Combined effects of retinoic acid and histone deacetylase inhibitors on human neuroblastoma SH-SY5Y cells

Histone deacetylase inhibitors (HDACi) constitute a promising treatment for cancer therapy due to their low toxicity, and first-generation HDACi are currently being tested in phase I/II clinical trials (1, 2). There are distinct classes of HDACi, including, among others, short-chain fatty acids such as butyrate and hydroxamic acids such as trichostatin A (TSA) and suberoylanilide hydroxamic acid (SAHA) that function by binding to the catalytic site of the enzyme (3).

During tumorigenesis, histone hypoacetylation can result in the silencing of genes involved in regulation of cell growth, differentiation, and apoptosis. HDACi increase acetylation of histones and transcription factors (4), which can reverse gene silencing facilitating gene expression (5). However, not all genes are up-regulated by treatment with HDACi, and the ratio of up-regulated to down-regulated genes has been found to be close to 1:1 (2). HDACi have been shown to induce G₁-phase cell cycle arrest with up-regulation of p21^Waf1/Cip1 (6, 7), which was associated with an Sp1 site in the gene promoter (8–11). These changes can result in cell type–specific effects, including reduced proliferation and metabolic activity, induction of apoptosis, and differentiation (12–15).

Neuroblastoma accounts for more than 15% of cancer-related deaths in early childhood (16). Thus, new drugs are needed for the treatment of this disease. The finding that all-trans retinoic acid (RA) causes differentiation of neuroblastoma cells (17) led to clinical trials that showed a substantial clinical benefit for retinoid therapy in advanced neuroblastoma, although 50% of children still relapsed (18).

The effects of RA are mediated by nuclear receptors. In the absence of ligand, these receptors act as transcriptional repressors due to the binding of corepressor complexes that contain histone deacetylases (HDAC). Ligand binding allows the release of corepressors and facilitates the ordered recruitment of coactivator complexes, some of which possess histone acetylase activity, that cause transcriptional stimulation (19). Several retinoids are able also to inhibit the activator protein 1 (AP-1) transcription pathway, which is activated upon growth factor signaling (19, 20), or even to have non-genomic effects that lead to stimulation of signaling pathways involved in neuronal differentiation (21).

Combination therapy of RA with HDACi may improve efficacy while reducing side effects. An enhanced inhibitory effect on neuroblastoma cell growth when the HDACi m-carboxycinnamic acid bishydroxamide (CBHA) was combined with RA has been found (22). To further investigate the effects of deacetylase inhibition, we compared the potency of the HDACi sodium butyrate, TSA, and SAHA alone and in combination with RA to inhibit growth of human SH-SY5Y neuroblastoma cells. This cell line has proven to be a valuable model for studying the effects of RA on neuronal differentiation, including extension of neuritic processes and growth arrest (23, 24).

SH-SY5Y cells were grown in RPMI containing 10% FCS (24). Twenty-four hours before the beginning of treatments, cells were shifted to medium containing AGX100 resin-charcoal–treated serum to eliminate retinoids. Cytotoxicity was determined by colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay as recommended by the manufacturer (Roche, Mannheim, Germany). Absorbance was read at 550 and 690 nm. Cells were plated in 96-well plates (7 × 10³ cells per well), and assays were done after incubation for 24 and 48 h with RA and HDACi. When the caspase family inhibitor (fluoromethylketone) Z-Val-Ala-Asp (OMe)-FMK (Z-VAD) was used, cells were inoculated at a density of 20 × 10³ cells per well. DNA synthesis was determined by [³H]thymidine incorporation in cells inoculated in six-multiwell plates (6 × 10⁵ cells per well). Cells were treated for 24 h with the different compounds and for the last 2 h with 2 μCi per well of [³H]thymidine. Cells were disrupted, and incorporated radioactivity was determined in a liquid scintillation Beta Wallac Counter. For counting, cells were inoculated in six-well plates at 5 × 10⁴ cells per well, treated 24 h later with the different compounds and counted every 2 days in Neubauer chambers. Cell morphology was assessed at 200× in a Zeiss inverted phase contrast microscope.

Triplicate cultures of SH-SY5Y cells grown in 60-mm Petri dishes were transferred to the medium containing depleted serum and, after 24 h, incubated with the different compounds for 48 h. Both floating and adherent cells were collected, washed twice with cold PBS, fixed with chilled 70% ethanol, and centrifuged. Pellets were incubated for 30 min at 37°C with 0.1 μg/mL RNase A and stained with propidium iodide (50 μg/mL) for sorting in FACScan (Becton Dickinson, San Jose, CA) cell sorter. Percentages of cells in sub-G₁, G₁, S, and G₂-M phases were calculated with WinMDI and Cylchred software for Windows.

