Calreticulin in T-lymphocytes. Identification of calreticulin in T ...

THEJOURNAL OF BIOLOGICAL CHEMl8TRK Vol. 267 No. 27, Issue of September 25 pp. 19039-19042,1992 0 1992 by The America; Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

Communication

Changes in the levels of free Ca2+ have been shown to be importatlt in signal transduction ina variety of cellular systems. Many hormones exert their effects, at least in part, by IDENTIFICATIONOFCALRETICULININ raising the Ca2+concentration in the cytoplasm which leads T-LYMPHOCYTES AND DEMONSTRATION THAT to activation of a variety of Ca2+-dependent enzymes that ACTIVATIONOF T CELLSCORRELATESWITH control cellular responses (Carafoli, 1987). When the antigen INCREASEDLEVELSOFCALRETICULINmRNA receptors on T-lymphocytes are stimulated by either mitogen, ANDPROTEIN* cross-linking antibodies, or their cognate ligands, a rise in (Received for publication, March 9, 1992) cytosolic Ca2+is also observed (Tsien et al., 1982; Hesketh et al., 1983; Gelfand et al., 1984; Lederman et al., 1984; for review Kimberley Burns$, Cheryl D. Helgasont, see Gardner (1989)). This increase of intracellular Ca2+ conR. Chris Bleackleyll, and Marek Michalakll From the Cardiovascular Disease Research Group and the centrationis believed to be a key stepinT-lymphocyte stimulation because it results in the activation of a variety of Department of Biochemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2S2 Ca2+-dependent kinases and phosphatases. Variousligands that bind T cell receptorsinduce rapid Ca2+is an essential second messenger for T cell ac- hydrolysis of inositol phospholipids (Taylor et al., 1984; Bertivation, but the exact mechanisms of its action are ridge, 1990). This leads to the production of diacylglycerol, poorly understood. The cytosolic Ca2+concentration is which activates protein kinaseC, and Imp3,’which causes a significantly increased upon the stimulation of T cells release of Ca2+ from the ER-associated intracellular store. with either mitogen, cross-linking antibodies, or their The Ca2+ signalof T cell activation may be separated into at cognate ligands. In this study, expressionof calreticu- least two distinct spatialpools: one associated with the intralin, a major Ca2+-binding(storage), KDEL protein of the endoplasmic reticulum was examined in resting cellular store (i.e. ER membranes) and one associated with and concanavalin A (ConA)-stimulated mouse and hu- enhanced Ca2+ influx via plasma membrane Ca2+ channels. man T-lymphocytes. Both resting, mouse and human Recent work from several laboratories suggests that the Ca2+ lymphocytes contain verylow levels of calreticulin content of an internal Ca2+store may effect the Ca2+ permemRNA and protein. Mouse splenocytes stimulated with ability of plasma membrane (Putney,1990; Mason et al., 1991; ConA exhibitedan induction in calreticulin mRNA Clementi et al., 1992; Demaurex et al., 1992). This coupling which peaked by Day 4. A &fold increase in the im- mechanism may be importantinmediatingthe effects of munoreactive calreticulin protein band was also ob- mitogen and other agonists in the control of Ca2+ homeostasis served in the cells during this period of stimulation. in T cells. Therefore, intracellular Ca2+ stores must play a Similarly when human lymphocytes were cultured very important role as a sources of Ca2+during T cell activawith ConA a significant increase in the levels of the tion. Ca2+-binding proteins, especially those associated with calreticulin mRNA and protein was observed. The the intracellular Ca2+ stores,would, therefore be expected to peak of calreticulin mRNA was observed at Day 1 rather than Day 4 as seen for the mouse. These results play a major role in controlling intracellularCa2+concentraclearly demonstrate the presence ofcalreticulin, a tions during thisprocess, but very little is known about their Ca2+-bindingprotein originally characterized in mus- involvement. Calreticulin is a major Ca2+-binding protein found in the cle tissue, in activated T-lymphocytes. Furthermore, we show that expression of calreticulin correlates with lumen of ER of a variety of different cellular systems (Opas T-lymphocyte activation. Our results suggest that cal- et al., 1991; Michalak et al., 1991; Milner et al., 1991). Because reticulin may be involved in the signaling pathway for of its Ca2+ binding capacity it was proposed to be a Ca2+ the induction of Ca2+-dependent processes and may storage protein in the lumen of ER (Milner et al., 1991). represent one regulatory mechanism operating in ac- Calreticulin binds -25 mol of Ca2+/mo1of protein with low tivation of T-lymphocytes. affinity and -1 mol of Ca2+/mo1of protein with high affinity (Ostwald and MacLennan,1974; Baksh and Michalak,1991)., The Ca2+ binding sites have been localized to different domains in the proteins (Baksh and Michalak, 1991), suggesting that they may perform distinct functions. Recently, Opas et * This work was supported by grants (to M. M.) from the Medical al. (1991) demonstrated that, althoughcalreticulin is concenResearch Council of Canada and the Heart and Stroke Foundation of Alberta and by grants (to R. C. B.) from the Medical Research trated in therough ER in non-musclecells, it is also found in Council of Canada and the National Cancer Institute. The costs of the nucleus andnuclear envelope. Thisisnotsurprising publication of this article were defrayed in part by the payment of considering that calreticulin has both the KDEL ER retention page charges. This article must therefore be hereby marked “aduer- signal and the nuclear targeting signal (Fliegel et al., 1989; tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate McCauliffe et al., 1990). this fact. $ Recipient of a studentship from the Alberta Heritage Foundation The present study was designed to investigate the expression of calreticulin, a KDELCa2+-bindingprotein, in Tfor Medical Research. 3 Recipient of a Province of Alberta Graduate StudentFellowship. lymphocytes. First, we identifiedcalreticulin proteinand

