Insulin like growth factor 1 supports a pulmonary niche that promotes type 3 innate lymphoid cell development in newborn lungs.

Summary:

Type 3 innate lymphoid cells (ILC3s) are critical for lung defense against bacterial pneumonia in the neonatal period, but the signals that guide pulmonary ILC3 development remain unclear. Here, we demonstrated that pulmonary ILC3s descended from ILC precursors that populated a niche defined by fibroblasts in the developing lung. Alveolar fibroblasts produced insulin-like growth factor 1 (IGF1), which instructed expansion and maturation of pulmonary ILC precursors. Conditional ablation of IGF1 in alveolar fibroblasts or deletion of the IGF-1 receptor from ILC precursors interrupted ILC3 biogenesis and rendered newborn mice susceptible to pneumonia. Premature infants with bronchopulmonary dysplasia, characterized by interrupted postnatal alveolar development and increased morbidity to respiratory infections, had reduced IGF1 concentrations and pulmonary ILC3 numbers. These findings indicate that the newborn period is a critical window in pulmonary immunity development, and disrupted lung development in prematurely born infants may have enduring effects on host resistance to respiratory infections.

Graphical Abstract

eTOC:

The signals that guide pulmonary ILC3 development remain unclear. Oherle et al. show that developmentally programmed expression of IGF1 in alveolar fibroblasts links postnatal lung growth with development of pulmonary ILC3s, coordinating concurrent development of lung and mucosal defenses. These data help explain why perturbations incurred during development, such as premature birth, are associated with increased risk of respiratory infections.

One sentence summary:

Fibroblast derived IGF1 directs ILC3 development in newborn lung

Introduction:

Bacterial pneumonia kills more than one million infants around the world each year (Walker et al., 2013). Robust helper T (T_H) cell-mediated adaptive immune response confers protection against pneumonia in adult mice and humans (Chen and Kolls, 2013). However, adaptive immune cells are scarce in the newborn lungs (dos Santos et al., 2013) and exhibit limited functionality (Opiela et al., 2009). In the context of inadequate T cell responses, innate immunity plays a dominant role in protection against respiratory infections in newborns (Kollmann et al., 2017). Type 3 Innate lymphoid cells (ILC3s), a group of sentinel mucosal immune cells, are among the first lymphocytes to colonize the mucosal tissues, including the newborn lungs (Diefenbach et al., 2014). ILC3s represent the initial responders to bacterial pathogens in the lungs (Van Maele et al., 2014). ILC3s are the numerically dominant ILC subset in the human lungs (De Grove et al., 2016). ILC3s can modulate the adaptive immune response by promoting proliferation and differentiation of T_H cells (von Burg et al., 2014) or by supporting the homeostasis of regulatory T (T_reg) cells at mucosal sites. Thus, ILC3s are in a prime position to potentially shape the emerging mucosal immunity (Bando and Colonna, 2016).

A wave of ILC3s infiltrates the newborn lungs during the first week of life (Huang et al., 2018). Newly arriving ILC3s are indispensable for newborn’s defense against bacterial pneumonia (Gray et al., 2017). It is hypothesized that ILC precursors disseminate from primitive yolk-sac-derived precursors in close concordance with fetal liver monocyte dissemination and seed peripheral tissues (Kotas and Locksley, 2018). A second wave of expansion extending from after birth through weaning is presumed to be marked by regional expansion and maturation in response to cues from niche stromal cells (Koga et al., 2018). Nevertheless, the anatomical location of these precursors in the developing lungs, the identity of specific stromal cells responsible for the expansion of pulmonary ILC precursors, and the knowledge of specific factors sustaining local division and differentiation of pulmonary ILC precursors are unknown. Understanding the mechanisms involved in the development and maturation of pulmonary ILC3s is of paramount importance to augment ILC3 response and develop lung protective strategies for the newborn.

Results:

IL-22 producing pulmonary ILC3 were essential for host defense against respiratory pathogens in the newborn.

