Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Apr 11.
Published in final edited form as: Cell Rep. 2021 Sep 21;36(12):109728. doi: 10.1016/j.celrep.2021.109728

Single-cell secretion analysis reveals a dual role for IL-10 in restraining and resolving the TLR4-induced inflammatory response

Amanda F Alexander 1,4, Ilana Kelsey 1, Hannah Forbes 1, Kathryn Miller-Jensen 1,2,3,5,*
PMCID: PMC8995750  NIHMSID: NIHMS1742916  PMID: 34551303

SUMMARY

Following Toll-like receptor 4 (TLR4) stimulation of macrophages, negative feedback mediated by the anti-inflammatory cytokine interleukin-10 (IL-10) limits the inflammatory response. However, extensive cell-to-cell variability in TLR4-stimulated cytokine secretion raises questions about how negative feedback is robustly implemented. To explore this, we characterize the TLR4-stimulated secretion program in primary murine macrophages using a single-cell microwell assay that enables evaluation of functional autocrine IL-10 signaling. High-dimensional analysis of single-cell data reveals three tiers of TLR4-induced proinflammatory activation based on levels of cytokine secretion. Surprisingly, while IL-10 inhibits TLR4-induced activation in the highest tier, it also contributes to the TLR4-induced activation threshold by regulating which cells transition from non-secreting to secreting states. This role for IL-10 in restraining TLR4 inflammatory activation is largely mediated by intermediate interferon (IFN)-β signaling, while TNF likely mediates response resolution by IL-10. Thus, cell-to-cell variability in cytokine regulatory motifs provides a means to tailor the TLR4-induced inflammatory response.

In brief

Alexander et al. use multiplexed single-cell secretion analysis to disentangle TLR4-induced autocrine feedback loops. They show that IL-10 raises the threshold of inflammatory activation in macrophages while also negatively regulating highly activated cells. The restraining role of IL-10 is mediated by IFN-β signaling, while TNF-induced IL-10 likely mediates response resolution.

Graphical Abstract

graphic file with name nihms-1742916-f0006.jpg

INTRODUCTION

Tight regulation of the inflammatory response is necessary for immune homeostasis (Medzhitov, 2008; Murray and Smale, 2012). To control the inflammatory response, innate immune cells use both positive and negative feedback motifs, some of which are mediated through cell-cell communication (Gottschalk et al., 2019; Hu et al., 2008; Lee et al., 2009). An important example is interleukin-10 (IL-10), a secreted negative regulator necessary for inflammatory control and resolution (Spits and de Waal Malefyt, 1992). IL-10 suppresses production of proinflammatory cytokines and chemokines (C/Cs) and restricts activation of antigen-presenting cells and the adaptive immune response (Conaway et al., 2017; Dagvadorj et al., 2008; Mittal and Roche, 2015; Murray, 2005). Dysregulation of IL-10 is associated with inflammatory and autoimmune pathologies (Iyer and Cheng, 2012), but attempts to therapeutically harness the anti-inflammatory activity of IL-10 have been largely unsuccessful due to an incomplete understanding of how IL-10 negative feed-back is regulated (O’Garra et al., 2008; Saxena et al., 2015).

Mechanisms of IL-10 negative regulation are difficult to discern due to its complex interactions with other regulatory motifs. For instance, following Toll-like receptor 4 (TLR4) stimulation, tumor necrosis factor (TNF) and interferon (IFN)-β are key paracrine signals that regulate the inflammatory response in macrophages and dendritic cells, including IL-10 production (Figure 1A). IFN-β paracrine signaling is necessary for robust IL-10 secretion (Chang et al., 2007; Howes et al., 2016; Iyer et al., 2010), and a subset of high IFN-β-producing cells appear to drive IL-10 production in neighboring cells (Shalek et al., 2014). TNF also promotes the secretion of IL-10 following TLR4 stimulation (Caldwell et al., 2014; Muldoon et al., 2020) and, analogous to IFN-β, a subset of high TNF-secreting cells amplifies IL-10 secretion in surrounding macrophages (Xue et al., 2015). The biological significance of two intermediate paracrine signals that exhibit high cell-to-cell variability controlling a critical resolving cytokine is unclear.

Figure 1. IL-10 suppresses TLR4-induced proinflammatory activation in a dose-dependent manner.

Figure 1.

Open in a new tab

(A) Schematic of paracrine signaling cascades activated by TLR4 stimulation.

(B) Time course of TNF secretion following stimulation of BMDMs at the indicated LPS dose. TNF secretion was measured by ELISA and is presented as mean ± standard error of the mean (SEM) of 3 biological replicates.

(C) Area under the curve (AUC) calculated from TNF secretion over time for each dose in (B).

(D) Dose response of TNF and IL-10 secretion following LPS stimulation of BMDMs for 8 h (left) and 24 h (right). Protein secretion was measured by ELISA and is presented as mean ± SEM of 3 biological replicates. Hill slopes calculated from 4-parameter logistic curves fitted to the dose-response data.

Here, we investigated the role of IL-10 in shaping the TLR4-induced macrophage response in primary murine bone marrow-derived macrophages (BMDMs) treated with TLR4 agonist lipopolysaccharide (LPS). To minimize the influence of paracrine signaling, we used a microwell assay that allows auto-crine signaling without interference from neighboring-cell signals to measure multiplexed C/C secretion in individual BMDMs. We found that activated macrophages lie along a trimodal proinflammatory axis, with IL-10 negative feedback modulating the distribution of cells within each activation tier. While IL-10 restrained activation of highly activated macrophages, its primary effect was to raise the threshold of activation for low and non-responder cells. IL-10’s role in setting the TLR4 activation threshold was largely mediated by IFN-β, and together these cytokines regulated the fraction of cells that mount a robust inflammatory response. By contrast, TNF positive feedback increased IL-10 secretion by highly activated macrophages, which contributed to resolution of the response. Thus, our study disentangles IL-10 signaling to reveal two distinct roles in negatively regulating the TLR4 inflammatory response.

RESULTS

IL-10 suppression of TLR4-induced proinflammatory activation increases with LPS dose

IL-10 negatively regulates the TLR4-induced inflammatory response (Figure S1A). However, IL-10 negative feedback is complicated by the fact that IL-10 is induced by the same signals it regulates, specifically TNF and IFN-β (Figure 1A). To characterize the relationship between TNF and IL-10 in the TLR4 response, we stimulated BMDMs in population for up to 24 h with a range of LPS doses (Figure 1B). Total TNF secretion, as calculated by the area under the curve of the time course, increased steeply between 1 and 10 ng/ mL LPS and then dropped significantly at higher doses (Figure 1C). This switch-like induction of TNF is consistent with a known threshold of TLR4-induced inflammatory activation mediated by mitogen-activated protein kinase (MAPK) intra-cellular signaling (Figure S1B) (Gottschalk et al., 2016). The decrease in TNF at high doses is likely due to negative feedback from anti-inflammatory factors such as IL-10.

