Arabinogalactan Proteins: Focus on the Role in Cellulose Synthesis and Deposition during Plant Cell Wall Biogenesis

Arabinogalactan proteins (AGPs) belong to a family of glycoproteins that are widely present in plants. AGPs are mostly composed of a protein backbone decorated with complex carbohydrate side chains and are usually anchored to the plasma membrane or secreted extracellularly. A trickle of compelling biochemical and genetic evidence has demonstrated that AGPs make exciting candidates for a multitude of vital activities related to plant growth and development. However, because of the diversity of AGPs, functional redundancy of AGP family members, and blunt-force research tools, the precise functions of AGPs and their mechanisms of action remain elusive. In this review, we put together the current knowledge about the characteristics, classification, and identification of AGPs and make a summary of the biological functions of AGPs in multiple phases of plant reproduction and developmental processes. In addition, we especially discuss deeply the potential mechanisms for AGP action in different biological processes via their impacts on cellulose synthesis and deposition based on previous studies. Particularly, five hypothetical models that may explain the AGP involvement in cellulose synthesis and deposition during plant cell wall biogenesis are proposed. AGPs open a new avenue for understanding cellulose synthesis and deposition in plants.


AtAGP40
A. thaliana AG peptide √ pollen T-DNA insertion mutant no alteration in pollen grain development but a reduction in pollen grain fitness prevents premature pollen grain germination [74] agp6 agp11 agp40 triple mutant a significant reduction in seed production and a higher number of early germinating pollen tubes inside the anthers

Cellulose Synthesis and Deposition during Plant Cell Wall Biogenesis
Plant cell walls are largely composed of cellulose, hemicelluloses, and pectins, along with a small amount of proteins and other compounds [147][148][149]. As the most abundant and main load-bearing biopolymer of the cell wall, cellulose is synthesized by cellulose synthase (CesA) proteins, integral plasma membrane proteins arranged into a unique hexagonal rosette complex called the cellulose synthase complex (CSC) [149,150].

AGPs Implicated in Cellulose Synthesis and Deposition
It has been proposed that some AGPs contribute to different biological processes, such as fiber development, microspore formation, and root growth via their impacts on cellulose synthesis and deposition. Cotton fibers are highly specialized and extremely elongated single-cell trichomes from seed epidermis, which are mainly composed of cellulose (>90%) [169,170]. Abundant AGP carbohydrate epitopes have been detected during the formation of cotton fibers, and several fiber-preferential genes encoding FLAs were isolated from cotton (Gossypium hirsutum L.) [123][124][125]171], implying that AGPs are probably implicated in the synthesis of cellulose. Direct evidence that cross-linking of AGPs with β-Yariv inhibits cellulose deposition on cultured tobacco protoplasts also gives a hint that AGPs are related to cellulose deposition [172]. In an increasing volume of evidence, this assumption has been further supported by phenotyping of loss-of-function and gain-of-function mutants. RNA interference (RNAi) of GhAGP4 inhibits fiber initiation and elongation in cotton and affects cellulose deposition of fiber cells. Suppression of GhAGP4 downregulates the expression level of the cellulose biosynthesis-related gene celA1, providing the direct proof that FLAs may affect the cell wall synthesis through cellulose deposition [126]. Overexpression of GhFLA1 in cotton promotes fiber elongation, whereas suppression of GhFLA1 slows down fiber initiation and elongation. In addition, expression levels of the genes involved in cellulose biosynthesis are remarkably enhanced in the GhFLA1 overexpression transgenic fibers, leading to a higher rate of cellulose. In contrast, the transcripts of these genes are dramatically reduced in GhFLA1 RNAi transgenic fibers with a lower rate of cellulose [124]. The intine of nearly half of the pollen grains in AtFLA3 RNAi transgenic plants appears to have some abnormalities, with an abnormal cellulose distribution, indicating that AtFLA3 may affect the pollen wall development by influencing cellulose deposition [84]. BcMF18 in B. campestris, encoding a classical AGP, is specifically expressed in pollen grains. Antisense transgenic pollen also shows intine layer development defects similar to FLA3 RNAi transgenic plants [72]. The case in Arabidopsis with a T-DNA insertion mutation of FLA1, showing a change of cellulose deposition in fla1, is also in support of this view [108]. AtFLA11/IRX13 and AtFLA12 participate in the formation of secondary cell walls, and double mutant shows reduced cellulose content, increased cellulose microfibril angle (refers to the microfibril deviation in the cell wall layer from the long axis of the cell), and impaired structure and composition of cell walls [98,99,158,173]. AtSOS5/AtFLA4 is found to cooperate in the cell wall sensing system and facilitate cellulose synthesis [38,[109][110][111][112][113][114][115][116]]. An atfla16 mutant shows that loss of FLA16 leads to reduced levels of cellulose and reduced stem length [100]. Unfortunately, because AGPs form a large family and a single-knockout mutant rarely results in a detectable phenotype, the precise functions of AGPs and their mechanisms of action in cellulose biosynthesis remain unclear.