We and others have previously shown that interleukin (IL)-22 is essential for newborn’s defense against respiratory pathogens, including Streptococcus pneumoniae (serotype 19 A), a leading cause of pneumonia in human newborns (Kaplan et al., 2010). Unlike adults, neither macrophages (CD45⁺F4/80⁺) or T cells (CD45⁺CD5⁺) or NK cells (CD45⁺NK.1⁺) were the principal sources of IL-22 in the newborn lungs (postnatal day [PN3]) (Fig. 1A). In contrast, the majority of IL-22 producing cells in the newborn lung (PN3) were lineage negative lymphocytes expressing transcription factor RAR related orphan receptor (RORγ)t, but not GATA binding protein (GATA) 3 or T box transcription factor 21 (TBX21 or T-bet), which identified them as ILC3 (CD45⁺CD3ε⁻CD5⁻CD11b⁻CD19⁻MHCII⁻F4/80⁻Ly6G SiglecF⁻) CD127⁺ α4β7⁺ RORγt⁺ GATA3⁻T-bet⁻ cells) (Fig. 1A, Fig. S1A). The repertoire of IL-22 producing ILC3s in the newborn lungs (PN3), included NCR⁺ILC3s (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]⁻ CD127⁺RORγt⁺GATA3⁻CCR6-NKp46⁺ cells) or NCR⁻ILC3s (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]⁻ CD127⁺RORγt⁺GATA3⁻CCR6⁻NKp46⁻ cells) or CCR6⁺ILC3s (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]⁻ CD127⁺RORγt⁺GATA3⁻NKp46-CCR6⁺ cells) (Fig. 1B). Furthermore, the frequencies of IL-22 producing NCR⁺ILC3s or NCRILC3s remained unchanged during postnatal development (PN1-PN21) (Fig. 1C). We confirmed these observations using transgenic Rorc^GFP/+ mice reporter newborn mice (PN3) (Fig. S1B).

Figure 1: IL-22 producing pulmonary ILC3s were essential for host defense against respiratory pathogens in the newborn.

A) Bivariate plots of live cells pooled from eight PN3 newborn mice. Numbers represent the frequency of respective cells.

B) Bivariate density plots of ILC3s (defined as live CD45⁺Lin⁻CD127⁺RORγt⁺GATA3⁻ cells) from six PN3 newborn mice. Numbers represent the frequency of IL-22⁺ ILC3 or frequencies of different ILC3 subsets. Lineage panel includes CD3ε,CD5,F4/80,CD11b,CD19 and Ly6G. Absolute number of live CD45⁺Lin⁻CD127⁺cells (ILCs) were 1609 ± 170 cells / lung. Absolute number of ILC3 were 370 ± 73 cells / lung.

C) Frequency of ILC3 subsets in lungs of age-defined newborn B6 mice.

D) Generation of Rorc^ΔIl22 mice. Representative histograms of CD45⁺Lin⁻ [CD3ε,CD5,F4/80,CD11b,CD19,Ly6G]⁻ CD127⁺RORγt⁺GATA3⁻ cells from lungs of newborn Rorc^Cre or Rorc^ΔIl22 or Il22 ^fl/fl mice stained for IL-22. Isotype antibody depicted in gray. Live cells were pooled from six PN3 mice.

E) Newborn Rorc^Cre or Rorc^ΔIl22 or Il22 ^fl/fl mice were examined for bacterial load in the lungs or F) susceptibility after intratracheal (i.t.) inoculation with S. pneumoniae or sham on PN5.

G) Newborn Rorc^Cre or Rorc^ΔIl22 or Il22 ^fl/fl (PN3) mice treated with i.t. IL-22 were examined for susceptibility after i.t. inoculation with S. pneumoniae or sham on PN5.

