Conceived and designed the experiments: DF JD JNT. Performed the experiments: DF. Analyzed the data: DF CT AQH JD JNT. Contributed reagents/materials/analysis tools: CT. Wrote the paper: DF JD JNT.
The authors have declared that no competing interests exist.
Dendritic Cells (DC) represent a key lung immune cell population, which play a critical role in the antigen presenting process and initiation of the adaptive immune response. The study of DCs has largely benefited from the joint development of fluorescence microscopy and knock-in technology, leading to several mouse strains with constitutively labeled DC subsets. However, in the lung most transgenic mice do express fluorescent protein not only in DCs, but also in closely related cell lineages such as monocytes and macrophages. As an example, in the lungs of CX3CR1+/gfp mice the green fluorescent protein is expressed mostly by both CD11b conventional DCs and resident monocytes. Despite this non-specific staining, we show that a shape criterion can discriminate these two particular subsets. Implemented in a cell tracking code, this quantified criterion allows us to analyze the specific behavior of DCs under inflammatory conditions mediated by lipopolysaccharide on lung explants. Compared to monocytes, we show that DCs move slower and are more confined, while both populations do not have any chemotactism-associated movement. We could generalize from these results that DCs can be automatically discriminated from other round-shaped cells expressing the same fluorescent protein in various lung inflammation models.
The lung immune system is very efficient: constantly exposed to pathogens and pollutants, the lower respiratory airways are nevertheless maintained sterile, while inflammation is kept at the lowest level
Among the most important immune cells in the lungs are monocytes, alveolar macrophages and dendritic cells (DCs)
A: Maximum intensity projection of a lung slice z-stack. Pulmonary CX3CR1-GFP cells (green) and alveolar collagen mesh detected by collection of SHG signal (gray). Two-photon excitation wavelength = 896 nm. B: Dashed red squares show the optic field imaged by the microscope (CX3CR1-GFP in green and SHG in grey). The realignment phase consists in calculating the tissue drift using the maximization of SHG signal cross-correlation.
The aim of the present study is to show how to overcome the non-discrimination of different subsets sharing the same fluorescent tag expression in dynamic studies. Here, we demonstrate the feasibility of an automated discrimination of two main CX3CR1-positive cell populations using a criterion based on the cell shape: the roundness. In order to separate Round-shaped cells (RSCs) and Dendritic-shaped Cells (DSCs), we suggest to introduce two novel coefficients: the Instantaneous Roundness Coefficient (IRC) measured in each frame and the Mean Roundness Coefficient (MRC) calculated as the mean of the IRC on the total tracking time for each cells. Using this strategy implemented in a cell tracking code, we show that different behaviour can be observed between the Round-shaped Cell (RSCs) and Dendritic-shaped Cell (DSCs) subsets. This novel approach may be generalized to other transgenic animal strains (e.g. MHCII-EGFP and CD11c-YFP knock in mice). This could lead to a better understanding of DC behaviour and a better analysis of the lung immune system during infection.
A: Edge detection of two CX3CR1+ pulmonary cells and their roundness coefficient. Scale bar = 10 µm. B: The relevant parameters used in this work are: i) the Mean Roundness Coefficient (MRC), calculated for each cell by meaning Instantaneous Roundness Coefficient (IRC) at each consecutive observable time; ii) the Maximal Distance (MD) of a cell (red arrow) is the longest distance covered from the first position; iii) the Meandering Index (MI) is the final distance from the first position
All experimental procedures were performed in accordance with the French Government guidelines for the care and use of laboratory animals and were approved by the
CX3CR1+/gfp mice (further referred as CX3CR1 mice) were maintained under specific pathogen-free conditions at the
A: Total lung cells of CX3CR1+/gfp mice were gated on CX3CR1 and analyzed for NK1.1, CD3e, CD11c, and CD11b expressions. B: Autofluorescence, CD80 and MHCII expressions on gate G1, G2 and G3 of panel A. Black histogram, isotype control; grey histogram, positive staining. C: Total lung cells were pre-gated on CD11c+ low autofluorecent cells and analyzed for the expression of CD11b and CD103. The expression of CX3CR1 is shown on the left panel for gate G4 (CD11b−CD103+ DCs, grey histogram) and for gate G5 (CD11b+CD103− DCs, black line). D: Total cells were pre-gated on CD45 cells and analyzed for the expression of F4/80 and CD11c. Expression of CX3CR1 and CD11b is shown on the left panels for gate G6 (CD11clowF4/80high), G7 (CD11chighF4/80high) and G8 (CD11chighF4/80low).Data from flow cytometry, performed on one CX3CR1+/gfp mouse lung harvested 30 minutes after intratracheal PBS injection. Data are representative of two distinct experiments.
