Identification of fasciclin-like arabinogalactan proteins in textile hemp (Cannabis sativa L.): in silico analyses and gene expression patterns in different tissues

Abstract

Background

Arabinogalactan proteins (AGPs) are cell surface glycoproteins belonging to the hydroxyproline-rich glycoprotein superfamily ([1] and references therein) which are involved in many aspects of plant development, i.e. pattern formation, phytohormone interaction, tissue differentiation, reproduction, response to (a)biotic stresses, cell expansion and secondary cell wall deposition [2, 3]. These heavily glycosylated proteins are subdivided into four main classes: classical AGPs, AG peptides, Lys-rich AGPs, fasciclin-like AGPs (FLAs) [3–6].

FLAs are characterized by the occurrence of one or two AGP domains, as well as one or two fasciclin (FAS) domains [7]. FAS domains were first identified in the fruit fly Drosophila melanogaster and later found in many other organisms, from bacteria to higher plants to animals [5]. Although a consensus sequence for the FAS domains is lacking, two regions are highly conserved, named H1 and H2 (of ca. 10 amino acids) [5]. Additionally, most FLAs show an N-terminal signal peptide and a C-terminal glycosylphosphatidylinositol (GPI) membrane anchor [5, 7], mediating attachment to the cell surface.

FLAs constitute multigene families in plants: for example, 21 FLAs have been identified in thale cress, 24 in rice, 35 in poplar, 34 in wheat, 19 in cotton, 33 in chinese cabbage and 18 in eucalypt [5, 7–11]. Molecular studies focused on FLAs are important, since they increase our understanding of the molecular functions of this protein family: the available literature on the topic has shown that FLAs in plants are not only related to tissue-specific functions, but also involved in generalized responses to environmental constraints, both biotic and abiotic [3, 7, 11, 12].

Additionally, a strong body of evidence in the literature has highlighted the importance of FLAs in regulating aspects linked to cell wall biosynthesis and, more generally, to stem mechanics in both herbaceous and woody species, as well as fibre growth. For instance, in Arabidopsis, insertional mutants of Atfla11 and Atfla12 and Atfla11/fla12 double mutants show modified stem mechanics, due to a decrease in cellulose, arabinose and galactose in secondary cell walls [12]. Likewise, in Eucalyptus, FLAs belonging to the subgroup A [5, 12] are involved in stem mechanics [11]: in particular EgrFLA2 is linked to cellulose microfibril angle. In poplar, antisense expression of PtFLA6 alters secondary cell wall composition in the xylem, by affecting the biosynthesis of lignin and cellulose [13]. In cotton, GhFLA1 is involved in fibre initiation and elongation: its overexpression increases fibre length, while its silencing results in shorter fibres with an altered primary cell wall composition [14]. In the fibre crop flax, some FLAs were shown to be upregulated at the snap point, a physical region marking the transition from elongation to cell wall thickening, hence confirming the potential function of these genes in the regulation of fibre development [15, 16].

Textile hemp (Cannabis sativa L.) is an economically important bast fibre-producing crop, with several applications in industry, namely the biocomposite, textile, construction sector [17]. This plant is not only important as a multi-purpose crop, but also useful for fundamental studies centered on cell wall biosynthesis/remodeling [18], because its stem tissues show strong differences in cell types and cell wall composition [19, 20]. The core of hemp stems (a.k.a hurd/shiv) is indeed woody, while the cortex harbors long gelatinous fibres, the bast fibres, with a high content in crystalline cellulose and poor in lignin [21]. The different stem heights correspond to distinctive stages of bast fibre development (from intrusive growth to thickening; Fig. ​Fig.1a).1a). It is hence possible to study the mechanisms involved in the development of cellulosic and woody fibres by separating the stem tissues of the same plant. The cortex can be peeled from the hurds and the bast fibres can be separated from the surrounding parenchymatic cells with the use of 80% ethanol, a mortar and a pestle [20, 22, 23].

Fig. 1

Sampling strategy of the stem tissues. a Stem cross-sections of the sampled internodes showing the progressive thickening of bast fibres and the development of the xylem tissue. b The segments excised from the sampled internodes are depicted, together …

The molecular steps involved in the regulation of bast fibre initiation, development and intrusive growth comprise many still unexplored aspects [24–26]; hence an increased knowledge in these mechanisms would favor the development of biotechnological tools focused on bast fibre improvement.