H) Newborn Rorc^ΔIl22 received adoptive transfer of ILC3s (sorted as Live CD45⁺Lin-CD127⁺(RORγt)GFP⁺ KLRG⁻ NK1.1⁻ cells) from age-matched from Rorc^GFP/+ mice [on CD45.1 background] or sham on PN3. Representative histograms from animals which received adoptive transfer (top panel) or sham (lower panel). Numbers represent the frequency of CD45.1⁺ (adoptively transferred) or CD45.2⁺ (host: Rorc^ΔIl22) cells after 48 hours. Data represents cells pooled from six PN3 newborn mice.

I) Newborn Rorc^Cre or Rorc^ΔIl22 that received adoptive transfer of ILC3s from age-matched from Rorc^GFP/+ mice were examined for susceptibility after i.t. inoculation with S. pneumoniae or sham on PN5.

Data is representative of three independent experiments. Results shown as the means ± s.e.m (ANOVA [Fig. 1E] or Kaplan-Meier log-rank test [Fig. 1F, 1G and 1I], *P ≤ 0.05; **P ≤ 0.01 and number of individual animals [n] are indicated in the figures). See also Fig. S1.

Intratracheal challenge with S. pneumoniae activated the ILC3s, as evidenced by the expanded frequency of IL-22 producing ILC3s (Fig. S1C) and increased mean fluorescent intensity (MFI) of IL-22 in pulmonary ILC3s (Fig. S1D) from S. pneumoniae challenged newborn Rorc^GFP/+ mice (PN5), consistent with prior observations (Felton et al., 2018; Van Maele et al., 2014). The IL22 MFI and frequency of IL-22⁺ ILC3s in the small intestine (SI) remained unchanged after infectious challenge (Fig. S1D, E), which suggested that ILC3 activation was restrained in compartment-specific fashion.

Next, we interrogated the importance of ILC3s in the newborn’s resistance to pneumonia. We bred mice harboring the Il22 conditional allele (Il22^fl/fl) with Rorc^Cre mice (Eberl and Littman, 2004). The progeny, referred to here as Rorc^ΔIl22 mice lacked IL-22 in all RORγt expressing cells, including the ILC3 (Fig. 1D). Newborn Rorc^ΔIl22 mice (PN3) were indistinguishable from either Rorc^Cre or Il22^fl/fl littermates in the number of ILC3s or the frequencies of various ILC3 subsets within the lungs or SI (Fig. S1F,S1G,S1H). Newborn Rorc^ΔIl22 mice (PN5) had increased (>3 fold) pathogen burden (Fig. 1E) and demonstrated an expanded susceptibility post challenge with S. pneumoniae (Fig. 1F). Heightened susceptibility to S. pneumoniae was reversed by intratracheal instillation of recombinant IL-22 (Fig. 1G).

To confirm the critical role of ILC3-derived IL-22 in newborn’s defense we adoptively transferred congenically (CD45.1) marked ILC3s via intratracheal route into age-matched Rorc^ΔIl22 newborn mice (PN3). Adoptively transferred ILC3 persisted in the newborn lung (Fig. 1H) and reversed the susceptibility to pneumonia in Rorc^ΔIl22 newborn mice (Fig. 1I). Our data, coupled with prior reports (Aujla et al., 2008; Van Maele et al., 2014; Xiong et al., 2016) established an essential role for ILC3 and IL-22 in host defense against bacterial pneumonia. Having shown that pulmonary ILC3 were protective against bacterial pathogens in the newborn, we next examined how the lungs were populated by protective ILC3s.

Developmental maturation and differentiation of pulmonary ILC3s occurred locally in the lungs.

