ASYMMETRIC LEAVES1 and REVOLUTA are the key regulatory genes associated with pitcher development in Nepenthes khasiana

Nepenthes develops highly specialized insect-eating organs called pitchers that provide adequate insect-derived nutrients to the plants to offset low nutrient availability in their natural habitat. But so far, the molecular basis of Nepenthes pitcher development remains largely unknown. In an attempt to unravel the underlying mechanisms of pitcher formation, we made morphological observations of the developing N. khasiana leaf and performed RNA-seq to identify genes controlling pitcher development. Histology and scanning electron microscopy photomicrographs show that pitcher formation in N. khasiana occurs early in development and shares anatomical features with the young in-rolled leaf base lamina. Analysis of the RNA-seq data indicated that the modification of the leaf into a pitcher is associated with the altered expressions of leaf polarity genes ASYMMETRIC LEAVES1 (AS1) and REVOLUTA (REV). In fact, both genes displayed exclusive or relatively higher expressions in the tip of the leaf that later developed into a pitcher. We propose that NkAS1 may act to inhibit lamina outgrowth and promote the formation of the tendril. Increased NkREV expression may have been involved in the formation of the N. khasiana pitcher. This dataset will allow further research into this area and serve as the basis for understanding Nepenthes pitcher development.


Results and Discussion
Because Nepenthes pitchers develop at the tips of tendrils 6 , we examined growth and development of the N. khasiana leaf from the initial stages of development, visible to the naked eye (Fig. 1). Leaf development in young N. khasiana plants is initiated as a slender structure with prominent white hairs at the apex, gradually declining in density along one end of the slender structure (Fig. 1a). At this stage, the emerging leaf was seen fairly covered by the preceding leaf base lamina. After a week, the developing leaf became more apparent with increased length and the presence of a pointed structure at the apex called a 'spur' (Fig. 1b). The emergence of the spur in N. khasiana occurred early in development as compared to the one in N. alata 6 , but this could be due to the differences in the age of the plants studied. The developing leaf increases further in length and an opening at another end of the slender structure, lacking the visible white hairs, started appearing (Fig. 1c). At the third week, an outgrowth of the developing leaf base can be seen and the apex region begins to swell, appearing as a small, juvenile pitcher (Fig. 1d). In addition, the intervening region or 'tendril' separating the pitcher from the leaf base became evident at this stage. At the fifth week, leaf base lamina outgrowth continued and became flattened with a slight increase in pitcher size (Fig. 1e). Further increase in the expansion of the leaf base, tendril length and pitcher size was observed at the sixth week (Fig. 1f). Here, differentiation of the pitcher into distinct zones -digestive zone and waxy zone -was visible. As was observed by Gaume et al. 12 in N. alata, a transitional zone was observed between the digestive and waxy zones of N. khasiana. At the seventh week, expansion of the leaf base ceases whereas elongation of the tendril, as well as the pitcher, continued (Fig. 1g). We also noticed the emergence of vertically-oriented structures on one side of the pitcher tube called 'wings' (red arrowhead in Fig. 1g), and the lid at the top of the pitcher became evident at this stage. At the eighth week, the developing leaf showed marked swelling of the pitcher tube and slight increase in tendril length (Fig. 1h). At the ninth week, a considerable increase in pitcher size was observed (Fig. 1i), but the swelling was more pronounced at the bottom of the pitcher, representing the digestive zone. The pitcher lid started detaching, initiating from the top adjacent to the spur (Fig. 1j). At this stage, the peristome can be seen appearing as a ring that surrounds the entrance of the pitcher and arranged in the form of ribs. At the tenth week, the pitcher lid was fully opened (Fig. 1k). Change in pitcher colouration started appearing, most prominently at the waxy zone moving towards the pitcher lid. At this stage, the N. khasiana leaf comprising the leaf base, tendril and the pitcher has attained maturity (Fig. 1l).