A: Expression of CX3CR1
Mice were euthanized either 30 minutes (‘early stage’ group) or 4 hours (‘late stage’ group) post administration of PBS or LPS. Left lobes of lung explants were cut in the middle with a vibratome (Leica). The bottom of lung lobes was carefully glued on a Petri dish filled up with phenol-red free RPMI medium (RPMI 1640, PAN Biotech GmBH) at 37°C. Medium was refreshed every hour. Explants were kept for one hour at 37°C in a 5% CO2 environment before imaging and kept at 37°C during the whole experiment. This phase was aimed to stabilize the explant by emptying out most of the air from the alveoli.
Lungs were harvested after mouse euthanasia, mechanically disrupted using gentleMACS™ Dissociator (Miltenyi Biotec) according to manufacturer instructions, enzymatically digested with 1 mg/mL Collagenase I (Worthington) for 30 min at 37°C in 50U/mL DNase I (Sigma)-containing DMEM. Then, the solution was filtered with 70 μm cell strainers (Becton Dickinson) to obtain single-cell suspensions.
A: Dendritic-shaped cells and B: Round-shaped cells at an early stage (average values from 1h30 to 2h30 post injection, closed symbols) and a late stage (average values from 5h to 6h post injection, open symbols) after injection of PBS (rounds) or LPS (squares). Three mice in each group, one symbol by cell. * for p<0.05; ** for p<0.01; *** for p<0.0001; ns for not significant.
Inhibition of nonantigen-specific binding of immunoglobulins to Fc receptors was performed using a rat antimouse CD16/CD32 antibody (2.4G2 BD Biosciences). Cells were subsequently stained for 30 min at 4°C with the following monoclonal antibodies: Alexa Fluor 700 conjugated anti-CD45 (30-F11; Biolegend), PE-Cy7 conjugated anti-CD11b (M1/70; eBioscience), APC conjugated anti-CD11c (HL3; BD Biosciences), PE conjugated anti-NK1.1 (PK136; BD Biosciences), PercPCy5.5 anti-CD3e (145-2C11; eBioscience), Alexa Fluor 700 conjugated anti-MHC II (MC-114-15.2; eBioscience), PercPCy5.5 anti-F4/80 (BM8; Biolegend), PE conjugated anti-CD80 (16-10A1; BD Biosciences), PE conjugated anti-CD103 (2E7; eBioscience).
A: Dendritic-shaped cells and B: Round-shaped cells at an early stage (average values from 1h30 to 2h30 post injection, closed symbols) and a late stage (average values from 5h to 6h post injection, open symbols) after injection of PBS (rounds) or LPS (squares). Three mice in each group, one symbol by cell. C, D: Overlay of Round-shaped cell tracks after late PBS (C) and LPS injection (D), after aligning their first coordinates. One color by track. Values of black circle radii in µm, equal to average cell Maximal Distance, are indicated ± standard deviation. * for p<0.05; ** for p<0.01; *** for p<0.0001; ns for not significant.
Dead cells were excluded by staining for 30 min at 4°C with Blue LIVE/DEAD® Fixable Dead Cell Stains (Invitrogen) following the manufacturer instructions.
