key: cord-0309567-ljt9pea4 authors: Stevens, Joseph; Steinmeyer, Shelby; Bonfield, Madeline; Wang, Timothy; Gray, Jerilyn; Lewkowich, Ian; Xu, Yan; Du, Yina; Guo, Minzhe; Wynn, James L.; Zacharias, William; Salomonis, Nathan; Miller, Lisa; Chougnet, Claire; O’Connor, Dennis Hartigan; Deshmukh, Hitesh title: Balance between protective and pathogenic immune responses to pneumonia in the neonatal lung enforced by gut microbiota date: 2021-09-27 journal: bioRxiv DOI: 10.1101/2021.09.27.461705 sha: 2c5f3f6666d295d0fe415db926f810caa715d0f8 doc_id: 309567 cord_uid: ljt9pea4 While modern clinical practices like cesarean sections and perinatal antibiotics have improved infant survival, treatment with broad-spectrum antibiotics alters intestinal microbiota and causes dysbiosis. Infants exposed to perinatal antibiotics have an increased likelihood of life-threatening infections, including pneumonia. Here, we investigated how gut microbiota sculpt pulmonary immune responses, promoting recovery and resolution of infection in newborn rhesus macaques. Early-life antibiotic exposure, mirroring current clinical practices, interrupted the maturation of intestinal commensal bacteria and disrupted the developmental trajectory of the pulmonary immune system as assessed by single-cell proteomic and transcriptomic analyses of the pulmonary immune response. Early-life antibiotic exposure rendered newborn macaques susceptible to bacterial pneumonia, mediated by profound changes in neutrophil senescence, inflammatory signaling, and macrophage dysfunction. Pathogenic reprogramming of pulmonary immunity was reflected by a hyperinflammatory signature in all pulmonary immune cell subsets. Distinct patterns of immunoparalysis, including dysregulated antigen presentation in alveolar macrophages, impaired costimulatory function in T helper cells, and dysfunctional cytotoxic responses in natural killer (NK) cells, were coupled with a global loss of tissue-protective, homeostatic pathways in lungs of dysbiotic newborns. Fecal microbiota transfer corrected the broad immune maladaptations and protected against severe pneumonia. These data demonstrate the importance of intestinal microbiota in programming pulmonary immunity. Gut microbiota promote balance between pathways driving tissue repair and inflammatory responses, thereby leading to clinical recovery from infection in infants. One sentence summary Gut microbiota promote clinical recovery by reinforcing the balance between regenerative pathways driving tissue homeostasis and inflammatory responses limiting pathogens in infected neonatal lungs. microbiota. Infants treated with antibiotics experience more significant morbidity in 23 response to sepsis and pneumonia, (13-17) compared to milder disease and spontaneous 24 resolution in infants not exposed to antibiotics. 25 In the present study, we define how gut microbiota sculpt immune responses that 27 mediate recovery and resolution rather than severe disease. Such an understanding is 28 vital to develop effective treatments for respiratory infections in this vulnerable 29 population. Non-human primate lung more closely resembles human infant lung in 30 structure, developmental stage, physiology, and mucosal immune mechanisms (18), 31 compared to the murine lungs (19) . In addition, like humans, non-human primates 32 develop lobar pneumonia and demonstrate a heterogeneous clinical response (20). Their 33 larger size and similarity to human newborns enable integration of clinical signs, 1 longitudinal assessment of disease progression, and the ability to distinguish mild, self-2 resolving pneumonia from severe, often fatal pneumonia. 3 4 Here, we define a tiered immunologic development program, anchored by stepwise 5 engagement of effector cells to limit tissue damage and promote recovery, that was 6 experimentally disrupted by early-life antibiotic exposure. Remodeling of the pulmonary 7 immune response was anchored by the appearance of a population of senescent, Antibiotic exposure during the first postnatal week delays microbiota maturation and 22 is associated with pro-inflammatory signatures in the peripheral blood: We treated a 23 cohort of vaginally delivered, nursery raised rhesus macaques (Supplemental Table 1 ) 24 with a cocktail of antimicrobials from postnatal days (PN) 1 to 7 or with saline (n = 4 in 25 each experimental group) (Fig. 1a) and profiled the fecal bacterial communities daily. 26 Stools in saline-treated newborn macaques, referred hereafter as control newborns, were 27 dominated by facultative aerobic Enterobacteriaceae during the first week of life before 28 changing to strict anaerobes, principally Bifidobacterium, Bacteroides, and Clostridium (Fig. 29 1b, Supplemental Table 2) in the second week of life, similar to the pattern seen in human 30 neonates (21, 22) . Dynamic changes in the abundance of peripheral blood neutrophils, 31 CD4 + and CD8 + T helper cells, NK cells, and migratory cytokines occurred 32 contemporaneously with changes in the stool microbiota ( Fig. 1c-e, Supplemental Fig. 33 dysbiotic newborns (Supplementary Table 6 ). Collectively, these data demonstrate that 23 disruption of commensal microbes by early-life antibiotic use causes a maladapted state 24 marked by both a robust, pro-inflammatory bias and neutrophil and T cell exhaustion. 