responses were first recognized for their role in antiviral immunity, but it is now widely appreciated that IFN-Is have many immunomodulatory functions, influencing antitumor responses, autoimmune manifestations, and antimicrobial defenses. Given these pivotal roles, it may be surprising that multilayered stochastic events create highly heterogeneous, but tightly regulated, all-or-nothing cellular decisions. Recently, mathematical models have provided crucial insights into the stochastic nature of antiviral IFN-I responses, which we critically evaluate in this review. In this context, we emphasize the need for innovative single-cell technologies combined with mathematical models to further reveal, understand, and predict the complexity of the IFN-I system in physiological and pathological conditions that may be relevant to a plethora of diseases. From population averages to single-cell resolutionProviding a robust first line of host defense, type I interferon (IFN-I; see Glossary) responses were first recognized for their role in antiviral immunity by Isaacs and Lindemann [1]. In addition, IFN-Is are known to have numerous immunomodulatory functions that go beyond the scope of antiviral immunity, with potent roles in antitumor responses, autoimmune diseases, and microbial infections [2][3][4]. The expression and regulation of IFN-I has been extensively studied in mammals, and important molecular regulators have been characterized (Box 1;). Population-averaged measurements have generated not only a wealth of knowledge on IFN-I responses and regulation, but also contradictory results that remain hard to interpret or understand [7]. Over the past few years, an overwhelming wave of single-cell data has changed the dogma on how systemic immune responses are generated, challenging the traditional assumption that all cells of a given lineage display nearly identical behaviors upon stimulation. HighlightsThe rise of single-cell technologies has highlighted a massive degree of cellular heterogeneity during mammalian interferon (IFN)-I responses.

Plasmacytoid dendritic cells (pDCs) are a rare type of highly versatile immune cells that besides their specialized function of massive type I interferon (IFN-I) production are able to exert cytotoxic effector functions. However, diversification upon toll like receptor (TLR)-induced activation leads to highly heterogeneous responses that have not been fully characterized yet. Using droplet-based microfluidics, we showed that upon TLR7/8 and TLR9-induced single-cell activation only 1-3% secretes IFNα, and only small fractions upregulate cytotoxicity markers. Interestingly, this 1-3% of early IFN-producing pDCs, also known as first responders, express high levels of programmed death-ligand 1 (PD-L1) and TNF-related apoptosis-inducing ligand (TRAIL), which makes these hybrid cells similar to earlier described IFN-I producing killer pDCs (IKpDCs). IFN-I priming increases the numbers of IFNα producing cells up to 40%, but does not significantly upregulate the cytotoxicity markers. Besides, these so-called second responders do not show a cytotoxic phenotype as potent as observed for the first responders. Overall, our results indicate that the first responders are the key drivers orchestrating population wide IFN-I responses and possess high cytotoxic potential.

Cytotoxicity is a vital effector mechanism used by immune cells to combat pathogens and cancer cells. While conventional cytotoxicity assays rely on averaged end-point measures, crucial insights on the dynamics and heterogeneity of effector and target cell interactions cannot be extracted, emphasizing the need for dynamic single-cell analysis. Here, we present a fully automated droplet-based microfluidic platform that allowed the real-time monitoring of effector-target cell interactions and killing, allowing the screening of over 60,000 droplets identifying 2000 individual cellular interactions monitored over 10 h. During the course of incubation, we observed that the dynamics of cytotoxicity within the Natural Killer (NK) cell population varies significantly over the time. Around 20% of the total NK cells in droplets showed positive cytotoxicity against paired K562 cells, most of which was exhibited within first 4 h of cellular interaction. Using our single cell analysis platform, we demonstrated that the population of NK cells is composed of individual cells with different strength in their effector functions, a behavior masked in conventional studies. Moreover, the versatility of our platform will allow the dynamic and resolved study of interactions between immune cell types and the finding and characterization of functional sub-populations, opening novel ways towards both fundamental and translational research.

Moving from the optimalization of single-cell technologies to the interpretation of the multi-complex single-cell data, the field of immunoengineering is granted with numerous important insights into the coordination of immune cell activation and how to modulate it for therapeutic purposes. However, insights come with additional follow-up questions that challenge our perception on how immune responses are generated and fine-tuned to fight a wide array of pathogens in ever-changing and often unpredictable microenvironments. Are immune responses really either being tightly regulated by molecular determinants, or highly flexible attributed to stochasticity? What exactly makes up the basic rules by which single cells cooperate to establish tissue-level immunity? Taking the type I IFN system and its newest insights as a main example throughout this review, we revise the basic concepts of (single) immune cell coordination, redefine the concepts of noise, stochasticity and determinism, and highlight the importance of single-cell variation in immunology and beyond.