In this current work, we propose some assumptions about the potential mechanisms of AGPs to participate in complex biological processes via their impacts on cellulose synthesis and deposition based on previous studies. 4.2.1. AGPs Are Involved in Cellulose Synthesis via the 1-Aminocyclopropane-1-Carboxylic Acid (ACC)-Mediated Pathway ACC is the direct precursor of ethylene, and the majority of the regulatory mechanisms of ethylene biosynthesis act at the level of ACC production by ACC synthases (ACSs) [174]. In addition to its role as the central molecule of ethylene biosynthesis, ACC is also capable of functioning in some biological processes via an ethylene-independent way. Tsang et al. found that the effect of ACC on primary root elongation in acute response to cell wall stress was partially independent of its conversion to ethylene or ethylene signaling in Arabidopsis [175]. The inhibition of cell elongation caused by disturbed cellulose biosynthesis can be fully restored in the short term by blocking ACC signaling despite the presence of visible cell wall damage [175].
It has been suggested that ACC might also be involved in AGP-related cell wall formation via an ethylene-independent pathway. A loss-of-function mutant of AtSOS5/AtFLA4, which lacks a GPI-anchored extracellular FLA, presents an impaired root growth and radial root tip swelling phenotype under high salt conditions [38,109,114,116]. What is particularly interesting is that double mutants of two AGP-specific galactosyltransferase genes (GALT2 and GALT5) and two leucine-rich repeat receptor-like kinase (RLK) genes (FEI1 and FEI2) phenocopy this mutant of AtSOS5/AtFLA4, respectively [110,116]. It has been demonstrated that these five proteins act linearly in the same signaling pathway of cellulose synthesis, in which AtSOS5/AtFLA4, glycosylated by GALT2 and GALT5 in the Golgi, helps to sense turgor pressure and transmits signals to plasma membrane-localized FEI1 and FEI2 [116]. An in-depth study on FEI1 and FEI2 brings ACC into play, where inhibition of ACSs suppresses the expansion defect in fei1 fei2 mutant by the disruption of an ethylene-independent pathway. As FEIs do not alter ACS activity and FEIs interact directly with ACS5 in a nonphosphorylation-dependent manner, it has been proposed that FEIs may form a scaffold to localize ACS or may complex ACS with other proteins and that ACC itself may act as a signaling molecule in cellulose synthesis during cell expansion rather than ethylene [110]. Thus, in this model, GPI-anchored AGPs, such as AtSOS5/AtFLA4, may act as a signal sensor to relay information to FEI proteins; then FEI proteins interact directly with ACSs and, as a consequence, collaborate on cellulose synthesis, possibly via an ACC-mediated signaling pathway ( Figure 1).

AGPs as Structural Components Affect Cellulose Deposition through Cross-Linking to Other Cell Wall Components
Cellulose associates with hemicelluloses to form a framework embedded in a matrix of pectins and proteins, allow the cellulose microfibrils to move apart during cell wall loosening, and trap them in place when cell wall growth stops [147,176,177]. Pectins, defined as a heterogeneous group of polysaccharides, are major components of the primary cell wall [176,178]. The complex and dynamic pectin network consists of homogalacturonans (HGs), rhamnogalacturonans type I (RG-I), and RG-II, with a small amount of xylogalacturonans, arabinans, and AG I, which are covalently linked to each other [147]. Hemicelluloses are cross-linking polymers of diverse structures, including xyloglucans, xylans, arabinoxylans, mannans, glucomannans, and β-glucans [179].
