Histone deacetylase inhibitors (HDACi) are potential candidates for therapeutic approaches in cancer and neurodegenerative diseases such as spinal muscular atrophy (SMA)—a common autosomal recessive disorder and frequent cause of early childhood death. SMA is caused by homozygous absence of SMN1 . Importantly, all SMA patients carry a nearly identical copy gene, SMN2 , that produces only minor levels of correctly spliced full-length transcripts and SMN protein. Since an increased number of SMN2 copies strongly correlates with a milder SMA phenotype, activation or stabilization of SMN2 is considered as a therapeutic strategy. However, clinical trials demonstrated effectiveness of the HDACi valproate (VPA) and phenylbutyrate only in <50% of patients; therefore, identification of new drugs is of vital importance. Here we characterize the novel hydroxamic acid LBH589, an HDACi already widely used in cancer clinical trials. LBH589 treatment of human SMA fibroblasts induced up to 10-fold elevated SMN levels, the highest ever reported, accompanied by a markedly increased number of gems. FL-SMN2 levels were increased 2–3-fold by transcription activation via SMN2 promoter H3K9 hyperacetylation and restoration of correct splicing via elevated hTRA2-β1 levels. Furthermore, LBH589 stabilizes SMN by reducing its ubiquitinylation as well as favouring incorporation into the SMN complex. Cytotoxic effects were not detectable at SMN2 activating concentrations. Notably, LBH589 also induces SMN2 expression in SMA fibroblasts inert to VPA, in human neural stem cells and in the spinal cord of SMN2 -transgenic mice. Hence, LBH589, which is active already at nanomolar doses, is a highly promising candidate for SMA therapy.
Spinal muscular atrophy (SMA) is a common autosomal recessively inherited disorder characterized by the degeneration of α-motoneurons in the anterior horns of the spinal cord resulting in muscular atrophy and weakness of proximal muscles. The incidence of SMA is 1:6000 live births and the carrier frequency 1:35 in the western European population ( 1 , 2 ). Depending on the age of onset and motor milestones achieved, SMA has been subdivided into four types. Approximately 50% of all SMA patients develop type I SMA, are never able to sit and walk unassisted, and usually die before the age of 2 years. Patients with type II SMA are able to sit but never to walk, while patients with type III SMA succeed in walking, but often become wheelchair bound during progression of the disease ( 3 ). The mildest form, type IV SMA, is characterized by the occurrence of first symptoms after the third decade of life and slow disease progression ( 4 ).
The disease determining survival motor neuron gene (SMN) is present in a telomeric and a centromeric version on chromosome 5q13 ( 5 ). While homozygous absence or mutation of the telomeric copy ( SMN1 ) has been determined to cause SMA, each patient retains at least one nearly identical centromeric copy gene, termed SMN2 ( 5 , 6 ). SMN2 differs from SMN1 by five single nucleotides, only one of which is located in the translated region ( 5 ). Although the C to T transition in exon 7 is a silent mutation, it modifies SMN2 pre-mRNA splicing ( 7 , 8 ). In contrast to SMN1 , which almost exclusively produces full-length transcripts ( FL-SMN1) , SMN2 merely produces 10% FL-transcripts whereas 90% of transcripts lack exon 7 ( SMN2Δ7 ). Correct SMN2 splicing is abolished since the C to T transition within exon 7 destroys an exonic splicing enhancer (ESE) recognition site for SF2/ASF ( 9 , 10 ) and instead creates a novel exonic splicing silencer (ESS) recognized by hnRNP A1 ( 11 , 12 ). In the central part of exon 7 of both SMN genes an additional ESE is present which is recognized by a network of splicing factors ( 13–16 ). Among these, hTRA2-β1, which directly binds to exon 7 SMN pre-mRNA, has been shown to be the most important one, since its over-expression restores the correct SMN2 splicing pattern to almost 80% FL-SMN2 transcripts ( 13 , 14 ).
In contrast to FL-SMN2 , SMN2Δ7 encodes a truncated and unstable protein ( 17 ), which is unable to fully compensate for the loss of SMN1 .
The SMN protein is ubiquitously expressed in all tissues, although particularly high in spinal cord ( 18 , 19 ). It is still not entirely understood why reduced SMN level affects exclusively α-motoneurons but no other cell types. In concordance with the fundamental role of SMN in snRNP-biogenesis [reviewed in ( 20 )], recent studies suggest that severely reduced SMN levels cause extensive splicing defects in a large number of transcripts ( 21 ), without indicating which transcripts may impair motoneuron survival. It is therefore more likely that SMA is caused by a rather specific involvement of SMN in motoneuron biology such as axonal growth and pathfinding ( 22–24 ), mRNA transport along processes to neuronal growth cones ( 25 , 26 ) or its essential role in maintaining neuromuscular junctions ( 27 , 28 ).
SMN2 is the most important SMA modifying gene that influences the SMA phenotype in all SMA patients ( 1 , 29 ). The more SMN2 copies a patient has, the milder the SMA phenotype is ( 1 , 29–31 ). Since SMN2 is present in each patient and encodes for the same protein as SMN1 , it is considered as the major target for a potential SMA therapy. Enhancing SMN2 transcription and/or correction of its pathological splicing pattern as well as SMN protein stabilization are assumed to be effective strategies achieving clinical improvements in patients with SMA [reviewed in ( 32 )]. In rare cases, an additional modifier, plastin 3, has been found to strongly protect against SMA ( 24 ) but there is not yet any therapeutic strategy targeting it.
