The structured mRNA element 45ABC mediates auto- and cross-regulation of RBP45 genes via alternative splicing

Alternative splicing, RNA structure, RNA motif, RNA-binding protein, RBP45, 53 Arabidopsis thaliana 54 55 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 4

56 During their life cycle, plants need to continuously adapt to environmental fluctuations, 57 involving precise genetic control to fine-tune cellular and developmental processes. A critical 58 layer of this regulation is the splicing of precursor messenger RNAs (pre -mRNAs), a co - or 59 post-transcriptional process that can occur in an either constitutive or alternative manner. 60 While during constitutive splicing the same exon sequences are incorporated in the same 61 order in the mature mRNA, multiple mRNA isoforms are generated by alternative splicing (AS) 62 due to differential 5' and 3' splice site selection. For Arabidopsis thaliana, an average of 4.4 63 transcript isoforms per gene was reported to be generated by AS and the usage of alternative 64 transcript start and end sites (Zhang et al., 2022). Numerous studies have provided evidence 65 that plants exploit the potential of AS in fine-tuning gene expression to modulate development, 66 responses to stress, and adaptation to environmental changes (Alhabsi et al., 2025; Reddy et 67 al., 2013; Staiger and Brown, 2013). AS can expand both transcriptomic and proteomic 68 diversity. A substantial proportion of the AS variants is likely not translated into proteins but 69 instead may influence gene expression by altering other processes such as mRNA stability 70 and localization. One widespread mechanism represents the coupling of AS and nonsense -71 mediated decay (NMD), which causes turnover of non-productive isoforms and thereby allows 72 quantitative gene control (Drechsel et al., 2013; Kalyna et al., 2012). 73 Eukaryotic pre-mRNA splicing is regulated by the interplay of cis-regulatory elements, specific 74 sequences and structures within the target RNAs, and trans-acting factors, including RNA and 75 protein molecules. The splicing regulatory proteins include two major classes: the 76 serine/arginine-rich (SR) proteins and heterogeneous nuclear ribonucleoprotein (hnRNP) 77 proteins. Originally, SR and hnRNP proteins, respectively, have been mainly considered as 78 activators and repressors of splicing sites, however, it is now clear that their action is highly 79 context-dependent and shaped by parameters such as the binding location and the presence 80 of other regulatory factors and elements (Fu and Ares, 2014; Reddy et al., 2013). Among the 81 diverse group of hnRNP proteins from A. thaliana (Wachter et al., 2012), the glycine-rich RNA-82 binding proteins GRP7 and GRP8 (Streitner et al., 2012) and polypyrimidine tract -binding 83 proteins (PTBs; Rühl et al., 2012) have been intensively studied and linked to AS. Both GRPs 84 and PTBs are part of auto- and cross-regulatory circuits, triggering formation of NMD-targeted 85 splicing variants from their own and the homologs’ pre -mRNAs when the levels of the 86 respective proteins are elevated. This type of balancing mechanism is common among 87 splicing regulators and other RBPs in plants and other organisms (Müller-McNicoll et al., 2019; 88 Reddy et al., 2013; Wachter et al., 2012). 89 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 5 The RNA-binding protein (RBP) 45 family also belongs to the hnRNP proteins and comprises 90 in A. thaliana the four members RBP45A, RBP45B, RBP45C, and RBP45D, each of which 91 contain three RNA recognition motifs (RRMs ; Lorković et al. , 2000) . So far, their specific 92 functions and in particular their possible role in AS regulation remain largely unresolved. 93 Previous work has shown that RBP45D interacts with pre -mRNA-processing factor 39a 94 (PRP39a), a component of the U1 small nuclear ribonucleoprotein (snRNP), suggesting a 95 function in 5′ splice site selection (Chang et al. , 2022) . Moreover, RBP45D was found to 96 promote flowering (Wang et al., 2022). In vitro interaction studies with RBP45B confirmed its 97 RNA binding (Peal et al., 2011), and RBP45 proteins were found to associate with proteins 98 involved in various aspects of RNA processing, including U1 snRNP components (Mangilet et 99 al., 2024) , cap binding protein 20 (CBP20) and the poly (A)-binding protein PAB8 100 (Muthuramalingam et al., 2016), and the RNA helicase up-frameshift 1 (UPF1; Chicois et al., 101 2018; Sulkowska et al., 2020). Recently, the presence of an evolutionary conserved structured 102 RNA (strucRNA), 45ABC, was reported for the pre -mRNAs of RBP45A, RBP45B, and 103 RBP45C, but not RBP45D (Sack et al., 2025). 104 In addition to trans-acting factors, RNA sequence and structural motifs play a crucial role in 105 facilitating exon and intron definition (Buratti and Baralle, 2004; Georgakopoulos -Soares et 106 al., 2022; Ullah et al., 2018). For instance, local RNA structures can alter the accessibility of 107 cis-regulatory elements, thereby either hindering or enhancing spliceosomal assembly 108 (Shepard and Hertel, 2008) . The ability of RNA to form dynamic structures gives rise to a 109 structural complexity that can be responsive to various external factors such as temperature 110 (Chung et al., 2020), subcellular localization (Sun et al., 2019), and interactions with RBPs 111 (Gosai et al., 2015) or metabolite ligands (Steinert et al., 2017). RNA secondary structures 112 have been shown to influence diverse steps of gene expression such as AS (Wachter, 2014), 113 RNA localization (Fernández-Moya et al. , 2021) , and RNA stability (Fischer et al. , 2020; 114 Goodarzi et al., 2012). While AS regulation through strucRNAs is well documented in animals 115 (Raker et al., 2009; Shepard and Hertel, 2008) , few such instances have been reported in 116 plants. One example is the thiamine pyrophosphate (TPP) -binding riboswitch, a conserved 117 RNA structural element found across all three domains of life, including fungi and plants, but 118 being absent from animals (Bocobza et al., 2007; Sudarsan et al., 2003; Wachter et al., 2007). 119 Upon binding of its ligand TPP, the riboswitch’s conformation changes, demasking an 120 alternative splice site and thereby altering the splicing pattern and downregulating the thiamin 121 biosynthesis gene THIC in a negative feedback loop. Another example for an RNA structure 122 that is conserved in plants and involved in AS control is the plant 5S ribosomal RNA structural 123 mimic (P5SM; Hammond et al., 2009). It interacts with a ribosomal protein to modulate AS of 124 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 6 a transcription factor pre -mRNA, adjusting 5S rRNA production to the level of its binding 125 partner L5 (Hammond et al., 2009). 126 In this study, we have characterized the AS-regulatory function of a structured mRNA element 127 that was recently found in a bioinformatic survey of the non-coding regions of plant pre-mRNAs 128 (Sack et al., 2025). The strucRNA 45ABC consists of two hairpins, which are highly conserved 129 among homologs of RBP45A, RBP45B, and RBP45C from monocots and dicots. It 130 consistently overlaps with a cassette exon (CE), the skipping of which generates a splicing 131 variant encoding the full -length RBP45 protein, while its inclusion introduces a premature 132 termination codon and triggers degradation via NMD (Sack et al., 2025). Using splicing 133 reporters based on the three At-RBP45 genes, we revealed negative feedback regulation of 134 their pre-mRNAs via AS. Disruption of the motif's first hairpin resulted in a strong increase in 135 the CE variant, while compensatory mutations restored the authentic splicing output. Further 136 mutational studies identified within this stem I a G-rich sequence that is required for efficient 137 CE inclusion and is proposed to serve as an RBP45 binding site. Moreover, we have 138 established in A. thaliana individual overexpression lines as well as single and higher order 139 rbp45 knockout mutants to examine these genes’ regulatory crosstalk as well as functions in 140 global AS control and plant growth. Transcriptome-wide analyses provided evidence that out 141 of these three RBP45 genes from A. thaliana only RBP45B plays a broader role in AS control 142 beyond the cross -regulatory circuit. Furthermore, all of the tested loss -of-function mutants 143 exhibited significantly shorter primary roots compared to wild -type plants. In summary, our 144 study has unravel led novel functions of RBP45 genes in A. thaliana and identified the 145 strucRNA 45ABC as a mediator of an AS -based gene regulatory circuit that achieves 146 specificity due to a combination of sequence and structural features. 147 148

149 The 45ABC motif overlaps with alternative splicing sites of RBP45 genes 150 Bioinformatic analysis of non-coding regions in available plant genome data revealed several 151 conserved strucRNA candidates that overlap with alternative splice sites (Sack et al., 2025). 152 One promising candidate from this study had previously been identified as a conserved 153 intronic element in RBP genes from angiosperms (Burgess and Freeling, 2014), however, its 154 functional role remained unknown. Since it is associated with RBP45A (AT5G54900), RBP45B 155 (AT1G11650), and RBP45C (AT4G27000) from A. thaliana, it was named 45ABC motif. In the 156 data sets analysed, the motif has 448 matches in genes from 112 mono- and dicotyledonous 157 species that are annotated as RBP45 homologs or are related to RBP45 (Supplemental Data 158 Set 1). Interestingly, for all RBP45 homologs, 45ABC is positioned in a long intron and 159 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 7 encompasses a CE (Figure 1a). Moreover, the two predicted hairpin loops of the motif are in 160 the vicinity of or overlap with the alternative splice sites (Figure 1b), with the alternative 3 ’ 161 splice site being located ~10 nt upstream of stem I and the alternative 5 ’ splice site and the 162 U1 snRNP binding motif (consensus sequence: AG/GUAAGU according to Sheth et al. (2006)) 163 being embedded in stem II. Analysis of the RBP45 transcripts revealed the generation of two 164 major AS variants: Skipping of the CE leads to an isoform (cd) that is predicted to encode the 165 full-length protein. CE inclusion introduces a premature termination codon (PTC) and a long 166 3’ untranslated region (UTR) making this isoform ( nc) probably unproductive. Accordingly, 167 NMD targeting of the CE inclusion variants from these three RBP45 genes from A. thaliana 168 was experimentally confirmed (Drechsel et al., 2013; Sack et al., 2025). RT-PCR analysis of 169 RBP45 splicing in different tissues revealed an overall similar splicing output, with the coding 170 variant being the predominant amplification product in all samples (Figure 1c, Figure S1). 171 Arabidopsis RBP45 homologues regulate AS of their own pre-mRNAs 172 The association of the strucRNA 45ABC with the alternative splice sites in the corresponding 173 pre-mRNAs indicated a functional relationship. To explore a potential regulatory circuit, we 174 designed splicing reporters based on the genomic DNA of RBP45A, RBP45B, and RBP45C 175 spanning the region from the 5’ UTR to the beginning of the second exon downstream of the 176 CE (Figure 2a). These reporters included an in-frame fusion of the CDS of green fluorescent 177 protein (GFP), resulting in a fluorescent fusion protein upon reporter splicing to the cd variant 178 (Figure 2a). Infiltration assays in Nicotiana benthamiana leaves were used to examine reporter 179 splicing, in the presence of co -expressed CDS constructs of RBP45A, RBP45B, or RBP45C 180 compared to a control construct containing the CDS of Luciferase (LUC). In the LUC samples, 181 all reporters showed splicing into both transcript variants, with cd being the predominant one 182 (Figure 2b). Co -expression of any of the RBP45 CDS constructs shifted splicing patterns 183 towards the respective nc isoform (Figure 2b, c). This shift in AS towards the nc variants 184 correlated with the reduced fluorescence levels of all reporters to less than 50% in the 185 presence of the CDS constructs of RBP45A, RBP45B, or RBP45C (Figure 2d). Taken 186 together, our observations revealed an AS -mediated feedback mechanism in which RBP45 187 genes negatively auto- and cross-regulate each other. 