Posttranscriptional Regulation of Prolactin (PRL) Gene Expression in PRL-Deficient Pituitary Tumor Cells
William
M. Billis,
Beverly
C. Delidow,
and Bruce
A. White
Graduate Program in Developmental Biology Department of Anatomy University of Connecticut Health Center Farmington, Connecticut 06030
hypophyseal cells which, like many other endocrine cell types, are responsive to alterations in the endocrine status of the organism (3). Thus, the number of lactotropes can change in certain physiological states (e.g. pregnancy; 3). Furthermore, there appears to be a degree of plasticity in the phenotype of acidophils, in that lactotropes, somatotropes, and lactosomatotropes may be interchangeable, with the final proportions of each acidophilic subpopulation dependent on the developmental or reproductive state of the animal (3, 4). There exist several clonal strains of rat pituitary tumor cells which produce PRL and GH and show regulatory responses similar to those of pituitary cells in situ (1). As such, these cell lines provide model systems for identifying the molecular mechanisms involved in the regulation of PRL and GH gene expression (1). Interestingly, the basal level production of PRL and GH is not the same in all of these cell lines and shows predictable shifts. For example, GH3 cells synthesize both PRL and GH at relatively high levels (i.e. >2% of total protein synthesis), whereas GC cells were cloned from a culture of GH3 cells which spontaneously ceased PRL production (1). Long-term culture of GH3 cells typically leads to a loss in PRL but not GH production (see below). Comparative studies of pituitary tumor cell lines have been undertaken by several investigators for the purpose of identifying underlying mechanisms which determine basal PRL production. For example, Zhang et al. (5) investigated the causes of the undetectable levels of PRL mRNA in GH12C, cells. (5). The findings that PRL mRNA levels increased after 5azacytidine treatment and that methylation patterns in the region of the PRL gene differed between DNA samples from GH12CI cells and GH&, led these investigators to conclude that hypermethylation of a specific site in the fourth exon of the PRL gene is involved in the transcriptional suppression of the PRL gene in GH,2C, cells (5). A significant limitation of this study is that the transcriptional rate of the PRL gene was not directly measured. Nevertheless, the results indicate a correlation between the relative degree of DNA methylation within the region of the PRL gene and the apparent level of transcriptional activity of the PRL gene. Similarly, Arnold et al. (6) attributed the inefficient use of the
Rat pituitary acidophils consist of somatotropes (GH+/PRL-), lactotropes (GH-/PRL+), and lactosomatotropes (GH+/PRL+). Studies have indicated interconversion of these cell types in response to changing hormonal status. Representative tumor cell lines have been obtained for each acidophil cell type, and some display spontaneous interconversions. We examined whether the switch from GH3 cells (GH+/PRL+) to GH3LP and GC cells (both GH+/ PRL-) involves repression of PRL gene expression at a transcriptional VS. posttranscriptional level. PRL mRNA is undetectable or barely detectable in GH3LP and GC cells. In contrast, nuclear extracts from these cells transcribe the PRL promoter in Y&O, and their Pit-l mRNA levels are comparable to those in GH3 cells. Nuclear run-on transcription assays demonstrated that the PRL gene is transcribed in GH3LP and GC cells at a rate of about 60% of that observed in GH3 cells. No evidence was obtained for a block to transcriptional elongation or for transcription in the antisense direction across the PRL gene. Northern blot analysis of nuclear RNA revealed partially degraded and undetectable PRL gene transcripts in GH3LP cells and GC cells, respectively. These findings indicate that PRL gene transcripts are specifically degraded in tumor cells which display a pure somatotrope phenotype and raise the possibility that the frans-differentiation of lactosomatotropes to somatotropes involves posttranscriptional regulation of PRL gene expression. (Molecular Endocrinology 6: 1277-1284,1992)
INTRODUCTION
The PRL gene is expressed pituitary and uterine tissue lactotropic cells are not a terms of their abundance or tropes represent a dynamic 088&8809/92/l 277-1284$03.00/O Molecular Endocrinology Copyright 0 1992 by The Endocrme
in a cell-specific manner in (1, 2). In the pituitary, the rigidly fixed population in phenotype. Instead, lactosubpopulation of adeno-
Scmety
1277
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 25 June 2015. at 01:40 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO. 1992 1278