Directly comparing TNF and IL-10 secretion 8 and 24 h post-LPS stimulation, we found that IL-10 was induced between 1 and 10 ng/mL similar to TNF (Figure 1D). However, while TNF secretion peaked at 10 ng/mL LPS, IL-10 secretion was highest at 1,000 ng/mL. Relatedly, we observed diminished TNF secretion at 100 and 1,000 ng/mL LPS, coinciding with peak IL-10 production. Together, these data show that IL-10-mediated suppression of TNF and other proinflammatory C/Cs increases with LPS dose.

TLR4-activated macrophages exhibit trimodal activation of the inflammatory response

Substantial cell-to-cell variability characterizes TLR4-stimulated macrophage secretion (Lu et al., 2015; Ramji et al., 2019), and this variability plays a role in permitting rapid and reproducible innate immune responses (Eberwine and Kim, 2015). To explore how TLR4-induced activation varied across individual BMDMs, we conducted multiplexed, single-cell secretion profiling (Lu et al., 2013). In this microwell format, cells are isolated such that functional autocrine signaling proceeds without interference from paracrine signals produced by neighboring cells.

We stimulated BMDMs with a range of LPS doses (0, 10, 100, and 1,000 ng/mL) for 8 h in the microwell device and measured 12 TLR4-induced secreted C/Cs (Figure 2A; Table S1). We observed LPS dose-dependent increases in the production of TNF and IL-10 in both the fraction of cells secreting each cytokine and the intensity of that secretion (Figure 2B). Interestingly, the average observed fraction of BMDMs secreting TNF was approximately 35%, while fewer than 20% secreted IL-10, even at the highest doses (Figure 2C). Hill coefficients for the fraction of cells responding with increasing LPS dose were >1, indicating switch-like behavior. However, TNF and IL-10 induction in microwells required higher LPS doses than the population response, because loss of paracrine signaling between cells reduces LPS activation in the microwells (Xue et al., 2015).

Figure 2. Single-cell secretion analysis reveals tiered activation of the TLR4 secretion program.

Figure 2.

Open in a new tab

(A) Schematic of workflow for single-cell secretion profiling. BMDMs were seeded onto the microwell chip and then stimulated with 0, 10, 100, or 1,000 ng/mL LPS for 8 h. High-dimensional secretion data were visualized by PHATE.

(B) Violin plots of single-cell secretion from BMDMs stimulated with LPS in the microwell assay. Black bar indicates fluorescent threshold of detection. Data are pooled from 3 independent biological replicates.

(C) Four-parameter logistic curves were interpolated from the mean percentage of cells secreting TNF and IL-10 above threshold in response to the indicated LPS dose. Data presented as mean ± 95% confidence intervals (CIs) calculated by bootstrapping.

(D and E) 2D PHATE visualization of 10-dimensional single-cell secretion from BMDMs stimulated with LPS as in (A) (four conditions). Data include activated macrophages (i.e., secreting at least 1 measured cytokine above threshold) colored by (D) relative intensity of the indicated cytokine (non-zero cells brought to front for visualization) or (E) the KDE for cells at the indicated LPS dose (other doses shown in gray).

(F)Scatterplots of the number of proteins co-secreted in the microwell device versus PHATE 1 coordinate. Spearman correlation = 0.85 (p value < 0.0001).

(G)KDE for individual BMDMs along the PHATE 1 axis calculated from data (E).

To further explore TLR4-mediated single-cell secretion, we visualized our single-cell secretion profiling data in two dimensions using an unsupervised dimensionality reduction algorithm, PHATE (potential of heat diffusion for affinity-based transition embedding) (Moon et al., 2019). PHATE uses a diffusion mapping algorithm to visualize continual progressions within biological data. We aggregated all LPS doses (0–1,000 ng/mL) and excluded cells that did not secrete at least one protein from Table S1. Our final PHATE analysis included secretion measurements often C/Cs (Figure 2A). CCL2 and Chitinase 3-like 3 (Chi3l3) were excluded because they were not clearly regulated by LPS.

Projecting the single-cell data onto two dimensions revealed that most BMDMs lay parallel to the PHATE 1 axis. At one end of the axis were cells secreting few proteins, while on the other end were cells secreting high levels of multiple C/Cs (Figure 2D). The PHATE 2 axis was largely defined by secretion of CCL3. When we overlaid LPS dose labels onto BMDMs projected in PHATE, we found that unstimulated cells mostly occupied the low end of the PHATE 1 activation axis and the high end of the PHATE 2 axis, characterized by secretion of only CCL3. Stimulating with LPS moved cells into the PHATE 1 activation axis, with cells secreting more inflammatory C/Cs at higher intensities (Figure 2E). Overall, BMDMs demonstrated a strong correlation between polyfunctionality (i.e., number of C/Cs being cosecreted) and their PHATE 1 coordinates (Figure 2F).

The kernel density estimation (KDE) of TLR4-stimulated cells across PHATE 1 identified a trimodal response with three distinguishable levels of graded activation as determined by fitting a Gaussian mixture model (Figure 2G; Figures S2A and S2B). Unstimulated cells were located almost exclusively in the low activation tier, while LPS moved cells into the intermediate and high tiers. Notably, a greater proportion of cells resided in the high tier at 100 versus 1,000 ng/mL LPS, suggesting that in the presence of high levels of LPS, regulatory mechanisms prevent excessive proinflammatory activation, possibly through secretion of IL-10.

IL-10 negative feedback modulates heterogeneity in the macrophage responder population

Although LPS-stimulated BMDMs increased secretion poly-functionality along the PHATE 1 axis, the strength of the correlation between the magnitude of secretion (i.e., signal intensity) from individual cells and the PHATE 1 coordinate varied by C/C (Figure S2C). C/Cs associated with the proinflammatory response (TNF, CCL5, CXCL1, and IL-6) exhibited strong corre lations (Spearman R ≥ 0.55), suggesting that PHATE 1 predominantly captures the degree of proinflammatory activation. By contrast, we observed weaker correlations for IL-10, and for IL-27 and IFN-β (Spearman R < 0.34), which mediate the secretion of IL-10 (Fitzgerald et al., 2013; Iyer et al., 2010). The weak correlation between IL-10 and the proinflammatory PHATE 1 axis suggested that IL-10 may not be acting solely to resolve the response.

