A crucial role for β2 integrins in podosome formation, dynamics and Toll-like-receptor-signaled disassembly in dendritic cells

Abstract

The dynamic properties of podosomes, their ability to degrade the underlying matrix and their modulation by Toll-like receptor (TLR) signaling in dendritic cells (DCs) suggests they have an important role in migration. Integrins are thought to participate in formation and dynamics of podosomes but the multiplicity of integrins in podosomes has made this difficult to assess. We report that murine DCs that lack β2 integrins fail to form podosomes. Re-expression of β2 integrins restored podosomes but not when the membrane proximal or distal NPxF motifs, or when an intervening triplet of threonine residues were mutated. We show that β2 integrins are remarkably long-lived in podosome clusters and form a persistent framework that hosts multiple actin-core-formation events at the same or adjacent sites. When β2 integrin amino acid residues 745 or 756 were mutated from Ser to Ala, podosomes became resistant to dissolution mediated through TLR signaling. TLR signaling did not detectably modulate phosphorylation at these sites but mutation of either residue to phospho-mimetic Asp increased β2 integrin turnover in podosomes, indicating that phosphorylation at one or both sites establishes permissive conditions for TLR-signaled podosome disassembly.

Keywords: Dendritic cells; Integrins; Podosomes.

Fig. 1.

β2-integrin-null DCs are podosome deficient. Wild type (WT) and Itgb2-null SDCs plated on glass coverslips were fixed and stained for β2 integrin (green; FITC), F-actin (red; Alexa-Fluor-555) and vinculin (grey; Alexa-Fluor-633). (A) WT cells contained podosome clusters with clear actin cores, and β2 integrin- and vinculin-rich podosome rings and/or plaques. Itgb2-null DCs adhered but did not form podosomes. (B) WT and Itgb2-null DCs, can both form focal adhesions (white arrows). Single optical sections of 0.7 µm, taken at the ventral surface of the cells were acquired using Zen 2009 software on a Carl Zeiss 700 confocal laser-scanning microscope with a 100x Plan Apochromat/NA 1.46 oil immersion objective. Scale bars: 10 µm (A), 5 µm (B). (C) Percentage of integrin-null cells containing podosomes confirms the dramatic lack of podosomes compared to WT DCs (*P = 0.01, unpaired t-test), whereas the percentage of cells containing focal adhesions is normal. (D) Individual podosomes were also counted in WT and Itgb2-null SDCs and the results demonstrate that individual Itgb2-null cells have less podosomes compared with WT (29–57 cells scored per sample, error bars are s.e.m. for triplicate biological samples, ***P<0.001, unpaired t-test). (E) DCs were plated for 75 min on substrates as indicated and the % of adherent cells assessed after washing. Only when Itgb2-null cells were plated on β2 substrate, ICAM-1, was there a significant reduction in adhesion (**P = 0.002, unpaired t-test). (F) DCs were plated for 2 hours onto coverslips coated with gelatin, fibronectin, laminin or fibrinogen (all at 10 µg/ml), or with HA (100 µg/ml), and cells with podosomes quantified after staining for F-actin and vinculin. There was no significant rescue of podosomes in Itgb2-null DCs when plated on the various substrates compared with plating on glass alone, except when cells were incubated on laminin, in which case podosome levels were reduced rather than rescued (paired t-tests).

Fig. 2.

Ex vivo cells that lack β2 integrin have podosome formation defects. (A) Resident lung cells (>95% alveolar macrophages) collected from wild type (WT) or Itgb2-null mice by bronchoalveolar lavage were plated on coverslips and stained for β2 integrin (green; FITC), F-actin (red; Alexa-Fluor 555) and vinculin (grey; Alexa-Fluor 633). Images were acquired as described for Fig. 1. Scale bars: 5 µm. (B) Adherent cells were scrutinized for podosome formation using systematic scanning of the coverslips. Quantification of the percentage of cells with podosomes indicated a strong defect in podosome formation in lung-derived cells (**P = 0.001, unpaired t-test). (C) Cellular composition of bronchoalveolar lavage in WT and Itgb2-null mice was determined morphologically by differential counts of DiffQuik-stained cytospin preparations.

