key: cord-0836606-fj1d4bff
authors: Licona‐Limón, Ileana; Garay‐Canales, Claudia A.; Muñoz‐Paleta, Ofelia; Ortega, Enrique
title: CD13 mediates phagocytosis in human monocytic cells
date: 2015-05-01
journal: J Leukoc Biol
DOI: 10.1189/jlb.2a0914-458r
sha: fd2f45bcb31d6d727c246838563b4e2b0b2b4e69
doc_id: 836606
cord_uid: fj1d4bff

CD13 is a membrane‐bound ectopeptidase, highly expressed on monocytes, macrophages, and dendritic cells. CD13 is involved in diverse functions, including degradation of peptide mediators, cellular adhesion, migration, viral endocytosis, signaling, and positive modulation of phagocytosis mediated by FcγRs and other phagocytic receptors. In this work, we explored whether besides acting as an accessory receptor, CD13 by itself is a primary phagocytic receptor. We found that hCD13 mediates efficient phagocytosis of large particles (erythrocytes) modified so as to interact with the cell only through CD13 in human macrophages and THP‐1 monocytic cells. The extent of this phagocytosis is comparable with the phagocytosis mediated through the canonical phagocytic receptor FcγRI. Furthermore, we demonstrated that hCD13 expression in the nonphagocytic cell line HEK293 is sufficient to enable these cells to internalize particles bound through hCD13. CD13‐mediated phagocytosis is independent of other phagocytic receptors, as it occurs in the absence of FcγRs, CR3, and most phagocytic receptors. Phagocytosis through CD13 is independent of its enzymatic activity but is dependent on actin rearrangement and activation of PI3K and is partially dependent on Syk activation. Moreover, the cross‐linking of CD13 with antibodies rapidly induced pSyk in human macrophages. Finally, we observed that antibody‐mediated cross‐linking of hCD13, expressed in the murine macrophage‐like J774 cell line, induces production of ROS. These results demonstrate that CD13 is a fully competent phagocytic receptor capable of mediating internalization of large particles.

Phagocytosis plays a critical role in innate and adaptive immunity by facilitating the removal and killing of pathogens while priming the adaptive immune response. Phagocytosis is a receptor-mediated event. Direct or indirect (opsoninmediated) recognition of the target particle by receptors on the surface of professional phagocytes triggers signaling events that mediate actin-dependent particle internalization. Receptors able to mediate phagocytosis are structurally diverse and include a variety of receptor families [1, 2] , such as FcgRs (FcgRI, -IIa, -IIc, and -IIIa), integrins (such as CR3, CR4, a 5 b 1 , a v b 3 , and LFA-1) [3] [4] [5] [6] , scavenger receptors (such as scavenger receptor I, CD36, MARCO) [7] [8] [9] , C-type lectin receptors (such as Dectin-1, dendritic cell-cell-specific intercellular adhesion molecule-3-grabbing nonintegrin, mannose receptor), glycoproteins (such as CD44) [10] , and adhesion molecules (such as carcinoembryonic antigen-related cell adhesion molecule 3) [11] . This list has expanded in the last years, and it is expected to continue growing, as the receptors involved in certain phagocytic processes have not been defined yet.

Membrane peptidases are a multifunctional group of ectoenzymes involved in several processes, including growth, differentiation, activation, cell-cell interactions, and trafficking. Interestingly, their participation in such processes is not always dependent on their peptidase activity [12] . Aminopeptidase N/CD13 is a transmembrane ectoenzyme expressed in several tissues. Among hematopoietic cells, CD13 expression is restricted to stem cells and most developmental stages of myeloid cells, to such an extent that CD13 has been considered a myelomonocytic marker [13] . CD13 cleaves N-terminal neutral amino acids of small peptides, and its substrates and functions vary according to the tissue where it is expressed. However, additional functions, independent of its enzymatic activity, have been attributed to CD13 [14] . Inhibition of CD13 enzymatic activity by pharmacological inhibitors, active-site mutations or inhibitory antibodies, is evidence that at least 3 types of functions of CD13 do not require its enzymatic activity: 1) as a viral receptor [15] [16] [17] , 2) as an adhesion molecule [18] [19] [20] , and 3) as a signaling molecule [20] [21] [22] [23] . Interestingly, all of these functions rely on CD13 crosslinking on the plasma membrane by antibodies or viral ligands, which presumably leads to signal transduction.

Structurally, CD13 is a heavily glycosylated type II membrane protein with a large extracellular domain, a single transmembrane region, and a short cytoplasmic domain of 8 aa. Despite having a very short cytoplasmic region, CD13 cross-linking by antibodies or viral ligands induces intracellular signals, such as intracellular Ca 2+ increases; activation of Src, PI3K, FAK, and the Ras/MAPK pathway; cytokine secretion [20, 23, 24] ; and CD13 association to actin fibers [20, 25] . Given that the short intracellular domain of CD13 contains no classic signaling motifs, it has been proposed that CD13 might require an auxiliary protein to transduce signals. In line with this, CD13 coimmunoprecipitates with the adaptor molecules Grb2/Sos in U-937 monocytes [18] , providing a possible link between CD13 aggregation and MAPK signaling. Furthermore, a recent report showed that CD13 is constitutively associated to the scaffolding protein IQGAP1 and that upon CD13 aggregation, a-actinin is recruited into the complex, linking membrane-bound CD13 to the cytoskeleton [20] . However, that same study showed that upon aggregation, CD13 is phosphorylated on the Tyr6 of its cytoplasmic domain in a Src-dependent manner and that this phosphorylation is required for ERK and FAK activation, as well as for CD13-mediated adhesion [20] . This observation emphasizes that despite being very short, the cytoplasmic region of CD13 participates directly in cellular signaling.

We [18, 19] and others [20, 26] have shown that antibodyinduced CD13 aggregation mediates monocyte-monocyte and monocyte-endothelial cell adhesion in vitro and in vivo in a tyrosine kinase-dependent manner. Phagocytosis shares biochemical mechanisms with adhesion, spreading, and migration, as all of these processes require coordinated cytoskeletal reorganization. Indeed, there are several examples of membrane molecules that work as cell adhesion molecules and phagocytic receptors, including CR3 [27, 28] , CD44 [10, 29] , CD36 [30] , and a V b 3 integrin [5] , among others [31] . The first suggestion that CD13 was possibly involved in phagocytosis dates back to 1996, when a positive correlation between CD13 expression and phagocytosis of microspheres was observed in hMDMs [32] . We have reported previously that CD13 cross-linking by antibodies during phagocytosis of zymosan particles or heat-killed Escherichia coli led to a more efficient uptake in macrophages and dendritic cells [33] . Furthermore, an increase in FcgRI-mediated phagocytosis was also observed in monocytes and macrophages when particles interacted simultaneously with CD13 and FcgRI compared with the phagocytosis of particles interacting with FcgRI alone. Likewise, the cocross-linking of CD13 with FcgRI by specifc mAb increases the pSyk level and duration compared with crosslinking FcgRI alone [34] . These results suggest that CD13 aggregation can trigger signaling pathways required for phagocytosis. Finally, some data in our previous works also suggested that besides acting as a positive modulator of phagocytosis, CD13 might function as a phagocytic receptor in U-937 [34] and dendritic cells [33] , although this was not studied thoroughly. In this paper, we examined whether CD13 can function as a primary phagocytic receptor. We found that CD13 is a competent phagocytic receptor capable of mediating phagocytosis of large particles independently of other phagocytic receptors, the signaling pathway required for phagocytosis through CD13 involves Syk and PI3K, and finally, that CD13 cross-linking induces ROS production.

