The composition of Australian Plantago seeds highlights their potential as nutritionally-rich functional food ingredients

When wetted, Plantago seeds become covered with a polysaccharide-rich gel called mucilage that has value as a food additive and bulking dietary fibre. Industrially, the dry husk layer that becomes mucilage, called psyllium, is milled off Plantago ovata seeds, the only commercial-relevant Plantago species, while the residual inner seed tissues are either used for low value animal feed or discarded. We suggest that this practice is potentially wasting a highly nutritious resource and here describe the use of histological, physicochemical, and chromatographic analyses to compare whole seed composition/characteristics of P. ovata with 11 relatives already adapted to harsh Australian conditions that may represent novel commercial crop options. We show that substantial interspecific differences in mucilage yield and macromolecular properties are mainly a consequence of differences in heteroxylan and pectin composition and probably represent wide differences in hydrocolloid functionality that can be exploited in industry. We also show that non-mucilage producing inner seed tissues contain a substantial mannan-rich endosperm, high in fermentable sugars, protein, and fats. Whole seed Plantago flour, particularly from some species obtained from harsh Australian environments, may provide improved economic and health benefits compared to purified P. ovata psyllium husk, by retaining the functionality of the seed mucilage and providing additional essential nutrients.

www.nature.com/scientificreports/ unique functional properties for industry 19,[36][37][38][39][40][41] . The limited studies that are currently available have shown that whole seed flour (WSF) from P. ovata and P. psyllium are useful hydrocolloid replacements in baking 42,43 with significant differences in product quality between the two species. These studies show the potential of WSF from different Plantago species to be a less wasteful, more nutritious alternative to psyllium husk for selectively manipulating food qualities based on interspecific differences in functionality. Therefore, we hypothesise that different Plantago species may yield correspondingly diverse functionalities and Australia may boast valuable untapped resources, with the added advantage of already being adapted to demanding climatic constraints that limit production of psyllium from P. ovata 10 . Thus, the aim of this work was to compare the composition and morphology of seeds of commercial psyllium, P. ovata, to eleven related species naturalised or native to Australia to assess their nutritional and functional potential as novel food ingredients. The results demonstrate great diversity in mucilage properties and composition that likely represent exploitable differences in hydrocolloid functionality and have provided the first comprehensive overview of Plantago seed composition, showing them to be rich in beneficial sugars, fats and protein. We suggest that the use of WSF from diverse Plantago species studied here will simultaneously allow selective manipulation of quality and nutritional enrichment of food products.

Materials and methods
Plant growth for seed bulking. Seeds of twelve Plantago species were obtained from the sources listed in Supplementary Table 1, and where applicable, collection complied with relevant institutional, national, and international guidelines and legislation. Seeds were stratified dry for 48 h at − 20 °C prior to germination then imbibed in filtered (0.22 µm) sterilisation agent (50:50, 50% ethanol:4% bleach with 0.05% Triton X-100) for 1 min before replacing the sterilisation agent and incubating for another minute. This was repeated until the seeds had been imbibed in fresh sterilisation agent 5 times, after which the seeds were washed 5 times with filtersterilised (0.22 µm) Milli-Q water. Seeds were spread onto pre-wetted autoclaved Whatman No. 1 paper in a sterile petri dish. Dishes were sealed, aluminium foil-wrapped and stratified for another 48 h at 4 °C. After stratification, plates were moved to a glasshouse with a day/night temperature of 23 °C/18 °C. Seeds were germinated for 10 days (3 days dark then 7 days exposed to the glasshouse day/night light cycle) then transferred to coco-peat soil mixture in tall citrus pots. Plants were grown to maturity from June to December (Adelaide, Australia) with no supplemental light. Once mature (capsules freely dehiscing), inflorescences were harvested, dried for 48 h at 37 °C, and then threshed to obtain seed. Where required, whole seeds were ground using a MM400 Mixer Mill (Retsch, Germany) and graded to 0.5 mm to produce whole seed flour (WSF).
