Auxin involvement in tepal senescence and abscission in Lilium: a tale of two lilies

Summary Strong evidence is presented for auxin regulating lily tepal abscission timing in relation to senescence. Transcriptome data was used to correlate auxin levels with expression of auxin-related genes.


Introduction
Floral lifespan is tightly regulated and is species specific (Rogers, 2013). The ecological function of petals is in attracting pollinators, and hence petal lifespan is often tightly linked to pollination (van Doorn, 1997). However, even in the absence of pollination, petals have a limited lifespan that is terminated either by wilting or abscission. This can be further subdivided into species where petals are abscised fully turgid and those in which some wilting occurs first. In monocotyledonous plants, petals (or tepals) usually show some signs of wilting (McKenzie and Lovell, 1992;van Doorn and Stead, 1997). This ranges from Hemerocallis (daylily) in which 66% of dry weight (DW) is lost (Lay-Yee et al., 1992) to Alstroemeria where only 20% is lost (Chanasut et al., 2003). In many species, petal senescence is coordinated by the growth regulator ethylene; ethylene biosynthesis increases dramatically during petal senescence, and exposure to exogenous ethylene accelerates the process (van Doorn, 2001). However in an important group of flowers including the lilies, ethylene does not appear to play a major role in petal senescence (Rogers, 2013). A number of other plant growth regulators have been implicated in the regulation of floral senescence in ethyleneinsensitive species (Arrom and Munné-Bosch, 2012a). In particular, auxin and cytokinin levels rose in Lilium L.A. var. 'Courier' prior to anthesis, falling thereafter (Arrom and Munné-Bosch, 2012a). Consistent with this, treatment of Iris cut flowers with cytokinins increased vase life slightly (van der Kop et al., 2003), as did treatment of Narcissus (Hunter et al., 2004) with gibberellic acid (GA 3 ).
Abscission is a well-characterized developmental process, occurring in leaves, fruit, and floral organs. In cut flowers, it is an important factor in their post-harvest quality (van Doorn and Stead, 1997). In all abscising tissues studied, the process can be divided into four stages (Niederhuth et al., 2013). The first stage involves the formation of an abscission zone (AZ) composed of a variable number of layers of small, cytoplasmically dense, cells. The structure of the AZ varies among species but is consistent within a single species (Taylor and Whitelaw, 2001). The timing of AZ formation also varies among species: for example, in tomato (Malayer and Guard, 1964) and Arabidopsis (Cai and Lashbrook, 2008) it forms long before abscission, while in cotton the AZ is formed only just before the organ is shed (Bornman et al., 1967). Once formed, the AZ is competent to respond to abscission signals initiating the second stage of abscission. In flowers, these are only normally activated during senescence of the organ. However, in ethylene-sensitive flowers such as geranium, application of exogenous ethylene can result in very rapid petal abscission within 1-2 h (Eversen et al., 1993). This activation of the pre-formed AZ results in the third stage of abscission: specific degradation of middle lamellae between the AZ cells by the action of a suite of hydrolytic enzymes including polygalacturonases, cellulases, and endoglucanases (Roberts et al., 2002). This allows the AZ cells to separate and the organ to abscise (Rogers and Stead, 2011). In the fourth stage, post-abscission, a protective layer forms over the site of abscission.
