Arabinogalactan protein–rare earth element complexes activate plant endocytosis

Significance

Due to the vital role for eukaryotes, clathrin-mediated endocytosis (CME) has attracted increasing attention, but great challenges still stand. Here, by overcoming the biggest challenge in CME research, we visualized the dynamic of extracellular cargo receptor protein by using stimulated emission depletion microscopy. We identified an unconventional mechanism for extracellular cargoes, here specifically rare earth elements (REEs). We showed that arabinogalactan proteins (AGPs) act as extracellular cargo receptors and move across the plasma membrane to initiate endocytosis. REEs promote the cross-membrane translocation of its extracellular cargo receptor AGPs to activate their endocytosis. Our data thus provide insights into the mechanism for the activation of CME, the biological role of AGPs, and the cellular mechanisms of REE actions in plants.

Abstract

Endocytosis is essential to all eukaryotes, but how cargoes are selected for internalization remains poorly characterized. Extracellular cargoes are thought to be selected by transmembrane receptors that bind intracellular adaptors proteins to initiate endocytosis. Here, we report a mechanism for clathrin-mediated endocytosis (CME) of extracellular lanthanum [La(III)] cargoes, which requires extracellular arabinogalactan proteins (AGPs) that are anchored on the outer face of the plasma membrane. AGPs were colocalized with La(III) on the cell surface and in La(III)-induced endocytic vesicles in Arabidopsis leaf cells. Superresolution imaging showed that La(III) triggered AGP movement across the plasma membrane. AGPs were then colocalized and physically associated with the μ subunit of the intracellular adaptor protein 2 (AP2) complexes. The AGP-AP2 interaction was independent of CME, whereas AGP’s internalization required CME and AP2. Moreover, we show that AGP-dependent endocytosis in the presence of La(III) also occurred in human cells. These findings indicate that extracellular AGPs act as conserved CME cargo receptors, thus challenging the current paradigm about endocytosis of extracellular cargoes.

Endocytosis, including clathrin-mediated endocytosis (CME), is a fundamental cellular process in plants, animals, and microorganisms (1, 2). By internalizing extracellular cargoes and membrane-integral or -associated proteins, CME plays an essential role in various cellular processes, such as cell signaling, cell-polarity formation, cell-fate determination, cell division, and cell movement (1, 2). Consequently, the mechanisms for cargo recognitions and CME machinery and processes have been extensively studied (1–4). A suite of adaptor proteins have been shown to be involved in the recognition of cargoes or cargo receptors at the cytoplasmic side of the plasma membrane (PM). One of the best-characterized and conserved adaptor proteins belongs to adaptor protein 2 (AP2) complexes consisting of 4 subunits, α, β, σ, and µ subunits (2–4). Upon activation by protein–protein interaction or modification such as phosphorylation, membrane cargo proteins are recognized by AP2, which then recruits clathrin subunit proteins for clathrin coat assembly (2–4). It is well established that extracellular cargoes are recognized by transmembrane receptors, whose cytoplasmic domains interact with CME adaptor proteins to initiate cargo endocytosis (2–4). Despite these advances, major gaps remain in our understanding of CME, particularly regarding the mechanisms underpinning the recognition of cargoes and the regulation of CME. Furthermore, the current research on CME faces major challenges, such as in the identification and visualization of cargo receptor proteins and their interaction with cargoes in vivo (1–4). By visualizing the dynamic of cargo or its receptor protein using stimulated emission depletion (STED) microscopy, and investigating the mechanism for the endocytosis of rare earth elements (REEs), e.g., lanthanum [La(III)], we have identified a mechanism for the activation of CME and cargo recognition.