Quite a few substances, including the HDACi sodium butyrate ( 33 ), valproic acid (VPA) ( 34 , 35 ), phenylbutyrate ( 36 , 37 ), suberoylanilide hydroxamic acid (SAHA) ( 38–40 ), M344 ( 41 ) and trichostatin A (TSA) ( 42 , 43 ), have been shown to activate SMN2 transcription and to modulate its splicing thereby increasing FL-SMN2 amounts in vitro . Moreover, the promising in vitro data of sodium butyrate, VPA and TSA have already been corroborated in transgenic SMA mouse models carrying the human SMN2 on an Smn null background ( 33 , 43 , 44 ). In all three cases, treatment with HDACi led to improved motor abilities and attenuated neurodegeneration. However, the use of VPA in a pilot clinical trial in SMA patients revealed elevated FL-SMN2 levels in blood only in one-third of them ( 45 , 46 ). Moreover, clinical improvements were recorded only in some but not in all treated patients, emphasizing that indeed a certain percentage of SMA patients are non-responders to VPA ( 47 ). Similar results have also been obtained in a first clinical trial with phenylbutyrate ( 37 ). Therefore, the search for novel and more potent drugs is essential to advance causal SMA therapy.
LBH589 (Panobinostat), a novel hydroxamic acid-derived HDACi, already demonstrated its high potential as a putative anti-cancer drug in numerous cancer cell lines and reached orphan drug status for the treatment of cutaneous T-cell lymphoma (CTCL) by the FDA in 2007 [reviewed in ( 48 )]. In the present study, we identified LBH589 to markedly increase SMN protein amounts at nanomolar doses in SMN1 -deleted fibroblast cell lines derived from SMA patients, in human neural stem cells (NSC) and in murine fibroblasts derived from SMN2 -transgenic mice. We performed a detailed analysis of different levels of SMN2 gene and protein regulation which may be affected by LBH589 treatment including expression, splicing and promoter acetylation. Furthermore, we were able to show that SMN is post-translationally stabilized via reduced ubiquitinylation as well as enhanced incorporation into the SMN complex. Most importantly, LBH589 also triggers SMN expression in cell lines from SMA patients who are non- or negative responders to VPA, underlining the potential of LBH589 as a putative SMA drug. Finally, LBH589 passes the murine blood brain barrier as shown by increased SMN expression in neuronal tissues of SMN2 transgenic mice injected with LBH589.
Treatment with LBH589 increases SMN protein levels in fibroblast cell lines derived from SMA patients
To assess whether LBH589 is able to influence SMN2 protein expression, SMN1 -deleted fibroblasts from three different SMA patients, ML16, ML17 and ML5 ( 34 ), were treated with LBH589. First, we determined the optimal drug concentration and treatment time. To do so, fibroblasts were treated in a kinetic experiment for time periods between 24 and 96 h with different concentrations of LBH589 ranging from 10 n m to 1 µ m and analysed for SMN expression (data not shown). It turned out that treatment for 64 h with LBH589 resulted in the highest up-regulation of SMN when compared with mock-treated cells, which were treated with solvent only (DMSO). On the other hand, concentrations below 100 n m did not increase SMN amounts within all time periods tested (data not shown). On the basis of these findings, cells were treated in all subsequent experiments with a single dose of LBH589 for 64 h, resulting in final concentrations between 100 n m and 1 µ m LBH589. Furthermore, we checked several housekeeping genes such as GAPDH or β-Actin for their expression under LBH589 regimen and found the latter one to be stably expressed (data not shown).
Next, three passages from each cell line were treated under the above-mentioned conditions and analysed via quantitative western blotting (Fig. 1 A and B). We observed a significant up-regulation of SMN with highest values between 6- and 8-fold at concentrations ranging between 300 n m and 750 n m when compared with mock-treated cells. Of note, a significant increase in SMN amounts of up to 5-fold was already observed at concentrations of 200 n m LBH589. Moreover, 100 n m LBH589 already led to a doubling of the SMN amounts (Fig. 1 A).
To further ascertain that the increased SMN amounts represent indeed functional protein, the intracellular localization of SMN in LBH589-treated cells was analysed using fluorescence microscopy. In general, SMN is uniformly distributed in the cytoplasm, whereas in the nucleus distinct foci, called gems, are formed ( 49 ). Since the number of gems correlates with the amount of SMN protein ( 19 ), the increase of SMN after LBH589 treatment should be reflected by an increased number of gems. Comparison of LBH589 to mock-treated cells revealed an approximately 7-fold increase of the overall number of cells containing gems (Fig. 1 C). Moreover, not only the proportion of cells containing a single gem increased, but also the number of cells containing two or more gems (Fig. 1 C). These data corroborate the previous western blot results showing similar increases in SMN levels.
On the basis of these data we next checked whether other SMN complex components, namely Gemin2 and Gemin3, are likewise up-regulated under LBH589 regimen or whether exclusively SMN is affected. Analysing RNA extracted from treated SMN1 -deleted fibroblasts (ML5) by quantitative real-time RT-PCR (qRT-PCR) we found that following LBH589 treatment Gemin2 transcripts were elevated up to 3-fold whereas those derived from Gemin3 doubled (Fig. 1 D). To test whether Gemin2 and -3 are generally induced by LBH589 or if their up-regulation is a mere consequence of the vast SMN up-regulation in SMN1 -deleted fibroblasts, we also checked their expression on RNA level in fibroblasts derived from a healthy person (ML6). We found that both Gemin2 and -3 are induced to a similar extent (Fig. 1 D) although SMN protein amounts merely tripled (data not shown). This suggests that induction of Gemin2 and -3 is rather a general feature of LBH589 than a consequence of massive SMN up-regulation seen in SMN1 -deleted fibroblasts.
To further strengthen our data we checked Gemin protein amounts in ML5 following LBH589 treatment (Fig. 1 E) and were able to show that indeed Gemin2 and 3 protein amounts are 1.5-to 2-fold up-regulated (Fig. 1 F). Since Gemin3 mRNA amounts were not markedly increased by LBH589, altered post-translational factors influencing protein turnover cannot be excluded. Taken together, we found that not only SMN but also other components of the SMN complex are induced by LBH589.