Auxin was identified as a key negative regulator of floral abscission over 50 years ago (Jacobs, 1962). This was demonstrated recently by manipulating auxin levels in AZ cells through activation of the bacterial auxin biosynthetic genes iaaL and iaaM specifically in these cells (Basu et al., 2013). However, levels of auxin are also modulated by the balance between the free, active form, and conjugated inactive storage form (Ludwig-Müller, 2011). It was postulated that an influx of auxin into the AZ prevents abscission from taking place (Taylor and Whitelaw, 2001), making auxin a key regulator in the final decision to abscise. Auxin is synthesized in young tissues, and in Arabidopsis petals there is a transient increase in auxin levels in buds (Aloni et al., 2006). The auxin is transported to other parts of the plant via a chemiosmotic mechanism mediated by PIN efflux carriers that determine the direction of flux (Leyser, 2006). In the leaf, it is generally accepted that the maintenance of constant polar indole-3-acetic acid (IAA) flux through the AZ prevents abscission (Osborne and Morgan, 1989;Taylor and Whitelaw, 2001;Roberts et al., 2002). Polar auxin transport inhibitors, such as 1-N-naphthylphthalamic acid (NPA), provide useful tools for analysing the importance of auxin transport during developmental processes (Nemhauser et al., 2000). The mode of action of NPA has not been fully elucidated, but AtAPP1, encoding a plasma-membrane metalloprotease, has been identified as a protein with a high affinity for NPA and a probable role in processing of PIN1 efflux carriers on the plasma membrane (Murphy et al., 2002).
In Arabidopsis, cell-wall dissolution is modulated by auxin through the action of members of the auxin response factor (ARF) transcription factor family. ARF proteins are required for an auxin response: they bind to cis-elements in promoters of auxin-responsive genes resulting in their activation or repression (Ulmasov et al., 1997). In Arabidopsis, there are 23 ARF genes (Wang et al., 2007) and four of them have a role in organ abscission. ARF1, ARF7, and ARF19 are directly upregulated by auxin, and these in turn upregulate ARF2, which acts to inhibit the expression of the hydrolytic enzymes (Ellis et al., 2005) responsible for middle lamella breakdown. In rice, there are 25 ARF genes; OsARF7 and OsARF9 show the closest homology to AtARF1 and OsARF16 shows closest homology to AtARF7 and AtARF9, while OsARF4 is the closest homologue to AtARF2 (Wang et al., 2007).
Lilium species include commercially important cut flowers, especially as hybrids (Benschop et al., 2010). The first group of hybrids produced were the Asiatic hybrids derived from species native to Central and East Asia. These have been further crossed to Lilium longiflorum to produce L. longiflorum×Asiatic (Lilium L.A.) hybrids. Lily hybrids include both abscising and non-abscising cultivars (van Doorn, 2001), although most cultivars show some wilting, with a longer time between wilting and tepal fall in Asiatic cultivars. The senescence patterns of both L. longiflorum (Battelli et al., 2011) and the Lilium L.A. hybrid var 'Courier' Munné-Bosch, 2010, 2012a, b) have been recently studied. Tepals of L. longiflorum wilt substantially during senescence but remain attached, whereas tepals of Lilium L.A abscise following less severe wilting. This offers the opportunity to compare the senescence process in these two closely related genotypes. Specifically, the aim of this work was to test to what extent paradigms for the role of auxin in abscission, and upregulation of ARF genes developed with model species such as Arabidopsis can be applied to this taxonomically divergent, ethylene-insensitive genus.