Fig. 3.

Retroviral expression of Itgb2-EGFP rescues podosomes in Itgb2-null DCs. Itgb2-null BMDCs were infected with retrovirus encoding a Itgb2-EGFP fusion protein, or GFP alone as a control. (A) The infected DCs were allowed to adhere to glass, then fixed and stained for F-actin (red; Alexa Fluor-555) and vinculin (grey; Alexa Fluor-633). Images were acquired using a Zeiss LSM700 as in Fig. 1. Scale bars: 5 µm. (B) Percentage of infected (GFP⁺) cells that contain podosomes, demonstrating significant reconstitution of podosomes with WT-β2 integrin compared to GFP alone (**P = 0.007, paired t-test). (C) Retrovirus encoding both Itgb2-EGFP and Lifeact-mCherry was used to infect Itgb2-null DCs. The cells were plated onto glass dishes and images collected every 10 seconds at 37°C using a Nikon Eclipse Ti TIRF microscope with an ApoTIRF 100x/NA1.49 objective as in Materials and Methods. A section through a podosome cluster was selected and Imaris software used to convert time to display on the z-axis, to generate a kymograph (Itgb2, green; actin, red; sequence represents 491 seconds), revealing the transient nature of the actin cores compared with the long-lived Itgb2-EGFP. To measure the lifetime of podosome cores, 682 podosomes were observed in six cells from at least three separate experiments. Scale bar: 5 µm. (D) A series of selected images (50-second intervals) shows that stable EGFP-labelled Itgb2 structures can support the reoccurrence of podosome cores in the same location after long periods of time (circled areas; images were cropped and circles were added to aligned layers using Photoshop CS5); ∼17±3% of sites (810 observed podosomes in six plaques from three experiments) hosted the return of actin cores. Scale bars: 0.5 µm.

Fig. 4.

FRAP analysis of Itgb2-EGFP lifetime in podosomes. BMDCs were infected with retroviruses for expression of actin-EGFP, EGFP-kindlin3, paxillin-mCherry or Itgb2-EGFP (Itgb2-null cells). The cells were plated into glass-bottomed dishes and an area within a podosome cluster was photobleached using a Zeiss LSM700 confocal microscope as in described in Materials and Methods. Cells were then imaged over time to follow fluorescence recovery (A). For the Itgb2-EGFP-expressing cells an additional area of cell was also bleached to assess integrin turnover outside of podosomes (Itgb2PM). The fluorescence recovery in podosomes of actin-EGFP, EGFP-kindlin-3, paxillin-mCherry and Itgb2-EGFP in the plasma membrane were all relatively rapid compared to the recovery of Itgb2-EGFP in podosomes. (B) Recovery curves for each tagged protein were normalized for comparison and mean t_1/2 values calculated, each from three independent experiments (three individual BMDCs cultures and viral infections), analyzing a minimum of ten cells per experiment (P = 0.04, paired t-test, comparing β2 integrin in podosomes versus plasma membrane). Scale bars: 2 µm.

Fig. 5.

Mutation of key residues in the cytoplasmic tail of Itgb2 and their effect on podosome formation. (A) Amino acid sequence of the cytoplasmic tail of the mouse β2 integrin with residues of interest highlighted; Ser745, Ser756 (red), Thr758, Thr 759, Thr 760 (TTT; blue), Phe754 and Phe766 (green). Amino acid positions are based on the human β2 integrin. (B) Itgb2-null cells were reconstituted with either wild type Itgb2-EGFP (WT-Itgb2) or NPXF-Ax2-Itgb2-EGFP (NPXF-Ax2) by retroviral infection (green). After fixation, the cells were stained for F-actin (red; Alexa-Fluor-555). The Itgb2-NPXF-A mutant was unable to rescue podosome formation even though it localized to the plasma membrane. (C) Percentage of podosomes in control WT cells or Itgb2-null cells reconstituted with GFP alone or with Itgb2 constructs containing the following cytoplasmic tail mutations as indicated. NPxF-Ax2: F754A, F766A. TTT-AAA: T758A, T759A, T760A, and double and triple combinations thereof. Data are from 100–150 GFP-positive cells per sample. Podosomes were not significantly reconstituted in Itgb2-null cells expressing the NPXF-Ax2 or TTT-AAA mutants compared to GFP alone (paired t-tests). (D) Cell surface expression of the WT-Itgb2 and NPxF-Ax2 and TTT-AAA mutants in the infected cells was confirmed by flow cytometry using an anti-β2 integrin antibody. (E) BMDCs cultured from Itgb2WT or Itgb2 TTT-AAA β2 cytoplasmic tail knock-in mice were stained for β2 integrin (green; FITC) and F-actin (red; Alexa-Fluor-555). Scale bars: 5 µm (B) and 10 µm (E). (F) Percentage of cells with podosomes indicates a loss of the podosome phenotype in the TTT-AAA knock-in DCs (***P<0.001, unpaired t-test). Images were acquired using a Zeiss LSM700, as in Fig. 1.