Cell wall components, including polysaccharides cellulose, hemicelluloses, and pectins, as well as structural proteins (such as AGPs, the protagonists of this review), interact covalently and noncovalently to form the functional cell wall [147][148][149]180]. Hijazi et al. proposed an overview of the interactions assumed or demonstrated between HRGPs and cell wall polysaccharides, highlighting the linkages of AGPs with pectins and hemicelluloses and their contribution to cell wall architecture [180]. The classical AGP, AtAGP57C, has been revealed to covalently attach to hemicellulosic and pectic polysaccharides, with RG-I and HG linked to Rha residues in AG polysaccharides and with arabinoxylan attached to either a Rha residue in the RG-I domain or directly to an arabinosyl residue in the AG glycan domain, to form ARABINOXYLAN PECTIN ARABINOGALACTAN PROTEIN1 (APAP1) in Arabidopsis cell suspension cultures [96]. AtAGP31 is a nonclassical AGP member with an N-terminus histidine (His)-rich stretch, a repetitive Pro-rich domain, and a C-terminus Cys-rich PAC (PRP and AGP containing Cys) domain [101]. AtAGP31 has been demonstrated to interact in vitro with galactans, which are lateral chains of RG-I through its PAC domain, bind to methylated polygalacturonic acids through its His-rich stretch, and show in vitro self-assembly, providing evidence for the model of noncovalent networks between AGPs and other cell wall components [103]. AGPs may sense extracellular signals by carbohydrate moieties and transmit signals to some receptor kinases, thereby regulating cell wall formation by promoting cellulose synthesis through an ethylene-independent ACC pathway. Cellulose microfibrils are synthesized by cellulose synthase complexes (CSCs) that are present at the plasma membrane. GALT2 localized to the endoplasmic reticulum (ER) and the Golgi and GALT5 localized to Golgi vesicles function in AGP O-glycosylation [40]. AtSOS5/AtFLA4, FEI1, and FEI2 are localized to the plasma membrane [109,110]. The GALT2 GALT5/AtSOS5/FEI1 FEI2 pathway is represented according to Basu et al. [116].

AGPs as Structural Components Affect Cellulose Deposition through Cross-Linking to Other Cell Wall Components
Cellulose associates with hemicelluloses to form a framework embedded in a matrix of pectins and proteins, allow the cellulose microfibrils to move apart during cell wall loosening, and trap them in place when cell wall growth stops [147,176,177]. Pectins, defined as a heterogeneous group of polysaccharides, are major components of the primary cell wall [176,178]. The complex and dynamic pectin network consists of homogalacturonans (HGs), rhamnogalacturonans type I (RG-I), and RG-II, with a small amount of xylogalacturonans, arabinans, and AG I, which are covalently linked to each other [147]. Hemicelluloses are cross-linking polymers of diverse structures, including xyloglucans, xylans, arabinoxylans, mannans, glucomannans, and β-glucans [179].
Cell wall components, including polysaccharides cellulose, hemicelluloses, and pectins, as well as structural proteins (such as AGPs, the protagonists of this review), interact covalently and noncovalently to form the functional cell wall [147][148][149]180]. Hijazi et al.
proposed an overview of the interactions assumed or demonstrated between HRGPs and cell wall polysaccharides, highlighting the linkages of AGPs with pectins and hemicelluloses and their contribution to cell wall architecture [180]. The classical AGP, AtAGP57C, has been revealed to covalently attach to hemicellulosic and pectic polysaccharides, with A hypothetical model of AGP involvement in cellulose synthesis via the 1-aminocyclopropane-1-carboxylic acid (ACC)-mediated pathway. AGPs may sense extracellular signals by carbohydrate moieties and transmit signals to some receptor kinases, thereby regulating cell wall formation by promoting cellulose synthesis through an ethylene-independent ACC pathway. Cellulose microfibrils are synthesized by cellulose synthase complexes (CSCs) that are present at the plasma membrane. GALT2 localized to the endoplasmic reticulum (ER) and the Golgi and GALT5 localized to Golgi vesicles function in AGP O-glycosylation [40]. AtSOS5/AtFLA4, FEI1, and FEI2 are localized to the plasma membrane [109,110]. The GALT2 GALT5/AtSOS5/FEI1 FEI2 pathway is represented according to Basu et al. [116].
Arabidopsis seed coat mucilage is an excellent model to study cellulose synthesis and its interactions with other cell wall polymers [165]. AtSOS5/AtFLA4 and FEI2 are found to not only participate in root growth, but also act in a similar pathway to regulate seed coat mucilage synthesis and deposition of cellulose rays during the hydration process of Arabidopsis seeds [110,111]. Previously, AtSOS5/AtFLA4 was suggested to affect cellulose synthesis on the seed coat surface, which, in turn, influences the anchoring of pectin components in seed coat mucilage [111]. However, further studies on atsos5/atfla4 revealed that the formation of cellulosic rays in the adherent mucilage layer was disrupted, with a significantly reduced pectin content, while the cellulose content in mucilage was hardly affected [112,113,165]. The pectin matrix is implicated in the deposition of cellulose microfibrils [181,182]. A hypothesis was proposed that AtSOS5/AtFLA4 could act as a structural component independently of cellulose biosynthesis and signaling, instead organizing cellulose microfibrils through interconnections with pectins or hemicelluloses, and that FEI2 would be required to localize AtSOS5/AtFLA4 in the plasma membrane [112,113,165].