25 26 Early-life antibiotic exposure is associated with clinically severe pneumonia. After 27 discontinuing antibiotics for one week, we challenged the newborn macaques in each 28 experimental group with S. pneumoniae (serotype 19F), a common and often fatal 29 respiratory pathogen in human infants (31). Vital signs, including heart rate, respiratory 30 effort, blood pressure, urine output, oxygen saturation, and overall clinical condition of 31 these newborn macaques, were monitored every 6 hours. In addition, chest radiographs 32 were obtained daily. The pediatric early warning score (PEWS) (32), an extensively 33 validated clinical decision-making and severity-scoring tool, was used to guide therapy, 1 including intravenous fluids and supplemental oxygen (Supplemental table 7) . Thus, we 2 provided these newborn macaques with intensive care comparable to what a human 3 infant with pneumonia would receive, with the exception of additional antibiotics after 4 challenge with S. pneumoniae. 5 6 The clinical status of dysbiotic newborns deteriorated, evidenced by higher peak PEWS 7 score and radiographic evidence of consolidation ( Fig. 1h and j) . The progression of 8 symptoms was also more rapid in the dysbiotic newborns (Fig. 1i) . The majority required 9 supportive therapy, an objective measure of clinical well-being. Almost all dysbiotic 10 newborns received supportive treatment by 60 hours (Fig. 1k, Supplemental table 7) , 11 recapitulating the rapid progression and increased morbidity seen in dysbiotic human 12 infants (33-35). These data together demonstrate that this newborn macaque model 13 mirrors a clinically relevant disease process and allows for profiling of the immune 14 response in target tissues, such as lungs, which is often impossible in human newborns. alone may not thoroughly explain the differential clinical response to pneumonia in 20 dysbiotic newborns. We, therefore, used a combination of single-cell RNA-sequencing 21 (scRNAseq) and high-parameter (>20 markers) CyTOF to then profile both innate and 22 adaptive immune cell responses in the lungs of control or dysbiotic newborns 60 hours. 23 after infection with S. pneumoniae. At this time, dysbiotic newborns had more severe 24 disease (PEWS = 8 ± 1.5) than control newborns (PEWS = 5 ± 0.6), despite similar 25 pathogen burden in the lungs (Supplemental Fig. 1g and h) . 26 27 Cell clusters identified in the pulmonary tissues after infection with S. pneumoniae in 28 control or dysbiotic newborns were annotated using signature genes from published 29 scRNAseq atlases (37) , (38) ( Fig. 2a and b, Supplemental Table 8 ). The proportion of 30 neutrophils identified by scRNAseq decreased in dysbiotic newborns, whereas the 31 T cell proportion increased ( Fig. 2c and d) . This mirrors the high-parameter CyTOF 32 cytometry where the frequency of neutrophils decreased approximately 40% in the lungs 1 of dysbiotic newborns, compared to control newborns after infection with S. pneumoniae 2 ( Fig. 2e and f, Supplemental Fig. 2a-c) . In contrast, the frequency of CD4 + T cells 3 expanded 2-fold in the lungs of dysbiotic newborns after infection with S. pneumoniae 4 ( Fig. 2e and f, Supplemental Fig. 2a-c) . Proportions of other immune cells, for example, 5 alveolar (AM) and interstitial (IM) macrophages, NK cells, B cells, and dendritic cells 6 (DC) subsets, were generally unchanged. 7 8 Canonical immune programs anchored by cell migration, differentiation, activation, and 9 tissue repair were broadly disrupted in the pulmonary immune cells of dysbiotic 10 newborns after infection (Fig. 2g, Supplemental Fig. 2d) . Conversely, immune programs 11 dominated by inflammation (NF-κB activity, IL-1β activation, and glycolysis) and 12 dysfunction (apoptosis, cellular stress, and ubiquitination) were active in all pulmonary 13 immune cells of dysbiotic newborns (Fig. 2g, Supplementary Table 9 ). Furthermore, 14 consistent with transcriptomic changes, cytokines associated with inflammation (CXCL8, 15 CXCL10, and TGFβ) and activation (IL6, IL8, and TNF-α) were increased. In contrast, 16 cytokines associated with tissue repair (GMCSF, IL10, and PDGF) were decreased in the 17 bronchial lavage fluid of dysbiotic newborns, with these cytokine levels correlating with 18 disease severity (Supplemental Fig. 2e and f, Supplemental Table 10) . 19 20 Neutrophils, AMs, and, to a lesser extent, IMs, were the principal sources of 21 inflammatory cytokines after infection with S. pneumoniae (Supplemental Fig. 2g) , 22 consistent with their critical roles in lung defense against pneumonia. Neutrophils and 23 AMs also exhibited the greatest transcriptomic remodeling (Fig. 2h) and most 24 differentially expressed transcripts related to inflammation and tissue repair (Fig. 2i , transcriptionally distinct neutrophil populations in the lungs after infection with S. 33 pneumoniae ( Fig. 3a and b, Supplemental Fig. 3a, Supplemental Table 11 ). Neutrophil 1 heterogeneity is influenced by developmentally-encoded cell programs and by 2 environment and pathogen-specific factors (45). To identify distinct neutrophil 3 populations, we used module scores (38), reflecting the average expression of all genes 4 related to neutrophil development, maturation, and activation using published gene 5 signatures from granulocytes during homeostasis and in the setting of sepsis (46) 6 (Supplemental Table 12 ). The population identified by expression of genes related to 7 'pathogen response' and 'cytokine signaling' reflected neutrophil heterogeneity caused 8 by pathogen exposure (Fig. 3b and c, Supplemental Fig. 3a) represents mature 9 neutrophils (cluster 1). Neutrophils characterized by high expression of CXCR4 and CD63 10 and reduced expression of CXCR2 and SELL ( Fig. 3b and c, Supplemental Fig. 3b ) 11 represent stressed, hyperinflammatory neutrophils (47) (cluster 2). These cells also had 12 increased gene transcripts associated with glycolysis, a pathway associated with 13 hyperinflammatory responses (48) (Supplemental Fig. 3b and c) . Immature neutrophils 14 (cluster 3) were identified based on expression of gene transcripts related to neutrophil 15 maturation (secretory vesicles, lysozymes, and phagocytosis) ( Fig. 3b and c, 16 Supplemental Fig. 3b) . The developmental relationship among these three clusters was 17 predicted by cellular trajectory analysis (49-52) (Fig. 3d) . Pseudotime analysis of granule 18 proteins, OLFM4 and SELL, and trafficking receptors, CXCR2 and CXCR4, known to be 19 involved in neutrophil maturation, supported a continuum of differentiation from 20 immature neutrophils (cluster 3) to mature neutrophils (cluster 1) to senescent, 21 hyperinflammatory neutrophils (cluster 2) (Fig. 3e) . 