CD13 Restricts TLR4 Endocytic Signal Transduction in Inflammation

Abstract

Dysregulation of the innate immune response underlies numerous pathological conditions. The Toll-like Receptor 4 (TLR4) is the prototypical sensor of infection or injury that orchestrates the innate response via sequential activation of both cell-surface and endocytic signaling pathways that trigger distinct downstream consequences. CD14 binds and delivers LPS to TLR4 and has been identified as a positive regulator of TLR4 signal transduction. It is logical that negative regulators of this process also exist to maintain the critical balance required for fighting infection, healing damaged tissue and resolving inflammation. We showed that CD13 negatively modulates receptor-mediated Ag uptake in dendritic cells to control T-cell activation in adaptive immunity. Here we report that myeloid CD13 governs internalization of TLR4 and subsequent innate signaling cascades, activating IRF-3 independently of CD14. CD13 is co-internalized with TLR4, CD14 and dynamin into Rab5⁺ early endosomes upon LPS treatment. Importantly, in response to TLR4 ligands HMGB1 and LPS, pIRF-3 activation and transcription of its target genes is enhanced in CD13^KO DCs while TLR4 surface signaling remains unaffected, resulting in a skewed inflammatory response. This finding is physiologically relevant as ischemic injury in vivo provoked identical TLR4 responses. Finally, CD13^KO mice showed significantly enhanced IFNβ-mediated signal transduction via JAK-STAT, escalating iNOS transcription levels and promoting accumulation of oxidative stress mediators and tissue injury. Mechanistically, inflammatory activation of macrophages upregulates CD13 expression and CD13 and TLR4 co-immunoprecipitate. Therefore, CD13 negatively regulates TLR4 signaling, thereby balancing the innate response by maintaining the inflammatory equilibrium critical to innate immune regulation.

Introduction

In response to injury or infection, danger signals and pathogenic stimuli are sensed by germline-encoded families of receptors Pattern Recognition Receptors (PRRs) to initiate the tightly controlled innate immune response. Dysregulation of this response can rapidly develop into systemic inflammation with uncontrolled and overwhelming release of proinflammatory mediators, triggering extensive tissue injury and eventual multiple organ failure. Advanced conditions have an unusually high mortality rate and unfortunately, despite medical advances its incidence is increasing dramatically in the US. Thus identification of novel mechanisms to constrain the response to infection or injury and temper development and progression of sepsis is critical.

Shared signals displayed by invading pathogens (Pathogen-associated Molecular Patterns or PAMPs) are recognized by PRRs to initiate signal transduction pathways to fight infection. Importantly, these PRRs also recognize endogenous signals of tissue injury (Danger-associated Molecular Patterns, DAMPs) released by dead and dying cells and trigger largely identical immune responses to remove dead cells and heal the wound. Members of the prototypical Toll-like Receptor (TLR) family recognize subclasses of bacterial and viral pathogens and include TLR4 that recognizes the Gram-negative bacterial cell wall LPS and a number of endogenous ligands released from injured tissue to activate the immune response. In response to infection, the absence of TLR4 activation results in death due to septicemia, while excessive activation can lead to serious complications of dysregulated inflammation and further tissue damage. Similarly, repair of sterile tissue damage also requires a well-controlled innate immune response in diseases such as myocardial infarction, stroke and peripheral artery disease (1–3). Therefore, tight control of TLR4 signaling is imperative for a balanced and effective immune response.

Ligation of TLR4 induces two independent signaling pathways initiated from different cellular locations (4, 5). The specific localization of TLR4 to these compartments is critical to the regulation of innate immunity in response to injury and infection, but the mechanisms that oversee receptor passage along various channels are not well understood. TLR4 engagement at the cell-surface triggers the TIRAP-MyD88 mediated pathway to activate p38MAPK and IKKβ-α-inflammatory cytokine genes (4–6). The TLR4-ligand complex is then internalized via a dynamin-dependent process where it initiates signaling from the endosome, eliciting a TRAM-TRIF-dependent response to promote transcription of type I IFNs via the interferon regulatory factor, IRF3 (4, 5). Recent studies have shown that aberrant signaling from either of these pathways result in both pro- and anti-healing effects in inflammatory disease models (7, 8).

The CD13/Aminopeptidase N membrane metallopeptidase is highly expressed on myeloid cells and activated endothelial cells and participates in various innate and adaptive immune mechanisms (9–14). Recently we found that CD13 negatively regulates receptor-mediated, dynamin-dependent antigen uptake by DCs to control the adaptive immune response (9). While DC development, maturation, antigen processing and presentation were normal in CD13^KO mice, lack of CD13 significantly enhanced cytotoxic T cell activation due to increased antigen internalization. Because the TLR4 endocytic signaling pathway requires TLR4 internalization in endosomes, we hypothesized that CD13 may also regulate this essential immune bridge between innate and adaptive immunity. In the current investigation we explored the role of CD13 in regulation of TLR4 signal transduction pathways. Our results demonstrate that CD13 regulates the balance between pro-inflammatory and type I IFN-generating signal transduction in myeloid cells, thus identifying myeloid CD13 as a functional regulator of innate immunity. These observations increase our understanding of the immune response to injury and infection, information essential to deciphering the pathways and endogenous mediators regulating ligand-induced receptor transport. This knowledge is fundamental to understanding the capabilities and limitations of innate immune detection systems involving TLR-mediated signal transduction pathways and to the design of strategies to promote healing by manipulating novel regulatory molecules.

