Two paralogous PHD finger proteins participate inParamecium tetraurelia’s natural genome editing

genome reorganization; PHD finger proteins; small RNAs .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 2

1 The unicellular eukaryote Paramecium tetraurelia contains functionally distinct nuclei: 2 germline micronuclei (MICs) and a somatic macronucleus (MAC). During sexual 3 reproduction, the MIC genome is reorganized into a new MAC genome and the old 4 MAC is lost. Almost 45,000 unique Internal Eliminated Sequences (IESs) distributed 5 throughout the genome require precise excision to guarantee a functional new MAC 6 genome. Here, we characterize a pair of paralogous PHD finger proteins involved in 7 DNA elimination. DevPF1, the early-expressed paralog, is present in only some of the 8 gametic and post-zygotic nuclei during meiosis. Both DevPF1 and DevPF2 localize in 9 the new developing MACs, where IESs excision occurs. In DevPF2 knockdown (KD) 10 long IESs are preferentially retained and late-expressed small RNAs decrease; no 11 length preference for retained IESs was observed in DevPF1-KD and development-12 specific small RNAs were abolished. The expression of at least two genes from the new 13 MAC with roles in genome reorganization seems to be influenced by DevPF1- and 14 DevPF2-KD. Thus, both PHD fingers are crucial for new MAC genome development, 15 with distinct functions, potentially via regulation of non-coding and coding transcription 16 in the MICs and new MACs. 17 18

19 A unique feature shared by all ciliates is the presence of nuclear dimorphism. In 20 Paramecium tetraurelia (henceforth Paramecium) the two micronuclei (MICs) resemble 21 the germline of multicellular organisms, being transcriptionally silent throughout most of 22 the life cycle and generating haploid nuclei during meiosis that develop and give rise to 23 all nuclei in the subsequent generation. Also similar to the multicellular soma, the 24 macronucleus (MAC) is optimized for most gene expression, and originates from a MIC 25 copy. The old MAC is fragmented during sexual division and subsequently diluted 26 across cell divisions, with the new MAC completely taking over somatic expression. The 27 development from the MIC genome to the MAC genome in Paramecium is a natural 28 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 3 form of genome editing that requires extensive reorganization, including genome 29 amplification (~800n), chromosome fragmentation and the elimination of about 25% of 30 the sequence content (Arnaiz et al., 2012; Guérin et al., 2017). These MIC genome-31 specific sequences comprise repeats, transposable elements and Internal Eliminated 32 Sequences (IESs). 33 34 In contrast to other elimination events, IES elimination requires precise excision in 35 Paramecium. Precise IES excision is not characteristic of all ciliates. Notably, in 36 Paramecium’s oligohymenophorean relative Tetrahymena, IESs are predominantly 37 imprecisely excised and only tolerated in intergenic regions (Hamilton et al., 2016). The 38 roughly 45,000 IESs in Paramecium are scattered throughout the genome in both non-39 coding and coding regions and vary from tens to thousands of base pairs in length 40 (Arnaiz et al., 2012). Since the coding density of the Paramecium MAC genome is high, 41 most IESs are intragenic (Arnaiz et al., 2012). Paramecium IESs are flanked by 42 conserved 5’-TA-3’ dinucleotides (Klobutcher & Herrick, 1995) and excised by 43 PiggyMAC (Pgm). Pgm is a domesticated transposase derived from PiggyBac 44 transposases (Baudry et al., 2009) like the excisase responsible for IES excision in 45 Tetrahymena (Cheng et al., 2010). The weakly conserved ~5 bp long inverted repeats at 46 Paramecium IES ends (Klobutcher & Herrick, 1995) fail to provide enough specificity for 47 reliable Pgm recruitment (Arnaiz et al., 2012). This suggests that other factors are 48 needed for precise IES targeting. 49 50 The targeting of MIC-specific sequences for elimination is thought to be assisted by 51 small non-coding RNAs, first characterized in Tetrahymena (Chalker & Yao, 2001; 52 Mochizuki et al., 2002). Like Tetrahymena, the biogenesis of the 25 nucleotide (nt) scan 53 RNAs (scnRNAs) occurs during meiosis in the Paramecium MICs. Bidirectional non-54 coding transcription of the MIC genome is thought to be initiated by the putative 55 transcription elongation factor Spt5m (Gruchota et al., 2017) and followed by the 56 cleavage of long double-stranded RNA (dsRNA) by the closely related Dicer-like protein 57 paralogs Dcl2 and Dcl3 (Hoehener et al., 2018; Lepère et al., 2009; Sandoval et al., 58 2014). Argonaute/Piwi proteins Ptiwi01/09 (also close paralogs) process the resulting 59 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 4 short dsRNAs, removing one of the two strands, and stabilize single-stranded scnRNAs 60 throughout the selection process in the parental MAC and targeting of MIC-specific 61 sequences in the new MACs (Bouhouche et al., 2011; Furrer et al., 2017). In the 62 parental MAC, Gtsf1 was recently proposed to promote ubiquitination and subsequent 63 degradation of the Ptiwi01/09 complexes harboring MAC-matching scnRNAs (Charmant 64 et al., 2023; Wang et al., 2023). In the new MACs, the putative transcription elongation 65 factor TFIIS4 was proposed to promote non-coding transcription required for scanning 66 the developing genome (Maliszewska-Olejniczak et al., 2015). 67 68 In Tetrahymena, H3K9 and H3K27 methylation precede IES excision (Y. Liu et al., 2007; 69 Taverna et al., 2002) and it was shown in Paramecium that development-specific 70 H3K9me3 and H3K27me3 histone mark deposition by the PRC2 complex depends on 71 scnRNAs and is essential for the elimination of transposons and IESs (Frapporti et al., 72 2019; Ignarski et al., 2014; Lhuillier-Akakpo et al., 2014; Miró-Pina et al., 2022; Wang et 73 al., 2022). We recently showed that the ISWI1 chromatin remodeling complex is 74 necessary for IES excision precision and Ptiwi01/09 co-immunoprecipitated with ISWI1 75 in a crosslinked treatment (Singh et al., 2022, 2023). After the initial onset of IES 76 excision, additional single-stranded sRNAs, iesRNAs, ranging in size from ~26 to 30 bp, 77 are produced by Dcl5 from excised IES fragments and stabilized on Ptiwi10/11 (Furrer 78 et al., 2017; Sandoval et al., 2014). iesRNAs were proposed to participate in a positive 79 feedback loop for the efficient removal of all IES copies (Sandoval et al., 2014). 80 Nevertheless, only a fraction of IES excision appears to depend on scnRNAs or 81 iesRNAs (Nowacki et al., 2005; Sandoval et al., 2014). 82 83 Despite the knowledge gained in the past decades, the picture of IES excision is far 84 from complete. To identify novel genes involved in IES excision, we examined proteins 85 potentially associated with ISWI1, a chromatin remodeler we recently showed to 86 facilitate precise IES excision (Singh et al., 2022). 87 88 89 90 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 5

91 Identification of a novel protein involved in IES excision 92 Recently, we reported evidence supporting the formation of a protein complex involving 93 ISWI1 and the ICOP proteins (Singh et al., 2023). We conducted an RNAi screen of 94 additional genes that were unique in the ISWI1 co-immunoprecipitation (IP)-mass 95 spectrometry (MS) and exhibited upregulation in a developmental gene expression time 96 course from ParameciumDB (Arnaiz et al., 2017) (Fig. S1A). 97 98 In the screening, we sought phenotypic evidence for failed genome reorganization in the 99 form of growth defects (assessed by survival tests), and substantial IES retention 100 (assessed by IES retention PCRs). ND7, a gene involved in trichocyst discharge 101 (Lefort-Tran et al., 1981), was used as a negative control as its silencing does not impair 102 genome reorganization (Nowacki et al., 2005). Nowa1-KD, which affects the excision of 103 scnRNA-dependent IESs (Nowacki et al., 2005), was used as a positive control. 104 Candidate 2 (PTET.51.1.G0620188) displayed both IES retention and lethality in the 105 new progeny, whereas candidate 1 (PTET.51.1.G0990120) showed high lethality 106 without IES retention (Fig. S1B,C). Therefore, candidate 2 was selected for further 107 investigations. 108 109 DevPF2 and DevPF1 are paralogous PHD finger proteins 110 The Paramecium aurelia species complex, to which P. tetraurelia belongs, underwent 111 multiple whole-genome duplications, with many closely related paralogs generated from 112 the most recent of these (Sellis et al., 2021). The chosen candidate has a closely related 113 paralog (PTET.51.1.G0240213) with which it shares 86.6% identity at both the 114 nucleotide and amino acid levels. The paralog is upregulated during sexual 115 development as well, although earlier (Fig. 1A). HMMER3 searches of the Pfam 116 database (Finn et al., 2003) predicted two domains in both proteins: a PHD and a PHD 117 zinc-finger-like domain (Fig. 1B,D,E). 118 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 6 119 The highly conserved PHD domain has often been reported to mediate the interaction of 120 nuclear proteins with histone modifications (Sanchez & Zhou, 2011), but other binding 121 affinities have also been described (see Discussion). PHD domains possess a well-122 conserved motif consisting of eight cysteine and histidine residues (C4HC3) that 123 coordinate two zinc ions, thereby providing it with structural stability. The presence of 124 the C4HC3 motif in both paralogs was confirmed using a multiple sequence alignment 125 with PHD domains from well-established PHD finger proteins from Homo sapiens and 126 Drosophila melanogaster (Fig. 1C). 127 128 AlphaFold2 predicted the structures of both paralogs with high confidence for the 129 domains (Fig. 1F,G). We compared the PHD predictions with the published structure of 130 the WSTF (Williams Syndrome Transcription Factor) PHD finger (Pascual et al., 2000). 131 WSTF, associated with the Williams Syndrome (Lu et al., 1998), is a subunit of the 132 ISWI-containing chromatin remodeling complex WICH (Bozhenok et al., 2002). The 133 superimposition confirmed the orientation of the eight C4HC3 residues in the DevPFs 134 towards the two zinc ions (Fig. 1H), supporting the idea that both paralogs function as 135 PHD finger proteins. Since they show development-specific upregulation (Fig. 1A), we 136 named the paralogs development-specific PHD finger 1 (DevPF1; early-expressed 137 paralog) and 2 (DevPF2; late-expressed paralog). 138 139 DevPF1 and DevPF2 show distinct nuclear localization 140 To determine the localization of both paralogs, we injected DNA constructs encoding 141 DevPF1 and 2 C-terminally tagged with green fluorescent protein (GFP) into MACs of 142 vegetative paramecia. The cells were collected during Paramecium sexual development 143 for confocal microscopy. The injected cultures displayed no growth defects compared to 144 non-transformed cells (Fig. S2A). However, we observed variable numbers for gametic 145 MICs (Figs 2, 3) and new MACs (Fig. S2C) in some cells, which has been observed 146 frequently for transgenes (e.g, Nowa1-GFP fusion; (Nowacki et al., 2005)). 147 148 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 7 Consistent with DevPF1’s early peak in mRNA expression from the developmental time 149 course in ParameciumDB, DevPF1-GFP was expressed during the onset of sexual 150 development, but not in vegetative cells with food vacuoles containing bacteria (Figs 2A, 151 S2B). DevPF1-GFP was distributed throughout the cytoplasm and localized in both 152 MICs before and during the S-phase of meiosis, when these nuclei swell (Fig. 2A). 153 Throughout the subsequent meiotic divisions, DevPF1-GFP localized to only a few of 154 the gametic MICs (Fig. 3A). Its micronuclear localization appeared independent of 155 nuclear division as detected by the presence of the spindle apparatus (Fig. 3A,B). 156 During post-zygotic mitotic divisions, DevPF1-GFP was observable in certain post-157 zygotic nuclei, but not in all of them (Fig. 3B). Later during development, DevPF1-GFP 158 was present in the early new MACs and remained in the new MACs throughout 159 development up to very late stages (Fig. 2A) despite the drop in its mRNA levels (Fig. 160 1B). During new MAC development there was also comparatively little cytoplasmic 161 DevPF1-GFP compared to that during meiosis. 162 163 Consistent with its mRNA expression profile, DevPF2-GFP emerged after the onset of 164 new MAC development and localized within the new MACs, where it remained up to the 165 late stages (Fig. 2B). 166 167 Silencing constructs partially co-silence both paralogs 168 We utilized RNAi by feeding to investigate the influence of the DevPFs on IES excision. 169 Two silencing regions were selected (a and b) on each DevPF1 and DevPF2 (Fig. 4A). 170 Due to the lack of regions with sufficient specificity for either of the paralogs, co-171 silencing was predicted (see Methods). Hence, we first experimentally verified the 172 possibility of co-silencing with mRNA and protein levels using silencing region a, since it 173 exhibited less off-target hits. 174 175 The mRNA levels of DevPF1 and DevPF2 were examined during a time course 176 experiment (more details and further analysis follow in subsequent sections) (Fig. 4B). 177 Consistent with the published expression profiles (Arnaiz et al., 2017), DevPF1 178 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 8 expression in the ND7 control knockdown (KD) was highest during onset of 179 development and gradually declined to almost no expression at the “very late” time 180 point. The late-expressed DevPF2 peaked at the “late” time point in the control KD. The 181 expression of both genes was strongly reduced upon their respective KDs (DevPF1 182 mRNA levels were reduced upon DevPF1-KD; DevPF2 mRNA levels were reduced 183 upon DevPF2-KD). A lesser reduction was also observed upon silencing of the 184 respective paralog (DevPF1 levels were reduced in DevPF2-KD and vice versa). Thus, 185 the DevPF1 and DevPF2 silencing constructs lead to co-silencing which is less efficient 186 than the target gene silencing. 