PRL promoter in PRL-deficient, ethyl methanesulfonateinduced GH3 variants (GH3B3) to the increased methylation state of the PRL gene. Interestingly, Laverriere et al. (7) observed that PRL production was enhanced by 5-azacytidine treatment of PRL-deficient GH&DL cells, but PRL synthesis and PRL mRNA remained undetectable in 5-azacytidine-treated GC cells. These results indicate that factors other than, or in addition to, DNA methylation are involved in silencing the PRL gene in GC cells. Since the transcriptional rate of the PRL gene has apparently not been directly measured in GC cells, this raises the possibility that a defect exists at some postinitiation or posttranscriptional step during transcriptional elongation, PRL RNA processing, splicing, or nuclear-cytoplasmic transport. Although the relative transcriptional rate of a specific gene often determines the steady state level of the corresponding mRNA, there is an increasing number of studies in which large alterations in mRNA levels cannot be accounted for by similar changes in transcription. For example, liver/ bone/kidney alkaline phosphatase (LBK AP) is produced at high levels in osteoblastic cells, but at significantly lower levels in nonosteoblastic cells such as hepatocytes (8). Although the relative steady state levels of cytoplasmic mRNA and nuclear RNA transcripts reflect the difference in LBK AP activity between these cell types, the LBK AP gene is transcribed at essentially the same rate in both cell types (8). Similarly, nuclear run-on transcription levels for insulin receptor are virtually equal in human liver cells (HepG2), human lymphocytes (IM9), monkey kidney (CVl), and normal human fibroblasts (NI Fib), despite dramatic differences in steady state levels of insulin receptor mRNA in these different cells (9). Furthermore, there is evidence for posttranscriptional regulation of PRL mRNA levels. The basal level of PRL mRNA in GH3 cells cultured in a chemically defined serum-free medium (SFM) is strongly dependent on the Ca2+ concentration of the culture medium. PRL mRNA levels in cells cultured in the presence of 0.4 mM CaCI, are typically 5- to lo-fold higher than those in cultures which have no added CaC12 (10). Our laboratory has recently reported that the Ca’+-induced, 5 to lo-fold increase in basal PRL mRNA levels is accompanied by a 0- to 2-fold change in PRL gene transcription (10, 11). Thus, it was of interest to us to examine whether the undetectable levels of PRL mRNA in GC cells are due to transcriptional or posttranscriptional defects in PRL gene expression. We report herein that the PRL gene is transcriptionally active in PRL-deficient GC and GH3LP cells at a level close to that of PRL-producing GH3 cells. The finding that nuclear PRL RNA precursors are partially degraded in GH3LP cells and are undetectable in GC cells suggests the progressive loss or gain of a factor(s) involved in the specific posttranscriptional regulation of PRL gene expression.
RESULTS Characterization of GC Cells Used for This Study The GC cells used for this study were generously provided by Dr. M. G. Rosenfeld (University of California
Vol6 No. 8
San Diego School of Medicine, La Jolla, CA). Since phenotypic alterations can appear in cell lines, it was important to first characterize the GC cells in terms of their levels of PRL mRNA and Pit-l mRNA and PRL promoter-dependent transcriptional activity in vitro relative to GH3 cells. Northern blot hybridization showed that Pit-l mRNA levels were virtually identical in GC and GHs cells, but that PRL mRNA levels in GC cells were at least 50-fold lower than in GH3 cells (Fig. 1A). Using the more sensitive technique of reverse transcription/polymerase chain reaction (PCR) with oligodeoxynucleotide primer combinations spanning the length of the PRL cDNA, DNA fragments corresponding to regions of the PRL mRNA were still barely detectable (Fig. 1B). Consistent with their equivalent Pit-l mRNA levels, PRL promoter-dependent transcriptional activity of GC cell nuclear extracts were similar to those of GH3 cell nuclear extracts (Fig. 1 C). These data are similar to
A
C PRL
GH3
PIT-1
GC
GH3
GC
GH3
GC
B E2-b
GH3
E3
GC
E3+E5
GH3
E4-b
GC
GH3
E5
GC
Fig. 1. Comparison of Steady State PRL and Pit-l mRNA Levels and the Transcription Activity of Nuclear Extracts in GC Cells vs. GHB Cells Cultures of GC and GH3 cells were treated in SFM with 10% horse serum and 2.5% fetal bovine serum for 24 h. A, PRL and Pit-l mRNA levels were determined by Northern blot hybridization. Arrows indicate the sizes of 18s (lower) and 28s (upper) ribosomal RNA. Note the major bands at the position of the mature cytoplasmic PRL and Pit-l mRNAs at 1 .O and 2.3 kb, respectively. B, PCR products corresponding to mRNA species containing various segments of the PRL gene were obtained using sequential upstream and downstream oligodeoxynucleotide primer combinations along the length of the PRL gene (exons 2 and 3, exons 3-5, and exons 4 and 5) as described in Materials and Methods. Arrows indicate the positions of correctly sized DNA fragments as determined by control PCR reactions using the PRL cDNA as a template. C, In vitro transcription of a PRL promoter construct by GC vs. GH3 nuclear extracts. Duplicate 30-119 samples of nuclear extracts from GC and GH3 cells cultured in serum-containing medium were assayed for their ability to transcribe a PRL promoter construct (pPRL 1957-&AT).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 25 June 2015. at 01:40 For personal use only. No other uses without permission. . All rights reserved.