Calreticulin in T-lymphocytes

ll Medical Scientist of the Alberta Heritage Foundation for Medical Research. 11 Medical Research Council Scientist and a Scholar of the Alberta Heritage Foundation for Medical Research.

The abbreviations usedare: InsP3, inositol 1,4,5-trisphosphate; ER, endoplasmic reticulum; PAGE, polyacrylamide gel electrophoresis; ConA, concanavalin A; Grp, glucose-regulated protein.

19039


19040

mRNA in mouse T-lymphocytes, and, second, showed that the levels of both were elevated upon T cell stimulation.

-

A

EXPERIMENTAL PROCEDURES

ConA Stimulation-Mouse spleen cells were suspended to a density of 5 X lo" cells/ml in RHFM buffer (RPMI-1640 containing 10% fetal bovine serum, 20 mM Hepes, pH 7.5,lOO p M 2-mercaptoethanol, and antibiotics) (GIBCO) containing20 units/ml recombinant human interleukin-2 and 10 pg/ml ConA. Cultures were incubated a t 37 "C, 5% Cotin Corning 75-cm2tissue culture flasks. Aliquots were taken on each of the 6 following days for RNA isolation. Human peripheral blood lymphocytes isolated by centrifugation over Ficoll-Paque were suspended to a density of 3 X 106cells/ml in RHFM (containing 10 pJml ConA and 500 units/ml recombinant human interleukin-2). RNA was isolated from samples representing 24 and 48 h of activation. Northern Blot Analysis-Total RNA was preparedusing the method of Chirgwin et al. (1979). Ten micrograms of total cellular RNA was separated on a formaldehyde-agarose gel then transferred t o Hybond nylon membranes. Blots were prehybridized (20% or 50% formamide, 5 X SSC, 5 X Denhardt's, 50 mM NaPO,, pH 6.5, 1 mM NaP4P04,0.1% SDS, 100 p~ ATP, 2.5 mM EDTA) overnight at 42 "C. The blots were hybridized in the same buffer with 32P-labeled probes. As a probe for calreticulin the 700-base pair BamHI cDNA fragment from pGEX-CRT plasmid was used (Baksh and Michalak, 1991). The blots were normalized with y-actin and in some experiments with 28 S ribosomal RNA (Tiemeier et al., 1977), ribosomal protein L32 cDNA probe (Dudov and Perry, 1984) or human glyceraldehyde-3-phosphate dehydrogenase cDNA probe (Clontech Laboratories). The filters were washed to a stringency of 0.1 X SSC and 0.1%SDS at 42 "C (at 65 "C for y-actin). The blots were stripped with a 0.1% SDS solution. The relative abundance of each message was determined by densitometry (Chromoscan 3 Joyce, Inter Technology, Don Mills, Ontario, Canada). Imnunoblotting A~lysis-Cellular extracts from ConA-stimulated and resting lymphocytes were prepared for immunoblotting as described by Opas et al. (1991). Proteins present in cellular extracts were separated by SDS-PAGE on 10% polyacrylamide gels as described by Laemmli (1970), then stained with Coomassie Blue or transferred to nitrocellulose membranes (Towbin et al., 1979). The nitrocellulose filters were incubated with goatanti-rabbit calreticulin as described previously (Milner et al., 1991). Standards were Bio-Rad prestained markers; phosphorylase b(135,000),bovine serum albumin (85,000), ovalbumin (50,000), carbonic anhydrase (39,000), soybean trypsin inhibitor (27,000), and lysozyme (17,000). RESULTS

