In the current work we studied the relevance of post-translational histone modifications for gene activation, analysed their dynamics on nanochromosomes specifically activated during sexual reproduction and their fate during a nuclear differentiation process. In our analysis of the gene-wide distribution of histone modifications we selected six macronuclear nanochromosomes of which four, the actin I gene, the DNA polymerase alpha gene, the histone H4 gene and a 1.1 kb gene of unknown function are transcribed during vegetative growth and two, mdp1 and mdp2, are silenced in the vegetative macronucleus and become expressed exclusively during conjugation [10
These nanochromosomes show significant differences in the length of sequences flanking the open reading frame (Figure ). While the actin I nanochromosome [GenBank accession number DQ108617
] and the nanochromosome encoding the DNA polymerase alpha gene [GenBank accession number AF194338
] have short flanking sequences, the nanochromosome encoding a histone H4 gene [GenBank accession number X16018
] was chosen because of its unusually long 5'-flanking region. Neither the protein encoded on the 1.1 kb nanochromosome [pCE7, GenBank accession number X72958
] nor its precise open reading frame are characterised, therefore the lengths of the flanking sequences are unknown. For analysis of conjugation-specific gene activation the genes encoding mdp1 and mdp2 were chosen. Mdp1 [GenBank accession number AY261996
] encodes a PIWI-homologue [10
], a protein which is a member of the RNAi-pathway in many organisms and is known to play a crucial role in the course of conjugation in ciliates. The protein expressed by mdp2 [GenBank accession number AY261997
] is not known. The PTM pattern on the mdp1 and mdp2 nanochromosomes was studied both in their silenced form in vegetative cells and after activation during conjugation. For further details on sequences see Additional file 1
. Finally, we studied histone modifications on genes in the differentiating macronucleus after conjugation. We analysed the PTM patterns on the actin I [GenBank accession number DQ108616
], mdp2 [GenBank accession number GU111958
] and the 1.1 kb gene [GenBank accession number X72958
, pCE7] sequences in the macronuclear anlage (Figure ) in the early polytene chromosome stage (30 hours post conjugation).
Figure 1 Schematic illustration of macronuclear nanochromosomes and their micronuclear sequences. (A) Macronuclear nanochromosomes, transcribed during vegetative growth. (B) Macronuclear nanochromosomes, silenced during vegetative growth. (A,B) Total length of (more ...)
Histone modifications shown to be present in macronuclei and associated with macronuclear-specific DNA sequences in the developing macronucleus are H3K14ac, H3K4me3 und H3K4me1 [11
] and therefore used in this study. Histone acetylations are very dynamic and generally correlate with an open chromatin state and active transcription [1
]. High-resolution analysis revealed that H3K14ac accumulates predominantly at promoter regions and transcriptional start sites (TSSs) [4
], in many cases co-localizing with H3K4me3 [4
]. Likewise, trimethylated H3K4 is associated with actively transcribed genes and decorates their promoters as well as the 5'-ends of coding regions [13
]. Although H3K14ac and H3K4me3 are known to be generally associated with actively transcribed genes, these modifications have also been found to decorate repressed genes in humans [15
To date, little is known about the role of H3K4me1 in the regulation of gene expression. H3K4me1 is found to accumulate towards the 3'-end of genes and is commonly linked to active transcription [15
The PTM patterns of actively transcribed genes
To identify patterns of PTMs on nanochromosomes encoding actively transcribed genes, chromatin was digested with micrococcus nuclease to yield mononucleosomes. Chromatin immunoprecipitation (ChIP) experiments and subsequent quantitative real-time PCR (qRT-PCR) analyses revealed that on nanochromosomes with short flanking sequences, e.g. actin I and polymerase? alpha, H3K14ac accumulated at the 5'-ends of genes and decreased towards the 3'-end (Figure ). On the actin I nanochromosome the level of acetylated H3K14 increased steadily in the first half of the nanochromosome before declining in the second half, reaching the minimum at the 3'-end (Figure ). Interestingly, the amount of H3K14ac in the 5'-flanking region was lower than at the start of the coding sequence. The relative amount of H3K14ac in the polymerase alpha