Cell lysates were obtained as previously described (25). Proteins were separated in SDS-PAGE and transferred to polyvinylidene diflouride membranes (Immobilon; Millipore, Bedford, MA) that were blocked for 1 h at room temperature with 4% bovine serum albumin. Incubation with primary antibodies (25, 26) was done overnight at 4°C and with the secondary antibody for 1 h at room temperature. Blots were visualized with enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ). Antibodies against cyclin D1, p21^Waf1/Cip1, p27^Kip1, and poly(ADP-ribose) polymerase (PARP) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and used at a 1:2,000 dilution. The antibody for acetylated histone H3 in Lys⁹ and Lys¹⁴ (Upstate, Charlottesville, VA) was used at a 1:5,000 dilution. Membranes were stripped, blocked, and reprobed with antihuman actin (Santa Cruz Biotechnology) or β-catenin (BD Biosciences, Bedford, MA) antibodies.

Total RNA from cells grown in 90-mm dishes was isolated with TriReagent (Sigma, St. Louis, MO), following the manufacturer's protocol. Total RNA (15 μg) was run in 1% formaldehyde-MOPS agarose gels and transferred to nylon Nytran-N (Schleicher & Schuell, Dassel, Germany) as described (24–26). Then, RNA was linked to membrane using UV-Stratalinker and stained with 0.02% methylene blue to assess equal loading. Probes were labeled by random primer with Ready-To-Go DNA Labeling kit and [α-³²P]dCTP (Amersham). The labeled probes were hybridized with specific probes for cyclin D1, p21^Waf1/Cip1, and p27^Kip1 as described (25, 26).

SH-SY5Y cells (1 × 10⁶) were plated in 60-mm wells, changed to serum-free medium 24 h later and transfected with 5 μg of luciferase reporter plasmids containing 2.3 kb of the human p21^Waf1/Cip1 promoter and 1.9 kb of the human p27^Kip1 promoter (26). Transfection was obtained by incubation with a mixture of cationic liposomes (1.5 μL/μg DNA) for 6 h. Cells were then treated for 48 h with RA and/or HDACi, and activity was determined. Each experiment was done in triplicate and was repeated at least thrice. Data are mean ± SD and are expressed as fold induction over the values obtained in the untreated cells.

Incubation of SH-SY5Y cells during 6 h with TSA, sodium butyrate, and SAHA caused a dose-dependent increase of histone H3 acetylation. A detectable increase was observed with the lower dose of TSA used, 50 nmol/L, and 100 nmol/L TSA caused an acetylation at least as strong as that produced by 2 mmol/L sodium butyrate or 1 μmol/L SAHA (Fig. 1A). However, the effect of these TSA concentrations on H3 acetylation is transient. Whereas sodium butyrate and SAHA produced a sustained increase in histone H3 acetylation that was very strong after 48 h, concentrations of 200 nmol/L or lower TSA had little effect at this time point (Fig. 1A). As shown in Fig. 1B, 100 nmol/L TSA caused a stronger acetylation than 2 mmol/L sodium butyrate after 6 h of incubation, but the opposite was true after 24 h. Re-addition of TSA after 24 h again caused a strong increase in the levels of acetylated H3 histone (Fig. 1C), suggesting that this reduction results from a rapid metabolism of TSA in SH-SY5Y cells rather than being secondary to changes in HDAC sensitivity to this compound.

By using MTT assays to measure viability, the dose- and time-dependent effects observed showed that significant killing of SH-SY5Y cells by HDACi required 48 h. At 24 h (Fig. 2A, left), the different HDACi had little effect even at high concentrations both in the presence and absence of RA. However, after 48 h, concentrations of 100 nmol/L TSA, 1 mmol/L sodium butyrate, or 0.5 μmol/L SAHA were ineffective, but higher concentrations (200 and 500 nmol/L TSA; 2 and 5 mmol/L sodium butyrate; and 1 and 2 μmol/L SAHA) caused a strong dose-dependent decrease of cell viability. Whereas RA (1 μmol/L) did not reduce this parameter, the retinoid was able to increase the effect of low HDACi concentrations, although a strong synergistic effect was not found (Fig. 2A, right). Direct cell counting also showed a marked effect of HDACi on SH-SY5Y cell viability. Figure 2B shows that whereas control cells proliferate rapidly, at 48 h, 200 nmol/L TSA, 2 mmol/L sodium butyrate, and 1 μmol/L SAHA caused a strong inhibition of cell proliferation, and that essentially, no viable cells remained after 4 days of treatment with the HDACi. An inhibitory effect of HDACi on DNA synthesis was already observed at 24 h (Fig. 2C). Concentrations higher than 200 nmol/L TSA, 2 mmol/L sodium butyrate, or 0.5 μmol/L SAHA reduced thymidine incorporation, and RA, which only caused a weak dose-dependent inhibition, again potentiated HDACi action.