We noticed that at a particular period in the growth and development of young N. khasiana plants, developing leaves showed different developmental stages as observed in Fig. 1. In Fig. 2a, a young N. khasiana shoot can be seen developing several leaves, each attaining distinct stages of development. We considered the topmost slender leaf (L1) as stage 1 (Fig. 2a,b) while the leaf (L2) with the expanded leaf base was considered stage 2 (Fig. 2a,c). The leaf (L3) showing complete expansion of the leaf base and the appearance of the pitcher tube characterized stage 3 (Fig. 2a,d) while the leaf (L4) showing elongation of the tendril and expansion of the pitcher tube represented stage 4 (Fig. 2a,e). The leaf (L5) showing pitcher expansion with the lid remaining unopened was considered as stage 5 (Fig. 2a). We usually see this pattern continue for a considerable period of time as the plant grow, but is lost as the plant reaches a certain height. We then examined the ultrastructure of each developmental stage using scanning electron microscopy (SEM) (Fig. 2f-t). Our results show that in stage 1, the apex region is fully covered with hairs (non-glandular trichomes), as can be seen in stages 2 and 3 ( Fig. 2f,i,l). The slender structure of stage 1 at 2-3 cm from the apex is covered with hairs at one end representing the midvein region, while the flat abaxial surface is covered with glandular trichomes (Fig. 2g). Glandular trichomes are also present on the adaxial and abaxial epidermal surfaces of the leaf base lamina of stages 2 and 3, respectively (Fig. 2j,k,m,n). Besides other functions, glandular and non-glandular trichomes are known to offer protection against insect herbivory (Serna and Martin 13 and references therein). As most Nepenthes plants rely on insects for their nutrients, these trichomes may shield young developing leaf tissues from herbivory by visiting insects. The epidermal surfaces of the adaxial and abaxial leaf base lamina of stages 2 and 3 are made up of irregularly shaped cells (Fig. 2j,k,m,n), but in the adaxial epidermis, cells are much larger in size (Fig. 2m). As the leaf develops further to form a tiny pitcher separated from the leaf base by the tendril (stage 4), the outer epidermal surface of the pitcher is still covered with hairs while the inner epidermal surface remained smooth ( Fig. 2o-q). Differentiation of the inner pitcher epidermal cells into the digestive glands, the lunate cells and the nectary glands can be seen only in stage 5 ( Fig. 2r-t).
We then compared SEM images of the leaf base lamina midvein (abaxial side) and the tendril of stages 3-5, and found that both structures are characterized by long, slender epidermal cells, on top of which develop glandular and non-glandular trichomes ( Fig. 2u-w). The density of both glandular and non-glandular trichomes decreases as the leaf mature. This observation suggests that the tendril is an extended structure of the midvein. Is it likely then that the Nepenthes pitcher represents an extension of the leaf base midvein which then expands to form the pitcher tube, as suggested by Juniper and Burras? 7 A comparison between the epidermal cells that make up the tendril/midvein (Fig. 2u-w) and those of the outer (abaxial) epidermal layer of the pitcher (right panels of Fig. 2r,s) suggested otherwise. To confirm, we analyzed the leaf of an in vitro raised Nepenthes plantlet. The SEM images clearly define the distinction between the epidermal cells of the midvein and the pitcher (Fig. S1, Supplementary information). Figure 2h shows that the leaf base lamina outgrowth occurred prior to stage 1, but remained rolled in. In-rolling of the adaxial leaf surface followed by marginal fusion was proposed as one of the mechanisms adopted by Nepenthes plants to modify their leaves 8 . Do pitchers really represent a product of the epiascidiation process? To address this question, we performed cross-sections of the stage 1 leaf, one at the apex and the other at the www.nature.com/scientificreports www.nature.com/scientificreports/ middle portion of the in-rolled leaf