Functional Divergence of GIGANTEA paralogs in Petunia x hybrida Reveals Uncoupling of Vegetative Growth and Flowering Time

29 The circadian clock gene GIGANTEA (GI) coordinates vegetative growth and floral 30 transition in plants, yet the roles of its paralogs in Solanaceae remain unclear. In Petunia 31 × hybrida, the GI locus is duplicated as PhGI1 and PhGI2. We investigated whether this 32 duplication uncouples vegetative growth from flowering time. Using RNA interference 33 and CRISPR/Cas9 mutagenesis, we dissected the function of PhGI2. RNAi silencing of 34 PhGI2 produced complex phenotypes, including enhanced vegetative growth, early floral 35 abortion, and delayed flowering, raising questions about cosuppression effects. To 36 resolve this, we generated three independent CRISPR alleles of PhGI2. These mutants 37 exhibited extreme late flowering without changes in vegetative architecture or scent 38 emission, demonstrating that PhGI2 specifically controls floral transition. Expression 39 analyses revealed that PhGI2 is required to maintain rhythmicity and amplitude of PhGI1 40 and core clock genes, including PhTOC1 and PhLHY. Our findings uncover 41 subfunctionalization of GI paralogs. Whilst PhGI1 represses vegetative growth and 42 influences floral development, PhGI2 promotes flowering and coordinates circadian 43 transcription. This genetic separation of vegetative growth and flowering time provides a 44 framework for manipulating crop architecture and phenology. 45 46

47 GIGANTEA; circadian clock, vegetative growth; flowering time; floral volatiles ; 48 flower development 49 50 51 52 53 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 3 54

55 56 The coordination of plant development with the environment is partly 57 orchestrated by the plant circadian clock, which regulates processes such as 58 flowering time, stomatal opening, cell proliferation and expansion, and floral scent 59 emission (Liu et al., 2001; Nusinow et al., 2011; Fenske et al., 2015; Fung-Uceda 60 et al., 2018). A common set of clock-associated genes, found in the picoeukaryote 61 Ostreococcus and conserved in plants, include a MYB transcription factor LATE 62 ELONGATED HYPOCOTYL (LH Y), a PSEUDO RESPOSE REGULATOR 63 (TOC1), and a blue light receptor similar to ZEITLUPE (ZTL) (Corellou et al., 64 2009; Bouget et al. , 2011) . Early in the evolution of land plants, the genetic 65 architecture of the plant circadian clock became more complex, incorporating 66 additional clock components and interlocking feedback loops (McClung, 2006; 67 Staiger et al. , 2013) . One such gene is GIGANTEA (GI) which is found in 68 Marchantia polymorpha and some charophytes but is absent in Physcomitrium 69 patens or Selaginella moellendorffii (Linde et al. , 2017) . AtGI was originally 70 identified in Arabidopsis as a mutant with delayed flowering and increased 71 vegetative size (Rédei, 1962; Fowler et al., 1999). 72 The biological functions of AtGI in flowering time occur part ly via its role in 73 the circadian clock. Loss of function of AtGI causes a shorter circadian rhythm with 74 dampened rhythmic expression of clock genes such as AtLHY under normal and 75 free running conditions (Mizoguchi et al., 2005; Sawa et al., 2007). It is also a 76 positive regulator of the flowering time gene AtFT in Arabidopsis (Sawa and Kay, 77 2011). This function is conserved in Late Bloomer1, the pea ortholog of GI (Hecht 78 et al. , 2011) Marchantia polymorpha (Kubota et al. , 2014) , soybean, or poplar 79 (Watanabe et al., 2011; Dong et al., 2022; Wang et al., 2023; Alique et al., 2024). 80 It is also responsible for seasonal adaptation in barley (Zakhrabekova et al., 2012). 81 Notably, the two soybean E2 maturity genes are GI orthologs that redundantly 82 regulate photoperiodic flowering and yield (Hayama et al. , 2003; Wang et al. , 83 2023). Thus, GI has a dual biological function in controlling both vegetative growth 84 and flowering time. 85 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 4 86 The GIGANTEA (GI) locus is duplicated in several plant lineages, including 87 members of the Solanaceae, with reported copy numbers ranging from two in 88 Petunia × hybrida to three in P. integrifolia and Solanum lycopersicum (Bombarely 89 et al., 2016). However, the specific roles of individual GI paralogs remain largely 90 unexplored. Functional analysis of PhGI1 in P. hybrida through gene silencing 91 revealed a complex phenotype, including enhanced vegetative growth, ectopic 92 floral development, early floral senescence, and a slight reduction in flower size. 93 Floral scent emission was only mildly affected (Brandoli et al., 2020). Interestingly, 94 silencing PhGI1 also led to a significant decrease in the expression of PhGI2, 95 raising the question of whether the observed phenotypic changes are due solely to 96 the loss of PhGI1, or whether downregulation of PhGI2 also contributes. Moreover, 97 it remains unclear whether the effect on PhGI2 expression is an off-target 98 consequence of the silencing construct, or whether PhGI1 and PhGI2 are involved 99 in a regulatory feedback mechanism governing each other’s expression. However, 100 silencing of PhGI1 does not cause a delayed flowering phenotype suggesting a 101 possible functional divergence of GI in Petunia. 102 Here, we present a comprehensive functional analysis of the Petunia x hybrida 103 PhGI2 paralog, using RNAi silencing and CRISPR/Cas9 -mediated targeted 104 mutagenesis. Our results show that silencing of PhGI2 caused a series of 105 phenotypes previously found in RNAi::PhGI1. However, RNAi::PhGI2 also 106 showed delayed flowering , a phenotype not found in RNAi::PhGI1. We 107 hypothesized that by using CRISPR alleles of PhGI2 we would identify the specific 108 functions of PhGI2. Our results show that PhGI1 and PhGI2 have undergone 109 subfunctionalization and neofunctionalization. PhGI2 retained the ancestral 110 function of promoting floral transition and is required for canonic circadian clock 111 gene expression , while PhGI1 retained the ancestral function of repressing 112 vegetative growth and acquired new roles in flower development. 113 114 115 116 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 5