Materials and methods
Plant material L. longiflorum cv. 'White Heaven' was grown in a commercial greenhouse and Lilium L.A. var.'Courier' (L. longiflorum×Asiatic hybrid) plants were obtained from greenhouse-grown bulbs. Individual flowers were harvested at the stage of closed bud by cutting above the last leaf. Flowers were placed in distilled water and kept in a growth chamber at 22 °C and 50% relative humidity. Under the conditions used, flower development and senescence progressed in a very predictable way from closed bud to full senescence (L. longiflorum) or abscission (Lilium L.A.).
Samples were collected from comparable developmental stages for the two species: closed bud (CB), full bloom flowers at anthesis (FB), flowers showing the first visible signs of senescence on outer tepals (early senescence, ES), and flowers at the end of their vase life (full senescence, FS), which was full dryness and wilting for L. longiflorum and abscission for Lilium L.A. Where appropriate, flowers were also harvested 1 or 2 d after either ES or FS.
Exogenous treatments with IAA and NPA Flowers at the stage of CB were treated with 10 µM IAA or with 50 µM NPA to inhibit auxin transport, and were added to the water in which the stems were immersed. Both treatments were applied throughout the experiment from CB to FS. Only outer tepals were sampled for all the analyses.

Ion leakage
Discs (8 mm diameter) were cut from each side of the central vein of the outer tepals about half way from the tip (20 discs per tepal) and placed in 10 ml of distilled water in Petri dishes. After a 2 h wash to remove ions from cut surfaces, the water was aspirated and fresh distilled water was added. Following incubation for 6 h, the conductivity of the bathing solution (sample conductivity) was measured with a conductivity meter (HI-8733; HANNA Instruments). Fresh distilled water (10 ml) was then added to the tepal discs and boiled for 15 min. After cooling to room temperature, the conductivity was measured again to obtain the subtotal conductivity.
Ion leakage was expressed as relative conductivity, which was calculated as sample conductivity divided by total conductivity (the sum of sample conductivity and subtotal conductivity).
RNA extraction and cDNA preparation RNA was extracted from the outer tepals with TRI Reagent (Sigma, St Louis, MO, USA) according to the manufacturer's instructions. RNA was subjected to DNase treatment using a TURBO DNA-free kit (Ambion, Austin, TX, USA) to remove contaminating genomic DNA. Two micrograms of RNA was reverse transcribed into cDNA using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) in accordance with the manufacturer's instructions.
Illumina reference transcriptome sequencing of L. longiflorum tepals Total RNA was extracted from the FB and ES stages of L. longiflorum tepals as described above and pooled. The quality of the RNA was checked by gel electrophoresis and capillary electrophoresis using a Bioanalyzer (RNA Nano Chip) and the concentration determined using absorption spectroscopy using a NanoDrop. A cDNA library was constructed by BaseClear (http://www.baseclear.com) following a Tru-Seq (Illumina) protocol and sequenced on an Illumina Hi-Seq 2000 using a 50-cycle paired-end protocol. A reference transcriptome was assembled using CLC Genomics Workbench (CLC bio). The validity of the assembly was performed by remapping (CLC Genomic Workbench) the original read onto the assembly and through functional annotation, with the former also providing a relative abundance of transcripts in the original sample. Functional annotation was performed using BLASTX and was used to interrogate the non-redundant protein sequence data downloaded from NCBI (1 November 2012) together with the Uniprot database (1 November 2012). Putative annotation assignment was performed at an E-value cut-off of <1E-05. BLAST results were integrated into BLAST2GO (Conesa and Götz, 2008), which was used to generate putative gene function and ontological assignments. Transcriptome data were deposited with the Sequence Read Archive (SRA) database at NCBI (SRA Experiment: SRX690392) Cloning of Lilium L.A. aminopeptidase P1 (APP1) Degenerate primers were designed based on sequences from an Alstroemeria auxin-responsive gene APP1 (Wagstaff et al., 2010) and similar sequences from rice, Ricinus, and Sorghum. These were used to amplify a 195 bp product. This was purified using a Qiaquick kit and ligated into pGEM-TEasy (Promega). Based on the sequence from the cloned fragment, specific primers were designed for quantitative real-time PCR (all primers are listed in Supplementary Table  S1 at JXB online). Table S1) were designed with Primer3 software (Rozen and Skaletsky, 2000) for the Lilium APP1, ARF6/8, ARF 7/19, and AUX1-like sequences derived from degenerate PCR or contigs from the L. longiflorum reference tepal transcriptome. PCR products from all the primer pairs were sequenced and compared with the available sequences to verify the specificity of the primers (sequences have been deposited in the EMBL data base under accession numbers LN606581, LN606582, LN606583, and LN606584). qPCR was carried out in a 7300 real-time PCR system (Applied Biosystems) using 50 ng of cDNA and a SYBR ® Green PCR master mix (Applied Biosystems). The PCR product was further analysed by a dissociation curve programme from 95 to 60 °C. Expression of the 18S rRNA gene was used for internal normalization, using the PUV1 and PUV2 primers, which amplify a 226 bp fragment (Dempster et al., 1999). Data were analysed using the 2 -ΔΔCT method (Livak and Schmittgen, 2001) and are presented as the relative level of gene expression. All real-time qPCRs were run in triplicate with cDNAs synthesized from RNA extracted from three biological replicates.