The REEs in the periodic table of elements comprise 15 lanthanide elements plus scandium and yttrium, which have similar properties in atomic radius and charge (5) and are considered to be nonessential elements of living organisms. REEs are extensively used in agriculture, industry, national defense, environmental protection, medicine, etc. (6, 7). For decades, REEs have been ingredients of fertilizers for the improvement of plant growth and crop yields mainly via foliage spraying, but the mechanisms for REEs to enter plant cells and their action mechanisms to promote plant growth remain poorly characterized. On the other hand, the widespread applications of REEs have also resulted in the massive accumulation of REEs in the global environment (including soil, water, and atmosphere) and living organisms at an unprecedented speed (8–10). Therefore, the pollution of REEs is rapidly emerging as a universal threat to ecological integrity and function, as well as human health (11), highlighting the urgent need for establishing guidelines to limit the concentration of REEs in the ecosystem. To accomplish this goal, it is imperative that we have a clear qualitative and quantitative examination about how REEs are absorbed by and act on plants, especially in leaves, which are directly sprayed with REEs in agricultural application. Recently, by using interdisciplinary techniques, including electron microscopic autoradiography (EMARG) of radioactive La, cerium (Ce), and terbium (Tb) [¹⁴⁰La(III), ¹⁴¹Ce(III), and ¹⁶⁰Tb(III)], we directly observed the life cycle of REEs in plants and surprisingly found that REEs, firstly anchored on the PM in the form of the nanoscale particles, activated endocytosis in leaf cells and then entered leaf cells with the help of some biomolecules (12, 13). Our findings have attracted much attention from a wide range of disciplines, e.g., medicine (14), agriculture (15), ecology (16), and cell biology (17, 18). At the same time, new questions have been raised about the mechanisms for the REE regulation of endocytosis. How do REEs cause the activation of endocytosis, what are the receptors for REEs (the primary target of REEs) to initiate the endocytosis, and how do these receptors interact with adaptors and then activate endocytosis in leaf cells?

In this study, by employing a double-imaging method involving isotopically labeled ¹⁴⁰La(III) and immunogold-labeled potential La(III) receptors, we show that arabinogalactan proteins (AGPs) were colocalized with La(III) on the cell surface and in La(III)-induced endocytic vesicles in Arabidopsis leaf cells. Superresolution imaging with STED microscopy, combined with immunocytochemical, biochemical, and genetic analyses, show that La(III) triggered the movement of AGPs from the outer face of the PM across the PM into the cytoplasmic face to interact with the μ subunit of the intracellular AP2, a CME adaptor complex. AGP-mediated La(III) endocytosis was also observed in human cells. These findings thus establish AGPs as the primary conserved targets of La(III) to initiate CME of La(III) and challenge the current paradigm about the CME of extracellular cargoes. Furthermore, our findings provide insight into the mechanisms by which REEs impact biological processes, a subject that has puzzled scientists for over a century.

Results and Discussion

AGPs Play a Critical Role in Mediating La(III)-Induced CME in Leaf Cells.

In our previous studies, we found that REEs (La, Ce, and Tb) could activate endocytosis in horseradish leaf cells (12, 13). We found that La(III) treatment can activate endocytosis in Arabidopsis (wild type, Col-0) leaves as well; therefore, in this study we choose La(III) as a representative REE to study the cellular mechanism underlying REEs-activated endocytosis in the model plant Arabidopsis. To identify the primary target of REEs, we use a double-imaging method that involves isotopically labeled ¹⁴⁰La(III) and immunogold-labeled potential targets. Based on the EMARG observations that La(III) treatment can activate endocytosis (Fig. 1A), the expression of several genes encoding AGPs (SI Appendix, Fig. S1) and the postulation that AGPs and La(III) might form complexes (5), we asked whether AGPs could be the potential REE targets. We colabeled La(III) with ¹⁴⁰La(III) and AGPs with immunogold in Col-0 leaf cells and visualized them using a transmission electron microscope equipped with an energy-dispersive spectrometer (TEM-EDS). By using this double-imaging method, we observed that AGPs were distributed on the cell surface and in endocytic compartments in the La(III)-treated Col-0 leaf cells together with La(III), in which the outlines of the endocytic compartments were drawn by AGPs labeled by immunogold particles (Fig. 1B and SI Appendix, Table S1). The observed sizes of the endocytic compartments were in the range of 100 to 200 nm (Fig. 1B), which were consistent with those of the EMARG images (Fig. 1A). The results were also consistent with the previously published size ranges of CME determined by TEM that has a resolution down to the nanometer level and thus is commonly used to determine the size of CME vesicles in living organisms (1, 2, 19). Meanwhile, compared with untreated cells, La(III) treatment greatly increased the labeling of AGPs on the PM (especially the invaginated PM domains) apparently undergoing endocytosis (Fig. 1 B and D and SI Appendix, Table S1), suggesting that the colocalization of La(III) and AGPs on the PM may induce AGP internalization.

Fig. 1.

