Expression of the Growth Hormone Gene and the Pituitary-Specific Transcription Factor GHF-1 in Diabetic Rats
Gabriela Bedo, Pilar Santisteban, Trinidad Jolin, and Ana Aranda Instituto de Investigaciones Biomedicas Consejo Superior de Investigaciones Cieftificas 28029 Madrid, Spain
Diabetes in the rat is associated with poor growth and decreased GH in the pituitary. In this study we have examined whether this reduction reflects an impairment of GH gene expression. Diabetes was induced by the administration of streptozotocin (7 mg/100 g BW), and 18 days later, GH content, GH mRNA, and GH transcription rate were determined. GH mRNA levels were reduced by more than 80% in the pituitaries of diabetic rats, which had a similarly reduced GH content. The differences observed in transcription fully account for the changes in mRNA concentration, since the transcription rate of the gene was also reduced by a factor of 10 in the diabetic pituitaries. Insulin therapy (3 U/15 days) partially restored these parameters. The expression of the specific transcription factor GHF-1 /Pit-1 in diabetic rats was also analyzed. Both GHF-1 mRNA levels and the binding of nuclear proteins to an oligodeoxynucleotide conforming to the GHF-1 proximal binding site in the promoter of the GH gene were normal in the diabetic pituitaries, thus excluding the possibility that decreased availability of this factor could be responsible for the decreased GH transcription. Since diabetes produced an approximately 3-fold reduction of circulating T3, the potential role of thyroid hormones on GH gene expression was also evaluated in thyroidectomized and thyroidectomized diabetic rats. Thyroidectomy decreased GH and GH mRNA to less than 5% of the values found in intact animals, and a single saturating injection of T3 (250 Mg/100 g BW) resulted in a 8- to 10-fold induction of GH mRNA after 6 h. This response was markedly depressed in thyroidectomized diabetic rats, in which T3 produced only a minor increase in GH mRNA. Administration of insulin alone to these animals did not alter GH mRNA, but partially restored the response to T3, and GH mRNA levels increased 3- to 4-fold in this group. The changes in GH mRNA 0888-8809/91/1730-1739$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
were not accompanied by concomitant changes in the abundance of GHF-1. These results show that transcription of the rat GH gene is altered in diabetes, and that insulin is required for a normal regulation of the GH gene by thyroid hormone. (Molecular Endocrinology 5:17301739, 1991)
INTRODUCTION
Diabetes in the rat is accompanied by growth impairment (1, 2). It has been reported that streptozotocin (STZ)-induced diabetes is associated with a significant decrease in GH secretion (1, 2) and a marked depression of pulsatile GH release (3). Previously, our laboratory has reported that the pituitary GH content is also greatly reduced in this experimental situation, probably as a consequence of a diminished GH synthesis by the pituitary gland (2), but the molecular mechanisms responsible for this finding are presently unknown. The synthesis and release of GH are controlled by complex hormonal interactions (4). It is well known that the thyroid hormone T3 is one of the main regulators of GH gene expression. It rapidly increases GH gene transcription in both normal pituitaries (5) and pituitary tumor cells (6, 7). We have previously reported that diabetes is associated with a deficiency of circulating thyroid hormones (1) and a decrease in their nuclear receptor concentration (8, 9). This observation suggests that the effect of diabetes on GH may be related to the low circulating thyroid hormone levels in these animals. A thyroid hormone response element has been mapped within the 5' flanking region of the rat GH gene (10-12), and this promoter region also contains other cis elements that are responsible for the strict pituitaryspecific expression of this gene. These elements consist of two binding sites for the transcription factor GHF-1 (13), also called Pit-1 (14). Strong experimental evidence indicates that GHF-1 is probably the only cell type-specific factor responsible for activation of the GH
1730
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1731
GH and GHF-1 Expression in Diabetes
gene in the anterior pituitary (13-18). It appears that the cell-specific and thyroid hormone response elements cooperate in the hormonal stimulation of GH gene expression (11). In the present study we have examined GH mRNA levels and GH gene transcription rates as well as GHF1 expression in the pituitary of diabetic rats and have found that its decreased GH content is due to a reduction in the transcription of the gene without alteration of GHF-1 expression. To analyze the relationship between thyroid hormones and insulin in the regulation of GH gene expression, we have studied the GH response to T3 in thyroidectomized diabetic rats and have found that the transcriptional response to thyroid hormones is profoundly blocked in the diabetic hypothyroid rats.
RESULTS Influence of Diabetes on Body Weight, Plasma Glucose, and T3 The criteria for diabetes was hyperglycemia that increased almost 6-fold after 18 days of treatment with STZ (Table 1). In agreement with our previous results (1,2), diabetes in the rat resulted in growth impairment, as shown by a decrease of about 50% in body weight with respect to that of intact animals. Insulin therapy started 3 days after the induction of diabetes, almost totally reversed the hyperglycemia and attenuated the effect of diabetes on growth, although it did not totally normalize body weight (Table 1). Circulating T3 levels are also shown in Table 1. Plasma T3 levels were markedly reduced in the STZ-treated rats and after insulin treatment increased significantly above the corresponding values in untreated diabetic rats.
GH Response to T3 in Diabetic Rats
Control
Diabetic
Diabetic + Insulin
As shown in Table 1, diabetes is associated with a deficiency in circulating thyroid hormone levels. Since rat GH gene transcription is highly dependent on thyroidal status, the potential role of T3 in the decreased GH gene expression in the diabetic animals was evaluated. For this purpose, a saturating dose of T3 was administered to control and diabetic animals. Circulating plasma T3 levels after treatment reached similar and very high values in all groups (>1 /xg/dl plasma). Figure 3 illustrates that T3 did not increase the very low GH or GH mRNA levels found in the diabetic rats. In the insulintreated animals, T3 elicited a clear GH response, although GH and GH mRNA levels did not reach control values. In the control animals, an excess of T3 did not induce major increases in the already high levels of GH mRNA.
