Regulation of Calreticulin Expression during Induction of Differentiation in Human Myeloid Cells

Induction of differentiation of HL-60 human myeloid cells profoundly affected expression of calreticulin, a Ca2+-binding endoplasmic reticulum chaperone. Induction with Me2SO or retinoic acid reduced levels of calreticulin protein by ∼60% within 4 days. Pulse-chase studies indicated that labeled calreticulin decayed at similar rates in differentiated and undifferentiated cells (t 1 2 ∼4.6 days), but the biosynthetic rate was <10% of control after 4 days. Differentiation also induced a rapid decline in calreticulin mRNA levels (90% reduction after 1 day) without a decrease in transcript stability (t 1 2 ∼5 h). Nuclear run-on analysis demonstrated rapid down-regulation of gene transcription (21% of control at 2 h). Differentiation also greatly reduced the Ca2+ content of the cells (25% of control), although residual Ca2+ pools remained sensitive to thapsigargin, ionomycin, and inositol trisphosphate. Progressive decreases were also observed in levels of calnexin and ERp57, whereas BiP/GRP78 and protein disulfide isomerase were only modestly affected. Ultrastructural studies showed a substantial reduction in endoplasmic reticulum content of the cells. Thus, terminal differentiation of myeloid cells was associated with decreased endoplasmic reticulum content, selective reductions in molecular chaperones, and diminished intracellular Ca2+ stores, perhaps reflecting an endoplasmic reticulum remodeling program as a prominent feature of granulocytic differentiation.

Biophysical studies show calreticulin to be a highly asymmetric molecule consistent with this predicted three domain structure. Calreticulin binds Ca 2ϩ with both high and low affinities and with high capacity. A single high affinity site (K d ϳ1 M) is located within the P domain of the molecule and multiple low affinity sites (K d ϳ2 mM, 20 -25 mol of Ca 2ϩ /mol of protein) are present in the highly acidic C domain (3,4). The N domain contains at least one site for binding Zn 2ϩ , which appears to involve 4 of the histidine residues in this region (5,6). Binding of these ions affects the conformation and stability of the protein (7).
Numerous and apparently unrelated functions have been ascribed to calreticulin since it was first identified, but its roles in the regulation of Ca 2ϩ homeostasis and as a molecular chaperone of nascent glycoproteins are the two functions that have been most extensively characterized. Calreticulin appears to regulate intracellular Ca 2ϩ homeostasis through more than one mechanism. The high Ca 2ϩ -binding capacity of calreticulin suggests that it acts as a Ca 2ϩ store or buffer within the ER, and there is evidence that in at least some cell types it is a main source of inositol 1,4,5-trisphosphate (IP 3 )-releasable Ca 2ϩ (3,5,8,9). In support of these observations, overexpression of calreticulin in some cells is accompanied by a significant increase in the Ca 2ϩ storage pool, as well as an alteration in IP 3 -mediated Ca 2ϩ release and influx (9,10). Moreover, the thapsigargin-and ionomycin-sensitive Ca 2ϩ pools were markedly reduced in calreticulin-null mouse embryonic stem cells, defects that were corrected by ectopic expression of calreticulin (11). In contrast, calreticulin-deficient murine embryonic stem cells and fibroblasts exhibited normal levels of IP 3 -releasable Ca 2ϩ (12), suggesting that dependence of Ca 2ϩ release on calreticulin may vary among cell types and signaling pathways. Calreticulin may also regulate Ca 2ϩ levels by direct interaction with the ER uptake mechanism. Luminal Ca 2ϩ released into the cytosol through the IP 3 receptor channel is taken back up into the ER by the sarco-endoplasmic reticulum calcium ATPases (SERCA). One isoform of this family of Ca 2ϩ pumps (SERCA2b) has an additional transmembrane segment and a C-terminal domain that extends into the lumen of the ER and contains a putative N-glycosylation site. In elegant studies using the Xenopus oocyte model, Camacho and colleagues (13,14) showed that calreticulin inhibited the activity of SERCA2b and altered the temporal and spatial patterns of IP 3 -mediated Ca 2ϩ release. This activity was dependent on Ca 2ϩ concentration and was mediated by the lectin-binding P domain of calreticulin, rather than its high capacity Ca 2ϩ -binding C domain.
Thus, calreticulin may modulate agonist-stimulated Ca 2ϩ mobilization through multiple pathways.
In addition to its role in Ca 2ϩ signaling, calreticulin has an important function as a lectin-like chaperone for newly synthesized glycoproteins (3,(15)(16)(17). In this respect, it displays many features in common with calnexin, another ER molecular chaperone. Calnexin and calreticulin share regions of structural homology and show lectin-like selectivity for mono-glucosylated N-linked glycoproteins (16, 18 -20). The two proteins differ in topography; calreticulin is a soluble luminal ER protein, and calnexin has a transmembrane segment and a cytoplasmic domain. The interactions of calreticulin with unfolded glycoproteins and with other ER chaperones are Ca 2ϩ -dependent, suggesting a potential functional link between the chaperone and Ca 2ϩ -modulating roles of calreticulin (6,20,21). Moreover, the 5Ј-flanking sequence of the calreticulin gene and the genes for the chaperone proteins BiP/GRP78 and GRP94 show regions of sequence homology, suggesting that they are coordinately regulated (22).
