What Can Boron Deficiency Symptoms Tell Us about Its Function and Regulation?

On the eve of the 100th anniversary of Dr. Warington’s discovery of boron (B) as a nutrient essential for higher plants, “boronists” have struggled to demonstrate a role beyond its structural function in cell walls dimerizing pectin molecules of rhamnogalacturonan II (RGII). In this regard, B deficiency has been associated with a plethora of symptoms in plants that include macroscopic symptoms like growth arrest and cell death and biochemical or molecular symptoms that include changes in cell wall pore size, apoplast acidification, or a steep ROS production that leads to an oxidative burst. Aiming to shed light on B functions in plant biology, we proposed here a unifying model integrating the current knowledge about B function(s) in plants to explain why B deficiency can cause such remarkable effects on plant growth and development, impacting crop productivity. In addition, based on recent experimental evidence that suggests the existence of different B ligands other than RGII in plant cells, namely glycolipids, and glycoproteins, we proposed an experimental pipeline to identify putative missing ligands and to determine how they would integrate into the above-mentioned model.


Introduction
Boron (B) is a chemical element with atomic number 5. It has two isotopes, 10 B and 11 B, with a relative abundance of 20% and 80%, respectively, giving an atomic weight of 10.81 [1]. Together with Silicon and Germanium, B is considered a metalloid because it has intermediate properties between metals and non-metals [1,2].
It is considered a light and scarce element at Cosmos (16.9 ± 2.2 atoms/10 6 Si), bypassed during normal thermonuclear reactions, which suggests that most B is formed by spallation reactions [1, and references therein]. Although B is widely distributed throughout the lithosphere and hydrosphere, its abundance varies in marine and terrestrial environments [3,4]. In marine environments, B concentration is approximately 0.4 mM, and it is homogenously distributed in a depth-independent (non-nutrient-like) concentration profile [4]. By contrast, in terrestrial environments, B is considered a scarce element, although its distribution here is heterogeneous, ranging from areas with deficiency to zones with concentrations that are considered toxic [3].
Besides being an essential element for plant growth widely studied by Plant Physiologists, during the last few years, B and B-containing minerals are attracting researchers of prebiotic chemistry. Borate was apparently essential for the early synthesis of ribofuranose [5,6], nucleic acid bases, biogenic carboxylic acids, and amino acids [7]. Furthermore, it was shown that borosilicate increased the efficiency and yielding of the Miller-Urey famous experiment [8] Within plants, the requirement of B is characterized by occurring within a narrow range of concentrations, out of which deficiency or toxicity symptoms begin to appear. These will be highly determined by the type of crop and edaphoclimatic conditions [3,9]. For instance, B is easily lixiviated in soils under high rainfall conditions, although the presence of organic matter increases retention of boric acid (H 3 BO 3 ) by esterification with cis-diol-containing residues [10], a property that can be on the basis of the physiological functions of B [11], as will be stated below.
Although Agulhon confirmed in 1910 that B existed in plants and proposed that it might be a catalytic fertilizer [12], the first studies about its essentiality began in the early 1920s when Katherine Warrington demonstrated in Vicia faba (and other plant species) that the absence of B led to a reduced root growth [13]. In fact, it is now known that B availability in many soils dedicated to agriculture is low, making the element the most limiting mineral micronutrient for plant productivity, perhaps with the exception of iron (Fe) and zinc (Zn). Indeed, B nutritional stress in plants affects many crops worldwide, limiting both yield quantity and quality [3,14].
During almost one century since Warrington's studies, a plethora of biochemical, physiological, or anatomical plant processes affected by B deficiency, which we listed in the following section, have been reported and reviewed in the past years by different authors, including Brown et al. [15], Goldbach & Wimmer [16], or Camacho-Cristóbal et al. [9]. Even more, its essentiality for plant growth has been questioned with the exciting hypothesis that B is certainly toxic for plants, and plants increase tolerance via phenolic sequestering [17]. However, the fact that organisms that do not synthesize phenolics, including animals, also suffer malformations under B-starvation [18][19][20][21], in many cases shared with those occurring in plants [22], still supports its role as an essential element.
In the last years, molecular biology approaches and high-throughput methodologies have focused B research on (i) the identification of genes increasing Boron Acquisition Efficiency (BAE) and Boron Use Efficiency (BUE); (ii) the description of omic changes in response to B deficiency and toxicity; or (iii) the characterization of signalling pathways in response to B deficiency, and have proven the participation of B in plant cellular roles not necessarily related to cell wall functioning (see IJMS Special Issue: Novel Aspects of Boron Biology in Plants. Boron and Plant Interaction, 2021) [23].
In this review, we summarized the existing knowledge about B physiology, and the demonstrated or proposed biological roles of this micronutrient in plants based on the described B deficiency symptoms. We also considered the molecules that are proposed to be ligands of boric acid/borate in plants (and other organisms) and the physiological responses of these organisms to changing B levels. Finally, we proposed and discussed a unifying model of B functions based on the reactivity of boric acid and borate to molecules containing diol groups.

One Hundred Years of Boron Research
Since Agulhon (1910) [12], a huge number of reports have explored different aspects of B plant physiological responses, which reflect the great interest aroused by this plant mineral nutrient. During the following lines, we summarized a century of knowledge in B research and honoured the authors who have greatly contributed to it. We apologize in advance for missed contributions, as there is no space to include all of them in a very few pages.
Starting with Warrington's studies [13], the first reports on B nutrition described that major symptoms of deficiency in plants were growth arrest, root swelling [24][25][26][27][28], or the accumulation of 'brown' substances, concomitant to necrosis and plant death [29][30][31]. Following the improvement of histological and biochemical techniques, different evidence that growth arrest is due to the inhibition of cell elongation [32,33] or/and arrest of cell  [34][35][36] was provided, and electron microscopy revealed defects in cell wall thickness and architecture [37].

Boron and the Cell Wall: The Only Demonstrated Primary Role
The attention of 'boronists' very soon focused on the cell wall. Katherine Warrington already indicated that B is fixed by the plant [13], and later, Skok provided evidence that B's role was related to the complexing capacity of borate ions [56].
Mazurek & Perlin described boric acid/borate complexes with diols containing compounds [57], and Loomis & Durst reported that 90% of B is associated with the cell wall fraction [58]. Years later, Matoh et al. raised the amount of B in the cell wall up to 98% [59], and Hu & Brown localized B associated with cell wall pectin and proposed that " . . . B plays a critical, although poorly defined, role in the cell wall structure of higher plants" [60].
That poorly understood role of B in the cell wall began to be defined when Findeklee & Goldbach showed that the elasticity of the cell wall is reduced under B-deficiency [61], which pointed to a B function in anchoring cell wall macromolecules and to the capacity of borate cross-linking two chains of rhamnogalacturonan II (RGII) through diol-ester bonds [62]. Matoh et al. found later that such RGII-dimer (dRGII) was ubiquitous in the cell wall of higher plants [63]. Finally, O'Neill et al. convincingly demonstrated that the growth of Arabidopsis thaliana dwarf mutants mur1 relies on the presence of apiose-borate complexes mediating the dimerization of RGII [64,65], which is described in detail in Section 3. Although RGII is a minor part of the pectin fraction [66], more than 90% appear dimerized, which is critical for plant growth. Specifically, this dRGII-borate complex is important to determine pore size and cell wall mechanic properties influencing cell wall expansion and, therefore, plant growth [67,68].
Although the basic structure of RGII and the B-mediated dRGII-B are highly conserved in vascular plants, some variability has been described among different plant species [69]. The relative abundance of the pectin fraction is lower in monocots than in dicots, likely explaining its lower B requirement [70]. Primitive plants contain traces of dRGII-B also with the same conserved structure, which suggests that genes involved in RGII biosynthesis appeared early in the evolution of land plants and that RGII dimerization was crucial for such evolution [71].