base lamina (Fig. 2f,g). The cross-sectioned photomicrograph of the apex region reveals the presence of a hollow structure with two protruding outgrowths at one end reminiscent of the leaf base margins (Fig. 2x,y), indicating that pitcher formation occurs early in development. Cells in the apex region are differentiated into two distinct types separated by a darkly-stained layer of vascular bundles. Cells formed away from the hollow region are bigger in size and irregularly shaped while those cells facing toward the hollow region are smaller in size and rectangularly shaped. Interestingly, both cells are identical, in structure and arrangement, to those present in the leaf base lamina (magnified images in Fig. 2x,y). Thus, our findings indicated that both structures -the apex region (pitcher) and the in-rolled leaf base lamina -share anatomical features. These observations together provide evidence to the assertion by Juniper et al. 8 that the modification of the Nepenthes leaf into a pitcher involves an epiascidiation process followed by marginal fusion. More research is needed to determine how and what drives the fusion of the Nepenthes leaf margins. We adopted the same strategy of selecting the different stages of N. khasiana leaf development, as depicted in Fig. 2a, to investigate transcriptional changes during the development of the highly specialized N. khasiana leaf. In total, we generated around 270 million high quality cleaned paired-end reads (Table S1, Supplementary information). Reads were pooled, normalized and assembled using the freely available software Trinity. De novo assembly yielded 576,563 transcripts, having mean contig length of 730.89 bp and maximum contig length of 21,003 bp with an N50 length of 1,374 bp. All assembled transcripts were found to be of length more than 200 bp (Fig. S2, Supplementary information). The assembled transcripts (≥200 bp) were compared with NCBI non-redundant protein database using BLASTX program. Matches with E-value ≤ 10 −5 and similarity score ≥ 50% were retained for further annotation. Around 60% of the transcripts found using BLASTX have confidence level of at least 1e-5, which indicated high protein conservation (Fig. S3a, Supplementary information). Close to 56% of the assembled transcripts found using BLASTX have similarity of more than 60% at protein level with the existing proteins available at the NCBI database (Fig. S3b, Supplementary information). We then aligned individual reads from each stage to the reference transcriptome to estimate transcript abundance among the five different stages of N. khasiana leaf development. On average, 93.8% of reads were properly aligned to the reference transcriptome with 308,800 unique transcripts possessing FPKM ≥ 1. Out of 308,800 unique transcripts, we identified 103,301 (33.45%) transcripts having significant BLASTX hit against the NCBI non-redundant database, of which 65,535 (21.22%) transcripts matched UniProt proteins. Out of 65,535 transcripts, 7,047 were found to be significantly differentially expressed. A total of 62,020 transcripts were commonly expressed in all five developmental stages (Fig. 3a). Among the uniquely expressed transcripts, stage 1 recorded 275 transcripts followed by stage 3 (28), stage 5 (7), stage 2 (6) and stage 4 (3). Transcripts exclusively expressed in each developmental stage are listed in Table S2, Supplementary information. From the correlation analysis of the five developmental stages of N. khasiana leaf development, stages 1 and 2 showed high correlation (Fig. 3b). This is expected as both stages are of continuous developmental processes and share morphological characteristics, except for the fact that in stage 1 the leaf base lamina is in-rolled while in stage 2, the leaf base lamina is becoming flattened (Fig. 2c,h). Surprisingly, stage 3 showed high correlation with stage 5 rather than stage 4 (Fig. 3b). The emergence of the pitcher tube becomes prominent at stage 3 while in stage 5 the pitcher is highly expanded and elongated (Fig. 2a,d). It is difficult to draw a conclusion from