n-Butyrate Effects Thyroid Hormone Stimulation of Prolactin Production and mRNA Levels in GH1 Cells*

Using cultured GH, cells, a growth hormone and prolactin-producing rat pituitary cell line, we have shown that n-butyrate and other short chain carboxylic acids stimulate histone acetylation and elicit a reduction of thyroid hormone nuclear receptor which is in- versely related to the extent of acetylation (Samuels, H. H., Stanley, F., Casanova, J., and Shao, T. C. (1980) J. Biol. Chem. 255, 2499-2508). In this study, we compared the n-butyrate and propionate modulation of receptor levels to regulation of the growth hormone and prolactin response by 3,5,3’-triiodo-~-thyronine (L-T~). n-Butyrate (0.1-10 mM) did not stimulate growth hormone production. L - T ~ stimulated the growth hormone response 4- to 5-fold and n-butyrate (0.5-1 mM) increased L - T ~ stimulation of growth hormone production 1.5- to 2-fold compared to L - T ~ alone. L - T ~ stimulation of growth hormone production at higher n-butyrate concentrations decreased in parallel with the n-butyrate-mediated reduction of receptor levels. In contrast with the growth hormone response, n-butyrate (0.5 mM) increased basal prolactin production about &fold. Prolactin production, which is inhibited 25 to 50% by L-T~,

Using cultured GH, cells, a growth hormone and prolactin-producing rat pituitary cell line, we have shown that n-butyrate and other short chain carboxylic acids stimulate histone acetylation and elicit a reduction of thyroid hormone nuclear receptor which is inversely related to the extent of acetylation (Samuels, H. H., Stanley, F., Casanova, J., and Shao, T. C. (1980) J. Biol. Chem. 255,[2499][2500][2501][2502][2503][2504][2505][2506][2507][2508]. In this study, we compared the n-butyrate and propionate modulation of receptor levels to regulation of the growth hormone and prolactin response by 3,5,3'-triiodo-~-thyronine (L-T~). n-Butyrate (0.1-10 mM) did not stimulate growth hormone production. L -T~ stimulated the growth hormone response 4-to 5-fold and n-butyrate (0.5-1 mM) increased L -T~ stimulation of growth hormone production 1.5to 2-fold compared to L -T~ alone. L -T~ stimulation of growth hormone production at higher n-butyrate concentrations decreased in parallel with the n-butyrate-mediated reduction of receptor levels. In contrast with the growth hormone response, n-butyrate (0.5 mM) increased basal prolactin production about &fold. Prolactin production, which is inhibited 25 to 50% by L -T~, was stimulated between 20and 70-fold by L -T~ + n-butyrate (0.5-1 mM) and this decreased at higher n-butyrate levels. Prolactin mRNA and growth hormone mRNA levels paralleled the changes in prolactin and growth hormone production rates. These effects of L -T~, n-butyrate, or L -T~ + nbutyrate appeared unrelated to changes in CAMP levels or global changes in DNA methylation of the growth hormone or prolactin genes. Propionate elicited the same effects as n-butyrate but at a 5to 10-fold higher concentration consistent with their relative effect on stimulating acetylation of chromatin proteins. These results suggest that prolactin gene expression is under partial regulatory repression which is reversed by a carboxylic acid-mediated postsynthetic modification event which allows for stimulation of the prolactin gene by thyroid hormone. GH, cells are a useful cell model to study the mechanism of thyroid hormone action in cultured cells (1)(2)(3)(4)(5)(6). In GH1 cells, L-T~' inhibits prolactin production by 25 to 50% (1) and stimulates a 3-to 10-fold increase in growth hormone synthesis and mRNA accumulation (2)(3)(4). Stimulation of growth * This research was supported by Grants AM 16636 and AM 21566 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
hormone mRNA levels by L -T~ appears to solely reflect increased growth hormone gene transcription rates (7-9). Abundant evidence indicates that these responses are mediated by the binding of hormone to a nuclear associated receptor (4,5). The receptor is an intrinsic chromatin-associated protein (15, Studies in intact cells and isolated nuclei have shown that n-butyrate increases the level of histone and nonhistone protein acetylation by inhibiting the rate of deacetylation rather than altering the rate of acetylation (10,11). This results from the n-butyrate inhibition of a ubiquitous chromatinassociated deacetylaseb) (12). This effect is not unique to nbutyrate but also occurs with other short chain aliphatic carboxylic acids with the potency of n-butyrate > propionate = valerate > acetate (13). Since a variety of studies in other systems suggested a relationship between histone acetylation and "gene" activation, we previously examined the effect of n-butyrate on thyroid