CD169+ macrophages orchestrate plasmacytoid dendritic cell arrest and retention for optimal priming in the bone marrow of malaria-infected mice

  1. Jamie Moore-Fried
  2. Mahinder Paul
  3. Zhixin Jing
  4. David Fooksman  Is a corresponding author
  5. Gregoire Lauvau  Is a corresponding author
  1. Albert Einstein College of Medicine, Department of Microbiology and Immunology, United States
  2. Albert Einstein College of Medicine, Department of Pathology, United States

Abstract

Plasmacytoid dendritic cells (pDCs) are the most potent producer of type I interferon (IFN), but how pDC is primed in vivo is poorly defined. Using a mouse model of severe malaria, we have previously established that upon priming by CD169+ macrophages (MPs), pDC initiates type I IFN-I secretion in the bone marrow (BM) of infected mice via cell-intrinsic TLR7 sensing and cell-extrinsic STING sensing. Herein we show that CD169+ MP and TLR7 sensing are both required for pDC arrest during priming, suggesting CD169+ MP are the source of TLR7 ligands. We establish that TLR7 sensing in pDC and chemotaxis are both required for pDC arrest and functional communication with CD169+ MP in the BM. Lastly, we demonstrate that STING sensing in CD169+ MP control pDC initiation of type I IFN production while also regulating pDC clustering and retention/egress from the BM. Collectively, these results link pDC acquisition of type I IFN-secreting capacity with changes in their motility, homing and interactions with CD169+ MP during infection. Thus, targeting this cellular interaction may help modulate type I IFN to improve outcomes of microbial infections and autoimmune diseases.

Editor's evaluation

This study presents valuable mechanistic findings demonstrating the requirement of CD169+ macrophage intrinsic STING signaling in regulating plasmacytoid DC motility/arrest and their activation. The evidence supporting the claims of the authors is solid, and their data suggest that adequate type I IFN production by pDC may also require macrophage intrinsic STING signaling. The work will be of great interest to the immunology community.

https://doi.org/10.7554/eLife.78873.sa0

Introduction

Plasmacytoid dendritic cells (pDCs) (Colonna et al., 2004; Liu, 2005; Reizis, 2019) are equipped with the unique ability to rapidly secrete up to 1000 times more type I interferon (IFN) – both IFNα and IFNβ – than any other cell (Cella et al., 1999; Siegal et al., 1999; Asselin-Paturel et al., 2001). This is accounted for by several pDC-intrinsic and -extrinsic mechanisms. pDC-intrinsic features include the constitutive expression of the endosomal Toll-like receptors (TLR-) 7 and 9, which, respectively, recognize single stranded RNA and unmethylated CpG motifs containing DNA, leading to type I IFN production via MyD88 signaling and IRF7-mediated transcription, the constitutively highly expressed master regulator of their type I IFN producing capacity (Honda et al., 2005b; Honda et al., 2005a). Several pDC-extrinsic mechanisms are also involved in pDC activation, during which pathogen-derived material is provided to pDC by other cells, allowing further checkpoints of pDC activation (Reizis, 2019). However, the precise cell–cell interactions and dynamic requirements are still not well understood.

Type I IFN is a potent immunoregulatory cytokine that modulates hundreds of genes (Stetson and Medzhitov, 2006; Ivashkiv and Donlin, 2014; Barrat et al., 2019). It promotes early innate immune responses by favoring cell-autonomous defense mechanisms that rapidly restrict viral escape and growth. Type I IFN signaling induces the production of proinflammatory cytokines that enable the mobilization of innate immune defenses and set the stage for adaptive immune responses by enhancing antigen processing, presenting, and costimulatory capacities of antigen presenting cells. It also directly signals to activated T and B cells, promoting expression of anti-apoptotic molecules and orchestrating their differentiation into robust effector cells. In contrast to these beneficial effects, excess or dysregulated type I IFN production that occurs in the context of auto-immune diseases, suggests that type I IFN can also promote pathology (Barrat and Su, 2019). While in viral infections, type I IFN generally appears as an absolute requirement for effective antiviral immune responses, in bacterial (Mycobacterium tuberculosis, Listeria monocytogenes) and parasitic (Plasmodium) infections, type I IFN can favor deleterious outcomes (Auerbuch et al., 2004; Spaulding et al., 2016; Sharma et al., 2011; Donovan et al., 2017; Mooney et al., 2017). This is often associated with the production of an inappropriate set of inflammatory mediators that hampers or skews the development of an effective host response and promotes damaging inflammation, as also exemplified during SARS-CoV-2 infection (Tay et al., 2020).

Consistent with the described roles of type I IFN, pDC-derived type I IFN has been reported to be essential for host resistance against a range of viral infections (Reizis, 2019; Barrat and Su, 2019). It augments antiviral defense through the rapid activation of innate immune effector responses such as that of NK cells, monocytes, and neutrophils (Cervantes-Barragan et al., 2012; Cervantes-Barragan et al., 2007; Swiecki et al., 2013; Webster et al., 2018; Davidson et al., 2011; Lynch et al., 2018). It also modulates long-term CD8+ and CD4+ T cell survival and differentiation (Swiecki et al., 2010). pDC-derived type I IFN also promotes autoimmunity in SLE and dermatitis, systemic sclerosis, and type I diabetes (Barrat and Su, 2019; Hansen et al., 2015; Ah Kioon et al., 2018; Guiducci et al., 2010). Similarly, in a mouse model of severe lethal malaria, pDC-derived type I IFN can drive innate immune cell activation and robust inflammatory responses (Spaulding et al., 2016).

Several mechanisms are implicated in pDC initiation of type I IFN production in vivo. These include TLR7/9 pDC-intrinsic sensing of viral nucleic acids occurring as a result of direct pDC infection by viruses (Lee et al., 2007; Décembre et al., 2014; Macal et al., 2012; Hornung et al., 2004). Indirect sensing of pathogen-derived materials from non-pDC-infected cells also occurs via exosomes, shedding of immature virus-containing particles or even internalization of immune complexes (Webster et al., 2016; Webster et al., 2018; Décembre et al., 2014; Dreux et al., 2012; Takahashi et al., 2010; Björck et al., 2008; Rönnblom and Alm, 2001). pDC may therefore interact with other cells or cell-derived materials, for effective pathogen sensing and subsequent initiation of type I IFN production. In line with this concept, several studies have reported that pDC forms rapid clusters upon TLR7/9-mediated activation (Asselin-Paturel et al., 2005; Mittelbrunn et al., 2009; Brewitz et al., 2017). Systemic injection of synthetic TLR7/9 ligands or murine CMV infection induces pDC to migrate via CXCR3 and CCR7 to the marginal zone of spleens where they achieve peak production of type I IFN (Asselin-Paturel et al., 2005). pDC can also form stable immunological synapses with CD4+ T cells in vitro and upon TLR9-induced activation in the marginal zone/red pulp area (Asselin-Paturel et al., 2005). More recently, lymph node (LN) pDC behavior monitored during viral infection showed that they were recruited via CXCR3 or CCR5, respectively, by infected subcapsular CD169+ MP or activated CD8+ T cells to the site of priming and interaction with XCR1+ DC (Brewitz et al., 2017). Here, type I IFN from pDC promoted (i) local macrophage (MP)-antiviral response preventing virus spreading and more efficient antigen cross-presentation by DC to CD8+ T cells. Consistent with these findings, the formation of a dynamic TLR7-dependent interferogenic synapse between pDC- and virus-infected cells was also described in vitro, in which viral RNA and type I IFN were exchanged as possible mechanisms potentiating pDC scanning of infected cells while providing local and targeted delivery of type I IFN (Assil et al., 2019). These results therefore suggested that tissue trafficking and localization of pDC in situ is essential for them to undergo activation and for their optimal function. However, the precise spatiotemporal dynamics of pDC behavior in vivo at steady state and during infection or inflammation, and which cells and molecular events may regulate these events still remain poorly understood. Whether pDC initiation of type I IFN production is linked to their motile behavior in vivo is also largely uncharacterized.

We have previously established that (1) CD169+ MP prime pDC to initiate rapid type I IFN production in the bone marrow (BM) of mice infected with lethal Plasmodium yoelii (Py) YM (Spaulding et al., 2016), (2) pDC priming involved pDC-intrinsic TLR7 and pDC-extrinsic STING sensing, and (3) pDC arrested in the BM of infected mice at the time of peak type I IFN production (~1.5 days) using intravital microscopy (IVM) while they were highly motile in the BM of naive mice. pDC arrested near or potentially in contact with CD169+ MP, however, these MPs were also densely distributed throughout the BM making it difficult to establish a functional interaction. CD169+ MP are tissue-resident MP, which localizes at entry ports of lymphatics in dLNs and in splenic marginal zones. They capture and sample antigens from dead cells, and act as entry points for viruses and bacteria, subsequently initiating immune defenses (Sung et al., 2012; Kastenmüller et al., 2012; Perez et al., 2017; Miyake et al., 2007). Building on our prior work, we directly tested the hypothesis that pDCs need to arrest and communicate with CD169+ MP in order to properly activate and initiate type I IFN production. We assess if CD169+ MP take up Py-infected red blood cells (iRBCs), acting as a primary source of parasite-derived materials and signals for pDCs. We also explore the molecular sensing mechanisms underlying the establishment of this functional interaction. Our results show that both CD169+ MP and TLR7 sensing in pDC control pDC arrest. We also establish that STING sensing in CD169+ MP is necessary to prime pDC production of systemic type I IFN, and to allow pDC egress from the BM.

Results

CD169+ MP take up P. yoelii-iRBC in the BM

Since CD169+ MP are required to activate pDC in the BM of blood stage malaria-infected mice (Spaulding et al., 2016), we hypothesized that CD169+ MP initiate pDC activation after taking up Py-iRBC. Flow cytometric analysis revealed the presence of two subpopulations of CD169+ MP in the BM, which we defined phenotypically as F4/80hi CD169int (F4/80hi MP subset) and F4/80int CD169hi (CD169hi MP subset) (Figure 1—figure supplement 1A). To explore uptake of Py-iRBC by BM MP, we isolated RBC either from naive (uninfected RBCs, uRBC) or Py-infected mice, and labeled them with the red fluorescent dye tetramethylrhodamine (TAMRA-Red+) before transfer to WT recipient mice for FACS(Fluorescence Activated Cell Sorting) and IVM analyses of the BM (Figure 1—figure supplement 1B). This strategy enabled quantification and visualization of the initial uptake phase of Py-iRBC versus uRBCs in the BM. For in vivo visualization of CD169+ MP by IVM, mice were injected i.v. with anti-CD169-FITC (or CD169-PE in other experiments) mAb 16 hr prior to IVM imaging (Spaulding et al., 2016). This allowed to identify CD169+ MP in the BM as FITChi or PEhi cells, and labeled the same cell subsets as when the Cre recombinase is expressed under the promoter of the gene encoding for CD169 (Siglec1Cre) crossed to conditional Rosa26-lox-STOP-lox (LSL)-tomato or -YFP mice (Figure 1—figure supplement 1C). As early as 6 hr post-transfer of TAMRA-Red+ RBC, both F4/80hi and CD169hi BM MP took up significantly increased proportions of Py-infected iRBCs compared to uRBCs, with a ~10-fold higher uptake of infected relative to uninfected RBC in both CD169+ MP subsets (Figure 1A). Of note, we did not detect direct uptake of Py-iRBCs by pDC (Figure 1—figure supplement 2A). Py-iRBC uptake was also comparable in mice injected with anti-CD169 mAb or control isotype Ab prior to the transfer of TAMRA-Red+ iRBC, ruling out that the in vivo labeling of CD169+ MP interfered with Py-iRBC uptake (Figure 1—figure supplement 2B). Time-lapse IVM imaging confirmed Py-iRBC attachment and accumulation on CD169+ MP in the BM, and possibly internalization (Figure 1B and Figure 1—video 1). Thus, taken together these data show that both populations of CD169+ MP take up Py-iRBC in the BM of infected mice in vivo, consistent with the hypothesis that MP can present parasite-derived materials and provide activating signals to pDC.

Figure 1 with 3 supplements see all
CD169+ MP uptake Plasmodium yoelii-infected RBC (iRBC) in the bone marrow and undergo activation and loss.

F4/80hi and CD169hi MP were gated as depicted in Figure 1—figure supplement 1A. (A) Representative dot plots of frequency of TAMRA-Red+ F4/80hi or CD169hi MP from control (no transfer), uninfected RBC (uRBC), and infected RBC (iRBC) transferred recipient mice are shown. Bar graphs show the mean frequencies ± standard error of the mean (SEM) of TAMRA-Red+ MP subsets in a pool of six independent replicate experiments with each symbol representing one mouse (n = 4–15 mice/condition). (B) Intravital image of TAMRA-Red+ Py-iRBC (red) associated with anti-CD169-FITC labeled CD169+ MP (green) in tibial BM 6 hr post-transfer. Sixteen hours prior to imaging, mice were i.v. administered CD169-FITC mAb to label CD169+ MP. Yellow arrowheads show Py-iRBC that are associated with CD169+ MP. Images are representative of two replicate experiments. (C, E, F) BM cells from uninfected or Py YM-infected (5 × 105 Py-iRBC) WT mice were harvested at indicated time points and stained with cell-surface expression of linneg (CD3, CD19, NK1.1, CD19), F4/80, CD169, and several activation markers (MHC-II, ICAM-1, CD80, Sca-1). In (C), representative FACS dot plots of CD169hi and F4/80hi MP in the BM. The relative proportion and absolute numbers CD169hi and F4/80hi MP in mouse leg of WT B6 mice are shown in a pool of two replicate experiments with SEM (n = 3–7 mice for each time point). (D) Representative image of tibial BM of PTCRA-EGFP mice administered with CD169-PE mAb 16 hr prior to imaging either uninfected or 36 hr post Py infection in 1 of >10 imaged BM. (E, F) Representative overlay FACS histograms for indicated activation marker on CD169hi and F4/80hi MP from indicated mice and conditions. When relevant, p values were calculated with *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test.