188 To assess this regulatory circuit in its native context, we modulated RBP45A, RBP45B, and 189 RBP45C expression in stably transformed A. thaliana plants. This was accomplished either 190 through their overexpression (OE, 35S:RBP45CDS) or by CRISPR/Cas9-mediated knockout 191 of one, two, or all three corresponding RBP45 genes (Figure S2). Analysis of total RBP45 192 transcript levels confirmed the overaccumulation of RBP45 mRNAs in the corresponding OE 193 lines (Figure 3a). Notably, the OE-B lines showed the strongest extent of overexpression with 194 ~17 times higher RBP45 amounts compared to WT. The two independent OE-A lines showed 195 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 8 an only ~5 -fold overexpression, while for RBP45C only one OE line with a less than 2 -fold 196 increase in RBP45C could be obtained. The other RBP45C-OE line probably showed silencing 197 as the level of the coding RBP45C transcript from the endogenous locus was decreased and 198 total RBP45C levels were WT-like (Figure 3a, b). In general, an increase in the total transcript 199 levels of one of the three RBP45 genes led to a downregulation of the other two. This effect 200 was particularly pronounced for the paralogs RBP45A and RBP45C. Furthermore, coding 201 levels of the endogenous genes decreased consistently for all OE lines (Figure 3b). 202 Interestingly, an upregulation of the non-coding transcript variants was only observed in case 203 of the OE-B lines, and this effect was consistent among all three genes (Figure 3c), indicating 204 a unique role of RBP45B in alternative splicing control. In contrast, overexpression of RBP45A 205 or RBP45C resulted in downregulation of the coding and non -coding AS variants from the 206 homologous endogenous genes . This decrease of the non -coding variants in the RBP45A- 207 and RBP45C-OE lines may be a consequence of their reduced levels of RBP45B.cd. 208 Accordingly, the non -coding variants of all three genes may predominantly arise through 209 RBP45B activity in A. thaliana . Moreover, the negative auto - and cross -regulation of the 210 RBP45 genes might involve AS-independent mechanisms. 211 To complement these findings and further dissect the roles of RBP45A, RBP45B, and 212 RBP45C in AS regulation, we also analysed transcript levels and splicing patterns of the 213 RBP45 genes in the three mutant lines generated via CRISPR/Cas9 mutagenesis: rbp45b, 214 rbp45bc, and rbp45abc (Figure S2). Given that the mutations give rise to PTCs at early 215 positions within the CDS, the corresponding mutants are considered to be knockout lines. 216 Total transcript levels of the respective mutated genes were significantly reduced compared 217 to those in WT seedlings, and a slight overaccumulation of the non-targeted homolog(s) was 218 seen (Figure 4a). In the rbp45b mutant, a decrease of non-coding isoforms was observed for 219 all three RBP45 genes, whereas the coding transcript variants of RBP45A and RBP45C 220 showed an opposite trend (Figure 4b, c). This is in line with our conclusion based on the data 221 from the OE lines that RBP45B is a major AS regulator for all three RBP45 genes. 222 Interestingly, knocking out RBP45C in addition to RBP45B in the double mutant rbp45bc 223 restored levels of non -coding RBP45B transcript and caused an additional decrease and 224 increase, respectively , of the non -coding and coding RBP45A variant (Figure 4b, c). 225 Accordingly, the expression of RBP45C and RBP45A is tightly linked. Furthermore, the low 226 levels of non-coding variants for all three RBP45 genes in the triple mutant suggested that the 227 cross-regulatory AS circuit might be restricted to this set of genes. 228 The structure of 45ABC is decisive for AS control of RBP45 genes 229 Our experiments revealed that the three RBP45 genes are subject to negative auto- and cross-230 regulation via AS. Given the association between the cassette exon and the 45ABC motif, we 231 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 9 next addressed the possible role of this strucRNA in controlling the AS output. To this end, we 232 created the following mutations in the 45ABC motif in the context of the RBP45C splicing 233 reporter (Figure 5a): The disruptive mutants DM1 and DM3 in stem I and stem II, respectively, 234 each involving the exchange of four nucleotides to eliminate base pairing and destabilize the 235 corresponding pairing elements. We ensured that the alternative 5' splice site within stem II 236 was preserved at its authentic location. To further examine the contribution of the stem I 237 structure, we designed an additional mutant DM2, containing three further nucleotide 238 substitutions and resulting in an almost complete loss of the pairing potential within this part 239 of the element. To ensure that any changes observed for the DMs are due to a change in 240 structure and not the altered sequence, compensatory mutations CM1, CM2, and CM3 were 241 created that are expected to restore the original structure in the context of the new sequence 242 (Figure 5a). Accordingly, if the splicing outcome is primarily dependent on structural features 243 of the mutated regions of the 45ABC motif, the splicing pattern of the DM and CM constructs, 244 respectively, should be disrupted and restored in comparison to the WT reporter. 245 We initially tested these splicing reporters upon transient expression in N. benthamiana leaves 246 and observed that the disruptive mutants DM1 and DM2 in stem I resulted in almost exclusive 247 splicing to the nc isoform, even in the absence of RBP45 CDS co-expression (Figure S3a). In 248 case of DM3 in stem II, only a partial splicing shift towards the nc variant was seen, and this 249 effect was further enhanced upon RBP45 co-expression. All compensatory mutations reverted 250 this splicing pattern by causing a relative increase of the cd variant, however, in comparison 251 to the WT construct, CM1 and CM2 did not fully compensate while CM3 overcompensated. In 252 case of the stem I mutants, the response to RBP45 co-expression was almost completely lost 253 and restored, respectively, for the DM and CM constructs (Figure S3b). As the transient 254 analysis in N. benthamiana leaves represents a heterologous system with artificially high 255 expression, we next tested splicing of the various reporter constructs in the endogenous 256 context. In stably transformed A. thaliana lines, DM1 led to a significant shift towards the nc 257 isoform (Figure 5b). In contrast to the transient system, in the native context CM1 was able to 258 restore the splicing pattern to a WT-like AS ratio. These findings further support that the stem I 259 structure contributes to proper splicing regulation. The more extensive mutation DM2 in stem I 260 also resulted in a shift towards the nc isoform, although this effect was quantitatively less 261 pronounced than in case of DM1. Nevertheless, CM2 also fully restored the WT splicing 262 pattern, being in line with the critical role of the stem I structure. In contrast to our observations 263 for the stem I mutants, the stem II -targeting DM3 showed only a slight increase in the nc/cd 264 ratio. Still, the CM3 mutation reverted this minor splicing shift back to WT levels, indicating 265 that stem II may also contribute to splice site selection (Figure 5b). Collectively, these results 266 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 10 demonstrated that the structure of 45ABC, particularly of stem I, plays a critical role in 267 modulating the AS outcome of the corresponding RBP45 genes. 268 We hypothesized that the base -pairing within stem I may affect the splicing outcome by 269 controlling the accessibility of an RBP45 binding site. A prior study revealed that recombinantly 270 expressed RBP45 protein from Nicotiana plumbaginifolia exhibits in vitro binding to poly(U) 271 and poly(G) ribohomopolymers (Lorković et al., 2000). Interestingly, a consecutive stretch of 272 guanines is present in stem I of 45ABC in the case of RBP45C. For RBP45A and RBP45B, 273 one of the guanines is replaced by adenine. Analysis of all available instances of this motif 274 identified in this region the purine-rich sequence RGRG (Sack et al., 2025). To experimentally 275 test the relevance of this G-stretch within stem I, additional mutations were introduced into this 276 strucRNA in the context of the RBP45C splicing reporter. In the binding mutant BM2, the 277 consecutive guanines were substituted with adenines to assess the role of binding sequence 278 specificity. Moreover, three C nucleobases on the complementary strand were mutated to Us 279 to preserve the pairing potential. As a control, we generated the additional mutant BM1, in 280 which the complementary poly(U) stretch was included while the G -stretch and the pairing 281 potential were preserved (Figure 6a). Upon transient co-expression with the LUC control in N. 282 benthamiana leaves, BM1 displayed a WT -like splicing pattern, whereas BM2 exhibited a 283 near-complete loss of splicing to the CE -containing variant (Figure 6b, c). This observation 284 provided evidence that the G -stretch is critical for splicing to the nc isoform. Upon co -285 expression with the CDS constructs of any of the three RBP45 homologs, BM2 still showed 286 an AS response, however, the splicing shift was markedly less pronounced compared to the 287 WT and the BM1 mutant. This responsiveness of the BM mutants can also be seen at the 288 protein level of the reporters, with reduced fluorescence upon RBP45 co-expression (Figure 289 S3c). In line with the increased splicing of BM2 to the coding variant, this mutant resulted in 290 higher reporter fluorescence than the WT construct (Figure S3d). The observation that splicing 291 of the BM2 mutant reporter is still responsive to RBP45 co-expression might be explained by 292 massive accumulation of the corresponding RBP45 proteins and rather unspecific RNA 293 binding in the N. benthamiana system. Moreover, the introduced U-stretch might also enable 294 RBP45 recruitment given the above -mentioned binding preference observed for a homolog 295 from N. plumbaginifolia under in vitro conditions (Lorković et al. 2000). Together with the 296

for the structural mutant reporters (Figure 5 and S3), we propose a combined impact 297 of structure and sequence on AS control. RBP45 binding to the G -rich sequence in stem I is 298 assumed to trigger CE inclusion, e.g. due to structural changes and/or recruitment of further 299 regulators. Disruption of the corresponding pairing element might facilitate RBP45 binding and 300 thereby shift the AS ratio towards more CE inclusion, as observed in DM1 and DM2. The 301 higher number of G residues in the 3’ part of stem I for DM1 compared to DM2 might result in 302 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 11 more efficient RBP45 binding, explaining the stronger shift towards the nc variant despite less 303 extensive stem weakening in case of DM1. 304 Misexpression of RBP45 can cause reduced primary root length and altered flowering 305 time 306 To gain insights into the biological relevance of the RBP45 genes, we compared the 307 development of the RBP45 misexpression lines and WT plants. A previous study reported 308 reduced primary root lengths for rbp45d mutants, while no such phenotype was seen for 309 rbp45a and rbp45c T-DNA insertion lines (Huang et al., 2022; Muthuramalingam et al., 2016). 310 Measuring root lengths of our mutant lines revealed that the loss of RBP45B also resulted in 311 shorter roots, and this phenotype was even more pronounced in the corresponding double 312 and triple mutants (Figure 7a). A diminished root length was also observed for the OE-C line, 313 whereas the two other overexpression lines showed WT -like growth. To test whether the 314 altered root phenotype of the corresponding lines might be a consequence of diminished seed 315 size, we measured the seed surface area of the mutant lines and different batches of WT 316 seeds. The variation observed among the mutants and different batches of WT seeds were in 317 a similar range (Figure S4). Accordingly, the altered root length can at least not solely be 318 explained by varying seed filling and resource availability. Another developmental parameter 319 affected in some of the mutants was the flowering time. For the knockout mutants, only rbp45b 320 showed a slightly delayed onset of flowering (Fig ure 7b). Later flowering had also been 321 described for rbp45d mutants (Huang et al., 2022; Muthuramalingam et al., 2016), further 322 indicating that RBP45 genes can share common functions in development. Interestingly, the 323 effect of RBP45 overexpression on flowering time differed between the three homologs. While 324 OE-A lines showed only a minor delay in flowering onset compared to the WT, OE-B and OE-325 C lines, respectively, flowered significantly earlier and later. These findings suggested that 326 RBP45 genes might have specific and redundant regulation targets and that proper expression 327 control of RBP45 genes is critical for normal plant development. 