nanochromosome was very high at the 5'-end, corresponding to the start of the coding region and decreased significantly within the following 2 kb (Figure ). The distribution of H3K14ac on the 1.1 kb nanochromosome (Figure ) was very similar to that seen on the actin I (Figure ) and polymerase alpha nanochromosome (Figure ). In contrast to the data described above results obtained from the histone H4 nanochromosome differed remarkably. Whereas only one peak was detectable for nanochromosomes with short flanking sequences, we found two peaks of H3K14ac within the histone H4 nanochromosome. A first accumulation of H4K14ac was observed in the 5'-flanking region, followed by a second peak within the coding region (Figure ). We repeated these chromatin immunoprecipitation experiments using a combined antibody directed against H3K9/14ac. The results obtained were entirely consistent with those for the H3K14ac antibody (see Additional file 2A-D
). Results obtained from ChIP analyses for trimethylated H3K4 were very similar to those obtained for H3K14ac. Again the examined nanochromosomes with short flanking untranslated regions, actin I (Figure ) and polymerase alpha (Figure ) showed an accumulation of H3K4me3 at the 5'-ends close to the start of the coding region and a significantly lower amount at the 3'-ends. A similar pattern was observed for the 1.1 kb nanochromosome (Figure ). In contrast, as for H3K14ac, two peaks of H3K4me3 were detected on the histone H4 nanochromosome, one in the 5'-flanking sequence and the other in the coding region (Figure ). Data obtained from ChIP and qRT-PCR analyses of H3K4me1 were less consistent compared to the data described for H3K14ac and H3K4me3. In the actin I nanochromosome the distribution of H3K4me1 seemed to mirror image the H3K4me3 distribution (Figure ). The level of H3K4me1 was low at the 5'-end and continuously increased towards the 3'-end reaching its highest concentration near the end of the coding region. The 1.1 kb gene exhibited a very similar distribution (Figure ). In contrast, the distribution of H3K4me1 on the polymerase alpha nanochromosome showed low amounts at both 5'-end and 3'-end of the coding region while the maximum level of H3K4me1 was found to reside in the middle of the gene (Figure ). H3K4me1 distribution in the histone H4 nanochromosome was different again in that its level was low at the 5'-end of the nanochromosome but peaked in the 5'-flanking sequence close to the coding region. Within the coding region this modification was almost evenly distributed (Figure ).
Figure 2 H3K14ac, H3K4me3 and H3K4me1 distribution on actively transcribed macronuclear nanochromosomes. Nanochromosomes encoding actin I (A-C), DNA polymerase alpha (D-F), 1.1 kb gene (G-I) and histone H4 (J-L) were examined. X-axis shows total length of gene, (more ...)
H3K14ac and H3K4me3 do not associate with 5'-ends of silenced genes
To identify the PTM pattern on transcriptionally silent genes, ChIP and qRT-PCR experiments were performed using the same antibodies as described above. As shown in Figure and Figure the distribution of histone modifications differed significantly from those on actively transcribed genes (Figure ). We found inactive genes to also associate with PTMs typical for open and permissive chromatin, although to a significantly lower amount, an observation also described in other organisms. However, not only the quantity differed but, more surprisingly, the PTM distribution also showed remarkable qualitative differences. While in active genes H3K14ac accumulated at the 5'-end of the nanochromosome, the relative amount of this modification was considerably higher at the 3'-end than at the 5'-end of the mdp1 (Figure ) and the mdp2 (Figure ) nanochromosomes. The amount of H3K14ac steadily increased over the nanochromosome to reach its maximum in the 3'-flanking region. Essentially the same was true for the distribution of H3K4me3 (Figure ). In contrast, the distribution of H3K4me1 in untranscribed nanochromosomes partly resembled that observed on actively transcribed genes. The level of monomethylated H3K4 was low at the 5'-end and increased towards the 3'-end. In both nanochromosomes the highest amount of H3K4me1 was found towards the 3'-end of the coding region (Figure ).
Figure 3 Patterns of H3K14ac, H3K4me3 and H3K4me1 on silenced macronuclear nanochromosomes during vegetative growth and upon activation during conjugation. The nanochromosome encoding mdp1 was examined during vegetative growth (A-C), 7 hpc (D-F) and 13 hpc (G-I). (more ...)