To explore this possibility, we compared LPS stimulation alone and in combination with either an IL-10 receptor blocking antibody (IL-10R Ab) to block IL-10 negative feedback or with recombinant IL-10 to uniformly impose IL-10 negative feed-back. Combining the BMDM secretion profiling datasets (i.e., LPS dose response and IL-10 perturbations) and visualizing with PHATE produced a similar proinflammatory axis along PHATE 1 (Figure S3A). Overlaying conditions revealed that blocking IL-10 negative feedback resulted in more highly activated cells at the far-right end of the axis for 1,000 ng/mL compared with LPS alone (Figure S3B). However, the more striking effect was a shift in BMDMs from the low-responding region of the PHATE 1 axis into the more activated region (Figure 3A; Figure S3B). Interestingly, in the absence of IL-10 signaling, the trimodal nature of the KDE for cells along the PHATE 1 axis was more apparent, as co-treatment with LPS and IL-10R Ab produced a higher proportion of cells in the inter-mediate and highly activated states (Figure 3B). Adding recombinant IL-10 dampened the proinflammatory response, with activated BMDMs exhibiting very little spread into the PHATE 1 activation axis at any LPS dose (Figure 3B; Figure S3B). Altogether, these data suggest that while IL-10 constrains highly activated macrophages, IL-10 negative feedback also acts to restrain low-activated cells.

Figure 3. IL-10 negative feedback modulates heterogeneity in the TLR4 secretion program.

Figure 3.

Open in a new tab

(A) 2D PHATE visualization of 9-dimensional single-cell secretion data for nine conditions: BMDMs stimulated for 8 h with LPS alone (10, 100, or 1,000 ng/mL), co-stimulated with IL-10R Ab (30 μg/mL) at each dose, or co-stimulated with recombinant IL-10 (10 ng/mL) at each dose. Data are colored by KDE for the 100 ng/mL LPS conditions (other cells grayed out for visualization).

(B) KDE for individual BMDMs along the PHATE 1 axis calculated from 100 ng/mL LPS stimulation shown in (A) and from 1,000 and 10 ng/mL LPS stimulations (Figure S3B). Arrows indicate inter-mediate and high modes of the trimodal PHATE 1 KDE distribution. The p values were calculated using the Kolmogorov-Smirnov test to determine whether two samples come from the same distribution.

To more closely explore the effect of IL-10 on TNF, we converted single-cell protein secretion intensities from the microwell device to concentrations using recombinant protein standard curves (Figure S4A). Concentrations below the threshold of detection in the device were set to 1 pg/mL (i.e., 0 on a log scale) for visualization, which was just below the lowest extrapolated concentrations detected. These data revealed that blocking IL-10 signaling increased the fraction of cells secreting TNF, mostly affecting intermediate-level responders (Figure 4A). By contrast, adding recombinant IL-10 significantly decreased the fraction of TNF-secreting cells in response to LPS stimulation and their average level of secretion. IL-10 similarly affected IL-6 and CCL5 (Figure 4B) as well as IL-27 and IFN-β (Figure S4B). Interestingly, blocking IL-10 signaling did not increase the average secretion of responder cells at 100 ng/mL LPS and even decreased the average secretion of responder cells at 1,000 ng/mL LPS (Figure 4B), presumably due to recruitment of low-level secretors into the activated population. These data suggest that IL-10 negative feedback acts to limit the total number of cells secreting TNF and other proinflammatory C/Cs in response to LPS, rather than acting solely to inhibit proinflammatory secretion in activated cells.

Figure 4. Low versus high levels of IL-10 negative feedback differentially modulate TNF activation.

Figure 4.

Open in a new tab

(A) Probability histograms for TNF secretion after 8 h of the indicated stimulation in the microwell device. Non-responding cells were set to 1 (“OFF”). Inset graphs show TNF+ subpopulation.

(B) Bargraphs of average percentage of cells secreting the protein above threshold in response to the indicated stimulation cues after 8 h in the microwell assay (top) or mean secretion level from the responding subpopulation of cells (only cells secreting indicated cytokine/chemokine above detection threshold in microwell device) (bottom). Data presented as mean ± 95% CIs calculated by bootstrapping. Significance determined by non-overlap of CIs.

(C) Conditional density re-scaled visualization (DREVI) plots showing the relationship between IL-10 and TNF. Data shown are pooled to include all stimulation doses (0, 10, 100, and 1,000 ng/mL LPS with or without IL-10R Ab at 30 μg/mL) where the 2 indicated cytokines were co-secreted during the 8-h stimulation (LPS alone: n = 516, +IL-10R Ab: n = 212). Arrows indicate areas of low and high IL-10 secretion most affected by IL-10 negative feedback.

(D) Edge response functions for DREVI plots shown in (C) were fit according to the conditional mean at the region of highest conditional density.

Low versus high levels of IL-10 negative feedback differentially modulate TLR4-mediated proinflammatory activation

A surprising observation upon blocking IL-10 negative feedback is that the affected responder population is larger than the sub-population of cells secreting IL-10. For example, at 100 ng/mL LPS stimulation, an average of 13.1% [11.6, 14.7] (95% CI) of cells secreted IL-10 above background in the microwell device (Figure 2C). However, blocking IL-10 at this same dose increased the fraction of cells secreting IL-6 and CCL5 to 17.6% [15.1, 19.8] and 16.1% [14.4, 17.8], respectively (Figure 4B), and also increased the average per cell secretion of IL-10 nearly 3-fold (Figure S4C), indicating that cell uptake of IL-10 is significant. Together, these data suggest that IL-10 may elicit negative feedback at concentrations below the limit of detection in the microwell device.

To investigate this further, we examined how the concentration of secreted TNF varied with the concentration of secreted IL-10 using conditional density re-scaled visualization (DREVI) (Krishnaswamy et al., 2014). DREVI re-scales multiplexed single-cell data points by their conditional density in order to robustly visualize relationships between proteins across a larger dynamic range. In this case, DREVI allowed us to emphasize the small number of cells that exhibited low levels of IL-10 secretion (i.e., <10 pg/mL) and look at the effect of this secretion on TNF. We pooled our interpolated microwell secretion data from all stimulation doses (0, 10, 100, and 1,000 ng/mL LPS) with and without IL-10 receptor blocking and then confined our DREVI analysis to individual BMDMs producing both cytokines.