22 Early-life antibiotic exposure strongly influenced the activation state of the lung 24 neutrophil compartment after infection with S. pneumoniae. Pseudobulk RNAseq analysis 25 of these pulmonary neutrophils identified distinct signatures associated with dysbiosis 26 (Supplemental Fig. 3c) . Senescent, hyperinflammatory neutrophils were unique to the 27 lungs of dysbiotic newborns, while mature neutrophils were absent in the lungs of 28 dysbiotic newborns (Fig. 3f) . CyTOF then also showed a consistent remodeling of the 29 pulmonary neutrophil pool with the emergence of stressed, senescent (CXCR2 lo , CXCR4 hi , 30 CD62L lo ) neutrophils in the lungs of dysbiotic newborns after infection with S. pneumoniae 31 ( Fig. 3g and h) . 32 There was broad induction of NFκB (a pro-inflammatory cytokine-encoding gene 1 regulator), enrichment of phagocytosis and degranulation gene sets, and increased 2 expression of epigenetic regulators associated with inflammatory neutrophils, 3 including PADI4, which is required for NETosis (53, 54) and CD274 (encoding for PD-L1 4 (55)), a marker of cell exhaustion in pulmonary neutrophils from dysbiotic newborns 5 (Supplemental Table 13 ). We noted consistent changes in neutrophil activating and 6 chemotactic cytokines in the bronchial washings of dysbiotic newborns after infection 7 with S. pneumoniae (Supplemental Fig. 3d) . 8 9 Remodeled pulmonary neutrophil compartment is associated with increased 10 pneumonia-related morbidity in dysbiotic newborns. Dysfunctional neutrophils 11 contribute to pulmonary damage in experimental models of acute lung injury (56, 57) , (58). 12 Conversely, neutrophil depletion is protective in several pneumonia models, and severe 13 disease is frequently associated with an increased pulmonary neutrophil influx. We 14 therefore hypothesized that these dysfunctional, senescent neutrophils contributed to 15 severe pneumonia in dysbiotic newborns. We identified the five most differentially-16 expressed genes between senescent, hyperinflammatory neutrophils and all other cells in 17 our dataset: HIF1A, CXCR4, CD274, LTF, and S100A8. We then used publicly available 18 whole-blood bulk transcriptomic datasets of infants and children with severe sepsis and 19 pneumonia (59). We scored each sample in these datasets by the aggregated expression 20 of these five genes. We used these scores to construct a receiver operating characteristic 21 (ROC) curve using the gene score as a predictor and severity as the response variable. 22 Our senescent, hyperinflammatory neutrophil gene score predicted mortality and disease 23 severity (AUC = 0.79) in infants with severe sepsis and pneumonia (Fig. 3i, Predicted regulatory networks active in pulmonary neutrophils from dysbiotic 32 newborns after infection with S. pneumoniae. Regulatory networks anchored by the 33 transcription factors C/EBPγ (encoded by CEBPG) and Kruppel-like factor (KLF) 6, which 1 are essential for neutrophil development (60-65), were enriched in immature neutrophils 2 (cluster 3) (Fig. 3j, Supplemental Table 15 ). Regulatory networks anchored by defense 3 response-associated transcription factors (66-68), for example, NFKB, IRF7, STAT5A, 4 BATF3, and HIF1A, were enriched in activated neutrophils (cluster 1) ( Fig. 3j and k, 5 Supplemental Table 15 ). In contrast, regulatory networks anchored by epigenome 6 modifying enzymes (69), lysine demethylase KDM5A, histone deacetylase HDAC2, and 7 nuclear factor interleukin 3 (NFIL3), a component of the circadian clock, were enriched 8 in senescent, hyperinflammatory neutrophils (cluster 2) ( Fig. 3j and k, Supplemental 9 which is informed by developmental programs and tissue and stress-specific signals (74-19 76), we used MacSpectrum(77), an analytical tool to stratify macrophage maturation and 20 activation. The dominant population enriched for gene transcripts associated with 21 'antigen processing and presentation' represents terminally differentiated mature AM 22 (cluster 1) ( Fig. 4a-e) . The population enriched for transcripts associated with 23 'inflammation', 'purinergic-inflammasome signaling' and 'IL1 receptor activation' was 24 identified as polarized, inflammatory AMs (cluster 2) ( Fig. 4a-e) . Immature AMs were 25 identified based on expression of gene transcripts associated with 'DNA replication and 26 cell division' (cluster 3) (Fig. 4a-e, Supplemental Table 16 ). Cellular trajectory analysis 27 identified a relationship between differentiation and activation as the AM developed 28 from immature (cluster 3) to mature (cluster 1) to polarized inflammatory AM (cluster 2) 29 ( Fig. 4f) . 30 Early-life antibiotic exposure strongly influenced the development and activation state of 1 AMs after infection with S. pneumoniae. Mature AMs (cluster 1), known to maintain 2 noninflammatory states by promoting tolerance and facilitating tissue repair (78), were 3 decreased in dysbiotic newborns. In contrast, the frequency of polarized, inflammatory 4 AMs (cluster 2) was increased also in dysbiotic newborn macaques (Fig. 4g) . Also, CyTOF 5 cytometry demonstrated that the frequency of M1-activated AMs (identified as live 6 MHCII + , CD11C + , CD11B + , CD86 + cells) was increased in dysbiotic newborns after 7 infection with S. pneumoniae, consistent with scRNAseq findings (Fig. 4h) , and correlated 8 with disease severity (Supplemental Fig. 4i) . 