328 RBP45B is the major regulator of AS among the three RBP45 paralogs 329 Since we observed that the three RBP45 genes are subject to auto - and cross-regulation 330 involving splicing control, we next investigated to which extent the RBP45 genes can influence 331 gene expression and AS at the whole transcriptome level. Therefore, RNA sequencing 332 analyses were performed using A. thaliana seedling samples of the higher order mutants 333 rbp45bc and rbp45abc as well as the constitutive OE lines for RBP45A, RBP45B, and 334 RBP45C. Based on the altered root phenotype, we also performed transcriptome -wide 335 analyses using root tissue of 7 -day-old WT seedlings and those mutants displaying shorter 336 primary roots, namely rbp45bc, rbp45abc, and OE-C. Data were analysed via the 3D RNA -337 seq pipeline ( Guo et al., 2019), resulting for both datasets in the detection of over 16,000 338 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 12 genes that were expressed in at least one sample type (Figures S5 and S6). The gene 339 expression analysis revealed rather few changes, with 36 and 52 differentially expressed (DE) 340 genes, respectively, in the whole seedlings and root samples (Figure 8a). Only eight of the DE 341 genes were shared between the two datasets, indicating the involvement of tissue -specific 342 regulation. Most of the DE genes were downregulated in the mutants, both for whole seedlings 343 and roots (Figure 8b and Supplemental Data Set 2) and with the largest number of DE genes 344 being detected in rbp45abc roots. This observation is in line with the finding that the triple 345 mutant showed the most pronounced reduction in root length. Among the DE genes was a 346 copia-like retrotransposon (AT5G35935), whose function is uncharacterized yet and which 347 was consistently downregulated in all RBP45 misexpression lines relative to WT samples in 348 both seedlings and roots (Figure S7). Given that the alteration in the transcript level of this 349 retrotransposon showed no correlation with RBP45 expression level, we concluded that this 350 change is probably not directly linked to RBP45 function. 351 The separate DE analyses for seedling and root samples indicated that only few genes are 352 regulated in an RBP45-dependent manner. We next extended our analysis to the line -wise 353 comparison of tissue-specific expression patterns, which might differ between the mutants and 354 the WT and could have been missed in the previous comparisons within one tissue type due 355 to small changes. As we used in our study different number of replicates for the seedling and 356 root samples, we first performed pair-wise comparisons for the WT and three types of mutants 357 (Figure S8a-d). The comparison of the replicates and genotypes gave consistent results, with 358 an overlap of 4,277 (68.4%) DE genes that show differential expression in seedlings and roots 359 of both WT and the two knockout mutant backgrounds rbp45bc and rbp45abc (Figure S8e). 360 Additionally including the OE-C line reduced the number of common DE genes only slightly 361 (Figure S8f; 4,1 00 genes, 63. 2%), representing a robust signature of tissue -specific gene 362 expression that showed in a GO term analysis among others an enrichment of photosynthesis-363 related genes (Supplemental Data Set 3), as expected in the comparison of phototrophic 364 seedlings and heterotrophic roots. The DE genes not shared between the WT and mutants 365 might point at processes that are disturbed upon RBP45 misexpression and therefore could 366 contribute to the altered root growth. GO analysis for the corresponding gene sets showed no 367 significant enrichment. Further research will be needed to validate these possible differences 368 in tissue-specific gene expression and to examine whether those are linked to the altered root 369 growth upon RBP45 misexpression. 370 Next, we determined the numbers of differentially alternatively spliced (DAS) genes by 371 comparing the mutant lines to WT, resulting in the seedling and root datasets, respectively, in 372 51 and 45 affected genes (Figure 8c and Supplemental Data Set 2). Considering the AS 373 changes upon RBP45 overexpression, RBP45B resulted in most changes among the three 374 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 13 corresponding mutants with 1 4 DAS genes in whole seedlings (Figure 8d). No DAS genes 375 were detected for the RBP45A overexpression seedlings, whereas the RB45A gene was the 376 only DAS target in the OE-C line, further supporting the cross-regulatory mechanism (Figure 377 8d). For seedlings of the knockout mutants rbp45bc and rbp45abc, respectively, 8 and 40 DAS 378 genes were detected, with 6 DAS genes shared between the two mutants (Figure 8d) . A 379 similar extent of AS changes was found in the root dataset, with 13 and 39 DAS genes, 380 respectively, in the double and triple mutant, and 7 overlapping genes (Figure S9a). Upon 381 RBP45C overexpression, two DAS genes were found in the roots, of which one each 382 overlapped with the dataset for the rbp45bc and the rbp45abc mutant. To further characterize 383 the alternative splicing changes, we assigned differential transcript usage (DTU) transcripts to 384 the identified DAS genes. In cases where expression levels of two transcripts from the same 385 gene were significantly and reciprocally altered in at least one type of misexpression line 386 compared to the WT, AS event types and their genomic locations were determined 387 (Supplemental Data Set 2) . This annotation was performed using the A. thaliana TAIR 10 388 genome and AtRTD2 -QUASI reference transcriptome. Considering the AS event positions 389 relative to the open reading frame, approximately half of the significantly altered events were 390 located in the 5′ and 3′ UTR (Figure 8e and S9). This enrichment compared to the reference 391 data set suggests that RBP45-mediated AS to a major extent affects non-coding regions. 392 To validate the transcriptome -wide findings, RT -PCR analyses were performed for seven 393 candidate DAS events (Figure 9 and S10 ). Out of those, three events were significantly 394 changed in both RNA -seq datasets ( AVT6, HYH, and VAB2), three ( ABC1K8, DEP1, and 395 ALAD1) were unique to whole seedlings and one ( XBAT35) was only detected in the root 396 dataset. According to the RNA-seq data, only DEP1 and AVT6 exhibited significant reciprocal 397 splicing changes in comparison of the two knockout lines and the OE-B lines. In our validation 398 experiments, all seven AS events were tested across all genotypes, and the significant 399 changes from the RNA -seq were independently confirmed (Figures 9 and S10). In case of 400 rbp45bc and rbp45abc, all candidate events were significantly and consistently altered both in 401 the two mutants and the two tissue types. With respect to the quantitative change, the effect 402 was slightly but consistently more pronounced in the triple mutant, suggesting an additive 403 effect of RBP45 paralog loss. Interestingly, an in general opposite AS shift was seen for OE-404 B, while overexpression of the other two RBP45 paralogs did not have such an effect (Figure 405 9 and S10). This further supports a prominent function of RBP45B in AS regulation among the 406 RBP45 family, and is consistent with its central role in auto- and cross-regulation. 407 408 409 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 14

410 Crosstalk of RBP45 homologs via AS regulation 411 Here, we characterized the role of the strucRNA 45ABC in AS regulation of RBP45A, RBP45B, 412 and RBP45C from A. thaliana. Our results revealed negative auto - and cross-regulation of 413 these three genes via AS, balancing the ratio of protein-coding cd variants to non-productive 414 nc variants that are targeted by NMD (Sack et al. , 2025) . This regulatory mechanism is 415 common among RBP genes and enables rapid and dynamic fine -tuning of protein levels 416 (Müller-McNicoll et al. , 2019; Reddy et al. , 2013; Wachter et al. , 2012) . In line with 417 autoregulation, constitutive overexpression of any RBP45 member led to reduced levels of the 418 cd isoform for the corresponding endogenous gene (Figure 3). With respect to cross -419 regulation, RBP45B seems to play a dominant role among the three homologs. Its 420 overexpression resulted in an increase of the nc variants for all three RBP45 genes, whereas 421 in lines overexpressing RBP45A or RBP45C both the cd and nc isoform levels from all three 422 endogenous RBP45 genes were diminished (Figure 3b, c). Accordingly, formation of the non-423 coding variants from these genes is to a major extent caused by RBP45B, which in turn is 424 suppressed in the RBP45A and RBP45C overexpression lines. This regulatory hierarchy is 425 further supported by the analysis of the knockout mutants (Figure 4). Knocking out only 426 RBP45B reduced nc isoforms for all three genes, whereas total levels of RBP45A and 427 RBP45C transcripts were elevated. In the rbp45bc double mutant, however, RBP45B nc levels 428 were restored to a WT -like level. In parallel, the AS pattern of RBP45A was altered in the 429 double mutant, with an increase of cd and decrease of nc transcripts, pointing t owards a 430 release of negative cross -regulation by RBP45C. As a consequence, RBP45A was de -431 repressed in the double mutant, and triggered enhanced production of nc transcripts from the 432 RBP45B gene (Figure S11). Moreover, the rbp45abc triple mutant had strongly diminished 433 amounts of nc isoforms for all three genes, suggesting that this regulatory circuit is probably 434 confined to the three RBP45 genes. 435 Similar auto- and cross -regulatory circuits have been identified in plants before, including 436 studies of the hnRNP genes GRP7 and GRP8 (Schöning et al., 2007; Schöning et al., 2008) 437 and the three PTB genes (Rühl et al., 2012; Stauffer et al., 2010) from A. thaliana. AS-based 438 cross-regulation was observed for GRP7/GRP8 and PTB1/PTB2, whereas the more distantly 439 related PTB3 had no effect on the splicing outcome of its two paralogs. What is then the 440 molecular mechanism underlying the hierarchy of the more complex regulatory circuit shown 441 in this study for the RBP45 genes? And why does it involve a conserved structured mRNA 442 motif as opposed to a simpler pre-mRNA architecture containing single or several binding sites 443 for the corresponding RBP as has been demonstrated in other instances? 444 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 15 Evolutionary analyses indicate that RBP45B is the most ancient one of the three paralogs, 445 while RBP45C and RBP45A emerged more recently and are more closely related (Lorković et 446 al., 2000; Park et al. , 2006) . This relationship is also reflected by the different domain 447 organization of the proteins. All three RBP45 proteins contain three RRMs; while RBP45A and 448 RBP45C have in addition a glutamine -rich N -terminus, RBP45B is enriched in glutamine 449 residues near the C-terminus and exhibits a proline-rich N-terminus. These differences might 450 be responsible for or contribute to the distinct roles of the three RBP45 genes in AS regulation. 451 Moreover, varying total and spatial expression patterns might play a role, and we observed in 452 WT plants that the transcript levels for RBP45B were highest among the three homologs 453 (Supplemental Data Set 2). A major role of RBP45B in AS regulation is not only supported by 454 its function in the auto- and cross-regulatory circuit, but also our transcriptome-wide analyses. 455 RBP45B overexpression caused the highest number of significant AS changes (Fig ure 8) as 456 well as reciprocal AS shifts compared to the double and triple knockout mutants (Fig ure 9). 457 Given that even for the RBP45B overexpression line and the triple mutant rbp45abc relatively 458 few global AS changes were detected, redundancy with other RBPs and/or additional 459 functions of these RBP45 genes besides AS regulation can be assumed. Moreover, o ur 460 observation of specificity among the RBP45 genes indicates unequal genetic redundance, 461 which has been previously discussed to be a more common evolutionary process of functional 462 gene diversification in A. thaliana (Briggs et al., 2006). 463 Critical and interconnected roles of 45ABC sequence and structure in balancing RBP45 464 expression 465 The exclusive presence of the strucRNA 45ABC in the three RBP45 genes and the associated 466 cross-regulatory network suggested that their expression is coordinated in an RNA structure-467 dependent manner. RNA folds can affect RBP interactions by various means, e.g. through 468 exposing or occluding binding sites, while RBP binding in turn can remodel the RNA structure. 