218 ± 1 7 117 ± 8
101 ± 9 705 ± 43 18±2
143 ±11 164 ± 15 54 ± 5
GH Response to T3 in Thyroidectomized Diabetic Rats
GH Expression in Diabetic Rats Pituitary GH content was greatly reduced in STZtreated rats, representing less than 10% of that in the intact age-paired controls. Insulin therapy increased the GH concentration by 5-fold, but the treatment did not normalize GH levels, which were still lower than those in control rats (Fig. 1a). To analyze whether the decreased hormonal content in the pituitaries of diabetic
Table 1. Effect of Diabetes and Insulin Replacement on Body Weight, Blood Glucose, and Circulating T3 Levels
BW(g) Glucose (mg/dl) Plasma T3 (ng/dl)
rats is due to a decrease in the expression of the gene, we measured GH mRNA. Northern Wot analysis of total pituitary RNA shows that the 32P-labeled GH cDNA probe yielded two major signals, one corresponding to a mRNA of 1 kilobase (kb) and the other approximately 2.8 kb in length (Fig. 2). The larger band was absent when cytoplasmic instead of total RNA was used, suggesting that this band represents the nuclear precursor. The mobility of both mRNAs was similar in all experimental groups, but their intensity was significantly decreased in the diabetic rats, indicating that diabetes decreases the abundance but does not affect the size of the mature gene transcript. Quantification of GH mRNA levels (relative to /3-actin content) is illustrated in Fig. 1b. The values obtained indicate that the changes in mRNA levels parallel those in GH concentration (Fig. 1 a), showing that the effect of diabetes is produced at the pretranslational level. Similar findings were observed in other catabolic situations; the GH mRNA concentration was also low in the pituitaries of rats fasted for 48 h, and the levels in food-restricted rats were in between those obtained in intact and diabetic animals (data not shown). To determine whether the decrease in GH mRNA in diabetes is secondary to a reduction in the rate of transcription, we conducted run-on experiments. Figure 1c illustrates that GH gene transcription rates were reduced by a factor of 10 in the diabetic rats, which is in very good agreement with the reduction in the amount of mRNA and protein. Insulin administration increased transcription to values close to those in nondiabetic animals, although they were not totally normalized. The transcription rate of the /3-actin gene was not altered in diabetic animals (data not shown)
63 ±6
Diabetes was induced with STZ, and insulin therapy was started 4 days later. The rats were killed 19 days after administration of STZ. Data are the mean ± SD of five or six animals/ group.
The difficulty in separating the deficiency in thyroid hormones from other changes occurring in diabetic rats as well as the problem of finding an increase in GH
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1732
c)
b)
•a
1
D
D+l
D+ I
D+l
Fig. 1. Expression of GH in the Pituitaries of Diabetic Rats a, GH content was measured by RIA in the pituitaries of intact animals (C), STZ-treated diabetic rats (D), and diabetic rats treated with insulin (D+l). Treatments are described in Materials and Methods. Each point represents the mean of six to eight rats ± SD. b, GH mRNA levels were measured by cDNA hybridization, as described in Materials and Methods. Values represent the mean ± SD of densitometric quantification of three dot blots and two Northern blots, including RNA from three to five animals each, c, GH gene transcription rates were determined in the same groups by run-on analysis, as described in Materials and Methods. The values represent the mean transcriptional activity of two experiments with six to eight animals per experimental group.
kb 2,8
•
•f C
0
D+l
C D
D+l
Fig. 2. Northern Blot Analysis of GH mRNA in Diabetic Rats Northern blot analysis was performed as described in Materials and Methods, with 12 ng total RNA (a) or 7 ^g cytoplasmic RNA (b). RNA was prepared from pituitaries of intact (C), diabetic (D), and insulin-treated diabetic (D+l) animals.
above high basal levels prompted us to use a better experimental model to analyze the GH response to thyroid hormones. Table 2 shows the body weight and plasma glucose levels of thyroidectomized diabetic rats treated with or without insulin. In this experiment, a group of nondiabetic animals receiving a restricted diet, so that their body weight gain remained as close as possible to that of diabetic rats, was also included. Thyroidectomized rats stopped growing soon after surgery, so their final weight was approximately the same as that 2.5 months previously. Diabetic rats lost approximately 15% of their body weight, and the magnitude of the weight loss in food-restricted rats was comparable to that in diabetic animals. Insulin therapy almost totally prevented the reduction in weight produced by STZ treatment, but did not increase body weight above the values found in thyroidectomized rats. As expected, diabetic rats were hyperglycemic,
whereas the food-restricted animals had low circulating glucose levels. Plasma T3 was very low (•
_18 S
1.2 *-
Fig. 6. Expression of Pituitary GHF-1 mRNA A representative Northern blot of GHF-1 mRNA is illustrated. Total pituitary RNA (7 ^g) from intact (C) and diabetic (D) animals was hybridized with [32P]GHF-1 cDNA. The migration of ribosomal 18S and 28S RNA as well as the size of the GHF1 transcripts are indicated.