Taken together, we propose a model in which AGPs act as structural components affecting cellulose deposition through interconnections with other cell wall components, such as hemicelluloses and pectins (Figure 2). components in seed coat mucilage [111]. However, further studies on atsos5/atfla4 revealed that the formation of cellulosic rays in the adherent mucilage layer was disrupted, with a significantly reduced pectin content, while the cellulose content in mucilage was hardly affected [112,113,165]. The pectin matrix is implicated in the deposition of cellulose microfibrils [181,182]. A hypothesis was proposed that AtSOS5/AtFLA4 could act as a structural component independently of cellulose biosynthesis and signaling, instead organizing cellulose microfibrils through interconnections with pectins or hemicelluloses, and that FEI2 would be required to localize AtSOS5/AtFLA4 in the plasma membrane [112,113,165].
Taken together, we propose a model in which AGPs act as structural components affecting cellulose deposition through interconnections with other cell wall components, such as hemicelluloses and pectins (Figure 2).  [96]. Noncovalent networks between AtAGP31 and cell wall polysaccharides refer to Hijazi et al. [103]. AtSOS5/AtFLA4 and pectin interconnections in a FEI2-dependent manner are represented according to [112,113,165].

AGPs Participate in the Deposition of Cellulose Microfibrils through the Microtubule as an Intermediary
The length, deposition angle, and crystallinity of cellulose microfibrils show a decisive effect on the physical properties of the cell wall [183]. AGPs have been shown to affect cellulose deposition in plant cell walls. In poplar, PtFLAs are found to be expressed in the xylem, of which 10 genes are specifically expressed in tension wood (TW). Some of these for the APAP1 complex [96]. Noncovalent networks between AtAGP31 and cell wall polysaccharides refer to Hijazi et al. [103]. AtSOS5/AtFLA4 and pectin interconnections in a FEI2-dependent manner are represented according to [112,113,165].

AGPs Participate in the Deposition of Cellulose Microfibrils through the Microtubule as an Intermediary
The length, deposition angle, and crystallinity of cellulose microfibrils show a decisive effect on the physical properties of the cell wall [183]. AGPs have been shown to affect cellulose deposition in plant cell walls. In poplar, PtFLAs are found to be expressed in the xylem, of which 10 genes are specifically expressed in tension wood (TW). Some of these genes are upregulated in TW (PtFLA1-10), which might be related to mechanical properties of TW [184]. Two FLA-encoding genes in Eucalyptus grandis W. Hill ex Maiden, EgrFLA1 and EgrFLA2, exhibit higher expression levels in the xylem of TW in the upper sides of branches that possesses a higher cellulose content and a low microfibril angle but, instead, a lower expression level in xylem below these branches, deeply implying an accordance between FLA expression level and cellulose content as well as microfibril angle [185]. Arabidopsis AtFLA11 and AtFLA12 are highly expressed in stems, mainly distributed in vascular bundles, surrounding parenchyma and vessels. In Atfla11/fal12 double mutants, the decreased cellulose content leads to a reduction of tensile strength, while the increased cellulose microfibril angle gives rise to a decrease in tensile stiffness, indicating that AtFLA11 and AtFLA12 could interfere with the deposition of cellulose microfibrils during the formation of the secondary cell wall [99].
Cortical microtubules can guide CSCs to move along the microtubule array in a cellulose synthase interactive 1 (CSI1)-dependent manner and, as a consequence, to affect the cellulose microfibril angle [154,155,158,173]. The close linkage between AGPs and cytoskeletal structures, including microfilaments and microtubules, has shed light on the potential role of AGPs in cellulose deposition through a cytoskeletal network. REB1/RHD1 encodes a UDP-D-Glc 4-epimerase, which is involved in the galactosylation of AGPs and xyloglucans [186]. The trichoblasts of mutant reb1-1 are highly swollen with cortical microtubules that are disordered or even completely absent and lack certain AGP epitopes, suggesting a connection between the organization of cortical microtubules and the deposition of AGPs [186,187]. Sardar et al. demonstrated that β-Yariv treatment in tobacco tissue culture cells triggers depolymerization/disorganization of microtubules and F-actin, and cytoskeletal disruptors alter LeAGP1 localization along the Hechtian strands (a stretched plasma membrane extending from the plasmolyzed protoplast to the cell wall in plants), implying that GPI-anchored AGPs play a role in the plasma membrane-cytoskeleton connection [138]. Further evidence that cortical microtubules' disorganization is induced by β-Yariv reagent and two mABs (JIM13 and JIM14) in root epidermal cells substantiates the hypothesis that cell surface AGPs influence the organization of cortical microtubules inside the cell [188]. In addition, the distance between cortical microtubules and the plasma membrane is increased significantly with β-Yariv reagent treatment [188]. All these findings lead to the hypothesis that altered AGP status impacts the mechanical properties of the cell wall, transmits the flow of communication from the cell wall to the microtubules by unknown transmembrane protein(s), and results in altered microtubule organization or dissociation from the membrane [138,188].