In adult mice, ILC3s emerge from common lymphoid precursors (CLP) in the bone marrow (BM) after a series of differentiation stages (Yu et al., 2014) (Seehus et al., 2015) into ILC precursors expressing transcription factor, promyelocytic leukemia zinc finger (ZBTB16), also known as zinc finger and BTB domain containing (ZBTB)16, (Constantinides et al., 2014; Klose et al., 2014) hereafter referred to as ZBTB16⁺ ILC precursors. The ZBTB16⁺ ILC precursors complete their maturation into ILC3s (Constantinides et al., 2014; Walker et al., 2019). The newborn BM (PN3) harbored several ILC3 developmental intermediates, including the CLPs (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]^- CD127⁺CD25^-ICOS^-CD135⁺α4β7^- cells and ZBTB16⁺ ILC precursors (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]^-CD127⁺CD25^-ICOS^-CD135^-α4β7⁺ cells) (Fig. 2A). ZBTB16⁺ ILC precursors in the newborn (PN3) BM expressed programmed cell death protein (PD) 1 and transcription factor inhibitor of DNA binding (ID2), but not RORγt or GATA3 or T-bet (Fig. 2A). Recent studies found that, co-expression of α4β7, PD1, ID2 and ZBTB16 marks ILC precursors (Walker et al., 2019; Xu et al., 2019) which harbor the potential to generate all helper ILC lineages. In contrast, mature ILC3s (defined as CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]^-CD127⁺RORγt⁺ cells) were rare in the newborn BM (Fig. 2A). Conversely, CLPs were scarce in the newborn (PN3) SI or the lung (Fig. 2A). However, the newborn (PN3) SI and lung was populated by several ZBTB16⁺ ILC precursors and mature ILC3s. The relative frequencies of early ILC3 developmental intermediates, for instance CLPs, remained unchanged in the BM during postnatal development (PN1–P28) (Fig. S1I). These data, together with recent studies (Koga et al., 2018), suggested that a part of developmental maturation and differentiation of pulmonary ILC3s occurred locally in the lungs.

Figure 2: Developmental maturation and differentiation of pulmonary ILC3 occurred locally in the lungs.

A) Bivariate density plots of live cells from bone marrow or small intestine or lungs of newborn B6 (PN3) mice. CLP were identified as CD45⁺Lin⁻CD127⁺CD25⁻ICOS⁻CD117⁺Sca-1⁻α4β7-CD135⁺RORγt⁻ZBTB16⁻ cells. ZBTB16⁺ ILC precursors were identified as CD45⁺Lin-CD127⁺CD25⁻ICOS⁻CD117⁺Sca-1⁻α4β7⁺CD135⁻ID2⁺RORγt⁻ZBTB16⁺PD1⁺ cells. ILC3s were identified as CD45⁺Lin⁻CD127⁺RORγt⁺GATA3⁻ cells. Lineage panel includes CD3ε,CD5,F4/80,CD11b,CD19 and Ly6G. Cells pooled from six mice. Numbers represent the frequencies of respective cell populations.

B) Live CD45⁺Lin⁻ ZBTB16 (GFP)⁺ cells from the lungs of newborn Zbtb16^GFP mice were examined for expression of indicated cell surface markers or transcription factors. Numbers represent the frequencies of respective populations. GFP gate is indicated in black.

C) Representative bivariate density plots of ZBTB16⁺ ILC precursors (defined as CD45⁺Lin-CD25⁻ICOS⁻CD127⁺α4β7⁺CD135⁻Id2⁺PD1⁺GATA3⁻RORγt⁻ZBTB16 (GFP)⁺ cells) in age-defined Zbtb16^GFPCre newborn or adult (PN28) mice. Cells pooled from six mice. Numbers represent the frequency of ZBTB16⁺ ILC precursors at respective ages.

D) Gating strategy used to quantify pulmonary ILC3s (defined as live CD45⁺ Lin⁻ CD127⁺RORγt (GFP)⁺GATA3⁻ cells) in Rorc^GFP/+ mice. GFP gate is indicated in red. Cells pooled from six PN7 mice. Numbers represent the frequency of respective cell types.

E) Representative bivariate density plots of ILC3s in age-defined Rorc^GFP/+ newborn or adult (PN28) mice. Numbers represent the frequency of total ILC3s or IL-22⁺ ILC3s at respective ages. Cells pooled from six mice.