AGPs are the targets of La(III) that activates endocytosis in Arabidopsis leaf cells. (A) The representative EMARG images of Col-0 leaf cells treated with 0, 30, or 80 μM La(III) containing radioactive ¹⁴⁰La(III) for 12 h from 3 biological replicates. White points are La(III). Bar = 100 nm. (B, top row) The representative TEM images of Col-0 leaf cells treated with 0, 30, and 80 μM La(III) and successively immunostained with anti-AGP antibody (JIM13) and gold-labeled anti-JIM13 antibody from 3 biological replicates. Irregular small dots indicated with white arrows are La(III). Large dark dots display AGPs labeled with immunogold. (Scale bar, 100 nm.) TEM allowed unambiguous observation of the distribution sites for both La(III) and AGPs. (B, middle row) Enlarged view showing more clear morphologies from the area marked with the black square in the top row. (Scale bar, 20 nm.) The TEM images show colocalization of AGPs and La(III) on the PM. (B, bottom row) The contents of La and gold (Au) labeling AGPs were determined by EDS. Among the possible sites where La(III) and AGPs were colocated on the PM, such as regions marked with red squares in the middle row, we randomly selected 10 sites to detect the contents of La and Au. The contents of La and Au shown in this row were the average values of the detected La and Au contents in the 10 sites. EDS detected La and Au from their same distribution sites, further showing that La(III) and AGPs were colocalized on the PM and might form their complexes on the PM. (C) The representative TEM images of β-glucosyl Yariv reagent-treated Col-0 leaf cells treated with 0, 30, or 80 µM La(III) for 12 h and successively immunostained with JIM13 and gold-labeled anti-JIM13 antibody from 3 biological replicates. Large dark dots display AGPs labeled with immunogold. (Scale bar, 100 nm.) (D) Quantitative analysis of gold particles on the PM in Col-0 leaf cells treated without or with β-glucosyl Yariv reagent. Data are represented as means ± SEM. *P < 0.05 (2-tailed unpaired t test). Mean values were obtained from 15 images taken from 3 independent experiments. Chl, chloroplast; CW, cell wall.

AGPs belong to a large family of glycoproteins ubiquitously found in plants (20, 21). Normally, AGPs are tethered to the outer leaflet of the PM by a glycosylphosphatidylinositol (GPI) anchor and may cross-link with some components of the cell wall (GPI-anchored AGPs) in plant cells (20, 21). AGPs play crucial roles in cell growth, embryo patterning, reproductive development, postembryonic patterning, signal transduction, and pollen tube growth in plants (20, 21). To date, no evidence is available for a role of AGPs in the regulation of endocytosis in plants, animals, and microorganisms. We thus asked whether AGPs could initiate the endocytosis in the presence of La(III) in plant leaves. The high functional redundancy is anticipated for AGPs due to the large AGP gene family (i.e., 85 AGP genes in the Arabidopsis genome) encoding AGPs with the incredible heterogeneity of backbone and carbohydrates (20, 21). To circumvent the difficulties in genetic approaches to analyze the redundant functions, we suppressed the functions of AGPs by using β-glucosyl Yariv reagent, a widely used inhibitor of AGPs (22, 23). Endocytosis was visualized by staining the cell membrane with N-(3-triethylammoniumpropyl)-4-(4-diethylaminophenylhexatrienyl) (FM4-64), a membrane-impermeable fluorescent styryl dye (24, 25), and GFP-tagged components of the CME machinery, e.g., GFP-tagged clathrin light chain (CLC1-GFP) (26). CLC1-GFP and FM4-64 dye colabeled both of the PM and the La(III)-induced endomembrane compartment visualized by confocal laser-scanning microscopy (CLSM), by which the dynamic process of endocytosis could be observed, although CLSM cannot determine the sizes of endocytic vesicles due to its technical limitation (e.g., diffraction limit) (27). The endomembrane compartment was only observed when leaf cells were treated with La(III) (Fig. 2A). Time-lapse imaging using spinning-disk confocal microscopy showed that La(III) treatment induced the budding of CLC1-GFP labeled large vesicles from the PM (Movie S1). These observations suggest that La(III) treatment may induce CME, resulting in the formation of the endomembrane compartments. These compartments were found in ∼20% of leaf cells in the presence of 30 μM La(III), with an average of 5 to 7 compartments (red circles) per cell (Fig. 2B, Top), and in ∼80% of leaf cells, with 3 to 17 compartments per cell in the presence of 80 μM La(III) (Fig. 2B, Top). Interestingly, treatments with the β-glucosyl Yariv reagent drastically blocked La(III)-activated endocytosis (Fig. 2B, below). Only 1 to 2 endocytic compartments in each cell were observed in cells treated with β-glucosyl Yariv reagent (Fig. 2B, below). These results suggest that AGPs play a critical role in initiating CME in leaf cells induced by certain concentrations of La(III). Moreover, larger vesicles were observed in the CLSM images than those in the EMARG and TEM images (Fig. 1 A and B) due to the fusion of the endocytic vesicles in CLSM images and lower spatial resolution of CLSM (19).

