Proteomic Characterisation of the Salt Gland-Enriched Tissues of the Mangrove Tree Species Avicennia officinalis

Plant salt glands are nature’s desalination devices that harbour potentially useful information pertaining to salt and water transport during secretion. As part of the program toward deciphering secretion mechanisms in salt glands, we used shotgun proteomics to compare the protein profiles of salt gland-enriched (isolated epidermal peels) and salt gland-deprived (mesophyll) tissues of the mangrove species Avicennia officinalis. The purpose of the work is to identify proteins that are present in the salt gland-enriched tissues. An average of 2189 and 977 proteins were identified from the epidermal peel and mesophyll tissues, respectively. Among these, 2188 proteins were identified in salt gland-enriched tissues and a total of 1032 selected proteins were categorized by Gene Ontology (GO) analysis. This paper reports for the first time the proteomic analysis of salt gland-enriched tissues of a mangrove tree species. Candidate proteins that may play a role in the desalination process of the mangrove salt glands and their potential localization were identified. Information obtained from this study paves the way for future proteomic research aiming at elucidating the molecular mechanism underlying secretion in plant salt glands. The data have been deposited to the ProteomeXchange with identifier PXD000771.


Introduction
Plants growing in saline environment have to cope with the constant exposure to high levels of salt and limited availability of freshwater. In some halophytic plant species (i.e., plants that are able to tolerate salt concentrations as high as 500-1000mM), there exists specialized microscopic structures located predominantly on the leaves and stems that are able to remove salts from the internal tissues and deposit them on the leaf surfaces [1,2]. Known as the salt glands, they are nature's desalination devices offering alternative routes for excess ion elimination through secretion, an adaptive feature that favours species inhabiting saline environment.
Previous studies by us [9,10] focused on the salt glands of a commonly found mangrove tree species in Singapore (Avicennia officinalis L) [11,12]. By making use of an epidermal system developed, we discovered a unique secretory pattern [10] that had been considered as an example of high resolution measurements of secretions that may lead to a general understanding on the mechanism of fluid secretion in both plant and animal systems [13]. This species grows in intertidal zones and has to cope with periodic exposure to fluctuating salinities [14]. We hypothesize that salt glands function as salt and water bi-regulatory units and the salt glands of this species offer an excellent platform to investigate their dynamic responses and molecular underpinnings to fluctuating salinities. Modern high throughput proteomics tools allow more detailed quantitative information, both temporal and spatial expression of proteins, to be obtained [15]. Recent studies in search of salt-responsive proteins in mangroves have also adopted a proteomic approach [16][17][18]. These studies, however, focused on the non-secretors (i.e., Bruguiera gymnorhiza, Kandelia candel) that do not have salt glands on their epidermal surfaces. Due to technical challenges faced in obtaining proteins directly from salt glands, a recent proteomic paper published by our group [19] focused on the plasma membrane and tonoplast proteins extracted from the leaves of A. officinalis. All these studies reported thus far have adopted a gel-based analysis approach.
In this study, as a continuous effort to better understand how water and salt are transported via the salt glands, we have adopted a shotgun approach to look into the proteome of salt-gland enriched tissues of the mangrove tree species A. officinalis. By removing the bulk of the leaf tissues (i.e., mesophyll tissues) devoid of salt glands, proteins from tissues that are rich in salt glands could be successfully obtained. To compensate for the technical limitations in obtaining large amounts of proteins from salt-gland-rich tissues, the shotgun approach adopted in this study allows simplified handling of samples with more exhaustive digestion and avoidance of sample loss in the gel matrix [20]. The data obtained via this approach offers a glimpse into the proteome of salt gland-rich materials and serves as a platform for identifying pool of proteins that could be involved in the desalination process of the mangrove salt glands.

