Macrophage Receptor with Collagenous Structure (MARCO) Is Processed by either Macropinocytosis or Endocytosis-Autophagy Pathway

The Macrophage Receptor with COllagenous structure (MARCO) protein is a plasma membrane receptor for un-opsonized or environmental particles on phagocytic cells. Here, we show that MARCO was internalized either by ruffling of plasma membrane followed by macropinocytosis or by endocytosis followed by fusion with autophagosome in CHO-K1 cells stably transfected with GFP-MARCO. The macropinocytic process generated large vesicles when the plasma membrane subsided. The endocytosis/autophagosome (amphisome) generated small fluorescent puncta which were visible in the presence of glutamine, chloroquine, bafilomycin, ammonia, and other amines. The small puncta, but not the large vesicles, co-localized with LC3B and lysosomes. The LC3-II/LC3-I ratio increased in the presence of glutamine, ammonia, and chloroquine in various cells. The small puncta trafficked between the peri-nuclear region and the distal ends of cells back and forth at rates of up to 2–3 μm/sec; tubulin, but not actin, regulated the trafficking of the small puncta. Besides phagocytosis MARCO, an adhesive plasma membrane receptor, may play a role in incorporation of various extracellular materials into the cell via both macropinocytic and endocytic pathways.


Introduction
Particles opsonized with IgG or with a complement protein (C3bi) are phagocytosed by macrophages via FcγR or CR3, respectively. In addition to the receptor-mediated internalization of opsonized particles and microorganisms, macrophages recognize and take up non-opsonized or environmental particles such as silica, iron oxide, carbon soot, and polystyrene beads via Macrophage Receptor with COllagenous structure (MARCO) protein [1,2]. The cytosolic domain of MARCO is very short [3] and no signal transduction pathway via MARCO has been proposed yet. Moreover, the metabolic fate of this plasma membrane protein has not been reported, although MARCO is known to play a pivotal role in the uptake of non-opsonized particles [4]. pcDNA6.2/N-EmGFP using the Gateway 1 LR Clonase Enzyme Mix (Life Technologies) according to the manufacturer's instructions. Chinese hamster ovary (CHO) -K1 cells and human embryonic kidney HEK293 cells were stably transfected with GFP-MARCO using Lipofectamine LTX-PLUS (Life Technologies). Unless otherwise mentioned, GFP-MARCO-CHO cells were cultured in F12 medium and HEK293 and GFP-MARCO-HEK cells were cultured in DMEM medium containing 100 units/mL penicillin, 100 μg/mL streptomycin, 20 μg/mL blasticidin-S, and 10% heat-inactivated fetal bovine serum (FBS). Jurkat and J774.1 cells were cultured in RPMI1640 medium containing 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% FBS.

Fluorescent staining
GFP-MARCO-CHO cells cultured in an 8-well glass chamber slide were fixed in phosphatebuffered 3.7% formaldehyde solution and then permeabilized with 0.1% Triton X-100. The slides were pre-incubated in 1% bovine serum albumin (BSA), washed in PBS, and stained with rhodamine phalloidin (Life Technologies) or anti-α-tubulin antibody followed by Alexa Fluor1 594-labeled goat anti-mouse IgG with enhancer solution (Life Technologies). Samples were mounted onto slides using a DAPI-containing anti-fade reagent (ProLong Gold1; Life Technologies) and the cells were observed by fluorescence microscopy. For immunostaining with anti-hamster LAMP2 antibody, the cells were permeabilized with PBS containing 0.1% saponin and 3% BSA, with subsequent processing as described above. The anti-LAMP-2 antibody was purified from the culture medium of the hybridoma (UH3, DSHB, Iowa City, IA) using the MAbTrap TM Kit (GE Healthcare). The cells were observed by fluorescence microscopy (Eclipse 80i, Nikon, Tokyo, Japan).

Cell growth
GFP-MARCO-CHO cells were suspended in complete F12 medium at 2 x 10 4 cells/mL and aliquotted at 100 μL/well in a 96-well culture dish. The cells were pre-cultured for 15 hr and then cultured in various conditions for 24 or 48 hr. The cell monolayers were washed twice with HBSS and the number of viable cells was evaluated by a modified MTT assay method using the WST-8 Cell Counting kit. After chromophore development, the reaction was quenched by adding one-tenth volume of 0.1 mol/L HCl solution. The O.D. at 450 nm was measured using a microplate reader (POLARstar OPTIMA, BMG Labtech, Offenburg, Germany).