THP-1 and J774 cells (originally obtained from ATCC, Manassas, VA, USA) were cultured in RPMI-1640 medium (Gibco, Grand Island, NY, USA). HEK293 cells (ATCC) were cultured in DMEM-high glucose (Gibco). All media were supplemented with 10% heat-inactivated FBS (Invitrogen, Carlsbad, CA, USA) and 1 mM sodium pyruvate, 0.1 mM nonessential amino acids solution, 0.1 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin (complete media; all purchased from Gibco). Cultures were maintained in a humidified atmosphere at 37°C with 5% CO 2 . Murine monoclonal anti-hCD13 (clone 452, IgG1) was purified in our laboratory from culture supernatants of the hybridoma, kindly donated by Dr. Meenhard Herlyn (Wistar Institute of Anatomy and Biology, Philadelphia, PA, USA). Murine monoclonal anti-human FcgRI (mAb 32.2, IgG1) was purified from supernatants of the corresponding hybridoma obtained from ATCC. Fab fragments of the antibodies were prepared with immobilized ficin (Pierce, Rockford, IL, USA). Biotin-F(ab)9 2 fragments of goat anti-mouse IgG (H+L) were from Zymed (Invitrogen) and from Life Technologies (Eugene, OR, USA). F(ab)9 2 fragments of goat anti-mouse IgG were purchased from Jackson ImmunoResearch (West Grove, PA, USA). Goat anti-mouse-FITC, used as secondary antibody for immunostaining, was from Zymed (Invitrogen). PE-labeled mouse anti-human pSyk (pY348), Fix Buffer I, and Perm Buffer II solutions for intracellular staining were from BD Phosflow (BD Biosciences, San Diego, CA, USA). hMDMs Buffy coats from healthy male donors were obtained from the Central Blood Bank of the Centro Médico Nacional La Raza, IMSS, which also approved of their use for these experiments. All experiments carried out with cells from human donors were performed following the Ethical Guidelines of the Instituto de Investigaciones Biomédicas, UNAM (Mexico D.F., México). MDMs were obtained from human PBMCs, as described previously [33] . In brief, mononuclear cells were isolated from buffy coats from healthy donors by density gradient centrifugation by use of Ficoll-Paque Plus (GE Healthcare Bio-Science, Uppsala, Sweden). PBMCs were washed 4 times with PBS, pH 7.4, and cultured in RPMI-1640 medium, supplemented with 10% (v/v) heatinactivated autologous plasma-derived serum, 1 mM MEM sodium pyruvate solution, 2 mM MEM nonessential amino acid solution, 0.1 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin, for 30 minutes at 37°C to allow monocytes to adhere to the plastic plate. Nonadherent cells were eliminated by washing, and adherent cells, enriched for monocytes (.90% purity, as determined by flow cytometry by use of CD14 as a marker of the monocytic population), were cultured for 7 days in RPMI-1640 medium, supplemented with 10% (v/v) heat-inactivated FBS, in a humidified atmosphere at 37°C with 5% CO 2 . The resulting MDMs were harvested by mechanical scrapping, washed, and used for experiments.

Modified SRBCs were prepared, as described previously [34] . In brief, erythrocytes (at 1.2 3 10 9 /ml in PBS-BSA 0.1%) were stained with 10 mM CFSE (Life Technologies). The stained erythrocytes were incubated with 250 mg/ml sulfo-NHS-biotin (Pierce) for 20 minutes at 4°C. After washing, they were coated with 35 mg/ml Streptavidin (Calbiochem, San Diego, CA, USA) for 20 minutes at 4°C. The biotin-streptavidin-coated erythrocytes were washed and incubated with biotinylated F(ab)9 2 fragments of goat anti-mouse IgG for 30 minutes at 4°C. SRBCs labeled with CFSE and coated with biotin, streptavidin, and F(ab)' 2 fragments of biotinylated anti-IgG antibodies are henceforth designated EBS-Fab. For phagocytosis assays, 1 3 10 6 MDMs or THP-1 cells were incubated with 2 mg of Fab fragments of mAb452 (anti-CD13), 8 mg Fab fragments of mAb32.2 (anti-FcgRI), 8 mg IgG1 (isotype-matched control), or without treatment (control) for 30 minutes at 4°C, washed, and incubated with EBS-Fab at a ratio of 1 monocytic cell:20 EBS-Fab, at 37°C for 45 minutes in the case of MDMs or for 105 minutes for THP-1 cells. Identical samples were incubated at 4°C as controls. Noninternalized erythrocytes were lysed by hypotonic shock. Phagocytosis was quantified by flow cytometry (Attune acoustic focusing cytometer; Applied Biosystems Carlsbad, CA, USA, or FACSCalibur; BD Biosciences), with and without addition of Trypan blue 0.02% in PBS (pH 4.5), to differentiate between attached and ingested erythrocytes. Data are expressed as the percentage of CFSE-positive cells (cells that have ingested at least 1 erythrocyte). In experiments evaluating the effect of pharmacological inhibitors of kinases or actin polymerization, MDMs or THP-1 cells were preincubated for 2 hours at 37°C, with 50 mM piceatannol, 50 mM LY, 200 nM cytochalasin D, or 1 or 10 mM BAY (all Sigma-Aldrich, St. Louis, MO, USA) or DMSO as vehicle control, and the phagocytosis assay proceeded as described above, always in the presence of the selected inhibitor. For experiments evaluating the effect of the pharmacological inhibitor of CD13 enzymatic activity, THP-1 cells were preincubated for 40 minutes at 37°C with bestatin 4 mg/ml (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and the phagocytosis assay proceeded as described above. For confocal microscopy, cells were labeled previously with PKH26 (Sigma-Aldrich) or labeled with DiI (Molecular Probes, Eugene, OR, USA) after the phagocytosis experiments. Cells were examined in a Zeiss LSM5 confocal microscope.

For the selective phagocytosis assays in J774 and HEK293 expressing hCD13, the EBS-Fab were incubated with Fab fragments of anti-CD13 mAb452 or Fab fragments of anti-FcgRI mAb32.2 or PBS-BSA alone (referred to as EFab452, EFab32.2, or Ec, respectively) for 40 minutes at 4°C. After washing, the EFab452, EFab32.2, or Ec were mixed with the cells without treatment at a ratio of 1 cell:20 EFab452, EFab32.2, or Ec and incubated at 37°C for 1¾ hours. Hypotonic lysis of noningested erythrocytes and analysis by flow cytometry were performed as described above. Cells were examined in a Zeiss LSM5 confocal microscope.

CD13 enzymatic activity was determined by the colorimetric measurement of the hydrolysis of the substrate L-alanine-4-nitroanilide hydrochloride (Sigma-Aldrich), as described previously. THP-1 cells (1 3 10 6 ) were pretreated with 4 ng/ml bestatin for 40 minutes at 37°C, followed by incubation at 4°C for 30 minutes. Then, cells were washed and incubated with bestatin for 1 hour at 37°C. Substrate was added to a final concentration of 6 mM, and incubation was continued for 45 minutes at 37°C. Absorbance at 405 nm was determined immediately.

For the determination of the effect of mAb on CD13 enzymatic activity, MDMs were preincubated for 30 minutes on ice with the different Fab fragments (Fab452, Fab C, or Fab 32.2), antibodies (WM-15 or IgG1), or bestatin at 4, 40, and 400 mg/ml and washed. Cells were then incubated for 105 minutes at 37°C, substrate was added for the last 45 minutes of incubation, and absorbance at 405 nm was determined.