Plant materials and protein extraction
The leaves of A. officinalis were required for the isolation of salt gland-enriched tissues (i.e., adaxial epidermal peels) for subsequent protein extractions. Shoots of A. officinalis were first collected from the mangrove swamp at Berlayer Creek (Sungei Berlayer, Labrador, Singapore; 1.27°N; 103.80°E; permit for collection granted by Keppel Club, Singapore). For each biological replicate, the adaxial epidermal peels, which harbour the salt glands, were separated from the mesophyll tissues of~20 leaves collected from several shoots according to Tan et al. [9]. Briefly, abaxial epidermal layers of excised leaves were removed, the leaves cut into segments before the leaf strips floated on enzyme mixture (pH 5.7; filter-sterilized) containing 0.1% (w/v) Pectolyase Y-23 (Seishin Pharmaceutical, Japan), 1.0% (w/v) Driselase (Sigma-Aldrich, USA) and 1.0% (w/v) Cellulase Y-C (Kikkoman Corporation, Japan) were vacuum infiltrated for 10 min and incubated in the dark at 30°C, 30 rpm for 1 h. The adaxial epidermal peels were easily detached from the mesophyll tissues after enzyme treatment. These peels were then rinsed, the remnants of mesophyll-palisade layers gently scraped off using a scalpel to obtain adaxial peels devoid of chlorophyll-containing cells and were collected separately from the mesophyll tissues. Three biological replicates were prepared. Total protein was extracted separately from these tissues by grinding them in liquid nitrogen and resuspending in buffer containing 25mM triethylammonium bicarbonate, 8M urea, 2% Triton X-100 and 0.1% sodium dodecyl sulphate [21]. The samples were then sonicated on ice for 30min, centrifuged (16000×g) at 15°C for 1h before supernatants were collected. Proteins were estimated using RCDC Protein Assay Kit (BioRad, Hercules, CA, USA) to compensate for interfering compounds in the samples.
Desalted samples were reconstituted with 15μl diluent [2% ACN, 0.05% formic acid (FA)], 5μl of which was loaded on 200μm × 0.5mm trap column and eluted onto analytical 75μm × 150mm column. Both columns were made of Repro-Sil-Pur C18-AQ, 3μm (Eksigent, Dr. Maisch, Germany). Peptides were separated by gradient formed by mobile phase A (2% ACN, 0.1% FA) and B (98% ACN, 0.1% FA): 5-12% mobile phase B in 20min, 12-30% mobile phase B in 40min, 30-90% mobile phase B in 2min, each at a flow rate of 300nl/min. The MS analysis was performed on TripleTOF 5600 system (AB SCIEX, Foster City, CA, USA) in Information Dependent Mode. MS spectra were acquired across mass range of 400-1800m/z in high resolution mode (> 30000) using 250ms accumulation time/spectrum. A maximum of 20 precursors/cycle was chosen for fragmentation from each MS spectrum with 100ms minimum accumulation time for each precursor and dynamic exclusion for 15s. Tandem mass spectra were recorded in high sensitivity mode (resolution > 15000) with rolling collision energy on.
Peptide identification and quantification was performed with ProteinPilot 4.5 software Revision 1656 (AB SCIEX) using the Paragon database search algorithm (4.5.0.0) and integrated false discovery rate (FDR) analysis function. The obtained MS/MS spectra were then searched against a database created [i.e., derived from transcriptome sequencing of salt glandenriched tissues (i.e., adaxial epidermal peels of A. officinalis, with a total 174552 entries including both normal and decoy sequences)]. The following search parameters were adopted: Sample Type-Identification; Cys Alkylation-MMTS; Digestion-trypsin; Special Factors-None; Species-None. The processing was specified as follows: ID Focus-Biological Modifications; Search Effort-Thorough; Detected Protein Threshold-0.05 (10.0%). Identified proteins for each biological replicate were selected based on a false discovery rate (FDR) of < 1%.
For Gene Ontology (GO) studies [22], proteins identified in the salt gland-enriched tissues and that are present in at least two of the biological replicates were selected for further analysis.
These selected proteins were first submitted to the UniProt Knowledgebase (UniProtKB) website (http://www.uniprot.org/help/uniprotkb) to retrieve the corresponding UniProtKB/Swiss-Prot entries and only annotated entities (i.e., with matched Swiss-Prot ID) were consolidated for GO analysis.