The Fine-Tuning of Boron Homeostasis. Does It Support Other Primary Roles of This Micronutrient?
Being the structural role the only demonstrated primary role of B in plants, it is proposed that many of the B deficiency symptoms result from the drastic changes in the cell wall structure and properties resulting from the decrease in dRGII-borate complexes [65]. Even more, the rapid cell death after inducing B deficiency has been attributed to defects in newly forming cell walls [72]. Nevertheless, it should not preclude the possibility that other primary functions of B may exist, particularly those related to the maintenance of membrane activities or/and regulation of developmental events, perhaps influencing cell signalling and transduction pathways [15,16,73].
Although there is still a lack of convincing evidence supporting those mechanisms underlying alternative roles of B in plant cells, the discovery of different transporters involved in B uptake and inner transport may support that this micronutrient s role goes beyond being a cell wall structural element.
Indeed, although most B is associated with RG-II in land plants [63], it can be found in the cytosol and vacuole [74] or associated with the plasma membrane [75,76]. Inde- pendently of the concentration of B in soils, its concentration in cell walls is kept almost constant, while B cytosolic levels vary with soil concentrations. This has resulted in the interesting hypothesis that a minimal level not bound to RG-II is needed to avoid several early B-deficiency symptoms [77] and that the reaction of the non-RG-II-linked B with other 'ligands' might be responsible for roles of B beyond the cell wall [11].
The uptake of B by plants was considered an unregulated passive process until the discovery of the presence of a complex transport system that was acting to maintain B homeostasis [78]. Today, B passive diffusion through the plasma membrane is considered to occur only for B uptake from soil to roots [79], but it is accompanied by transport mediated by channels [80] that have also been described in growing shoot tissues [81] and in reproductive organs [82,83]. In summary, plants sense internal and external conditions of B and rapidly regulate the expression of channels of the NIP (Nodulin26-like Intrinsic Protein) subfamily of MIPs (MAJOR INTRINSIC PROTEIN) family and transporters of the Borate Exporter family (BOR) to control B homeostasis [84]. In Arabidopsis, AtNIP5;1 is responsible for facilitating the uptake of B from soil to roots [85], and the exporter BOR1, which is localized toward the stele, is key for xylem loading under low-B conditions [86], and BOR2 seems to export B from the symplast to the apoplast to ensure efficient RG-II crosslinking [87]. Meanwhile, BOR4 is induced under high-B concentrations and is involved in the exclusion of B, enhancing B toxicity tolerance [88].
Additionally, other borate channels have been described to be expressed in different tissues and at particular developmental stages. AtNIP6;1 is expressed mainly in the node region of shoots and is involved in xylem-phloem translocation of B to growing leaves [81]; AtNIP7;1 seems to be required for pollen development [83]; and AtNIP4;1 and AtNIP4;2 expressions are related to pollen tube elongation [89]. Moreover, evidence that X Intrinsic Proteins (-XIPs), another subfamily of MIPs, can facilitate B transport to young tissues has also been reported [90]. Altogether, the complex mechanism of B uptake and translocation must guarantee continuous supply to grow cell walls, but, at the same time, it is preferentially distributed to developing meristems, which supports that B is not a merely structural element but plays a potentially key role in developmental processes.
Reinforcing the importance of B for plant development, orthologs and paralogs of Arabidopsis BORs and NIPs have been described in many plant species [91][92][93][94][95][96][97]. Sequences similarities indicate that these transporters belong to conserved gene families that show developmental stage-dependent expression patterns in different tissues to reach B requirements that ensure the proper execution of blueprints for the plant building [84].

Boron and Cell Membranes
Before demonstrating B function in RGII crosslinking, researchers focused their attention on cell membranes describing that B deficiency impairs membrane transport, membrane-associated enzymatic activities, or membrane composition, and interestingly, during the last years, the evidence of B roles as a linker element in the cell membrane and endomembranes has re-emerged.
Robertson & Loughman demonstrated a reduced absorption of phosphate under B deficiency [98], and Goldbach showed that phosphate and glucose uptake and efflux rates were decreased under B-limiting conditions [99]. Also, K-Cl stimulated ATPase [43], and ATP-dependent H + pumping and vanadate-sensitive ATPase activities were demonstrated to be inhibited by B deficiency [100]. In all cases, these effects were quickly reverted by B addition, suggesting that the membrane properties rely on B nutrition. Furthermore, B has also been involved in redox activities and the maintenance of membrane potential [101].
A structural role of B in the membrane was proposed by several researchers to explain a large number of reported effects of B on membrane processes. For instance, the fluidity of liposomes prepared from low B treated cells of sunflower was lower than in liposomes coming from B-sufficient cells [100]; deficiency reduced both total lipid and phospholipid contents in roots and leaves of Lycopersicon esculentum (tomato) and Abelmoschus esculentus (okra) [102], and Cakmak et al. (1995) showed increments of solutes leakage under B deficiency [103]. Altogether, these findings indicate that B could primarily affect membrane function by playing a structural role that protects membrane integrity. This is supported by Tanada analyses that showed a major part of B localized in membranes [104]. Nevertheless, although membranes harbour glycans as good candidates for binding B, the difficulty identifying B complexes formed with membrane components does not allow us to discard that the observed effects are secondary events of the affected cell wall.
At the end of the 20th and the beginning of the 21st Century, new techniques, and new biological models for plants, animals, and prokaryotes, have allowed the development of new approaches to unravel membrane-related B functions. For instance, the use of phenylboronic acid (PBA) as a competitor for B-binding sites caused the disassembly of transvacuolar cytoplasmic strands and cell collapse [105], suggesting either a structural role of B in the cytoskeleton or, more likely, a disruption of cytoskeletal proteins anchored to membrane glycolipids or/and glycoproteins. Also, following a similar experimental procedure, it was shown that PBA induced abnormal internalization of PIN1, blocking auxin transport and generating abnormal Arabidopsis embryo early development, which supported that PBA competes with B for membrane proteins [106].
A large amount of information relating B with membrane glycans has come from the research performed on the legume-rhizobia symbioses [107][108][109]. This symbiotic interaction triggers the development of a new organ, the root nodule, that follows a unique process of organogenesis characterized by events in which exists an intense membrane synthesis. It is estimated that the membrane synthesis rate is about 30 to 50-fold higher than in other plant-growing organs [110]. Matching with the analyses reported by Tanada, the content of B in nodules is higher than in roots or shoots, likely because it is demanded by such an amount of membrane synthesis [111]. Therefore, the legume-rhizobia symbioses is a very suitable model to investigate the role of B in membrane-located processes. The bacteria proliferate inside the nodule and differentiate into N 2 -fixing bacteroids enclosed by a plantderived peribacteroid membrane (PBM), which differentiates a glycocalyx composed of new glycolipids and glycoproteins [112,113] involved in bacteria-plant cell surface interactions important to ensure the success of the symbiosis [114]. Some of those components are either abnormally glycosylated [107] or not detected [108] in B-deficient nodules, resulting in cell cycle and the cell division-cell differentiation transition misregulation that leads to a tumour-like development [109]. Recently, an abnormal N-glycosylation during early development under B deficiency has been described in pea nodules, Arabidopsis roots, and Dario rerio [22]. Like in nodules, aberrant root apical meristem in Arabidopsis and a failure of zebrafish organogenesis occurred. Although the described aberrant development could be due to defects in the cell wall structure, the effects observed in the animal model support a primary role of B in membranes, likely related to the synthesis and stability of glycan moieties of glycoproteins and glycolipids. As mentioned above, cis-diol-containing sugar residues, harboured by the cell glycocalyx in membranes and matrices, are potential candidates to be ligands of B. Several of them have already been identified in studies that will be described later [75,76,115].