469 Recent transcriptome-wide structure-profiling studies in plants further highlighted the critical 470 interplay between RNA sequence and structural features. For example, Gosai et al. (2015) 471 observed that RBP binding sites in nuclear mRNAs are generally less structured, and Liu et 472 al. (2021) reported that the 5’ splice site and branch point need to be single stranded for 473 efficient splicing. In plants, few examples exist where the impact of RNA structure and its 474 dynamical behavio ur on AS or other steps of gene expression has been functionally 475 characterized. One such case is the TPP riboswitch mechanism that functions via occlusion 476 of the alternative 5’ splice site by the interaction with an aptamer region in the absence of 477 ligand binding (Wachter et al., 2007). AS regulation via changes in splice site availability was 478 also reported for the heat shock factor HsfA2 gene from tomato, involving temperature -479 dependent structural changes at a 3’ splice site (Broft et al., 2022). An alternative mechanism 480 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 16 has been proposed in AS control of the TFIIIA gene via the strucRNA P5SM (Hammond et al., 481 2009), where L5 binding may prevent binding of a splicing regulatory protein independent of 482 an RNA structural change. 483 Using a mutational approach, w e showed in this work that opening the structure of 45ABC 484 resulted in increased usage of the alternative splice sites (Figure 5). The most pronounced 485 shift towards the nc isoform including the CE was observed upon disruption of stem I. 486 Compensatory mutations that restored base-pairing also rescued the splicing pattern to a WT-487 like AS ratio, underscoring the structural requirement for proper splicing regulation by this 488 element. The important role of stem I in AS regulation combined with the localization of the 489 alternative 5′ splice site and the respective U1 snRNP-binding motif (Sheth et al., 2006) within 490 stem II suggest that the assembly of the two hairpins is required for proper splice site selection. 491 Sequence and structure conservation of 45ABC may therefore be explained by its several 492 roles; as a binding site for RBP45 and U1 components, its structural constraints, and potential 493 crosstalk between the elements. Notably, all three RBP45 proteins have been shown to 494 interact with the U1 snRNP component U1-C (Huang et al., 2022). Their human ortholog TIA-495 1 is described to bind downstream of the 5’ splice site and to recruit U1 snRNP through a 496 direct interaction with U1 -C, thereby influencing splice site selection (Förch et al., 2002). In 497 the case of the RBP45 proteins, their binding to stem I of 45ABC might facilitate U1 recruitment 498 to the alternative 5’ splice site embedded in stem II and thereby promote its usage. 499 Accordingly, this strucRNA may act as a sensor for RBP45 protein levels that allows balancing 500 RBP45 expression in an AS-dependent manner (Figure 10). Closely related proteins, namely 501 RBP45D and RBP47 (Chang et al., 2022), could potentially also bind 45ABC and affect the 502 splicing outcome. RBP45D has been reported to be involved in alternative splice site selection 503 via interaction with U1 components (Chang et al., 2022), whereas the RBP47 proteins have 504 been described as regulators of stress granule dynamics and therefore may primarily act in 505 other aspects of RNA metabolism (Kosmacz et al., 2019). Functional redundancy between 506 RBP45 and possibly also RBP47 proteins in U1 recruitment to 5’ splice sites could also explain 507 why even in the triple mutant rbp45abc only relatively few AS changes were detected in our 508 transcriptome-wide analysis. Besides being part of the spliceosome, U1 snRNP also performs 509 splicing-independent functions. In a process termed telescripting , U1 binding blocks 510 recognition of cryptic polyadenylation signals and thereby prevents premature cleavage and 511 polyadenylation (Berg et al., 2012; Kaida et al., 2010). This additional U1 snRNP function has 512 recently also be demonstrated for A. thaliana (Mangilet et al., 2024) and future experiments 513 should address a possible involvement of RBP45 proteins. 514 We also examined if the conserved purine -rich motif (RGRG) within stem I might serve as a 515 binding site for the three RBP45 proteins and is important for the regulation. In line with this 516 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 17 hypothesis, transient reporter assays revealed that mutating the five consecutive guanines to 517 adenines caused an almost complete loss of the nc isoform under control conditions (Figure 518 6). Upon co-expression of any RBP45 CDS, a splicing response was still detectable, but it was 519 markedly attenuated compared to the WT construct. This residual responsiveness may be 520 explained by RBP45 binding to the introduced U-stretch, consistent with previous findings that 521 recombinantly expressed RBP45 from N. plumbaginifolia (Lorković et al., 2000) and RBP45D 522 from A. thaliana (Huang et al., 2022) interacted with U -rich RNAs in vitro. Alternatively, or 523 additionally, it might reflect non -specific interactions caused by high levels of RBP45 524 expression in the transient expression system. With regard to the proposed binding preference 525 of the RBP45 proteins, the U -rich loop of stem I (Sack et al., 2025) may also contribute to 526 RBP45 binding. Furthermore, it is well known that RBPs differ not only in their sequence 527 preferences but also with respect to the structural requirements at the binding site and its 528 context (Dominguez et al. , 2018; Foley et al. , 2017; Gosai et al. , 2015) . Accordingly, 529 conservation of the stem I structure of 45ABC might be a consequence of the RBP45 proteins’ 530 binding preferences. Taken together, both sequence and structural features of 45ABC 531 contribute to this RBP45-responsive regulatory mechanism, providing also an explanation for 532 the relatively low level of covariation of this motif (Sack et al., 2025). 533 Our findings raise the question whether related mechanisms of strucRNA -dependent AS are 534 more common. Interestingly, Sack et al. (2025) also identified conserved strucRNAs within the 535 pre-mRNAs of GRP7, GRP8, and PTB2. The GRP7&8 strucRNA consists of a hairpin 536 including an alternative 5’ splice site in the predicted stem (Sack et al., 2025). Usage of this 537 splice site leads to NMD-sensitive isoforms as part of negative feedback regulation (Schöning 538 et al., 2008), and both GRPs have been shown to bind downstream of this motif (Leder et al., 539 2014; Staiger et al., 2003). Accordingly, the corresponding strucRNA may suppress usage of 540 this splice site, while GRP binding could promote it (Sack et al., 2025). Similarly, the PTB2 541 motif is a single hairpin that includes the 5’ splice site and a polypyrimidine stretch that are 542 used for AS -mediated negative auto- and cross-regulation of PTB2 expression (Burgardt et 543 al., 2024; Sack et al. , 2025; Stauffer et al. , 2010) . So far experimental evidence for the 544 functional relevance of these novel strucRNAs is lacking. However, their discovery suggests 545 that such RNA folds might be more common in AS regulation. Reasons for their sequence and 546 structural conservation can include the presence of binding sites for splicing factors and 547 regulators, specific requirements for structuredness, and regulatory interactions between 548 individual elements as proposed for 45ABC. Shorter motifs such as single hairpins are more 549 difficult to detect by bioinformatics approaches searching for conserved structural elements 550 and thus might be more common than reflected by our current knowledge. 551 552 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 18 Physiological roles and functions of RBP45 proteins 553 Our transcriptome-wide analyses of rbp45 knockout and overexpression lines revealed only 554 limited changes in global gene expression and AS. Interestingly, most observed AS changes 555 occurred in UTRs (Figure 8 and S9) . This enrichment could point at additional functions of 556 RBP45 association with its target RNAs that occur downstream of AS control, such as the 557 established role of UTR-binding proteins in processes such as mRNA translation, localization, 558 and stability (Hardy and Balcerowicz, 2024) . Consistent with such an idea, RBP45 proteins 559 localize to both the nucleus and cytoplasm (Huang et al., 2022; Lorković et al., 2000) and 560 interact with mRNA decay components such as UPF1 (Chicois et al., 2018; Sulkowska et al., 561 2020). Protein-protein interaction studies further support a role of RBP45 proteins in regulating 562 the mRNA fate, e.g. RBP45B was shown to interact with the cap -binding protein CBP20 and 563 the poly(A) -binding protein PAB8 (Muthuramalingam et al. , 2016) . Furthermore, RBP45A, 564 RBP45B, and RBP45C were reported to be associated with the UTRs of the GRP7 mRNA 565 (Reichel et al., 2024). These interactions and the concomitant regulatory potential might be 566 particularly important under stress. Accordingly, expression of RBP45A and RBP45B has 567 been reported to be induced upon ozone exposure (Peal et al., 2011). 568 Critical functions of the RBP45 genes in plant development are supported by the mutants’ 569 phenotypes. With an increasing number of knocked out RBP45 genes, we observed a 570 progressively reduced primary root length (Figure 7a). This observation also suggests 571 functional redundancy among the three paralogs. Interestingly, overexpression of RBP45C 572 similarly disturbed root growth. Inspecting the transcript levels in this overexpression line from 573 the RNA-seq data revealed a massive overaccumulation of RBP45C specifically in the root, 574 pointing at tissue-specific regulation (Supplemental Data Set 2). This overaccumulation may 575 have caused suppression of the other two RBP45 paralogs and therefore resulted in a similar 576 phenotype as for the knockout lines. Further research is needed to define the responsible 577 target genes or processes. Interestingly, a previous study by Foley et al. (2017) identified an 578 interaction of RBP45 proteins with a TG -rich motif in the 3’ UTRs of root hair -specific 579 transcripts, potentially linking RBP45 function to root cell fate decisions. Beyond changes in 580 root length, rbp45 mutants showed altered flowering time. Overexpression of RBP45A caused 581 a mild delay, while RBP45B and RBP45C overexpression led to earlier and later flowering, 582 respectively (Figure 7b). Delayed flowering upon loss of RBP45B is consistent with previous 583 observations (Muthuramalingam et al. , 2016) . Interestingly, this phenotype was absent in 584 higher-order mutants, suggesting a potential antagonism or compensatory functions among 585 RBP45 homologs in this process. 586 In conclusion, our work has identified novel physiological functions of the RBP45 genes in 587 plant development. Furthermore, we have characterized an intricate auto - and cross -588 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 19 regulatory circuit that balances the expression of the three RBP45 genes via the strucRNA 589 45ABC. The conserved sequence and structural features of this element point towards an AS 590 regulatory mechanism that involves RBP45 -mediated recruitment of U1 snRNP to an 591 alternative splice site, resulting in negative feedback control of RBP45 expression. 592 593 594