For cross-linking of CD13 on MDMs, 1 3 10 6 cells were incubated either with Fab fragments of mAb452 anti-CD13 (4 mg), Fab fragments of anti-FcgRI (10 mg), anti-FcgRII (10 mg), or without primary antibody for 30 minutes at 4°C. Subsequently, cells were washed and incubated with F(ab)9 2 fragments of goat anti-mouse IgG for 10 minutes at 4°C, followed by incubation at 37°C (cross-linking) for the indicated time in complete media. Signaling was stopped by fixation in Fix Buffer I (BD Phosflow; BD Biosciences) for 10 minutes at 37°C, followed by washing and permeabilization in Perm Buffer II BD Phosflow (BD Biosciences) for 40 minutes at 4°C. Afterwards, cells were washed and stained with anti-pSyk antibody coupled to PE. Lentiviral shRNA-mediated silencing of CD13 in THP-1 cells THP-1 cells at 70% confluency in a 96-well plate were transduced by MISSION Lentiviral Transduction Particles against hCD13 (SHVRS-NM_001150 Virus titer: .1 3 10 6 transducing units/ml; Sigma-Aldrich). We examined the efficiency of gene silencing by 5 clones designed to target CD13 (TRCN0000050238, TRCN0000050239, TRCN0000050240, TRCN0000050241, TRCN0000050242) or control transduction particles (GFP-target shRNA; produced in our lab). After overnight transduction with 8 mg/ml polybrene, the medium containing viral particles was replaced with fresh medium. From day 2 after infection, selection medium with 2.5 mg/ml puromycin (Sigma-Aldrich) was used in subsequent cultures for the following 2 months. Stable knockdown of CD13 was determined by flow cytometry. Out of the 5 clones tested, clone TRCN0000050239, showed the best knockdown efficiency. We worked with cells transduced with clone TRCN0000050239 only (herein referred as L2 cells).

To induce hCD13 expression in epithelial (HEK293) and murine macrophage-like (J774) cells, hCD13 was cloned into a lentiviral expression vector, pLenti-suCMV(hANPEP)-Rsv(RFP-Bsd; GenTarget, San Diego, CA, USA). An empty vector, suCMV(empty)-Rsv(RFP-Bsd) plasmid, was generated by excising the hANPEP sequence from pLenti-suCMV(hANPEP)-Rsv(RFP-Bsd). The pLKO.1-puro with shRNA anti-GFP insert (kindly provided by Dr. Julián Valdés, Instituto de Fisiología Celular, Universidad Nacional Autónoma de México) was used as control for silencing experiments.

All recombinant lentiviruses were produced by transient transfection of 293T cells, according to standard protocols. A third-generation packaging system (kindly provided by Dr. Luis Vaca, Instituto de Fisiología Celular, UNAM) was used for maximum biosafety. In brief, subconfluent 293T cells were cotransfected with 20 mg of the plasmid vector, 15 mg pMDLg/RRE, 15 mg pRSV-Rev, and 15 mg pVSVGg by calcium phosphate precipitation. Supernatants containing the lentiviral particles were harvested at 36 hours after transfection. Lentiviral particles were concentrated by ultracentrifugation and used immediately for cell transduction.

Cultures of J774 and HEK293 cells in complete media at 70% confluency in 12-well plates were transduced by concentrated lentiviral particles of suCMV (hANPEP)-Rsv(RFP-Bsd) or suCMV(Empty)-Rsv(RFP-Bsd) with 8 mg/ml polybrene. After 12 hours incubation with viral particles, the medium was replaced with fresh medium. From day 3 after transduction, selection medium with 2.5 and 10 mg/ml blasticidine (Sigma-Aldrich) for J774 and HEK293, respectively, was used for subsequent selection cultures. Membrane expression of hCD13 was determined by flow cytometry, whereas cytosolic RFP expression was determined by confocal microscopy.

CM-H 2 DCFDA, for 20 minutes at 37°C in HBSS. After washing, CD13 was cross-linked for 30 minutes at 37°C, as described above for determination of pSyk in MDMs or incubated with PMA as control. Cells were analyzed immediately by flow cytometry.

To assess the expression of CD13 or FcgRI on the cell surface, cell lines and MDMs (0.5 3 10 6 ) were labeled with Fab fragments of mAb452 (anti-CD13) or Fab fragments of mAb32.2 (anti-FcgRI) for 30 minutes at 4°C, washed, and incubated with a FITC-conjugated goat anti-mouse IgG secondary antibody for 20 minutes at 4°C. Cells were then washed, fixed with paraformaldehyde 1% in PBS, and analyzed by flow cytometry. In experiments aimed at evaluating cell-surface expression of molecules, selective phagocytosis, pSyk, or ROS production, the fluorescence intensity of a total of 15,000 cells was measured by use of a flow cytometer (Attune acoustic focusing cytometer; Applied Biosystems, or FACSCalibur; BD Biosciences).

Graphs contain data from at least 4 equivalent and independent experiments. Data are presented as means 6 SEM. Data were analyzed by use of paired, 2-tailed Student's t-test. Differences were considered significant at P , 0.05.

We first examined whether CD13 is able to mediate phagocytosis independently of other phagocytic receptors in THP-1 cells and hMDMs. We used an experimental system to direct sheep erythrocytes (as the phagocytic prey) to individual receptors on the cell, as reported previously [34] . Erythrocytes loaded with CFSE and labeled with biotin-streptavidin, were coated with biotin-F(ab)9 2 fragments of goat anti-mouse IgG (EBS-Fab). These EBS-Fabs would only interact with molecules on the cell surface with bound Fab fragments of specific murine mAb. As the specificity of the system is based on antibodies, and several cells used in this work express FcgRs, we used Fab fragments to exclude any possible contribution of FcgRs to the phagocytosis (Fig. 1A) . Cells were preincubated on ice with Fab fragments of anti-CD13 mAb452 (Fab452), Fab fragments of anti-FcgRI mAb32.2 (Fab32.2) as positive control, an isotype control antibody (murine IgG1), or no treatment (basal phagocytosis). Cells were washed, mixed with EBS-Fab, and incubated at 37°C or 4°C for the time periods indicated. After incubation, noninternalized erythrocytes were lysed by hypotonic shock. The percentage of cells with internalized erythrocytes (CFSE-positive cells) was determined by flow cytometry after quenching any extracellular fluorescence by use of Trypan blue. We found that the binding of EBS-Fab to the cells through CD13 induced its efficient internalization by the human monocytic cell line THP-1 (mean 28.5%) compared with isotype-matched control antibody (mean 2.5%) or no treatment (mean 3.6%; Fig. 1B and C; n = 25). As a positive control, we measured the phagocytosis through the high-affinity receptor FcgRI (mean 33.5%). As expected, no significant internalization was observed when identical samples were kept at 4°C (Fig. 1B, lower) .

To exclude the possibility that the observed, CD13-mediated internalization of erythrocytes constituted a phenomenon specific of the monocytic cell line, we performed the same experiments on hMDMs. We found similar results to those observed in THP-1 cells, where aggregation of CD13 by the EBS-Fab induced significant particle internalization (mean 31.3%) compared with basal phagocytosis (mean 9.9%). The level of CD13-mediated phagocytosis is comparable with that mediated by FcgRI (mean 38.5%; Fig. 2A and B; n = 14). No significant internalization was observed when identical samples were kept at 4°C ( Fig. 2A, lower) The percentage of cells internalizing erythrocytes through CD13 was somewhat lower than the percentage of cells internalizing erythrocytes through FcgRI, both in THP-1 cells and in hMDMs. This was not a result of a lower binding of Fab fragments to CD13 compared with fragments bound to FcgRI, as anti-CD13 Fab fragments and anti-FcgRI Fab fragments showed similar levels of binding to the cells at the concentrations used in these assays (Figs. 1D and 2C). When we compared the PI (PI = % CFSEpositive cells multiplied by mean fluorescence intensity), we obtained similar results to those from comparing the positive cell percentage, as the mean PI of internalization through FcgRI is slightly higher than the mean PI through CD13, both in THP-1 cells (Supplemental Fig. 1A ) and in macrophages (Supplemental Fig. 1B) . Hence, in this system, although phagocytosis through the canonic phagocytic receptor FcgRI is higher, phagocytosis through CD13 is fully competent, reaching levels of .80% of FcgRI-mediated phagocytosis. To ascertain further that cellassociated fluorescence truly represents CD13-mediated internalization of EBS-Fabs, we verified particle uptake by light (not shown) and confocal microscopy. Typical confocal microscopic images of the internalization of EBS-Fab through CD13 in THP-1 and MDMs are shown in Figs. 1E and 2D, respectively. The number of internalized particles/cell was similar after internalization through CD13 or FcgRI in THP-1 cells and MDMs, in line with the comparison between PIs. Importantly, particles visualized inside of the cells after internalization through CD13 or FcgRI are of similar size, both in THP-1 cells (Fig. 1E ) and in MDMs (Fig.  2D ). Phagocytosed particles are ;4 mm in size, which rules out other forms of internalization, such as caveolae-mediated endocytosis [characterized by small (50-80 nm) vesicles) or clathrinmediated endocytosis (characterized by 100 nm vesicles)]. Altogether, these data indicate that CD13 is capable of mediating phagocytosis of large ($4 mm) particles.