Fig. 2.

AGPs initiate the CME of Arabidopsis leaf cells after treatment La(III). (A) The representative CLSM images of Col-0 leaf cells expressing CLC1-GFP treated with 0, 30, or 80 µM La(III) for 12 h and stained with FM4-64 from 3 biological replicates. Induced endocytic vesicles are indicated with white arrowheads. (Scale bar, 10 µm.) The further dynamic results from time-lapse imaging are shown in Movie S1. (B) The representative CLSM images of Col-0 leaf cells treated with 0, 30, or 80 µM La(III) for 12 h and stained with FM4-64. Induced endocytic vesicles are indicated with white arrowheads. The leaf cells were treated without or with β-glucosyl Yariv reagent before La(III) treatment. (Scale bar, 10 µm.) (C and D) The representative CLSM images of the leaf cells in Col-0 and DEX-inducible Venus–CANTH lines treated with 0, 30, or 80 µM La(III) for 12 h from 3 biological replicates. (Scale bar, 6 µm.) The leaves of Col-0 and DEX-inducible Venus–CANTH lines treated with 0, 30, or 80 µM La(III) for 12 h were stained with FM4-64 (red) and JIM13 coupled with secondary antibody (rabbit anti-rat IgG heavy chain & light chain (H&L) [FITC]) (green) and observed under CLSM for evaluation of AGP’s CME. (E) The frequency of endocytosis and the amount of endocytic vesicles in Col-0 and DEX-inducible Venus–CANTH lines treated with 0, 30, or 80 µM La(III) for 12 h. A representative result of 3 independent experiments is presented (n ≥ 10). Data are represented as means ± SEM, *P < 0.05 (2-tailed unpaired t test).

AGPs and La(III) Act Together to Trigger CME in Leaf Cells.

The colocalization of La(III) with AGPs on the PM and the requirement of AGPs for La(III)-triggered endocytosis suggest that AGPs act on the early events of the endocytosis, possibly as direct targets or cocargoes of La(III). If so, AGPs would be internalized by endocytosis together with La(III). To visualize AGP endocytosis, AGPs were successively immunostained with anti-AGP antibody (JIM13) and fluorescence-labeled anti-JIM13 antibody in cells costained with FM4-64 (Fig. 2 C and D). This method allowed the simultaneous visualization of endocytic events and AGP dynamics using CLSM. Immunofluorescence labeling of AGPs showed that La(III) treatment dramatically induced the accumulation of AGPs along the PM as punctates and in endocytic vesicles (Fig. 2 C and E); these effects were stronger with 80 μM La(III) treatment compared with 30 μM La(III) (Fig. 2 C and E). These results are consistent with the results from immunogold labeling of AGPs (Fig. 1B). Furthermore, treatment with β-glucosyl Yariv reagent blocked the accumulation of AGPs on the PM (Fig. 1 C and D). The green fluorescence that labeled punctate AGPs was superimposed on the FM4-64 red fluorescence that labeled both of the PM and the endosomal compartment (Fig. 2C, Right columns in middle and bottom rows), indicating that AGPs can serve as a marker for endocytosis in these cells. This result was consistent with that from the TEM for immunogold-labeled AGPs (Fig. 1B). In addition, after 30 μM La(III) treatment, all of the endosomal compartments showed AGP signals (Fig. 2C, Right columns in middle row). After 80 μM La(III) treatment, most of endosomal compartments showed AGP signals, while some endosomal compartments showed no AGP signals (Fig. 2C, Right columns in bottom row). These vesicles without AGP signal might result from other La(III) receptors, which awaits further investigation. Taken together, our results suggest that La(III) and AGPs act together to trigger endocytosis in leaf cells.