H3K14ac and H3K4me3 are relocated upon gene activation
Since we observed significant differences in the PTM patterns in genes either actively transcribed or repressed during vegetative growth it seemed of considerable interest to analyse the PTM patterns upon gene activation. Mdp1 and mdp2 are not expressed during vegetative growth but mRNAs of both genes can be detected 6-8 hours after the initiation of conjugation [10
]. We therefore isolated macronuclear chromatin from conjugating cells at 7 hpc (hours post conjugation) and performed ChIP and qRT-PCR analyses. The distribution of PTMs observed on these genes after activation of gene expression differed significantly from those in a silenced state. Upon activation, the pattern of H3K14ac and H3K4me3 was inverted compared to their silent status. On mdp1 the highest levels of acetylated H3K14 were found at the 5'-end at the beginning of the coding region and its concentration decreased towards the 3'-end (Figure ), reminiscent of the pattern on transcribed genes during vegetative growth. A similar observation was made for mdp2 (Figure ). These ChIP experiments were repeated using a combined antibody directed against H3K9/14ac. The results obtained were entirely consistent with those for the H3K14ac antibody (see Additional file 3A-F
). The distribution of H3K4me3 (Figure ) was similar to that of H3K14ac. ChIP data obtained for H3K4me1 were not as consistent as the other modifications. While on mdp2 no change in H3K4me1 distribution was detected (Figure ) a significant redistribution on mdp1 was observed (Figure ). In the repressed mdp1 gene H3K4me1 was predominantly found at the 3'-end but spread after activation almost over the entire nanochromosome. As a control, we analysed genes expressed during both, vegetative growth and conjugation. None of the PTM patterns on the actin I (see Additional file 4A-C
) or the polymerase alpha nanochromosome (see Additional file 4D-F
) changed during conjugation, indicating that relocation of PTMs is induced only upon gene activation. It has been reported that in the course of conjugation overall gene expression increases at 6-8 hpc [18
] before it declines and reaches a low level. Therefore, we also analysed PTM patterns on the mdp1 and mdp2 nanochromosomes at a later time point after conjugation (13 hpc) when the overall transcription rate had already decreased. At this point PTM patterns were not as clearly structured as described for 7 hpc. H3K14ac, H3K4me3 and H3K4me1 were more or less evenly distributed over the entire nanochromosomes of both mdp1 and mdp2 (Figure ), suggesting that we observe a transitional stage in which transcription rate already becomes reduced.
During nuclear development genes exhibit PTM patterns different from those of actively transcribed genes
Chromatin was isolated from macronuclear anlagen (30 hpc) and the distribution of 3 PTMs on macronuclear-specific sequences was analysed by ChIP and qRT-PCR. To avoid amplification of macronuclear contaminations in ChIP with antibodies directed against H3K14ac, H3K4me3 and H3K4me1 anlagen-specific primers were used for qRT-PCR analyses (Figure ). At this stage of anlagen differentiation the patterns of H3K14ac and H3K4me3 unambiguously differed from those in actively transcribed genes during vegetative growth. In the actin I gene the level of H3K14ac was low near the 5'-end and steadily increased towards the 3'-end of the gene (Figure ). The same was true for H3K4me3 (Figure ). H3K4me1 in contrast was evenly distributed over the gene (Figure ). Unfortunately, due to the lack of suitable primer combinations, the very 5'-end of the actin I sequence could not be analysed. With the exception of H3K4me3 which is evenly distributed over the sequence, results from the analyses of PTM distributions on the mdp2 (Figure ) gene resemble those from actin I (Figure ). The data obtained from the 1.1 kb nanochromosome also suggest a similar distribution of single PTMs on this gene (Figure ). However, although few anlagen-specific sequences could be found within the 1.1 kb sequence, many of them were not suited for qRT-PCR. Therefore, PCR fragment c (Figure ) had to be amplified using macronucleus-specific primers so that simultaneous amplification of contaminating macronuclear DNA can not be excluded. Similar analyses of other genes were technically impossible due to micronuclear complexity of the polymerase alpha gene [19
], the micronuclear simplicity of the histone H4 gene and limited knowledge of micronuclear sequence of mdp1.
Figure 4 Patterns of H3K14ac, H3K4me3 and H3K4me1 on macronuclear-destined sequences during macronuclear development. Sequences encoding actin I (A-C), mdp2 (D-F) and the 1.1 kb gene (G-I) were examined. X-axis shows total length of gene, Y-axis shows percent (more ...)