Transmission electron microscopy (TEM)
GFP-MARCO-CHO cells were cultured on a cover slip placed in a culture dish and incubated under various conditions. The cells were pre-fixed for 2 hr in 2.5% glutaraldehyde in 0.1 M phosphate buffer containing 0.2 M sucrose and post-fixed with 1% osmium tetroxide for 1 hr. After sequential dehydration using increasing concentrations of ethanol, samples were incubated in butyl glycidyl ether (BGE) and then embedded in Quetol 812 resin in a BEEM 1 capsule (Okenshoji, Tokyo, Japan). Thin sections (80-100 nm) were prepared and stained with uranium acetate and lead citrate. The samples were observed using a transmission electron microscope (JEM-2010, JEOL, Tokyo, Japan).

Scanning electron microscopy (SEM)
GFP-MARCO-CHO cells were pre-fixed in 2.5% glutaraldehyde and post-fixed with 1% osmium tetroxide as described above. After sequential dehydration using increasing concentrations of ethanol, the samples were subjected to critical point drying and sputter-coated with Pt/ Pd. The cells were observed by scanning electron microscope (JSM-5800, JEOL).

Statistical analyses
Cell growth data, LC3-II/LC-I ratios, and densitometric data were presented as mean ± SEM. Statistical analyses were performed by Analysis of Variance (ANOVA) followed by Tukey's post-hoc comparison. Statistical significance was accepted when p <0.05.

Results
Macropinocytic vesicles and glutamine-induced small puncta GFP-MARCO was internalized by macropinocytosis in GFP-MARCO-CHO cells, as shown in Fig 1A and 1B and S1 Movie. Dense GFP labeling or local ruffling of plasma membrane appeared in peripheral sites of the cell as plasma membrane ruffled and the GFP-labeled membrane was pinched off and internalized as vesicles. The vesicles disappeared within several minutes.
The cells were cultured either in F12 or DMEM culture medium overnight and the live cells were observed by confocal microscopy. Surprisingly, small fluorescent puncta appeared when the cells were cultured in DMEM (Fig 2A). We examined all nutrients and supplements contained differentially between F12 and DMEM and found that the higher concentration of glutamine (4 mM) in DMEM was responsible for the formation of the small fluorescent puncta. The small fluorescent puncta (ca. 0.5 μm) observed in DMEM-grown GFP-CHO-MARCO cells were distinct from the large macropinocytic retractive vesicles (ca. 3 μm) (Fig 1A). Glutamine is catabolized to glutamate by glutaminase, and glutamate is catabolized to α-ketoglutarate by glutamate dehydrogenase, generating two ammonia molecules per glutamine in a process termed glutaminolysis. The small fluorescent puncta appeared when the cells were cultured in F12 culture medium supplemented with additional glutamine (final concentration 4 mM) or with 4 mM (NH 3 ) 2 CO 3 , although supplementation of F12 with 3 mM L-alanyl-L-glutamine did not induce the puncta (Fig 2B). We additionally noted that the proliferation of GFP-MARCO-CHO cells was significantly reduced when cultured in DMEM (Fig 2C). However, 4 mM glutamine did not change the growth rate significantly, suggesting that the generation of small fluorescent puncta was independent of the cell proliferation. Endocytosis was involved in generation of the fluorescent small puncta because dynasore, an inhibitor of dynamin, reduced the number of ammonia-induced small puncta ( Fig 2D).
Glutamine-induced small puncta are related to autophagosome generated by lysosomal dysfunction The small fluorescent puncta induced by glutamine or ammnonia were autophagic puncta, because chloroquine, an autophagy inhibitor, also induced those small fluorescent puncta in  GFP-MARCO-CHO cells cultured in F12 medium and the puncta were co-localized with wildtype LC3B, an autophagosome marker ( Fig 3A). Those small puncta (ca. 0.5 μm) had a doublemembrane structure as observed by TEM ( Fig 3B). Bafilomycin, a specific inhibitor of vacuolar-type H + -ATPase and another autophagy inhibitor, also induced the fluorescent small puncta (S1 Fig). Ammonia is membrane-permeable and neutralize the acidity of lysosomes. The lysosomes were distinctly stained by pHrode TM Red. However, they are not differentially stained in the presence of (NH 4 ) 2 CO 3 , indicating that the GFP-derived fluorescent small puncta were produced by lysosomal alkalinization (Fig 4A). The fluorescent small puncta also co-localized with lysosomal protein LAMP-2 (Fig 4C), while macropinocytic vesicles were not co-localized with LAMP-2-containing vesicles (Fig 4B), suggesting that GFP-MARCO internalized by macropinocytosis is probably recycled without digestion in lysosomes and GFP-MARCO internalized by endocytosis eventually merged with the lysosome regardless of the function of this organelle.
It is known that GFP is resistant to lysosomal proteolysis and appearance of the free GFP moiety reflects cargo delivery of GFP-LC3 from autophagosome to lysosome [22]. The appearance of a band at 27 kD in the (NH 3 ) 2 CO 3 -exposed cells, which is corresponding to the free GFP moiety, indicates that the lysosomal function was compromised by alkalinization and GFP cleaved off GFP-MARCO was not digested in lysosomes (Fig 4D).