Boron and Developmental Events
Back in 1985, Lovatt published an interesting hypothesis stating that the evolution of the xylem resulted in the acquisition of the essentiality of B for apical meristem activity and conferred the advantage of preventing B toxicity [116], coinciding with the recent consideration of Lewis that affirmed that B is a toxic element for vascular plants [17]. Lovatt suggested that a threshold concentration of B must reach meristematic cells to promote division and subsequent expansion to ensure growth, preventing the accumulation of B in meristems to a toxic concentration. The evolution of the xylem ensured a gradient of B, lowering its concentration in the growing cells in the elongation zone and reaching a critical minimum content in the meristematic cells to elicit mitosis. The hypothesis was supported by the fact that B is toxic to most organisms at relatively very low concentrations, being vascular plants the most tolerant, and by observing that DNA synthesis, cell division, and elongation are inhibited under B deprivation and soon reversed after B supply. The review also proposed that regulation of cell division by B may be potentially common to other organisms than vascular plants.
More recent studies have shown that induced B deficiency or loss of function mutations on B translocators alters cell cycle regulatory pathways, cell differentiation, and the development of vegetative and reproductive structures [91,[117][118][119][120]. Particularly, responses to deficiency described in those and other studies recently reviewed [121,122] suggest that B can induce molecular pathways that determine meristem fate and place B nutrition as a key regulator of developmental processes.
Indeed, the earliest defects following B-starvation are growth arrest and aberrant meristem formation [119,123,124]. This growth arrest can be attributed to defects in cell elongation or differentiation due to abnormal cell wall architecture [65,119,125], although evidence of cell division inhibition has also been reported. Actually, at the initial steps of B nutrition research, the works performed by Sommer and Sorokin supported that B deficiency caused root tip malformations and impairment of cell division [29]. And as mentioned above, using PBA to mimic B-deficiency, it was shown that root apical meristem (RAM) formation was disrupted in embryos as early as the first asymmetric cell division of the hypophysis appeared [106], placing B as crucial for embryo formation. Furthermore, Poza-Viejo et al. reported a reduction of cell division 4h after transferring Arabidopsis seedlings to severe B deficiency media due to inhibition of the G1-DNA replication phase transition [120]. Cell division inhibition was accompanied by a later loss of identity of the quiescent centre (QC) that could be attributed to the down-regulation of CCS52A2 that controls QC and the maintenance of surrounding stem cells [126].
As previously stated, the interaction of legumes with soil rhizobia triggers an interesting process of plant organogenesis in which cell division, cell elongation, and cell differentiation must be finely regulated [127,128]. Particularly crucial is the activation by CCS52A of the transition from mitosis to endoreduplication to gain polyploidy required for cell elongation prior to bacteria invasion [129]. Interestingly, CCS52A is also downregulated during early organogenesis of B-deficient nodules, leading to failure of cell elongation and cell differentiation [117]. Both in the QC and in nodule cells, expression of CCS52A(A2) promotes ubiquitination and proteolysis of the anaphase-promoting complex, resulting in cell polyploidy. This is crucial to maintain QC identity and mitotic activity of surrounding stem cells of RAM and to induce nodule cell elongation, bacterial invasion/spreading, and cell differentiation, respectively.
Development of reproductive organs is often more sensitive to B deficiency than vegetative growth [123]; therefore, it is not surprising that specific mechanisms of B homeostasis are induced at a particular moment in which shoot meristems transition from vegetative to floral development [130]. Apparently, BRAHMA (BRM) protein, which is degraded in response to high B [131], maintains the juvenile phase [132]. The transition to a mature phase prior to reproductive development may be the consequence of the reduction in BRM activity in response to the B translocation increase driven by B transporters [122].
Altogether, it seems that this micronutrient could play central roles in molecular regulatory pathways of the embryo and post-embryo plant development.

