Expression of endomembrane calcium pumps in colon and gastric cancer cells. Induction of SERCA3 expression during differentiation.

Calcium mobilization from the endoplasmic reticulum (ER) into the cytosol is a key component of several signaling networks controlling tumor cell growth, differentiation, or apoptosis. Sarco/endoplasmic reticulum calcium transport ATPases (SERCA-type calcium pumps), enzymes that accumulate calcium in the ER, play an important role in these phenomena. We report that SERCA3 expression is significantly reduced or lost in colon carcinomas when compared with normal colonic epithelial cells, which express this enzyme at a high level. To study the involvement of SERCA enzymes in differentiation, in this work differentiation of colon and gastric cancer cell lines was initiated, and the change in the expression of SERCA isoenzymes as well as intracellular calcium levels were investigated. Treatment of the tumor cells with butyrate or other established differentiation inducing agents resulted in a marked and specific induction of the expression of SERCA3, whereas the expression of the ubiquitous SERCA2 enzymes did not change significantly or was reduced. A similar marked increase in SERCA3 expression was found during spontaneous differentiation of post-confluent Caco-2 cells, and this closely correlated with the induction of other known markers of differentiation. Analysis of the expression of the SERCA3 alternative splice isoforms revealed induction of all three known iso-SERCA3 variants (3a, 3b, and 3c). Butyrate treatment of the KATO-III gastric cancer cells led to higher resting cytosolic calcium concentrations and, in accordance with the lower calcium affinity of SERCA3, to diminished ER calcium content. These data taken together indicate a defect in SERCA3 expression in colon cancers as compared with normal colonic epithelium, show that the calcium homeostasis of the endoplasmic reticulum may be remodeled during cellular differentiation, and indicate that SERCA3 constitutes an interesting new differentiation marker that may prove useful for the analysis of the phenotype of gastrointestinal adenocarcinomas.


Introduction
Cellular calcium concentration gradients and calcium ion fluxes are important components of several signaling networks controlling cell growth, differentiation or apoptosis (1)(2). In a resting cell, the cytosolic free calcium concentration is approximately 50-100 nM, whereas the endoplasmic reticulum (ER) or the extracellular medium contains calcium in the high micromolar to low millimolar range. Binding of several growth factors, hormones, chemokines or bioactive peptides to their cell surface receptors leads to the formation of the second messenger inositol-1,4,5-trisphosphate (IP3), which induces calcium release from the endoplasmic reticulum into the cytosol through IP3 receptor calcium channels. The ensuing decrease of the calcium content of the ER lumen induces the opening of calcium channels in the plasma membrane, allowing calcium influx into the cytosol from the extracellular space.
Calcium release from the ER and ensuing calcium influx lead to the augmentation of the cytosolic free calcium concentration. As many key components of intracellular signaling networks, such as various calmodulin-activated kinases (3), protein kinase C (4), calcineurin (5), calpains (6), as well as the PYK-2 tyrosine-kinase (7), the Ras guanine nucleotide exchange factor Ras-GRF (8), the apoptosis-associated kinase DAP-2 (9) or the apoptosislinked calcium-binding protein ALG-2 (10) are directly activated by increased cytosolic calcium concentrations, cellular calcium fluxes constitute an important component of several signal transduction networks of the cell.
In addition to its role played in signaling in the cytosolic compartment, calcium stored whithin the ER lumen is required for the posttranslational modification and processing of by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 5 newly synthesized proteins transiting across the organelle (11). It is becoming increasingly clear that calcium stored in the ER is involved in homeostatic, synthetic as well as signaling functions also within the lumen of the organelle (12)(13)(14).
Refilling of calcium into the ER from the cytosol by active, ATP-driven ion transport is assured by Sarco-Endoplasmic Reticulum Calcium transport ATPases, also called SERCA enzymes (12). These enzymes, by pumping calcium into the ER against a steep concentration gradient, decrease cytosolic calcium levels after an episode of activation, and make calcium available in the ER lumen for intra-ER calcium dependent functions, as well as for being released into the cytosol during a next signaling event. Three SERCA genes are known, which by alternative splicing can give rise to several protein isoforms. SERCA1a and 1b are expressed in adult and neonatal skeletal muscle, respectively (15). SERCA2a is found in cardiac and smooth muscle, whereas SERCA2b has been found in all non-muscle cell types studied so far (16,17). Expression of SERCA3 has been detected in a selected group of cell types, including cells of hematopoietic origin, where this enzyme is constitutively expressed (18)(19)(20)(21), and the existence of three SERCA3 alternative splice isoforms has been reported (22)(23)(24). Although SERCA3 mRNA has been detected in various tissues, including normal intestinal epithelium (25,26) the expression of SERCA-type enzymes has not been studied in colon and gastric cancer so far.
The calcium content of the ER lumen is a key determinant controlling apoptosis induced by physiologic stimuli (14,27). The modulation of the calcium content of the ER by Bcl-2 is involved in the regulation of the apoptotic potential of the cell (14,28), and the regulation of SERCA expression and function by Bcl-2 is thought to be involved in this process (29). In short term experimental settings the direct pharmacological inhibition of calcium pumping activity has been shown to lead to growth arrest, differentiation or caspase-12-dependent apoptosis, depending on the cell type (30)(31)(32)(33), and highly specific SERCA by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 6 inhibitors such as thapsigargin or 2,5-di-tert-butyl-1,4-hydroquinone are known tumor promoters in vivo, as well as in vitro (34,35) when applied chronically. Moreover, endogenously expressed truncated SERCA variants have recently been implicated in the modulation of the apoptotic potential of the cell by interfering with SERCA-dependent calcium transport (36). In addition, peptide hormone receptors that mobilize calcium from the endoplasmic reticulum have been shown to be involved in positive feedback mechanisms regulating colon cancer cell proliferation and behavior (37,38). All these data taken together suggest that cellular phenotype, proliferation status, apoptotic potential and stage of differentiation are intricately connected to ER calcium homeostasis. However, the mechanisms involved in these processes are poorly understood. In order to better understand the role of SERCA enzymes in epithelial maturation, and to shed light on the involvement of the calcium homeostasis of the ER in epithelial malignancies, in this work we investigated the expression of SERCA enzymes in a series of human colon and gastric cancer cell lines, carcinoma tissue and primary cells, and studied the modulation of SERCA expression and function during cell differentiation.

