforming capacity and avid HSP70 binding of specific p53 mutants in in vitro assays suggest a role for HSP70 in transformation [15]. The exposure of specific HSP70 binding sites in the C-terminus and N-terminus occur in aberrant p53 protein conformations but not wild-type conformation [ 16,17]. HSP70 binding prolongs the half-life of mutant p53 proteins which contributes to accumulation of mutant p53 protein and dense immunocytochemical staining compared to normal tissue [11,12]. In some cases, HSP70 binding may not effectively sequester p53 to cytoplasm, since similar proportions of HSP70-p53 complexes have been found in both cytoplasm and nucleus [18]. An alternate explanation for HSP70-p53 binding might be as a complex cellular response to improperly folded protein. Several studies indicate that the number of HSP's or other proteins bound to p53 may be more numerous than the HSC70-HSP70 complexes previously reported [19,20] and their identification might provide insight into sequestration phenomenon. We reasoned that the identity of additional components in p53-HSP70 complexes might be heat shock proteins, since different chaperone families often engage in cooperative binding [21,22]. In this study we hypothesized that members of the HSP70 and HSP90 families and perhaps other proteins might be involved in binding to aberrant p53 from transformed cells. Unlike our initial study [20], we sought to identify individual HSP70 members and to determine if soluble and organelle-resident HSP70 members were complexed to p53 and, finally, to localize these complexes to subcellular structures. The possible relationship of HSP

complexation to p53 as a cellular stress response to abnormal protein is discussed.

2. Materials and methods

2.1. Cell culture and radiolabeling Human HT1080 and murine C3H10T1/2 cell lines were grown in 100 mm dishes in DMEM and BME media, respectively. HT1080 fibrosarcoma cells (6TG c5 line) contain two mutant p53 alleles (codons 245 and 277) [23]. A murine fibroblast line (XR-III) created by X-ray irradiation of C3H10T1/2 cells was isolated from transformed Type III foci and expresses immunologically mutant p53 protein [24,25]. Complete medium was created by adding 10% FCS, 2 mM glutamine and 5 p,g/ml gentamicin. Cells at 80-90% confluence were placed in methionine-free medium for 1 h, followed by replacement with fresh complete methionine-free medium containing 35S-Met ( > 9000 Ci/mmol, ICN, Irvine, CA) at 0.5 mCi in 4 ml per plate (0.125 mCi/ml) for either 3 or 6 h. Plates were gassed with a 95%:5% O2:CO 2 mixture in a sealed plastic enclosure and incubated at 37°C. 2.2. lmmunoprecipitation Radiolabeled cells were prepared for immunoprecipitation from whole cell, nuclear or cytoplasmic lysates as described previously [20,26] using isotonic lysis buffer (25 mM Tris, pH 8, 150 mM NaC1, 0.5% NP-40, 0.5% sodium deoxycholate, 10 Ixg/ml aprotinin, 5 p~g/ml leupeptin, 1 mM EDTA, pH 8, 1 mM PMSF), low-salt lysis buffer (20 mM Hepes, 5 mM NaC1, 5 mM MgC12, 0.5% NP40, 0.1% sodium deoxycholate plus inhibitors) and wash buffer (isotonic lysis buffer with 0.1% SDS). In some experiments, cells were radiolabeled and mitochondria were isolated by sucrose gradient from the 10000 × g fraction of cell homogenates as previously described [27]. Mitochondria were dissolved in isotonic lysis buffer prior to immunoprecipitation. Immunoprotein complexes were formed with 6.5 ~g of primary antibody per ml of lysate overnight at 4°C with gentle rotation. Antibodies recognizing native proteins included p53 protein (PAb421, Oncogene Science, Uniondale, NY), HSP70 (Clone W27, Oncogene Science), HSP90 (Clone 3G3, Affinity BioReagents, Neshanic Station, NJ) and GRP75 (carboxy-peptide rabbit antisera; Section 2.7). Immunoprotein complexes were immunoprecipitated with either goat anti-mouse IgG, anti-mouse IgM or anti-rabbit IgG attached to r-Protein G agarose (Oncogene Science). Negative control primary antibodies were from the appropriate antibody isotypes from myeloma cells. Non-specifically bound proteins were removed by repeated washing with isotonic lysis buffer. Isolated radiolabeled proteins were dissolved in electrophoresis buffer and stored at - 8 0 ° C prior to analysis.

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68 2.3. Subcellular fractionation f o r Western blotting

Cytoplasmic and nuclear proteins were fractionated for Western blot analysis as previously described [26,28]. Briefly, cytoplasmic fractions were prepared by 1000 × g centrifugation of Dounce homogenates. The cytoplasmic supernatants were dialyzed against water and lyophilized. Extracellular membranes from 1000 × g pellets were dissolved and nuclei were isolated by centrifugation. Nuclear proteins were prepared hy nuclear shearing, centrifugal removal of DNA fragments, dialysis of the supernatant and lyophilization.

59

our lab [28]. A 1 mg sample of cytoplasmic proteins derived from HT1080 or C3H10T1/2 cells was separated and blotted onto PVDF-P sQ membranes (Millipore, Bedford, MA). Proteins were localized by staining with R-250 Coomassie brilliant blue (CBB). Other blots were immunostained with HSP antibodies or spiked with radiolabeled protein from immunoprecipitions to determine candidate proteins for sequencing. CBB-stained proteins from 2 or 3 blots were excised, destained in methanol and inserted into a membrane-blot cartridge of an ABI Model 473A protein sequencer using pulsed-liquid flow. Automated sequencing was programmed for 2 standard and 23 residue cycles.