To fully demonstrate that the phagocytosis of EBS-Fab observed in our system is a consequence of particle interaction with CD13 and to exclude the possibility that Fab fragments of mAb452 might be binding to membrane proteins different from CD13, we generated a stable, CD13-deficient THP-1 subline by employing an RNA interference approach. We transduced siL2 or siC+ as control, into THP-1 cells by lentiviral particle delivery. Cells transduced with siL2 showed a significant reduction of nearly 80% in CD13 expression on the cell membrane compared with control or parental cells, as measured by flow cytometry (Fig. 3A) . In contrast, cells transduced with siC+ did not show a detectable change in CD13 expression compared with the untransduced parental cell line (Fig. 3A) . We also monitored the expression of FcgRI in these cells and found a small increase in FcgRI expression in L2 compared with C+ or parental THP-1 cells (Fig. 3A) . We did not observe any other phenotypic or morphologic differences between the CD13-deficient L2 cells and the parental THP-1 cells.

When parental THP-1, control (C+), and CD13-deficient (L2) cells were treated with Fab452 and incubated with EBS-Fab, no significant internalization of EBS-Fab was observed in the L2 cells, in contrast to (C+) or parental cells ( Fig. 3B and C; n = 7). Notably, phagocytosis of EBS-Fabs through FcgRI ( Fig. 3B and C) was not impaired in CD13-deficient cells (L2), demonstrating that these cells are otherwise competent phagocytes. The increase in phagocytosis through FcgRI in L2 cells compared with control cells is not statistically significant and might be explained by the fact that FcgRI expression is higher in L2 cells. These results are in line with recent work in murine CD13 null animals [35] , which showed that lack of CD13 does not affect FcgR-mediated phagocytosis. However, it is important to stress that although the absence of CD13 does not impair phagocytosis through FcgRs, cross-linking of CD13 positively regulates phagocytosis through FcgRs, as we reported previously [34] .

Importantly, this experiment confirmed that the phagocytosis of EBS-Fabs through CD13, observed in Figs. 1 and 2, is truly mediated by CD13, as L2 cells express FcgRI and FcgRII (not shown), but they do not internalize EBS-Fab when Fab anti-CD13 is used. Altogether, these data support that CD13 aggregation, induced by a particle binding to the cell solely through CD13, triggers internalization of the particle.

Phagocytosis, through all bona fide receptors, requires dynamic reorganization of the actin cytoskeleton. To determine whether CD13-mediated particle internalization is dependent on actin cytoskeleton rearrangements, we evaluated CD13-mediated phagocytosis in THP-1 cells treated with 200 nM cytochalasin D, an inhibitor of actin polymerization. Treatment of cells with cytochalasin D resulted in a statistically significant impairment of particle internalization through CD13 (mean 6.9%, n = 9), compared with treatment with vehicle (mean 33.0%; Fig. 4A and B). As expected, FcgRI-mediated phagocytosis was also abrogated after cytochalasin D treatment of the cells (Fig. 4A and B) . These data demonstrate that CD13-mediated phagocytosis depends on reorganization of the actin cytoskeleton.

Actin dynamics for phagocytosis are regulated by signaling pathways activated downstream of ligated phagocytic receptors. The best-understood signaling pathways during phagocytosis are those triggered by the opsonic receptors FcgRs and CR3. The nonreceptor tyrosine kinase Syk becomes activated by phosphorylation upon FcgR or CR3 cross-linking by antibodies or ligand-bearing particles [1] . Several studies have demonstrated that Syk is required for FcgR-mediated phagocytosis, but the role of Syk in CR3-mediated phagocytosis has been somewhat controversial, as CR3 phagocytosis is normal in Syk-deficient macrophages [36] but is severely impaired by transfer of Syk-siRNA or dominant-negative Syk into macrophage-like, differentiated HL-60 cells [37] . In the case of the b-glucan receptor Dectin-1, it has been shown that although Dectin-1 activates Syk, phagocytosis through Dectin-1 is independent of Syk in macrophages but dependent on Syk in fibroblasts, suggesting that Dectin-1-mediated phagocytosis involves novel signaling kinases as well as Syk [38, 39] . Therefore, it seems that Syk activation is not indispensable for all types of phagocytic receptors, even though they are able to activate Syk upon ligation. To investigate whether CD13-mediated phagocytosis is dependent on Syk activation, we used the pharmacological inhibitor of Syk, piceatannol. We preincubated THP-1 cells or hMDMs in the presence of 50 mM piceatannol or DMSO as vehicle control and measured phagocytosis of EBS-Fabs. Treatment of cells with piceatannol completely abolished phagocytosis mediated through CD13 in hMDMs ( Fig. 5A ; mean DMSO 24.7% vs. piceatannol 6.0%, n = 5) and in THP-1 cells (Supplemental Fig. 2A ; mean DMSO 25.3% vs. piceatannol 6.3%, n = 8). As shown in Fig. 5A (and Supplemental Fig. 2A) , treatment of cells with Fab452 and piceatannol resulted in CFSE-positive cell percentages similar to background phagocytosis (No Fab; dotted line), both in hMDMs and THP-1 cells. Phagocytosis, through FcgRI, as anticipated, was also abolished by treatment of cells with piceatannol in hMDMs and THP-1 cells (Fig. 5A and Supplemental Fig. 2A ). To be fully certain about the role of Syk in CD13-mediated phagocytosis, we used a second, more specific Syk inhibitor, BAY, reported to have .600-fold selectivity for Syk than against other kinases, as piceatannol, although used extensively as a Syk inhibitor, has only ;10-fold selectivity for Syk over other kinases involved in phagocytosis, including PI3K and Src kinases [40, 41] . We preincubated hMDMs in medium with 1 or 10 mM BAY or DMSO as vehicle control and measured phagocytosis of EBS-Fabs. As shown in Fig. 5B , CD13-mediated phagocytosis was reduced by 35% in the cells treated with BAY (mean DMSO 30.2% vs. BAY 10 mM 19.9% or BAY 1 mM 19.4%, n = 7). The partial inhibition of phagocytosis through CD13 was not a result of incomplete Syk inhibition, as phagocytosis through FcgRI was abolished completely at both BAY concentrations used ( Fig. 5B ; mean DMSO 35.6% vs. BAY 10 mM 2.2% or BAY 1 mM 3.3%, n = 7). Thus, activation of Syk contributes to, but is not essential for, CD13-mediated phagocytosis. This result suggests that CD13-mediated phagocytosis is likely to involve different early signaling pathways, one that is Syk dependent and another Syk independent. Next, we determined whether cross-linking of CD13 induces Syk phosphorylation. hMDMs were stimulated by aggregation of CD13 or FcgRI for 45 seconds at 37°C. pSyk (pY348) was analyzed by flow cytometry, and a representative contour plot is shown in Fig. 5C . Phosphorylation of Syk, following CD13 aggregation, was statistically different from nonstimulated cells (secondary Fabs only) and also from cells stimulated by FcgRI aggregation for 45 seconds (Fig. 5D ; n = 5). Interestingly, we observed that CD13 aggregation induces stronger Syk phosphorylation than FcgRI aggregation when analyzed shortly after ligation (45 seconds), suggesting that pSyk-induced by CD13 or FcgRI cross-linking follows different kinetics. Indeed, in preliminary experiments to determine the optimal time-point to measure Syk activation after CD13 cross-linking, we found that CD13 aggregation induces a rapid phosphorylation of Syk that peaks at 40 seconds and starts to drop soon after. In contrast, FcgRI-induced phosphorylation of Syk peaks after 60 seconds of receptor aggregation, and the phosphorylation is maintained for several minutes [42] . Together, these data indicate that CD13 aggregation induces phosphorylation of Syk and that CD13-mediated phagocytosis in macrophages is partially dependent on Syk activation.

