Multifunctional evolution of B and AGL6 MADS box genes in orchids

We previously found that B and AGL6 proteins form L (OAP3-2/OAGL6-2/OPI) and SP (OAP3-1/OAGL6-1/OPI) complexes to determine lip/sepal/petal identities in orchids. Here, we show that the functional L’ (OAP3-1/OAGL6-2/OPI) and SP’ (OAP3-2/OAGL6-1/OPI) complexes likely exist and AP3/PI/AGL6 genes have acquired additional functions during evolution. We demonstrate that the presumed L’ complex changes the structure of the lower lateral sepals and helps the lips fit properly in the center of the flower. In addition, we find that OAP3-1/OAGL6-1/OPI in SP along with presumed SP’ complexes regulate anthocyanin accumulation and pigmentation, whereas presumed L’ along with OAP3-2/OAGL6-2/OPI in L complexes promotes red spot formation in the perianth. Furthermore, the B functional proteins OAP3-1/OPI and OAGL6-1 in the SP complex could function separately to suppress sepal/petal senescence and promote pedicel abscission, respectively. These findings expand the current knowledge behind the multifunctional evolution of the B and AGL6 genes in plants.

Based on the conserved principle of the perianth (P) code in orchids, the higher-order SP complex (OAP3-1/OAGL6-1/ OAGL6-1/OPI) specifies sepal/petal identity, whereas the L complex (OAP3-2/OAGL6-2/OAGL6-2/OPI) is exclusively required for lip formation 6 . Since the L complex gene OAGL6-2 is also expressed in lateral sepals whereas OAP3-2 is also expressed in petals 6 , these genes may have other, still unknown roles in determining the morphological identity of lateral sepals or petals beyond the P code. Furthermore, L and SP complex genes are continuously expressed throughout flower development 6 , suggesting that OAP3, OPI and OAGL6 may have additional functions during organ morphogenesis. To uncover these questions, we used virus-induced gene silencing (VIGS)-based gene knockdown to comprehensively functionally characterize the genes from the L and SP complexes and their target networks in Phalaenopsis orchids.
Here, we show the existence of L′ (OAP3-1/OAGL6-2/OPI) and SP' (OAP3-2/OAGL6-1/OPI) complexes. The L′ complex has the function in regulating the structure of the lower part of the lateral sepals and helps the lips fit properly in the center of the flower. We also find that B and AGL6 proteins in P-code harbor additional function during orchid flower development. OAP3-1/ OAGL6-1/OPI in SP complexes regulate anthocyanin accumulation and pigmentation whereas OAP3-2/OAGL6-2/OPI in L complexes promotes red spot formation in the perianth. Furthermore, SP complex proteins OAGL6-1 is found to be able to promote pedicel abscission whereas OAP3-1/OPI could function to suppress sepal/petal senescence.

Results
The finding of L′/SP′ complexes in the advanced P code. OAGL6-2 in the L complex is not only highly expressed in the lip but also expressed at a very low level in the lateral sepals of orchid flowers 6 ( Fig. 1a-c). OAGL6-2 is mainly expressed in the lower part of the lateral sepals in different Phalaenopsis varieties (Fig. 1d) (Supplementary Fig. 1a-o) and in various species of orchids ( Supplementary Fig. 2a-r), whereas the SP complexassociated gene OAGL6-1 is expressed evenly in all parts of the lateral sepals (Fig. 1e, f) (Supplementary Fig. 1p-t). Interestingly, OAGL6-2-VIGS not only caused the conversion of the lip into a petal-like structure (Fig. 1k) but also caused the loss of the distinct characteristics such as red spots or a greenish color (Fig. 1b) ( Supplementary Fig. 1b, e, h, k, n) and the curved edge (Fig. 1b, j) ( Supplementary Fig. 1x) on the lower part of the lateral sepals (Fig. 1l, m) (Supplementary Fig. 1y). In addition to converting lip cells into upper sepal epidermal cells (Fig. 1i, middle) (Supplementary Fig. 1w, middle), the epidermis in OAGL6-2-VIGS lower lateral sepals (Fig. 1h, middle) ( Supplementary Fig. 1v, middle) was clearly converted into upper sepal epidermis (Fig. 1g, left and middle) ( Supplementary Fig. 1u, left and middle). This result indicated that OAGL6-2 must function in determining lower lateral sepal identity.
