Tissue-specific regulation of the human acute-phase serum amyloid A genes, SAA1 and SAA2, by glucocorticoids in hepatic and epithelial cells

2.1 Transcriptional regulation of transfected SAA1 and SAA2 promoters by cytokines and glucocorticoids in an epithelial cell line

Numerous reports have documented the induction of A-SAA mRNA, accompanied by an increase in A-SAA protein synthesis, after treatment of various hepatic cell lines with cytokines. It has further been shown that the cytokine-driven induction of A-SAA mRNA in such cell lines can be enhanced by co-treatment with dexamethasone 9, 23, 24. However, the detection methods employed in the above studies were based on Northern blotting and the use of hybridization probes that cannot distinguish between the products of the SAA1 and SAA2 genes. Consequently, none of the published reports have assessed the magnitude or kinetics of induction of the individual SAA1 and SAA2 mRNA. In a recent study using an RT-PCR assay that could discriminate between the two A-SAA mRNA, we were able to establish that dexamethasone alone could not induce either SAA1 or SAA2 transcriptional activity in HepG2 cells, an observation which was in agreement with previous studies using Northern blotting. However, published reports that dexamethasone alone could induce A-SAA mRNA accumulation in KB epithelial cells 9 suggested that there is tissue specificity in the mechanisms governing A-SAA transcriptional regulation. We therefore characterized and compared the kinetics of A-SAA mRNA isoform induction in response to cytokines and GC in both KB and HepG2 cells.

KB epithelial cells were subjected to single and various combination treatments with 10 ng/ml IL-1β, 10 ng/ml IL-6 and 50 nM dexamethasone for 2, 4, 6, 8, 12 and 24 h following transfection with pGL3-SAA1pt (Fig. 1A) or pGL3-SAA2pt (Fig. 1B). The SAA1 and SAA2 promoters both responded moderately to IL-1β alone, and more modestly to IL-6 alone; for each promoter the addition of IL-1β plus IL-6 was synergistic. In KB cells, as is the case for HepG2 cells 22, IL-1β-induced and combination IL-1β/IL-6 readouts from the SAA1 promoter, but not the SAA2 promoter, could be greatly enhanced by dexamethasone at all time points. However, the two cell lines exhibited markedly different responses to dexamethasone alone. The SAA1 promoter could be up-regulated by dexamethasone alone in KB cells, a response which is not observed in the hepatoma cell line 22. Interestingly, the magnitude of the dexamethasone-mediated transcriptional readout from the SAA1 promoter in KB cells was even greater than that driven by IL-1β. The SAA2 promoter is not transcriptionally activated in response to dexamethasone alone. This result, together with the above observation that exogenous IL-6 can induce SAA2 transcription, indicates thatdexamethasone does not act merely by stimulating the production of biologically active amounts of IL-6, or other cytokines, in KB cells.

2.2 Comparison of the transcriptional regulation of the endogenous SAA1 and SAA2 genes by cytokines and glucocorticoids in cultured epithelial and hepatoma cells

The responses of the endogenous A-SAA genes, in HepG2 and KB cells, to treatment with cytokines and/or dexamethasone for 0, 1, 3, 6, 9, 12, 24 and 48 h were evaluated by direct analysis of extracted RNA. Proportional RT-PCR was performed using a single primer-pair that binds to common complementary regions of both SAA1 and SAA2 cDNA. The primers generate products from the SAA1 and SAA2 mRNA that differ in size by 26 nucleotides, thereby allowing the relative levels of SAA1 and SAA2-derived transcripts to be assessed. In HepG2 cells treated with IL-1β (Fig. 2A), the percentages of total A-SAA product derived from SAA1 mRNA ranged from 47% at the 6 h (at which the SAA1 and SAA2 products are first observed) to 36% and 34% at 24 h and 48 h, respectively (Fig. 2F). This indicates that SAA2 has a transcriptional advantage over SAA1 that becomes established at a ratio of approximately 2:1 by 24 h. In contrast, co-administration of dexamethasone to IL-1β–treated HepG2 cells (Fig. 2B) reverses the ratio of RT-PCR products, clearly favoring those derived from SAA1 mRNA. At 3 h, SAA1 mRNA constitutes >80% of the total mRNA, suggesting that dexamethasone, in the presence of IL-1β, confers a marked early transcriptional advantage to the SAA1 gene. This transcriptional advantage is maintained at a slightly lower level from 9–48 h, such that the ratio of SAA1:SAA2-derived products is approximately 2:1. Thus, in the context of cytokine induction dexamethasone effects a switch in the relative transcriptional activities of SAA1 and SAA2.

