A calcium responsive element that regulates expression of two calcium binding proteins in Purkinje cells

MATERIALS AND METHODS

Parasagittal slices 250 μm in thickness were prepared from P11 to P12 mouse cerebellum using a tissue chopper (McIlwain, Mickle Laboratory Engineering, Surrey, United Kingdom). The slices then were cultured in the same manner as previously described (14) except that a serum-free medium consisting of Earle’s salts (GIBCO), basal medium eagle (BME) amino acids (Sigma), BME vitamins (GIBCO), 20 mM NaHCO₃, 10 mg/ml BSA (Sigma), 32 mM glucose (Sigma), 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml sodium selenite, 20 units/ml penicillin/streptomycin, and 2 mM l-glutamine was used. After 4 hr in culture, the slices were transfected using the biolistic particle delivery system (Bio-Rad) (14). Plasmids were precipitated onto 1.0 μm in diameter gold beads by placing them in a solution of 1.25 M CaCl₂ and 20 mM spermidine free base. The particle delivery system created a helium shock wave with a pressure gradient of 1,300 psi to accelerate the coated beads into a cerebellar slice that had been placed in a vacuum of 25 in of Hg (84 kPa). When cotransfected, all plasmids were mixed before precipitation onto the gold beads. Subsequent to transfection, slices were cultured for 48 hr, after which they were fixed and stained. Quantitations of expression levels were done using either immunocytochemistry or chromogenic color reactions. The latter were done as in ref. 14.

Immunocytochemistry was performed on the slices following procedures very similar to those in ref. 15. Anti-β-galactosidase Gal-13 clone (Sigma) at 1:100 and rabbit anti-alkaline phosphatase (Zymed) at 1:50 as well as fluorescein isothiocyanate-conjugated goat anti-rabbit at 1:300 and CY3-conjugated goat anti-mouse at 1:300 (Jackson ImmunoResearch) were used.

The calbindin, calmodulin, and major late promoter (pML) constructs were all made with the pnLacf vector (16), which contains LacZ fused to a nuclear localization sequence from the simian virus 40 large T antigen. The GW1-calbindin construct was made by ligating a rat cDNA clone of the calbindin D28k gene, which was a gift from M. Thomassett (Hôpital Robert Debre, Paris), into the GW1 expression vector, which was a gift from the British Biotechnology Corporation. Calretinin was amplified from a rat cDNA library and ligated into the GW1 expression vector in a similar manner to calbindin. All constructs were verified by cleavage with restriction enzymes and by sequencing. Cytomegalovirus (CMV) immediate early promoter driving alkaline phosphatase (CMVAP) was constructed by inserting the human placental alkaline phosphatase gene into the pCMVβ expression vector (Stratagene) in the NotI sites.

RESULTS

Cotransfection of CMVAP and the same promoter driving nuclear targeted β-galactosidase (CMVnLacZ) resulted in coexpression of these markers in virtually all transfected cells (Fig. 1A). In contrast, when CMVAP and ≈1 kb of 5′ flanking DNA from the calbindin gene driving nLacZ were cotransfected into the slices, only Purkinje cells expressed nLacZ (ref. 14, Fig. 1B).

After cotransfection with CMVAP and the 1-kb calbindin promoter driving nLacZ, Purkinje cells were found to express nLacZ at one of two levels: either at a relatively high level, or at such a low level as to be undetectable (Fig. 1C). This “all or nothing” pattern of expression of nLacZ in individual transfected cells was seen after each transfection in this study. It allowed simple quantitation of promoter activity as the percentage of transfected Purkinje cells expressing the nLacZ marker, and promoter specificity as the ratio of the promoter activity in transfected Purkinje cells versus that in transfected non-Purkinje cells.

