Abstract
Responsiveness of genes to steroid hormones is a complex process involving synergistic and/or antagonistic interactions between specific receptors and other nonreceptor transcription factors. Thus, DNA recognition elements for steroid hormone receptors are often located among binding sites for other trans-acting factors. The hormonal form of vitamin D, 1,25-dihydroxyvitamin D3, stimulates transcription of the tissue-specific osteocalcin (OC) gene in osteoblastic cells. The rat OC vitamin D response element contains an internal acitvating protein-1 (AP-1) site. Here, we report for the first time that this AP-1 site is critical for the transcriptional enhancement of rat osteocalcin gene expression mediated by vitamin D. Precise mutations were introduced either in the steroid half-elements or in the internal AP-1 sequences. One mutation within the internal AP-1 site retained vitamin D receptor/retinoid X receptor binding equivalent to that of the wild-type sequence, but resulted in complete loss of vitamin D inducibility of the OC promoter. These results suggest a functional interaction between the hormone receptor and nuclear oncoproteins at the rat OC vitamin D response element. This cooperation of activities may have important consequences in physiological regulation of osteocalcin transcription during osteoblast differentiation and bone tissue development in vivo.
STEROID hormones regulate eukaryotic development, differentiation, and cellular homeostasis through the transcriptional control of tissue-specific genes. This control is mediated by intracellular receptor proteins. The ligand-activated receptors act as transcriptional activators or repressors through direct interaction with DNA cognate motifs within promoters of responsive genes. The presence of activating protein-1 (AP-1) sites within or in the close proximity of a variety of hormone-response elements suggests either inhibitory or synergistic interactions between steroid hormone receptors and AP-1 members. The AP-1 class of transcription factors can inhibit the activation of the receptors for glucocorticoid, progesterone, estrogen, androgen, thyroid, and retinoids (1–7). In addition, cooperative interactions between steroid hormone receptors and AP-1 members have been observed for the ovalbumin gene (8), the proliferin promoter (9), and the neurotensin/neuromedin promoter (10). Inhibitory or synergistic effects of AP-1 on steroid hormone responsiveness can occur by direct protein/protein interaction between Fos/Jun proteins and steroid hormone receptors as well as by protein/DNA interactions involving adjacent or overlapping recognition motifs of AP-1 and steroid hormone receptors.
The bone-specific osteocalcin (OC) gene promoter is modularly organized and exhibits both positive and negative regulatory elements (11). AP-1 sites in the rat osteocalcin promoter have been found within or in the vicinity of other regulatory elements of the rat OC gene, including OC box 1 at nucleotides (nt) −100 to −77 controlling basal OC expression (12), the transforming growth factor-β response element at nt −146 to −139 (13), the distal glucocorticoid response element at nt −769 to −763 (14), and at nt −469 to −462 within the vitamin D responsive element (12). These sites selectively bind different AP-1 family members (15). Variations in the core AP-1 or flanking nucleotides affect the binding affinities of specific dimer formation, as demonstrated with both recombinant proteins (16, 17) and endogenous AP-1 factors (15).
The vitamin D response element (VDRE) is a key component of steroid-mediated transcriptional regulation of osteocalcin gene expression both in vivo and in vitro. Characterization of regulatory motifs within and in the vicinity of the OC VDRE illustrates the potential for multiple integrated activities that contribute to vitamin D responsiveness of the OC gene (18–22), including the vitamin D receptor (VDR), retinoid X receptors (RXR), AP-1, YY1, and Cbfa factors (12, 18–24). The functional VDRE is flanked by two sites (A and B) that bind the Cbfa1-containing osteoblast-specific complex (24). The transcription factor YY1 binds to a recognition motif that overlaps the proximal steroid half-element (SHE) and mediates repression of vitamin D-enhanced OC gene promoter activity (23).
Sequence analysis of rat and human osteocalcin VDREs (nt −468 to −440 and nt −513 to −483, respectively) revealed a binding site for AP-1 transcription factors located between the two SHEs and partly overlapping the distal RXR-binding half-element (12, 19–22). In addition to this internal AP-1 motif, the human VDRE contains an AP-1 site immediately upstream of the VDR complex binding site (25). This adjacent AP-1 site has been shown to contribute to enhancement of OC transcription by vitamin D (25, 26). By contrast, the internal AP-1 site in the rat OC VDRE was suggested to be involved in blocking vitamin D responsiveness of the OC gene in proliferating osteoblasts (12, 27), but functional contributions of this internal AP-1 site to vitamin D responsiveness were not experimentally addressed.
