Cellulose is one of the most common polymers found in living organisms. It is found in cyanobacteria as within the eukaryotic lineages, indicating an ancient origin (Nobles et al., 2001; Popper et al., 2011). Cellulose is a polysaccharide composed of β-D-glucopyranose (Glc) linked by (1–4) glycosidic bonds. In all land plants and some green algae, cellulose microfibrils are synthesized by a complex of 18 cellulose synthases A (CESA), according to computational analyses, instead of the 36 originally proposed, forming a rosette in the plasma membrane (McNamara et al., 2015; Oehme et al., 2015; Nixon et al., 2016; Jarvis, 2018). The resulting cellulose microfibrils are composed of probably 18 glucan chains (Vandavasi et al., 2016) with crystalline and desordered form of cellulose co-existing in each cross-section of the microfibril (Jarvis, 2018). Phylogenetic analyses have shown that P. patens and S. moellendorffii CESA protein sequences constitute their own clade while in the seed plants the phylogenetic tree is separated into 6 major clades indicating ancient gene duplication events (Carroll and Specht, 2011). The CESA6 clade is divided into 2 subclades: the first one (6A) is found only in eudicots while the other one (6B) is present in both monocots and eudicots. Arabidopsis seemed to have lost the 6B clade (Carroll and Specht, 2011).

The moss P. patens genome contains seven CESA genes named PpCESA3-8 and 10 and three CESA pseudo-genes (PpCESA1-2, and 9) (Roberts and Bushoven, 2007; Yin et al., 2009; Wise et al., 2011). Moreover, Funaria hygrometrica (Reiss et al., 1984; Rudolph and Schnepf, 1988; Rudolph et al., 1989) and P. patens (Roberts et al., 2012; Nixon et al., 2016) have rosette-type CES complexes. This reveals that cellulose synthesis in non-vascular plants is similar to Charophycean green algae and land plants than other algae which, depending on the species, have linear or other types of complexes (Tsekos, 1999; Saxena and Brown, 2005). Among the seven CESA expressed in P. patens protonemal cells, several may be involved in tip-growth (Tran and Roberts, 2016) and others are involved in the secondary cell wall biosynthesis of sclereid cells in the leaf midribs of P. patens (Norris et al., 2017). In ferns, analyses of genomes and transcriptomes have shown that C. richardii, Pteridium aquilinum and Adiantum capillus-veneris possess ortholog genes that clustered with AtCESA1-4 and AtCESA7-8 (Yin et al., 2014), but tissue specific expression has not been carried out, yet. In the genome of A. thaliana, ten CESA genes and six CELLULOSE SYNTHASE LIKE D (CSLD) are found (Richmond and Somerville, 2001; Bernal et al., 2008), while the rice genome contains seven CESA and five CSLD (Wang et al., 2010). In A. thaliana, CESA6, CSLD1 and CSLD4 were detected at the plasma membrane of the whole pollen tube (Wang et al., 2011; Chebli et al., 2012). CESA1, 3, and 9 are also expressed suggesting a possible role during pollen tube growth (Persson et al., 2007; Chebli et al., 2012). Similarly, in Nicotiana tabacum pollen tubes, CESA and CSLD were localized at the plasma membrane of the entire pollen tube with the highest level at the tip, driven by microtubules (Cai et al., 2011). To date, no information on the expression of CESA genes in pollen tubes of gymnosperms has been reported.

Even if callose is the main component of the cell wall in angiosperm pollen tubes, cellulose, often detected with calcofluor white (which does not discriminate between cellulose and callose) (Supplemental Table 1), is generally located all along the pollen tube with a weaker detection at the tip than in the shank. This pattern was observed in Pyrus communis (Aloisi et al., 2017), N. tabacum (Ferguson et al., 1998), lily (Roy et al., 1997) and A. thaliana (Lennon and Lord, 2000; Derksen et al., 2002) (Figure 1). However, using CBM3a, a specific probe for cellulose (Supplemental Table 1), two populations of A. thaliana pollen tubes were observed: 70% were strongly labeled in the entire tube and 30% did not display any labeling at the tip (Chebli et al., 2012), probably depending on the growth status and/or the developmental stages of the pollen tube. At the sub-cellular level, cellulose was located in the inner cell wall layer of the pollen tube and in vesicles (Chebli et al., 2012). This suggests that cellulose biosynthesis might be initiated in vesicles carrying CES that will be deposited at the tip. Alternatively, these vesicles could originate from endocytosis (Chebli et al., 2012) suggesting cellulose recycling.