Boron, Cell Signaling Mechanisms, and Gene Expression Regulation
Recent studies using microarrays or RNAseq have provided increasing evidence supporting B nutrition's effect on the regulation of gene expression affecting metabolism, cell wall, and membrane integrity and function, stress response, or micronutrient homeostasis [118,[133][134][135][136][137]. As summarized in the following lines, almost every signalling and transduction pathway is activated in response to B deficiency, which allows us to hypothesize that the micronutrient may be involved in cell signalling. Nevertheless, many experimental data support that most of the responses are linked to the effect on cell walls. Briefly, based on our and other author s studies, Kobayashi et al. [72] proposed that the disturbed pectin network decreases the tensile strength of the cell wall leading to an in-crease of turgor pressure that stretches the cell membrane triggering a stress response that resembles hypersensitive responses to pathogens. Transduction of such mechanical signal involves a rapid influx of Ca 2+ , ROS production, and MAPK cascades that result in auxin/ethylene-mediated cessation of growth and cell death [125,[138][139][140][141]. This is supported by the fact that blocking Ca 2+ channels in B-deprived cells largely inhibited the expression of stress-responsive genes [139] or adding antioxidant reagents can prevent the death of B-deficient cells [138] by restoring cell elongation. Similarly, blocking ethylene biosynthesis or perception or using mutants defective in ethylene or auxin response can restore B deficiency molecular responses [125]. Furthermore, destabilization of the cell wall under Ca deficiency but not under other nutrient deficiencies such as K or Mg that are also involved in pectin cross-linking triggers a similar response [140]. On the contrary, preincubating B-deficient cells with a supplement of extra Ca increases cell wall strengthening, attenuating the expression of B-deficiency-responsive genes [140]. Similarly, the addition of Ca partially restored the impaired development of B-deficient legume nodules [111] and the expression of 75% of genes affected by stress [118].
In roots, Ca influx, ROS production, and cell death occur preferentially in the elongation zone [72]. In line with this, maize mutants affected in the synthesis of B transporters, which are involved in the transition to the reproductive phase, develop defective inflorescences with reduced RG-II dimerization, which is largely restored by adding B into the media [91]. Therefore, failure of meristem formation and functioning can also be explained by the mechanosensitive response to B deficiency. But is that all?
Based on the comparison among different gene expression profiles, the mechanosensitive hypothesis proposes a pathogen-like response under B deficiency. However, Bdeficiency disturbs cell wall structure, and it is expected that the upregulation of genes is involved in cell wall functioning. Nevertheless, while genes related to the cell wall structure, included in different transcriptomic analyses, are upregulated after pathogen attack [142], they appear downregulated under B deficiency [72,134], suggesting that B might be required to induce the expression of cell wall synthesis and assembly-related genes [74]. Using the legume nodule model, the first visible symptoms of B deficiency appear early after root inoculation with rhizobia [143], and although pathogenesis-related proteins were synthesized [144], oxidative damage was not detected even 3 weeks postinoculation [145]. At this developmental stage, cell death is not observed, but it appears an abnormal cell division resembling tumour behaviour [109], indicating an early failure of development and suggesting that B deficiency is not necessarily associated with cell death. Furthermore, low-B results in abnormal embryonic development in animals [19], also leading to a tumour-like amorphous structure when B deprivation occurs at the early cleavage stage [22]. Interestingly, boric acid can inhibit cell proliferation in different cancer cell lines [146,147]. Such implications of B nutrition in animal physiology/development, together with the organogenesis failure in plants and animals, claim for alternative or/and additional sensing/response mechanisms to low B conditions.
The existence of a putative B sensor molecule in plants able to detect external B concentration remains still unknown. Certainly, the loss of cell wall integrity that can alter turgor pressure could be the cellular signal triggering the B stress response. In agreement with such a hypothesis, the literature offers other possible sensing mechanisms associated with soluble, not necessarily external, B. Regarding B deprivation sensing mechanisms, a computational model for B distribution in roots localized the highest concentration of soluble B around the QC, which might be likely used to keep RAM activity [148]. Also, the fact that the 5 -untranslated region (UTR) of NIP5;1 responds to the increase in cytosolic B promoting mRNA degradation [149] led to the assumption of a sensor mechanism acting in the cytosol resulting in the development of biosensors of cytosolic B [150]. Interestingly, this 5 -UTR response to B seems to function also in animal cells [149]. Additionally, there are different cell wall receptors that act in response to stresses regulating cell wall dynamics [151]. Also, the arabinogalactan-proteins (AGP) have been proposed as sensors of soluble periplasmic B because, as we described later, they contain sugar residues suscep- tible to interaction with B [152]. Afterwards, the signal would be conveyed to the nucleus, and several cell signalling transduction pathways can be implicated. In agreement with this hypothesis, Dumont et al. demonstrated that the inhibition of root cell elongation induced by the fucose analogue 2-fluoro 2-L-fucose (a chemical inhibitor of RG-II biosynthesis) was partially restored by boric acid supplementation without rescuing RG-II synthesis nor dimerization [153]. These observations suggest that B itself, rather than the RG-II dimer, is an essential component of the cell wall integrity-sensing mechanism that controls cell elongation, perhaps due to its ability to bind to the cis-diol motifs of signalling molecule(s).
The potential role of phytohormones in the regulation of B stress responses has been widely studied. By applying pharmacological approaches combined with reverse genetics using mutant lines affected in hormone synthesis or hormonal perception/transduction pathways, it was described that B deprivation alters plant development affecting synthesis, transport, or/and reception of auxins [106,154], ethylene [144,155], cytokinins [119,120,156], brassinosteroids [157], jasmonic acid [158], and the cross-talk among hormones [159]. Thus, phytohormone regulatory pathways are considered crucial in regulating cell signalling in B nutrition, although other cell signalling mechanisms common in plants and animals must also be important.
As previously mentioned, blocking sites of B binding with PBA led to cytoskeleton disruption [105], and levels of actin and tubulin increased in response to short-term B starvation due to the altered cytoskeleton polymerization [160]. Therefore, B might be involved in signalling through a cascade of signals via the cell wall-plasma membrane-cytoskeleton continuum [16], through endocytosis of signalling elicitors. Supporting this hypothesis, the abundance of homogalacturonan and RGII rapidly increased in cell walls shortly after B-deprivation in Zea mays, and their endocytosis was also inhibited [161]. In animals, the maintenance of the membrane-cytoskeleton continuum and, hence, endocytosis-mediated signalling also support the essentiality of B in the development of these organisms.
Also, the fact that the increase of cytosolic Ca 2+ (cytCa 2+ ) is a rapid response to B deficiency could explain why B may be involved in signalling through transduction pathways activated by Ca 2+ [152], which could also be extended to animals. Many abiotic and biotic stresses induce an increase of cytCa 2+ following the activation of cyclic nucleotidegated Ca 2+ channels (CNGCs) [162]. A plasma membrane-localized CNGC was found to be upregulated in Arabidopsis in response to B deficiency [163]. Therefore, it has been proposed that membrane sensors of B-deficiency could induce activation of CNGCs resulting in Ca 2+ increments that could activate Ca-related proteins, such as calmodulin (CaM). Ca-CaM regulates then different transcription factors and B-responsive genes [152].
Right after discovering that B is part of a signalling molecule in bacteria (AI-2 quorumsensing autoinducer) that interacts with the sensor protein LuxP [164], another tentative working hypothesis to explain the possible function of B is that it is a cellular signal itself or that it is implicated in a soluble B-complex that interacts with different transcription factors. In line with this, Kasajima et al. described that WRKY6 is a transcription factor involved in response to B deficiency in Arabidopsis [165], which also plays an important role in embryogenesis [166]. Besides, B could interact with -hydroxyl groups (OH) of amino acid residues (as serine or threonine) of different transcription factors, as in the bacterial LuxP. However, there is no evidence to date of this type of binding.

Boron Complexing Molecules
Boric acid and borate form cyclic diesters with cis-diols containing molecules, being the most stable those on a furanoid ring structure [57]. Prior demonstration of RGII cross-linking, Loomis & Durst postulated the existence of borate-apiose complexes in the plant cell wall, or borate-mannose/galactose links that stabilize cyanobacterial heterocyst envelope, explaining the structural roles of B and, also, the B toxicity effects as borate can form diesters with ribose of ribonucleotides and RNA [58]. After the "dRGII-B" discovery, two more findings prompted the idea that the structure or/and function of other molecules depends on their interaction with the borate. The AI-2 quorum-sensing autoinducer and subsequent transformation in a borate-diester furanosyl structure generate the active form complex [164], which implies that B is involved in signalling pathways; furthermore, the demonstration that borate is required for the synthesis of stable pentoses, including ribose [5], suggested that B was involved in the prebiotic synthesis of biomolecules. Later, it was described that the in vitro synthesis of ribonucleotides was enhanced by B binding to 2 and 3 positions [167]. Therefore, it was considered that B could be a "staple" element that stabilizes the synthesis and/or activate molecules. Thus, finding molecules whose activity/function depends on borate cross-linking is considered key to demonstrating the primary roles of B (for example, furanoses, adenylates, inositides, polysaccharides, and glycans) [11]. Along this section, we shortly reviewed the main research focused on unravelling B complexing molecules in plants (summarized in Table 1) and their relationship with B deficiency in order to link particular B-complexes with functions in plant growth and development.