Because La(III)-induced endosomal compartments costained with CLC1-GFP, we hypothesize that the AGP-dependent internalization of La(III) is mediated by CME, which is known to be the main endocytic pathway in plants (28–30). We further tested whether CME is involved in La(III)-activated AGP internalization. To inhibit CME, we generated transgenic lines harboring a dexamethasone (DEX)-inducible Venus–CANTH construct, in which the C-terminal region of AtAP180 proteins (CANTH) was fused with the Venus protein (31–33). The conserved AP180 is essential for the early stage of CME (31, 33); thus, overexpression of CANTH is known to specifically block CME (31, 32). DEX-induced overexpression of Venus–CANTH blocked the internalization of AGPs but had no effect on their accumulation on the PM induced by La(III) treatment (Fig. 2D). Furthermore, no endocytic compartments were observed in La(III)-treated Venus–CANTH cells (Fig. 2D), indicating that Venus–CANTH effectively blocked La(III)-induced CME in these cells. The induction of Venus–CANTH also abolished La(III)-triggered CME in leaf cells without AGP antibody (JIM13) staining, excluding the possibility that this antibody could interference with La(III)-triggered CME (SI Appendix, Fig. S2). The results further demonstrated that AGPs are involved in the activation of CME, which is in agreement with the inhibition of La(III)-triggered endocytosis by β-glucosyl Yariv reagent (Fig. 2B).

La(III)-Triggered AGP Movement across the PM during the Initiation of CME.

Because AGPs are secretory proteins localized to the outer leaflet of the PM via a GPI anchor (20, 21), they may act as cocargoes or cargo receptors to initiate CME. Based on the existing models of CME in plants and other systems (28–30), the extracellular cargoes must be recognized by a transmembrane cargo receptor, whose intracellular domain interacts with the intracellular adaptor proteins. To understand how AGPs may work as cocargoes or cargo receptors for CME, we used the superresolution STED microscopy to visualize the dynamic behaviors of AGPs during La(III)-induced CME (27, 34, 35). To overcome the technical challenges presented by plant cells such as the cell wall (>250 nm thickness) for superresolution imaging (36), we performed the STED imaging of AGPs by using the Arabidopsis protoplasts. La(III)-treated protoplasts were incubated with (secondary antibody)-labeled anti-AGP antibody, and the location of AGPs in protoplast was shown by observing the fluorescence signals of the secondary antibody. Then, time-lapse imaging of stained AGPs was conducted. As shown in Fig. 3 and Movie S2, AGPs were localized on the surface of the protoplasts and rarely crossed the PM in the control cells that were not treated with La(III). In La(III)-treated Col-0 protoplasts, PM-associated AGPs were dramatically increased, and the majority of them were found to move across the membrane and enter the cytoplasm (Fig. 3A, Movie S2, and SI Appendix, Fig. S3). It appears that AGPs were spanning the PM on their way into the cytoplasm (Fig. 3A, Movie S2, and SI Appendix, Fig. S3). Intriguingly, in the DEX-induced protoplasts from the Venus–CANTH lines, La(III) treatment also induced AGPs to cross the PM (or inserted into the membrane), but most of them did not completely enter the cytoplasm (Fig. 3B, Movie S2, and SI Appendix, Fig. S3) and were apparently stuck in the PM (Fig. 3B, Movie S2, and SI Appendix, Fig. S3). Quantitative analysis (SI Appendix, Fig. S3 and Table S2) showed that ∼20% of leaf cells in La(III)-treated Col-0 displayed AGP movement across the PM, and in these cells, ∼90% AGPs on the cell surface crossed the PM. In leaf cells of Venus–CANTH lines treated with La(III), almost all AGPs on the cell surface spanned the PM but failed to enter the cytoplasm.

Fig. 3.

AGPs migrate across the PM after La(III) treatment. (A and B) The leaves of Col-0 and DEX-inducible Venus–CANTH lines were treated with 0 or 30 µM La(III). Twelve hours later, the protoplasts were isolated from the respective leaves, stained with JIM13 and secondary antibody (goat anti-rat IgG [Alexa Fluor 555]), and observed under STED microscopy (∼108 s) to visualize the dynamic behavior of AGPs during La(III)-induced CME. The images presented in A and B are a representative time series of images taken from Movie S2 (n = 10 frames; t = 108 s), which showed the time-lapse imaging of AGPs moving across the PM in the leaf cells of Col-0 and DEX-inducible Venus–CANTH lines at the indicated times after 12-h treatment with 0 or 30 µM La(III). (Scale bar, 2.5 µm.)