2.4. Electrophoresis 2.7. Production o f rabbit antibodies to GRP75

SDS-PAGE of protein,; was performed by the standard Laemmli method on 10% acrylamide resolution gels. Separation of proteins by 2D-PAGE was conducted as reported previously [28]. Briefly, proteins were solubilized in urea lysis buffer (9 M urea, .4% NP-40, 2% ampholytes pH 9-11, 1% DTT and 0.1% SDS), loaded onto 13.5 cm X 1.5 mm tube gels and separated by charge using isoelectric focusing (pH 4 - 8 ampholytes, BDH, Poole, England) for 11,219 volt hours. Extruded tube gels were loaded onto polyacrylamide slab gels cast in a 10% to 16% linear gradient. Radiolabeled proteins were fixed in acid-ethanol solution overnight and detected after 1 week by film fluorography. 2.5. Western blotting

After electrophoretic separation, proteins were electrotransferred to nitrocellulose, and proteins were detected by chemiluminescence. Dilutions of primary antibodies were 1 p~g/ml for anti-p53 (PAb240, Oncogene Science), 1:5000 for anti-HSP70 (Clone BRM-22, Sigma, St. Louis, MO), 1:500 for anti-HSP90 (Clone 3G3, Affinity BioReagents) and 1 txg/ml of anti-GRP75 (carboxy-peptide rabbit antisera; Section 2.7). Excess primary antibody was removed by four TBS-T washes (Tris-buffered saline, pH 7.6, with 0.1% Tween 20). Bound primary antibody was localized by a 1:100000 dilution of appropriate HRP-conjugated secondary antibody (anti-mouse IgG; anti-mouse IgM; anti-rabbit IgG; Boehinger Mannheim). Non-specific binding by the secondary antibody was determined by use of an appropriate non-specif~ic mouse myeloma isotype as a substitute for the primary monoclonal antibody or by use of preimmune rabbit sera IgG to substitute for anti-peptide antibodies. Proteins were visualized by chemiluminescent detection (Amersham, Arlington Heights, IL). 2.6. Protein sequencing

Preparative 2D-PAGE procedures were used to isolate sufficient amounts of heat shock proteins for sequencing from cytoplasmic preparations by procedures developed in

An amino-terminal peptide, ASEAIKGAVVGIDLGTT N S C V A V C - , and a carboxy-terminal peptide, -CNKLKEEISKMRELLARKDSETGE-, were synthesized and corresponded to residues 47-69 (plus a terminal cysteine) and residues 608-627, respectively, of human GRP75. Both peptides corresponded to consensus regions in human and mouse GRP75 in order to produce cross-reactive antibodies. The peptides were dissolved in conjugation buffer (83 mM sodium phosphate buffer, 0.1 M EDTA, 0.9 M NaC1) and solubilizers. The cysteine terminus was conjugated to maleimide-activated KLH (Pierce, Rockford, IL), separated from unreacted peptide by gel filtration, and mixed with an equal volume of Complete Freund's adjuvant prior to injection. New Zealand white female rabbits (2-2.5 kg) were subdermally injected with 1 ml of emulsion containing 0.25 mg of conjugated amino-terminal peptide. Thereafter, rabbits were re-exposed to the same amount of antigen in Incomplete Freund's adjuvant at weeks 4 and 8 and were bled at week 10. Serum from exposed animals was tested for reactivity to the peptide. Pre-exposure serum served as a control. The IgG fraction was isolated from positive-responding rabbit serum or pre-exposure serum with commercial kits (Middlesex, Mansfield, MA) and then diluted to 1 Ixg/txl in pH 7.4 PBS. Carboxy-peptide antisera was used for immunoprecipitation and Western blotting of Grp75 but only the amino-peptide antisera worked well for immunoelectron microscopy. Serum was stored frozen at - 2 0 ° C until use. 2.8. Ultrastructure and immunoelectron microscopy

Normal ultrastructure was visualized by classical electron microscopy. HT1080 cells were cultured in two-welled Nunc chamber slides (Nunc, Naperville, IL) in which all processing through encapsulation was performed. Cultured cells were rinsed in PBS and fixed for 1 h in 1% paraformaldehyde + 1% glutaraldehyde in PBS. Two 5 min rinses in PBS preceded and followed postfixation for 1 h in 1% aqueous osmium tetroxide. Following a brief rinse in distilled water, en bloc staining for 30 min was per-

60

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

formed using 2% aqueous uranyl acetate. Dehydration through a graded series of alcohols preceded standard embedment in gelatin capsules in the resin, LX-112 (Ladd, Burlington, VT). Postembedding immunoelectron microscopy using ice polymerization of LR White (Polysciences, Warrington, PA) was begun by centrifuging HT1080 cells after trypsin removal from plastic culture dishes. Cells were fixed in a PBS suspension for 60 rain in 4% paraformaldehyde with 0.02% glutaraldehyde. After removal of fixative, pellets were dehydrated for 10 min in separate aqueous solutions of 50% and 70% ethanol followed by a 60 min en b l o c staining in 2% uranyl acetate in 70% alcohol. A 10 min dehydration in 75% ethanol followed. Ten min in 2 parts LR White to 1 part 75% ethanol preceded four 20 min changes in pure LR White. Embedment was performed on ice using 10 pA of accelerator per 5 ml of resin. Ultrathin (80 I~m) sections for immunolabeling were cut from LR White blocks and collected on 1 × 2 mm nickel slot grids (Ted Pella, Redding, CA). Grids with the thin sections were then placed sequentially on drops of the following solutions: 5% normal goat serum in TBS (pH 7.6) for 15 min; anti-p53 (rabbit CMI antibody; Vector, NovoCastra, Burlingame, CA) at a 1:50 or 1:100 dilution, or anti-GRP75 (amino-peptide antisera) at a 1:500 dilution in TBS overnight at 25°C; ten rinses in TBS (pH 8.2); goat anti-rabbit IgG antibody conjugated to 10 nm colloidal

gold particles (Amersham, Arlington Heights, IL) at a 1:20 dilution in TBS (pH 8.2), 2 h; ten rinses in TBS (pH 8.2); three rinses in PBS; 2% glutaraldehyde in PBS, 15 min; 10 rinses in distilled water; 2% aqueous uranyl acetate, 15 rain; 10 rinses in distilled water; Reynolds' lead citrate, 3 min; 10 rinses in distilled water. Finally, the grids were placed on Formvar support film (Fullam, NY) cast upon slides with 0.5 cm holes and examined in a Zeiss 10CR transmission electron microscope.