CD169+ MP undergo massive loss and activation in Py-infected BM

If uptake of Py-iRBCs is indeed important for initial sensing of parasite-derived molecular patterns by MPs, we hypothesized that their activation status should be significantly modulated. Thus, we monitored both F4/80hi and CD169hi MP numbers and activation using various cell-surface markers related to type I IFN signaling (Sca-1), antigen presentation (MHC-II, CD80, CD86), and adhesion/uptake (ICAM-1, CD11a) following Py infection (Figure 1C–F and Figure 1—figure supplement 2C). We found a significant loss (>50%) of both MP subsets in the BM of Py-infected mice, that occurred between 30 and 36 hr post-infection as quantified by FACS analysis (Figure 1C) and IVM imaging of the tibial BM where the decrease in the overall BM MP numbers was also visualized in situ (Figure 1D). In addition, Sca-1, CD80 costimulatory molecule, and ICAM-1 integrin ligand expression were increased while CD11a was diminished, consistent with robust activation of both subsets of MP in response to Py infection (Figure 1E, F and Figure 1—figure supplement 1D). In summary, CD169+ MP in the BM of Py-infected mice are rapidly lost as they undergo robust activation.

CD169+ MP are required for pDC arrest in Py-infected BM

To directly link CD169+ MP priming of pDC to pDC motile behaviors, we next assessed whether Py-activated after Py infection control pDC arrest using time-lapse IVM of the tibial marrow. As an initial step, we established baseline pDC motile behaviors using several readouts comparing uninfected versus Py-infected WT mice at the time pDC produce peak type I IFN (i.e., 36 hr). We took advantage of the PTCRA-EGFP reporter mice, that express the enhanced green fluorescent protein (EGFP) under the control of the human pre-T cell receptor α (PTCRA) promoter. This promoter is selectively expressed in BM pDC (>95% of GFP+ cells, Figure 2—figure supplement 1A, Shigematsu et al., 2004) enabling us to visualize pDCs in vivo. For CD169+ MP visualization, mice were injected intravenously with the anti-CD169-PE mAb prior to IVM imaging (Figure 1—figure supplement 1C). To quantify pDC dynamic behavior, we utilized several metrics, which include track velocity (TV) that measures cell path length over time, displacement rate (DR) that measures overall distance cells moved (irrespective of path) over time, and mean square displacement (MSD) that classifies cell movement as random, confined, or processive and inform on the cell behavior. In WT uninfected mice, pDC was motile, exhibiting long tracks and a TV of 7.14 μm/min (Figure 2A, B and Figure 2—video 1), faster than their reported speed of 5 μm/min in LNs (Mittelbrunn et al., 2009), but slower than naive T cells in the LN that travel at 10–15 μm/min (Shigematsu et al., 2004). Based on their DR (1.5 μm/min) and MSD, pDC migrated through and explored a large volume of parenchymal tissue (~1350 μm2/hr), similar to LN lymphocytes (Mittelbrunn et al., 2009) and consistent with a scanning role in the BM (Figure 2B–D). In contrast, CD169+ MP were sessile, whether in naive or infected mice (Figure 2—figure supplement 1B). After Py infection, pDC tracks, TV (5.13 μm/min), DR (0.93 μm/min), and MSD were all significantly reduced. These quantitative measures are also consistent with their arrest visualized by IVM. Average mean values of these distinct measurements from independent mice reflected these findings. We next tested whether CD169+ MP were involved in pDC arrest after Py infection. We used (1) PTCRA-EGFP reporter mice crossed to siglec1DTR/DTR mice and (2) sublethally irradiated siglec1DTR/WT or control WT recipient mice (500 rads) reconstituted with WT PTCRA-EGFP reporter BM cells. These chimeras allowed us to rule out any substantial impact of the DTR transgene in pDC since WT PTCRA-EGFP+ pDC did not express the transgene. In siglec1DTR mice, whether chimeric or not, diphtheria toxin (DT) injection induced the selective depletion of both subsets of CD169+ MP (Figure 2—figure supplement 1C). Interestingly, IVM imaging of Py-infected DT-treated mice revealed much longer pDC tracks compared to untreated control mice (Figure 2E). Moreover, pDC DR and MSD were significantly increased compared to untreated mice (by ~50%, Figure 2F, G and Figure 2—video 2). As expected, in Py-infected, DT-treated PTCRA-EGFP/WT control chimeras, pDC had significantly reduced motility compared to uninfected. Average mean values from independent mice were also consistent with these conclusions. Thus, these data collectively establish that CD169+ MP control pDC dynamics during Py infection and suggest that pDC arrest is linked to their ability to initiate type I IFN production.

Figure 2 with 3 supplements see all
CD169+ MP control pDC arrest in Py-infected mice.

PTCRA-EGFP reporter mice (WT, siglec1DTR/WT, or partial bone marrow chimeras) either received PBS (uninfected) or were infected with 5 × 105 Py-iRBCs, and administered with CD169-PE mAb 16 hr prior to intravital imaging microscopy (IVM) of tibial BM. (A–D) Analysis of pDC dynamic behavior in uninfected or 36 hr Py-infected WT PTCRA-EGFP mice. (A) Flower plots of 2D tracks of individual pDCs, superimposed after normalization of their starting coordinates to the origin over 30 min of imaging. (B) Track velocity (TV) and displacement rates (DRs) of individual pDC with (C) average in all mice, and (D) mean square displacement (MSD) analysis of pDC over time and slope of MSD/motility. For bar graphs, each symbol features one mouse with mean ± standard error of the mean (SEM) (n = 3–7 mice were imaged for each experimental condition and genotype). (E–G) PTCRA-EGFP siglec1DTR/WT or PTCRA-EGFP/siglec1DTR/WT and PTCRA-EGFP/WT partial BM chimeras were injected with diphtheria toxin (DT) or control PBS 12 hr prior to Py infection. (E) Tracks of individual pDC in the indicated experimental conditions over 1 hr of imaging. (F) DRs of individual pDC and average in all mice. (G) MSD analysis of pDC over time and slope of MSD/motility. For all bar graphs, each symbol features one mouse with mean ± SEM (n = 2–4 mice were imaged for each experimental condition and genotype). p values were calculated with *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test for flow-cytometry comparisons and Welch’s t-test for comparison of IVM quantifications. Multiple linear regression analyses were applied for statistical analysis of the MSD plots.

pDC arrest in Py-infected BM depends on TLR7-intrinsic signaling

To achieve type I IFN production, pDC requires TLR7-intrinsic sensing (Spaulding et al., 2016). If pDC arrest is indeed linked to their robust activation, we predicted that TLR7 sensing in pDC would also be required for their arrest. We first generated Tlr7−/y and WT PTCRA-EGFP reporter mice and monitored pDC motility in the tibial BM of Py infected (36 hr post-infection), or control uninfected live mice using IVM imaging. In both Tlr7−/y and WT uninfected mice, pDC was highly motile with long tracks and a comparable TV (respectively, 6.52 and 7.16 μm/min, Figure 3A, B and Figure 3—video 1). Interestingly, pDC in Tlr7−/y uninfected mice exhibit greater motility, suggesting that tonic TLR7 signals regulate pDC motility under steady-state conditions (Figure 3D). Following Py infection in WT mice, pDC motility was reduced by 31%, in terms of TV (4.92 μm/min) and 45% based on DR (~0.87 μm/min), as well as for MSD and slope of MSD averages (Figure 3B, D and Figure 3—video 2). In contrast, pDC in Py-infected Tlr7−/y mice did not arrest and continued migrating with similar speeds in terms of TV (7.05 μm/min) and DR (1.55 μm/min, Figure 3B, C) despite detectable measurable MSD reductions (Figure 3D) reflecting a more restricted migration pattern, but not arrest. Average mean values from independent mice also largely confirmed these conclusions, overall suggesting that while other factors than TLR7 signals control pDC motility, these remain insufficient to completely arrest them. Representing the proportion of pDC moving or arrested based on WT pDC threshold values (both for TV and DR) further highlights a comparable proportion of pDC moving or arrested in Py-infected Tlr7−/y mice and naive WT or Tlr7−/y mice, in contrast to Py-infected WT counterpart (Figure 3—figure supplement 1A). To further assess if cell-intrinsic TLR7 sensing mediated pDC arrest, we generated partial mixed BM chimera mice as above, in which WT recipient mice (CD45.1+) were grafted with either Tlr7−/y- or WT-PTCRA-EGFP reporter BM after sublethal irradiation. In these mice, ~30% of hematopoietic-derived cells originated from the donor reporter BM, allowing for tracking of Tlr7−/y or control WT PTCRA-EGFP+ pDC in an environment where ~70% of HSC-derived cells and all radioresistant cells, were intact for TLR7 signaling (Figure 3—figure supplement 1B). Consistent with the prior result (Figure 3A–D), Tlr7−/y pDC in infected chimeras exhibited long tracks and did not arrest while WT pDC had much shorter tracks and arrested, establishing that TLR7 signaling intrinsic to pDC, is required for them to stop during infection (Figure 3E, F and Figure 3—video 3). We also quantified pDC/CD169+ MP contact durations using a Kiss and Run algorithm (Imaris) as a measure of their possible interactions in chimeric mice and found that WT pDC significantly increased their contact times with the MP during infection (from ~2 to 16 min) whereas Tlr7−/y pDC interacted with the MP for comparable amounts of time (~4–5 min) whether mice were infected or not (Figure 3G). Thus, taken together, these results show that TLR7 sensing in pDC is required for their arrest and potential interactions with CD169+ MP in the BM of malaria-infected mice at the time pDC produce peak type I IFN. Since CD169+ MP are required for pDC arrest, it further suggests that MP are providing TLR7 ligands to pDC.

Figure 3 with 4 supplements see all
TLR7 sensing in pDC is required for their arrest and clustering in Py-infected mice.

WT and Tlr7−/y PTCRA-EGFP reporter mice, and WT PTCRA-EGFP/WT and Tlr7−/y PTCRA-EGFP/WT partial bone marrow (BM) chimeras either received PBS (uninfected) or were infected with 5 × 105 Py-iRBCs, and administered with CD169-PE mAb 16 hr prior to intravital microscopy (IVM) of tibial BM. (A) Tracks of individual pDC dynamic behavior in uninfected or 36 hr Py-infected mice with starting positions in the same origin over 30 min of imaging. (B) Track velocity (TV) and displacement rates (DRs) of individual pDC with (C) average in all mice, and (D) mean square displacement (MSD) analysis of pDC over time and slope of MSD/motility. Each symbol feature one mouse with mean ± standard error of the mean (SEM). Four to six individual mice were imaged for each experimental condition and genotype. (E) Tracks of individual pDC in uninfected or Py-infected WT PTCRA-EGFP/WT and Tlr7−/y PTCRA-EGFP/WT partial BM chimeras over 1 hr of imaging. (F) DR of individual pDC and (G) quantification of individual pDC contact time with CD169+ MP are shown. Three to four individual mice were imaged for each experimental condition and genotype. p values were calculated with *p < 0.05; **p < 0.01; ; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test for flow-cytometry comparisons and Welch’s t-test for comparison of IVM quantifications. Multiple linear regression analyses were applied for statistical analysis of the MSD plots. The WT PTCRA-EGFP data depicted in (A) are the same as in Figure 2A; (B, C) are identical as in Figure 2B, C; (D) are the same as in Figure 2D; (F) are the same as in Figure 2F. This enabled comparisons across mutant mouse conditions relative to WT.

Altered chemotactic receptor and adhesion molecule expression on pDCs from Py-infected Tlr7−/y compared to WT mice

Because TLR7 sensing alters pDC dynamic behavior in vivo, we next hypothesized that TLR7 signaling likely modulated both their migratory and adhesion characteristics. Several chemotactic receptors (CXCR3, CXCR4, CX3CR1, CCR2, CCR5, CCR9) and adhesion molecules (CD11a, ICAM-1, VCAM-1) were reported to be expressed on pDC in naive mice and to contribute to their trafficking and ability to cluster upon activation (Asselin-Paturel et al., 2005; Brewitz et al., 2017; Sawai et al., 2013). We monitored cell-surface expression of chemokine receptors (CXCR3, CCR9, CXCR4, CX3CR1) and integrin ligands (ICAM-1, VCAM-1) on pDC in uninfected compared to Py-infected WT and Tlr7−/y mice (Figure 3—figure supplement 1C). While basal expression of these receptors and integrins was comparable in all groups of uninfected mice, CCR9 and CX3CR1 were significantly upregulated on pDC after infection in WT but not in Tlr7−/y mice. Significant increases in CXCR4 expression were not reliably detectable. In contrast, CXCR3 expression was significantly downregulated on pDC in both WT and Tlr7−/y mice, but to different extents (90% and 50%, respectively). Expression of ICAM-1, a ligand for the integrin LFA-1, was significantly upregulated in both WT and Tlr7−/y after infection, but the magnitude of upregulation remained substantially greater in WT mice (factor of ~3). Expression of CD11a, the alpha chain of LFA-1, was slightly decreased in Py-infected WT mice whereas expression did not change in Tlr7−/y mice. Lastly, expression of VCAM-1, ligand for the integrin VLA-4, was significantly reduced in both WT and Tlr7−/y during the infection, but to a lesser extent in Tlr7−/y mice. Collectively, these data link TLR7-mediated sensing by pDCs to changes in cell-surface expression of several chemotactic receptors and adhesion molecules that may contribute to modifications in pDC motility, homing, and/or arrest that occur following Py infection.