F) Absolute number of pulmonary ZBTB16⁺ ILC precursors or mature ILC3s in age-defined-newborn or adult mice quantified by flow cytometry.

G) Representative histograms or H) relative proportions of pulmonary ZBTB16⁺ ILC precursors (blue) or mature ILC3s (red) incorporating BrdU in newborn (PN3) or adult (PN28) B6 mice. Broken black line represents isotype control antibody.

Data is representative of three independent experiments. Results shown as the means ± s.e.m (ANOVA [Fig. 2H, 2J], *P ≤ 0.05; **P ≤ 0.01 and number of individual animals [n] are indicated in the figures). See also Fig. S2.

Next, we quantified the number of ZBTB16⁺ ILC precursors in the newborn (PN3) lung (Fig. 2B) or the BM (Fig. S1J) using Zbtb16^GFPCre reporter mice (Constantinides et al., 2014). As expected, in Zbtb16^GFPCre mice, GFP was preferentially expressed in pulmonary ZBTB16⁺ ILC precursors (defined as Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]⁻ α4β7⁺CD127⁺CD135⁻CD25⁻eGFP⁺ cells) (Fig. 2B). Furthermore, a significant proportion of these pulmonary ZBTB16⁺ ILC precursors (Lin⁻ α4β7⁺CD127⁺CD135⁻CD25⁻eGFP⁺ cells) co-expressed PD1 and ID2 (Fig. 2B). The frequency of other GFP-expressing lymphocyte populations, for instance, mature ILCs, CD4⁺ T cells, γδ T cells or NK cells in the newborn lungs was very low (Fig. S1L), consistent with a prior report (Constantinides et al., 2014). Zbtb16^GFPCre is a hypomorphic allele and potentially under-reports the expression of ZBTB16 (Park et al., 2019). We therefore affirmed the expression of ZBTB16 by intracytoplasmic staining in newborn (PN3) B6 mice (Fig. S1L). The number of pulmonary ZBTB16⁺ ILC precursors increased six-fold from PN1(262 ± 31 cells/mouse lung) to PN7 (1344 ± 74/mouse lungs), decreased marginally by PN14 (1791 ± 136/mouse lung) and remained unchanged hereafter to adulthood (1156 ± 153/mouse lung) (Fig. 2C,F).

Next, we used Rorc^GFP/+ mice (Eberl et al., 2004) to quantify the mature ILC3s (Fig. 2D). The increase in pulmonary ZBTB16⁺ ILC precursors was followed by a dramatic expansion in the number of mature pulmonary ILC3s (defined as live CD45⁺ Lin⁻ CD127⁺ RORγt (GFP)⁺ cells) between PN3 (310 ± 102 per mouse lung) and PN14 (1818 ± 142 per mouse lung). The number of ILC3s decreased to adult levels by PN21 (1292 ± 123 per mouse lung) (Fig. 2E, F). The accrual of immature ZBTB16⁺ ILC precursors and developmentally mature ILC3s in successive waves (Fig. 2F) was reminiscent of ontogeny of alveolar macrophages. The newborn lung is populated by a wave of developmentally immature pre-alveolar macrophages which proliferate and differentiate into mature alveolar macrophages (Guilliams et al., 2013). We therefore investigated whether the expanded pool of pulmonary ILC3s in the newborn lungs was similarly derived from proliferation and maturation of local ZBTB16⁺ ILC precursors. ZBTB16⁺ ILC precursors incorporated BrdU, while mature pulmonary ILC3s showed minimal BrdU incorporation (Fig. 2G, H), suggesting that ZBTB16⁺ ILC precursors, but not differentiated mature ILC3s were actively proliferating in the lungs. We therefore sought to determine whether actively proliferating pulmonary ZBTB16⁺ ILC precursors contributed to the expanded pool of mature pulmonary ILC3s in the newborn mice.

