Heat-shock inducible clonal analysis reveals the stepwise establishment of cell fates in the rice stem

The stem, consisting of nodes and internodes, is one of the major organs in seed plants. In contrast to other organs, however, processes of stem development remain elusive, especially when nodes and internodes are initiated. By introducing an intron into the Cre recombinase gene, we established a heat-shock inducible clonal analysis system in a single binary vector and applied it to the stem in the flag leaf phytomer of rice. With detailed characterizations of stem development, we show that cell fate acquisition for each domain of the stem occurs stepwise. Cell fates for a single phytomer and the foot (non-elongating domain at the stem base) were established in the shoot apical meristem by one plastochron before the leaf initiation. The fate acquisition for the node occurred just before the leaf initiation, separating cell lineages for leaves and stems. Subsequently, fates for the axillary bud were established in early leaf primordia. Finally, cells committed to the internode emerged from, at most, a few tiers of cells when the stem epidermis was at the 12∼25 celled stage. Thus, the internode is the last part of the stem whose cell fate is established. This study provides a groundwork to unveil underlying molecular mechanisms in stem development and a useful tool for clonal analysis, which can be applied to various species.

. Alternatively, in plants homozygous for 55 recessive alleles with DNA transposon insertions, spontaneous reversions to the wild type 56 in the presence of autonomous factors were utilized (Dawe and Freeling, 1990). In this 57 case, the revertant sectors with pigmentation can be visualized. By using these methods, 58 numbers and locations of cells in the embryonic shoot meristem contributing to adult 59 organs such as leaves, internodes, ears (axillary branches bearing female inflorescence in 60 maize), and tassels (male inflorescence) have been estimated (Johri and Coe, 1983; the Cre recombinase removes the roadblock by recombining two loxP sites and allows 126 the expression of GUS reporter. Because these recombination events occur by chance, the 127 GUS reporter is activated in certain cells randomly, and such cells will generate clonal 128 GUS-positive sectors. 129 To save time and effort due to two rounds of transformation, we aimed to combine 130 the components into a single binary vector (Figure 2, A and B). It will be significant, 131 especially in crop species with longer life cycles. Besides, we replaced the 35S promoter 132 with the maize ubiquitin (UBQ) promoter because we previously found that the 35S 133 promoter is often silenced and the UBQ promoter shows much more stable activity in rice 134 (Tsuda et al., 2022). We also inserted a GFP coding sequence at the C terminus of GUS 135 to allow non-destructive monitoring of spontaneous reporter activation. Initially, we tried 136 to transfer this proUBQ-loxP-tpCRT1-loxP-GUS-GFP fragment into the Cre-containing 137 binary vector, but it was unsuccessful. When we digested the resultant plasmid, the band 138 pattern indicated that the plasmid lacked the tpCRT1 roadblock ( Figure 2C). This is 139 possibly due to the misexpression of Cre recombinase in Escherichia coli and unwanted 140 8 loxP recombination. Therefore, we inserted the first intron of the castor bean catalase 141 gene cat-1 into the Cre coding region. This intron is widely used in the intron-GUS 142 reporter gene in binary vectors (Tanaka et al., 1990). As we expected, the resultant 143 plasmid pHS_iCre_LGG_ver.1 (LGG stands for lox, GUS and GFP) showed a predicted 144 band pattern after restriction digestion, indicating that the introduction of the intron 145 stabilized the plasmid structure ( Figure 2C). 146 The initial version of the intron-Cre gene had a low splicing efficiency. 147 We introduced pHS_iCre_LGG_ver.1 into rice calli (hereafter we call the transgenic 148 plants LGG_ver.1) and monitored spontaneous reporter activation by observing GFP 149 fluorescence ( Figure 2D). Spontaneous activation in LGG_ver.1 was rare (only one in 34 150 independent transgenic T0 calli). Next, we regenerated transgenic shoots from these calli 151 and tested the induction rate of GUS sectors. Because it has been reported that a heat for 30 minutes. Among 15 independent transgenic T0 lines that we examined, only 9 155 sectors in 5 lines were found. Thus, even in the lines capable of induction, the induction 156 rate was very low: one or two GUS sectors per plant. To determine a possible cause of 157 this low induction rate, we checked the Cre gene expression by RT-PCR. We found that 158 the spliced form of the Cre gene product was very faint, and the majority was in an 159 unspliced form ( Figure 2E). Thus, LGG_ver.1 had a very low induction rate, possibly 160 due to inefficient splicing of the intron introduced into the Cre gene.