Materials and Methods

Mice

Global CD13^KO mice were generated at the Gene Targeting and Transgenic Facility as published (11) and crossed for 10 generations to the FVB/N background. All animals were housed at the University of Connecticut Health Center animal facilities and all procedures were performed in accordance with Institutional and Office of Laboratory Animal Welfare guidelines.

Generation of bone marrow-derived dendritic cells (BMDC) and bone marrow-derived macrophages (BMDM)

Total bone marrow cells were obtained by flushing the femur and tibia, followed by lysis of red blood cells and BMDCs were isolated as described (9). BMDM were isolated by culturing for 6 days in DMEM medium supplemented with 10% fetal bovine serum and antibiotic cocktail containing penicillin and streptomycin and 20ng/ml M-CSF. CD8⁺ CD11c⁺ splenic DCs were isolated as described (9).

Stimulation of myeloid cells with LPS or HMGB1

BMDMs or BMDCs were serum starved in plain DMEM medium for 2h followed by stimulation with 100ng/ml LPS (from E. coli, serotype 055:B5, TLRgrade; Enzo Life Sciences) or 3μg/ml purified HMGB1 (Endotoxin content<0.1EU/μg protein from E. coli; Adipogen) for 0–90 min at 37°C.

Flow cytometry

Primary macrophages or DCs were treated with LPS for 0–40 min at 37°C. 0.5×10⁶ cells were fixed with 2% paraformaldehyde, permeabilized with 0.1% TritonX-100, blocked with 5% normal goat serum and stained with rat anti-mouse TLR4 monoclonal antibody (1:500), Sa15-21 or rat anti-mouse CD14 monoclonal antibody (1:500) (clone Sa14-2; Hycult Biotech), at 4°C for 30 min followed by secondary goat anti-rat Alexa 488 antibody (1:5000). Rat IgG-Alexa 488 (1:500) was used as isotype control. Flow cytometry was performed on live cells using LSRII (Becton Dickinson) and the data analyzed with FlowJo software (Tree Star).

Quantitative RT-PCR analysis

RNA was isolated using Trizol according to manufacturer’s instructions (Invitrogen Corporation, Carlsbad, CA). Results were expressed relative to the internal control gene glyceraldehyde-3-phosphate dehydrogenase. Sequence of PCR primers were either obtained from the PrimerBank database (http://pga.mgh.harvard.edu/primerbank/) or designed with Primer Express software (Applied Biosystems).

Western blot analysis

Cell lysates from primary macrophages or muscles from ischemic murine hind limbs were separated by SDS-PAGE and probed for indicated proteins. GAPDH, β-actin or tubulin were used as loading controls.

Co-immunoprecipitation

BMDM cell lysate was immunoprecipitated with protein G-conjugated agarose beads (Invitrogen) by constant rotation for 1 h at 4°C. The beads were washed and treated with control IgG or CD13 monoclonal Ab. The antigen-antibody complex was mixed on rotating shaker at 4°C for overnight. The precipitated protein was analyzed by immunoblot with either anti-TLR4 or anti-CD13 antibodies.

Immunofluorescence Analysis

0.5 × 10⁶ BMDM were grown on coverslips and serum starved for 2h in plain DMEM medium. Cells were treated with either vehicle or 100ng/ml LPS for various time intervals. Cells were fixed in 4% paraformaldehyde solution for 30 min at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature, washed and blocked with normal donkey serum for 30 min at room temperature followed by rat anti-mouse CD13 monoclonal Ab (1:1000) in 5% BSA/PBS, overnight at 4 degree. The cells were washed and treated with donkey anti-rat Alexa 594 as secondary Ab (1:1000) for 1h at room temperature. Cells were washed, blocked in normal goat serum, stained with rat anti-mouse TLR4 antibody (1:2000) or rat anti-mouse CD14 Ab (1:1000) for 2h at room temperature then treated with secondary goat anti-rat Alexa 488 (1:5000) for 1h at room temperature. For endosomal proteins, cells were blocked with normal goat serum, followed by addition of rabbit anti-mouse Rab5 or Rab11a or Dynamin monoclonal Abs for 2h at room temperature (1:500) then treated with goat anti-rabbit Alexa 488 as secondary Ab (1:5000). Unless otherwise stated, all primary and secondary antibodies were purchased from Cell Signaling and Molecular Probes respectively. All antibody dilution were made in 5% BSA/PBS. For all co-localization studies, staining were performed sequentially followed by treatment with TOPRO3 (nuclear stain), mounted with Prolong gold antifade mounting medium (Life Technologies) and analyzed by a Zeiss Axiovert fluorescence confocal microscope. Specimens were visualized at the excitation wavelength of 488 nm for Alexa 488, 543 nm for Alexa 594 and 633 nm for Topro-3. Images were photographed with Optronics camera attached to Zeiss Axioskop 2 plus microscope using the Zeiss Achroplan 63X or 100X; oil objective and photographed with Axiocam MRC. Images were analyzed by LSM software (Zeiss) and quantified by MetaMorph software (Molecular Devices).

Pulse chase experiment

LPS treated cells were pulsed in the presence of rat anti-CD13 antibody or rat anti-TLR4 antibody at 4°C followed by a chase at 37°C for 30 min in the presence of LPS in plain DMEM medium, allowing internalization of antibody-bound CD13 or TLR4. At different time intervals cells were fixed, permeabilized and stained with rabbit anti-Rab11 Ab followed by goat anti-rabbit Alexa 488 and subjected to confocal imaging.