The SMN2 splicing pattern is shifted towards FL-SMN2 by LBH589 treatment
To assess whether the increased SMN amounts after LBH589 treatment were due to transcription activation or splicing correction or both, FL-SMN2 and SMN2Δ7 transcript levels were analysed by qRT-PCR following treatment with 100 n m , 400 n m and 1 µ m of LBH589 for 64 h.
For all three tested cell lines, we found a significant increase in FL-SMN2 amounts whereas SMN2Δ7 amounts stayed unchanged or slightly decreased. Almost 3-fold increased FL-SMN2 levels were recorded for ML16 and ML5 at a concentration of 400 n m and 1 µ m , whereas ML17 showed a doubling at both concentrations (Fig. 2 A). SMN2Δ7 decreased by 20–50% in ML5, but no significant changes could be detected in ML16 and ML17 (Fig. 2 B). Among many other possible explanations as for example altered promoter architecture or pre-mRNA processing, the shift in the SMN2 splicing pattern towards FL-SMN2 (Fig. 2 C) could also be attributed to an altered expression of proteins functioning in SMN splicing. Since several splice factors are well known to bind SMN mRNA and influence splicing ( 50 ), we next checked their expression after LBH589 treatment (Fig. 2 D and E). hTRA2-β1, a splicing factor promoting SMN2 exon 7 inclusion ( 13 ), revealed 2–3-fold elevated levels following treatment with concentrations between 400 n m and 1 µ m of LBH589. In contrast, the expression of SF2/ASF and SRp20, both having been shown to influence SMN splicing ( 9 , 51 ), did not noticeably change their expression (Fig. 2 D and E). Strikingly, LBH589 treatment of fibroblasts transfected with an hTRA2-β1 siRNA failed to significantly increase SMN protein amounts ( Supplementary Material, Fig. S1A and B ), supporting the idea that one of the mechanisms by which LBH589 up-regulates SMN is correction of SMN2 splicing, although a general effect of hTRA2-β1 on SMN expression cannot be excluded.
LBH589 induces acetylation of the SMN2 promoter thereby boosting promoter activity
Since we could show that treatment with the HDACi LBH589 increases total SMN2 transcription, we next focused on the SMN2 promoter itself. Many covalent histone modifications like acetylation and methylation are known to modulate transcription rates. Among these acetylation at lysine 9 of histone H3 (H3K9) has been linked to active transcription ( 52 ). Therefore, we performed chromatin immunoprecipitation (ChIP) analysis of the SMN2 promoter with regard to H3K9 acetylation in LBH589 and mock-treated fibroblast ( 39 ). Following treatment with 400 n m LBH589 for 64 h, DNA–histone complexes were precipitated with an anti-acetyl-H3K9 antibody and analysed by qRT-PCR. LBH589 increased acetylation at all four tested regions ( 39 ) of the SMN2 promoter. The highest change in H3K9 acetylation was found at human SMN2 promoter region 2 (huSP2) with an augmentation of 3.3-fold followed by huSP3 with 2.6-fold (Fig. 3 A).
To test whether the increase in H3K9 acetylation following LBH589 treatment could be attributed to an isoenzyme specific or a pan-HDAC inhibition, we performed an HDAC inhibition assay ( 53 ). By using increasing concentrations of LBH589 ( Supplementary Material, Fig. S2 ), we observed a concentration-dependent inhibition of HDAC activity to a level of 1% activity at 10 µ m ; inhibition to 90, 50 and 10% of control HDAC activity was observed at IC 10 = 12 n m , IC 50 = 106 n m and IC 90 = 953 n m , respectively. These data indicate that at concentrations regularly used in this manuscript, LBH589 acts as a pan-HDACi.
Compared with other potent HDACi, the IC 50 value found for LBH589 is very low; e.g. the IC 50 value for SAHA is almost five times higher [∼500 n m ( Supplementary Material, Fig. S2 ), reviewed in ( 53 )]. VPA, another compound with a perspective for SMA therapy, achieved half-maximum HDAC inhibition at around 7.24 m m ( 53 ).
On the basis of our ChIP data, we next checked whether increased H3K9 acetylatation is indeed accompanied by an elevated promoter activity. For this purpose we treated an NSC-34 reporter cell line ( 42 ) with various concentrations of LBH589 for 64 h. This reporter cell line is stably transfected with the β-lactamase under the control of the SMN2 promoter. Addition of the fluorescent dye CCF2-AM allows monitoring SMN2 promoter activity, since the dye is cleaved by the β-lactamase in an concentration-dependent manner. Following LBH589-treatment we recorded a concentration-dependent increase in promoter activity. The half maximal increase was observed at EC 50 = 108 n m (Fig. 3 B) whereas the maximum promoter activation by LBH589 amounts to 2.5 au. This is further corroborated by the finding that the EC 50 value for the LBH589-induced SMN2 promoter activity increase matches the IC 50 value for HDAC inhibition by the compound ( Supplementary Material, Fig. S2 ). The respective curve slopes nH were not different from unity, suggesting that LBH589 reacts in a 1:1 stoichiometry during promotor activity stimulation and HDAC inhibition.
Taken together, we were able to show that increased H3K9 acetylation and SMN2 promoter activity under LBH589 regimen indeed go hand in hand and that concentrations of >200 n m LBH589 induce maximum SMN2 promoter activation.
LBH589 reduces SMN protein turnover but does not affect mRNA turnover
While SMN protein increased up to 8-fold, promoter activity as well as amounts of FL-SMN2 merely tripled upon LBH589 treatment. Hence, the question arises, which mechanisms may account for this discrepancy: enhanced SMN2 mRNA or protein stability? To analyse the SMN2 mRNA stability we blocked de novo mRNA synthesis in fibroblasts either treated with 400 n m LBH589 or mock by means of actinomycin D. By analysing transcript amounts over time by qRT-PCR, we calculated mRNA half-life periods ( T1/2 ) for both FL - SMN2 and SMN2Δ7 transcripts as well as three control transcripts. For none of the tested transcripts did we find any notable difference in stability (Fig. 4 A and Supplementary Material, Table S2 ).