161
Adjusting the intron structure improved the splicing efficiency and sector induction 162 rates. 163 To improve the splicing efficiency, we compared the sequences around the splicing donor 164 9 and acceptor sites of the inserted intron with those of the consensus in the rice genome 165 ( Figure 2F) (Campbell et al., 2006). Although the dinucleotides at the donor (GT) and 166 acceptor (AG) sites were the same as the consensus, outside sequences in flanking exons 167 differed (red underbars in Figure 2F). The upstream exons adjacent to the donor site in 168 the consensus frequently end with a dinucleotide "AG", whereas our case had "GG". In 169 addition, the downstream exons in the consensus frequently start with "G", but our case 170 did with "A". This comparison suggested that the sequences around the intron insertion 171 site were not optimal for efficient splicing in rice. It was also possible that the internal 172 sequence of this cat-1 intron was not suitable for efficient splicing in this specific case.

173
Based on these considerations, we shifted the intron insertion site 1 bp to the 5' side 174 to match the flanking exon sequences with those of consensus in the rice genome ( Figure   175 2F). We also replaced the intron of cat-1 gene with that of the rice actin gene, RAc1, 176 which is highly and constitutively expressed (McElroy et al., 1990). We named this 177 second version pHS_iCre_LGG_ver.2 and introduced it into rice calli. The frequency of 178 the spontaneous reporter activation was similarly low as the first version (two in 46 179 independent transgenic T0 calli). Importantly, among the 30 T0 lines which were  To test the performance of this system, we selected three independent lines and examined 186 the induction of the Cre gene in response to varying degrees of heat shock treatments.  PCR showed that the spliced Cre transcript in LGG_ver.2 was induced upon heat shock, 188 although we still detected a significant amount of the unspliced form ( Figure 2E). In line 189 #33, the Cre transcript was undetectable before induction, and the induction level was 190 likely to be the lowest among the three. Upon heat shock treatments, the transcript level 191 increased in the first 15 minutes, and the level further increased with longer incubation 192 and/or at higher temperatures (Figure 2, E and G). Similar tendencies were found in the 193 other two lines (#37 and #18), although they had leaky and higher expression levels 194 ( Figure 2E). 195 Next, we examined the efficiency of sector induction under various conditions 196 ( Table 1). We treated germinating seedlings 3 days after germination (DAG) and stained 197 GUS sectors at 6 DAG. We counted the numbers of seedlings and roots (in both seminal 198 and crown roots) with GUS sectors to evaluate the induction frequency. Importantly, 199 without induction, germinating seedlings of these three lines at T2 generation showed no  highest Cre expression level, the induction rate was even higher (Table 1). Thus, the Cre 210 expression levels and GUS-sector induction rates correlated. In contrast, the induction 211 rate was very low in LGG_ver.1, indicating that the improvement of the intron was 212 essential for efficient induction (Table 1). 213 Sector inducibility in various tissues and organs. 214 Next, we examined whether GUS sectors could be induced in various tissues and organs 215 using LGG_ver.2 #33 at T2 generation. In crown roots, as shown earlier, longitudinal cell 216 files of GUS sectors were often observed (Figure 2, J and K). By closely examining their 217 root apices, we identified sectors induced in the putative stem cell regions ( Figure 2L). 218 In the vegetative shoots, longitudinal cell files and small patches of GUS sectors were 219 found in the young leaf sheath and stem, respectively (Figure 2M, Supplemental Table   220 1). Clonal analyses described in the following sections also showed that GUS sectors 221 could be induced in the SAM during the reproductive transition (Figures 4 and 5). 222 Furthermore, sectors could also be induced in developing panicles and embryos ( Figure   223 2, N-P). Thus, these observations proved the utility of this system in studying various 224 tissues and organs in rice.

225
Characterization of the stem structure in rice 226 Before conducting clonal analyses, we characterized the structure and development of the 227 stem in the flag leaf phytomer (Figures 1 and 3). Upon reproductive transition, the SAM 228 produces the flag leaf and turns into the inflorescence meristem. The internode, whose 229 elongation is limited during the vegetative phase, starts rapid elongation simultaneously 230 in common rice cultivars. Elongating internodes are numbered from the top; the 231 uppermost internode beneath the panicle is internode I, and the second one beneath the 232 flag leaf is internode II (Figure 3A) (Takane and Hoshikawa, 1993;Yamaji and Ma, 2014).