Materials and Methods.
Cells and treatments.
Exponentially growing cells were trypsinized and seeded into 20 cm 2 cell culture dishes at a density of 2x10 4 cells/cm 2 . When cells reached 80% confluency by microscopic examination (day 2 or 3 post-plating, depending on the rate of growth), medium was renewed and drugs were added from concentrated stock solutions. Cells that grow in suspension or in a semiadherent manner (KATO-III, NCI-SNU-1, NCI-SNU-16, RF-48) were seeded at an initial density of 2x10 5 cells/ml at the beginning of treatments. Sodium salts of short chain fatty acids and of their analogs were dissolved in phosphate buffered saline at a concentration of 0.3 M. When only the free acid forms were available commercially (Sigma-Aldrich, France), these were neutralized by dissolving in 0.3 M sodium bicarbonate at the same concentration and sterile filtered. Due to their hydrophobicity, 1,2,3-tributyrlyl-glycerol (tributyrin, Fluka, Germany) and pyvaloyloxy-methylbutyrate (Calbiochem, Darmstadt, Germany) were dispersed to the desired final concentrations as fine emulsions in complete medium by sonication immediately prior experiments. Suberoylanilide hydroxamic acid (SAHA) was purchased from Upstate Biotechnology (Lake Placid, NY), and apicidin from Calbiochem (Darmstadt, Germany). Herbimycin-A, HC-toxin and thapsigargin were from Sigma Aldrich France. Drugs were added to the cell cultures from concentrated stock solutions made in dimethyl sulfoxyde (DMSO). The final concentration of DMSO vehicle did not exceed 0.1%, was included in control experiments and did not interfere with the assays.
Untreated control cells were harvested in the exponential phase of non-confluent growth.
After treatments as indicated on figures, the cells were quickly washed with ice cold NaCl (150 mM) twice, precipitated with 5% trichloroacetic acid overnight at 4°C and Rabbit polyclonal antibodies used in this work that specifically recognize the various SERCA3 isoenzymes (SERCA3a, 3b and 3c) have been characterized previously in detail (24). Immunostaining for the detection of carcinoembryonic antigen and of dipeptidyl peptidase IV was performed using the C6G9 (Sigma) and the HBB 3/775/42 (43) monoclonal antibodies, respectively, at a 1000-fold dilution of the ascites using the electrophoresis and immunostaining system as outlined above. ZO-1 protein was immunostained using an affinity-purified rabbit polyclonal antibody obtained from Zymed (CA) that recognizes both the αand α+ isoforms, after electrophoresis of cellular proteins in 6.5% polyacrylamide gels.
Luminescent signal obtained using the Amersham Enhanced Chemiluminescene system was quantified by scanning non-saturated luminograms (on Kodak Biomax ML films) with an Epson Perfection Photo 1240U scanner using the Adobe Photoshop software (Adobe Systems Inc., Mountain View, CA) and quantitated using the Scion Image software (version 4.0.2, Scion Corporation, www.scioncorp.com). Due to the absence of detectable SERCA3 in untreated KATO-III or Caco-2 cells, SERCA expression in these cells is expressed on figures in percentages, with endpoint signal being taken arbitrarily as 100%. 9 Primary cells.
Primary colon cancer cells were obtained from fresh surgical specimens. Homogenous tumor tissue was carefully separated from adjacent structures, cut into sub-millimeter sized pieces with a scalpel and placed into 24 well plates in a medium consisting of a mixture of equal volumes of RPMI-1640 and Ham's F12 nutrient medium supplemented with glutamax-I, glutamine, sodium pyruvate, nonessential amino acids, vitamins and reduced glutathione (all reagents obtained from Gibco-Life Technologies, France) plus 20% decomplemented fetal calf serum, 50 U/ml penicillin, 50 µg/ml streptomycin and 2.5 µg/ml amphotericin-B. After 3 days in culture, floating tissue debris were aspirated, and growth of adherent cells and morphology, as well as the absence of fibroblasts or microbial contamination was monitored microscopically. Only cultures devoid of fibroblast contamination were used. CEA synthesis was detected by immunoblotting of total cell lysates as indicated above. The cells were first treated with short chain fatty acids at day 6 post-plating in the wells in which they had been originally plated, and then after four months of continuous growth and regular subculturing in 20 cm 2 Petri dishes.