595 Plant cultivation and transformation 596 Arabidopsis thaliana ecotype Columbia-0 plants were either grown on soil or in sterile culture 597 on agar plates. Therefore, seeds were surface-sterilized using 3.75% NaClO and 0.01% Triton 598 X-100 and stratified at 4 °C for 2 - 4 days. If not specified otherwise, they were germinated on 599 ½ Murashige and Skoog medium including vitamins (Duchefa) and containing 2% sucrose 600 and 0.8% phyto agar. Segregating mutant lines were grown on plates containing additionally 601 25 µg/mL kanamycin. For experiments in which root lengths were measured, 1.2% phyto agar 602 was used and plates were placed not in horizontal but vertical orientation. The standard 603 settings of the climate chambers were 16 h white light (~100 µE)/22 °C and 8 h darkness/20 °C 604 at 60% relative humidity. Nicotiana benthamiana was grown on soil in a climate chamber (16 h 605 white light (~120 µE)/24 °C, 8 h dark/22 °C at 60% relative humidity). 606 Stable transformation of A. thaliana was achieved by the floral dip method described by 607 Clough and Bent (1998). Selection of primary transformants was done on half -strength 608 Murashige and Skoog medium (including vitamins) containing 0.8% plant agar, 25 µg/mL 609 kanamycin, and 200 µg/mL cefotaxime. Resistant plants were transferred to soil and upon 610 further growth subjected to PCR -based genotyping. Selection of CRISPR lines was carried 611 out via the fluorescence-accumulating seed technology (FAST; Shimada et al., 2010) under a 612 Leica M205FCA fluorescence stereomicroscope (excitation: 470 nm, emission: 585 nm) and 613 positive seeds directly germinated on soil. Cas9 -free plants in later generations were 614 confirmed by the absence of fluorescence. 615 Nicotiana benthamiana leaves were transformed via Agrobacteria-mediated leaf infiltration as 616 described by Wachter et al. (2007) with a normalization construct based on mOrange2 617 (Shaner et al., 2008). After two days, leaf material was harvested two and thre e days after 618 infiltration, respectively, for analysis on RNA and protein level. 619 Genotyping of plant mutants 620 The genetic status of the generated mutants was confirmed by isolating genomic DNA 621 followed by genotyping PCR. To 100 mg ground plant tissue, 500 µL extraction buffer (200 622 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 20 mM Tris-HCl pH 9, 400 mM LiCl, 25 mM EDTA, 1% SDS) was added and vortexed thoroughly 623 at room temperature. After 5 min centrifugation at 15 000 g, the supernatant was mixed with 624 the same volume of isopropanol, incubated for 5 min, and then centrifuged for 10 min at 625 15;000 g to precipitate the gDNA. The pellet was washed with 70% Ethanol, dried at 37 °C, 626 and resuspended in ½ TE buffer (5 mM Tris-HCl pH 8, 0.5 mM EDTA). Genotyping PCR was 627 performed with homemade Taq polymerase according to standard procedures with primers 628 listed in Supplemental Data Set 4. For CRISPR lines, a proof -reading polymerase was used 629 for PCR amplification, followed by purification of PCR fragments (GeneJET PCR Purification 630 Kit, Thermo Fisher Scientific) and Sanger sequencing of mutated sites. 631 Plant phenotyping 632 Seeds were sorted using Boxeed 2.1 (Labdeers) with the following criteria: pixel count range 633 200 – 2000, SSE 0 – 95, and LS ratio 0 – 25. Measurements and pictures of matching seeds 634 were taken, and size distribution was analyzed. 635 For root length measurements, seeds were also stratified for 96 h in darkness at 4 °C before 636 plating them on square ½ MS plates singly in one row. The plates were placed vertically into 637 racks and kept under beforementioned conditions for 7 days. The plates were scanned, and 638 primary root length was measured using ImageJ software (Schindelin et al. 2012). 639 For the determination of flowering time, seeds were stratified for 96 h in darkness at 4 °C and 640 then cultivated on soil under the conditions mentioned above. Flowering time was defined as 641 the time span at which the inflorescence reached a length of 1 cm. 642 Cloning procedures 643 The splicing reporters for RBP45A (AT5G54900), RBP45B (AT1G11650), and RBP45C 644 (AT4G27000) were constructed by PCR amplification of the corresponding sequences from 645 A. thaliana genomic DNA using the oligonucleotides indicated in Supplemental Data Set 4 and 646 sub-cloning into pGEM -T (Promega, www.promega.de) . All final reporter constructs were 647 cloned in the pBinAR vector (Höfgen and Willmitzer, 1990) using the previously published 648 TFIIIA reporter (Hammond et al., 2009) by replacing the target gene region via KpnI/XbaI 649 restriction digest. Motif mutations were introduced via PCR mutagenesis in pGEM -T using 650 oligonucleotides as indicated in Supplemental Data Set 4. The CDS constructs used for leaf 651 infiltration assays and generating the stable constitutive overexpression lines were also cloned 652 in the pBinAR vector. Upon PCR amplification from cDNA, the fragments were first subcloned 653 into pGEM-T and subsequently ligated into pBinAR via KpnI/XbaI. All final constructs were 654 confirmed by sequencing of the inserts. The CRISPR-Cas9 constructs for generation of rbp45 655 knockout plants were created using the GoldenGate cloning system reported by Stuttmann et 656 al. (2020). The sgRNAs indicated in Supplemental Data Set 4 were designed using 657 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 21 CHOPCHOP (Labun et al., 2019) and cloned into the pDGE652 or pDGE347 vector under 658 control of the A. thaliana U6-26 promoter fragment as described by Stuttmann et al. (2020). 659 The vector additionally features an intron -containing Cas9 under control of the RPS5A 660 promoter. 661 RNA isolation & analysis of splicing patterns 662 RNA was isolated from ~100 mg ground plant tissue using EURx Universal RNA Purification 663 Kit (Roboklon) with an on -column DNase treatment (RNase free DNaseI, NEB) as specified 664 in the manufacturer’s protocol. The RNA concentration was determined using a 665 spectrophotometer (DeNovix DS11 FX+). Reverse transcription was carried out with 666 SuperScript II (Invitrogen) using a dT 20 primer with a 5 min incubation step at 65 °C prior to 667 adding the reaction buffer and enzyme. The cDNA obtained was subjected to analysis of 668 splicing patterns via co-amplification PCR or quantitative PCR (qPCR). 669 Co-amplification PCR was performed with primers (Supplemental Data Set 4) encompassing 670 the alternative spliced regions and a homemade Taq polymerase. The resulting DNA products 671 were separated on horizontal agarose gels. Either a 1 kb plus (New England Biolabs) or a 672 100 bp plus GeneRuler ladder (Thermo Scientific) was used to estimate the molecular weight. 673 The bands were stained after electrophoresis in ethidium bromide solution at 0.5 µg/mL. The 674 bands were visualized with UV light and documented with the Quantum -CX5 Edge (Vilber). 675 Quantification of co -amplification PCR products was performed with a Bioanalyzer 2100 676 (Agilent, www.agilent.com) using the DNA1000 chip according to manufacturer’s protocol. 677 qPCR was performed using the Biorad CFX384 Real -Time PCR system (Biorad, 678 www.biorad.com) and the MESA -BLUE 2x qPCR MasterMix Plus for SYBR® Assay 679 (Eurogentec, www.eurogentec.com). Primers are listed in Supplemental Data Set 4 and 680 primer efficiencies were determined with serial dilutions of template. All reactions were done 681 in triplicates, and a melting curve analysis was included. Analysis of the data was done with 682 the use of the relative standard curve method. Transcript level of PP2A (AT1G13320) served 683 as reference. 684 Whole transcriptome sequencing 685 150 ng total RNA isolated as described before and derived from two biological replicates of 686 10-day old A. thaliana WT, rbp45abc, rbp45bc, and RBP45A/RBP45B/RBP45C 687 overexpression seedlings was utilized. For root-specific analysis, RNA was used from clipped 688 roots of 7-day-old seedlings with three biological replicates each for WT, rbp45bc, rbp45abc, 689 and the RBP45C overexpression line. The concentration and quality of the RNA was assessed 690 based on the RNA integrity number (RIN) measured with an Agilent Bioanalyzer 2100 using 691 the RNA6000 Nano protocol. Eurofins Genomics performed poly-A-selection and generation 692 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 22 and sequencing of strand-specific cDNA libraries on a NovaSeq 6000 S4 system, generating 693 a minimum of 25 million read pairs (2 x 150 nt) per sample. BBDuk (Bushnell B., 694 www.sourceforge.net/projects/bbmap) was used for trimming the reads, which were then 695 mapped to the reference transcriptome AtRTD2 -QUASI (Zhang et al., 2017) using Salmon 696 (Patro et al., 2017). Data was analyzed using the 3D RNA-seq App (Guo et al., 2021). Further 697 details are given in Supplemental Methods. 698 To determine relative positions of AS events within transcripts, first the longest possible ORF 699 from each transcript in AtRTD2-QUASI was determined. For each gene, the transcript with the 700 longest ORF was set as reference to define coordinates of translation start and end sites. 701 Based on these coordinates, AS events were localized in all transcript isoforms per gene. Only 702 the first event from the 5’ end was considered. The same events in multiple isoforms were 703 counted only once. Further details are provided in Supplemental Methods. 704 Venn diagram displays were created with DeepVenn (Hulsen, 2022). Gene ontology term 705 analysis was performed using ShinyGO 0.8 5 (Ge et al., 2020) with as FDR cutoff of 1 x 10 -5 706 using the KEGG pathway database (Kanehisa et al., 2021). 707 Protein extraction and fluorescence assay 708 Total protein was isolated from ~100 mg ground tissue of infiltrated N. benthamiana leaves. 709 After adding 300 µL buffer (50 mM Tris pH 7.5, 150 mM NaCl, 0.1% Tween, 0.1% ß -710 mercaptoethanol), the samples were vortexed and centrifuged at 4 °C for 15 min at 15 000 g. 711 Supernatant was transferred to a new tube and then 100 µL of each sample pipetted into a 712 96 well-plate (Greiner, black, flat bottom) for fluorescence measurement. The fluorescence 713 was detected using the TECAN infinite M1000, with excitation at 478 - 492 nm and emission 714 at 515 - 525 nm for EGFP measurement. For fluorescence measurement of the reference 715 mOrange2, the settings 525 - 530 nm for excitation and 590 - 610 nm for emission were used. 716 Measurements were taken with flash frequency of 400 Hz and integration time of 20 µs. The 717 measured values were read out in TECAN i -control and transferred to Microsoft Excel 718 (Microsoft Office 2016) for evaluation. 719 Statistical analyses 720 Statistics were carried out with Prism GraphPad 9.4.0 (GraphPad; www.graphpad.com) or, in 721 case of seed surface data, using R Statistical Software (v4.1.2; R Core Team 2021). It was 722 assumed that data follow a normal distribution. Results from statistical analysis are listed in 723 Supplemental Data Set 5. 724 725 726 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 23 Funding 727 This research project was supported by the Deutsche Forschungsgemeinschaft (DFG, 728 German Research Foundation) to AW and ZW (Project No 453961148). 729 730 731 Author contributions 732 Conceptualization: A.W., M.R, Z.W..; Investigation: M.R., R.B., C.E., M.S.; Writing: M.R. and 733 A.W.; Supervision: A.W. and Z.W.; Funding acquisition: A.W. and Z.W. 734 735

736 We are grateful to Christine Wendler and Celine Denrath for technical assistance. 737 738 739 Data statement 740 RNA-seq data have been deposited in the European Nucleotide Archive (ENA) repository 741 (https://www.ebi.ac.uk/ena) under accession number PRJEB102517. 742 743 744 Short legends for Supporting Information 745 Supplemental Figure 1. RBP45 splicing patterns in different tissues. 746 Supplemental Figure 2. Models of RBP45 genes and sites of mutations in knockout lines. 747 Supplemental Figure 3. 45ABC mutations change splicing of RBP45C reporter in N. 748 benthamiana 749 Supplemental Figure 4. Seed sizes in RBP45 misexpression lines and WT. 750 Supplemental Figure 5. Principal component analyses of transcriptome datasets. 751 Supplemental Figure 6. RBP45 misexpression causes relatively few changes in AS and 752 gene expression. 753 Supplemental Figure 7. A Copia-like retrotransposon is downregulated in all rbp45 754 misexpression lines 755 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 24 Supplemental Figure 8. Tissue-specific gene expression in WT and rbp45 misexpression 756 lines. 757 Supplemental Figure 9. DAS genes and AS event location in root samples upon RBP45 758 misexpression. 759 Supplemental Figure 10. Validation of splicing patterns for candidate genes in RBP45 760 knockout and overexpression lines. 761 Supplemental Figure 11. Model of crosstalk and hierarchy in RBP45-mediated splicing 762 regulation. 763 764 Supplemental Methods 765 766 767 Supplemental Data Set 1. Representatives of the 45ABC motif. 768 Supplemental Data Set 2. RNA-seq data analysis. 769 Supplemental Data Set 3. GO term analysis. 770 Supplemental Data Set 4. List of oligonucleotides and constructs. 771 Supplemental Data Set 5. Statistical analyses. 772 773 774 775 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 25 Figure legends 776 Figure 1: Alternative splicing of RBP45A, RBP45B, and RBP45C genes is associated 777 with the structured element 45ABC. 778 (a) Schematic depiction of alternatively spliced region of RBP45A (AT5G54900), RBP45B 779 (AT1G11650), and RBP45C (AT4G27000) giving rise to coding ( cd) and non -coding ( nc) 780 isoforms with the CE encompassed by the strucRNA 45ABC (rounded red rectangle). PTC 781 indicated by asterisk; exons, introns, CDS and UTRs are depicted by boxes, lines, black and 782 white shading, respectively. 783 (b) Predicted secondary structure of strucRNA 45ABC based on the consensus sequence 784 from A. thaliana . Red letters show alternative splice sites used for CE inclusion. Letters 785 indicate nucleotides (nt); W, A or U; R, A or G; Y, C or U; lines show variable -length regions. 