In phagocytic leukocytes, calreticulin serves several important functions. It acts as a molecular chaperone for the enzyme myeloperoxidase (15), a catalytic component of oxygen-dependent microbicidal systems. In addition, membranous structures containing calreticulin, as well as the Ca 2ϩ -ATPase SERCA2b, accumulate in the actin-rich filamentous regions surrounding developing phagocytic vacuoles (23), suggesting involvement of these vesicles in the modulation of Ca 2ϩ -dependent phagolysosomal functions. Calreticulin has also been reported to be released from activated neutrophils (24) and may bind and alter the activity of the C1q component of complement (25). A recent report (26) demonstrates that C1q-calreticulin interactions are involved in the uptake of apoptotic cells by phagocytes.
Because of its importance in the biology and function of myeloid cells, we have investigated the biosynthesis and regulation of calreticulin in this cell type. Our previous studies (27) of the biosynthesis and post-translational processing of calreticulin in the HL-60 and PLB-985 myeloid cell lines showed that the primary translation product undergoes co-translational signal peptide cleavage and post-translational N-linked glycosylation to form the mature molecule. In the current study, we have investigated the transcription and translation of calreticulin and characterized the mechanisms of the dramatic down-regulation of its expression during in vitro induction of myeloid cell differentiation.
Protein Biosynthesis-Biosynthetic labeling of cultured cells was carried out as described previously (15,27) using a 30-min preincubation in methionine-free medium followed by pulse-labeling with [ 35 S]methionine (ϳ1000 Ci/mmol, Amersham Biosciences). The duration of pulses and subsequent chases with excess unlabeled methionine are indicated in the text and figure legends. Following labeling, the cells were disrupted in an anti-protease buffer and immunoprecipitated with specific anti-human calreticulin (27). In some experiments, immunoprecipitation was also carried out on aliquots of cell culture medium. Immunoprecipitates were analyzed by SDS-PAGE and fluorography of the dried gels, and band intensity was quantitated by scanning densitometry (CS-9000U, Shimadzu, Kyoto, Japan).
Northern Blot Analysis-Poly(A) ϩ RNA was isolated from the cells using SDS lysis and oligo(dT) selection (FastTrack, Invitrogen). The RNA was dissolved in nuclease-free water (Invitrogen) and quantitated by absorption at 260 nm. Samples of RNA (3 g) were denatured in formamide/formaldehyde buffer at 65°C for 15 min and then separated by electrophoresis (360 V-h) on 1.2% agarose-formaldehyde gels. Separated RNA was transferred overnight to nylon membranes (Nytran Plus, Schleicher & Schuell) by capillary blotting in 20ϫ SSC and then fixed by UV cross-linking. The membranes were pre-hybridized for 2 h at 42°C in a solution containing 5ϫ SSC, 5ϫ Denhardt's solution, 0.1% SDS, 50% formamide, 50 mM sodium phosphate, pH 6.8, and 100 g/ml heat-denatured salmon sperm DNA. Probes used in this study (cloned from an HL-60 cell human cDNA library (33)) were the cDNAs for calreticulin (1.9 kb), p47 phox (1.36 kb), p67 phox (2.2 kb), and a PstI restriction fragment of type I IP 3 receptor (1.5 kb; some sequence in common with types II and III IP 3 R isoforms). Probes were labeled with [␣-32 P]dCTP (ϳ3000 Ci/mmol, PerkinElmer Life Sciences) using a random-prime labeling kit (Roche Molecular Biochemicals), heat-denatured, and 10 6 cpm/ml added to the membrane in fresh pre-hybridization buffer supplemented with 10% dextran. Hybridization was carried out at 42°C overnight. The membranes were then washed in a stepwise manner with a final stringency wash of 0.2ϫ SSC plus 0.1% SDS at 68°C for 20 min. Following autoradiography the membranes were stripped by soaking in 50% formamide, 2ϫ SSPE at 65°C for 1 h, and then re-probed as above with DNA for human ␤-actin (CLONTECH Laboratories, Palo Alto, CA), which served as a sample-loading control. Quantitation was done by scanning densitometry as above. Where indicated, transcripts were also examined in a slot-blot format, using the same probes and hybridization conditions as noted above.
RNA Stability-Transcript stability was assessed by addition of actinomycin D (50 g/ml) to cultures of either undifferentiated or induced cells at 0, 1, 2, 3, 4 and 6 h. At the 6-h time point, poly(A) ϩ RNA was isolated from all sets of cells and analyzed in a slot-blot format, using the same procedures, probes, and hybridization conditions as noted above. Scanning densitometry was used to determine the time required for a 50% decrease from base-line transcript levels, designated the t1 ⁄2 .