On the basis of these findings, we then asked whether LBH589 treatment affects SMN protein rather than mRNA turnover. Since it has previously been shown that SMN is degraded via the ubiquitin–proteasome system (UPS) ( 54 , 55 ), we tested for altered protein degradation under LBH589 regimen. Inhibition of proteasomal degradation with MG132 resulted—as expected—in a doubling of SMN amounts since it is not longer degraded (Fig. 4 B and Supplementary Material, Fig. S3 ). However, adding MG132 to LBH589-treated cells failed to further increase SMN levels (Fig. 4 B), suggesting that either LBH589 directly inhibits the proteasome or that ubiquitinylation of SMN—a prerequisite for its proteasomal degradation—is reduced whereby SMN is prevented from its degradation.
To address the question which steps of the UPS-pathway are modified by LBH589, we next tested whether the proteasome itself is inhibited by LBH589. The proteasomal degradation system encompasses three different proteolytic activities: caspase-, trypsin- and chymotrypsin-like ( 56 ). Using a luciferin-coupled degradation assay to measure catalytic rates for each of these proteolytic activities (Fig. 4 C), we did not detect any significant inhibitory effect of LBH589 on any of them. On the other hand, MG132, a positive control for proteasome inhibition, led to a profound inhibition of all three proteolytic activities ranging from 50 to 80%. In contrast, the negative control NH 4 Cl, a known inhibitor of the lysosome but not of the proteasome ( 54 ), had—as expected—only minimal effects on the proteasome. VPA slightly decreased trypsin- and chymotrypsin-like rates, but led to considerable activation of caspase-like proteasomal degradation (Fig. 4 C). Taken together, these data suggest LBH589 to have no inhibitory effect on the proteasome.
To further clarify how SMN degradation is abated, we next tested whether ubiquitinylation of SMN is reduced by LBH589 treatment. To do so, we precipitated tri- and tetraubiquitinylated proteins from LBH589-treated fibroblasts using an ubiquitin-interaction-motif (UIM) from the proteasome subunit Rpn10 ( 57 ) coupled to agarose beads. Ratios between ubiquitinylated and non-ubiquitinylated SMN were determined by resolving equal SMN amounts by western blot.
As depicted in Figure 4 D, we could show that following LBH589 treatment the portion of ubiquitinylated SMN compared with non-ubiquitinylated SMN was reduced to 30% (Fig. 4 E). This finding suggests that reduced ubiquitinylation of SMN may be the critical factor that accounts for the discrepancy between the 8-fold increased SMN protein amounts and the merely tripled mRNA amounts. We ascertained the absence of non-ubiquitinylated proteins in the precipitate by staining against ODC, a protein known not to be regularly ubiquitinylated ( Supplementary Material, Fig. S4A ) and proved as well the effectiveness of LBH589 treatment by western blotting using β-Actin as loading control ( Supplementary Material, Fig. S4B ).
Since western blot analysis of overall ubiquitinylation after LBH589 treatment revealed no general reduction (Fig. 4 F), it is obvious that the decreased SMN ubiquitinylation is rather a secondary than a primary effect of LBH589. It has previously been shown that SMN stability is facilitated by incorporation of SMN into the SMN complex ( 55 ). Therefore, we argue that the observed reduction in SMN ubiquitinylation is rather a consequence of enhanced SMN complex formation, which in turn keeps SMN out of the UPS pathway. Taken together, these results indicate that in addition to the transcriptional activation and splicing correction, LBH589 also acts post-translationally by decreasing SMN protein ubiquitinylation and favouring its inclusion into the SMN complex.
LBH589 displays low cytotoxicity up to 500 n m
If LBH589 is to be considered as a potential drug for SMA therapy, toxic side effects should be minimal. For this reason, we analysed cell viability following 96 h LBH589 treatment by an MTT-assay as described previously ( 41 ). As depicted in Figure 5 , cell viability was not considerably affected, except for concentrations above 10 µ m LBH589. Lower concentrations of LBH589 slightly decreased cell viability by 10–15%. To assess whether LBH589 is indeed cytotoxic or whether the decreased cell viability is a result of a reduced cell metabolism, cytotoxicity of LBH589 was assessed by an LDH assay in fibroblasts treated for 96 h. We were able to show that concentrations up to 500 n m did not affect cell survival, whereas concentrations ranging from 750 n m to 10 µ m slightly decreased cell survival. Higher concentrations of LBH589, however, reduced cell survival by 60%. Taken together, LBH589 does not seem to severely impair cell survival at the concentrations regularly used in our experimental settings. Moreover, already low concentrations of around 400 n m , which have a profound effect on SMN levels, are clearly well tolerated.
LBH589 activates SMN expression in neuronal tissues in vitro and in vivo
On the basis of the data obtained from our MTT and LDH assay, we further corroborated our previous results from human SMN1 -deleted fibroblasts in a neural cell type with concentrations clearly inducing SMN2 expression but not compromising cell viability. To do so, neural stem cells (NCSs) derived from two epilepsy patients who had undergone surgery were isolated and treated with 200 and 400 n m LBH589 for 48 h. In both NSC lines (H23 and H30), LBH589 increased SMN amounts by 2–5-fold (Fig. 6 A and B) with highest values at 200 n m , thus confirming the data previously obtained from SMN1 -deleted fibroblasts.
Furthermore, as a first step towards an in vivo examination of LBH589, we used murine embryonic fibroblasts (MEFs) derived from SMN2 transgenic mice carrying four human SMN2 copies on a null Smn background [ Smn−/− ; SMN2tg/tg ( 58 )] to answer the question whether LBH589 is also able to induce SMN2 expression in a murine cellular context. Following 64 h of LBH589 regimen, we found that SMN expression was elevated significantly up to 12-fold at 400 n m LBH589 and up to 7-fold at all other concentrations between 200 and 750 n m (Fig. 6 C and D). These data emphasize the great perspective of LBH589 and its potential use in an SMA-like mouse model for preclinical tests.