In gymnosperm pollen tubes, it has been reported that cellulose, stained with calcofluor white, is present all along the pollen tube in Picea abies (Lazzaro et al., 2003) and Picea wilsonii (Sheng et al., 2006). However, it is noteworthy that calcofluor labels β-glucans, and does not discriminate cellulose from callose (Supplemental Table 1). In Pinus sylvestris (Figure 1), microfibrils appear to be less dense in the pollen tube tip compared to the rest of the pollen tube wall as shown from alkaline or acid-treated pollen tubes observed by transmission electron microscopy (Derksen et al., 1999).

In mosses, cellulose is present in the cell wall of all species that have been investigated, including P. patens protonemal cells (Kremer et al., 2004; Moller et al., 2007; Nothnagel and Nothnagel, 2007; Goss et al., 2012; Roberts et al., 2012). In contrast, P. patens rhizoids were very weakly labeled with CBM3a (Berry et al., 2016). Cellulose probed with CBM28 (Supplemental Table 1) was not detected in the protonemal cells or rhizoids (Berry et al., 2016). However, it was shown that those two probes recognize different types of cellulose, display different binding affinity and that other cell wall polymers such as pectin can reduce their interactions with cellulose (Blake et al., 2006). In ferns, such as C. richardii, calcofluor white staining showed a uniform labeling on the entire rhizoid cell wall, suggesting the presence of β-glucan, cellulose or callose, or both all along the cell (Eeckhout et al., 2014). Finally, in the cell wall of C. richardii antheridia, CBM3a staining confirmed also the presence of cellulose (Lopez and Renzaglia, 2018). Altogether, the data demonstrate that cellulose is broadly present in all these tip-growing cells at different levels.

In fast-growing pollen tubes, despite its low abundance compared to sporophytic cell wall tissues, studies using A. thaliana mutants have shown that cellulose/hemicellulose synthesis is of main importance. AtCSLD1 and AtCSLD4 are two membrane-localized proteins in pollen tubes. Csld1 and csld4 homozygous mutants are sterile, and mutant pollen grains showed abnormal germination with high rates of ruptured tubes in vitro, and arrest of growth in vivo (Wang et al., 2011). Mutation in AtCSLC6 resulted in a strong reduction of pollen tube growth (Boavida et al., 2009). However, it has been reported that CSLC5 was involved in XyG backbone synthesis (Liepman et al., 2010). Similar results were observed in the tip-growing rhizoids of M. polymorpha (Honkanen et al., 2016). Generation of mutant lines in M. polymorpha showed that 33 genes were required for rhizoid growth (Honkanen et al., 2016). Among them, mutations in MpCSLD1 and MpCSLD2 which have been suggested to function in synthesis of mannan, xylan and cellulose (Gu et al., 2016), induce very short and burst tips or short rhizoids, respectively (Honkanen et al., 2016), supporting the importance of the cellulose-hemicellulose network in the mechanical strengthening of the cell wall.

Cellulose synthesis inhibitors 2,6-dichlorobenzonitrile (DCB) and isoxaben strongly affect the growth of tip-growing cells, particularly pollen tubes. In pollen tubes of Lilium auratum, Petunia hybrida (Anderson et al., 2002) and Pinus bungeana (Hao et al., 2013), DCB induced distortion of the cell wall, changes in the cell wall composition and tip-bursting (Anderson et al., 2002). Isoxaben, induced tip swelling in conifer pollen tubes (Lazzaro et al., 2003). Interestingly, in P. patens, DCB and isoxaben, treatments did not have a strong effect on protonemal growth rates, and only DCB caused tip rupture (Tran et al., 2018). Those treatments seemed to modify the CESA mobility but not the density and probably have an impact in the patterning of cellulose biosynthesis in protonemal cells (Tran et al., 2018). These data reveal a clear difference between pollen tubes and protonemal cells in response to those cellulose synthesis inhibitors, probably due to the different levels of cellulose found in those cells. However, moderate enzymatic treatment with cellulase resulted in larger pollen tube diameters, promoted tip swelling and eventually bursting (Aouar et al., 2010). All these data support that despite the low abundance of cellulose in the cell wall of pollen tubes, it has an important function together with hemicelluloses in maintaining the cell wall integrity (Mollet et al., 2013; Tran et al., 2018).

Because of their crystalline nature, the formation of cellulose microfibrils is generally associated with the stiffening and the reduction of extensibility of the cell wall. In most cells with cylindrical geometry, cellulose microfibrils are oriented perpendicular to the growth axis (Baskin, 2005) in order to strengthen the cell wall against turgor-induced tensile stress in a circumferential direction and to ensure elongation in the longitudinal direction. In tip-growing cells, cellulose microfibrils are organized differently. In pollen tubes from Petunia and Pinus, the principal orientation of cellulose microfibrils is at 45° to the long axis (Sassen, 1964; Derksen et al., 1999). It is observed in Lilium and Solanum pollen tubes an orientation at 20° and 15°, respectively (Aouar et al., 2010; Geitmann, 2010). In regards to this pattern of deposition, it is admitted that cellulose does not play an important role in resisting circumferential tensile stress in the distal tubular region of the pollen tube and in tip-growing cell in general. But the fact that inhibitors of cellulose synthesis cause apical swelling in pollen tubes indicates that cellulose microfibrils affect cell wall stability in the transition region between the tip and the sub-apical dome.