GαI-protein signaling is required for inducing pDC production of type I IFN

Since we found important modulations in chemokine receptor expression driven by TLR7 signaling, we hypothesized that chemotaxis was involved in pDC motility and arrest. To test this possibility, we injected Py-infected mice with either PBS (control) or pertussis toxin (PT), which blocks GαI-protein-coupled receptor signal transduction and related chemotaxis, and subsequently analyzed pDC motility, morphology, and type I IFN production (Figure 4 and Figure 4—video 1). In Py-infected PTCRA-EGFP reporter mice, PT treatment led to stronger immobilization of pDC compared to control mice, with shorter tracks and significantly lower DR (Figure 4A, B). pDC arrest was also concomitant with distinct morphological changes (Figure 4C). Notably, pDCs in PT-treated mice exhibited a round shape contrasting with the elongated dendritic shape observed in PBS-treated counterparts. Quantitative analysis of individual pDC length, surface, and shape factor (a measure of cell sphericity) further revealed significant differences in PT- versus PBS-treated Py-infected mice (Figure 4D). While after PT-treatment pDCs were immobile and unable to migrate, they were still viable in the BM and could potentially sense freely diffusing parasite-derived ligands. To determine whether PT treatment impacted the ability of pDC to make type I IFN, we carried out the same experiment in IfnbYFP/YFP reporter mice (Scheu et al., 2008), which revealed an 80% reduction in type I IFN-expressing pDC (YFP+) compared to control infected group (0.44% vs. 2.1% YFP+ pDCs, Figure 4E). Taken together with the imaging findings, this indicated that GαI-protein-dependent transduction was required to achieve maximal type I IFN production. Despite their drastic immobilization and morphological changes, pDC in PT-treated infected mice still modulated cell-surface expression of costimulatory (CD86) and adhesion (CD62L, ICAM-1, VCAM-1) molecules, as well as chemotactic receptors (CXCR3), compared to PBS-treated infected and uninfected control mice (Figure 4E). Specifically, expression of CD62L, ICAM-1, VCAM-1, and CXCR3 but not CD86, were significantly altered in PT- compared to PBS-treated Py-infected mice. These results show that blocking GαI-protein-coupled receptor signal transduction including chemotaxis, alters expression of most but not all of these markers, and the acquisition of type I IFN-secreting capacity by the pDC.

Figure 4 with 1 supplement see all
pDC arrest is not sufficient to achieve robust type I IFN production during Py infection.

WT PTCRA-EGFP (A–D) or IfnbYFP/YFP (E, F) reporter mice either received PBS or pertussis toxin (PT) and were infected with 5 × 105 Py-iRBCs. In (A–D), mice were injected with PT or control PBS at the time of Py infection and administered with CD169-PE mAb 16 hr prior to intravital microscopy (IVM) of tibial bone marrow (BM). The analysis of pDC dynamic behavior of PT-treated or untreated, 36 hr Py-infected mice are shown. (A) Flower plots of 2D tracks of individual pDCs, superimposed after normalization of their starting coordinates to the origin over 30 min of imaging. (B) Displacement rate (DR) of individual pDC. (C) Representative IVM images of tibial BM. (D) Quantification of cell morphology of arrested pDC in PBS versus PT-treated infected mice. (E, F) IfnbYFP/YFP reporter mice received indicated treatments (PBS, PT, Isotype Ab, anti-CXCR3+/−CXCL9+/−LFA-1, anti-LFA-1+/−VLA-4) and the time of Py infection, and 36 hr later BM cells were stained for cell-surface expression of linneg (CD3, CD19, NK1.1, CD11b, Gr1), BST-2, Siglec-H, CD86, CD62L, CXCR3, VCAM-1, and ICAM-1. Bar graphs quantify the expression of YFP, CD86, and ICAM-1 by pDCs with mean ± standard error of the mean (SEM) across two to five independent replicate experiments, with each symbol featuring one mouse (n = 3–10 mice). p values were calculated with *p < 0.05; **p < 0.01; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test for flow-cytometry comparisons and Welch’s t-test for comparison of IVM quantifications.

The fact that PT treatment prevented pDC motility, and both CD169+ MP and TLR7 sensing were required for pDC arrest and type I IFN production, suggested that pDCs do not encounter freely circulating TLR7 ligands in the BM but rather that such ligands must be presented by Py-activated CD169+ MP through close and durable contacts with pDC. However, as PT treatment has pleiotropic activity and may induce indirect effects on other cells, we next attempted to determine more specifically which chemotactic and adhesion receptors mediated pDC homing and activation. CXCR3 is significantly downregulated after Py YM (Figure 3—figure supplement 1B) and has been shown to promote chemotaxis of pDC to sites of priming in the LN (Brewitz et al., 2017). CXCR3 has two ligands, CXCL10 and CXCL9, which are highly expressed by monocytes and MP during infections (Soudja et al., 2014; Kastenmüller et al., 2013). Thus, we tested if CXCR3 was functionally contributing to pDC activation in the BM. We injected IfnbYFP/YFP mice at the time of infection both with anti-CXCR3 (CXCR3-173 clone) and anti-CXCL9 2A6.9.9 mAb, which blocks the CXCL10-binding site and neutralizes soluble CXCL9, respectively (Figure 4F). pDC in these mice expressed YFP at similar levels relative to control-isotype Ab-treated or non-treated mice (1.5–2%). Furthermore, Ab blocking of LFA-1 and VLA-4 integrins, with or without anti-CXCR3 mAb, also failed to decrease expression of type I IFN by the pDC, suggesting that none of the tested reagents were sufficient to recapitulate the in vivo effects of PT treatment on pDC activation. Collectively, these results suggested that pDC motility requires GαI-dependent chemotaxis. This process controls pDC homing to where TLR7 ligands are, and direct or indirect functional interactions with CD169+ MP to initiate type I IFN production.

STING signaling is not required for pDC arrest during Py infection

Since (1) STING signaling extrinsic to pDC is required for pDC to secrete type I IFN during Py YM infection (Spaulding et al., 2016), and (2) pDC arrest and possible functional interaction with MP, is linked to their full activation, we hypothesized that STING sensing could also contribute to pDC homing, arrest and communication with MP to initiate type I IFN production. To investigate these possibilities, we analyzed pDC motility in Sting1Gt/Gt PTCRA-EGFP reporter mice by IVM. In uninfected Sting1Gt/Gt mice, most pDC exhibited high motility while they arrested in Py-infected counterparts (Figure 5A and Figure 5—video 1). The majority of pDC tracks were confined and with reduced displacement from origin following the infection. While pDC DR and MSD slopes were significantly diminished compared to naive mice, the average pDC TV remained elevated (6.52 μm/min), similar to that of uninfected mice (6.33 μm/min) (Figure 5B–D). Such discordance between TV and DR usually means that cells exhibit a more confined migration pattern. Consistent with this interpretation, the MSD of pDC was significantly higher in Sting1Gt/Gt compared to WT-infected mice (Figure 5D). We next infected partial BM chimeras in which sublethally irradiated Sting1Gt/Gt or control WT mice were reconstituted with WT PTCRA-EGFP donor BM, allowing to monitor WT pDC motility in a Sting1Gt/Gt host where the majority of HSC-derived cells (~70%) and all radioresistant cells are deficient for STING. In these mice where we visualized WT pDC in a Sting1Gt/Gt host, pDC arrested and had longer contact durations with MP compared to control WT chimera and non-chimera mice (Figure 5E, F and Figure 5—video 2). Most interestingly, we observed that in both non-chimera Py-infected Sting1Gt/Gt mice, pDC clusters were significantly larger in volume than in WT control mice, by threefold on average, and in some cases exceeding WT clusters by tenfold (Figure 5G, H and Figure 5—videos 2 and 3). We also enumerated by FACS analysis higher numbers of pDC in the BM of Sting1Gt/Gt compared to WT IfnbYFP/YFP reporter Py-infected mice (Figure 5I), correlating with the significantly larger clusters of pDC quantified in the BM of Sting1Gt/Gt mice. As expected, the number of Ifnb-expressing pDC (YFP+) was significantly reduced compared to WT counterparts, despite increased pDC numbers. In addition, while still significantly modulated, expression of several chemotactic receptors (CXCR3, CCR9, CXCR4, CX3CR1) and the integrin VCAM-1, failed to undergo as much cell-surface modulation as in WT mice (Figure 5—figure supplement 1). This suggested that pDC in Sting1Gt/Gt mice may not receive appropriate signals to enable them to resume motility, leave clusters and egress from the BM. Taken together, these data establish that pDC-extrinsic, STING-dependent signals do not control pDC homing and arrest in the BM during Py infection, but that they may control pDC functional interactions with CD169+ MP and egress from the BM, thereby limiting their accumulation and/or retention in the BM.

Figure 5 with 4 supplements see all
pDCs arrest, form larger clusters and accumulate in the bone marrow of STING-deficient compared to WT Py-infected mice.

WT or Sting1Gt/Gt PTCRA-EGFP reporter mice either received PBS (uninfected) or were infected with 5 × 105 Py-iRBCs, and administered with CD169-PE mAb 16 hr prior to intravital microscopy (IVM) imaging of tibial BM. The analysis of pDC dynamic behavior of uninfected or 36 hr Py-infected mice is shown. (A) Tracks of individual pDC in Sting1Gt/Gt PTCRA-EGFP reporter mice with starting positions in the same origin over 30 min of imaging. (B) Track velocity (TV) and displacement rate (DR) of individual pDC with (C) average in all mice, and (D) mean square displacement (MSD) analysis of pDC over time and slope of MSD/motility. Each symbol features one mouse with mean ± standard error of the mean (SEM). Three to seven individual mice were imaged for each experimental condition and genotype. (E) DR of individual pDC and (F) individual pDC contact time with CD169+ MP, in uninfected or Py-infected Sting1Gt/Gt PTCRA-EGFP/WT and WT PTCRA-EGFP/WT partial BM chimeras. (H) Quantification of cluster volume in WT, Sting1Gt/Gt, Sting1Gt/Gt PTCRA-EGFP/WT, and WT PTCRA-EGFP/WT partial BM chimeras. (G) Image from IVM of tibial BM showing pDC clusters 36 hr post Py infection. For (E–H), three to four individual mice were imaged for each experimental condition and genotype. (I) Number of pDC per leg and relative proportion of YFP+ pDC in the BM of uninfected or Py-infected (36 hr) WT- and Sting1Gt/Gt-IfnbYFP/YFP reporter mice, as quantified by FACS after staining for cell-surface expression of linneg (CD3, CD19, NK1.1, CD11b, Gr1), BST-2, Siglec-H. Bar graphs quantify the expression of YFP in pDCs with mean ± SEM across two independent replicate experiments (n = 4–6 mice/group). p values were calculated with *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test for flow-cytometry comparisons and Welch’s t-test for comparison of IVM quantifications. Multiple linear regression analyses were applied for statistical analysis of the MSD plots. Data of WT PTCRA-EGFP depicted in (B, C) are the same as in Figures 2B, C, 3B, C; (D) are also in Figures 2D and 3D. WT PTCRA-EGFP/WT partial BM chimera in (E) are also in Figures 2F and 3F. These are included for ease of comparisons across conditions relative to WT.

STING signaling in CD169+ MP controls pDC production of type I IFN and egress from the BM

Next, we assessed if STING signaling specifically in CD169+ MP during Py YM infection controlled (1) initiation of type I IFN production by pDC, and (2) pDC accumulation versus egress from the BM. We bred siglec1Cre/Cre knock-in mice to Sting1Gt/Gt or Sting1F/F mice (where exons 3–5 of Sting1 were flanked with two LoxP sites) to generate siglec1Cre/WTSting1Gt/F mice, in which CD169+ MP lacked expression of functional STING. All these mice were also homozygous for the IfnbYFP reporter transgene, allowing us to monitor type I IFN expression by FACS. This reporter faithfully tracks with the production of IFNα and IFNβ cytokines as we established before (Spaulding et al., 2016). Upon Py YM infection, while ~2.2% of pDC (~4200 pDC/leg) in WT and Sting11Gt/WT control mice expressed IFNβ (YFP+ ICAM-1hi), we quantified a ~60% reduction in YFP+ pDC when STING was lacking in CD169+ MPs (Figure 6A), demonstrating that STING signaling in CD169+ MP was required for pDC priming of type I IFN producing capacity. Other markers strongly modulated on pDC upon activation (ICAM-1, CXCR3, CD86) underwent similar changes as in WT mice, suggesting either multistep activation or independent processes (Figure 6—figure supplement 1A). Interestingly, in these mice where CD169+ MP lacked STING, the number of pDC in BM was ~1.5 times higher than in WT or Sting11Gt/WT control groups (p value = 0.1, Figure 6B). Similar increases in numbers of pDC were also enumerated in the BM of whole-body STING-deficient (Sting1Gt/Gt) or CD169+ MP-depleted (DT-treated siglec1DTR/WT) infected mice, consistent with retention of pDC in BM when CD169+ MP or STING-dependent signals were absent (Figure 5G–I and Figure 5—videos 2 and 3). Collectively, these results show that STING signaling in CD169+ MP during Py infection, provide pDC with key signal(s) to initiate type I IFN production and egress from the BM.

Figure 6 with 1 supplement see all
STING signaling in CD169+ MP is required for plasmacytoid dendritic cells (pDCs) production of type I interferon (IFN) and egress from the bone marrow (BM).

WT-, Sting1Gt/WT-, or CD169Cre/CreSting1Gt/F-IfnbYFP/YFP reporter mice were infected with 5 × 105 Py-iRBCs. Thirty-six hours later BM cells were stained with live/dead and for cell-surface expression of linneg (CD3, CD19, NK1.1, Gr1), CD11b, BST-2, Siglec-H, CD169, F4/80, and indicated cell-surface markers. (A) Representative FACS dot plots of YFP/ICAM-1+ pDC are shown, together with the proportion and numbers of YFP+ pDC/leg in summary bar graphs. (B) Number of pDC per leg in indicated mouse genotype. (C) Representative dot plots of CD169hi and F4/80hi MP in indicated mice. Bar graphs show numbers per leg and proportions. Data represent the pool of three independent replicate experiments with each symbol featuring one mouse (n = 3–18 mice). p values were calculated with *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant, using two-tailed unpaired Student’s t-test. (D) Working model: during Py YM malaria infection, CD169+ MP rapidly take-up Py-infected RBC (step 1). Next, pDC, which is highly motile in the BM of naive mice, home to and establish interactions with CD169+ MP at peak production of type I IFN by pDC (steps 2 and 3). This process is dependent on GaI-mediated chemotaxis. CD169+ MP and TLR7 sensing in pDC are both required for pDC arrest. Lastly, STING signaling in CD169+ MP is needed for pDC to initiate type I IFN secretion, regulate pDC clustering/retention, and egress from the BM to the blood (step 4).