786 Grey shading marks U1 snRNP binding region. Stems I and II are labelled. 787 (c) AS patterns of RBP45 genes based on RT -PCR products of samples from 11 -day-old 788 whole seedlings (s), roots of 7-day-old seedlings (r), rosette leaves (rl), and flower buds (fb). 789 Binding sites of corresponding primer pairs indicated in Figure S2a. L, size marker in 100 bp 790 increments from 0.3 to 0.7 kb. 791 792 Figure 2: RBP45A, RBP45B, and RBP45C can induce inclusion of the cassette exon 793 overlapping with the 45ABC strucRNA. 794 (a) Design of splicing reporter constructs for RBP45 genes (display based on RBP45C) with 795 boxes and lines representing exons and introns, respectively; arrowheads show the 796 approximate binding sites of primers used for co-amplification PCR shown in (b). 797 (b) RT-PCR products of splicing variants from reporters based on RBP45A (top), RBP45B 798 (middle), and RBP45C (bottom) upon transient expression in N. benthamiana leaves co -799 infiltrated with LUC (-), RBP45A CDS (A), RBP45B CDS (B), or RBP45C CDS (C) constructs. 800 The topmost visible band for the RBP45A and RBP45B reporters likely represents a gel 801 running artefact, as it is absent from Bioanalyzer runs and cannot be identified as distinct 802 splicing variant using sequencing. L: size marker, 100 bp increments. 803 (c, d) Quantification of reporter output based on AS ratios of RT -PCR products analysed via 804 Bioanalyzer ( c) and GFP fluorescence ( d). Bars indicate mean values, error bars show 805 standard deviations, and open circles represent individual data points. AS ratio and 806 fluorescence of LUC samples each was set to 1. Asterisks indicate significant differences 807 compared to LUC control (one sample t test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). 808 809 Figure 3: Analysis of negative auto - and cross-regulation in RBP45 overexpression 810 lines unveils a major function of RBP45 in splicing control. 811 Quantitative PCR analysis of total (a), endogenous coding (b), and endogenous non -coding 812 (c) RBP45 transcript levels in 11-day-old RBP45 overexpression (OE) A. thaliana seedlings. 813 All values are expressed relative to reference transcript PP2A and normalized to the 814 respective WT mean. Circles represent individual data points from biological replicates and 815 standard deviations are depicted. Asterisks indicate significant change compared to WT (one-816 way ANOVA followed by Dunnett’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 817 0.001, ****p < 0.0001); nd, not determined. 818 819 Figure 4: RBP45 knockouts reveal the crosstalk and hierarchy in RBP45 -mediated 820 splicing regulation. Quantitative PCR analysis of total (a), coding (b), and non -coding (c) 821 RBP45 transcript levels in 11-day-old A. thaliana seedlings in rbp45 single and higher order 822 mutants. All values are expressed relative to PP2A and normalized to the respective WT 823 mean. Circles represent individual data points from biological replicates and standard 824 deviations are depicted. Asterisks indicate significant change compared to WT ( one-way 825 ANOVA followed by Dunnett’s multiple comparisons test, * p < 0.05, ** p < 0.01, ***p < 0.001, 826 **** p < 0.0001). 827 828 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 26 829 Figure 5: Disrupting the 45ABC structure causes an AS shift towards cassette exon 830 inclusion. 831 (a) Schematic representation of the base pairing potential of 45ABC in its wild-type (WT) form 832 of RBP45C, as well as disruptive (DM) and compensatory (CM) mutations. Stem I and II areas 833 shaded in pink and grey, respectively. Alternative splice site is indicated in red letters and 834 mutated nucleotides are shown in purple (DM) or blue (CM). 835 (b) Bioanalyzer quantification of WT and mutant 45ABC reporter AS in 11 -day-old stably 836 transformed A. thaliana seedlings. Reporter constructs based on RBP45C sequence. Upper 837 part shows reporter AS ratios with mean value of WT reporter set to 1. Bars show mean values 838 and individual data points correspond to independent transformant lines. Standard deviations 839 are depicted, and asterisks indicate significant change compared to the WT reporter (one-way 840 ANOVA followed by Dunnett’s multiple comparisons test, * p < 0.05 , ** p < 0.01). Lower part 841 shows agarose gel analyses of representative samples with grey and red arrowheads 842 indicating RT-PCR bands from coding and non -coding splice variants, respectively. L: size 843 marker, from bottom to top: 0.5 – 0.8 kb, in 0.1 kb increments. 844 845 Figure 6: A purine stretch within stem I of 45ABC promotes inclusion of the cassette 846 exon. 847 (a) Schematic representation of the base pairing potential within stem I of 45ABC in its wild-848 type (WT) form as well as potential binding motif mutations (BM1, BM2) in the context of an 849 RBP45C splicing reporter. Mutated nucleotides are depicted in blue or purple as in Figure 5a. 850 (b) Agarose gel analysis of RT-PCR products of splicing variants derived from RBP45C WT, 851 BM1, and BM2 reporter upon co -expression in N. benthamiana leaves with LUC (-) or CDS 852 constructs of RBP45A (A), RBP45B (B), or RBP45C (C). L: size marker, from bottom to top: 853 0.4 – 1.0 kb in 0.1 kb increments. 854 (c) Log-scale quantification of reporter splicing for the samples described in (b). Mean values 855 with standard deviation are depicted; open circles represent individual data points. Asterisks 856 indicate significant differences compared to the WT construct (two -way ANOVA followed by 857 Dunnett’s multiple comparisons test; **p < 0.01, ****p < 0.0001). 858 859 Figure 7: Altered root length and flowering time upon RBP45 misexpression. 860 (a) Primary root lengths of 7-day-old A. thaliana WT and RBP45 misexpression lines depicted 861 by representative pictures and mean values with standard deviations. White scale bar 862 corresponds to 1 cm and numbers of analysed seedlings from three independent experiments 863 are indicated. Asterisks indicate significant change compared to WT ( one-way ANOVA 864 followed by Dunnett’s multiple comparisons test; **** p < 0.0001). 865 (b) Flowering time in days after sowing. Lines, display details, and statistical analysis ( * p < 866 0.05, **** p < 0.0001) as described for (a). 867 868 Figure 8: RBP45 misexpression affects expression and AS of few genes in A. thaliana. 869 (a, b) Differential gene expression in rbp45 mutants compared to WT for 10 -day-old whole 870 seedlings or roots from 7 -day-old plants. Venn diagram (a) shows numbers of combined 871 significant genes, while bar plot (b) provides line-wise comparisons and direction of changes. 872 (c, d) Genes showing differential AS for samples as defined in (a, b). Venn diagram of DAS 873 genes cumulated for seedling and root samples (c) or for individual mutant comparisons in 874 seedlings (d) are displayed. Numbers correspond to DAS genes, with number in parenthesis 875 in (d) indicating the overlap with the root data. 876 (e) Positioning of DAS events (RBP45 -dep., top) compared to all events (AtRTD2 -QUASI, 877 bottom) in proportions and total numbers (in parentheses). Based on their location relative to 878 translational start and stop sites, events were assigned to the 5’ UTR, CDS, or 3’ UTR; “nd” 879 refers to events that were overlapping or could not be located, as further described in Figure 880 S9c and the method section. 881 882 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 27 883 Figure 9: Reciprocal AS shifts upon RBP45B knockout and overexpression confirm its 884 function in splicing regulation. AS ratios were analy sed from 10 -day-old A. thaliana WT, 885 rbp45bc, rbp45abc, OE -A#15, OE -B#1, and OE -C#8 for ALAD1 (AT1G69740), VAB2 886 (AT4G38510), DEP1 (AT5G53850), ABC1K8 (AT5G64940), HYH (AT3G17609), and AVT6 887 (AT3G30390). For each AS event, bar chart based on Bioanalyzer quantification (top) and 888 representative gel picture (middle) of RT -PCR co -amplification products, and the 889 corresponding gene models (bottom) are displayed. Mean value (bars), standard deviation 890 (error bars), and individual data points (circles: based on samples used for RNA sequencing; 891 triangles: additional replicates) are depicted each; mean AS ratio of WT was set to 1. Asterisks 892 indicate significant change compared to WT (one-way ANOVA followed by Dunnett’s multiple 893 comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Size ladder (L) for gels 894 consisted of DNAs in 100 bp increments with the strongest band corresponding to 500 bp 895 (marked with gray dot). In the gene models, exons, introns, CDS, and UTRs are depicted by 896 boxes, lines, black, and white shading, respectively; double dashes indicate cropped regions; 897 arrowheads show primer binding sites and scale bar is individually adjusted to 100 bp for each 898 model. 899 900 Figure 10: Model of 45ABC-mediated AS regulation. 901 The strucRNA 45ABC consists of two stem loops, with the second one encompassing the 902 alternative 5’ splice site used for CE inclusion. When RBP45 protein (brown cloud) is not 903 bound to stem I (left), the CE is removed resulting in the cd variant that is translated into 904 RBP45 protein. RBP45 binding to stem I may facilitate U1 snRNP recruitment to the alternative 905 5‘ splice site in stem II (right), thereby promoting its usage. The CE -containing nc variant 906 contains a premature termination codon (asterisk), triggering NMD turnover as part of the 907 negative feedback regulatory loop. Gray and white boxes in the transcript models correspond 908 to coding and non-coding regions. 909 910 911 912 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 28 Figure 1 913 914 915 916 Figure 1: Alternative splicing of RBP45A, RBP45B, and RBP45C genes is associated 917 with the structured element 45ABC. 918 (a) Schematic depiction of alternatively spliced region of RBP45A (AT5G54900), RBP45B 919 (AT1G11650), and RBP45C (AT4G27000) giving rise to coding ( cd) and non -coding ( nc) 920 isoforms with the CE encompassed by the strucRNA 45ABC (rounded red rectangle). PTC 921 indicated by asterisk; exons, introns, CDS and UTRs are depicted by boxes, lines, black and 922 white shading, respectively. 923 (b) Predicted secondary structure of strucRNA 45ABC based on the consensus sequence 924 from A. thaliana . Red letters show alternative splice sites used for CE inclusion. Letters 925 indicate nucleotides (nt); W, A or U; R, A or G; Y, C or U; lines show variable -length regions. 926 Grey shading marks U1 snRNP binding region. Stems I and II are labelled. 927 (c) AS patterns of RBP45 genes based on RT -PCR products of samples from 11 -day-old 928 whole seedlings (s), roots of 7-day-old seedlings (r), rosette leaves (rl), and flower buds (fb). 929 Binding sites of corresponding primer pairs indicated in Figure S2a. L, size marker in 100 bp 930 increments from 0.3 to 0.7 kb. 931 932 933 934 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 29 Figure 2 935 936 937 938 Figure 2: RBP45A, RBP45B, and RBP45C can induce inclusion of the cassette exon 939 overlapping with the 45ABC strucRNA. 940 (a) Design of splicing reporter constructs for RBP45 genes (display based on RBP45C) with 941 boxes and lines representing exons and introns, respectively; arrowheads show the 942 approximate binding sites of primers used for co-amplification PCR shown in (b). 943 (b) RT-PCR products of splicing variants from reporters based on RBP45A (top), RBP45B 944 (middle), and RBP45C (bottom) upon transient expression in N. benthamiana leaves co -945 infiltrated with LUC (-), RBP45A CDS (A), RBP45B CDS (B), or RBP45C CDS (C) constructs. 946 The topmost visible band for the RBP45A and RBP45B reporters likely represents a gel 947 running artefact, as it is absent from Bioanalyzer runs and cannot be identified as distinct 948 splicing variant using sequencing. L: size marker, 100 bp increments. 949 (c, d) Quantification of reporter output based on AS ratios of RT -PCR products analysed via 950 Bioanalyzer ( c) and GFP fluorescence ( d). Bars indicate mean values, error bars show 951 standard deviations, and open circles represent individual data points. AS ratio and 952 fluorescence of LUC samples each was set to 1. Asterisks indicate significant differences 953 compared to LUC control (one sample t test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). 954 955 956 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 30 Figure 3 957 958 959 960 Figure 3: Analysis of negative auto - and cross-regulation in RBP45 overexpression 961 lines unveils a major function of RBP45B in splicing control. 962 Quantitative PCR analysis of total (a), endogenous coding (b), and endogenous non -coding 963 (c) RBP45 transcript levels in 11-day-old RBP45 overexpression (OE) A. thaliana seedlings. 