Phylogenetic analysis. Mature leaf tissue from Plantago plants was frozen at − 80 °C and ground by stainless steel ball bearing for 30 s at 30 Hz using a MM400 Mixer Mill (Retsch, Germany) fitted with 2 ml tube adapter. DNA was extracted from ground leaf tissue following Healey et al. 44 . Nuclear ribosomal internal transcribed spacer (ITS) regions were amplified by PCR using primers 45 and conditions listed in Supplementary  Table 2. Amplified ITS regions were sequenced by AGRF (Adelaide, Australia). Where possible, sequence identities were confirmed by MUSCLE nucleotide alignment with published sequences from Rønsted et al. 46 and Tay et al. 47 . The ITS sequence of Veronica salicornioides, FJ024624.1 47 , was downloaded from Genbank and used as an outgroup as per Rønsted et al. 46 .
ITS sequences were trimmed using BMGE 48 and the maximum-likelihood tree constructed in Geneious v8.1.3 (Biomatters Ltd, NZ) with the RAxML tree builder 49 using the GTR GAMMA nucleotide model with 500 rapid bootstrapping replicates.
Seed morphometric measurements. Seed length and width measurements were determined by image analysis. Images of seeds were taken at 1 × magnification on an AxioImager M2 (Zeiss, Germany) fitted with an AxioCam 105 colour camera (Zeiss, Germany). Length and width of 20 seeds per species were measured using ZEN 2012 software (Zeiss, Germany). To determine 1000 seed weight, seeds were manually counted and weighed.
Mucilage microscopy. Mucilage architecture was observed by whole mount immunolocalisation of heteroxylan using the LM11 antibody 50 and ruthenium red staining following Phan et al. 33 . Ruthenium red used in this work (C075) was purchased from ProSciTech (Australia) and prepared as per Cowley et al. 51 .
Whole seed thin section microscopy and immunolabelling. Whole seeds were transversely halved with a razor blade and fixed and embedded in LR White resin as per Burton et al. 52 modified to use the 80% ethanol fixative described in Phan et al. 9 . Fixed and embedded tissue was sectioned at 1 µm on an Ultramicrotome (Leica, EM UC6) using a diamond knife (DiATOME, Nidau, Switzerland). After staining thin sections with toluidine blue, internal seed structures were imaged with an AxioImager M2 (Zeiss, Germany) fitted with an AxioCam 105 colour camera (Zeiss, Germany). Endosperm mannan polysaccharides were immunolocalised using a tenfold dilution of LM21 anti-(1,4)-β-mannan primary antibody (Kerafast, US). The secondary antibody, Alexa Fluor 555 goat anti-rat IgG, was applied at a 100-fold dilution. All fluorescent images were taken with a Zeiss M2 AxioImager with an AxioCam 506 mono black and white camera. Images were processed using ZEN 2012 software (Zeiss, Germany).
Seed mucilage fractionation and yield analysis. Mucilage was extracted and fractionated and yield traits were determined following Cowley et al. 51 with no deviation from the described procedure. Briefly, 1 mL www.nature.com/scientificreports/ of water was added to 30 mg of seed and extracted at 25 °C for 1.5 h with agitation. After brief centrifugation, the supernatant was transferred to a new tube (cold water extractable (CWE) mucilage fraction) and the volume of the pellet-containing tube was returned to 1 mL. The extraction was repeated again, but at 65 °C to obtain the hot water extractable (HWE) mucilage fraction. Finally, the pellet (seeds with the most extraction resistant mucilage fraction) were agitated on a tissue disruptor-type mill at 30 Hz for 10 min to obtain the intense agitation extractable (IAE) fraction. Collected fractions were freeze-dried to a constant weight and compared to starting mass of seeds to determine the yield: Water absorption capacity assay. The water absorption capacity was determined following Cowley et al. 51 with modifications. As the seeds were of variable size, 30 mg were weighed into 2 mL microcentrifuge tubes and 1 g of deionized water was added to each tube. After briefly vortexing to break surface tension and submerge seeds, the seed mucilage was allowed to expand, undisturbed, for 45 min at room temperature (25 °C). After 45 min, a 1000 μL laboratory pipette was used to remove unabsorbed water, avoiding disturbing the seeds and their mucilage. Removed water was weighed and the water absorption capacity was determined using the following equation: Monosaccharide analysis. Monosaccharide profiles of fractionated mucilage (redispersed at 1 mg/mL in Milli-Q water) and WSF were determined using reverse phase high performance liquid chromatography (RP-HPLC) of 1-phenyl-3-methyl-5-pyrazoline (PMP) derivatives following Cowley et al. 51 . Area under the peaks was compared to standard curves of mannose, ribose, rhamnose, glucuronic acid, galacturonic acid, glucose, galactose, xylose, arabinose and fucose 53 .