On the basis of our promising in vitro data, we then tested LBH589 in vivo . Single injections of 40 mg/kg [bodyweight] LBH589 were applied subcutaneously into the neck of 3-month-old SMN2 transgenic mice [ Smn−/− ; SMN2tg/tg ( 58 )]. RNA or proteins were isolated over time from different tissues as shown in Figure 6 E and F. For all three analysed time points 2- to 3-fold increased SMN2 transcript amounts in the spinal cord and a 1.6-fold increase of SMN protein in brain were found. This implies that LBH589 is able to cross the murine blood brain barrier. No changes in SMN expression were recorded for the liver of these mice (Fig. 6 F and Supplementary Material, Fig. S5B ).
To further ascertain that LBH589 is capable to pass the murine blood brain barrier, a prerequisite for the evaluation of LBH589 in SMA-like mice, we also checked global histone H3 acetylation as a hallmark of active transcription in the brain. LBH589 indeed rapidly increased H3 acetylation following subcutaneous injection ( Supplementary Material, Fig. S5B ) as shown by ELISA, suggesting that LBH589 indeed is capable of passing the blood brain barrier, which is imperative for a putative SMA drug.
LBH589 increases SMN protein amounts in SMA fibroblasts, which failed to respond to VPA
Many different HDACi have already been studied as potential SMA drugs [(reviewed ( 59 )]. Among these, VPA and phenylbutyrate are being tested in several ongoing clinical trials in SMA patients [reviewed ( 60 )]. However, in a first pilot trial with 20 SMA patients, only one-third of patients were positive responders to VPA presenting increased FL-SMN2 levels in blood, whereas the remaining of patients have either been non- or negative responders, showing no change in FL-SMN2 transcript levels or even a decrease ( 45 ). From three of the latter patients as well as two further patients in which SMN was up-regulated in blood upon VPA regimen, skin biopsies were obtained, fibroblast cultures established and subsequently treated in vitro with either VPA or LBH589. By treating these cell lines (two non-responders: ML79 and ML82, one negative responder: ML73, and two positive responders: ML60 and ML67) with VPA we were able to confirm the in vivo responses detected in blood (Fig. 7 ): SMN levels were unchanged in cell lines derived from VPA non responders (ML79 and ML82) or even down-regulated as in the case of cell line ML73, which was derived from a negative responder. On the other hand, SMN protein amounts were increased in the cell lines ML60 and ML67—both derived from patients reacting positive to VPA in blood. Taken together, the in vitro system fibroblasts allowed us to confirm the in vivo data produced afore.
Strikingly, LHB589 treatment increased SMN amounts in all five tested fibroblast cell lines (Fig. 7 ) independent of the response to VPA. Notably, treatment with LBH589 also increased SMN amounts in the fibroblast cell line ML73, which showed decreased SMN amounts upon VPA treatment. Moreover, in ML60 SMN amounts increased 10-fold which is the highest up-regulation we were able to detect and clearly outscores the effect of VPA ( Supplementary Material, Table S3 ). These findings emphasize the potential of LBH589 for SMA therapy and even for patients who are non- or negative responders to other drugs, such as VPA.
SMA is a unique disorder in the field of human genetics for several reasons: ∼96% of all SMA patients carry the same mutation in SMN1 , which allows rapid and easy identification by molecular testing [reviewed in ( 6 )]. However, although SMN1 is homozygously absent in patients with SMA, SMN2 is present in all patients producing minor protein amounts identical to SMN1 . This unique scenario provides the therapeutic opportunity to compensate for the loss of the disease determining gene by activation of the nearly identical copy gene by pharmacological means. Already a fair number of drugs has been identified so far and shown to significantly elevate FL-SMN2 mRNA and SMN protein levels in vivo in SMA patients as well as SMA-like mice [reviewed in ( 60 , 61 )].
The HDACi LBH589, brought to market as Panobinostat, has already proved its effectiveness in a wide variety of cancer cell lines by inducing growth inhibition and cell death [reviewed in ( 48 )]. LBH589 achieved orphan drug status for the treatment of CTCL by the FDA in October 2007. In a phase I study orally administered LBH589 was well tolerated and induced positive clinical response in 6 of 10 CTCL patients ( 62 ). Phase I (e.g. Hodgkin's lymphoma) and II (refractory multiple myeloma) studies are ongoing to further evaluate the potential of LBH589 as an anticancer drug [reviewed in ( 48 )]. Because of these findings, the analysis of LBH589 as a putative SMA drug is self-evident.
In our experiments, already doses of 200 nM LBH589 were able to increase SMN protein amounts 5-fold after 64 h of treatment (Fig. 1 A). Moreover, up to 10-fold elevations were recorded at concentrations of 500 or 750 n m , which is the highest up-regulation of SMN ever reported and of such high extent that almost theoretical SMN levels of unaffected carriers with one SMN1 copy are reached. Among other HDACi like sodium butyrate ( 33 ), VPA ( 35 , 45 ), phenylbutyrate ( 36 ), SAHA ( 38–40 ), M344 ( 41 ) and TSA ( 42 , 43 ), which are also able to enhance SMN2 expression, LBH589 shows the highest effect on SMN levels at comparatively low concentrations.
Moreover, we were able to show that not only the SMN protein amount ascended but also the number of gems increased by the same extent (Fig. 1 C). In line with this, also SMN complex components like Gemin2 and Gemin3 were up-regulated under LBH589 regimen. This suggests that the observed increase in the number of gems could rather be attributed to a de novo generation of SMN complexes than to the incorporation of SMN into the existing structures.