964 All values are expressed relative to reference transcript PP2A and normalized to the 965 respective WT mean. Circles represent individual data points from biological replicates and 966 standard deviations are depicted. Asterisks indicate significant change compared to WT (one-967 way ANOVA followed by Dunnett’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 968 0.001, ****p < 0.0001); nd, not determined. 969 970 971 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 31 Figure 4 972 973 974 975 Figure 4: RBP45 knockouts reveal the crosstalk and hierarchy in RBP45 -mediated 976 splicing regulation. Quantitative PCR analysis of total (a), coding (b), and non -coding (c) 977 RBP45 transcript levels in 11-day-old A. thaliana seedlings in rbp45 single and higher order 978 mutants. All values are expressed relative to PP2A and normalized to the respective WT 979 mean. Circles represent individual data points from biological replicates and standard 980 deviations are depicted. Asterisks indicate significant change compared to WT ( one-way 981 ANOVA followed by Dunnett’s multiple comparisons test, * p < 0.05, ** p < 0.01, ***p < 0.001, 982 **** p < 0.0001). 983 984 985 986 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 32 Figure 5 987 988 989 990 Figure 5: Disrupting the 45ABC structure causes an AS shift towards cassette exon 991 inclusion. 992 (a) Schematic representation of the base pairing potential of 45ABC in its wild-type (WT) form 993 of RBP45C, as well as disruptive (DM) and compensatory (CM) mutations. Stem I and II areas 994 shaded in pink and grey, respectively. Alternative splice site is indicated in red letters and 995 mutated nucleotides are shown in purple (DM) or blue (CM). 996 (b) Bioanalyzer quantification of WT and mutant 45ABC reporter AS in 11 -day-old stably 997 transformed A. thaliana seedlings. Reporter constructs based on RBP45C sequence. Upper 998 part shows reporter AS ratios with mean value of WT reporter set to 1. Bars show mean values 999 and individual data points correspond to independent transformant lines. Standard deviations 1000 are depicted, and asterisks indicate significant change compared to the WT reporter (one-way 1001 ANOVA followed by Dunnett’s multiple comparisons test, * p < 0.05 , ** p < 0.01). Lower part 1002 shows agarose gel analyses of representative samples with grey and red arrowheads 1003 indicating RT-PCR bands from coding and non -coding splice variants, respectively. L: size 1004 marker, from bottom to top: 0.5 – 0.8 kb, in 0.1 kb increments. 1005 1006 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 33 Figure 6 1007 1008 1009 1010 Figure 6: A purine stretch within stem I of 45ABC promotes inclusion of the cassette 1011 exon. 1012 (a) Schematic representation of the base pairing potential within stem I of 45ABC in its wild-1013 type (WT) form as well as potential binding motif mutations (BM1, BM2) in the context of an 1014 RBP45C splicing reporter. Mutated nucleotides are depicted in blue or purple as in Figure 5a. 1015 (b) Agarose gel analysis of RT-PCR products of splicing variants derived from RBP45C WT, 1016 BM1, and BM2 reporter upon co -expression in N. benthamiana leaves with LUC (-) or CDS 1017 constructs of RBP45A (A), RBP45B (B), or RBP45C (C). L: size marker, from bottom to top: 1018 0.4 – 1.0 kb in 0.1 kb increments. 1019 (c) Log-scale quantification of reporter splicing for the samples described in (b). Mean values 1020 with standard deviation are depicted; open circles represent individual data points. Asterisks 1021 indicate significant differences compared to the WT construct (two -way ANOVA followed by 1022 Dunnett’s multiple comparisons test; **p < 0.01, ****p < 0.0001). 1023 1024 1025 1026 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 34 Figure 7 1027 1028 1029 1030 Figure 7: Altered root length and flowering time upon RBP45 misexpression. 1031 (a) Primary root lengths of 7-day-old A. thaliana WT and RBP45 misexpression lines depicted 1032 by representative pictures and mean values with standard deviations. White scale bar 1033 corresponds to 1 cm and numbers of analysed seedlings from three independent experiments 1034 are indicated. Asterisks indicate significant change compared to WT ( one-way ANOVA 1035 followed by Dunnett’s multiple comparisons test; **** p < 0.0001). 1036 (b) Flowering time in days after sowing. Lines, display details, and statistical analysis (* p < 1037 0.05, **** p < 0.0001) as described for (a). 1038 1039 1040 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 35 Figure 8 1041 1042 1043 1044 Figure 8: RBP45 misexpression affects expression and AS of few genes in A. thaliana. 1045 (a, b) Differential gene expression in rbp45 mutants compared to WT for 10 -day-old whole 1046 seedlings or roots from 7 -day-old plants. Venn diagram (a) shows numbers of combined 1047 significant genes, while bar plot (b) provides line-wise comparisons and direction of changes. 1048 (c, d) Genes showing differential AS for samples as defined in (a, b). Venn diagram of DAS 1049 genes cumulated for seedling and root samples (c) or for individual mutant comparisons in 1050 seedlings (d) are displayed. Numbers correspond to DAS genes, with number in parenthesis 1051 in (d) indicating the overlap with the root data. 1052 (e) Positioning of DAS events (RBP45 -dep., top) compared to all events (AtRTD2 -QUASI, 1053 bottom) in proportions and total numbers (in parentheses). Based on their location relative to 1054 translational start and stop sites, events were assigned to the 5’ UTR, CDS, or 3’ UTR; “nd” 1055 refers to events that were overlapping or could not be located, as further described in Figure 1056 S9c and the method section. 1057 1058 1059 1060 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 36 Figure 9 1061 1062 1063 1064 Figure 9: Reciprocal AS shifts upon RBP45B knockout and overexpression confirm its 1065 function in splicing regulation. 1066 AS ratios were analysed from 10-day-old A. thaliana WT, rbp45bc, rbp45abc, OE-A#15, OE-1067 B#1, and OE -C#8 for ALAD1 (AT1G69740), VAB2 (AT4G38510), DEP1 (AT5G53850), 1068 ABC1K8 (AT5G64940), HYH (AT3G17609), and AVT6 (AT3G30390). For each AS event, bar 1069 chart based on Bioanalyzer quantification (top) and representative gel picture (middle) of RT-1070 PCR co-amplification products, and the corresponding gene models (bottom) are displayed. 1071 Mean value (bars), standard deviation (error bars), and individual data points (circles: based 1072 on samples used for RNA sequencing; triangles: additional replicates) are depicted each; 1073 mean AS ratio of WT was set to 1. Asterisks indicate significant change compared to WT (one-1074 way ANOVA followed by Dunnett’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 1075 0.001, ****p < 0.0001). Size ladder (L) for gels consisted of DNAs in 100 bp increments with 1076 the strongest band corresponding to 500 bp (marked with gray dot). In the gene models, 1077 exons, introns, CDS, and UTRs are depicted by boxes, lines, black, and white shading, 1078 respectively; double dashes indicate cropped regions; arrowheads show primer binding sites 1079 and scale bar is individually adjusted to 100 bp for each model. 1080 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 37 Figure 10 1081 1082 1083 1084 Figure 10: Model of 45ABC-mediated AS regulation. 1085 The strucRNA 45ABC consists of two stem loops, with the second one encompassing the 1086 alternative 5’ splice site used for CE inclusion. When RBP45 protein (brown cloud) is not 1087 bound to stem I (left), the CE is removed resulting in the cd variant that is translated into 1088 RBP45 protein. RBP45 binding to stem I may facilitate U1 snRNP recruitment to the alternative 1089 5‘ splice site in stem II (right), thereby promoting its usage . The CE-containing nc variant 1090 contains a premature termination codon (asterisk), triggering NMD turnover as part of the 1091 negative feedback regulatory loop. Gray and white boxes in the transcript models correspond 1092 to coding and non-coding regions. 1093 1094 1095 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 38

1096 1097 Alhabsi, A., Ling, Y., Crespi, M., Reddy, A.S.N. and Mahfouz, M. (2025) Alternative Splicing 1098 Dynamics in Plant Adaptive Responses to Stress. Annual Review of Plant Biology, 14, 687–717. 1099 Berg, M.G., Singh, L.N., Younis, I., Liu, Q., Pinto, A.M., Kaida, D., Zhang, Z., Cho, S., Sherrill -1100 Mix, S., Wan, L. and Dreyfuss, G. (2012) U1 snRNP determines mRNA length and regulates 1101 isoform expression. Cell, 150, 53–64. 1102 Bocobza, S., Adato, A., Mandel, T., Shapira, M., Nudler, E. and Aharoni, A. (2007) Riboswitch-1103 dependent gene regulation and its evolution in the plant kingdom. Genes & development, 21, 1104 2874–2879. 1105 Briggs, G.C., Osmont, K.S., Shindo, C., Sibout, R. and Hardtke, C.S. (2006) Unequal genetic 1106 redundancies in Arabidopsis—a neglected phenomenon? Trends in Plant Science, 11, 492–498. 1107 Broft, P., Rosenkranz, R., Schleiff, E., Hengesbach, M. and Schwalbe, H. (2022) Structural 1108 analysis of temperature-dependent alternative splicing of HsfA2 pre-mRNA from tomato plants. 1109 RNA biology, 19, 266–278. 1110 Buratti, E. and Baralle, F. (2004) Influence of RNA Secondary Structure on the Pre-mRNA Splicing 1111 Process. Molecular and Cellular Biology, 24, 10505–10514-10505–10514. 1112 Burgardt, R., Lambert, D., Heuwieser, C., Sack, M., Wagner, G., Weinberg, Z. and Wachter, A. 1113 (2024) Positioning of pyrimidine motifs around cassette exons defines their PTB -dependent 1114 splicing in Arabidopsis. The Plant journal : for cell and molecular biology, 118, 2202–2218. 1115 Burgess, D. and Freeling, M. (2014) The most deeply conserved noncoding sequences in plants 1116 serve similar functions to those in vertebrates despite large differences in evolutionary rates. Plant 1117 Cell, 26, 946–961. 1118 Chang, P., Hsieh, H. and Tu, S. (2022) The U1 snRNP component RBP45d regulates temperature -1119 responsive flowering in Arabidopsis. The Plant Cell, 34, 834–851. 1120 Chicois, C., Scheer, H., Garcia, S., Zuber, H., Mutterer, J., Chicher, J., Hammann, P., Gagliardi, 1121 D. and Garcia, D. (2018) The UPF1 interactome reveals interaction networks between RNA 1122 degradation and translation repression factors in Arabidopsis. Plant Journal, 96, 119–132. 1123 Chung, B.Y., Balcerowicz, M., Di Antonio, M., Jaeger, K.E., Geng, F., Franaszek, K., Marriott, P., 1124 Brierley, I., Firth, A.E. and Wigge, P.A. (2020) An RNA thermoswitch regulates daytime growth in 1125 Arabidopsis. Nature Plants, 6, 522–532. 1126 Clough, S.J. and Bent, A.F. (1998) Floral dip: a simplified method for Agrobacterium-mediated 1127 transformation of Arabidopsis thaliana. The Plant journal : for cell and molecular biology, 16, 735–1128 743. 1129 Dominguez, D., Freese, P., Alexis, M.S., Su, A., Hochman, M., Palden, T., Bazile, C., Lambert, 1130 N.J., van Nostrand, E.L., Pratt, G.A., Yeo, G.W., Graveley, B.R. and Burge, C.B. (2018) 1131 Sequence, Structure, and Context Preferences of Human RNA Binding Proteins. Molecular Cell, 1132 70, 854-867.e9. 1133 Drechsel, G., Kahles, A., Kesarwani, A.K., Stauffer, E., Behr, J., Drewe, P., Rätsch, G. and 1134 Wachter, A. (2013) Nonsense-mediated decay of alternative precursor mRNA splicing variants is 1135 a major determinant of the Arabidopsis steady state transcriptome. Plant Cell, 25, 3726–3742. 1136 Fernández-Moya, S.M., Ehses, J., Bauer, K.E., Schieweck, R., Chakrabarti, A.M., Lee, F.C.Y., 1137 Illig, C., Luscombe, N.M., Harner, M., Ule, J. and Kiebler, M.A. (2021) RGS4 RNA Secondary 1138 Structure Mediates Staufen2 RNP Assembly in Neurons. International Journal of Molecular 1139 Sciences, 22. 1140 Fischer, J.W., Busa, V.F., Shao, Y. and Leung, A.K. (2020) Structure-Mediated RNA Decay by 1141 UPF1 and G3BP1. Molecular Cell, 78, 70-84.e6. 1142 Foley, S.W., Gosai, S.J., Wang, D., Selamoglu, N., Sollitti, A.C., Köster, T., Steffen, A., Lyons, E., 1143 Daldal, F., Garcia, B.A., Staiger, D., Deal, R.B. and Gregory, B.D. (2017) A Global View of 1144 RNA-Protein Interactions Identifies Post-transcriptional Regulators of Root Hair Cell Fate. 1145 Developmental cell, 41, 204-220.e5. 1146 Förch, P., Puig, O., Martínez, C., Séraphin, B. and Valcáreö, J. (2002) The splicing regulator TIA-1 1147 interacts with U1-C to promote U1 snRNP recruitment to 5’ splice sites. The EMBO Journal, 21, 1148 6882–6892. 1149 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 39 Fu, X.D. and Ares, M. (2014) Context-dependent control of alternative splicing by RNA-binding 1150 proteins. Nature Reviews Genetics, 15, 689–701. 1151 Ge, S.X., Jung, D. and Yao, R. (2020) ShinyGO: a graphical gene-set enrichment tool for animals 1152 and plants. Bioinformatics (Oxford, England), 36, 2628–2629. 