hormone nuclear receptor levels and excision from chromatin by nuclease digestion (14). We found that n-butyrate elicited a decrease of thyroid hormone receptor abundance which was inversely related to the extent of histone acetylation. The reduction of receptor levels occurred over the same concentration range at which n-butyrate inhibited a chromatin-associated deacetylase in GH1 cells. Other carboxylic acids also elicited a reduction of receptor which was directly related to the extent of inhibition of deacetylase activity as reflected in the extent of histone acetylation (14). Yen and Tashjian (15) have recently shown that n-valerate and n-butyrate can increase growth hormone and prolactin production in GH, cells. In this paper, using GH, cells, we have examined the relationship between the carboxylic acid alteration of thyroid hormone receptor levels and the L-T3 regulation of growth hormone and prolactin production and mRNA abundance. With low concentrations of n-butyrate (0.5-1.0 mM), L -T~ stimulation of growth hormone production was 1.5-to 8-fold greater than with L -T~ alone. At higher nbutyrate concentrations, L -T~ stimulation of the grodh hormone response decreased in parallel with the carboxylic acidmediated reduction of receptor. With 0.5-1.0 mM n-butyrate, L -T~ markedly stimulated rather than inhibited prolactin production and mRNA levels. This response also decreased in parallel with the carboxylic acid-mediated decrease of thyroid hormone receptor levels. These studies suggest that partial repression of prolactin gene expression is reversed by a carboxylic acid-mediated postsynthetic modification event. Low concentrations of carboxylic acid, which do not decrease Regulation of Growth Hormone and Prolactin Production 9769 thyroid hormone receptor levels, appear to allow for positive control of the prolactin gene by thyroid hormone.

EXPERIMENTAL PROCEDURES
Materiuls-~-[3'-'~I]T~ (1200 pCi/pg), immunoreagents for cAMP determination, and DNA polymerase I were from New England Nuclear. [a-32P]dCTP and (1261)iodide (carrier-free) were from Amersham. Triton X-100 and propionate were from Eastman, n-butyrate was from Sigma, and all cell culture media and sera were from Gibco. Anion exchange resin (AG 1-XS) was from Bio-Rad, oligo(dT)-eellulose (Type 3) was from Collaborative Research, Proteinase K was from Beckman, SeaKem LE-agarose was from FMC, and forskolin was from Calbiochem-Behring. DNase I was from Worthington and all restriction endonucleases were from New England Biolabs. Rat growth hormone and prolactin for iodination, reference standards, and antisera for radioimmunoassay determination were obtained through the generosity of the Rat Pituitary Hormone Program of the National Institutes of Health. pRGH-1, a rat growth hormone cDNAbearing plasmid (16), was generously provided by John D. Baxter (University of California, San Francisco). pPRL-2, a rat prolactin cDNA-bearing plasmid (17), was generously provided by Richard A. Maurer (University of Iowa, Iowa City). Both plasmids were propagated and amplified in bacteria according to standard procedures (18). All other reagents were of the highest purity available and were obtained from Sigma, British Drug House, Pharmacia, or Fisher Scientific.
Cell Culture Conditions and Estimation of Growth Hormone and Prolactin Production Rates-GH1 cells were grown in monolayer culture in l-cm' multiwell dishes, 25-cm2 flasks (Falcon Plastics) or 490-cmZ roller bottles (Corning) as previously described (3,4). Cells were depleted of thyroid hormone before each experiment by culturing them for 12 to 24 h in Ham's F-10 medium supplemented with AG 1-X8 resin-charcoal-treated calf serum (lo%, v/v) which has been shown to be depleted of thyroid hormone (19). This medium was replaced with serum-free Ham's F-10 with or without the indicated concentrations of n-butyrate or propionate for 24 h prior to the addition of 0.5 nM L-T~. This concentration of hormone results in maximal growth hormone synthetic rates in serum-free medium (4). Each day the medium from one set of cells was sampled and the cells were washed twice with 0.14 M NaCl at 4 "C and frozen for later determination of cell protein (20). In the remaining cultures, the medium was exchanged, the concentration of carboxylic acid and/or L -T~ was maintained, and the incubation continued. Growth hormone and prolactin synthesis were estimated from the accumulation of growth hormone and prolactin in the culture medium over 24-h periods by radioimmunoassay as described (1,3,4). This provides an estimate of the average 24-h synthetic rate since GH1 cells rapidly release the synthesized peptides into the medium and growth hormone and prolactin are not degraded in the medium or cells (2,21).