Based on the abnormalities of cellulose deposition in AGP mutants described above, it is speculated that AGPs may regulate the deposition of cellulose microfibrils by affecting the arrangement of cortical microtubules and/or the connection between cortical microtubules and the plasma membrane through transmembrane protein(s) (Figure 3). that AtFLA11 and AtFLA12 could interfere with the deposition of cellulose microfibrils during the formation of the secondary cell wall [99].
Cortical microtubules can guide CSCs to move along the microtubule array in a cellulose synthase interactive 1 (CSI1)-dependent manner and, as a consequence, to affect the cellulose microfibril angle [154,155,158,173]. The close linkage between AGPs and cytoskeletal structures, including microfilaments and microtubules, has shed light on the potential role of AGPs in cellulose deposition through a cytoskeletal network. REB1/RHD1 encodes a UDP-D-Glc 4-epimerase, which is involved in the galactosylation of AGPs and xyloglucans [186]. The trichoblasts of mutant reb1-1 are highly swollen with cortical microtubules that are disordered or even completely absent and lack certain AGP epitopes, suggesting a connection between the organization of cortical microtubules and the deposition of AGPs [186,187]. Sardar et al. demonstrated that β-Yariv treatment in tobacco tissue culture cells triggers depolymerization/disorganization of microtubules and F-actin, and cytoskeletal disruptors alter LeAGP1 localization along the Hechtian strands (a stretched plasma membrane extending from the plasmolyzed protoplast to the cell wall in plants), implying that GPI-anchored AGPs play a role in the plasma membrane-cytoskeleton connection [138]. Further evidence that cortical microtubules' disorganization is induced by β-Yariv reagent and two mABs (JIM13 and JIM14) in root epidermal cells substantiates the hypothesis that cell surface AGPs influence the organization of cortical microtubules inside the cell [188]. In addition, the distance between cortical microtubules and the plasma membrane is increased significantly with β-Yariv reagent treatment [188]. All these findings lead to the hypothesis that altered AGP status impacts the mechanical properties of the cell wall, transmits the flow of communication from the cell wall to the microtubules by unknown transmembrane protein(s), and results in altered microtubule organization or dissociation from the membrane [138,188].
Based on the abnormalities of cellulose deposition in AGP mutants described above, it is speculated that AGPs may regulate the deposition of cellulose microfibrils by affecting the arrangement of cortical microtubules and/or the connection between cortical microtubules and the plasma membrane through transmembrane protein(s) (Figure 3).

AGPs Act as Potential Signal Molecules during Cell Wall Biogenesis
Almost two decades ago, Showalter envisioned some likely scenarios for AGPs in molecular interactions and cellular signaling at the cell surface [2]. Since AGPs are proteoglycans and their protein backbone is decorated by AG polysaccharides, AG polysaccharides determine the characters of AGPs and affect their functions [3,[38][39][40], as previously mentioned in this review. So far, a series of evidence has been provided to emphasize the importance of AG polysaccharides for AGP signaling. GhGalT1 is implicated in the biosynthesis of the β-1,3-galactan backbone of AGPs and is responsible for the glycosylation of AGPs in cotton [170]. The length of cotton fibers in GhGalT1 RNAi silencing lines becomes longer. Interestingly, the level of JIM8 (a mAB)-responsive carbohydrates epitopes is decreased [170]. Prolyl 4-hydroxylases in tomato (SlP4Hs) are involved in Pro hydroxylation of AGPs. The level of JIM8-bound epitopes in SlP4H-silenced tomato plants is also altered, inferring phenotypes of root tip and branch lengthening and leaf enlargement [189]. This similarity leads to an assumption that particular carbohydrate epitopes related to JIM8 in AGPs may be associated with cell elongation and expansion. In addition, the Arabidopsis mutant mur1, with blocked biosynthesis of L -fucose in the AG polysaccharides of AGPs, displays a dwarf phenotype and a decreased root cell elongation, implying that AGPs modified by L -fucose participate in cell elongation and growth [190]. Defects in the synthesis of AG glycans of AGPs, caused by the functional disruption of KNS/UPEX1 (a type II GALT), results in pollen aggregation and reduced fertility [191]. Furthermore, GlcA residues have also been demonstrated to be essential for the biosynthesis of type II AG and normal function of AGPs [192,193]. The Arabidopsis β-glucuronosyltransferases participate in the process of grafting GlcA on AGP glycans. Mutation in AtGlcAT14A leads to a reduction of GlcA substitution and an enhanced cell elongation during seedling growth [193]. A knockout mutant of the Arabidopsis β-glucuronidase (GUS) gene AtGUS2, atgus2-1, has decreased GlcA content and shortened hypocotyl, consistent with a role for the AG polysaccharides of AGPs in cell growth [192].
