Rice LEAFY COTYLEDONE1 regulates embryonic envelope development and chlorophyll biogenesis in embryo

The scutellum, coleoptile, coleorhiza, and epiblast (if it exists) consist of a complex embryonic 20 envelope to protect the plumule and radicle inside a grass embryo. Controversies have been 21 provoked for centuries regarding homologies of the grass embryonic structures. Here we found 22 that the rice LEAFY COTYLEDONE1 ( LEC1 ) gene, OsNF-YB7 , is vital for embryo development. A 23 leaf-like structure (LL) was developed from the scutellum of osnf-yb7 to replace the embryonic 24 envelope that formed in wild-type. Additionally, osnf-yb7 developed chloroembryos due to 25 overactivated chlorophyll biosynthesis. Thus, OsNF-YB7 likely plays a dual role in chlorophyll 26 biogenesis in rice embryos: (1) OsNF-YB7 directly represses genes, such as rice GOLDED-LIKE1 27 ( OsGLK1 ), involving chlorophyll biosynthesis; (2) OsNF-YB7 binds to OsGLK1 to repress the 28 downstream genes of OsGLK1. Parallel phenotypes shown in osnf-yb7 and lec1 suggest functional 29 conservation of the LEC1-type genes in plants. Both lec1 cotyledons and osnf-yb7 LL displayed 30 true leaf characteristics. Our morphological and transcriptional evidence implied that LL replaces 31 the embryonic envelope in osnf-yb7 , raising the hypothesis that the grass embryonic envelope is 32 an analog of Arabidopsis cotyledon. This study demonstrates that OsNF-YB7 acts as a negative 33 regulator in chlorophyll biogenesis and is important for embryonic envelope formation.


Introduction 36
LEAFY COTYLEDON1 (LEC1) of Arabidopsis encodes a subunit of the nuclear family Y (NF-Y) 37 transcription factor (TF) with specialized function in plants (Lotan et al., 1998;Lee et al., 2003).  Table S8). Altogether, we assumed that the breakdown of osnf-yb7 dormancy was caused by 240 activating GA biosynthesis, coupled with the repression of the ABA-induced dormancy signals. 241

OsNF-YB7 is essential for the acquisition of desiccation tolerance 242
Although osnf-yb7 embryo is more active after embryogenesis ( Fig. 3G-K), the TTC staining 243 analysis showed that mature seeds of osnf-yb7 were less vigorous than the WT after imbibition 244 for 12 h ( Fig. 4G and H). Therefore, most of the dehydrated osnf-yb7 seeds are not germinable 245 (Niu et al., 2021). However, we found that if the seeds were not desiccated, osnf-yb7 could 246 germinate. Furthermore, using the seeds harvested before drying ( 25 DAF), osnf-yb7 247 germinated much faster than WT seeds (Fig. 4I). However, the subsequent seedling development 248 of osnf-yb7 was arrested (Fig. 4J). 249 The findings suggested that OsNF-YB7 is required for the acquisition of desiccation tolerance in 250 rice. In favor of this, the genes encoding late embryogenesis abundant (LEA) proteins that 251 accumulate during late stages of embryogenesis and associate with dehydration were 252 significantly repressed in the 10-DAF-old mutant embryo. However, there was no expression 253 difference of these genes between the WT and osnf-yb7 embryos at 5 DAF (Fig. S5A). Among 254 these LEA-encoding genes, OSEM, Rab16A, and REG2 are marker genes activated at the Em8 255 phase (Itoh et al., 2005). However, their expressions were largely repressed in osnf-yb7 embryos 256 at 10 DAF ( Fig. S5B-D). 257