XylGalA has a galacturonan backbone with substituting Xyl residues (Figure 1). XylGalA probed with LM8 (Supplemental Table 1) has been detected in the entire A. thaliana pollen tube (Dardelle et al., 2010). It seems that XylGalA is not present in P. patens protonemal cells, as no signal was observed with the same mAb (Moller et al., 2007). Moreover, out of 16 gene families involved in pectin synthesis and remodeling analyzed by McCarthy et al. (2014), orthologs of A. thaliana XylGalA xylosyltransferase genes were not found in the genome of P. patens, suggesting that XylGalA appeared later. Unfortunately, no information is currently available on XylGalA of C. richardii or gymnosperm pollen tubes. In A. thaliana, the study of xylogalacturonan deficient 1 (xgd1) which gene codes a XYLOSYLTRANSFERASE involved in XylGalA biosynthesis did not display peculiar phenotype on pollen grains or pollen tubes (Jensen et al., 2008), suggesting that this gene is probably not expressed in pollen.

AGPs are found throughout the entire plant kingdom from bryophytes (Lee et al., 2005) to angiosperms (Seifert and Roberts, 2007; Ellis et al., 2010; Ma et al., 2017). They have been involved in a large number of biological functions including tip-growth (Nguema-Ona et al., 2012). The protein backbone of AGPs is extensively O-glycosylated accounting for more than 90% (w/w) of the mass of the molecule (Figure 5). AGPs consist of a type-II arabinogalactan with a β-(1,3)-galactan backbone and both β-(1,6)-galactosyl and α-(1-5)-arabinosyl residues (Figure 5). They also possess some less abundant sugars including Xyl, Fuc, Rha, and GlcA (Nothnagel, 1997; Showalter, 2001; Tan et al., 2010; Nguema-Ona et al., 2014). Despite its low level, GlcA was shown to bind calcium and thus, AGPs could be an external source of Ca²⁺ necessary for the intracellular machinery of pollen tube growth including Golgi vesicle exocytosis (Lamport et al., 2014). AGPs may also play a role in plasticizing the cell wall (Lamport et al., 2018). In bryophytes such as Sphagnum sp., Polytrichastrum formosum and P. patens, and the monilophytes like Equisetum arvense, Dryopteris filix-mas and Pteridium aquilinum, the typical type-II arabinogalactan linkages found in AGPs were detected (t-Araf , 4-GlcAp, 3-Galp, 3,6-Galp). However, unusual sugars were also analyzed such as a terminal 3-O-methyl-L-Rha in ferns and mosses which has not been found in angiosperms so far (Fu et al., 2007; Moller et al., 2007; Bartels and Classen, 2017), or even 1,2,3-linked Gal residues never described before in AGPs (Bartels and Classen, 2017). Interestingly, in the lycophytes like Lycopodium annotinum, the terminal 3-O-methyl-L-Rha was not detected. Instead high levels of t-Arap were found (Bartels and Classen, 2017). In A. thaliana pollen tubes, one study has highlighted the presence of linkages that can be found in AGPs such as t-Ara, 3-Gal and 3,6-Gal (Dardelle et al., 2010). Similar results were obtained in N. alata pollen tube cell walls with the detection of t-Araf , t-Arap, 3-Galp, and 3,6-Galp (Lampugnani et al., 2013).