Nuclear Run-on Analysis-All solutions were made using diethyl pyrocarbonate-treated sterile distilled and deionized water. Undifferentiated or induced cells (ϳ5 ϫ 10 8 in 5 ml of relaxation buffer) were treated with 2 mM diisopropyl fluorophosphate for 20 min and then disrupted by nitrogen cavitation at 350 pounds/square inch for 20 min at 4°C with evacuation into relaxation buffer containing 1.25 mM EGTA, all as described previously (34). Following centrifugation at 500 ϫ g for 10 min at 4°C, the nuclear pellet was resuspended on ice in nuclear wash buffer (50 mM Tris-HCl, pH 8.0, 1.25 mM EGTA, 25 mM KCl, 1 mM spermidine), and an aliquot was placed in 1 g/ml ethidium bromide for determination of the concentration of nuclei under fluorescence microscopy. After centrifugation at 1000 ϫg for 10 min at 4°C, the nuclei were resuspended at 2 ϫ 10 8 /ml in storage buffer (50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 0.1 mM EDTA, 2 mM dithiothreitol, 40% glycerol). Aliquots were then rapidly frozen in a methanol/dry ice bath and stored in liquid N 2 . For the RNA labeling reaction, 100 l of freshly thawed nuclei in storage buffer were mixed with an equal volume of labeling mix (200 mM KCl, 8 mM MgCl 2 , 1 mM each of ATP, UTP, and CTP, 100 M GTP, and 100 Ci of [␣-32 P]GTP (800 Ci/mmol, PerkinElmer Life Sciences)). Labeling was carried out at 30°C for 30 min in a shaking water bath and then 20 l (4 g) of RQ1 RNase-free DNase I (Promega Corp., Madison, WI) was added and incubation continued for 10 min. After dilution with 220 l of digestion buffer (1% SDS, 50 mM Tris-HCl, pH 7.0, 50 mM EDTA), 20 l (60 g) of proteinase K (Invitrogen) was added, and incubation was carried out at 42°C for 30 min. Samples were extracted twice with phenol/chloroform/isoamyl alcohol and then once with chloroform. The final aqueous phase was ethanol-precipitated in the presence of 20 g of carrier RNA (Escherichia coli transfer RNA, RNase-free, Roche Molecular Biochemicals). The pellet was dissolved in 70 l of nuclease-free water, and unincorporated label was removed by centrifugation through a Sephadex G-50 spin column (Roche Molecular Biochemicals). The amount of 32 P activity incorporated into RNA was determined by scintillation counting of trichloroacetic acid (10% for 10 min and then 5% twice for 5 min each) precipitates.
The DNA probes used were the human cDNAs for calreticulin (1.9 kb), myeloperoxidase (3.3 kb), and p47 phox (1.36 kb), all constructed in pBluescript (Stratagene, La Jolla, CA). The pBluescript vector was used as a control probe. For detection of ␤-actin transcripts, we used a linear DNA probe containing the complete 2-kb cDNA for human ␤-actin (CLONTECH). Plasmid DNA was purified using an anion exchange protocol (Qiagen Inc., Chatsworth, CA.) and linearized by digestion with HindIII. Linearized plasmids were denatured in a boiling water bath for 5 min, then treated with NaOH to a final concentration of 0.2 M, neutralized with an equal volume of 2 M ammonium acetate, and applied to Nytran Plus membranes using a slot-blot apparatus (5 g of DNA/slot). The applied DNA was fixed to the nylon membrane by UV cross-linking, and the strips were stored over desiccant at 4°C. Strips containing the fixed DNA were pre-hybridized at 52°C overnight in 2 ml of hybridization solution (5ϫ SSC, 5ϫ Denhardt's solution, 0.1% SDS, 50% formamide, 50 mM sodium phosphate, pH 6.8, and 100 g/ml E. coli transfer RNA). The solution was discarded and replaced with 2 ml of fresh hybridization solution containing ϳ10 6 cpm/ml of the labeled RNA. Hybridization was carried out at 52°C for 3-4 days. The membranes were then washed twice in 2ϫ SSC at room temperature for 15 min, twice in 2ϫ SSC at 65°C for 60 min, once at 37°C for 30 min in 2ϫ SSC containing 100 g/ml RNase A, and finally in 0.3ϫ SSC containing 0.1% SDS at 65°C for 30 min. Bands visualized by autoradiography were quantitated by densitometry as above. 45 Ca 2ϩ Measurements-HL-60 cells were cultured for 54 h in the presence of 10 Ci/ml 45 Ca 2ϩ in complete medium with or without 1.25% Me 2 SO. The cells were centrifuged and then resuspended in Ca 2ϩ -free medium (138 mM NaCl, 6 mM KCl, 1 mM MgCl 2 , 20 mM glucose, 20 mM HEPES, pH 7.4). For the measurement of total cellassociated 45 Ca 2ϩ , 10 6 cells in 200 l of buffer were centrifuged; the radioactivity in the supernatant was counted and subtracted from the radioactivity measured for the same volume of cell suspension. To measure thapsigargin-sensitive and ionomycin-sensitive 45 Ca 2ϩ content, 10 6 45 Ca 2ϩ -labeled cells in Ca 2ϩ -free medium were exposed for 10 min to either 100 nM thapsigargin or 2 M ionomycin, and the radioactivity was counted in the supernatant after centrifugation of cells. Samples were placed in a vial containing liquid scintillation mixture (Ultima Gold, Packard Instrument Co.), and the radioactivity was measured using a Packard 1900 TR scintillation counter.
Transmission Electron Microscopy-Untreated and Me 2 SO-treated HL-60 cells were centrifuged at 400 ϫ g for 5 min and fixed in a phosphate-buffered mixture of 2% paraformaldehyde and 0.5% glutaraldehyde for 45 min, then 1% osmium tetroxide for 30 min, washed, and dehydrated in sequential ethanols followed by propylene oxide. Thin sections were cut, stained with uranyl acetate followed by lead citrate, and then viewed in a Phillips 301 TEM at 60 kV.
Statistical Analysis-Standard error was used as an estimate of variance, and means were compared using Student's t test for independent variables.