It has been reported that MAPK activation and intracellular Ca 2+ rise, induced by CD13 aggregation, are inhibited by PI3K inhibitors [24] , suggesting that CD13 aggregation induces PI3K activation. During FcgR-mediated phagocytosis, PI3K activation lies downstream of Syk, and its activation is essential for phagocytosis. As we observed that CD13-mediated phagocytosis is partially dependent on Syk and that CD13 cross-linking induces Syk activation, we next examined the requirement of PI3K during CD13-mediated phagocytosis. THP-1 cells or hMDMs were preincubated with the PI3K inhibitor LY, and phagocytosis was measured. Treatment of cells with LY resulted in a strong impairment of CD13-mediated phagocytosis in hMDMs ( Fig. 5E ; mean DMSO 27.4% vs. LY 7.1%, n = 5) and in THP-1 cells (Supplemental Fig. 2B ; mean DMSO 25.3% vs. LY 5.6%, n = 9). As reported extensively, phagocytosis through FcgRI was inhibited completely in the presence of LY, both in human macrophages ( Fig. 5E ; mean DMSO 31.6% vs. LY 6.6%) and in THP-1 cells (Supplemental Fig. 2B ; mean DMSO 29.3% vs. LY 2.7%). These data suggest that PI3K is involved in CD13mediated phagocytosis.

There are several well-characterized mAb against CD13, with varying effects on enzymatic activity and epitope specificity. Among the panel of different antibodies against CD13, mAb WM-15 binds to the active site of CD13 and has been reported to be the best inhibitor of its enzymatic activity [18, 43] . In contrast, mAb452 has a small but significant effect on enzymatic activity and has been shown to bind to a different epitope form that recognized by the mAb WM-15 [18] . A third anti-CD13 mAb (mAb C) was developed in our laboratory and binds to an epitope distinct from the one recognized by mAb452 but partially overlapping the epitope recognized by mAb WM-15 (data not shown). The effect of this antibody on CD13 enzymatic activity was previously unknown. We used these 3 different anti-CD13 antibodies and measured their ability to mediate phagocytosis. We found that the 3 different anti-CD13 antibodies were capable of inducing phagocytosis to a similar extent ( Fig. 6A ; mean Fab452 32.0%, mean Fab C 29.8%, mean WM-15 26.04%, n = 7). Although CD13-mediated phagocytosis of cells treated with WM-15 seems to be somewhat lower, we found no statistically significant difference from the phagocytosis of cells treated with Fab452. Of note, the 3 anti-CD13 antibodies showed similar levels of binding (Fig. 6B) . These results suggest that the epitope recognized by the individual anti-CD13 mAb is not important for CD13-mediated phagocytosis to occur. The same phenomenon has been described for CD13-mediated adhesion, where CD13 cross-linking by different anti-CD13 mAbs (452, WM-15, Y2-K, or WM-4.7) resulted in CD13-mediated adhesion, but none of them was able to induce adhesion when only Fabs, without secondary antibody, were used [19] . These data suggest that cross-linking of CD13 rather than the specific epitope recognized by a particular mAb, is the determinant factor during CD13-mediated adhesion and phagocytosis.

We also determined the effect of these different antibodies on CD13 enzymatic activity at concentrations used in phagocytosis assays. As control for the enzymatic activity determinations, we used bestatin, a well-characterized chemical inhibitor of aminopeptidase activity, which preferentially targets CD13. As shown in Fig. 6C , bestatin inhibited CD13 enzymatic activity in a dosedependent manner. In line with previous reports, treatment of cells with Fab452 resulted in a small, yet statistically significant enzymatic activity inhibition of 20% (mean Fab452 79.3%, n = 6), and treatment of cells with WM-15 inhibited CD13 enzymatic activity by 40% (mean WM-15 59.9%, n = 6). Treatment of cells with Fab C significantly inhibited CD13 enzymatic activity by 30% (mean FabC 67.1%, n = 6). As expected, treatment with Fab32.2 or IgG1 had no effect on enzymatic activity relative to control cells (mean Fab32.2 94.6%, mean IgG1 97.0%, n = 6). This experiment shows that the different anti-CD13 antibodies inhibit enzymatic activity to different extents, and this difference might be related to their binding to different epitopes, as suggested previously [18, 43] .

Several functions attributed to CD13 depend on its enzymatic (peptidase) activity [14] . However, as mentioned above, some other functions are independent of its enzymatic activity. Given that the 3 different antibodies anti-CD13 showed a different effect on enzymatic activity (Fig. 6C ), but we found no difference in phagocytosis induction by the 3 different antibodies (Fig. 6A) , our data suggest that CD13-mediated phagocytosis is independent of its enzymatic activity.

To ascertain whether the enzymatic activity of CD13 is required for the phagocytosis observed, we chemically blocked CD13 with bestatin and carried out phagocytosis in cells treated with Fab452. As shown in Fig. 6D , treatment of THP-1 cells with bestatin inhibited the enzymatic activity of CD13 by 40%. In contrast, phagocytosis through CD13 in cells treated with bestatin remained unaltered (Fig. 6E) . These data further support the notion that the CD13-mediated phagocytosis is not related to its enzymatic activity.

A powerful strategy to demonstrate the ability of a membrane receptor to mediate phagocytosis is to express the protein of interest in a different cellular context, usually in nonphagocytic cells, and to show that it confers phagocytic capacity to the cell. To further confirm the role of CD13 as a primary phagocytic receptor, we expressed CD13 in the epithelial cell line HEK293. These cells lack the expression of most phagocytic receptors and CD13. With the use of a lentiviral delivery system, we infected HEK293 cells with a vector to express hCD13 and RFP under separate promoters. As control, we excised hCD13 from the vector, leaving only the RFP construct. After selection and cloning, we obtained clones with high (clone A6) and medium (clone A4) hCD13 expression (Fig. 7A) . Cells infected with RFP virus (clone M6) did not express hCD13 (Fig. 7A) . As HEK293 cells grow as monolayers that attach to the plastic surface, and we noted that repeated washing disrupts the monolayers and affects cell viability, we modified the construction of phagocytic preys so that they carried the specific Fab [Fab anti-CD13 (EFab452) or Fab anti-FcgRI (EFab32.2) or no Fab (Ec; Supplemental Fig. 3) ]. With the use of this modified construction of the phagocytic prey, the cell monolayer is not subjected to repeated washing steps before incubation with the particles. We verified phagocytosis of these modified phagocytic preys in THP-1 cells by cytometry ( Fig. 7B ; n = 4) and confocal microscopy (not shown), obtaining similar results to those found with the use of the original construction of phagocytic preys. HEK293 hCD13 high (A6) and HEK293 hCD13 med (A4) clones are capable of internalizing Fab452-opsonized but not Fab32.2-opsonized phagocytic preys ( Fig. 7C ; n = 5). HEK293 hCD13 neg (M6) cells did not internalize any phagocytic prey. It is interesting to note that the extent of internalization of phagocytic preys correlated with the level of hCD13 expression, as HEK293 hCD13 high cells (A6) are more competent internalizing erythrocytes (28%) than HEK293 hCD13 med (A4;15%). Cytometry results were confirmed by confocal microscopy. As shown in Fig. 7D , we observed internalized erythrocytes in A6 cells only when they were opsonized with Fab anti-CD13 (lower left panels) but no internalization when erythrocytes were opsonized with anti-FcgRI (lower right panels). The upper panels in Fig. 7D show the same conditions for M6 cells. Importantly, this experiment allowed us to rule out the contribution of FcgRs and other major phagocytic receptors in phagocytosis mediated by CD13, as HEK293 cells lack expression of FcgRs, CR3 [44] [45] [46] , and most scavenger receptors [47, 48] . These results further support the role of CD13 as a receptor capable of inducing internalization, independently of other phagocytic receptors. Altogether, these data provide strong evidence that CD13 aggregation by interaction with particles mediates phagocytosis and that CD13-mediated phagocytosis is independent of FcgRs.