kb in size. The intensities of the bands were similar in control and diabetic rats. Quantification of Northern and dot blots showed that GHF-1 mRNA levels were not altered in the same diabetic animals in which GH mRNA levels were decreased, and that insulin treatment did not alter this mRNA (data not shown). Figure 7 shows that thyroidectomy did not decrease GHF-1 mRNA levels and that the lack of response to T3 in the thyroidectomized diabetic animals is not due to a deficiency in the expression of the GHF-1 gene, since its levels were as least as high as those in nondiabetic rats. To exclude the possibility that the protein product of this mRNA could be functionally altered in diabetes, we conducted mobility shift assays with a 32P-labeled oligodeoxynucleotide conforming to the GHF-1 proximal binding site in the promoter of the GH gene and with nuclear extracts from control and diabetic rats treated with or without insulin and T3. Figure 8 illustrates that two concentrations of nuclear extract proteins (3 and 8 ng) were used. Incubation of the DNA fragment with 3 HQ nuclear protein resulted in the formation of a band (A) with retarded electrophoretical mobility. The use of higher extract concentrations led to the appearance of a second protein-DNA complex (B) of slower mobility than complex A. Both bands were competed in the presence of excess unlabeled DNA, although higher concentrations of competitor were needed for the disappearance of complex A than B. Competition was specific, since other unrelated oligodeoxynucleotides did not compete. Additionally, the retarded bands were observed only in pituitary glands and in pituitary cell lines that produce GH and/or PRL (GHi, GH4Ci, or 235-1 cells), but were absent in nonpituitary cells, such as C6 glioma cells or HeLa cells (data not shown). Both radioactive bands were present in all experimental groups, and in several assays we did not find consistent differences in the intensities of the retarded bands among the diabetic and nondiabetic rats, showing that
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1735
GH and GHF-1 Expression in Diabetes
C
Tx
Tx
f3
TxD TxD
i
TxD
#
GH $
GHF-1 Fig. 7. Dot Blot Hybridization of Pituitary RNA with GH and GHF-1 Cytoplasmic RNA (4, 2, and 1 jug) from intact (C), thyroidectomized (Tx), and thyroidectomized diabetic (TxD) rats treated with or without insulin (I) or T3 was sequentially hybridized with cDNAs for GH, GHF-1, and /3-actin. Details of the treatments are described in Materials and Methods.
(-)
Fig. 8. Mobility Shift Assay with Nuclear Extracts from Pituitaries of Normal and Diabetic Rats and the Proximal GHF-1 -Binding Site of the GH Promoter The 32P-labeled oligodeoxynucleotide whose sequence is shown in Materials and Methods was used for gel retardation experiments with the amounts of nuclear proteins (3 or 8 jug) indicated in the figure. Pituitary nuclear extracts from intact (C), diabetic (D), and D rats treated with insulin (I) or T3 were assayed. The lane (-) indicates migration of the free DNA in the absence of nuclear extracts, and cp shows competition of the preceeding lane (C; 8 jug) with an exccess of unlabeled oligodeoxynucleotide. A and B indicate the presence of retarded complexes formed in the presence of nuclear proteins.
a deficit in GHF-1 does not seem to be responsible for the low GH transcription rates in these animals. DISCUSSION These results confirm and extend earlier reports that diabetes causes marked growth retardation and de-
creases pituitary GH content (1, 2). We have previously shown that GH synthesis, as determined by the in vitro incorporation of [3H]leucine into specific immunoprecipitates, was decreased in pituitary glands from rats with STZ-induced diabetes (1), and we show here that this decreased synthesis can be attributed to a marked reduction in expression of the GH gene, rather than to
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1991 1736
a translational inhibition or a decreased hormone stability. The skeletal growth-promoting effects of GH appear to be mediated largely via generation of insulinlike growth factor-l (IGF-I). A decrease in circulating levels of IGF-I has been reported in diabetic rats, and poor growth in diabetes is restored to normal after appropriate therapy with the factor (29, 30). Our data indicate that the decreased pituitary GH content in diabetic rats was directly related to a parallel reduction in GH mRNA levels. This was not a nonspecific effect, since diabetes had no influence on the level of expression of the /3-actin gene, confirming the results of others (30). The possibility that the reduction in GH could result from altered pituitary somatotroph representation has been excluded by Olchovsky et al. (31), who have shown that this cell type represents about 50% of the total pituitary cell population in both normal and diabetic rats. The reduction in GH mRNA could occur due to shortening of the mRNA half-life and/or a decrease in the rate of GH gene transcription and mRNA synthesis. Our results clearly demonstrate that diabetes affects transcription of the gene and that, as determined by run-on analysis, the reduction in the rate of GH gene transcription can totally account for the decrease in mRNA and GH content in the pituitaries. Treatment with insulin resulted in a considerable improvement in the expression of the GH gene, although the doses of insulin used did not produce total normalization. Partial reversal by insulin treatment as well as the finding that GH levels are also low in spontaneous diabetes in the rat (32) suggest that the reduction in GH gene expression described here is not a consequence of a nonspecific effect of STZ. The possible causes of the impairment of GH transcription could be directly related to the lack of insulin or could be a secondary result of an altered metabolic state. Insulin can directly influence the expression of the GH gene on the somatotrophs. It has been decribed that insulin (and IGF-I) can decrease GH mRNA and promoter expression in pituitary cell lines (33, 34). Accordingly, we would expect an opposite effect, i.e. an increase in GH, in the diabetic rats. However, it has also been shown that the effect of insulin is complex and is modulated by other factors, since insulin can have both stimulatory and inhibitory effects on expression of the GH gene in these cells depending on the media and growth conditions (35). Since regulation of GH by insulin can be dependent on the metabolic state of the cell, it is likely that the endocrine and metabolic disturbances accompanying diabetes are contributing to the effect that we observed in the diabetic rats. In this respect we found that starvation for 48 h decreases the abundance of GH mRNA to levels similar to those found in diabetic rats, and that food-restricted animals have levels in between those of diabetic and normal animals. These data suggest that the carential metabolic state in the diabetic animals may contribute to the reduced GH expression and that specific factors