In the light of the above-mentioned relationships between FLAs and cell wall-related processes and considering the industrial applications of C. sativa, we here sought to identify and study the expression patterns of hemp FLA genes in the different stem tissues, as well as in other organs. By using bioinformatics coupled to RT-qPCR, we show that some FLA genes are highly expressed in bast fibres. Moreover, we identify groups of FLAs, upregulated either at the top or the bottom of the stem, which share putative conserved elements in their promoters. Our study therefore lays the foundation to further molecular analyses on a unique family of proteins in an important herbaceous crop.

Methods

Results

Identification of putative FLAs in C. sativa: Protein architecture and phylogenetic analysis

BLAST/BLAT analyses of the 21 A. thaliana sequences (AtFLAs) performed against the Medicinal Plant Genomics Resource, the NCBI EST and the Cannabis Genome Browser Gateway databases led to the identification of 23 CsaFLAs (Additional file 3: Table S3). It should be noted that, during the database queries, a contig, i.e. csa_locus_44222_iso_1_len_407_ver_2, which was initially called CsaFLA22 and retrieved at the Medicinal Plant Genomics Resource (MPGR), was also found. However, we believe that this partial gene was erroneously attributed to C. sativa, since we never amplified any product with different primers designed on it and the reported FPKM values at the MPGR are 0 for all the tissues examined. We discarded this gene from our analyses, but kept the original nomenclature given to the hemp FLA genes (i.e. CsaFLA1–24), as at this stage we cannot rule out the existence of this gene in textile hemp.

The intron-exon structure analysis highlighted the presence of 5 genes (CsaFLA2–17–8–12–4) containing one intron: of these genes, CsaFLA12 and 4, both possessing a small intron, group together according to the maximum likelihood (ML) phylogenetic tree, while CsaFLA2–17–8, containing longer introns, are in a different branch (Additional file 4: Figure S1). To check that the putative CsaFLAs belong to the FLA family, the occurrence of the following features was checked: the presence of at least one FAS domain, a signal peptide at the N-terminus, (in some cases) a GPI anchor at the C-terminus and the presence of AGP domains (Additional file 5: Table S4; Additional file 6: Figure S2). The identified FLAs show a PAST% ranging from 19.4 to 44.5% (Table ​(Table1),1), which is in agreement with the values reported for poplar FLAs [7].

Table 1

Proline, Alanine, Serine, and Threonine (PAST) proportions in CsaFLAs

In order to investigate the evolutionary relationship between known plant FLA proteins, a maximum-likelihood phylogenetic tree was built using the identified 23 CsaFLAs, 21 AtFLAs, 18 Eucalyptus (EgrFLAs) and 35 poplar sequences (PtrFLAs) (Fig. ​(Fig.22 and Additional file 6: Figure S2). It should be noted that we chose to perform the phylogenetic analysis using full-length CsaFLA sequences to conform to the previously published tree of poplar FLAs [7]. Different results may be obtained if the mature protein sequences are used.

Fig. 2

The phylogenetic tree of FLA proteins from hemp, poplar, eucalypt and thale cress. The maximum likelihood tree was constructed with 23 FLA proteins in hemp, 18 in eucalypt [11], 35 in poplar [7] and 21 in Arabidopsis [5]. Only bootstrap values higher …

CsaFLAs cluster into four major classes (A-D). Class A is the largest clade, with 11 hemp members (CsaFLA3/6/7/9/11/12/13/15/16/18/19) containing a single FAS domain flanked by two AGP domain and a GPI anchor at the C-terminus. Class B includes CsaFLA5/8/17 which contains two FAS domains and a single AGP region. Class C comprises 5 members (CsaFLA1/2/4/10/14), characterized by two FAS, two AGP domains and a GPI anchor. The last class includes 4 proteins (CsaFLA20/21/23/24) with no distinctive protein architecture (although CsaFLA20 and CsaFLA23 have 2 FAS domains like group B, however their lengths are smaller). The percentage of CsaFLA identity with the putative orthologs from thale cress is between 32 and 76% (Table ​(Table22).