9 We hypothesized that dysfunctional inflammatory macrophages were associated with 10 severe pneumonia in dysbiotic macaques. Using published gene signatures from AMs in 11 acute respiratory distress syndrome (ARDS) in humans (46), we found that gene 12 transcripts predicting recovery were enriched in cluster 1 (Fig. 4i , Supplemental Table 13 17). Conversely, gene transcripts predicting severe ARDS/death were enriched in cluster 14 2 ( Fig. 4i, Supplemental Table 17 ). Consistent with our hypothesis, expression of genes 15 related to tissue damage and vascular inflammation were differentially enriched in 16 dysbiotic macaques ( Fig. 4j and Supplemental Table 18 ). ATP released from damaged 17 epithelium and endothelium activates an ATP-driven purinergic-inflammasome 18 signaling pathway and is associated with fatal pneumonia and severe ARDS (79). 19 Consistent with the above observations, we found increased ATP levels in the bronchial 20 lavage fluid of dysbiotic macaques that did not reach statistical significance (p = 0.06) 21 (Supplemental Fig. 4g) . Finally, decreased expression of tolerance promoting MHC class 22 II genes further supports the global dysfunction of AM in dysbiotic newborns (Fig. 4j) . 23 24 Antibiotic exposure also remodeled the IM pool after infection with S. pneumoniae. 25 Analogous to AM, we identified three unique populations of mature IM marked by 26 expression of genes related to macrophage migration, phagocytosis, tolerance promotion, 27 and wound repair (Cluster 1); activated IMs identified by their expression of genes 28 related to antigen presentation, IL1 receptor activation, and T helper cell differentiation 29 (Cluster 2); and a population marked by expression of genes associated with ER stress, 30 exhaustion, and apoptotic clearing, identified as exhausted and stressed IM (Cluster 3) 31 (Supplemental Fig. 4a-d) . In addition, antibiotic exposure was associated with 32 contraction of pro-repair IMs and expansion of stressed, exhausted pro-inflammatory 33 IMs (Supplemental Fig. 4e , f, h, and i, Supplemental Table 19 ). Collectively, these data 1 suggest that the loss of pro-phagocytic, tolerance-promoting, and antigen-presentation 2 programs, which facilitate protective functions of lung macrophages and expansion of 3 ATP-purinergic inflammasome signaling and pro-inflammatory programs are associated 4 with lung damage and increased pneumonia-related morbidity in dysbiotic macaques. Table 20) . 14 15 A regulatory network anchored by ARNTL (encoded by BMAL1), a component of the 16 circadian clock, was uniquely overrepresented in the stressed and hyperinflammatory 17 macrophages (cluster 2) that was expanded in dysbiotic newborns ( Fig. 4k and l) . Diurnal 18 oscillations of intestinal microbiota are thought to drive the programming of host 19 immune responses via ARNTL (83) and other components of the circadian clock. In 20 addition, RXRA and NFIL3 were overrepresented in regulatory networks for cluster 2 21 (Fig. 4k) . These data, coupled with similar observations in neutrophils ( Fig. 3j and k) , 22 suggest a role for shared regulatory networks anchored by circadian clock components 23 in the transcriptional remodeling of pulmonary neutrophil and alveolar macrophage 24 compartments in dysbiotic newborns. 25 26 with S. pneumoniae. In contrast to the significant remodeling of the myeloid 28 compartment, we observed more modest changes in the distribution of various 29 pulmonary T cells between dysbiotic and control newborns (Supplemental Fig. 5a-c) . 30 Cytokines associated with T cell differentiation and effector responses, such as IL6, IL8, 31 IL17, CXCL8, and CXCL10, increased in the bronchial lavage fluid of dysbiotic macaques 32 (Supplemental Fig. 2d, Supplemental Fig. 5d) . Hyper-or hypoactivation of T cells, or 33 skewing towards an ineffective differentiation state, such as T H 17 cells, exhausted T cells, 1 or terminally differentiated T cells, are associated with severe viral pneumonia in animal 2 models (84). The majority of pulmonary T helper cells from dysbiotic newborns co-3 expressed CD279 (PD-1) and CD38, markers linked to T cell exhaustion (85-88), and 4 showed decreased expression of costimulatory molecules, CD28 and CD40 5 (Supplemental Fig. 5e ). The frequency of dysfunctional CD4 + T helper cells (co-6 expressing CD279 and CD38) correlated with disease severity (Supplemental Fig. 5f) . and proliferation markers, such as MCM, PCNA, and EIF4A1, identified CD56 + NK cells 30 (Cluster 2) and proliferating NK cells (Cluster 3), respectively (Supplemental Fig. 5i and 31 j). In contrast, a cluster expressing transcripts associated with cell survival, cellular stress, 32 and inflammation represented cytotoxic and stressed NK cells (Cluster 1) (Supplemental 33 Fig. 5i and j) . The numbers of cytotoxic and stressed NK cells expanded in dysbiotic 1 newborns, consistent with CyTOF analysis (Supplemental Fig. 5k) . Next, the 2 transcriptomic analysis further revealed an increased abundance of transcripts associated 3 with T cell activation (97), inflammation, and exhaustion in dysbiotic newborns 4 (Supplemental Fig. 5l, Supplemental Table 22 ). These data, coupled with reports 5 implicating dysfunctional cytotoxic NK cell responses in severe COVID-19(41), suggest 6 that defects in NK cell cytotoxicity may be associated with adverse outcomes caused by 7 pneumonia in dysbiotic newborns. NOTCH and SEMA4, were dominant in control newborns ( Fig. 5a and b) . In contrast, 17 signaling pathways related to inflammation(105, 106), such as CD86 and RESISTIN, and 18 cell exhaustion(107, 108), such as PDL1 and PDL2, were dominant in dysbiotic newborns 19 ( Fig. 5a and b) , suggesting a global rewiring of immune cell-to-cell communication 20 network in dysbiotic newborn macaques (Fig. 5c, Supplemental Figure 6a -c). 