187 188 To investigate how the changes in mRNA levels affect protein levels, we checked the 189 localization of the GFP-tagged DevPFs upon KDs. Since DevPF1 is expressed 190 throughout the whole development, multiple developmental time points were collected 191 (Fig. S3A). For the late-expressed DevPF2, only cell stages with clearly visible new 192 MACs were considered (Fig. S3B). In addition to ND7-KD, the knockdown of PGM, the 193 gene encoding the PiggyMac IES excisase (Baudry et al., 2009), was performed to test 194 whether the disturbance of IES excision alters DevPF localization. Neither the 195 localization of DevPF1-GFP nor of DevPF2-GFP was impaired by either of the control 196 KDs. In contrast, the GFP signals were almost completely lost upon DevPF1- or 197 DevPF2-KD. To quantify this observation, GFP fluorescence signals were measured in 198 new MACs (Fig. 4C) as both paralogs exclusively localize to the new MACs during late 199 stages. In line with the observed reduction in mRNA levels, DevPF1-GFP expression 200 was efficiently reduced upon DevPF1-KD, whereas DevPF2-KD led to a weaker 201 reduction. For DevPF2-GFP, the levels were almost equally reduced in DevPF1- and 202 DevPF2-KD. Thus, we confirmed co-silencing on both mRNA and protein levels with 203 reduced silencing efficiency compared to the targeted KD. Therefore, all results 204 obtained in KD experiments must be considered, at least in part, as a combined effect 205 of silencing both DevPF1 and DevPF2, albeit with only a partial contribution from the 206 non-targeted gene silencing. 207 208 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 9 To further investigate the impact of co-silencing on the KD analysis we examined IES 209 retention score (IRS) correlations of multiple KD replicates (more details and further 210 analysis in subsequent sections). The DevPF2-KD replicates showed high to moderate 211 correlation among each other while they correlated less well with two out of four 212 DevPF1-KD replicates (Fig. 4D). This suggests that despite the partial co-silencing, 213 individual KD effects might be observed. 214 215 DevPF1 and DevPF2 affect IES excision genome-wide 216 The influence of the DevPFs on genome reorganization was initially investigated with 217 survival tests and IES retention PCRs upon KDs. Reduced protein levels during sexual 218 development can induce errors including IES retention, impacting the survival of the 219 subsequent generation. For survival tests, the growth of the cells that completed their 220 sexual development was followed for several divisions. IES retention PCRs test for the 221 presence (failed excision) of specific IESs in the new MAC genome. ND7-KD and PGM-222 KD were used as negative and positive control, respectively. To investigate the 223 possibility that the observed effects result from off-target silencing of an unrelated gene, 224 two silencing probes (a and b) were tested for each paralog (Fig. 4A). DevPF1 and 225 DevPF2 KDs with either of the silencing probes resembled PGM-KD, with high lethality 226 in the new progeny (Fig. 5A) and retention of selected IESs (Fig. 5B). This indicates that 227 both DevPF1 and DevPF2 contribute to IES excision. 228 229 Next, we tested genome-wide IES retention in enriched new MAC DNA samples. We 230 observed considerably elevated levels of retained IESs in both DevPF1- and DevPF2-231 KD (Fig. 5C,D). Notably, differences between replicates of the same KDs were 232 observed, whereas replicate pairs processed in parallel (see Methods) exhibited similar 233 profiles. Correlations among the paralog replicates indicated that despite varying IES 234 retention distributions, DevPF2-KD replicates demonstrated high correlations among 235 themselves (Fig. 4D). DevPF1-KD replicates correlated less well with each other, and 236 DevPF1-KD replicate 3 (3) showed a high correlation with the DevPF2-KDs. This 237 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 10 indicates that DevPF2-KD replicates were more consistent than the DevPF1-KD 238 replicates. 239 240 Genes that work closely together are expected to show similar KD effects on IES 241 retention. To identify functionally related genes, DevPF1-KD and DevPF2-KD IRS data 242 was correlated with published data from other gene KDs (Fig. 5E). DevPF2-KD (4) was 243 selected from the DevPF2 replicates. DevPF1-KD (2) and DevPF1-KD (4) were 244 selected as representative of the variability observed in the DevPF1-KDs. DevPF2-KD 245 (4) displayed high correlation with other KDs, such as TFIIS4 and DCL2/3/5 (Fig. 5E). 246 Moderate correlation was observed for DevPF1-KD (4) with SPT5m, whereas DevPF1-247 KD (2) did not correlate well with any of the tested KDs. 248 249 Short IESs are proposed to predominantly rely on the excision complex (specifically 250 Pgm (Baudry et al., 2009) and Ku80c (Marmignon et al., 2014)) for removal, while long 251 IESs tend to require additional molecules for excision (Sellis et al., 2021). To determine 252 whether DevPF1- and DevPF2-KD preferentially affect long IESs, the length distribution 253 of the top 10% of highly retained IESs in each KD was plotted (Fig. S4A,B, Table S1). In 254 comparison to the length distribution of all IESs, DevPF2-KD (4) showed an 255 overrepresentation of long IESs, similar to observations in EZL1-KD, silencing of the 256 catalytic subunit of the PCR2 complex (Frapporti et al., 2019; Lhuillier-Akakpo et al., 257 2014), or DCL2/3/5-KD, silencing of the scnRNA and iesRNA biogenesis proteins 258 (Lepère et al., 2009; Sandoval et al., 2014), (Fig. S4A). Conversely, the highly retained 259 IESs in DevPF1-KD (2) did not show the same overrepresentation and resembled the 260 profile in PGM- and KU80c-KD, silencing of two members of the excision complex. 261 Again, the replicates of the DevPF KDs exhibited variation in the extent of the observed 262 effect (Fig. S4B). 263 264 Defects in IES excision not only result in the retention of IESs but can also lead to 265 excision at alternative TA boundaries. So far, alternative excision above background 266 levels has only been reported for silencing of ISWI1 and its complex partners (Singh et 267 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 11 al., 2022, 2023). Neither DevPF1-KD nor DevPF2-KD resulted in elevated levels of 268 alternative excision (Fig. S4; Table S2). 269 270 DevPF1- and DevPF2-KD alter the small RNA population 271 The early-produced 25 nt scnRNAs and the late-produced 26-30 nt iesRNAs have been 272 proposed to assist MIC-specific sequence targeting in the new MACs (Sandoval et al., 273 2014). Therefore, the small RNA populations across developmental time points upon 274 DevPF KDs were analyzed (Figs 6A, S6A). In DevPF1-KD (2), scnRNA production was 275 completely abolished, an effect also observed in the KD of genes proposed to be 276 involved in scnRNA production: the two scnRNA-processing genes DCL2 and DCL3 277 (Sandoval et al., 2014), and STP5m, involved in the generation of the transcripts serving 278 as substrates for Dcl2/3 cleavage (Gruchota et al., 2017). The KD of the late-expressed 279 DevPF2 showed a much weaker reduction of scnRNA production, which might be 280 caused by co-silencing of DevPF1. 281 282 To further investigate DevPF1’s effect on the scnRNA pathway, we observed Ptiwi09-283 GFP localization upon DevPF1-KD. Ptiwi09, together with Ptiwi01, stabilizes the 284 scnRNAs throughout scnRNA selection in the parental MAC and targeting of MIC-285 specific sequences in the new MACs (Bouhouche et al., 2011; Furrer et al., 2017). As 286 previously described (Bouhouche et al., 2011; Singh et al., 2023), Ptiwi09-GFP localizes 287 first to the cytoplasm and parental MAC with a transient localization in the swelling MICs 288 before shifting to the new MAC (Fig. 6B). Upon DevPF1-KD (Fig. 6B), the localization to 289 the MICs before meiosis I is not impaired; however, the translocation into the parental 290 MAC is strongly reduced and Ptiwi09-GFP predominantly remains in the cytoplasm 291 throughout meiosis II and MAC fragmentation. We have reported a similar change in 292 Ptiwi09-GFP localization upon DCL2/3-KD (Singh et al., 2023), suggesting that the loss 293 of scnRNAs is responsible for the failed protein transfer into the parental MAC. Similar 294 to DCL2/3-KD, DevPF1 depletion does not affect Ptiwi09-GFP’s localization to the new 295 MACs (Fig. 6B). 296 297 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 12 Interestingly, DevPF1-HA IP at two developmental time points (early: about 30% 298 fragmentation; late: visible new MACs in fragmented cells) identified Ptiwi01/09 as 299 potential interaction partners of DevPF1 with a higher enrichment in the early than the 300 late time point (Fig. S6B, Table S3). None of the other small RNA-related proteins were 301 detected (Dcls, Spt5m, TFIIS4 or Ptiwi10/11). 302 303 For both DevPF1- and DevPF2-KD, iesRNA production was impaired. iesRNAs are 304 proposed to derive from dsRNAs transcribed from excised IESs (Allen et al., 2017; 305 Sandoval et al., 2014). Hence, failed excision of IESs in DevPF1- or DevPF2-KD 306 contributes to reduced iesRNA levels, as has consistently been observed for many 307 other KDs of genes involved in Paramecium genome editing (Charmant et al., 2023; de 308 Vanssay et al., 2020; Ignarski et al., 2014; Maliszewska-Olejniczak et al., 2015; Singh et 309 al., 2022; Wang et al., 2023). The lack of scnRNAs in the DevPF1-KD cannot explain 310 the absence of iesRNAs, as these accumulate even if the preceding scnRNA production 311 is blocked (Sandoval et al., 2014). In the late time point analyzed for DevPF IPs, 312 peptides mapping to Ptiwi10/11/06 were detected in DevPF2-IP (Fig. S6C, Table S3), 313 but not DevPF1-IP (Fig. S6B). Therefore, DevPF2 might contribute to iesRNA 314 biogenesis by an interaction with Ptiwi proteins. 315 316 DevPF1- and DevPF2-KD affect mRNA expression 317 Since PHD fingers have often been reported to be involved in gene expression 318 regulation (Aasland et al., 1995; Sanchez & Zhou, 2011) we sought to investigate 319 whether the DevPF KDs alter mRNA expression levels during development. Batch 320 effects had a major influence on the variance within the replicates (Fig. S7A), as 321 observed for IES retention (Fig. 5C,D). 322 323 DevPF1-KD showed almost no differentially expressed genes compared to ND7-KD 324 during onset of development (Fig. 7A). During this early stage, genes are transcribed 325 solely from the parental MAC, where DevPF1-GFP does not localize (Figs 2A, 3). 326 Surprisingly, in DevPF2-KD, a high number of genes were differentially expressed 327 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 13 during the onset of development (Fig. 7A). Since DevPF2 is late-expressed and 328 DevPF1-KD showed no effect, the observed difference might be caused by differing cell 329 stages within the collected populations of DevPF2-KD and ND7-KD. During the “early”, 330 “late” and “very late” time points, DevPF1- and DevPF2-KD showed similar changes in 331 mRNA expression. 332 333 The abolishment of development-specific small RNAs in the DevPF-KDs might result 334 from downregulation of genes involved in scnRNA or iesRNA production. We observed 335 no general trend indicating a drastic reduction of expression of scnRNA-related genes, 336 like DCL2, PTIWI01 or SPT5m (Figs 7B, S7B, Table S4, S5). However, these trends in 337 expression should be considered with the caveat of considerable expression variability 338 and limitation of the number of replicates that could practically be obtained. At least for 339 Ptiwi09, the localization experiments upon DevPF1-KD confirmed no loss in protein 340 levels (Fig. 6B). 341 342 The expression of iesRNA-related genes was altered in both DevPF1- and DevPF2-KD 343 compared to ND7-KD (Figs 7D, S7B, Table S4, S5). DCL5, the Dicer-like protein 344 responsible for the initial cleavage of IES derived dsRNAs into small iesRNAs 345 (Sandoval et al., 2014), was downregulated (Table S4, S5) in early stages, but tended to 346 be upregulated in the very late stage (Table S4, S5). PTIWI10 and PTIWI11, the Piwi 347 proteins responsible for further processing and stabilization of iesRNAs during the 348 positive feedback loop (Furrer et al., 2017), were downregulated in both DevPF1- and 349 DevPF2-KD (Table S4, S5). Successful expression of PTIWI10/11 has been proposed 350 to depend on IES excision since both genes are expressed from the new MAC and 351 harbor IESs in their flanking/coding regions (Furrer et al., 2017) (Fig. S7C). If IES 352 retention was the only cause for downregulation, one would expect higher IRSs for 353 these IESs in KDs with lower mRNA levels. While the mRNA reduction is stronger in 354 DevPF1-KD than in DevPF2-KD (Fig. 7D, Table S4, S5), this trend is not reflected in the 355 IRSs of the IESs whose retention is proposed to interfere with PTIWI10/11 expression 356 (Table 1). In most of the KD replicates, there is no or low retention (IRS < 0.1) and the 357 replicates showing moderate to high retention (0.1 < IRS < 0.3) belong to both DevPF1- 358 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 14 and DevPF2-KD. Hence, the reduced mRNA levels of PTIWI10/11 cannot only be 359 explained by IES retention. 360 361