OsNF-YB7 represses chlorophyll biogenesis and photosynthesis in the embryo 258
The embryos produced by osnf-yb7 embryos were greenish owing to the activation of Chl 259 biogenesis ( Fig. 5A-D). Intriguingly, the quantification analysis showed that the content of Chlb 260 was not strikingly different between osnf-b7 and WT, but the mutant embryos accumulated 261 more Chla, at either 15 DAF or maturation stage ( Fig. 5C and D). Furthermore, chlorophyll 262 autofluorescence could be observed in an osnf-yb7 embryo at as early as 5 DAF ( Fig. S6A and B). 263 Additionally, many genes responsible for rice chlorophyll biogenesis were activated in the 264 mutant embryos at either 5-or 10 DAF ( Fig. S6E and F). In consequence, as revealed by the 265 Mapman analysis, genes involving photosynthesis were dramatically enriched for DEGs, and the 266 vast majority of the genes was upregulated (Fig. S7A-D). 267 Several TFs have been identified in plants for regulating chlorophyll biogenesis or chloroplast 268 development (Jarvis and López-Juez, 2013). Some TFs, such as rice GOLDEN 2-like1 (OsGLK1), 269 Carbon-metabolism-involved (OsGNC), were significantly upregulated in the osnf-yb7 embryos; 271 among which, OsGLK1 was the most activated gene (Fig. 5E-H). Overexpression of OsGLK1 can 272 significantly increase chlorophyll content in rice (Nakamura et al., 2009). By surveying the genes 273 upregulated in an OsGLK1 overexpression line (Nakamura et al., 2009), we found that most of 274 the OsGLK1 activated genes ( 70.2%) were also upregulated in the osnf-yb7 embryos at either 5-275 or 10-DAF (Fig. 5I). To explore the possibility that OsNF-YB7 directly represses the expression of 276 OsGLK1, we conducted a dual-luciferase (LUC) assay to test the hypothesis. The results showed 277 that OsNF-YB7 significantly repressed the activities of the LUC reporter driven by the OsGLK1 278 promoter ( Fig. 5J and K). 279 Meanwhile, we examined the interaction between OsNF-YB7 and OsGLK using yeast two-hybrid 280 assays (Y2H). We fused full-length OsGLK1 with the GAL4 DNA-activation domain and full-length 281 OsNF-YB7 with the GAL4 DNA-binding domain. The results showed that OsNF-YB7 interacts with 282 OsGLK1 in yeast (Fig. 6A). The proteins also showed strong interactions in the epidermal cells of 283 tobacco, as indicated by the split complementary LUC assay (Fig. 6B). The bimolecular 284 fluorescent complementary (BiFC) analysis suggested that the interaction occurred exclusively in 285 the nuclei (Fig. 6C). Moreover, we transiently coexpressed OsNF-YB7 that tagged with green 286 fluorescent protein (YB7-GFP), and OsGLK1 that tagged with 3x flag (GLK1-flag) in rice protoplast. 287 Also, the co-immunoprecipitation (CoIP) analysis showed that GLK1-flag could be 288 co-immunoprecipitated in rice using the anti-GFP antibody (Fig. 6D). Collectively, these findings 289 suggest that OsNF-YB7 interacts with OsGLK1 both in vivo and in vitro. 290 The primary target genes of GLK1 in Arabidopsis are light-harvesting and Chl biosynthesis-related 291 genes (Nagatoshi et al., 2016). For example, the genes encoding protochlorophyllide 292 oxidoreductase (POR), which catalyzes the reactions of protochlorophyllide (Pchlide) to 293 chlorophyllide for Chl synthesis, were transcriptionally activated by GLK1 in Arabidopsis (Waters 294 et al., 2009). Therefore, OsPORA and LHCB4 (Light-Harvesting Complex B4), a Chla-b binding 295 protein-encoding gene that activated by OsGLK1 in rice , were selected to 296 analyze whether OsNF-YB7 disturbs the OsGLK1-mediated chlorophyll biosynthesis, given their 297 expression was substantially activated in the osnf-yb7 embryos ( Fig. 6E and F). As expected, the 298 transient expression of OsGLK1 in rice protoplast substantially activated the reporter gene driven 299 by the promoters of OsPORA and LHCB4, indicating that these genes are transcriptionally 300 targeted by OsGLK1 ( Fig. 6H-I). When we coexpressed OsNF-YB7 with OsGLK1, the activation 301 ability of OsGLK1 was significantly repressed (Fig. 6H-I). However, OsNF-YB7 alone did not show 302 influences on the expression of the reporter (Fig. 6H-I) OsGLK1, and OsGLK2, respectively, we successfully obtained the osnf-yb7;osglk2 double mutant 310 and the osnf-yb7;osglk1;osglk2 triple mutant at the T 0 generation (Fig. S8). By analyzing embryos 311 the mutants produced, we found that osnf-yb7;osglk2 and osnf-yb7;osglk1;osglk2 displayed 312 embryo morphology similar to that of osnf-yb7 ( Fig. 6J-M). However, the Chl accumulated in 313 osnf-yb7;osglk1;osglk2 was less than in osnf-yb7 and osnf-yb7;osglk2 ( Fig. 6J-M). In comparison 314 to the achlorophyllous embryo of WT, the osnf-yb7;osglk1;osglk2 triple mutant still showed 315 somewhat green coloration in apical part of the embryos (Fig. 6M), suggesting that in addition to 316 OsGLKs, there are some other players contribute to the Chl biogenesis in the embryos of 317 osnf-yb7. 318