Immunohistochemistry.
Staining of 5 µm thick cryostat sections of freshly frozen normal and malignant colon and stomach tissue with the SERCA3 specific PLIM430 monoclonal antibody was performed as follows: slides were allowed to dry overnigh at room temperature at the sections were then fixed in acetone for ten minutes at room temperature, were allowed to dry and were rehydrated in Tris-buffered saline (TBS, pH=7.4) containing 0.1% Tween-20 (TBS-Tween) for ten minutes. Inhibition of nonspecific protein binding was performed by incubation for 30 min. in TBS-Tween supplemented with 5% nonfat dry milk. The PLIM430 antibody, previously purified by protein-A affinity chromatography, was then applied upon the sections 10 in the above solution at 1 µg/ml concentration and incubated at room temperature for 90 minutes. The slides were then rinced with distilled water three times and incubated in TBS-Tween-milk for ten minutes. After repeating this washing step once, the slides were rinsed three times with TBS-Tween, and incubated for ten min. in TBS containing 1/30 vol. normal horse serum. Slides were incubated with biotinylated anti-mouse secondary antibody (Vectastain ABC kit, Vector Laboratories, CA) for one hour, followed by incubation with avidin-biotin-peroxidase complex (Vectastain ABC kit) for 45 min. according to the protocol of the manufacturer. As chromogen 3,3'-diaminobenzidine was used and the slides were performed for ten cycles with an annealing temperature decrement from 65°C to 56°C in order to increase the specificity of priming during initial cycles of amplification. PCR was then conducted essentially as described (45) (45). The amplification products were separated in 1.5% agarose gels and visualized by ethidium bromide staining. The apparent molecular masses of the PCR products corresponded to those calculated based on the reported sequences, moreover the identity of the PCR products was also confirmed by direct sequencing (Genome Express, Grenoble, France). Quantitative data were obtained using the Scion Image software (see above) on digitalized images of ethidium bromide stained gels. Data in this work correspond to at least three independent experiments and are presented as means +/-S.E.