233
Nodes are also numbered in a similar way; the node at the flag leaf insertion (i.e., beneath 234 the internode I) is called node I, and the next node below is node II. The foot, which is a 235 non-elongating domain at the bottom (explained later), was numbered in a similar way in 236 this study; that in between internode II and node II was named foot II ( Figure 1A, 3B). 237 Thus, the flag leaf phytomer consists of the flag leaf, node I, internode II, an axillary bud, 238 and foot II. survival. In rice, there is another non-elongating portion beneath internodes, which has 253 not been well characterized yet (Figure 1, A and B). We recently named this region "foot" 254 (Tanaka et al., 2023). Although the foot shares a central cavity with the above internode 255 at maturity (Figure 3B), it is a chlorophyll-less and non-elongating tissue in which 256 vascular bundles from the axillary bud connect to those of the stem (Figure 1A, 3D,G). 257 The pulvinus surrounds the foot at the leaf base ( Figure 3B). Therefore, it is appropriate 258 to treat the foot as a structure distinct from the internode above.  Because a new leaf is produced per 5 days on average in rice (Itoh et al., 1998), we took 267 tissue samples every 5 days from 10 days before (-10 SD) until 15 days after the transition 268 (+15 SD) from the top four shoots ( Figure 4A). 269 At +5 SD, the flag leaf primordium had just been initiated (P1 primordium in

293
To investigate the timing and the order of cell fate establishment in the stem, we applied 294 our clonal analysis system to the stem development of the flag leaf phytomer. Here, we 295 follow the notion of cell identity acquisition; it is not the lineage but the position of cells 296 relative to their surroundings that determines cellular identities in many aspects of plant 297 development (Poethig et al., 1986;McDaniel and Poethig, 1988;Steeves and Sussex, 298 1989;Scheres, 2001). If a certain sector was confined to a single domain (organ or tissue), 299 we consider that cells within the sector did not proliferate beyond the domain boundaries 300 and hence that the cell fates had been established at the point of induction.

301
Along with the growth scheme of the above-mentioned reproductive transition, we 302 prepared six sample groups at intervals of 5 days ( Figure 4A). The heat shock conditions 303 were at 42℃ for 30 or 45 minutes because treatments of lower temperature or shorter 304 time did not induce enough sectors in the stem (Supplemental Table 2). After the last 305 treatment at +15 SD, we let LGG_ver.2 #33 plants grow for additional 3~4 weeks and 306 harvested culms for GUS staining. Each sample group comprised the top 4 or 5 culms 307 from at least 10 individuals. Because the internode II was much longer than other domains,  Table 2,3).

320
In samples treated at -10 SD, most sectors (91.3%) were confined to the flag leaf 321 phytomer (Table 2 and Figure 5C). Among these, sectors spanning the entire flag leaf 322 phytomer were the most prominent (32.8%), and those across multiple tissue types were 323 also frequent ( Table 2). This indicates that, at two plastochrons prior to the flag leaf 324 initiation, the cell fates for this phytomer but not for specific organs had been largely 325 established in the SAM. At significant frequencies, however, we observed sectors 326 confined to the bottom of foot II (12.5%, Table 3), suggesting that foot II may be the 327 earliest domain whose cell fate is established. Occasionally, we also observed sectors that 328 spanned across two successive phytomers (8.7%, Table 2). This reflects the existence of 329 cells whose fate had not been destined to a single phytomer in the sample population at a 330 certain frequency.

331
In samples treated at -5 SD, sectors that spanned across two phytomers disappeared,  Table 3). Therefore, the fate of the flag leaf phytomer had been 334 completely established but has yet to be for specific organs. In addition, sectors confined 335 to foot II became more evident (20% , Table 3), supporting the hypothesis that foot II is 336 the earliest domain whose cell fate is established.

337
Determination of nodal cell fates split the flag leaf and the subtending stem. 338 In samples treated at 0 SD, sectors that extended for the entire phytomer almost 339 disappeared (1%), and those confined either to the stem or the flag leaf became evident 340 (88.8%, Table 2, Supplemental Table 3 and Figure 5E). Therefore, cell lineages 341 specifically contributing to the flag leaf or the stem had been largely established at the 342 flank of the SAM, or in the P0 region. This divergence of cell lineages for the flag leaf 343 and the stem was accompanied by the frequent emergence of sectors confined in node I 344 or the pulvinus (25.5%, Table 3). These tendencies became more evident in the samples 345 treated at +5 SD ( Figure 5F). Therefore, node I and the associated leaf structure pulvinus 346 may be second domains whose cell fate was established, splitting neighboring cell 347 lineages into the flag leaf and subtending stem. It is noteworthy that, at 0 SD, sectors 348 confined to the axillary bud and internode II were rare and absent, respectively; sectors 349 found in axillary buds often extended to internode II, and those in the internode II always