Uptake of LPS-FITC

CD11c⁺ BMDC or CD8⁺ CD11c⁺ Splenic DCs treated with different doses of FITC-labeled LPS for 45 min at 37°C. DCs treated with vehicle served as controls. To examine the amount of ligand internalized, cells were briefly washed in Glycine/NaCl buffer, pH 2.5 twice, to eliminate surface fluorescence and stained with anti-CD11c Ab or anti-CD8 antibodies and % live CD11c⁺ LPS-FITC⁺ or CD11c⁺ CD8⁺ LPS-FITC⁺ cells were analyzed by flow cytometry. Cells were incubated at 4°C to determine the extent of nonspecific binding which was deducted from the experimental intracellular uptake.

Uptake of irradiated OVA-splenocytes

Splenocytes from wild type mice were incubated with OVA-FITC, UV-irradiated, and increasing doses of OVA-loaded splenocytes were incubated with wild type or CD13^KO CD11c⁺ CD8⁺ splenic DC for 45 min at 37°C. Cells were washed and uptake by CD11c⁺ CD8⁺ OVA-FITC⁺ DCs was measured by flow cytometry.

Nitrite measurement

Nitric oxide is converted to nitrate and nitrite in the cell and released into the media. BMDM were treated with LPS and supernatants collected. Nitrite concentrations in the media or serum were determined using the Griess reagent according to manufacturer’s protocol and sodium nitrite as a reference standard (Abcam). Optical density was measured at 540 nm.

Femoral artery ligation

The femoral artery was dissected in age- and sex- matched wild type and CD13KO mice and muscles isolated at 3 days following ischemic injury as described (13).

Statistical Analysis

Results are presented as mean±SEM. Statistical analysis was performed using unpaired, two-tailed t test. Differences were considered significant at p<0.05.

Results

Loss of CD13 decreases TLR4 surface expression and increases LPS uptake in LPS-stimulated splenic DCs

Ligand-induced receptor endocytosis is a fundamental cellular mechanism that regulates diverse functions such as uptake of essential metabolites, control of receptor availability and redirection of signal transduction pathways (15). In innate immunity, TLR4 endosomal signaling is critical to maintain the balance between pro- and anti-inflammatory responses to tissue injury and infection (4). Based on our previous findings that CD13 regulates dynamin-dependent, mannose receptor-mediated antigen uptake in DCs to shape the adaptive immune response (9), we hypothesized that CD13 may mediate innate immunity as well by controlling ligand-induced TLR4 endocytosis. While TLR4 surface expression was equivalent in unstimulated wild type and CD13^KO splenic DCs by flow cytometric and immunofluorescence analysis (Fig. 1A–C) and Supplemental Fig. S1B, loss of TLR4 surface expression was higher over time in CD13^KO compared to wild type DCs upon treatment with the TLR4 ligand LPS (Fig. 1B, C). Similarly, flow cytometric analysis illustrated that BMDCs or CD8⁺ splenic DCs from CD13^KO mice were significantly more efficient in the internalization of labeled LPS over a range of doses shown to induce TLR4 endocytosis (6) (0.05–1μg/ml, Fig. 1D and data not shown) suggesting that endocytic internalization of TLR4 and its ligand LPS is increased in the absence of CD13. Importantly, pretreatment of cells with 100μg/ml bestatin, an inhibitor of CD13 peptidase activity showed no significant effect on LPS-induced TLR4 endocytosis or uptake of LPS-FITC in either genotype (Fig. 1B, D), suggesting that CD13 regulation of TLR4 internalization is independent of its peptidase function.

Figure 1. LPS-induced TLR4 endocytosis is increased in CD13KO BMDM.

(A). Representative histogram of TLR4 surface expression on wild type and CD13^KO BMDM at basal level assessed with rat anti-mouse TLR4-specific antibody Sa15-21 followed by goat anti-rat Alexa 488 as secondary and data analyzed by flow cytometry. Cells with no staining, cells treated with secondary Ab only or Rat IgG-Alexa 488 isotype served as controls. (B) LPS induced TLR4 endocytosis in CD13WT and CD13^KO splenic DCs stimulated with either vehicle or LPS (100 ng/ml) for indicated time intervals. Cells were pretreated with 100μg/ml Bestatin (BE) for 30 min followed by LPS addition. a; CD13WT+LPS/20 min vs. CD13^KO +LPS/20 min, b; CD13WT+LPS/40 min vs. CD13^KO +LPS/40 min (C) Immunofluorescence analysis of TLR4 expression in response to LPS for 30 min. Cells were fixed, permeabilized and stained with Sa15-21 antibody. TLR4 (green) and TOPRO3; nuclear (blue). Original magnification 63x oil. Arrows indicate internalization. Scale bar; 10μM. (D) CD13WT and CD13^KO DCs were treated with or without 100μg/ml Bestatin (BE) for 30 min followed by addition of indicated doses of LPS-FITC for 45 min and % LPS⁺ CD11c⁺ cells were measured by flow cytometry. a, b, c, d; CD13WT/LPS vs. CD13^KO/LPS at doses 0.05, 0.1, 0.5, 1.0 μg/ml LPS-FITC respectively. E) Immunofluorescence analysis of CD13WT BMDM treated with LPS for 30 min followed by staining with CD13, TLR4 and other endosomal markers as depicted. (a) Vehicle treated CD13 (red) and TLR4 (green). (b) LPS treated CD13 (red) and TLR4 (green) or CD13 (red) and Dynamin (green). (c) LPS treated TLR4 (red) and Rab5 (green) or CD13 (red) and Rab5 (green). CD13, TLR4, Dynamin and Rab5 co-internalize under similar experimental conditions. (d) TLR4 (red) or CD13 (red) and Rab11a (green) co-localize upon LPS stimulation in a pulse-chase experiment. TOPRO3; nuclear (blue). Original magnification 100x oil. Arrows indicate co-localization. Scale bar; 10μM (F) Bars represent average Pearson’s correlation coefficient (r) between proteins as indicated from individual cell in each image (# of images=10) from three independent experiments. Data represent % Cell ± SEM (n = 3 isolates/genotype, three independent experiments). *p < 0.05; **p < 0.01.