AGPs are classified into 9 families based on the protein backbone structure: classical AGPs, AG peptides, fasciclin-like AGPs (FLA), Lys-rich AGPs, early nodulin- (ENOD) like AGPs, non-specific lipid transfer protein-like AGPs (nsLTP-like AGPs), phytocyanin-like AGPs, xylogen-like AGPs and chimeric AGP-extensins (Ma and Zhao, 2010; Showalter et al., 2010; Tan et al., 2012; Johnson et al., 2017a; Ma et al., 2017). Among them, some can be anchored to the plasma membrane via GlyceroPhosphatidyl insositol (GPI) (Youl et al., 1998; Svetek et al., 1999). Classical AGPs, AG-peptides, FLAs, phytocyanins, and lipid transfer-like proteins are among the GPI-anchored proteins identified in A. thaliana pollen tubes by transcriptomic and proteomic analyses (Lalanne et al., 2004). The importance of GPI-anchored proteins was highlighted by the disruption of SETH1 and SETH2 genes that encode subunit homologs of a GPI-N-acetylglucosaminyltransferase (GPI-GnT) complex involved in GPI anchor synthesis (Lalanne et al., 2004). The mutant plants display a strong reduction of pollen germination and pollen tube growth, as well as abnormal callose deposition possibly by affecting simultaneously a large number of GPI-anchored proteins. Transcriptome data analyses suggested that GPI anchored AGPs were already quite abundant in plant lineages such as mosses and liverworts (Johnson et al., 2017b). The question remains however to what extent the GPI-anchor of a protein affects its function (Nguema-Ona et al., 2012). Using a bioinformatics approach, Ma et al. (2017) have predicted that classical AGPs, AG-peptides, and Lys-rich AGPs first emerged in P. patens, S. moellendorffii and the gymnosperm P. abies, respectively. They also revealed that the number of predicted genes varied greatly across plant lineages: 104 in the bryophyte P. patens, 49 in the pteridophyte S. moellendorffii, 129 in the gymnosperm P. abies, 48 in the basal angiosperm A. trichopoda (Figure 1) (Amborella Genome Project, 2013), in monocots between 82 in the orchid Phalaenopsis equestris and 305 in Zea mays and in eudicot plants between 77 in Carica papaya and 313 in Glycine max (Ma et al., 2017).

Several mAbs and reagents have been used to localize AGPs (Table 2; Supplemental Table 1). Among them, the β-D-Glucosyl Yariv (βGlcY), able to bind and precipitate AGPs (Yariv et al., 1967), was used to study protonemal expansion of M. polymorpha (Shibaya et al., 2005), P. patens (Lee et al., 2005) and pollen tube growth in several species including the angiosperm Magnoliales: Asimina triloba, the monocot L. longiflorum or the eudicots N. tabacum, Malus domestica and others (Figure 1) (Mollet et al., 2002; Nguema-Ona et al., 2012; Fang et al., 2016b; Losada et al., 2017). N. tabacum pollen tube growth was not affected by a treatment with βGlcY and AGPs were not detected at the tip using several mAbs such as MAC207 and JIM8 (Supplemental Table 1). However, a ring-like pattern was observed in the shank of the tube after cellulase or pectinase treatments (Li et al., 1992).

On the other hand, a treatment with βGlcY strongly reduced or arrested the tip expansion of P. patens protonemata (both caulonemal and chloronemal cells) (Lee et al., 2005), the growth of M. polymorpha protonemata (Shibaya et al., 2005) or pollen tubes that displayed AGPs at their tips (Mollet et al., 2002; Leszczuk et al., 2019) and in the shank as a periodic ring-like deposition after a pectinase treatment (Jauh and Lord, 1996) (Table 2). In lily and strawberry (Fragaria x ananassa) pollen tubes and in P. patens protonemata, the growth arrest was fast but reversible after removal of the reagent. In the case of lily pollen tubes, despite the growth arrest, vesicular secretion at the tube tip was persisting. Disorganized cell wall architecture and composition were observed due to an abnormal callose deposition (Roy et al., 1998; Leszczuk et al., 2019). This change of cell wall composition and organization in the tip induced a change in the mechanical properties of the cell wall, being stiffer in the βGlcY treated pollen tube tip and even more in the new emerged pollen tube compared to the untreated control (Leszczuk et al., 2019). The authors suggested that this hardening may be related to the appearance of weakly methylesterified HG in the tip of the newly-grown pollen tube. Similarly, an abnormal cell wall deposition was also observed at the tube tip of M. polymorpha protonemata treated with βGlcY (Shibaya et al., 2005). The main difference between both tip-growing cells is that the protonema arrested-tip slowly resumed growth 3h after the removal of βGlcY (Lee et al., 2005) whereas in lily and strawberry pollen tubes, the arrested-tips did not resume growth but a new tip emerged back from it (Mollet et al., 2002,Leszczuk et al., 2019).