Whole seed soluble sugar extraction and profiling. Soluble sugars were extracted from 10 mg of WSF following Vespreet et al. 54 with modifications based on Matros et al. 55 . Due to substantial mucilage gelling in water when attempting a sequential 80% ethanol and water extraction, soluble sugars were twice extracted from flour in 80% ethanol at 85 °C for 30 min on a mixer (700 rpm) at a final dilution of 1:40 (w/v, mg/µl). Supernatants were combined, diluted with water to 1:25,000 (w/v, mg/µl) and 25 µl per sample were analysed by high pH anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) on a Dionex ICS-5000 system using a DionexCarboPAC TM PA-20 column (3 × 150 mm) with a guard column (3 × 50 mm) kept at 30 °C and operated at a flow rate of 0.5 mL min −1 . The eluents used were (A) 0.1 M sodium hydroxide and (B) 0.1 M sodium hydroxide with 1 M sodium acetate. The gradient used was: 0% (B) from 0 to 2 min, 20% (B) from 2 to 35 min, 100% (B) from 35 to 36.5 min, 0% (B) from 37.5-38.5 min. Detector temperature was maintained at 20 °C, data collection was at 2 Hz and the Gold Standard PAD waveform (std. quad. potential) was used. Data acquisition, processing, and peak integration were performed using the Chromeleon™ version 7.1.3.2425 software (Thermo Scientific). Compounds were annotated and quantified based on peaks of available analytical standards. Glucose, fructose, sucrose, raffinose, 1-kestose, maltose, nystose and stachyose analytical standards were purchased from Sigma-Aldrich, while 1,1,1-kestopentaose was obtained from Bio-Strategy. A standard of planteose was purified and prepared from Salvia hispanica seed mucilage as per Xing et al. 56 . Schematic structures were prepared using the DrawGlycan-SNFG online tool ( http:// www. virtu algly come. org/ DrawG lycan/) 57 .
Protein analysis. Protein content of 150 mg of WSF was determined by the Dumas method using a Rapid N Exceed bench top nitrogen analyser (Elementar, USA). Conversion factor was 6.25. Protein content was determined in triplicate.
Lipid analysis. Total lipid in 500 mg of WSF was determined by modified Folch method 58 and fatty acid profiles were determined by gas chromatography of transesterified lipids following Liu et al. 59 . Analyses were performed in duplicate at SAHMRI (Adelaide, Australia). Multivariate principle component analysis for separation was performed in PAST software (version 3.25) 60 .
Energy calculation. Energy content of whole seeds was estimated from the energy density of protein, carbohydrate, dietary fibre and fat components in whole Plantago seed 61 . Average protein, carbohydrate, dietary fibre, and fat contents (% w/w) were obtained from whole seed protein, soluble sugar, monosaccharide, and lipid profiling analyses, respectively.

Consent for publication.
All authors give consent for the data to be published.
Yield (%) = mass of freeze dried mucilage mass of seeds pre-extraction × 100 Water absorption capacity g/g = Initial weight of water added − weight of unabsorbed water Initial weight of seeds added Seed mucilage characteristics. There is significant variation in expanded seed mucilage architecture of the Plantago species studied here (Fig. 2). We corroborate the same characteristics and differences in ruthenium red phenotype of P. coronopus, P. lanceolata, P. ovata, P. major, P. cunninghamii, P. debilis and P. varia described in detail by Phan et al. 33 . Here we studied five additional Australian native species and found, owing to their relatedness ( Fig. 1), P. turrifera and P. bellidioides to be strikingly similar to P. cunninghamii and P. debilis, and P. gaudichaudii to be most similar to P. varia. These similarities were also found when immunolabelling the mucilage ( Fig. 2a-l). Two further native species, P. paradoxa and P. triantha, despite their apparent genetic similarity to other native species (Fig. 1), produced a negligible mucilage envelope that was poorly distinguishable with ruthenium red staining ( Fig. 2F and 3J) and only slightly more apparent with xylan-directed immunolabelling ( Fig. 2f and 3j).
