Highlights
- •Timing of granuloma formation influences local microenvironment and bacterial burden
- •Mast cells, type 2 immunity, and tissue remodeling underlie early, high-burden granulomas
- •Type1-type17 and cytotoxic T cells associate with late-forming, low-burden granulomas
- •Distinct interaction circuits across granuloma phenotypes nominate therapeutic targets
Summary
Mycobacterium tuberculosis lung infection results in a complex multicellular structure: the granuloma. In some granulomas, immune activity promotes bacterial clearance, but in others, bacteria persist and grow. We identified correlates of bacterial control in cynomolgus macaque lung granulomas by co-registering longitudinal positron emission tomography and computed tomography imaging, single-cell RNA sequencing, and measures of bacterial clearance. Bacterial persistence occurred in granulomas enriched for mast, endothelial, fibroblast, and plasma cells, signaling amongst themselves via type 2 immunity and wound-healing pathways. Granulomas that drove bacterial control were characterized by cellular ecosystems enriched for type 1-type 17, stem-like, and cytotoxic T cells engaged in pro-inflammatory signaling networks involving diverse cell populations. Granulomas that arose later in infection displayed functional characteristics of restrictive granulomas and were more capable of killing Mtb. Our results define the complex multicellular ecosystems underlying (lack of) granuloma resolution and highlight host immune targets that can be leveraged to develop new vaccine and therapeutic strategies for TB.
Introduction
Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains a major global health threat (WHO, 2019). Mtb infection is characterized by the formation of granulomas predominantly in the lungs and lymph nodes (Flynn and Klein, 2011; Lin et al., 2014b; Russell et al., 2010; Ulrichs and Kaufmann, 2006). These spatially organized structures, composed of a mixture of immune and non-immune cells (Ehlers and Schaible, 2013; Flynn and Klein, 2011; Gideon et al., 2019; Lin et al., 2006; Mattila et al., 2013; Pagan and Ramakrishnan, 2014; Phuah et al., 2012; Reece and Kaufmann, 2012; Ulrichs and Kaufmann, 2006), are key sites of host-pathogen interactions that can either restrict or facilitate bacterial survival. Delineating protective responses in humans has been challenging given the limited accessibility of affected lung tissue and difficulty determining the true extent of bacterial control. The cynomolgus macaque model of Mtb infection recapitulates the diversity of human outcomes and granuloma pathologies and enables detailed studies of the features of immunologic success and failure in Mtb granulomas (Canetti, 1955; Flynn and Klein, 2011; Lin et al., 2006).A spectrum of granuloma types, organization, and cellular composition has been described in both humans and non-human primates (NHPs) (Canetti, 1955; Flynn and Klein, 2011; Hunter, 2011, Hunter, 2016; Lin et al., 2006). The bacterial burden in individual granulomas is highest early in infection and then decreases due to increased bacterial killing as the immune response matures, even in macaques that ultimately develop active TB (Cadena et al., 2016; Lin et al., 2014b; Maiello et al., 2018). Strikingly, however, individual granulomas within a single host follow independent trajectories with respect to inflammation, cellular composition, reactivation risk, and ability to kill Mtb (Coleman et al., 2014b; Gideon et al., 2015; Lenaerts et al., 2015; Lin et al., 2013, Lin et al., 2014b; Malherbe et al., 2016; Martin et al., 2017). We and others have profiled immune responses among individual cell types in macaque lung granulomas, including those of T cells (Diedrich et al., 2020; Foreman et al., 2016; Gideon et al., 2015; Lin et al., 2012; Mattila et al., 2011; Wong et al., 2018), macrophages (Mattila et al., 2013), B cells (Phuah et al., 2012, Phuah et al., 2016), and neutrophils (Gideon et al., 2019; Mattila et al., 2015) and have also examined the instructive roles of cytokines, including interferon (IFN)-γ, interleukin (IL)-2, tumor necrosis factor (TNF), IL-17, and IL-10 (Gideon et al., 2015; Lin et al., 2010; Wong et al., 2020). Although these analyses have led to insights into how specific canonical cell types and effector molecules relate to bacterial burden, they have not yet revealed how the integrated actions of diverse cell types within individual granulomas influence control.High-throughput single-cell genomic profiling methods afford new opportunities to define the cell types, phenotypic states, and intercellular circuits that comprise granulomas and inform their dynamics (Prakadan et al., 2017). Here, we developed and applied a multifactorial profiling pipeline—integrating longitudinal positron emission tomography and computed tomography (PET-CT) imaging, single-cell RNA sequencing (scRNA-seq), and molecular measures of bacterial killing with immunohistochemistry and flow cytometry—to identify features of TB lung granulomas that correlate with bacterial clearance in cynomolgus macaques. We defined the cellular compositions and cell-cell signaling networks associated with bacterial persistence or control. Collectively, our data define the cellular ecosystems within TB lung granulomas in which Mtb is controlled or alternatively survives and multiplies, uncovering therapeutic and prophylactic targets for future investigation.
Results
Profiling longitudinal TB granuloma dynamics, bacterial burden, and bacterial killing
We sought to define the complex cellular ecosystems of granulomas that manifest different degrees of bacterial control in NHPs. Four cynomolgus macaques were infected with a low dose of Mtb (<10 CFU; Erdman strain) and followed for 10 weeks (Figure 1A ). Ten weeks post-infection (p.i.) was chosen as a pivotal time point at which bacterial killing could be identified in some but not all granulomas during the course of immune activation and mobilization, even in macaques that would eventually progress to active TB (Figures S1A–S1C). Progression of Mtb infection and individual granuloma dynamics were monitored at 4, 8, and 10 weeks p.i. by using PET-CT imaging of FDG avidity as a proxy for inflammation (Figures S1D and S1E; Table S1) (Coleman et al., 2014b; White et al., 2017). At necropsy, individual PET-CT identified lung granulomas were excised and dissociated to obtain a single-cell suspension; viable bacterial burden (CFU, colony forming units—i.e., culturable live bacterial burden) and cumulative (live + dead) bacterial load (chromosomal equivalents, CEQ) were measured to define the extent of bacterial growth and killing in each granuloma (Lin et al., 2014b; Munoz-Elias et al., 2005).