In SH-SY5Y cells, RA treatment for 48 h induced morphologic differentiation with neurite extension (Fig. 3). In contrast, morphology after treatment with HDACi (200 mmol/L TSA, 2 mmol/L sodium butyrate, or 1 μmol/L SAHA) was typical of cells undergoing cell death, with neurite shortening and the appearance of cells in suspension. This occurred both in the presence and absence of RA (Fig. 3).

Flow cytometry analysis confirmed that HDACi produced SH-SY5Y cell death, manifested by accumulation of sub-G₁ cell debris. Approximately 60% of cells treated with 2 mmol/L sodium butyrate or 1 μmol/L SAHA were in sub-G₁ both in the presence and absence of 1 μmol/L RA after 48 h of incubation (Fig. 4A). These compounds also caused a significant decrease in the number of cells in the S phase. In contrast, 100 nmol/L TSA had little effect on cell death or cell cycle progression, showing that a sustained increase in histone acetylation is required for these HDACi actions.

Proteolysis of PARP after incubation for 48 h with 200 nmol/L TSA, 2 mmol/L sodium butyrate, or 1 μmol/L SAHA (Fig. 4B) showed that HDACi caused apoptotic cell death. In contrast, incubation with 1 μmol/L RA for the same time period did not alter cell cycle distribution and did not induce apoptotic SH-SY5Y cell death.

To determine if cell death caused by HDACi is caspase dependent, we first determined PARP proteolysis in the presence and absence of the caspase inhibitor Z-VAD. As shown in Fig. 5A, the 89-kDa band representing cleavage of the protein was found after 24 h of incubation with 5 mmol/L sodium butyrate or 2 μmol/L SAHA alone, but this band was absent or highly reduced in cells treated with the combination of HDACi and Z-VAD. The effect of the caspase family inhibitor was also examined in MTT assays done in cells treated with sodium butyrate (2 and 5 mmol/L) and SAHA (1 and 2 μmol/L). Figure 5B shows that at 24 h incubation, these compounds had little effect on cell viability, in agreement with the results shown in Fig. 2A. However, after 48 h, both inhibitors caused a dose-dependent reduction of cell viability that was significantly reversed in the presence of Z-VAD, indicating the participation of caspases in HDACi-mediated apoptosis of SH-SY5Y cells.

The effect of 24 h incubation with the various HDACi on histone H3 acetylation paralleled closely the regulation of cyclin D1 and cyclin kinase inhibitors (CKI) levels in SH-SY5Y cells (Fig. 6A). TSA was effective at 200 nmol/L and, at 500 nmol/L that induced a very strong histone H3 acetylation, caused a marked reduction of cyclin D1 and a strong increase of p21^Waf1/Cip1. Sodium butyrate was effective from 0.5 mmol/L, causing a progressive decrease of cyclin D1 and a concomitant induction of the CKI, and the same was true for SAHA between 0.5 and 2 μmol/L. RA did not increase the levels of acetylated H3 and did not reduce cyclin D1 expression. However, an increase in p21^Waf1/Cip1 expression was also found after treatment with 10 μmol/L RA (Fig. 6A). Moreover, the combination of SAHA or sodium butyrate with 1 μmol/L RA did not cause a further reduction of cyclin D1 expression, but up-regulation of p21^Waf1/Cip1 levels by the HDACi was further increased in the presence of the retinoid (Fig. 6B), suggesting that regulation of this CKI plays a more important role than cyclin D1 on the repression of SH-5Y5Y cell growth found with the combined treatment. Because other CKIs could also contribute to the arrest of cell growth observed after HDACi and RA treatment, the levels of p27^Kip1 were also measured. As shown in Fig. 6C, the levels of this inhibitor were also elevated after the treatments, although less strongly than p21^Waf1/Cip1.