In this study we used our previously published mucilage fractionation method 51 to investigate mucilage yield and fractionation traits (Fig. 3). There was significant interspecific variation in total yield (Fig. 3A) and fractionation profile (Fig. 3B) of seed mucilage. Total yield of mucilage was highest in P. ovata and comparatively lower in P. lanceolata, P. paradoxa, and P. triantha. Australian native species, P. cunninghamii, P. turrifera, P. bellidioides, P. debilis, P. gaudichaudii and P. varia had quite similar total yield (Fig. 3A) and fractionation profiles (Fig. 3B Figure 2. Expanded mucilage architecture of 12 Plantago species. Ruthenium red (dark pink) stains acidic polysaccharides at the mucilage periphery (A-L). Scale = 1 mm. Anti-xylan antibody LM11 (green) probes for xylan backbone epitopes in the inner mucilage layers (a-l) while the seed is counterstained with propidium iodide (red/yellow). Scale = 500 µm. www.nature.com/scientificreports/ and less mucilage could be extracted using cold water (CWE). To complement mucilage extractability traits, the water absorption capacity by whole Plantago seeds was determined (Fig. 3C). Water absorption capacity (WAC) was found to vary substantially with P. turrifera and P. bellidioides having a significantly higher WAC than other species, while absorption by P. lanceolata, P. paradoxa and P. triantha was very low. A correlation study was performed to determine if there was a link between mucilage fractionation and yield traits and the observed differences in WAC (Fig. 3D). Logically, significant (p < 0.05) correlations between WAC and total mucilage content and WAC and the more gel-like extraction resistant mucilage fractions were found. The strongest correlation (r = 0.73, p = 0.006) with WAC was the proportion of hot water-and intense agitation extractable (HWE + IAE) mucilage. The chemical composition of fractionated mucilage was determined by monosaccharide profiling which revealed both interspecific differences and trends (Fig. 4). Monosaccharide profiles were similar between all fractions of all species, with xylose and arabinose (constituents of heteroxylan) comprising the majority of monosaccharides quantified along with rhamnose and galacturonic acid (constituents of pectin) and minor quantities of glucose and galactose (Fig. 4A). Other monosaccharides for which standards were included, mannose,    www.nature.com/scientificreports/ ribose, glucuronic acid, and fucose, were minute or not detected (data not shown). The relative abundance of each monosaccharide detected in the mucilage fractions, however, differed significantly. The arabinose to xylose ratio differed interspecifically and also between fractions. In almost all species, the arabinose to xylose ratio rises with increasingly harsh extraction steps (Fig. 4B). In P. paradoxa, the inverse was observed and in the CWE fraction of P. triantha the value was outside the expected range of less than 1 (1.28). These species produced little mucilage and displayed poor monosaccharide recovery which explains some of the discrepancies. In particular, the CWE mucilage fraction from P. triantha was very small and the detected arabinose and xylose are likely not heteroxylan-derived. Values for the more abundant HWE and IAE fractions are within the expected range (< 1). The ratio of rhamnose to galacturonic acid was stable between fractions and differed minimally between species (Fig. 4A). Rhamnose and galacturonic acid, however, were found to be highly-enriched in the CWE fraction of all species (Fig. 4C) but much lower or even absent in subsequent fractions.
Plantago endosperm and embryo morphology and composition. Monosaccharide profiling revealed that mannose is a major non-cellulosic monosaccharide in whole seeds of all species, being the most abundant in 8 of 12 species studied here (Fig. 5A). Using a monoclonal antibody recognising β-(1,4)-mannan 62 it was observed that Plantago species have a substantial endosperm with thick (grand median = 8.206 μm, Supplementary Fig. 1) mannan-rich walls ( Fig. 5C-L). Minimal amounts of mannan were detected in the embryo but it was completely absent in the mucilage layer of all species except P. cunninghamii (Fig. 5G). However, mannose is completely absent in P. cunninghamii mucilage, even that tightly adhered to the seed 51 suggesting that the mannan labelling seen in the mucilage layer of P. cunninghamii is likely an artefact of non-specific binding in poorly-fixed mucilage. The endosperm cells also all contained tightly-packed spherical bodies ( Supplementary  Fig. 2) which are likely to be aleurone grains and/or oil bodies, similar to those seen in seed tissues of other species like Cannabis sativa 63 .