How can OAGL6-2 function in determining lower lateral sepal identity without the presence of its L complex partner OAP3-2? Is it possible that OAGL6-2 interacts with OAP3-1 in the SP complex to form a new L′ complex (OAGL6-2 + OAP3-1) that has a similar function as the L complex? To test this assumption, a FRET analysis for orchid proteins was performed in tobacco leaf cells 1,9 . The result indicated that the OAGL6-2 could not only form a L complex (OAGL6-2 + OAP3-2 + OPI) (Supplementary Fig. 3a) but also likely interacts with OAP3-1 in the SP complex to form a new L′ complex (OAGL6-2 + OAP3-1) ( Supplementary  Fig. 3b). Thus, the lower lateral sepals likely have both high SP (OAGL6-1 + OAP3-1+OPI) and low L′ (OAGL6-2 + OAP3-1 + OPI) complex functions ( Fig. 1n-1). In OAGL6-2-VIGS, the reduced OAGL6-2 in the lower lateral sepals likely caused reduced L′ complex activity and proportionally increased SP complex activity, resulting in a more upper lateral sepal structure/ morphology ( Fig. 1n-2) and turning the whole lateral sepals to a nearly identical dorsal sepal structure (Fig. 1n-3).
In the OAGL6-1-VIGS flowers, only the structure/morphology of the lower lateral sepals was affected (Fig. 1o), in which SP complex activity was reduced and in which the proportion of the L′ complex increased (Fig. 1n-4 Fig. 1z). FRET analysis showed that OAGL6-1 could not only form an SP complex with OAP3-1 (Supplementary Fig. 3c) but also interacts with OAP3-2 (an L complex component) to form the SP' complex (OAGL6-1 + OAP3-2 + OPI) ( Supplementary Fig. 3d).
Our results revealed that OAGL6 proteins are the determining factors that form either the L or SP complex in controlling orchid perianth identity. OAGL6-1 determines SP (with OAP3-1) or SP′ (with OAP3-2) complexes, and OAGL6-2 determines L (with OAP3-2) or L′ (with OAP3-1) complexes. In orchid flowers (Fig. 1q), the dorsal and upper parts of lateral sepals are exclusively regulated by the SP complex. The lower part of the lateral sepals showed a sepal/lip-like morphology since it was regulated by a strong SP complex and a weak L′ complex. Petals are regulated by both SP and SP' complexes. The lips were regulated mainly by the L complex and weakly by L′/SP/SP′ complexes.
One interesting question is why the L′ complex coexists with the SP complex in the lower lateral sepals of orchid flowers. We found that the flat lip-like epidermis ( Fig. 1h-1) altered the structure/morphology and caused a curved edge in both lower lateral sepals (Fig. 1b, j, n-1) ( Supplementary Fig. 1n, x), which created space to fit the lip more properly in the center, without being too prominent (Fig. 1b, j). Thus, the production of the lips should have coevolved with the emergence of the L′ complex in the lower lateral sepals during orchid flower evolution. This structural change does not occur in the dorsal sepals since their position in the flower does not affect the development of the lips.