STING-mediated activation of pDC is independent of CD169+ MP loss

Upon Py infection, a majority of CD169+ MP are rapidly lost (Figure 1). We hypothesized that STING signaling may control their loss and thus contribute to providing activating ligands and signals to pDC. We therefore assessed whether STING contributed to MP loss during Py infection. In whole-body Sting1Gt/Gt mice, CD169hi but not F4/80hi MP, were partially protected from loss compared to WT control mice (by ~50%, Figure 6C), suggesting that this may prevent optimal pDC priming. However, in siglec1Cre/WTSting1Gt/F mice where STING is selectively lacking in CD169+ MP, both MP populations underwent similar loss as in WT mice, suggesting STING deficiency extrinsic to MP, protects CD169hi MP from loss. However, in siglec1Cre/WTSting1Gt/F mice, pDC failed to initiate type I IFN transcription and accumulated in the BM compared to control groups (Figures 5I and 6B), consistent with a model in which STING signaling in CD169+ MP provide activating signals to pDC that are unrelated to their loss. Of note, MP activation 36 hr post-infection, as measured by cell-surface modulation of MHC-II, CD80, ICAM-1, and Sca-1, was mostly comparable in all infected groups, likely ruling out a general activation defect of STING-deficient MP (Figure 6—figure supplement 1B). In summary, these data uncouple MP loss that occurs during Py infection from STING-mediated priming of pDC by CD169+ MP.

Discussion

Here, we reveal a complex multi-cell choreography that orchestrates pDC production of type I IFN during malaria infection. We establish the existence of a functional interaction (direct or indirect) between CD169+ MP and pDC in vivo that is needed to license pDC priming of type I IFN production in vivo during this infection. We link pDC arrest to their optimal priming. We show that CD169+ MP take up parasite-infected RBCs in the BM. We find that both CD169+ MP and pDC-intrinsic TLR7 sensing are required for pDC arrest, suggesting that MP is providing TLR7 ligands to pDC. We also show that MP-intrinsic STING sensing is required to activate pDC production of type I IFN, and enable their egress/release from the BM after priming (Figure 6D).

These findings are therefore consistent with a model in which CD169+ MP in the BM act as first-line responders to infection, akin to other CD169+ MP found in the LN and spleen (Grabowska et al., 2018). Depletion of these MP prior to infection prevents type I production by pDCs (Spaulding et al., 2016), indicating that these antigen presenting cells are non-redundant and uniquely capable of providing Py ligands and/or other innate signals to pDC in the BM. The fact that depletion of CD169+ MP and cell-intrinsic lack of TLR7 prevent pDC arrest establishes that pDC motility, homing and functional interactions with MP are prerequisites to achieve type I IFN production. These results suggest that TLR7 ligands, which trigger pDC arrest and type I production, are also not freely diffusing in the BM and/or that in vivo, pDC require nucleic acids to be properly presented or packaged to access TLR7-containing endosomes in pDC. In either case, we propose that direct or indirect interactions with CD169+ MPs are required to deliver Py-derived ligands to pDCs in the BM. This finding is consistent with prior studies showing pDC sensing of hepatitis C-infected hepatocytes via viral RNA-containing exosomes (Dreux et al., 2012; Takahashi et al., 2010) or cellular particles containing Dengue virus (Décembre et al., 2014). pDC was also reported to be directly infected by specific viruses such as herpes simplex virus and arenaviruses, and to utilize specialized intracellular processes like autophagy to sense viral nucleic acids and initiate type I IFN production (Lee et al., 2007; Macal et al., 2012; Assil et al., 2019). Given the very low frequency and numbers of pDC but their rapid response, it is likely that pDC are largely using such indirect but more effective modes of recognition. The formation of specialized cell–cell structures between pDC and infected cells or cells presenting pathogen-derived molecules, is indeed proposed to mediate both uptake and sensing of TLR ligands and the local delivery of type I IFN in interferogenic synapses (Assil et al., 2019).

Proper clustering and arrest of pDCs is a TLR7 guided process. TLR7 is required in pDC for their arrest, and type I production, as well as extensive remodeling of pDC expression of chemokine receptors and adhesion molecules, likely promoting stable contacts with activated, Py-loaded MPs. In vivo, pDC are recruited via CXCR3- and CCR5-dependent chemotaxis to virus-infected subcapsular draining LN MP or CD8+ T cells, respectively, for local delivery of type I IFN, rapidly restricting early virus spreading while promoting effective T cell priming (Brewitz et al., 2017). Given that pDC expresses multiple chemotactic receptors, and some are significantly up- or downregulated during Py infection (e.g., CCR9, CXCR3, CX3CR1), these processes are likely to be redundant. Consistent with this interpretation, blockade of CXCR3-mediated chemotaxis failed to recapitulate all GαI-mediated chemotaxis blockade. Even co-blockade of several adhesion molecules (LFA-1, VLA-4) involved in classical leukocyte adhesion and clustering, did not abrogate the production of type I IFN by pDC, suggesting that other chemotactic and adhesion mechanisms are involved and further investigations are needed.

Finally, we establish TLR7-dependent arrest and interaction of pDC with CD169+ MP. We also show that full activation of pDC requires STING signaling in CD169+ MP and that MP control pDC arrest. It was therefore unexpected that we did not find STING controls pDC arrest. However, we revealed that in STING-deficient mice, pDC formed significantly greater size clusters and accumulated in higher numbers in the BM compared to WT counterparts. This result was confirmed in mice lacking STING in the radioresistant compartment and only in CD169+ MP, consistent with a model where STING-dependent signals control pDC egress from the BM, and as a consequence, the organization of pDC clusters. In our previous work (Spaulding et al., 2016), we reported that pDC numbers in the BM of Py-infected mice decreased after peak type I IFN production while that of blood pDC increased. Larger clusters and increased numbers of pDC in STING-deficient mice, and in mice with STING-deficient CD169+ MP, may indicate that pDC are not fully activating and thus remain docked, awaiting for additional signals. These large clusters of pDC are further evidence that pDC priming is spatially segregated in the BM and that activation signals are not freely diffusing. These results connect the choreography of the cell–cell signal transmissions with pDC changes in arrest and migration for efficient regulation. Other signals that need to be elucidated are likely to regulate these cellular interactions in vivo.

In particular, which STING-dependent signal(s) are delivered from Py-activated CD169+ MP to pDC remains unknown. STING was originally identified as a cytosolic nucleic-acid sensor, which triggering leads to the induction of a robust type I IFN response via IRF3 and NFkB, and a rapid antiviral state that enables the control of viral infections (Zhong et al., 2008; Ishikawa and Barber, 2008). While pDC can secrete type I IFN independently of type I IFN signaling during VSV and MCMV infections (Tomasello et al., 2018; Barchet et al., 2002), pDC are also known to require type I IFN autocrine signaling when responding to TLR7/9 synthetic ligands in vivo (Asselin-Paturel et al., 2005). Thus, type I IFN may be the STING-dependent missing signal given by CD169+ MP during malaria infection. Several studies provide evidence that CD169+ subcapsular marginal zone and sinus MP in the spleen and LN, respectively, may produce type I IFN in response to viral infections (VSV, Vaccinia, MCMV) (Gui et al., 2019; Brault et al., 2018; Gaidt et al., 2017). However, since STING also triggers the proinflammatory transcriptional regulator NFkB, other cytokines could be the missing signal. STING signaling can also initiate distinct inflammatory cell deaths and autophagy as a defense mechanism. Among STING-mediated cell death, lysosomal cell death, pyroptosis, necroptosis, and NETosis were described to occur in myeloid cells in response to cGAS-STING sensing/signaling of viral DNA (Gui et al., 2019; Brault et al., 2018; Gaidt et al., 2017; Lood et al., 2016). While we observed a major loss of CD169+ MP in the BM (>60%) of Py-infected WT mice when type I IFN levels peak (30–36 hr), STING-deficient MP underwent similar loss, most likely excluding STING involvement in MP death. However, STING-driven autophagy could provide live CD169+ MP with Py-derived nucleic acids and other products for optimal presentation to pDC.

STING is generally shown to require cGAS for cytosolic double stranded DNA recognition (Sharma et al., 2011) and the production of STING-activating cyclic dinucleotides second messengers (Chen et al., 2016). Our results are discrepant with a prior study that also used lethal murine Py YM, and concluded that STING sensing of Py parasites in pDC prevented strong type I IFN responses by pDC through pDC-intrinsic SOCS1-mediated inhibition of TLR7/MyD88 signaling (Yu et al., 2016). In our previous work, as well as that of other groups (Spaulding et al., 2016; Sharma et al., 2011), blocking type I IFN production and signaling, increased mouse resistance to severe Py YM and P. berghei and non-lethal P. chabaudi and Py 17XNL murine malaria, respectively. We neither noted increased type I IFN levels in Sting1Gt/Gt mice, nor increased IFNβ production by STING-deficient pDC, and Sting1Gt/Gt mice did not exhibit greater resistance to Py YM either. Also, while the STING pathway may be active in pDCs, current experimental evidence is scarce (Reizis, 2019; Bruni et al., 2015), consistent with our finding that STING sensing occurs in the CD169+ MP, extrinsically to pDCs. We favor the interpretation that the Py YM strain used in this previous work, is likely to exhibit genetic differences with the repository Py YM strain MRA-755 we have used, which may account for the different outcomes. However, taken together, our results that type I IFN signaling negatively impact malarial infection outcomes, appear to be more consistent with the general consensus across several murine models of infection and laboratories.

In conclusion, our analysis of the dynamic behavior of pDC in vivo by IVM directly in the BM of mice undergoing severe murine malaria, reinforce a model in which pDC sensing of their ligands in vivo is a highly coordinated and orchestrated process. We establish that CD169+ MPs, a subset of tissue-resident MP that act as essential sentinels of the immune system constantly sampling the lymph and blood (Sung et al., 2012; Kastenmüller et al., 2012; Perez et al., 2017; Miyake et al., 2007), play a key role in regulating pDC activation in vivo. While we establish the existence of a functional interaction between these MP and the pDC in the BM of malaria-infected mice, further investigations will be needed to determine (1) the nature of the STING-dependent signals derived from CD169+ MP that allow pDC to achieve type I IFN-producing capacity and egress the BM to the blood and (2) whether this interaction is direct or not, and may also take place in other tissues and infections or autoimmune diseases.

Materials and methods

Mice

C57BL/6J wild-type (WT B6), Tlr7−/y (Lund et al., 2004) (strain 8380, Jackson Labs), Ifnbmob/mob/Ifnb-yfp+/+ (denoted IfnbYFP/YFP) (Gaya et al., 2015) (strain 10818, Jackson Labs), Sting1Gt/Gt (Sauer et al., 2011) (strain 17537, Jackson Labs), Sting1F/F (strain 031670, Jackson Labs), Rosa26LoxP-STOP-LoxP(LSL)-tTomato (strain 007914, Jackson Labs), and Rosa26LSL-YFP (strain 006148, Jackson Labs) mice have been described and were purchased from the Jackson laboratory. PTCRA-EGFP mice (129 background) (Shigematsu et al., 2004) and were obtained from Dr. Boris Reizis (NYU, New York, NY) and backcrossed to WT B6 mice for 10 generations. Tlr7−/y and Sting1Gt/Gt PTCRA-GFP as well as Tlr7−/y and Sting1Gt/Gt, IfnbYFP/YFP reporter mice were obtained by intercross, respectively, for use in IVM and FACS assays. siglec1DTR/DTR and siglec1Cre/Cre mice, a kind gift from Dr. Masato Tanaka (Riken, Japan) (Miyake et al., 2007) expressed the human diphtheria toxin receptor (DTR) or the Cre recombinase knocked-in to the exon 1 of the CD169-encoding gene the mice. Mice were used as heterozygous for experiments (siglec1DTR/WT, siglec1Cre/WTSting1Gt/FIfnbYFP/YFP, Sting1Gt/FIfnbYFP/YFP, siglec1Cre/WTRosa26LSL-tTomato/WT, siglec1Cre/WTRosa26LSL-YFP/WT), obtained by intercrossing of the relevant strains above. All mice were housed and bred in specific pathogen-free conditions at the Albert Einstein College of Medicine. Eight- to twelve-week-old and sex-matched males and female mice were used for all experiments.

Partial BM chimeras siglec1DTR/WT or WT B6 recipient mice were exposed to 5 Gy total body irradiation (500 rads) and reconstituted with 2 × 106 BM cells isolated from either Tlr7−/y or WT PTCRA-EGFP reporter mice. Chimerism of reconstituted mice was checked ~6 weeks later in the blood, before infections were conducted. Blood chimerism of donor-derived HSC was ~30%.

Plasmodium infections

Request a detailed protocol

Plasmodium yoelii yoelii (17X), clone YM parasites (stock MRA-755) was obtained from BEI Resources. P. yoelii YM-infected red blood cells (iRBCs) from a frozen stock (stored in liquid nitrogen in Alsever’s solution and 10% glycerol) were injected intraperitoneally (i.p.) into a eight- to twelve-week-old WT B6 mouse, and grown for ~4 days. When parasitemia reached 2–10%, 5 × 105 . yoelii YM iRBCs were inoculated intravenously (i.v.) in a volume of 200 μl PBS into each experimental mouse.

Preparation of cell suspensions for flow cytometry (FACS) analysis

Request a detailed protocol

Blood was collected in K2 EDTA(Ethylenediaminetetraacetic) tubes and red blood cells lysed with NH4Cl buffer (0.83%, vol/vol). BM was harvested from the femur and tibia by removing epiphyses to expose marrow cavity and placing into a 0.5-ml Eppendorf with a punched hole in the bottom made with an 18.5 G needle, nested into a 1.5-ml Eppendorf tube. Tubes were next centrifuged at 10,000 × g, 4°C, and cells resuspended in media for further used for further analyses (see below).

Cell staining for FACS analysis

Request a detailed protocol

Cell suspensions were stained with LIVE/DEAD Fixable Aqua (Thermo Fisher Scientific) or Ghost Dye Red 780 (Tonbo) in PBS for 30 min at 4°C, then incubated with Fc-block CD16/32 (clone 2.4G2) for 20 min at 4°C. Cell suspensions were next incubated in Ab mixes in FACS buffer (2 mM EDTA, 2% fetal bovine serum [FBS], 0.02% sodium azide in PBS) for 20 min at 4°C. For cell counts, we utilized 50,000 counting beads (SpheroTech) added to cells isolated from one leg or one spleen. Stained cells were collected either on a FACS BD LSR-II or Aria III. A list of all mAbs and sera used in the study with full information is provided in Supplementary file 1. Data were analyzed using FlowJo version 9.6.2 (TriStar).