Relationship of CAMP Production to Peptide Hormone Synthesis- The influence of n-butyrate and/or L -T~ on cellular cAMP levels was examined as above except that after the 0.14 M NaCl wash, cells were treated with 0.5 ml of 5% perchloric acid (w/v). After 2 h of extraction at 4 "C, the perchloric acid extracts were transferred and neutralized with KHC03 using bromcresol purple as a pH indicator. The samples were acetylated (22) and cAMP was determined by radioimmunoassay (23) using immunoreagents obtained from New England Nuclear. The perchloric acid-extracted cells were saved for analysis of cell protein (18). The effect of forskolin on L -T~ regulation of growth hormone and prolactin synthesis was studied as described above except that the medium contained 10% (v/v) AG 1-X8 resin-charcoaltreated calf serum instead of serum-free F-10 medium. The concentration of L-T3 was 5 nM, which stimulates maximal rates of growth hormone synthesis in medium supplemented with 10% serum that contains hormone-binding proteins (2). Details of the experimental protocol are given in the legend to Table I. Analysis of Prolactin mRNA and Growth Hormone mRNA Abundance-GH1 cells, grown in 490-cmZ roller bottles, were incubated with serum-free Ham's F-10 medium with 0.75 mM n-butyrate for 24 h while parallel control cell cultures received no additions. At 24 h, half of the control and n-butyrate-incubated cells received 0.5 nM L-T3. The cells were refed with medium containing the same components 24 h later and were incubated for an additional 24 h. The medium was saved for growth hormone and prolactin determination and the cells were harvested with a rubber policeman into 0.14 M NaCl, 1% sodium dodecyl sulfate, 5 mM EDTA, 50 mM Tris, pH 8.3 at 25 "C, with 600 pg/ml Proteinase K (24). The cell lysate was sheared by several passages through a 25-gauge needle and an aliquot was saved for DNA determination (25). Following DNase I digestion (26), pheno1:chloroform extraction, and ethanol precipitation, the poly(A)+ RNA was isolated by two successive oligo(dT)-cellulose chromatographic steps (27).
Preparation of High Molecuinr Weight DNA and Restriction Endonuclease Digestion-GH1 cell cultures, incubated as described for mRNA isolation, were harvested at 4 "C with a rubber policeman in 5 ml of 0.25 M sucrose, 20 mM Tris, pH 7.85 at 25 "c, and 1.1 mM MgC12 and homogenized gently in a Teflon-glass homogenizer for 15 strokes at 500 rpm. After centrifugation, the nuclear pellet was washed in the same buffer containing 0.2% Triton X-100. The nuclei were then suspended in 5 ml of 0.25 M EDTA, 0.5% sarcosyl, and 1.5 mg of Proteinase K, and incubated at 37 "C for 48 h. This incubation was continued an additional 72 h at 37 "C after the addition of 500 pglsample of Proteinase K, and at 50 "C for 2 h after the addition of another 500 pg of Proteinase K. The chromatin was then extracted three times with phenol (containing 0.1% 8-OH quinoline, w/ v):chloroform:isoamyl alcohol (2524:l) and dialyzed extensively at 25 'C against 10 mM Tris, pH 7.5, and 1 mM EDTA. Restriction endonuclease digestion was performed as described by New England Biolabs and the DNA samples were electrophoresed in 1% agarose gels (28). The gels were blotted to nitrocellulose (Schleicher and Schuell) and hybridized as described above. EcoRI and PuuII-BstNI restriction fragments of pBR322 and a Hind111 digest of h phage were electrophoresed as standards.