Phagocytes produce ROS conjunctly with phagocytosis or after stimulation with various agents. ROS generation has been implicated in a variety of physiologic responses and is a critical component of the antimicrobial repertoire of macrophages. Therefore, we investigated if CD13 aggregation results in production of ROS. To do so, we generated murine macrophage-like J774 cells expressing hCD13 by use of the system described above for HEK293 cells. As J774 cells express murine CD13 [49] and murine FcgRs, we first verified that the Fabs used for the phagocytosis assay did not bind to murine cells. Neither the Fab452 nor the Fab32.2 recognized molecules on the murine cells. After lentiviral infection, we obtained a stable cell line (R3) expressing high levels of hCD13 and a control cell line (M3) with no hCD13 expression (Fig. 8A) . In these cells, we measured the internalization of phagocytic preys opsonized with Fab anti-CD13 (EFab452) and Fab anti-FcgRI (EFab32.2) or without specific Fab (Ec). Whereas a high percentage of R3 cells phagocytosed EFab452 (26.5%), they were unable to phagocytose EFab32.2 or Ec, as expected (Fig. 8B ). M3 cells were unable to phagocytose any particle tested. R3 cells internalize erythrocytes very efficiently (Fig. 8C ) compared with internalization by hCD13 high HEK293 cells (Fig. 7D) , which might be explained by the fact that whereas J774 are professional phagocytes, HEK293 cells are not so.

Having demonstrated that in murine macrophage, like J774 cells, hCD13 functions as a fully phagocytic receptor, we investigated whether CD13 ligation is capable of inducing ROS. With the use of soluble ligands to induce hCD13 aggregation in R3 cells (hCD13 high ), we measured ROS generation by use of CM-H 2 DCFDA as indicator. We found that treatment with Fab anti-CD13 plus secondary F(ab)9 2 induces ROS in R3 but not in M3 cells ( Fig. 8D ; n = 5). Altogether, these results suggest that phagocytosis through CD13 leads to the generation of ROS that could potentially promote the degradation of ingested material.

We have shown previously that CD13 aggregation positively regulates phagocytosis mediated by distinct receptors in monocytes, macrophages, and monocyte-derived dendritic cells [33, 34] . However, whether CD13 functions only as an accessory receptor or whether it could act as a primary phagocytic receptor mediating internalization of particles remained unknown. In the present work, we provide strong evidence that CD13 can act as a primary phagocytic receptor even in the absence of other major phagocytic receptors. CD13-mediated phagocytosis depends on actin cytoskeleton rearrangement and on PI3K activity and is only partially dependent on Syk activity. A primary purpose of phagocytosis is to isolate, kill, and degrade microbial preys. In this work, we also provide data indicating that CD13-mediated phagocytosis induces production of ROS that may facilitate microbial killing.

CD13 was shown to mediate phagocytosis of large particles in the monocytic cell line THP-1 as well as in primary hMDMs. In these cells, CD13 is capable of mediating phagocytosis to an extent similar to that of the canonical phagocytic receptor FcgRI, as in our experimental system, the percentage of cells internalizing particles through both receptors is similar, and the number of particles ingested per cell is also similar.

CD13 can mediate internalization of large particles (4 mm), which strongly argues in favor of phagocytosis and excludes other forms of internalization. This is of particular importance, as CD13 has been implicated in the endocytosis of coronavirus [25, [50] [51] [52] , proteins [53] , targeted peptides [54] , or antibodies [25, 53] in several cells types. The reported forms of CD13-mediated endocytosis are caveolae-mediated endocytosis [25, 55] and clathrinmediated endocytosis [50, 56] . Therefore, it seems that CD13 can mediate different mechanisms of internalization, where the nature of the ligand and the cell type dictate the path of endocytosis. In this regard, CD13 is similar to other phagocytic receptors capable of mediating different forms of internalization, depending on the size of the ligand-bearing entity, such as FcgRs, mannose receptor, Dectin-1, and scavenger receptors.

The natural ligand(s) of CD13, capable of inducing its aggregation and the phagocytosis of large particles in vivo, are currently unknown. Both exogenous and endogenous polyvalent ligands for CD13 have been identified. Human CMV [17, 57] and an important subgroup of coronaviruses [15, 58] , including the HCoV-229E [52] , are exogenous polyvalent ligands for CD13 identified to date. The binding of HCoV-229E to human fibroblasts results in CD13 clustering, and more importantly, CD13 aggregation induced by HCoV-229E mirrors the consequences of antibody-induced CD13 aggregation regarding adhesion [19] and CD13 distribution on the cell membrane [25] . As for endogenous ligands, several lines of evidence suggest that galectin-3 might constitute an endogenous ligand for CD13. Yang et al. [59] used a human cDNA phage display biopanning method to identify proteins interacting with CD13 and found that 70% of the isolated clones encoded galectin-3, which suggests that galectin-3 is a likely endogenous ligand of CD13. Furthermore, in a study that uses affinity chromatography to identify ligands of galectin-3 in seminal fluid, CD13 was found to be 1 of the 9 proteins identified [60] . Moreover, galectin-3 coimmunoprecipitates with CD13 in resting U-937 cells [61] , and CD13 and galectin-3 colocalize in HUVECs after addition of exogenous galectin-3 [59] . All of these data suggest that galectin-3 is an endogenous ligand of CD13 in vivo. Galectin-3 is a b-galactosidebinding lectin, constitutively expressed and secreted by various cell types, including macrophages [62] . Galectin-3 expression is highly up-regulated during monocyte-to-macrophage differentiation [63] and during various infectious processes [64] [65] [66] , and galectin-3 is present in considerable amounts in the extracellular space at inflammatory sites [62, 65] . Extracellular galectin-3 can function in an autocrine or paracrine fashion to mediate cell-cell and cell-ECM interactions [67, 68] and monocyte and macrophage migration [69] and promotes cellular responses, such as respiratory burst [63, 70] and phagocytosis [65, 71, 72] . Importantly, galectin-3 binds to surface glycans of bacteria [73, 74] , protista [75, 76] , fungi [77, 78] , and parasites [72] . Thus, the potential role of galectin-3 as a pattern recognition receptor has become an area of intense attention. Galectin-3 can oligomerize, forming multivalent molecules capable of interacting with several ligands at a time and of cross-linking cell-surface glycoproteins [79] . Therefore, it is enticing to propose that galectin-3 might be functioning as an opsonin mediating the interaction of pathogens with CD13 and phagocytosis in vivo. In a recent paper, a role for galectin-3 as an opsonin enhancing the phagocytosis of apoptotic cells by macrophages has been proposed [71] , raising the possibility that the galectin-3-CD13 pair is functional in pathogenic and stressful settings. Further studies are needed to determine whether galectin-3 can act as an opsonin to promote CD13-mediated phagocytosis in vivo.

During phagocytosis mediated by FcgRs, IgG-opsonized particles induce receptor aggregation, resulting in Src kinasesmediated tyrosine phosphorylation of ITAMs in the cytoplasmic domains of the receptor chains. Syk is recruited to the phosphorylated ITAMs, and this binding promotes its tyrosine phosphorylation and activation. Syk activation is essential for the activation of downstream signaling molecules, including PI3K, ERK, Vav, and phospholipase C g, among others, which regulate cytoskeleton rearrangements and membrane-remodeling events necessary for pseudopod extension, phagosomal closure, and internalization.

