Arabinogalactan proteins – becoming less mysterious


  • Barbara G. Pickard

    Corresponding author
    • Gladys Levis Allen Laboratory of Plant Sensory Physiology, Department of Biology, Washington University in St Louis, St Louis, MO, USA
    Search for more papers by this author

(tel +1 314 721 2080; email


As Lamport & Várnai point out in this issue of New Phytologist (pp. 58–64), the arabinogalactan proteins (AGPs) of the cell wall have long been characterized with the word ‘mysterious’. Study of large carbohydrate-rich polymers is of course generally fraught with technical difficulties, but efforts to understand the AGPs have been persistent because many members of the class do seem to regulate so many important aspects of growth, differentiation, and resistance to stresses. Now, it can be argued that such persistence is carrying us into an era in which high-technology tools will rapidly provide diverse kinds of data to generate new ideas for why the plant relies so extensively on AGPs. Genomic analyses will, of course, contribute heavily to such progress (e.g. Tan et al., 2012). But the contribution by Lamport & Várnai in this issue seems outstanding among the indicators foretelling how enhanced interdisciplinary coupling of technical capabilities and insightful biological modeling will broaden the base for understanding biophysical mechanisms.

‘Despite years of study, the characterization of the pumps and channels in the cell membrane remains challenging and incomplete.’

The physical evidence

Following elucidation by Tan et al. (2010) of the primary structure of the large arabinogalactan unit distributed repetitively and fairly uniformly along the c. 200-residue peptides of the important ‘classical’ (glypiated – hence membrane-attached) AGPs, Lamport & Várnai focused on the possibility for binding Ca2+ into a negatively charged pocket formed by a pair of glucuronic acids in each moiety. It is clear since the work of Arif & Newman (1993) that Ca2+ is bound by wall polymers at relatively high pH and released at lower values. The specific binding polymer and its mode of interaction with Ca2+ is of general interest in that Ca2+ is a ‘universal integrator’ coordinating a vast array of biochemical reactions, some in the wall per se but vastly more in the cytoplasm, which it enters if it encounters open Ca2+ channels that permit it to pass inward down its electrochemical gradient.

Performing dynamic molecular simulations, Lamport & Várnai found that the divalent calcium ion bound tightly in the pocket, and visualized how the ionic attraction rigidifies the arabinogalactan: dramatic loss of vibrational movement was seen with Ca2+ as compared with monovalent Na+. They then evidenced massive ‘real life’ binding of Ca2+ in solution by saturating extracted AGP with Ca2+ and titrating the resulting complex with acid, showing that each glucuronic acid pair could hold one Ca2+. There are two to three dozen such binding options in a single glycoprotein (of which there are several kinds), and a large fraction of their Ca2+ is released when a pH of 3 or even lower is attained. Up to now the major Ca2+ reservoir of the wall has been considered to be pectins, which do indeed reversibly bind considerable amounts. Adjusting pH of pectin solutions and comparing binding coefficients, however, AGPs are shown to exchange Ca2+ much more freely than do pectins.

The calcium capacitor: key to plant cell oscillations?

Proton pumps (H+-ATPases) and calcium channels have long been considered to act in rhythmic feedbacks to control concentrations of the ions inside and outside the plasma membrane. Lamport & Várnai propose that charging and discharging of surface-bound AGP with Ca2+ plays an important and indeed necessary role in such feedbacks. To popularize the concept of readily reversible Ca2+ storage by AGP, they establish the simple term ‘calcium capacitor’. It is easy to believe that the capacitor is loaded with Ca2+ when the periplasm is at pH 5, a widely quoted average steady-state wall value, and they offer reasons to believe that pH values low enough to release Ca2+ are in fact realized in small volumes just outside the cell membrane during growth oscillations. Given this persuasive evidence, they put together a heuristic model for how the capacitor might interact with pumps, channels and exocytotic vesicles to create the oscillations. (Parenthetically, defining the AGP-based capacitor brings to mind the phenomenology of pH-dependent concentration of Ca2+ close to the carboxyl head-groups of the plasma membrane – see Abraham-Shrauner, 1975.)

All such models must be taken with a grain of salt. Despite years of study, the characterization of the pumps and channels in the cell membrane remains challenging and incomplete. The model of Lamport & Várnai need not precisely identify the specific molecular players in order to argue the possibilities for roles of such entities in capacitor-promoted oscillations. The model is useful even if it only puts placeholders in a feedback model. At a minimum it can invigorate fresh thinking about the H+ and Ca2+ exchanges manifesting as oscillations. The overall proposal provides an extraordinarily interesting frame within which very complex data sets about the exchanges can be viewed and new data and models created.