IAA extraction and analysis
Frozen tepal samples from control and NPA-treated flowers were homogenized in 5 vols of cold 80% (v/v) methanol and then stirred for 4 h at 4 °C, before centrifugation at 2000g for 15 min. The pellet was reextracted twice; the supernatants were pooled and reduced to the aqueous phase under vacuum, and the pH of the supernatant was adjusted to 2.8 and partitioned four times against equal volumes of ethyl acetate. The samples were dried and dissolved in a small volume of 10% (v/v) aqueous acetonitrile containing 0.5% (v/v) acetic acid just before highperformance liquid chromatography. IAA was purified by high-performance liquid chromatography and quantified by gas chromatography/ mass spectrometry as described previously (Mariotti et al., 2011).

Light microscopy
Abscission zones were excised from flowers at different stages of development. The pieces of tissue, approximately one-eighth of the cross-sectional area, were immersed in primary fixative comprising 3% glutaraldehyde, 4% formaldehyde in 0.1 M PIPES buffer, pH 7.2, for a minimum of 1 h. Specimens were then rinsed in 0.1 M PIPES buffer, post-fixed in 1% buffered osmium tetroxide (1 h), rinsed in buffer, block stained in 2% aqueous uranyl acetate (20 min), dehydrated in an ethanol series, and embedded in Spurr resin (Agar Scientific, Stansted, UK) in the standard way. The polymerized blocks were then reoriented in order to be able to section perpendicular to the abscission zone. Semi-thin 0.5 µm sections were cut on a Leica OMU 3 ultramicrotome and stained with 1% toluidine blue in 1% borax and photographed on a light microscope using a Nikon Coolpix 4500 digital camera.

Detachment force
For L. longiflorum, the free portion of the tepals was removed and the filaments and ovary trimmed further to ensure that when clamped only the corolla tissue was held. The clamp, with attached flower, was connected to a strain gauge (M/no DFG-1K; Shimpo), the pedicel was grasped firmly, and a single straight pull was employed to remove the corolla. Where the corolla tore, or was pulled from the clamp, the break strength was recorded as 'in excess of the recorded value'; hence, for some stages the values presented in Fig. 3B and C are underestimates of the force needed to detach the corolla. For Lilium L.A., individual tepals were trimmed by about one-third of their length, and then clamped and the detachment force determined using a single straight pull of the pedicel. The process then was repeated for the remaining tepals of that flower.