3. Results

Anti-p53 antibody PAb240 recognizes an exposed site present in aberrant conformations of p53 protein during immunoprecipitation reactions [23]. P53 could be immunoprecipitated from nuclear lysates of radiolabeled HT1080 and C3H10T1/2 transformed lines with PAb240 as shown in Fig. 1, panel A, indicating the presence of aberrant conformation, as well as anti-p53 PAb421 antibody which can react with native, aberrant conformation p53 proteins. p53 bands in SDS-PAGE blots were confirmed by immunostaining (panel C) after concentration of p53 by immunoprecipitation from whole cell lysates. Presence of aberrant conformation suggested that p53 might be associated with heat shock proteins [11]. Panel B shows that anti-p53 immunoprecipitates from whole cell lysates con-

Whole Cell Lysate, IP

Nuclear Lysate, IP HT1080

+'°

HT1080

C3H10T1/2

I

C3H10T1/2

_#+ I , .

oa"

Western Blot HT1080

,,~.~a ¢0° ,o+.

~"

iiiii i ==~ iii iiiiiii= 17.4-

~

.=. .!. . . . . .

!iii i

~i Ii i

ii;iii

9IkD

i i11iiili!i!!i

p53~

43-

~

~

m

~

......... ~

7O_kO[ p53->

i

I

p53.->

~"

iii

] ....

IE~'II

i

ii

68-

C3H10T112

........ : -->

~'

i

29-

-

18.4o

Panel A

Panel B

Panel C

Fig. I. Detection of p53 protein and p53-associated proteins in HTI080 and C 3 H I 0 T I / 2 cells. Cells were radiolabeled with [3SS]methionine for 6 h. Nuclear or whole cell lysates were isolated as described in Section 2. Lysates were analyzed for p53 by immunoprecipitation (IP) and were separated on 10% SDS-PAGE gels. Bands at 53 kDa were observed in nuclear (panel A) and whole cell (panel B) lysates in both lines with PAb421 and PAb240, but not in non-specific (N.S.) binding lanes. In panel C nuclear p53 was concentrated by immunoprecipitation with PAb421, separated by SDS-PAGE and was verified as p53 by chemiluminescent detection with PAb240 after Western blotting. In addition to p53 protein, whole cell lysates in panel B show a broad band in the 70 kDa region and a 90 kDa band accompanying p53 immunoprecipitation.

61

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

tained associated proteins which migrated in a wide band in the 70 kDa region, and also at 90 kDa, consistent with the presence of HSP70 and HSP90 class proteins. That these 70 kDa and 90 kDa bands were not observed in nuclear lysates suggested ~ cytoplasmic origin. In order to improve resolution and to identify these bands, further studies were conducted in cytoplasmic and nuclear protein lysates separated by 2D-PAGE. Cytoplasmic lysates we~;e obtained by dissolving whole cells in a low-salt lysis buffer and separating nuclei by centrifugation. Immunoprecipitations of p53 were performed with HT1080 and C3H10T1/2 cytoplasmic and nuclear lysates after radiolabeling for 6 h under the same conditions as SDS-PAGE analysis shown in Fig. 1. Several proteins coprecipitated with p53 in cytoplasmic lysates as shown in Fig. 2, panel A. 2D-PAGE separation of p53 immunoprecipitates shows that the broad SDS-PAGE '70 kDa' band is composed of several acidic proteins which range from 70 to 80 kDa as well as a weakly basic protein at 90 kDa. In the 70 kDa region, three groups of proteins could be distinguished, while the 90 kDa protein consisted of several charged species. Non-specific binding showed

actin and tubulin as principal contaminants, in addition to other minor proteins. Nuclear p53 protein shown in panel B of Fig. 2 was devoid of associated proteins; therefore, we attempted to identify the 70 to 80 kDa proteins and the weakly basic 90 kDa protein associated with p53 in the cytoplasm of HT1080 and C3H10T1/2 cells. HSP proteins were identified by immunostaining 2DPAGE blots of cytoplasmic fractions as shown in Fig. 3. The top two panels show immunostaining of HSP70 members at 73 kDa (HSC70) and 78 kDa (GRP78) and the inducible form (HSP70) at 72 kDa detected by antibody BRM-22. One notable aspect is that HSP70 is constitutively expressed in HT1080 cells but is barely visible in C3H10T1/2 cells. Notably, after thermal stress, only murine C3H10T1/2 cells showed large increases in HSP70 as well as HSC70 in cytoplasm and nuclei while HSP70 and HSC70 immunostains in human HT1080 cells were the same as normothermic controls without any accumulation in nuclei (data not shown). In HT1080 cells, HSP70 and HSC70 had similar masses at 73 kDa and were distinguishable only by charge. Recognition of these HSP70 family proteins were reconfirmed using anti-HSP70 Clone

Nuclear Lysate, IP

Cytoplasmic Lysate, IP

-200 -97.4 -68

97.4-

-43

68-

-29 L_

43-

=Z

-18.4 -200

200- N.S

-97.4

~i~iii~i~;!~!!!!iiii~iii~iiiiiiiiiii!iiii!iii~i~!i!!!i~!~!~• I ••¸¸¸¸¸~¸~¸¸¸¸~¸~ i i l ~ i

i i i!!!!~i!i i!i!!iil iii i ii ii~ii ii ! i i i !i i i !i i ilJ

97.4-

-68

i~iiii~!~!i~i~iiiiii~i~iiiiiiiiiiiiiiiiiiiiiii~iiiiiiiiiiiiiiiiiiiiiiiiii!iii~9;~! iiii~i~ii!i1i~i~i!ii~!iiiiiiiiii~i1~ii1iii1iiiiiiiii~i!!i