Different from that observed in dorsal/lateral sepals, a significant and relative weak reduction in full-red pigmentation, downregulation of PeMYB12 expression and accumulation of anthocyanins in petals was observed in OAGL6-1-VIGS (Fig. 2d, left 3, f, i) and OPI-VIGS We also examined the VIGS of Phalaenopsis KA38501, which has yellow/red-orange perianth pigmentation (Supplementary The L′ complex regulates red spot formation. In addition to causing the conversion of lips to sepal/petal-like structures (Fig. 2a, e, left 7), OAGL6-2-VIGS also resulted in a significant reduction in red spots (Fig. 2k, left) in the lateral sepals (Figs. 1l, 2c, left 7, k, right), similar to that observed in PeMYB11-VIGS (Fig. 2c, left 6, k, middle) Phalaenopsis F894 flowers. Interestingly, similar to that which occurred in OAGL6-2 (Fig. 1c), PeMYB11 was also exclusively expressed in the lips (high) and lateral sepals (low) (Fig. 2l), and its expression was significantly downregulated in the OAGL6-2-VIGS flowers (Fig. 2l). This result revealed that OAGL6-2 formed an L′ complex with OAP3-1 in the lateral sepals; in turn, this complex controls the activation of PeMYB11 (Fig. 2j), which has been thought to be associated with red spot formation 22,27 . Silencing OAGL6-2 caused a severe reduction in L ′ complex function, subsequently downregulating PeMYB11 expression ( Fig. 2l) and reducing red spot formation (Fig. 2c, left 7, k, right). Since OAP3-1 was expressed at a much higher level than OAGL6-2 was in the lateral sepals, the reduced amount of OAP3-1 in the OAP3-1-VIGS lateral sepals is still able to form enough numbers of L′ complexes with a low amount of OAGL6-2, with no effect on the red spot formation (Fig. 2c, left 4). This result supported that OAGL6-2 is the major component in the formation of L/L′ complexes in the control of PeMYB11 expression in lateral sepal and lips (Fig. 2j). Red spot formation was not affected in the OAGL6-1-VIGS lateral sepals (Fig. 2c, left 3, 1o), indicating the SP complex is not involved in regulating this process (Fig. 2j).
OAP3-1 and OPI suppress senescence of sepals/petals. In addition to see the alteration in pigmentation, significantly earlier senescence for the dorsal/lateral sepals compared with that of the control (Fig. 3a-c) was observed for the OAP3-1-VIGS ( Fig. 3d-f) F894 flowers. In the controls, the senescence of the sepals and petals occurred simultaneously. The senescence of the OAP3-1-VIGS petals occurred later than that of the dorsal/lateral sepals ( Fig. 3d-f), although it occurred earlier than that of the control F894 petals (Fig. 3g). Similar early senescence for both dorsal/ lateral sepals and petals was observed in OPI-VIGS flowers ( Fig. 3h-j). This early senescence phenotype was not observed in OAGL6-1-VIGS (Fig. 3j), OAP3-2-VIGS or OAGL6-2-VIGS flowers. These results indicated that only the B class proteins (OAP3-1 + OPI) from the SP complex may form an independent senescence-associated complex (S complex) in suppressing sepal/ petal senescence in orchids (Fig. 3k). Since the senescence of the petals occurred later than that of the sepals in OAP3-1-VIGS flowers ( Fig. 3d-f), OAP3-2+OPI from the SP′ complex might form an S' complex in petals with a complementary function in suppressing petal senescence with the SP complex, albeit in a weaker manner (Fig. 3k).
To avoid any influence on flower senescence due to the alteration of flower organ identity in VIGS of OAP3-1/OPI, a VIGS experiment was performed by late infection after flower Fig. 1 The emergence of L' and SP' complexes in orchid flowers revealed modified protein interactions in the P-code model. a A flower of wild-type control (mock) Phalaenopsis F894. LP lips, P petals, DS dorsal sepals, LS lateral sepals. Bars = 15 mm. b Close-up of the lateral sepals from (a), which are divided into a lower part (#1, 2) and an upper part (#3, 4). The red spots (red arrow) and curved edge (blue arrow) were observed in portion #1. c-f The expression of OAGL6-2 (c) and OAGL6-1 (e) in lips (Lip), petals (P), dorsal sepals (DS) and lateral sepals (LS). The expression of OAGL6-2 (d) and OAGL6-1 organs were evaluated during the later stage (VIGS-late). When all the flowers were senescent and had abscised from the OAP3-1-VIGS-late (Fig. 3l, m) or OPI-VIGS-late (Fig. 3l, n) F894 inflorescences, the flowers of the control inflorescences did not yet show senescence (Fig. 3l). A similar early senescence phenotype was also observed in the OAP3-1-VIGS-late and OPI-VIGS-late V3 ( Supplementary Fig. 5) and KA38501 ( Supplementary Fig. 6) Phalaenopsis flowers. This result supports that orchid B class proteins (OAP3-1 + OPI) suppress the senescence of sepals/petals (Fig. 3k).