362 Implications of the PHD domain for DevPF1 and DevPF2 functions 363 Genome reorganization is a fundamental process underlying cell and immune system 364 development and some diseases (Bassing et al., 2002; Forment et al., 2012; Mani & 365 Chinnaiyan, 2010; Rooney et al., 2004). Ciliates undergo massive genome 366 reorganization during the maturation of their somatic genome. This makes them 367 excellent models for studying the complex mechanisms involved in the targeted 368 elimination of genomic sequences (Beisson et al., 2010d). In the present study, we 369 described two paralogous PHD finger proteins, DevPF1 and DevPF2, involved in IES 370 excision in Paramecium. Both paralogs harbor a PHD and a PHD zinc finger-like 371 domain (Fig. 1). These domains belong to the zinc-finger family and the PHD domain is 372 characterized by a well-conserved C4HC3 motif (Aasland et al., 1995; Schindler et al., 373 1993). The eight core amino acids of this motif coordinate two zinc ions and thereby 374 provide structural stability to the domain (Pascual et al., 2000). Among other histone-375 binding domains, such as bromodomains or PWWP domains, PHD fingers are the 376 smallest (Fleck et al., 2021; Miller et al., 1985). Multiple sequence alignment and 377 structure predictions confirmed the presence of the characteristic C4HC3 motif in both 378 DevPF1 and DevPF2 (Fig. 1), suggesting that both PHDs might be functional. 379 380 PHD fingers, mainly nuclear proteins, are often considered epigenetic readers, 381 recognizing histone modifications, primarily on the histone 3 (H3) N-terminal tail 382 (Sanchez & Zhou, 2011). Peptides matching to histones were enriched in the DevPF-383 IPs of late developmental time points (Fig. S6B,D, Table S3), however none of them 384 was specific to H3. PHD fingers have been reported to bind non-H3 partners, like DNA, 385 histone 4, or other proteins (Bienz, 2006; Black & Kutateladze, 2023; Gaurav & 386 Kutateladze, 2023; L. Liu et al., 2012; Oppikofer et al., 2017). The combination of the 387 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 15 PHD and PHD-zinc-finger-like domain in the DevPFs may enable the paralogs to 388 simultaneously recognize two adjacent histone modifications, as demonstrated for 389 tandem PHD domains (Zeng et al., 2010). PHD domains are also found in various 390 chromatin associated proteins involved in gene regulation. Notably, ISWI-containing 391 chromatin remodeling complexes often include a subunit with a PHD domain, such as 392 the ACF (Eberharter et al., 2004), NURF (Haitao Li et al., 2006; Wysocka et al., 2006) or 393 WICH (Bozhenok et al., 2002) complexes. 394 395 DevPF2 was initially identified in pulldowns of the ISWI1 protein, and, thus far, no PHD-396 containing protein has been shown to be a part of this remodeling complex (Singh et al., 397 2022, 2023). It is intriguing to consider that DevPF2 might contribute PHD functionality 398 to the ISWI1 chromatin remodeling complex. However, DevPF2-KD does not show 399 elevated levels of alternative excision (Fig. S4C-E) that are characteristic of other 400 members of the complex so far (Singh et al., 2022, 2023) and ISWI1 was not identified 401 as a potential interaction partner in the DevPF2-IP (Fig. S6C). If DevPF2 interacts with 402 the ISWI1 complex, we infer that it may not be a core complex component, particularly 403 as it does not contribute to excision precision. 404 405 A potential role for DevPF1 and DevPF2 as transcription factors? 406 A potential role in non-coding transcription in the MICs (for scnRNA production) 407 DevPF1’s localization in the MICs (Figs 2A, 3) and its importance for scnRNA 408 production (Fig. 6A) could point towards its involvement in the bidirectional transcription 409 of the MIC genome for scnRNA production. Spt5m (Gruchota et al., 2017) and TFIIS2/3 410 (Maliszewska-Olejniczak et al., 2015) are proposed to be involved in this micronuclear 411 transcription. One of the DevPF1-KD replicates showed moderate IRS correlation with 412 SPT5m (Fig. 5E) (to our knowledge, no IRS data exists for TFIIS2 or TFIIS3) and 413 SPT5m-KD also reduces scnRNA production. The localization of Dcl2-GFP (Lepère et 414 al., 2009), Ptiwi09-GFP (Fig. 6B) and DevPF1-GFP (Fig. 2A) in the swelling MICs 415 suggests that scnRNA biogenesis occurs during the S-phase of meiosis. Ptiwi09 and 416 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 16 DevPF1 may interact in the MICs or the cytoplasm. Non-crosslinked IP’s would be 417 needed to further verify this interaction. However, PTIWI01/09-KD does not completely 418 abolish scnRNAs (Furrer et al., 2017), indicating that DevPF1 acts upstream of scnRNA 419 loading and guide strand removal. Future investigations of bi-directional transcription 420 and scnRNA biogenesis will allow to identify how all these molecules cooperate. 421 422 Spt5m-GFP, TFIIS2/3-GFP and DevPF1-GFP are present in the MICs beyond S-phase 423 and localize to the new MACs at later stages (Gruchota et al., 2017; 424 Maliszewska-Olejniczak et al., 2015). Their role in the MIC during meiotic divisions 425 remains unknown. It was speculated that Spt5m might be involved in co-transcriptional 426 deposition of epigenetic marks that sustain meiotic processes, ultimately aiding in IES 427 targeting. The potential of PHD domains to bind histone modifications raises a similar 428 possibility for DevPF1. However, its role appears to be more specific, as DevPF1 is not 429 present in all gametic and zygotic nuclei simultaneously (Fig. 2&3). 430 431 Msh4/5, homologs of proteins essential for crossover, are also present in all gametic 432 nuclei during the first and second meiotic division, and their silencing leads to 433 substantial IES retention (Rzeszutek et al., 2022). However, their non-canonical 434 functions that lead to IES retention are not yet fully understood (Rzeszutek et al., 2022). 435 Since new MACs develop in DevPF1-KD (Figs 6B,S3) and MSH5-KD cells, neither of 436 the genes are essential for crossover or karyogamy. More research will be needed in 437 future to decipher the functions of the DevPF proteins in the gametic nuclei. 438 439 A potential role in non-coding transcription in the new MAC (for scnRNA-based 440 targeting and iesRNA production) 441 Non-coding transcription in the new MAC, which is hypothized to generate substrates 442 for scnRNA pairing, was proposed to be regulated by the putative transcription 443 elongation factor TFIIS4 that specifically localizes to the early new MACs 444 (Maliszewska-Olejniczak et al., 2015). DevPF2-KD IRSs of some replicates correlated 445 most strongly with TFIIS4-KD (Fig. 5E), pointing towards a shared functionality. Both 446 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 17 DevPF1 and DevPF2 have the potential to act in the same regulatory process as TFIIS4 447 because both their GFP-fusions localize to the new MACs. In fact, there are reports of 448 transcription factors that combine the TFIIS and PHD domains: Bypass of Ess1 (Bye1) 449 protein in Saccharomyces cerevisiae harbors a PHD and a TFIIS-like domain, with the 450 former recognizing histone 3 lysine 4 trimethylation and the latter establishing contact 451 with polymerase II for transcriptional regulation (Kinkelin et al., 2013; Pinskaya et al., 452 2014). It is possible that similar functionality is separated on two individual proteins in 453 Paramecium. However, TFIIS4 was not detected in either of the DevPF-IPs in the late 454 developmental stage. 455 456 The production of iesRNAs was also proposed to depend on the non-coding 457 transcription of concatenated excised IES fragments (Allen et al., 2017; Sandoval et al., 458 2014). Although it was established that IES concatemers are likely formed by DNA 459 ligase 4 (Lig4) (Allen et al., 2017), little is known about the proposed bidirectional 460 transcription to produce substrates for Dcl5 cleavage. Allen et al. speculated on the 461 involvement of TFIIS4. Since iesRNA production is almost completely abolished in 462 DevPF1- and DevPF2-KD, a contribution to this transcription is plausible. 463 464 The potential function of the DevPFs may extend far beyond TFIIS4-dependent 465 transcription: whereas TFIIS4-GFP localizes transiently to early new MACs 466 (Maliszewska-Olejniczak et al., 2015), DevPF2-GFP and DevPF1-GFP remain in the 467 new MACs for much longer (Fig. 2). 468 469 A potential role in gene transcription in the parental and the new MAC 470 Early in development, the parental MAC is solely responsible for gene expression and, 471 after genome reorganization progresses, the new MAC contributes at later stages 472 (Berger, 1973). In Tetrahymena, E2F family transcription factors were shown to control 473 the cell cycle through gene expression during meiosis (Zhang et al., 2018). DevPF1 and 474 DevPF2 are unlikely to be active in the parental MAC since none of the GFP-fusion 475 proteins localized there (Fig. 2). Consistently, DevPF1-KD showed no differential gene 476 expression compared to ND7-KD during the onset of development (Fig. 7C) and 477 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 18 Ptiwi09-GFP expression was not impaired upon DevPF1-KD (Fig. 6B). However, it is 478 difficult to reach a definite conclusion for other genes due to the high variability in 479 expression between the replicates (Figs 7D, S7B) and the high number of differentially 480 expressed genes in DevPF2-KD (Fig. 7C) observed during the onset of development. 481 Cells in the “onset” time point are challenging to collect because cell staging relies on 482 MAC morphology changes visualized by DAPI staining. Truly vegetative cells cannot be 483 distinguished from cells initiating meiosis since their MACs look the same; however, the 484 gene expression profiles are expected to differ substantially (Figs 2A, S2B). The 485 collection of subsequent time points is more reliable because the alteration of old MAC 486 shape as development progresses is pronounced. 487 488 At the subsequent stages, DevPF1- and DevPF2-KD affected similar genes. Either, the 489 changes are nonspecific to the DevPF-KDs and result from the proposed nuclear 490 crosstalk to adjust transcription levels to accommodate for failed IES excision 491 (Bazin-Gélis et al., 2023) or they are specific to the DevPF-KDs and both paralogs 492 exhibit similar functions in the regulation of gene expression. Interestingly, differential 493 expression was observed at the “early” time point (Fig. 7C). GTSF1-KD, also causing 494 substantial IES retention, hardly shows any differentially expressed genes at a 495 comparable stage (DevPF1/2-KD: 282/231 differentially expressed genes, respectively, 496 at about 30% fragmentation (Fig. 7C); GTSF1-KD: 10 differentially expressed genes at 497 about 30-50% fragmentation; (Wang et al., 2023)). This indicates that the early change 498 in gene expression might be specific to DevPF-KDs, potentially mediated by other 499 proteins shuttling into the parental MAC. Since Ptiwi09-GFP translocates efficiently to 500 the parental MAC upon GTSF1-KD (Wang et al., 2023) but not upon DevPF1-KD (Fig. 501 6B), it might be worth investigating differential expression upon PTIWI01/09-KD. 502 503 Late in development, gene expression starts from the new MACs (Berger, 1973), where 504 both DevPF paralogs localized (Fig. 2). Some late-expressed genes, like PTIWI10, are 505 expressed only from the new MAC after the initial onset of IES excision (Furrer et al., 506 2017). Indeed, PTIWI10/11 mRNA levels are downregulated in DevPF1-KD or DevPF2-507 KD (Fig. 7D, S7B, Table S4, S5). This trend cannot be explained solely by the strength 508 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 19 of retention observed for the IESs interfering with PTIWI10/11 expression (Table 1). It 509 suggests that DevPF1 and DevPF2 may regulate gene expression in the new MAC, 510 albeit specifically for some genes like PTIWI10 and PTIWI11. The extent of gene 511 expression regulation by the DevPFs beyond these genes remains uncertain. To further 512 investigate if the DevPFs serve as transcription factors, and if so, which genes they 513 regulate, genes associated with DevPF binding could be identified by techniques like 514 Cut-and-Run (Skene et al., 2018) and compared to mRNA expression changes upon 515 DevPF-KDs. 516 517 Potential cytoplasmic functions 518 In contrast to the other putative transcription factors discussed so far (Spt5m, 519 TFIIS2/3/4, DevPF2), DevPF1-GFP exhibits a pronounced cytoplasmic distribution in 520 the early stages of development (Fig. 2A). While most described PHD fingers are 521 nuclear proteins, some can be recruited to the cytoplasm or plasma membrane by 522 binding partners (Betz et al., 2004; Gozani et al., 2003). DevPF1 may play a role in 523 transmitting signals of sensed starvation to the MICs, initiating sexual development. As 524 DevPF1 is not constitutively expressed during vegetative growth (Figs 1B, S2B), 525 another factor is needed to first initiate DevPF1’s gene expression in the parental MAC. 526 However, DevPF1 might interact with specific markers of starvation in the cytosol, 527 promoting early sexual processes. If that is the case, DevPF1 is not essential for 528 general meiotic processes, as meiosis and new MACs development show no defects in 529 DevPF1-depleted cells (Figs 6B,S3). Since peptides matching Ptiwi01/09 were identified 530 in the DevPF1-IP, the Ptiwi01/09 complex is a potential binding partner of DevPF1 in 531 the cytoplasm. However, since Ptiwi01/09 are highly expressed proteins (Bouhouche et 532 al., 2011), further IP experiments would be needed to verify this interaction. 