Because the specific antibody of AGPs (JIM13) may not interfere with La(III)-triggered CME (SI Appendix, Fig. S2), these observations suggest that La(III) induced AGPs to move across the PM, but these AGPs in the PM were not internalized if the assembly of clathrin coats was blocked by Venus–CANTH. The behavior of AGPs translocating across the PM drives the transmembrane transport and internalization of primary and secondary antibodies bound to AGPs. These observations raised a very exciting question: how can AGPs move across the PM without the assembly of clathrin coats, i.e., without the membrane invagination after the clathrin proteins are recruited to the inner face of the membrane? This unexpected discovery apparently contradicts with our conventional knowledge about AGPs (20, 21). Given GPI anchoring of AGPs to the outer leaflet of the PM, the most likely interpretation of these findings is that La(III) induced the conformational changes in AGPs, causing AGPs to flip-flop to the inner side of the PM.

La(III) Induced the Physical Association of AGPs with the μ Subunit of the Intracellular AP2 to Initiate CME.

If AGPs move across the PM, they could recruit adaptor proteins to initiate the assembly of clathrin coat proteins (2, 3). To date, the primary adaptor proteins in plant CME are the AP2 family proteins; other adaptor proteins, such as ENTH/ANTH family proteins, and accessory proteins, such as TPLATE complex (TPC), have also been reported (2–4). We first costained AGPs (green) and the µ subunit of intracellular AP2 (blue) in both Col-0 and DEX-inducible Venus–CANTH lines using anti-AGP and anti-µ antibodies conjugated to fluorescein isothiocyanate (FITC) and Alexa Fluor405, respectively. In the absence of La(III), only small amounts of AGPs and AP2 were localized on the cell surface, while they exhibit distinct localization sites in Col-0 leaf cells (Fig. 4A, Right image in first row). La(III) treatment induced the expression of AGPs and AP2, respectively (Fig. 1B and SI Appendix, Fig. S4). Consequently, La(III) treatment induced the colocalization of AGPs and AP2 on both the membrane and the endocytic compartments (Fig. 4A, second row). In the DEX-treated Venus–CANTH lines, AGPs and AP2 were colocalized only on the PM of leaf cells in the presence of La(III) (Fig. 4B). These results demonstrate that La(III) induced its codistribution with AGPs and AP2 in a physical proximity on the PM and that their subsequent internalization is dependent upon clathrin assembly.

Fig. 4.

La(III) triggers AGPs endocytosis by interacting with AP2 in leaf cells. (A) The representative CLSM images of coimmunostained AGPs (green) and AP2 (blue) in the cells of Col-0 leaves treated with 0 or 30 µM La(III) for 12 h from 3 biological replicates. The leaves after 12-h treatment with 0 or 30 µM La(III) were stained with JIM13 coupled with secondary antibody (rabbit anti-rat IgG H&L [FITC]) and monoclonal rabbit anti-AP2M1 antibody (EP2695Y) coupled with donkey anti-rabbit IgG H&L (Alexa Fluor 405), and observed under CLSM for evaluation of AGP’s CME. (Scale bar, 6 µm.) (B) The representative CLSM images of coimmunostained AGPs (green) and AP2 (blue) in the leaf cells of DEX-inducible Venus–CANTH lines treated with 30 µM La(III) for 12 h. (Scale bar, 6 µm.) (C) CLSM images of the leaf cells in the mutant of AP2 μ subunit (ap2μ-1, SALK-083693C) and its complementated line treated with 0 or 30 µM La(III) for 12 h and stained with FM4-64. Induced endocytic vesicles are shown with white arrowheads. The leaf cells of ap2μ-1 were treated without or with β-glucosyl Yariv reagent before La(III) treatment. (Scale bar, 10 µm.) (D) AGPs directly interacted with AP2 as tested by co-IP from Col-0, DEX-inducible Venus–CANTH lines and ap2μ-1, and Western blot analysis. In this assay, the anti-AP2M1 antibody (EP2695Y) that specifically binds to the μ subunit of AP2 was used to precipitate AP2. The + and − symbols mean that protein extracts were incubated with and without anti-AP2M1 antibody, respectively, before being incubated with protein G Sepharose beads. Input factions were the extracted total proteins, and co-IP fractions were the immunoprecipitated AGPs and AP2 in the extracted proteins. (E) The content of La(III) in the immunoprecipitates (by anti-AP2M1 antibody) obtained from the co-IP assay of Col-0 leaves treated with 0 or 30 µM La(III) for 12 h. *P < 0.05 (2-tailed unpaired t test).