Rhamnogalacturonan II
Rhamnogalacturonan II is, perhaps, structurally, the most complex polysaccharide identified on Earth. It has a main chain similar to homogalacturonan and four branches (named A to D) composed of 13 different residues and more than 20 different linkages. RGII requires a complex process of synthesis that involves more than 50 genes, of which only a few have been identified (for review, see [66,173]). In parallel with the characterization of RGII structure, B was identified to mediate the dimerization of two RGII molecules (dRGII-B) in different plants [62,63,174,175] and was shown its implication in the in vitro dimer formation treating, under certain conditions, monomeric RGII (mRGII) with boric acid [176]. Later, it was determined that dimerization is due to borate esters at O-2 and O-3 of the apiosyl residue of side chain A in each RG-II [177,178].
Once demonstrated the in vivo dRGII-borate complex formation it remained the question of whether cross-linking occurs during the synthesis of RGII in the endomembrane system, after secretion to the periplasmic space, or in muro once it is incorporated into the cell wall. Supported by in vitro boric acid mediated dimerization of mRGII obtained by acid hydrolysis of cell walls [176,177], the hypothesis of in muro dimerization was first accepted and maintained for about 15 years. Studies on cultured cells confirmed a correlation between B levels and RGII dimerization, since the addition of B to B-deprived cells increased the amount of dRGII-B, but not the amount of mRGII retained in walls of Bdeficient cells [179,180]. These results suggest that dimerization occurs in newly synthesized RGII that could occur during intracellular synthesis or during secretion. Chormova et al. observed that B-deficiency acclimated cells resupplied with B form dRGII-B, but when treatment to inhibit de novo synthesis of polysaccharides preceded B resupply, dRGII was not detected despite maintaining mRGII amounts [181,182]. Therefore, the authors concluded that B-bridging must occur in Golgi or during exocytosis, but not after secretion into the apoplast. Recently, in a finely performed work using [ 14 C]glucose to label the newly synthesized RGII in Rosa and Arabidopsis cell cultures, the same team concluded that B-bridging of RGII "occurs predominantly within the Golgi system, prior to release into the apoplast, and continue at a much-reduced rate after secretion" [183].

Cell Wall and Extracellular Matrix Glycoproteins
Although demonstrated dimerization of RGII by B seems to occur mainly in the endomembrane system, is still unclear the precise mechanism controlling the in vivo dimerization. There is huge evidence that suggests that it must be accompanied by interaction, likely mediated also by B, with the cell wall and matrix glycoproteins. Extensins and other cationic histidine-rich-, proline-hydroxyproline-rich (HPRG)-glycoproteins have been proposed as natural chaperones that facilitate the formation of dRGII-B [182]. Extensins can be located on the inner face of the cell wall, interacting with B and "waiting" for the secretion of newly synthesized RGII, perhaps as B-donors, to immediately catalyse cross-linking or, alternatively, it could be located inside the Golgi cisternae or Golgi-derived vesicles to dimerize RGII before secretion. Although it is not demonstrated that wall glycoproteins interact, at least transiently, with B, immunocytochemical analysis of Phaseolus vulgaris B-deficient nodules revealed that cortical cells have walls with not covalently linked HPRG [169]. Since the synthesis was not apparently affected by B deficiency, it was interpreted that B has a role in the assembly of wall components prior to secretion.
Periplasmic arabinogalactan proteins (AGPs) comprise a major and highly diverse group of plant cell matrix glycoproteins, many of which have GPI-lipid membrane anchors and have been considered analogous to extracellular matrices of animal cells (reviewed in [184,185]). The glycan moiety might act as a soluble signal and as a coreceptor of morphogens regulating embryo [186][187][188] and post-embryo pattern formation [189][190][191], and many other processes of plant growth and development [184,185]. The three mannose residues and the galactose residues of AGPs with GPI-lipid anchor are potential sites of interaction with B that have led AGPs to be postulated as sensors of B deficiency [152].
The first evidence of interaction between B and matrix glycoproteins came from using the specific boric acid-chelating resin Amberlite IRA743 [192] (which can "capture" potential molecules complexed with B) following affinity chromatography of fractions derived from Pisum sativum symbiotic nodules and immunostaining with specific antibodies. Besides RGII, a legume-specific glycoprotein from nodule infection threads and extracellular matrices was identified as a putative borate ligand [145]. This glycoprotein, first termed root nodule extensin (RNE), was structurally characterized as a heteropolymer that alternated glycol motifs of extensin and AGP. Later, it was renamed arabinogalactan protein-extensin (AGPE) [193]. This protein is implicated in the apical growth of infection threads which seems to be regulated by the transition from fluid to a solid state of AGPE through peroxide-driven tyrosine cross-linking at the tip growing point [194]. In B-deficient nodules, infection threads appeared swollen with inhibited apical growth and aborted prematurely [195], suggesting that the interaction of AGPE with borate is crucial. These results share similarities with the previously known requirement of B for the growth of pollen tubes, one of the most characteristic events of cell apical growth in angiosperms [196,197]. Very interestingly, AGPs secreted to the stigma, and the style are crucial for pollen tube guidance [198].
The legume AGPE structure suggests that it is covalently associated with cell wall components reinforcing the infection thread's structure [193]. Indeed, epitope tag modifications altered its behaviour in the extracellular matrix, preventing crosslinking in the cell walls of transformed tobacco cells [199]. After the demonstration that AGPE is a potential ligand of B, co-immunoprecipitation assays allowed identifying of a stable association between AGPE and RGII in B-fed legume nodules that was not detected in B-deficient nodules [115]. Therefore, although there is still a lack of evidence of the nature and specific