Influence of n-Butyrate and Propionate on Thyroid Hormone Nuclear Receptor Leoels-Experiments to estimate thyroid hormone nuclear receptor were performed by culturing cells in 25-emZ flasks with F-10 medium containing 10% (v/v) hormone-free calf serum for 24 h. The flasks were then incubated with 5 nM L -[ '~I ] T~ for 1.5 h at 37 "C and the cells were harvested and nuclei prepared as previously described (14). The radioactivity in the nuclear pellet was quantitated using a refrigerated Packard y spectrometer at 55% efficiency followed by DNA determination (25). Nonspecific binding of L-['~I]T, to GHI cell nuclei was estimated using a 1000-fold molar excess of nonradioactive L-T~. This value, always less than 5% of total binding, was subtracted from results obtained using radioactive hormone alone.
StatkticalAnalysis-Wherever appropriate, statistical analysis was performed using Student's t test.

RESULTS
Effect of n-Butyrate and Propionate on Thyroid Hormone Nuclear Receptor Levels- Fig. 1 shows the effect of a 24-h incubation of n-butyrate (0.05 to 10 mM) and propionate (0.5 to 25 mM) on the level of GHI cell thyroid hormone nuclear receptors. Both compounds decreased receptor levels with a maximal receptor reduction of 82.5% for n-butyrate and 72.5% for propionate. Both also elicit a slight increase in receptor levels at low concentrations. Propionate is approximately 5to 10-fold less effective than n-butyrate in eliciting receptor depletion although the shapes of the response curves are almost identical. These results are in good agreement with our previous observations (where only one concentration of propionate was examined) that propionate is approximately 5-to 10-fold less effective than n-butyrate in eliciting receptor reduction and histone acetylation (14). Receptor reduction occurs without inhibition of protein synthesis by n-butyrate (14) or propionate (data not shown).  24 h while one set of cells which received no additions served as a control. The medium was then replaced with medium containing the same carboxylic acid concentration with or without 0.5 nM L-T3 as indicated. After another medium exchange at 48 h, the cells were incubated for an additional 24 h. Growth hormone production rates were determined by assaying the growth hormone which accumulated in the medium during the last 24-h period using radioimmunoassay. The results were normalized per 100 pg of cell protein and are expressed as a percentage of the control cells which did not receive either L-T3 or carboxylic acid. Growth hormone production in the control cells was 250 ng/100 pg of cell protein/24 h. Each point represents the average of triplicate cell cultures and each culture varied less than * 10% from the mean.
Effect of n-Butyrate and Propionate on Growth Hormone Production- Fig. 2 demonstrates that 5 to 10 mM n-butyrate, which substantially lowers receptor levels, reduces the L-T~induced growth hormone response approximately 80% compared to cells incubated with only hormone. However, lower concentrations of n-butyrate (0.5 and 1 mM), which can cause substantial postsynthetic modification of chromatin proteins (ll), resulted in a 36 and 26% increase in the growth hormone response to L-T~. This increased response occurred at nbutyrate concentrations which either had no effect on receptor levels (0.5 mM) or partially reduced receptor abundance (3 mM). Growth hormone production with L -T~ + n-butyrate at 0.5-10 mM are each significantly different than the control flasks which received only L -T~ (p < 0.001). Without L-T~, high concentrations of n-butyrate also decreased the basal rate of growth hormone synthesis which at 10 mM was 65% of the control cells which received no n-butyrate. Without hormone, n-butyrate did not stimulate an increase in growth hormone production at any concentration. Fig. 2 also shows an identical experiment performed with propionate. Sufficiently high concentrations of propionate could not be achieved (due to its solubility) in hormonetreated cells to elicit a complete inhibition in growth hormone production. However, as with n-butyrate, propionate alone stimulated no increase in growth hormone synthesis while L-T3 with 2.5 and 5 mM propionate resulted in 22 and 41% increases compared to L -T~ alone 0, < 0.001). The propionate dose-response curve was shifted 7.5-fold rightward of the nbutyrate response curve.