The expression of OAGL6-1 continuously increased 90 days AFO (Fig. 3o) during flower development, which is opposite that of OAP3-1/OPI, and OAGL6-1 expression only slightly decreased or even increased in OAP3-1/OPI-VIGS-late senescent flowers (Fig. 3q), respectively, further supporting that OAGL6-1 in the SP complex is not involved in the regulation of orchid flower senescence.
In the control abscised pedicel, a fully developed abscission zone (AZ) with cells displaying a fully rounded, intact appearance was observed (Fig. 4l) (Supplementary Fig. 7i). In contrast, upon manual pedicel removal from the senescent flower, broken cells at the fracture plane and a flattened cavity were observed in the OAGL6-1-VIGS pedicel AZs (Fig. 4m). The expression levels of PeBOP1/2, which are necessary for the formation of the floral AZ 40,41 , were significantly downregulated in the OAGL6-1-VIGS pedicel (Fig. 4n) (Supplementary Fig. 7j). Further analysis indicated that small differentiated cells that formed in the AZ of the control flower pedicel (Fig. 4o, Supplementary Fig. 8c, e) were also observed in the junction of the pedicel and stem in the OAGL6-1-VIGS pedicel (Fig. 4p, Supplementary Fig. 8d); however, their formation was clearly altered by the penetration of lignin throughout the pedicel to the stem (Fig. 4p, Supplementary Fig. 8d, f). These results revealed that the defect in abscission of the OAGL6-1-VIGS pedicel was likely due to the strong increase in lignin formation in the pedicel, which tightly connected the pedicel to the stem and subsequently prevented the abscission of the pedicel from the stem. During this process, AZ was likely formed, although its further development and function were altered.
When the expression levels of all the P-code genes in the pedicels were analyzed, the expression levels were OAGL6-1 (SP) = OPI (SP, L) > OAP3-1 (SP) > OAGL6-2 (L), and expression was not detected for OAP3-2 (L) (Fig. 4q). Similar expression patterns for OAGL6-1/OAP3-1/OAP3-2 were also observed in the pedicels of various species of orchids ( Supplementary Fig. 9). Since defects in pedicel abscission were not observed in VIGS Phalaenopsis flowers with the silencing of the other P-code genes (OAP3-1/ OPI/OAGL6-2/OAP3-2), OAGL6-1 is likely exclusively required for the regulation of pedicel abscission. Our results revealed that the OAGL6-1 protein has an additional function to ensure the final abscission of the pedicel through the downregulation of lignin formation and subsequent normal AZ development (Fig. 4r). In the OAGL6-1-VIGS pedicel, OAGL6-1 knockdown caused the upregulation of lignin biosynthetic genes and a high level of lignin in the pedicel, which altered AZ development and resulted in the strong blocking of pedicel abscission.

Discussion
B and AGL6 proteins form L (OAP3-2/OAGL6-2/OPI) and SP (OAP3-1/OAGL6-1/OPI) complexes to determine lip and sepal/ petal identities in orchids 6 (Fig. 5). This study further showed the involvement of the proposed L′ complex (OAP3-1/OAGL6-2/ OPI) in regulating the lower part of the lateral sepals and lip identity/morphology during orchid flower evolution to help the evolved lips fit properly in the center of the flowers (Fig. 5). In addition, the SP complex (OAP3-1/OAGL6-1/OPI) was found to regulate anthocyanin accumulation and pigmentation in sepals/ petals, whereas an SP' complex (OAP3-2/OAGL6-1/OPI) with the complementary function was identified in the petals (Fig. 5). L′/L complexes regulate the formation of red spots in the sepals/ lips (Fig. 5). The roles of B and AGL6 genes in the regulation of flower color have never been experimentally verified in plants.
We found that anthocyanin accumulation is correlated with LMYB12 but not LAGL6/LAP3/LPI expression in lily tepals ( Supplementary Fig. 10), suggesting a possibly unique role for orchid OAGL6-1/OAP3-1/OPI in regulating flower pigmentation during evolution.