1153 Georgakopoulos-Soares, I., Parada, G.E. and Hemberg, M. (2022) Secondary structures in RNA 1154 synthesis, splicing and translation. Computational and structural biotechnology journal, 20, 2871–1155 2884. 1156 Goodarzi, H., Najafabadi, H.S., Oikonomou, P., Greco, T.M., Fish, L., Salavati, R., Cristea, I.M. 1157 and Tavazoie, S. (2012) Systematic discovery of structural elements governing stability of 1158 mammalian messenger RNAs. Nature, 485, 264–268. 1159 Gosai, S.J., Foley, S.W., Wang, D., Silverman, I.M., Selamoglu, N., Nelson, A.D.L., Beilstein, 1160 M.A., Daldal, F., Deal, R.B. and Gregory, B.D. (2015) Global analysis of the RNA-protein 1161 interaction and RNA secondary structure landscapes of the Arabidopsis nucleus. Molecular Cell, 1162 57, 376–388. 1163 Guo, W., Tzioutziou, N.A., Stephen, G., Milne, I., Calixto, C.P., Waugh, R., Brown, J.W.S. and 1164 Zhang, R. (2021) 3D RNA-seq: a powerful and flexible tool for rapid and accurate differential 1165 expression and alternative splicing analysis of RNA-seq data for biologists. RNA biology, 18, 1166 1574–1587. 1167 Hammond, M.C., Wachter, A. and Breaker, R.R. (2009) A plant 5S ribosomal RNA mimic regulates 1168 alternative splicing of transcription factor IIIA pre-mRNAs. Nature structural & molecular biology, 1169 16, 541–549. 1170 Hardy, E.C. and Balcerowicz, M. (2024) Untranslated yet indispensable-UTRs act as key regulators 1171 in the environmental control of gene expression. Journal of Experimental Botany, 75, 4314–4331. 1172 Höfgen, R. and Willmitzer, L. (1990) Biochemical and genetic analysis of different patatin isoforms 1173 expressed in various organs of potato (Solanum tuberosum). Plant Science, 66, 221–230. 1174 Huang, W., Zhang, L., Zhu, Y., Chen, J., Zhu, Y., Lin, F., Chen, X. and Huang, J. (2022) A genetic 1175 screen in Arabidopsis reveals the identical roles for RBP45d and PRP39a in 5 ’ cryptic splice site 1176 selection. Frontiers in Plant Science, 13, 1086506. 1177 Hulsen, T. (2022) DeepVenn—a web application for the creation of area-proportional Venn diagrams 1178 using the deep learning framework Tensorflow.js. 1179 Kaida, D., Berg, M.G., Younis, I., Kasim, M., Singh, L.N., Wan, L. and Dreyfuss, G. (2010) U1 1180 snRNP protects pre-mRNAs from premature cleavage and polyadenylation. Nature, 468, 664–668. 1181 Kalyna, M., Simpson, C.G., Syed, N.H., Lewandowska, D., Marquez, Y., Kusenda, B., Marshall, 1182 J., Fuller, J., Cardle, L., McNicol, J., Dinh, H.Q., Barta, A. and Brown, J.W.S. (2012) Alternative 1183 splicing and nonsense-mediated decay modulate expression of important regulatory genes in 1184 Arabidopsis. Nucleic Acids Research, 40, 2454–2469. 1185 Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M. and Tanabe, M. (2021) KEGG: 1186 integrating viruses and cellular organisms. Nucleic Acids Research, 49, D545-D551. 1187 Kosmacz, M., Gorka, M., Schmidt, S., Luzarowski, M., Moreno, J.C., Szlachetko, J., Leniak, E., 1188 Sokolowska, E.M., Sofroni, K., Schnittger, A. and Skirycz, A. (2019) Protein and metabolite 1189 composition of Arabidopsis stress granules. The New phytologist, 222, 1420–1433. 1190 Labun, K., Montague, T.G., Krause, M., Torres Cleuren, Y.N., Tjeldnes, H. and Valen, E. (2019) 1191 CHOPCHOP v3: expanding the CRISPR web toolbox beyond genome editing. Nucleic Acids 1192 Research, 47, W171-W174. 1193 Leder, V., Lummer, M., Tegeler, K., Humpert, F., Lewinski, M., Schüttpelz, M. and Staiger, D. 1194 (2014) Mutational definition of binding requirements of an hnRNP-like protein in Arabidopsis using 1195 fluorescence correlation spectroscopy. Biochemical and biophysical research communications, 1196 453, 69–74. 1197 Liu, Z., Liu, Q., Yang, X., Zhang, Y., Norris, M., Chen, X., Cheema, J., Zhang, H. and Ding, Y. 1198 (2021) In vivo nuclear RNA structurome reveals RNA-structure regulation of mRNA processing in 1199 plants. Genome Biol, 22, 11. 1200 Lorković, Z.J., Kirk, D.A.W., Klahre, U., Hemmings-Mieszczak, M. and Filipowicz, W. (2000) 1201 RBP45 and RBP47, two oligouridylate-specific hnRNP-like proteins interacting with poly(A)1 RNA 1202 in nuclei of plant cells. Cambridge University Press, 6, 1610–1624. 1203 Mangilet, A.F., Weber, J., Schüler, S., Adler, M., Mjema, E.Y., Heilmann, P., Herold, A., 1204 Renneberg, M., Nagel, L., Droste-Borel, I., Streicher, S., Schmutzer, T., Rot, G., Macek, B., 1205 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 40 Schmidtke, C. and Laubinger, S. (2024) The Arabidopsis U1 snRNP regulates mRNA 3’-end 1206 processing. Nature Plants, 10, 1514–1531. 1207 Müller-McNicoll, M., Rossbach, O., Hui, J. and Medenbach, J. (2019) Auto-regulatory feedback by 1208 RNA-binding proteins. Journal of molecular cell biology, 11, 930–939. 1209 Muthuramalingam, M., Wang, Y., Li, Y. and Mahalingam, R. (2016) Interacting protein partners of 1210 Arabidopsis RNA-binding protein AtRBP45b. Plant Biology, 19, 327–334. 1211 Park, J. in, Endo, M., Kazama, T., Saito, H., Hakozaki, H., Takada, Y., Kawagishi -Kobayashi, M. 1212 and Watanabe, M. (2006) Molecular characterization of two anther-specific genes encoding 1213 putative RNA-binding proteins, AtRBP45s, in Arabidopsis thaliana. Genes and Genetic Systems, 1214 81, 355–359. 1215 Patro, R., Duggal, G., Love, M.I., Irizarry, R.A. and Kingsford, C. (2017) Salmon provides fast and 1216 bias-aware quantification of transcript expression. Nature Methods, 14, 417–419. 1217 Peal, L., Jambunathan, N. and Mahalingam, R. (2011) Phylogenetic and expression analysis of 1218 RNA-binding proteins with triple RNA recognition motifs in plants. Molecules and Cells, 31, 55–64. 1219 Raker, V.A., Mironov, A.A., Gelfand, M.S. and Pervouchine, D.D. (2009) Modulation of alternative 1220 splicing by long-range RNA structures in Drosophila. Nucleic Acids Research, 37, 4533–4544. 1221 Reddy, A.S., Marquez, Y., Kalyna, M. and Barta, A. (2013) Complexity of the alternative splicing 1222 landscape in plants. Plant Cell, 25, 3657–3683. 1223 Reichel, M., Schmidt, O., Rettel, M., Stein, F., Köster, T., Butter, F. and Staiger, D. (2024) 1224 Revealing the Arabidopsis AtGRP7 mRNA binding proteome by specific enhanced RNA 1225 interactome capture. BMC plant biology, 24, 552. 1226 Rühl, C., Stauffer, E., Kahles, A., Wagner, G., Drechsel, G., Rätsch, G. and Wachter, A. (2012) 1227 Polypyrimidine tract binding protein homologs from Arabidopsis are key regulators of alternative 1228 splicing with implications in fundamental developmental processes. The Plant Cell, 24, 4360–4375. 1229 Sack, M., Reinhardt, M., Burgardt, R., Berg, P., Bauer, J., Wachter, A. and Weinberg, Z. (2025) 1230 Identification and characterization of new structured RNA classes in plants. RNA biology, 22, 1–16. 1231 Schöning, J.C., Streitner, C., Meyer, I.M., Gao, Y. and Staiger, D. (2008) Reciprocal regulation of 1232 glycine-rich RNA-binding proteins via an interlocked feedback loop coupling alternative splicing to 1233 nonsense-mediated decay in Arabidopsis. Nucleic Acids Research, 36, 6977–6987. 1234 Schöning, J.C., Streitner, C., Page, D.R., Hennig, S., Uchida, K., Wolf, E., Furuya, M. and 1235 Staiger, D. (2007) Auto-regulation of the circadian slave oscillator component AtGRP7 and 1236 regulation of its targets is impaired by a single RNA recognition motif point mutation. Plant Journal, 1237 52, 1119–1130. 1238 Shaner, N.C., Lin, M.Z., McKeown, M.R., Steinbach, P.A., Hazelwood, K.L., Davidson, M.W. and 1239 Tsien, R.Y. (2008) Improving the photostability of bright monomeric orange and red fluorescent 1240 proteins. Nature Methods, 5, 545–551. 1241 Shepard, P.J. and Hertel, K.J. (2008) Conserved RNA secondary structures promote alternative 1242 splicing, 14, 1463–1469. 1243 Sheth, N., Roca, X., Hastings, M.L., Roeder, T., Krainer, A.R. and Sachidanandam, R. (2006) 1244 Comprehensive splice-site analysis using comparative genomics. Nucleic Acids Research, 34, 1245 3955–3967. 1246 Shimada, T.L., Shimada, T. and Hara-Nishimura, I. (2010) A rapid and non-destructive screenable 1247 marker, FAST, for identifying transformed seeds of Arabidopsis thaliana. The Plant journal : for cell 1248 and molecular biology, 61, 519–528. 1249 Staiger, D. and Brown, J.W.S. (2013) Alternative splicing at the intersection of biological timing, 1250 development, and stress responses. The Plant Cell, 25, 3640–3656. 1251 Staiger, D., Zecca, L., Wieczorek Kirk, D.A., Apel, K. and Eckstein, L. (2003) The circadian clock 1252 regulated RNA-binding protein AtGRP7 autoregulates its expression by influencing alternative 1253 splicing of its own pre-mRNA. The Plant journal : for cell and molecular biology, 33, 361–371. 1254 Stauffer, E., Westermann, A., Wagner, G. and Wachter, A. (2010) Polypyrimidine tract-binding 1255 protein homologues from Arabidopsis underlie regulatory circuits based on alternative splicing and 1256 downstream control. The Plant journal : for cell and molecular biology, 64, 243–255. 1257 Steinert, H., Sochor, F., Wacker, A., Buck, J., Helmling, C., Hiller, F., Keyhani, S., Noeske, J., 1258 Grimm, S., Rudolph, M.M., Keller, H., Mooney, R.A., Landick, R., Suess, B., Fürtig, B., 1259 Wöhnert, J. and Schwalbe, H. (2017) Pausing guides RNA folding to populate transiently stable 1260 RNA structures for riboswitch-based transcription regulation. eLife, 6. 1261 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint 41 Streitner, C., Köster, T., Simpson, C.G., Shaw, P., Danisman, S., Brown, J.W.S. and Staiger, D. 1262 (2012) An hnRNP-like RNA-binding protein affects alternative splicing by in vivo interaction with 1263 transcripts in Arabidopsis thaliana. Nucleic Acids Research, 40, 11240–11255. 1264 Stuttmann, J., Barthel, K., Martin, P., Ordon, J., Erickson, J.L., Herr, R., Ferik, F., Kretschmer, 1265 C., Berner, T., Keilwagen, J., Marillonnet, S. and Bonas, U. (2020) Highly efficient multiplex 1266 editing: One-shot generation of 8x Nicotiana benthamiana and 12x Arabidopsis mutants . 1267 Sudarsan, N., Barrick, J.E. and Breaker, R.R. (2003) Metabolite-binding RNA domains are present 1268 in the genes of eukaryotes. Rna, 9, 644–647. 1269 Sulkowska, A., Auber, A., Sikorski, P.J., Silhavy, D.N., Auth, M., Sitkiewicz, E., Jean, V., Merret, 1270 R.M., Bousquet-Antonelli, C.C. and Kufel, J. (2020) RNA Helicases from the DEA(D/H)-Box 1271 Family Contribute to Plant NMD Efficiency. Plant & cell physiology, 61, 144–157. 1272 Sun, L., Fazal, F.M., Li, P., Broughton, J.P., Lee, B., Tang, L., Huang, W., Kool, E.T., Chang, H.Y. 1273 and Zhang, Q.C. (2019) RNA structure maps across mammalian cellular compartments. Nature 1274 structural & molecular biology, 26, 322–330. 1275 Ullah, F., Hamilton, M., Reddy, A.S.N. and Ben-Hur, A. (2018) Exploring the relationship between 1276 intron retention and chromatin accessibility in plants. BMC Genomics, 19, 21. 1277 Wachter, A. (2014) Gene regulation by structured mRNA elements. Trends in Genetics, 30, 172–181. 1278 Wachter, A., Rühl, C. and Stauffer, E. (2012) The role of polypyrimidine tract-binding proteins and 1279 other hnRNP proteins in plant splicing regulation. Frontiers in Plant Science, 3, 1–11. 1280 Wachter, A., Tunc-Ozdemir, M., Grove, B.C., Green, P.J., Shintani, D.K. and Breaker, R.R. (2007) 1281 Riboswitch control of gene expression in plants by splicing and alternative 3 ’ end processing of 1282 mRNAs. The Plant Cell, 19, 3437–3450. 1283 Wang, L., Xu, D., Scharf, K., Frank, W., Leister, D. and Kleine, T. (2022) The RNA-binding protein 1284 RBP45D of Arabidopsis promotes transgene silencing and flowering time. The Plant journal : for 1285 cell and molecular biology, 109, 1397–1415. 1286 Zhang, R., Calixto, C.P.G., Marquez, Y., Venhuizen, P., Tzioutziou, N.A., Guo, W., Spensley, M., 1287 Entizne, J.C., Lewandowska, D., Have, S. ten, Frei Dit Frey, N., Hirt, H., James, A.B., Nimmo, 1288 H.G., Barta, A., Kalyna, M. and Brown, J.W.S. (2017) A high quality Arabidopsis transcriptome 1289 for accurate transcript-level analysis of alternative splicing. Nucleic Acids Research, 45, 5061–1290 5073. 1291 Zhang, R., Kuo, R., Coulter, M., Calixto, C.P.G., Entizne, J.C., Guo, W., Marquez, Y., Milne, L., 1292 Riegler, S., Matsui, A., Tanaka, M., Harvey, S., Gao, Y., Wießner-Kroh, T., Paniagua, A., 1293 Crespi, M., Denby, K., Hur, A.B., Huq, E., Jantsch, M., Jarmolowski, A., Koester, T., 1294 Laubinger, S., Li, Q.Q., Gu, L., Seki, M., Staiger, D., Sunkar, R., Szweykowska -Kulinska, Z., 1295 Tu, S., Wachter, A., Waugh, R., Xiong, L., Zhang, X.-N., Conesa, A., Reddy, A.S.N., Barta, A., 1296 Kalyna, M. and Brown, J.W.S. (2022) A high-resolution single-molecule sequencing-based 1297 Arabidopsis transcriptome using novel methods of Iso-seq analysis. Genome biology, 23, 149. 1298 1299 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 27, 2025. ; https://doi.org/10.1101/2025.11.25.690383doi: bioRxiv preprint