(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Divergent transcriptional and transforming properties of PAX3-FOXO1 and PAX7-FOXO1 paralogs [1]

['Line Manceau', 'Université Paris Cité', 'Cnrs', 'Institut Jacques Monod', 'Paris', 'Julien Richard Albert', 'Pier-Luigi Lollini', 'Department Of Experimental', 'Diagnostic', 'Specialty Medicine']

Date: 2022-07

PAX3-FOXO1 and PAX7-FOXO1 harbour divergent genomic occupancy and transactivation potential

To compare the activities of the PAX3-FOXO1 and PAX7-FOXO1 fusion proteins, we sought to design a system in which their expression can be controlled and expressed at similar levels in an identical cellular context. Human foreskin fibroblasts (HFF) were engineered to express a copy of a FLAG-tagged version of these TFs in a doxycycline (DOX)-inducible manner from the AAVS1 safeguard locus (S1B Fig). Three independent cell lines expressing PAX3-FOXO1 or PAX7-FOXO1 were generated (S1C Fig). All three PAX3-FOXO1 and PAX7-FOXO1 lines expressed similar fusion protein levels in bulk and at the single cell level 48h after exposure to DOX (S1C and S1E Fig).

Using these cell lines, we first mapped the genomic targets of PAX3-FOXO1 and PAX7-FOXO1 using Cleavage Under Targets and Tagmentation (CUT&Tag) [24] (Figs 1 and S2 and S3). We used two separate antibodies to detect the fusion TFs directed against either the N-terminal FLAG tag or the C-terminal FOXO1 domain. The latter could be used, as wild-type FOXO1 levels were low in control HFF compared to that of the transgenic PAX-FOXO1 proteins (S1C and S1D Fig). Data obtained with these two antibodies and in separate cell lines gave similar results, validating the approach (S2A Fig). In total, we selected 6000 cis-regulator modules (CRMs) with the highest PAX3-FOXO1 and/or PAX7-FOXO1 binding signals (Figs 1A, 1B and S2A and S1 Table). The majority of PAX-FOXO1-bound CRMs occupied intronic or intergenic regions (S2B Fig). These regions were mostly enriched for DNA motifs known to be recognized by PAX3 and PAX7 DNA-binding domains and poorly enriched for FOXO1 DNA binding motifs (Figs 1C, 1D and S2C and S2 Table). Hence, as shown for PAX3-FOXO1, PAX7-FOXO1 DNA sequence specific recognition is mainly mediated via its PAX7 DNA binding domains [16,17,23]. Finally, approximately 18% of these regions were previously identified as bound by PAX3-FOXO1 in an RMS cell line, some of which are in close proximity to iconic PAX3-FOXO1 target genes in RMS, including MYOD1 and PIPOX (S2D and S2E Fig) [16]. Thus, the recruitment of PAX-FOXO1s to the genome is preserved in part across cellular contexts in which they act, and examination of PAX-FOXO1 activity on the fibroblast genome is likely to shed some light on the contribution of both paralogs in establishing the molecular states of the RMS.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. PAX3-FOXO1 and PAX7-FOXO1 recruitment to the genome and impact on H3K27ac deposition. (a) Examples of IGV tracks showing normalized FLAG and H3K27ac CUT&Tag reads distribution in cells expressing either PAX3-FOXO1 (P3F) or PAX7-FOXO1 (P7F) 48h post-DOX treatment. Scales in counts per million (CPM). (b) Heatmaps of normalized FLAG and H3K27ac CUT&Tag signals obtained in control (Ctl), PAX3-FOXO1 (P3F) and PAX7-FOXO1 (P7F) expressing HFF 48h post-DOX treatment in three cis-regulatory modules (CRMs) clusters. The ranking of the CRMs is the same on all heatmaps and follows the occupancy levels of PAX3-FOXO1 in descending order (arrow). Cluster 1 contains CRMs onto which the occupancy rate of PAX3-FOXO1 and PAX7-FOXO1 are similar. Cluster 2 CRMs are more bound by PAX3-FOXO1 than PAX7-FOXO1, and cluster 3 is the converse of cluster 2. (c) Logos of position weight matrices of DNA recognition motifs for the PAX paired (PrD) and/or homeodomain (HD) and by bHLH TFs (E-BOX). (d) Left panels: Enrichment for DNA binding motifs displayed in c in the CRMs belonging to the clusters defined in (b) (bars: -log2(p-value)). Right panels: Average profiles of normalized H3K27ac CUT&Tag signals at CRMs belonging to the clusters defined in (b) in the indicated control (Ctl), P3F or P7F cell lines. Note that the two P7F purple curves are superimposed. https://doi.org/10.1371/journal.pgen.1009782.g001

For approximately half of the total bound CRMs, the occupancy of PAX3-FOXO1 and PAX7-FOXO1 were comparable (Fig 1A–1I, cluster 1 in Fig 1B). In contrast, 1429 CRMs showed increased occupancy for PAX3-FOXO1 relative to PAX7-FOXO1 (Fig 1A-II, cluster 2 in Fig 1B) and 1475 CRMs the inverse (Fig 1A-III, cluster 3 in Fig 1B). Whereas PAX7-FOXO1 binding was barely detected in CRMs preferentially bound by PAX3-FOXO1, PAX3-FOXO1 recruitment was detected on most CRMs preferentially bound by PAX7-FOXO1 (Fig 1B). Thus, the two paralog proteins have divergent recruitment patterns in the genome, with PAX3-FOXO1 being more widespread than PAX7-FOXO1.