LPS treatment induced endosomal co-localization of CD13 with TLR4

LPS stimulation has been shown to induce accumulation of TLR4 in endosomes (16). To determine if CD13 undergoes a similar fate following LPS ligation, we treated wild type macrophages with LPS and assessed CD13 localization by immunofluorescence using antibodies to TLR4 and markers of the various endosomal compartments. In contrast to vehicle treatment (Fig. 1Ea), LPS stimulation indeed led to internalization of CD13 in TLR4⁺ and dynamin⁺ vesicles (Fig. 1Eb). Furthermore, similar to TLR4, CD13 was found in Rab5⁺ early endosomes [Fig. 1Ec, and ref. (4)]. Importantly, pulse-chase experiments with anti-TLR4 or anti-CD13 antibodies demonstrated that surface-labeled TLR4 or CD13 trafficked to Rab11⁺ recycling endosomes, peaking at 30 minutes following LPS treatment (Fig. 1Ed), suggesting that CD13 may participate in receptor recycling as well. Collectively, this data is consistent with our hypothesis that CD13 regulates TLR4 endocytosis and suggesting that CD13 may also impact TLR4 endosomal signal transduction pathways. Quantification of images by MetaMorph software indicated that compared to vehicle treatment (r=0.3), a strong positive correlation (r=0.5–0.7) between proteins upon LPS stimulation was observed as determined by Pearson’s analysis (Fig. 1F). Furthermore, microscopic images indicate that the majority of CD13⁺ or TLR4⁺ vesicles are also Rab5⁺ or Rab11a⁺ while Rab5⁺ or Rab11a⁺ vesicles do not always contain CD13 or TLR4, probably indicating that the system is not saturated.

CD13 controls LPS induced TLR4 endosomal signal transduction

In DCs, blocking TLR4 internalization directly impairs the TRIF-dependent endosomal signal transduction pathway while MyD88-mediated TNF-production is unaffected, suggesting that TRIF-dependent signaling is primarily regulated at the level of endocytosis (6). Since LPS uptake is increased in the absence of CD13, we predicted that the alternate TRIF-dependent endosomal signaling would be preferentially induced in CD13^KO myeloid cells. Immunoblot analysis of lysates from wild type and CD13^KO macrophages (Fig. 2A, C) or DCs (Fig. 2B, D) demonstrated that lack of CD13 in either cell type resulted in a significant activation of the alternate pathway-specific transcription factor, IRF3 in response to LPS in a time dependent manner. Pretreatment of LPS-stimulated DCs with Dynasore, a small molecule inhibitor of the scission GTPase dynamin (17), abrogated the increase suggesting that CD13 plays a role at the level of dynamin-dependent endocytosis (Fig. 2B, D). Alternatively, pretreatment of wild type macrophages with the CD13 blocking mAb SL13 prior to LPS induction recapitulated the increased IRF3 activation seen in CD13^KO cells, further implicating CD13 in the endocytic process (Fig. 2E). Finally, the rise in IRF3 phosphorylation correlates with functional hyperactivation in CD13^KO DCs as evidenced by increases in mRNA levels of the pIRF3 target genes IFNβ and IP-10 in response to LPS [Fig. 2F, G and refs (18, 19)]. In addition to pathogenic ligands such as LPS, TLR4 also responds to molecules released by dead and dying cells, DAMPs. The predominant TLR4-binding DAMP is the nuclear protein HMGB1, which is secreted by inflammatory macrophages and monocytes (20). Importantly, CD13^KO cells treated with purified, endotoxin-free HMGB1 protein also showed significantly increased levels of IRF3 targets IFNβ and IP-10 (Figs. 2F and G, respectively), which was abolished by pretreatment with Dynasore (Fig. 2G). Collectively, these results indicate that CD13 modulates the uptake of TLR4 and its ligands LPS and HMGB1 in a dynamin-dependent manner, thereby controlling the innate immune response to pathogenic or danger-associated antigens.

Figure 2. Lack of CD13 enhances TLR4 signaling via the alternate endocytic pathway.