Type I Interferon (IFN-I)-mediated antiviral responses are central to host defense against viral infections. Crucial is the tight and well-orchestrated control of cellular decision-making leading to the production of IFN-Is. Innovative single-cell approaches revealed that the initiation of IFN-I production is limited to only fractions of 1-3% of the total population, both found in vitro, in vivo, and across cell types, which were thought to be stochastically regulated. To challenge this dogma, we addressed the influence of various stochastic and deterministic host-intrinsic factors on dictating early IFN-I responses, using a murine fibroblast reporter model. Epigenetic drugs influenced the percentage of responding cells. Next, with the classical Luria-Delbrück fluctuation test, we provided evidence for transient heritability driving responder fates, which was verified with mathematical modeling. Finally, while studying varying cell-densities, we substantiated an important role for cell density in dictating responsiveness, similar to the phenomenon of quorum sensing. Together, this systems immunology approach opens up new avenues to progress the fundamental understanding on cellular decision-making during early IFN-I responses, which can be translated to other (immune) signaling systems.

Type I Interferon (IFN-I)-mediated antiviral responses are central to host defense against viral infections. Crucial is the tight and well-orchestrated control of cellular decision-making leading to the production of IFN-Is. Innovative single-cell approaches revealed that the initiation of IFN-I production is only limited to a fraction of 1-3% of the total population, both found in vitro and in vivo, which were thought to be stochastically regulated. To challenge this dogma, we addressed the influence of various host-intrinsic factors, both stochastic and epigenetic, on dictating early IFN-I responses. Hypomethylating drugs increased the percentage of responding cells. Next, with the classical Luria-Delbrück fluctuation test, we provided evidence that the fate of becoming a responding cell is transiently heritable. Finally, while studying varying cell-densities, we substantiated an important role for quorum sensing, which was verified by mathematical modeling. Together, this systems immunology approach opens up new avenues to progress the fundamental understanding of cellular decision-making during early IFN-I responses, which can be translated to other (immune) signaling systems, and ultimately will improve IFN-I based immune therapies.

Increasing evidence suggest that Natural killer (NK) cells are composed of distinct functional subsets. This multi-functional role displayed by NK cells have made them an attractive choice for anti-cancer immunotherapy. A functional NK cell repertoire is generated through cellular education, resulting in heterogeneous NK cell population with distinct capabilities to respond to different stimuli. The application of a high-throughput droplet-based microfluidic platform allows monitoring of NK cell-target cell interactions at single-cell level and in real-time.Through fluorescence-based screening of around 80,000 droplets, with different Effector:Target ratios, a fully automated image analysis allows for the assessment of individual killing events in each droplet over time. We observed a variable response of single NK cells towards different target cells and identified a distinct population of NK cells capable of inducing multiple target lysis, coined as serial killers. To meet the increasing clinical demand for NK cells several sources, such as umbilical cord blood (UCB), have successfully been explored. By assessing the cytotoxic dynamics, we showed that single UCB-derived CD34+ hematopoietic progenitor (HPC)-NK cells display superior anti-tumor cytotoxicity.Additionally, with an integrated analysis of cytotoxicity and cytokine secretion we showed that target cell interactions augmented cytotoxic as well as secretory behavior of NK cells. By providing an in-depth assessment over NK cell functions, this study provides crucial information on diversity and functional characteristics of peripheral blood NK cells and ex vivo-generated HPC-NK cells to develop and improve of NK cell-based cancer immunotherapy.

Type-I interferon (IFN-I) subtypes signal through the same IFNalpha receptor (IFNAR), and initiate temporal STAT1/2 activation to orchestrate innate and adaptive immunity. It remains unknown how IFNAR discriminates between subtypes (e.g., IFNalpha and IFNbeta), and how STAT1/2 signaling is affected by time-varying inputs. Here, we utilize our microfluidic system and live-cell imaging to quantify STAT1/2 activation dynamics in a reporter fibroblast model. Population-averaged and single-cell analyses reveal distinct STAT1/2 responses to various IFNalpha and IFNbeta inputs. Upon continuous stimulation, cells show less sensitivity but more sustained responses to IFNalpha over IFNbeta. A short IFNalpha pulse induces nearly homogeneous STAT1/2 dynamics, in contrast to heterogeneous responses in IFNbeta-pulsed cells. Distinct STAT1/2 refractory states emerge upon exposure to repeated IFN-I pulses, while alternating pulse stimulation reveals that IFNbeta can revoke STAT1/2 refractoriness caused by IFNalpha, but not vice versa. These findings highlight the differences between IFNalpha and IFNbeta signaling and how they can elicit distinct temporal cellular behaviors during viral infection.