Inactivation of P. patens AGP1 encoding a classical AGP resulted in reduced cell expansion suggesting its role in tip-growth (Lee et al., 2005). In A. thaliana, four genes: two classical GPI anchored AGPs (AtAGP6 and AtAGP11) and two AG-peptides (AtAGP23 and AtAGP40) are expressed only in pollen grains and pollen tubes. The double mutant agp6/11 has reduced pollen tube growth and seed set (Levitin et al., 2008; Coimbra et al., 2010) indicating that AGPs are important during pollen tube growth. Using yeast two-hybrids, interactors of AGP6 and AGP11 were mostly involved in recycling of cell membrane components via endocytosis (Costa et al., 2013). It revealed that endosomal trafficking pathways are fundamental regulators of tip-growing cells. In a search of pollen specific AtAGP6/11 orthologs from 1,000 plant transcriptomes, Johnson et al. (2017b) revealed that they could be confidently detected in the basal angiosperms (A. trichopoda), basal eudicots (Ranunculales), Rosids (Brassicales, Fabales, Myrtales), Asterids (Apiales, Solanales, Asterales, Ericales), the non-commelinid (Liliales, Arecales, and Asparagales) and commelinid (Poales including rice) monocot lineages, suggesting the existence of an ancestral gene (Figure 1). In rhizoids of P. patens, labeling with LM2 and JIM13 was either weak on the entire rhizoid or absent (Figure 2) (Berry et al., 2016). In contrast, the labeling of AGPs (with LM2 and LM6) (Supplemental Table 1) was very strong on the whole rhizoid of the monilophyte C. richardii (Table 2) (Eeckhout et al., 2014), suggesting either a different carbohydrate composition of AGPs and/or a different accessibility of the mAb to the epitopes thus supporting a different cell wall organization. Cell-surface immuno-labeling or βGlcY coloration revealed a strong labeling at the tip in the protonemata of M. polymorpha and P. patens and pollen tubes from lily (Mollet et al., 2002), Asinima (Losada et al., 2017) and A. thaliana (Dardelle et al., 2010). In gymnosperms like P. wilsonii or P. meyeri, labeling with LM2 was located in the entire pollen tube wall or as a periodic ring-like pattern (Chen et al., 2007, 2008). With the same mAb, a ring-like pattern was also observed in the cell wall of apple pollen tubes (Table 2) (Fang et al., 2016a).

Extensins (EXTs) are repeated pentapeptide Ser(Hyp)₄ that are moderately O-glycosylated with short Ara and Gal oligosaccharides on Hyp and Ser residues respectively (Figure 6) (Kieliszewski and Lamport, 1994; Velasquez et al., 2015). In the cell wall, EXTs play an important role in development, and cross linking of EXTs is generally associated with the arrest of cell expansion (Johnson et al., 2017b). The cross-linked EXTs were predicted to occur in most land plants except in commelinid monocot grasses and arose in non-vascular plants, particularly in hornworts (Dendrocerotales, Notothyladales…). They were less represented in liverworts (Marchantiales…) and mosses (Funariales including P. patens, Polytrichales, Sphagnales…). Indeed, classical EXTs are predicted to be absent from P. patens as well as from the gymnosperm Pinus taeda (Liu et al., 2016; Johnson et al., 2017b). Accordingly, using microarrays, almost no detectable signal was observed with LM1, JIM19 and JIM20 (Supplemental Table 1) in fractionated cell wall extracts from P. patens protonemata (Moller et al., 2007). Similarly, no detectable signal was observed with LM1 in A. thaliana pollen tubes (Dardelle et al., 2010) suggesting either a very low abundance, epitope masking or different carbohydrate epitopes. In the spikemoss, S. moellendorffii, very weak signal was also observed with LM1 but clear signal was detected with JIM20 on an alkali-extracted cell wall extract (Harholt et al., 2012).

Non-classical EXTs include short EXTs and chimeric proteins such as LRX (Leucine-Rich repeat EXT), proline-rich EXT-like receptor kinases (PERKs), formin-homolog EXTs (FH EXTs) (Liu et al., 2016), AGP-EXTs, which are highly represented in gymnosperms (conifers, gnetophytes, Ginkgo and Cycadales) (Johnson et al., 2017b) and others. LRXs, PERKs and FH EXTs derived earlier than classical EXTs and were predicted to be present also in P. patens and S. moellendorffii (Liu et al., 2016).

Genes involved in EXT glycosylation such as β-arabinosyltransferases of the GT77 family were found in S. moellendorffii and P. patens (Harholt et al., 2012). The importance of the O-glycosylation status of EXT on the conformation of the glycoprotein has been described (Stafstrom and Staehelin, 1986) and was further investigated by functional genomics on the tip-growing protonemata of P. patens and A. thaliana pollen tubes (MacAlister et al., 2016). Mutations on genes coding the enzymes responsible for initiating oligoarabinose chains on the protein backbone, the hydroxyproline O-arabinosyltransferases (HPATs) (Ogawa-Ohnishi et al., 2013) revealed opposite effects on those two models (MacAlister et al., 2016). Athpat1/3 double mutant pollen tubes showed both in vivo and in vitro a strong reduction of pollen tube expansion. In contrast, among the two genes in P. patens (HPATa and HPATb), mutation in HPATa displayed the most striking phenotype by inducing an increase in the number and length of filaments which was correlated with a faster growth of the protonema in the hpata/b double mutant than in the wild type. Interestingly, the double mutant did not display any defects on the length of the other tip-polarized cells: the rhizoids. This suggests that other HPAT genes are expressed in this structure. The authors suggested that the different responses between the two models may be related to the growth rates (which are faster in pollen tubes), the cell wall compositions (which are quite different), or the different target proteins of these enzymes (MacAlister et al., 2016).