Induction of Differentiation Reduces the Level and Biosynthetic Rate of Calreticulin Protein
Protein Level-Me 2 SO induces granulocytic differentiation of myeloid cell lines (29). To determine the effect of differentiation on calreticulin protein level, HL-60 cells were incubated for 4 days in complete medium supplemented with 1.25% Me 2 SO; aliquots were removed at 0, 2, and 4 days, and cell lysates were analyzed by immunoblotting with monospecific rabbit antihuman calreticulin (27). HL-60 cell calreticulin protein levels decreased over time (Fig. 1A), with quantitative analysis of 5 independent experiments (Fig. 1A, table inset) indicating more than a 50% decline at 4 days. In a separate experiment, five different sets of HL-60 cells were analyzed in parallel at 0 and 4 days after Me 2 SO addition (Fig. 1B). There was a consistent decrease in calreticulin levels to a mean of 38.8% (Ϯ3.7, p Ͻ 0.001) of the zero time controls.
Protein Catabolic Rate-Possible mechanisms that could account for the observed decreases in steady-state calreticulin levels accompanying the induction of differentiation were an increase in the catabolic rate or a decrease in the biosynthetic rate. To determine whether differentiation influenced the rate of disappearance of calreticulin, undifferentiated and 4-day Me 2 SO-induced HL-60 cells were metabolically pulse-labeled with [ 35 S]methionine for 30 min and then chased in fresh complete medium with unlabeled methionine for up to 7 days. For the Me 2 SO-induced cells, the chase medium included Me 2 SO, although essentially similar results were obtained if Me 2 SO was omitted from the chase medium. Fixed volume aliquots of the cultures were removed at intervals for cell centrifugation, immunoprecipitation, SDS-PAGE, and fluorography. Following densitometry and normalization to the corresponding levels obtained immediately after labeling (Fig. 2), we determined that labeled calreticulin in the undifferentiated and Me 2 SO-treated cells decayed at approximately equal rates during the chase period. The calculated calreticulin t1 ⁄2 for untreated cells was 4.5 days and for Me 2 SO-treated cells was 4.6 days. Thus, an increased rate of loss of calreticulin does not appear to contribute to the reduction of cellular levels of calreticulin observed during HL-60 cell differentiation. An interesting feature of this experiment was the unusually high stability of calreticulin. Our data suggest a fractional catabolic rate in HL-60 myeloid cells of only about 13% per day. In additional experiments, we determined both cell-associated and extracellular 35 S-calreticulin. The results indicated that release of calreticulin from the cells contributed only modestly to cellular calreticulin homeostasis (Ͻ20%) under our experimental conditions (data not shown). Protein Biosynthetic Rate-In view of the lack of any change in the catabolism of calreticulin to account for its decreased levels in differentiated cells, we next investigated whether the biosynthetic rate was changed in response to Me 2 SO-induced differentiation. Parallel experiments were also carried out with other differentiating agents. Thus, HL-60 cells were cultured in Me 2 SO or retinoic acid to induce granulocytic differentiation or in PMA or 1,25-dihydroxyvitamin D 3 to induce monocytic differentiation. At intervals, aliquots of cells were taken for biosynthetic pulse labeling with [ 35 S]methionine, immunoprecipitation, and SDS-PAGE analysis (Fig. 3). Induction of differentiation by each agent decreased the amount of nascent calreticulin labeled with [ 35 S]methionine, indicating reductions in the biosynthetic rate. Similar results were observed in studies on PLB-985 cells (data not shown). The effect was progressive over time and showed differences in rate and extent with the various differentiating agents. Initially, the effect was most rapid in the Me 2 SO-treated cells, which exhibited a reduction of calreticulin synthesis to 36% of zero time levels after only 1 day of treatment. In 4 replicate experiments, the mean normalized values for calreticulin biosynthesis rates (relative to zero time controls) after 1, 2, 4, and 6 days of culture in Me 2 SO were 44.2 (p Ͻ 0.05), 33.8 (p Ͻ 0.01), 9.5 (p Ͻ 0.001), and 2.0% (p Ͻ 0.001), respectively. These experiments clearly indicate that the reduction in calreticulin protein levels observed during differentiation of myeloid cells resulted largely from a major decrease in the rate of calreticulin biosynthesis.

Induction of Differentiation Reduces Calreticulin mRNA Level and Transcriptional Rate
Transcript Level-The studies described above showed that the induction of differentiation of HL-60 cells by a variety of agents resulted in a marked reduction in the biosynthesis of calreticulin. To determine the effect of myeloid cell differentiation on calreticulin mRNA levels, Northern analysis was carried out on the poly(A) ϩ RNA from cells induced to differentiate (Fig. 4). Calreticulin mRNA was highly expressed in untreated cells but was markedly reduced in both HL-60 and PLB-985 cells within 1 day of Me 2 SO treatment, and further reduction was seen after 2 days (Fig. 4A). Following normalization to the levels of ␤-actin mRNA, it was estimated that Me 2 SO treatment induced an approximate 80% reduction in the level of calreticulin transcripts within 1 day. In 6 replicate experiments, the mean normalized values for calreticulin message levels (relative to zero time controls) after 1, 2, and 4 days of culture in Me 2 SO were 19.6, 16.2, and 9.1%, respectively (p Ͻ 0.001 for each).