-43

68o

-29

43I

5

I

r

I

i

7

5

6

7

-18.4

~,

~

~

pl

pl

Panel A

Panel B

.~

Fig. 2. Isolation and 2D-PAGE separation of p53-associated proteins from cytoplasm of HT1080 and C3H10T1/2 cells. Cultures were radiolabeled with [ 35S]methionine for 6 h. Anti-p53 (PAb421) or mouse IgG2a (non-specific binding) were used for immunoprecipitation. Immune complexes were analyzed by 2D-PAGE. Panel A shows the 70 kDa (70-80 kDa) and 90 kDa protein groups which co-precipitated with p53. Various isoforms were apparent for each protein group. Non-specific binding (N.S.) showed actin and tubulin as principal contaminants in addition to other minor proteins in cytoplasm (panel A) with similar non-specific bindLng results in nuclear lysates (data not shown). Panel B shows p53 immunoprecipitation from nuclear lysates where no associated proteins were observed.

62

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

7.10. The middle panels of Fig. 3 show immunostaining of GRP75 (3 charge variants) with rabbit antibody raised to a peptide from a conserved region on the carboxy end of the protein. GRP75 was not detectable with BRM-22 but when blots in the middle panels were stripped and reprobed with BRM-22 anti-HSP70, immunostaining of the same blot showed the presence of all other HSP70 proteins observed in the top panels. HSP90 has often been observed in 2D-PAGE blots as an acidic protein, [4,21] focusing directly above GRP78, which we have also observed in various human and rodent cell lines (data not shown). Western blotting of 2DPAGE separated cytoplasmic proteins with anti-HSP90 clone 3G3 did not reveal any slightly basic 90 kDa proteins unless we first immunoprecipitated to concentrate reactants and then immunostained with this antibody as shown in the lower panels of Fig. 3. The major protein detected was a neutral to slightly basic 90 kDa

series of 3-4 identical mass proteins in an isoelectric range of 6.9-7.2 for both cell types. In related experiments, cytoplasmic protein fractions were spiked with radiolabeled anti-p53 immunoprecipitation reactions and then blotted onto nitrocellulose. P53 coprecipitants on X-ray film exposed to 2D-PAGE blots corresponded to HSC70, GRP75, GRP78 and the 90 kDa protein but not HSP70 (data not shown). The identity of proteins that coprecipitate with p53 was also confirmed by protein sequencing. We were unable to load sufficient amounts of protein from anti-p53 immunoprecipitations due to breakage of IEF tube gels. Therefore, 1 mg cytoplasmic protein samples were separated by preparative 2D-PAGE, blotted onto PVDF membranes and stained. Selection of stained proteins for sequencing was guided by overlay of autorads containing samples spiked with 35S-labeled anti-p53 immunoprecipitants or by over-

Cytoplasm HT1080

C3H10T1/2 - 97.4

- 68

i

- 43

s - 97.4

i - 68

'~

- 43

I

6

-

97.4

- 68

I

I

I

I

6

7

6

7

pl Fig. 3. Western blotting of HSP70 and 90 kDa proteins separated by 2D-PAGE from cytoplasm of HTI080 and C3H10T1/2 cells. For HSP70 class proteins, a 25 I-~gsample of cytoplasmic protein was separated by 2D-PAGE and blotted onto nitrocellulose. Blots were immunostained with anti-HSP70 (clone BRM-22), anti-GRP75 (rabbit IgG antisera), and anti-HSP90 (clone 3G3). Clone BRM-22 recognizes HSC70, HSP70 and GRP78. GRP75 was found as three isoforms. The 90 kDa protein associated with p53 could only be visualized by first concentrating via immunoprecipitation followed by Western blotting with clone 3G3. Unlike acidic HSP90 protein, the 90 kDa protein (series of 3 - 4 isoforms) associated with p53 focused in a basic region which was not detectable by Western blot of cytoplasmic proteins without prior concentration by immunoprecipitation. Immunostaining of non-specific primary antibodies did not show reactivity to any heat shock proteins (data not shown).

63

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

HTI080 Cytoplasm - 97.4

- 68

- 43

I=1 - 97.4

GRP~ ..... r"

- 68

- 43 1

I

I

I

I

I

5

6

7

5

6

7

pl

C3H10T1/2 Cytoplasm ,~,~

~ ~iii!~;i ¸¸ ~l!il!~ii .

.

.

.

'

'

~

-

9714

- 68

iiiiiiiii~iiii,,, iiiiiii~iiiiiiiiiiii~ii~

I!i

ig

- 43 |

|

|

i

!

! A

- 97.4

- 68

- 43

I

I

I

I

I

I

5

6

7

5

6

7

pl Fig. 4. Immunoprecipitation of HSP and p53 proteins from HT1080 and C3H10T1/2 cytoplasmic lysates. Cultures were radiolabeled with [35S]methionine for 3 h. Immune complexes wele formed in cytoplasmic lysates with anti-HSP70 (clone W27) in panels A,E; anti-GRP75 (rabbit IgG antisera) in panels B,F; anti-HSP90 (clone 3G3) in panels C,G; and anti-p53 (PAb421) in panels D,H. Immunoprecipitated proteins were separated in 2D-PAGE gels which were dried and fluorographed for 1 week. Panels A - D show results from HT1080 cells, and C3H10T1/2 data are indicated in panels E-H. Non-specific binding (not shown) with appropriate primary antibodies indicated that actin (large arrow) and tubulin were non-specific reactants. Although HSP proteins were separated as multiple isofcrms, the principal species are indicated by arrowheads. The multiple isoforms of p53 are underlined.

64

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

C3H10T1/2 97.4-

HT1080 -97.4

Mitochondr|a

e

68-

-68

43-

-43

a~ 97.4-

Nucleus

Nucleus

-97.4 68I

-68

........