Table 2

Identities of CsaFLAs with AtFLAs

CsaFLA expression patterns in hemp tissues

An immunohistochemical analysis carried out with the LM14 antibody (recognizing AGPs) revealed that the epitope is distributed in different tissues of the hemp stem (Additional file 7: Figure S3): this result shows the broad distribution of these proteins in the different hemp stem tissues. In particular, in the bottom internode, AGPs are present in the core tissues (cell walls of fibres and vessels), cortical parenchyma/collenchyma and, notably, in the inner region of the fibres, i.e. the layer (plasma membrane) delimiting the fibre cell lumen.

Of the 23 CsaFLAs identified, 22 were expressed in the stem tissues (Fig. ​(Fig.3;3; Additional file 8: Figure S4). CsaFLA14 was detected at very low levels in the stem tissues (Ct > 32).

Fig. 3

Expression profiles of CsaFLAs in hemp stems. Hierarchical clustering of expression profiles of 22 CsaFLAs in the three regions of stem. Data in the form of bar graph are shown in Additional file 8: Figure S4

In hemp bast fibres, the heat-map hierarchical clustering shows 5 major expression trends (Fig. ​(Fig.3).3). These are the following: 1) a group of genes (CsaFLA2–6–24) is upregulated at the middle internode containing the snap point (in the core the expression decreases towards the base of the stem); 2) CsaFLA1–4–7–8–10–20–23 are expressed at higher levels in the top and decreased towards the bottom internode; 3) two FLAs, CsaFLA5 and 21, are downregulated at the snap point; 4) three genes, CsaFLA9–11–17, show a tendency to upregulation at the snap point, although the pattern is less marked with respect to group I (and in the core the expression increased towards the stem base); 5) the last group comprises FLAs upregulated at the bottom (CsaFLA3–12–13–15–16–18–19).

Gene expression analysis was carried out on other hemp tissues, i.e. leaves, roots, male/female flowers and seedlings, to check whether hemp FLAs expressed at low levels in the stem showed a distinctive expression pattern in other hemp tissues (Fig. ​(Fig.4;4; Additional File 9: Figure S5). The heat-map hierarchical clustering reveals the presence of 4 main expression patterns (Fig. ​(Fig.4).4). More specifically, a group of genes, comprising CsaFLA1–4–5–6–8–10–17, is expressed at higher levels in hemp leaves; in group II, two genes specifically upregulated in male flowers are present (CsaFLA14 and 23); group III comprises 9 FLAs expressed at higher levels in hemp roots (CsaFLA3–7–11–12–13–15–16–18–19); the fourth group is represented by CsaFLA2–9–20–21–24, which are overall more expressed in female flowers.

Fig. 4

Expression profiles of CsaFLAs in the other tissues. Hierarchical clustering of expression profiles of 23 CsaFLAs in the leaves, roots, seedlings, male/female flowers. Data in the form of bar graph are shown in Additional file 9: Figure S5

The expression of some FLAs belonging to the above-mentioned 5 stem groups has also been investigated in hypocotyls, aged 6-9-12-15-17-20 (H6 to H20). The hemp hypocotyl was proven to be a suitable model to study cell wall-related processes accompanying secondary growth [18], therefore the goal was to verify whether their expression pattern highlighted the same trend observed in adult stems. CsaFLA1 –2–8–21 were more expressed in young hypocotyls (H6); CsaFLA3 –9–11–13 were more expressed in H15, H17 and H20 (Fig. ​(Fig.55).

Fig. 5

Expression analysis of 8 CsaFLAs in the hypocotyls of C. sativa. Error bars indicate the standard error of the mean (n = 3). Different letters indicate statistically significant values at the one-way ANOVA test (p < 0.05) …