Wild type and CD13^KO BMDM (A) or splenic DCs (B) or CD13 mAb, SL13 treated wild type BMDM (E) treated with either vehicle, 100ng/ml LPS or pretreated with 80μM Dynasore, followed by LPS stimulation for 30 min as indicated. Cells were lysed and immunoblotted for pIRF3 levels. (C) and (D) represent quantification of (A) and (B) (pIRF3/IRF3) respectively. (F, G) Quantitative RT-PCR analysis of pIRF3 targets, IFN-β and IP-10 mRNA normalized to GAPDH in wild type and CD13^KO BMDM treated with Vehicle (V), 100ng/ml LPS (L) or 3μgml/ml HMGB1 (H) or Dynasore (D) or pretreated with Dynasore followed by LPS (D+L) or HMGB1 (D+H) for indicated time period. Data represent relative fold change of CD13^KO and wild type mRNA levels ± SEM (n = 3 isolates/genotype, three independent experiments). **p < 0.01; *p < 0.05.

MyD88-dependent TLR4 surface signaling is independent of CD13

MyD88-dependent TLR4 signal transduction from the plasma membrane is independent of its internalization (6) and thus we would expect that loss of CD13 would have a minimal effect on this primarily pro-inflammatory cascade. To address this issue, we analyzed p-p38/MAPK levels in wild type and CD13^KO BMDMs and found that levels of activated ERK, p38 and Akt were equivalent in response to LPS over time regardless of genotype (Fig. 3A–D). As expected, the increase in phosphorylated-p38 was not sensitive to Dynasore, further confirming that TLR4 signaling from the cell surface is independent of dynamin and is not mediated by CD13 (Fig. 3C). However, Akt phosphorylation was blocked by Dynasore treatment (Fig. 3D), consistent with the late phase activation of Akt via dynamin-dependent endosomal signaling described in previous studies (4, 5). Indeed, phosphorylation of the upstream mediator of NF-κB/MAPK signaling, pIRAK4, was comparable between genotypes (Figure 3E) in agreement with pERK and pp38 levels and supporting late-phase Akt activation (4, 5). Finally, quantitative RT-PCR analysis indicated a time dependent increase in the mRNA levels of downstream targets IL-6, TNF-α, IKKβ-α and IL-1β in response to either LPS or HMGB1 in cells from either genotype (Figs. 3F and S1A). In contrast to endosomal targets, there was no significant difference in the levels of NFκB/MAPK downstream target genes between wild type and CD13^KO BMDM at time points prior to 40 minutes. However, later time points (60 min) showed a modest but not significant increase in these targets in CD13^KO cells, again suggesting late-phase endosomal NF-κB/MAPK activation as has been reported (4, 5). Finally, Dynasore pretreatment had no effect on NF-kB target gene levels, further confirming that TLR4 surface signaling is dynamin independent (Supplemental Fig S1A). Taken together, these data further confirmed that CD13 specifically participates in the endosomal TLR4-TRIF-pIRF3 pathway.

Figure 3. CD13 does not mediate MyD88-dependent TLR4 surface signaling.

Wild type and CD13^KO BMDM were treated with either vehicle, 100ng/ml LPS or pretreated with 80μM Dynasore followed by LPS stimulation. Figure shows representative immunoblot analysis of phosphor-ERK (A), -p-38 (B and C), -Akt (D) and –IRAK4 (E) -. V; Vehicle, L; LPS, D; Dynasore, D+L; Dynasore followed by LPS. (F) Quantitative RT-PCR analysis of NFκB/MAPK signaling targets, IL-6, TNF-α and IKKβα. Cells were treated with 100ng/ml LPS or 3μg/ml HMGB1 over the indicated time period. Data represent relative transcript level of CD13^KO and wild type normalized to GAPDH ± SEM (n = 3 isolates/genotype, three independent experiments). Treatment of cells of both genotypes with either ligand significantly (p < 0.05) increased transcript levels at each time point when compared to untreated controls, however transcript levels at each time point were not statistically different between genotypes.

CD13 regulates the TLR4 response to DAMPs released in ischemic injury

We have identified CD13 as an inflammatory adhesion molecule that is required for optimal monocyte trafficking to sites of ischemic injury in numerous tissues (10, 12–14, 21). In particular, in a model of ischemic peripheral muscle injury, the ratios of infiltrating monocytes in ischemic tissues from animals lacking CD13 were markedly altered, resulting in a distinctly ‘pro-healing’ cytokine milieu (13, 14, 22). However, healing of damaged tissue and muscle regeneration was unambiguously impaired in these mice despite the favorable inflammatory environment. We hypothesized that the CD13 regulation of the TLR4 response to endogenous DAMPs, in ischemic tissues might contribute to compromised repair. In agreement with this notion, isolated calf and tibia muscles from CD13^KO mice subjected to femoral artery dissection showed a significant induction of IRF3 activation and increased mRNA levels of its target genes IFN-β (Fig. 4A, B), consistent with a physiologic basis for our in vitro observations.

Figure 4. IRF3 and its targets are hyperactivated in vivo in response to ischemic injury in CD13^KO mice.

3 days following femoral artery dissection of the hind limbs of wild type and CD13^KO mice, cells isolated from ischemic muscles were lysed and analyzed for pIRF3 protein levels by immunoblot analysis (A) and its targets, IFN-β and IP-10 by quantitative real-time PCR analysis (B). Data represent relative mRNA expression (Ro) ± SEM (n = 2 isolates/genotype). *p < 0.05. Nitrite concentration in the serum (C) and muscle extract (D) were significantly enhanced in CD13^KO mice compared to wild type mice, 3 days after femoral artery ligation of the hind limb (n = 3 isolates/genotype, three independent experiments). *p < 0.05. (E) Lethally irradiated OVA-FITC loaded splenocytes were added to wild type and CD13^KO splenic DCs and uptake measured by flow cytometry. Data represent mean fluorescence intensity or % Cell ± SEM (n = 3 isolates/genotype, three independent experiments). **p < 0.01; *p < 0.05.