Detailed studies of both the distal AP-1 site in the human OC VDRE domain (28) and the internal spacer comprising the AP-1 motif within the rat OC VDRE (26–28) have identified specific nucleotides that are critical for vitamin D responsiveness. However, in these studies either the AP-1 point mutations resulted in loss of VDR binding (27, 28) or VDR binding was not addressed (26). Furthermore, these studies did not attempt to compare the activity of the VDRE when bound by a VDR complex alone and its activity when bound by both VDR and AP-1 complexes. Here, we generated a precise mutation in the AP-1 site within the rat OC VDRE that abolishes AP-1 binding without affecting VDR/RXR interactions. The consequential effects of this mutation provide new insight for precise vitamin D-dependent regulation of the rat osteocalcin gene that requires AP-1 interactions for functional activity of the VDR/RXR heterodimer.
Materials and Methods
Plasmid construction
pAP-1mtOCCAT is a derivative of pOCZCAT (29) in which the chloramphenicol acetyltransferase (CAT) gene is regulated by the 1.1-kb rat OC promoter. pAP-1mtOCCAT was constructed by PCR with pOCZCAT as the template, using either the rat VDRE-AP-1mt or VDRE-SHEmt oligonucleotides as primers (see Table 1). These oligonucleotides carry mutations in the internal AP-1 site and in the SHEs, respectively.
Sequences of the oligonucleotides used for gel shift assays, cloning and comparison of rat oc VDRE sequences with mouse and human VDREs
 |
| |
Steroid half-elements (SHEs) are overlined; AP-1 and YY1 sites are underlined. Nucleotides in lower case show mutations in the steroid half-elements and AP-1 site. Numbers indicate the promoter location of the sequences.
Sequences of the oligonucleotides used for gel shift assays, cloning and comparison of rat oc VDRE sequences with mouse and human VDREs
| |
| |
Steroid half-elements (SHEs) are overlined; AP-1 and YY1 sites are underlined. Nucleotides in lower case show mutations in the steroid half-elements and AP-1 site. Numbers indicate the promoter location of the sequences.
For heterologous promoter studies, the rat OC VDRE, VDRE-AP-1mt, and VDRE-SHEmt oligonucleotides (Table 1) were each cloned in pCK-16 (a gift from Dr. Sturzenbecker, Hoffmann-La Roche, Nutley, NJ), a plasmid containing the herpes simplex thymidine kinase (TK) promoter upstream of the luciferase gene. The double stranded oligonucleotides were cloned in the unique HindIII and BglII sites of pCK-16 by blunt end ligation. Constructs were subjected to restriction digests and dideoxy sequencing to confirm the integrity and orientation of the cloned oligonucleotides.
Cell culture
ROS 17/2.8 osteosarcoma cells (provided by Dr. S. Rodan, Merck, Sharp, and Dohme Research Laboratories, West Point, PA) were maintained in Ham’s F-12 medium supplemented with 5% FCS, 2 mml-glutamine, and 1.1 mm Ca+2. 1,25-Dihydroxyvitamin D3[ 1,25-(OH)2D3] was provided by Dr. M Uskokovic (Hoffmann-La Roche).
Transfections
ROS 17/2.8 cells were transfected by the diethylaminoethyl-dextran method (30) 24 h after plating. For CAT reporter plasmids, cells were plated at a density of 0.5 × 106/100-mm plate and transfected with 20 μg total DNA containing 10 μg reporter plasmid and 10 μg salmon sperm DNA (Sigma Chemical Co., St. Louis, MO). For the luciferase reporter plasmid, cells were plated in 35-mm six-well plates at a density of 8 × 104 cells/well and transfected with 5μ g reporter plasmid. Three hours posttransfection, cells were glycerol shocked and incubated in Ham’s F-12 medium containing either 10−8m 1,25-(OH)2D3 or vehicle. Cells were harvested after 48–60 h, lysed in 1 × reporter lysis buffer (Promega Corp., Madison, WI), and subjected to either CAT or luciferase assay. Each transfection experiment was repeated three or four times with two or more DNA preparations.