The early steps in the Lamport & Várnai model are compared (Fig. 1, left) with those in one such alternative model (Fig. 1, right). At the left, there are two postulated stretch sensing elements. One is an H+ pump; it releases Ca2+ from the capacitor. The second is a Ca2+ channel, which opens in synchrony with the H+ pump and admits the released Ca2+ to the cytosol. Once Ca2+ enters the cytosol, it can do all the things that it famously does there – which can include shipment of new materials into the wall (not drawn). But meanwhile the acid causes the wall to stretch. This feeds forward to open more H+ pumps: can this produce the runaway feedback illustrated in the figure?

Figure 1.

Modeling contribution of the arabinogalactan protein (AGP) calcium capacitor in growth oscillations. Left: runaway feedback with postulated stretch-activated proton pumps plus calcium channels? Right: need-responsive cycling with stretch sensed exclusively by mechanosensitive calcium channels? Inset: the regulation of one mechanosensitive cation channel by pH, from Ding et al. (1993), with permission. At this stage in our knowledge it is premature to model except to stimulate a new generation of modeling!

Moreover, while further identification and characterization of proton pumps and Ca2+ channels will be required for evaluating the generality of the paired-channel scheme, it is notable that activity of the suggested mechanosensory channel (Ding et al., 1993; Pickard & Fujiki, 2005; cf. Kurusu et al., 2012) declines markedly as pH drops (Ding et al., 1993). If Ca2+ entry shuts down, this might terminate a network of Ca2+-stimulated actions required for growth, including release of vesicles with new wall precursors. Possible participation by other channels (cf. Dutta & Robinson, 2004; Kurusu et al., 2012) could be subject to experimentation even on the basis of presently available information. Is it a problem for the scheme on the left that no proton pump has been rigorously proven to have mechanosensory function? (Some such pumps are, of course, activated downstream of Ca2+ entry.) And will the scheme be susceptible to the runaway feedforward loop that is sketched?

In the sketch on the right (Fig. 1), the single stretch sensitive element is again the Ca2+-permeable channel alluded to by Lamport & Várnai. When opened by wall stretch transmitted to the membrane, entering Ca2+ promotes action by the H+ pump – which elevates release of Ca2+ from the capacitor with the result of more Ca2+ entry. But then the increase in acid inhibits channel action to close the feedback loop. The H+ diffusing further into the wall leads to stretch, which along with various downstream actions of Ca2+ starts the cycle again. Thus the action is controlled by need. In simple form the model was proposed upon discovery of the mechanosensory channel in question – without knowledge of the AGP capacitor. It is greatly improved by the capacitor!

These two schemes illustrate first how the existence of the capacitor inspires fresh thought about how mechanosensing and oscillations can work. However, they also provide fresh impetus to the already active study of the pumps and channels that must be involved in a membrane-centered oscillatory feedback, as well as the returns of the ions that must occur for perpetuated activity (Bose et al., 2011) and the Ca2+-stimulated participation of vesicles ultimately essential for adding mass to the wall.

Does the capacitor model account for all fast calcium oscillations?

Perhaps the expectation of Lamport & Várnai that cells of plants differ from those of animals in needing no Ca2+ compartmentation just inside the cell membrane – that their model can account for all the relatively rapid Ca2+-based oscillations that the plant cell makes – is overly enthusiastic. Masaaki Fujiki when in my laboratory years ago (cf. Pickard & Fujiki, 2005) revealed Ca2+ oscillations arising from the cortex. The experiments are unfortunately unpublished but should be known so that others can expand on them. Touching an onion epidermis cell containing cytosolic cameleon with a fine probe releases a pulse of Ca2+ in the adjacent cytoplasm. The pulse is quickly and cleanly swept away by the surprisingly undiminished rotary cytoplasmic streaming, disappearing at the far end of the huge cell. Without discernable return of the initial pulse and without another touch, a new pulse suddenly appears in the original submembrane volume and is swept away again in the same manner. This happens a third time during the c. 15-min experiments. Thus, after the submembrane compartment is indirectly stimulated by one quick transient mechanostimulated Ca2+ influx, it oscillates on its own.

The calcium capacitor model as general inspiration

Not only does it seem that there may be multiple loci generating Ca2+-based oscillations, but it seems likely that the capacitor may well help control many regulatory actions of Ca2+ besides the oscillations. These have been hard to model in detail based on data generated with earlier technology; but diverse new tools are at hand, such as high-resolution atomic force microscopy and various high-resolution forms of confocal microscopy. With increasing combination of biological insight with high-technology tools, one expects further roles for the AGP capacitor to emerge in the near future. Reversible Ca2+ binding by membrane-associated AGPs likely features in a range of diverse structural and regulatory activities of these now somewhat less mysterious periplasmic glycoproteins.


Thanks to Professor Eric McLamore (University of Florida, USA) for critical reading and for preparing the figure. Support is appreciated from Grant 1102803 from the Civil, Mechanical and Manufacturing Innovation Division of the National Science Foundation and from Gladys Levis Allen.