Among the AP2 subunits, cargoes commonly bind directly to the μ subunit of AP2, thus initiating the CME (2). To determine whether the μ subunit of AP2 is involved in AGP-mediated initiation of CME in leaf cells, we examined endocytosis in the mutant of AP2 μ subunit (ap2μ-1) with or without β-glucosyl Yariv reagent treatments in the absence or presence of La(III). We found that La(III)/AGP-initiated endocytosis was greatly diminished in the knockout (transfer DNA insertional mutant) of the μ subunit (Fig. 4C). Moreover, the complementation of the μ subunit of AP2 recovered La(III)-activated endocytosis (Fig. 4C). Thus, AGPs may initiate the endocytosis via their direct interaction with the μ subunit of AP2. Indeed, coimmunoprecipitation (co-IP) assay data showed that AGPs and AP2 physically interacted in both Col-0 and DEX-induced Venus–CANTH lines but only after La(III) treatment (Fig. 4D). The distinct band for AGPs was not observed in coimmunoprecipitated proteins from ap2μ-1 mutant after La(III) treatment (Fig. 4D), which strongly supports the conclusion that AGPs and AP2 physically interacted in Col-0 after La(III) treatment. Moreover, by bioinformatics prediction and molecular dynamic simulation, we found that some AGPs expressed in leaves contain the recognition site of AP2 (SI Appendix, Table S3) and might directly interact with AP2 to form the stable complexes (SI Appendix, Fig. S5), which awaits further investigation. We next used anti-AP2 antibody to immunoprecipitate the active AGPs that formed complexes with AP2 after La(III) treatment and determined the content of La(III) in the active AGP molecules by using inductively coupled plasma mass spectrometry. We found that La(III) was present in the active AGP molecules, further demonstrating the formation of AGP-La(III) complexes (Fig. 4E). Taken together, our results demonstrate that La(III) treatment induced the formation of the La(III)-AGP-AP2 complexes and their subsequent internalization by CME.

AGP-Dependent Endocytosis in the Presence of La(III) Also Occurred in Human Cells.

REEs are ubiquitous in atmospheric particulate matters, especially in developing countries (9), and are found to accumulate in human organs and cells in the form of nanoparticles to a harmful level (11, 37). The residual REEs (such as gadolinium) in human body results in nephrogenic systemic fibrosis (37). AGPs in human are obtained through diet, mostly plant-based diet (38, 39), bind or adhere to the cell surface (40, 41), and perform important functions such as reducing the cholesterol in the blood (42), regulating immunity (43), and serving as targeted-drug carriers to deliver drug to target organs rapidly (39). Therefore, we asked whether La(III) induced the activation of AGP-dependent endocytosis in human cells. To test this, we used human embryonic kidney 293 cells (HEK293) and observed that the endocytosis was indeed activated in HEK293 in the presence of La(III), and the activation was more evident when the concentrations of La(III) were increased (SI Appendix, Fig. S6A). Treatments with β-glucosyl Yariv reagent drastically blocked La(III)-activated endocytosis (SI Appendix, Fig. S6B). In mammalian cells, except for AGPs, no glycoproteins have been reported that can be inhibited by β-glucosyl Yariv (39, 42, 43), while fetal bovine serum containing AGPs (44, 45) was added in the culture medium (46). So, it seems likely that AGPs also activate and regulate REEs’ endocytosis in human cells. The co-IP assay data showed that AGPs and AP2 physically interacted in HEK293 cells after La(III) treatment (SI Appendix, Fig. S6C), further demonstrating that AGP-mediated endocytosis also occurred in human cells. Thus, AGP-dependent endocytosis in the presence of La(III) appears to be a universal mechanism. Moreover, we did not observe the big ring structures formed in HEK293 cells upon La(III) treatments. We reason that the structures of endocytic vesicles may differ across species or even in different organs of the same organism (1, 47, 48).