RBC labeling

Request a detailed protocol

Red blood cells were harvested from uninfected or Py YM-infected WT B6 mice with ~20% parasitemia by submandibular puncture into K2 EDTA tubes and washed twice with 10 ml 1× Mg2+Ca2+-free PBS. Peripheral blood was pelleted by centrifugation at 800 × g and 4°C for 5 min. RBCs were gently resuspended up to a volume of 5 ml in 1× Mg2+Ca2+-free PBS, and overlaid on top of 5 ml room temperature Ficoll-Plus in a 15 ml Falcon tube which was centrifuged at 1000 × g for 15 min at 20°C without breaks. RBC pellet was washed in 5 ml Mg2+Ca2+-free PBS, before staining with 5 μM 5-carboxytetramethylrhodamine (TAMRA-Red, ThermoFisher Scientific, Catalog#C6121) in PBS 2% FBS for 20 min at 37°C, 5% CO2 and in the dark. Labeled RBCs were then washed in 10 ml PBS, and 50 μl of packed RBCs recovered in 200 μl PBS were injected i.v. by retro-orbital route into recipient mice.

PT and in vivo mAb treatments

Request a detailed protocol

PT (Sigma Millipore #516560) was reconstituted in 100 mM sodium phosphate, 500 mM sodium chloride, pH 7.0 and stored at −80°C. Mice were treated with 2.5 µg/µl PT in 100 µl of 1× PBS, i.v. at the time of infection. For mAb injections, mice received 250 μg of each mAb as specified in the figure panel, i.v. 30 min prior infection (Supplementary file 1).

Two-photon preparation, imaging, and analysis

Request a detailed protocol

In vivo imaging of BM was conducted as described previously (Pitt et al., 2015). PTCRA-EGFP reporter mice were anesthetized with isoflurane (2.5–3% for induction, 1–2% for surgery/imaging) admixed in 1:1 O2: air mixture at a flow rate of 1 l/min. Once induction was achieved, a gas mask was secured over the nose and stabilized using Velcro on a stage maintained at 37°C. The mouse anesthetic plane was ensured through the lack of response to toe and/or tail pinches. Level of isoflurane was adjusted accordantly to ensure the mouse maintained appropriate sedation with a stead, non-labored respiratory rate between 60 and 80 breath/min. Leg hair was then shaved using a surgical blade and a small amount of soap and water to facilitate hair removal. Loose hair and residual soap/water were removed with a kimwipe and the leg dried. Subsequently, the mouse torso was secured on the stage with Velcro under the diaphragm so as to not interfere with breathing, and the leg stabilized with a piece of scotch tape with the mouse is a supine position and the foot everted to reveal the medial surface of the leg. Then, the medial aspect of the tibia was surgically exposed by making a small incision at the level of the medial malleolus, avoiding the saphenous artery, to expose the tibial bone. A window was secured to the leg using vet bond, sliding a small tooth of the window under the tibial bone and then fixed to the stage using screws. Then, the exposed bone was shaved using a microdrill to a 100-μm thickness. Once optimal thinness was achieved, a ring of vacuum grease was drawn around the imaging area, then filled with a drop of 37°C lactated ringers (Thermo Scientific Cat#R08432), and the mouse transferred to the microscope chamber maintained at 37°C. Images were collected on an Olympus FV-1000MPE upright laser scanning microscope using a MaiTai DeepSee Laser (Spectraphysics) set to a wavelength of 910–920 nm. BM was visualized using a ×25 NA 1.05 objective and magnified 1.5–2×. Time lapse stacks with 5 μm slices spanning a total depth of 50 μm were captured at an acquisition of 4.0 μs/pixel using. To visualize CD169+ MP, mice were intravenously injected with 5 μg of anti-CD169 (SER4) antibody (BD Pharmingen) conjugated to PE (or FITC for Py-iRBC uptake experiments), 16 hr prior to imaging (Gaya et al., 2015).

Movie analysis: 4D image sets were analyzed on Volocity 6.3 (Improvision). Tracking was conducted using semi-automated scripts using spot function. For pDC identification, Channel 3 subtracted from Channel 2 using imaging processing -> channel arithmatics function. XY diameter set to 10 μm, threshold 7 for brightness, max distance 8 μm, max gap 0. Filtered to include only pDCs tracked >300 s. For cell-contact duration between pDC and CD169+ MP, analysis was conducted using Kiss and Run Analysis built-in plugin extension between pDC spots and MP surfaces (Imaris, Matthew Gastinger). Distance transformation function used with threshold set to 5 μm for distance. Data obtained from contact durations were tallied up for all pDC in the field in contact with CD169+ MP. Data were pooled from three or more mice from independent experiments. Supplementary movies were assembled using After Effects (Adobe). Contrast and saturation were enhanced to allow for greater ease of pDC visualization above MP for inclusion in publication (post-analysis).

Two-dimensional plots were created at a higher resolution to our specification using Matlab. Instructions and script hyperlinked: ./XYZ plots/Read Me.docx; ./XYZ plots/XY_Plots_ZB_10082019.m.

Statistics

All statistical tests were run with GraphPad Prism 8. Significance is depicted as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. All graphs show mean ± standard error of the mean. The unpaired t-test was used for comparing two groups related to FACS data. Welch’s t-test was used for comparing two groups related to IVM analyses. Multiple linear regression analyses were applied to test if the factor conditions (time and group) predicted displacement for statistical analysis of the MSD plots, assuming a gaussian distribution of residual and least squares.

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting file; Source Data files have been provided for all figures. There is no restriction of access.

References

Decision letter

  1. Florent Ginhoux
    Reviewing Editor; Agency for Science Technology and Research, Singapore
  2. Tadatsugu Taniguchi
    Senior Editor; University of Tokyo, Japan

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Spatio-temporal organization of plasmacytoid dendritic cell priming by CD169+ macrophages in malaria-infected mice" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Michael L Dustin (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The reviewers commend you on your efforts to extend your previous work and design a robust methodology to assess selective interactions of parasitized RBCs with macrophages. However the reviewers also noted major concerns with conflicting data and overstated results that significantly detracted from their enthusiasm. Significant concerns were expressed regarding the demonstration of interactions between CD169+ macrophages and pdcs, which is a major conclusion of the paper. Furthermore, two reviewers were concerned that there was not a clearly defined delineation between this work and your previous work. Specific comments are listed below.

Reviewer #1:

In this study, the authors extend previous work published from their lab showing that CD169+ macrophages license plasmacytoid dendritic cells (pDCs) to make type I interferon in the bone marrow during P. yoelii infection. Though there are some interesting points made in the manuscript, overall the manuscript reads more as a compilation of findings following up on their previous PLOS Pathogens paper, rather than a coherent story. Furthermore, some of the more novel conclusions in the manuscript are not well supported by the data.

1. In several parts of the manuscript, it is difficult to distinguish by the writing what was previously published by this group and what is new to this manuscript.

2. From Figure 1, the authors conclude that CD169 is involved in the binding and uptake of infected RBC. However, the authors mis-state what the data show. They state "we found that CD169 het mice, BM MPs had significantly higher frequencies of both extra and intracellular TAMRA-Red+…" However, Figure 1D shows a lower frequency of extracellular iRBCs, but no difference in intracellular RBCs. Though the "representative" flow is picked to show a >5x difference in extracellular RBCs, the number of both intra and extracellular RBCs bound/internalized by macrophages is significantly different, showing that the number of BM MPs is what is changing. In fact, the authors show the F4/80hi cells go down in number after infection in Figure 2, consistent with this interpretation.

3. Much of the manuscript uses CD169-/- mice and the authors state that the CD169hi macrophages still exist in these mice, however it is critical to show this to be able to interpret the data using these mice as the authors have already shown that these cells are critical for pDC clustering and IFNa production.

4. In Figure 2, there are two contradictory pieces of data. In Figure 2A, CD11a is show to be downregulated in infected mice, while in Figure 2D, CD11a is unchanged compared to uninfected mice.

5. Although the authors acknowledge that PT treatment has indirect effects, they still conclude that pDC motility is required for pDC IFNa production and fail to address the possibility that motility of other cell types, including CD169+ macrophages, being involved.

6. In Figure 7, the authors demonstrate that STING is required independently of cGAS for efficient type I interferon production by pDC. However, the authors had already demonstrated a role for STING extrinsic to pDC in their previous publication. Therefore, the only new piece of data here is that it is independent of cGAS, which is interesting, but needs to be followed up on in its own right.

7. This manuscript would be greatly enhanced by a model of what had been shown previously and what is shown in this manuscript.

8. Use of statistics throughout the manuscript is not consistent and tests used are not noted in figure legends. Sometimes only 2 groups in a figure are compared and sometimes all groups, even within one figure panel, with no justification.

Reviewer #2:

Moore et al. characterize the phenotype and movement of pDCs in the bone marrow of infected mice, and test several pathways (TLR7, STING, cGAS, GPCRs) for their role in regulating pDC motility, phenotype, and interferon production. There are several nice experiments here. Unfortunately, the authors grossly misrepresent or overinterpret their findings and extrapolate to a model of cell-cell interaction for which they provide little evidence. This is clearly a follow-up study to their 2016 PLoS Pathogens paper, but I'm not convinced it stands on its own; at a minimum it needs extensive textual revision to accurately represent the data, and once it has done that, I'm not sure it meets broad enough interest for an eLife paper.

My main concern is that the text misrepresents what the data actually show. E.g., the title and abstract claim that CD169+ macrophages coordinate the localization and function (IFN secretion) of pDCs in a STING- and TLR7-dependent manner, but there is not a single figure in the paper that quantifies interactions between macs and pDCs. (The handful of microscopy images are not sufficient to make conclusions about these interactions-and the authors do not even attempt to do this, e.g. by using arrowheads to mark interacting cell clusters). The authors do state (p. 9) that pDC did not cluster around CD169+ macs in infected TLR7 KO mice compared to WT mice, but they show no data, and cite their previous paper for the WT observation. Further, there are no data to show that STING and TLR7 are acting within macs, rather than pDC (or another cell type), to affect arrest or clustering of pDC. I gather that the authors may have shown in their 2016 paper a requirement for mac-expressed STING in pDC production of interferon, but that doesn't mean the same is true for pDC motility/ clustering. (And it doesn't tell us anything about the TLR7 requirement.) The authors establish in Figure 1 that macs take up infected RBCs, but they do not show that pDCs do NOT take up iRBCs (which could perhaps demonstrate indirectly that pDCs must get their activating signals from macs). There is precedent for direct pDC uptake of malaria parasites (Wykes et al. PNAS 2011), though controversial… still, in this paper at least, the authors have not demonstrated that any interactions with CD169+ macs are controlled by STING/ TLR7/GPCR, or required for IFN production by pDCs.

I also think the involvement of TLR7 in pDC arrest-a key point of the paper-is overstated. The section heading (p.8) states that "pDC arrest in Py-infected bone marrow is TLR7-dependent." But as the authors note, TLR7 KO pDCs have much higher baseline activity. They measure motility in four ways: track velocity, displacement rate, MSD, and slope of MSD. The latter two clearly show that TLR7 KO pDC still arrest after infection; the stats are not there to show whether the magnitude of their arrest is significantly different from WT. One measurement of motility (velocity) shows that TLR7 KO cells do differ from WT in that they don't slow down after infection. The final measure, displacement rate (3C- bottom) is not convincing: WT mice decrease displacement rate significantly after infection and TLR7 KO mice don't, but the WT infected group has an n of 7, and the TLR7 group only has 3 mice and is trending downward. It seems very possible that with more mice, this would become significant. So. at least two, and possibly 3 of the 4 measures of motility show NO difference between WT and TLR7-/- pDC. Given this, I think the finding is way overstated. The authors can still say that TLR7 plays some role in pDC motility, but they need to qualify this quite a bit (which they do on p. 9, but not throughout much of the paper, including the abstract).

A third unsupported conclusion is that "PT treatment led to immobilization of pDC, and their lack of visible clustering with MP" (p. 10, Figure 5A, B). The authors show that pDCs are largely nonmotile in infected, PT-treated mice, but they do NOT compare to untreated controls, in which we know that pDCs are arresting and largely nonmotile (Figure 3A second panel). Is there a difference between motility in infected, PT-treated versus untreated mice? This is critical to show. Further, as mentioned before, there is no measurement of clustering with macs, so the authors should not assert that PT treatment alters this. Even the assertion that pDC are "morphologically distinct" after PT treatment seems overstated (p. 10); Figure 5C shows no difference in shape factor and a trending but non-significant difference in longest axis.

Reviewer #3:

pDC are important cells in the innate immune system that produce type I IFN (IFNab) in response to infection or injury. This group published in 2016 that pDC are highly motile in the bone marrow, but arrest in proximity to macrophages in the context of experimental malaria. Here, they further investigate the role of the CD169 receptor, TLR7, STING and cGAS innate sensors and GPCR motility receptors on this response. The work more clearly defines the role of the innate sensing pathways and chemokine receptors in the motility phenotype and the functional endpoint of IFNab production. There are a number of very nice aspects to this study. They have developed a robust method to detect the selective interaction of parasitized RBC with macrophages and to assess whether or not they are internalized on captures on the surface of macrophages. The baseline motility is completely pertussis toxin sensitive, although it was not possible to link the deceleration to an IFNab responsive chemokine receptor/chemokine pair. The interesting result is that the control of motility is necessary for the response, but not sufficient for IFNab production. A weakness of the paper is that the genetic manipulations effect both the macrophages and the pDC, so it's not possible to determine, for example, if TLR7 or STING are mainly operating in the pDC or the macrophage, or both.

The number of experiments is not quite enough to nail down the pDC migration phenotype of the TLR7 deficient mice. Three of the 4 mice analysed have more persistent migration such that the MSD is clearly greater. While the trend is clear the statistical significance is marginal, but this seems to be due to a limited samples size. So a couple of more experiments would be helpful. The data on the STING KO confirm that they can nail this when the migration phenotype is similar between two mouse strains.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "CD169+ macrophages orchestrate plasmacytoid dendritic cell arrest and retention for optimal priming in the bone marrow of malaria-infected mice" for further consideration by eLife. Your revised article has been evaluated by Tadatsugu Taniguchi (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Essential revisions:

1) CD169+ Macs a) capture infected RBC and b) interact with PDC.