Differentiation of HL-60 and PLB-985 cells is accompanied by the induction of a number of myeloid-specific genes, the products of which carry out essential functions in the terminally differentiated granulocytic cell. Two such genes are p47 phox and p67 phox , cytosolic components of the phagocyte NADPH oxidase (35). To place the differentiation-induced changes in calreticulin mRNA in context, they were compared with those for p47 phox and p67 phox in HL-60 cells treated with Me 2 SO (Fig. 4B). The cellular level of calreticulin mRNA was again rapidly reduced, whereas the p47 phox and p67 phox transcripts increased from undetectable levels in untreated cells to maximal expression after 2-4 days of Me 2 SO treatment. Parallel studies were done comparing calreticulin with the type 1 IP 3 receptor, a protein that, like calreticulin, is localized to the endoplasmic reticulum and is involved in calcium release (Fig.  4C). We observed that both Me 2 SO and retinoic acid induced a 3-4-fold increase in IP 3 R mRNA within 1 day of treatment, whereas calreticulin mRNA was again significantly reduced.
The detailed kinetics of the effects of various inducing agents on calreticulin mRNA expression in HL-60 cells were examined in a slot-blot hybridization assay. RNA was isolated from the cells at 0, 3, 6, 9, 12, 24, and 48 h after initiation of treatment and probed for calreticulin and ␤-actin. After scanning densitometry, calreticulin mRNA was normalized to ␤-actin levels, and results were expressed as a percentage of the initial value (Fig. 5). Me 2 SO treatment produced the most rapid reduction (50% in ϳ5 h) of calreticulin mRNA (as was observed for protein synthetic rates, see Fig. 3). Retinoic acid and PMA were also very active, reducing the mRNA levels by 50% in ϳ10 h, whereas 1,25-dihydroxyvitamin D 3 was less active in downregulation of calreticulin transcripts (50% reduction in ϳ25-30 h).
Transcript Stability-Possible mechanisms that could account for the observed decreases in steady-state levels of calreticulin transcripts accompanying the induction of differentiation were a decrease in transcript stability (i.e. an increase in the catabolic rate) or a decrease in the transcriptional rate. To determine whether calreticulin mRNA stability was affected by Me 2 SO treatment, studies were carried out using actinomycin D as an inhibitor of transcription. HL-60 cells were cultured in Me 2 SO for 0, 1, or 2 days and then incubated in fresh medium; actinomycin D was added at various times up to 6 h (see "Experimental Procedures"). At the end of these incubations, poly(A) ϩ RNA was extracted, and slot-blot hybridization analyses were carried out. Levels of calreticulin and ␤-actin mRNA were expressed as a percentage of the levels at 0 time, and t1 ⁄2 was calculated from these values. Because the t1 ⁄2 averaged between 3 and 5 h, analysis focused especially on the 4-h time point. Treatment of the cells with Me 2 SO resulted in significant and similar increases in the normalized 4-h transcript levels for both calreticulin and actin (Table I). In keeping with this, the calculated t1 ⁄2 values for both calreticulin and actin transcripts were increased by ϳ50% (Table I), indicating stabilization of these mRNAs in Me 2 SO-induced HL-60 cells. Thus, the down-regulation of calreticulin mRNA in differentiating HL-60 cells could not be explained by increased transcript catabolism. Indeed, enhanced mRNA stability provided some counter-balance to decreased transcription (see below).
Transcriptional Rate-To determine whether reduction in the rate of gene transcription was responsible for the rapid down-regulation of calreticulin mRNA in differentiating HL-60

FIG. 4. Northern hybridization analysis of calreticulin transcript levels during myeloid cell differentiation.
A, HL-60 or PLB-985 cells were cultured in the presence of 1.25% Me 2 SO for 0, 1, or 2 days, as indicated. Poly(A) ϩ RNA was isolated, separated (3 g/lane) on 1.2% agarose-formaldehyde gels, transferred, and probed for calreticulin mRNA, followed by stripping and probing for ␤-actin, which was used as a loading control. After densitometric quantitation, the content of calreticulin transcripts was corrected for actin and expressed as a percent of the zero time values. The studies shown are representative of five independent experiments for HL-60 cells and two experiments for PLB-985 cells. B, HL-60 cells were cultured and analyzed as in A, except that the blots were also probed for p47 phox and p67 phox . The data for calreticulin mRNA levels are expressed in the same fashion as above. The study shown is representative of two independent experiments. C, HL-60 cells were cultured and analyzed as in A, except that differentiation was induced with either Me 2 SO (Me) or retinoic acid (R) for 1 or 3 days, as indicated, and the blots were also probed for type 1 IP 3 receptor (IP 3 R). The data for calreticulin and IP 3 R mRNA levels are expressed in the same fashion as above. The study shown is representative of two independent experiments. CRT, calreticulin. HL-60 cells were cultured in the various differentiating agents as indicated, and aliquots were removed at the times shown. Poly(A) ϩ RNA was isolated and analyzed by slot-blot hybridization with a calreticulin probe, following which the blots were stripped and probed for ␤-actin, which was used a loading control. After densitometric quantitation, the content of calreticulin transcripts was corrected for actin and expressed as a percent of the zero time values. The results shown are from a single study, which is representative of two independent experiments. CRT, calreticulin. cells, nuclear run-on analyses were carried out on undifferentiated and Me 2 SO-induced cells. As illustrated in the left-hand panel of Fig. 6, transcription of the calreticulin gene was essentially eliminated following 96 h of Me 2 SO treatment, as was transcription of myeloperoxidase, another gene that is downregulated in differentiation-induced HL-60 cells. In the same cells, the transcription of the internal control ␤-actin remained largely unchanged, whereas transcription of p47 phox was undetectable initially but was strongly induced at 96 h. To determine more precisely the period during which calreticulin transcription was affected, nuclear run-on analysis was carried out at shorter intervals after initiation of Me 2 SO treatment. These studies (Fig. 6, right-hand panel) showed that calreticulin gene transcription was down-regulated very rapidly after Me 2 SO exposure (79% decrease in 2 h). Myeloperoxidase transcription appeared to be reduced even more quickly, whereas actin was unaffected. These experiments indicated that a dramatic decrease in the rate of gene transcription explained the downregulation of calreticulin mRNA in differentiating HL-60 cells.