DREVI analysis of the LPS-stimulated co-secretion relation-ship between TNF and IL-10 identified three tiers of TNF activation—low, intermediate, and high—that tracked with increasing concentrations of IL-10. Within each tier, as IL-10 increased, the concentration of TNF plateaued or decreased (Figure 4C, left). Blocking IL-10 negative feedback altered the functional relationship between these two cytokines, specifically at low (<10 pg/mL) and high (>300 pg/mL) concentrations of IL-10 (Figure 4C, right). Without IL-10 negative feedback, we observed co-production of low levels of IL-10 secretion with both low and intermediate levels of TNF secretion, suggesting that low concentrations of IL-10 prevented low TNF responders (or non-responders) from becoming intermediate responders via switch-like activation (Figure 1). Notably, we do not observe inhibition of TLR4-stimulated TNF production in BMDMs in population at IL-10 concentrations below 10 ng/mL (Figure S4D), lead-ing us to hypothesize that IL-10 may be concentrated locally at the cell surface. We further observed that the highest tier of TNF activation no longer plateaued with increasing IL-10, but instead continued to increase (Figure 4D). Overall, our analysis shows that low amounts of IL-10 are sufficient to regulate heterogeneity in the TLR4 response.

IFN-β largely mediates IL-10 regulation of the threshold for TLR4 activation, while TNF activates the resolving role of IL-10

Finally, we sought to determine whether TNF and IFN-β had distinct roles in mediating IL-10 negative feedback. To explore this, we again performed multiplexed single-cell secretion profiling on BMDMs stimulated with 100 ng/mL LPS alone, or co-stimulated with either soluble TNF receptor (sTNFR) to block TNF autocrine signaling, or with interferon-α/β receptor blocking antibody (IFNAR Ab) to block IFN-β autocrine signaling. Both sTNFR and IFNAR Ab significantly reduced the fraction of cells secreting IL-10 as well as the magnitude of that secretion (Figures S5A and S5B), confirming roles for TNF and IFN-β in promoting IL-10 production.

We combined our sTNFR and IFNAR Ab results with our previous IL-10R Ab experiment and visualized the data with PHATE (Figure 5A; Figure S5C). Co-stimulation with LPS and sTNFR substantially diminished the proinflammatory response as evidenced by the lack of cells throughout the PHATE 1 proinflammatory axis (Figures 5B and 5C) and by significant decreases in the percentage of cells producing inflammatory cytokines IL-6, IL-27, CXCL1, and CCL5 (Figure 5D). While these results confirm TNF as a strong positive regulator of the TLR4 response, the attenuated response prevented a direct analysis of how TNF-stimulated IL-10 impacted TLR4 activation.

Figure 5. IFN-β largely mediates IL-10 regulation of the threshold for TLR4 activation, while TNF positive feedback activates the resolving role of IL-10.

Figure 5.

Open in a new tab

(A and B) 2D PHATE projection of 9-dimensional single-cell secretion data from BMDMs stimulated with 100 ng/mL LPS alone or co-stimulated with 30μg/mL IL-10R Ab, 5 μg/mL IFNAR Ab, or 5 μmg/mL sTNFR for 8 h in the microwell device (four conditions). Data colored by relative secretion intensity of the indicated protein (A) or cell density (B) for the indicated subpopulation of cells (other cells grayed out for visualization).

(C) KDE for BMDMs along the PHATE 1 axis stimulated as indicated and calculated from data shown in (A) and (B). Arrows indicate intermediate and high modes of the trimodal PHATE 1 distribution. Number of biological replicates vary by experiment: LPS alone (n = 3); sTNFR (n = 2); IFNAR and IL-10R (n = 1).

(D) Bar graphs of percentage of cells secreting the indicated protein above threshold in response to the indicated cues after 8 h in the microwell assay. Data presented as mean ± 95% CIs calculated by bootstrapping. Significance determined by non-overlap of CIs.

(E) Dose response of TNF secretion following LPS stimulation of BMDMs for 24 h. TNF secretion was measured by ELISA and presented as mean ± SEM of 2 biological replicates.

(F) Schematic model illustrating dual roles for IL-10 in the TLR4 response.

By contrast, cells co-stimulated with LPS and IFNAR Ab increased their proinflammatory activity, as evidenced by movement of cells farther into the proinflammatory activation axis (Figure 5B), and by significant increases in the percentage of cells secreting TNF, IL-6, CXCL1, and CCL5, similar to what we observed with LPS+IL10R Ab (Figure 5D). However, the fraction of BMDMs secreting IL-27 and CCL3 remained unchanged with IL-10 signaling blocked, but significantly increased when IFNAR was blocked (Figure 5D). Despite these differences, we found significant functional crossover between negative feedback from IL-10 and IFN-β in both the digital and analog components of the TLR4 response.

Analyzing the KDE for cells along the PHATE 1 axis revealed that blocking TNF moved most BMDMs into the lowest tier of TLR4 activation, consistent with its primary role in positive feed-back (Figure 5C). Blocking IFNAR significantly increased BMDMs in the intermediatetier of activation, but did not increase activation in the high tier. By contrast, blocking IL-10R increased BMDMs in both the intermediate and high activation tiers (Figure 5C). Taken together, these data raise the possibility that TNF, or another paracrine signal, is promoting IL-10 secretion primarily in the highest activation tier.

Our evidence that IL-10 and IFN-β negative feedback largely act to restrain cells from responding to LPS led us to postulate that IL-10 may play a role in setting the threshold for TLR4 activation. To explore this possibility, we stimulated BMDMs in a plate with increasing doses of LPS alone or co-stimulated with IL-10R Ab or IFNAR Ab for 24 h and measured TNF concentration by ELISA. With LPS stimulation alone, we observed TNF induction above 1 ng/mL LPS stimulation dose, but with IL-10 or IFN-β extracellular signaling blocked, we saw TNF induced between 0.1 and 1 ng/mL LPS (Figure 5E). Thus, IL-10, mediated by IFN-β paracrine signaling, contributes to a threshold of TLR4-induced proinflammatory activation.

DISCUSSION

Macrophages use paracrine communication networks to appropriately respond to an immune threat while avoiding tissue damage due to hyperinflammation (Murray and Smale, 2012). In this study, we combined single-cell and cell-population assays to investigate how heterogeneity in proinflammatory C/C secretion and in IL-10 negative feedback shape the TLR4 response. Over-all, we found that the TLR4 response displays tiered activation and that IL-10 regulates the distribution of macrophages within those tiers.