533 534 DevPF1’s selective localization to gametic and post -zygotic nuclei 535 The selective localization of DevPF1 to certain gametic and post-zygotic nuclei (Fig. 3) 536 raises intriguing questions about its potential role in nuclear fate decisions. The survival 537 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 20 and destruction of the gametic nuclei depends on their subcellular positioning 538 (Grandchamp & Beisson, 1981). DevPF1 may play a role in either promoting their 539 movement or preparing for their degradation. However, the observed number of nuclei 540 simultaneously containing DevPF1-GFP (zero to four) neither fits the number of nuclei 541 selected for survival (one) nor for degradation (seven). DevPF1 may either contribute to 542 this process successively or may not be directly related to the nuclear fate itself. The 543 fate of the post-zygotic nuclei is decided during the second mitotic division by the 544 subcellular localization of the division products (Grandchamp & Beisson, 1981). This 545 means, from each post-zygotic nucleus, one of the division products will remain as MIC 546 and one develops into a new MAC. During the second mitotic division, DevPF1-GFP 547 was observed in one of the two dividing nuclei. Its localization in the precursor of one 548 MIC and one MAC without being present in the precursor of the other MIC and MAC, 549 does not imply its involvement in the nuclear fate decision. Furthermore, DevPF1-KD 550 neither impaired the selection of gametic nuclei nor the differentiation of the new MACs. 551 552 The specific localization of nuclear proteins to certain nuclei in multinuclear cells has 553 been studied extensively in insect embryos. In Drosophila, the transcription factors 554 Bicoid (Driever & Nüsslein-Volhard, 1988) and Dorsal (Roth et al., 1989) establish the 555 anterior-posterior, and dorsal-ventral axis, respectively, by initiating gene expression 556 depending on the cytoplasmic localization of the nuclei. The activity of the transcription 557 factors is restricted by gradients to a certain cytoplasmic region (Morisato & Anderson, 558 1995; Spirov et al., 2009). However, DevPF1-GFP’s nuclear localization does not 559 appear associated with subcellular localization of the nuclei and it remains unclear how 560 DevPF1-GFP is specifically recruited. 561 562 As only fixed cells were examined, the dynamics of DevPF1-GFP localization were not 563 captured. The fact that DevPF1-GFP localization is independent of nuclear divisions 564 (Fig. 3B), combined with observations of cells at the meiotic or mitotic division stage 565 with an absence of DevPF1-GFP in all nuclei (Fig. S2C), suggests that DevPF1 566 localization might be asynchronous and transient. Possibly it is recruited to each of the 567 gametic nuclei at some point before the completion of the second meiotic division and to 568 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 21 each of the post-zygotic nuclei before completion of the second mitotic divisions. Live 569 cell imaging could illuminate the dynamics of DevPF1 localization and its correlation 570 with nuclear fate. However, this approach presents challenges, as it requires confocal 571 imaging to capture the DevPF1-GFP signal in the MICs, and the observation time scale 572 would need to span across multiple hours of Paramecium development. 573 574 DevPF1: a general factor for IES excision 575 DevPF1 plays a role throughout sexual development: from the early stages before 576 meiosis to the very late stages (Fig. 2A). It appears to influence various aspects of 577 genome reorganization in the MICs and the new MACs, including scnRNA production 578 and potentially expression of certain genes. Consequently, the depletion of DevPF1 579 affects the excision of a wide range of IESs (Fig. 5C). However, it is important to 580 reiterate that we observed high batch-to-batch variability in the DevPF replicates in both 581 IES retention (Fig. 5C, D) and mRNA expression (Figs 7D, S7B). The time point 582 collection had a major influence on mRNA levels (Fig. S7A). Variable new MAC 583 enrichment by a sucrose gradient might introduce variation into the IRS analysis, as 584 fragments of the parental MAC add unexcised IES sequences, diluting the effect of IES 585 retention (Charmant et al., 2023). Fluorescence-activated nuclear sorting (FANS) 586 enables better nuclear separation in Paramecium (Charmant et al., 2023; Guérin et al., 587 2017) and should be able to eliminate most of such variation. Additionally, 588 microinjection of DNA into macronuclei before RNAi experiments can be used to control 589 for contaminating DNA from old MAC fragments. 590 591 Revisiting previous KD experiments with additional replicates would be worthwhile to 592 explore the extent of batch-to-batch IRS and expression variance for other KDs. 593 Noteworthy, variability in IES retention across replicates has recently been shown for 594 GTSF1 (Charmant et al., 2023; Wang et al., 2023), suggesting this phenomenon is not 595 restricted to DevPF1 and DevPF2. In general, KD experiments are challenging to tightly 596 control for reproducibility, and more effort should be invested in generating knockouts in 597 Paramecium, as established in Tetrahymena (Chalker, 2012). 598 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 22 It has been shown that evolutionarily old IESs tend to be short and are excised early in 599 development, independent of additional factors apart from the excision machinery 600 (Sellis et al., 2021). On the other hand, evolutionarily young IESs tend to be long, later 601 excised and dependent on the scnRNA pathway and the deposition of histone 602 modifications in the new MAC for their excision (Sellis et al., 2021; Swart et al., 2014; 603 Zangarelli et al., 2022). In line with this, most gene KDs tested in this study exhibited an 604 overrepresentation of long IESs among their most highly retained IESs, including 605 DevPF2 (Fig. S4A,B). Only PGM-KD, KU80c-KD and two of the DevPF1-KD replicates 606 showed no preference for long IESs. Pgm and Ku80c are components of the excision 607 machinery and are therefore expected to affect all IESs. While DevPF1 may not be a 608 direct part of the excision machinery, it appears to have a general contribution to IES 609 excision, regardless of the length of the IES. Consequently, we propose that DevPF2 610 contributes to the excision of long IESs, while DevPF1 may serve as a more general 611 factor. 612 613

614 Paramecium tetraurelia cultivation 615 Mating type 7 (MT7) cells from strain 51 of Paramecium tetraurelia were grown in 616 Wheat Grass Powder (WGP, Pines International) medium supplemented with 10 mM 617 sodium phosphate buffer (pH 7.3). WGP medium was bacterized with E.coli strain 618 HT115 to feed paramecia, and the cultures were maintained either at 27°C or at 18°C 619 according to the standard protocol (Beisson et al., 2010b, 2010c). 620 621 Protein localization imaging by fluorescence microscopy 622 Plasmids for microinjection were generated by amplifying the coding and flanking 623 sequences from MT7 genomic DNA and introducing them with the PCR-based method 624 CPEC (Quan & Tian, 2011) into the L4440 plasmid (Addgene, USA). DevPF1 was 625 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 23 expressed with its endogenous flanking regions (304 bp upstream of the DevPF1 start 626 codon and 272 bp downstream of the DevPF1 stop codon). DevPF2 endogenous 627 flanking regions (455 bp upstream the DevPF2 start codon and 273 bp downstream of 628 the DevPF2 stop codon) yielded no expression. Therefore, as PGM exhibits a similar 629 expression profile to DevPF2 (Fig. 1B), DevPF2 genomic coding sequence was inserted 630 between the PGM flanking regions (96 bp upstream of the PGM start codon and 54 bp 631 downstream of the PGM stop codon). Before the stop codon, the GFP coding sequence 632 was connected to the protein coding sequences via a glycine-serine-linker 633 (SSGGGSGGSGGGS). 60 μg of plasmid DNA was linearized with AhdI (New England 634 Biolabs, UK) and extracted with phenol-chloroform for injection. 635 636 Paramecia were microinjected with either C-terminally GFP-tagged DevPF1 637 (endogenous regulatory regions) or C-terminally GFP-tagged DevPF2 (PGM regulatory 638 regions) following the standard protocol (Beisson et al., 2010a). Sexual development 639 was induced by starvation and cells of different developmental stages were collected 640 and stored in 70% ethanol at -20°C. To stain cells with DAPI (4,6-diamidino-2-2-641 phenylindole), cells were dried on a microscopy slide, washed twice with phosphate-642 buffered saline (PBS) and permeabilized for 10 min at RT (room temperature) with 1% 643 Triton X-100 in PHEM (PIPES, HEPES, EGTA, magnesium sulfate), fixed with 2% 644 paraformaldehyde (PFA) in PHEM and washed once for 5 min at (RT) with 3% BSA 645 (bovine serum albumin, Merck-Sigma, Germany) in Tris-buffered saline with 10 mM 646 EGTA and 2 mM MgCl2 (TBSTEM). After DAPI (2 μg/ml in 3% BSA) incubation for 7-10 647 min at RT, the cells were mounted 40 µl of ProLong Gold Antifade mounting medium 648 (Invitrogen, USA) or ProLong Glass Antifade mounting medium (Invitrogen, USA). For 649 α-tubulin staining, after permeabilization and fixation, cells were blocked for 1 h at RT 650 with 3% BSA and 0.1% Triton X-100. Primary rat anti-α-tubulin antibody (Abcam, UK) 651 was diluted 1:200 in 3% BSA and 0.1% Triton X-100 in TBSTEM and incubated 652 overnight at 4°C. After 3 washes with 3% BSA, the goat anti-rat secondary antibody 653 conjugated to Alexa fluorophore 568 (Abcam, UK) was diluted 1:500 in 3% BSA and 654 0.1% Triton X-100 in TBSTEM and incubated for 1 h at RT. After two washes, cells 655 were stained with DAPI and mounted with Prolong Glass Antifade mounting medium. 656 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 24 Images were acquired on a confocal SP8 Leica fluorescence microscope (60x/1.4 oil 657 objective) with constant laser settings. The detector (photon multiplier) gain for the DAPI 658 signal (430-470 nm) varied to accommodate differences in signal strength (500-550 V). 659 Postprocessing was done in Fiji (version 2.14.0/1.54f) (Schindelin et al., 2012). 660 Brightness and contrast in the GFP channel was set the same in all the images to be 661 compared (Figs 2, S2B: DevPF1-GFP: Min 0, Max 681 and DevPF2-GFP: Min 0, Max 662 170; Figs 3, S2C: DevPF1-GFP: Min 0, Max 703; Fig. S3: constant settings for each cell 663 stage). 664 665 Knockdown efficiency validation using fluorescence intensity 666 Cells injected with either DevPF2-GFP or DevPF1-GFP were subjected to KDs of ND7, 667 PGM, DevPF2 and DevPF1 genes. Cells during new MAC development were collected 668 (for details see methods on silencing experiments), then stained with DAPI and 669 mounted on ProLong Glass Antifade as described above. Images of a single z-plane 670 through the new MAC were acquired on a SP8 Leica Confocal microscope with 60x/1.4 671 oil objective using the same laser settings for all images. For each KD, 10 cells were 672 imaged. In Fiji software (version 2.14.0/1.54f), the brightness and contrast in the GFP 673 channel was set the same values for all images compared in the same analysis 674 (DevPF1-GFP injected cells: Min 0, Max 1078; DevPF2-GFP injected cells: Min 0, Max 675 298). Fluorescent signal was measured in a constant area in 1 MAC of each cell and 676 the area mean was used as intensity for this nucleus. The area was set in the DAPI 677 channel and the fluorescence was measured in the GFP channel. Since the same area 678 was measured for each nucleus, no normalization was used to account for nuclear size 679 variation. To account for background fluorescence, GFP fluorescence in non-680 transformed wild type cells was measured and the mean of all wild type cells was 681 subtracted from all measured intensities. All intensities were normalized to the mean of 682 all ND7-KD cells in the corresponding injection. All scripts are available from 683 https://github.com/Swart-lab/DevPF_code. 684 685 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 25 Co-immunoprecipitation 686 Paramecia were injected with either Human influenza hemagglutinin (HA)-tagged 687 DevPF1 (same cloning strategy as described before) or GFP-tagged DevPF2. For 688 DevPF1-HA, an early time point (about 30% fragmentation) and late time point (new 689 MACs clearly visible in fragmented cells) was collected, while for DevPF2-GFP, only the 690 late time point was collected. Non-transformed wild type cells were collected as 691 controls. Cells were washed twice with 10 mM Tris and as much liquid was removed as 692 possible. For 300 ml initial culture volume, cells were fixed with 1 ml 1% PFA for 10 min 693 at RT and quenched with 100 µl of 1.25 M glycine for 5 min at RT. After one wash with 694 PBS (centrifugation for 1 min at 4°C and 1000 g), 2 ml lysis buffer (50 mM Tris, 150 mM 695 NaCl, 5 mM MgCl2, 1% Triton X-100, 10% Glycerol and cOmplete protease inhibitor 696 EDTA-free (Roche, Germany)) were added and cells were sonicated using an MS72 tip 697 on a Bandelin Sonopulse device with 52% amplitude for 15 s on ice. The pellet and 698 input fraction were separated by centrifugation (13,000 g, 4°C, 30 min). 699 700 To enrich HA-tagged proteins, 50 µl beads (Anti-HA-affinity matrix, Merck-Sigma, 701 Germany) were washed thrice (500 g, 4°C, 2 min) in ice-cold IP buffer (10 mM Tris pH 702 8, 150 mM NaCl, 1 mM MgCl2, 0.01% NP-40, 5% Glycerol, cOmplete protease inhibitor 703 EDTA-free (Roche, Germany) and incubated with 1 ml of cleared input lysate overnight 704 at 4°C. After four washes with ice-cold IP buffer, the bound proteins were eluted from 705 the beads in 50 µl 2× PLB (10% SDS, 0.25 M Tris pH 6.8, 50% Glycerol, 0.2 M DTT, 706 0.25% Bromophenol blue) at 98°C for 20 min (IP fraction). 707 708 To enrich GFP-tagged proteins, 25 µl beads (GFP-Trap Agarose beads, Chromotek, 709 Germany) were washed once with ice-cold 20 mM Tris pH 7.5 with 100mM NaCl (2,500 710 g, 4°C, 5 min) and thrice in ice-cold IP buffer. Beads were incubated with 1 ml cleared 711 input lysate for 1 to 2 h at 4°C and washed four times with ice-cold IP buffer. Bound 712 proteins were eluted in 30 µl 2× PLB at 98°C for 20 min (IP fraction). 