Gel mobility shift assays
Nuclear extracts were prepared as described previously (31) from confluent ROS 17/2.8 cells that were plated at 0.8 × 106/100-mm plate and treated for 24 h with either 10−8m 1,25-(OH)2D3 or vehicle. Probes were prepared by phosphorylating gel-purified oligonucleotides (Table 1) using [γ-32P]ATP and T4 polynucleotide kinase (New England Biolabs, Inc., Beverley, MA). DNA binding reactions (20 μl) contained 5 μg nuclear proteins, 1 mm dithiothreitol, and 1.5 μg poly(dI-dC)·poly (dI-dC) and were incubated for 20 min at room temperature. Supershift experiments were performed by preincubating the nuclear proteins with antibodies for 1 h on ice before the DNA binding reaction. The immunoreactivity of gel shift complexes was monitored with the monoclonal VDR antibody F2A10, provided by Dr. H. F. DeLuca (University of Wisconsin, Madison, WI). The monoclonal RXR antibodies used in these studies were provided by Dr. P. Chambon (IGBMC, Illkrich, France). Antibody 4RX-D12 (in this study also referred to as antibody RXRα, -β, and -γ) recognizes the D/E region of RXR and cross-reacts with RXRα, RXRβ, and RXRγ. Antibodies 4RX3A2, 16RX3E8, and 12RX2D3 specifically recognize RXRα, RXRβ, and RXRγ, respectively. To control for specificity of antibody-induced complexes, antibodies were also incubated with probe alone. Antihuman glucocorticoid receptor antibody was used as a nonspecific antibody to serve as an additional control. Protein/DNA complexes were subjected to electrophoresis in 4% polyacrylamide gels and visualized by autoradiography.
Results
The AP-1 site within the rat OC VDRE is critical for vitamin D induction. The OC VDRE represents a potential site for interaction between AP-1 and VDR signaling pathways. To determine the relative functional contributions of the SHEs vs. the AP-1-binding site of the VDRE to vitamin D-mediated transcriptional enhancement, we introduced mutations in either the SHEs (Table 1, VDRE-SHEmt) or the internal AP-1 site (VDRE-AP-1mt). OC promoter-CAT constructs containing these mutations were assayed for vitamin D responsiveness in transiently transfected ROS 17/2.8 cells. In contrast to the 5-fold enhancement of the wild-type pOCZCAT construct and consistent with other studies in which the SHEs were mutated (26–28), vitamin D-mediated trans-activation was completely lost with pSHEmtOCCAT (Fig. 1). In fact, transcription of pSHEmtOCCAT was actually inhibited 2.3-fold by vitamin D (see Discussion). Strikingly, complete loss of vitamin D-mediated transcriptional enhancement was also observed when the internal AP-1 site within the VDRE was selectively disrupted (Fig. 1A, pAP-1mtOCCAT). These findings were reproducible in four independent studies (Fig. 1B). This result demonstrates that the intact internal AP-1 site is functionally required for vitamin D-mediated activation of OC gene transcription.
The internal AP-1 site within the rat OC VDRE is essential for vitamin D-mediated transcriptional enhancement. A, ROS 17/2.8 cells were transiently transfected with wild-type pOCZCAT, pSHEmtOCCAT carrying mutations in the SHEs, or pAP-1mtOCCAT carrying a mutation in the internal AP-1 site of the rat OC VDRE. Cells were incubated for 48–60 h in the absence (control) or presence of 10−8m 1,25-(OH)2D3. The expression of reporter gene was measured by CAT assays and is presented as the percent acetylation of chloramphenicol substrate. Each bar represents the mean ± sd (n = 3). The results shown represent one of three experiments with similar results. B, Graph representing vitamin D/control fold effects observed for four independent transfection experiments comparing the pOC2CAT, pSHEmtOCCAT, and pAP-1mtOCCAT promoter-reporter activities described in A.