Unexpectedly, B functional OAP3-1 and OPI could function separately from SP complex in suppressing the senescence of the sepals/petals of orchids, whereas OAP3-2/OPI with complementary function to that of the OAP3-1/OPI is present in the petals (Fig. 5). The ability of AP3/PI orthologs to regulate flower organ senescence is surprising. A similar downregulation of AP3/ PI expression during petal/tepal development/senescence was also observed in other various species of orchids ( Supplementary  Fig. 11), the dicot Arabidopsis and the monocot lily (Supplementary Fig. 12), revealing the possibly conserved role of AP3/PI orthologs in regulating flower senescence in flowering plants throughout evolution. The finding of the functions for B genes outside of floral organ identity specification is interesting. In addition to the orchid OAP3-1/OPI in this study, it has been reported that mutation in the apple B-function PI ortholog MdPI resulted in the production of parthenocarpic fruit 42 . Thus, evolutionary modification of the functions for the B class genes should occur and remains to be explored in more plant species.
We found that OAGL6-1 in the SP complex also has additional features. The orchid OAGL6-1 could function to promote the abscission of the pedicel (Fig. 5). This function is clearly independent from that of the B proteins OAP3-1/OPI in the SP complex (Fig. 5). Since orchid flowers abscised from the pedicel, which is different from the process in plant species such as Arabidopsis and lily (in which abscission occurs from the base of the flower organs), OAGL6-1 orthologs may have evolved an additional expression pattern in the pedicel to control lignin levels during the evolution of this type of flower and to further develop this additional function to control pedicel abscission. Interestingly, it has been reported that two E-function SEP proteins (SLMBP21 and SEP4/J2) were involved in the regulation of pedicel abscission zone development in tomato 43,44 . Mutations in SLMBP21 or SEP4/J2 caused the lack of the abscission zone on the pedicels (jointless pedicels) of flowers 43,44 . However, the mechanism regulated by OAGL6-1 in orchids is clearly different from that for tomato SLMBP21, since OAGL6-1 suppressed lignin levels in pedicels whereas SLMBP21 affected common meristem activity genes such as LeWUS and LATERAL SUPPRESSOR in the AZ 43 .
In conclusion, a scenario was proposed to elucidate the evolutionary modification of the functions of the B class AP3/PI and AGL6 genes. In the eudicot Arabidopsis, AP3/PI proteins form an SP-like complex with A/E proteins in regulating petal formation. In the monocots lily (Liliales) and orchid (Orchidaceae), the AP3/ PI proteins specifically interact with the AGL6-like protein to form SP-like complexes in lily or SP/SP′L/L′ complexes in orchid to further regulate tepal (lily) or sepal/petal/lip (orchids) identity. AP3/PI proteins also likely evolved to have a conserved function in Arabidopsis, lily and orchid in regulating petal (Arabidopsis), tepal (lily) and sepal/petal (orchid) senescence. During orchid flower evolution, AP3/PI/AGL6 in SP/SP′/L/L′ complexes further developed a function involved in the specific regulation of pigmentation in the perianth, and AGL6-1 in the SP complex acquired an additional function in controlling pedicel abscission, which is absent in Arabidopsis and lily.
Anthocyanin content analysis. Total anthocyanin content in the flower samples was determined according to the spectrophotometric method described previously 50 with slight modifications. Briefly, 100 mg of fresh flower tissues were homogenized in liquid nitrogen and added to 1 ml of 3 M HCl:H 2 O:MeOH (1:3:16, by volume). The crude extracts were agitated gently in the dark for 24 h at 4°C and then centrifuged at 21,000 × g for 15 min. Anthocyanin levels in the supernatant were measured as A 530 -0.24 A 653 by the spectrophotometer TECAN Infinite M200PRO with The i-control microplate reader software (Mannedorf, Switzerland) 51 .