43-

o

I 5

I 6

I 7

I 5

-43

A

I 6

I 7

pl Fig. 5. lmmunoprecipitation of p53 from nuclei and sucrose gradient enriched mitochondria of HTI080 and C3H10T1/2 cells. Cells were radiolabeled with [35S]methionine for 3 h. Nuclear and mitochondrial lysates were prepared with isotonic lysis buffer as described in Section 2. P53 (clone PAb421) was isolated by immunoprecipitation, separated by 2D-PAGE, and detected by fluorography of dried gels after 1 week.

lay of E C L films o f a n t i - H S P immunostains. A m i n o terminal sequencing identified G R P 7 5 f r o m the first 24 residues from the a m i n o terminus for h u m a n and murine samples while G R P 7 8 was identified f r o m the first 14 residues for the h u m a n sample and f r o m the initial 17 residues for the murine sample. C o m p a r e d to the predicted s e q u e n c e f r o m c D N A , the N - t e r m i n u s o f G R P 7 5 begins at residue 46 and for G R P 7 8 at residue 19, reflecting the signal sequence processing k n o w n for these proteins. R e p e a t e d attempts to obtain sequence f r o m other p53-associated proteins at 73 k D a and 90 k D a masses by preparative 2 D - P A G E or larger i m m u n o p r e c i p i t a t i o n reactions were not successful and the proteins appear to be N-terminal blocked. A l t h o u g h we believe that the clone 3G3-reactive protein is a variant o f

H S P 9 0 we designated it as a ' 9 0 k D a ' , p53-associated protein. In our next set of experiments, we p e r f o r m e d i m m u n o precipitations for HSP70, G R P 7 5 and H S P 9 0 in cytoplasmic lysates to identify p53-associated heat shock proteins and to explore associations a m o n g H S P proteins in their native state (Fig. 4, panels A - C , E - G ) . Results suggested that H S P 7 0 m e m b e r s were c o m p l e x e d to the 90 k D a protein in the c y t o p l a s m i c fraction. A n t i - H S P 7 0 (Clone W 2 7 ) in panels A and E, w h i c h r e c o g n i z e s epitopes specific for cognate and inducible forms o f HSP70, included all H S P 7 0 m e m b e r s and the 90 k D a protein. Analysis of anti-GRP75 and a n t i - H S P 9 0 b o u n d proteins f r o m H T 1 0 8 0 cytoplasmic lysates in panels B and C, respectively, re-

Fig. 6. Ultrastructure of HT1080 cells with immunoelectron microscopic localization of p53 and GRP75. Panels A and B show ultrastructure under normal fixation conditions. Cells were prepared for immunoelectron microscopy using 10 nm colloidal gold for localization of p53 protein (panels C-E) and GRP75 (panel F) as described in Section 2. As seen in panels A (6295 X ) and B (47 500 X ), a typical cultured cell contains a large, convoluted nucleus (n), often displaying a prominent nucleolus (nu); numerous mitochondria (m) often clustered near the nucleus; endoplasmic reticulum (er); Golgi (g); abundant microfilaments (f, arrows, panel B) frequently located near the periphery; and foot-like processes (arrows, Panel A) radiating outward as extensions of the plasma membrane. Immunolabeling in panels C-F is indicated by the presence of small, round black colloidal gold particles. In nearly all cells immunolabeled for p53, nuclear localization was apparent and was generally associated with the electron dense heterochromatin, but not the nucleolus, as shown in Panel C (47 500 x ). Panels D (47500 X ) and E (76 000 X ) show p53 immunolabeling in the nucleus and mitochondria. In panel D, a p53-immunolabeled mitochondrion is adjacent to some unlabeled mitochondria near the immunolabeled nucleus. Immunolabeling of mitochondria but not nucleus with GRP75 rabbit antisera in panel F (47500 X ) demonstrated specificity of the technique. Background immunolabeling of subcellular structures was essentially absent under these conditions.

B.A. Merrick et al. / Biochimica et Biophysiea Acta 1297 (1996) 57-68

vealed an association among HSC70, GRP75 and 90 kDa proteins in which GRP78 did not appear. In C3H10T1/2 cells (panel F), anti-GRP7'i reactants were GRP75, HSC70 and a small amount of 90 kDa. Anti-HSP90 (panel G) in C3H10T1/2 lysates similarly revealed the weakly basic 90

65

kDa protein accompanied by HSC70, GRP75 and GRP78. Poor detection of the murine 90 kDa protein may either be due to low expression in C3H10T1/2 cells or reduced entry of this weakly basic protein into the IEF gel. When nuclear lysates from radiolabeled human or murine cells