Induction of Expression of Calreticulin mRNA in T CellsEach day over a period of 6 days, total cellular RNA was isolated from the stimulated mouse splenic cells. The RNA was subjected to Northern blotanalysis and therelative levels of calreticulin mRNA determined. The blots were normalized using a y-actin cDNA probe (Fig. 1B). Resting lymphocytes contain very low to undetectable levels of calreticulin mRNA (Fig. LA, lane 1 ) . In contrast the level of mRNA encoding calreticulin was increased in stimulated cells (Fig. lA, lanes 2-5). The expression of calreticulin mRNA was maximal by Day 4 (lane5 ) and declined by Day 5 (Fig. lA, lane 6). The relative abundance of calreticulin and y-actin mRNA was determined by densitometry. Fig. 1C shows that the mRNA encoding calreticulin was increased in stimulated cells (Day 2-5) and that theexpression of this mRNA was maximal by Day 4. Comparison of the levels of calreticulin mRNA in Day 1 to that observed in Day 4 (Day 0 levels of calreticulin mRNA were almost undetectable) revealed approximately 2fold increase in the levels of the mRNA (Fig. 1C). In three separate experiments approximately 3.2 & 1.3-fold increase in the level of the calreticulin mRNA was observed. In contrast the signal observed for y-actin was increased by less than 1.6-fold in the same period of time. In a separate experiment, in addition to calreticulin and y-actin, the Northern blots were probed with 28 S ribosomal RNA, ribosomal pro-

C 20

CRT

0

1

2

3

4

5

6

Days

FIG. 1. Induction of calreticulin mRNA during ConA stimulation of lymphocytes. Total cellular RNA was purified from resting ( l a n e I ) and ConA-stimulated (lanes 2-6) lymphocytes. RNA was electrophoretically separatedona formaldehyde-agarose gel, blotted onto Hybond N nylon filters, and hybridized to cDNA encoding calreticulin (A) or y-actin ( B )as described under "Experimental Procedures." Portions of autoradiograms containing all hybridizable RNA species are shown. The abundance of calreticulin and y-actin mRNA was determined by densitometry scanning of blots (A) and ( B )and is shown in C.

tein L32 cDNA,and glyceraldehyde-3-phosphate dehydrogenase cDNA probes (data not shown). When levels ofRNA encoding these probes were compared between Day 1and Day 4 in the stimulated T cells, 6-fold induction of calreticulin mRNA was observed, and only 2-, 3-, 3-, and 4.0-fold increase in mRNA for y-actin, 28 S ribosomal RNA, ribosomal protein L32, and glyceraldehyde-3-phosphate dehydrogenase, respectively. Immunoblotting Analysis of Calreticulin-In order to establish if the increased levels of the calreticulin mRNA lead to a concomitant increase in the amountof the protein synthesized we carried out Western blot analysis. After Day 4 of stimulation with ConA when the highest level of calreticulin mRNA was observed, cell lysates were analyzed for the presence of calreticulin using a specific goat anti-calreticulin antibodies. The specificity of this antibody was established earlier (Milner et al., 1991). A 60-kDa immunoreactive calreticulin protein was detected in resting lymphocytes (Fig, 2, lane 2). When mouse lymphocytes were stimulated with ConA, a significant increase in the amount of immunoreactive calreticulin was observed (Fig. 2). Stimulated cells (Fig. 2, lane 1 ) were estimated by densitometry to contain -4.0 & 2.0-fold (n = 3) greater levels of proteins than resting cells (lane2). Expression of Calreticulin in Human T Cells-We also tested resting and stimulated human peripheral blood lymphocytes for the presence of calreticulin mRNA and protein. Unstimulated human peripheral blood lymphocytes (Day 0) expressed undetectable amounts of calreticulin mRNA (Fig. 3, A and C ) ; however, when the lymphocytes were cultured with ConA significant levels of the calreticulin mRNA became apparent (Fig. 3, A and C). The induction profile was relatively rapid as calreticulin mRNA was decreased slightly by Day 2 (Fig. 3C). Comparison of the levels of calreticulin mRNA in Day 0 when levels of calreticulin mRNAwere almost undetectable to that observed in Day 1 revealed approximately 17.6 +. 5.6-fold increase ( n = 3). The increased level of calreticulin mRNA was observed in the stimulated human T cells regardless of whether blots were normalized with y-actin, 28 S ribosomal RNA, ribosomal protein L32