Phloroglucinol-HCl staining. The junctions for the Phalaenopsis flower pedicel and stem were longitudinally cut, whereas 2-mm-thick Phalaenopsis pedicle slices were crossly cut from the base of the pedicel using a blade and put into 24-well plates. The staining solution contained 0.3 g phloroglucinol in 10 ml absolute ethanol and mixed with 5 ml concentrated HCl (37%). The sample slices were filled with staining solution and covered for 5 min, followed by changing the staining solution with water. Fig. 3 The OAP3-1 and OPI B functional proteins regulated sepal/petal senescence. a-c The inflorescence (a), front view (b) and back view (c) of flowers of the control (Mock) Phalaenopsis F894. d-f The inflorescence (d), front view (e) and back view (f) of flowers of OAP3-1-VIGS F894, which showed early senescence of the sepals (red arrows). g The senescence of the OAP3-1-VIGS petal (white arrow) occurred later than did that of the dorsal/ lateral sepals (red arrow). h-i The inflorescence (h) and front view (i) of flowers of OPI-VIGS F894, which showed early senescence of the sepals (red arrow) and petals (white arrow). j The dorsal (DS) and lateral (LS) sepals of control (Mock), OAGL6-1-VIGS, OAP3-1-VIGS, and OPI-VIGS F894 (from left to right) at the same stage. k Summary of the protein complexes that regulate sepal/petal senescence in orchid flowers. The OAP3-1/OPI B functional proteins may form a senescence complex (S), which is exclusively required for the suppression of PeSAG39/PeEDF1/2 expression and senescence in dorsal (DS) and lateral (LS) sepals. S and minor S' (OAP3-2/OPI) complexes have complementary functions in regulating petal (P) senescence. l-n At the same stage, all the flowers were senescent and abscised from the OAP3-1-VIGS-late (l, middle) (m) and OPI-VIGS-late (l, right) (n) inflorescences, whereas the   (g, h).
The arrows indicate the AZ region with the small differentiated cells formed. The cells in the AZ region in (p) were altered by the penetration of lignin from the pedicel to the stem. Bars = 300 µm. In (j, l, m, o, p), each experiment was repeated twice independently with similar results. q Analysis of the expression levels of OAP3-1, OAP3-2, OAGL6-1, OAGL6-2, and OPI in 15 mm and mature flower pedicels of control (Mock) and OAGL6-1-VIGS V3 flowers. The data were derived from one NGS data. r In wild-type orchid flowers, OAGL6-1 proteins may form an abscission complex (A) to promote the abscission of pedicels by suppressing the expression of VND1/MYB46/63/85 and lignin formation and to ensure the normal BOP1/2 expression and subsequent AZ formation. Source data underlying Figs. 4k, 4n, and 4q are provided as a Source Data file.
Cryo-scanning electron microscopy. The procedure used for Cryo-SEM was described in a previous study 6 . Briefly, the perianths were dissected and frozen using liquid nitrogen and transferred to the sample preparation chamber at −160°C. The samples were etched for 10 min at −85°C and were observed under a cryo-scanning electron microscope after gold coating (FEI Quanta 200 SEM, Quorum Cryo System PP2000TR FEI; FEI Company, Hillsboro, Oregon).
FRET analysis. The procedure used for FRET analysis was described in the previous studies 6,18 . Briefly, to fuse OAGL6-1/OAP3-1/OAGL6-2/OAP3-2/OPI with CFP or YFP, the cDNAs for Phalaenopsis OAGL6-1/OAP3-1/OAGL6-2/OAP3-2/ OPI were obtained by PCR amplification using gene-specific primers (Supplementary Table 3) and cloned into the pEpyon-36K and pEpyon-37K vectors upstream of the CFP or YFP sequence under the control of the CaMV 35S promoter. The above constructs were transformed into the Agrobacterium strain C58C1. Different ectopic proteins were expressed in tobacco leaf cells through Agrobacterium-infiltrated N. benthamiana leaves by being infiltrated in 10 mM MgCl2 at room temperature until immersed. A confocal microscope was used to detect the fluorescence signals in the nucleus. The procedure used to visualize fluorophores was described in a previous study 9 . Briefly, an Olympus FV1000 confocal microscope (Olympus FV1000, Tokyo, Japan) and the soft FV-ASW 3.0 software were used to visualize fluorophores and to calculate the raw FRET and FRET efficiency values. The mean value of FRET efficiency in the nucleus was calculated to evaluate the variation in protein interaction distances among different protein complexes (n > 4).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of this work are available within the paper and the Supplementary Information files. A reporting summary for this article is available as a Supplementary Information file. The data sets generated and analyzed during this study are available from the corresponding author upon request. Source data are provided with this paper.