The 1-kb calbindin promoter fragment drove expression of nLacZ in greater than 90% of the transfected Purkinje cells but in less than 2% of the transfected non-Purkinje cells (Fig. 2A). Thus, the activity of the 1-kb calbindin promoter in Purkinje cells is ≈90% and its ratio of specificity is at least 50-fold. The −155-bp calbindin promoter construct drove expression in 80% of the transfected Purkinje cells and in less than 3% of the transfected non-Purkinje cells (30-fold specificity). Thus, the vitamin D response element, which drives expression of the calbindin gene in non-CNS cells, is not required for Purkinje cell specific expression. The −85-bp calbindin promoter construct showed both high transcription activity (80%) and cell specificity (>50-fold). However, deletion of this promoter to −45 bp completely destroyed its activity (Fig. 2A). These results allow two important conclusions. First, it is evident that the sequence between −85 bp and −45 bp, termed the Purkinje cell element (PCE1), is required for Purkinje cell specific expression in situ because transcription activity was lost by deleting these sequences. Second, it is likely that the calbindin promoter does not contain elements that repress transcription in non-Purkinje cells in situ, because the cell specificity of the calbindin promoter was retained in all active constructs.

PCE1 was found to contain three repeats of a GA-rich 10-bp motif. Because transcriptional regulation often involves relatively simple DNA elements that are repeated in the 5′ flanking regions of specific genes, we tested whether these repeats are essential for Purkinje cell specific expression of the calbindin promoter. When two of the repeats were mutated, the 1-kb promoter could no longer mediate transcription (Fig. 2A). These results clearly demonstrate that PCE1 is required for calbindin transcription in situ.

To determine whether PCE1 is sufficient to drive Purkinje cell specific expression, a single copy was inserted 20 bp upstream from the TATA box of the minimal adenovirus major late promoter (pML) and assayed in slice preparations. The pML has been extensively used as a minimal promoter containing all of the basal elements required for transcription initiation in vitro but none that can program efficient or cell-specific transcription in vivo (17). This was confirmed in cerebellar slice preparations, because pML was not expressed in any cell type after biolistic transformation (Fig. 2B). In contrast, pML plus PCE1 (PCE1/pML) caused expression in 54% of the transfected Purkinje cells and <4% of the non-Purkinje cells (Fig. 2 B and C). Thus, PCE1 conferred both high activity (54%) and cell specificity (≈20-fold) to pML. These results are comparable to the activity of other well characterized transcriptional regulatory elements, allowing us to conclude that PCE1 is both necessary and sufficient for Purkinje cell specific expression in organotypic slices of P12 cerebellum.

PCE1 appears to be novel, because it does not match any of the known transcription factor binding sites (18). However, a sequence containing three 10 nucleotide repeats very similar to those of PCE1 (referred to as PCE1*) was found in the 5′ flanking region of calmodulin II. Like calbindin, calmodulin II is a Ca²⁺ binding protein that is expressed abundantly in Purkinje cells (19), although unlike calbindin, it can activate enzymes and modulate activity of ion channels when it is in a calcium-bound state (6). Furthermore, it has been shown that 360 bp of DNA upstream from the calmodulin II gene, which includes PCE1*, can mediate expression in cerebellar Purkinje cells in transgenic mice (20).

Three-hundred-thirty bp of 5′ flanking DNA of calmodulin II was found to cause expression in 82% of the transfected Purkinje cells and in 22% of the transfected non-Purkinje cells (Fig. 3). Thus, this promoter is approximately equally active for Purkinje cell transcription, but much less specific than the calbindin promoter (≈4-fold vs. >50-fold). This is in agreement with the in vivo expression pattern of calmodulin II, which is expressed strongly in cerebellar Purkinje cells and at much lower levels in other cerebellar cell types (20). When PCE1* was mutated in the context of the 330-bp promoter fragment (Fig. 3) it caused expression in 14% of both Purkinje and non-Purkinje cells. Thus, PCE1* is required for the Purkinje cell specificity of the calmodulin II promoter in situ, and it has very little effect on transcription of this gene in other cell types. Apparently, a second element that is not cell specific is responsible for the residual activity of this promoter. While we have not mapped this element, it has been reported that sequences between −280 bp and −260 bp are required for the activity of the calmodulin II promoter in 3T3 cells (20), suggesting that the same region is responsible for the remaining transcriptional activity of this promoter after mutation of PCE1. Finally, when the PCE1* was inserted upstream of pML, this construct was expressed in ≈40% of the transfected Purkinje cells and was not expressed in other cerebellar cell types (Fig. 3). We conclude, therefore, that PCE1* also acts as a Purkinje cell specific transcription element in the calmodulin II promoter.