Effect of n-Butyrate and Propionate on Prolactin Production- Fig. 3 shows the results of an experiment which examined prolactin production in control and L -T~ (0.5 nM) cultured cells as a function of carboxylic acid concentration. In contrast with the growth hormone response, n-butyrate alone stimulated prolactin production 7.5-fold at 0.5 mM and a 2fold stimulation was evident even at the lowest concentration  Incubated with Forskolin-To further explore whether the L-T3 stimulation of prolactin production is mediated by CAMP, a study was performed using forskolin which stimulates cAMP by directly activating the catalytic subunit of the adenylate cyclase system (36). GH, cells were incubated with 10 PM forskolin and/or L -T~, a n d t h e medium obtained between 24 and 48 h of incubation was assayed for growth hormone and prolactin (Table I)   The peptide hormone production rates represent the mean of triplicate cell cultures each of which did not vary from the mean by more than 10%. cAMP which accumulated in the medium of parallel flasks was 4.5 pmo1/100 pg of cell protein f 15% in control cell cultures and 1850 pmo1/100 pg of cell protein f 15% in cells incubated with 10 p~ forskolin. L -T~ did not alter the extent of cAMP stimulation by forskolin.

Incubation conditions
Growth hormone Prolactin

TABLE I1
Growth hormone and prolactin production with L -T~ and n-butyrate GH, cells were incubated with the indicated concentrations of L-Ts f n-butyrate as described in Fig. 6. The medium obtained from the 48-72-h incubation time was assayed for growth hormone and prolactin by radioimmunoassay and normalized per 100 pg of cell DNA.  (Fig. 6A). Control cells (lanes 1 and 2) show a distinct 1-kb species which represents mature cytoplasmic prolactin mRNA. This was slightly decreased in L-T3-cultured cells (lanes 3 and 4). The cells incubated with n-butyrate alone (lanes 5 and 6) show increased prolactin mRNA levels while those cultured with both L -T~ and n-butyrate (lanes 7 and 8) show a striking increase in prolactin mRNA abundance. As estimated by densitometry, the intensity of the autoradiographic bands parallels the amount of prolactin produced during the 24-h period preceding mRNA isolation (Table 11). In addition, a 1.7-1.8-kb nuclear precursor of prolactin mRNA was observed when cells are incubated with L -T~ and nbutyrate (lanes 7 and 8). A 1.7-1.8-kb species has been reported to be a highly abundant processed form of the prolactin gene transcript in GH, cells (24).
The growth hormone mRNA response to n-butyrate is shown in Fig. 6B and represents the same blot as Fig. 6A which was hybridized to a probe for growth hormone mRNA ("2P-labeled pRGH-1) after elution of the labeled pPRL-2 probe. Control cell cultures (lanes I and 2) show a 1-kb species corresponding to growth hormone mRNA (16). As estimated by densitometry, L -T~ (0.5 nM) (lanes 3 and 4 ) stimulated a 5-to 10-fold increase in growth hormone mRNA. Lanes 5 and 6 show results with 0.75 mM n-butyrate alone which are essentially identical to the control cells. L -T~ + n-butyrate (lanes 7 and 8) stimulated an increase in growth hormone mRNA abundance which was 5-fold greater than L -T~ alone.
As with prolactin, the growth hormone production rates paralleled growth hormone mRNA abundance for each of the incubation conditions (Table 11).
Growth Hormone and Prolactin Gene Methylation in GH, Cells-Since the studies of Bird and Southern (37), it has been shown that many actively transcribed genes are undermethylated compared with those that are transcriptionally dormant (38). Recent studies have shown that n-butyrate incubation can decrease DNA methylation in Friend murine erythroleukemia cells (39). This suggested that the effect of n-butyrate and/or L -T~ might involve a mechanism which results in undermethylation of the growth hormone and prolactin genes in GH, cells. GH, cell DNA was isolated and incubated with several restriction endonucleases whose specificity with regard to methylated sequences are known (18). HpaII and MspI are isoschisomers which cleave the sequence CCGG. However, HpaII will not cleave this sequence if the second cytosine is methylated (40). MspI cleaves both the methylated and unmethylated sequence. AuaI cleaves the sequence C(T or C)CG(A or G)A but will not digest the sequence if the cytosine preceding the guanine is methylated (37). Fig. 7 (left) shows an autoradiograph from an MspI digest of GH, cell DNA electrophoresed in a 1% agarose gel and hybridized against "P-labeled pRGH-1. Based on the restriction map of the growth hormone gene cloned from rat liver DNA, MspI or HpaII should yield DNA fragments