Identification of conserved motifs in the promoters of some CsaFLAs

A bioinformatic analysis was carried out on the available promoter regions (between 116 and 1064 bp retrieved at the Cannabis Genome Browser Gateway) of the CsaFLA genes showing a distinctive expression pattern in the bast fibres. The promoters of the genes within group I, II and V were selected (Additional File 10), in the light of their upregulation at the snap point, at the younger and older stem regions respectively (Fig. ​(Fig.3).3). While no conserved motifs could be found with the MEME suite tool for the genes in group I, 1 motif was found for the genes within group II and V (Table ​(Table2).2). The search carried out with the conserved sequence of group II and V FLAs in the JASPAR CORE 2016 plants database identified SOC1 (lowest p-value among the matches retrieved) and MYB3 as candidates recognizing similar motifs (Table ​(Table3).3). In addition to that, the promoters of three representative CsaFLAs showing upregulation at the snap point, at the top, or at the bottom (CsaFLA2, CsaFLA7 and CsaFLA16) were amplified and cloned. The purpose was to verify the sequence conservation between the fibre variety Santhica 27, the Purple Kush strain and Finola variety [28]. The motifs identified by MEME were cross-verified with PlantPAN 2.0 (http://plantpan2.itps.ncku.edu.tw/) and are highlighted in yellow in the promoters of CsaFLA7 and CsaFLA16 from Santhica 27 (Additional File 11).

Table 3

Conserved motifs in the promoters of FLAs from group II and V

Domains, conserved regions, motifs and residues mediating adhesion in CsaFLAs from class A and C.

Homology models of one representative each from Class C (FLA 10) and Class A (FLA 11) showing FAS, AGP and GPI anchor domains, H1 and H2 conserved regions, putative amino acids involved in adhesion and [YFL]H motif are given in Fig. ​Fig.66 and Table ​Table4.4. The conserved sequences for FLA10 for the first FAS domain are LTVLVLSNGA and ISILEISAPII for H1 and H2 regions respectively and the [YF]H motif is AL-X-LH-VV (Fig. ​(Fig.6a;6a; Table ​Table4).4). The conserved sequences for FLA10 for the second FAS domain are LTLFAPNDEA and LVIFTVDNVL for H1 and H2 regions respectively and the [YF]H motif is VL-X-YH-SL (Fig. ​(Fig.6a;6a; Table ​Table4).4). The conserved sequences for FLA11 for a single FAS domain are ITVFAPTDSA and LSVFEVDQVL for H1 and H2 regions respectively and the [YF]H motif is LV-X-YH-VL (Fig. ​(Fig.6b;6b; Table ​Table44).

Fig. 6

Modeling of representative CsaFLAs. a Homology model of FLA10 showing right and left views. Left: ribbon model showing secondary structures. Right: space-filled model. Both fasciclin (FAS) domains are shown in turquoise and the H1 and H2 conserved regions …

Table 4

Solvent accessibility of residues within conserved motifs

The two hydrophobic residues preceding and after [YF]H motif are thought to be implicated in mediating adhesion of these proteins. It is interesting to note that these residues are generally aliphatic amino acids such as I, L and V [5], however in FLA10 an Ala and polar Ser are found in the first and second FAS domains respectively (Fig. ​(Fig.6a6a).

It is noteworthy that [YF]H residues are either completely or partially buried, however both residues that flank [YF]H motif on the N-terminal side (AL, VL, LV) are completely solvent-exposed in both FLA10 and FLA11 (Table ​(Table4).4). In contrast, aliphatic residues (VV) on the C-terminal side of [YF]H motif in H1 domain of FLA10 are either completely or partially buried, whereas polar Ser is partially buried, but Leu is exposed in the H2 domain (Fig. ​(Fig.6a,6a, space-filled). For FLA11 both VL residues on the C-terminal side of [YF]H motif are solvent exposed (Fig. ​(Fig.6b,6b, space-filled). In general, residues belonging to FLA11 and those located towards the N-terminal side are exposed more favorably to mediate adhesion than FLA10 and those located on the C-terminal side. This suggests that adhesion for both these proteins may be mediated via hydrophobic interactions.

Discussion

The FLAs identified in C. sativa group into the previously described four phylogenetic classes (Fig. ​(Fig.2)2) [5]. A nomenclature of CsaFLAs is hereby also proposed which follows the Arabidopsisclassification (i.e. when the phylogenetic tree highlighted clustering of a CsaFLA proteins with a specific AtFLA, the same number was assigned to C. sativa).