The carbohydrate components of AGPs contain a lot of structural information, which makes potential candidates for chemical signals. The carbohydrate moieties can be extracellularly processed by glycosidases, such as β-galactosidases, and detached from AGPs to form free AG glycans, therefore providing the possibility for AGPs in signaling [51,194]. Some excellent reviews have given detailed information of a number of glycoside hydrolases (GHs) involved in the metabolism of AGP carbohydrate moieties, including β-galactosidases, β-galactanases, α-arabinofuranosidases, β-arabinopyranosidases, β-glucuronidases, α-fucosidases, and α-rhamnosidases [55,195,196]. However, only a few plant GHs have been reported to hydrolyze AGP glycans relative to the well-characterized AGP-degrading GHs from microbial origin [196], and more exploration is still warranted to understand the role of AG polysaccharide structure towards the AGP function in plant growth and development.
Plant cells mainly undergo anisotropic growth, including diffusion and tip growth [197]. Cell growth is achieved through strictly controlled cell wall expansion. In this unique process, the influx of water from the extracellular space forms turgor pressure to act on cell wall elasticity and extensibility. Thus, wall stress relaxation may result from the loosening and shifting of load-bearing linkages between cellulose microfibrils. Subsequently, the cell wall expands, and newly synthesized cellulose microfibrils, as well as the pre-existing wall polymers, deposit on the thinned cell wall to further re-form cross-linking with matrix polysaccharides secreted into the wall [147]. The above-mentioned deficient mutants of AG polysaccharides or GlcA residues of AGPs display cell expansion alterations, a phenotype with a delayed elongation and growth. All this is reminiscent of cell expansion, but the deposition process of new cell wall components could be disturbed, resulting in abnormal anisotropic growth of cells. It has been speculated that AGPs may regulate the cellulose deposition process in the cell wall through their AG polysaccharides as signal molecules possibly recognized by plasma membrane receptors [29], thus achieving an anisotropic growth of cells (Figure 4).

AGPs Act as Putative Ca 2+ Capacitors to Regulate Cellulose Deposition Possibly through Pectin-Ca 2+ Cross-Links
The carbohydrate moieties of AGPs may not only act as potential chemical signals but also participate in the signal transduction process by chelation with calcium ions (Ca 2+ ). AGP6 and AGP11 are two classical AGPs with specific expression and functional redundancy in pollens and pollen tubes [65][66][67][68]. The double null mutant agp6 agp11 shows phenotypes that include collapsed pollen grains, inhibited pollen tube growth, and precocious pollen germination inside the anthers [67,68]. Costa et al. found that the expression of calcium-and signaling-related genes was altered in agp6 agp11 pollen tubes, indicating the putative involvement of AGPs in Ca 2+ signaling cascades [69]. Additional studies have provided evidence for this potential function of AGPs. The AG polysaccharides of AGPs have been verified to bind Ca 2+ at GlcA residues with a binding stoichiometry of 2:1 at pH = 5, to form an AGP-Ca 2+ oscillator, thereby activating H + ATPase on the plasma membrane and allowing the influx of Ca 2+ into cells [198]. AG isolated from glcat14 triple mutants deficient in the β-glucuronosyltransferases that transfer GlcA to the AG has lower Ca 2+ binding capacity in vitro, and the plants with this defective AG have multiple developmental defects, such as reduced trichome branching, and limited seedling growth [199]. Taken together, these findings imply that the binding of GlcA on AGP polysaccharides to Ca 2+ is important for cell elongation and growth. regulate the cellulose deposition process in the cell wall through their AG polysaccharides as signal molecules possibly recognized by plasma membrane receptors [29], thus achieving an anisotropic growth of cells ( Figure 4).

Figure 4.
A putative mechanism is an enzymatic release of AG polysaccharides from AGPs that may act as signal molecules possibly recognized by plasma membrane receptors. The sugars may be cleaved by glycoside hydrolases and may function as signal molecules binding to specific receptors, as proposed by Showalter [2].