Another study showing the important function of EXTs in pollen tube growth was conducted on the classical EXT18 (Choudhary et al., 2015). Despite pleiotropic phenotypes, mutation in EXT18 reduced significantly the pollen viability, pollen tube growth and increased the level of burst tubes leading to a strong reduction of seed set (Choudhary et al., 2015). Moreover, the level of expression of 12 other classical EXTs was modified in the mutant. Two were up-regulated, including EXT19 (closely related to EXT18), and ten were down-regulated, suggesting a coordinated network of expression between the EXT genes (Choudhary et al., 2015).

Evidence of the presence of LRX in pollen tubes was highlighted in maize more than 20 years ago by immunolocalization of PEX1 (Pollen EXT1) in the inner callosic cell wall layer (Rubinstein et al., 1995). PEXs were then also found in other monocots (Sorghum, rice, Brachypodium…) and eudicots (tomato, Arabidopsis, Brassica rapa, potato…) (Stratford et al., 2001; Liu et al., 2016). More recently, three studies have shown the important role of LRX in pollen tube growth and cell wall mechanics (Fabrice et al., 2018; Sede et al., 2018; Wang et al., 2018). In A thaliana, among the 11 LRX genes, four of them (LRX8-11, previously described as PEX1-4) are devoted to the reproductive development (Baumberger et al., 2003). The LRR domain is thought to bind an interaction partners which have been recently characterized as the pollen secreted signaling peptides RALF4 and RALF19 (RAPID ALKALINIZATION FACTOR). The two peptides maintain the pollen tube cell wall integrity and are translocated into the interior of the pollen tube via the CrRLK1L (Catharanthus roseus Receptor-Like Kinases1-Like subfamily (Mecchia et al., 2017) while the EXT domain anchors the protein in the cell wall (Ringli, 2010), reviewed by Marzol et al. (2018). Interestingly, abnormally short and burst rhizoid structures were also observed in M. polymorpha mutants defective in MpTHESEUS that belongs to the CrRLK1L family (Honkanen et al., 2016). These data reveal that cell wall integrity sensing is conserved in tip-growing of modern seedless land plants (Honkanen et al., 2016).

The studies from single, to quadruple mutants in LRX8-11 revealed a weak phenotype for the single mutants but severe in the double lrx8/9, and the triple lrx8/9/10, lrx8/9/11, or lrx9/10/11 mutants. It was observed a strong decrease of pollen germination, an increased level of bursting and abnormal tubes. Such mutants show wavy growth, widened tips, leakage in the sub-apical region of cytoplasmic contents enriched in cell wall polymers. It was observed the emergence of a bulge at the shank or at the pollen tube tip which causes tip bursting leading in vivo to a strong reduction of seed set (Fabrice et al., 2018; Sede et al., 2018; Wang et al., 2018). In those mutants, pollen tubes displayed abnormal cell walls with altered levels of pectins including HG and RG-I, fucosylated XyG and accumulation of callose (Fabrice et al., 2018; Sede et al., 2018; Wang et al., 2018). These defects in the cell wall composition and organization led to an increase of internal turgor pressure and cell wall stiffness of the pollen tube (Fabrice et al., 2018). Interestingly, the pollen tube growth defects observed in lrx8/9/11 triple mutant, were partially supressed by decreasing the calcium concentration, or adding LaCl₃, a calcium channel inhibitor. It suggests that LRX proteins influence Ca²⁺-related processes (Fabrice et al., 2018). Indeed, monitoring calcium level in the triple mutant revealed a very strong spike of calcium just before bursting (Fabrice et al., 2018).

All the data accumulated suggest that EXTs provide either structural support at the tip of the growing pollen tube and/or can serve as an intermediate in the communication between the cell wall and the cytoplasm (Bascom et al., 2018).

Little information is known about callose deposition in bryophytes and pteridophytes. However, cytochemical staining (Supplemental Table 1) have shown that callose was present in the aperture of developing P. patens spores and in protonemata of P. patens (Moller et al., 2007; Tang, 2007; Schuette et al., 2009; Berry et al., 2016) and F. hygrometrica (Bopp et al., 1991). In P. patens, the analysis of 3-Glc linkage confirmed the presence of callose (Moller et al., 2007). Callose deposition has also been observed to some extent in rhizoids of the liverwort M. polymorpha (Cao et al., 2014) and the C-fern C. richardii (Eeckhout et al., 2014). These different studies show that callose deposits constitute a permanent compound already present in spores and very probably associated to the development of both protonemata and rhizoids in bryophytes and pteridophytes.