Results
Senescence markers show significant differences in L. longiflorum and Lilium L.A.
Although flower life progresses through similar stages in L. longiflorum and Lilium L.A., the final destiny of the tepals diverged. In L. longiflorum, tepals wilted substantially during senescence with visible browning and dehydration; however, they remained attached to the flower. In contrast, Lilium L.A. tepals abscised without wilting while still relatively turgid (Fig. 1A). Tepal fresh weight (FW), DW, ion leakage, and protein content were compared between L. longiflorum and Lilium L.A. to establish first a benchmarking of the senescence progression between the two flowers and then to determine key differences that might be related to the different strategies of wilting versus abscission in these closely related genotypes. ES was defined in both genotypes as the earliest stage showing visible signs of senescence such as tepal browning and increased translucency, while FS was defined in Lilium L.A. as the time of abscission and in L. longiflorum as complete wilting and full dehydration of the tepals (Fig. 1A). In both genotypes, FW/DW fell as the tepals progressed from CB through FB to senescence (Fig. 1B). However, in the later stages of senescence, there was a marked difference. Whereas in Lilium L.A. tepals there was a gradual reduction in FW/ DW, in L. longiflorum there was a sharp decline between the early and late senescence stages. There were also marked differences in the pattern of ion leakage (Fig. 1C). In Lilium L.A., electrolyte leakage remained relatively low and constant throughout bud opening and senescence, while in L. longiflorum there was a sharp increase in ion leakage between early and late senescence. In contrast, the decline in protein levels followed essentially the same trend in the two genotypes, although protein levels in L. longiflorum were lower throughout (Fig. 1D).
An AZ forms in both L. longiflorum and Lilium L.A., but the detachment of L. longiflorum is anomalous In L. longiflorum, the outer tepals and inner tepals are fused over the lower third of the tepal length, with the margins of the outer tepals fused to the midrib of the inner tepals. Thus, in transverse section ( Fig. 2A-C), there can appear to be two whorls of tepals and, if looking even closer to the pedicel, the individual outer tepals may also overlap giving the appearance of multiple whorls of tepals. When examining the AZ in longitudinal section, this results in several structures being visible, each with its own putative AZ (Fig. 2D).
In section, the AZ of each L. longiflorum tepal was visible. Even in freshly opened flowers at FB (Fig. 2D), the putative AZ, comprising a series of smaller cells, could be recognized. At higher magnification, the beginning of a fracture line was discernible running between these cells (Fig. 2E). However, in older flowers, several days after FS, this fracture line had not progressed significantly (Fig. 2F). The AZ was identifiable at the base of Lilium L.A. tepals even in CBs (Fig. 2G), and cell-wall dissolution was visible in the AZ of those flowers approaching ES (Fig. 2H).
Detachment force was measured in both lilies to establish whether the AZs observed were fully functional. In L. longiflorum, small dark lines, which later developed into cracks (Fig. 3A), were visible on the outside of the tissue at the point where the AZ would be expected to be situated. The fusion of the tepals in L. longiflorum meant that, unlike in Lilium L.A., only the force to detach the corolla, not individual tepals, could be quantified (Fig. 3B, C).
In early-stage buds through to freshly opened flowers, the force to remove the corolla in L. longiflorum either exceeded 1000g (the limit of the force gauge) or more often the corolla tissue tore at approximately 700-800g, although in some fully open flowers detachment of the corolla at the AZ occurred at 500-600g (Fig. 3C). As the flowers senesced and wilted, the extent of the cracking (Fig. 3A) increased, concomitant with a decrease in the force needed to detach the corolla. However, in flowers with wilted, browning tepals (i.e. beyond FS) where the external cracks appeared to be both extensive and deep (FS+2 in Fig. 3A), a force of 100-200g was still required to remove the corolla (Fig. 3B), showing that the abscission process was far from complete, even in senescent flowers that had passed the end of their vase life. In younger, freshly opened flowers (FB), no external cracking was visible (Fig. 3A) and the corolla could not be forcibly be removed by the strain gauge (Fig. 3B).
In Lilium L.A., some tepals could be detached from flowers that appeared to be at the FB stage, although most tore rather than detached (Fig. 3C). By the ES stage, virtually all tepals could be detached, and the force needed was greatly reduced (~600g); by the FS stage, all tepals had already abscised, but in the time between ES and FS the force needed to detach the tepals decreased, as would be expected for an organ showing a typical abscission process (Fig. 3C).

Auxin levels differ between the two lilies during tepal senescence
As auxin appears to be a key regulator of abscission, the level of total IAA as well as active free IAA and inactive conjugated forms (ester-linked to sugars and amide-linked to amino acids and peptides) was determined for both lily genotypes. Around the time of flower opening (from CB to FB), IAA content was similar in the two lilies, at about 100-150 ng g -1 of DW. Then, as senescence progressed, in L. longiflorum both free IAA and conjugated IAA increased dramatically (Fig. 4A), although the ratio of free to conjugated IAA was essentially 1:1 throughout the flower lifespan ( Supplementary  Fig. S1 at JXB online). In contrast, free IAA levels in Lilium L.A. remained low at every stage, from CB to abscission, while the portion of inactive IAA-amide gradually increased (Fig. 4B). Note that in Lilium L.A. the ratio of free to conjugated IAA changed from about 1:1 to 1:2 between CB and abscission ( Supplementary Fig. S1).
The effect of exogenous application of IAA was tested on both lily genotypes. No effect was seen either on senescence progression or on timing of abscission (data not shown).