Posttranscriptional Control of PRL in GC Cells
1279
previously published findings (12) and are consistent with studies that also showed that PRL promoterreporter constructs were efficiently expressed in GC cells in transient transfection experiments (13). Taken together, findings from this and other laboratories attest to the competency of transacting factors in GC cells to promote PRL gene transcription. The Endogenous PRL Gene Is Transcribed Deficient GC Cells
in PRL-
Despite the findings stated above, the PRL gene could be silenced in GC cells do to an alteration in cis. Thus, we measured the transcriptional rate of the endogenous PRL gene in GC and GH3 cells by nuclear run-on transcription assay. Using a near full-length PRL cDNA probe, PRL gene transcription was detected at a rate of 1.3 f 0.24 ppm (n = 6) in GC cells, compared to 2.1 + 0.34 ppm (n = 6) in GH3 cells. Thus, the rate of PRL gene transcription was about 60% of that found in GHS cells even though the PRL mRNA level was less than 2% of that in GHB cells (Figs. 1A and 2A). In two subsequent experiments, several controls were included in order to confirm that the observed hybridization signal was an accurate indicator of PRL gene transcription. Since antisense transcripts have been detected for some genes (e.g. c-myc; 14, 15) labeled RNA was hybridized to strand-specific DNA probes. These data indicate that there is only significant transcription in the sense direction (Fig. 28). Little or no hybridization signal was detected in run-on samples from rat-l fibroblasts and human HeLa cells (Fig. 2B). These results demonstrate that the PRL gene is in fact transcribed in GC cells, and that the small (i.e.
William
M. Billis,
Beverly
C. Delidow,
and Bruce
A. White
Graduate Program in Developmental Biology Department of Anatomy University of Connecticut Health Center Farmington, Connecticut 06030
hypophyseal cells which, like many other endocrine cell types, are responsive to alterations in the endocrine status of the organism (3). Thus, the number of lactotropes can change in certain physiological states (e.g. pregnancy; 3). Furthermore, there appears to be a degree of plasticity in the phenotype of acidophils, in that lactotropes, somatotropes, and lactosomatotropes may be interchangeable, with the final proportions of each acidophilic subpopulation dependent on the developmental or reproductive state of the animal (3, 4). There exist several clonal strains of rat pituitary tumor cells which produce PRL and GH and show regulatory responses similar to those of pituitary cells in situ (1). As such, these cell lines provide model systems for identifying the molecular mechanisms involved in the regulation of PRL and GH gene expression (1). Interestingly, the basal level production of PRL and GH is not the same in all of these cell lines and shows predictable shifts. For example, GH3 cells synthesize both PRL and GH at relatively high levels (i.e. >2% of total protein synthesis), whereas GC cells were cloned from a culture of GH3 cells which spontaneously ceased PRL production (1). Long-term culture of GH3 cells typically leads to a loss in PRL but not GH production (see below). Comparative studies of pituitary tumor cell lines have been undertaken by several investigators for the purpose of identifying underlying mechanisms which determine basal PRL production. For example, Zhang et al. (5) investigated the causes of the undetectable levels of PRL mRNA in GH12C, cells. (5). The findings that PRL mRNA levels increased after 5azacytidine treatment and that methylation patterns in the region of the PRL gene differed between DNA samples from GH12CI cells and GH&, led these investigators to conclude that hypermethylation of a specific site in the fourth exon of the PRL gene is involved in the transcriptional suppression of the PRL gene in GH,2C, cells (5). A significant limitation of this study is that the transcriptional rate of the PRL gene was not directly measured. Nevertheless, the results indicate a correlation between the relative degree of DNA methylation within the region of the PRL gene and the apparent level of transcriptional activity of the PRL gene. Similarly, Arnold et al. (6) attributed the inefficient use of the