Vol 5 No. 11
needed for expression of the gene may be altered in these experimental situations. Pituitary-specific expression of the GH gene is governed by a transcription factor, GHF-1 /Pit-1, that binds to two sites within its promoter (13, 14). GHF-1 is a homeobox-containing protein (15) with a POL) domain and, therefore, is a member of a family of DNA-binding proteins that control development and differentiation. Since the expression of GHF-1 is essential for GH gene transcription (13-18), and diabetes produces a clear impairment of the transcription of this gene, we tested the possibility that decreased GHF-1 expression could be involved. However, the abundance of GHF-1 transcripts was normal in diabetic rats, although they have much lower GH mRNA levels than nondiabetic animals. To exclude the possibility that in diabetes the protein product could have altered DNA-binding properties, we analyzed the appearance of retarded protein-DNA complexes with the proximal GHF-1-binding site in the rat GH promoter. Previous work from different laboratories has shown that this retardation is observed only in pituitary cells and is due to GHF-1 binding (11,13-18). We observed the presence of one or two complexes, depending on the amount of protein used in the assay, which most likely are due to the binding of GHF-1. The more retarded complex could be interpreted as being formed by a protein different from GHF-1, which binds with less affinity or is present at lower concentrations, but recent data indicate that bacterially expressed Pit1 also forms two distinct retarded species, which could represent one or two molecules of the transcription factor bound to DNA (18). In any case, both retarded bands were produced with the nuclear extracts from diabetic and nondiabetic animals, and the abundance of both was similar, suggesting that the concentration of GHF-1 is normal in the pituitaries of diabetic rats. It is, thus, clear that other factors are determining the low rate of GH gene transcription in these animals. Transcription of this gene is additionally determined by other cis elements and trans-acting factors. The GH promoter contains elements for thyroid hormone, glucocorticoid, or retinoic acid receptors and AP-2 recognition sites and is regulated by GRH via cAMP (10-12, 36-38). Changes in GH expression are, therefore, not surprising in diabetes, which in rats produces hypoactivity of the thyroid gland (1), hyperactivity of adrenocortical function (39), and decreased GRH and increased somatostatin (31). Among the factors mentioned above, thyroid hormone seems to play a major role in determinating the rate of transcription of the rat GH gene, and consequently, the low circulating T3 could contribute to the reduced GH expression in diabetes. However, in the absence of insulin, treatment for 6 h with a receptorsaturating dose of T3, which ensures thyroid hormone availability for the pituitaries of diabetic rats, did not restore GH mRNA levels to normal, suggesting that the defect is not only the hypothyroidism of these animals. It is possible that a longer treatment with T3 would be able to restore GH mRNA levels to normal. However,
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1737
GH and GHF-1 Expression in Diabetes
we have previously shown that treatment for 10 days with T3 did not normalize the low pituitary GH content of diabetic rats (1). To analyze more clearly the relationship between thyroid hormones and insulin in GH gene expression, we determined GH and GHF-1 mRNA levels in pituitaries of thyroidectomized and thyroidectomized diabetic rats. As expected, hypothyroidism produced a profoundly marked reduction of GH, and our results also show that the concentration of GHF-1 mRNA was not decreased in thyroidectomized rats. This is compatible with a rapid response to thyroid hormone, which would find the transcriptional machinery ready for its effect. Actually, T3 administration to the thyroidectomized rats induced, after 6 h, a marked accumulation of GH mRNA, which has been previously shown to be due entirely to augmented transcription (5), although at this early time point T3 treatment does not return the GH mRNA to normal levels. In contrast, the concentration of GH mRNA did not significantly increase in the hypothyroid diabetic animals treated with T3. Insulin administration reestablished the response to T3, but insulin by itself did not increase GH mRNA in the thyroidectomized diabetic rats. Taken together, these results show that both hormones are required for a normal expression of the GH gene, and that the transcriptional response to thyroid hormones is severely impaired in diabetic animals. The transcriptional action of thyroid hormone in the GH gene is mediated by interaction of T3-receptor complexes with a response element located in the 5' flanking region (10, 12). We have previously reported a decrease in thyroid hormone receptor concentration in the liver of diabetic rats (8, 9), and it is likely that a reduction of receptor in the pituitary could contribute to the diminished response to thyroid hormones. However, other nuclear factors different from the receptor and GHF-1 are required for thyroid hormone regulation of the GH gene. Santos et al. (5) have shown that labile proteins are necessary for T3 induction of GH mRNA, since when protein synthesis is inhibited, T3 is unable to increase transcriptional activity of the GH gene, implying the involvement of transcriptional factors that turn over rapidly. More recent observations (40, 41) indicate that accessory nuclear proteins are involved in the high affinity binding of the receptor to the thyroid hormone response element present in the promoter of the GH gene. The nature of these factors is still unknown, but it is tempting to speculate that the concentration or function of these auxiliary proteins could be impaired in the diabetic animals. Future studies will hopefully shed light on this question.
MATERIALS AND METHODS Animals Male Wistar rats bred in our laboratory, weighing approximately 100 g at the start of each experiment, were used. The animals were kept under constant temperature (22 C) and a 12-h light, 12-h dark cycle (lights on at 0700 h).