117 Design of the RNAi and Crispr/Cas9 constructs 118 We obtained the PhGI2 coding region from the genome sequence of Petunia x hybrida 119 W115. The coding gene model corresponds to Peaxi162Scf00160g01744.1. We 120 developed a hairpin construct that would discriminate PhGI2 from PhGI1 for the vector 121 construction targeting the 3’UTR of PhGI2. Site-specific primers (Supplementary Table 122 S1) with the attB1 and attB2 sites for Gateway® recombination, were used to PCR -123 amplify a DNA fragment of 208 bp. To obtain a hairpin -like structure, the PhGI2 124 fragment was first recombined into the entry vector pDONR201 (Invitrogen) and then 125 into the destination vector pHELLSGATE 12 (Helliwell and Waterhouse, 2003). 126 127 The Crispr/Cas9 guide targeting PhGI2 was selected using the web tool CHOPCHOP 128 (Labun et al., 2019) which offers analysis of the target genome of Petunia x hybrida. A 129 vector VB211218 -1015kpn pPBV[CRISPR] -Neo/Kana-zCas9-AtU6-130 26>{PhGI2_exon4} of 15224 bp was build and purchased, with a guide RNA targeting 131 exon 4 of the gene PhGi2 (ACTGCCTTCAACTCCTAGGT). Integrity of all constructs 132 were confirmed through PCR amplification and visualization on 1% agarose gel. 133 134 Plant material, transformation and sampling 135 Seeds of Petunia x hybrida of the double haploid variety 'Mitchell W115' were 136 collected. In vitro germinated plants were used as the wildtype controls and as the source 137 of explants for plant transformation as described (Manchado-Rojo et al. , 2014) . The 138 disarmed Agrobacterium tumefaciens strain EHA105 was used for transformation as 139 described previously. Lines of T0 and T1 generation transformed with Crispr/Cas9 140 constructs, were confirmed through PCR detection of the ZmCAS9gene. 141 Four independent T0 lines RNAi::PhGI2 were selected for further studies. The T1 142 generation of wildtype plants as well as silenced lines of PhGI2, were grown in a growth 143 chamber under controlled conditions of 16 hours of light/8 hours of darkness (16:8 LD), 144 luminous intensity of 250 μE m -2 s -1 and a constant temperature of 26 ± 1° C. The T2 145 generations were cultured in a greenhouse under natural long-day conditions. 146 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 6 The T1 generation of seven independent T0 Crispr/Cas9 -PhGI2 lines were further 147 phenotyped. Plants were grown in a greenhouse a photoperiod ranging from 9.5 – 12.5 h 148 of day light. We obtained three independent alleles that were used further. 149 150 Circadian sampling 151 Samples were taken every three ( RNAi::PhGI2) to four (Crispr/Cas9-PhGi2) hours 152 from wildtype plants during 24 hours and one or two plants from independent silenced 153 lines as well as two early flowering control plants and two late flowering mutant lines 154 homozygous for the Crispr/Cas9-PhGI2 alleles. The collected tissues were immediately 155 frozen in liquid nitrogen and stored at –80°C until further analysis. 156 Under growth chamber conditions, ZEITGEBER Time 0 (ZT0) was considered as 157 the time when the light was turned on. Under natural greenhouse conditions, sampling for 158 expression analysis was conducted at sunrise under 12 hours day light condition with 159 ZEITGEBER Time 0 (ZT0) coinciding with sunrise. We performed experiments under 160 free running conditions using WT plants that had grown for several weeks under long day 161 conditions and were transferred to continuous dark. 162 163 Phenotypic analysis 164 For each selected RNAi:: PhGI2 T1 line, T2 plants were propagated after self -165 pollination. At least three T2 plants were characterized for each line to analyse the 166 phenotypes associated with GI2 RNA interference. We analysed vegetative growth 167 including plant height, internode length, number of leaves to first flower, number of 168 axillary stems, leaf length and width. We quantified number of flower buds and fully 169 developed flowers, corolla diameter, tube and petiole length. 170 T1 populations of 30 -40 individuals from Crispr/Cas9 -PhGI2 lines were grown. 171 Three lines (4,12 and 19) showed a clear segregation into early flowering, mid flowering 172 and late flowering phenotypes and were further phenotyped in detail. The GI2 region 173 targeted by the Crispr/Cas9 construct was amplified and sequenced for three late 174 flowering plants of each line. We analysed vegetative growth including stem length, 175 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 7 branch number, internode length, leaf length and width. We measured flower number, 176 days to first flower, corolla width and tube length. 177 178 Volatile organic compound analysis 179 The analysis of volatile organic emission compound (VOC) was performed sampling 180 three flowers from wildtype plants and two plants of two silenced lines after 2 -3 days 181 after anthesis as described in (Manchado-Rojo et al. , 2012) . For each of the three 182 Crispr/CAS9-PhGI2 lines, we analysed VOC emission of four flowers from early 183 flowering control plants and late flowering mutant plants during a period of 4 hours, 184 starting at sunrise. 185 186 Analysis of circadian gene expression 187 We used phenol:chloroform to isolate total RNA from leaves (Box et al. , 2011) 188 followed by spectrophotometric quantification (NanoDrop2000). Equal amounts of RNA 189 were used to synthesize cDNA according to the manufacturer’s instructions (Maxima 190 First Strand cDNA Synthesis Kit for RT -qPCR, with dsDNase, 191 https://www.thermofischer.com/, catalog number: K1641). Three biological samples in 192 form of young leaves and two technical replicas were analysed for each sample in 193 quantitative PCR analysis. The gene ACTIN 11 (ACT), was used as reference gene for 194 relative gene expression quantification, after selection as valuable housekeeping gene for 195 Petunia leaves and petals under circadian conditions (Terry et al., 2019a). All the primers 196 used for PhGI2, PhGI1 and other clock genes were designed using pcrEfficiency software 197 (Mallona et al., 2011) (Supplementary Table S1). 