In contrast to the 8–10-fold elevated protein levels, the FL-SMN2 transcripts augmented only 2–3-fold (Fig. 2 A) whereas SMN2Δ7 amounts remained stable or in the case of ML5 decreased (Fig. 2 B). This result suggests that LBH589 activates SMN2 transcription as well as corrects SMN2 splicing (Fig. 2 C). Indeed the SMN2 promoter assay ( 42 ) as well as the ChIP analysis proved that LBH589 induces transcription by H3K9-hyperacetylation of the human SMN2 promoter (Fig. 3 A) ( 39 ) with concentrations >200 n m of LBH589 increasing 2–2.5-fold SMN2 promoter activity (Fig. 3 B). A comparable activation of the SMN2 promoter has also been reported for TSA ( 42 ) and SAHA (our data, not shown, and ( 39 )), whereas VPA increases promoter activity only by a factor of 1.4–1.8 (our data, not shown, and ( 35 , 39 )). Combination of these data with an HDAC inhibition assay (Fig. S2) suggests that the 2–2.5-fold increase represents the maximum possible promoter activation; a value which was already reached at 200 n m of LBH589.
The doubling of the promoter activity alone is obviously insufficient to notably increase SMN amounts since the majority of SMN2 transcripts lack the exon 7, resulting in a truncated protein. We checked expression of several splice factors known to influence SMN splicing and could indeed show that hTRA2-β1, which is known to promote exon 7 inclusion ( 13 ), is up-regulated after treatment with LBH589. Levels of SF2/ASF, another SR-protein necessary for exon 7 splicing ( 9 , 10 ), and SRp20, which has recently been proposed to promote SMN 2 exon 7 inclusion ( 51 ), did not change at all (Fig. 2 D and E). In line with these results, treatment of fibroblasts transfected with an siRNA targeting hTRA2-β1 almost completely failed to up-regulate SMN ( Supplementary Material, Fig. S1A and B ), suggesting that the reversion of the SMN2 splicing pattern as well as promoter activation are both prerequisites for efficient SMN up-regulation by LBH589.
Based on the fact that SMN2 transcript levels were only tripled whereas protein amounts increased 8–10-fold, we further analysed post-transcriptional factors influencing SMN amounts: among these, mRNA half-life is an obvious possibility since it is well established that mRNA turnover is often used to post-transcriptionally regulate gene expression ( 63 ). Similar to other HDACi ( 64 ), we found no significant difference in mRNA half-life after LBH589 treatment for any tested transcripts although they tended to be slightly more stable (Fig. 4 A and Supplementary Material, Table S2 ). Unlike earlier reported data which suggested comparable half-lives for FL-SMN2 and SMN2Δ7 of around 5–6 h ( 65 ), we found in our experimental setup a T1/2 for FL-SMN2 of around 8.5 h, whereas SMN2Δ7 amounts halved after 11–12 h. However, these differences could most likely be attributed to inter-individual differences between the cell lines used.
Since we could exclude that mRNA turnover is affected by LBH589, we next assessed whether protein turnover is possibly affected by HDACi treatment as it has been described earlier ( 64 ). In agreement with the fact that SMN is degraded by the UPS ( 54 , 55 ), in a two-step process [first coupling to ubiquitin, then degradation by the proteasome ( 67 )], inhibition of this system with MG132 led to elevated SMN levels in mock treated cells (Fig. 4 B). However, administration of MG132 to LBH589-treated fibroblast cells failed to further increase SMN levels implying LBH589 directly affects SMN turnover (Fig. 4 B). Since we were not able to detect any inhibitory effect of LBH589 on the proteasome itself (Fig. 4 C), we next focused on ubiquitinylation of SMN. Precipitation of tetra- and tri-ubiquitinylated proteins revealed that in contrast to an overall increase in ubiquitinylated proteins under LBH589 treatment (Fig. 4 F) SMN ubiquitinylation was reduced to 30% (Fig. 4 D and E). Although differential expression of specific enzymes involved in SMN ubiquitinylation—which are not known yet—cannot fully be ruled out, it is more likely that the enhanced SMN complex formation may account for this observation, thus leading to the massive SMN accumulation.
Since it is well established that integration of SMN into the SMN complex is favoured by activation of cAMP-dependent protein kinase A (PKA), PKA expression and activity under LBH589 treatment was tested ( Supplementary Material, Fig. S6A and B ). In both instances, no activating effect of LBH589 was detected. We therefore exclude PKA activation as the trigger of intensified SMN complex formation and argue that the likewise up-regulation of Gemin2 and Gemin3 keeps SMN out of the UPS pathway by increased complex formation.
To corroborate the data from SMN1 -deleted fibroblasts, we further checked whether LBH589 is able to induce SMN expression in human NSCs. In line with our previous results, 200 and 400 n m LBH589 elevated SMN levels 2–5-fold in both NCS lines (Fig. 6 A and B). Moreover, treatment of MEFs derived from SMN2 transgenic mice ( Smn−/− ; SMN2tg/tg ) with LBH589 elevated SMN levels up to 12-fold, which is even higher than in human fibroblasts (Fig. 6 C and D). Because of these convincing results we then tested LBH589 in vivo in a first experiment by subcutaneous LBH589 injections in Smn−/− ; SMN2tg/tg mice. Although the half-life of LBH589 in mice [IV t1/2 = 1,37 h; oral t1/2 = 2,90 h ( 48 , 67 )] is relatively short, we found that both FL-SMN2 and SMN2Δ7 were elevated 2–3-fold in the spinal cord from 8 h up to 24 h post s.c. injection (Fig. 6 E). These data imply that LBH589 is also able to cross the murine blood-brain barrier, which is furthermore confirmed by the fact that global histone H3 acetylation in the brain of these animals was elevated ( Supplementary Material, Fig. S5A ). In addition, LBH589 is currently evaluated in a clinical trial for the treatment of glioma in humans (clinicaltrials.gov/NCT00848523), indicating that LBH589 is also able cross the human blood-brain barrier—a prerequisite for a putative SMA drug. Furthermore SMN protein levels increased almost 1.6-fold 8 and 24 h post s.c. injection in the brain of these mice (Fig. 6 F and Supplementary Material, Fig. S5B ). In the liver of these mice, we were not able to detect any change upon LBH589 injection, which most likely can be explained by its high metabolic rate. Taken together, these results provide first evidence that LBH589 also induces SMN expression in vivo . However, further optimization of dose and application is needed.
