MLL 3 / 4 prevents stem cell hyperplasia and controls differentiation 1 programs in a planarian cancer stem cell model

Background: The family of Mixed Lineage Leukaemia (MLL) histone methyltransferase proteins are often implicated in disease processes, particularly cancer. Here we focus on the MLL3 and MLL4 which are mutated in a high percentage of cancers implicating them as tumour suppressors, but very little is known about the underlying transcriptional and epigenetic changes that contribute to cancer. Results: Here we make use of the highly accessible planarians model system to uncover a role for MLL3/4 in controlling stem cell differentiation and proliferation, that suggests conservation tumour suppressor over a large evolutionary timescale function for this epigenetic regulator. Knockdown of the planarian Mll3/4 orthologs compromises stem cell differentiation and leads to hyperproliferation and tumour-like outgrowth formation. The planarian system allows as to investigate the epigenetic and transcriptional studies changes in cells that will go on to form tumours at an early stage after loss of MLL3/4 function, identifying genome-wide changes that occur early in the development of the pathology. This revealed mis-regulation of both conserved oncogenes and tumour suppressors, that together likely explain the cancer-like phenotype observed in planarians. We confirm MLL3/4 tumour suppressor function and uncover a deep conservation of this role in stem cells. We find potentially conserved mis-regulated downstream targets driving the effects of MLL3/4 loss of function. Our work demonstrates the suitability of planarians for the study of epigenetic phenotypes related to cancer and stem cell function, and for capturing early causative changes in a definitive population of tumour forming stem cells in vivo.


Background 21
The family of Mixed Lineage Leukaemia (MLL) histone methyltransferase 22 proteins is often implicated in disease processes, particularly cancer. We 23 focus on the tumour suppressors MLL3 and MLL4, which are mutated in a 24 high percentage of cancers, but very little is known about the underlying 25 transcriptional and epigenetic changes that contribute to malignancy. 26