352
The internode is the last part of the stem whose cell fate was established.

353
In contrast to earlier sample groups, sectors in axillary buds induced at +5 SD hardly  Sectors confined to internode II were first observed at a low frequency (2.8 %) in 361 samples treated at +10 SD ( Figure 5G and Table 3). In the sample group treated at +15 362 SD, most GUS sectors found in internode II (88.8%) did not extend into neighboring 363 tissues anymore ( Figure 5H, Table 3 and Supplemental Table 3). Therefore, cell fate 364 determination for the internode likely occurred between the 12-and 25-celled stages.

365
Notably, the majority of these sectors in internode II spanned more than half of the length 366 of internode II (Figure 5H), suggesting that the number of cells destined for internode II 367 was still very few at +15 SD. These results showed that the internode II was the last tissue 368 whose cell fate was determined in the flag leaf phytomer. Cre exceeded this level. Although the exact mechanism for this requirement of heat shock 390 is unclear, the insertion of an intron likely resulted in a faithful induction system.

391
The frequency of GUS-sector induction increased with higher temperatures and 392 longer treatments. In germinating seedlings, efficient induction of GUS sectors was 393 already observed at 41 o C, and upregulation of Cre was observed within 15 minutes after 394 the treatment. However, in the case of the flag leaf phytomer, the heat shock treatment at 395 42 o C for 30 minutes or longer was required for an efficient induction (Supplemental 396 table 2). This is probably due to the requirement of higher temperatures or a longer time 397 for heat conduction to reach the shoot apices enclosed by adult leaves. Thus, the optimal 398 condition for heat shock depends on how the tissue of interest is exposed. It is important 399 to select the lowest temperature and the shortest duration to avoid undesired side effects 400 such as growth retardation or tissue damage.

401
In every sample group, we observed significant variabilities in the extent that 402 sectors extended. For example, most sectors induced at -10 SD were confined to the flag 403 leaf phytomer, whereas some spanned to the neighboring node II. In axillary buds, 404 although most sectors extended to neighboring organs until +5 SD, those confined to this 405 organ were observed at low frequencies repeatedly (Table 4). Even in a single sample, 406 some sectors extended for more numbers of phytomers/organs/tissue types than their 407 neighboring sectors (arrowheads in Figure 5, B, I, L, M). According to the studies in 408 maize, there are two possible sources of these variabilities (Poethig et al., 1986;McDaniel 409 and Poethig, 1988). First, there must be variances in developmental stages among samples 410 at the point of induction. This is trivial but unavoidable. Second, more importantly, cells  Our experiments showed that the cell fates for distinct parts of the stem are 417 determined stepwise depending on the tissue type (summarized in Figure 6). Around -10 418 20 SD when the leaf primordium two plastochrons before the flag leaf is initiating, the cell 419 fate for the flag leaf phytomer is largely but incompletely established in the SAM, and it 420 becomes complete 5 days later. This is consistent with the observation in maize, in which 421 the fate for a single phytomer is established one plastochron before its initiation (Poethig 422 and Szymkowiak, 1995). In addition, a fraction of the cell population had already been   (Johri and Coe, 1996). Overall, 437 cell fate commitment for non-elongating tissues occurs first, in which the foot is 438 determined earlier than the node, and that for the internode takes place at the end. containing the neighboring SacI and KpnI cloning sites was named pHS_Cre2. 471 We also modified p35S_lox_GUS to achieve stable expression and non-destructive   For sector induction in the internode II, plants at 35,40,45,50,55,and 60 DAG 506 (for -10, -5, 0, +5, +10, and +15 SD treatments, respectively) were incubated in a water 507 bath at 41 or 42 o C for 30 or 45 minutes. To avoid the temporal drop of water temperature 508 at the beginning of the heat shock, up to 3 pots with soil were incubated in 35 L of water 509 at once, and the water temperature was monitored. After the heat shock, plants were 510 cooled down at room temperature for 30 minutes and grown in the growth chamber for 511 additional 3 to 4 weeks until the lamina joint distance between the flag leaf and a leaf 512 below reached 3 to 5 cm. At this stage, the length of internode II ranged from 1 to 8 cm.  After rows in this spreadsheet were sorted by the relative position of sectors, data were 532 plotted using the "heatmap" function in base R.        LGG_ver.  Numbers in parentheses are percentages by total sectors.

780
Sectors confined in node II were not considered.