of 0.4 and 0.5 kb if the DNA restriction sequence is unmethylated (41,42). If the sequence is methylated, HpaII will not generate these fragments but the 0.4-and 0.5-kb fragments will be excised by MspI. HpaII digestion of GH, DNA yielded hybridizable fragments of 9.2 and 6.3 kb (not illustrated) while MspI generated fragments of 4.9, 4.3, 0.5, and 0.4 kb. These results indicate that the HpaII-MspI restriction sites within the growth hormone gene are methylated. The identification of the 4.9 and 4.3 MspI and the 9.2-and 6.3-kb HpaII restriction fragments suggest that one or more HpaII-MspI restriction sites flank the growth hormone gene. These sites appear to be heterogeneously methylated as indicated by the different restriction fragments excised by MspI and HpaII. Fig. 7 (right) illustrates results from a HindIII-AuaI digestion of GH, cell DNA. In this case, AuaI will not cleave a methylated sequence. If the AuaI site is methylated, only a 5.98-kb fragment would be excised, while if completely unmethylated, bands of 3.05 and 2.95 kb should be observed (41,42). Only one band of 3.9 kb is identified on the gel. This is a DNA species larger than anticipated and may result from anomalous migration on the gel. However, the identification of a single band is not unexpected since the difference in size  Fig. 7, the plasmid probe was eluted at 65 "C with several changes of 5 mM Tris, pH 7.9, 0.2 mM EDTA, 0.05% sodium pyrophosphate, 0.002% polyvinylpyrrolidone, 0.002% Ficol, and 0.002% bovine serum albumin by the method of Thomas (54). The blot was then rehybridized to "P-labeled pPRL-2 plasmid. The MspI and the HindIII-AuaI digests and the lune designations are as in Fig. 7. of the predicted doublet fragments is too small to be resolved in this gel.
The blots hybridized to "P-labeled pRGH-1 in Fig.7 were also hybridized against labeled pPRL-2 (Fig. 8). Fig. 8 (left) shows the results of MspI digestion of GHI DNA while Fig. 8  (right) shows the HindIII-Am1 digest. If the prolactin gene is unmethylated at the recognition site for HpaII-MspI, both enzymes will yield fragments of 1.7 and 4.3 kb (43). However, if this site is methylated, only MspI will yield these fragments. Since bands of 1.7, 3.9, and about 6.0 kb are generated by MspI and no fragments were identified from a HpaII digest (data not shown), the three known HpaII-MspI sites in the prolactin gene are methylated. It also appears, from identification of the 6.0-kb fragment, that one other site probably exists flanking the prolactin gene that is also methylated. Alternatively, the 6.0-kb fragment may reflect an incomplete digestion to yield the 1.7-and 3.9-kb restriction fragments.
Simultaneous digestion of the prolactin gene using HindIII and Am1 would be expected to yield two fragments of 5.9 and 5.1 kb if the AuaI site in the prolactin gene is methylated but 5.1-, 3.1-, and 2.8-kb species if unmethylated (43). The blot in Fig. 8 (right) identified fragments estimated as 4.9, 4.7, 3.4, and 3.0 kb. The two larger species are in an area of the gel where molecular weight determinations are not precise while the smaller fragments are within 10% of their expected size. Therefore, the AuaI sites in the prolactin gene appear to be variably methylated. Of significance, no major difference in the methylation of the prolactin or the growth hormone genes in response to either L -T~ and/or n-butyrate was apparent by restriction enzyme analysis in Figs. 7 and 8.

DISCUSSION
In this study, we have examined the relationship between carboxylic acid alteration of thyroid hormone nuclear receptor levels and the L-T3 regulation of growth hormone and prolactin production. Fig. 1 shows that n-butyrate elicits a biphasic effect on thyroid hormone receptor levels with low concentrations (0.05-0.25 mM) stimulating a 20 to 30% increase and higher concentrations (>0.5 mM) decreasing receptor abundance. Propionate exhibited the same biphasic receptor response but was shifted 10-fold rightward of the n-butyrate response. Fig. 2 illustrates that n-butyrate or propionate alone did not stimulate growth hormone production at any concentration examined. With L-T~, however, n-butyrate between 0.5 and 1 mM resulted in a 25 to 30% stimulation of growth hormone production compared to L -T~ alone. With higher nbutyrate concentrations, L -T~ stimulation of growth hormone synthesis decreased in parallel with the n-butyrate concentration-dependent decrease in receptor levels (Fig. 1). Stimulation of growth hormone synthesis by L -T~ was also influenced by propionate except that the response was shifted approximately 10-fold rightward of the n-butyrate-mediated effect.