the correlation analysis, but as can be seen in Fig. 4, most transcripts that are highly expressed at stage 4 got reduced at both stages 3 and 5. Real-time qPCR validation of 25 randomly selected significantly differentially expressed genes (DEGs) corroborated the RNA-seq results with an overall Pearson correlation of 0.904 ( Fig. 5 and Table S3, Supplementary information). www.nature.com/scientificreports www.nature.com/scientificreports/ To identify genes enriched in each of the five stages of N. khasiana leaf development defined in the present study, we performed k-means clustering of the significantly DEGs using Cluster 3.0. The DEGs were grouped into 12 clusters (Fig. 4A). We then performed functional enrichment for each cluster using BLAST2GO PRO (Supplementary data S1). The top 10 GO terms for each cluster is represented in Fig. 4B. Table 1 shows a list of DEGs enriched in each cluster (top10 GO terms). A summary of the GO terms and representative genes enriched in each cluster in relation to the five different stages of N. khasiana leaf development is given in Note S1, Supplementary information. www.nature.com/scientificreports www.nature.com/scientificreports/ To identify which of these enriched genes play a role in the development of Nepenthes pitcher, we further dissected the first four developmental stages into different parts to separate the pitcher from the leaf base or tendril (Fig. 6a) and performed RT-PCR to check for their localized expressions (Fig. 6b). We also performed qPCR to accurately quantity the expression of selected genes. The RT-PCR results show two genes, NkLAC and NkAS1, which are expressed exclusively or at higher levels in the pitcher of stages 1-3 (Fig. 6b). In Arabidopsis, LAC is required for lignin polymerization 14 , and studies have reported the presence of lignin in Arabidopsis trichome cell walls 15 . Therefore, enhanced expression of NkLAC in the pitcher may be associated with lignin synthesis in the trichomes covering the apex region (Fig. 2f,i,l). The AS1 gene is expressed during initiation of leaf primordia and acts to exclude KNOX proteins from initiating leaves 16 . Overexpression of AS1 in Arabidopsis Col background causes plants to develop narrower leaves and longer petioles 17 ; in Ler background, however, plants showed stunted growth but developed normal leaves 18 . When AS1 is mutated in the Arabidopsis Ler background, leaf initiation persists; but in severe cases, the first two rosette leaves displayed novel leaf phenotype typical of a lotus leaf, in which the petiole is attached to the abaxial surface of the leaf lamina 18 . Interestingly, flattened leaves of the Ler-background as1 mutant lack a prominent midvein. Similar phenotypes were also observed in Arabidopsis as2 mutants, implying that both AS1 and AS2 function in the same regulatory pathway by forming protein complexes, which in turn act to regulate downstream genes during leaf development 18 . The AS1-AS2 complexes are localized in the adaxial domain and together with ER promote leaf polarity establishment 18 . To specify the role of AS1 in Nepenthes pitcher formation, we examined the expressions of AS2 and ER homologs in N. khasiana. The results show that NkAS2 is expressed in all the dissected tissues with relatively higher expression in the pitcher (Fig. 6b). NkER, on the other hand, displayed exclusive expression in the pitcher of stages 2-4 with almost same levels of expression in stage 1 (Fig. 6b). The qPCR results for NkAS1, NkAS2 and NkER match the results seen in the RT-PCR analysis, with a few exceptions (Fig. S4, Supplementary information). These exceptions include NkAS2 expression in the dissected tissues of stage 4 (higher in tendril) and NkER expression in the dissected tissues of stage 1 (higher in pitcher). Thus, it is safe to say that higher AS1 expression is correlated with an increased expression of NkAS2 and NkER in the pitcher of N. khasiana. But based on the information available for Arabidopsis, it is unlikely that AS1 may play a direct role in the formation of the N. khasiana pitcher.