To further explore this novel role for CD13 in ischemic tissue damage, we turned our attention to possible downstream effects of increased pIRF3 and its target gene IFN-β. Ligation of the type I- IFNαβ receptor, IFNAR, induces the JAK/STAT pathway to initiate production of the enzyme inducible nitric oxide synthase [iNOS, and ref (23)]. iNOS metabolizes L-arginine to produce nitric oxide (NO), which is a key component in microbial defense but is toxic when present at high levels (23, 24). A CD13-dependent overexpression of iNOS may lead to increased levels of NO and its reactive metabolites, which would be deleterious to the surrounding tissue. Indeed, CD13^KO mice exhibited significantly higher levels of the NO metabolite nitrite in both serum and muscle extracts when compared to the wild type mice (Fig. 4C–D) supporting a scenario where a CD13-dependent increase in IRF3 phosphorylation and IFNβ levels prompts autocrine activation of IFNAR and increased iNOS, exaggerated production of nitrite and other injurious reactive oxygen species in response to ischemic injury. Finally, to confirm that these observations were due to increased uptake of antigens from dying cells, wild type and CD13^KO DCs were incubated with irradiated wild type splenocytes labeled with OVA-FITC and uptake of cell-associated fluorescence by DCs was analyzed (25). Similar to the response to LPS, internalized cell-associated fluorescence was significantly higher in cells lacking CD13 (Fig. 4E), confirming that CD13 mediates internalization of dead cell antigens to impact the TLR4 response to sterile injury in vivo.

CD13-dependent IRF3 signaling leads to increased nitrite release in LPS stimulated macrophages

To more clearly define the mechanism underlying the increase in nitrite production in injured tissues of CD13^KO animals, we treated wild type and CD13^KO bone marrow macrophages with LPS and measured nitrite release in the culture medium by ELISA. Cells lacking CD13 produced increased levels of nitrite in a time dependent manner with a concomitant increase in iNOS transcripts in CD13^KO cell lysates when compared to its wild type counterpart (Fig. 5A, B), consistent with our in vivo results. To confirm that increased iNOS levels were mediated by IFNAR, we pretreated the cells with the synthetic protein tyrosine kinase inhibitor, AG490 to specifically inhibit JAK2-dependent signal transduction triggered downstream of IFNAR activation. Both nitrite concentrations and iNOS transcript levels were significantly and equivalently diminished by pretreatment with 25μM AG490 in both wild type and CD13^KO cells [(Fig. 5A, B and ref (26)]. Importantly, activation of NFκB MyD88-dependent signaling initiated from the cell surface also leads to production of iNOS. We measured the levels of the active subunit of NFκB, p65 and its downstream target IL-6 under similar conditions and determined that while LPS stimulation led to a two-fold increase in IL-6 transcript and p65 protein level in both genotypes, blocking JAK2 activity had no effect on levels of p65-NFκB or its downstream target IL-6 (Fig. 5C–E). Therefore, the increase in iNOS and nitrite in CD13^KO cells and tissues is due to CD13 regulation of the p-IRF3 signaling pathway leading to increased autocrine, IFNβ-induced JAK/STAT signal transduction and likely contributes to the impaired healing we observe in CD13^KO animals (13).

Figure 5. CD13-dependent pIRF3 signaling increases nitrite release.

(A) Wild type and CD13^KO BMDM were treated with LPS +/− JAK2 inhibitor, 25 μM AG490 (Calbiochem) pretreatment for 30 min and nitrite released in the culture medium was measured by ELISA. (B) iNOS and (C) IL-6 mRNA levels were measured in the cell lysate under similar experimental conditions by quantitative RT-PCR analysis. Data represent ± SEM (n = 3 isolates/genotype, three independent experiments). **p < 0.01; *p < 0.05. Immunoblot analysis (D) and quantification (E) of p65- NFκB protein levels under similar experimental conditions. a; WT/LPS vs. WT/Vehicle, b; CD13^KO/LPS vs CD13^KO/Vehicle. *p < 0.05.

CD13 functions independently of the positive regulator CD14 in LPS-induced TLR4 endocytosis

Myeloid CD14 binds LPS on the cell surface and delivers it to TLR4 to initiate signal transduction and thus has been classified as a positive regulator of TLR4 endocytosis (6). Since we have demonstrated that CD13 negatively regulates TLR4 endocytosis, it is possible that CD13 and CD14 act in concert to regulate the TLR4 response to infection and injury. Therefore, we investigated their possible association in LPS-stimulated receptor internalization. Similar to TLR4, surface CD14 expression is comparable in unstimulated wild type and CD13^KO splenic DCs (Figs. 6A and 1A). However, in contrast to the differential endocytosis of TLR4, CD14 internalization was modest with no difference in cells of either genotype upon LPS induction and is not sensitive to Bestatin (Figs. 6B, C and 1B, C). Analogously, CD13 and CD14 are not consistently co-localized in endosomes in response to LPS as demonstrated by a weak Pearson’s coefficient (r=0.33) (Fig. 6D, E). Taken together these results suggest that regulation of TLR4 internalization by CD13 is independent of CD14, and perhaps more importantly, that CD13 may not regulate endocytosis per se, but rather which proteins are internalized.