Soluble sugar profiling. Soluble sugar profiling was used to determine the content of rapidly-mobile non-starch reserve carbohydrates in the internal Plantago seed tissues (Fig. 6). Heat mapped chromatographic data (Fig. 6A) show that the soluble sugar profiles are similar between all species with common major components, with most of the interspecific variation coming from differing abundances of the major sugars, sucrose (β-d-fructofuranosyl-(2 → 1)-α-d-glucopyranoside) and planteose (α-d-galactopyranosyl-(1 → 6)-β-dfructofuranosyl-(2 → 1)-α-d-glucopyranoside) which account for up to 86% of the soluble sugar content (Supplementary Fig. 3). Some variation between profiles was evident between 2 and 5 min elution times but these components are (based on relative peak area; Supplementary Fig. 3) lowly abundant minor sugars, possibly unidentified mono-or di-saccharides. Quantification of extracted soluble sugars shows that the trisaccharide planteose is the major soluble sugar in seeds of the Plantago species studied here (up to 3.2% w/w), except in P. lanceolata where sucrose dominates, although followed by planteose (Fig. 6B). P. lanceolata also had the most unique soluble sugar profile in other ways: the trisaccharide raffinose (β-d-fructofuranosyl-(2 → 1)-α-dglucopyranoside-(1 → 6)-α-d-galactopyranoside) and its higher homologue, the tetrasaccharide stachyose (β-dfructofuranosyl-(2 → 1)-α-d-glucopyranoside-(1 → 6)-α-d-galactopyranosyl-(1 → 6)-α-d-galactopyranoside), which, present in minute quantities in other species, were additional major soluble sugars in P. lanceolata.     www.nature.com/scientificreports/ Currently unidentified, we also detected an additional sugar, Unknown 1 (Fig. 6A), eluting at approximately 14 min in extracts of all species. Based on the relative chromatographic peak area, Unknown 1 appears to be a major soluble sugar in seven of the species tested, comparable to planteose amounts in P. debilis or P. triantha ( Supplementary Fig. 3). Levels of other oligosaccharides included in the standard panel, 1-kestose, maltose, nystose, and 1,1,1-kestopentaose were minute or not detected in the species studied here. Figure 6. Soluble sugar profiling of Plantago seeds. (A) Chromatographic data presented in heatmap form to highlight differences in chromatographic profile between species. Chromatographic data has been cropped to an elution period that includes saccharides with a low degree-of-polymerisation (DP) (2-18 min). For simplicity, chromatographic data of standard runs (standard panel: a combined low DP saccharide suite, planteose and stachyose) were combined. (B) Quantification of key low DP saccharides and their schematic structure. A grey scale (dark grey, highest value; white, lowest value) has been used to display differences in abundance of one type between species. Values are means and standard deviation of three replicates. Abbreviations: 1,1,1-KPE = 1,1,1-Kestopentaose.  Fig. 4). Seed size is not origin-specific as both native and non-native species are included in the small-and large-seeded categories ( Supplementary Fig. 4).
To determine if there is any nutritional value in Plantago seeds that is normally overlooked, the seed nutrient (protein, carbohydrate, dietary fibre, and fat) composition was determined (Table 1). There is a significant amount of interspecific variation in the content of each nutrient. Compared to commercial psyllium (P. ovata) which lies at the extremes in three of four nutrients ( Supplementary Fig. 5), the other species studied have less dietary fibre but more fat and more protein, with native species tending to have the highest protein content. In carbohydrate content, P. ovata is close to the median. P. ovata and P. lanceolata have a lower energy content than all other species due to their low protein and fat content (~ 900 kJ vs. ~ 1100 kJ).