Low levels of SAA1-derived product could be detected in untreated KB cells (Fig. 2C–E), but not in untreated HepG2 cells (Fig. 2A, B). This observation in the former cell type is in accord with in situ hybridization studies by Urieli-shoval et al. 8, which documented the presence of A-SAA mRNA in normal (i.e. uninflamed) human epithelium. Taken together, the above findings suggest that the SAA1 gene has a low level of constitutive activity in epithelial cells. In KB cells treated with IL-1β alone (Fig. 2C), there is a clear induction of SAA1-derived product as early as 1 h post-stimulus. By 3 h, products derived from both SAA1 and SAA2 mRNA are evident. And through 48 h, the ratio is approximately equal (Fig. 2G), indicating equivalent transcriptional activity of SAA1 and SAA2. In contrast, when dexamethasone is also present (Fig. 2D), SAA1 has a 2:1 transcriptional advantage over SAA2 (Fig. 2G), i.e. the impact of dexamethasone in the context of IL-1β induction is similar to that in HepG2 cells.

No A-SAA-derived PCR product was detected in HepG2 cells treated with dexamethasone alone (data not shown), whereas in KB cells treated with dexamethasone alone only the SAA1-derived product was detected (Fig. 2E). These results are consistent with those generated using isoform-specific promoter reporter constructs and underscore the striking differences in the kinetics of cytokine and GC induction of SAA1 and SAA2 transcription between HepG2 and KB cells (Fig. 2F, G).

2.3 HepG2 and KB cells do not have biologically significant levels of constitutive or dexamethasone activated NF-κB

It has been reported that many cancer cell lines have constitutively active NF-κB (reviewed in 25). To confirm that the transcriptional responses of the SAA1 gene described above are determined by properties intrinsic to its promoter, rather than by a particular capacity (i.e. that is not shared by SAA2) to respond to constitutively active NF-κB, and/or dexamethasone-dependent (i.e. IL-1β–independent) NF-κB activation, we performed a series of experiments using the NF-κB inhibitor curcumin (diferuloylmethane). Curcumin has been shown to prevent the up-regulation of various NF-κB inducible genes via its inhibition of the pre-requisite activation of IKK 26. HepG2 and KB cells were transfected with the PathDetect^® NF-κB reporter plasmid, pNF-κB-Luc (Stratagene), which has five NF-κB responsive sequences located in an artificial promoter upstream of luciferase. They were treated with medium only, curcumin (50 mM), IL-1β, or IL-1β plus curcumin for 4 h (Fig. 3A, B). The addition of curcumin alone did not reduce the low level of constitutive transcriptional readout in either cell line. However, in both HepG2 and KB cells, 50 mM curcumin was sufficient to completely abolish the IL-1β–driven component of transcriptional readout. These data indicate that any constitutively active NF-κB present in the cell lines under test is biologically insignificant with respect to the engagement of targets in promoters that have been optimized to respond to this transcription factor.

The above experiment did not eliminate the trivial explanation that the capacity of the SAA1 promoter to respond to dexamethasone may be due to GC being able to activate NF-κB via an undefined mechanism. KB cells were, therefore, transfected with pGL3-SAA1pt and treated with medium only, dexamethasone alone, or dexamethasone plus IL-1β in the presence or absence of curcumin (Fig. 3C). There was no significant difference in transcriptional readout from cells treated with medium only and with curcumin. As established above (Fig. 1A, 2E), the administration of dexamethasone alone causes transcriptional up-regulation of the SAA1 promoter. However, this was unaffected by the addition of curcumin, strongly suggesting that the mechanism of GC-mediated induction in KB cells is independent of NF-κB. The addition of dexamethasone plus IL-1β resulted in a marked and synergistic activation of the SAA1 promoter. This synergy was completely abolished by the co-administration of curcumin, which yielded a readout equivalent to that observed following treatment with dexamethasone alone. Thus, in contrast to the GC-driven component, the IL-1-driven component of the synergistic response appears to be entirely NF-κB-dependent.

