| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
FJ
EXPRESS SUMMARY ARTICLE The Full-length version of this article is also available, published online May 18, 2005 as doi:10.1096/fj.04-3425fje. |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
,1,2,||
,3
* Department of Cancer Biology and
Department of Biostatistics, Dana Farber Cancer Institute and
Department of Genetics, Harvard Medical School, Boston, Massachusetts, USA;
Department of Medicine, Division of Hematology-Oncology and
|| Howard Hughes Medical Institute, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA
3 Correspondence: Department of Craniofacial Development, King’s College London, Floor 28 Guy’s Tower, Guy’s Hospital, London SE1 9RT, UK. E-mail: jeremy.green{at}kcl.ac.uk
SPECIFIC AIMS
The goal of this study was to elucidate the known teratogenic action of the anti-epileptic drug valproate and test the hypothesis that its effects are mediated by inhibition of the chromatin remodeling enzyme, histone deacetylase. A second aim was to apply a gene ensemble (rather than a "laundry list") approach to microarray data as a tool in phenotypic profiling.
PRINCIPAL FINDINGS
1. VPA effects on development: retardation, axial malformation, and pigmentation defects in zebrafish and Xenopus
VPA causes spina bifida in humans by unknown mechanisms and has several known biochemical activities, including histone deacetylase (HDAC) inhibition. Our strategy was to characterize the VPA-induced phenotype in multiple ways to compare it in detail with phenotypes induced by structurally related compounds and an unrelated but HDAC inhibitory compound. To this end, zebrafish and Xenopus embryos were treated with VPA from early gastrula and scored at 24 h and early tadpole stage, respectively. In both species, VPA induced gross retardation, defects in gut coiling, shorter A/P axes, bent tails, and some eye defects. Both species developed pericardial effusions seen as edema surrounding the heart region. Compared with zebrafish, Xenopus embryos were relatively insensitive to VPA.
2. The effects of different VPA analogs and TSA on development: correlation with HDAC inhibition
Hypothesizing that VPA is teratogenic because of its HDAC inhibitory activity, we predicted that less inhibitory VPA analogs would be less teratogenic than VPA itself. The relative effects on histone acetylation in vivo (Fig. 1
A, B) matched those previously reported in vitro. Thus, VPA and related sodium butyrate and the structurally unrelated trichostatin A (TSA) strongly induced histone acetylation. The analog VPM does not whereas analogs 4PA and 2EH have intermediate activity.
|
Abnormal development correlated closely with HDAC inhibitory activity. Embryos treated with NaBu or TSA had defects similar to those seen with VPA (Fig. 1C
). VPM had little phenotypic effect while 4PA and 2EH delayed development (Fig. 1B
). We examined the expression of cardiac troponin, a marker of myocardium. The level of expression was unaffected but the heart tube was elongated, indicating a defect in looping (Fig. 1D
). The relative degree of this looping defect correlated closely with both HDAC inhibitory activity and the gross phenotypic series for the various compounds (i.e., VPA=TSA=NaBu>2EH=4PA>VPM) (Fig. 1D
). An essentially identical phenotypic series was generated for zebrafish embryos (not shown).
3. Transcriptome-wide effects of VPA and TSA are highly concordant
To test whether phenotypic similarity at the gross anatomical level did not conceal multiple more subtle differences between the effects of different HDAC inhibitors, we needed an unbiased and comprehensive measurement of correlation between the effects of VPA vs. other HDAC inhibitiors. This was provided by microarray analysis. RNA was extracted from pools of treated embryos and cDNA probes applied to Affymetrix Gene Chips for comparison. We normalized the resulting intensity levels using the GC-RMA background correction (an improvement on the commonly used Affymetrix MAS algorithm). We filtered the expression values to remove weak expression signals, which are prone to experimental noise. Figure 2
presents resulting scatter plots of fold change with one compound compared with the other. The data are presented with and without all < 2-fold changes. This drastic filter removes batch-dependence (see online Fig. 3A) but may discard meaningful biological variation.
|
The striking result is the close match between the effects of the two HDAC inhibitors. For highly (>2-fold) perturbed genes, the correlation coefficients are well above 0.90; even including the less perturbed genes, the correlation coefficient is 0.90 ± 0.03 and 0.87 ± 0.03 for Xenopus and zebrafish, respectively. The degree of concordance between the different teratogens is strong enough to be similar to that between different doses or direct replicates of a given teratogen (Fig. 2A
and online Fig. 3A). Rank product analysis of the changes confirms the concordance even with a highly stringent False Detection Rate cutoff (supplementary data). This suggests that the effects of the two different compounds on a given species are as similar as replicates of the same compound.