Functional conservation of ZmLEC1 genes for embryo development in maize
There are three LEC1-type genes encoded by the maize genome, among which, 320 Zm00001d051697 and Zm00001d017898 are phylogenetically closer to OsNF-YB7, while 321 Zm00001d045772 is a homolog of rice OsNF-YB9 (Fig. 7A). Similar to their rice homologs, 322 Zm00001d051697 and Zm00001d017898 are predominantly expressed in the embryo, whereas 323 Zm00001d045772 is endosperm preferentially expressed in our previous report (E et al., 2018). 324 To detect whether Zm00001d051697 and Zm00001d017898 play a conserved role in embryo 325 development, we generated double mutants of the genes using a CRIPSR/Cas9 gene-editing 326 approach. Because of high similarities between Zm00001d051697 and Zm00001d017898, we 327 could use the same guard RNA to knock out the two genes simultaneously (Fig. S9A). To increase 328 the chance to obtain double mutants, we designed two targets for gene editing. We obtained 329 five independent transgenic events, two of which produced seeds showing similar phenotypes to 330 that of the osnf-yb7 mutants. By sequencing the T 2 individuals showing a phenotype, we found 331 that both Zm00001d051697 and Zm00001d017898 were mutated, while the ones with only one 332 gene-edited showed normal development (Fig. S9B). 333 All seeds produced by line A05-01-1 exhibited embryo defects. We, therefore, used this line for 334 subsequent phenotypic analyses. Like that of osnf-yb7, post-germination development of 335 A05-1-1 was arrested (Fig. 7B). No coleoptile was developed in the mature maize mutant seeds; 336 instead, we could observe a LL structure after imbibition ( Fig. 7C-H). The maize embryo does not 337 have an epiblast, but we could still find that the structure surrounding the plumule-radicle axis at 338 the ventral side was not developed in the lec1 mutant ( Fig. 7C, D, G, and H), resembling the 339 observation that epiblast was disappeared in the osnf-yb7 embryos. Similarly, the coleorhiza of 340 the lec1 mutant in maize was also mal-developed ( Fig. 7C and D). However, some of the dried 341 mutant seeds were germinable and can develop into a fertile plant (Fig. S9C). A close observation 342 of the germinated lec1 maize seeds suggested that a LL structure replaced the coleoptile, leaving 343 the inside leaves unprotected (Fig. 7I-L). The cross-sections of germinated maize embryos 344 showed that the LL structure of the lec1 mutant is a scutellum derivative ( Fig. 7K and L), 345 consistent with the phenotype that showed in the rice mutant. Notably, the maize lec1 embryos 346 were achlorophyllous ( Fig. 7C-H). This was possible because multiple layers of bracts inhibited 347 light from penetrating the ear. In favor of this hypothesis, blocking light perception by covering aluminum foils on a rice panicle resulted in achlorophyllous embryos of osnf-yb7 (Fig. S10). 349 Altogether, phenotypical similarities between the rice and maize lec1 mutants strongly suggested 350 that the LEC1-type genes are functionally conserved for grass embryo development. by investigating the expression profiles in the osnf-yb7 endosperm at 10 DAF, we found that only 368 a small number of genes showed differential expression (88 downregulated and 136 upregulated) 369 in comparison to the WT, further confirming that OsNF-YB7 and OsNF-YB9 are confined to 370 different compartments (i.e., embryo and endosperm, respectively) for function ( Fig. S11B and 371 Table S9). However, we did find expression changes of some genes. For example, OsNF-YB9 was 372 significantly activated in the osnf-yb7 endosperm, while several genes in response to auxin were 373 downregulated (Fig. S11C). 374 We previously reported that OsNF-YB7 and OsNF-YB9 could complement the lec1 defects in 375 Arabidopsis (Niu et al., 2021). Here we found that almost all developmental defects showed in lec1 could be observed in osnf-yb7, including loss of quiescence, weakened dormancy, and 377 desiccation intolerance (Figs. 3 and 4). Embryogenesis defects of lec1 and osnf-yb7 were also 378 comparable. Both LEC1-type mutants showed over-proliferation of suspensors; the hypocotyl of 379 the mutants was not as bent as the WT (Fig. 1I-T). A heterochronic conversion was induced in 380 lec1, for whose cotyledons acquired characteristics displayed in a true leaf (Meinke et al., 1994;381 West et al., 1994). Likewise, an LL was derived from the osnf-yb7 scutellum at the corresponding 382 regions where WT's coleoptile and epiblast originated. Therefore, we believed that the LL of 383 osnf-yb7 is not a coleoptile analog, but more like a true leaf in several aspects ( Fig. 2A-L): (1) 384 trichomes were developed at the tip and on the surface of the LL structure; (2) cell arrangement 385 of the LL structure was different from that of the coleoptile; (3) multiple vascular bundles were 386 developed rather than two. This evidence suggested a leaf identity of LL. Additionally, the maize 387 lec1 mutants, we generated also displayed similar embryo morphology: the coleoptile was 388 substituted by a structure similar to foliage (Fig. 7B-L). 389