Results 27
Here we make use of the highly accessible planarian model system to 28 uncover a role for MLL3/4 orthologs in controlling stem cell differentiation 29 and proliferation, suggesting conservation of tumour suppressor function 30 over a large evolutionary timescale for these epigenetic regulators. The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/126540 doi: bioRxiv preprint first posted online Apr. 11, 2017; binding domain, suggesting that it has chromatin-binding properties 49 (Figure  149 1a). There are two planarian genes homologous to Drosophila Trr and the C-150 terminus of mammalian Mll3/4 -Smed-trr-1 (KC262345) and Smed-trr-2 151 (DN309269, HO004937), both previously described 36  box at a non-conserved position and SMED-TRR-2 (TRR-2) has no NR boxes 156 (Figure 1a). This could indicate that the planarian members of the trithorax-157 related family are not broadly involved in recognition of nuclear receptors like 158 their arthropod and mammalian counterparts 33,35,50,51 . It is also possible that 159 some functional divergence exists between TRR-1 and TRR-2, where only 160 TRR-1 is capable of interacting with nuclear receptors. 161 We performed wholemount in situ hybridization (WISH) and found that LPT, 162 trr-1 and trr-2 are broadly expressed across many tissues and organs. 163 Gamma irradiation, used to remove all cycling cells in S. mediterranea within 164 24 hours, revealed that the three transcripts are also likely to be expressed in 165 stem cells (Figure 1b). This was supported by using an alternative method for Cell sorting (FACS) datasets 9 revealed that 28% of LPT's total expression is 179 in the X1 FACS fraction (S/G2/M stem cells), 44% in the X2 fraction (G1 stem 180 cells and stem cell progeny) and 28% in the X ins fraction (irradiation-181 . CC-BY-NC 4.0 International license not peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/126540 doi: bioRxiv preprint first posted online Apr. 11, 2017; insensitive; differentiated cells) of FACS-sorted planarian cell populations 182 (Figure 1e). Both trr-1 and trr-2 showed similar distribution in expression 183 through FACS-sorted cell populations. This is in agreement with the observed 184 ISH patterns, suggesting all 3 transcripts are widely expressed and co-185 expressed in cycling stem cells, stem cell progeny and in neuronal cells 186 (Figure 1b-e, Additional File 2d). These data support the hypothesis that 187 these proteins act together, with LPT binding chromatin to serve as a scaffold 188 for TRR methyltransferase activity 35,39 . In order to study the function of planarian Mll3/4, we investigated phenotypes 193 after RNAi-mediated knockdown. Following LPT(RNAi), there was a clear 194 failure to regenerate missing structures, including the eyes and pharynx, with 195 regenerative blastemas smaller than controls (Figure 2a-b). After 8 days of 196 regeneration we observed that, as well as failure to regenerate missing 197 structures, animals began to form tissue outgrowths (Figure 2c-  The copyright holder for this preprint (which was . http://dx.doi.org/10.1101/126540 doi: bioRxiv preprint first posted online Apr. 11, 2017; increases in cell division are seen in knockdown animals that are left 277 unwounded (Additional File 10b). 278 In 8 day-regenerating LPT(RNAi) worms the observed over-proliferation is a 279 result of localized clusters of mitotic cells (Figure 4b). Since 8 days of 280 regeneration is the last stage before outgrowth formation commences, these 281 clusters likely correspond to sites of future outgrowths (Figure 4c). Similar 282 mitotic clusters are also seen at later stages of regeneration in animals that 283 are yet to develop outgrowths (Figure 4d). We looked specifically in 284 outgrowths and found mitotic cells (Figure 4d (Additional File 13c), an effect similar to the pathology seen following 309 knockdown of the tumor suppressor SMG-1 56 . The epithelial layer in 310 LPT(RNAi) animals also appears less well-defined than that in control 311 animals, with a blurred distinction between epithelium and mesenchyme. 312 Another feature of the LPT(RNAi) phenotype, encountered in a variety of 313 malignancies 58 , is changed nuclear shape (Additional File 13d). 314 In summary, LPT controls NB proliferation and restricts stem cells to pre-315 defined tissue compartments. Experiments described earlier showed that  After observing the global changes in expression and histone modification 403 patterns following LPT(RNAi), we wanted to identify individual mis-regulated 404 genes that could potentially be major contributors to the differentiation and 405 tumor-like phenotype, and assess which of these were potentially direct or 406 indirect targets of MLL3/4 activity in stem cells. 407 The well-known tumor suppressor p53, where hypomorphic loss of function 408 has been previously shown to cause dorsal outgrowths in planarians 61 , was 409 found to be significantly down-regulated in X1 stem cells after LPT(RNAi) 410 While we observed an agreement between expression levels and changes in 441 H3K4me3 around the TSS for many mis-regulated genes, this was not the 442 case for all genes. One example where transcriptional expression is 443 significantly up-regulated, but H3K4me3 levels are slightly down-regulated is 444 utx (Figure 10a). This finding suggests that for some mis- TRR-2 control differentiation to form many (gut, eyes, brain, pharynx), but not 501 all lineages (cilia, protonephridia), suggesting that the MLL3/4 COMPASS-like 502 complex is not a universal and unilateral regulator of differentiation. This

Animal husbandry 584
Asexual freshwater planarians of the species S. mediterranea were used. The 585 culture was maintained in 1x Montjuic salts water 72 . Planarians were fed 586 organic calf liver once a week. After every feeding, the water was changed. 587 Planarians were starved for 7 days prior to each experiment. They were also 588 starved throughout the duration of each experiment.

Mean. 704
For analysis of RNA-seq data, Wald's test (as part of the Sleuth 82 software) 705 was used for assessing differential expression. Spearman's rank correlation 706 was used for assessing the correlation between RNA-seq and ChIP-seq data. 707 Hypergeometric tests were used for assessing enrichment in the RNA-seq 708 data. 709 710

Data availability 711
The ChIP-seq and RNA-seq datasets are deposited in the Short Read Archive 712 with accession numbers: PRJNA338116 and PRJNA338115 respectively). 713 The 'Pomeroy Brain' dataset 63 from the oncomine database 714 Declarations 720

Competing interests 721
The authors declare they have no competing interests. 722