Trafficking of ammonia-induced autophagic puncta
Neither inhibition of actin polymerization by cytochalasin D nor inhibition of microtubule polymerization by nocodazole was affected by (NH 4 ) 2 CO 3 (Fig 5A and 5B). Cytochalasin D crippled withdrawal movement of macropinocytosis, leaving several retreating macropinocytic vesicles at peripheral sites ( Fig 5C). However, small autophagic puncta were not affected by treatment with cytochalasin D. In contrast, most autophagic small puncta agglomerated in the peri-nuclear region in cells treated with nocodazole, an inhibitor of microtubule polymerization ( Fig 5C). Some autophagic small puncta moved between the peripheral and the peri-nuclear regions at speeds of up to 2-3 μm/sec. However, most of the puncta moved more slowly (S2 Movie). It is plausible that the movement of autophagic small puncta was hindered by nocodazole and the puncta were accumulated around the nucleus. Those results also suggest that macropinocytosis, which requires polymerization of actin, is not involved in the formation of autophagic puncta, if any.
Conversion of LC3-I to LC3-II in ammonia-and amine-exposed cells The conversion of LC3-I to its lipidated form LC3-II is a benchmark for autophagosomal maturation. Fig 6A shows that the lipidation of LC3 occurred in GFP-MARCO-CHO cells grown in F12 supplemented with (NH 4 ) 3 CO 3 time-and dose-dependently. The LC3 lipidation in (NH 4 ) 3 CO 3 -or chloroquine-supplemented medium was observed in other types of cells such as Jurkat and J774.1 cells (Fig 6B). It should be noted that the amount of LC3 varies among the cell lines and LC3-I was not remarkably changed as LC3-II was increased clearly by (NH 4 ) 3 CO 3 and chloroquine. LC3 was lipidated in GFP-MARCO-CHO cells grown in medium supplemented with other amines such as trimethylamine (8 mM) and trimethylamine hydrochloride (8 mM) (Fig 6C). Those amines also induced autophagic small puncta (Fig 6D). Rapamycin, an inhibitor of mTOR, neither induced autophagic small puncta nor attenuated a potency of (NH 4 ) 3 CO 3 to induce formation of the puncta (S2 Fig). The autophagic fluorescent small puncta were also observed in GFP-MARCO-HEK cells when these cells were cultured in complete DMEM medium (containing 4 mM glutamine). Those puncta disappeared 6 hr after the medium was replaced with glutamine-free DMEM medium and emerged again 4 hr after the glutamine-free medium was replaced with complete DMEM medium (Fig 7A). The LC3-II was decreased when the cells were cultured in glutamine-free DMEM. However, the LC3-II/LC3-I ratio remained unchanged or even decreased a little after a further culture medium change to complete DMEM for 4 hr (Fig 7B), although the autophagic fluorescent small puncta re-appeared under these conditions (Fig 7A). The level of ATG5 protein, a key molecule of autophagy, was unaltered by any of the medium changes.