Analysis of the lily tepal transcriptome reveals 50 genes related to auxin biosynthesis and perception including 12 ARF-like genes.
Forty million reads of a L. longiflorum reference transcriptome from ES/FB tepals were assembled into 13 000 unigenes. Fifty unigenes showed homology to auxin-related genes from other species (Table 1) consistent with the high levels of auxin detected. Twelve unigenes showed homology to genes belonging to the ARF family of transcription factors, and could be assigned to six Arabidopsis homology groups: AtARF6/8, AtARF9/18/11, AtARF19, AtARF3/4, AtARF7/19 and AtARF16 (Supplementary Table S2 at JXB online). Homology was assessed based on the inclusion in the contig sequence of motifs III and IV, which are consensus sequences shared by AUX/IAA proteins but are discriminatory between different ARF genes in rice (Wang et al., 2007) ( Supplementary Fig. S2 at JXB online). Primers were designed to two contigs with homology to AtARF6/8 and AtARF7/19 (contigs 651 and 7111, respectively) and were verified by PCR and sequencing of products.
The transcript level of the ARF6/8 homologue fell with tepal age in both lilies but with different patterns (Fig. 5A,  B). In Lilium L.A., there was a peak in transcript level at FB, which then essentially disappeared at abscission; in L. longiflorum, the highest expression was in CB and expression then declined gradually. The transcript level of the homologue of Arabidopsis ARF7/19 fell with development and senescence in Lilium L.A., while in L. longiflorum, which has the higher auxin content and delayed incomplete abscission, levels rose slightly during flower life and were much higher than in Lilium L.A. at both early and late stages of senescence (ES and FS) (Fig. 5C, D).

NPA has opposite effects in L. longiflorum and Lilium L.A.
When the flowers at stage CB were treated with NPA, a widely used auxin transport inhibitor, senescence progression in L. longiflorum and the time of abscission in Lilium L.A. did not change (data not shown). However, NPA induced remarkable changes in IAA concentration in the whole tepal (Fig. 4). In both genotypes, the levels of free IAA during the senescence stages were reduced by more than 50%. The effect of NPA treatment on the conjugated IAA pool differed substantially between the two genotypes. In L. longiflorum, IAA-ester and IAA-amide were reduced to about 30% of the control during early and late senescence. In contrast, in Lilium L.A., IAA-amide showed a dramatic increase at the time of abscission and IAA-ester levels did not change significantly.
The overall result was that, at the last stage of flower life, NPA treatment had the opposite effect on total IAA amount in the tepal in the two genotypes. In fact, in L. longiflorum, there was an almost 2-fold reduction in total IAA concentration, while in the abscised tepals of Lilium L.A., total IAA actually increased slightly, mainly due to the rise in IAA conjugates ( Supplementary Fig. S1). Consequently, the ratio of free to conjugated IAA went from 1:1 to 1:7 at the time of abscission.
To determine whether the distribution of auxin across the tepal differed between genotypes, IAA content in the different regions of the tepal was determined at the FS stage. Free IAA was not distributed equally along the tepal axis. In both genotypes, the IAA concentration was higher at the base compared with the tip (Table 2) with levels over 4-fold higher in L. longiflorum in each region. NPA treatment resulted in a fall in IAA concentration at the tip and a greater decrease in the middle section but a significant increase at the base of the tepal in both genotypes.

APP1 and AUX1 gene expression
To investigate further the role of auxin transport, expression of APP1, which is involved in the processing of PIN efflux carriers, and AUX1, an auxin influx transporter, was determined in the two lilies during tepal development and senescence.
Using degenerate primers designed to sequences from Alstroemeria, rice, Ricinus, and Sorghum, a 195 bp fragment homologous to Arabidopsis AtAPP1 was obtained from Lilium L.A. tepal cDNA ( Supplementary Fig. S3 at JXB online). Quantitative expression analysis showed that, in both lilies, APP1 transcript levels were highest in the CB, but then there was an opposite trend during senescence (Fig. 6A,  B). Expression was undetectable at abscission in Lilium L.A. while it was still high at the last stage of senescence in L. longiflorum, coincident with the highest levels of IAA (Fig. 3A). The expression of the homologue of AUX1 decreased with progression of senescence in both lilies (Fig. 6C, D), although levels were almost 3-fold higher at FB in Lilium L.A. Thus, ES in Lilium L.A. was accompanied by a 2-fold greater fall in expression compared with L. longiflorum.