21 Dysfunctional macrophages were the 'central hub' of misdirected cell-cell 23 communications ( Fig. 5d and e, Supplemental Fig. 6a and b) . In contrast, neutrophils 24 and, to a lesser extent, T cells were predicted to be 'targets' of such miscommunications, immune co-stimulation and complement activation, such as C3-C3AR1, and tissue repair, 32 such as NOTCH, SEMA4 and THY1 (Fig. 5d, Supplemental Fig. 6f and h) (109, 110) . 33 These data identify a potential mismatch between inflammatory and pro-repair 1 pathways anchored by dysfunctional macrophages in dysbiotic newborns. Although no specific bacterial taxa have been consistently associated with pulmonary 24 host resistance to pathogens, fecal microbiota transplantation, which transfers the entire 25 gut microbiota from one host to another, has demonstrated improved clinical outcomes 26 in immunotherapy trials (110, 112, 113) . We performed fecal transfer (FT); wherein, the 27 fecal contents of control newborns were transferred to dysbiotic newborns on postnatal 28 day 8. Dysbiotic newborns who received fecal transfer, referred hereafter as FT-recipient 29 newborns, were challenged with S. pneumoniae on postnatal day 14 (6 days post fecal 30 transfer). FT-recipients had lower PEW score post-infection, less rapid progression of 31 symptoms, and reduced need for supportive therapy (Fig, 6a-d) . Although all recipients 32 demonstrated clinical benefit, the benefit was variable ( Fig. 6e and f) . 33 1 Gut microbiota composition of the FT-recipients (post-treatment) differed from their 2 baseline (pre-FT) (Fig. 6e) . We observed a non-significant shift of FT-recipients' 3 microbiota toward donor microbiota and those FT-recipients that engrafted closer to the 4 donor displayed a more robust clinical response (Fig. 6e, Supplemental Table 23 ). After 5 FT, gut microbiota had a higher abundance of Bifidobacterium bifidum, a favorable 6 modulator of immune responses in humans (6, 22, 114) (Fig. 6g) . The probiotic 7 strain Bifidobacterium longum 5 1A was associated with a reduced pro-inflammatory 8 response, decreased neutrophil recruitment, and improved ability to combat pulmonary 9 infections induced by Klebsiella pneumoniae in mice(115, 116). While our study lacked the 10 power to establish a clear association between specific bacterial taxa and clinical response 11 to pneumonia, our results indicate FT could be feasible and potentially effective in 12 restoring host resistance in dysbiotic newborns. previously increased in neutrophils of dysbiotic macaques, were decreased in FT-28 recipient newborn macaques (Fig. 6i, Supplemental Fig. 7e , Supplemental Table 24 ). 29 Expression of genes related to inflammasome or IL1-signaling, such as 31 NLRP3, IL1B, IL10RA, and NFKB1, decreased in AMs of FT-recipient macaques 32 compared to dysbiotic macaques (Fig. 6i, Supplemental Fig. 7f , Supplemental Table 25) . 33 Gene transcripts related to macrophage migration, such as CCR2, CCL3, CX3CL1, 1 phagocytosis, such as CLEC7A, CD47, and C3, molecules promoting tolerance, such as 2 CCR2, ADA, IL10, and FCRL3, and wound repair, such as ANG, IL10, and VEGFA, 3 remained unchanged in FT-recipients (Fig. 6i, Supplemental Fig. 7f , Supplemental 4 Table 25 ). Transcripts associated with T cell activation and differentiation, such as IL7R, 5 CCR7, CD3D, CD3E, and antigen processing and presentation, such as CCR4, ICOS, and 6 LYN, which were severely decreased in T cells of dysbiotic newborns, increased, albeit 7 partially, after FT (Fig. 6i, Supplemental Fig. 7g , Supplemental Table 26 ). Transcripts 8 associated with tissue repair and growth, such as VEGF, PDGFA, and EGF, did not 9 recover (Fig. 6i, Supplemental Table 26 ). 10 11 Miscommunication between the innate and adaptive immune cells was reversed after FT 12 (Supplemental Fig. 7h ). Dysregulated signaling pathways related to inflammation (IL1, 13 IL6, IFN-γ) and immune co-stimulation (CD45) were corrected in FT-recipients (Fig. 6j) . 14 However, signaling pathways associated with tissue repair (NOTCH, IL10 and SEMA4) 15 and chemotaxis (CXCL) remained dysregulated in FT-recipients (Fig. 6j , Supplemental 16 The use of newborn rhesus macaques permitted a highly granular examination of 24 pulmonary and systemic immune responses to a common respiratory pathogen, not 25 possible in either murine or human neonates alone. Their larger size and similarity of 26 lung structure, developmental stages, physiology, and mucosal immune mechanisms 27 (18) to human infants, coupled with the longitudinal assessment of disease progression, 28 permitted us to distinguish mild, self-resolving pneumonia from severe pneumonia. In 29 this study, succession of microbial communities in vaginally delivered macaques was 30 similar to the pattern seen in human neonates (21, 22), with an abundance of 31 Gammaproteobacteria and Bifidobacterium bifidum, taxa reported to favorably modulate 32 immune response in humans. Early-life antibiotic exposure disrupted the evolution of 33 gut microbiota in newborn rhesus macaques. Bacteroides and Parabacteroides were 1 underrepresented, whereas Enterococcus and Clostridium were overrepresented, similar to 2 gut microbial communities in a large cohort of vaginally delivered infants exposed to 3 perinatal antibiotics (117, 118). This early-life antibiotic exposure was associated with 4 increased morbidity to S. pneumoniae, recapitulating epidemiological observations (33-5 35). 