66

B.A. Merrick et al. /Biochimica et Biophysica Acta 1297 (1996) 57-68

were probed with anti-HSP70 and anti-HSP90 antibodies, only minimal amounts of heat shock proteins were found, suggesting HSP complexes exist primarily in cytoplasm (data not shown). In other experiments, only C3H 10T 1 / 2 cells were thermal responsive when subjected to mild heat shock, showing induction HSC70 and HSP70 proteins in both cytoplasmic and nuclear lysates while HSC70 and HSP70 levels were unchanged by heat shock in HTI080 cells (data not shown). Finally, anti-p53 immunoprecipitations from cytoplasmic lysates (Fig. 4, panels D and H) revealed several p53 isoforms accompanied by HSC70 and GRP75, GRP78 and 90 kDa where inducible HSP70 was conspicuously absent. Levels of GRP78 complexed with anti-p53 antibody (Fig. 4, particularly panel D) were relatively lower compared to those seen in Fig. 2 which may be related to the shorter period of radiolabel incorporation (3 h) needed to resolve heat shock proteins. The presence of the mitochondrial-resident protein, GRP75, as an immunoprecipitant in anti-p53 immunocomplexes suggested that p53 might be present in this organelle. Initial experiments involved isolation of a 10000xg post-nuclear pellet from homogenates of 3SSlabeled HT1080 and C3H10T1/2 cells and observing the presence of p53 by immunoprecipitation. Further experiments in both cell lines were performed to enrich mitochondria by centrifugation of 10 000 × g post-nuclear pellets through a sucrose gradient prior to immunoprecipitation. Nuclei were isolated from 1000 × g pellets by dissolution of membranes in low salt lysis buffer followed by recentrifugation. Parallel preparations from unlabeled cells were also made for ultramicroscopy and enzyme analysis. Electron microscopy showed intact nuclei without other contaminants, while fine structure of intact mitochondria, isolated as a light brown band at 1.18 g / m l density, revealed double membranes and cristae. Respective marker enzyme activities for mitochondria (glutamate dehydrogenase) and cytosol (lactate dehydrogenase) were enriched over five-fold from cytoplasmic homogenates. Activities from these marker enzymes were absent from lysates prepared from isolated nuclei suggesting no contamination from cytoplasmic proteins. Cells labeled with 35S were homogenized, and nuclei and sucrose-gradient purified mitochondria were solubilized for p53 immunoprecipitation. Fluorographs in Fig. 5 show the presence of p53 protein in mitochondria as well as nuclei. It is noteworthy that GRP75 is the predominant heat shock protein associated with p53 in mitochondria, although some of the 90 kDa protein was still observed with HT1080 cells. In nuclei, p53 is generally devoid of associated HSP proteins and consists of charge separated series similar to results in Fig. 2, panel B. This experiment suggests that p53 protein is localized in mitochondria of these cells and is associated with GRP75. Immunoelectron microscopic localization of p53 was performed to substantiate biochemical results which indicated an association of p53 with subcellular structures.

Representative findings in HTI080 cells are shown in Fig. 6. The findings for C3H10T1/2 cells were similar (data not shown). Normal ultrastructure of a whole HTI080 cell including nucleus and cytoplasmic organelles in panel A and a magnified mitochondrion near the nuclear envelope in panel B show well preserved structure under normal fixation conditions. Immunodetection of p53 was most consistently detected on ultrathin EM sections with an anti-p53 polyclonal antibody (CM1) raised to the human recombinant protein. We found little reactivity occurred with anti-p53 monoclonal antibodies like PAb240, probably because of epitope loss during fixation. P53 protein was observed in nearly all nuclei and was most often associated with heterochromatin (arrows), but not nucleoli (panel C). P53-1abeled mitochondria which are adjacent to labeled nuclei are shown in panels D and E. Nearly, all cells contained p53-1abeled mitochondria but they were not uniformly labeled within each cell (panel D). The distribution of p53-1abeling in mitochondria occurred in clusters (panel D) or could also be dissipated (panel E). P53-1abeling was also observed within cytoplasmic contents but could not be definitively assigned to other organelles like endoplasmic reticulum due to reduced preservation of ultrastructure during fixation. Sections treated with control rabbit serum showed almost no reaction to immunogold and were easily distinguished as non-specific binding (data not shown). Abundant labeling of the matrix specific protein, GRP75, was found in mitochondria, but not nucleus or other structures (panel F), and served as a positive control for this technique.

4. Discussion

The interactions of aberrant p53 protein with cellular proteins are likely to be important in understanding the biology of the transformed phenotype. In this study we have identified cytoplasmic proteins, HCS70, GRP75, GRP78 and a 90kDa protein, bound with aberrant p53 using immunoprecipitation, immunoblotting and protein sequencing analyses. For the first time, p53 has been localized to mitochondria using biochemical methods and immunoelectron microscopy. The association of GRP78 in p53 complexes implicates the presence of p53 in endoplasmic reticulum (ER) as well but this remains to be confirmed. The present finding, localizing p53 to mitochondria and ER in these two cell lines could be indicative of a function of mutant p53 in the transformation process and it will be important to explore this phenomena in other tumors given the frequency of cytoplasmic immunohistochemical staining for p53 [29,30]. Immunoprecipitation studies in H T 1 0 8 0 and C 3 H 1 0 T I / 2 lines also suggested the presence of preformed complexes among HSP proteins, particularly among HSC70, GRP75 and the 90 kDa protein. Although the present studies did not determine the specific HSP associa-

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

tions, there is precedent for heteromeric HSP complexes as reported in other systems. For example, HSP70-HSP90 binding in murine liver cytosol was accompanied by 50 kDa, 56 kDa, 63 kDa and 188 kDa proteins [21]. In CHO cells, the unliganded glucocorticoid receptor is bound to HSP70 and HSP90 [31], and pp60 ..... is complexed with HSP90 and a 50 kDa protein in virally transformed fibroblasts [32]. Immunoprecipitation of murine testis proteins with anti-HSP90 (Clone 3G3) revealed HSP90 was bound to HSC70 [22]. Generally, HSP90 binding stabilizes a diverse number of proteins in an inactive or disassembled state [1] which may act in concert with HSP70. In our study, HSP proteins coprecipitated with anti-p53 but we did not observe p53 as a coprecipitant in anti-HSP immunoprecipitation reactions. We have considered several possibilities which irLclude masking of epitopes for anti-HSP antibodies in p53-HSP complexes, absence of a stabilizing influence of anti-p53 antibody (during anti-HSP reactions) with subsequent loss of p53 during the stringency of isolation, or the relatively small fraction that a p53-HSP complex comprises of total cellular HSP. Considering the multiplicity of HI~P involvement in cellular functions, one would have expected many other proteins as coprecipitants in anti-HSP reactions; however, we observed HSP proteins to be principally associated with each other in native lysate condlLtions. This observation suggests that HSP-HSP oligomers predominate in these cell lines against a diversity in HSP complexes where p53-HSP complexes are a fractional amount and are not readily observed during anti-HSP reactions. HSP90 has often been reported as a cytosolic protein with an acidic isoelectric value (pI 5.3) [4,21]. In our studies, a weakly basic 90 kDa protein is the major reactant with p53. It is possible that anti-HSP90 (Clone 3G3) may be reacting with another HSP90 variant in these tumor cells. Isolation of the 90 kDa protein and subsequent unsuccessful attempts in sequencing from PVDF blots suggest this protein is N-terminal blocked. The p53-associated 90kDa protein is detected more easily in human HT1080 cells compared to murine C3H10T1/2 cells. Although the reasons for this; are not clear they may involve a greater expression of the 90 kDa protein in human cells and the more basic nature of the murine 90 kDa protein, a factor which often prevents many basic proteins from effectively entering the IEF gel. Reactivity of the protein with clone 3G3 antibody, migration at 90 kDa and interaction with HSP70 proteins suggest the 90 kDa protein is related to HSP90. We are investigating how HSP complexes might recognize and bind aberrant p53. The exposure of cryptic sites normally unavailable on wild-type p53 may be a key factor in understanding cytoplasmic HSP binding and its affects on subcellular distribution of aberrant p53. Normally, p53 transport to the nucleus is thought to be programmed into the primary amino-acid sequence, which contains one major and two minor nuclear localization signals [33]. Nuclear transloca-