VDR interactions with the rat OC VDRE occur independently of the integrity of the AP-1 motif
Loss of vitamin D activation as a result of the VDRE-AP-1 mutation (Fig. 1) may reflect either 1) loss of VDR complex binding to the VDRE, or 2) requirement for AP-1 protein occupancy at the VDRE domain to facilitate functionality of a bound VDR complex. To address these possibilities, we studied VDR complex binding to the VDRE-AP-1 mutant by gel shift assays. Consistent with previous reports (12, 27, 32), Fig. 2A shows formation of VDR/RXR, YY1, and two AP-1 complexes at the wild-type rat OC VDRE with ROS 17/2.8 nuclear extracts. The intensity of the YY1 complex is dependent on nuclear extract preparation and is clearly distinguished in Fig. 2C. Note that the mutation in the VDRE SHE mt does not interfere with YY1 binding (Table 1). The upper and lower AP-1 complexes (arrows) are competed by an unlabeled rat OC VDRE oligonucleotide as well as the VDRE-SHE mutant oligonucleotide containing the native AP-1 sequence (Fig. 2A and Table 1), but not by the VDRE-AP-1 mutant oligonucleotide. These results are consistent with previous identification of the VDRE complexes by supershift analysis with antibodies to Fos and Jun family members (15). Competition with the VDRE-SHE mutant reveals that the lower AP-1 band comigrates with a non-AP-1 complex that remains. This unidentified minor complex is also present when the VDRE-AP-1mt is used as probe (Fig. 2B). The specificity of the AP-1 complexes formed at the rat OC VDRE was further confirmed by competition of these bands with an AP-1 consensus sequence oligonucleotide (Fig. 2C). The VDR complex was competed by the unlabeled rat OC VDRE oligonucleotide as well as the VDRE-AP-1 mutant oligonucleotide, but not by the VDRE-SHE mutant oligonucleotide. Thus, the VDR complex forms effectively in the absence of the AP-1 site, and the AP-1 complex can interact in the absence of the SHEs. Taken together, these results demonstrate that interactions of the VDR and AP-1 with the rat OC VDRE can occur independently of each other. The ability of the VDR complex to interact with the VDRE independent of the AP-1 site is confirmed in Fig. 2B, which shows formation of the VDR complex with the AP-1 mutant VDRE probe. We conclude that loss of vitamin D-mediated transcriptional activation with pAP-1mtOCCAT (Fig. 1) cannot be attributed to failure of VDR complex formation at the mutated VDRE.
The integrity of the internal AP-1 site is not required for VDR complex interactions at the rat OC VDRE. A, Gel mobility shift assays were performed with 32P end-labeled wild-type rat OC VDRE probe and 5 μg nuclear protein isolated from ROS 17/2.8 cells that were treated for 24 h with 10−8m 1,25-(OH)2D3. Competition analysis was performed with unlabeled oligonucleotides, as indicated above the lanes, that were used at increasing concentrations of 0, 12, 25, and 50 nm, respectively (from left to right) in the presence of 0.5 nm probe. The large arrowhead indicates a specific VDR/RXR complex, the double arrows indicate the AP-1 complexes, and the small arrowhead shows the YY-1 interaction as evidenced by the competition analysis. The thin arrow designates an unidentified non-AP-1 complex that comigrates with the lower AP-1 complex and is revealed by competition with the SHE mutant B. Gel shift assays were performed as described in A, but with the VDRE-AP-1 mutant site (AP-1 mt) as probe. C, Gel shift assays were performed as described in A with the AP-1 consensus oligonucleotide (*) as competitor to confirm the specificity of the AP-1 bands formed with the VDRE probe.
The VDR complex formed at the VDRE-AP-1 mutant is primarily composed of VDR and RXRα
The inability of vitamin D to enhance transcription of pAP-1mtOCCAT (Fig. 1) might reflect an alteration in the composition of the VDR complex formed with the wild-type compared with that formed with the AP-1 mutant OC VDRE. To address this possibility, we performed supershift analyses using the VDRE-AP-1 mutant probe, vitamin D-treated ROS 17/2.8 nuclear extracts, and monoclonal antibodies against VDR and different RXR proteins. The VDR complex was effectively supershifted by the VDR antibody (Fig. 3). Using the RXR antibodies, a complete supershift was observed with an antibody that recognizes all three RXR subtypes, and a partial supershift was observed with each of the RXRα- and the RXRβ-specific antibodies. The RXRγ-specific antibody did not affect the VDR complex. Thus, the VDR complex(es) formed with the AP-1-mutant VDRE is mediated primarily by VDR and RXRα and to a limited extent by VDR/RXRβ heterodimers. This complex composition is identical to the composition of the VDR complex that forms with the corresponding wild-type OC VDRE probe, as previously reported (31–33) and confirmed in parallel supershift analysis in these studies with the wild-type probe (Fig. 3A). Therefore, loss of vitamin D enhancement with pAP-1mtOCCAT (Fig. 1) does not appear to result from an alteration in the composition of the VDR/RXR complex (Fig. 3).