Ligands in Cell Membranes: Glycolipids and Glycoproteins
An important amount of B was detected associated with membrane fractions [105]. Plant cell membranes have glycolipids and glycoproteins that contain different residues such as galactose, mannose, inositol, and hydroxyl-aminoacids that contain sites of potential interaction with borate. In 1977, Pollard et al. [43] hypothesized that the simplest explanation for impaired membrane functions under B deficiency is that B interacts directly with polyhydroxy components of membrane glycoproteins and glycolipids. However, the first evidence came up much later.
By studying legume nodule development using immunohistochemistry, it was circumstantially detected that αRGII labelled the peribacteroid membranes (PBM) in B-sufficient but not in B-deficient nodules [108]. Following fractionation, three glycoproteins sharing antigenicity with RGII were identified and appeared in both PBM and cell membranes of differentiating cells and disappeared once cell differentiation was completed. The fact that antibody labelling was concentrated at the interface membrane-apoplast or PBM-peribacteroid space and that the vesicle merging that is targeted to symbiosomes fails in B-starved nodules led to postulate that B stabilizes RGII-glycoproteins in Golgi-derived vesicles and facilitates vesicle fusion during symbiosome development and nodule organogenesis [109]. Nevertheless, it was not demonstrated that those glycoproteins were B interactors.
One year later, Dr Goldbach's group, using phenyl boronate affinity chromatography followed by 2-D electrophoresis and MALDI-TOF, isolated and identified membrane proteins able to interact in vitro with borate in root microsomal fractions of Arabidopsis and Zea mays [75]. Several of the identified proteins are especially interesting as they are related to processes affected by B deficiency. For example, the binding capacity of B to various H + -ATPases can be related to the early reported affected H + transport through membranes [100] since B may stabilize the enzymes in the membrane; several are involved in plant defence responses, which match well with the model of defence-like response to B-deficiency described by Kobayashi et al. [72]. Particularly relevant was the identification of endoplasmic reticulum (ER) B-binding proteins, including different luminal-binding proteins such as the chaperone BiP which facilitates the assembly of protein complexes within the ER and also acts in the ER quality control mechanism that recognizes and sends to degradation abnormally folded proteins [200]. Disruption of ER homeostasis led to ER stress and to the accumulation of misfolded proteins in its lumen. Furthermore, the accumulation of N-glycosylated proteins is a general feature of the development of legume nodules, roots, and also of Danio rerio embryos [22]. Using Amberlite IRA-743 in affinity chromatography experiments, several linked-glycan-linked glycoproteins were isolated from legume nodules developed with optimal B nutrition but not from B-starved nodules [201], suggesting that the interaction with B is important to ensure a proper glycosylation/folding. Besides, B-sufficient nodules also yielded BiP isoforms but not from B-deficient nodules. The use of an anti-BiP antibody revealed that these proteins were synthesized and even accumulated in B-deficient legume nodules and Arabidopsis roots, which indicated that B-deficiency can lead to ER stress.
Concerning glycolipids, glycosylinositol phosphoryl ceramides (GIPCs) ubiquitous in plants, animals, and bacteria were circumstantially isolated and identified as B-binding glycolipids from vegetative cell membranes of the cyanobacterium Anabaena cultivated in media with B, but not in the absence of this micronutrient [202]. Also, Voxeur and Fry discovered that disruption of borate ester linkages enhanced the extractability of glycosylinositol phosphoryl ceramides (GIPCs) from Rosa cells and showed that GIPCs are able to interact with RG-II, possibly forming a GIPC-B-RGII complex that can favour the dimerization of RG-II [76]. Interestingly, GIPCs are major components of lipid rafts in eukaryotic cells, and it is proposed that the alteration of lipid rafts can be responsible for the effects of B deficiency. Lipid rafts are functionally membrane microdomains that can bind GPI-anchored, transmembrane protein or linked-glycan-linked proteins that are likely transported from the endoplasmic reticulum and assembled in the Golgi apparatus [203]. They play important roles in vesicle biosynthetic and endocytic traffic, and GPI, N-glycans, or G-proteins associated with lipid rafts can activate different signal transduction pathways, also in animal development, which can be the basis to explain the failure of plant and animal organogenesis under B deficiency.
Overall, interactions mediated by B involving RGII-extensins-AGP-GPI might be essential to maintain structurally and functionally the cell-wall-membrane-cytoskeleton continuum. Moreover, the binding of B to GIPCs and glycoprotein associated with lipid rafts apparently stabilized the function of the endomembrane system, vesicular traffic, and signal transduction pathways associated with the membranes, which may be common for plants and animals.

Soluble Potential Ligands
Soluble borate complexes with sorbitol, mannitol, fructose, sucrose, and N-acetylserine have been identified in phloem sap as mechanisms of phloem B mobility [170,171].
Fructoborates have a positive impact on human health when used as a dietary supplement of B [204]. They are considered natural sources of B to satisfy B requirements of animals and humans, but there is no evidence of their regulatory roles.
Many regulatory and/or signalling molecules are ribonucleotides with the potential capacity of forming B complexes through diesters with 2 and 3 OH of its ribosyl residue. Ralston and Hunt used capillary electrophoresis to identify and quantify B binding to adenosine-containing molecules and ranked it from S-adenosylmethionine (SAM) with the highest affinity for B decreasing in NAD, adenosine phosphates (ATP, ADP, AMP), and cAMP with the lowest affinity [172]. A plausible explanation of the undoubtedly beneficial effects of dietary B for human and farm animal health partially holds on the formation of adenylate-borate complexes [205], which can be common to plants and has been inherited by "plant boronists" as a working hypothesis. However, to date, there is no experimental evidence of in vivo borate-complexing with nucleotides. Moreover, B was hypothetically important for the stable prebiotic synthesis of ribonucleotides [167], but once synthesized, the phosphate at the 5 -position destabilizes borate bound to ribose [206]. This can explain why SAM has a higher affinity for B binding and, at the same time, does not support the in vivo complexation of ribonucleotide phosphates. SAM decreased in the liver of rats fed with poor B diets [207], and it is the precursor for bacterial AI-2 synthesis [164]. Therefore, it is a more plausible ligand of B with biological importance. Nonetheless, the potential in vivo SAM-B complexes remain still to be demonstrated.
Other potential B ligands are phosphoinositides (PI) and soluble inositol phosphates (IP), which contain two hydroxyl groups in cis configuration at positions 2 and 3 [11,15,205]. Based on the observed increase in cytosolic Ca 2+ concentration in response to B deficiency [139,152], a potential complexation of B to PIs or IPs could prevent IP3-mediated calcium release and the subsequent signalling cascade in response to B. However, to date, B-inositol complexes have not been isolated. Moreover, it is likely that the proximity of several phosphate groups destabilized B bounding, as occurs in ribonucleotides.
An important group of regulatory molecules at the post-transcriptional level are miRNAs, which are also potential ligands of B. Several reports describe that miRNAs are important regulators for the adaptation of plants to B-deficiency [208,209] and toxicity [210]. However, to our knowledge, the formation of in vivo B-miRNAs has not yet been explored. On the contrary, it has been shown that borate favours the destabilization of polymeric RNA, and therefore, it is postulated that its participation in the prebiotic synthesis should be restricted to ribonucleotides and not extended to polynucleotides [211].
Altogether, there is a lack of evidence for the essential roles of soluble-borate complexes during growth and development. On the contrary, some studies reconsider the idea of Loomis & Durst, stating that the interaction of B with ribose of ribonucleotides may explain some effects of B toxicity [58] and support that B can disrupt metabolism and development by binding to ATP, NADH or RNA [9,212,213]. Nevertheless, the 'toxical' binding of B to ribonucleotides could be used to achieve a beneficial effect following B resupply after a B-deficient period. Indeed, it has been demonstrated that boric acid complexed to NAD + inhibits ADP-ribosyl cyclase that in eukaryotes converts NAD + to cyclic-ADP-ribose (cADPR), a second messenger that triggers the release of Ca 2+ from storage compartments [214]. This 'toxic' action of B inhibiting Ca 2+ release by complexing NAD inhibits the proliferation of prostate cell cancer [215]. Hypothetically, the resupply of B to B-deficient plants could inhibit Ca 2+ release by the same or similar mechanism, complexing cNMP precursors and inhibiting cytosolic Ca 2+ increase through cyclic nucleotide-gated Ca 2+ channels. This could therefore stop Ca 2+ -induced accumulation of ROS and prevent cell death that is activated in response to B deficiency, being a mechanism to return to a normal physiological status.