SERCA3 expression is lost in colon carcinomas.
In order to study SERCA3 expression in human tissue, we developed an immunohistochemical staining protocol using the SERCA3-specific PLIM430 monoclonal antibody. As illustrated in Fig. 1, Panel A SERCA3 was readily detected in normal colonic crypt epithelium in enterocytic, as well as mucus secreting cells, as well as in stomach mucosa (Panel B). In colonic crypts stronger SERCA3 staining could be seen in more mature cells residing in the luminal region. However, SERCA3 could also be detected in deeply located, less mature cells, suggesting that SERCA3 expression is induced early in colonic epithelial differentiation. On the other hand, when SERCA3 expression was investigated in colon cancer tissue, in 9 cases out of 12 examined, a complete lack of staining in the malignant cells was observed (Panels C-G), and in the remaining cases a faint staining could be observed in the apical region of the cells (Panel H). At the same time, adjacent normal epithelial tissue stained strongly positive for SERCA3 in all specimens in a highly reproducible manner.
These data show that SERCA3 expression is dramatically decreased or completely lost in colon adenocarcinomas, although present at high levels in normal epithelium.
In accordance with immunoblotting data presented in Figure 2, SERCA3 expression could also be detected by immunocytochemistry in the COLO-205 cell line (Panel J), whereas the KATO-III cell line was negative (Panel I). SERCA3 expression in COLO-205 cells was, however, markedly weaker than that observed in normal colonic epithelial cells. 13 SERCA3 expression in colon and gastric cancer cell lines.
In order to study SERCA3 expression in various colon and gastric cancer cell lines, cells in the exponential phase of growth were harvested, and total cellular protein was probed with the PLIM430 (pan-SERCA3 specific) monoclonal antibody in a Western blot format as in (20), using a very sensitive chemiluminescent detection method. As an internal control, SERCA2 was detected using the IID8 monoclonal antibody as well. As shown in As shown in Figure 6 Panel A, other short chain fatty acids such as propionate, valerate, isobutyrate, isovalerate, caproate and valproate also induced SERCA3 expression, as well as the butyrate generating prodrugs tributyrin (tributyryl-glycerol) and pivaloyloxymethyl butyrate (49)(50)(51). Other histone-deacetylase inhibitors such as HC-toxin, apicidin and SAHA also induced SERCA3 expression, although, in accordance with the literature (52-54) to a lesser extent than butyrate. On the other hand, and similarly to previous data, acetate, and other chemicals structurally related to butyrate, such as, crotonate, cyclopropane-carboxylate, pentenoate, pentinoate, trimethylacetate were only marginally active or had no effect, and heptafluorobutyrate, pivaloate, γ-aminobutyrate or lactate were inactive (not shown). Aryl-derivatives such as 3-phenylpropionate and 4-phenylbutyrate also 15 induced SERCA3 expression, whereas trans-cinnamate or phenoxyacetate were without effect (not shown). These observations are in agreement with previous data in the literature regarding the pharmacological profile of these molecules (55,56), and with the recently established three-dimensional tube-like pocket structure of the active site of histonedeacetylase where inhibitors bind (57).
Interestingly, as shown in Figure 6  When SERCA expression of post-confluent Caco-2 was analyzed, a marked induction of SERCA3 expression was seen (Figure 9, Panel A). While the expression of SERCA3 was undetectable in exponentially growing non-confluent and early post-confluent cultures, SERCA3 expression was manifest from day 5-6 post-confluency, and reached a plateau at day 20, while SERCA2 expression was only slightly increased. The induction of SERCA3 expression followed a time course very similar to that of carcinoembryonic antigen, a widely used marker of differentiation of this cell line (61). In addition, during this process the expression of dipeptidyl peptidase IV, another marker of differentiation (62) was induced, and the isoform switch from αto α+ of the tight junction protein ZO-1 was seen as described earlier (63,64).
SERCA expression was also investigated in HT29-5M21 cells, a well characterized, methothrexate-resistant clone of HT29. Although non-differentiated in the exponential phase of growth, in post-confluent culture these cells display a well differentiated, goblet cell-like, mucus secreting phenotype (39). As shown in Figure 9, Panel B, SERCA3 expression was induced during the differentiation of post-confluent HT29-5M21 cells.

SERCA3 isoenzymes.
Recent data in the literature, including ours, indicate that the SERCA3 gene can give rise by alternative splicing in the 3' region of the primary transcript to three mRNA isoforms, SERCA3a, b and c, coding for proteins that carry unique peptide sequences in their Cterminal region (22)(23)(24). In order to investigate SERCA3 expression on the isoform level, we performed semi-quantitative RT-PCR experiments using oligonucleotide primers that allow the specific amplification of the various SERCA3 transcripts (45), and studied the expression of the corresponding protein isoforms using recently developed antipeptide antibodies (24) that recognize unique peptide sequences in SERCA3a, b and c, respectively. Figure 10 18 and protein expression levels followed thereafter a plateau-type time course. The markedly earlier detection of SERCA3 mRNA is probably due to the higher sensitivity of RT-PCR when compared to that of Western blotting. In addition, as post-confluent differentiation of Caco-2 cells is accompanied by growth arrest, the plateau-type kinetics of the accumulation of SERCA3 protein is compatible with transitory induction of corresponding mRNA, assuming that SERCA3 protein is more stable in these cells than the corresponding mRNA.