Our results indicate that CD13-mediated phagocytosis is partially dependent on activation of Syk, as pretreatment of THP-1 cells or hMDMs with BAY resulted in an impairment of phagocytosis through CD13. This finding suggests that Syk might be activated by aggregation of CD13. Indeed, we were able to show that in hMDMs, cross-linking of CD13 with antibodies results in a rapid phosphorylation of Syk that peaks at 45 seconds and decreases rapidly. This fast phosphorylation/dephosphorylation kinetics could account for differences between these results and our previous studies, where we were unable to detect pSyk from 3 to 20 minutes after CD13 cross-linking in U-937 cells, assayed by immunoprecipitation, followed by Western blotting. Possibly, our inability to detect pSyk in our previous studies was related to Figure 8 . hCD13 expressed on murine macrophagelike cells is able to mediate phagocytosis and for ROS production. (A) J774 cells transduced with lentiviral particles containing plasmids for CD13 expression pLenti-suCMV(hANPEP)-Rsv(RFP-Bsd; clone R3) and transfection control pLenti-suCMV (empty)-Rsv(RFP-Bsd; clone M3) were incubated with 2 mg Fab fragments of mAb452 (anti-CD13) and 8 mg Fab fragments of mAb32.2 (anti-FcgRI) or without antibody for 30 minutes at 4°C. Cells then were washed and incubated with secondary goat anti-mouse FITC. A representative histogram of the expression of hCD13 on these clones is shown. (B) J774 clones R3 and M3 were incubated with EFab452, EFab32.2, or Ec for 120 minutes at 37°C. Noningested erythrocytes were lysed, and samples were analyzed by flow cytometry to determine the percentage of CFSE-positive cells. Average of CFSEpositive cells of 7 independent experiments is shown. ***P , 0.0001. (C) Confocal microscopy representative images of phagocytosis in J774 clones M3 and R3 (red) incubated with EFab452 and EFab32.2 (green). (D) J774 clones R3 and M3 were preloaded with CM-H 2 DCFDA for 20 minutes at 37°C. After washing, CD13 was cross-linked for 30 minutes at 37°C, as described above for MDMs, or incubated with PMA as control. Cells were analyzed immediately by flow cytometry. Average mean fluorescence intensity of 5 independent experiments relative to control is shown. ***P = 0.0009. differences in detection timing, cell type, and technique used [34] . The results in this paper show that CD13 aggregation induces rapid pSyk and that phagocytosis through CD13 likely involves activation of Syk as well as other early kinases that are necessary for the recruitment and activation of downstream signaling.

How Syk is activated after CD13 aggregation is currently unclear, as the short cytoplasmic tail of CD13 does not have a classic ITAM motif. However, it was recently shown that in the U-937 monocytic cell line, CD13 cross-linking by antibodies induced phosphorylation of several proteins, including FAK, ERK, and remarkably, the phosphorylation of CD13 on its cytoplasmic tail in a Src-dependent manner. The pharmacological inhibition of Src-, ERK-, or FAK-abrogated, CD13-dependent adhesion and tyrosine phosphorylation of CD13 was found to be essential for Src, ERK, and FAK activation and for CD13mediated adhesion [20] . This study identified the phosphorylation of the tyrosine residue in the cytoplasmic part of CD13 as a crucial event in CD13-mediated intracellular signaling, and it is possible that this phosphorylation event is in some way related to the activation of Syk observed by us after CD13 aggregation. This same study reported that anti-CD13 antibody-induced adhesion does not depend on Syk activation, as treatment of U-937 with piceatannol (10 mM) did not impair adhesion [20] . These results are in agreement with our results that use the Syk inhibitor, BAY, where we found that CD13-mediated phagocytosis is impaired but able to proceed in the absence of Syk activation. The partial effect of Syk inhibition by BAY on CD13-mediated phagocytosis could suggest that other kinases, not inhibited by BAY but inhibited by piceatannol, participate along with Syk in CD13mediated phagocytosis. The identity of the other kinases involved in CD13-mediated phagocytosis is currently unknown. However, given that CD13 aggregation recruits and activates FAK [20] , it seems plausible that FAK might be responsible for PI3K recruitment and activation during CD13-mediated phagocytosis, as has been observed in signaling by CR3 and other integrins.

PI3K activation appears to be essential for the extension of pseudopods around the ingested particle and for phagosomal closure during phagocytosis of most phagocytic receptors. We found that CD13-mediated phagocytosis also depends on PI3K, as treatment of THP-1 cells or hMDMs with the PI3K inhibitor LY strongly impaired phagocytosis through CD13.

All phagocytic processes are driven by a finely controlled rearrangement of the actin cytoskeleton, and we also showed that CD13-mediated phagocytosis depends on actin cytoskeleton rearrangements. It has been shown previously that CD13 crosslinking by antibodies induces actin rearrangements leading to adhesion [19, 20] . FcgR-and CR3-mediated phagocytosis are accompanied by the local recruitment of a-actinin [80] , a protein that can bundle actin filaments. In this way, a-actinin links the force generated by the actin network to the bound particle. In this regard, CD13 aggregation by antibodies leads to recruitment of a-actinin, linking membrane-bound CD13 to cytoskeleton [20] . Therefore, we propose that during phagocytosis through CD13, particle-induced CD13 cross-linking induces actin rearrangements and recruitment of a-actinin to link the CD13-bound particle to the cytoskeleton.

In this work, we started dissecting the signaling events involved in CD13-mediated phagocytosis. Although much remains to be defined, we found that CD13-mediated phagocytosis involves activation of Syk as well as other initiating kinases, PI3K, and actin cytoskeleton rearrangements. As the signaling pathways involved in CD13-mediated phagocytosis coincide with the signaling pathways of most phagocytic receptors, it is plausible that in vivo CD13-induced signals can synergize in vivo with those activated by other phagocytic receptors, as reported previously in in vitro systems [33, 34] .

Our present work provides compelling evidence that CD13 is a primary phagocytic receptor, fully capable of inducing phagocytosis independently of other receptors. We demonstrated that the expression of CD13 on nonphagocytic HEK293 cells is sufficient to enable them to phagocytose through CD13. Moreover, as these cells do not express FcgRs (nor most phagocytic receptors), these data further support the notion that CD13 is a bona fide phagocytic receptor. It has been shown that CD13 and FcgRI coredistribute on the cell membrane after the cross-linking of either one of them in monocytes and macrophages [34] . Therefore, one concerning possibility existed: that particle-induced CD13 aggregation during phagocytosis would also aggregate FcgRI, and the observed CD13-mediated internalization of particles resulted from the signaling triggered by aggregation of FcgRI, making CD13-induced phagocytosis dependent on FcgRI. As HEK293 cells do not express FcgRs, it is clear that CD13 does not depend on FcgRI to mediate phagocytosis, as HEK293 cells expressing hCD13 internalize erythrocytes through CD13 very efficiently. These data show that CD13 is a primary phagocytic receptor, capable of acting independently of other phagocytic receptors. However, it is extremely difficult to rule out completely the possibility that other unknown cosignaling molecules expressed on HEK293 can be participating in CD13-mediated phagocytosis. Under physiologic conditions, several types of receptors are activated in parallel, and multiple signaling cascades are triggered concomitantly. Hence, it is possible that in more natural settings, where CD13 is expressed on the cell membrane along with other phagocytic receptors sharing common signaling pathways with CD13, phagocytosis, through CD13, might be regulated by other receptors engaged by the phagocytic prey, just like we have shown that CD13 modulates phagocytosis by other receptors [33, 34] . CD13-deficient THP-1 cells or bone marrowderived macrophages from CD13-deficient mice [35] showed no alteration in phagocytosis through FcgRs, demonstrating that although CD13 aggregation enhances phagocytosis through FcgRs, this phagocytosis is not dependent on CD13. Likewise, phagocytosis through CD13 is not dependent on FcgRI, although it might be enhanced by the coaggregation of FcgRI.

In summary, our current observations provide evidence that CD13 is a primary phagocytic receptor. The signaling involved in CD13-mediated phagocytosis shares common elements with other phagocytic receptors, which might allow for cooperation between CD13 and other receptors in the signaling and internalization of prey. 