Discussion
MARCO is a key molecule that recognizes unopsonized particles in macrophages, although MARCO's signal transduction cascade upon attachment to particles is unknown. Uptake of polystyrene particles was highly enhanced by transient overexpression of MARCO irrespective of the particle size (20 nm, 200 nm, and 1 μm) [23], suggesting that MARCO functions as a "fly ribbon" on the plasma membrane of macrophages. MARCO is probably processed and digested in the cells, because MARCO was not detected as a single band and many fragments were observed in western blot analysis of GFP-MARCO proteins using anti-MARCO antibody [21]. As we demonstrate in the present work, MARCO was internalized by macropinocytic ruffling of plasma membrane in GFP-MARCO-CHO cells, and the resulting vesicles moved centripetally and disappeared in several minutes. Additionally, we observed that GFP-MARCOcontaining macropinocytic vesicles or macropinosomes did not co-localize with lysosomes. The fate of macropinosomes is not well understood. In the current study, there was no sign that macropinosomes fused with lysosomes in GFP-MARCO-CHO cells, suggesting that GFP-MARCO-containing macropinosomes probably were recycled by fusion with the plasma membrane, as previously reported in A431 and 3T3 cells [24]. However, it is possible that part of the macropinocytic membrane in our cells was incorporated into the autophagic small puncta; indeed, phagocytic membrane has been detected in autophagosomes in HeLa cells [25].
Surprisingly, small fluorescent puncta were formed in GFP-MARCO-CHO and GFP-MAR-CO-HEK cells when the cells were cultured in DMEM. These small puncta were observed only rarely when the cells were cultured in F12 culture or glutamine-free DMEM medium. We compared the composition of F12 with that of DMEM, and noted that a number of components (including pyruvate, vitamins, trace metals such as zinc and copper, putrescine, linoleic acid, and lipoic acid) are present in F12 culture medium and are missing or present only at lower concentrations in DMEM. None of these components abrogated the generation of the small puncta in GFP-MARCO-CHO cells when used to supplement DMEM. In contrast, DMEM contains L-glutamine at a level 4 times higher than that of F12 culture medium (4 mM vs. 1 mM), and culture in F12 supplemented with L-glutamine induced the formation of small puncta in GFP-MARCO-CHO cells. GlutaMAX TM (L-alanyl-L-glutamine), which is gradually hydrolyzed to release L-glutamine, was less effective than L-glutamine; 4 mM ammonium carbonate, a direct ammonia source, induced the puncta rapidly. These results suggest that intracellular ammonia generated by glutaminolysis induced the small puncta when GFP-MAR-CO-CHO cells were cultured in DMEM. The puncta were not formed in glutamine-free DMEM. However, supplementation of glutamine-free DMEM with 1 mM glutamine (i.e., to the level contained in F12 culture medium) was sufficient to induce the formation of small puncta after 24 hr (data not shown). These results suggest that although glutamine is necessary for the generation of puncta, other medium components modulate the generation and/or stability of the puncta. Intriguingly GFP-MARCO-CHO cells grew faster in F12 than in DMEM, and glutamine supplementation did not further enhance the cell proliferation, suggesting that the small fluorescent puncta were formed in the presence of glutamine regardless of growth conditions.
Macropinocytosis is an actin-dependent process that permits uptake of extracellular fluid and proteins. Endocytic vesicles are continuously formed by invagination and scission of plasma membranes [26]. In the process of endocytosis, plasma membrane pinches off and fuses with ATG16L1 to form phagophores [14]. Two different classes of endosomes (ATG9-associated and ATG16L1-associated) appear to be incorporated into LC3-negative phagophores, with the Soluble NSF Attachment protein REceptor (SNARE) mediating the coalescence of those two different endosomes [27][28][29]. Our work revealed that the autophagic small fluorescent puncta were still observed even when macropinocytosis was inhibited by cytochalasin D and dynasore inhibited the production of the small puncta, suggesting that the endocytic pathway was the dominant contributor to the generation of MARCO-containing autophagic membranes.
It has been reported that autophagosomes are formed randomly in the peripheral regions in mammalian cells and move centripetally and centrifugally at a speed of 0.1 μm/sec in normal rat kidney cells, gradually distributing in the peri-nuclear region [30]. It also has been shown that the centripetal movement of autophagosomes is dependent on dynein, the microtubule motor protein. In our current study, some small puncta traveled to and fro between the peripheral and peri-nuclear sites at a speed of up to 2-3 μm/sec, although other small puncta appeared to move more slowly. The movement of puncta was impaired upon disruption of tubulin by nocodazole, suggesting that the small fluorescent puncta movement was controlled via microtubules. It has been reported that mTOR is activated and lysosomes are maintained at the periphery of the cell through microtubule-dependent mechanisms when nutrients are sufficient [31]. In contrast, nutrient-deprivation increases intracellular pH from 7.1 to 7.7, resulting in accumulation of lysosomes in the peri-nuclear area and deactivation of mTOR [32]. The lysosomal positioning regulates mTORC1 signaling, which in turn influences autophagosome formation and autophagosome-lysosome fusion rates [32]. The present study suggests that generation of ammonia by exposure of cells to excess glutamine and/or amines increases intracellular pH; intracellular alkalinization presumably impairs lysosomal functions, resulting in accumulation of endocytosis-derived MARCO-containing autophagosomes.
The conversion of LC3-I to LC3-II reflects conjugation of phosphatidylserine and phosphatidylethanolamine to LC3-I [33,34], a process that is inhibited by protein kinase A-mediated phosphorylation [35]. Glutamine has been shown to increase the levels of LC3-II and autophagosomal puncta in parallel to inactivation of mTOR and MAP p38 in rat intestinal epithelial and human colonic epithelial cells [36]. Glutamine synthetase synthesis is transcriptionally regulated by PI3K-PKB-FOXO signaling, and intracellular glutamine up-regulates autophagy by preventing the translocation of mTORC1 to lysosomal membranes [37]. Those reports suggest that intracellular glutamine induces autophagy via mTOR, a traditional autophagy key molecule that is inhibited by rapamycin and nutrient starvation. In contrast, glutamine also induces autophagy via ammonia [22], a by-product of glutamine metabolism. Glutaminolysis-or ammonia-induced pro-autophagic signals have been reported not to be attributable to nutrient depletion or to inhibition of mTOR [22,38]. Ammonia, which is generated in mouse embryonic fibroblasts when the cells are deprived of glucose, is known to induce ULK1/2-mTORindependent and ATG5-dependent autophagy [39]. The present study suggests that glutamine at a typical culture medium concentration (4 mM) is sufficient to induce accumulation of autophagic puncta via ammonia production, and that ammonia is an effective signal for the conversion of LC3-1 to LC3-II.
It has been reported that ectopically expressed MARCO was observed on the cell surface in Jurkat and L929 cells, whereas it was retained intracellularly in J774.1, EL-4, WEHI-3B, and P338D 1 cells. The different behavior of MARCO in those cell lines may be due to the presence or absence of molecular chaperons required for the correct folding and assembly of the trimeric and functional MARCO [40]. Our current study indicates that MARCO molecules on the cell membrane were continually internalized by macropinocytosis and endocytosis in GFP-MAR-CO-CHO cells. It remains to be answered whether the dynamic behavior of exogenously overexpressed MARCO in GFP-MARCO-CHO cells is also seen in endogenous MARCO molecules in primary macrophages.