2) Show FN Type 1 secretion.

The strategies proposed by reviewers could include:

Using breeding of PTCRA-eGFP mice and CD169Cre x Rosa fluorescent reporters.

Use of CD169-Knockin reporter.

Use of CD169 Cre x Rosa stop flox-reporter to image CD169+ Macs and perform shield chimera and graft the BM of PTCRA-eGFP mice into CD169 Cre x Ai14 mice for example.

Reviewer #1:

This is a revised manuscript by a group that has published on this topic before in 2016 (Plos Pathogen paper). They largely extend their previous findings and add some mechanistic studies demonstrating the requirement of CD169+ macrophage intrinsic STING signaling in regulating pDC motility/arrest. Furthermore their data suggest that adequate type I IFN production by pDC may also require macrophage intrinsic STING signaling.

Since I was not part of the initial review I evaluated the previous reviewer comments and assessed how well the authors responded to the criticisms. I agreed with the majority of the concerns raised by the previous reviewers. Furthermore, the authors have attempted to strengthen the study in good faith and have included new studies. In particular, in inclusion of experiments where CD169 conditional STING-KO mice are used adds to the novelty of this study.

The study is largely an extension of their previous work and in the revised manuscript the results for the most part support their conclusions. The use of dynamic imaging to assess pDC dynamics in the bone marrow is interesting and will add new avenues for further studies.

This study also adds to our knowledge of how pDCs get activated and produce type I IFNs in vivo. The mechanisms of how pDCs get activated in vivo still remain poorly understood, however, this report begins to shed some light.

Some deficiencies still remain, which will be discussed below:

Concerns that still remain:

1. Can the authors comment on the concern that use of anti-CD169 antibody for imaging the macrophages in vivo may prevent adequate uptake of infected RBCs or parasites in vivo?

2. Could the authors comment more on the physiological significance of type I IFN during malaria infection. Are pDCs needed for clearance of the infection and if type I IFN is required at all for the benefit of the host. This is a rather controversial topic and a clearer discussion taking their own studies in to account is warranted.

3. The fact that the motility of pDCs in TLR7 KO mice even under a steady state is so much higher still makes the interpretation of the data in Figure 3 difficult.

4. The authors make the claim that CD169+ macrophage intrinsic STING signaling is required for the production of Type I IFN by pDCs. However, they only use the reporter mouse for these studies, and the imaging is not very clear. Thus, the authors should show ELISA data in the serum and bone marrow to complement their imaging studies.

5. The CD169+ macrophages themselves can produce copious amounts of Type I IFNs. Can the authors eliminate the possibility of paracrine IFN secreted by the macrophages as another mechanism by which pDCs respond to and produce more IFN – especially in studies where CD169-DTR mice are used or in STING conditional KO studies? Here too, assessing IFN production will be valuable.

6. Finally, the authors should be careful about making any substantive conclusions with respect to pDC – CD169+ macrophage interactions. In this respect the dynamic imaging is not as convincing. The authors themselves admit that due to the high density of macrophages in the BM "it makes it difficult to establish a functional interaction"

Reviewer #2:

This revised version investigates the role of CD169 BM macrophages in the activation of pDC following the injection of P.yoelii infected RBCs. It aims at demonstrating that BM CD169+ macrophages are key regulators of pDC retention and activation, in a TLR7 and STING dependant manner.

Overall, the take home message appears valid but some experiments need to be performed in order to exclude technical artifacts:

1. TAMRA and PE appear to have very similar excitation and emission spectra. In addition, they both emit in the "Red" channel, which is typically highly autofluorescent. Why did the authors chose these two dyes to specifically label RBC and CD169+ Macs? The use of green/red (or blue/red or blue/green) fluorochromes might have improved the quality of the movies.

2. Reviewer 1 is correct when he/she states that PT treatment has indirect effects than can alter pDC motility and IFNa production. As the authors fail to identify the chemokines (or PT sensitive signalling pathways) that are at play, it is incorrect to state that pDC motility is required for pDC IFNa production. The authors should make this very clear in the manuscript.

3. Figure 1 does not convince me that CD169 Macs pick up infected RBC. It is well documented that CD169+ macs are fragile (at least in lymphoid organs) and that most CD169+ cells analyzed by flow are not even macrophages (PMID: 22675532 and more recently PMID: 34818538). As IVM lacks resolution, the authors should perform the same experiment in a PTCRA-eGFP mouse and stain fixed bone sections for CD169/GFP expression and then quantify the amount of fluorescent RBC captured by pDC and CD169+ macs. In addition, as the authors possess CD169Cre mice, I am wondering why they did not cross this strain to a fluorescent reporter such as the Rosa-RFP line to (a) visualize CD169+ macs by IVM and (b) prove that pDC are not fate-mapped in this model, and thus not targeted in CD169DTR mice.

4. Related to the previous points, I am concerned by FigS2C and sup movie 3. Both are showing data obtained in CD169 DTR mice treated with DT. While flow cytometry showed a 6 fold reduction of CD169+ macs upon DT injection, IVM does not show a similar decrease. If true, it suggests that one technique is not reliable to observe CD169+ macs. Can the authors comment on this?

5. In their reply to the reviewer's comments, the authors seem to have addressed the other concerns of the referees, including the use of a CD169Cre sting FF model to ablate STING in CD169+ Macs.

https://doi.org/10.7554/eLife.78873.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

In this study, the authors extend previous work published from their lab showing that CD169+ macrophages license plasmacytoid dendritic cells (pDCs) to make type I interferon in the bone marrow during P. yoelii infection. Though there are some interesting points made in the manuscript, overall the manuscript reads more as a compilation of findings following up on their previous PLOS Pathogens paper, rather than a coherent story. Furthermore, some of the more novel conclusions in the manuscript are not well supported by the data.

1. In several parts of the manuscript, it is difficult to distinguish by the writing what was previously published by this group and what is new to this manuscript.

The revised version now clarifies what was published before and what is new. Briefly, our prior study published in PLoS Pathogens in 2016, established that type I IFN receptor deficient mice resist lethal P. yoelii (Py) infection and that pDC are the major early producers of type I IFN in response to this infection. This work also showed that type I IFN production is initiated in the BM of infected mice via pDC-intrinsic MyD88/TLR7 signaling and upon priming by CD169+ BM MP. It further established the requirement for STING sensing in radioresistant cells, presumably the CD169+ MP. Lastly, we also provided in this work the first evidence, using intravital mutiphoton imaging (IVM) of the BM of living mice, that pDC arrested next to CD169+ MP at their peak production of type I IFN (36 hrs post Py infection).

In the current work, we investigated if CD169+ MP and parasite sensing via TLR7 and STING regulates pDC dynamic behavior and their interactions with CD169+ BM MP in vivo. Since CD169+ MP are required to prime pDC production of type I IFN, we also explored if these MP take up P. yoelii (Py)-infected RBCs, and provide STING-dependent signals to the pDC for full activation. Conclusions of this work are that:

i) CD169+ MP uptake Py-infected red blood cells (iRBCs) by FACS and IVM.

ii) A majority of CD169+ MP in the BM undergo activation and loss (>70%) between 30 and 36 hrs post infection, consistent with their implication in iRBC sensing and uptake (Revised Figure 1)

iii) CD169+ MP control pDC arrest. In Py-infected mice in which CD169+ MP are depleted, pDC fail to arrest (new Figure 2).

iv) TLR7 signaling in pDC is also required for their arrest and augmented interactions with the MP (revised Figure 3). v)Homing of pDC to CD169+ MP and acquisition of type I IFN producing capacity requires GI-mediated chemotaxis (revised Figure 4).

v) STING signaling in CD169+ MPs is required to enable pDC production of type I IFN and for their egress from the BM (revised Figure 5 and new Figure 6).

The revised version streamlines the original message and includes many new experiments and data that directly support the conclusions outlined above. We have added a new set of experiments that directly link CD169+ MPs to pDC arrest during Py infection (new Figure 2). We have generated novel partial bone marrow (BM) chimera mice to demonstrate that (i) TLR7intrinsic signaling in pDC is required for pDC arrest (revised Figure 3) and (ii) pDC-extrinsic STING-signaling regulates their retention in the BM (revised Figure 5). We also produced novel conditional knockout mice in which CD169+ MP lack STING. With these mice we established that STING signaling intrinsic to CD169+ MP is required to prime pDC production of type I IFN and to allow pDC egress from the BM (new Figure 6).

2. From Figure 1, the authors conclude that CD169 is involved in the binding and uptake of infected RBC. However, the authors mis-state what the data show. They state "we found that CD169 het mice, BM MPs had significantly higher frequencies of both extra and intracellular TAMRA-Red+…" However, Figure 1D shows a lower frequency of extracellular iRBCs, but no difference in intracellular RBCs. Though the "representative" flow is picked to show a >5x difference in extracellular RBCs, the number of both intra and extracellular RBCs bound/internalized by macrophages is significantly different, showing that the number of BM MPs is what is changing. In fact, the authors show the F4/80hi cells go down in number after infection in Figure 2, consistent with this interpretation.

We agree that the data presented in Figure 1D was confusing. The labelled uninfected and infected RBC uptake experiments followed a different set up than that of the natural infection which was depicted in Figure 1A (now in revised Figure S1B). Briefly, RBCs (~5x107) were collected either from uninfected mice or from Py-infected mice when they reached ~20% blood parasitemia (~3-4 days post infection). RBCs were next labelled with the TAMRA-red and transferred to naïve mice. Six hrs later, we (i) harvested the BM cells and quantified TAMRAred+ RBCs associated with the BM MP populations by FACS and (ii) conducted IVM on the BM of recipient living mice. With such experimental setting we focused on the RBC uptake phase before MP loss occurs (revised Figure 1C). The data show (by FACS and IVM) that CD169+ MP (both CD169hi and F4/80hi) are involved in the uptake of Py-infected but not uninfected RBC (revised Figure 1A, B and Movie S1). This has also been clarified in the revised manuscript.

CD169 is a sialoadhesin known to bind RBCs. While our original data supported the hypothesis that CD169 is involved in Py-iRBC uptake, we decided to remove these data from the revised version in an effort to streamline the manuscript and focus on the most important take home message.

3. Much of the manuscript uses CD169-/- mice and the authors state that the CD169hi macrophages still exist in these mice, however it is critical to show this to be able to interpret the data using these mice as the authors have already shown that these cells are critical for pDC clustering and IFNa production.

The experiments using the CD169-/- mice (prior Figures 1D and 2D) have been removed from the revised paper, following the rationale outlined above. The gating strategy we used throughout the manuscript to identify BM CD169+ MPs is depicted in revised Figure S1A.

4. In Figure 2, there are two contradictory pieces of data. In Figure 2A, CD11a is show to be downregulated in infected mice, while in Figure 2D, CD11a is unchanged compared to uninfected mice.

We have pooled 2 independent replicate experiments now presented in the revised Figure S1D.

Overall, we found a consistent downregulation of CD11a expression on CD169+ MP (CD169hi and F4/80hi MP subsets). Additional replicate experiments showed the same result, but could not be pooled as a result of staining (MFI) differences, so we are confident these data are representative.

5. Although the authors acknowledge that PT treatment has indirect effects, they still conclude that pDC motility is required for pDC IFNa production and fail to address the possibility that motility of other cell types, including CD169+ macrophages, being involved.

We certainly agree with this reviewer that whole body PT treatment is not ideal and may affect other cells’ motility. One possibility is to treat purified pDC with PT in vitro and transfer them to naïve or infected hosts. However, adoptive transfers of pDC is very ineffective (they only represent less than 0.5% of DC) and we could not find transferred pDC in the BM (a point also noted by Reviewer #3). Another approach would involve expression of PT in pDC, but we are not aware of a pDC-specific inducible promoter we could use for this purpose, therefore this seemed beyond the scope of the current work. We also note that CD169+ MP were not motile in our IVM experiments. This observation has been quantified and is now shown in revised Figure S2B. In light of these comments, we have nevertheless carefully re-worded this section to reflect such limitations.

6. In Figure 7, the authors demonstrate that STING is required independently of cGAS for efficient type I interferon production by pDC. However, the authors had already demonstrated a role for STING extrinsic to pDC in their previous publication. Therefore, the only new piece of data here is that it is independent of cGAS, which is interesting, but needs to be followed up on in its own right.

In our previous work, we had established that STING signaling extrinsic to pDC and in the radiosensitive compartment, was important for pDC to secrete type I IFN. In the revised paper, we generated novel conditional knockout mice in which CD169+ MP lacked STING (new Figure 6). The data show that STING signaling in CD169+ MP is required for pDC production of type I IFN and egress from the BM. We further confirmed that when STING lacked only in the radioresistant compartment, pDC formed large clusters and accumulated in the BM (revised Figure 5H).

We have removed the cGAS data from the revised paper as we agree that this would need further investigations, and currently distracts from the main message of this work.

7. This manuscript would be greatly enhanced by a model of what had been shown previously and what is shown in this manuscript.

This is a great suggestion, and we provide a schematic to further clarify this aspect below. We have incorporated the final model in Figure 6D of the revised paper.

8. Use of statistics throughout the manuscript is not consistent and tests used are not noted in figure legends. Sometimes only 2 groups in a figure are compared and sometimes all groups, even within one figure panel, with no justification.

We apologize for this, and have now carefully checked that all statistical analyses were done and that the tests used were specified.

Reviewer #2:

Moore et al. characterize the phenotype and movement of pDCs in the bone marrow of infected mice, and test several pathways (TLR7, STING, cGAS, GPCRs) for their role in regulating pDC motility, phenotype, and interferon production. There are several nice experiments here. Unfortunately, the authors grossly misrepresent or overinterpret their findings and extrapolate to a model of cell-cell interaction for which they provide little evidence. This is clearly a follow-up study to their 2016 PLoS Pathogens paper, but I'm not convinced it stands on its own; at a minimum it needs extensive textual revision to accurately represent the data, and once it has done that, I'm not sure it meets broad enough interest for an eLife paper.

We thank this reviewer for her/his time and critiques. Following up on these comments and those from the other reviewers, we have conducted many additional experiments to address key weaknesses. We have also significantly revised the manuscript to incorporate these new experiments (two new figures and four substantially revised figures), and clarify what is new in this work compared to our previous report (please also check our response to Reviewer#1 comment #1 on this major point, and the new schematic provided above and in revised Figure 6D). Detailed answers to the specific critiques of this reviewer are further provided below.