198 199 Data analysis 200 The relative gene expression of the circadian genes, relative to the reference gene 201 ACT, was calculated applying the comparative CT method (Schmittgen and Livak, 2008) 202 as well as using group -wise comparison with the REST Program (Pfaffl et al., 2002). 203 Periodicity and significance were evaluated using JTK_CYCLE and Lomb–Scargle (LS) 204 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 8 as implemented in MetaCycle (R version 4.3.2) (Glynn et al., 2006; Hughes et al., 2010; 205 Wu et al., 2016). P-values were adjusted with the Benjamini-Hochberg FDR (Benjamini 206 and Hochberg, 1995). 207 Significance differences among data were determined based on Fisher´s F -test and 208 Student´s T-Test after data transformation to fit to a normal distribution. Volatile organic 209 compound profiles were analyzed using the R -package GCprofileMaker (Perez-Sanz et 210 al., 2021). 211 212 3. Results 213 Generation of RNAi and CRISPR/Cas9 PhGI2 Lines 214 To investigate the biological role of PhGI2, we generated and analysed two types of loss-215 of-function lines: four independent RNA interference (RNAi) lines (RNA i::PhGI2.4.2, 216 4.4, 6.1, and 6.2) and three CRISPR/Cas9-induced mutant alleles (PhGI2.12, PhGI2.19, 217 and PhGI2.4). Initial RNAi constructs targeting PhGI1 3’-UTR resulted in ~50% 218 suppression of PhGI2 (Brandoli et al. , 2020) , suggesting either cosuppression effects 219 (Angenent et al., 1994) or potential transcriptional regulation of PhGI2 by PhGI1. To 220 resolve this, we designed RNAi fragments that specifically targeted PhGI2 3’-UTR, 221 avoiding sequence overlap with PhGI1. 222 To further isolate PhGI2-specific functions, we developed CRISPR/Cas9 alleles using 223 guide RNAs targeting unique sequences in exon 4 of PhGI2. We characterized three 224 alleles. PhGI2.19 had a three-base pair in-frame deletion removing the highly conserved 225 Proline 215 (P215) (Supplementary Fig. S1). PhGI2.4 had a single base insertion causing 226 a frameshift at position 216, generating five altered amino acids followed by a premature 227 stop codon . Finally, PhGI2.12 had a four -base deletion leading to an 11 -amino acid 228 frameshift starting at position 215, also followed by a stop codon (Fig. 1A-C). 229 The predicted wild-type PhGI2 protein is 1162 amino acids, consistent in length with GI 230 homologs in Oryza sativa (1160), Arabidopsis thaliana (1173), and Marchantia 231 polymorpha (1187). In contrast, PhGI2.12 and PhGI2.4 encode truncated proteins of only 232 215–216 amino acids. These premature truncations likely disrupt protein function. 233 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 9 Similar truncations in Arabidopsis GI alleles, such as gi-2 (aa144), gi-6 (aa492), and gi-234 3 (aa963) result in late flowering (Araki and Komeda, 1993; Fowler et al., 1999). Notably, 235 the deleted P215 in PhGI2.19 is evolutionarily conserved among GI proteins across plant 236 species (Fig. 1C), supporting the hypothesis that these alleles represent loss -of-function 237 mutations. 238 Collectively, these silenced and mutant lines enabled us to investigate the role of PhGI2 239 within the context of Petunia circadian regulation. 240 241 PhGI1 and PhGI2 lose rhythmicity under free-running conditions 242 We had previously shown that both PhGI1 and PhGI2 appear to lose expres sion under 243 continuous dark (DD) free running conditions (Brandoli et al. , 2020) . These results 244 suggest that PhGI1 and PhGI2 do not function as self-sustained oscillators. Nevertheless, 245 we analysed their expression profiles under continuous dark (DD) free -running 246 conditions. JTK_CYCLE and Lomb –Scargle analyses revealed robust rhythmic 247 expression of PhGI1 and PhGI2 under LD conditions, whereas no significant rhythmicity 248 was detected under free -running conditions (Supp lementary Table S 2). This contrasts 249 with Arabidopsis thaliana , where AtGI maintains rhythmic expression under constant 250 light or darkness (Fowler et al., 1999). These results indicate that Petunia GI genes require 251 environmental cues to sustain rhythmic expression, revealing a species -specific 252 difference in circadian clock architecture. 253 254 PhGI2 affects the peak expression and rhythmicity of clock genes 255 We performed a time -series gene expression relative quantification at 3–4 h intervals. 256 Expression patterns of PhGI2 and PhGI1 were analysed in the CRISPR line PhGI2.12 257 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 10 and in RNAi-silenced lines 4.2 (Fig. 2A–D), 4.4, 6.1, and 6.3. In PhGI2.12, PhGI2 peaked 258 at ZT8, coinciding with WT, and its expression level was unchanged. In contrast, PhGI1 259 peaked earlier (ZT8) than in WT (ZT12), although overall expression levels were 260 maintained. In RNAi::PhGI2 lines, the temporal expression pattern was preserved, but 261 transcript levels of both PhGI1 and PhGI2 were significantly reduced, with up to an 262 eightfold decrease for PhGI2 and an approximately 50% reduction for PhGI1. This 263 downregulation of PhGI1 may result from cross -silencing by the RNAi construct or 264 reflect a regulatory role of PhGI2 in PhGI1 transcription. 265 Silencing of PhGI2 did not lead to major alterations in the expression profiles of the 266 morning-phased genes PhLHY, nor of the evening phased genes PhCHL, PhTOC1 and 267 PhELF4, whose temporal patterns remained largely comparable to wild type across the 268 diel cycle (Fig 2. E,G,I,K). 