AGPs Act as Putative Ca 2+ Capacitors to Regulate Cellulose Deposition Possibly through Pectin-Ca 2+ Cross-Links
The carbohydrate moieties of AGPs may not only act as potential chemical signals but also participate in the signal transduction process by chelation with calcium ions (Ca 2+ ). AGP6 and AGP11 are two classical AGPs with specific expression and functional redundancy in pollens and pollen tubes [65][66][67][68]. The double null mutant agp6 agp11 shows phenotypes that include collapsed pollen grains, inhibited pollen tube growth, and precocious pollen germination inside the anthers [67,68]. Costa et al. found that the expression of calcium-and signaling-related genes was altered in agp6 agp11 pollen tubes, indicating the putative involvement of AGPs in Ca 2+ signaling cascades [69]. Additional studies have provided evidence for this potential function of AGPs. The AG polysaccharides of AGPs have been verified to bind Ca 2+ at GlcA residues with a binding stoichiometry of 2:1 at pH = 5, to form an AGP-Ca 2+ oscillator, thereby activating H + ATPase on the plasma membrane and allowing the influx of Ca 2+ into cells [198]. AG isolated from glcat14 triple mutants deficient in the β-glucuronosyltransferases that transfer GlcA to the AG has lower Ca 2+ binding capacity in vitro, and the plants with this defective AG have multiple developmental defects, such as reduced trichome branching, and limited seedling growth [199]. Taken together, these findings imply that the binding of GlcA on AGP polysaccharides to Ca 2+ is important for cell elongation and growth.
Ca 2+ signaling is involved in abiotic stress, wound response, stomatal movements, self-incompatibility, interaction with pathogenic microorganisms, tip growth (pollen tube growth and root hair growth), and other vital processes in plants [43,200,201], in which AGPs are also widely involved. This opens the possibility that an AGP-Ca 2+ oscillator may participate in multiple signal transduction processes in cells. Boron deficiency in A. thaliana causes Ca 2+ influx in root cells and induces the expression of calcium signaling-related genes [202]. It is speculated that boron could interact with Gal residues in the GPI anchor structure of AGPs to stabilize the anchorage of AGPs to the plasma membrane. At the same time, GPI anchors could be used as boron receptors to release Ca 2+ by an AGP-Ca 2+ oscillator in the periplasm after sensing boron deficiency and then initiate a series of downstream signal transduction processes [203]. Like AGPs, auxin is also implicated in Figure 4. A putative mechanism is an enzymatic release of AG polysaccharides from AGPs that may act as signal molecules possibly recognized by plasma membrane receptors. The sugars may be cleaved by glycoside hydrolases and may function as signal molecules binding to specific receptors, as proposed by Showalter [2]. Ca 2+ signaling is involved in abiotic stress, wound response, stomatal movements, self-incompatibility, interaction with pathogenic microorganisms, tip growth (pollen tube growth and root hair growth), and other vital processes in plants [43,200,201], in which AGPs are also widely involved. This opens the possibility that an AGP-Ca 2+ oscillator may participate in multiple signal transduction processes in cells. Boron deficiency in A. thaliana causes Ca 2+ influx in root cells and induces the expression of calcium signalingrelated genes [202]. It is speculated that boron could interact with Gal residues in the GPI anchor structure of AGPs to stabilize the anchorage of AGPs to the plasma membrane. At the same time, GPI anchors could be used as boron receptors to release Ca 2+ by an AGP-Ca 2+ oscillator in the periplasm after sensing boron deficiency and then initiate a series of downstream signal transduction processes [203]. Like AGPs, auxin is also implicated in many processes of plant growth and development, and it is also capable of triggering an intracellular Ca 2+ signal response [204]. Based on these findings, Lamport et al. proposed a novel concept of an AGP-Ca 2+ -auxin signaling cascade model: first, auxin-activated plasma membrane H + -ATPase could release H + , thus lowering extracellular pH; subsequently, AGP-Ca 2+ oscillator would release Ca 2+ that enters the cytosol through Ca 2+ channels; then, Ca 2+ recycled from the cytosol via Golgi vesicle exocytosis would recharge the AGP capacitors to form a reservoir again [43].
In addition to acting as a Ca 2+ reservoir, AGPs may also associate with RLKs to mediate various signaling transductions [92]. The Arabidopsis AtENDOL14 is a GPI-anchored AGP with a plastocyanin-like domain, which has strong and specific physical interaction with the extracellular domain of FERONIA [94]. As a plasma-membrane-localized receptor kinase, FERONIA has been recently proved to induce Ca 2+ signaling to maintain cell wall integrity during salt stress [205]. The trio of AtENDOL14, FERONIA, and Ca 2+ signaling suggests a possibility that GPI-anchored AGPs are involved in FERONIA-dependent Ca 2+ signaling [92,205].