Further, we show that the expression of NkREV was higher in the pitcher than the leaf base lamina in stages 2-4, with similar levels of expression in the two dissected tissues of stage 1 (Fig. 6b). Similar expression pattern was also observed for NkATHB15; but in the tendril and pitcher of stage 4, NkATHB15 expressions were of the same levels (Fig. 6b). Although the expression of NkATHB15 is of interest, overexpressing ATHB15 in Arabidopsis, however, do not show any phenotype 19 . In Arabidopsis, reports have shown that a single nucleotide change in the putative lipid/sterol-binding START domain of the REV gene results in gain-of-function mutation 20,21 . This gain-of-function mutation causes leaves to become trumpet-shaped. In severe cases, these trumpet-shaped leaves grew out from the abaxial midvein 21 . Unlike gain-of-function mutations in PHB and PHAVOLUTA (PHV) 22 , the inside surface of the trumpet-shaped leaf in gain-of-function rev mutants is adaxial while the outside surface is abaxial, similar to the Nepenthes pitcher. The rev phenotype is caused as a result of the failure on the part of miR165/166 to regulate REV 20 . Failure in miRNA regulation may be due to a reduced level of AGO1, which is known to direct miRNAs to target sequences 23 . Alternatively, AGO10 may compete with AGO1 for miR165/166 to prevent post-transcriptional modification of the REV transcript 24 . In light of these findings, we performed RT-PCR to determine the expression patterns of Nepenthes homologs of AGO1 and AGO10. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Contrary to our expectations, the results show reduced levels of expressions for both genes, in comparison to NkREV expressions, across all the dissected samples (Fig. 6b). To test whether reduced levels of NkAGO1 and NkAGO10 expressions is a result of primer failure to amplify the targeted genes, we increased the cycles of PCR amplification and found that NkAGO1 expression levels increases but remained uniform throughout the different dissected tissues whereas NkAGO10 displayed relatively higher expression in the pitcher parts of stages 2-4 ( Fig. S5, Supplementary information). The qPCR results for NkREV and NkATHB15 expressions across the dissected tissues corroborated the RT-PCR results (Fig. S6, Supplementary information). Barring a few exceptions (higher expression in the pitcher of stage 1 and similar expression in both tissues of stage 4), the qPCR result for NkAGO10 matches the one conducted at increased cycles of RT-PCR amplification (35 PCR cycles). For NkAGO1, the qPCR result indicated higher level of expression in the pitcher than leaf base of stages 1-4, which contradicts with the RT-PCR result (Fig. S6, Supplementary information). It is likely then that other regulatory mechanisms, other than AGO1/AGO10, may be controlling the expression of REV in the developing pitchers of N. khasiana.

Conclusion
In Insectivorous Plants, Darwin imagined that the origin of carnivory in angiosperms is a result of natural selection acting on sticky gland-bearing plants that are then adapted over time to digest the captured prey 25 . This proposition raises the question of what strong selection pressure could have directed the evolution of such extraordinary events of morphological innovation. We now know that the origin of megaphyllous leaves in land plants is linked with a 90% reduction in atmospheric CO 2 and the concurrent increase in stomatal densities to prevent overheating 26 . It can be assumed then that the nutrient-poor habitat in which carnivorous plants grow might have imposed strong selective pressure resulting in the evolution of innovative leaf morphologies. In the event of such an adaptive transformation, recruitment of common developmental programmes evident in typical angiosperm leaves occurred but is altered in a way to generate morphological novelty. In the present study, we identified a number of candidate genes that might play a role in the development of the Nepenthes pitcher. Of these, AS1 and REV are of significant interest as evidenced by higher NkAS1 and NkREV expressions in the developing pitchers of N. khasiana. How these genes are recruited to modify a N. khasiana leaf into a pitcher require further investigation. This involves the use of RNA in situ hybridization techniques, functional validation of candidate genes in Arabidopsis mutants and/or silencing of the candidate genes in N. khasiana by RNA interference.