Callose is found in the cell wall of pollen grains in gymnosperms such as Ginkgo/cycads, but not in their pollen tube cell walls (Yatomi et al., 2002; Abercrombie et al., 2011). Using aniline blue staining (Supplemental Table 1), callose was not detected in the pollen tube wall of several Pinus species but detected in Podocarpus nagi and Chamaecyparis obtusa pollen tube (Yatomi et al., 2002). In contrast, using a callose-specific mAb (Supplemental Table 1), the pollen tube cell walls of six tested Pinus species were labeled to various degrees: a slight labeling was observed in most of the examined conifers (Yatomi et al., 2002; Fernando et al., 2010), none in Pinus taeda (Abercrombie et al., 2011), but in the entire wall of two Podocarpus species, Cryptomeria japonica and C. obtusa. Callose is also uniformly distributed along the tube shank and a very faint fluorescence could be detected in the tip region of pollen tubes from P. meyeri (Chen et al., 2007). Similarly, callose was found transiently in the tip of P. sylvestris pollen tubes (Derksen et al., 1999). This tip-localized callose deposition is generally not observed in angiosperm pollen tubes grown in vitro in normal culture conditions (Mollet et al., 2013). The role of these transient callose deposits at the gymnosperm pollen tube tip is unknown. However, Pacini et al. (1999) suggested that callose observed at the pollen tip may not have a structural function but rather could serve as a reserve polysaccharide, although this hypothesis needs to be confirmed.

Interestingly, in the shank of gymnosperm pollen tubes studied so far, the cell wall is composed of one layer (Derksen et al., 1999; Yatomi et al., 2002; Fernando et al., 2010) whereas in angiosperms, two layers are present, an outer non-callosic cell wall layer and an inner callose-enriched layer (Meikle et al., 1991; Ferguson et al., 1998; Dardelle et al., 2010; Derksen et al., 2011; Chebli et al., 2012; Mollet et al., 2013). Consequently, in contrast with other cells callose is one of the main cell wall components of the pollen tube in angiosperms representing about 35% in 16-h old A. thaliana or N. alata pollen tubes, along with arabinans (Dardelle et al., 2010; Lampugnani et al., 2013) .

An interesting difference between gymnosperms and angiosperms pollen tubes is a lack of callose plug formation in the gymnosperm studied, so far (Fernando et al., 2010; Abercrombie et al., 2011). In angiosperms, callose plugs are synthesized periodically to maintain the vegetative cell in the tip-region. These plugs are present in the early-divergent angiosperm A. trichopoda (Williams, 2009). Their functions is to seal off the pollen tube from the pollen grain and to separate the young from the older parts of the tube maintaining a manageable cytosolic volume. These callosic barriers reduce the risk of damage and allow the tube to grow longer distances (Taylor and Hepler, 1997). In the eudicot Camellia japonica, the average degree of polymerization (DP) of callose found in the plugs was at least 90 whereas in the inner pollen tube cell wall the DP was much lower (DP ≈ 21) suggesting a possible more cohesive structure of callose deposits in the plugs than in the cell wall (Nakamura et al., 1980, 1984). Within angiosperms, the morphology of callose plugs and the patterns of their depositions were previously shown to vary among species (Mogami et al., 2006; Qin et al., 2012). Variability in the plug position has also been observed in pollen tubes among ecotypes and even within an ecotype (Qin et al., 2012). In this study covering 14 A. thaliana ecotypes, it was demonstrated that the position of the first callose plug and subsequently the number of callose plugs within a pollen tube will have an impact on its length. Thus, pollen tubes with the first callose plug near the pollen grain were on average longer than those with the first callose plug farther from the grain. This pattern of callose plug deposition was heritable (Qin et al., 2012). This indicates that the periodic deposition of callose plugs can impact pollen tube length.

Callose biosynthesis is catalyzed by a multi-subunit enzyme complex associated to the plasma membrane in numerous types of plant cells (Verma and Hong, 2001). The key components of this complex are grouped under the name of CALLOSE SYNTHASES (CalS) / GLUCAN SYNTHASE-LIKE (GSL) (Li et al., 2003; Farrokhi et al., 2006; Abercrombie et al., 2011), that encode GT48 β-(1-3)-glucan synthase enzymes (Tucker et al., 2018). The involvement of CalS in spore development is a plesimorphic feature of terrestrial plants. Callose is synthesized from UDP-Glc, that binds directly to the catalytic subunit of CalS/GSL (Drbkov and Honys, 2017). Twelve CalS have been identified in A. thaliana distributed in four clades (Richmond and Somerville, 2000; Verma and Hong, 2001). These first identified genes were crucial for further investigation of CalS-like orthologs in other land plant models. Among these 12 genes, CalS5, CalS11, and CalS12 are devoted to pollen development (Chen and Kim, 2009). CalS5 was suggested to be involved in the synthesis of the callose wall during pollen grain germination, pollen tube growth and callose plug formation (Nishikawa et al., 2005). Knockout mutations of CalS5 did not perturb vegetative growth significantly but resulted in a male sterility (Dong et al., 2005). The mutant cals5 displayed a complete loss of callose plugs in the growing pollen tubes (Dong et al., 2005). However, another study on a cals5 mutant obtained by point mutations showed also defect in callose deposition but the pollen tube growth was normal (Nishikawa et al., 2005). The authors suggested that callose was important for pollen wall patterning but not for the tube growth (Nishikawa et al., 2005). This report sows doubts about the exact role of callose plugs in the pollen tube growth. It seems highly surprising, that mutant pollen tubes without callose plugs may have the same performance and grow similarly as wild-type pollen tubes in vitro and in vivo. This finding questions the exact role of callose in pollen tubes while this polymer appears to be an evident evolutionary hinge at the sight of the phylogenetic studies. Overexpression of CalS5 under the CaMV 35S promoter in A. thaliana also causes abnormal callose deposition and resulted in the precocious pollen germination prior to anthesis, suggesting that the expression level and localization pattern of CalS5 are important for pollen development (Xie et al., 2010).