By visualizing high-dimensional single-cell secretion data in 2D using PHATE, we identified trimodal activation of proinflammatory secretion (Figures 2 and 3). Previous studies showed bimodal cytokine production when measuring a single protein, such as TNF or IL-6 (Muldoon et al., 2020; Shalek et al., 2013), which is consistent with our observed intermediate and high activation tiers. Our analysis expands on those observations by simultaneously analyzing 10 C/Cs secreted in response to TLR4 stimulation. This multiplexed analysis revealed an additional mode of low activation in which cells secrete few cytokines at low intensity or only robustly secrete CCL3. Our earlier study of human macrophage single-cell secretion stimulated with various TLR agonists identified loosely defined functional subgroups that were mostly determined by varied degrees of activation as opposed to the secretion of entirely different proteins (Lu et al., 2015). We hypothesize that additional doses of the TLR4 agonist LPS allowed visualization of the full range of macrophage activation, revealing that the previously identified subgroups may have represented snapshots of different activation tiers as opposed to functionally distinct clusters.

We found that IFN-β acts via IL-10 to raise the threshold for TLR4 activation in macrophages (Figure 5). While the relationship between IL-10 and IFN-β and their anti-inflammatory function is known (Chang et al., 2007; McNab et al., 2014; Shalek et al., 2014), a role for IFN-β in preventing macrophages from responding to TLR4 stimulation was not previously described. Interestingly, activation of IFN-β is stochastic in response to viral infection, and paracrine signaling to neighboring cells is key to establishing an antiviral state (Rand et al., 2012; Zhao et al., 2012). Our study suggests a similar mechanism might establish a state refractory to proinflammatory activation.

A major unanswered question is how molecular motifs activated by TLR4 stimulation combine to produce tiered activation states. We hypothesize that these states result from the interaction of TNF positive feedback and IL-10 negative feedback motifs at both low and high levels of TLR4 stimulation (Figure 5F). Such interacting motifs produce multiple stable states in other biological systems (Zhang et al., 2014; Zhou et al., 2018). We anticipate that future work exploring mathematical models of TNF-IL-10 interactions would be a productive way to investigate these mechanisms.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kathryn Miller-Jensen (kathryn.miller-jensen@yale.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

  • All data reported in this paper will be shared by the lead contact upon request.

  • All original code for the single-cell secretion profiling analysis and data processing are publicly available at https://github.com/Miller-JensenLab/Single-Cell-Analysis as of the date of publication.

  • Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact, Kathryn Miller-Jensen (kathryn.miller-jensen@yale.edu) upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice

Wild-type C57BL/6J mice (female, 6 weeks) were purchased from Jackson Laboratories. All mice were housed in the Yale Animal Resources Center in specific pathogen-free conditions. All animal experiments were performed according to the approved protocols of the Yale University Institutional Animal Care and Use Committee.

Cultured Cells

Bone marrow derived macrophages (BMDMs) were generated as previously described (Trouplin et al., 2013). In brief, bone marrow was extracted from the tibias and femurs of the hind legs of mice 8–12 weeks of age with a syringe. Afterward, red blood cells were lysed with ammonium-chloride-potassium lysis buffer (Lonza) and the cells were incubated for 4 hours at 37°C with 5% CO2 in a non-tissue culture treated Petri-dish with BMDM media (RPMI supplemented with 10% FBS, 100 U/mL penicillin, 100 μmg/mL streptomycin, 1% sodium pyruvate, 25 mM HEPES buffer, 2mM L-glutamine, and 50 mM 2-mercaptoethanol). After 4 hours, the non-adherent cells were transferred to a new non-tissue culture treated Petri dish and incubated with BMDM media supplemented with 20 ng/mL M-CSF (PeproTech). On day 3 after plating, 10 mL of BMDM media supplemented with 20 ng/mL M-CSF was added to the dish. After 6 days, BMDMs were lifted with PBS+5mM EDTA and gentle scraping. Cell suspensions were seeded onto plates at a density of 250,000 cells/ml for population ELISA experiments or onto the microwell device at a density of 125,000 cells/mL.

METHOD DETAILS

In vitro treatments and quantification of secretion in population

BMDMs were plated and stimulated in BMDM media with 10 ng/mL M-CSF (PeproTech). Cells were plated at the appropriate density and allowed to adhere overnight. Cells were stimulated with LPS (Invivogen) alone, or co-stimulated with LPS and IL-10R blocking antibody (BioLegend, clone: 1B1.3a), or LPS and recombinant IL-10 (PeproTech), or LPS and IFNAR1 blocking antibody (Invitrogen, clone: MAR1–5A3) for the times indicated in the legends. Cell culture medium was collected at the end of the incubation period and was assayed by ELISA according to the manufacturer’s instructions. The antibody pairs used in the ELISAs were the same as those used for the microwell assay and can be found in the Key resources table.