After determining that Plantago seeds contain a modest fat content, we investigated the fatty acid composition of the seed oil ( Table 1). The ratio of saturated (SFA) to unsaturated fatty acids (UFA) is relatively similar between species (Median = 1:5.2) but six of eight native species were slightly higher in UFA content, up to 33% more than P. ovata in P. gaudichaudii (1:6.4). The saturated fatty acid profiles are very similar between species with palmitic (16:0) and stearic acid (18:0) accounting for at least 94% of SFA species. The most variation in fatty acid profiles comes from differences in the UFA profile. Multivariate analysis shows that 89% of variance between the species is due to differences in the ratio of omega-3 to omega-6 UFAs ( Supplementary Fig. 6). All native species except P. triantha separate from commercial psyllium and the naturalised species because they contain a much higher quantity of omega-3 UFA in the seed oil; a significant difference between the non-native and native groups (p = 0.008, Supplementary Table 3). Commercial psyllium seed contains the least omega-3 UFA (3.2% w/w) of the species studied followed by the naturalised species (at most 28% w/w in P. coronopus). This is in clear contrast to the native species which contain an average of 39% omega-3 UFA (w/w), up to 54.5% in P. turrifera. When calculated as the omega-3 to omega-6 ratio, all native species have ratios greater than 1 (1.37-4.85) except P. triantha which has a very low ratio similar to non-native species which are all less than 1 (0.11-0.76). The yield of the major omega-3 UFA, alpha-linolenic acid (18:3) is particularly high in native species where, again, P. turrifera has the highest yield of 6.27% (w/w). There is also a small amount of variation (8%) from differences in the omega-6 to omega-9 UFA ratio. The omega-6 to omega-9 ratio was fairly similar between native species and much greater in naturalised species (up to 2.6 times greater than the native species average). Transaturated fatty acids were not present (data not shown).

Discussion
In this study we present a comprehensive profiling of major nutritional and functional components of mucilageproducing seeds of 12 Plantago species that grow in Australia.
Mucilage comprises a major part of Plantago seeds and is the main factor currently influencing its use as a food additive and digestive health ingredient. Natural variation in Plantago mucilage properties is apparent even in the expanded mucilage architecture (Fig. 2). When extracting the seed mucilage in sequential fractions there are significant differences in WAC, yield and resistance to extraction (Fig. 3) and such variation may be explained by compositional differences (Fig. 4). The most easily-extracted Plantago seed fraction (CWE) contains abundant pectic monosaccharides, rhamnose and galacturonic acid, which are minimal in subsequent fractions (Fig. 4C). This is in line with our previous findings which suggest that the pectin in highly hydrophilic first mucilage fractions acts to 'prime' and initiate mucilage expansion 9 . Subsequent mucilage fractions which are obtained using more intense extraction conditions contain mostly xylose and arabinose, constituents of heteroxylan, in varying molar ratios. Yu et al. 64 showed that the innermost layer(s) of mucilage in P. ovata was most gel-like as a result of high heteroxylan substitution complexity, corroborated in our previous publication 9 and by others 65 www.nature.com/scientificreports/ fractions (akin to the inner layers described by Yu et al. 64 ) had the highest AX ratios, an indication of heteroxylan substitution complexity. Similarly, the HWE and IAE mucilage fractions of P. cunninghamii, P. turrifera, P. bellidioides, P. debilis, P. gaudichaudii and P. varia had particularly high AX ratios suggesting even more complex heteroxylan substitution (Fig. 3B), as indicated by minimal immunolabelling by LM11 because the unsubstituted/ lowly-substituted backbone epitope is not present (Fig. 2e, g, h, i, k, and l). The inner mucilage fractions in these species also represented a large proportion of the total mucilage (Fig. 3B) and analysis of the WAC of mucilage on intact seeds (Fig. 3C) found WAC to be strongly correlated with both HWE (r = 0.62, p = 0.032) and IAE content (r = 0.69, p = 0.013) but the strongest correlation was a combination of the two (r = 0.73, p = 0.006) (Fig. 3D). Furthermore, the mucilage of these species is still apparent as retained, swollen envelopes even after extraction (compared to P. lanceolata, for example) (Supplementary Fig. 7) showing that the robustness of seed mucilage in Australian Plantago species is likely due, at least in part, to heteroxylan structure and resultant intermolecular bonds. It should be noted however that mucilage characteristics and composition appear to be at least partly independent of phylogenetic lineage and non-native/native status (p > 0.05, Supplementary Table 3). Instead, preliminary findings associating mucilage content and water absorption capacity with