2.4 Onset of SAA1 and SAA2 promoter induction by cytokines and glucocorticoids in cultured epithelial cells

To study the kinetics of A-SAA induction at very early time points post-stimulus, KB cells were transfected with pGL3-SAA1pt (Fig. 4A) or pGL3-SAA2pt (Fig. 4B) and subjected to single and various combination treatments with cytokines and dexamethasone for 15, 30, 45, 60, 90 and 120 min. The induction of both promoters by IL-6 began to become apparent at 30 min and became significant at 60 min (p=0.02 for SAA1, p=0.01 for SAA2). In HepG2 cells, it has previously been shown that the SAA2 promoter is induced more rapidly following IL-6 treatment than following IL-1β treatment 25. In contrast to the later time points shown in Fig. 1, there was no significant induction of either promoter by IL-1β alone, and the transcriptional readout in response to the combined treatment of IL-1β plus IL-6 was not significantly different from that observed in response to treatment with IL-6 alone at these early time points. The increase in transcriptional readout from the SAA1 promoter in response to treatment with GC was evident as early as 60 min, and reached significance by 90 min (p=0.006).

2.5 SAA1 glucocorticoid responsiveness is dose-dependent

In KB cells transfected with the SAA1 promoter luciferase construct and treated with increasing amounts of dexamethasone (from 10 nM to 1 μM), there was a clear dose response even in the absence of cytokines (Fig. 5A). In contrast, in HepG2 cells transfected with pGL2-SAA1pt, even very high concentrations of dexamethasone alone failed to activate the SAA1 promoter (Fig. 5B). However, in the presence of 10 ng/ml IL-1β plus 10 ng/ml IL-6, the GC-mediated enhancement of transcriptional activity was clearly dose-dependent (Fig. 5B).

It has been reported that HepG2 cells have low endogenous levels of GR 26. To eliminate the possibility that the lack of SAA1 response to GC alone in HepG2 cells was due to insufficient GR, or to treatment with doses of dexamethasone that failed to reach a threshold needed to activate GR in this cell type, HepG2 cells were co-transfected with a GR expression construct 27 and pGL2-SAA1pt, and treated with the highest concentration (1 μM) of dexamethasone (Fig. 5C). Even this superphysiological dose failed to induce transcription in the absence of cytokines. However, in HepG2 cells transfected as above and treated with the same dexamethasone concentration in the presence of fixed concentrations of IL-1β plus IL-6, there was an enhanced response, indicating that the lack of SAA1 response to GC alone could not be attributed to inadequate levels of GR.

2.6 SAA1 glucocorticoid responsiveness is GR and GRE-dependent

To confirm that the GC responsiveness of the SAA1 promoter in KB cells occurs via direct interaction of GR with the previously identified GRE positioned –208 to –194 bp upstream of the transcription start site, experiments were carried out using the GR antagonist RU486 (mifepristone) and cells transfected with native and mutant promoter constructs. KB cells were transfected with the wild-type SAA1 promoter luciferase construct and treated with 50 nM dexamethasone and/or 100 nM RU486 for 6 h (Fig. 6A). RU486 has been shown to exhibit partial agonist activity that is dependent on which corepressors are present 28. There was a slight, but non-significant, increase in transcriptional readout in the presence of RU486 (p=0.2). Treatment with dexamethasone resulted in an approximately 10-fold increase in transcriptional activity, an effect that was blocked by the addition of RU486. This establishes that GC-mediated induction of the SAA1 promoter in KB cells occurs via the GR.