269 We next investigated the role of PhGI2.12 in regulating the rhythmic expression of the 270 previously mentioned genes PhCHL, PhLHY, PhTOC1, and PhELF4 (Fig. 2 F,H,J,L; 271 Supplementary Table S4). All genes displayed robust rhythmic expression in WT plants. 272 Homozygous PhGI2.12 mutants exhibited advanced expression phases of PhGI1 and 273 PhLHY (4 h) and PhTOC1 (6 h), accompanied by a loss of rhythmicity in PhGI1, PhCHL, 274 and PhELF4. In addition, expression amplitude was reduced across all analysed genes, 275 with the strongest relative reductions observed for PhLHY (66%) and PhTOC1 (79%). 276 These results demonstrate a strong effect of PhGI2 on overall circadian gene expression. 277 278 PhGI2 promotes flowering time but does not affect floral development 279 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 11 We found that RNAi::PhGI2 lines exhibited significant alterations in flowering time. In 280 the T1 generation, flowering was delayed by approximately 2–3 weeks, while in T2 plants 281 the delay ranged from 1 to 4 weeks (Fig. 3 A). Similarly, CRISPR/Cas9 PhGI2 lines 282 showed a notable delay in flowering time in the T1 generation, with a range of 5 –10 283 weeks between early - and late -flowering individuals, displaying a Mendelian 1:2:1 284 segregation pattern (Fig. 3B; Supplementary Table S3). 285 In RNAi::PhGI2 plants, delayed flowering was accompanied by a reduction in the total 286 number of flower buds at the end of the flowering period ranging between 32 and 86% 287 compared to WT (Supplementary Information Table S4). Additionally, 38.5% of these 288 buds emerged ectopically as a third lateral organ positioned between the terminal flower 289 and the inflorescence shoot (Fig. 4A-D). Most ectopic flowers aborted prematurely, 290 ultimately resulting in a reduction of fully developed flowers by 35% compared to wild -291 type controls. The corolla diameter was reduced on average by 31% and floral tube length 292 by 13%. This suggested that PhGI2 could have a similar role to PhGI1 preventing floral 293 abortion. Alternatively, the observed phenotypes were the result of cosuppression of 294 PhGI1. 295 The previous assumption that PhGI2 may be involved in suppressing early floral abortion 296 was not confirmed. Indeed, no aborted or ectopic flowers were observed in any of the 297 CRISPR/Cas9-derived alleles or their segregating populations (data not shown). 298 Furthermore, corolla width and tube length were not affected (Supplementary Table S5). 299 These findings suggest that the floral abnormalities are specifically associated with 300 reduced PhGI1 transcript levels due to RNA interference. These data effectively rule out 301 a function of PhGI2 on floral development . Furthermore, as three independent PhGI2 302 alleles exhibited a clear dosage -dependent effect on flowering time, we conclude that 303 PhGI2 plays a central role in regulating flowering time in Petunia. This dosage sensitivity 304 likely reflects a broader biological function of GI across plant species (see Discussion). 305 306 PhGI2 is not involved in vegetative growth 307 To determine whether PhGI2 influences vegetative development, we analysed both 308 RNAi-silenced lines and CRISPR/Cas9-induced mutants. As flowering time differed by 309 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 12 months, we took all measures when the first flowers appeared. RNAi lines exhibited 310 marked alterations in shoot growth as the main stem showed a significant reduction of 311 54.7% (p = 0.0059) (Fig. 5A, Supplementary Table S6). These architectural changes were 312 associated with shortened basal, median and apical internodes. The number of leaves 313 before the first flower was also significantly reduced by 46.7% (0=0.000). 314 In contrast, PhGI2 CRISPR mutants did not exhibit any significant changes in vegetative 315 morphology. Despite the clear dosage -dependent effect on flowering time where WT 316 plants flowered early, heterozygotes intermediate, and homozygous mutants late (Fig. 5 317 B), all late flowering genotypes were morphologically indistinguishable from each other. 318 No differences were observed in plant height, branch number, internode length, or leaf 319 size (Supplementary Information Table S 7). These findings collectively indicate that 320 PhGI2 primarily regulates flowering time, and that vegetative development can be 321 genetically uncoupled from floral induction. 322 323 PhGI2 is not involved in scent control 324 We measured the emission of VOCs from the flowers of four plants of the independent 325 lines 4 and 6 of RNAi::PhGI2, and in wildtype. We distinguished main VOCs 326 characterized by a threshold level of 2% of total VOCs (Table S8), and minor VOCs. 327 RNAi::PhG2 lines showed a reduction on average of 19.42% in total VOC emission (Fig. 328 6A). We analysed rhythmic VOC emission during 24 hours in 3-hour intervals (Fig. 6B). 329 In silenced lines, lowest emission was recorded at ZT9, three hours later than in wildtype 330 flowers. Both wildtype and silenced lines showed peak emission during the dark phase at 331 ZT18. 332 We compared VOC emission between early flowering WT siblings to late flowering 333 homozygous mutants of PhGI2.4 and PhGI2.12 in a 4 -hour sampling period between 334 ZT0-4. There were no significant differences between the samples in the amount of 335 methyl benzoate, main and total VOCs (Fig. 7A). The VOC pattern of the main volatiles 336 excluding methyl benzoate did not change consistently between late and early siblings or 337 the WT (Fig. 7B). Thus, we conclude that PhGI2 is not involved in coordinating quality 338 or quantitative scent emission. The observed decrease in total VOC production and 339 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 13 changes in rhythmic emission in RNAi lines are probably an effect of combined reduction 340 of PhGI1 and PhGI2. 341 342