My main concern is that the text misrepresents what the data actually show. E.g., the title and abstract claim that CD169+ macrophages coordinate the localization and function (IFN secretion) of pDCs in a STING- and TLR7-dependent manner, but there is not a single figure in the paper that quantifies interactions between macs and pDCs. (The handful of microscopy images are not sufficient to make conclusions about these interactions-and the authors do not even attempt to do this, e.g. by using arrowheads to mark interacting cell clusters). The authors do state (p. 9) that pDC did not cluster around CD169+ macs in infected TLR7 KO mice compared to WT mice, but they show no data, and cite their previous paper for the WT observation.

We have added new/revised figures in which:

i) pDC dynamic behavior in the presence or absence of CD169+ MPs has been quantified in naïve versus Py-infected mice (new Figure 2A-D). We now also link the presence of CD169+ MP to pDC arrest during Py infection. When CD169+ MP are depleted, pDC fail to arrest and cluster.

ii) In Py-infected TLR7-/- mice, pDC slow down but never arrest like in WT mice, therefore they do not form clusters. The new sets of experiments assess how STING and TLR7 signaling in pDC affect their motility (revised Figures 3, 5 and our answer below). We have also developed new scripts to quantify pDC and MP interaction by measuring their contact times (revised Figures 3G, 5F).

Further, there are no data to show that STING and TLR7 are acting within macs, rather than pDC (or another cell type), to affect arrest or clustering of pDC. I gather that the authors may have shown in their 2016 paper a requirement for mac-expressed STING in pDC production of interferon, but that doesn't mean the same is true for pDC motility/ clustering. (And it doesn't tell us anything about the TLR7 requirement.)

The revised Figures 3 and 5 incorporate new IVM experiments using partial BM chimera models in which we could visualize pDC lacking TLR7 or STING in WT host mice. When pDC lacked TLR7, they failed to arrest and interact with CD169+ MP (Figure 3E-G). In contrast, when pDC lacked STING, they arrested and interacted with the MP like WT pDC (Figure 5E, F). These results therefore (i) confirm our original observations in TLR7-/- and STING-/- mice, and (ii) directly link pDC motility (arrest, clustering) and interaction with CD169+ MP to their ability to secrete type I IFN (shown in our prior study).

The authors establish in Figure 1 that macs take up infected RBCs, but they do not show that pDCs do NOT take up iRBCs (which could perhaps demonstrate indirectly that pDCs must get their activating signals from macs). There is precedent for direct pDC uptake of malaria parasites (Wykes et al. PNAS 2011), though controversial… still, in this paper at least, the authors have not demonstrated that any interactions with CD169+ macs are controlled by STING/ TLR7/GPCR, or required for IFN production by pDCs.

We found no evidence for pDC uptaking Py-iRBCs neither in our FACS uptake experiments (See revised Figure S1C), nor in IVM. The new experiments shown in the revised manuscript establish that CD169+ MP and TLR7 sensing in pDC are both required for pDC arrest and clustering. We also show that STING sensing in pDC do not play a role in their arrest, clustering or production of type I IFN. Furthermore, using a new set of conditional knockout mice we generated, in which CD169+ MPs lack STING, we report that pDC fail to secrete type I IFN and egress from the BM during the infection (new Figure 7), as observed in whole body STINGdeficient mice (revised Figure 5I). Together with the new data presented in the revised Figure 2E-G, where we link the presence of CD169+ MP to pDC arrest, the results support the original model that CD169+ MP provide STING-dependent signal(s) to prime pDC production of type I IFN and enable their egress from the BM.

I also think the involvement of TLR7 in pDC arrest-a key point of the paper-is overstated. The section heading (p.8) states that "pDC arrest in Py-infected bone marrow is TLR7-dependent." But as the authors note, TLR7 KO pDCs have much higher baseline activity. They measure motility in four ways: track velocity, displacement rate, MSD, and slope of MSD. The latter two clearly show that TLR7 KO pDC still arrest after infection; the stats are not there to show whether the magnitude of their arrest is significantly different from WT. One measurement of motility (velocity) shows that TLR7 KO cells do differ from WT in that they don't slow down after infection. The final measure, displacement rate (3C- bottom) is not convincing: WT mice decrease displacement rate significantly after infection and TLR7 KO mice don't, but the WT infected group has an n of 7, and the TLR7 group only has 3 mice and is trending downward. It seems very possible that with more mice, this would become significant. So. at least two, and possibly 3 of the 4 measures of motility show NO difference between WT and TLR7-/- pDC. Given this, I think the finding is way overstated. The authors can still say that TLR7 plays some role in pDC motility, but they need to qualify this quite a bit (which they do on p. 9, but not throughout much of the paper, including the abstract).

This point is well taken. We have conducted several new experiments to address these weaknesses (revised Figure 3). In addition to analyzing pDC motility in additional TLR7-/- mice, we have also generated partial BM chimeras in which only TLR7-/- pDC can be visualized in a WT host environment, allowing to assess if TLR7 signaling intrinsic to pDC alters their motility. While in uninfected mice, TLR7-/- pDC exhibited higher motility than in WT counterparts (as measured by TV, DR, MSD, revised Figure 3), the most important finding was in Py-infected mice where TLR7-/- pDC, while slowing down, remained as motile as in uninfected WT mice. This was in contrast to WT pDC that arrested compared to that of uninfected mice. We have added statistical analyses comparing all groups for each parameter measured. Moreover, by quantifying the contact time between pDC and CD169+ MP, we show an increase only for WT but not TLR7-/- pDC, consistent with this finding. This new set of experiments therefore strongly support the idea that TLR7 signaling in pDC control their arrest and activation.

A third unsupported conclusion is that "PT treatment led to immobilization of pDC, and their lack of visible clustering with MP" (p. 10, Figure 5A, B). The authors show that pDCs are largely nonmotile in infected, PT-treated mice, but they do NOT compare to untreated controls, in which we know that pDCs are arresting and largely nonmotile (Figure 3A second panel). Is there a difference between motility in infected, PT-treated versus untreated mice? This is critical to show.

We thank this reviewer for this suggestion. We have quantified pDC motility in PT-treated or untreated Py-infected mice (revised Figure 4A, B and Supplementary movie 7). The data show that DR of pDC in PT-treated mice was significantly reduced compared to that of untreated, showing that blocking GI-protein signaling and chemotaxis, induced stronger immobilization of pDC than Py infection only.

Further, as mentioned before, there is no measurement of clustering with macs, so the authors should not assert that PT treatment alters this.

The revised Figures 3G and 5F provide a quantification of pDC and CD169+ MP contact times.

However, in PT-treated mice, pDC arrest was drastic compared to untreated mice (revised Figure 4A, B and new Supplementary movie 7), making it unlikely that pDC and MP can establish functional interactions. We have carefully re-written this section of the results to highlight these findings.

Even the assertion that pDC are "morphologically distinct" after PT treatment seems overstated (p. 10); Figure 5C shows no difference in shape factor and a trending but non-significant difference in longest axis.

We have conducted further analyses of these data using software package Volocity 6.3 to include the monitoring of more pDC (revised Figure 4D). The data reveal statistically significant differences in pDC from PT-treated and untreated, Py-infected mice, using multiple metrics related to pDC shape including Longest Axis, Surface Area and Shape Factor. The Shape factor is a numerical indication of how similar 3D shape is to a perfect sphere with shape factor 1 indicating a perfect sphere.

Reviewer #3:

pDC are important cells in the innate immune system that produce type I IFN (IFNab) in response to infection or injury. This group published in 2016 that pDC are highly motile in the bone marrow, but arrest in proximity to macrophages in the context of experimental malaria. Here, they further investigate the role of the CD169 receptor, TLR7, STING and cGAS innate sensors and GPCR motility receptors on this response. The work more clearly defines the role of the innate sensing pathways and chemokine receptors in the motility phenotype and the functional endpoint of IFNab production. There are a number of very nice aspects to this study. They have developed a robust method to detect the selective interaction of parasitized RBC with macrophages and to assess whether or not they are internalized on captures on the surface of macrophages. The baseline motility is completely pertussis toxin sensitive, although it was not possible to link the deceleration to an IFNab responsive chemokine receptor/chemokine pair. The interesting result is that the control of motility is necessary for the response, but not sufficient for IFNab production. A weakness of the paper is that the genetic manipulations effect both the macrophages and the pDC, so it's not possible to determine, for example, if TLR7 or STING are mainly operating in the pDC or the macrophage, or both.

We thank this reviewer for his positive comments and fair assessment of our work. The revised manuscript now includes new experiments in which we show that CD169+ MP control pDC arrest (new Figure 2). We also generated partial BM chimeras to visualize TLR7-/- or STING-/- pDC in a WT environment (revised Figures 3, 5). Lastly, we have produced new conditional knockout mice in which STING is lacking in CD169+ MPs (new Figure 6). These new data confirm the original phenotypes and establish that:

i) CD169+ MP are needed for pDC to arrest during Py infection (new Figure 2).

ii) TLR7 sensing and signaling in pDC is also required for pDC to arrest, cluster and interact with CD169+ MP (revised Figure 3).

iii) STING signaling in pDC is not important for them to arrest (revised Figure 5).

iv) STING signaling is important in CD169+ MPs to provide STING-dependent signal(s) to pDCs so that they can initiate type I IFN secretion. The data also show that if such signal(s) is(are) missing, pDC accumulate in the BM, form larger clusters and fail to egress from the BM (revised Figures 5 and 6).

The number of experiments is not quite enough to nail down the pDC migration phenotype of the TLR7 deficient mice. Three of the 4 mice analysed have more persistent migration such that the MSD is clearly greater. While the trend is clear the statistical significance is marginal, but this seems to be due to a limited samples size. So a couple of more experiments would be helpful. The data on the STING KO confirm that they can nail this when the migration phenotype is similar between two mouse strains.

The revised paper includes additional IVM experiments to strengthen the original phenotype, as well as all the new partial BM chimera experiments that confirm and extend our original findings and cell-intrinsic phenotypes.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Reviewer #1:

This is a revised manuscript by a group that has published on this topic before in 2016 (Plos Pathogen paper). They largely extend their previous findings and add some mechanistic studies demonstrating the requirement of CD169+ macrophage intrinsic STING signaling in regulating pDC motility/arrest. Furthermore their data suggest that adequate type I IFN production by pDC may also require macrophage intrinsic STING signaling.

Since I was not part of the initial review I evaluated the previous reviewer comments and assessed how well the authors responded to the criticisms. I agreed with the majority of the concerns raised by the previous reviewers. Furthermore, the authors have attempted to strengthen the study in good faith and have included new studies. In particular, in inclusion of experiments where CD169 conditional STING-KO mice are used adds to the novelty of this study.

The study is largely an extension of their previous work and in the revised manuscript the results for the most part support their conclusions. The use of dynamic imaging to assess pDC dynamics in the bone marrow is interesting and will add new avenues for further studies.

This study also adds to our knowledge of how pDCs get activated and produce type I IFNs in vivo. The mechanisms of how pDCs get activated in vivo still remain poorly understood, however, this report begins to shed some light.

We thank this reviewer for his time and thorough assessment of our revised work.

Some deficiencies still remain, which will be discussed below:

Concerns that still remain:

1. Can the authors comment on the concern that use of anti-CD169 antibody for imaging the macrophages in vivo may prevent adequate uptake of infected RBCs or parasites in vivo?

We have conducted a new experiment where we quantified the uptake of labelled iRBC in mice injected with anti-CD169 or control isotype mAbs (new Figure 1—figure supplement 2B). Data confirmed the greater uptake in F4/80hi MP compared to CD169hi MP, but did not show any differences between anti-CD169 or isotype mAb-treated mice, suggesting that labelling CD169+ MP with the anti-CD169 mAb in vivo does not interfere with MP uptake of Py-iRBC.

2. Could the authors comment more on the physiological significance of type I IFN during malaria infection. Are pDCs needed for clearance of the infection and if type I IFN is required at all for the benefit of the host. This is a rather controversial topic and a clearer discussion taking their own studies in to account is warranted.

Type I IFN enhances inflammation and lethal outcomes in the P. yoelii YM (Spaulding et al., PLoS Pathogens 2016) and in the P. berghei (Sharma et al., Immunity 2011) models of lethal infections. It also prevents effective control of non-lethal P. yoelii (17XNL) and P. chabaudi chabaudi AS by preventing optimal Tfh and GB B cell responses as a result of inadequate inflammation (Zander et al., PLoS Pathogens 2016 ; Sebina et al., PLoS Pathogens 2016). Overall, these studies are consistent with the idea that type I IFN are detrimental in malaria infections. While in our study, we reported that pDC were the major producers of type I IFN during Py YM infection, pDC depletion only mildly rescued mice survival (Figure S3E in our PLoS Pathogens 2016), suggesting type I IFN from other cells significantly contributed to the lethal outcome. However, and as pointed out by this reviewer and in our original discussion, results from Yu et al. (Immunity 2016) differed from ours and from the general consensus. These authors reported that STING deficiency in pDC led to higher secretion of type I IFN as a result of the loss of SOCS1-mediated inhibition of TLR7/MyD88 signaling, and to greater resistance to malaria infection. They also revealed that pDC were required for type I IFNdependent mouse survival, a finding consistent with their model but discrepant with the broader consensus. We have further expanded and clarified this section of the discussion accordingly (page 20).

3. The fact that the motility of pDCs in TLR7 KO mice even under a steady state is so much higher still makes the interpretation of the data in Figure 3 difficult.

Indeed, it is notable that pDC are hyper motile in TLR7-deficient mice, suggesting that tonic TLR7 signaling may be regulating pDC motility rates at steady state. While infection does lower motility of TLR7-deficient pDC in a cell autonomous fashion, they never arrested like WT pDC, by all 3 measures (TV, DR and MSD). This suggested that other factors are also controlling pDC speed, but that these are insufficient to completely arrest these cells. As our previous work found that TLR7-deficient pDC cannot produce type I, slowing these cells during infection is insufficient for proper activation. Taken together, these results suggest that reduced motility is insufficient to lead to pDC activation and only complete arrest, with TLR7 signaling, can lead to proper activation. As another way to convey this finding, we have also represented the data by compiling the proportion of WT versus TLR7-/- pDC arrested or moving in the various experimental settings, using TV and DR threshold from WT mice (new Figure 3-supplement figure 1A). The data show a comparable proportion of pDC moving or arrested in Py-infected Tlr7-/y mice and naïve WT or Tlr7-/y mice, in contrast to Py-infected WT counterpart.