The vitamin D-dependent complex formed with the VDRE-AP-1mt probe is primarily composed of a VDR/RXR heterodimer. A, Gel mobility shift assays were performed as described in Fig. 2A using the OC VDRE oligonucleotide as probe and nuclear extracts from vitamin D-treated ROS 17/2.8 cells (control; CTR-1). Before the binding reactions, the nuclear extracts were incubated with specific antibodies. The left panel shows that the VDR-specific complex interacting with the OC VDRE probe is supershifted with monoclonal VDR antibody F2A10 (VDR lane). The right panel displays immunoshifts of the VDR complex by an RXR antibody recognizing all three RXR subtypes (RXRα, -β, and -γ) or by RXR antibodies specific for RXRα or RXRβ. The RXRγ antibody does not mediate a supershift. Antibodies were used either undiluted or in a 1:100 dilution as indicated. The CTR-2 lane included an antibody recognizing the glucocorticoid receptor (14 ) as a control for VDR and RXR antibody specificity. B, Gel supershift assays were performed as described in A, using VDRE-AP-1mt oligonucleotide as probe.
Requirement of an intact AP-1 site for VDRE-mediated activation of a heterologous promoter
The OC VDRE can mediate vitamin D-dependent activation of the heterologous TK promoter (20, 22, 34, 35). We addressed the requirement for the internal AP-1 site in this context. Wild-type, VDRE-SHE mutant, and VDRE-AP-1 mutant oligonucleotides corresponding to the −468 to− 440 rat OC promoter sequence were inserted upstream of a TK promoter-luciferase fusion gene, and responsiveness to vitamin D was assessed in transiently transfected ROS 17/2.8 cells (Fig. 4). Consistent with previous reports (22, 34, 35), the wild-type OC VDRE-TKLUC construct exhibited 1.8- to 2-fold increased expression in response to vitamin D. As anticipated, the VDRE-SHEmt-TKLUC mutant did not confer vitamin D enhancement on the heterologous TK promoter. More importantly, there was no vitamin D enhancement with the VDRE-AP-1mt-TKLUC construct lacking the AP-1-binding site. Hence, the internal AP-1 site of the rat OC VDRE is required to confer vitamin D responsiveness on both the native OC (Fig. 1) and the heterologous TK promoter (Fig. 4). Taken together, our results suggest that VDR/RXR binding to the rat OC VDRE is not sufficient for vitamin D-mediated trans-activation, but, rather, requires the intact AP-1 site.
The internal AP-1 site within the rat OC VDRE is required for vitamin D-mediated transcriptional activation in the context of a heterologous promoter. ROS 17/2.8 cells were transiently transfected with TK-luciferase plasmids containing a single copy of the wild-type rat OC VDRE, the VDRE-SHEmt, or the VDRE-AP-1mt. Cells were treated with either vehicle (control) or 10−8m 1,25-(OH)2D3 for 48 h and extracted for luciferase assay as described in Materials and Methods. The values plotted on the graph represent the mean ± sd (n = 6). Similar results were obtained in two additional experiments.
Discussion
In this study we have demonstrated for the first time that the AP-1 binding motif within the rat OC VDRE is critical for vitamin D-mediated OC gene activation, but does not affect VDR/RXR interaction with the VDRE. The rat OC VDRE represents a multipartite gene regulatory element and contains recognition motifs for VDR/RXR heterodimers (31–33), YY1 (23), and an AP-1 motif internal to the SHEs that binds AP-1 factors in in vitro assays (12, 15). It has been shown that VDR/RXR activates rat OC gene transcription in response to vitamin D (20, 23, 28, 35) and that YY1 antagonizes this response (23). However, the contribution of the AP-1 motif to vitamin D-mediated enhancement of rat OC promoter activity was not experimentally established. The AP-1 family of proteins can mediate either suppressor (c-Fos/c-Jun) or enhancer (Fra-2/Jun D) activity on the OC promoter (15).