Figure 6. CD13-mediated TLR4 endocytosis and endosomal signaling is independent of CD14.

(A). Representative histogram of CD14 surface expression in unstimulated wild type and CD13^KO BMDM measured by flow cytometry. (B) % of surface CD14 remaining following LPS treatment in wild type and CD13^KO CD11c⁺ CD8⁺ splenic DCs stimulated with vehicle, LPS (100 ng/ml) or 100μg/ml Bestatin (BE) for 30 min followed by LPS addition for indicated time intervals by flow cytometry. (C) Immunofluorescence analysis of CD14 expression in response to LPS for 30 min. Cells were fixed, permeabilized and stained with anti-mouse CD14 antibody (clone Sa14-2; Hycult Biotech). CD14 (green) and TOPRO3; nuclear (blue). Original magnification 63x oil. Scale bar; 10μM (D) Co-localization and co-internalization of CD13 (red) and CD14 (green) in LPS stimulated BMDM by confocal microscopy. Arrows indicate co-localization. Original magnification 100x oil. Scale bar; 10μM (E) Pearson’s correlation coefficient between CD13 and CD14 upon vehicle or LPS treatment. Data represents average of 10 images each from three independent experiments.

CD13 expression is upregulated in response to LPS and directly interacts with TLR4

In an effort to uncover potential clues into the mechanism by which CD13 may regulate endocytosis in the context of infection or injury, we investigated the impact of inflammatory signals on CD13 itself. Previous studies had suggested that CD13 expression levels were increased in human peripheral blood monocytes in response to injury in vivo or LPS treatment in vitro (27). In agreement with this finding, we found that LPS stimulation of WT macrophages induced CD13 protein expression over four-fold in a time dependent manner (Fig. 7A), suggesting that maintaining a high level of CD13 expression is advantageous during inflammation. Additionally, our study characterizing CD13 in the regulation of antigen uptake demonstrated that while the lack of CD13 enhanced mannose receptor (MR) endocytosis, CD13 and MR did not occupy the same complex (9). Surprisingly and in contrast to MR, immunoblot analysis of macrophage and dendritic cell lysates immunoprecipitated with CD13 mAb demonstrated that TLR4 was indeed present in the CD13 immunoprecipitate (Fig. 7B), suggesting they are components of the same protein complex and perhaps that CD13 regulates internalization by different, receptor-dependent mechanisms.

Figure 7. CD13 expression is induced and CD13 and TLR4 co-immunoprecipitate in macrophages and DCs.

(A) Wild type BMDM were treated with 100ng/ml LPS for increasing time intervals and probed for CD13 protein expression. Data represent relative fold change of CD13/β-actin levels ± SEM (n = 2 isolates/WT, two independent experiments). *p < 0.05. (B). Cell lysates from wild type BMDC or BMDM precipitated with agarose beads were pulled down with control IgG or CD13 mAb and probed for TLR4 and CD13 protein expression. (C) Schematic of TLR4 endocytic regulation by CD13. CD13 and TLR4 are present in a protein complex under basal conditions. In response to TLR4 ligands, TLR4 signals via two pathways. Surface TLR4 signals via MyD88 to activate NfκB and proinflammatory cytokines and is independent of CD13 expression. TLR4-CD13 then undergoes endocytosis in a dynamin and CD14 dependent manner resulting in IRF3 activation and upregulation of its target genes, type I IFNs. In response to LPS, CD13 modulates TLR4 receptor-mediated endocytosis where lack of CD13 skews cytokine production towards IRF3 target, IFNβ, ultimately leading to toxic iNOS and ROS production via activation of IFNAR and JAK/STAT signaling.

Discussion

We have recently shown that CD13 negatively regulates dynamin-dependent endocytosis of distinct ligand-receptor pairs in dendritic cells (9). Here we extend these observations to demonstrate that CD13 maintains the prototypical pattern recognition receptor TLR4 at the dendritic cell surface, thus holding endocytic signal transduction in check and avoiding deleterious tissue destruction. We find that in resting dendritic cells and macrophages CD13 and TLR4 associate in a protein complex and that LPS induces CD13 internalization and co-localization with TLR4 in dynamin⁺ and Rab5⁺ early endosomes in vitro. In the absence of CD13, TLR4 signal transduction is skewed toward the endocytic TRIF-biased pathway, inducing hyperactivation of IRF3 and its target cytokine, IFNβ and autocrine production of iNOS, thus indicating an unequivocal role for CD13 in the regulation of the innate immune response to LPS (Fig. 7C). We had previously shown that tissue repair and healing were impaired in CD13^KO animals in response to ischemic injury in vivo (13) and asked if altered TLR4 signal transduction in response to release of endogenous DAMPs may contribute to the tissue destruction we observed in vivo. Indeed, assay of ischemic skeletal muscle from CD13^KO animals showed exaggerated levels of IFNβ. While physiologic levels of IFNβ normally induce pro-angiogenic and anti-inflammatory cytokines, uncontrolled IFNβ production can lead to autocrine activation of its receptor IFNAR and subsequent JAK/STAT signaling cascade, ultimately resulting in pro-inflammatory iNOS and toxic reactive oxygen species (Fig. 7C, (23, 24, 28). Indeed serum and muscle extracts from CD13^KO mice exhibited increased levels of the NO metabolite nitrite (28). Thus, we have identified a novel function for CD13 in maintaining the balance between pro- and anti-inflammatory cytokines in the innate immune response to PAMPs and DAMPs.