Conclusions
A plasma membrane receptor MARCO was incorporated into both macropinosomes and endosomes, although only the latter vesicles were further processed into autophagosomes. Ammonia generated from glutamine inhibited autophagy. Although MARCO was first reported to be a phagocytic receptor for environmental particles and microorganisms, the present study suggests that MARCO is internalized during macropinocytic and endocytic processes. It is also possible that the uptake of unopsonized particles via MARCO is brought about significantly by macropinocytic activity of phagocytes, if MARCO functions mainly as a "fly ribbon". GFP-MARCO-CHO cells were cultured in F12 complete medium to 50% confluence, and rapamycin was added to concentrations of 0, 100 nM, or 1 μM. After 24 hr, (NH 3 ) 2 CO 3 was added to each well to a final concentration of 0 or 4 mM and the cells were further cultured for 5 hr. S2 Movie. The CHO-GFP-MARCO cells were cultured in a cell culture plastic dish and incubated with 4 mM (NH 4 ) 2 CO 3 for 15 hr. The movie was taken in real time (5 frames/sec) using an inverted fluorescence microscope. Some vesicles moved quickly from the peri-nuclear region to the distal area of the cells or vice versa. The puncta shown by the arrows moved along the radial direction with a velocity of 2.6 (upper arrow) and 1.9 μm/sec (lower arrow). The scale bar shows 50 μm. (WMV)