343 In this work we have performed a comprehensive analysis of PhGI2 and its biological 344 functions. GI expression was originally elucidated in Arabidopsis where it shows a strong 345 circadian oscillation. Most of the transcriptomic studies of circadian clock have been 346 performed using leaves. While the exact peak of expression may not be conserved among 347 plant species, GI is one of the clock genes with a cycling expression pattern (Izawa et al., 348 2011; Berns et al., 2014; Terry et al., 2019a). In wildtype plants, the expression pattern 349 of PhGI2 was characterized by an evening phased peak expression , coinciding with 350 PhGI1 expression and similar to the expression pattern observed for GI orthologs in other 351 species, including Arabidopsis (Fowler et al., 1999), or soybean (Marcolino-Gomes et 352 al., 2014). In contrast to Arabidopsis, where AtGI shows strong oscillation under free 353 running conditions (Park et al. , 1999) , PhGI2 cyclic expression in leaves depend s on 354 photoperiod and is identical to PhGI1 (Fenske et al., 2015; Brandoli et al., 2020). Classic 355 work showed that Arabidopsis hypocotyl elongation , leaf growth and movement are 356 circadian regulated i.e. they are somehow resilient to free running conditions of 357 continuous light or dark (Millar et al., 1995; Apelt et al., 2017). We had previously shown 358 that petunia leaf movement depends on light inputs (Díaz-Galián et al., 2019). As PhGI1 359 and PhGI2 expression is also light dependent, we conclude that the Petunia circadian 360 clock has a different configuration, both in terms of external cues and internal 361 coordination. 362 363 Although we attempted to down regulate PhGI2 in a specific manner, designing RNAi 364 constructs targeting the 3’UTR , this was not achieved and PhGI1 was readily down 365 regulated to over 50%. Classic work in Petunia has shown that RNAi based 366 downregulation may cause a strong cosuppression of genes with high DNA homology 367 (Angenent et al., 1994; Jorgensen, 1995; Paoli et al., 2009). Our data indicates that the 368 common phenotypes found in downregulated plants of PhGI1 or PhGI2 are in fact an 369 effect of PhGI1 silencing and not PhGI2 as most of them are not found in the PhGI2 370 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 14 CRISPR alleles identified. 371 We had previously hypothesized that PhGI1 and PhGI2 expression maybe mutually 372 regulated. Indeed, the PhGI2.12 mutant has a drastic effect on the amplitude and 373 rhythmicity of PhGI1, PhTOC, PhLHY and its own expression. Considering the lack of 374 data concerning GI as a transcriptional regulator, the molecular mechanism of this 375 uncovered autoregulation needs further molecular analysis. Recent work suggests that GI 376 may cooperate with transcription factors in controlling target genes, but the data available 377 is correlational (Siemiatkowska et al., 2022). 378 Despite the dramatic shift in flowering time, vegetative growth remained unaltered in 379 PhGI2 mutants. This is different from Arabidopsis, where late -flowering genotypes, 380 including AtGI, named for its vegetative phenotype, typically display increased biomass. 381 The functions of GI in flowering occur via the FT-TFL1 loci. Extensive research in day-382 neutral tomato and other species has shown that the fine-tuning of FT (downstream of GI) 383 and SP/TFL1/FD regulates determinate growth (Lifschitz et al., 2006; Shalit et al., 2009). 384 Supporting this model, the loss of function of FT and TFL1 paralogs in Petunia, as well 385 as a QTL in common bean ( PvTFL1), has been shown to coordinate flowering and 386 vegetative development (González et al. , 2016; Abdulla et al. , 2024) . Importantly, 387 vegetative growth has two layers. One is determinate versus indeterminate growth and 388 the second, lateral organ size. CRISPR -generated lines PhGI2.12, PhGI2.4, and 389 PhGI2.19 reveal that in Petunia × hybrida, flowering time and lateral organ size can be 390 genetically uncoupled. Although increased biomass may correlate with delayed 391 flowering, it is not directly caused by it. Moreover, PhGI1 appears to have undergone 392 subfunctionalization to repress vegetative growth, while PhGI2 retains a role in 393 promoting floral transition. These findings support the idea that floral induction and 394 reduced vegetative growth are parallel, rather than causally linked, developmental 395 processes. They also highlight the two distinct aspects of vegetative growth, lateral organ 396 size and growth termination, as genetically and functionally separate components. 397 398 It is worth noting that homozygous PhGI2 CRISPR alleles may take up to 140 days vs 70 399 to flower, compared to segregating WT siblings. The GI locus is involved in control of 400 floral transition in many other species. The quantitative function of GI appears to be 401 conserved in a variety of plants as OsGI has a dosage effect in rice under field conditions 402 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 15 (Izawa et al. , 2011) . Furthermore, in soybean GmGI has three paralogs and they are 403 involved in domestication and adaptation to different environmental conditions (Wang et 404 al., 2023). Our finding of a dosage effect of PhGI2 on flowering time indicates that it is 405 a major regulator of this trait across land plants. 406 407 In Petunia x hybrida, the emission of floral fragrance is dominated by benzenoids . The 408 lack of consistent changes in VOC profiles in PhGI2.4 and PhGI2.12 suggest that other 409 clock components may be involved in coordinating scent emission. Indeed work 410 performed in N.attenuata, snapdragon and Petunia shows that LHY and ZTL/CHL are 411 responsible for the circadian regulation of scent (Fenske et al., 2015; Yon et al., 2016; 412 Terry et al., 2019b). 