KEY RESOURCES TABLE

REAGENT OR RESOURCE SOURCE IDENTIFIER
Antibodies
Anti-mouse CD210 (IL-10R) (for blocking) BioLegend Clone: 1B1.3a, Cat#: 112708; RRID: AB_313521
Anti-mouse IFNAR-1 (for blocking) Invitrogen Clone: MAR1–5A3, Cat#: 127304; RRID: AB_1089156
CD16/CD32 eBioscience Clone: 93, Cat#:14–0161-85; RRID: AB_467134
Phospho-p38 MAPK (Thr180/Tyr182) (3D7) Rabbit mAb - Alexa Flour 488 Cell Signaling Technology Cat#: 41768; RRID: AB_2799209
Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (D13.14.4E) Rabbit mAb Cell Signaling Technology Cat#: 4094; RRID: AB_10694057
Goat anti-Rabbit IgG (H+L), Alexa Fluor 488 Thermo Fisher Scientific Cat#: A-11034; RRID: AB_2576217
Chemicals, peptides, and recombinant proteins
sTNFR R&D Systems Cat#: 425-R1
Recombinant Murine IL-10 PeproTech Cat#: 210–10
Recombinant Murine M-CSF PeproTech Cat#: AF-315–02
LPS-EK (LPS from E. coli K12) InvivoGen Cat#: tlrl-peklps
BD Phosflow Fix Buffer I BD Biosciences Cat#: 557870; RRID: AB_2869102
BD Phosflow Perm Buffer III BD Biosciences Cat#: 558050; RRID: AB_2869118
Critical commercial assays
TNF-α mouse Thermo Fisher Scientific Cat#: 88–7324-88; RRID: AB_2575080
IL-10 mouse R&D Systems Cat#: DY417
IL-6 mouse R&D Systems Cat#: DY406
IL-27 mouse R&D Systems Cat#: DY1834
CXCL1 mouse R&D Systems Cat#: DY453
CCL2 mouse R&D Systems Cat#: DY479
CCL3 mouse R&D Systems Cat#: DY450
CCL5 mouse R&D Systems Cat#: DY478
GM-CSF mouse R&D Systems Cat#: DY415
Chi3l3 mouse R&D Systems Cat#: DY2446
IFN-β mouse R&D Systems Cat#: DY8234
IL-12p40 mouse BD Biosciences Cat#: 555165; RRID: AB_2869032
Experimental models: Organisms/strains
C57BL/6J Jackson Laboratory Stock#: 000664; RRID: IMSR_JAX:000664
Software and algorithms
GraphPad Prism GraphPad Software https://www.graphpad.com/; RRID: SCR_002798
MATLAB MathWorks https://www.mathworks.com/products/matlab.html; RRID: SCR_001622
FlowJo FlowJo LLC https://www.flowjo.com/; RRID: SCR_008520
PHATE Moon et al., 2019 https://github.com/KrishnaswamyLab/PHATE
DREVI Krishnaswamy et al., 2014 https://dpeerlab.github.io/dpeerlab-website/dremi.html
Single-Cell Secretion Profiling Lu et al., 2013 https://github.com/Miller-JensenLab/
Single-Cell-Analysis
CellProfiler Broad Institute https://cellprofiler.org/; RRID: SCR_007358
ImageJ NIH https://imagej.nih.gov/ij/; RRID: SCR_003070
ZEN ZEISS https://www.zeiss.com/microscopy/en_us/products/microscope-software/zen.html; RRID: SCR_013672
Other
Attune NxT Flow Cytometer Thermo Fisher Scientific Cat#: A24858
ZEISS Axio Observer ZEISS https://www.zeiss.com/microscopy/us/products/light-microscopes/axio-observer-for-biology.html
Genepix 4200A Molecular Devices https://www.moleculardevices.com/products/additional-products/genepix-microarray-systems-scanners
Open in a new tab

Microwell assay for single-cell secretion profiling

Single-cell secretion profiling experiments were performed as previously described (Lu et al., 2013; Xue et al., 2015). Briefly, capture antibodies (Key resources table) were flow patterned onto epoxy silane-coated glass slides (Super-Chip; ThermoFisher). The polydimethylsiloxane (PDMS) microwell arrays and antibody barcode glass slides were blocked using complete RPMI. BMDMs were suspended in complete RPMI supplemented with 10 ng/mL M-CSF and were added to the PDMS microwell array and allowed to adhere overnight. The next day BMDMs were stimulated as indicated with complete RPMI supplemented with 125 nM of live cell marker (Calcein AM; ThermoFisher) to allow automatic live cell detection. The BMDMs in the PDMS microwell array were then covered with the antibody barcode slide, secured with plates and screws, and allowed to incubate for 8 hours. At the end of the in-cubation period, the device was imaged with an automated inverted microscope (Axio Observer; ZEISS) to record well position and cell locations. The device was then disassembled, and the sandwich immunoassay was performed: The glass slide was incubated with a mixture of detection antibodies (Key resources table) for 2 hours, followed by incubation with 20 mg/mL streptavidin-APC (eBio-science) for 30 minutes, rinsed with PBS and deionized water, and scanned with a Genepix 4200A scanner (Molecular Devices).

Multidimensional PHATE analysis

To further visualize high-dimensional secretion data from the microwell assay, the dimensionality reduction algorithm known as potential of heat diffusion for affinity-based transition embedding (PHATE) was used. PHATE analysis was performed using the available MATLAB and Python packages from the Krishnaswamy lab (https://github.com/KrishnaswamyLab/PHATE). Standard parameters were used to make 2-dimensional PHATE projections from the 9 and 10-dimensional BMDM datasets Manual adjustments to the t parameter were made for optimal visualization. PHATE plots were colored by kernel density estimation (KDE) using Gaussian kernels, or by relative secretion levels of each protein as indicated; cells below the secretion threshold were greyed out for visualization. Extracted PHATE parameters were analyzed using custom software written in Python. KDE plots of the PHATE 1 coordinates were plotted using Seaborn. We fitted a Gaussian mixture model with three components to the distribution of PHATE 1 coordinates. We chose 3 components based on minimization of the Akaike information criteria (AIC) and Bayesian information criteria (BIC) scores.

Recombinant protein standard curves

To convert measured fluorescent intensities from the microwell assay to concentrations of cytokines and chemokines, we used recombinant protein calibration curves Figures S3 (Figure S4A). Recombinant protein standard curves were derived by measuring the intensity values of recombinant proteins of concentrations between 39 and 5000 pg/mL in the microwell assay. The 4 Parameter Logistic (4PL) nonlinear regression model was used to fit the standard curves, and the 95% confidence intervals were calculated with Prism8 (GraphPad). Concentration values for intensities less than the detection limit of the calibration curve were set to zero or set to 1 for histogram visualization.

Conditional density rescaled visualization (DREVI)

For further analysis, interpolated concentration data from the microwell assay was analyzed using conditional density re-scaled visu-alization (DREVI). The DREVI software was used to visualize 2D stochastic relationships in single-cell data using conditional density estimation, representing how a variable X influences a variable Y, by depicting the distribution of Y for each value of X. DREVI algo-rithm was downloaded from the Pe’er Lab website (https://dpeerlab.github.io/dpeerlab-website/dremi.html).

Flow Cytometry

Plated BMDMs were stimulated for 15 minutes with LPS at varying doses (0, 0.1, 1, 10, 100, 1000 ng/mL). Afterward, BMDMs were lifted in ice-cold PBS+EDTA with gentle scraping and immediately fixed with Phosflow Fix buffer (BD Biosciences) for 10 minutes at 37°C. Cells were then transferred to a 96 well u-bottom plate. After washing, cells were permeabilized with Phosflow Perm Buffer III (BD Biosciences). For intracellular phospho-protein staining, cells were blocked with Fc receptor antibody (eBioscience, CD16/32,1:200 dilution) on ice for 15 mins in FACS buffer (PBS + 2% FBS). Cells were then stained in 50 uL for 1 hour at room temperature with anti-Phospho-p38 MAPK (Thr180/Tyr182, 3D7, Alexa Fluor 488 Conjugate) Rabbit mAb (Cell Signaling Technology #41768, 1:50 dilution) or anti-Phospho-p44/42 MAPK (Erk1/2, Thr202/Tyr204, D13.14.4E) Rabbit mAb (Cell Signaling Technology #4370) as a primary followed by staining in 50 uL using Goat anti-Rabbit IgG (H+L, Alexa Fluor 488) as a secondary antibody. All samples were acquired on an Attune NxT Flow Cytometer, and analyzed with FlowJo (FlowJO, LLC).