Further studies were carried out to determine whether the reduced transcription of the calreticulin gene resulted from an inhibition of transcription initiation or from a blockade of elongation of nascent transcripts. Nuclear run-on analysis was carried out on nuclear extracts of cells treated with Me 2 SO for 6 and 12 h and using DNA probes corresponding to different regions of the calreticulin gene (exons 1-9 (i.e. full-length), 1-5, 7-9, and 9 alone). Determination of signal intensity showed that transcription decayed at similar rates across the length of the calreticulin gene (data not shown), suggesting that Me 2 SO treatment induces a reduction in transcription initiation, rather than an elongation blockade of ongoing gene transcription.

Induction of Differentiation Reduces Ca 2ϩ Storage Pools
Current evidence indicates that calreticulin functions as a principal ER store or buffer for intracellular Ca 2ϩ (1,3,9). To determine whether the decrease in calreticulin content observed during differentiation of HL-60 cells was accompanied by a change in Ca 2ϩ storage capacity, cells were labeled to isotopic equilibrium with 45 Ca 2ϩ for 54 h in the presence or absence of Me 2 SO. Differentiation had a profound effect on Ca 2ϩ content, reducing total cell-associated 45 Ca 2ϩ by about 75% (Fig. 7). Nonetheless, the Ca 2ϩ pool in the differentiated cells, although greatly reduced, was sensitive to both thapsigargin and ionomycin treatment, suggesting that differentiation resulted in a quantitative effect on intracellular Ca 2ϩ stores, rather than an alteration in the nature of the stores themselves. This was supported in further studies, where it was observed that the fraction of total cellular Ca 2ϩ released after stimulation with the physiologic agonist IP 3 was similar in undifferentiated and differentiated cells (data not shown).

Induction of Differentiation Causes Selective Reductions in Levels of ER Proteins
To investigate the effect of differentiation on other ER resident proteins besides calreticulin, and to control for the possibility of a generalized Me 2 SO-induced impairment of protein homeostasis, the effect of Me 2 SO treatment on a number of other proteins was investigated (Fig. 8). Among the ER proteins examined there was a hierarchy of responses. Calreticulin was consistently and strongly down-regulated as shown above, and calnexin levels were very similarly reduced over the course of 6 days of differentiation. ERp57 exhibited substantial downregulation, although more slowly and to a lesser extent than for calreticulin and calnexin (Fig. 8). In contrast, protein-disulfide isomerase and BiP/GRP78 remained unchanged or decreased only modestly. In the same experiments, the levels of the NADPH oxidase cytosolic components p67 phox and p47 phox increased markedly during differentiation, and the levels of actin were essentially unchanged. Thus, the induction of granulocyte differentiation in HL-60 cells was associated with selective reduction of a subset of ER chaperone proteins.

Induction of Differentiation Reduces ER Content
The decrease in expression of the ER resident proteins calreticulin, calnexin, and ERp57 suggests that the induction of differentiation in HL-60 cells is associated with a change in the subcellular ER compartment itself. To investigate morphological changes in cellular ultrastructure, HL-60 cells were treated with Me 2 SO for up to 6 days and then fixed and processed for transmission electron microscopy. Marked changes in morphology consistent with the granulocytic maturation/ differentiation of the cells were observed in both the nuclear and extranuclear compartments. The control untreated cells displayed a rounded morphology, with large non-lobulated nuclei containing prominent nucleoli and dispersed nuclear chromatin (Fig. 9, A-C). The extranuclear compartment of the untreated cells contained an abundant network of rough ER. In  contrast, the nuclei of cells treated with Me 2 SO for 6 days demonstrated a pyknotic accumulation of nuclear chromatin and marked lobulation and ER profiles were much less prevalent than in undifferentiated cells (Fig. 9, D-E).

DISCUSSION
Calreticulin is a widely expressed and highly conserved Ca 2ϩ -binding ER protein. It is essential for normal embryonic development (36) and participates in a wide range of cellular functions, including Ca 2ϩ homeostasis and signaling, nascent protein folding, regulation of steroid-sensitive genes, and modulation of cell adhesion (1,3,4). Despite these important roles, relatively little is known about the regulation of expression of the calreticulin gene and how such regulation may affect cell function.
Pharmacologic depletion of ER Ca 2ϩ stores has been shown to up-regulate expression of calreticulin in HeLa and NIH 3T3 cells through a mechanism involving transcriptional activation (37)(38)(39). Transcription of the calreticulin gene is also activated by heat shock in several cell types (37,40,41). Redox mechanisms may be involved in calreticulin regulation, although whether oxidant stress results in decreased (42) or increased (43) expression appears to depend on the experimental system employed. Several examples of tissue-specific regulation of calreticulin expression have been reported. Activation of T lymphocytes by concanavalin A was associated with increased levels of calreticulin transcripts and protein (44). Ionizing radiation resulted in enhanced amounts of calreticulin in the nucleus of radio-resistant squamous carcinoma cells, but whether this was due to translocation from ER or increased biosynthesis was not clear (45). Calreticulin transcripts and protein levels in rat prostate and seminal vesicles were decreased by castration and restored in a tissue-specific manner by androgen replacement therapy (46).