6 7 We found that early-life antibiotic exposure created a maladapted immune state, 8 characterized by a solid pro-inflammatory signature in the peripheral blood. 9 Longitudinal high-dimensional data from next-generation sequencing, plasma proteins, 10 and high-parameter CyTOF provided a further opportunity to integrate diverse data 11 related to individual immune cell populations and plasma proteins in neonates (119). 12 However, prior studies were limited to healthy newborns (120) and could not assess 13 either rapid changes in the neonate's immune system during the first two weeks of life 14 transcripts were enriched in dysbiotic macaques. We speculate that extended neutrophil 33 lifespan and the resulting exhaustion, coupled with failure to remove senescent and 1 exhausted neutrophils from the infected lungs, caused severe tissue damage due to the 2 release of proteases, cationic peptides, and cytokines, which were increased in dysbiotic 3 newborn macaques. An extended neutrophil lifespan was observed in patients with 4 sepsis (125), ARDS (126), severe asthma (127), or acute coronary artery disease (128) and 5 was associated with disease progression and poor prognosis. Our data suggest that 6 therapeutic approaches targeting hyperinflammation, neutrophil clearance from 7 inflamed tissues, or induction of neutrophil apoptosis may have the potential to improve 8 clinical outcomes in infected, dysbiotic infants. suggests that neutrophils and CD4 + T cells transform the first-order cytokine signals 22 into second-order cytokines that enhance the trafficking and extravasation (OSM-LIFR), 23 immune co-activation (Complement C3-C3AR1), and effector function, such as 24 phagocytosis (PTPRC-MRC1), to eliminate pathogens. At the same time, we report that 25 reciprocal interactions limit macrophage activation (MIF-CD74/CXCR2 and CD83-26 PECAM1) and promote tissue-repair factors (CSF1-CSFR1, TGFBR3-TGFB1, SEMA4-27 NRP). Sequential engagement of these communication circuits ensures that the minimum 28 necessary response to a microbe is engaged (Supplemental Fig.6j) . newborns. Similarly, macrophage-anchored signaling pathways related to neutrophil 2 migration, such as CXCL-CXCR2 and THBS1-CD47, neutrophil extravasation, such as 3 ITGB2-ICAM1, and neutrophil activation, such as SELPLG-SELL, were disrupted in 4 dysbiotic newborns (Supplemental Fig.6j) . These changes in the communication circuits 5 can potentially explain the shared hyperinflammatory signature in all immune cells and 6 global loss of tissue-protective, homeostatic pathways. It may also explain the observed 7 distinct patterns of immunoparalysis, such as dysregulated antigen presentation in the 8 macrophages and impaired costimulatory function in T cells. first-order cytokines, which activate other innate and adaptive immune cells via second-29 order cytokines. FT may not completely mitigate the disruption of such first-and second-30 order effectors, most likely due to persistent macrophage dysfunction, contributing to 31 suboptimal benefit. Nevertheless, our study provides proof-of-concept evidence for the 32 ability of FT to improve clinical outcomes in 'at-risk' dysbiotic newborns. 33 1 Our study has some limitations. The necessity of frequent clinical examination, sample 2 collection, and invasive procedures precluded us from using dam-reared infants. 3 Therefore, infants in our study received a diet consisting of formula, not breastmilk. As 4 infants grow, feeding practices play an increasing role in determining the composition of 5 the infant gut microbiota (142-144); however, delivery mode and perinatal antibiotic use 6 have a stronger influence on the composition of the microbial community immediately 7 after birth (145). Nevertheless, further studies are needed to delineate the relative 8 contribution of infant diet to pulmonary immune maturation during infancy. Finally, 9 mechanistic studies in murine models are necessary to test the hypotheses presented in 10 this work. 11 In summary, our data suggest that divergence from the canonical innate and adaptive 13 immune responses and tissue-repair programs, which are typically associated with 14 resolution of pneumonia, leads to the clinical morbidity seen in dysbiotic newborns. We 15 show that the immune response in dysbiotic newborns is marked by multifaceted newborn mice was used to estimate that four animals in each group would be sufficient 9 to detect a 20% difference in morbidity with 80% power and an α of 0.05. Twelve 10 vaginally delivered Indian origin rhesus macaque infants (Macaca mulatta) 11 (Supplemental Table 1 ) were used in this study, which was conducted per NIH's Guide 12 for the Care and Use of Laboratory Animals. Infant macaques were separated from their 13 dams, raised in a nursery from the day of birth, and exposed to a normal light cycle (lights visualization, aided by laryngoscope, S. pneumoniae (10 6 CFU in 1 ml of saline) was 28 instilled into the trachea via an 8F feeding tube. After recovery, newborns were returned 29 to the nursery. Vital signs, including heart rate, respiratory effort, blood pressure, urine 30 output, oxygen saturation, and overall clinical condition of these newborn macaques 31 were monitored every 6 hours. Chest radiographs were obtained daily. Sixty hours post-32 infection, the newborn macaques were euthanized. 33 Fecal transplant: Fecal transplants (FT) were performed in newborn macaques who had 1 received the cocktail of vancomycin, gentamicin, and ampicillin from PN days 1-7. Pooled 2 fecal contents from PN7 saline-treated newborn infants were homogenized in phosphate-3 buffered saline (PBS), and fibrous material was filtered out using a 70-&m-pore-size filter. 