Methods
Plant material and growth conditions. Nepenthes khasiana plants were collected from their natural habitat located at Jaraiñ, Jaiñtia Hills District, Meghalaya (25° 18.651″ N, 92° 07.786″ E), transferred into pots containing Soilrite Mix (Keltech Energies Ltd, Bangalore) and kept at the greenhouse of the School of Life Sciences, Jawaharlal Nehru University, New Delhi with temperature and humidity maintained at 28 ± 2 °C and 60 ± 5%, respectively. examining growth and development of the N. khasiana leaf. We examined growth and development of the N. khasiana leaf from the initial stages of development visible to the naked eye (Fig. 1). Following the examination, five stages of N. khasiana leaf development were selected for RNA-seq (Fig. 2a-e). This selection is based on the observation that at a particular period in the growth and development of young N. khasiana plants, five distinct stages of leaf development can be seen in different developing leaves (Fig. 2a). The topmost slender leaf (L1) was considered stage 1 while the leaf (L2) with the expanding leaf base was considered stage 2 ( Fig. 2a-c). The third leaf from the top (L3), showing complete expansion of the leaf base and the emergence of the pitcher tube, characterized stage 3 (Fig. 2a,d). The leaf (L4) showing elongation of the tendril and expansion of the pitcher tube represented stage 4 (Fig. 2a,e). The leaf (L5) with the expanded unopened pitcher was considered stage 5 (Fig. 2a). www.nature.com/scientificreports www.nature.com/scientificreports/ Scanning electron microscopy (SEM). Tissues from each stage were cut into smaller pieces and fixed in 0.1 M phosphate buffer (0.1 M Na 2 HPO4, pH adjusted to 7.2 by adding 0.1 M NaH 2 PO4) containing 2.5% glutaraldehyde. As depicted in Fig. 2a-e, these tissue pieces represent portions of the leaf base lamina as well as the tip of the developing leaf of stages 1-3. The developing pitchers of stages 4 and 5 were cut into separate pieces at three different locations -top, a little above the middle and bottom -which correspond to the lid, waxy zone and digestive zone, respectively. In addition, the leaf base lamina midvein and the tendril of the later stages of N. khasiana leaf development (stages 3, 4 and 5) were also cut into smaller pieces and fixed in the aforesaid fixative. Histological analysis. Tissues from stage 1 were cut into two separate pieces: one representing the apex and another from the middle portion of the leaf base lamina (Fig. 2f,g), and fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2). Fixed tissue samples were processed, cross-sectioned, stained with 0.05% toluidine blue, viewed and photographed using a Nikon Eclipse Ti-S Inverted Microscope. RNA extraction, library preparation and sequencing. For RNA extraction, tissues from each of the five different stages of N. khasiana leaf development defined in the present study were dissected using a sterile scalpel blade. As depicted in Fig. 2a, leaves of stages 1 and 2 were dissected at the base of each leaf. On the other hand, leaves of stages 3, 4 and 5 were dissected near the tip (Fig. 2a). Thus, a portion of the leaf base, tendril and pitcher make up stage 3 whereas stage 4 comprises the tendril and pitcher. Stage 5 represents the unopened pitcher (Fig. 2a). Total RNA from these dissected tissues was then isolated using the Raflex Kit (Bangalore Genei, India) and/or the Spectrum Plant Total RNA Kit (Sigma, USA), following the manufacturer's instructions. Libraries were prepared using the TruSeq stranded total RNA library preparation kit (Illumina). Generated libraries were validated on the Agilent 2100 Bioanalyzer and sequenced using the Illumina HiSeq. 2000 platform following the manufacturer's recommended protocol to generate 2 × 100 bp paired-end data. The RNA-seq data from two biological replicates has been submitted to NCBI short read archive and can be accessed under accession number SRR4340048. Two biological replicates represent two separate individual plants, from which tissue samples comprising the different stages were harvested and sequenced separately.
Data pre-processing and de-contamination. Illumina adapter sequences were removed from the data using Cutadapt v1.3 27 . Low-quality data (Q < 20) were filtered using Sickle v1.33 28 . From the trimmed paired-end reads, unwanted sequences which include mitochondrial genome sequence, ribosomal RNAs, transfer RNAs,