Results
The salt glands of Avicennia officinalis are microscopic (20-40 μm) structures found on the epidermal leaf surfaces (Fig 1A). They can secrete droplets of salt solutions, which appear circular in shape above the salt glands under a layer of oil when the adaxial (upper) epidermal peel (which harbours the salt glands) was viewed from the top (Fig 1B). These adaxial epidermal peels that are enriched with salt glands (Fig 1C) thus serve as good starting materials for the study of the salt gland proteome. To achieve this, proteins from both the adaxial epidermal peels (salt gland-enriched) and mesophyll tissues (salt gland-deprived) were extracted and compared (Fig 2A). For each extraction, approximately 2 mg proteins/g tissues and 9 mg proteins/g tissues were obtained from the epidermal peels and mesophyll tissues, respectively. A 2DLC/MS/MS analysis was performed on each of the trypsin-digested samples and identified proteins for each biological replicate were selected (< 1% FDR; S1-S6 Tables). An average of 2189 ± 128 (Table 1 and S1-S3 Tables) and 977 ± 150 (Table 1 and S4-S6 Tables) proteins were observed from the epidermal peels and mesophyll tissues, respectively. To obtain a list of proteins from salt gland-enriched tissues, only proteins that are found in epidermal peels but not in mesophyll tissues were considered. Data were sorted using nwCompare [23], with proteins found in any biological replicates for each tissue type taken into consideration and those extracted from epidermal peels and can be identified from mesophyll tissues eliminated. Using this approach, 2188 proteins are identified in salt gland-enriched tissues (Fig 2B and see S7  Table). Of these, 496 proteins were commonly found in all biological replicates, 536 proteins observed in two out of three biological replicates while remaining 1156 were present in one of the biological replicates ( Fig 2B). Among the 496 proteins that were commonly found in all three biological replicates analysed, more than 25% of the proteins with at least one unique peptide was identified, while~50% of the proteins with 2-5 unique peptides and the rest with at least 6 unique peptides identified ( Fig 2C). By looking at the distribution of protein sequence coverage, more than 65% of the proteins identified showed a sequence coverage of 15-40% ( Fig  2D).
To better understand the functions of proteins identified in salt gland-enriched tissues, proteins that are present in at least two of the biological replicates were selected from the list of 2188 proteins for GO analysis [22]. A total of 1032 selected proteins (see S8 Table) were analysed and information were retrieved from the UniProt Knowledgebase website (http://www. uniprot.org/help/uniprotkb). The proteins were annotated based on three organizing principles of GO (Fig 3). They were characterised by their function in diverse biological processes, with 11 sub-categories identified ( Fig 3A). Majority of these proteins were predicted to participate in metabolic (24%), cellular (22%) or single-organism (13%) processes or were responding to stimulus (9%), if not involved in localization (12%) (Fig 3A).
Among the list of proteins, many heat shock proteins (HSPs) or proteins related to carbohydrate and energy metabolism (e.g., ATPases, ATP synthases, aconitate hydratases, GTP-binding proteins) were identified ( Table 2 and S8 Table). Proteins (Table 2). Membrane proteins such as aquaporins, transporters/exchangers, channels and pumps were also observed in the salt gland-rich tissues (Table 2). Kinases, leucine-rich repeat proteins, 14-3-3-like protein and calreticulin commonly involved in signal transduction pathways had also been identified in this study (Table 2 and S8 Table). Candidate proteins that are of interest to us pertaining to the secretory process of salt glands include ATPases (e.g., Swissprot ID: Q03194, Q2QY12), transporters (e.g., Swissprot ID: Q96303, Q9LKW9, Q9LVM1 Q9FY75), aquaporins (e.g., Swissprot ID: Q7XLR1) and GTP-binding proteins (e.g., Swissprot ID: O04266). Based on GO analysis, most of these proteins were predicted to be localized to the plasma membrane while some were expected to be found in the tonoplast, mitochondria, Golgi apparatus or endoplasmic reticulum (Fig 4).