4 The solution was centrifuged (100 x g), and the pellet was resuspended in PBS with 10% 5 glycerol and frozen at -80°C for later use as donor microbiota. For transplantation, 25 g 6 of fecal donor material was thawed for each FT-recipient animal, resuspended in 20 ml 7 PBS, and gavaged into the duodenum via an endoscope after sedating the infant 8 macaques on PN8. FT-recipient infants (n=4) were inoculated with S. pneumoniae 9 (10 6 CFU) on PN14. (http://press.igsb.anl.gov/earthmicrobiome/protocols-and-standards/16s/) and 21 sequenced using an Illumina MiSeq. We merged ~21 million paired reads using fastq-22 join, to obtain ~12 million assembled 16S amplicon sequences. 23 After quality filtering, demultiplexing with split_libraries_fastq.py of QIIME v. 1.9.1, and 25 further processing with vsearch v. 1.11.1(146), we identified ~90,000 operational 26 taxonomic units (OTUs). Subsequent taxonomic assignment against the 16S reference 27 sequence set of SILVA, v. 1.28(147), OTU sequence alignment, and generation of an 28 unrooted phylogenetic tree, was performed in QIIME. A basic statistical diversity 29 analysis was performed, using core_diversity_analysis.py of QIIME, including alpha-30 /beta-diversity and relative taxa abundances in sample groups. The determined relative 31 taxa abundances were further analyzed with LEfSe (Linear discriminant analysis effect 32 size)(148), to identify differential biomarkers in sample groups. Alpha-diversity analysis 33 was performed on samples after rarefaction to 10000 sequences/sample (minimum 1 sampling depth). Rarefaction curves were generated for the phylogenetic distance 2 between two groups. Phylogenetic distance was calculated at a rarefaction depth of 10000 3 sequences per sample. 4 The Analysis of Composition of Microbiomes (ANCOM) test(149) was used for 6 differential taxonomical analysis with abundance datasets of all taxonomic levels 7 between groups using the FDR cutoff of 0.05. The ANCOM test provides a W-statistic 8 score for each taxon and logical indicators demonstrating whether a taxon is differentially 9 abundant. The ANCOM test also provides a CLR (conditional likelihood ratio test) F-10 statistic score, which accounts for the effect-size difference of each taxon between two 11 groups. Higher F-and W-scores indicate a higher probability for the taxa to be truly 12 different across groups. Volcano plots were generated with W-statistic on the y-axis, and 13 the F-score (CLR mean difference) on the x-axis using an R package ggplot2(150). 14 Heatmaps were created using "pheatmap" for R (RStudio version 1.1.463, based on R 15 tubes for whole blood) and processing the following day. Remaining whole blood was 21 centrifuged (2,000g, 10 min, 4˚C) without brake to separate the cells and plasma within 6 22 hours of collection. Cells were resuspended in RPMI with 10% FCS, slow frozen using a 23 Mr. Frosty device, and stored at -80˚C till further analysis. Plasma separated from the 24 cells was frozen at -80˚C till further analysis. Alveolar wash was centrifuged (2,000 x g, 25 10 min, 4˚C) and cells were spun onto glass slides using a Cytospin 4 Cytocentrifuge 26 (Thermo Scientific), dried for 10 min, fixed in methanol, and stained with the Hema 3 27 manual staining system (Fisher Diagnostics) to identify different immune cells. transformed data using a grid size of 10x10. Eleven main cell lineages in the data were 10 identified (Naïve CD4 + and CD8 + T cells, cytotoxic CD8 + T cells, B cells, CD56 bright NK cells, 11 CD56 dim NK cells, CD14 + and CD16 + monocytes, neutrophils and dendritic cells) and 12 shown as an overlay on the tSNE projection (Supplemental Fig. 1a ) and as Minimum 13 Spanning Tree projection (Supplemental Fig. 1b) . To precisely describe the phenotypic 14 landscape of the neutrophils and CD4 + T cells, the dataset was further 15 partitioned/clustered using the FlowSOM algorithm, resulting in several clusters. Omnibus, Accession number GSE176408) were aligned to the Rhesus macaque reference 32 Mmul_10 with Cell Ranger 1.3 (10X Genomics), generating expression count matrix files 33 (see Supplemental Table 28 for dataset metrics). Cells that had fewer than 750 UMIs or 1 greater than 15,000 UMIs, as well as cells that contained greater than 20% of reads from 2 mitochondrial genes or rRNA genes (RNA18S5 or RNA28S5) or hemoglobin genes, were 3 considered low quality and removed from further analysis. Putative multiplets were 4 removed with DoubletFinder (version 2.0). Genes that were expressed in fewer than 10 5 cells were removed from the final count matrix. 6 7 Data analysis: The Seurat package (version 3.1.0, https://satijalab.org/seurat/) was 8 used to identify common cell types across different experimental conditions, differential 9 expression analysis, and most visualizations. Percentages of mitochondrial, ribosomal 10 genes, and hemoglobin genes were regressed during data scaling to remove unwanted 11 variation due to cell quality using the SCTransform () function in Seurat. PCA was 12 performed using the 3,000 most highly variable genes, and the first 20 principal 13 components (PCs) were used to perform UMAP to embed the dataset into two 14 dimensions. Next, the first 20 PCs were used to construct a shared nearest neighbor graph 15 (SNN; FindNeighbors ()) and this SNN used to cluster the dataset (FindClusters ()). 