A

A

E

.

19041

=

. m y .

r,

c

s

1

2

3

s

1 2

FIG.2. Immunoblotting analysis of calreticulin in resting and ConA-stimulated T-lymphocytes. Cellular extracts of resting and stimulated lymphocytes were prepared as described by Opas et al. (1991). The proteins were separated by SDS-PAGE and then stained with Coomassie Blue ( A ) or transferred electrophoretically t o nitrocellulose membranes ( B )and incubated with goat anti-calreticulin antibody as described under “Experimental Procedures.” Lane I , ConA-stimulated T-lymphocytes, Day 4; lane 2, resting lymphocytes; lane 3, purified calreticulin. Lane S in A, Bio-Rad low range molecular weight standards: phosphorylase b (97,400), bovine serum albumin (66,200), ovalbumin (42,700), bovine carbonicanhydrase (31,000), soybean trypsin inhibitor (21,500). In B purified calreticulin standard was co-electrophoresed with the molecular weight standards (Bio-Rad prestained markers) (lane S ) : phosphorylase b (135,000), bovine serum albumin (85,000), ovalbumin (50,000), carbonic anhydrase (39,000), soybean trypsininhibitor (27,000), and lysozyme (17,000). Protein amounts were 15 pg/lane. The arrow indicates calreticulin.

s

1

2

3

FIG.4. Immunoblotting analysis of calreticulin in resting and ConA-stimulated human T-lymphocytes. Cellular extracts of resting and stimulated lymphocytes were prepared as described by Opas et al. (1991), and proteins were separated by SDS-PAGE. Gels were stained with Coomassie Blue ( A ) or transferred electrophoretically to nitrocellulose membranes ( B ) and incubated with goat anticalreticulin antibody as described under “Experimental Procedures.” Lane 1 , resting lymphocytes; lane 2, ConA-stimulated T-lymphocytes, Day 4; lane 3, purified calreticulin. Lane S in A, Bio-Rad low range molecular weight standards: phosphorylase b (97,400), bovine serum albumin (66,200), ovalbumin (42,700), bovine carbonicanhydrase (31,000), soybean trypsin inhibitor (21,500). Lane S in B, purified calreticulin standard was co-electrophoresed with the Bio-Rad prestained molecular weight standards: phosphorylase b (135,000), bovine serum albumin(85,000), ovalbumin (50,000), carbonic anhydrase (39,000), soybean trypsin inhibitor (27,000), and lysozyme (17,000). Protein amounts were 15 pgllane. The arrow indicates calreticulin.