We have previously documented the sequence and position of the GRE in SAA1 that is solely responsible for the GC-mediated enhancement of cytokine-driven transcription in HepG2 cells. When key residues in this sequence are mutated, the in vitro binding of GR is greatly reduced, and the enhancement of cytokine induction of the SAA1 promoter in HepG2 cells is completely abolished (data not shown). KB cells were transfected with a SAA1 promoter luciferase construct (GREM) in which critical residues in the GRE had been mutated. This construct exhibits a level of transcriptional activity in response to IL-1β and IL-6 that is the same as that of the native SAA1 promoter (Fig. 6B). However, in IL-1β plus IL-6-treated KB cells, there is no dexamethasone-mediated enhancement of transcriptional activity; nor does the mutant promoter respond to dexamethasone alone. Taken together, these data indicate that the GRE at positions –208 to –194 in the SAA1 promoter is functionally active in epithelial cells and is an absolute requirement for SAA1 engagement by GC in all cell types in the presence or absence of concomitant cytokine stimulation.

Kumon et al. 29 recently proposed the involvement of a C/EBP site in the GC-mediated up-regulation of SAA1 in human aortic smooth muscle cells. Here, we have shown that the up-regulation of SAA1 by GC in epithelial cells occurs exclusively via a GRE between positions –208 and –194 bp, the site which is also responsible for GC enhancement of cytokine induction of SAA1 transcription in HepG2 cells. Furthermore, we have also established that a SAA1 promoter luciferase reporter construct with a mutated C/EBP site is subject to the same GC-mediated enhancement of cytokine induction in HepG2 cells as the native SAA1 promoter (data not shown). It is, therefore, possible that the decrease in GC responsiveness observed by Kumon et al. 29 following deletion of the region from –252 to –174 bp of the SAA1 promoter was, in fact, due to coincident deletion of the GRE together with the C/EBP site.

4.1 Plasmids

The human A-SAA promoter luciferase reporter constructs pGL3-SAA1pt and pGL3-SAA2pt were generated by digestion of pGL2-SAA1pt and pGL2-SAA2pt 22 with Nco1 and Mlu1, and ligation of approximately 700 bp of upstream regulatory sequence and 5′UTR of the respective genes into pGL3 (Promega). The PathDetect^® NF-κB reporter plasmid, pNF-κB-Luc, was purchased from Stratagene. A luciferase reporter construct containing the SAA1 promoter with a mutated GRE (GREM) was generated by site-directed mutagenesis using the primers 5′-CATCTTTTGGGCTCAGGTTGC-3′ and 5′-AGCCCAAAAGATGTGCCTGG-3′. The constitutive human Glucocorticoid Receptor-α expression plasmid, CMX-GR 27, was a gift of Dr Ron Evans, The Salk Institute, La Jolla, CA.

4.2 Cell culture and transient transfection

Human HepG2 hepatoma cells and KB epithelial cells (ATCC) were cultured in DMEM containing 10% FCS, gentamycin, sodium pyruvate and non-essential amino acids (Gibco BRL). Cells were seeded into 24-well plates, 24 h prior to transfection using FuGENE® (Roche Biomolecular) as previously described 35. Cells transfected with A-SAA promoter luciferase reporter constructs together with renilla control plasmid, were incubated for 16–20 h before replacement of culture medium with fresh medium alone or fresh medium containing 10 ng/ml cytokine(s) and/or 50 nM dexamethasone and/or 100 nM RU486 (Mifepristone). IL-1β was purchased from Peprotech. IL-6 was obtained from Astra Zeneca. Dexamethasone, curcumin and RU486 were from Sigma.

4.3 Luciferase assays

Cells were harvested at various times post-treatment, washed in PBS and resuspended in Passive Lysis Buffer (Promega). Lysates were assayed for luciferase and renilla activity using the LLR and Stop and Glo reagents (Promega) in a dual injection luminometer (Turner Designs). Each treatment was carried out in triplicate and the mean ratio of luciferase to renilla activity and standard deviations were calculated. The ratios are expressed relative to untreated controls and are representative of three independent experiments.

4.4 RT-PCR

Total RNA was prepared using RNeasy columns (Qiagen) from HepG2 and KB cells treated for up to 72 h under various experimental conditions. A-SAA RT-PCR was as described previously 22. PCR samples were run on 9% polyacrylamide minigels, stained with ethidium bromide and visualized using a UVP imaging system and Labworks 4.0 software.