key environmental factors ( Supplementary Fig. 8) suggest that these strong macromolecular properties may have developed to ensure germination success in climatically different Australian environments, as has already been demonstrated for Plantago 67 and other species 68,69 . Importantly, the strong macromolecular properties in some Australian Plantago species like P. turrifera or P. bellidioides may also give them improved functionality over commercial psyllium in food and health applications where strong rheological/viscous properties are advantageous. Along with the seed mucilage, we have also shown here that the internal tissues of Plantago seeds contain a wide array of saccharide families that represent dietary fibres that are typically underutilised during commercial psyllium production. After a review of the literature, we found that while Plantago species are known to have a substantial endosperm 70,71 , its exact composition has not been defined. The hard endosperm cell walls of P. ovata were suggested to be comprised solely of reserve cellulose 72 but crystalline cellulose levels were found to be, at most, 4% (w/w) 73 . A clue is found during P. ovata seed development where mannan synthases are highly expressed along with the accumulation of mannose monosaccharides 74 . Here we are the first to confirm that the endosperm of all Plantago species studied contains thickened cell walls that are rich in mannan (Fig. 5). The substantial endosperm cell walls, like other species 75 , gives Plantago a 'hard-seeded' quality 76,77 (for example, seeds of P. ovata require over four times the specific grinding energy compared to wheat 42,78 ) but also results in a larger proportion of dietary fibre carbohydrates in the absence of starch 73 . In addition to the abundant endosperm polysaccharides, soluble sugars are also abundant in the Plantago seeds studied here (Fig. 6), in agreeance with previous reports 79 . Planteose (originally isolated from Plantago and named for the genus 80 ) is the major soluble sugar in seeds of all Plantago species studied here except P. lanceolata where only sucrose is higher. While minute quantities of the oligosaccharides raffinose and stachyose were detected in all species, these sugars are only abundant in P. lanceolata seeds where stachyose levels were even similar to planteose. This agrees with previous findings that raffinose and stachyose were only detected in seeds of Plantago Sect. Arnoglossum of which P. lanceolata is the only member included here 81 . An additional sugar, Unknown 1 (Fig. 6A), was also found in all species. Based on relative chromatographic peak area, Unknown 1 appears to be a major soluble sugar in seven of the species tested, with levels even comparable to planteose in P. debilis or P. triantha (Supplementary Fig. 3). As the chromatographic method used here elutes soluble sugars roughly by their degree-of-polymerisation (DP), it is likely that Unknown 1 is a tetra-or pentasaccharide (DP4-5) as it elutes after nystose (DP4) but before 1,1,1-kestopentaose (DP5). While the exact identity of this compound will be elucidated by mass spectrometry, we hypothesise that Unknown 1 is likely to be the tetrasaccharide sesamose (α-d-galactopyranosyl-(1 → 6)-α-dgalactopyranosyl-(1 → 6)-β-d-fructofuranosyl-(2 → 1)-α-d-glucopyranoside), a higher DP homologue of planteose, which while not reported previously for Plantago, is found in tandem with planteose in other species 82 . The array of soluble sugars and mannan polysaccharides in internal tissues of Plantago seeds are likely to be the primary carbon source fuelling germination 83,84 , however they also have medicinal and nutritional significance as fermentable dietary fibres. Fermentable dietary fibres are consumed by beneficial bacteria in the human gut, producing short chain fatty acids like butyrate which is required to maintain colon health 85 and there is also mounting evidence that modulating the microbiome through dietary fibre supplementation has significant effects on a wide range of disease indications (for review see Lynch & Pedersen 86 ). While P. ovata husk is not a fermentable dietary fibre and thus provides limited fermentation-and microbiome-related benefits 87 , whole P. ovata seeds are reported to be fermentable, increasing faecal butyrate content in a clinical trial 88 and improving numerous digestibility markers compared to the husk in a pre-clinical study 89 . These findings demonstrate that at least one non-husk component in Plantago seeds is fermentable by gut microbiota. While its exact prebiotic properties are currently unknown 56 , planteose, along with its homologues and other soluble sugars in Plantago seeds are likely fermentation candidates as many oligosaccharides are readily fermentable by beneficial bacteria in the gastrointestinal tract 90,91 . Endosperm mannan is also possibly fermentable compared to husk as diverse mannans and mannooligosaccharides released from mannan digestion have been shown to be well-fermented by a human faecal inoculum 92,93 .