Diabetes was induced, as previously described (1, 2), by a single ip injection of 7 mg/100 g BW STZ (Upjohn Research Laboratory, Kalamazoo, Ml). Four days after receiving STZ, diabetic rats received either saline or 3 U/100 g BW bovine insulin (Novo Lente, Copenhagen, Denmark), sc, twice daily for 15 days until death. The effect of T3 was assesed by a single injection of a saturating dose of 250 Mg/100 g BW during the last 6 h. Groups of rats starved for 48 h or fed 50% of the diet of the controls for 15 days were also used. In other experiments the effects of diabetes and thyroid hormones were analyzed in hypothyroid rats. The animals were surgically thyroidectomized when weighing 80-100 g, and 1 week later received an ip injection of 100 ^C\ 131I to eliminate vestiges of thyroid tissue. The criteria for total thyroidectomy was body weight stasis and low plasma T4 and T3 levels (
Gabriela Bedo, Pilar Santisteban, Trinidad Jolin, and Ana Aranda Instituto de Investigaciones Biomedicas Consejo Superior de Investigaciones Cieftificas 28029 Madrid, Spain
Diabetes in the rat is associated with poor growth and decreased GH in the pituitary. In this study we have examined whether this reduction reflects an impairment of GH gene expression. Diabetes was induced by the administration of streptozotocin (7 mg/100 g BW), and 18 days later, GH content, GH mRNA, and GH transcription rate were determined. GH mRNA levels were reduced by more than 80% in the pituitaries of diabetic rats, which had a similarly reduced GH content. The differences observed in transcription fully account for the changes in mRNA concentration, since the transcription rate of the gene was also reduced by a factor of 10 in the diabetic pituitaries. Insulin therapy (3 U/15 days) partially restored these parameters. The expression of the specific transcription factor GHF-1 /Pit-1 in diabetic rats was also analyzed. Both GHF-1 mRNA levels and the binding of nuclear proteins to an oligodeoxynucleotide conforming to the GHF-1 proximal binding site in the promoter of the GH gene were normal in the diabetic pituitaries, thus excluding the possibility that decreased availability of this factor could be responsible for the decreased GH transcription. Since diabetes produced an approximately 3-fold reduction of circulating T3, the potential role of thyroid hormones on GH gene expression was also evaluated in thyroidectomized and thyroidectomized diabetic rats. Thyroidectomy decreased GH and GH mRNA to less than 5% of the values found in intact animals, and a single saturating injection of T3 (250 Mg/100 g BW) resulted in a 8- to 10-fold induction of GH mRNA after 6 h. This response was markedly depressed in thyroidectomized diabetic rats, in which T3 produced only a minor increase in GH mRNA. Administration of insulin alone to these animals did not alter GH mRNA, but partially restored the response to T3, and GH mRNA levels increased 3- to 4-fold in this group. The changes in GH mRNA 0888-8809/91/1730-1739$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
were not accompanied by concomitant changes in the abundance of GHF-1. These results show that transcription of the rat GH gene is altered in diabetes, and that insulin is required for a normal regulation of the GH gene by thyroid hormone. (Molecular Endocrinology 5:17301739, 1991)
INTRODUCTION
Diabetes in the rat is accompanied by growth impairment (1, 2). It has been reported that streptozotocin (STZ)-induced diabetes is associated with a significant decrease in GH secretion (1, 2) and a marked depression of pulsatile GH release (3). Previously, our laboratory has reported that the pituitary GH content is also greatly reduced in this experimental situation, probably as a consequence of a diminished GH synthesis by the pituitary gland (2), but the molecular mechanisms responsible for this finding are presently unknown. The synthesis and release of GH are controlled by complex hormonal interactions (4). It is well known that the thyroid hormone T3 is one of the main regulators of GH gene expression. It rapidly increases GH gene transcription in both normal pituitaries (5) and pituitary tumor cells (6, 7). We have previously reported that diabetes is associated with a deficiency of circulating thyroid hormones (1) and a decrease in their nuclear receptor concentration (8, 9). This observation suggests that the effect of diabetes on GH may be related to the low circulating thyroid hormone levels in these animals. A thyroid hormone response element has been mapped within the 5' flanking region of the rat GH gene (10-12), and this promoter region also contains other cis elements that are responsible for the strict pituitaryspecific expression of this gene. These elements consist of two binding sites for the transcription factor GHF-1 (13), also called Pit-1 (14). Strong experimental evidence indicates that GHF-1 is probably the only cell type-specific factor responsible for activation of the GH
1730
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1731
GH and GHF-1 Expression in Diabetes
gene in the anterior pituitary (13-18). It appears that the cell-specific and thyroid hormone response elements cooperate in the hormonal stimulation of GH gene expression (11). In the present study we have examined GH mRNA levels and GH gene transcription rates as well as GHF1 expression in the pituitary of diabetic rats and have found that its decreased GH content is due to a reduction in the transcription of the gene without alteration of GHF-1 expression. To analyze the relationship between thyroid hormones and insulin in the regulation of GH gene expression, we have studied the GH response to T3 in thyroidectomized diabetic rats and have found that the transcriptional response to thyroid hormones is profoundly blocked in the diabetic hypothyroid rats.
RESULTS Influence of Diabetes on Body Weight, Plasma Glucose, and T3 The criteria for diabetes was hyperglycemia that increased almost 6-fold after 18 days of treatment with STZ (Table 1). In agreement with our previous results (1,2), diabetes in the rat resulted in growth impairment, as shown by a decrease of about 50% in body weight with respect to that of intact animals. Insulin therapy started 3 days after the induction of diabetes, almost totally reversed the hyperglycemia and attenuated the effect of diabetes on growth, although it did not totally normalize body weight (Table 1). Circulating T3 levels are also shown in Table 1. Plasma T3 levels were markedly reduced in the STZ-treated rats and after insulin treatment increased significantly above the corresponding values in untreated diabetic rats.
GH Response to T3 in Diabetic Rats
Control
Diabetic
Diabetic + Insulin
As shown in Table 1, diabetes is associated with a deficiency in circulating thyroid hormone levels. Since rat GH gene transcription is highly dependent on thyroidal status, the potential role of T3 in the decreased GH gene expression in the diabetic animals was evaluated. For this purpose, a saturating dose of T3 was administered to control and diabetic animals. Circulating plasma T3 levels after treatment reached similar and very high values in all groups (>1 /xg/dl plasma). Figure 3 illustrates that T3 did not increase the very low GH or GH mRNA levels found in the diabetic rats. In the insulintreated animals, T3 elicited a clear GH response, although GH and GH mRNA levels did not reach control values. In the control animals, an excess of T3 did not induce major increases in the already high levels of GH mRNA.