As shown in
Calcium homeostasis of differentiating KATO-III cells.
In order to study cellular calcium homeostasis in functional terms during drug-induced cell differentiation, untreated, as well as butyrate treated KATO-III cells were loaded with Fura-2 and analyzed by calcium spectrofluorimetry. Resting cytosolic calcium levels and calcium mobilization from the ER into the cytosol upon complete inhibition of cellular SERCA activity using supramaximal concentrations of the specific SERCA inhibitor thapsigargin (32,65) were quantified. As shown in Figure 11, the resting cytosolic calcium concentration of butyrate treated cells was significantly higher than that of controls (Panel A), whereas the amount of calcium released from the ER into the cytosol upon SERCA inhibition by thapsigargin (Panel B) was decreased in butyrate treated cells, indicating decreased calcium storage in the ER. These data show, that the calcium homeostasis of the KATO-III cell line undergoes a significant remodeling during drug-induced differentiation. Short chain fatty acids are produced by the fermentation of dietary fibers by the colonic flora, and the induction of the differentiation followed by apoptosis, of microscopic precancerous lesions by these molecules (and in particular by butyrate) is considered as being a main mechanism of the protective effect of a fiber-rich diet against colorectal cancer (67,68). From these data it is tempting to speculate, that cellular calcium homeostasis may be modulated by short chain fatty acids in the colonic epithelium. In addition, our data suggest that this effect may also operate in the case of gastric cancer, as well.

Discussion
Inhibition of histone-deacetylases is a key component of the mechanism of action of differentiation induction by butyrate (69). Highly specific histone-deacetylase inhibitors such as HC-toxin, apicidin or SAHA also induced SERCA expression in our hands, indicating that this mechanism may be involved in the modulation of SERCA3 expression by short chain fatty acids. Butyrate appeared, however, to be a more potent inducer of SERCA3 expression. This is in accordance with data in the literature obtained in various experimental systems (52)(53)(54), and may be due to differences in stability of the drugs in cell culture conditions, and to that in addition to histone-deacetylase inhibition, butyrate and other SCFAs also interact with other intracellular targets as well (70). The functional implications of the modulation of SERCA expression during epithelial differentiation are complex. The pattern of a calcium signal is shaped by the concerted and coordinated action of mechanisms that increase cytosolic calcium levels (i.e. calcium channels) and that decrease it (calcium pumps), leading in many instances to oscillatory calcium signals. SERCA enzymes are actively resequestering calcium into the endoplasmic reticulum even during calcium mobilization from this organelle, and thus clearly contribute to the shaping of calcium transients (71,72). The various geometrical characteristics of a calcium transient, and the frequency and amplitude of repetitive calcium oscillations convey key information to calcium-activated intracellular targets. The modulation by the cell of the spatiotemporal characteristics of calcium oscillations confers specificity and selectivity to calcium signals, because calcium activated target molecules such as calmodulin-dependent protein kinases, calcineurin or protein kinase C isoenzymes are optimally activated at distinct frequencies and amplitudes of calcium oscillations (73)(74)(75). This can lead to differential activation of transcription factors such as NF-AT, NF-κB and others, leading to the modulation of gene expression by calcium oscillations (76)(77)(78). The biochemical characteristics of SERCA2b and of SERCA3 enzymes are distinct (79). In particular, the calcium affinity of SERCA3 is lower (K Ca = 1.2 µM) than that of SERCA2b (K Ca = 0.2 µM).
Quantitative changes of the relative abundance of various SERCA isoenzymes thus may alter resting cytosolic calcium levels, and may modify the shape of a calcium transient and of    Expression of SERCA-type calcium pumps in colon-and gastric cancer cell lines.        Induction of SERCA3 expression in freshly isolated primary colon cancer cells.
Cells at an early stage following plating (day 6, left) and after four months of continuous growth and subculturing (right) were treated with 3 mM butyrate or valerate for 5 days. In both cases a strong induction of SERCA3 expression was obtained by both SCFA.

Panel B).
Resting cytosolic calcium concentration was significantly higher in butyrate treated cells than in controls, and the amount of calcium stored in intracellular pools that could be released into the cytosol by thapsigargin was significantly decreased in butyrate treated cells.
. by guest on March 24, 2020