413 Taking all our data together we propose a model where the ancestral GI biological 414 functions include repression of biomass, and induction of floral transition (Fig. 8A). The 415 triple duplication of Solanaceae genomes gave rise to two paralogs in Petunia x hybrida, 416 where PhGI2 maintained the functions controlling floral transition and shared with PhGI1 417 the circadian clock coordination. In contrast the repression of vegetative growth and all 418 the newly uncovered functions in flower development resulted from a 419 subfunctionalization and neofunctionalization of PhGI1 (Fig. 8 B). Coupling of floral 420 transition and vegetative development in Arabidopsis may differ from Petunia in its 421 regulatory coordination as Arabidopsis inflorescences are racemes while Petunia and 422 other Solanaceae have cymose inflorescences (Prusinkiewicz et al., 2007; Rebocho et al., 423 2008). Our work describes for the first time the functional evolution of circadian clock 424 genes in Solanaceae and shows a genetic separation of vegetative growth from flower 425 transition. 426 Acknowledgments 427 The current research was funded by project PID2021 -127933OB-C21 funded by 428 Ministerio de Ciencia, Innovación y Universidades from Spain. 429 The authors declare that no generative AI tools were used to create scientific content, 430 data, or interpretations in this manuscript. Generative AI (ChatGPT) was used 431 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 16 exclusively for language editing to improve clarity and grammar. All authors reviewed 432 and approved the final text. 433 434 Competing Interests 435 None declared 436 Author contributions 437 C.B, design of the research; performance of the research; data analysis, collection, 438 interpretation, figure development, writing the manuscript; MEC design of the research; 439 performance of the research; data analysis, interpretation, writing the manuscript, funding 440 acquisition, project management; JW design of the research; performance of the research; 441 data analysis, collection, interpretation, figure development, writing the manuscript, 442 funding acquisition, project management. 443 444 Data Availability 445 446 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 17

Abdulla MF, Mostafa K, Kavas M. 2024. CRISPR/Cas9-mediated mutagenesis of FT/TFL1 in petunia improves plant architecture and early flowering. Plant Molecular Biology 114, 69. Alique D, Redondo López A, González Schain N, Allona I, Wabnik K, Perales M. 2024. Core clock genes adjust growth cessation time to day-night switches in poplar. Nature Communications 15, 1784. Angenent GC, Franken J, Busscher M, Weiss D, Vantunen AJ, van Tunen AJ. 1994. Co- Suppression of the Petunia Homeotic Gene Fbp2 Affects the Identity of the Generative Meristem. Plant Journal 5, 33–44. Apelt F, Breuer D, Olas JJ, Annunziata MG, Flis A, Nikoloski Z, Kragler F, Stitt M. 2017. Circadian, Carbon, and Light Control of Expansion Growth and Leaf Movement. Plant Physiology 174, 1949–1968. Araki T, Komeda Y. 1993. Analysis of the Role of the Late-Flowering Locus, Gl, in the Flowering of Arabidopsis-Thaliana. Plant Journal 3, 231–239. Benjamini Y, Hochberg Y. 1995. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society: Series B (Methodological) 57, 289–300. Berns MC, Nordstrom K, Cremer F, Toth R, Hartke M, Simon S, Klasen JR, Burstel I, Coupland G. 2014. Evening Expression of Arabidopsis GIGANTEA Is Controlled by Combinatorial Interactions among Evolutionarily Conserved Regulatory Motifs. The Plant Cell 26, 3999–4018. Bombarely A, Moser M, Amrad A, Bao M, Bapaume L, Barry CS, Bliek M, Boersma MR, Borghi L, Bruggmann R. 2016. Insight into the evolution of the Solanaceae from the parental genomes of Petunia hybrida. Nature plants 2, 1–9. Bouget FY, Troein C, Corellou F, Dixon LE, van Ooijen G, O’Neill JS, Millar AJ. 2011. Multiple light inputs to a simple clock circuit allow complex biological rhythms. Plant Journal 66, 375–385. Box MS, Coustham V, Dean C, Mylne JS. 2011. Protocol: A simple phenol-based method for 96-well extraction of high quality RNA from Arabidopsis. Plant methods 7, 7. Brandoli C, Petri C, Egea-Cortines M, Weiss J. 2020. The clock gene Gigantea 1 from Petunia hybrida coordinates vegetative growth and inflorescence architecture. Scientific reports 10, 1–17. Corellou F, Schwartz C, Motta JP, Djouani-Tahri E, Sanchez F, Bouget FY. 2009. Clocks in the Green Lineage: Comparative Functional Analysis of the Circadian Architecture of the Picoeukaryote Ostreococcus. Plant Cell 21, 3436–3449. Díaz-Galián MV, Perez-Sanz F, Sanchez-Pagán JD, Weiss J, Egea-Cortines M, Navarro PJ. 2019. A Proposed Methodology to Analyze Plant Growth and Movement from Phenomics Data. Remote Sensing 11, 2839. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 18 Dong L, Hou Z, Li H, et al. 2022. Agronomical selection on loss-of-function of GIGANTEA simultaneously facilitates soybean salt tolerance and early maturity. Journal of Integrative Plant Biology 64, 1866–1882. Fenske MP, Hewett Hazelton KD, Hempton AK, Shim JS, Yamamoto BM, Riffell J a., Imaizumi T. 2015. Circadian clock gene LATE ELONGATED HYPOCOTYL directly regulates the timing of floral scent emission in Petunia. Proceedings of the National Academy of Sciences 112, 9775–9780. Fowler S, Lee K, Onouchi H, et al. 1999. GIGANTEA: a circadian clock-controlled gene that regulates photoperiodic flowering in Arabidopsis and encodes a protein with several possible membrane-spanning domains. EMBO Journal 18, 4679–4688. Fung-Uceda J, Lee K, Seo PJ, Polyn S, De Veylder L, Mas P. 2018. The circadian clock sets the time of DNA replication licensing to regulate growth in Arabidopsis. Developmental cell 45, 101–113. Glynn EF, Chen J, Mushegian AR. 2006. Detecting periodic patterns in unevenly spaced gene expression time series using Lomb–Scargle periodograms. Bioinformatics 22, 310–316. González AM, Yuste-Lisbona FJ, Saburido S, Bretones S, De Ron AM, Lozano R, Santalla M. 2016. Major Contribution of Flowering Time and Vegetative Growth to Plant Production in Common Bean As Deduced from a Comparative Genetic Mapping. Frontiers in Plant Science 7. Hayama R, Yokoi S, Tamaki S, Yano M, Shimamoto K. 2003. Adaptation of photoperiodic control pathways produces short-day flowering in rice. Nature 422, 719–722. Hecht V, Laurie RE, Vander Schoor JK, Ridge S, Knowles CL, Liew LC, Sussmilch FC, Murfet IC, Macknight RC, Weller JL. 2011. The pea GIGAS gene is a FLOWERING LOCUS T homolog necessary for graft-transmissible specification of flowering but not for responsiveness to photoperiod. The Plant cell 23, 147–61. Helliwell C, Waterhouse P. 2003. Constructs and methods for high-throughput gene silencing in plants. Methods (San Diego, Calif.) 