Pulmonary ZBTB16⁺ ILC precursors contribute to the homeostatic pool of mature ILC3 in the newborn lung.

Global lack of ZBTB16 differentially affects the homeostasis of different ILC subsets at extra-intestinal sites, such as the lungs (Constantinides et al., 2014), potentially confounding their interpretation. We therefore generated Zbtb16^iDTR mice to selectively deplete pulmonary ZBTB16⁺ ILC precursors in the newborn mice (Fig. 3A). Newborn Zbtb16^iDTR mice or Rosa26^iDTR were treated with diphtheria toxin (DT) via the intratracheal route on PN1 as done previously (GeurtsvanKessel et al., 2008) (Fig. S2A–H, Supplementary Video 5). Newborn Zbtb16^iDTR mice, which received a single treatment of intratracheal DT on PN1 (Fig. 3B) had significantly decreased number of pulmonary ZBTB16⁺ ILC precursors (90% of control) (Fig. 3B, Fig. 3C). ZBTB16⁺ ILC precursors in the BM and SI were unchanged in newborn Zbtb16^iDTR mice (PN3) treated with intratracheal DT, which confirmed the preferential depletion of pulmonary ZBTB16⁺ ILC precursors compared to their gut-resident counterparts (Fig. 3B). Loss of pulmonary ZBTB16⁺ ILC precursors in DT-treated Zbtb16^iDTR newborn mice (PN3) was accompanied by a severe reduction (>70% loss) in the number of mature pulmonary ILC3s (defined as live CD45⁺ Lineage [CD3ε, CD5, CD11b, CD11c, CD19, F4/80, Ly6G]⁻CD127⁺RORγt⁺GATA3⁻ cells) (Fig. 3B) and increased susceptibility after intratracheal challenge with S. pneumoniae (Fig. 3D). Finally, adoptively transferred ILC3 from age-matched B6 mouse reversed the susceptibility to pneumonia in DT-treated Zbtb16^iDTR mice newborn mice (PN5) (Fig. S3A,B). These data unveiled a critical role for pulmonary ZBTB16⁺ ILC precursors in the newborn’s host defense against pneumonia.

Figure 3: Pulmonary ZBTB16⁺ ILC precursors contributed to the homeostatic pool of mature ILC3 in the newborn lung.

A) Generation of Zbtb16^iDTR mice.

B) Zbtb16^iDTR or Rosa26^iDTR newborn mice (PN1) treated with intratracheal diphtheria toxin (i.t. DT) were examined for the presence of ZBTB16⁺ ILC precursors (defined as CD45⁺Lin⁻CD25-ICOS⁻CD127⁺α4β7⁺CD135⁻Id2⁺PD1⁺GATA3⁻RORγt⁻ZBTB16 (GFP)⁺ cells) or ILC3s (defined CD45⁺Lin⁻CD127⁺RORγt⁺GATA3⁻ cells) in lungs, small intestine (SI) or bone marrow (BM) on PN7. Representative bivariate density plots of cells pooled from six mice. Numbers represent the frequency of ZBTB16⁺ ILC precursors or ILC3s in respective organs.

C) Absolute number of ZBTB16⁺ ILC precursors in the lungs of newborn Zbtb16^iDTR or Rosa26^iDTR (PN1) mice treated with i.t. DT or sham.

D) Newborn Zbtb16^iDTR or Rosa26^iDTR mice (PN1) treated with i.t. DT were examined for susceptibility after i.t. inoculation with S. pneumoniae on PN5.

E) Newborn Zbtb16^iDTR or Rosa26^iDTR mice (PN1) treated with i.t. DT were examined for the presence of ILC3s in lungs at defined-ages.

F) Newborn Zbtb16^iDTR or Rosa26^iDTR mice (PN1) treated with i.t. DT were examined for susceptibility after i.t. inoculation with S. pneumoniae on PN28.