713 714 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 26 For western blots, 0.5% of total input and 15% of total IP fraction were resolved on 10% 715 SDS-PAGE gels and wet transferred onto a 0.45 µm nitrocellulose membrane for 2 h at 716 80 V and 4°C (Bio-Rad, Germany). The membrane was blocked for 1 h in 5% BSA in 717 PBST (PBS + 0.2% Tween20). HA-tagged proteins were detected with an HRP-718 conjugated anti-HA antibody (sc-7392 HRP, Santa Cruz, USA) diluted 1:500 in PBST 719 and incubated overnight at 4°C. GFP-tagged proteins were detected with an primary 720 anti-GFP antibody (ab290, Abcam, UK) diluted 1:2000 and incubated overnight at 4°C 721 followed by an secondary anti-rabbit HRP conjugated antibody (12-348, Merck Millipore, 722 Germany) diluted 1:5000 in PBST and incubated for 1 h at RT. Membranes were 723 screened using AI600 (GE Healthcare, Germany). 724 725 Samples were sent to EMBL’s Proteomics Core Facility in Germany for mass 726 spectrometry experiments and analysis. Using R, contaminants were removed from the 727 FragPipe output files (protein.tsv, (Kong et al., 2017)), and only proteins quantified with 728 a minimum of two razor peptides were included for subsequent analysis. After log2 729 transformation of raw TMT reporter ion intensities, batch effect correction (limma 730 package’s (Ritchie et al., 2015) ‘removeBatchEffects’ function), and variance 731 stabilization normalization (vsn) with vsn package (Huber et al., 2002), the abundance 732 difference in WT and DevPF samples was maintained by determining different 733 normalization coefficients. To investigate differential protein expression (limma 734 package), replicate information was incorporated in the design matrix with the ‘lmFit’ 735 limma function. “hit” annotation: false discovery rate (FDR) smaller 5% and a fold 736 change of at least 100%. “candidate” annotation: FDR smaller 20% and a fold change of 737 at least 50%. Scripts to generate volcano plots are available from 738 https://github.com/Swart-lab/DevPF_code. 739 740 Silencing experiments, survival test and IES retention PCR 741 Silencing constructs for DevPF2 and DevPF1 were generated by cloning genomic gene 742 fragments into a T444T plasmid (Sturm et al., 2018) (Addgene, USA) using CPEC 743 (Quan & Tian, 2011). For both DevPF1 and DevPF2, two silencing regions were 744 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 27 selected: DevPF1 silencing region a (525 bp fragment from 3-527; position 1 is the first 745 nucleotide of the start codon); DevPF1 silencing region b (733 bp fragment from 532-746 1264); DevPF2 silencing region a (525 bp fragment from 3-527); DevPF2 silencing 747 region b (731 bp fragment from 532-1262). Co-silencing was predicted with the RNAi 748 off-target tool from ParameciumDB (Heng Li & Durbin, 2009) for both silencing regions 749 (DevPF1 silencing region a and b: 19 and 30 hits, respectively, in DevPF2 gene; 750 DevPF2 silencing region a and b: 19 and 30 hits, respectively, in DevPF1 gene). The 751 plasmids were transformed into HT1115 (DE3) E. coli strain and expression was 752 induced overnight at 30°C with Isopropyl ß-D-1-thiogalactopyranoside (IPTG; Carl Roth, 753 Germany). Paramecia were seeded into the silencing medium at a density of 100 754 cells/ml to induce sexual development by starvation after 4 to 6 divisions. KD 755 experiments were performed as previously described (Beisson et al., 2010e). 756 757 After the paramecia finished sexual development, 15 cells were transferred into a 758 regular, non-induced, feeding medium for the survival test. Paramecia were monitored 759 for three days to observe growth effects. For IES retention PCRs, genomic DNA was 760 extracted from cultures that finished sexual development using GeneElute – Mammalian 761 Genomic DNA Miniprep Kit (Merck-Sigma, Germany). PCRs were done on specific 762 genomic regions flanking an IES (Table S6) to check for the retention of IESs. 1-12.5 ng 763 DNA was used as input andPCR products were resolved on 1-2% agarose gels. 764 765 Time course silencing experiments 766 The time course experiments were conducted in three batches, each processing two KD 767 replicates in parallel (batch A: replicates 1 and 2 of ND7-, DevPF1- and DevPF2-KD; 768 batch B: replicates 3 and 4 of ND7-, DevPF1- and DevPF2-KD; batch C: replicates 5 769 and 6 of ND7- and DevPF2-KD). In batch A and B, cells were collected as soon as the 770 first meiotic cells were observed in the population (onset), between 20 to 40% 771 fragmentation (early), at 80-90% fragmentation (late) and 6 h after the late time point 772 (very late). In batch C, cells were collected before the onset of autogamy (vegetative), at 773 50% fragmentation (early), at 100% fragmentation + visible anlagen (very late) and 6 h 774 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 28 later (very late + 6h). Since batch C was collected at different stages, only the “very late” 775 time point of batch C was considered for differential expression analysis. For all time 776 course replicates, enriched new MAC DNA was analyzed for IES retention and total 777 RNA was collected from the collected time points for sRNA and/or mRNA analysis. 778 779 Macronuclear isolation and Illumina DNA-sequencing 780 Samples for new MAC isolation were collected from the KD cultures of all time course 781 experiments three days after completion of sexual development as described previously 782 (Arnaiz et al., 2012). DNA library preparation (350 bp fragment sizes) and Illumina 783 sequencing (paired-end, 150 bp reads) were done at Novogene (UK) Company Limited, 784 Cambridge according to their standard protocols. 785 786 IES retention and alternative boundary analysis 787 For IES retention score analysis, whole genome sequencing reads of enriched new 788 MAC DNA after KD were adaptor trimmed using TrimGalore (Krueger, 2019) if 789 significant Illumina adapter content was observed using FastQC v0.11.9 (Andrews, 790 2010) (see Table S7 for adapter sequences). The “Map” module of ParTIES v1.05 791 pipeline was used to map the reads on MAC and MAC+IES reference genomes with 792 changes in the /lib/PARTIES/Map.pm file as described in (Singh et al., 2023). The IES 793 retention scores (IRS) were calculated by the “MIRET” module (provided as 794 DevPF_IRS.tab.gz). All scripts are available from https://github.com/Swart-795 lab/DevPF_code. IRS correlations were calculated as described previously (Swart et al., 796 2014). 797 798 Alternative excision was analyzed as described previously (Singh et al., 2023). In brief, 799 properly paired and mapped reads were selected from the output from the ParTIES 800 "Map" module for the MAC+IES reference genome and downsampled to the same 801 library size (DevPF1-KD (1) and DevPF2-KD (2) were excluded due to small library 802 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 29 size). We then employed the "MILORD" module of a pre-release version of ParTIES (13 803 August 2015) with default parameters to annotate alternative and cryptic IES excision. 804 All scripts are available from https://github.com/Swart-lab/DevPF_code. 805 806 The data generate in this study was compared with data of previously published KDs: 807 PGM-KD (Arnaiz et al., 2012), TFIIS4-KD (Maliszewska-Olejniczak et al., 2015), 808 SPT5m-KD (Gruchota et al., 2017), PTCAF1-KD (Ignarski et al., 2014), DCL2/3/5-KD 809 (Sandoval et al., 2014), KU80c-KD (Abello et al., 2020), EZL1-KD (Lhuillier-Akakpo et 810 al., 2014) and ISWI1-KD (Singh et al., 2022). 811 812 RNA extraction and sequencing 813 Total RNA was either extracted with phenol-chloroform followed by Monarch Total RNA 814 Miniprep kit (New England Biolabs) or with the Quick-RNA Miniprep kit (Zymo). For 815 phenol-chloroform extraction (batch C), 300 ml cells subjected to RNAi were washed 816 twice with 10 mM Tris pH 7.5 (RT, 280 g, 2 min) and shock frozen by dropping them 817 directly into liquid nitrogen. 500 μl of 2× DNA/RNA protection reagent from the Monarch 818 kit were added to the frozen pellet and the cells thawed by vortexing. After adding 10 μl 819 proteinase K and 1 ml RNA lysis buffer, the manufacturer's instructions (RNA Binding 820 and Elution (Cultured Mammalian Cells)) were followed. On-column DNase I treatment 821 was included. 822 823 For RNA extraction with Quick-RNA Miniprep kit (batch A and B), 100 ml of 824 Paramecium cultures subjected to RNAi by feeding were washed twice in 10 mM Tris 825 pH 7.5 in pear-shaped oil flasks by centrifugation (RT, 280 g, 2 min). After the final 826 wash, cells were collected on ice and spun at 2,000 g for 2 min and 4°C and as much 827 liquid as possible was removed. 3× volume of 1× DNA/RNA Shield (Biozym) was mixed 828 with the cells and the samples were stored at -70°C until further processing. For RNA 829 extraction, samples were thawed at RT and mixed with 1× volume of RNA lysis buffer. 830 The manufacturer’s instructions were followed (section: (III) Total RNA Purification). 831 832 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 30 Extracted total RNA was send to Azenta Life Sciences for library preparation (sRNA: 833 NEBNext Small RNA Library Prep Set for Illumina; mRNA: NEBNext Ultra II RNA 834 Library Prep Kit for Illumina) and paired-end Illumina sequencing (NovaSeq 2×150bp). 835 836 Small RNA analysis 837 Small RNA sequencing reads were trimmed using cutadapt (Martin, 2011) version 3.2 838 with the parameter -a “AGATCGGAAGAGCACACGTCTGAACTCCAGTCA” to remove 839 the relevant Illumina adaptor sequence. Trimmed reads were mapped to the 840 Paramecium tetraurelia strain 51 MAC + IES genome and L4440 (ND7-KD) or T444T 841 (DevPF1/DevPF2-KD) silencing vector with bwa version 0.7.17-r1188 (Heng Li & 842 Durbin, 2009). GNU grep (version 2.14) was used to select 10-49 bp long, uniquely 843 mapped reads (possessing the SAM file format flags “XT:A:U”) and sRNA length 844 histograms were generated by a Python script. All scripts are available from 845 https://github.com/Swart-lab/DevPF_code. 846 847 mRNA analysis 848 Illumina adapter sequences (Table S7) were trimmed from reads with TrimGalore 849 (Krueger, 2019). Reads were mapped to the Paramecium tetraurelia strain 51 850 transcriptome with hisat2 (Kim et al., 2019) allowing 20 multimappings (-k 20). Using 851 samtools (Heng Li et al., 2009), the properly paired and mapped reads were filtered (-f2 852 flag) and sorted by the read name (-n flag). Unique mapping reads were acquired with 853 eXpress (Roberts & Pachter, 2013) with 5 additional online expectation-maximization 854 rounds to perform on the data after the initial online round (-O 5 flag) to improve 855 accuracy. Scripts are available from https://github.com/Swart-lab/DevPF_code. 856 857 Read counts were normalized with DEseq2 (Love et al., 2014) package in R (version 858 3.6.3). For plotting, DEseq2 in-build functions plotPCA, plotMA and plotCounts were 859 combined with ggplot2 (Villanueva & Chen, 2019) package (version 3.4.3). Differentially 860 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 31 expressed genes were identified for each time point with a Wald test (false discovery 861 rate (alpha) = 0.1). Differentially expressed genes were filtered with an absolute 862 log2(Fold Change) > 2 (corresponding to a 4-fold change) and an adjusted p-value < 863 0.01. The time point, KD and batch were known sources of variation in the dataset 864 (design = ~ batch + timepoint + KD+ timepoint:KD). All scripts are available from 865 https://github.com/Swart-lab/DevPF_code. 866 867 Structure prediction with AlphaFold 868 Protein structures were predicted with AlphaFold2 multimer (Evans et al., 2021; Jumper 869 et al., 2021) using the ColabFold v1.5.2-patch (Mirdita et al., 2022) in Google Colab with 870 default parameters. 871 872 Sequence alignment 873 Domains were predicted using InterProScan (Paysan-Lafosse et al., 2023). The 874 nucleotide sequence of DevPF2 and DevPF1 (including introns) were aligned with 875 clustalOmega (Sievers et al., 2011) (version 1.2.3) pairwise sequence alignment tool in 876 Geneious prime (version 2023.2.1) with default parameters (Fig 4A). 877 878 Multiple sequence alignment of PHD domains was done with clustalOmega (version 879 1.2.1) using the MPI bioinformatics toolkit’s web interface (Zimmermann et al., 2018) 880 with default parameters. 881 882 Manuscript writing 883 Grammar and language refinement were assisted by an AI language model developed 884 by OpenAI (GPT-3.5 architecture) (OpenAI, 2023). 885 886 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 32

887 We thank the BioOptics core facility and Genome center of MPI for Biology (Tübingen, 888 Germany) for their assistance and Andre Noll for computer system administration. 889 890 Competing interests 891 The authors declare no competing interests. 892 893 Funding 894 This work was funded by the Max Planck Society. 895 896 Data availability 897 Supplementary files, including uncropped blot images, microcopy raw files and IES 898 retention scores have been deposited to the open research data repository of the Max 899 Planck Society EDMOND (https://doi.org/10.17617/3.VKJBJ0). Sequencing raw files 900 have been deposited to the European Nucleotide Archive (ENA; 901 https://www.ebi.ac.uk/ena/browser/home) (Leinonen et al., 2011) (accession number: 902 PRJEB67678). The mass spectrometry proteomics data have been deposited to the 903 ProteomeXchange Consortium (Deutsch et al., 2023) via the PRIDE (Perez-Riverol et 904 al., 2022) partner repository (accession number: PXD046704). 905 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 33