ting, was also increased (3.1 f 1.2-fold; n = 3) (Fig. 4). DISCUSSION

Wereport here that calreticulinmRNA and protein is present in mitogen-stimulated mouse and human T-lymphocytes. As calreticulin isa major Ca2+-binding (storage) protein found in the lumen of ER membranes (Milner et al., 1991), the increased synthesis of this protein may be directly related to T cell Ca2+ homeostasisduring mitogenic stimulation. C 30 The activation of T cells with antigen ormitogen has been ICRT 1 studied in the past andshown to lead to a series of morphological and biochemical changes (Ashman, 1984). Synthesis of specific mRNA and proteinis oneof the major biochemical events during the T cell activation process (Ahern and Kay, 1975; Cooper and Braverman, 1981; Ashman, 1984). In addition, changes in intracellular Ca2+concentration appear tobe an obligatory step in themetabolic cascade leading eventually to DNA synthesis andcell division (Tsien et al., 1982; Hesketh 0 1 2 et al., 1983; Gelfand et al., 1984; Lederman et dl., 1984; Days Ashman, 1984). Very little is known about the Ca2+-binding FIG.3. ConA induction of calreticulin mRNA in human proteins present in T-lymphocytes and their physiological lymphocytes. Total cellular RNA was isolated from resting and role during T cell activation. The increasedsynthesis of ConA stimulated lymphocytes. Lane 1 , resting lymphocytes; lane 2, calreticulin mRNA and protein, reported here, suggests that stimulated lymphocytes, Day 1; lane3, ConA-stimulated lymphocytes, calreticulin may be a part of the Ca2+-dependent transduction Day 2. RNA was electrophoretically separated on a formaldehydeagarose gel, blotted onto Hybond N nylon filters and hybridized to pathway(s) in stimulatedcells. Increased amounts of calreticDNA encoding calreticulin ( A ) or y-actin ( B ) as described under culin will increase the storage capacity of the ER andin turn modulate activities of both the ER Ca2+-ATPase and InsP3“Experimental Procedures.” Portions of autoradiograms containing all hybridized RNA species are shown. The abundance of calreticulin sensitive Ca2+release pathways. Rapid and efficient sequesand y-actinmRNA was determined by densitometry scanning of blots tration of Ca2+in the lumen of ER by calreticulin could lead ( A ) and ( B )and is shown in C. to the increased Ca2+ uptake by ER Ca2+-ATPase, dramatically enhancing theremoval of Ca2+from the cytoplasm. This cDNA, or glyceraldehyde-3-phosphate dehydrogenase cDNA hypothesis is supported by the finding that both calreticulin probes. Synthesis of calreticulin protein in stimulated human and the InsP3-receptor/Ca2+-channel are found in the same peripheral blood lymphocytes, as determined by Western blot- intracellular location, likely associated with the ER mem-