n-Butyrate and propionate incubation can elicit a paradoxical prolactin response to L -T~ (Fig. 3). In contrast with the growth hormone response, n-butyrate alone (0.5 mM) increased the basal prolactin production 7.5-fold with a decrease in the response occurring at higher n-butyrate concentrations (Fig. 3). Surprisingly, prolactin production and mRNA levels, which are modestly inhibited by L -T~, were stimulated about 20-to 70-fold by L -T~ + n-butyrate (0.5-1 mM) (Figs. 3 and 6; Table 11). A similar response to propionate or L -T~ + propionate was also observed except that the response was again shifted 5-to 10-fold rightward of the n-butyrate effect.
Stimulation of each response by L -T~ + n-butyrate (0.75 mM) is secondary to an increase in the accumulation of prolactin and growth hormone mRNA (Fig. 6). Since growth hormone mRNA has an estimated half-life of 50 h (7), it is unlikely that further prolongation of the half-life could account for the extent of growth hormone mRNA accumulation observed in Fig. 6. In addition, a 1.7-1.8-kb nuclear precursor of prolactin mRNA (24) was identified in GH, cells incubated with L -T~ + n-butyrate (Fig. 6). These observations suggest that the L-T3 + n-butyrate stimulation of prolactin and growth hormone synthesis is secondary to mRNA accumulation and likely reflects mRNA synthesis rather than prolongation of mRNA half-life.
The kinetics of stimulation of growth hormone and prolac-tin production in cells cultured with L -T~ + n-butyrate shows a significant lag period (24 h) before the effect of L -T~ is observed (Fig. 4). After this initial lag period, L -T~ stimulation of both growth hormone and prolactin production increased progressively over the remaining 48 h of the experiment. The long lag period suggests several described n-butyrate effects which may alter thyroid hormone-regulated gene expression. These include alteration in DNA methylation (39), stimulation of CAMP production (34), control of cell cycle progression and arrest of the cells in the G1 period (44), and modification in the extent of phosphorylation (45), methylation (45), ADPribosylation (46), or acetylation (45) of chromatin-associated proteins.
Since n-butyrate incubation results in DNA hypomethylation and globin synthesis in Friend erythroleukemic cells (39), we examined whether n-butyrate incubation resulted in hypomethylation of growth hormone and prolactin gene sequences in GHl cells. Figs. 7 and 8 demonstrate that the HpaII-MspI restriction sites in both the growth hormone and prolactin genes are methylated and the AuaI site in both genes is unmethylated. We are not aware of a previous study which examined prolactin gene methylation in relation to stimulation or inhibition of the prolactin response. However, our findings for the growth hormone gene are in agreement with those of Moore et al. (47). L -T~ or n-butyrate f L -T~ induced no alteration in the extent of methylation of prolactin or growth hormone gene sequences based on this analysis. These studies exclude global but not site-specific effects of n-butyrate on methylation of these genes since our analysis is limited to those sequences which allow for isoschisomer restriction enzyme analysis.
Since Murdoch et al. (35) reported that forskolin, which markedly elevates cAMP levels, stimulates prolactin gene transcription and mRNA accumulation in GH4 cells, we examined the long term effect of forskolin on the prolactin response in GH1 cells. These studies (Table I) show that forskolin stimulated a 400-fold increase in cAMP production but had no effect on basal prolactin production and did not allow for stimulation of prolactin synthesis by L-T~. In the study by Murdoch et al. (35), the effect of cAMP was attributed to the phosphorylation of a M, = 23,000 basic nonhistone protein and it is possible that this regulatory protein is absent from GHl cells.