PCE1 is conserved in the calbindin and calmodulin II promoters from different species, which is consistent with PCE1 being important for transcriptional control (Fig. 4). It was found that the 10-nucleotide repeats from the calbindin and calmodulin promoters of different species have the consensus sequence 5′-GAGGG(G/A)GGAG-3′, with the two nucleotides in the fifth and sixth positions more variable than the others. In addition, the first four nucleotides of the most 5′ repeat and the last four nucleotides of the most 3′ repeat tended to show more variation (Fig. 4). In both the mouse calbindin and calmodulin II promoters the three 10-bp repeats are arranged so that they are on the same face of the double helix as the TATA box, although there is 20 bp of sequence between the TATA box and the calbindin repeats, whereas the calmodulin II elements are immediately adjacent to this sequence (Fig. 4). This spacing also is conserved between different species, indicating that it may be important for the function of PCE1. It is for this reason that in the constructs PCE1/pML and PCE1*/pML, the elements were placed exactly 20 nucleotides upstream of the TATA box of pML. It is interesting to note that the 10 bp of sequence in each promoter that separates the first two repeats is different between calbindin and calmodulin II, however it is conserved between different species, indicating that it may have functional significance. Also note that the second repeat is not present in the dog and human calmodulin II promoters despite the fact that the first, the third, and the intervening sequence are preserved (Fig. 4).

To see whether Ca²⁺ could modulate PCE1-mediated transcription, the ability of two calbindin promoter constructs to mediate transcription in the context of different levels of intracellular Ca²⁺ was assessed. When the intracellular free Ca²⁺ concentration was increased by supplementing the medium with 1 μM calcium ionophore ionomycin, a reproducible increase in PCE1 mediated Purkinje cell-specific expression was observed without an associated increase in the intensity of pCMV-mediated transcription. It was found in three experiments that PCE1/pML expressed in ≈50% more transfected Purkinje cells when ionomycin was added to the medium (Fig. 5A), a difference that is statistically significant (P < 0.01, t test). Consequently, the presence of ionomycin can increase the activity of PCE1/pML to roughly the level of the native 1-kb calbindin promoter, consistent with the hypothesis that increasing free Ca²⁺ can increase the efficiency with which PCE1 mediates transcription. The addition of ionomycin did not appreciably change the expression level in non-Purkinje cells (5% vs. 3%) (Fig. 5A), indicating that while Purkinje cell specific expression was significantly increased, nonspecific expression levels remained very low.

If PCE1-mediated transcription is Ca²⁺ sensitive, we reasoned that lowering the intracellular level of free Ca²⁺ should decrease PCE1 activity. To investigate this, we took advantage of the demonstrated Ca²⁺ buffering capacity of the calbindin and calretinin proteins and the ability to achieve nearly 100% efficiency of cotransfection of multiple DNA constructs using the biolistic methodology. In these experiments the biolistic carriers were coated with three different DNAs: the CMVAP marker gene to identify transfected cells; the test promoter construct to measure Purkinje cell specific transcription mediated by PCE1; and either an empty CMV vector control, or the CMV vector driving expression of the calbindin or calretinin proteins to lower free intracellular Ca²⁺ in the transfected cells. This experimental approach allowed us to test the effects of buffering free Ca²⁺ in individual cells rather than trying to alter intracellular Ca²⁺ levels using pharmacologic agents to treat the entire organotypic slice preparation. As shown in Fig. 5B, cotransfection of the empty vector (GW1) with either the 85-bp calbindin promoter construct (80%) or the PCE1/pML construct (50%) did not change their activity from that observed in the absence of the CMV empty vector (see also Fig. 2 A and B). In contrast, when the 85-bp calbindin promoter was cotransfected with the CMV promoter driving expression of calbindin (GW1-calbindin) (30%) or calretinin (GW1-Calretinin) (25%) there was a dramatic decrease in its activity (Fig. 5B). A similar ≈2.5-fold decrease in activity was observed when these Ca²⁺ buffering proteins were cotransfected with the PCE1/pML test promoter (Fig. 5B). These results demonstrate that transcription via PCE1 is sensitive to free intracellular Ca²⁺ concentrations, suggesting that the regulation of the Ca²⁺ buffering capacity of cerebellar Purkinje cells is at least partially regulated at the transcriptional level.