Within class A, the largest, it is possible to observe a separate clade represented by CsaFLA3-12-13-15-16-18-19 which is highly expressed at the snap point and in the older stem regions, both in the bast fibres and the shivs (Fig. ​(Fig.3).3). A subset of class A genes (CsaFLA3-9-11-13) was more expressed in the old hypocotyls (peaking at H17 with high values at H15 and H20). As previously shown [18], the hypocotyl undergoes secondary growth in H9 and later time-points. The phylogenetic position of this cluster of FLAs, together with their common expression pattern, might indicate a specific role in secondary growth. This group of genes may indeed represent hemp-specific single FAS domain FLAs specialized in secondary growth, in a manner analogous to what was previously shown in eucalypt and thale cress [11, 12]. Hemp is unfortunately recalcitrant to transformation, therefore homologous testing, as previously performed on e.g. eucalypt FLAs [11], is cumbersome. However heterologous testing in a more amenable system, e.g. Nicotiana tabacum, can confirm or refute the hypothesis.

It is here worth discussing also the phylogenetic position of CsaFLA11 in a clade grouping AtFLA11, EgrFLA2b and EgrFLA3b (Fig. ​(Fig.2).2). These genes were shown to affect stem mechanics, as well as cell wall architecture [11, 12]. The AtFLA11 transcript was detected in the xylem and interfascicular fibres in inflorescence stem, preceding the lignification of those two tissues [37]; CsaFLA11 also shows a gradual increase in expression towards the older regions of the stem and it is slightly more expressed in the older hypocotyl too (Fig. ​(Fig.5).5). This FLA represents another interesting candidate putatively involved in cell wall-related processes in textile hemp.

Within class C, CsaFLA4 and CsaFLA1 group together with the characterized orthologs from thale cress. AtFLA4 (SOS5) is involved in cell expansion [38] and AtFLA1 was shown to regulate root and shoot development in tissue culture [39]. CsaFLA8 was more expressed in the TOP region of the stem, as well as in H6, suggesting a role in elongating tissue. However, it remains to be shown whether the hemp genes are involved in the same regulatory networks as in Arabidopsis.

The expression of the 23 CsaFLAs was first investigated in the different tissues of the stem, because we wanted to identify those genes specifically associated with a tissue-type and a stem region. Among them, we would like to draw the reader’s attention on the first group of genes, represented by CsaFLA2–6–24, because they show a different expression profile in the bast fibres and the shivs. The expression in the shivs shows a decrease from the top to the bottom of the stem, while in the bast fibres their expression peaks at the snap point. This is quite interesting if we consider that the snap point is the region marking a shift in the stem mechanical properties, as it determines the transition from cell elongation to thickening [27]. It was shown that the young stem regions of hemp at the vegetative stage of growth are characterized by the presence of ca. 66% glucose, while older regions have about 82%: this result confirms that during their transition from elongation to thickening, bast fibres require great amounts of glucose for the synthesis of cellulose [40]. The 3 FLAs may therefore be involved in cell wall-related processes occurring during this transition. Additionally, this is in agreement with the flax microarray data showing upregulation of certain FLAs around the snap point [15] and with the increased expression of poplar FLAs in tension wood, which, like bast fibres, is composed of a cellulosic G-layer [41, 42]. As previously discussed for poplar tension wood, specific FLAs with a GPI-anchor might be involved in the cytoskeleton-cell wall connections during fibre expansion/elongation [41]. This would be the case of CsaFLA2 and CsaFLA6, which possess a GPI-anchor (Additional file 5). In the hypocotyl, CsaFLA2 was significantly more expressed in H6 (Fig. ​(Fig.5).5). FLAs might also be involved in triggering a cellular signal inducing the formation of the G-layer, via the cleavage of their GlcNAc oligosaccharides by the action of chitinases [22, 41]. It was shown that in flax stems, specific chitinases are highly expressed in bast fibres and may regulate G-layer formation in these cell types [22]. Therefore, it is reasonable to assume that the concerted action of specific FLAs and chitinases may be involved in the transition from elongation to G-layer formation in hemp.