As in A. thaliana, the genome of P. patens contains also 12 putative PpCalS that cluster in three clades with AtCalS genes (Schuette et al., 2009). PpCalS5, being the ortholog of AtCalS5, was suspected to be involved in the spore development (Schuette et al., 2009). In 2011, a study on CalS expression patterns brought evidence that AtCalS5 orthologs were expressed in mature pollen grains and pollen tubes of many different early-divergent angiosperms including the basal angiosperm A. trichopoda. In gymnosperms, Ginkgo CalS5 and Gnetum CalS5 were only expressed in pollen grains whereas CalS5 orthologs were only expressed in fast-growing pollen tubes (i.e., angiosperms) suggesting that CalS5 plays different but crucial roles during pollen formation and/or pollen tube growth (Abercrombie et al., 2011). Finally, an ortholog of AtCalS5 was also found in S. moellendorffii while the gymnosperm Pinus taeda pollen expressed another ortholog PtCalS13 (Abercrombie et al., 2011).

NaGsl1, a CalS gene, is specifically and abundantly expressed in pollen grains of N. alata (Doblin et al., 2001). Levels of NaGsl1 transcripts were high in mature pollen, suggesting that pollen grains are released with the appropriate transcripts for wall formation before germination. Transcript levels remained high during pollen tube growth (Doblin et al., 2001). Most CalS are expressed under the dependence of the bZIP transcriptor factor family as seen in A. thaliana (Drbkov and Honys, 2017). In tobacco pollen tubes, NaCalS was localized in the distal region and at the tip of the pollen tube and accumulated, driven by microtubules, at the plasma membrane where callose plugs are synthesized (Cai et al., 2011).

Genes involved in callose hydrolysis and recycling, putatively involved in the pollen tube wall remodeling, includes the β-(1,3)-glucan hydrolase genes (GH17 family). These genes are notably conserved across land plant species (Tucker et al., 2018). Exogenous application of lyticase, an enzyme able to degrade callose, on pollen tubes can increase the pollen tube diameter more drastically in the distal region. As a consequence, the stiffness decreased while viscoelasticity increased in the distal region of the tube (Parre and Geitmann, 2005a). In contrast, in the tip region where callose is not present, no such effect was observed (Parre and Geitmann, 2005a). In the angiosperm L. longiflorum, exo-glucanases were described to play an important role in the regulation of pollen tube elongation (Takeda et al., 2004).

Callose represents about 80% of the wall mass of the tube shank (Schlupmann et al., 1994) and the bulk of wall volume in most angiosperms (Williams et al., 2016). There is evidence that building amorphous callose-based wall found in angiosperms is faster and more energy efficient than elaborating a cellulose microfibril-based wall produced by gymnosperms (Kudlicka and Brown Jr, 1997; Williams, 2012). Early-divergent angiosperms like A. trichopoda, Nuphar polysepala or Austrobaileya scandens have pollen tubes with a secondary callose wall and produce callose plugs compared to gymnosperms. The apparition of these two features in early-divergent angiosperms combined with novel secretory carpel tissues (Williams, 2009) are probably the reason of the difference of growth rates. The fast rate of pollen tube growth is sustained in angiosperms by efficient exocytosis of new wall material at the tip or synthesis at the plasma membrane in a small region at, or just behind, the tube apex (Williams et al., 2016) and endocytosis of cell wall components are probably of importance. In the shank of the tube, the inner callose layer is essential for insuring an efficient polarized tip-growth and resistance to turgor pressure. All these studies have highlighted that short-lived and fast-growing pollen tubes associated with the apparition of an inner callose wall and callose plugs were the fundamental innovations that have trigger the ecological success of angiosperms.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.