906 Aasland, R., Gibson, T. J., & Stewart, A. F. (1995). The PHD finger: implications for 907 chromatin-mediated transcriptional regulation. Trends in Biochemical Sciences, 20(2), 908 56–59. https://doi.org/10.1016/s0968-0004(00)88957-4 909 Abello, A., Régnier, V., Arnaiz, O., Le Bars, R., Bétermier, M., & Bischerour, J. (2020). 910 Functional diversification of Paramecium Ku80 paralogs safeguards genome integrity 911 during precise programmed DNA elimination. PLoS Genetics, 16(4), e1008723. 912 https://doi.org/10.1371/journal.pgen.1008723 913 Allen, S. E., Hug, I., Pabian, S., Rzeszutek, I., Hoehener, C., & Nowacki, M. (2017). 914 Circular Concatemers of Ultra-Short DNA Segments Produce Regulatory RNAs. Cell, 915 168(6), 990-999.e7. https://doi.org/10.1016/j.cell.2017.02.020 916 Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data. 917 (0.11.9) [Computer software]. Babraham Bioinformatics. 918 Arnaiz, O., Mathy, N., Baudry, C., Malinsky, S., Aury, J.-M., Denby Wilkes, C., Garnier, 919 O., Labadie, K., Lauderdale, B. E., Le Mouël, A., Marmignon, A., Nowacki, M., 920 Poulain, J., Prajer, M., Wincker, P., Meyer, E., Duharcourt, S., Duret, L., Bétermier, 921 M., & Sperling, L. (2012). The Paramecium germline genome provides a niche for 922 intragenic parasitic DNA: evolutionary dynamics of internal eliminated sequences. 923 PLoS Genetics, 8(10), e1002984. https://doi.org/10.1371/journal.pgen.1002984 924 Arnaiz, O., Van Dijk, E., Bétermier, M., Lhuillier-Akakpo, M., de Vanssay, A., 925 Duharcourt, S., Sallet, E., Gouzy, J., & Sperling, L. (2017). Improved methods and 926 resources for Paramecium genomics: transcription units, gene annotation and gene 927 expression. BMC Genomics, 18(1), 483. https://doi.org/10.1186/s12864-017-3887-z 928 Bassing, C. H., Swat, W., & Alt, F. W. (2002). The mechanism and regulation of 929 chromosomal V(D)J recombination. Cell, 109 Suppl, S45-55. 930 https://doi.org/10.1016/s0092-8674(02)00675-x 931 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 34 Baudry, C., Malinsky, S., Restituito, M., Kapusta, A., Rosa, S., Meyer, E., & Bétermier, 932 M. (2009). PiggyMac, a domesticated piggyBac transposase involved in programmed 933 genome rearrangements in the ciliate Paramecium tetraurelia. Genes & 934 Development, 23(21), 2478–2483. https://doi.org/10.1101/gad.547309 935 Bazin-Gélis, M., Eleftheriou, E., Zangarelli, C., Lelandais, G., Sperling, L., Arnaiz, O., & 936 Bétermier, M. (2023). Inter-generational nuclear crosstalk links the control of gene 937 expression to programmed genome rearrangements during the Paramecium sexual 938 cycle. BioRxiv. https://doi.org/10.1101/2023.04.16.537068 939 Beisson, J., Bétermier, M., Bré, M.-H., Cohen, J., Duharcourt, S., Duret, L., Kung, C., 940 Malinsky, S., Meyer, E., Preer, J. R., & Sperling, L. (2010a). DNA microinjection into 941 the macronucleus of paramecium. Cold Spring Harbor Protocols, 2010(1), 942 pdb.prot5364. https://doi.org/10.1101/pdb.prot5364 943 Beisson, J., Bétermier, M., Bré, M.-H., Cohen, J., Duharcourt, S., Duret, L., Kung, C., 944 Malinsky, S., Meyer, E., Preer, J. R., & Sperling, L. (2010b). Maintaining clonal 945 Paramecium tetraurelia cell lines of controlled age through daily reisolation. Cold 946 Spring Harbor Protocols, 2010(1), pdb.prot5361. https://doi.org/10.1101/pdb.prot5361 947 Beisson, J., Bétermier, M., Bré, M.-H., Cohen, J., Duharcourt, S., Duret, L., Kung, C., 948 Malinsky, S., Meyer, E., Preer, J. R., & Sperling, L. (2010c). Mass culture of 949 Paramecium tetraurelia. Cold Spring Harbor Protocols, 2010(1), pdb.prot5362. 950 https://doi.org/10.1101/pdb.prot5362 951 Beisson, J., Bétermier, M., Bré, M.-H., Cohen, J., Duharcourt, S., Duret, L., Kung, C., 952 Malinsky, S., Meyer, E., Preer, J. R., & Sperling, L. (2010d). Paramecium tetraurelia: 953 the renaissance of an early unicellular model. Cold Spring Harbor Protocols, 2010(1), 954 pdb.emo140. https://doi.org/10.1101/pdb.emo140 955 Beisson, J., Bétermier, M., Bré, M.-H., Cohen, J., Duharcourt, S., Duret, L., Kung, C., 956 Malinsky, S., Meyer, E., Preer, J. R., & Sperling, L. (2010e). Silencing specific 957 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 35 Paramecium tetraurelia genes by feeding double-stranded RNA. Cold Spring Harbor 958 Protocols, 2010(1), pdb.prot5363. https://doi.org/10.1101/pdb.prot5363 959 Berger, J. D. (1973). Nuclear differentiation and nucleic acid synthesis in well-fed 960 exconjugants of Paramecium aurelia. Chromosoma, 42(3), 247–268. 961 https://doi.org/10.1007/BF00284774 962 Betz, C., Schlenstedt, G., & Bailer, S. M. (2004). Asr1p, a novel yeast ring/PHD finger 963 protein, signals alcohol stress to the nucleus. The Journal of Biological Chemistry, 964 279(27), 28174–28181. https://doi.org/10.1074/jbc.M401595200 965 Bienz, M. (2006). The PHD finger, a nuclear protein-interaction domain. Trends in 966 Biochemical Sciences, 31(1), 35–40. https://doi.org/10.1016/j.tibs.2005.11.001 967 Black, J. C., & Kutateladze, T. G. (2023). Atypical histone targets of PHD fingers. The 968 Journal of Biological Chemistry, 299(4), 104601. 969 https://doi.org/10.1016/j.jbc.2023.104601 970 Bouhouche, K., Gout, J.-F., Kapusta, A., Bétermier, M., & Meyer, E. (2011). Functional 971 specialization of Piwi proteins in Paramecium tetraurelia from post-transcriptional 972 gene silencing to genome remodelling. Nucleic Acids Research, 39(10), 4249–4264. 973 https://doi.org/10.1093/nar/gkq1283 974 Bozhenok, L., Wade, P. A., & Varga-Weisz, P. (2002). WSTF-ISWI chromatin 975 remodeling complex targets heterochromatic replication foci. The EMBO Journal, 976 21(9), 2231–2241. https://doi.org/10.1093/emboj/21.9.2231 977 Chalker, D. L., & Yao, M. C. (2001). Nongenic, bidirectional transcription precedes and 978 may promote developmental DNA deletion in Tetrahymena thermophila. Genes & 979 Development, 15(10), 1287–1298. https://doi.org/10.1101/gad.884601 980 Chalker, D. L. (2012). Transformation and strain engineering of Tetrahymena. Methods 981 in Cell Biology, 109, 327–345. https://doi.org/10.1016/B978-0-12-385967-9.00011-6 982 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 36 Charmant, O., Gruchota, J., Arnaiz, O., Zangarelli, C., Bétermier, M., Nowak, K., 983 Legros, V., Chevreux, G., Nowak, J., & Duharcourt, S. (2023). The nuclear PIWI-984 interacting protein Gtsf1 controls the selective degradation of small RNAs in 985 Paramecium. BioRxiv. https://doi.org/10.1101/2023.09.19.558372 986 Cheng, C.-Y., Vogt, A., Mochizuki, K., & Yao, M.-C. (2010). A domesticated piggyBac 987 transposase plays key roles in heterochromatin dynamics and DNA cleavage during 988 programmed DNA deletion in Tetrahymena thermophila. Molecular Biology of the 989 Cell, 21(10), 1753–1762. https://doi.org/10.1091/mbc.e09-12-1079 990 de Vanssay, A., Touzeau, A., Arnaiz, O., Frapporti, A., Phipps, J., & Duharcourt, S. 991 (2020). The Paramecium histone chaperone Spt16-1 is required for Pgm 992 endonuclease function in programmed genome rearrangements. PLoS Genetics, 993 16(7), e1008949. https://doi.org/10.1371/journal.pgen.1008949 994 Deutsch, E. W., Bandeira, N., Perez-Riverol, Y., Sharma, V., Carver, J. J., Mendoza, L., 995 Kundu, D. J., Wang, S., Bandla, C., Kamatchinathan, S., Hewapathirana, S., Pullman, 996 B. S., Wertz, J., Sun, Z., Kawano, S., Okuda, S., Watanabe, Y., MacLean, B., 997 MacCoss, M. J., … Vizcaíno, J. A. (2023). The ProteomeXchange consortium at 10 998 years: 2023 update. Nucleic Acids Research, 51(D1), D1539–D1548. 999 https://doi.org/10.1093/nar/gkac1040 1000 Driever, W., & Nüsslein-Volhard, C. (1988). A gradient of bicoid protein in Drosophila 1001 embryos. Cell, 54(1), 83–93. https://doi.org/10.1016/0092-8674(88)90182-1 1002 Eberharter, A., Vetter, I., Ferreira, R., & Becker, P. B. (2004). ACF1 improves the 1003 effectiveness of nucleosome mobilization by ISWI through PHD-histone contacts. The 1004 EMBO Journal, 23(20), 4029–4039. https://doi.org/10.1038/sj.emboj.7600382 1005 Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A. W., Green, T., Žídek, A., 1006 Bates, R., Blackwell, S., Yim, J., Ronneberger, O., Bodenstein, S., Zielinski, M., 1007 Bridgland, A., Potapenko, A., Cowie, A., Tunyasuvunakool, K., Jain, R., Clancy, E., … 1008 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 37 Hassabis, D. (2021). Protein complex prediction with AlphaFold-Multimer. BioRxiv. 1009 https://doi.org/10.1101/2021.10.04.463034 1010 Finn, R., Griffiths-Jones, S., & Bateman, A. (2003). Identifying protein domains with the 1011 Pfam database. Current Protocols in Bioinformatics, Chapter 2, Unit 2.5. 1012 https://doi.org/10.1002/0471250953.bi0205s01 1013 Fleck, K., Nitz, M., & Jeffers, V. (2021). “Reading” a new chapter in protozoan parasite 1014 transcriptional regulation. PLoS Pathogens, 17(12), e1010056. 1015 https://doi.org/10.1371/journal.ppat.1010056 1016 Forment, J. V., Kaidi, A., & Jackson, S. P. (2012). Chromothripsis and cancer: causes 1017 and consequences of chromosome shattering. Nature Reviews. Cancer, 12(10), 663–1018 670. https://doi.org/10.1038/nrc3352 1019 Frapporti, A., Miró Pina, C., Arnaiz, O., Holoch, D., Kawaguchi, T., Humbert, A., 1020 Eleftheriou, E., Lombard, B., Loew, D., Sperling, L., Guitot, K., Margueron, R., & 1021 Duharcourt, S. (2019). The Polycomb protein Ezl1 mediates H3K9 and H3K27 1022 methylation to repress transposable elements in Paramecium. Nature 1023 Communications, 10(1), 2710. https://doi.org/10.1038/s41467-019-10648-5 1024 Furrer, D. I., Swart, E. C., Kraft, M. F., Sandoval, P. Y., & Nowacki, M. (2017). Two Sets 1025 of Piwi Proteins Are Involved in Distinct sRNA Pathways Leading to Elimination of 1026 Germline-Specific DNA. Cell Reports, 20(2), 505–520. 1027 https://doi.org/10.1016/j.celrep.2017.06.050 1028 Gaurav, N., & Kutateladze, T. G. (2023). Non-histone binding functions of PHD fingers. 1029 Trends in Biochemical Sciences, 48(7), 610–617. 1030 https://doi.org/10.1016/j.tibs.2023.03.005 1031 Gozani, O., Karuman, P., Jones, D. R., Ivanov, D., Cha, J., Lugovskoy, A. A., Baird, C. 1032 L., Zhu, H., Field, S. J., Lessnick, S. L., Villasenor, J., Mehrotra, B., Chen, J., Rao, V. 1033 R., Brugge, J. S., Ferguson, C. G., Payrastre, B., Myszka, D. G., Cantley, L. C., … 1034 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 38 Yuan, J. (2003). The PHD finger of the chromatin-associated protein ING2 functions 1035 as a nuclear phosphoinositide receptor. Cell, 114(1), 99–111. 1036 https://doi.org/10.1016/s0092-8674(03)00480-x 1037 Grandchamp, S., & Beisson, J. (1981). Positional control of nuclear differentiation in 1038 paramecium. Developmental Biology, 81(2), 336–341. https://doi.org/10.1016/0012-1039 1606(81)90297-9 1040 Gruchota, J., Denby Wilkes, C., Arnaiz, O., Sperling, L., & Nowak, J. K. (2017). A 1041 meiosis-specific Spt5 homolog involved in non-coding transcription. Nucleic Acids 1042 Research, 45(8), 4722–4732. https://doi.org/10.1093/nar/gkw1318 1043 Guérin, F., Arnaiz, O., Boggetto, N., Denby Wilkes, C., Meyer, E., Sperling, L., & 1044 Duharcourt, S. (2017). Flow cytometry sorting of nuclei enables the first global 1045 characterization of Paramecium germline DNA and transposable elements. BMC 1046 Genomics, 18(1), 327. https://doi.org/10.1186/s12864-017-3713-7 1047 Hamilton, E. P., Kapusta, A., Huvos, P. E., Bidwell, S. L., Zafar, N., Tang, H., 1048 Hadjithomas, M., Krishnakumar, V., Badger, J. H., Caler, E. V., Russ, C., Zeng, Q., 1049 Fan, L., Levin, J. Z., Shea, T., Young, S. K., Hegarty, R., Daza, R., Gujja, S., … 1050 Coyne, R. S. (2016). Structure of the germline genome of Tetrahymena thermophila 1051 and relationship to the massively rearranged somatic genome. ELife, 5. 1052 https://doi.org/10.7554/eLife.19090 1053 Hoehener, C., Hug, I., & Nowacki, M. (2018). Dicer-like enzymes with sequence 1054 cleavage preferences. Cell, 173(1), 234-247.e7. 1055 https://doi.org/10.1016/j.cell.2018.02.029 1056 Huber, W., von Heydebreck, A., Sültmann, H., Poustka, A., & Vingron, M. (2002). 1057 Variance stabilization applied to microarray data calibration and to the quantification 1058 of differential expression. Bioinformatics, 18 Suppl 1, S96-104. 1059 https://doi.org/10.1093/bioinformatics/18.suppl_1.s96 1060 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 39 Ignarski, M., Singh, A., Swart, E. C., Arambasic, M., Sandoval, P. Y., & Nowacki, M. 1061 (2014). Paramecium tetraurelia chromatin assembly factor-1-like protein PtCAF-1 is 1062 involved in RNA-mediated control of DNA elimination. Nucleic Acids Research, 1063 42(19), 11952–11964. https://doi.org/10.1093/nar/gku874 1064 Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., 1065 Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., Bridgland, A., Meyer, C., 1066 Kohl, S. A. A., Ballard, A. J., Cowie, A., Romera-Paredes, B., Nikolov, S., Jain, R., 1067 Adler, J., … Hassabis, D. (2021). Highly accurate protein structure prediction with 1068 AlphaFold. Nature, 596(7873), 583–589. https://doi.org/10.1038/s41586-021-03819-2 1069 Kim, D., Paggi, J. M., Park, C., Bennett, C., & Salzberg, S. L. (2019). Graph-based 1070 genome alignment and genotyping with HISAT2 and HISAT-genotype. Nature 1071 Biotechnology, 37(8), 907–915. https://doi.org/10.1038/s41587-019-0201-4 1072 Kinkelin, K., Wozniak, G. G., Rothbart, S. B., Lidschreiber, M., Strahl, B. D., & Cramer, 1073 P. (2013). Structures of RNA polymerase II complexes with Bye1, a chromatin-1074 binding PHF3/DIDO homologue. Proceedings of the National Academy of Sciences of 1075 the United States of America, 110(38), 15277–15282. 1076 https://doi.org/10.1073/pnas.1311010110 1077 Klobutcher, L. A., & Herrick, G. (1995). Consensus inverted terminal repeat sequence of 1078 Paramecium IESs: resemblance to termini of Tc1-related and Euplotes Tec 1079 transposons. Nucleic Acids Research, 23(11), 2006–2013. 1080 https://doi.org/10.1093/nar/23.11.2006 1081 Kong, A. T., Leprevost, F. V., Avtonomov, D. M., Mellacheruvu, D., & Nesvizhskii, A. I. 1082 (2017). MSFragger: ultrafast and comprehensive peptide identification in mass 1083 spectrometry-based proteomics. Nature Methods, 14(5), 513–520. 1084 https://doi.org/10.1038/nmeth.4256 1085 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 40 Krueger, F. (2019). TrimGalore: A wrapper around Cutadapt and FastQC to consistently 1086 apply adapter and quality trimming to FastQ files, with extra functionality for RRBS 1087 data (0.6.10) [Computer software]. GitHub. 1088 Lefort-Tran, M., Aufderheide, K., Pouphile, M., Rossignol, M., & Beisson, J. (1981). 1089 Control of exocytotic processes: cytological and physiological studies of trichocyst 1090 mutants in Paramecium tetraurelia. The Journal of Cell Biology, 88(2), 301–311. 