The cell biology of phagocytosis

Macrophage receptors and immune recognition

Macrophage cytoskeleton association with CR3 and CR4 regulates receptor mobility and phagocytosis of iC3b-opsonized erythrocytes

Cellular invasion by Staphylococcus aureus involves a fibronectin bridge between the bacterial fibronectin-binding MSCRAMMs and host cell beta1 integrins

Identification of a factor that links apoptotic cells to phagocytes

Novel role of ICAM3 and LFA-1 in the clearance of apoptotic neutrophils by human macrophages

Protection from lethal gram-positive infection by macrophage scavenger receptor-dependent phagocytosis

CD36 and TLR interactions in inflammation and phagocytosis: implications for malaria

Regulation and functional involvement of macrophage scavenger receptor MARCO in clearance of bacteria in vivo

CD44 is a phagocytic receptor

Granulocyte CEACAM3 is a phagocytic receptor of the innate immune system that mediates recognition and elimination of human-specific pathogens

Membrane bound members of the M1 family: more than aminopeptidases

CD13-not just a marker in leukemia typing

The moonlighting enzyme CD13: old and new functions to target

Molecular determinants of species specificity in the coronavirus receptor aminopeptidase N (CD13): influence of N-linked glycosylation

Mutational analysis of aminopeptidase N, a receptor for several group 1 coronaviruses, identifies key determinants of viral host range

CD13 (human aminopeptidase N) mediates human cytomegalovirus infection

CD13 in cell adhesion: aminopeptidase N (CD13) mediates homotypic aggregation of monocytic cells

CD13 is a novel mediator of monocytic/endothelial cell adhesion

Tyrosine phosphorylation of CD13 regulates inflammatory cell-cell adhesion and monocyte trafficking

Aminopeptidase N/CD13 is associated with raft membrane microdomains in monocytes

Enzymatic activity is not a precondition for the intracellular calcium increase mediated by mAbs specific for aminopeptidase N/CD13

Human cytomegalovirus induces inhibition of macrophage differentiation by binding to human aminopeptidase N/CD13

Aminopeptidase N/CD13 is directly linked to signal transduction pathways in monocytes

Human coronavirus 229E binds to CD13 in rafts and enters the cell through caveolae

CD13 is essential for inflammatory trafficking and infarct healing following permanent coronary artery occlusion in mice

ICAM-1 (CD54): a counter-receptor for Mac-1 (CD11b/CD18)

Complement receptor type three (CD11b/CD18) of human polymorphonuclear leukocytes recognizes fibrinogen

CD44 and its interaction with extracellular matrix

CD36 gene transfer confers capacity for phagocytosis of cells undergoing apoptosis

Integrin-dependent phagocytosis: spreading from microadhesion to new concepts

) 1,25-Dihydroxyvitamin D3 stimulates phagocytosis but suppresses HLA-DR and CD13 antigen expression in human mononuclear phagocytes

Aminopeptidase N (CD13) is involved in phagocytic processes in human dendritic cells and macrophages

Aminopeptidase N (CD13) functionally interacts with FcgammaRs in human monocytes

CD13 is dispensable for normal hematopoiesis and myeloid cell functions in the mouse

The Syk protein tyrosine kinase is essential for Fcgamma receptor signaling in macrophages and neutrophils

Protein-tyrosine kinase Syk is required for pathogen engulfment in complement-mediated phagocytosis

Dectin-1 activates Syk tyrosine kinase in a dynamic subset of macrophages for reactive oxygen production

Dectin-1 uses novel mechanisms for yeast phagocytosis in macrophages

Phosphoinositide 3-kinase is a novel target of piceatannol for inhibiting PDGF-BB-induced proliferation and migration in human aortic smooth muscle cells

Role of Src kinases and Syk in Fcgamma receptor-mediated phagocytosis and phagosome-lysosome fusion

Syk and Lyn phosphorylation induced by FcgammaRI and FgammaRII crosslinking is determined by the differentiation state of U-937 monocytic cells

Deletion of the zincbinding motif of CD13/aminopeptidase N molecules results in loss of epitopes that mediate binding of inhibitory antibodies

The apoptotic-cell receptor CR3, but not alphavbeta5, is a regulator of human dendritic-cell immunostimulatory function

Involvement of complement receptor 3 (CR3) and scavenger receptor in macrophage responses to wear debris

Role of the beta-subunit arginine/lysine finger in integrin heterodimer formation and function

Class B scavenger receptor types I and II and CD36 mediate bacterial recognition and proinflammatory signaling induced by Escherichia coli, lipopolysaccharide, and cytosolic chaperonin 60

) MARCO, TLR2, and CD14 are required for macrophage cytokine responses to mycobacterial trehalose dimycolate and Mycobacterium tuberculosis

Expression, regulation and functional activities of aminopeptidase N (EC 3.4.11.2; APN; CD13) on murine macrophage J774 cell line

The coronavirus transmissible gastroenteritis virus causes infection after receptor-mediated endocytosis and acid-dependent fusion with an intracellular compartment

Cells of human aminopeptidase N (CD13) transgenic mice are infected by human coronavirus-229E in vitro, but not in vivo

Human aminopeptidase N is a receptor for human coronavirus 229E

Targeting aminopeptidase N, a newly identified receptor for F4ac fimbriae, enhances the intestinal mucosal immune response

Peptides for cell-selective drug delivery

The reversion-inducing cysteine-rich protein with Kazal motifs (RECK) interacts with membrane type 1 matrix metalloproteinase and CD13/aminopeptidase N and modulates their endocytic pathways

Basolateral internalization of GPI-anchored proteins occurs via a clathrinindependent flotillin-dependent pathway in polarized hepatic cells

Productive cytomegalovirus (CMV) infection exclusively in CD13-positive peripheral blood mononuclear cells from CMV-infected individuals: implications for prevention of CMV transmission

Aminopeptidase N is a major receptor for the enteropathogenic coronavirus TGEV

Aminopeptidase N/CD13 induces angiogenesis through interaction with a pro-angiogenic protein, galectin-3

Investigation of galectin-3 function in the reproductive tract by identification of binding ligands in human seminal plasma

A role for galectin-3 in CD13-mediated homotypic aggregation of monocytes

Regulation of secretion and surface expression of Mac-2, a galactoside-binding protein of macrophages

Expression and function of galectin-3, a beta-galactosidebinding lectin, in human monocytes and macrophages

Role of galectin-3 as an adhesion molecule for neutrophil extravasation during streptococcal pneumonia

Galectin-3 reduces the severity of pneumococcal pneumonia by augmenting neutrophil function

Upregulation of galectin-3 and its ligands by Trypanosoma cruzi infection with modulation of adhesion and migration of murine dendritic cells

Interactions between galectin-3 and Mac-2-binding protein mediate cell-cell adhesion

Galectin-3 promotes adhesion of human neutrophils to laminin

Human galectin-3 is a novel chemoattractant for monocytes and macrophages

Galectin-3 activates the NADPH-oxidase in exudated but not peripheral blood neutrophils

Galectin-3 functions as an opsonin and enhances the macrophage clearance of apoptotic neutrophils

LacdiNAc-glycans constitute a parasite pattern for galectin-3-mediated immune recognition

Pseudomonas aeruginosa lipopolysaccharide binds galectin-3 and other human corneal epithelial proteins

Galectin-3 binds to Helicobacter pylori O-antigen: it is upregulated and rapidly secreted by gastric epithelial cells in response to H. pylori adhesion

Specific recognition and cleavage of galectin-3 by Leishmania major through species-specific polygalactose epitope

Novel mechanism that Trypanosoma cruzi uses to adhere to the extracellular matrix mediated by human galectin-3

beta-1,2-Linked oligomannosides from Candida albicans bind to a 32-kilodalton macrophage membrane protein homologous to the mammalian lectin galectin-3

Galectin-3 induces death of Candida species expressing specific beta-1,2-linked mannans

Seeing strangers or announcing "danger": galectin-3 in two models of innate immunity

Molecular definition of distinct cytoskeletal structures involved in complement-and Fc receptormediated phagocytosis in macrophages

KEY WORDS: aminopeptidase N • macrophages • phagocytic receptor