Cells were then fixed with Cellfix (BD Biosciences). Cell acquisition was directly performed with an LSR-II machine using FACSDiva software (BD Biosciences) and the data were analysed with FlowJo software (TreeStar). Cell doublets were excluded using FSC-A and FSC-H.
Both second-harmonic generation (SHG) and two-photon excitation fluorescence (TPEF) imaging were performed on a Zeiss LSM 710 microscope equipped with a W Plan-Apochromat 20× NA 1.0 DIC M27 75mm water immersion objective (Zeiss). Two-photon excitation was produced at 896 nm by a femtosecond Ti: Sa laser (Chameleon Ultra, Coherent). SHG and EGFP signals were both epidetected by two dedicated non-descanned detector, one is coupled with a 500–550 nm band-pass system for EGFP and the other with a 448 nm ±20 nm band-pass filter for SHG. Z-stacks were acquired every two minutes during one hour. Images size was 512 by 512 pixels, corresponding to a field of view of 280 by 280 µm.
Data processing was performed under Matlab using a multiple particle tracking code by Blair and Dufresne (available on
For the four experimental conditions (mixing early/late stage and PBS/LPS delivery), results from three mice were pooled. Comparisons between groups were performed using Mann-Whitney test, using GraphPad Prism Software (GraphPad Software, Inc.).
Sample drift due to the presence of air in the lung is a major issue
The Matlab code was implemented with an edge detection custom-made routine aimed to determine the roundness coefficient of each cell. Instantaneous Roundness Coefficient (IRC) is defined as follow:
The IRC indicates the index of circularity of any object: from 0 (line-shaped) to 1 (round-shaped) (
The Meandering Index (MI) yields information about the directionality of the cell movement
This parameter being inappropriate to fully characterize a random walk path
First, we assessed the different lung phagocyte subpopulations by flow cytometry analysis (
To validate our imaging analysis approach, we first looked at the representation of distribution frequency of MRC at homeostasis (
Therefore, subpopulations of DC and monocytes in flow cytometry were in the same range as DSC and RSC in microscopic analysis. We decided to set the MRC threshold to 0.35 for the rest of the analysis.
Previous studies by two-photon excited fluorescence (TPEF) showed that LPS could activate tracheal DCs
The Maximal Distance (MD) of DSCs was not affected by LPS (
Finally, we compared the Meandering Index (MI) of RSCs and DSCs. Even though LPS increased MD for RSCs, neither of the two populations presented a MI altered by LPS. Furthermore, MI values for all groups were less than 0.4, suggesting that movements observed were not directed by chemotactism
In this study, we show for the first time that we could systematically discriminate two cell populations present in an organ explant and sharing the same staining, using a shape criterion. This may be of paramount importance for the DC subset analysis, as so far no fluorescent protein knock-in mouse strain available is expressed only in a defined DC lineage. To our knowledge the separation of different subsets expressing the same fluorescent tag has been seldomly studied, although it can be a major issue.
In a very recent report on lung exploration by TEPF of CD11c-YFP mouse, Veres
In fact, the shape is not a poorly defined character of a cell, as stressed by the fact that DCs were identified and described for the first time only by their shape
Thanks to that, we analyze here the effects of LPS instillation by intra-tracheal route on cell motility. Intra-tracheal instillation was chosen in spite of its invasivity (a consequence may be the increase of velocity of DSCs at the early
LPS is a Gram-negative outer membrane component known to elicit strong immune responses via the Toll-like receptor (TLR) 4
Our results prove the feasibility of the shape-based discrimination of two functionally distinct cell populations, monocytes and DCs, both expressing EGFP via the promoter CX3CR1, under LPS mediated inflammatory conditions. The results we present are in accordance with the expected functionality of these two major immune cell populations. DCs movements induced by LPS have been already reported
The new imaging tool we have developed in this study more accurately discriminates DC population from closely related lineages. It could be more universally applied to improve our knowledge of the lung immune system.
The authors would like to thank Dr. Jean-Claude Vial for his precious advice and mentorship and Drs. Bradley Stiles and Jeffrey Froude for reading and editing the manuscript.