When the time course of the effect of sodium butyrate (2 mmol/L) on CKI expression was analyzed, it was found that the induction of p21^Waf1/Cip1 and p27^Kip1 levels followed similar kinetics. No CKI induction was observed after 1 and 3 h of incubation with the inhibitor, but a detectable increase was found between 5 and 7 h (Fig. 6D). The influence of increasing concentrations of the different HDACi, as well as RA, was then examined after 6 h of treatment (Fig. 6E). Whereas at 24 h, an increase in CKI expression was only detected at high TSA concentrations (see Fig. 6A), at 6 h of incubation, the increase in p21^Waf1/Cip1 and p27^Kip1 levels were maximal at a TSA concentration as low as 50 nmol/L. These results match again the stimulation of histone H3 acetylation because the actions of low TSA concentrations were transient, and a strong acetylation after 24 h was only obtained at high concentrations of this inhibitor. This contrasts with the effects of butyrate that caused a sustained induction of CKI and histone acetylation at all concentrations used. However, RA was also effective in increasing the CKIs after 6 h of treatment even at 0.1 μmol/L.

In agreement with previous reports indicating that HDACi can induce p21^Waf1/Cip1 gene transcription, treatment with these compounds at concentrations that caused H3 acetylation also increased the levels of p21^Waf1/Cip1 transcripts in SH-SY5Y cells (Fig. 7A). In addition, these levels were further induced in the presence of RA. p27^Kip1 mRNA levels also increased after incubation with 2 mmol/L sodium butyrate and 1 μmol/L SAHA, and RA further potentiated this increase. Cooperation of the retinoid with HDACi to stimulate activity of a luciferase reporter plasmid containing the p21^Waf1/Cip1 promoter in transient transfection assays was also found (Fig. 7B). RA increased by 2-fold promoter activity, and the various HDACi were able to further induce this stimulation. TSA at 200 nmol/L was less potent than 2 mmol/L sodium butyrate or 1 μmol/L SAHA for this induction. Induction of p27^Kip1 promoter activity was also found after the treatments. RA was able to induce by 3- to 4-fold reporter activity and to cooperate with the HDACi to stimulate the promoter (Fig. 7C).

We have compared the effect of the HDACi sodium butyrate with that caused by two hydroxamic acid derivatives, TSA and SAHA, on human SH-SY5Y neuroblastoma cells. It has been shown that TSA at nanomolar concentrations causes a strong accumulation of acetylated histones in cells and blocks HDAC activity in vitro (27). Our work confirms that low concentrations of TSA cause a strong increase in histone H3 acetylation in SH-SY5Y cells. However, the effect of this inhibitor is transient when used at doses lower than 200 nmol/L. This could result from a short life span of this compound in these cells because re-addition of TSA again increases acetylation. The effects of sodium butyrate, at millimolar concentrations, or SAHA, a second-generation HDACi (28) at micromolar concentrations, are more stable, being able to increase acetylation even after 48 h. Interestingly, SAHA has been successful in phase II trials for the treatment of cutaneous T cell lymphoma (29) and is the first HDACi approved by the Food and Drug Administration to enter the clinical oncology market.

The concentrations of HDACi that produced long-standing histone hyperacetylation also caused strong growth inhibition. Inhibition is a consequence of increase in apoptotic cell death, as indicated by the appearance of cells in sub-G₁, and the induction of PARP proteolysis, a late apoptotic event. HDACi-induced apoptosis seems to be at least partially caspase dependent because cell death was significantly attenuated in the presence of a caspase family inhibitor. Besides inducing apoptosis, in the surviving cells, HDACi also caused a significant decrease in the number of cells in G₁ and S phases. Because cyclin D1 is essential for G₁ entry, this reduction could be related to the decrease of this protein found in HDACi-treated cells. It has been shown that the HDACi BL1521A reduces cyclin D1 transcripts in neuroblastoma cells (30), and we also observed this reduction in SH-SY5Y cells treated with sodium butyrate, TSA, or SAHA. However, the possibility that HDACi could also alter cyclin D1 stability in agreement with the very recent finding that TSA induces ubiquitin-dependent degradation of cyclin D1 in breast cancer cells (31) cannot be excluded.