Reconsideration of the homologies of complex embryo structures in grass 390
Several distinct concepts have been proposed to interpret what the grass embryonic structures 391 represent. The scutellum is considered equivalent to the cotyledon, either entire or partial of it 392 (Brown, 1965(Brown, , 1960Xu et al., 1999;Chandler, 2008;Satoh et al., 1999). However, no agreements 393 have been achieved regarding the homologies of the epiblast and coleorhiza, which have been 394 debated for centuries (see reviews Brown, 1965Brown, , 1960. The homology of coleoptile is a more 395 controversial topic. The grass coleoptile has been proposed as a part of the cotyledon, a leaf 396 (first, second, or third leaf), a leaf sheath, or an innovation of grass that does not have a 397 counterpart in the dicot embryo (see reviews Brown, 1965Brown, , 1960. 398 The coleoptile primordium was formed from the ventral side of the scutellum; the epiblast and 399 coleorhiza, and the lateral and ventral scales that are covering the coleoptile, are directly derived 400 from the scutellum, and there was no primordium differentiated for forming these structures (Xu 401 et al., 1999;Itoh et al., 2005;Brown, 1960Brown, , 1965. Mutants that showed specific embryo defects, 402 such as osnf-yb7, may provide new insights for the debating. The entire embryonic envelope 403 development was impaired in osnf-yb7: coleoptile, coleorhiza, and epiblast were completely lost ( Fig. 1I-T); the scutellum accumulated less storage reserves, and the palisade-shaped epithelial 405 cells were not differentiated (Fig. S2A-D). A LL structure (with somewhat foliage identities) was 406 developed from where the coleoptile and epiblast originated. LEC1 defects resulted in a 407 heterochronic conversion for the Arabidopsis cotyledons. Given the parallel phenotypes of lec1 408 and osnf-yb7, we inferred that the structures of WT that transformed into the LL structure in 409 osnf-yb7 are equivalent to the cotyledons in Arabidopsis (Fig. S12 A-C). For this consideration, 410 the coleoptile and epiblast are most likely the cotyledon analogs in rice. The epiblast is an 411 extension of the coleorhiza (Brown, 1960(Brown, , 1965 and the midrib of the LL structure is extended 412 from the main scutella vascular bundle (Fig. 1Q-T). Considering these, we propose a hypothesis 413 that the entire embryonic envelope is the cotyledon of grass (Fig. S12 A-C). The expression 414 profiles of the laser-microdissected rice embryo tissues favor the idea that the embryonic 415 envelope is a continuum sharing similar gene expression. By reanalyzing the data generated by 416 Sato et al. (2016), we found that mRNA profiles of the scutellum and epiblast/coleorhiza were 417 similar but distinct from those of the shoot and root (Fig. S13). produced by the osnf-yb7 mutants (Figs. 5A-D and S6A and B). These findings indicated that the 431 LEC1-type gene likely acts as a negative regulator in rice, but a positive regulator in Arabidopsis, 432 for Chl biosynthesis.
By surveying the literature, we noticed that Meinke (1992) reported that the cotyledons of lec1 434 mutant remained green unusually late in development. Although there was no significant Chl 435 content difference when using whole seeds for quantification, Parcy et al. (1997) did observe 436 that the tip of lec1 cotyledons accumulates more Chl. Moreover, the lec1;abi3 double mutant 437 embryos produced a much higher Chl than the abi3 single mutant (Parcy et al., 1997). These 438 findings challenge the concept that LEC1 positively regulates Chl biosynthesis and photosynthesis 439 in Arabidopsis. It is worth noting that in the osnf-yb7 embryos, Chl also preferentially 440 accumulated in the tip-end of the LL structure ( Fig. S6A and B). Therefore, LEC1 may play a role 441 Here, we found that OsNF-YB7 interacted with OsGLK1 in vitro and in vivo (Fig. 6A-459   D). OsGLK1 activated the expression of OsPORA and LHCB4, whereas OsNF-YB7 alone showed no 460 impact on these genes (Fig. 6G-I). However, Coexpression of OsGLK1 and OsNF-YB7 in 461 protoplasts repressed the ability of OsGLK1 to activate OsPORA and LHCB4 (Fig. 6G-I). Most of 462 the OsGLK1-induced genes were also activated in osnf-yb7 (Fig. 5I). Therefore, we believe that 463 OsNF-YB7 plays a dual role in regulating Chl biosynthesis and photosynthesis in rice embryos 464 ( Figs. S14A and B). First, it represses the downstream genes achieving by its function as a 465 transcriptional inactivator; second, OsNF-YB7 can interact with TFs such as OsGLK1 to disturb 466 their activation abilities for Chl biosynthesis and photosynthesis-related genes (Figs. S14A and B). 467 Intriguingly, a previous study showed that OsNF-YB2 positively regulates chloroplast 468 development in rice (Miyoshi et al., 2003), which suggests that the LEC1-type and non-LEC1-type 469 NF-Y members can play a distinct role in Chl biogenesis. The spikelets were marked on the day of anthesis for sampling different aged embryos and 488 endosperm. The maize used in this study was grown in the experimental field of Biogle GeneTech in Xishuangbanna, Yunnan Province, China. Because the seeds produced by osnf-yb7 were 490 inviable, the mutants used were propagated asexually by ratooning as described previously 491 (Cheng et al., 2020). The rice and maize seeds were germinated in a growth chamber. The 492 chamber temperature was maintained at 28°C with 12 h/12 h light/dark cycle. 493