30, 289–95. Hughes ME, Hogenesch JB, Kornacker K. 2010. JTK_CYCLE: an efficient nonparametric algorithm for detecting rhythmic components in genome-scale data sets. Journal of Biological Rhythms 25, 372–380. Izawa T, Mihara M, Suzuki Y, et al. 2011. Os-GIGANTEA confers robust diurnal rhythms on the global transcriptome of rice in the field. The Plant cell 23, 1741–1755. Jorgensen RA. 1995. Cosuppression, Flower Color Patterns, and Metastable Gene-Expression States. Science 268, 686–691. Kubota A, Kita S, Ishizaki K, Nishihama R, Yamato KT, Kohchi T. 2014. Co-option of a photoperiodic growth-phase transition system during land plant evolution. Nature Communications 5, 3668. Labun K, Montague TG, Krause M, Torres Cleuren YN, Tjeldnes H, Valen E. 2019. CHOPCHOP v3: expanding the CRISPR web toolbox beyond genome editing. Nucleic acids research 47, W171–W174. Lifschitz E, Eviatar T, Rozman A, Shalit A, Goldshmidt A, Amsellem Z, Alvarez JP, Eshed Y. 2006. The tomato FT ortholog triggers systemic signals that regulate growth and .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 19 flowering and substitute for diverse environmental stimuli. Proc Natl Acad Sci U S A 103, 6398–6403. Linde A, Eklund DM, Kubota A, et al. 2017. Early evolution of the land plant circadian clock. New Phytologist 216, 576–590. Liu XL, Covington MF, Fankhauser C, Chory J, Wanger DR. 2001. ELF3 encodes a circadian clock-regulated nuclear protein that functions in an Arabidopsis PHYB signal transduction pathway. Plant Cell 13, 1293–1304. Mallona I, Weiss J, Egea-Cortines M. 2011. pcrEfficiency: a Web tool for PCR amplification efficiency prediction. BMC Bioinformatics 12, 404. Manchado-Rojo M, Delgado-Benarroch L, Roca MJ, Weiss J, Egea-Cortines M. 2012. Quantitative levels of Deficiens and Globosa during late petal development show a complex transcriptional network topology of B function. The Plant Journal 72, 294–307. Manchado-Rojo M, Weiss J, Egea-Cortines M. 2014. Validation of A integumenta as a gene to modify floral size in ornamental plants. Plant biotechnology journal 12, 1053–1065. Marcolino-Gomes J, Rodrigues FA, Fuganti-Pagliarini R, Bendix C, Nakayama TJ, Celaya B, Molinari HBC, de Oliveira MCN, Harmon FG, Nepomuceno A. 2014. Diurnal Oscillations of Soybean Circadian Clock and Drought Responsive Genes. (N Cermakian, Ed.). PLoS ONE 9, e86402. McClung CR. 2006. Plant circadian rhythms. The Plant cell 18, 792–803. Millar AJ, Straume M, Chory J, Chua N-H, Kay SA. 1995. The Regulation of Circadian Period by Phototransduction Pathways in Arabidopsis. Science 267, 1163–1166. Mizoguchi T, Wright L, Fujiwara S, et al. 2005. Distinct roles of GIGANTEA in promoting flowering and regulating circadian rhythms in Arabidopsis. The Plant cell 17, 2255–70. Nusinow DA, Helfer A, Hamilton EE, King JJ, Imaizumi T, Schultz TF, Farré EM, Kay SA, Farre EM. 2011. The ELF4-ELF3-LUX complex links the circadian clock to diurnal control of hypocotyl growth. Nature 475, 398-U161. Paoli ED, Dorantes-acosta A, Zhai J, Paoli EDE, Dorantes-acosta ANA, Accerbi M. 2009. Distinct extremely abundant siRNAs associated with cosuppression in petunia Distinct extremely abundant siRNAs associated with cosuppression in petunia. 1965–1970. Park DH, Somers DE, Kim YS, Choy YH, Lim HK, Soh MS, Kim HJ, Kay SA, Nam HG. 1999. Control of circadian rhythms and photoperiodic flowering by the Arabidopsis GIGANTEA gene. Science (New York, N.Y.) 285, 1579–82. Perez-Sanz F, Ruiz-Hernández V, Terry MI, Arce-Gallego S, Weiss J, Navarro PJ, Egea- Cortines M. 2021. gcProfileMakeR: An R Package for Automatic Classification of Constitutive and Non-Constitutive Metabolites. Metabolites 11. Pfaffl MW, Horgan GW, Dempfle L. 2002. Relative expression software tool (REST(C)) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucl. Acids Res. 30, e36-. Prusinkiewicz P, Erasmus Y, Lane B, Harder LD, Coen E. 2007. Evolution and development of inflorescence architectures. Science 316, 1452–1456. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 20 Rebocho AB, Bliek M, Kusters E, Castel R, Procissi A, Roobeek I, Souer E, Koes R. 2008. Role of EVERGREEN in the development of the cymose petunia inflorescence. Dev Cell 15, 437–447. Rédei GP. 1962. Supervital Mutants of Arabidopsis. Genetics 47, 443–460. Sawa M, Kay SA. 2011. GIGANTEA directly activates Flowering Locus T in Arabidopsis thaliana. Proceedings of the National Academy of Sciences of the United States of America 108, 11698–11703. Sawa M, Nusinow DA, Kay SA, Imaizumi T. 2007. FKF1 and GIGANTEA complex formation is required for day-length measurement in Arabidopsis. Science 318, 261–265. Schmittgen TD, Livak KJ. 2008. Analyzing real-time PCR data by the comparative CT method. Nature Protocols 3, 1101–1108. Shalit A, Rozman A, Goldshmidt A, Alvarez JP, Bowman JL, Eshed Y, Lifschitz E. 2009. The flowering hormone florigen functions as a general systemic regulator of growth and termination. Proc Natl Acad Sci U S A 106, 8392–8397. Siemiatkowska B, Chiara M, Badiger BG, et al. 2022. GIGANTEA Is a Negative Regulator of Abscisic Acid Transcriptional Responses and Sensitivity in Arabidopsis. Plant and Cell Physiology 63, 1285–1297. Staiger D, Shin J, Johansson M, Davis SJ. 2013. The circadian clock goes genomic. Genome Biology 14, 208. Terry MI, Carrera-Alesina M, Weiss J, Egea-Cortines M. 2019a. Transcriptional Structure of Petunia Clock in Leaves and Petals. Genes 10, 860. Terry MI, Pérez-Sanz F, Navarro PJ, Weiss J, Egea-Cortines M. 2019b. The snapdragon