A unifying Model for B Function(s) in Plants and Concluding Remarks
Boron deficiency in plants has been widely explored for 100 years at cellular, physiological, and molecular levels. Considering the main purpose of this review of proposing a unifying model for B deficiency responses and for B primary roles, we have focused on the early events triggered by B deficiency which are temporally and spatially relevant at the cell/tissue scale trying to connect potential primary B functions with potential ligands. Based on it, we summarized our proposal in Figure 1.
However, to our knowledge, the formation of in vivo B-miRNAs has not yet been ex-plored. On the contrary, it has been shown that borate favours the destabilization of polymeric RNA, and therefore, it is postulated that its participation in the prebiotic synthesis should be restricted to ribonucleotides and not extended to polynucleotides [211].
Altogether, there is a lack of evidence for the essential roles of soluble-borate complexes during growth and development. On the contrary, some studies reconsider the idea of Loomis & Durst, stating that the interaction of B with ribose of ribonucleotides may explain some effects of B toxicity [58] and support that B can disrupt metabolism and development by binding to ATP, NADH or RNA [9,212,213]. Nevertheless, the 'toxical' binding of B to ribonucleotides could be used to achieve a beneficial effect following B resupply after a B-deficient period. Indeed, it has been demonstrated that boric acid complexed to NAD + inhibits ADP-ribosyl cyclase that in eukaryotes converts NAD + to cyclic-ADPribose (cADPR), a second messenger that triggers the release of Ca 2+ from storage compartments [214]. This 'toxic' action of B inhibiting Ca 2+ release by complexing NAD inhibits the proliferation of prostate cell cancer [215]. Hypothetically, the resupply of B to Bdeficient plants could inhibit Ca 2+ release by the same or similar mechanism, complexing cNMP precursors and inhibiting cytosolic Ca 2+ increase through cyclic nucleotide-gated Ca 2+ channels. This could therefore stop Ca 2+ -induced accumulation of ROS and prevent cell death that is activated in response to B deficiency, being a mechanism to return to a normal physiological status.