An essential core of auxiliary molecules are required for signaling from the TLR4 complex in response to LPS; these include LPS binding protein, LBP, that carries serum LPS to CD14 on the cell surface which then delivers it to the TLR4/MD2 complex to initiate signaling via the adaptor proteins MyD88, TIRAP, TRIF and TRAM [reviewed in ref (29)]. Further studies have identified other associated molecules that regulate either cell surface or endocytic TLR4 signaling such as mediation of LPS binding to CD14 by CD33 (30), delivery of LPS to TLR4 by CD14 (6), Syk phosphorylation of TLR4 (31), enrichment of TIRAP at the plasma membrane by CD11b to regulate the MyD88-dependent pathway (32) as well as TRIF-mediated signal transduction from endosomes (33) and others whose functions are unknown (34). We find that loss of CD13 has a similar phenotype in response to both PAMPs and DAMPs, but no influence on LPS-induced pro-inflammatory MyD88 signaling; arguing that CD13’s control of TLR4 signaling is independent of ligand availability and is primarily at the level of endocytosis.

Endocytosis is an extremely complex process that regulates the internalization of a vast array of cargo molecules which are identified, organized and assembled by adaptor proteins into a single endocytic compartment [reviewed in refs (15, 35)]. These accessory molecules range in their specificity for a particular cargo(es) and have different sub-cellular/tissue locations and importantly, operate in a modular fashion such that adding or subtracting adaptors to a complex can specifically modify the uptake of receptors or their ligands. Thus adaptors can control the composition of proteins in the endosome without interfering with vesicle formation. We have demonstrated that in addition to TLR4, the internalization of a number of receptors via mechanisms requiring the fission GTPase dynamin is CD13-dependent (9, 36). A potential mechanistic insight to CD13’s function in endocytosis may be provided by the fact that dynamin-dependent endocytosis primarily initiates in lipid rafts (37, 38) and that CD13 localizes in lipid rafts in monocytes (39) and endothelial cells (36). We have shown that CD13 regulates plasma membrane organization by determining the composition of proteins included in lipid rafts (36), suggesting it may act as an endocytic adaptor protein. While the CD13 cytoplasmic or transmembrane domains do not contain the motifs identified as adaptor ‘sorting signals’ (40), we have found that CD13 is present in a physical complex with one of the receptors it regulates but not another (9). It is possible that depending on the target, CD13 can act as either a direct adaptor or in an indirect manner (perhaps with other adaptors) to internalize target receptors. We are currently dissecting these possibilities by molecular mapping and mutation of the CD13/TLR4 interacting domains to determine if disrupting the interaction impacts endocytosis of TLR4 or MR as well as identifying other additional CD13-interacting proteins by proteomic analysis. Additionally, we and others (41) find that CD13 expression levels are remarkably upregulated by inflammatory activation signals and tissue damage, supporting the notion that a high level of CD13 expression is important during the immune response. Understanding the contribution of these and other CD13-mediated mechanisms to endocytosis is the subject of active investigation in our laboratory.

While we find that CD13 negatively controls TLR4 internalization, CD14 has been identified as a positive regulator of this process since it delivers LPS to TLR4 (6), perhaps indicating a potential functional interaction between these two opposing mediators. However in untreated cells CD13 and CD14 do not extensively co-localize on the membrane, particularly when compared to CD13 and TLR4. Similarly, although CD13 and CD14 appear to occupy the same endosomal compartments in response to LPS treatment, the majority of the CD14 remains autonomous while TLR4 and CD13 are nearly completely co-localized in intracellular vesicles. Furthermore, the loss of CD13 has no effect on LPS-induced CD14 uptake, suggesting these are separate processes. Alternatively, it has been proposed that in response to DAMPs CD14 guides TLR4 into lipid rafts to promote downstream signaling (42), raising the possibility that CD13 and CD14 may not regulate endocytic processes, but rather act as opposing adaptors that mediate which proteins are available at the endosomal ‘loading station’.

We have found that CD13 also negatively regulates internalization of the bradykinin receptor in endothelial cells (36), uptake via the mannose and transferrin receptors in DCs and macrophages (9) in addition to TLR4 endocytosis. In each of these cases, downstream signal transduction cascades are altered, illustrating endocytic control over signal transduction. As an extracellular peptidase, CD13 cleaves single amino acid residues from small peptides and therefore its substrates vary depending on the tissue where it is expressed. However, our observations have clearly shown that CD13 is a surprisingly multifunctional molecule that often acts independently of its enzymatic activity to regulate diverse processes such as angiogenesis, inflammatory trafficking, antigen presentation, cell-cell adhesion, coronavirus infection and maintenance of the stem cell niche (9–14). While it is difficult to reconcile how a single molecule can regulate such apparently disparate functions, the fact that CD13 participates in the endocytosis of various receptors may underlie its wide range of activities. This finding emphasizes that endocytosis is far more complex than a mechanism to transport molecules across a membrane; it is responsible for the organization and communication of all eukaryotic cells and is intimately linked to essentially every aspect of cellular regulation including receptor availability, signal transduction, adhesion, migration, membrane integrity and cell polarity (15). When viewed in this context, CD13 and other endocytic regulators become critical biological monitors.

Abbreviations

spDC

splenic derived dendritic cell

BMDC

bone marrow-derived dendritic cell

BMDM

bone marrow-derived macrophages

WT

wild type

KO

knockout