218 ± 1 7 117 ± 8
101 ± 9 705 ± 43 18±2
143 ±11 164 ± 15 54 ± 5
GH Response to T3 in Thyroidectomized Diabetic Rats
GH Expression in Diabetic Rats Pituitary GH content was greatly reduced in STZtreated rats, representing less than 10% of that in the intact age-paired controls. Insulin therapy increased the GH concentration by 5-fold, but the treatment did not normalize GH levels, which were still lower than those in control rats (Fig. 1a). To analyze whether the decreased hormonal content in the pituitaries of diabetic
Table 1. Effect of Diabetes and Insulin Replacement on Body Weight, Blood Glucose, and Circulating T3 Levels
BW(g) Glucose (mg/dl) Plasma T3 (ng/dl)
rats is due to a decrease in the expression of the gene, we measured GH mRNA. Northern Wot analysis of total pituitary RNA shows that the 32P-labeled GH cDNA probe yielded two major signals, one corresponding to a mRNA of 1 kilobase (kb) and the other approximately 2.8 kb in length (Fig. 2). The larger band was absent when cytoplasmic instead of total RNA was used, suggesting that this band represents the nuclear precursor. The mobility of both mRNAs was similar in all experimental groups, but their intensity was significantly decreased in the diabetic rats, indicating that diabetes decreases the abundance but does not affect the size of the mature gene transcript. Quantification of GH mRNA levels (relative to /3-actin content) is illustrated in Fig. 1b. The values obtained indicate that the changes in mRNA levels parallel those in GH concentration (Fig. 1 a), showing that the effect of diabetes is produced at the pretranslational level. Similar findings were observed in other catabolic situations; the GH mRNA concentration was also low in the pituitaries of rats fasted for 48 h, and the levels in food-restricted rats were in between those obtained in intact and diabetic animals (data not shown). To determine whether the decrease in GH mRNA in diabetes is secondary to a reduction in the rate of transcription, we conducted run-on experiments. Figure 1c illustrates that GH gene transcription rates were reduced by a factor of 10 in the diabetic rats, which is in very good agreement with the reduction in the amount of mRNA and protein. Insulin administration increased transcription to values close to those in nondiabetic animals, although they were not totally normalized. The transcription rate of the /3-actin gene was not altered in diabetic animals (data not shown)
63 ±6
Diabetes was induced with STZ, and insulin therapy was started 4 days later. The rats were killed 19 days after administration of STZ. Data are the mean ± SD of five or six animals/ group.
The difficulty in separating the deficiency in thyroid hormones from other changes occurring in diabetic rats as well as the problem of finding an increase in GH
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1732
c)
b)
•a
1
D
D+l
D+ I
D+l
Fig. 1. Expression of GH in the Pituitaries of Diabetic Rats a, GH content was measured by RIA in the pituitaries of intact animals (C), STZ-treated diabetic rats (D), and diabetic rats treated with insulin (D+l). Treatments are described in Materials and Methods. Each point represents the mean of six to eight rats ± SD. b, GH mRNA levels were measured by cDNA hybridization, as described in Materials and Methods. Values represent the mean ± SD of densitometric quantification of three dot blots and two Northern blots, including RNA from three to five animals each, c, GH gene transcription rates were determined in the same groups by run-on analysis, as described in Materials and Methods. The values represent the mean transcriptional activity of two experiments with six to eight animals per experimental group.
kb 2,8
•
•f C
0
D+l
C D
D+l
Fig. 2. Northern Blot Analysis of GH mRNA in Diabetic Rats Northern blot analysis was performed as described in Materials and Methods, with 12 ng total RNA (a) or 7 ^g cytoplasmic RNA (b). RNA was prepared from pituitaries of intact (C), diabetic (D), and insulin-treated diabetic (D+l) animals.
above high basal levels prompted us to use a better experimental model to analyze the GH response to thyroid hormones. Table 2 shows the body weight and plasma glucose levels of thyroidectomized diabetic rats treated with or without insulin. In this experiment, a group of nondiabetic animals receiving a restricted diet, so that their body weight gain remained as close as possible to that of diabetic rats, was also included. Thyroidectomized rats stopped growing soon after surgery, so their final weight was approximately the same as that 2.5 months previously. Diabetic rats lost approximately 15% of their body weight, and the magnitude of the weight loss in food-restricted rats was comparable to that in diabetic animals. Insulin therapy almost totally prevented the reduction in weight produced by STZ treatment, but did not increase body weight above the values found in thyroidectomized rats. As expected, diabetic rats were hyperglycemic,
whereas the food-restricted animals had low circulating glucose levels. Plasma T3 was very low (•
_18 S
1.2 *-
Fig. 6. Expression of Pituitary GHF-1 mRNA A representative Northern blot of GHF-1 mRNA is illustrated. Total pituitary RNA (7 ^g) from intact (C) and diabetic (D) animals was hybridized with [32P]GHF-1 cDNA. The migration of ribosomal 18S and 28S RNA as well as the size of the GHF1 transcripts are indicated.