16 Manual annotation of cellular identity was performed by finding differentially expressed 17 genes for each cluster using Seurat's implementation of the Wilcoxon rank-sum test 18 (FindMarkers()) and comparing those markers to known cell type-specific genes from 19 published studies (37) , (38). Global differential gene expression profiles between all cell 20 types were identified and organized with the software cellHarmony (152), using these 21 Seurat clusters (fold change > 1.2, empirical Bayes t-test p-value <0.05, FDR corrected) 22 Automated annotation of T cell subsets: This annotation was performed using web 24 application (https://azimuth.hubmapconsortium.org) (153). Anchors between the 25 reference (multimodal reference dataset of > 100,000 PBMC) and our dataset were 26 identified using a precomputed, supervised PCA of the reference dataset that maximally 27 captures the structure of the weighted n neighbor (WNN) graph. Cell type labels from 28 the reference dataset were transferred to each cell of the query data set through 29 previously identified anchors. signature of senescent neutrophils could be used to predict mortality, we first developed 32 a five-gene signature of the senescent neutrophils by identifying the most differentially 33 enriched genes in cluster 2 neutrophils in our transcriptomic dataset relative to all other 1 cells. Next, we downloaded normalized transcript counts from a publicly available whole 2 blood bulk transcriptomic dataset (GSE696686) (59). We then scored each sample in this 3 dataset by the expression of the five genes enriched in our senescent neutrophil cluster 4 (HIF1A, CXCR4, CD274, LTF, and S100A8). Finally, we used these gene signature scores 5 as a predictor variable and disease severity metadata reported by Wynn et al.(59) as the 6 response variable to construct an ROC curve to quantify and visualize the sensitivity and 7 specificity of the prediction. Nectins and nectin-like molecules: roles 1 in contact inhibition of cell movement and proliferation Involvement of notch signaling in wound healing Semaphorin-Plexin Signaling Controls Mitotic Spindle Orientation 6 during Epithelial Morphogenesis and Repair Differential role for CD80 and CD86 in the regulation of the innate 8 immune response in murine polymicrobial sepsis Resistin, an 10 adipokine with potent proinflammatory properties Growth arrest in human T-cells is controlled by the non-coding RNA growth-14 arrest-specific transcript 5 (GAS5) PD-L2 is a second ligand for PD-1 and inhibits T cell activation Frontline Science: Defects in immune function in patients with 18 sepsis are associated with PD-1 or PD-L1 expression and can be restored by 19 antibodies targeting PD-1 or PD-L1 Gut microbiome modulates response to anti-PD-1 21 immunotherapy in melanoma patients PD-L1 blockade improves survival in experimental sepsis by 23 inhibiting lymphocyte apoptosis and reversing monocyte dysfunction The commensal microbiome is associated with anti-PD-1 efficacy 26 in metastatic melanoma patients Gut microbiome influences efficacy of PD-1-based immunotherapy 28 against epithelial tumors Interaction between microbiota and immunity in 30 health and disease Oral treatment with Bifidobacterium longum 51A reduced 32 inflammation in a murine experimental model of gout Control of Klebsiella pneumoniae pulmonary infection and 35 immunomodulation by oral treatment with the commensal probiotic 36 Bifidobacterium longum 5(1A) Effects of microbiota-directed foods in gnotobiotic animals and 38 undernourished children Persistent gut microbiota immaturity in malnourished 40 Bangladeshi children A Call for Blood Stereotypic Immune System Development in Newborn Children Perinatal inflammation influences but does not arrest rapid 46 immune development in preterm babies Dynamic molecular changes during the first week of human life 48 follow a robust developmental trajectory Delay of neutrophil apoptosis in acute coronary syndromes Phagocytosis of apoptotic cells in 13 homeostasis Diverse macrophage populations 15 mediate acute lung inflammation and resolution Soluble PD-L1 improved direct ARDS by reducing monocyte-derived 18 macrophages Severe COVID-19 Is Marked by a Dysregulated 20 Transkingdom control of microbiota diurnal oscillations 22 promotes metabolic homeostasis Haematopoietic stem cell 24 release is regulated by circadian oscillations Circadian gene Bmal1 regulates diurnal oscillations of 26 Ly6C(hi) inflammatory monocytes A Neutrophil Timer Coordinates Immune Defense and 28 Vascular Protection An epithelial circadian clock controls pulmonary inflammation and 30 glucocorticoid action A Neutrophil Timer Coordinates Immune Defense and 32 Vascular Protection Circadian Features of Neutrophil 34 Clocking in to immunity Desynchronization of the molecular clock contributes to the 38 heterogeneity of the inflammatory response Antibiotics, birth mode, and diet shape microbiome 40 maturation during early life Association of Exposure to Formula in the Hospital and 42 Subsequent Infant Feeding Practices With Gut Microbiota and Risk of Overweight 43 in the First Year of Life Breast-fed and bottle-fed infant rhesus macaques develop 45 distinct gut microbiotas and immune systems Delivery Mode Affects Stability of Early Infant Gut 47 Microbiota VSEARCH: a versatile open 1 source tool for metagenomics The SILVA ribosomal RNA gene database project: improved data 3 processing and web-based tools Metagenomic biomarker discovery and explanation Analysis of composition of microbiomes: a novel method for 7 studying microbial composition Human eosinophils recognize endogenous 12 danger signal crystalline uric acid and produce proinflammatory cytokines 13 mediated by autocrine ATP cellHarmony: cell-level matching and holistic 15 comparison of single-cell transcriptomes Integrated analysis of multimodal single-cell data. bioRxiv ToppGene Suite for gene list 18 enrichment analysis and candidate gene prioritization GO-Elite: a flexible solution for pathway and ontology over-21 representation