4. The authors make the claim that CD169+ macrophage intrinsic STING signaling is required for the production of Type I IFN by pDCs. However, they only use the reporter mouse for these studies, and the imaging is not very clear. Thus, the authors should show ELISA data in the serum and bone marrow to complement their imaging studies.

In our 2016 study, we established that the IFNbYFP reporter mouse faithfully tracks for type I IFN production by pDC, both for IFN and IFN in the serum and BM, as quantified by ELISA. We have clarified this aspect in the revised manuscript (pages 14-15).

5. The CD169+ macrophages themselves can produce copious amounts of Type I IFNs. Can the authors eliminate the possibility of paracrine IFN secreted by the macrophages as another mechanism by which pDCs respond to and produce more IFN – especially in studies where CD169-DTR mice are used or in STING conditional KO studies? Here too, assessing IFN production will be valuable.

It is very much a possibility that we considered, and further emphasized in the revised discussion (page 19). STING leads to TBK/IRF3 activation and IFNβ transcription, thus IFNβ produced by CD169+ MP via STING, could enhance pDC production of type I IFN. Depleting CD169+ MP prevents pDC to produce type I IFN (our 2016 study, using the reporter and ELISA) and we do not detect Ifnb gene transcription in CD169+ MP. In addition, removing type I IFN receptor on pDC is known to prevent them from secreting more type I IFN (Asselin-Paturel et al., JEM 2005). Also, we do not know of conditional Irf3 Irf7 double knockout mice that we could cross to Cd169Cre expressing mice to abrogate any type I IFN production by CD169+ MP. Thus, we feel that these important limitations in the experimental systems readily available to us, make the formal demonstration of the STING-dependent mechanism by which MP activate pDC largely beyond the scope of the current study.

6. Finally, the authors should be careful about making any substantive conclusions with respect to pDC – CD169+ macrophage interactions. In this respect the dynamic imaging is not as convincing. The authors themselves admit that due to the high density of macrophages in the BM "it makes it difficult to establish a functional interaction"

Our prior work showed that deletion of these MP prevented the production of type I IFN by pDC. The first revision of this manuscript incorporated new sets of experiments in which we directly investigated whether CD169+ MP regulated pDC motility (Figure 2E-G). These new data clearly established that selective depletion of CD169+ MP prevented pDC arrest during infection, directly linking the presence of CD169+ MP to pDC stopping. While in vivo imaging can never fully rule out that these functional interactions are indirect, or that an additional component is involved, it is more likely than not that they are linked, until new information supports an alternative model. We have now clarified these notions and revised the text to avoid any overstatements.

Reviewer #2:

This revised version investigates the role of CD169 BM macrophages in the activation of pDC following the injection of P.yoelii infected RBCs. It aims at demonstrating that BM CD169+ macrophages are key regulators of pDC retention and activation, in a TLR7 and STING dependant manner.

Overall, the take home message appears valid but some experiments need to be performed in order to exclude technical artifacts:

We thank this reviewer for his time and evaluation of our work.

1. TAMRA and PE appear to have very similar excitation and emission spectra. In addition, they both emit in the "Red" channel, which is typically highly autofluorescent. Why did the authors chose these two dyes to specifically label RBC and CD169+ Macs? The use of green/red (or blue/red or blue/green) fluorochromes might have improved the quality of the movies.

We apologize for the confusion between the text and the figure legend. In the IVM uptake experiments presented in Figure 1B and Figure 1-video 1, Py-iRBC were labeled with TAMRA-Red while CD169+ MP were labeled with anti-CD169-FITC mAb, allowing for easy discrimination between CD169+ MP (green) and Py-iRBC (red). In the FACS uptake experiments presented in Figure 1A, transferred Py-iRBC were also labeled with TAMRA-Red and CD169+ MP were stained ex vivo with anti-CD169-APC and F4/80-FITC. In all other IVM experiments of the paper, pDCs constitutively expressed GFP+ (PTCRA-eGFP) and CD169+ MP were labelled in vivo using anti-CD169-PE mAb injection. We have further clarified these technical details in the revised manuscript (revised text, Figure 1B and Figure 1—figure supplement 1B legends, and in the Materials and methods).

2. Reviewer 1 is correct when he/she states that PT treatment has indirect effects than can alter pDC motility and IFNa production. As the authors fail to identify the chemokines (or PT sensitive signalling pathways) that are at play, it is incorrect to state that pDC motility is required for pDC IFNa production. The authors should make this very clear in the manuscript.

We have carefully edited the revised version to reflect such limitations.

3. Figure 1 does not convince me that CD169 Macs pick up infected RBC. It is well documented that CD169+ macs are fragile (at least in lymphoid organs) and that most CD169+ cells analyzed by flow are not even macrophages (PMID: 22675532 and more recently PMID: 34818538). As IVM lacks resolution, the authors should perform the same experiment in a PTCRA-eGFP mouse and stain fixed bone sections for CD169/GFP expression and then quantify the amount of fluorescent RBC captured by pDC and CD169+ macs. In addition, as the authors possess CD169Cre mice, I am wondering why they did not cross this strain to a fluorescent reporter such as the Rosa-RFP line to (a) visualize CD169+ macs by IVM and (b) prove that pDC are not fate-mapped in this model, and thus not targeted in CD169DTR mice.

We have crossed the Cd169Cre mouse to both Rosa26LSLYFP and Rosa26LSLTd mice, as requested by this reviewer (new Figure 1—figure supplement 1C). The data compare both reporters and the in vivo anti-CD169 mAb labelling of CD169+ MP. In all three settings, we observe cells expressing low and high levels of CD169. Only the high expressors corresponded to CD169+ MP (>80%) while the low expressors likely resulted from non-specific staining or expression (YFP or tomato) of CD11b+ cells. This suggested that in vivo anti-CD169 mAb labelling is as effective to track CD169+ MP as the reporter systems are. Thus, we do not believe that redoing all IVM experiments in Cd169Cre reporter mice would change any of the major findings.

As far as pDC targeting in the Cd169DTR mouse, we have reported in our 2016 study (Figure S6A) that DTtreatment in uninfected and in Py-infected Cd169DTR mice did not decrease pDC proportion in BM, blood and spleen. The finding that pDC proportion and numbers (current Figure 7B) are significantly increased in the BM of DT-treated Pyinfected Cd169DTR mice is consistent with our the results showing increased retention of pDC in the BM of STINGdeficient mice (Figure 5I) and in mice where CD169+ MP lack STING (Figure 7B). Altogether, these results strongly argue against pDC targeting in the Cd169Cre or Cd169DTR models.

4. Related to the previous points, I am concerned by FigS2C and sup movie 3. Both are showing data obtained in CD169 DTR mice treated with DT. While flow cytometry showed a 6 fold reduction of CD169+ macs upon DT injection, IVM does not show a similar decrease. If true, it suggests that one technique is not reliable to observe CD169+ macs. Can the authors comment on this?

Thank you for noticing this. We went back to the original movie settings and realized that the two movies had distinct gain settings to favor the visualization of pDC. Unfortunately, postprocessing oversaturated signal in the PE (red) channel, making CD169- autofluorescent cells appear PE+ in the DT-treated mice. We have reduced the gain in the re-processed corresponding movie without altering the gain (“version 2”) so that we can provide side by side with the prior version, together with the two different export settings. We also provide for this reviewer (Author response image 1) an example image projection of the expected anti-CD169-PE punctated staining (red arrows) in control versus DT-treated Py-infected mice, which is very different from that of red-autofluorescent cells (orange arrows).

Gain settings optimized for visualization of MP:

o RDX2 settings

Mix: 104

Max: 805.67

Gamma: 1.28

o RDX3 settings

Mix 128.98

Max: 4095.00

Gamma: 1.34

Author response image 1
https://doi.org/10.7554/eLife.78873.sa2

Article and author information

Author details

  1. Jamie Moore-Fried

    Albert Einstein College of Medicine, Department of Microbiology and Immunology, Bronx, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    Contributed equally with
    Mahinder Paul
    Competing interests
    No competing interests declared
  2. Mahinder Paul

    Albert Einstein College of Medicine, Department of Microbiology and Immunology, Bronx, United States
    Contribution
    Formal analysis, Investigation, Methodology, Project administration, Writing - review and editing
    Contributed equally with
    Jamie Moore-Fried
    Competing interests
    No competing interests declared
  3. Zhixin Jing

    1. Albert Einstein College of Medicine, Department of Microbiology and Immunology, Bronx, United States
    2. Albert Einstein College of Medicine, Department of Pathology, New York, United States
    Contribution
    Software, Formal analysis, Supervision, Validation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  4. David Fooksman

    1. Albert Einstein College of Medicine, Department of Microbiology and Immunology, Bronx, United States
    2. Albert Einstein College of Medicine, Department of Pathology, New York, United States
    Contribution
    Conceptualization, Resources, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Project administration, Writing - review and editing
    For correspondence
    david.fooksman@einsteinmed.edu
    Competing interests
    No competing interests declared
  5. Gregoire Lauvau

    Albert Einstein College of Medicine, Department of Microbiology and Immunology, Bronx, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    gregoire.lauvau@einsteinmed.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3050-2664

Funding

NIAID (AI103666)

  • David Fooksman
  • Gregoire Lauvau

NIH (T32 GM7288)

  • Jamie Moore-Fried

NIH (F31 HL147470)

  • Jamie Moore-Fried

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the AECOM FACS core facilities. We thank Ilseyar Akhmetzyanova for helping with initial training on intravital microscopy and Zachary Benet for providing the MATLAB script to make the displacement charts featured in this manuscript.

This work was funded by the National Institutes of Health Grants (NIH/NIAID) AI103666 and Hirschl Caulier Award to GL and AI128735 to DF and GL. JM was supported by NIH MSTP training grant T32 GM7288-46 and F31 HL147470.

Ethics

This study was carried out in strict accordance with the recommendations by the animal use committee at the Albert Einstein College of Medicine under protocol number #20171113 and 00001375. The institution is accredited by the 'American Association for the Use of Laboratory Animals' (DHEW Publication No. (NIH) 78-23, Revised 1978), and accepts as mandatory the NIH 'Principles for the Use of Animals'. All efforts were made to minimize suffering and provide humane treatment to the animals included in the study.

Senior Editor

  1. Tadatsugu Taniguchi, University of Tokyo, Japan

Reviewing Editor

  1. Florent Ginhoux, Agency for Science Technology and Research, Singapore

Publication history

  1. Received: March 23, 2022
  2. Preprint posted: April 3, 2022 (view preprint)
  3. Accepted: October 23, 2022
  4. Accepted Manuscript published: October 24, 2022 (version 1)
  5. Version of Record published: November 10, 2022 (version 2)

Copyright

© 2022, Moore-Fried, Paul et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 372
    Page views
  • 92
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jamie Moore-Fried
  2. Mahinder Paul
  3. Zhixin Jing
  4. David Fooksman
  5. Gregoire Lauvau
(2022)
CD169+ macrophages orchestrate plasmacytoid dendritic cell arrest and retention for optimal priming in the bone marrow of malaria-infected mice
eLife 11:e78873.
https://doi.org/10.7554/eLife.78873

Further reading

    1. Immunology and Inflammation
    Sara E Vazquez, Sabrina A Mann ... Joseph L DeRisi
    Research Advance Updated

    Phage immunoprecipitation sequencing (PhIP-seq) allows for unbiased, proteome-wide autoantibody discovery across a variety of disease settings, with identification of disease-specific autoantigens providing new insight into previously poorly understood forms of immune dysregulation. Despite several successful implementations of PhIP-seq for autoantigen discovery, including our previous work (Vazquez et al., 2020), current protocols are inherently difficult to scale to accommodate large cohorts of cases and importantly, healthy controls. Here, we develop and validate a high throughput extension of PhIP-seq in various etiologies of autoimmune and inflammatory diseases, including APS1, IPEX, RAG1/2 deficiency, Kawasaki disease (KD), multisystem inflammatory syndrome in children (MIS-C), and finally, mild and severe forms of COVID-19. We demonstrate that these scaled datasets enable machine-learning approaches that result in robust prediction of disease status, as well as the ability to detect both known and novel autoantigens, such as prodynorphin (PDYN) in APS1 patients, and intestinally expressed proteins BEST4 and BTNL8 in IPEX patients. Remarkably, BEST4 antibodies were also found in two patients with RAG1/2 deficiency, one of whom had very early onset IBD. Scaled PhIP-seq examination of both MIS-C and KD demonstrated rare, overlapping antigens, including CGNL1, as well as several strongly enriched putative pneumonia-associated antigens in severe COVID-19, including the endosomal protein EEA1. Together, scaled PhIP-seq provides a valuable tool for broadly assessing both rare and common autoantigen overlap between autoimmune diseases of varying origins and etiologies.

    1. Immunology and Inflammation
    2. Structural Biology and Molecular Biophysics
    Rui Liu, Kangcheng Song ... Lei Chen
    Research Article Updated

    Phagocyte oxidase plays an essential role in the first line of host defense against pathogens. It oxidizes intracellular NADPH to reduce extracellular oxygen to produce superoxide anions that participate in pathogen killing. The resting phagocyte oxidase is a heterodimeric complex formed by two transmembrane proteins NOX2 and p22. Despite the physiological importance of this complex, its structure remains elusive. Here, we reported the cryo-EM structure of the functional human NOX2-p22 complex in nanodisc in the resting state. NOX2 shows a canonical 6-TM architecture of NOX and p22 has four transmembrane helices. M3, M4, and M5 of NOX2, and M1 and M4 helices of p22 are involved in the heterodimer formation. Dehydrogenase (DH) domain of NOX2 in the resting state is not optimally docked onto the transmembrane domain, leading to inefficient electron transfer and NADPH binding. Structural analysis suggests that the cytosolic factors might activate the NOX2-p22 complex by stabilizing the DH in a productive docked conformation.