DISCUSSION

The identification of PCE1 provides an example of a transcriptional regulatory element that is both necessary and sufficient for expression in a specific CNS neuron in situ. Previously, the only precisely defined DNA sequence elements mediating neural specific gene expression were the inhibitory elements, RE-1 or NRSE 1, which were originally identified in the promoters of the type II sodium channel (22) and SCG10 (23), respectively. In combination with activator elements, repressor elements similar to RE-1 and NRSE 1 have been shown to cause neuronal-specific expression of numerous genes (24–26). The fact that PCE1 can mediate Purkinje cell-specific expression in the absence of any silencer elements indicates that it is part of a mechanism for achieving neuron-specific gene expression that is different from the mechanism involving RE-1 or NRSE 1.

Although PCE1 was found to mediate Purkinje cell-specific expression in both the calbindin and calmodulin II promoters, it is not involved in all cases of Purkinje cell specific expression. For example, PCE1 is not found in the promoters of either calmodulin I or calmodulin III. While this may seem unexpected because all three isoforms of calmodulin have similar expression patterns and, specifically, all three express at high levels in Purkinje cells, previous experiments have suggested that expression of these three isoforms is mediated through different mechanisms. It was shown that in transgenic mice, expression in Purkinje cells could be driven both by 360 bp of 5′ flanking DNA of the calmodulin II promoter (20) and 980 bp of 5′ flanking DNA of the calmodulin III promoter (27); however, there is no sequence identity between these two promoters or with that of calmodulin I. PCE1 is also absent from the 5′ flanking DNA of L7 (28) and of aldolase C (Zebrin II) (29), which is not surprising because the expression patterns of the two genes differ from that of calbindin both spatially and temporally. Specifically, calbindin is expressed in all Purkinje cells both during development and through adulthood, whereas L7 is expressed only in subsets of Purkinje cells during development (30) and Zebrin II is expressed in subsets of Purkinje cells through adulthood (31).

Because the calbindin and calmodulin II genes are expressed early and abundantly in Purkinje cells, it is probable that PCE1 and its cognate transcription factor(s) are critical to the function of these neurons. Identification of PCE1 provides an avenue for identification of this transcription factor and investigation of its biological role in Purkinje cell development. Furthermore, it is likely that the PCE1 pathway plays an important role in several other neuronal cell types. For example, both calbindin and calmodulin II are expressed at very high levels in dentate gyrus granule cells (20), and we expect that PCE1 also will be involved in directing expression of these genes in the hippocampus. On the other hand, both of these genes are expressed elsewhere in the CNS, and calbindin is expressed in various places outside the brain. Our results indicate that sequences controlling expression at these sites are distinct from PCE1.

The facts that PCE1 is required for the expression of calbindin, which participates in homeostatic control of intracellular Ca²⁺, and that the activity of this element is sensitive to intracellular Ca²⁺ levels, strongly suggest that PCE1 is part of a pathway that determines the Ca²⁺ buffering capacity of cerebellar Purkinje cells. Two properties of Ca²⁺-sensitive transcription via PCE1 suggest that regulation of the Ca²⁺ buffering capacity of Purkinje cells is distinct from these previously characterized Ca²⁺-sensitive transcriptional control mechanisms. First, PCE1 bears no significant sequence similarity to either the serum response element or cyclic AMP response element/calcium response element (Fig. 4; ref. 32). Given the strong conservation of both of these previously characterized calcium response elements in target genes, and the high homology of PCE1 across species in both the calbindin and calmodulin II promoters (Fig. 4), it is very unlikely that either serum response factor (SRF) or cyclic AMP response element binding factor (CREB) plays a role in transcriptional activation through PCE1. Second, unlike the SRF and CREB pathways, which are active in many different cell types in many different physiologic situations, the activity of PCE1 is cell specific. It seems probable, therefore, that PCE1 is a component of a novel Ca²⁺-sensitive signaling mechanism. Identification of transcription factors that operate through PCE1 and direct measurement of their sensitivity to free intracellular Ca²⁺ levels will be a critical next step in the analysis of this pathway.