QUANTIFICATION AND STATISTICAL ANALYSIS

Single-cell secretion profiling and data processing

Device images were analyzed using a custom script in MATLAB (MathWorks) to automatically detect well location and number of cells per well, extract all signals from each well, and process the data (https://github.com/Miller-JensenLab/Single-Cell-Analysis). In brief, after automatic well and live cell detection, signal image registration, and manual curation, the software automatically extracted the intensity signal from each antibody for all the microwells in the device. This signal across the chip for each individual anti-body was normalized by subtracting a moving Gaussian curve fitted to the local zero-cell well intensity values. A secretion threshold for each antibody was set at the 99th percentile of all normalized zero-cell wells. Data was transformed using the inverse hyperbolic sine with cofactor set at 0.8× secretion threshold.

Statistics

Data were presented as means ± SEM unless otherwise specified. Statistical analysis was performed by ordinary 2-way ANOVA and the Dunnett method for correction of multiple comparisons as specified in the figure legends. All analyses were performed using Prism 8.4.1 software (GraphPad). For single-cell distributions, statistics were performed using a bootstrapping procedure to calculate the confidence intervals associated with sampling error in single-cell data. To obtain confidence intervals through bootstrapping, the single-cell datasets for each condition were sampled 10,000 times with replacement, and the metric of interest was calculated for each resampled dataset. We then calculated a 95% confidence interval for these resampled datasets, and statistical significance was assigned to pairwise comparisons with non-overlapping confidence intervals. This bootstrapping procedure was done using custom scripts in MATLAB and Python.

Supplementary Material

1
2

Highlights.

  • TLR4 activation in macrophage populations is resolved by IL-10 negative feedback

  • IL-10 secretion is decoupled from TLR4 proinflammatory activation in isolated cells

  • IL-10 activity contributes to a threshold that restrains TLR4-induced activation

  • IL-10’s role in restraining TLR4 activation is largely mediated by IFN-β signaling

ACKNOWLEDGMENTS

We thank Andre Levchenko, Smita Krishnaswamy, and David Hafler, as well as members of the Miller-Jensen lab, for insightful advice and helpful discussion. This work was supported by the National Institutes of Health (R01-GM123011 to K.M.-J.) and the National Science Foundation (DGE1122492 to A.F.A.).

Footnotes

DECLARATION OF INTERESTS

The authors declare no competing interests.

INCLUSION AND DIVERSITY

One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in science.

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2021.109728.