Rat pituitary acidophils consist of somatotropes (GH+/PRL-), lactotropes (GH-/PRL+), and lactosomatotropes (GH+/PRL+). Studies have indicated interconversion of these cell types in response to changing hormonal status. Representative tumor cell lines have been obtained for each acidophil cell type, and some display spontaneous interconversions. We examined whether the switch from GH3 cells (GH+/PRL+) to GH3LP and GC cells (both GH+/ PRL-) involves repression of PRL gene expression at a transcriptional VS. posttranscriptional level. PRL mRNA is undetectable or barely detectable in GH3LP and GC cells. In contrast, nuclear extracts from these cells transcribe the PRL promoter in Y&O, and their Pit-l mRNA levels are comparable to those in GH3 cells. Nuclear run-on transcription assays demonstrated that the PRL gene is transcribed in GH3LP and GC cells at a rate of about 60% of that observed in GH3 cells. No evidence was obtained for a block to transcriptional elongation or for transcription in the antisense direction across the PRL gene. Northern blot analysis of nuclear RNA revealed partially degraded and undetectable PRL gene transcripts in GH3LP cells and GC cells, respectively. These findings indicate that PRL gene transcripts are specifically degraded in tumor cells which display a pure somatotrope phenotype and raise the possibility that the frans-differentiation of lactosomatotropes to somatotropes involves posttranscriptional regulation of PRL gene expression. (Molecular Endocrinology 6: 1277-1284,1992)
INTRODUCTION
The PRL gene is expressed pituitary and uterine tissue lactotropic cells are not a terms of their abundance or tropes represent a dynamic 088&8809/92/l 277-1284$03.00/O Molecular Endocrinology Copyright 0 1992 by The Endocrme
in a cell-specific manner in (1, 2). In the pituitary, the rigidly fixed population in phenotype. Instead, lactosubpopulation of adeno-
Scmety
1277
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 25 June 2015. at 01:40 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO. 1992 1278
PRL promoter in PRL-deficient, ethyl methanesulfonateinduced GH3 variants (GH3B3) to the increased methylation state of the PRL gene. Interestingly, Laverriere et al. (7) observed that PRL production was enhanced by 5-azacytidine treatment of PRL-deficient GH&DL cells, but PRL synthesis and PRL mRNA remained undetectable in 5-azacytidine-treated GC cells. These results indicate that factors other than, or in addition to, DNA methylation are involved in silencing the PRL gene in GC cells. Since the transcriptional rate of the PRL gene has apparently not been directly measured in GC cells, this raises the possibility that a defect exists at some postinitiation or posttranscriptional step during transcriptional elongation, PRL RNA processing, splicing, or nuclear-cytoplasmic transport. Although the relative transcriptional rate of a specific gene often determines the steady state level of the corresponding mRNA, there is an increasing number of studies in which large alterations in mRNA levels cannot be accounted for by similar changes in transcription. For example, liver/ bone/kidney alkaline phosphatase (LBK AP) is produced at high levels in osteoblastic cells, but at significantly lower levels in nonosteoblastic cells such as hepatocytes (8). Although the relative steady state levels of cytoplasmic mRNA and nuclear RNA transcripts reflect the difference in LBK AP activity between these cell types, the LBK AP gene is transcribed at essentially the same rate in both cell types (8). Similarly, nuclear run-on transcription levels for insulin receptor are virtually equal in human liver cells (HepG2), human lymphocytes (IM9), monkey kidney (CVl), and normal human fibroblasts (NI Fib), despite dramatic differences in steady state levels of insulin receptor mRNA in these different cells (9). Furthermore, there is evidence for posttranscriptional regulation of PRL mRNA levels. The basal level of PRL mRNA in GH3 cells cultured in a chemically defined serum-free medium (SFM) is strongly dependent on the Ca2+ concentration of the culture medium. PRL mRNA levels in cells cultured in the presence of 0.4 mM CaCI, are typically 5- to lo-fold higher than those in cultures which have no added CaC12 (10). Our laboratory has recently reported that the Ca’+-induced, 5 to lo-fold increase in basal PRL mRNA levels is accompanied by a 0- to 2-fold change in PRL gene transcription (10, 11). Thus, it was of interest to us to examine whether the undetectable levels of PRL mRNA in GC cells are due to transcriptional or posttranscriptional defects in PRL gene expression. We report herein that the PRL gene is transcriptionally active in PRL-deficient GC and GH3LP cells at a level close to that of PRL-producing GH3 cells. The finding that nuclear PRL RNA precursors are partially degraded in GH3LP cells and are undetectable in GC cells suggests the progressive loss or gain of a factor(s) involved in the specific posttranscriptional regulation of PRL gene expression.