Generation of CRISPR/Cas9 mutants 494
The osnf-yb7 mutant lines used in the study were previously generated in our lab (Niu et al., 495 2021 Table S10. 501

Sectioning, staining, and microscopic observation 502
For semithin section preparations, different aged embryos of WT and osnf-yb7 were dissected 503 under a dissecting microscope and were then fixed in FAA solution (60% (v/v) ethanol, 5% (v/v) 504 glacial acetic acid, and 5% (v/v) formaldehyde) and subjected to vacuum pumping for 40 min. 505 After dehydration through an ethanol series and infiltrated with xylene for embedding in resin, 506 the embedded samples were sectioned at 2.5 μm thickness using a rotary microtome (Leica). 507 Sections were then stained with 0.1% toluidine blue, coomassie brilliant blue, or I 2 -KI solution 508 (80-mg KI, 10-mg I2 per ml) and photographed using an Olympus IX71 microscope. 509 The glumes were removed from rice seeds for free-hand sectioning, and then the caryopsis was 510 carefully cut with a double-edged blade. For desiccated mature seeds, the caryopses were 511 soaked in water at 4°C overnight before the experiment. The sections were then photographed 512 or stained with TCC solution (Solarbio) according to the manufacture's protocol. 513 CLSM was performed as described previously (Cheng et al., 2020). Images of the PI-stained 514 embryos were taken using an LSM710 (Zeiss) microscope with excitation/emission wavelengths 515 of 559/619 nm. 516 For SEM analysis, the rice embryos were fixed overnight at 4°C in 2.5% glutaraldehyde in 517 phosphate buffer (0.1 M, pH 7.0), washed three times in the phosphate buffer (0.1 M, pH 7.0) for 518 15 min at each step, then, postfixed with 1% OsO4 in phosphate buffer for 2 h and washed three 519 times in phosphate buffer. The samples were dehydrated through an ethanol series, then 520 transferred to absolute ethanol. The dehydrated sample was coated with gold-palladium using 521 an ion sputterer (EM SCD500, Leica) and imaged using a SEM (GeminiSEM 300, Zeiss). 522