Cyclin-dependent kinase (CDK) activity is counteracted by CKI. The WAF/KIP family of CKIs that includes, among others, p21^Waf1/Cip1 and p27^Kip1 inhibits kinase activity of G₁-S CDKs (32). In agreement with observations in other cells (6–11, 33, 34), HDACi caused an important increase of p21^Waf1/Cip1 in SH-SY5Y cells. It has been suggested that this induction is a critical effector of HDACi for cell cycle arrest, and that cells lacking this CKI are insensitive to these compounds (35). However, butyrate can induce G₁ arrest in mouse embryonic fibroblasts lacking p21^Waf1/Cip1, suggesting that cyclin D1 could be the key factor in this inhibition (36). In any case, growth arrest promoted by HDACi in SH-SY5Y cells was accompanied by both cyclin D1 reduction and p21^Waf1/Cip1 induction. This CKI was not the only inhibitor induced by these agents because the levels of p27^Kip1 were also increased upon incubation with HDACi. The increase in p27^Kip1 most likely contributes to the inhibition of growth by HDACi in SH-SY5Y cells. In contrast with p21^Waf1/Cip1 that seems to be a common target for different HDACi, induction of p27^Kip1 seems to be more selective. For instance, the HDACi TSA and valproic acid, but not butyrate or BL1521, have been found to induce this protein in other neuroblastoma cell lines (7, 33, 34).

The influence of HDACi on p21^Waf1/Cip1 levels is due to transcriptional stimulation. Most likely, Sp1 binding motifs that mediate p21^Waf1/Cip1 induction in other cell types (9–11) also mediate induction in SH-SY5Y cells. An increase in the amount of acetylated histone H3 associated to the p21^Waf1/Cip1 promoter (8) as well as a reduction in HDAC1 binding (37) has been obtained by chromatin immunoprecipitation assays after treatment of other cells with HDACi. This should also occur in SH-5Y5Y cells, in which an increase in H3 acetylation is found. Our data also show that HDACi increase p27^Kip1 mRNA levels and stimulate p27^Kip1 promoter activity in transfection assays. Although the promoter elements that mediate p27^Kip1 gene transcription by HDACi have not been yet defined, most likely, increased association of acetylated histones with the promoter is involved in this stimulation.

Retinoids have been used in clinical trials in patients with neuroblastoma (16–18), although their therapeutic use is limited by their toxicity. Our data show that the effect of RA in SH-SY5Y cell growth is much weaker than that of HDACi. In contrast with HDACi, treatment of SH-SY5Y cells with RA did not cause apoptosis. This could be related to increased expression of Bcl-2 (38), an effective apoptosis suppressor. In addition, contrary to that found with HDACi, RA did not reduce cyclin D1. However, in agreement with previous results in neuroblastoma cells (24, 39), RA caused CKI induction. Our data show that the increase in p21^Waf1/Cip1 and p27^Kip1 seems to be at least partially due to transcriptional stimulation because RA increased CKI transcripts and caused promoter activation in reporter assays. In addition, it has been recently reported that RA induces p27^Kip1 phosphorylation that leads to its stabilization and accumulation into the nuclear compartment (40). The promoter elements responsible for the induction of p27^Kip1 gene expression by RA are still unknown, but the increase of p21^Waf1/Cip1 promoter activity by RA could depend on a promoter response element that mediates p21^Waf1/Cip1 induction by RA in leukemia cells (41).

It has been proposed that the combination of retinoids with other drugs could result in improved antitumorigenic activity, allowing the use of low retinoid concentrations and reducing its side effects (20). The HDACi CBHA and RA have been described to inhibit neuroblastoma cell growth in significantly lower concentrations when used together than when used individually (23). Our results show a clear cooperation of RA with other HDACi to increase p21^Waf1/Cip1 and p27^Kip1 levels in SH-SY5Y cells that might be involved in the stronger reduction in cell proliferation found when both agents were used together. This cooperation was also observed for mRNA induction and promoter activation, confirming their transcriptional effect on CKI gene expression. In contrast, RA did not increase HDACi-mediated cell death and did not further reduce cyclin D1 expression. This could explain why strong synergistic effects of both types of drugs on SH-SY5Y growth arrest were not found.

Our results, as well as the finding that HDACi can suppress growth in human-mouse neuroblastoma xenografts (13, 42), suggest that HDACi that cause a sustained increase of histone acetylation could be promising therapeutic agents for neuroblastoma treatment. Their antitumorigenic effects could be stronger than those of retinoids, and both types of drugs could cooperate to inhibit neuroblastoma growth.

We thank Aton Pharma (Tarrytown, NY) for the gift of SAHA. Maxy De los Santos is a Fellow of Fundación Carolina.