Ca 2+ is found in the cell wall ionically cross-linked to HGs in the pectin matrix [147]. In the presence of Ca 2+ , pectin cross-linking Ca 2+ occurs to form the "eggbox" structure, which has been proposed to be load-bearing components in cell walls [206]. It has been demonstrated that pectins can bind to cellulose during its synthesis and deposition through interactions with, for example, Ca 2+ -deficient regions of HGs and binding of the arabinan and galactan side chains to the cellulose, and these bindings are reversible [207,208]. After Ca 2+ chelation, pectin cross-linking Ca 2+ may be removed and pectins recycled [69,209]. In addition, a recent study indicates that the strength of the pectin-Ca 2+ hydrogels affects cellulose structure, crystallinity, and material properties [209].
Since AGPs have a higher affinity for Ca 2+ than pectin, a discharged AGP-Ca 2+ capacitor would be recharged by Ca 2+ recycled from the cytosol and possibly from the wall matrix (e.g., Ca 2+ -pectin) [198]. Cellulose/Ca 2+ -bound pectin interactions and the novel concept of dynamic Ca 2+ recycling by an AGP-Ca 2+ oscillator underlie an interesting possibility that AGPs may act as putative Ca 2+ capacitors to regulate cellulose deposition possibly through pectin-Ca 2+ cross-links ( Figure 5).
Ca 2+ is found in the cell wall ionically cross-linked to HGs in the pectin matrix [147]. In the presence of Ca 2+ , pectin cross-linking Ca 2+ occurs to form the "eggbox" structure, which has been proposed to be load-bearing components in cell walls [206]. It has been demonstrated that pectins can bind to cellulose during its synthesis and deposition through interactions with, for example, Ca 2+ -deficient regions of HGs and binding of the arabinan and galactan side chains to the cellulose, and these bindings are reversible [207,208]. After Ca 2+ chelation, pectin cross-linking Ca 2+ may be removed and pectins recycled [69,209]. In addition, a recent study indicates that the strength of the pectin-Ca 2+ hydrogels affects cellulose structure, crystallinity, and material properties [209].
Since AGPs have a higher affinity for Ca 2+ than pectin, a discharged AGP-Ca 2+ capacitor would be recharged by Ca 2+ recycled from the cytosol and possibly from the wall matrix (e.g., Ca 2+ -pectin) [198]. Cellulose/Ca 2+ -bound pectin interactions and the novel concept of dynamic Ca 2+ recycling by an AGP-Ca 2+ oscillator underlie an interesting possibility that AGPs may act as putative Ca 2+ capacitors to regulate cellulose deposition possibly through pectin-Ca 2+ cross-links ( Figure 5).

Conclusions
A number of features including the functional redundancy of AGP family members, the complex post-translational modification process involving many related genes, a high complexity of the carbohydrate side chain structure, and the inability of β-Yariv reagent to recognize a single specific AGP hinder our complete understanding of this gene family. We have been continuously looking for links in numerous research studies on AGPs and trying to find clues that can reasonably explain the functional mechanisms of AGPs in vital activities in plants, as well as connecting these data to compile a possible mechanistic scenario. On the basis of previous studies, five models of how AGPs may participate in cellulose synthesis and deposition during cell wall biogenesis have been proposed: (A) AGPs sense extracellular signals by carbohydrate side chains and transmit signals to some receptor kinases, thereby regulating cell wall formation by promoting cellulose synthesis through an ethylene-independent ACC pathway; (B) AGPs serves as structural components affecting cellulose deposition through cross-linking to other cell wall components, such as hemicelluloses and pectins; (C) AGPs regulate the deposition of cellulose microfibrils by affecting the arrangement of cortical microtubules and/or the connection between cortical microtubules and the plasma membrane through transmembrane protein(s); (D) AGPs act as potential chemical signals with their AG polysaccharides; and (E) AGP-Ca 2+ oscillator forms by chelating Ca 2+ to regulate cellulose deposition in the cell wall possibly through pectin-Ca 2+ cross-links. These hypothetical models can provide some clues for further research on the functions of AGPs in cellulose synthesis and deposition, without discarding other mechanistic pathways that might also be involved. Since members from different AGP subfamilies have fairly distinct characteristic domains, the exact molecular mechanisms of AGP action in complicated plant biological processes, not solely devoted to cellulose metabolism and deposition, will certainly require further in-depth investigations in the near future.