A unifying Model for B Function(s) in Plants and Concluding Remarks
Boron deficiency in plants has been widely explored for 100 years at cellular, physiological, and molecular levels. Considering the main purpose of this review of proposing a unifying model for B deficiency responses and for B primary roles, we have focused on the early events triggered by B deficiency which are temporally and spatially relevant at the cell/tissue scale trying to connect potential primary B functions with potential ligands. Based on it, we summarized our proposal in Figure 1. (a) At the cellular level, boron sufficiency (+B) ensures the dimerization of rhamnogalacturonan II (RGII) and the bridging of other cis-diol-containing molecules such as glycoproteins and glycolipids, so the cytoskeleton/plasma membrane/cell wall (C/PM/CW) continuum is maintained, and root development progress normally. In B deficiency (−B), cells cannot sustain RGII dimerization nor crosslinking of other cis-diol-containing molecules, resulting in the accumulation of vesicles likely containing monomeric RGII and glycoproteins in the cell. In parallel, the lack of dimeric RGII would lead to the loosening of the cell wall, which perceived by specific cell wall receptors, will trigger primary a Ca2+ influx and ROS burst, and then the accumulation of the stress phytohormones ethylene and jasmonic acid, root morphogens auxins and cytokinins, and the inhibition of brassinosteroid signalling. (b) At the root level, the above-mentioned signalling pathways will arrest growth by targeting numerous processes in root development: exhaustion of the quiescent centre, arrest of cell division, inhibition of cell elongation, and early cell differentiation.
To date, the only undoubtedly confirmed function of B in plants is the dimerization of RGII [65], which determines cell wall porosity and strength [67,68]. How to connect this primary function of B with the signalling pathways involved in the B deficiency response? The mechanical stimuli proposed by Kobayashi et al. [72] might not be sufficient to explain the diversity of symptoms of B-deficient plant growth and development and could not be the unique sensing mechanism. The description of cell wall receptors for particular stresses [151] opens the possibility that plants could perceive specifically dRGII-B status and trigger specific responses to low borate-mediated dimerization. Among the families of cell wall receptors, wall-associated kinases (WAKs) are the best-characterized pectin receptors [216]. The induction of WAK2-dependant activation of invertase is detected after in vitro addition of RGII to protoplast, but no RGII (nor RGI) binding was detected [217]. Other cell wall receptor family interacting with pectin comprises proline-rich extensin-like receptor kinases (PERKs). They are kinases located at the plasma membrane which contain a proline-rich, extracellular domain similar to that of extensins likely embedded in the cell wall, that likely cross-link to wall extensins [218], and that are not covalently bound to B-deficient walls [169]. Many AGPs with GPI-anchors bind cell wall pectins [219], and because the GPI anchor may be cleaved by phospholipases and the AGP moiety liberated from the plasma membrane [220] could also act as cell wall sensors. Since it was shown that legume-AGPE binds B and B promotes the formation of AGPE-RGII complex [115], it should be considered a candidate to sense B-deficiency.
These cell wall receptors would transmit B deficiency through the membrane and the cytoskeleton continuum to elicit several signalling pathways, which would explain the pleiotropic effects of B deficiency on plant growth and development. Briefly, several of the B-deficiency symptoms summarized in this review could be explained by the combined activation of several phytohormone regulatory pathways, which would include, at least, jasmonic acid-dependent wound/Cell Wall Damage (CWD) response [158], ethylene, auxin, and auxin/ethylene cross-talking [106,141,154], cytokinin (CK) and auxin/CK cross-talking [120,156,221], or brassinosteroids (BR) [157].
Although not mediated by phytohormones, transduction of B-deficiency sensing through ubiquitous pathways involving Ca 2+ and Ca 2+ -dependent activation of calmodulins would satisfactorily explain symptoms of B-starvation in plants and in organisms without RGII or devoid of cell walls [152], whenever a sensing mechanism of B not located in cell wall exists. A diversity of membrane glycolipids and/or glycoproteins common to plants, animals, and even bacteria, are already identified as potential ligands of B are good candidates [77,78,202], and their interactions with B can maintain membrane integrity and membrane transport and signalling mechanisms. The first evidence of B essentiality in animal development was dysplasia during the early cleavage period of zebrafish embryonic development, associated with membrane blebbing and cytoplasm extrusion on the animal pole [222]. Images of low B embryos resemble those when the release of extracellular vesicles (EVs) occurs and suggest uncontrolled extracellular dumping. EVs, and particularly exosomes, are naturally formed by cells, contain signalling or regulatory molecules, and participate in cell-to-cell communication [223].
EVs formation is associated with membrane lipids microdomains, or lipid rafts [224,225]. Therefore, observed membrane blebbing could be attributed to abnormal lipid raft function. Supporting this hypothesis, several glycolipids and glycoproteins with potential B-binding capacity are associated with lipid rafts [203]. Lipid rafts do not occur only in the plasma membrane but also are organized in intracellular membranes. They are essential for maintaining almost all cellular functions, including, besides EVs formation, the spatial organization of the plasma membrane, sensing and transduction, lipid, and protein trafficking from ER to Golgi apparatus, vesicle trafficking from Golgi to different target sites, and the formation of endocytic vesicles and endosome movements [224]. Cellular and biochemical approaches provided evidence that B is required for correct endomembrane trafficking. Short-term B deficiency led to rapid accumulation of both actin and tubulin in Arabidopsis [161]. This coincides with ultrastructural changes described in B-deficient cells as early as 1976, which included fast intracellular accumulation of swollen vesicles [226]. Abnormal accumulation of N-linked glycoproteins, which is a general feature of B deficiency in plants and in animals [22], seems to be also related to a failure of protein and glycan secretion through the endomembrane trafficking that could be related to glycolipids/glycoproteins interacting with B.
All this evidence points to a failure of the endomembrane mechanics in response to B deficiency, which is common to plants and animals. As mentioned, N-linked highmannose-rich glycoproteins (HMRG) accumulate during the development of Arabidopsis, legume nodules, and zebrafish embryos [22]. The HMR-glycan is added in the ER, and subsequent N-glycan processing involves trimming and substitution reactions to a more complex glycan in the ER and finally in Golgi [227]. Modifications of these N-linked HMRglycans are crucial for correct protein folding and serve as tags for quality control [228]. Failures in N-glycan trimming lead to an accumulation of HMRG proteins in the ER, and the subsequent activation of UPR (Unfolded Protein Response) and ERAD (ER-Associated Degradation) responses, growth arrest, and root swelling [229]. All of them are symptoms observed under B-deficiency, which suggests a failure of N-glycosylation, which is supported by the literature.
Indeed, B deficiency results in the abnormal glycosylation of particular proteins crucial for cell surface interactions regulating developmental events, as occurs with PsNLEC-1, a lectin involved in symbiosome development in pea nodules [230]. This protein is accumulated as a pre-N-linked glycosylated form in cytosolic vesicles and vacuole [107], being B-deficient nodules phenotypically similar to those developed in pea mutants with altered glycosylation [231]. Other glycoproteins components of the symbiosome and plasma membrane are not detected in B-deficient nodules [108]. Indirect evidence that the phenomenon of abnormal N-glycosylation is also common to animal cells is the fact that HMRGP also accumulates in zebrafish larvae when B-deficiency was induced just prior to 60h post-egg fertilization. At that time, an intense de novo synthesis of glycans targeted to regions of pectoral fins, jaws, and sensorial organs preceded its development [232].
Following fractionation of legume nodules and affinity chromatography with Amberlite IRA-743, several N-linked glycoproteins were isolated as potential B-ligands that were accumulated in B-deficient nodules [201]. Among them, it was found the already described Ps-NLEC-1 and, interestingly, several forms of BiP that were also isolated by boronate affinity chromatography from Arabidopsis and Zea mays microsomal fractions [75], which may be indicative of an activation of UPR due to underglycosylation leading to ER stress [233].
After one century of studies, more recent evidence claims to draw a scenario trying to find a model for common ubiquitous responses of living forms to B deficiency. To date, the most accepted primary function of B in plants is maintaining cell wall structure by the convincingly demonstrated cross-linking of RGII [64]. Sensing the perturbed pectin network, either both mechanistically or through pectin interacting cell wall receptors triggers a response to B deficiency that is conveyed to the nucleus, affecting several regulatory pathways that would explain all the described symptoms on growth and development [72]. But sensing at the cell wall site does not explain why B deficiency also has similar effects on animal development.
Trying to solve this puzzle with the aid of the whole research revisited for this review, an alternative model that can be complementary to the cell wall sensing mechanism is based on the potential interaction of B with membrane glycolipid and glycoproteins. The accumulation of HMRG in B-deficient plants and animals is surely reflecting underglycosylation due to the instability of ER membrane and ER stress, as a common symptom of B deficiency. The first immediate consequence is that vesicle trafficking is altered. Secondly, membrane or extracellular glycans involved in cell signalling, which are crucial for plant or animal organogenesis, are not accurately targeted or, if targeted, are abnormally glyco-sylated. Then, cell surface sensors are not functional, the formation of plasma membrane lipid rafts fails, the (wall)-plasma membrane-cytoskeleton continuum breaks, exosome and endocytic vesicle formation is uncontrolled, and finally, induction signalling from neighbouring cells 'short-circuits' leading to tumour-like early development or to defective organogenesis. As a feedback effect trying to restore membrane glycans and the continuum cell surface-cytoskeleton, enhanced synthesis of glycans and cytoskeletal proteins that keep accumulating in an ER stress environment occurs.
Going back to the primary roles of B in plants, the fact that a marginal amount of mRGII is found in B-deficient cells and that the addition of borate does not promote dimerization of previously synthesized mRGII [180] is consistent with this model. Indeed, dimerization exclusively occurs in de novo synthesized pectin or during the synthesis or prior to secretion, and it is facilitated by membrane glycoproteins and glycolipids [76,168]. Therefore, although RGII-crosslinking is undoubtedly a primary role of B in plants, it might be considered that it is not the primary function since correct dRGII-B formation seems to hold on the N-glycosylation machinery, which is apparently common in all eukaryotes. Supporting this, the Arabidopsis mur1-1 mutant, in which RG-II crosslinking was demonstrated essential for plant growth [64], shows defects in N-glycosylation accumulating HMGP despite a lack of fucose-complex N-glycans [234]. In this mutant, although boric acid or fucose partially restored growth, it did not reach wild-type rosette development [64], indicating the importance of N-glycosylation for dRGII-dimer formation and secretion.
To conclude this review, we proposed to further investigate the proposed role of B on N-glycosylation and secretion mechanisms and their relationship with the synthesis, complexing, secretion, and correct deposition of RGII. Arabidopsis reporter lines affected in different cell components' secretion have already been developed [235,236] and combining these tools with immunocytochemical approaches may be useful to validate and contextualize the failure of secretion of glycoproteins and/or dRGII-B under B deficiency. Also, it would be interesting to analyse the changes in N-glycosylation in the mur1-1 mutant line and other mutants affected by N-glycosylation in relation to B nutrition, which might help to identify and to confirm the existence of an in vivo linking of B to glycan residues that might be crucial for a proper N-glycosylation. In parallel, similar approaches on animal or animal cell lines would provide further knowledge of the common primary functions of B in N-glycosylation mechanisms in plants and animals.