Finally, our data show that LBH589 has not only the most pronounced impact on SMN protein expression reported so far, but it is also positively acting on SMN levels in fibroblast cell lines inert to VPA treatment (Fig. 7 ). The short-chain fatty acid VPA has already been used in clinical trials for SMA therapy, but turned out to be beneficial only in less than half of SMA patients, whereas the remaining patients were either non- or negative responders revealing no or even decreased FL-SMN2 levels ( 45 ). Using fibroblasts obtained from all three groups of patients ( Supplementary Material, Table S3 ), we now got similar results according to the FL-SMN2 levels in blood. This observation provides first evidence that similar intra-individual responses to VPA can be obtained in the two tissue systems, skin and blood. This knowledge may be used in future ahead of treatment to determine whether somebody is indeed responder to a certain drug or not. But of course this awaits confirmation in a much larger number of individuals.
In addition to the fact that LBH589 is able to stimulate SMN expression in VPA non-responders, the physiological compatibility of LBH589 seems to be suitable backing up its therapeutic potential. Only at high concentrations of 20 µ m cell viability was severely influenced, whereas cytotoxic effects were just detectable starting at 750 n m (Fig. 5 A). However, since already a 3–5-fold up-regulation was detected at 200 n m LBH589 (Fig. 1 A), which is clearly below toxic concentrations, LBH589 should be considered as a promising candidate for SMA therapy.
In conclusion, this study highlights the potential of LBH589 as a candidate for SMA drug therapy and gives insight into the exact mechanism(s) by which SMN is actually up-regulated. LBH589 action is characterized by tremendous SMN up-regulation at very low doses accompanied by good physiological tolerance in vitro . Most importantly, patients who failed or negatively responded to VPA as determined by SMN2 expression in vivo (blood) and in vitro (fibroblast cell lines treated with VPA) showed a significant SMN protein induction under LBH589 treatment.
MATERIALS AND METHODS
Cell culture and treatment of cell lines
Primary human SMN1 -deleted fibroblasts cell lines were established from skin biopsies derived from SMA patients who fulfilled the diagnostic criteria for SMA ( 68 ) and grown as described elsewhere ( 34 ). Informed written consent was obtained from all patients according to the Declaration of Helsinki and the study has been approved by the ethical committee of the University Hospital of Cologne.
For drug treatment the day before treatment either some 2 × 10 5 fibroblasts were seeded into 10 cm dishes or some 1 × 10 5 cells in 6-well plates, respectively. Cells were treated with various concentrations of LBH589 (Novartis Pharmaceuticals, Basel, Switzerland) dissolved in DMSO. Following 64 h of treatment (if not indicated otherwise), subconfluent fibroblasts were harvested. Valproic acid (Sigma-Aldrich, Munich, Germany) treatment was performed as described previously ( 34 ).
Human hippocampal tissue was obtained from two patients submitted to epilepsy surgery. Following mechanical dissociation and enzymatic digestion of hippocampal tissue, isolated adult neural stem cells (NSCs) were expanded in N5 medium supplemented with EGF, FGF2 and LIF on poly- l -ornithine/laminin-coated plates. Stemness of isolated hippocampal NSCs was proved by clonal growth after repetitive expansion in vitro . Human NSCs were treated for 48 h with the indicated LBH589 concentrations. Informed consent was obtained from each patient and the study was approved by the ethical committee of University Hospital of Erlangen.
Generation of primary MEFs
Primary MEFs were isolated from E13.5 embryos of SMN2 trangenic animals [ Smn−/− ; SMN2tg/tg ( 58 )] according to standard protocols. In brief, after removal of head, the heart and liver from the dissected embryo tissue was homogenized, cells sedimented and subsequently cultured and treated as described above.
Subcutaneous LBH589 injections of mice
Age-matched 3-month-old adult mice [ Smn−/− ; SMN2tg/tg ( 58 )] were subcutaneously injected once with 40 mg/kg [bodyweight] LBH589 dissolved in DMSO into the neck pucker. Animals were sacrificed after the indicated time points and organs were processed for either RNA or protein isolation following standard protocols. An 8 h DMSO-treated mouse served as negative control.
Western blot analysis
To analyse protein amounts, whole cell lysates were prepared with RIPA buffer (Sigma-Aldrich) and western blot analysis was performed as described previously ( 34 ). The following antibodies were used: mouse monoclonal anti-β-Actin (1:10.000, Sigma-Aldrich), mouse monoclonal anti-SMN (1:2.000, BD Transduction Laboratories, San Jose, CA, USA), rabbit polyclonal anti-hTRA2-β1 ( 14 ), mouse monoclonal anti-SRp20 (1:200, Abnova, Taipei City, Taiwan), rabbit polyclonal anti-SF2/ASF (1:1000 Abcam, Cambridge, UK), rabbit polyclonal anti-Gemin2 (1:200, CIND, Oswestry, UK), mouse monoclonal anti-Gemin3 (1:250, Santa Cruz, Heidelberg, Germany), rabbit polyclonal anti-PKA (1:200, Abcam, Cambridge, UK), mouse monoclonal anti-ODC (1:1000 Abcam, Cambridge, UK), mouse monoclonal anti-Ubiquitin (1:1000, Santa Cruz, Heidelberg, Germany) and mouse monoclonal anti-p53 (1:250, Abcam, Cambridge, UK). Ubiquitinylated SMN was detected with a rabbit polyclonal anti-SMN (1:250, Santa Cruz, Heidelberg, Germany).