RNA extraction and real-time PCR assay 523
Total RNA was isolated using the RNA-easy Isolation Reagent (Vazyme, R701-01). One microgram 524 of total RNA was used for cDNA synthesis with the First Strand cDNA Synthesis kit (Vazyme, 525 R123-01). Real-time RT-PCR was performed using the SYBR qPCR Master Mix (Vazyme, Q111-02) 526 in the CFX Connect Real-Time PCR Detection platform (Bio-Rad). The experiments were 527 performed using at least three biological replicates. The relative expression level of the tested 528 genes was normalized to the rice Ubiquitin gene (GenBank accession AF184280) and calculated 529 using the 2 ΔCt method. The primers used for qRT-PCR were listed in Table S10. 530

RNA-seq and differential expression analysis 531
Total RNA was extracted from the 5-and 10 DAF embryos and 10 DAF endosperm of the WT and 532 osnf-yb7. Three biological replicates for each sample were set. The qualified samples were 533 submitted to BGI for library preparation and sequencing. The CLC Genomics Workbench 12.

Hormone extraction and quantification 539
First, approximately 0.1-g embryo was collected from the WT and osnf-yb7 for hormone 540 extraction and quantification. Then, the contents of IAA, ABA, GA 1, and GA 4 were measured by 541 high-performance liquid chromatography-tandem mass spectrometry (Agilent, 1290) according 542 to the previously described procedures (Cheng et al., 2020). 543

Yeast two-hybrid assays 544
The coding sequences (CDS) of OsGLK1 and OsNF-YB7 were cloned into the pGADT7 and 545 pGBKT7, respectively. The constructs were cotransformed into yeast strain AH109 using 546 Frozen-EZ Yeast Transformation II kit according to the manufacturer's protocol. The empty 547 pGADT7 and pGBKT7 vectors were cotransformed in parallel as negative controls. The 548 transformants were first selected on synthetic dropout medium (SD/-Trp-Leu) plates. Then, we 549 tested protein-protein interactions using selective SD/-Trp-Leu-His and SD/-Trp-Leu-His-Ade 550 dropout medium. Interactions were observed after 3 d of incubation at 28°C. The primers used 551 for generating these constructs are listed in Table S10. 552

Split complementary LUC assays 553
Split complementary LUC assays were performed as previously described (Niu et al., 2020). 554 The CDS of OsGLK1 and OsNF-YB7 was cloned into JW771 and JW772 vectors to generate 555 nLUC-OsGLK and cLUC-OsNF-YB7, respectively. The constructs were introduced into 556 Agrobacterium tumefaciens strain GV3101 and then co-infiltrated into N. benthamiana leaves, 557 and the LUC activities were analyzed after 48-h infiltration using Tanon Imaging System (5200 558 Multi; Tanon). The primers used for vector construction are shown in Table S10. 559

Bimolecular fluorescence complementary assays 560
The CDS of OsGLK1 and OsNF-YB7 was cloned into the pSPYNE (nYFP) and PSPYCE (cYFP). 561 The prepared plasmids were transformed into Agrobacterium strain GV3101, and the indicated 562 transformant pairs were infiltrated into N. benthamiana leaves. After 48 h after infiltration, the 563 fluorescence signal of yellow fluorescent protein (YFP) was observed with confocal microscopy 564 (Carl Zeiss, LSM 710). Images were captured at 514 nm laser excitation and 519-620 nm emission 565 for YFP. The primers used for vector construction are shown in Table S10. 566

Chlorophyll measurement and confocal imaging 587
One hundred micrograms of embryo of the indicated genotypes were extracted in 3 ml of 100% 588 dimethyl sulphoxide (DMSO) and incubated at 65°C for 1 h. Then, the absorbance values at 648.2 589 and 664.9 nm wavelengths were measured by spectrophotometry, and then total chlorophyll 590 content was calculated (Barnes et al., 1992). 591 Chlorophyll autofluorescence signal was detected using confocal microscopy (Carl Zeiss, LSM 592 710), excitation 633 nm; emission 625-730 nm.