Discussion
In this study, we adopted the state-of-the-art shotgun technique to look into the proteome of salt gland-rich tissues. This approach allows the fast detection of proteins from complex mixtures and is rapidly replacing commonly gel-based methods (e.g., two-dimensional gel electrophoresis coupling with MS) [20,24,25]. Through this approach, many more proteins could be obtained (i.e, 2188 proteins identified in salt gland-rich tissues; Fig 2B and S7 Table). Among them, many of the proteins that are known to play a role in defence and stress responses in plants (e.g., stress proteins, HSPs, co-chaperones, dehydration-responsive/desiccation-related proteins, AAA-ATPase) were identified in the salt gland-rich tissues in this study (Table 2 and  S8 Table). HSPs, for examples, which are one of those proteins commonly identified in this study, are members of the molecular chaperones that help to protect the plants against stress (e.g., salt stress) by refolding the proteins and maintaining their native conformation, thus preventing irreversible protein aggregation during adverse conditions [17,19,26,27]. Fructose-1,6-bisphosphate aldolase that had been identified and suggested to play a role in salt tolerance mechanisms that are common to both the glycophytes and mangrove plants [16] was also found in our salt gland-rich protein pools (Table 2). Proteins involved in signal transduction, such as 14-3-3-like protein and calreticulin that had been shown to be upregulated during salt stress in K. candel [17] and reported in the plasma membrane proteins (i.e., 14-3-3-like protein) of A. officinalis leaves [19] had also been identified herein in the salt gland-rich tissues ( Table 2). Earlier studies by Hill and Hill [28] and Faraday et al [29] looked into the ion fluxes of Limonium salt glands and possible involvement of ion pumps and channels during secretion had been suggested by Vassilyev and Stepanova [30]. Inhibitor studies investigating on the different types of ATPase activities or looking at various membrane proteins (channels, antiporters) during secretion had also been reported (e.g., [5,10,[31][32][33]). Many membrane proteins (e.g., ABC transporters, sodium/potassium transporters, ATPases, aquaporins, proton pumps, sodium/hydrogen exchanger, ion channels) had been identified in the salt gland-rich protein pools (Table 2 and S8 Table) and could be involved during the desalination process. ABC transporters that are identified as one of the major transporters in our recent transcriptome study [34] and their abundance in the tonoplast and plasma membrane fractions of A. officinalis leaves [19] were also observed in the salt gland-rich protein pools in this study.
Secretion via the salt glands eliminates excess salts (predominantly Na + and Cl -) from the plant tissues and is believed to be energy-requiring [2]. The identification of proteins associated to carbohydrate and energy metabolism (e.g., ATPases, ATP synthases, aconitate hydratases, GTP-binding proteins) in the salt gland-rich protein pools (Table 2 and S8 Table) suggest high metabolic rate within these plant tissues. Many of these proteins found in the salt glandenriched tissues are involved in heterocyclic compound, ion and small molecule (e.g., ATP, GTP) binding (> 70%) and showed hydrolase, transferase and oxidoreductase activities (> 80%) (Figs 3 and 4, Table 2 and S8 Table) and the abundance of such proteins actually favours processes that are energy-dependent, including the desalination process in the salt glands. Determination of ATPase activities from leaves/leaf cells of salt gland-bearing species, for example, has been attempted in early studies [35,36]. Subsequent electrophysiological studies on Avicennia salt glands looked into possible ATPase activities [31,32]. Inhibitors of plasma membrane H + -ATPases (including orthovanadate) has been shown to inhibit salt secretion for both bicellular and multicellular glands [31,37]. High plasma membrane ATPase activity has been reported in the gland cells [38,39], suggesting possible role of plasma membrane Ptype H + -ATPase in salt secretion. Recent studies on A. marina further suggest some dependence on increased ATPase and antiporter gene expression in nitric oxide-enhanced salt secretion [5].
A. officinalis under study is a salt gland-bearing tropical mangrove tree species growing towards the sea and needs to cope with ever-fluctuating salinities (0.5-35ppt) [2,40]. Taking into consideration that secretion removes not just salts, but involves an inevitable loss of water, the identification of water channels (i.e., aquaporins) at the protein level in this study further reinforce our earlier studies on the involvement of aquaporin during secretion in this species [10,34]. Salt glands of this species thus offer an excellent platform for studying dynamic responses in regulating salt and water during secretion under rapidly changing salinities. The identification of major proteins that can respond to stimulus and are involved in cellular processes enabling them to cope with dynamic salinity changes will help us understand the process better.

Conclusion
In conclusion, we report the first proteomic analysis of salt gland-enriched tissues of a mangrove tree species. By comparing protein profiles of epidermal peels with mesophyll tissues, proteins found in salt gland-enriched tissues were identified, allowing GO analysis to be performed and a list of candidate proteins that could be involved in the desalination process identified. We believe that information obtained herein is valuable and can be used to dissect the molecular mechanisms that control the dynamics of secretion in mangrove salt glands.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository [41] with the dataset identifier PXD000771.
Supporting Information S1 Table. List of proteins identified in epidermal tissues of A. officinalis (first biological replicate) using AB SCIEX ProteinPilot Software 4.2, with false discovery rate (FDR) set at 1%. (XLSX) S2 Table. . List of proteins identified in epidermal tissues of A. officinalis (second biological replicate) using AB SCIEX ProteinPilot Software 4.2, with false discovery rate (FDR) set at 1%.