G) Newborn Zbtb16^iDTR (PN1) mice treated with i.t. DT, received adoptive transfer of ZBTB16⁺ ILC precursors (sorted as Live CD45⁺Lin⁻CD127⁺α4β7⁺NKp46⁻NK1.1⁻KLRG1⁻CCR6⁻CD25-PD1⁺GFP⁺ cells) from age-matched ZBTB16^EGFPCre/+ mice on CD45.1 background] (upper panel) or sham (lower panel) on PN5. Five days later, the relative proportion of ZBTB16⁺ ILC precursors or ILC3s carrying respective congenic markers were examined by FACS. Representative histograms of live CD45⁺ pooled from six PN10 newborn mice. Numbers represent the frequency of CD45.1⁺ (adoptively transferred) or CD45.2⁺ (host: ZBTB16^iDTR).

H) Newborn Zbtb16^iDTR or Rosa26^iDTR (PN1) mice treated with i.t. DT, received adoptive transfer of ZBTB16⁺ ILC precursors or sham from age-matched congenic (CD45.1) donor mouse on PN3. Two weeks later, the relative proportion of host-derived or adoptively transferred ZBTB16⁺ ILC precursor-derived ILC3s (CD45.1⁺ ILC3) were quantified by FACS (on PN17).

I) Newborn Zbtb16^iDTR or Rosa26^iDTR (PN1) mice treated with i.t. DT, received adoptive transfer of ZBTB16⁺ ILC precursors from age-matched donor mouse on PN3. Seven days later, the newborn mice were examined for susceptibility after i.t. inoculation with S. pneumoniae.

Data is representative of three independent experiments. Results shown as the means ± s.e.m (unpaired two-tailed ANOVA [Fig. 3C, 3E, 3H] or Kaplan-Meier log-rank test [Fig. 3D, 3F, 3I], *P ≤ 0.05; **P ≤ 0.01and number of individual animals [n] are indicated in the figures). See also Fig. S2.

An important limitation of our current experimental model was its potential to disrupt the homeostasis of other ILC subtypes. Indeed, intratracheal DT treatment depleted the pulmonary ILC2s (defined as Live CD45⁺Lin⁻CD127⁺KLRG⁺GATA3⁺RORγt⁻ cells) in Zbtb16^iDTR newborn mice (PN5) (Fig. S3C), which raised the possibility that increased susceptibility to pneumonia in Zbtb16^iDTR newborn mice could have been due to lack of ILC2s. We adoptively transferred congenically (CD45.1) marked WT ILC2s via intratracheal route into age-matched DT-treated Zbtb16^iDTR newborn mice (PN3). While adoptively transferred ILC2s persisted in the newborn lung (Fig. S3D), they failed to reverse the expanded susceptibility to pneumonia in DT-treated Zbtb16^iDTR newborn mice (PN5) (Fig. S3D). The extreme rarity of ILC1s in the newborn lung (Lambert, 2015) precluded us from evaluating whether these cells could mediate host defense as efficiently as pulmonary ILC3s in DT-treated Zbtb16^iDTR newborn mice. NK cells are important in host defense against respiratory pathogens (Paget and Trottein, 2013). While the frequency of NK cells decreased in DT-treated Zbtb16^iDTR newborn mice (PN5), the difference did not reach statistical significance (Fig. S3F). Taken together, these data confirmed that increased susceptibility to pneumonia in DT-treated Zbtb16^iDTR newborn mice was due to the relative numerical deficiency of ILC3s.

Loss of pulmonary ZBTB16⁺ ILC precursors in the newborn period had enduring consequences for the pulmonary ILC3 pool and host resistance to pneumonia beyond childhood.