1091 https://doi.org/10.1083/jcb.88.2.301 1092 Leinonen, R., Akhtar, R., Birney, E., Bower, L., Cerdeno-Tárraga, A., Cheng, Y., 1093 Cleland, I., Faruque, N., Goodgame, N., Gibson, R., Hoad, G., Jang, M., Pakseresht, 1094 N., Plaister, S., Radhakrishnan, R., Reddy, K., Sobhany, S., Ten Hoopen, P., 1095 Vaughan, R., … Cochrane, G. (2011). The european nucleotide archive. Nucleic 1096 Acids Research, 39(Database issue), D28-31. https://doi.org/10.1093/nar/gkq967 1097 Lepère, G., Nowacki, M., Serrano, V., Gout, J.-F., Guglielmi, G., Duharcourt, S., & 1098 Meyer, E. (2009). Silencing-associated and meiosis-specific small RNA pathways in 1099 Paramecium tetraurelia. Nucleic Acids Research, 37(3), 903–915. 1100 https://doi.org/10.1093/nar/gkn1018 1101 Lhuillier-Akakpo, M., Frapporti, A., Denby Wilkes, C., Matelot, M., Vervoort, M., 1102 Sperling, L., & Duharcourt, S. (2014). Local effect of enhancer of zeste-like reveals 1103 cooperation of epigenetic and cis-acting determinants for zygotic genome 1104 rearrangements. PLoS Genetics, 10(9), e1004665. 1105 https://doi.org/10.1371/journal.pgen.1004665 1106 Liu, L., Qin, S., Zhang, J., Ji, P., Shi, Y., & Wu, J. (2012). Solution structure of an 1107 atypical PHD finger in BRPF2 and its interaction with DNA. Journal of Structural 1108 Biology, 180(1), 165–173. https://doi.org/10.1016/j.jsb.2012.06.014 1109 Liu, Y., Taverna, S. D., Muratore, T. L., Shabanowitz, J., Hunt, D. F., & Allis, C. D. 1110 (2007). RNAi-dependent H3K27 methylation is required for heterochromatin formation 1111 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 41 and DNA elimination in Tetrahymena. Genes & Development, 21(12), 1530–1545. 1112 https://doi.org/10.1101/gad.1544207 1113 Li, Haitao, Ilin, S., Wang, W., Duncan, E. M., Wysocka, J., Allis, C. D., & Patel, D. J. 1114 (2006). Molecular basis for site-specific read-out of histone H3K4me3 by the BPTF 1115 PHD finger of NURF. Nature, 442(7098), 91–95. https://doi.org/10.1038/nature04802 1116 Li, Heng, & Durbin, R. (2009). Fast and accurate short read alignment with Burrows-1117 Wheeler transform. Bioinformatics, 25(14), 1754–1760. 1118 https://doi.org/10.1093/bioinformatics/btp324 1119 Li, Heng, Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., 1120 Abecasis, G., Durbin, R., & 1000 Genome Project Data Processing Subgroup. (2009). 1121 The Sequence Alignment/Map format and SAMtools. Bioinformatics, 25(16), 2078–1122 2079. https://doi.org/10.1093/bioinformatics/btp352 1123 Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and 1124 dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. 1125 https://doi.org/10.1186/s13059-014-0550-8 1126 Lu, X., Meng, X., Morris, C. A., & Keating, M. T. (1998). A novel human gene, WSTF, is 1127 deleted in Williams syndrome. Genomics, 54(2), 241–249. 1128 https://doi.org/10.1006/geno.1998.5578 1129 Maliszewska-Olejniczak, K., Gruchota, J., Gromadka, R., Denby Wilkes, C., Arnaiz, O., 1130 Mathy, N., Duharcourt, S., Bétermier, M., & Nowak, J. K. (2015). TFIIS-Dependent 1131 Non-coding Transcription Regulates Developmental Genome Rearrangements. PLoS 1132 Genetics, 11(7), e1005383. https://doi.org/10.1371/journal.pgen.1005383 1133 Mani, R.-S., & Chinnaiyan, A. M. (2010). Triggers for genomic rearrangements: insights 1134 into genomic, cellular and environmental influences. Nature Reviews. Genetics, 1135 11(12), 819–829. https://doi.org/10.1038/nrg2883 1136 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 42 Marmignon, A., Bischerour, J., Silve, A., Fojcik, C., Dubois, E., Arnaiz, O., Kapusta, A., 1137 Malinsky, S., & Bétermier, M. (2014). Ku-mediated coupling of DNA cleavage and 1138 repair during programmed genome rearrangements in the ciliate Paramecium 1139 tetraurelia. PLoS Genetics, 10(8), e1004552. 1140 https://doi.org/10.1371/journal.pgen.1004552 1141 Martin, M. (2011). Cutadapt removes adapter sequences from high-throughput 1142 sequencing reads. EMBnet.Journal, 17(1), 10. https://doi.org/10.14806/ej.17.1.200 1143 Miller, J., McLachlan, A. D., & Klug, A. (1985). Repetitive zinc-binding domains in the 1144 protein transcription factor IIIA from Xenopus oocytes. The EMBO Journal, 4(6), 1145 1609–1614. https://doi.org/10.1002/j.1460-2075.1985.tb03825.x 1146 Mirdita, M., Schütze, K., Moriwaki, Y., Heo, L., Ovchinnikov, S., & Steinegger, M. 1147 (2022). ColabFold: making protein folding accessible to all. Nature Methods, 19(6), 1148 679–682. https://doi.org/10.1038/s41592-022-01488-1 1149 Miró-Pina, C., Charmant, O., Kawaguchi, T., Holoch, D., Michaud, A., Cohen, I., 1150 Humbert, A., Jaszczyszyn, Y., Chevreux, G., Del Maestro, L., Ait-Si-Ali, S., Arnaiz, O., 1151 Margueron, R., & Duharcourt, S. (2022). Paramecium Polycomb repressive complex 1152 2 physically interacts with the small RNA-binding PIWI protein to repress 1153 transposable elements. Developmental Cell, 57(8), 1037-1052.e8. 1154 https://doi.org/10.1016/j.devcel.2022.03.014 1155 Mochizuki, K., Fine, N. A., Fujisawa, T., & Gorovsky, M. A. (2002). Analysis of a piwi-1156 related gene implicates small RNAs in genome rearrangement in tetrahymena. Cell, 1157 110(6), 689–699. https://doi.org/10.1016/s0092-8674(02)00909-1 1158 Morisato, D., & Anderson, K. V. (1995). Signaling pathways that establish the dorsal-1159 ventral pattern of the Drosophila embryo. Annual Review of Genetics, 29, 371–399. 1160 https://doi.org/10.1146/annurev.ge.29.120195.002103 1161 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 43 Nowacki, M., Zagorski-Ostoja, W., & Meyer, E. (2005). Nowa1p and Nowa2p: novel 1162 putative RNA binding proteins involved in trans-nuclear crosstalk in Paramecium 1163 tetraurelia. Current Biology, 15(18), 1616–1628. 1164 https://doi.org/10.1016/j.cub.2005.07.033 1165 OpenAI. (2023). ChatGPT [Large language model] (GPT-3.5) [Computer software]. 1166 https://chat.openai.com/chat. 1167 Oppikofer, M., Sagolla, M., Haley, B., Zhang, H.-M., Kummerfeld, S. K., Sudhamsu, J., 1168 Flynn, E. M., Bai, T., Zhang, J., Ciferri, C., & Cochran, A. G. (2017). Non-canonical 1169 reader modules of BAZ1A promote recovery from DNA damage. Nature 1170 Communications, 8(1), 862. https://doi.org/10.1038/s41467-017-00866-0 1171 Pascual, J., Martinez-Yamout, M., Dyson, H. J., & Wright, P. E. (2000). Structure of the 1172 PHD zinc finger from human Williams-Beuren syndrome transcription factor. Journal 1173 of Molecular Biology, 304(5), 723–729. https://doi.org/10.1006/jmbi.2000.4308 1174 Paysan-Lafosse, T., Blum, M., Chuguransky, S., Grego, T., Pinto, B. L., Salazar, G. A., 1175 Bileschi, M. L., Bork, P., Bridge, A., Colwell, L., Gough, J., Haft, D. H., Letunić, I., 1176 Marchler-Bauer, A., Mi, H., Natale, D. A., Orengo, C. A., Pandurangan, A. P., Rivoire, 1177 C., … Bateman, A. (2023). InterPro in 2022. Nucleic Acids Research, 51(D1), D418–1178 D427. https://doi.org/10.1093/nar/gkac993 1179 Perez-Riverol, Y., Bai, J., Bandla, C., García-Seisdedos, D., Hewapathirana, S., 1180 Kamatchinathan, S., Kundu, D. J., Prakash, A., Frericks-Zipper, A., Eisenacher, M., 1181 Walzer, M., Wang, S., Brazma, A., & Vizcaíno, J. A. (2022). The PRIDE database 1182 resources in 2022: a hub for mass spectrometry-based proteomics evidences. 1183 Nucleic Acids Research, 50(D1), D543–D552. https://doi.org/10.1093/nar/gkab1038 1184 Pinskaya, M., Ghavi-Helm, Y., Mariotte-Labarre, S., Morillon, A., Soutourina, J., & 1185 Werner, M. (2014). PHD and TFIIS-Like domains of the Bye1 transcription factor 1186 determine its multivalent genomic distribution. Plos One, 9(7), e102464. 1187 https://doi.org/10.1371/journal.pone.0102464 1188 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 44 Quan, J., & Tian, J. (2011). Circular polymerase extension cloning for high-throughput 1189 cloning of complex and combinatorial DNA libraries. Nature Protocols, 6(2), 242–251. 1190 https://doi.org/10.1038/nprot.2010.181 1191 Ritchie, M. E., Phipson, B., Wu, D., Hu, Y., Law, C. W., Shi, W., & Smyth, G. K. (2015). 1192 limma powers differential expression analyses for RNA-sequencing and microarray 1193 studies. Nucleic Acids Research, 43(7), e47. https://doi.org/10.1093/nar/gkv007 1194 Roberts, A., & Pachter, L. (2013). Streaming fragment assignment for real-time analysis 1195 of sequencing experiments. Nature Methods, 10(1), 71–73. 1196 https://doi.org/10.1038/nmeth.2251 1197 Rooney, S., Chaudhuri, J., & Alt, F. W. (2004). The role of the non-homologous end-1198 joining pathway in lymphocyte development. Immunological Reviews, 200, 115–131. 1199 https://doi.org/10.1111/j.0105-2896.2004.00165.x 1200 Roth, S., Stein, D., & Nüsslein-Volhard, C. (1989). A gradient of nuclear localization of 1201 the dorsal protein determines dorsoventral pattern in the Drosophila embryo. Cell, 1202 59(6), 1189–1202. https://doi.org/10.1016/0092-8674(89)90774-5 1203 Rzeszutek, I., Swart, E. C., Pabian-Jewuła, S., Russo, A., & Nowacki, M. (2022). Early 1204 developmental, meiosis-specific proteins - Spo11, Msh4-1, and Msh5 - Affect 1205 subsequent genome reorganization in Paramecium tetraurelia. Biochimica et 1206 Biophysica Acta. Molecular Cell Research, 1869(6), 119239. 1207 https://doi.org/10.1016/j.bbamcr.2022.119239 1208 Sanchez, R., & Zhou, M.-M. (2011). The PHD finger: a versatile epigenome reader. 1209 Trends in Biochemical Sciences, 36(7), 364–372. 1210 https://doi.org/10.1016/j.tibs.2011.03.005 1211 Sandoval, P. Y., Swart, E. C., Arambasic, M., & Nowacki, M. (2014). Functional 1212 diversification of Dicer-like proteins and small RNAs required for genome sculpting. 1213 Developmental Cell, 28(2), 174–188. https://doi.org/10.1016/j.devcel.2013.12.010 1214 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 45 Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., 1215 Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., Tinevez, J.-Y., White, D. J., 1216 Hartenstein, V., Eliceiri, K., Tomancak, P., & Cardona, A. (2012). Fiji: an open-source 1217 platform for biological-image analysis. Nature Methods, 9(7), 676–682. 1218 https://doi.org/10.1038/nmeth.2019 1219 Schindler, U., Beckmann, H., & Cashmore, A. R. (1993). HAT3.1, a novel Arabidopsis 1220 homeodomain protein containing a conserved cysteine-rich region. The Plant Journal, 1221 4(1), 137–150. https://doi.org/10.1046/j.1365-313x.1993.04010137.x 1222 Sellis, D., Guérin, F., Arnaiz, O., Pett, W., Lerat, E., Boggetto, N., Krenek, S., 1223 Berendonk, T., Couloux, A., Aury, J.-M., Labadie, K., Malinsky, S., Bhullar, S., Meyer, 1224 E., Sperling, L., Duret, L., & Duharcourt, S. (2021). Massive colonization of protein-1225 coding exons by selfish genetic elements in Paramecium germline genomes. PLoS 1226 Biology, 19(7), e3001309. https://doi.org/10.1371/journal.pbio.3001309 1227 Sievers, F., Wilm, A., Dineen, D., Gibson, T. J., Karplus, K., Li, W., Lopez, R., 1228 McWilliam, H., Remmert, M., Söding, J., Thompson, J. D., & Higgins, D. G. (2011). 1229 Fast, scalable generation of high-quality protein multiple sequence alignments using 1230 Clustal Omega. Molecular Systems Biology, 7, 539. 1231 https://doi.org/10.1038/msb.2011.75 1232 Singh, A., Häußermann, L., Emmerich, C., Nischwitz, E., Seah, B. K., Butter, F. K. B., 1233 Nowacki, M., & Swart, E. C. (2023). ISWI1 complex proteins facilitate developmental 1234 genome editing in Paramecium. BioRxiv. https://doi.org/10.1101/2023.08.09.552620 1235 Singh, A., Maurer-Alcalá, X. X., Solberg, T., Häußermann, L., Gisler, S., Ignarski, M., 1236 Swart, E. C., & Nowacki, M. (2022). Chromatin remodeling is required for sRNA-1237 guided DNA elimination in Paramecium. The EMBO Journal, 41(22), e111839. 1238 https://doi.org/10.15252/embj.2022111839 1239 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 46 Skene, P. J., Henikoff, J. G., & Henikoff, S. (2018). Targeted in situ genome-wide 1240 profiling with high efficiency for low cell numbers. Nature Protocols, 13(5), 1006–1241 1019. https://doi.org/10.1038/nprot.2018.015 1242 Spirov, A., Fahmy, K., Schneider, M., Frei, E., Noll, M., & Baumgartner, S. (2009). 1243 Formation of the bicoid morphogen gradient: an mRNA gradient dictates the protein 1244 gradient. Development, 136(4), 605–614. https://doi.org/10.1242/dev.031195 1245 Sturm, Á., Saskoi, É., Tibor, K., Weinhardt, N., & Vellai, T. (2018). Highly efficient RNAi 1246 and Cas9-based auto-cloning systems for C. elegans research. Nucleic Acids 1247 Research, 46(17), e105. https://doi.org/10.1093/nar/gky516 1248 Swart, E. C., Wilkes, C. D., Sandoval, P. Y., Arambasic, M., Sperling, L., & Nowacki, M. 1249 (2014). Genome-wide analysis of genetic and epigenetic control of programmed DNA 1250 deletion. Nucleic Acids Research, 42(14), 8970–8983. 1251 https://doi.org/10.1093/nar/gku619 1252 Taverna, S. D., Coyne, R. S., & Allis, C. D. (2002). Methylation of histone h3 at lysine 9 1253 targets programmed DNA elimination in tetrahymena. Cell, 110(6), 701–711. 1254 https://doi.org/10.1016/s0092-8674(02)00941-8 1255 Villanueva, R. A. M., & Chen, Z. J. (2019). ggplot2: Elegant Graphics for Data Analysis 1256 (2nd ed.). Measurement: Interdisciplinary Research and Perspectives, 17(3), 160–1257 167. https://doi.org/10.1080/15366367.2019.1565254 1258 Wang, C., Lv, L., Solberg, T., Wen, Z., Zhang, H., & Gao, F. (2023). Conservation of the 1259 ancestral function of GTSF1 in transposon silencing in the unicellular eukaryote 1260 Paramecium tetraurelia. BioRxiv. https://doi.org/10.1101/2023.10.06.561219 1261 Wang, C., Solberg, T., Maurer-Alcalá, X. X., Swart, E. C., Gao, F., & Nowacki, M. 1262 (2022). A small RNA-guided PRC2 complex eliminates DNA as an extreme form of 1263 transposon silencing. Cell Reports, 40(8), 111263. 1264 https://doi.org/10.1016/j.celrep.2022.111263 1265 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 47 Wysocka, J., Swigut, T., Xiao, H., Milne, T. A., Kwon, S. Y., Landry, J., Kauer, M., 1266 Tackett, A. J., Chait, B. T., Badenhorst, P., Wu, C., & Allis, C. D. (2006). A PHD finger 1267 of NURF couples histone H3 lysine 4 trimethylation with chromatin remodelling. 