n-Butyrate enhancement of the hormonal stimulation of certain proteins (e.g. alkaline phosphatase) may be secondary to a GI arrest where hormonal regulation may be more pronounced than at other stages of the cell cycle (44). However, it is unlikely that the marked stimulation of prolactin mRNA by L -T~ + n-butyrate (Fig. 6) is explained by arrest in the GI phase. First, the concentrations of n-butyrate used to elicit the L -T~ response (0.75 mM) is lower than that (X.5 mM) shown to arrest other cell types in G1 (44), and we have previously shown that 0.75 mM n-butyrate does not alter cell growth or the DNA content per culture compared to control cells (14). Furthermore, 6 mM propionate, a concentration at which maximal L -T~ stimulation of growth hormone and prolactin occurs (Figs. 2 and 3), has been reported not to inhibit cell growth (13). Second, without n-butyrate, at least 40% of growing cells would be expected to be in the GI period (48). Therefore, asynchronous cells cultured without n-butyrate but with L -T~ should show some stimulation of prolactin synthesis or mRNA accumulation. Since L -T~ decreases prolactin production and mRNA levels (Figs. 5 and 6), an nbutyrate increase in the fraction of cells in the G1 phase does not provide an explanation for the paradoxical regulation of the prolactin response by L-T~. Kimura et al. (49) recently reported that inhibition of ADPribosylation in GH, cells resulted in a 3-to 4-fold increase in basal growth hormone and prolactin production and also further enhanced the L -T~ induction of growth hormone synthesis. However, inhibition of ADP-ribosylation did not result in stimulation of prolactin synthesis by L -T~ as observed with n-butyrate in our studies. Stimulation of phosphorylation of chromatin-associated proteins and histone methylation has not been examined in GHl or GH3 cells but have been reported to occur in other cell types using concentrations of n-butyrate of 5 mM or higher (45). Although our studies do not exclude these postsynthetic modification events, n-butyrate concentrations greater than 5 mM decrease rather than enhance the effect of L-T3 on the growth hormone and prolactin response.
The acetylation of core histones and other nuclear proteins is the most extensively examined effect of n-butyrate in cells (50). First demonstrated by Riggs et al. (10) in HeLa cells and in erythroleukemic cells, this effect has been shown to occur in a wide variety of cell lines (50). The effects of n-butyrate and propionate described in this paper are consistent with a mechanism which involves increased acetylation of histones or other chromosomal proteins. First, the 5-to 10-fold shift in the n-butyrate and propionate dose-response curves (Figs. 1-3) is in agreement with the potency of these compounds in stimulating acetylation as a consequence of their relative inhibitory effects on chromatin-associated deacetylase activity (12). Furthermore, the long n-butyrate incubation period required for prolactin stimulation by L -T~ is also consistent with an effect on the acetylation process. Although a large fraction of histones is deacetylated with a tlh of minutes, acetylation of a subset of the histone population only occurs after very long n-butyrate incubation times (13).
Although our studies suggest a relationship between inhibition of chromatin-associated deacetylase activity and L -T~ stimulation of prolactin and growth hormone mRNA production, the precise mechanism surrounding these events remains unclear. Analysis of n-butyrate effects in cells indicates that the carboxylic acid only stimulates or inhibits the expression of a small fraction of the genome (51). Of considerable interest is that low concentrations of n-butyrate reverse the prolactin response to L-T~. In most experiments, L -T~ modestly inhibits the prolactin response 25 to 50%. n-Butyrate alone moderately stimulates the prolactin response and with n-butyrate thyroid hormone acts as a potent stimulator of prolactin synthesis and mRNA production (Figs. 3-6; Table 11). This suggests that prolactin gene expression is under partial regulatory repression which is reversed as a consequence of nbutyrate-mediated postsynthetic modification events. This could reflect an alteration of chromatin structure, or postsynthetic modification and inactivation of a regulatory repressor(s) of prolactin gene expression.
The growth hormone and prolactin genes are closely related and are presumed to have arisen from gene duplication or other recombinant events (52). Both genes share many common sequences and show some sequence homology in the 5' region flanking the structural gene where hormone-receptor control of genes has been shown to occur (53). Therefore, both the growth hormone and prolactin genes may contain similar thyroid hormone receptor promotor sequences in the 5"flanking regions. A working model to explain our results is that repression of prolactin gene expression is reversed by nbutyrate incubation which allows for stimulation of gene expression by the thyroid hormone-receptor complex. This is observed at low concentrations of n-butyrate where no decrease in thyroid hormone receptor occurs. Within this framework, the modest L -T~ inhibition of prolactin production could occur by direct or indirect enhancement of a putative repressor component(s). Although this model is speculative it can be tested by appropriate transfection studies with a recombinant fusion gene containing the 5"flanking region of the prolactin gene.