In group II and V are FLAs which, in the bast fibres, show a gradual decrease from the apical to the basal part of the stem and an increase in expression, respectively. A similar trend was observed in the hypocotyls: CsaFLA8 (belonging to the stem group II) was more expressed in H6; CsaFLA13 (belonging to the stem group V) was more expressed in H15, H17 and H20 (Fig. ​(Fig.5).5). In addition, the hypocotyl expression pattern of CsaFLA3 was similar to the one of CsaFLA13. Our study therefore identified specific FLAs likely involved in bast fibre elongation during intrusive growth (CsaFLA1–4–7–8–10–20–23) and others involved in secondary cell wall deposition during the thickening stage (CsaFLA3–12–13–15–16–18–19).

The expression of hemp FLAs was also investigated in other tissues, notably leaves, roots, male/female flowers and in seedlings (Fig. ​(Fig.44).

The genes belonging to group III in stem tissues (Fig. ​(Fig.3)3) are highly expressed in roots: within this cluster of FLAs are the orthologs of AtFLA7 and AtFLA11 (Fig. ​(Fig.2)2) for which a higher number of ESTs was retrieved in the roots of thale cress [5].

In reproductive organs, the RT-qPCR results show that some genes are highly expressed in male and female flowers. This suggests that some FLAs are involved in hemp inflorescence formation.

In seedlings, two genes showed higher expression, i.e. CsaFLA2 and 24 (Fig. ​(Fig.4).4). Notably, CsaFLA2 is the ortholog of AtFLA2 (Fig. ​(Fig.1),1), whose EST number is higher in Arabidopsis flower buds [5].

In order to investigate whether specific regulatory elements occurred in the promoters of the genes showing specific expression patterns in the stem tissues, we analyzed the genes from group I-II and V (Fig. ​(Fig.3).3). While for group I no conserved motifs could be obtained, 2 conserved sequences were found for group II and V (Table ​(Table2).2). A conserved motif recognized by the MADS box transcription factor SOC1 could be identified in the promoters of the genes upregulated in the apical stem regions: this finding suggests that they may be involved in a developmental program regulating the transition from vegetative to reproductive growth and/or the response to hormonal regulation (e.g. via gibberellin). In this respect it is noteworthy that in A. thaliana SOC1 was shown to control the annual growth habit [43]: soc1 ful mutants show indeed woody growth reminiscent of the perennial lifestyle. Hence the FLAs upregulated at the top of the stems might belong to a regulatory circuit controlling elongation and suppressing secondary growth.

The genes in group V show the presence of a conserved motif putatively recognized by MYB3, which is an R2R3 MYB transcriptional repressor belonging to subgroup S4 together with the characterized AtMYB4 [44]. MYB4 negatively regulates phenylpropanoid biosynthesis (more specifically, in thale cress it is a negative regulator of hydroxycinnamic acid metabolism and it exerts its silencing function by displacing the activators binding to the MYB motifs present in many promoters of genes involved in the phenylpropanoid metabolism; [44]). It is therefore possible that the identified element is involved in the coordination of phenylpropanoid biosynthesis in bast fibres and might regulate the hypolignification observed in these cells [45, 46]. In our recently-published transcriptomic dataset [47], we observed an upregulation of the SOC1 gene at the top (4-fold induction with respect to the bottom and 1.3-fold induction with respect to the middle) and MYB4 at the bottom (1.7-fold induction with respect to the top and 4.6-fold induction with respect to the middle). This result therefore strengthens the existence of a putative regulatory circuit (controlling, among other genes, the expression of CsaFLAs) at the top and bottom of adult hemp plants.

Conclusions

In conclusion, our work has identified (at least) 23 genes coding for FLAs in textile hemp, some of which specific to distinct stages of bast fibre development. Bioinformatics has highlighted the occurrence of conserved motifs in the promoters of genes upregulated either at the top or at the bottom of the stem. This finding points to the existence of a fine regulatory network controlling bast fibre elongation and cell wall composition. Future functional analyses carried out on heterologous systems will shed more light on the functions of the identified genes.

Additional files

Acknowledgements

Authors’ contributions

GG conceived and designed the experiments; LM-P performed the experiments; MB contributed to the RT-qPCR and bioinformatics analyses; KSS performed protein modeling; GG, LM-P, SL, MB, St.L, KSS and J-FH analyzed the data; GG, LM-P, SL, MB, St. L, KSS and J-FH wrote the paper. All authors read and approved the final manuscript.

Notes

Footnotes

References