To summarize, we found that extracellular AGPs, upon induction by La(III), move across the PM and physically interact with the AP2 complexes to initiate the endocytosis of AGPs/La(III) (Figs. 1–5 and SI Appendix, Figs. S1–S8 and Tables S1–S4). Our findings demonstrate an unprecedented mechanism for the activation of CME by proposing that AGPs serve as unconventional receptors for extracellular La(III) cargo to initiate CME (Fig. 5). Based on our previous studies that REEs have similar life cycles and cellular behaviors in plant cells (12, 13), and the chemical principle that the REEs in the periodic table of elements have similar physical and chemical properties (5), we suggest that our conclusions in this study may represent the general action mechanisms for all REEs in plant cells (Fig. 5). This mechanism for CME activation seem to occur across the eukaryotic kingdoms. Thus, our findings support a paradigm for the activation of CME by extracellular cargoes, in which extracellular cargo receptors migrate across the PM to interact with AP2. These findings also raise some exciting questions: how can AGPs, which are heavily modified by bulky arabinogalactans and are typically tethered to the extracellular cell surface via a GPI anchor, move across the PM, and how does REE binding initiate this transmembrane movement? The transmembrane migration occurs before clathrin assembly, as shown using the STED superresolution microscopy in cells where this process was blocked by CANTH expression (Fig. 3). Flip-flopping of REE-activated AGPs is a possible explanation for the transmembrane movement. From a structural perspective, it is possible that after AGPs containing transmembrane domain act as receptors to coordinate with the hydrous ions of REE(III), the introduction of hydrous ions of REE(III) increases the hydrophilicity of AGPs (SI Appendix, Fig. S5). This may speed up the flip-flopping of membrane lipids, resembling the rapid flip-flopping of endoplasmic reticulum membrane lipids driven by transmembrane proteins after their hydrophilicity is increased (49–51), and thus drive the transmembrane movement of AGPs (Fig. 3). Alternatively, La(III)-activated AGPs could interact with a transmembrane protein, which then may bind the AP2 complexes. Moreover, it was reported that some of the AGPs are expressed in specific tissues (such as AGP6 and AGP11 in pollen) (52). Whether AGPs in pollen tubes have the same role as AGPs in leaves should be investigated in future studies by using the related mutants.

Fig. 5.

An unprecedented mechanism for the initiation of CME. (A) Mechanism for CME initiated by traditional cargo. Traditional cargo is a transmembrane receptor protein. When cargo receives the endocytic signal, it exposes the recognition domain to recruit and interact with AP2. Subsequently, AP2 recruits clathrin to assemble clathrin-coated vesicle. (B) Unprecedented mechanism for CME initiated by AGPs in presence of REEs. Originally, AGPs are secreted proteins localized to the outer leaflet of the PM via a GPI anchor, but when they encounter La(III), they change their structure to transmembrane form. This form of AGPs exposes the recognition domain to recruit and interact with AP2 and then recruit clathrin to assemble clathrin-coated vesicle. These further showed that secretory AGPs anchored in the outer leaflet of the PM via a GPI anchor act as the transmembrane receptor proteins and perform the function of transmembrane receptor proteins.

Explosive usage of REEs in the modern industry and agriculture urgently demands our understanding of how REEs enter the biological systems and their impacts on ecosystems and human health. We found that AGPs and REEs also activate and regulate the endocytosis in human cells (SI Appendix, Fig. S6). Meanwhile, REEs are heavy metal elements in the periodic table of elements (5). Thus, REEs-initiated endocytosis opens floodgates for heavy metal elements to enter plant cells (11) because REEs and heavy metal elements are ubiquitous in atmospheric particulate matters (9). This action greatly increases the risk of heavy metal stress in plants and enhances the accumulation of pollutants in the human body via the food chain. Obviously, once pollutants accumulate in plant cells via REEs-initiated endocytosis in leaf cells, the health of animals and human beings will be seriously threatened. Our current study also solves the mystery of how REEs sprayed on the foliage of plants in agriculture may enter plant cells via endocytosis; that is, AGPs are the primary targets of REEs to initiate the CME in the leaf cells. This knowledge may help to design new strategies to manipulate plants for optimal REE accumulation and for limiting heavy metal accumulation in living organisms.

Materials and Methods

Plant materials, growth conditions and La(III) treatment, plasmid construction, chemical solutions and treatments, real-time PCR analysis, EMARG measurement of La(III) subcellular distribution, AGP immunolocalization and EDS measurement, FM4-64 internalization assay, STED microscopy determination, AGPs and AP2 immunolocalization assay, co-IP assay for the AGP-AP2μ interaction, the measurement of the La level in AGP-La(III) complexes, chemical calculation, endocytosis in HEK293 cells, and statistical analysis are described in SI Appendix, SI Materials and Methods.