LATE ELONGATED HYPOCOTYL plays a dual role in activating floral growth and scent emission. Cells 8, 920. Wang L, Li H, He M, Dong L, Huang Z, Chen L, Nan H, Kong F, Liu B, Zhao X. 2023. GIGANTEA orthologs, E2 members, redundantly determine photoperiodic flowering and yield in soybean. Journal of Integrative Plant Biology 65, 188–202. Watanabe S, Xia ZJ, Hideshima R, et al. 2011. A Map-Based Cloning Strategy Employing a Residual Heterozygous Line Reveals that the GIGANTEA Gene Is Involved in Soybean Maturity and Flowering. Genetics 188, 395-U260. Wu G, Anafi RC, Hughes ME, Kornacker K, Hogenesch JB. 2016. MetaCycle: an integrated R package to evaluate periodicity in large scale data. Bioinformatics 32, 3351–3353. Yon F, Joo Y, Cortés Llorca L, Rothe E, Baldwin IT, Kim S-G. 2016. Silencing Nicotiana attenuata and alters circadian rhythms in flowers. New Phytologist 209, 1058–1066. Zakhrabekova S, Gough SP, Braumann I, et al. 2012. Induced mutations in circadian clock regulator Mat-a facilitated short-season adaptation and range extension in cultivated barley. Proceedings of the National Academy of Sciences of the United States of America 109, 4326– 4331. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 21 Figure legends Fig.1. Mutations in PhGI2 obtained by CRISPR/Cas9. ( A) Sequence alignment of the GI2 exon4 region. Deletions and insertions are marked in green. ( B) Positions in the protein mutated in PhGI2.4, PhGI2.12 and PhGI.19. Amino acids marked in green correspond to the new sequence resulting from the frame shift. ( C) Conserved region in GI proteins across the plant kingdom. Fig. 2. Effect of knocking down and knocking out PhGI2 on the expression of (A,B) PhGI2, (C,D) PhGI1, (E,F) PhCHL, (G,H) PhLHY, (I,J) PhTOC and (K,L) PhELF4 during a 24 hour period in CRISPR/CAs9 PhGI2 line 12 and in iRNA::PhGI2 line 4.2, compared to expression in the wild-type. Expression represents the normalized expression NE according to the formula (NE) = 2^-(Ct experimental – Ctn). Three samples were analyzed for each time point and error bars indicate the standard deviation. Asterisks indicate statistical significance between wildtype and iRNA lines with *P < 0.05; **P < 0.01; ***P < 0.001 according to group-wise comparison with to Students T-test. Figure 3. Effect of PhGI2 downregulation and knockout on flowering time. ( A) Percentage of fully open flowers in weeks after transplanting from in vitro culture to substrate of T1 and T2 generation of iRNA::PhGI2 lines compared to wild type plant. (B) Segregation of flowering time from CRISPR/Cas9-GI2 lines 4, 12,and 19. Values are the average and deviation of 6 early flowering, mid flowering and late flowering plants. Letters indicate significant differences for each line according to Students T-test. Fig. 4. Effect of downregulation of PhGI2 on flower development. (A) Wild type Petunia flower. (B) RNAi::PhGI2 inflorescence showing ectopic flowers and floral abortion. (c) Inflorescence showing aborted flowers with distinct pedicel death (d) Close up of early aborted flower. Fig. 5. Effect of downregulation and knockout of PhGI2 on vegetative development. (A) Vegetative growth characteristics in WT (left) and iRNA::PhGI2 (right) plants under growth chamber conditions of 16 hours light/8 hours darkness. ( B) From left to right, heterozygote, homozygous PhGI2.12 and two WT plants grown in the greenhouse. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 22 Fig. 6. Volatile emission by flowers from wild type and iRNA::PhGI2 T1 lines. Flowers were excised at ZT0. ( A) Total VOC emission in wild type flowers compared to iRNA::PhGI2 lines in 24 hours and (B) VOCs emission In three hour intervals during 24 hours. Absolute total emission of VOCs per grams of fresh weight is given as sum of integrated peak area. Asterisks indicate statistical significance between wild type and iRNA lines with *P < 0.05; **P < 0.01; ***P < 0.001 according to Student’s T-test. Fig. 7. Effect of PhGI2 knockout on VOC emission. ( A) Amount of methyl benzoate, main VOCs and total VOCs and ( B) amount of main VOCs of wild type flowers and homozygote PhGI2.12 and PhGI2.4 flowers. Emission was recorded during 4 hours from ZT 0 to ZT4. and calculated based on the integrated peak area divided by flower fresh weight. Fig. 8. ( A) Model for the ancestral GI gene function and ( B) model for function of GI paralogs in Petunia after a genome triplication. PhGI2 retained the floral induction and circadian functions while PhGI1 retained growth repression and evolved new roles in flower development. Supplementary information Fig. S1 Multiple sequence alignment of GI proteins from land plants. Fig. S2 Expression of (a) PhGI2, (b) PhGI1, (c) PhZTL, (d) PhLHY, (e) PhTOC and (f) PhELF4 in iRNA::PhGI2 lines 4.4, 6.1 and 6.3 compared to expression in the wild-type, during a 24 hour period. Table S1 PCR primers. Table S2 Statistical analysis of rhythmicity of PhGI1 and PhGI2 gene expression data under normal (LD) and free running conditions (DD). Table S3. Statistical analysis of rhythmicity of gene expression data in RNAi lines Table S4. Statistical analysis of rhythmicity of gene expression data in PhGI2.12 mutant Table S5 Segregation of three CRISPR/Cas9 alleles of PhGI2 for flowering time. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint 23 Table S6 Comparison of floral parameters between Wild-Type and silenced PhGI2. Table S7 Comparison of floral parameters between WT (early flowering). mid (heterozygous) and late (homozygous) flowering plants of Crispr/Cas9 PhGI2 lines 12. 4 and 19. Table S8 Vegetative parameters of iRNA::PhGI2 in T2 generation. Table S9 Comparison of vegetative parameters between late flowering Crispr/Cas9 Gi2 mutants (late flowering) and early and medium flowering siblings. Table S10 Main Volatile Organic Compounds analyzed. .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted January 25, 2026. ; https://doi.org/10.64898/2026.01.20.700548doi: bioRxiv preprint