kb in size. The intensities of the bands were similar in control and diabetic rats. Quantification of Northern and dot blots showed that GHF-1 mRNA levels were not altered in the same diabetic animals in which GH mRNA levels were decreased, and that insulin treatment did not alter this mRNA (data not shown). Figure 7 shows that thyroidectomy did not decrease GHF-1 mRNA levels and that the lack of response to T3 in the thyroidectomized diabetic animals is not due to a deficiency in the expression of the GHF-1 gene, since its levels were as least as high as those in nondiabetic rats. To exclude the possibility that the protein product of this mRNA could be functionally altered in diabetes, we conducted mobility shift assays with a 32P-labeled oligodeoxynucleotide conforming to the GHF-1 proximal binding site in the promoter of the GH gene and with nuclear extracts from control and diabetic rats treated with or without insulin and T3. Figure 8 illustrates that two concentrations of nuclear extract proteins (3 and 8 ng) were used. Incubation of the DNA fragment with 3 HQ nuclear protein resulted in the formation of a band (A) with retarded electrophoretical mobility. The use of higher extract concentrations led to the appearance of a second protein-DNA complex (B) of slower mobility than complex A. Both bands were competed in the presence of excess unlabeled DNA, although higher concentrations of competitor were needed for the disappearance of complex A than B. Competition was specific, since other unrelated oligodeoxynucleotides did not compete. Additionally, the retarded bands were observed only in pituitary glands and in pituitary cell lines that produce GH and/or PRL (GHi, GH4Ci, or 235-1 cells), but were absent in nonpituitary cells, such as C6 glioma cells or HeLa cells (data not shown). Both radioactive bands were present in all experimental groups, and in several assays we did not find consistent differences in the intensities of the retarded bands among the diabetic and nondiabetic rats, showing that
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1735
GH and GHF-1 Expression in Diabetes
C
Tx
Tx
f3
TxD TxD
i
TxD
#
GH $
GHF-1 Fig. 7. Dot Blot Hybridization of Pituitary RNA with GH and GHF-1 Cytoplasmic RNA (4, 2, and 1 jug) from intact (C), thyroidectomized (Tx), and thyroidectomized diabetic (TxD) rats treated with or without insulin (I) or T3 was sequentially hybridized with cDNAs for GH, GHF-1, and /3-actin. Details of the treatments are described in Materials and Methods.
(-)
Fig. 8. Mobility Shift Assay with Nuclear Extracts from Pituitaries of Normal and Diabetic Rats and the Proximal GHF-1 -Binding Site of the GH Promoter The 32P-labeled oligodeoxynucleotide whose sequence is shown in Materials and Methods was used for gel retardation experiments with the amounts of nuclear proteins (3 or 8 jug) indicated in the figure. Pituitary nuclear extracts from intact (C), diabetic (D), and D rats treated with insulin (I) or T3 were assayed. The lane (-) indicates migration of the free DNA in the absence of nuclear extracts, and cp shows competition of the preceeding lane (C; 8 jug) with an exccess of unlabeled oligodeoxynucleotide. A and B indicate the presence of retarded complexes formed in the presence of nuclear proteins.
a deficit in GHF-1 does not seem to be responsible for the low GH transcription rates in these animals. DISCUSSION These results confirm and extend earlier reports that diabetes causes marked growth retardation and de-
creases pituitary GH content (1, 2). We have previously shown that GH synthesis, as determined by the in vitro incorporation of [3H]leucine into specific immunoprecipitates, was decreased in pituitary glands from rats with STZ-induced diabetes (1), and we show here that this decreased synthesis can be attributed to a marked reduction in expression of the GH gene, rather than to
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1991 1736
a translational inhibition or a decreased hormone stability. The skeletal growth-promoting effects of GH appear to be mediated largely via generation of insulinlike growth factor-l (IGF-I). A decrease in circulating levels of IGF-I has been reported in diabetic rats, and poor growth in diabetes is restored to normal after appropriate therapy with the factor (29, 30). Our data indicate that the decreased pituitary GH content in diabetic rats was directly related to a parallel reduction in GH mRNA levels. This was not a nonspecific effect, since diabetes had no influence on the level of expression of the /3-actin gene, confirming the results of others (30). The possibility that the reduction in GH could result from altered pituitary somatotroph representation has been excluded by Olchovsky et al. (31), who have shown that this cell type represents about 50% of the total pituitary cell population in both normal and diabetic rats. The reduction in GH mRNA could occur due to shortening of the mRNA half-life and/or a decrease in the rate of GH gene transcription and mRNA synthesis. Our results clearly demonstrate that diabetes affects transcription of the gene and that, as determined by run-on analysis, the reduction in the rate of GH gene transcription can totally account for the decrease in mRNA and GH content in the pituitaries. Treatment with insulin resulted in a considerable improvement in the expression of the GH gene, although the doses of insulin used did not produce total normalization. Partial reversal by insulin treatment as well as the finding that GH levels are also low in spontaneous diabetes in the rat (32) suggest that the reduction in GH gene expression described here is not a consequence of a nonspecific effect of STZ. The possible causes of the impairment of GH transcription could be directly related to the lack of insulin or could be a secondary result of an altered metabolic state. Insulin can directly influence the expression of the GH gene on the somatotrophs. It has been decribed that insulin (and IGF-I) can decrease GH mRNA and promoter expression in pituitary cell lines (33, 34). Accordingly, we would expect an opposite effect, i.e. an increase in GH, in the diabetic rats. However, it has also been shown that the effect of insulin is complex and is modulated by other factors, since insulin can have both stimulatory and inhibitory effects on expression of the GH gene in these cells depending on the media and growth conditions (35). Since regulation of GH by insulin can be dependent on the metabolic state of the cell, it is likely that the endocrine and metabolic disturbances accompanying diabetes are contributing to the effect that we observed in the diabetic rats. In this respect we found that starvation for 48 h decreases the abundance of GH mRNA to levels similar to those found in diabetic rats, and that food-restricted animals have levels in between those of diabetic and normal animals. These data suggest that the carential metabolic state in the diabetic animals may contribute to the reduced GH expression and that specific factors
Vol 5 No. 11