RESULTS Characterization of GC Cells Used for This Study The GC cells used for this study were generously provided by Dr. M. G. Rosenfeld (University of California
Vol6 No. 8
San Diego School of Medicine, La Jolla, CA). Since phenotypic alterations can appear in cell lines, it was important to first characterize the GC cells in terms of their levels of PRL mRNA and Pit-l mRNA and PRL promoter-dependent transcriptional activity in vitro relative to GH3 cells. Northern blot hybridization showed that Pit-l mRNA levels were virtually identical in GC and GHs cells, but that PRL mRNA levels in GC cells were at least 50-fold lower than in GH3 cells (Fig. 1A). Using the more sensitive technique of reverse transcription/polymerase chain reaction (PCR) with oligodeoxynucleotide primer combinations spanning the length of the PRL cDNA, DNA fragments corresponding to regions of the PRL mRNA were still barely detectable (Fig. 1B). Consistent with their equivalent Pit-l mRNA levels, PRL promoter-dependent transcriptional activity of GC cell nuclear extracts were similar to those of GH3 cell nuclear extracts (Fig. 1 C). These data are similar to
A
C PRL
GH3
PIT-1
GC
GH3
GC
GH3
GC
B E2-b
GH3
E3
GC
E3+E5
GH3
E4-b
GC
GH3
E5
GC
Fig. 1. Comparison of Steady State PRL and Pit-l mRNA Levels and the Transcription Activity of Nuclear Extracts in GC Cells vs. GHB Cells Cultures of GC and GH3 cells were treated in SFM with 10% horse serum and 2.5% fetal bovine serum for 24 h. A, PRL and Pit-l mRNA levels were determined by Northern blot hybridization. Arrows indicate the sizes of 18s (lower) and 28s (upper) ribosomal RNA. Note the major bands at the position of the mature cytoplasmic PRL and Pit-l mRNAs at 1 .O and 2.3 kb, respectively. B, PCR products corresponding to mRNA species containing various segments of the PRL gene were obtained using sequential upstream and downstream oligodeoxynucleotide primer combinations along the length of the PRL gene (exons 2 and 3, exons 3-5, and exons 4 and 5) as described in Materials and Methods. Arrows indicate the positions of correctly sized DNA fragments as determined by control PCR reactions using the PRL cDNA as a template. C, In vitro transcription of a PRL promoter construct by GC vs. GH3 nuclear extracts. Duplicate 30-119 samples of nuclear extracts from GC and GH3 cells cultured in serum-containing medium were assayed for their ability to transcribe a PRL promoter construct (pPRL 1957-&AT).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 25 June 2015. at 01:40 For personal use only. No other uses without permission. . All rights reserved.
Posttranscriptional Control of PRL in GC Cells
1279
previously published findings (12) and are consistent with studies that also showed that PRL promoterreporter constructs were efficiently expressed in GC cells in transient transfection experiments (13). Taken together, findings from this and other laboratories attest to the competency of transacting factors in GC cells to promote PRL gene transcription. The Endogenous PRL Gene Is Transcribed Deficient GC Cells
in PRL-
Despite the findings stated above, the PRL gene could be silenced in GC cells do to an alteration in cis. Thus, we measured the transcriptional rate of the endogenous PRL gene in GC and GH3 cells by nuclear run-on transcription assay. Using a near full-length PRL cDNA probe, PRL gene transcription was detected at a rate of 1.3 f 0.24 ppm (n = 6) in GC cells, compared to 2.1 + 0.34 ppm (n = 6) in GH3 cells. Thus, the rate of PRL gene transcription was about 60% of that found in GHS cells even though the PRL mRNA level was less than 2% of that in GHB cells (Figs. 1A and 2A). In two subsequent experiments, several controls were included in order to confirm that the observed hybridization signal was an accurate indicator of PRL gene transcription. Since antisense transcripts have been detected for some genes (e.g. c-myc; 14, 15) labeled RNA was hybridized to strand-specific DNA probes. These data indicate that there is only significant transcription in the sense direction (Fig. 28). Little or no hybridization signal was detected in run-on samples from rat-l fibroblasts and human HeLa cells (Fig. 2B). These results demonstrate that the PRL gene is in fact transcribed in GC cells, and that the small (i.e.