Effects of cellular growth and differentiation programs on calreticulin expression, the focus of our current work, have been examined in only limited fashion. At the whole organism and organ level, there are examples of relatively high expression of calreticulin during development in several species, including plants (47), zebrafish (48), frog (49), and mouse (36,50). However, at the cellular level, the two model systems reported to date exhibited divergent patterns. In the myogenic cell line L6, it was determined that calreticulin was not regulated during differentiation to mature myotubes (51). In contrast, differentiation of NG-108-15 neuroblastoma-glioma cells was associated with increased calreticulin content, primarily due to enhanced protein stability (52).
The current report represents the most complete and systematic analysis to date of the effects of cell differentiation on rates of calreticulin transcription and translation, as well as the catabolism of calreticulin mRNA and protein. We made use of two well characterized human myeloid cell lines, HL-60 and PLB-985, and four differentiation-inducing agents, two of which lead to the granulocytic pathway and two of which lead to the monocytic pathway. Given the substantial evidence for important functional roles of calreticulin in myeloid cells (15,(23)(24)(25), these models are expected to be biologically relevant. Initially, we observed that myeloid differentiation led to consistent and major decreases in the cellular content of calreticulin protein. Pulse-labeling experiments then demonstrated that the decreased calreticulin levels could not be ascribed to an increased catabolic rate of the protein in differentiated cells. Interestingly, calreticulin proved to be an exceedingly stable protein, at least in the myeloid cell lines tested, a finding that is in keeping with earlier qualitative observations in a muscle cell line (53). In our quantitative studies the calculated fractional catabolic rate was only 13% per day. Completing our

FIG. 8. Immunoblot analysis of selected ER and cytosolic proteins in HL-60 cells during Me 2 SO-induced differentiation.
The experimental protocol was essentially as in Fig. 1, except that Me 2 SO treatment was extended to 6 days, and the amount of the total cell lysate loaded was varied to accommodate differences in the relative cellular abundance of the proteins and the titer of the primary antibodies. The amounts loaded were 0.25 ϫ 10 6 cell equivalents (CE) for calreticulin (CRT); 0.125 ϫ 10 6 CE for calnexin (CLNX); 0.025 ϫ 10 6 CE for protein-disulfide isomerase (PDI); 1.25 ϫ 10 6 CE for immunoglobulin-binding protein (BiP/GRP78); 0.5 ϫ 10 6 CE for p47 phox and p67 phox ; and 0.25 ϫ 10 6 CE for actin. Dilutions of primary and secondary antibodies were as described under "Experimental Procedures." Detection was carried out by chemiluminescence. The results shown are representative of those obtained in three independent experiments using identical conditions. analysis at the protein level were pulse-labeling studies that demonstrated dramatic decreases in the rate of calreticulin biosynthesis during myeloid cell differentiation. Comparison of the synthetic and catabolic rates indicated that the decline in steady-state calreticulin levels due to the rapid decrease in synthesis was partially buffered by slow catabolism. For example, at a time during differentiation when the biosynthetic rate was only about 10% of the base-line rate, the protein content was still maintained at about 40% of base-line levels.
Turning to analysis of calreticulin mRNA, we found that myeloid differentiation resulted in very rapid decreases in transcript levels, approximately in parallel to the observed decreases in synthesis of the protein. This effect did not represent a generalized decrease in mRNA levels, because we found concomitant increases in transcripts for the NADPH oxidase components p47 phox and p67 phox . Interestingly, as calreticulin transcripts decreased, those for the type I IP 3 receptor actually increased. Up-regulation of the IP 3 receptor under these circumstances, although well described previously (54,55), is curious, given the co-localization of the receptor with calreticulin in the ER Ca 2ϩ storage organelles and the shared functional participation of the two proteins in stimulus-mediated Ca 2ϩ release. The decrease in calreticulin mRNA during differentiation of myeloid cells could not be ascribed to increased catabolism of the transcripts. Indeed, calreticulin mRNA, as well as actin mRNA, exhibited an increased half-life in differentiated cells. Finally, we assessed transcriptional rates by nuclear run-on analysis. These experiments demonstrated that transcription of the calreticulin gene decreased by about 80% within 2 h of HL-60 cell exposure to the differentiation-inducing agent, Me 2 SO. Comparisons with other genes showed similar transcriptional down-regulation of myeloperoxidase versus up-regulation of p47 phox and little or no change in actin. It appears that the extremely rapid decrease in calreticulin transcriptional rate is partly offset by an increase in mRNA stability but that steady-state transcript levels decrease rather quickly once transcription is essentially shut down. Putting together the entire sequence of events for calreticulin from gene to protein, myeloid cell differentiation is associated with very rapid down-regulation of transcription, whereas the impacts on steady-state transcript and protein levels are modulated by increased transcript stability and constitutively high protein stability.