needed for expression of the gene may be altered in these experimental situations. Pituitary-specific expression of the GH gene is governed by a transcription factor, GHF-1 /Pit-1, that binds to two sites within its promoter (13, 14). GHF-1 is a homeobox-containing protein (15) with a POL) domain and, therefore, is a member of a family of DNA-binding proteins that control development and differentiation. Since the expression of GHF-1 is essential for GH gene transcription (13-18), and diabetes produces a clear impairment of the transcription of this gene, we tested the possibility that decreased GHF-1 expression could be involved. However, the abundance of GHF-1 transcripts was normal in diabetic rats, although they have much lower GH mRNA levels than nondiabetic animals. To exclude the possibility that in diabetes the protein product could have altered DNA-binding properties, we analyzed the appearance of retarded protein-DNA complexes with the proximal GHF-1-binding site in the rat GH promoter. Previous work from different laboratories has shown that this retardation is observed only in pituitary cells and is due to GHF-1 binding (11,13-18). We observed the presence of one or two complexes, depending on the amount of protein used in the assay, which most likely are due to the binding of GHF-1. The more retarded complex could be interpreted as being formed by a protein different from GHF-1, which binds with less affinity or is present at lower concentrations, but recent data indicate that bacterially expressed Pit1 also forms two distinct retarded species, which could represent one or two molecules of the transcription factor bound to DNA (18). In any case, both retarded bands were produced with the nuclear extracts from diabetic and nondiabetic animals, and the abundance of both was similar, suggesting that the concentration of GHF-1 is normal in the pituitaries of diabetic rats. It is, thus, clear that other factors are determining the low rate of GH gene transcription in these animals. Transcription of this gene is additionally determined by other cis elements and trans-acting factors. The GH promoter contains elements for thyroid hormone, glucocorticoid, or retinoic acid receptors and AP-2 recognition sites and is regulated by GRH via cAMP (10-12, 36-38). Changes in GH expression are, therefore, not surprising in diabetes, which in rats produces hypoactivity of the thyroid gland (1), hyperactivity of adrenocortical function (39), and decreased GRH and increased somatostatin (31). Among the factors mentioned above, thyroid hormone seems to play a major role in determinating the rate of transcription of the rat GH gene, and consequently, the low circulating T3 could contribute to the reduced GH expression in diabetes. However, in the absence of insulin, treatment for 6 h with a receptorsaturating dose of T3, which ensures thyroid hormone availability for the pituitaries of diabetic rats, did not restore GH mRNA levels to normal, suggesting that the defect is not only the hypothyroidism of these animals. It is possible that a longer treatment with T3 would be able to restore GH mRNA levels to normal. However,
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 14 November 2015. at 16:17 For personal use only. No other uses without permission. . All rights reserved.
1737
GH and GHF-1 Expression in Diabetes
we have previously shown that treatment for 10 days with T3 did not normalize the low pituitary GH content of diabetic rats (1). To analyze more clearly the relationship between thyroid hormones and insulin in GH gene expression, we determined GH and GHF-1 mRNA levels in pituitaries of thyroidectomized and thyroidectomized diabetic rats. As expected, hypothyroidism produced a profoundly marked reduction of GH, and our results also show that the concentration of GHF-1 mRNA was not decreased in thyroidectomized rats. This is compatible with a rapid response to thyroid hormone, which would find the transcriptional machinery ready for its effect. Actually, T3 administration to the thyroidectomized rats induced, after 6 h, a marked accumulation of GH mRNA, which has been previously shown to be due entirely to augmented transcription (5), although at this early time point T3 treatment does not return the GH mRNA to normal levels. In contrast, the concentration of GH mRNA did not significantly increase in the hypothyroid diabetic animals treated with T3. Insulin administration reestablished the response to T3, but insulin by itself did not increase GH mRNA in the thyroidectomized diabetic rats. Taken together, these results show that both hormones are required for a normal expression of the GH gene, and that the transcriptional response to thyroid hormones is severely impaired in diabetic animals. The transcriptional action of thyroid hormone in the GH gene is mediated by interaction of T3-receptor complexes with a response element located in the 5' flanking region (10, 12). We have previously reported a decrease in thyroid hormone receptor concentration in the liver of diabetic rats (8, 9), and it is likely that a reduction of receptor in the pituitary could contribute to the diminished response to thyroid hormones. However, other nuclear factors different from the receptor and GHF-1 are required for thyroid hormone regulation of the GH gene. Santos et al. (5) have shown that labile proteins are necessary for T3 induction of GH mRNA, since when protein synthesis is inhibited, T3 is unable to increase transcriptional activity of the GH gene, implying the involvement of transcriptional factors that turn over rapidly. More recent observations (40, 41) indicate that accessory nuclear proteins are involved in the high affinity binding of the receptor to the thyroid hormone response element present in the promoter of the GH gene. The nature of these factors is still unknown, but it is tempting to speculate that the concentration or function of these auxiliary proteins could be impaired in the diabetic animals. Future studies will hopefully shed light on this question.
MATERIALS AND METHODS Animals Male Wistar rats bred in our laboratory, weighing approximately 100 g at the start of each experiment, were used. The animals were kept under constant temperature (22 C) and a 12-h light, 12-h dark cycle (lights on at 0700 h).
Diabetes was induced, as previously described (1, 2), by a single ip injection of 7 mg/100 g BW STZ (Upjohn Research Laboratory, Kalamazoo, Ml). Four days after receiving STZ, diabetic rats received either saline or 3 U/100 g BW bovine insulin (Novo Lente, Copenhagen, Denmark), sc, twice daily for 15 days until death. The effect of T3 was assesed by a single injection of a saturating dose of 250 Mg/100 g BW during the last 6 h. Groups of rats starved for 48 h or fed 50% of the diet of the controls for 15 days were also used. In other experiments the effects of diabetes and thyroid hormones were analyzed in hypothyroid rats. The animals were surgically thyroidectomized when weighing 80-100 g, and 1 week later received an ip injection of 100 ^C\ 131I to eliminate vestiges of thyroid tissue. The criteria for total thyroidectomy was body weight stasis and low plasma T4 and T3 levels (