67

tion signals in mutant p53 proteins might be overcome by cytoplasmic HSP binding and, in the cells studied here, result in localization to mitochondria and perhaps ER. The exact means by which p53 is recognized for import to subcellular organelles like mitochondria, and then combined with organelle-specific HSP's remains to be determined. Further, not all p53 is trapped in cytoplasm of HT1080 and C3H10T1/2 cells, since p53 was found in nuclei by immunoprecipitation and immunoelectron microscopy. In our studies, immunoprecipitated p53 was not bound by HSP's in the nucleus and was generally associated with heterochromatin as shown by immunoelectron microscopy. Western blotting and immunoprecipitation in HT1080 and C3H10T1/2 cells suggested that minor amounts of HSP proteins exist in the nucleus while HSP complexes are mainly localized in the cytoplasm, which may limit their ability to bind nuclear p53. Recent studies report that transactivation of human HSP70 gene is closely correlated with in vitro transforming activity and avid HSC70 binding of mutant p53, though p53 itself does not directly interact with the heat shock element in the promoter region [34]. If aberrant p53 does affect levels of free HSP70 or HSP70 oligomers, then this biochemical signal could have similarities to the powerful inductive signals produced during thermal stress when HSP70 binds to denatured proteins [35]. When HT1080 and C3H10T1/2 cells were subjected to heat stress, only C3H10T1/2 cells responded by producing the inducible HSP70 and increased levels of HSC70 while HT1080 cells, which constitutively produced HSP70 and HSC70, did not show an inductive response for either protein. These data probably reflect an altered thermal stress response system in HT1080 cells. The different reaction to thermal stress observed in HT1080 cells as compared to that observed in C3H10T1/2 cells suggests that the signals and pathways involved in HSP-p53 binding probably contain distinguishing features different from those of the thermal stress response. Components of the UPR signaling pathway for malformed or misfolded proteins [9] might be relevant, since GRP78 was found in complex with p53 in HT1080 and C3H10TI//2 cells. Evidence is accumulating that suggests that peptide binding properties of HSC70, GRP75, HSP90 may be related to antigen presentation pathways ultimately leading to an immune response [36,37]. Association of aberrant p53 with HSP70 and HSP90 family proteins may represent part of a larger program of HSP-mediated antigen processing [38,39]. The production of circulating antibodies to p53 in oncology patients is well correlated with formation of p53:HSP complexes at the cellular level and appears to be necessary for development of an immune response [40]. At present, intensive research in peptide immunization therapeutics is being directed at generation of mutant p53specific cytotoxic T lymphocytes (CTL), recognition of p53 CTL epitopes and identification of pertinent antigen processing pathways involving HSC70 and GRP75

68

B.A. Merrick et al. / Biochimica et Biophysica Acta 1297 (1996) 57-68

[37,41-43]. We are presently investigating whether mitochondrial localization of p53 is a general phenomena among mutant p53 expressing cell lines and whether there are possible linkages to antigen presentation pathways. Any analogous interactions of HSP proteins with p53 in terms of selective protein stabilization and involvement in cellular processing for antigen presentation may be important to understanding cancer cell biology of HT1080 and C3HIOT1/2 cells and development of therapeutic strategies for similarly responding tumor cells.

[16] [17] [18] [19] [20] [21] [22] [23]

Acknowledgements The authors wish to acknowledge transmission electron microscopic work performed by John L. Horton. We thank Page C. Myers and James Clark, of the NIEHS Animal Care Facility for animal care, peptide injection and rabbit serum collection. Manuscript reviews and comments by Dr. David J. Dix at the U.S. Environmental Protection Agency and Dr. Gloria A. Preston at NIEHS are greatly appreciated. We are also grateful for the HTI080 cell line from Dr. Preston.

[24] [25] [26] [27] [28] [29] [30] [31]

References [32] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]