Fig. 6.

Mutation of Ser756 to Ala blocks acute LPS-stimulated podosome loss. (A) Itgb2-null BMDCs were infected with retroviral constructs expressing WT-Itgb2-EGFP or S756A-Itgb2-EGFP (green). Cells were treated as indicated with (+LPS) or without (–LPS) 50 ng/ml LPS for 30 minutes, then fixed and stained to visualize F-actin (red; Alexa-Fluor-555) and α-actinin 4 (grey; Alexa-Fluor-633). Podosomes reconstituted with the S756A-Itgb2-EGFP mutant were resistant to the LPS-stimulated disassembly seen for WT-Itgb2-EGFP (A; green). Images were acquired using a Zeiss LSM700, as in Fig. 1. Scale bars: 5 µm. (B) Percentage of EGFP+ cells that contain podosomes when reconstituted with Itgb2-EGFP constructs containing double or single Ser to Ala mutations (S745A and S756A, or either S745A or S756A) with or without treatment with LPS (50 ng/ml) or prostaglandinE2 (10 µg/ml) for 30 minutes, before fixation and staining as above. LPS-driven podosome loss, though significant (**P = 0.005) for WT-Itgb2-EGFP, was not significant in cells expressing the single S745A or S756A β2 mutants (paired t-tests). (C) CD40 expression in cells expressing WT, S745A or S756A Itgb2 was assessed by flow cytometry in control cells (dashed line) and after 20 hours of LPS treatment (solid line).

Fig. 7.

Mutation of Ser756 to Asp rescues acute LPS-driven podosome loss. (A) Itgb2-null BMDCs were infected with retroviruses containing either the WT-Itgb2-EGFP or the S756D-Itgb2-EGFP mutant constructs (green EGFP staining) and treated with (+LPS) or without (–LPS) 50 ng/ml LPS for 30 minutes before fixation and then stained to visualize F-actin (red) and α-actinin 4 (grey). Images were acquired using a Zeiss LSM700, as described for Fig. 1. The images show that podosomes formed normally in S756D-Itgb2-EGFP expressing cells and that LPS induced podosome dissolution. Scale bars: 5 µm. (B) Percentages of infected (GFP positive) cells showing podosomes when reconstituted with empty vector (pBMN-I-GFP), or constructs for WT β2 integrin or the S745D and S756D mutants, with or without LPS treatment were quantitated. Podosomes in cells expressing the S745D and S756D mutants were responsive to LPS (**P = 0.001 and *P = 0.014, unpaired t-tests). (C) Itgb2-null BMDCs expressing EGFP fusion proteins of WT, S745A, S745D, S756A or S756D-Itgb2 were cultured in glass bottom dishes for FRAP analysis, as in Fig. 4. Cells expressing Itgb2-EGFP in a typical honeycomb shaped podosome ring pattern in the ventral plasma membrane were selected for photobleaching and the half-life for fluorescence recovery measured for each construct as in Materials and Methods (ten cells per experiment). S745D and S756D mutants show reduced half-life compared to corresponding Ala mutants (**P = 0.001 and **P = 0.006, respectively, paired t-test). Data were from three independent experiments from different bone marrow and virus preparations.