Discussion
Given the differences in senescence progression and final abscission between the two lily genotypes, the focus of this work was to discover whether contrasting auxin levels could explain the differences in progression of abscission in relation to senescence.
First, a comparison was made of the senescence patterns between the two genotypes: L. longiflorum, where petals remain attached and wilt, and Lilium L.A., where tepals abscise turgid. Visual similarities between the developmental stages from CB to ES were supported by measurements of FW/DW and ion leakage. In both genotypes, FW/DW fell gradually and ion leakage remained constant from the CB to the FB stage. However, as the flowers entered senescence, their programmes diverged. Whereas in Lilium L.A. FW/DW continued to fall gradually, there was a sudden 2-fold drop between early and late senescence stages in L. longiflorum. This was accompanied by a doubling in ion leakage. These parameters are indicative of a rapid increase in water loss and cell death in L. longiflorum as senescence progresses but not in Lilium L.A. This is common in many species, including other lilies such as Hemerocallis (Lay-Yee et al., 1992), where there is substantial weight loss during senescence. In Alstroemeria, as in L. longiflorum, ion leakage rose, sharply coincident with the first signs of visual senescence (Leverentz et al., 2002). The lack of increases in the rate of weight loss and ion leakage in Lilium L.A. are reminiscent of other species such as tulip where FW fell to only about 70% of its maximal value (Azad et al., 2008), and Prunus yedoensis petals where no signs of PCD were detected prior to turgid abscission (Yamada et al., 2007). Thus, not only do Lilium L.A. tepals abscise but they do so in a relatively intact state, suggesting a very different senescence programme to L. longiflorum.
One possibility for the lack of abscission in L. longiflorum would be the failure to develop an AZ at all (Rogers and Stead, 2011). However, examination of the base of the outer tepals of both genotypes revealed that there is what appears to be a functional AZ in both. This suggests that the failure to completely abscise L. longiflorum tepals must depend on the signals that control the timing and completion of abscission. Differences in auxin levels in the AZ are a key component in tipping the balance towards abscission (Taylor and Whitelaw, 2001). Therefore, levels of auxin were compared throughout development and senescence across whole tepals of the two lilies. The dramatic increase in auxin levels in late senescence in L. longiflorum tepals would be consistent with a role for this growth regulator in delaying activation and/ or completion of the AZ while senescence processes such as cell death and water loss are still progressing in other cells or tissues. The constant and low levels of auxin in Lilium L.A. are consistent with an earlier activation and completion of the AZ, allowing tepals to abscise turgid. Since the balance between free and conjugated auxin is also important in determining the activity of this growth regulator (Rosquete et al., 2012), the finding that at FS L. longiflorum tepals contained higher levels of free compared with the two inactive conjugated forms further indicates that the auxin is active in this tissue. In contrast, in Lilium L.A. tepals, there was substantially more IAA-amide at FS than free IAA, indicating that, as well as lower levels of auxin, more of it was also conjugated and therefore inactive. This suggests several possible regulatory methods: an increased biosynthesis or transport, or reduced metabolism of the auxin, as well as differences in the activity of enzymes regulating the balance between conjugated IAA and free IAA.
To understand further the role of auxin in these lilies, auxin-related genes were derived from an early senescent tepal reference transcriptome. Target sequences for lily ARF genes known in Arabidopsis to be important in regulating abscission were sought: specifically Arabidopsis ARF1, ARF7/19, and ARF2. No homologues to ARF1 or ARF2 were obtained, despite a good depth of sequencing. This could be due to low levels of expression at this stage of development: in Arabidopsis, ARF1 is expressed at very low levels during petal and leaf senescence, and ARF2 is upregulated in senescent leaves but not in older petals (eFP browser; Winter et al., 2007). However, putative homologues of ARF7/19 were identified, and their expression pattern in L. longiflorum during senescence was consistent with upregulation of the expression of these genes in Arabidopsis older petals. In Arabidopsis, ARF19 is induced by auxin (Wilmoth et al., 2005) and is part of a positive-feedback loop involving ARF7. Thus, the upregulation of the lily ARF7/19-like gene is also consistent with an increase in auxin in this genotype. In contrast, the fall in expression of the ARF7/19-like gene with senescence in Lilium L.A. is consistent with the slight fall in free auxin in this genotype and also with the completion of abscission. As a comparison, the expression of ARF6/8 was also analysed and found to be downregulated during senescence in both lily genotypes. In Arabidopsis, ARF8 is involved in petal growth  and expansion (Varaud et al., 2011), and ARF6 and ARF8 are strongly expressed in both young and older petals, but their expression declines with petal age.
Having established a clear correlation between auxin levels and abscission timing in the two lily genotypes, we also asked if auxin transport was involved, since this may be more important that absolute concentration of IAA. NPA treatment did not affect the timing of senescence in L. longiflorum or abscission in Lilium L.A. This is in contrast to leaf abscission in Mirabilis jalapa where treatment with NPA delayed ethylene-induced abscission (Meir et al., 2006). However, effects of NPA on senescence are consistent with those seen in Iris, another ethylene-insensitive flower (van Doorn et al., 2013). Nevertheless, NPA did affect IAA levels differentially in the two genotypes, especially the levels of conjugated IAA, suggesting that auxin transport is required for determining levels of free auxin. The fall in expression of APP1 in Lilium L.A. during senescence, but not in L. longiflorum, may suggest that auxin efflux is a factor in maintaining high levels of auxin in the latter. The distribution of auxin levels from the base to the tip of the petal fits with previous data showing higher levels at the base and is consistent with auxin transport in the petal during development. The effect of NPA suggests that the auxin transport is still active and is in the direction of the base to the tip since NPA treatment reduces auxin levels in the middle and tip sections and increases free auxin concentration at the base. The increase in auxin at the base with NPA treatment suggests a blockage of auxin transport due to the inhibitory effect of the NPA. This is probably into the base of the tepal from other floral organs, since during late tepal development and senescence, levels of auxin biosynthesis in the tepal itself are likely to be low (Aloni et al., 2006). However, since this affects both lilies, the conclusion is that while auxin levels correlate with timing of abscission, transport of auxin does not.