Fibroblast cells were grown and fixed as described elsewhere ( 41 ). After treatment with 400 n m LBH589 gems were visualized using a mouse monoclonal anti-SMN FITC-labelled antibody (BD Transduction laboratories). DAPI-staining (VectorLabs, Burlingham, CA, USA) ensured nuclear localization of gems. For each condition, 300 nuclei were analysed by fluorescence microscopy (Carl Zeiss, Germany).
Precipitation of ubiquitinylated proteins
Precipitation of ubiquitinylated proteins was carried out according to the manufacturer’s protocol using an Rpn10-UIM domain agarose conjugate (Biomol, Lörrach, Germany). In brief, cells were lysed by sonication in PBS containing 5% glycerol, 1% TX-100, 5 µ m MG132 and protease inhibitors (Sigma-Aldrich). Lysates were pre-cleaned using GSH-agarose (Sigma-Aldrich) and subsequently incubated over night with Rpn10-UIM domain agarose conjugates at 4°C on a rotating wheel. Precipitates were washed several times and analysed by western blotting.
Determination of transcript levels by qRT-PCR
Total RNA was isolated from 6-well plates using the RNeasy Kit (Qiagen, Hilden, Germany) and QIAshredder according to the manufacturer’s protocol. DNA contaminations were eliminated with the RNase-free DNase I set (Qiagen, Hilden, Germany). Subsequent analysis of FL-SMN2 and SMN2Δ7 transcripts by quantitative real-time RT–PCR on a LightCycler 1.5 (Roche, Basel, Switzerland) were performed as described in detail earlier ( 41 ). Further transcripts were analysed using the primers given in Supplementary Material, Table S1 .
Chromatin immunoprecipitation (ChIP)
Following the manufacturer’s instruction, chromatin was isolated from approximately 2 × 10 6 cells using the OneDay ChIP kit (Diagenode, Liège, Belgium). Immunoprecipitation was carried out with an anti-acetylH3K9-antibody (Diagenode, Liège, Belgium) at 4°C over night. Subsequent qPCR of the SMN2 promoter region using 100 ng DNA was performed using an ABI 7500 real-time PCR machine (Applied Biosystems, Foster City, CA, USA) as described elsewhere ( 39 ). PCR product amounts were calculated from the linear-logarithmic PCR phase and normalized to γ-Actin ( 69 ).
SMN2 promoter assay
To determine the SMN2 promoter activity, NSC-34 cells, stably transfected with β-lactamase under the control of the SMN2 promoter, were used ( 35 , 42 ). Following LBH589 treatment for 64 h, the activity of β-lactamase was measured by CCF2-AM turnover (Invitrogen, Paisley, UK) using a Tecan Safire 2 microplate reader (Tecan, Maennedorf, Switzerland).
RNA stability and protein degradation assay
To assess mRNA stability, fibroblast cells were treated with 5 µg/ml actinomycin D (Sigma-Aldrich) ( 65 ) and DMSO or LBH589, respectively. RNA isolation and analysis of transcript amounts was performed as mentioned above. The proteasome inhibitor MG132 (Sigma-Aldrich) was used to analyse protein degradation. Following treatment with 400 n m LBH589 for 24 h, MG132 was added at a final concentration of 5 µ m for 15 h ( 54 ). Proteins were analysed as described above.
Activity rates of different proteolytic activities associated with the proteasome were determined on a Glomax luminometer (Promega, Madison, WI, USA) using the Proteasome-Glo 3-Substrate Cell-Based Assay System (Promega) according to the manufacturer's protocol.
Cell viability and cytotoxicity assays
To assess the number of viable cells, an MTT assay was performed as described previously ( 41 ). MTT, a tetrazole, is reduced by mitochondrial reductases in active proliferating cells to formazan. Quantification of formazan amounts allows the determination of cell viability. Cytotoxicity was determined by measuring the release of the cytoplasmatic enzyme LDH into the culture medium by necrotic cells using the CytoTox 96 non-radioactive cytotoxicity assay (Promega).
Excel 2003 (Microsoft, USA) and Sigma Plot 9.0 (Systat Software, San Jose, CA, USA) were used to perform statistical analysis. Significant differences between data sets were identified using a two-sided Student's t -test. Three levels of statistical significance were distinguished: * P < 0.05; ** P < 0.01; *** P < 0.001. Each value is given as mean of three experiments ± SEM if not indicated differently. In box-plots 50% quartiles are represented by the box including median, whereas whiskers display 5 and 95% quantiles.
Nonlinear regression analysis
Data for the stimulation of SMN2 promoter activity by LBH589 in NSC-34 SMN2 promoter reporter cells were analysed nonlinearly applying a four parameter equation using Prism 5.02 (Graphpad software®, San Diego, USA). This allowed to obtain estimates for the top and bottom plateau of the curve as well as for EC50 , the concentration inducing a half maximal stimulation, and nH , the Hill-slope of the curve. It was tested whether the curve slope nH was different from unity by applying a F -test; P < 0.05 was taken as criterion for statistical significance.
This work was supported by grants provided from the Deutsche Forschungsgemeinschaft to B.W. [Wi-945/12-3], from Families of SMA to B.W. and E.H. [WIR07/09], from the Center for Molecular Medicine Cologne to B.W. and E.H. [D5], and the Initiative “Forschung und Therapie für SMA” to BW.
We are grateful to SMA patients and clinicians for their help in this study. We thank Katharina Zimmerman and Serjoscha Blick for establishing the fibroblast cell lines and determining the SMN genotype, and Iris Jusen (Institute of Pharmacy, Bonn) for performing the HDAC assay. We thank Jill Jarecki (FSMA) and DeCode for supplying us with the NSC34 β-lactamase reporter cell line, Glenn Morris (Wolfson Centre for Inherited Neuromuscular Disease, Oswestry) for the Gemin2 antibody and Novartis (Basel) for providing LBH589.
Conflict of Interest statement . None declared.