Within the cells of the endosperm are large aleurone grains and oil bodies which house the protein and fat content of the Plantago seeds. Protein levels are generally similar to previous reports of Plantago species (10-20% w/w) 42,94-96 , but Australian Plantago species are generally higher compared to these reports and all are higher than commercial psyllium (P. ovata) ( Table 1). P. paradoxa and P. triantha are very high in protein content, over 30% (w/w), which is higher than important grain legumes like chickpeas or lentil (24% and 26.1%, respectively) 97 . Plantago protein has previously been reported to contain essential amino acids and is well digested 95 showing the potential of Plantago seeds as a protein source. Seeds of Plantago species are also rich in fats and all Australian species contain more fat than commercial psyllium, P. ovata (Table 1). Of particular nutritional importance is the proportion of omega-3 and omega-6 UFAs which mammals are unable to synthesise and are thus essential. www.nature.com/scientificreports/ There is a growing body of evidence that omega-3 fatty acids, particularly long-chain fatty acids (LCFAs) EPA and DHA commonly obtained from consuming fish, are protective against cardiovascular diseases 98 . However many researchers conclude that the world's ecosystems could not sustainably supply the population with enough fish to provide the recommended EPA and DHA intake 99 and thus sufficient intake of plant-derived UFAs to allow endogenous LCFA production from these precursors is recommended as an alternative. As plant-derived omega-3 (anti-inflammatory) and omega-6 (pro-inflammatory) 100,101 UFAs are competitively desaturated and elongated into LCFAs by the same pathways, it is important that the ratio of omega-3 to omega-6 is at the very least 1:1 102 to reduce the risk of inflammation. However an omega-3 to omega-6 intake ratio of 4:1 is suggested to be ideal for promoting heart, liver and gut health 103 . Here we show that Australian native Plantago seeds have omega-3 to omega-6 ratios greater than the minimum adequate ratio of 1:1 (Table 1) and after extensive review of the literature we report that P. turrifera has the highest ratio of omega-3 to omega-6 fatty acids reported for any seed (4.85:1). This is higher than benchmark species Linum usitatissimum (flax) or Salvia hispanica (chia), with 3.44:1 and 3.08:1, respectively 104 , although total fat content was 50% and 30% lower in P. turrifera than these species, respectively. Consumption of Plantago seeds, particularly those of Australian natives like P. turrifera, may therefore contribute to improving the undesirable omega-3 to omega-6 ratio that is common in modern, western diets, and thus diminish the associated health risks 100,102 .

Conclusions and future directions
In this study we have provided a more complete understanding of the composition and morphology of seeds of twelve Plantago species that grow in Australia and outline implications these differences may have for food and nutritional applications. We have shown that among the species studied here there is variation in seed mucilage content and macromolecular properties with some native species having a higher capacity to absorb water than commercial psyllium, a key predictor of polysaccharide properties. We hypothesise that these differences are a result of fraction-and species-specific differences in heteroxylan and pectin content/structure and may represent wide differences in hydrocolloid functionality that can be exploited in industry. By profiling the nutrient content of the inner Plantago seed tissues (endosperm and embryo), we show that current commercial production of psyllium in particular underutilises the nutritional value of whole Plantago seeds. The use of a whole seed Plantago product, particularly from those species with favourable mucilage properties, and that are native and adapted to harsh Australian conditions, may provide a more nutritious but still functional alternative to commercial psyllium husk. In this way, the use of a Plantago WSF could confer multiple benefits: (1) hydrophilic mucilage polysaccharides could rheologically modify food systems and improve their quality; (2) minimal fermentation of the mucilage polysaccharides could benefit metabolic and gastrointestinal disorders that are improved when the viscosity and consistency of the gut digesta is increased; and (3) fermentable fibres and lipids along with other phytonutrients derived from the internal seed tissues could contribute to the improvement of a wide range of microbiome-, inflammatory-and oxidative stress-related disorders. Future work will investigate how the specific differences in composition and properties outlined here impact the functionality of Plantago WSFs in food technology and gut health and aim to unravel the drivers of natural variation. The differences in seed mucilage content and properties appear to be mostly independent of phylogenetic lineage, however preliminary data suggests that differences among Australian native Plantago species might have developed as an 'adaptive trait' to cope with water-deficit stress in arid Australian environments suggesting that climate might be an important factor. Future work will aim to validate these preliminary findings by scrutinising more Plantago accessions that encompass more of the genus' environmental range and provide insight into the intraspecific heterogeneity in the traits tested here. Probing this 'adaptive trait' hypothesis may aid in prospecting for ideally-adapted Plantago accessions with specific seed compositions for development as novel crops with tailored industrial end-uses.

Data availability
The datasets used and analysed during this work are available from the corresponding author upon reasonable request.