4. VPA and TSA concordance is maintained across multiple conditions
The dose-response design of the experiment provided an additional method of comparison: when effects of the drugs are plotted with respect to dose, the profiles are strikingly alike (see online Fig. 4). The vast majority show similar responses to the drug treatments. There are some exceptions, but only 4 genes of 2000 show such divergence. Several genes are concordant despite nonmonotonic dose-response expression profiles (i.e., the expression zigzags similarly for both compounds). Together with the close parallels in the dose-response profiles for the majority of genes, these reinforce the notion that these two structurally contrasting HDAC inhibitors act very similarly at the transcriptome level.
5. VPA transcriptomal and individual gene effects only weakly resemble retardation
Retardation of development is a prominent effect of VPA on several species of embryos. It might be argued that any agents that cause retardation would give similar effects, making correlation between VPA and TSA a nonspecific effect. To control for this, we compared the effect of the two compounds with pure retardation (i.e., with expression levels in younger, untreated sibling embryos) (Fig. 2B
). The correlation between younger ("retarded") embryos and the drug treatments is less than that between the two different drug treatments. Examination of in situ patterns of neural markers Krox20 and Pax2 reinforced this conclusion: patterns in treated embryos were unlike from those in control embryos at any earlier stages (see online Fig. 5).
6. Cross-species transcriptomal effect comparisons show relatively weak concordance
To test for cross-species concordance in HDAC inhibitor effects, we identified putative orthologs represented on both chips (
1277 genes) using the NIH Homologene database. With this gene set, there was much lower correlation between species than between different treatments within a species. Only about half the genes affected 1.5-fold or more by a given drug in one species were similarly affected in the other. We observed only a minimal increase in cross-species concordance when taking subsets of the genes with increasing sequence identity. Irreducible species expression differences, suboptimal stage matching and residual nonorthology of genes presumably limit cross-species concordance.
CONCLUSIONS AND SIGNIFICANCE
Our data show that the effects of VPA on Xenopus and zebrafish embryos include retardation, axial malformation (crooked tail and shortened axis), neural patterning defects, cardiac malformation, and a distinctive ensemble of transcriptional changes. First we show that the malformations are mimicked by TSA, sodium butyrate, and certain VPA analogs at doses that cause inhibition of HDAC, but not by VPA analogs that lack HDAC inhibition. Second, we show that VPA and TSA, two structurally unrelated HDAC inhibitors, have common effects on gene expression for thousands of genes across different doses and for two different species. We have provided a relatively unbiased, semiquantitative assay for similar vs. different effects across the transcriptome. While our data cannot rule out coincidentally convergent effects for VPA and TSA, they do show that the effects are so similar as to be likely due to a common activity, namely, HDAC inhibition. Our data therefore describe in depth a characteristic HDAC inhibition phenotype and provide evidence that HDAC inhibition is a major mechanism of VPA-induced teratogenesis in these model vertebrate embryos.
The significance of these results is that they narrow the likely target genes of the model teratogen VPA to those affected by HDAC; such genes may be involved in mechanisms of spina bifida and other birth defects in humans. This is a modest subset of all genes. Chromatin immunoprecipitation (ChIP) has been used to identify HDAC target genes and, coupled with microarray techniques ("ChIP-on-chip"), may help identify critical early VPA and TSA targets in model embryos.
Our multifactorial correlation of teratogenic VPA effects with HDAC inhibition points to histone acetylation status as the primary biochemical target upstream of the teratogenic action of VPA. Further mechanistic insight into teratogenic mechanisms will ultimately lead to reduction in spina bifida and other birth defects in the human population.
|
FOOTNOTES
1 These authors contributed equally to this work. 
2 Current address: Program in Computational Biology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA.
To read the full text of this article, go to http://www.fasebj.org/cgi/doi/10.1096/fj.04-3425fje; doi: 10.1096/fj.04-3425fje
This article has been cited by other articles:
|
|
 |
|
  F. Hong, R. Breitling, C. W. McEntee, B. S. Wittner, J. L. Nemhauser, and J. Chory RankProd: a bioconductor package for detecting differentially expressed genes in meta-analysis Bioinformatics, November 15, 2006; 22(22): 2825 - 2827. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Eikel, K. Hoffmann, K. Zoll, A. Lampen, and H. Nau S-2-PENTYL-4-PENTYNOIC HYDROXAMIC ACID AND ITS METABOLITE S-2-PENTYL-4-PENTYNOIC ACID IN THE NMRI-EXENCEPHALY-MOUSE MODEL: PHARMACOKINETIC PROFILES, TERATOGENIC EFFECTS, AND HISTONE DEACETYLASE INHIBITION ABILITIES OF FURTHER VALPROIC ACID HYDROXAMATES AND AMIDES Drug Metab. Dispos., April 1, 2006; 34(4): 612 - 620. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