1268 Nature, 442(7098), 86–90. https://doi.org/10.1038/nature04815 1269 Zangarelli, C., Arnaiz, O., Bourge, M., Gorrichon, K., Jaszczyszyn, Y., Mathy, N., 1270 Escoriza, L., Bétermier, M., & Régnier, V. (2022). Developmental timing of 1271 programmed DNA elimination in Paramecium tetraurelia recapitulates germline 1272 transposon evolutionary dynamics. Genome Research, 32(11–12), 2028–2042. 1273 https://doi.org/10.1101/gr.277027.122 1274 Zeng, L., Zhang, Q., Li, S., Plotnikov, A. N., Walsh, M. J., & Zhou, M.-M. (2010). 1275 Mechanism and regulation of acetylated histone binding by the tandem PHD finger of 1276 DPF3b. Nature, 466(7303), 258–262. https://doi.org/10.1038/nature09139 1277 Zhang, J., Yan, G., Tian, M., Ma, Y., Xiong, J., & Miao, W. (2018). A DP-like 1278 transcription factor protein interacts with E2fl1 to regulate meiosis in Tetrahymena 1279 thermophila. Cell Cycle, 17(5), 634–642. 1280 https://doi.org/10.1080/15384101.2018.1431595 1281 Zimmermann, L., Stephens, A., Nam, S.-Z., Rau, D., Kübler, J., Lozajic, M., Gabler, F., 1282 Söding, J., Lupas, A. N., & Alva, V. (2018). A Completely Reimplemented MPI 1283 Bioinformatics Toolkit with a New HHpred Server at its Core. Journal of Molecular 1284 Biology, 430(15), 2237–2243. https://doi.org/10.1016/j.jmb.2017.12.007 1285 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 1 Figure 1: Features of the PHD finger proteins DevPF1 and DevPF2 (A) mRNA expression profiles for DevPF1, DevPF2 and PGM during various developmental stages: VEG (vegetative growth), MEI (micronuclear meiosis and macronuclear fragmentation), FRG (~50% of the population with fragmented maternal MACs), DEV1 (significant proportion with visible anlagen), DEV2/3 (majority with visible 0 1000 2000 3000 4000 VEG MEI FRG DEV1 DEV2/3 DEV4 mRNA [A.U.] DevPF1 DevPF2 PGM DevPF2 PHD PHD-zinc-finger like domain DevPF2: DevPF1: cyan: DevPF1 green: DevPF2 yellow: WSTF magenta: C4HC3 grey: zinc ion A B D C E HPHD PHD-zinc-finger like domain pLDDT 0-50 90-100> > DevPF1 PHD PHD-zinc-finger like domain DevPF2 DevPF1 Jade-3.human p300.human WSTF.human NURF301.drosophila Jade-1.human MLL.human C4HC3 motif F G DevPF2DevPF1 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 2 anlagen), DEV4 (majority with visible anlagen). Expression data retrieved from ParameciumDB (Arnaiz et al., 2017). (B) Schematic representation of predicted domain architecture for DevPF1 and DevPF2. (C) Multiple sequence alignment (Clustal Omega) of DevPF1 and DevPF2 amino acid sequence with PHD domains of published human and Drosophila PHD finger proteins. (D) to (F): Predicted protein structure (AlphaFold2) for DevPF1 and DevPF2, colored by domain (PHD: orange; PHD-zinc-finger-like domain: green) in (D) and (E), and by prediction confidence (pLDDT: predicted local distance difference test) in (F) and (G). (H) Structure predictions of DevPF1 and DevPF2 PHD domain superimposed with NMR structure of WSTF PHD domain (PDB accession number 1F62). .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 3 GFP DAPI + GFP starved vegetativemeiosis IImeiosis II DevPF1-GFP S-phase new MAC development early new MAC development late new MAC development GFP DAPI + GFP DevPF2-GFPA B C MIC MAC gametic nuclei skein zygote post-zygotic nuclei frag- ments new MIC new MAC 1 2 3 4 5 6 7 1 2 3,4 5 6 7 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 4 Figure 2: Subcellular localization of DevPF-GFP proteins DevPF1-GFP (A) and DevPF2-GFP (B) localization at various developmental stages. DNA (stained with DAPI) in magenta. GFP signal in yellow. No image of DevPF2-GFP during S-phase was acquired. Green arrow: MIC. Cyan arrow: new MAC. Maximum intensity projections of multiple z-planes. Scale bar = 10 µm. (C) Schematic overview of nuclear morphology during sexual development, with corresponding cell stages in the images indicated by numbers. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 5 individual z-planes maximum intensity projection meiosis II meiosis II meiosis IImeiosis I meiosis II post-zygote maximum intensity projectionsindividual z-planes A B DAPIoverlayoverlay anti-α-tubulinGFP MIC with DevPF1-GFP MIC without DevPF1-GFP DevPF1-GFP DAPI anti-α-tubulin .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 6 Figure 3: Selective DevPF1-GFP localization in Paramecium MICs (A) Overlay of DAPI (DNA stain; pink) and GFP (yellow) signal in two DevPF1-GFP injected Paramecium cells during meiotic stages. Maximum intensity projections (left) and individual z-planes of the same stack (right). (B) DevPF1-GFP localization with visualization of nuclear spindle. DAPI (pink), GFP (yellow) and anti-α-tubulin staining (cyan). Maximum intensity projections (top) for DAPI and overlay (DAPI, GFP and anti- α-tubulin). Individual z-planes of the same stacks (bottom) for anti-α-tubulin, GFP and overlay. (A) and (B): Red arrows: MICs with DevPF1-GFP localization; White arrows: MICs without DevPF1-GFP localization. Scale bar = 10 µm. .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 7 D PGM DevPF2 (1) DevPF2 (2) DevPF2 (3) DevPF2 (4) DevPF2 (5) DevPF2 (6) DevPF1 (1) DevPF1 (2) DevPF1 (3) DevPF1 (4) PGM DevPF2 (1) DevPF2 (2) DevPF2 (3) DevPF2 (4) DevPF2 (5) DevPF2 (6) DevPF1 (1) DevPF1 (2) DevPF1 (3) DevPF1 (4) PGM DevPF2 (1) DevPF2 (2) DevPF2 (3) DevPF2 (4) DevPF2 (5) DevPF2 (6) DevPF1 (2) DevPF1 (1) DevPF1 (3) DevPF1 (4) ND7 PGM DevPF1 DevPF2 KD normalized mean fluorescence intensity GFP signal in new MACs DevPF1-GFP DevPF2-GFP A C onset early late very late 0 500 1000 1500 2000 st-veg early late very late timepoint mRNA [normalized counts] batch A B C KD ND7 PS17 AS17 DevPF2 ND7 DevPF1 DevPF20 500 1000 1500 2000 st-veg early late very late timepoint mRNA [normalized counts] batch A B C KD ND7 PS17 AS17 DevPF2 DevPF1 DevPF2 B 0 500 1000 1500 2000 onset early late very late timepoint mRNA [normalized counts] batch A B C KD ND7 PS17 AS17 DevPF2 DevPF2 0 1000 2000 onset early late very late timepoint mRNA [normalized counts] batch A B C KD ND7 PS17 AS17 DevPF1 DevPF1 cells 0% 50% 100% onset early late very late number of cells vegetative skein fragmented new MACs time point .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 8 Figure 4: Co-silencing effects observed in DevPF knockdowns (A) Nucleotide identity across DevPF1 (bottom) and DevPF2 (top) genes. Screenshot of pairwise sequence alignment in Geneious prime software. Silencing region (violet), exon (green), intron (white), perfect identity (gray) and mismatch/gap (black). Scale in base pairs at the top. (B) mRNA expression levels of DevPF1 (top) and DevPF2 (bottom) upon KDs (ND7 (control), DevPF1 and DevPF2) at different developmental time points (onset, early, late and very late). Lines represent the mean of all replicates for a given KD and time point. The cell stage composition of each time point averaged over all KDs is shown at the top (individual compositions in Fig. S5), along with schematic representations of the considered cell stages. (C) Protein expression upon KD: fluorescence intensities of DevPF1-GFP (top) and DevPF2-GFP (bottom). Red line: median. Whiskers: 1.5 times the interquartile range from the lower or upper quartile. Dots: data points outside the whiskers. Sample size = 10. (D) IES retention score (IRS) correlations between DevPF1- and DevPF2-KD replicates. Diagonal: IRS distributions of individual KDs. Below diagonal: correlation graphs of pairwise comparisons. Above diagonal: corresponding Spearman correlation coefficients. Red lines: ordinary least- squares (OLS) regression, orange lines: LOWESS, and gray lines: orthogonal distance regression (ODR). .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 9 263949 30904513560135cells = n = CA DB E PGM TFIIS4 DCL2/3/5 SPT5m PTCAF1 ISWI1 DevPF2 (4) DevPF1 (4) DevPF1 (2) PGM TFIIS4 DCL 2/3/5 SPT5m PTCAF1 ISWI1 DevPF2 (4) DevPF1 (4) DevPF1 (2) PGMTFIIS4DCL2/3/5SPT5mPTCAF1ISWI1DevPF2 (4) DevPF1 (4) DevPF1 (2) MT IES IES+ 460 bp IES- 265 bp IES+ IES- IES 5 IES+ 501 bp IES- 299 bp ND7-KD PGM-KD AS17-1-KD AS17-2-KD PS17-2-KD PS17-1-KD MT IES DevPF1-b-KD DevPF1-a-KD DevPF2-b-KD DevPF2-a-KD PGM-KD ND7-KD IES+ (501 bp) IES- (299 bp) IES+ (460 bp) IES- (265 bp) 51G4404 0% 20% 40% 60% 80% 100% ND7 PGM DevPF2-aDevPF2-bDevPF1-aDevPF1-b number of cells survival sickness death .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 10 Figure 5: Effects of DevPF knockdowns on genome-wide IES retention (A) Viability of new progeny after KDs (ND7 (negative control), PGM (positive control), DevPF1 and DevPF2) during sexual development. For DevPF1 and DevPF2, two silencing regions were targeted (a and b, see Fig. 4A). The numbers of experiments (n) and cells counted (cells) are indicated at the top. Survival: normal division. Sickness: reduced growth. Death: 3 or less cells after three days. (B) IES retention PCRs for two IESs on genomic DNA isolated from KD cells. (C) and (D): IES retention score (IRS) histograms for DevPF1 (C) and DevPF2 (D) KD replicates, indicated in parentheses. (E) IRS correlation between KDs. Diagonal: IRS distributions of individual KDs. Below diagonal: correlation graphs of pairwise comparisons. Above diagonal: corresponding Spearman correlation coefficients. Red lines: ordinary least-squares (OLS) regression, orange lines: LOWESS, and gray lines: orthogonal distance regression (ODR). .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 11 Aonsetearlylatevery late DevPF2-KD (1)ND7-KD (1) fraction of total reads small RNA length [nt] DevPF1-KD (2) siRNA (23 nt) scnRNA (25 nt) siRNA (23 nt) scnRNA (25 nt) iesRNA (~26-30 nt) 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% 100% 0% MAC IES silencing vector small RNAs matching to: cell stages: siRNA (23 nt) siRNA (23 nt) scnRNA (25 nt) siRNA (23 nt) siRNA (23 nt) scnRNA (25 nt) vegetative skein fragmented new MACs DevPF1-KD GFPoverlay GFPoverlay no KD S-phase meiosis I meiosis II fragments new MACsB S-Phasemeiosis Imeiosis IIfragmentsnew MACs C Ptiwi09-GFP .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 12 Figure 6: Changes of small RNA populations upon DevPF knockdowns (A) Small RNA populations (10-40 nt) at developmental time points (onset, early, late and very late) in different KDs (ND7 (control), DevPF1 and DevPF2), mapping to silencing plasmid backbone (vector), MAC or IES sequences. Individual cell stage compositions are indicated by the bar to the right of each diagram, along with schematic representations of the cell stages considered. (B) Ptiwi09-GFP localization at different developmental stages in the context of no (top) and DevPF1 KD (bottom). DAPI (pink) and GFP (yellow). Individual z-planes for GFP and overlay (DAPI and GFP). Green arrows: MICs. Cyan arrows: new MAC. Scale bar = 10 µm. (C) Schematic representation of cell stages in (B). .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 23, 2024. ; https://doi.org/10.1101/2024.01.23.576875doi: bioRxiv preprint 13 DevPF1vs ND7DevPF2vs ND7DevPF2vs DevPF1 C D onsetearlylatevery late cells: 100%20%40%60%80% vegetativeskeinfragmentednew MACs A upregulateddownregulatedother 0 500 1000 1500 2000 st-veg early late very late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 DevPF2 ND7DevPF1DevPF2 scnRNArelated iesRNArelated KD: DevPF2-KDDevPF1-KDND7-KDtimepoint +-+onsetscnRNAs +-+early +-+late +-+very late ---onsetiesRNAs ---early --+late --+very late B 0 10000 20000 30000 40000 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 PTIWI01 0 100 200 300 400 500 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 DCL2 0 5000 10000 15000 20000 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 PTIWI10 0 5000 10000 15000 20000 25000 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 NOWA1 0 300 600 900 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 DCL5 0 250 500 750 1000 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 SPT5m 0 500 1000 1500 onsetearlylatevery late timepoint mRNA[normalized counts] batch A B C KD ND7 PS17 AS17 TFIIS4 0%20%40%60%80%100%onsetearlylatevery late number of cells new MACsfragmentedskeinvegetative DCL2 PTIWI01 NOWA1 SPT5m DCL5 PTIWI10 TFIIS4 14 Figure 7: Differential gene expression in DevPF knockdowns (A) Cell stage composition of each time point averaged over all KDs (individual compositions in Fig. S5), along with schematic representations of the considered cell stages. (B) Presence or absence of scnRNAs and iesRNAs in different KDs (ND7, DevPF1 and DevPF2) and time points (onset, early, late, very late). (C) Differentially expressed genes in DevPF1- (top) or DevPF2- (middle) compared to ND7-KD or DevPF1- compared to DevPF2-KD (bottom) at different developmental time points (onset, early, late and very late). Thresholds for up-/downregulation: adjusted p-value 2. The number of up-/downregulated genes is indicated in each diagram. For all comparisons, 35777 transcripts were analyzed, except for: DevPF1-ND7 onset (33696), DevPF2-ND7 early (35083), and DevPF2-DevPF1 onset (34389). (D) Gene expression levels of selected genes upon KDs (ND7 (control), DevPF1 and DevPF2) at different developmental time points (onset, early, late and very late). The lines represent the mean of all replicates in a given KD and time point. 15 Table 1: IES retention scores of IESs at PTIWI10/11 genes The genes PTIWI10 and PTIWI11 contain IESs in their coding and/or flanking regions, which were proposed to impair their transcription when retained. The IRS values for the three relevant IESs (IDs with prefix IESPGM.PTET51.1) are provided for each KD. Rows are color-coded according to the KDs as shown in the mRNA read count diagrams (i. e. Figs 7D, S7B). KD Replicate PTIWI11 PTIWI10 coding region flanking region coding region IESPGM.PTET51.1.62.345420 IESPGM.PTET51.1.24.407807 IESPGM.PTET51.1.24.408279 ND7 3 0.00 0.00 0.00 4 0.00 0.00 0.00 5 0.00 0.00 0.00 DevPF1 1 0.09 0.29 0.11 2 0.08 0.15 0.06 3 0.04 0.03 0.06 4 0.02 0.01 0.01 DevPF2 1 0.10 0.07 0.01 2 0.02 0.24 0.15 3 0.00 0.00 0.00 4 0.00 0.00 0.01 5 0.03 0.00 0.00 6 0.01 0.00 0.00