We have now demonstrated the functionality of the AP-1 site in the rat OC VDRE by introducing mutations in the internal spacer comprising the AP-1 motif. Strikingly, a mutation that disrupts the interactions of AP-1 but not VDR/RXR or YY1, with the VDRE results in the loss of vitamin D-mediated trans-activation. The requirement for the intact AP-1 site for vitamin D-mediated trans-activation was confirmed in the context of a heterologous promoter. These findings suggest that the vitamin D responsiveness of OC transcription involves synergism between the internal AP-1 site and the VDR/RXR complex. This synergistic mode of regulation may be analogous to that of the TIMP-1 gene, where binding sites for AP-1 and c-Ets-1 are adjacent. Although c-Ets-1 alone did not activate transcription from this element, it enhanced transcription synergistically with AP-1 either in the context of the natural promoter or when the sequences were linked upstream of a heterologous promoter (36).
AP-1 proteins may interact directly or indirectly with the VDR/RXR complex to activate transcription in a vitamin D-dependent manner. As expression of AP-1 proteins has been reported to increase in response to vitamin D (37, 38), increased levels of AP-1 proteins may promote occupancy at the internal AP-1 site and interact with the VDR/RXR complex to regulate transcription. This observation is supported by other studies that suggest synergistic activity of AP-1 proteins with steroid hormone-dependent transcription factors. For example, c-Jun has been reported to synergize with androgen receptor (AR) to mediate AR-induced trans-activation, and c-Fos was reported to be a negative regulator of c-Jun action on AR (39). Our findings are consistent with recent reports suggesting cooperative interactions between several different families of transcription factors and steroid hormone receptors. Synergism for transcriptional activation has been observed between the VDR and SP-1, nuclear factor-1, octamer binding protein-1, and AP-1 using synthetic promoters (40), and between the glucocorticoid receptor and CCAAT-binding transcription factor/nuclear factor-1 for the mouse mammary tumor virus promoter (41). It is possible that functional interactions between AP-1 and VDR/RXR complexes are stabilized by other proteins, including general transcription factors or transcriptional adaptor proteins, such as TAFII30 (42), SPT6 (43), Src-1 (44), RIP140 (45), CCAAT-binding protein (46), and nuclear co-activator protein-62 (47).
Interestingly, expression of the OC promoter containing a mutation in the SHEs that prevents binding of VDR/RXR heterodimer reveals a 2-fold repression of OC promoter activity in response to vitamin D. This result most likely reflects direct or indirect effects of vitamin D on AP-1 binding to any of the multiple sites on the OC promoter. In osteoblasts, vitamin D increases c-Fos expression (37), thus accounting for the suppressor effects on OC promoter activity at any of several AP-1 sites. Vitamin D also modulates the expression of other OC promoter factors that down-regulate OC transcription, including GR (14), MSX-2 (48), and Dlx-5 (49). In the mouse it has been reported that vitamin D decreases messenger RNA of AML-3/Cbfa1 (50), a bone-specific trans-activator of OC (51–53). However, vitamin D does not effect binding of this osteoblast-specific complex at Cbfa regulatory elements in rat bone cells (24, 54) (data not shown).
The organizations of recognition motifs for VDR, RXR, YY1, and Fos/Jun-related proteins in the VDREs of the human, rat, and mouse OC genes are different. These molecular differences may mediate species-specific responses to vitamin D to modulate OC gene expression. VDREs of the rat and human osteocalcin genes each contain an internal AP-1 site, whereas the human OC gene contains a second AP-1 site adjacent to the VDRE. Both human and rat VDREs mediate vitamin D-dependent activation of the osteocalcin gene (18–22). In contrast, the homologous mouse OC promoter domain contains a VDR/RXR binding sequence, but not an associated AP-1 site. Vitamin D treatment does not up-regulate transcription of the mouse osteocalcin gene (55, 56), but the sequence binds the VDR/RXR heterodimer complex (55). Based on these observations, it is possible that failure of the mouse OC VDRE to activate transcription in response to vitamin D in vivo (50, 56) and in cells (55) may reflect the absence of an AP-1 site within the mouse VDRE domain. In summary, our results demonstrate that the internal AP-1 site of the rat VDRE is a necessary component of the functional VDRE.
The secretarial assistance of Judy Rask is gratefully acknowledged.
Author notes
Present address: Department of Physiology, Michigan State University, Lansing, Michigan 48824.