Gething, M.J. and Sambrook, J. (1992) Nature 355, 33-45. Rensing, S.A. and Maier, U.-G. (1994) J. Mol. Evol. 39, 80-86. Lee, A. S. (1992) Curr. Opin. Cell Biol. 4, 267-273. Mizzen, L. A., Chang, C., Garrels, J.I. and Welch, W.J. (1989) J. Biol. Chem. 264, 20664-20675. Mazzarella, R.A. and Green, M. (1987) J. Biol. Chem. 262, 88758883. Hickey, E., Brandon, S.E., Smale, G., Lloyd, D. and Weber, L.A. (1989) Mol. Cell. Biol. 9, 2615-2626. Moore, S.K., Kozak, C., Robinson, S.J., Ullrich, S.J. and Appella, E. (1989) J. Biol. Chem. 264, 5343-5351. Frydman, J., Nimmesgern, E., Erdjument-Bromage, H., Wall, J.S., Tempst, P. and Hartl, F.-U. (1992) EMBO J. 11, 4767-4778. McMillan, D.R., Gething, M.-J. and Sambrook, J. (1994) Curr. Opin. Biotechnol. 5, 540-545. Greenblatt, M.S., Bennett, W.P., Hollstein, M. and Harris, C.C. (1994) Cancer Res. 54, 4855-4878. Gannon, J.V., Greaves, R., Iggo, R. and Lane, D.P. (1990) EMBO J. 9, 1595-1602. Martinez, J., Georgoff, I., Martiniz, J. and Levine, A.J. (1991) Genes Dev. 5, 151-159. Hinds, P.W., Finlay, C.A., Frey, A.B. and Levine, A.J. (1987) Mol. Cell. Biol. 7, 2863-2869. Finlay, C.A., Hinds, P.W., Tan, T-H., Eliyahu, D., Oren, M. and Levine, A.J. (1988) Mol. Cell Biol. 8, 531-539. Hinds, P.W., Finlay, C.A., Quartin, R.S., Baker, SJ., Fearon, E.R.,

[33] [34] [35] [36] [37] [38] [39]

[40] [41]

[42] [43]

Vogelstein, B. and Levine, A.J. (1990) Cell Growth Differen. 1, 571-580. Hainaut, P. and Milner, J. (1992) EMBO J. 11, 3513-3520. Lam, K.T. and Calderwood, S.K. (1992) Biochem. Biophys. Res. Commun. 184, 167-174. Zerrahn, J., Deppert, W., Weidemann, D., Patschinsky, T., Richards, F. and Milner, J. (1992) Oncogene 7, 1371-138l. Maxwell, S.A. and Roth, J.A. (1993) Oncogene 8, 3421-3426. Selkirk, J.K, Merrick, B.A., Stackhouse, B.L. and He, C. (1994) Appl. Theor. Electrophor. 4, t l - 18. Perdew, G. and Wbitelaw, M., (1991) J. Biol. Chem. 266, 67086713. Gruppi, C.M. and Wolgemuth, D.J. (1993) Dev. Gen. 14, 119-126. Anderson, M.J., Casey, G., Fasching, C.L. and Stanbridge, E.J. (1994) Genes Chromosom. Cancer 9, 266-281. He, C., Merrick, B.A., Witcher, L.L., Patterson, R.M., Daluge, D.R. and Selkirk, J.K. (1994) Electrophoresis 15, 726-734. Merrick, B.A., He, C., Witcher, L.L. and Selkirk, J.K. (1993) Proc. Am. Assoc. Ca~c. Res. 34, 105. Rickwood, D., Wilson, M.T. and Darley-Usmar, V.M. (1987) in Mitochondria: A Practical Approach, pp. 1-16, IRL Press, Oxford. Merrick, B.A., Pence, P.M., He, C., Patterson, R.M. and Selkirk, J.K. (1995)Biotechniques 18, 292-299. Merrick, B.A., Patterson, R.M., Witcher, L.L., He, C. and Setkirk, J.K. (1994) Electrophoresis 15~ 735-745. Gudas, J. M., Nguyen, H., Klein, R.C., Katayose, D., Seth, P. and Cowan, K.H. (1995) Clin. Cancer Res. 1, 71-80. Ali, I.U., Schweitzer, J.B., Ikejiri, B., Saxena, A., Robertson, J.T. and Oldfield, E.H. (1994) Cancer Res. 54, 1-5. Sanchez, E.R., Hirst, M., Scherrer, L.C., Tang, H., Welsh, M.J., Harmon, J.M., Simons, S.S.Jr., Ringold, G.M. and Pratt, W.B. (1990) J. Biol. Chem. 265, 20123-20130. Whitelaw, M.L., Hutchison, K. and Perdew, G.H. (1991) J. Biol. Chem. 266, 16436-16440. Shaulsky, G., Goldfinger, N., Ben-Ze'ev, A. and Rotter, V. (1990) Mol. Cell. Biol. 10, 6567-6577. Tsutsumi-Ishii, Y., Tadokoro, K., Hanaoka, F. and Tsuchida, N. (1995) Cell Growth and Diff. 6, 1-8. Morimoto, R.I., Sarge, K.D. and Abravaya, K. (1992) J. Biol. Chem. 267, 21987-21990. Srivastava, P.K., Udono, H., Blachere, N.E. and Li, Z. (1994) Immunogenetics 39, 93-98. Kim, H.T., Nelson, E.L., Clayberger, C., Sanjanwala, M., Sklar, J. and Krensky, A.M. (1995) J. Immunol. 154, 1614-1623. Fruci, D., Rover, P., Butler, R.H., Sorrentino, R., Tosi, R. and Tanigaki, N. (1993) Immunogenetics 38, 41-46. Manara, G.C., Sansoni, P., Badiali-DeGiorgi, L., Gallinella, G., Ferrari, C., Brianti, V., Fagnoni, F.F., Ruegg, C.L., DePanfilis, G. and Pasquinelli, G. (1993) Blood 82, 2865-2871. Davidoff, A. M., Iglehart, J.D. and Marks, J.R. (1992) PNAS 89, 3439-3442. Yanuck, M., Carbone, D.P., Pendelton, C.D., Tsukui, T., Winter, S.F., Minna, J.D. and Berzofsky, J.A. (1993) Cancer Res. 53, 3257-3261. Wiedenfeld, E.A., Fernandez-Vina, M., Berzofsky, J.A. and Carbone, D.P. (1994) Cancer Res. 54, 1175-1177. Nijman, H.W., Houbiers, J.G.A., van der Burg, S.H., Vierboom, M.P.M., Kenemans, P., Kast, W.M. and Melief, C.J.M. (1993) J. Immunother. 14, 121-126.