To test whether the observed differences in fusion protein occupancy may stem from a distinct use of their DNA-binding domains [25,26], we probed the different sets of CRMs for several prototypical PAX3/7- and PAX3-FOXO1 PrD, PrD+HD, and HD DNA binding motifs (see Materials and Methods; Fig 1C and S2 Table). All PAX-FOXO1-bound CRMs were enriched in PrD DNA binding motifs, to a lesser extent, in PrD+HD DNA binding motifs (Fig 1DI-III). The HD DNA recognition motifs were detected in CRMs bound by both PAX-FOXO1s (Fig 1DI). However, this motif was sparsely present in PAX3-FOXO1 CRMs (Fig 1DII) and, in contrast, most represented in PAX7-FOXO1 CRMs (Fig 1DIII). Overall, these results suggest that PAX3-FOXO1 uses its PrD and HD to bind DNA, whereas PAX7-FOXO1 preferentially uses its HD [25,26]. We then scanned PAX-FOXO1-related CRMs for E-BOX motifs recognized by class II bHLH TFs (Fig 1C and S2 Table). Apart from the PAX3-FOXO1-specific CRMs, this E-BOX was strongly represented in the other PAX-FOXO1s-bound CRMs, supporting a common interaction between bHLH TFs and both PAX-FOXO1s [16,20–22] (Fig 1DI-III).

We then assessed the chromatin status of PAX-FOXO1-linked CRMs in control and PAX-FOXO1s expressing HFF using CUT&Tag by mapping the genomic distribution of H3K27ac and H3K27me3, which are respectively hallmarks of transcriptionally active and repressed CRMs (Figs 1A, 1B, 1DIV-DVI and S3). We also analysed the distribution profile of the heterochromatin histone mark H3K9me3, which borders a fraction of PAX3-FOXO1 bound genomic regions in Rh4 RMS tumour cells [23]. Both PAX-FOXO1 proteins were recruited predominantly in CRMs which are in the control cell line (Ctl) in absence of fusion protein recruitment, unmarked by these three histone modifications with their levels very low compared to the regions of the genome more enriched for these marks (S3A–S3C Fig). The distribution profile and the levels of the heterochromatic H3K27me3 and H3K9me3 marks in the vicinity of these CRMs remained unchanged upon PAX-FOXO1 recruitment (S3A–S3C Fig). On a genome-wide scale, the occupancy rate of PAX-FOXO1s correlated better with de novo deposition of the H3K27ac mark than with the deposition of the heterochromatin marks (S3B Fig). Hence, PAX7-FOXO1, as PAX3-FOXO1, induces chromatin signatures of transcriptional activation [16]. Interestingly, H3K27ac levels showed a better correlation with PAX7-FOXO1 occupancy than with PAX3-FOXO1 occupancy (S3C Fig). Accordingly, the H3K27ac level on PAX-FOXO1-bound CRMs were generally higher in the presence of PAX7-FOXO1 than PAX3-FOXO1 (Figs 1A, 1B, 1DIV-DVI, S3A and S3CI). This was particularly visible on CRMs onto which both PAX3-FOXO1 and PAX7-FOXO1 could be recruited to (cluster 1 in Fig 1B). Thus, PAX7-FOXO1 ability to establish an active enhancer chromatin signature is greater than that of PAX3-FOXO1. This is reinforced by a principal component analysis of H3K27ac occupancy in the 6000 PAX-FOXO1 bound CRMs showing that ectopic PAX7-FOXO1 generates a chromatin state more distant from that of control cells than PAX3-FOXO1 (S3F Fig).

Overall, our results demonstrate that the two RMS-associated paralog fusion TFs exhibit a divergent mode of recruitment to the genome, likely due to the previously identified discrete affinities of the DNA-binding domains of PAX3 and PAX7 for specific DNA motifs [27,28]. Thus, the fusion between the PAX DNA-binding domains and the FOXO1 transactivation domain would have preserved the diversification that has occurred during evolution in the genome recruitment patterns of PAX3 and PAX7. Our data also support the idea that PAX-FOXO1 chimeras act as pioneer factors as previously shown [16,23]. Both are able to bind unmarked chromatin and have the capacity to promote its transcriptional activation. While the recruitment of PAX7-FOXO1 in most cases results in the deposition of the active enhancer mark H3K27ac, this was much less the case for PAX3-FOXO1. As previously observed [24], some PAX3-FOXO1 bound regions (mainly in cluster 2) did not engage towards transactivation. This lack of activation could be due to a lower enrichment in DNA motifs recognised by both the PrD and HD domains of the PAX-FOXO1 or by the bHLH TFs. Supporting the former, it has been shown for the intact Pax7 protein that recruitment to a PrD-like motif is less likely to result in transcriptional activation than recruitment to motifs recognised by both PrD and HD. Finally, it is notable that the repertoire of DNA motifs in the PAX-FOXO1-bound regions we identified in HFF is much less enriched in juxtaposed PrD and HD motifs than the PAX3-FOXO1-bound regions in tumour cells [26]. This could be due to a temporal dynamic in the recruitment mode of the PAX-FOXO1 factors. Indeed, it has been shown that some pioneer factors begin to be recruited to DNA first on incomplete DNA recognition motifs and then on more complete motifs [27]. Supporting this idea, in myoblasts, PAX3-FOXO1 is less recruited to PrD+HD sites after 8 hours than after 24 hours [23]. Finally, in the tumour context, the specificities in the transcriptional activity of PAX3-FOXO1 and PAX7-FOXO1 that we have revealed in fibroblasts can be expected to be amplified or diminished by other parameters, such as the presence of several isoforms with variable transcriptional potential [28], variations in the expression levels of PAX-FOXO1s [29], or the cellular context (cells of origin and history).

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1009782

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.