What are the functional consequences of altered levels of calreticulin expression? There are several reports (9, 10, 56, 57) documenting significant effects of calreticulin overexpression on Ca 2ϩ signaling and homeostasis. In general, these studies demonstrated increases in IP 3 -or thapsigargin-sensitive Ca 2ϩ stores, as well as decreased store-operated (capacitative) Ca 2ϩ entry via plasma membrane channels. Conversely, antisense oligonucleotide-mediated suppression of calreticulin expression resulted in a decrease in IP 3 -dependent release of Ca 2ϩ stores (58). Calreticulin-deficient mouse embryonic fibroblasts exhibited decreased total Ca 2ϩ stores and proportionately decreased release by thapsigargin, ionomycin, or IP 3 (11). Consistent with these studies, the decreased cellular content of calreticulin that we observed during the physiologic process of myeloid cell differentiation correlated with a commensurate decrease in the Ca 2ϩ stores. Yet despite this reduction, the residual Ca 2ϩ stores remained sensitive to thapsigargin, ionomycin, and IP 3 .
In addition to the differentiation-induced change in the size of the intracellular Ca 2ϩ pools that we have described here, there is evidence that myeloid cell differentiation is also accompanied by a qualitative change in the mechanisms that regulate agonist-mediated Ca 2ϩ signaling and homeostasis. Induction of differentiation in HL-60 cells was reported to result in a redistribution of the subtypes of both the IP 3 receptors that regulate the release of Ca 2ϩ from the internal stores and the SERCA proteins that are responsible for refilling the depleted stores (59). A steady decline in the expression of the dominant SERCA2b isoform was observed, together with a rapid increase of the SERCA PLIM (SERCA3b) isoform. SERCA3b has a lower affinity for Ca 2ϩ than SERCA2b and presumably displays slower kinetics in refilling Ca 2ϩ stores, resulting in a more sustained elevation of cytosolic Ca 2ϩ following stimulation. With respect to IP 3 receptors, differentiation of HL-60 cells resulted in increased expression of the type 2 receptor (IP 3 R-2) compared with either the type 1 or type 3 isoforms. IP 3 R-2 has a higher binding affinity for IP 3 than either IP 3 R-1 or IP 3 R-3 (60), and recent functional studies indicated that, unlike IP 3 R-1, IP 3 R-2 is not inhibited by high cytosolic concentrations of Ca 2ϩ (61). Thus, the increased expression of SERCA3b and IP 3 R-2 during myeloid differentiation would be expected to result in increased sensitivity to Ca 2ϩ -mobilizing agonists, together with a more sustained response. This is consistent with the well known properties of terminally differentiated mature granulocytes as highly reactive and motile cells that respond rapidly to external stimuli by chemotaxis, phagocytosis, and the release of enzymes and inflammatory mediators. The rate of reduction of SERCA2b in differentiating HL-60 cells reported earlier (59) is similar to the rate of reduction in the levels of calreticulin that we have described here. Because it has been shown that both calreticulin and calnexin influence SERCA2b activity (13,14), it may well be that these proteins form a functional unit and occupy the same ER sub-compartment. It is tempting to speculate that their reduced expression during differentiation reflects selective atrophy of this subcompartment that may account for at least part of the reduced total cell Ca 2ϩ content that we observed. In support of this possibility is evidence that the SERCA2b and SERCA3b Ca 2ϩ transport proteins are associated with different functional pools of intracellular Ca 2ϩ stores, the SERCA3b pool being more sensitive to release by IP 3 (62,63). Therefore, it is possible that the differentiation-induced reduction of SERCA2b may be accompanied by a reduction in the size of its associated Ca 2ϩ pool.
In addition to its Ca 2ϩ storage role, calreticulin functions as an ER molecular chaperone for nascent glycoproteins (16 -19). In myeloid cells, for example, the microbicidal protein myeloperoxidase interacts with calreticulin during its biosynthesis (15). The implications for protein synthesis and processing of the differentiation-induced decrease in calreticulin levels in myeloid cells are uncertain. Of note, a recent report by Nakamura and colleagues (11) suggests malfolding of the bradykinin receptor in calreticulin-deficient mouse fibroblasts, supporting the potential importance of our findings for chaperone function in myeloid cells. Clearly, protein synthesis continues during myeloid cell differentiation, albeit with an altered program of gene expression. The gp91 phox component of the phagocyte NADPH oxidase is a good example of a glycoprotein that is strongly up-regulated during differentiation, although whether it interacts with calreticulin has not been reported. Interestingly, it was found recently that the selection of a particular chaperone pathway by a nascent glycoprotein was partly dependent on the position of the N-linked glycan. If the glycosylation site was within ϳ50 residues of the N terminus, the glycoprotein interacted preferentially with calreticulin and calnexin. In contrast, nascent glycoproteins with N-linked glycans more distant from the N terminus interacted initially with BiP/GRP78, a member of the Hsp70 family, and only post-translationally with calreticulin and calnexin (64). In the case of gp91 phox , the N-linked glycosylation sites are all Ͼ90 residues from the N terminus, suggesting that it might interact preferentially with BiP/GRP78.
The selective down-regulation that we have demonstrated for calreticulin, calnexin, and ERp57 during cell differentiation may influence the pathways and efficiency of glycoprotein folding in the ER. Our ultrastructural analysis of differentiating HL-60 cells, as well as earlier studies of mature neutrophils (65), support the concept that major remodeling of the ER is an inherent feature of granulocytic differentiation. Whether this ER plasticity is accompanied by specific reprogramming of glycoprotein processing and how alternative chaperones such as BiP/GRP78 and protein-disulfide isomerase, both of which we found to be relatively preserved, may compensate for the losses of calreticulin, calnexin, and ERp57 require further study.