19042


branes (Krause et al., 1990). Increased Ca2+ capacityof inter- these results also suggest that calreticulin may play a role in nal stores may have additional consequences for Ca2+homeo- T cell proliferation. Gene knock-out and/or anti-sense experstasis in T cells. Studies in various cell types, including T - iments may be able to distinguish between these possibilities. lymphocytes, have suggested a correlation between filling of Acknowledgments-We thank B. Petryk for excellent technical intracellular Ca2+pools and Ca2+ entry across the plasma membranes (Putney,1990; Mason et al., 1991; Clementi et al., assistance. We thank Drs. A. Zarain-Herzherg (University of Mani1992; Demaurex et al., 1992). Thus, according to this model toba), R. Godbout, and D. Brindley (University of Alberta) for generous gifts of 28 S ribosomal RNA, ribosomal protein L32 cDNA, and of Ca2+ entry, the filling state of the intracellular Ca2+stores human glyceraldehyde-3-phosphate dehydrogenase cDNA probes, redetermines the permeability of the plasma membranes. The spectively. Ca2+capacity of internal Ca2+ stores could depend on the REFERENCES availability of calreticulin, amajor Ca2+binding(storage) Ahern, T., and Kay, J. E. (1975) Erp. Cell Res. 92, 513-515 protein of ER membranes. In addition, Arber et al. (1992) Arber, S., Krause, K.-H., and Caroni,P. (1992) J. Cell Biol. 116, 113-125 have proposed that calreticulin co-localizes with s-cyclophilin Ashman, R. F. (1984) in Fundamental Immunology (Paul, W. E., ed.) pp. 267300, Raven Press, New York to the same intracellular compartment.Cyclophilin has been Baksh, S., and Michalak, M. (1991) J. Biol. Chem. 266, 21,458-21,465 Berridge, M. J. (1990) J. Bid. Chem. 265,9585-9586 shown recently to interact with Ca2+-bindingpolypeptides, the Carafoli, E. (1987) Annu. Rev. Biochem. 56,395-433 calcineurin (Ca2+/calmodulin-activated phosphatase),and Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J. (1979) Biochemistry 18,5294-5299 calmodulin (Liu et al., 1991; Friedman et al., 1991). These Clementi. E.. Scheer. H.. Zacchetti. D.. Fasolato. C.. Pozzan. T.. and Meldolesi. proteins, including calreticulin, may play a critical role in J. (1992) J . Biol. Chem. 267,2164-2172 ' ' Cooper, H. L., and Braverman, R. (1981) J. Biol. Chem. 2 5 6 , 7461-7467 signal transduction in stimulatedT cells. N., Lew, D. P., and Krause, K.-H. (1992) J . Biol. Chem. 267,2318Calreticulin is a highly conserved protein that belongs to Demaurex, 2.124 the KDEL family of proteins, whichinclude BiP (Grp78), Do-&;, A. J., Wasley, L. C:, Raney,P., Haugejorden, S., Green, M., and Kaufman, R. J. (1990) J. B~ol.Chem. 265, 22,029-22,034 Grp94, and protein disulfideisomerase (Pelham, 1989; Ge- Dudov, K. P., and Perry, R.P. (1984) Cell 37,457-468 L., Burns, K., MacLennan, D. H., Reithmeier, R. A. F., and Michalak, thing and Sambrook,1992). In lymphoid cell lines lines rep- Fliegel, M. (1989) J.Biol. Chem. 2 6 4 , 21,522-21,528 resenting various stages of B-lymphocyte development Friedman. J.. and Weissman. I. (1991) Cell 66. 799-806 Gardner, P . (1989) Cell 5 9 , i5-20 (Munro and Pelham,1986) or in B cells induced to differen- Gelfand, E. W., Cheung, R. K., and Grinstein, S. (1984) J. Cell. Physiol. 1 2 1 , 6.1.1-6.19 tiate, elevated expression of several proteins of the KDEL familyhas recentlybeen reported (Haas and Wabl, 1983; Lewis et al., 1985; Mazzarella and Green, 1987; Dorner et al., 1990; Mazzarella et al., 1990). No comparable information is available concerning expression of KDEL proteins in resting or stimulated T-lymphocytes. This is the first report documenting the presence of calreticulin, a KDEL protein, in resting lymphocytes and induction of its mRNA and protein upon mitogen stimulation. It is interesting to note that under certain stress conditions BiP and Grp94 have both been found in the ER and the nucleus (Welch et al., 1983; Lee et al., 1984). Kozutsumi et al. (1988) proposed that BiP is able to recognize the secretory load in the ER and may initiate a signaling pathway to the nucleus to regulate transcription from the ER stress protein genes. Similarlywe have recently found someimmunoreactive calreticulin localized to the nucleus (Opas et al., 1991). In stimulated cells calreticulin may also accelerate the rate of a manner synthesis of itself or other stress-inducedproteins in similar to that proposed for BiP (Lenny and Green, 1991; Gething and Sambrook, 1992). This hypothesis is currently ". . Ostwald. T. J.. and MacLennan. D. H. (1974) J. Bid. Chem. 249,974-979 under investigation in our laboratories. Pelham, H. R.'B. (1989) Annu. keu. Cell Biol. 6,1-23 In conclusion, in this report we have presented evidence Putney, J. W., Jr. (1990) Cell Calcium 11,611-624 M. V., Metcalfe, J. C., Hesketh, R. B., Smith, G. A., and Moore, J. P. that expression of calreticulin mRNA and protein isinduced Taylor, (1984) Nature 312,462-465 in stimulated T cells. By virtue of its cellular localization and Tiemeier, D. C., Tilghman, S. M., and Leder, P. (1977) Gene (Amst.) 2, 173191 its ability to bind and store Ca2+, calreticulinmay modulate Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. U. S. A. Ca2+-depev.dent events in the ER of stimulated T-lympho76,4350-4354 Tsien, R. T., and Rinsk, T. J. (1982) Nature 295.68-70 cytes. In this paperwe speculate that calreticulinexpression Welch, W.Y.,J.,Pozzan, Garrels, J. I., Thomas, G. P., Lin, J. J.-C., and Feramisco, J. R. may be directly related to lymphocyte activation; however, (1983) J. Biol. Chem. 258,7102-7111 I

~

~~~

i

~~~

~

~~~

~I