REFERENCES

  1. Caldwell AB, Cheng Z, Vargas JD, Birnbaum HA, and Hoffmann A. (2014). Network dynamics determine the autocrine and paracrine signaling functions of TNF. Genes Dev. 28, 2120–2133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Chang EY, Guo B, Doyle SE, and Cheng G. (2007). Cutting edge: involvement of the type I IFN production and signaling pathway in lipopolysaccharide induced IL-10 production. J. Immunol. 178, 6705–6709. [DOI] [PubMed] [Google Scholar]
  3. Conaway EA, de Oliveira DC, McInnis CM, Snapper SB, and Horwitz BH (2017). Inhibition of Inflammatory Gene Transcription by IL-10 Is Associated with Rapid Suppression of Lipopolysaccharide-Induced Enhancer Acti-vation. J. Immunol. 198, 2906–2915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Dagvadorj J, Naiki Y, Tumurkhuu G, Hassan F, Islam S, Koide N, Mori I, Yoshida T, and Yokochi T. (2008). Interleukin-10 inhibits tumor necrosis factor-alpha production in lipopolysaccharide-stimulated RAW 264.7 cells through reduced MyD88 expression. Innate Immun. 14, 109–115. [DOI] [PubMed] [Google Scholar]
  5. Eberwine J, and Kim J. (2015). Cellular Deconstruction: Finding Meaning in Individual Cell Variation. Trends Cell Biol. 25, 569–578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Fitzgerald DC, Fonseca-Kelly Z, Cullimore ML, Safabakhsh P, Saris CJM, Zhang G-X, and Rostami A. (2013). Independent and interdependent immunoregulatory effects of IL-27, IFN-β, and IL-10 in the suppression of human Th17 cells and murine experimental autoimmune encephalomyelitis. J. Immunol. 190, 3225–3234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Gottschalk RA, Martins AJ, Angermann BR, Dutta B, Ng CE, Uder-hardt S, Tsang JS, Fraser IDC, Meier-Schellersheim M, and Germain RN (2016). Distinct NF-kB and MAPK Activation Thresholds Uncouple Steady-State Microbe Sensing from Anti-pathogen Inflammatory Responses. Cell Syst. 2, 378–390. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Gottschalk RA, Dorrington MG, Dutta B, Krauss KS, Martins AJ, Uderhardt S, Chan W, Tsang JS, Torabi-Parizi P, Fraser ID, and Ger-main RN (2019). IFN-mediated negative feedback supports bacteria class-specific macrophage inflammatory responses. eLife 8, e46836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Howes A, Taubert C, Blankley S, Spink N, Wu X, Graham CM, Zhao J, Saraiva M, Ricciardi-Castagnoli P, Bancroft GJ, and O’Garra A. (2016). Differential Production of Type I IFN Determines the Reciprocal Levels of IL-10 and Proinflammatory Cytokines Produced by C57BL/6 and BALB/c Macro-phages. J. Immunol. 197, 2838–2853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Hu X, Chakravarty SD, and Ivashkiv LB (2008). Regulation of IFN and TLR Signaling During Macrophage Activation by Opposing Feedforward and Feed-back Inhibition Mechanisms. Immunol. Rev. 226, 41–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Iyer SS, and Cheng G. (2012). Role of interleukin 10 transcriptional regula-tion in inflammation and autoimmune disease. Crit. Rev. Immunol. 32, 23–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Iyer SS, Ghaffari AA, and Cheng G. (2010). Lipopolysaccharide-mediated IL-10 transcriptional regulation requires sequential induction of type I IFNs and IL-27 in macrophages. J. Immunol. 185, 6599–6607. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Krishnaswamy S, Spitzer MH, Mingueneau M, Bendall SC, Litvin O, Stone E, Pe’er D, and Nolan GP (2014). Systems biology. Conditional density-based analysis of T cell signaling in single-cell data. Science 346, 1250689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Lee TK, Denny EM, Sanghvi JC, Gaston JE, Maynard ND, Hughey JJ, and Covert MW (2009). A noisy paracrine signal determines the cellular NF-kappaB response to lipopolysaccharide. Sci. Signal. 2, ra65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Lu Y, Chen JJ, Mu L, Xue Q, Wu Y, Wu P-H, Li J, Vortmeyer AO,Miller-Jensen K, Wirtz D, and Fan R. (2013). High-throughput secretomic analysis of single cells to assess functional cellular heterogeneity. Anal. Chem. 85, 2548–2556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lu Y, Xue Q, Eisele MR, Sulistijo ES, Brower K, Han L, Amir AD, Pe’er D, Miller-Jensen K, and Fan R. (2015). Highly multiplexed profiling of single-cell effector functions reveals deep functional heterogeneity in response to pathogenic ligands. Proc. Natl. Acad. Sci. USA 112, E607–E615. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. McNab FW, Ewbank J, Howes A, Moreira-Teixeira L, Martirosyan A, Ghilardi N, Saraiva M, and O’Garra A. (2014). Type I IFN induces IL-10 pro-duction in an IL-27-independent manner and blocks responsiveness to IFN-g for production of IL-12 and bacterial killing in Mycobacterium tuberculosis-in-fected macrophages. J. Immunol. 193, 3600–3612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Medzhitov R. (2008). Origin and physiological roles of inflammation. Nature 454, 428–435. [DOI] [PubMed] [Google Scholar]
  19. Mittal SK, and Roche PA (2015). Suppression of antigen presentation by IL-10. Curr. Opin. Immunol. 34, 22–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Moon KR, van Dijk D, Wang Z, Gigante S, Burkhardt DB, Chen WS, Yim K, Elzen AVD, Hirn MJ, Coifman RR, et al. (2019). Visualizing struc-ture and transitions in high-dimensional biological data. Nat. Biotechnol. 37, 1482–1492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Muldoon JJ, Chuang Y, Bagheri N, and Leonard JN (2020). Macro-phages employ quorum licensing to regulate collective activation. Nat. Com-mun. 11, 878. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Murray PJ (2005). The primary mechanism of the IL-10-regulated antiinflam-matory response is to selectively inhibit transcription. Proc. Natl. Acad. Sci. USA 102, 8686–8691. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Murray PJ, and Smale ST (2012). Restraint of inflammatory signaling by interdependent strata of negative regulatory pathways. Nat. Immunol. 13, 916–924. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. O’Garra A, Barrat FJ, Castro AG, Vicari A, and Hawrylowicz C. (2008). Strategies for use of IL-10 or its antagonists in human disease. Immunol. Rev. 223, 114–131. [DOI] [PubMed] [Google Scholar]
  25. Ramji R, Alexander AF, Muñ oz-Rojas AR, Kellman LN, and Miller-Jen-sen K. (2019). Microfluidic platform enables live-cell imaging of signaling and transcription combined with multiplexed secretion measurements in the same single cells. Integr. Biol. 11, 142–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Rand U, Rinas M, Schwerk J, Nö hren G, Linnes M, Krö ger A, Flossdorf M, Ká ly-Kullai K, Hauser H, Hö fer T, and Kö ster M. (2012). Multi-layered stochasticity and paracrine signal propagation shape the type-I interferon response. Mol. Syst. Biol. 8, 584. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Saxena A, Khosraviani S, Noel S, Mohan D, Donner T, and Hamad ARA (2015). Interleukin-10 paradox: A potent immunoregulatory cytokine that has been difficult to harness for immunotherapy. Cytokine 74, 27–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Shalek AK, Satija R, Adiconis X, Gertner RS, Gaublomme JT, Ray-chowdhury R, Schwartz S, Yosef N, Malboeuf C, Lu D, et al. (2013). Single-cell transcriptomics reveals bimodality in expression and splicing in immune cells. Nature 498, 236–240. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Shalek AK, Satija R, Shuga J, Trombetta JJ, Gennert D, Lu D, Chen P, Gertner RS, Gaublomme JT, Yosef N, et al. (2014). Single-cell RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510, 363–369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Spits H, and de Waal Malefyt R. (1992). Functional characterization of human IL-10. Int. Arch. Allergy Immunol. 99, 8–15. [DOI] [PubMed] [Google Scholar]
  31. Trouplin V, Boucherit N, Gorvel L, Conti F, Mottola G, and Ghigo E. (2013). Bone Marrow-derived Macrophage Production. J. Vis. Exp. 81, e50966. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Xue Q, Lu Y, Eisele MR, Sulistijo ES, Khan N, Fan R, and Miller-Jen-sen K. (2015). Analysis of single-cell cytokine secretion reveals a role for paracrine signaling in coordinating macrophage responses to TLR4 stimulation. Sci. Signal. 8, ra59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zhang J, Tian X-J, Zhang H, Teng Y, Li R, Bai F, Elankumaran S, and Xing J. (2014). TGF-b-induced epithelial-to-mesenchymal transition proceeds through stepwise activation of multiple feedback loops. Sci. Signal. 7, ra91. [DOI] [PubMed] [Google Scholar]
  34. Zhao M, Zhang J, Phatnani H, Scheu S, and Maniatis T. (2012). Stochas-tic expression of the interferon-bgene. PLoS Biol. 10, e1001249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Zhou X, Franklin RA, Adler M, Jacox JB, Bailis W, Shyer JA, Flavell RA, Mayo A, Alon U, and Medzhitov R. (2018). Circuit Design Features of a Stable Two-Cell System. Cell 172, 744–757.e17. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1742916-supplement-1.pdf (5.6MB, pdf)
2
NIHMS1742916-supplement-2.pdf (10.7MB, pdf)

Data Availability Statement

  • All data reported in this paper will be shared by the lead contact upon request.

  • All original code for the single-cell secretion profiling analysis and data processing are publicly available at https://github.com/Miller-JensenLab/Single-Cell-Analysis as of the date of publication.

  • Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact, Kathryn Miller-Jensen (kathryn.miller-jensen@yale.edu) upon request.

RESOURCES