Conclusions
Despite their close genetic relationship, L. longiflorum and Lilium L.A. tepals age through different mechanisms, indicating that wilting and abscission strategies can differ in very closely related genotypes. The presence of a fully formed AZ in both genotypes suggests that the difference in abscission relates to the very last steps of AZ activation/completion. There is a clear correlation between auxin levels and abscission timing in relation to senescence markers. Furthermore, the role of auxin in abscission previously elucidated in Arabidopsis may involve the same mechanism through the ARF genes in this taxonomically unrelated group. Auxin transport does not affect senescence, and the effects of the auxin transport inhibitor NPA did not affect the two genotypes differentially, indicating that, although differing auxin levels may be responsible for the timing of tepal abscission, this is not due to differences in auxin transport during senescence.

Supplementary data
Supplementary data are available at JXB online.
Supplementary Table S1. All primers used for PCR.
Supplementary Table S2. L. longiflorum contigs showing homology to ARF genes. Supplementary Fig. S1. Total auxin levels as the sum of free and conjugated IAA. Supplementary Fig. S2. Alignment of ARF-like lily sequences with nearest rice ORF match based on BLASTX homology. Supplementary Fig. S3. Alignment of APP1-like lily sequence with the Arabidopsis AtAPP1 gene (AT4G36760).