To determine if the loss of pulmonary ZBTB16⁺ ILC precursors in the early newborn period has durable effects on pulmonary ILC3 pool, we treated neonatal Zbtb16^iDTR mice with DT via intratracheal route on PN1 and followed the animals into adulthood. DT-treated neonatal Zbtb16^iDTR mice demonstrated severely reduced number of pulmonary ILC3s throughout the newborn period (PN3, 7 and 14), that extended into adolescence (PN21) and adulthood (PN28) (Fig. 3E). Similarly, increased susceptibility to pneumonia after an early-life loss of pulmonary ZBTB16⁺ ILC precursors prevailed beyond the neonatal period, into adulthood (Fig. 3F). Next, we asked if restoring pulmonary ZBTB16⁺ ILC precursors in early life could replenish the pulmonary ILC3 pool and restore resistance to pneumonia. We isolated pulmonary ZBTB16⁺ ILC precursors from Zbtb16^EGFPCre reporter mice (backcrossed to CD45.1 background) using a combination of positive and negative selection of cell surface markers (Fig. S3G). We depleted the native pulmonary ZBTB16⁺ ILC precursors by treating Zbtb16^iDTR mice with intratracheal DT. Then, we adoptively transferred congenically labeled (CD45.1) pulmonary ZBTB16⁺ ILC precursors into DT-treated Zbtb16^iDTR mice on PN3. The adoptively transferred pulmonary CD45.1⁺ ZBTB16⁺ ILC expanded numerically in recipient newborn mice as early as 1-week post-transfer (Fig. 3G). Within two weeks, more than 80% of pulmonary ILC3s were CD45.1⁺ (Fig. 3H), which suggested that the majority of pioneering ILC3s in the newborn lung were derived principally from undifferentiated pulmonary ZBTB16⁺ ILC precursors. Importantly, the adoptively transferred ZBTB16⁺ ILC precursors restored the resistance to bacterial pneumonia in DT-treated Zbtb16^iDTR newborn mice (PN3) (Fig. 3I).

Differential requirements for development and maturation of pulmonary ZBTB16⁺ ILC precursors.

We performed genome-wide transcriptional profiling of ZBTB16⁺ ILC precursors from the BM and the lungs in newborn mice (PN3) (Fig. S4A). An unsupervised analysis revealed a differential abundance of transcripts (Fig. S4B) linked to growth factor signaling and chemotaxis in pulmonary ZBTB16⁺ ILC precursors (Fig. S4C). Transcripts for genes related to cell growth, for example, insulin-like growth factor receptor (Igf1r) 1, platelet-derived growth factor receptor (Pdgfr) alpha, and signaling molecules, for instance, ataxin (Atx) 1, phosphoinositide-3-kinase regulatory subunit 1 (Pik3r1), wingless (Wnt1), Wnt3A, Wnt5A and Wnt7A were differentially abundant in pulmonary ZBTB16⁺ ILC precursors compared to their BM counterparts (Fig. S4D). Similarly, transcripts for genes necessary for lung homing such as C-X-C chemokine receptor 5 (Cxcr5)(Chea et al., 2016), Cxcr6 (Chea et al., 2015), chemokine receptor 4 (Ccr4), adhesion molecules such as Semaphorin 5a (Sema5A) and Sema6 were enriched in pulmonary ZBTB16⁺ ILC precursors (Harly et al., 2018; Yu et al., 2016) (Fig. S4D). In contrast, we found no significant difference in transcripts involved in broader ILC development, for example, Nuclear factor, interleukin 3 regulated (Nflil3), Tox, Zbtb16, T cell-specific transcription factor 7 (Tcf7), Gata3, Rorc, Tbx21, Id2 and B Cell Lymphoma 11B (Bcl11b) in pulmonary ZBTB16⁺ ILC precursors compared to their BM-counterparts (Ishizuka et al., 2016) (Fig. S4D). Transcripts for homing receptors known to be expressed on ILC precursors during development, for example, selectin P ligand (Selplg), Cxcr3, Cxcr 4, Cx3cr1, Ccr1, Ccr7 and Ccr8 were comparable between pulmonary ZBTB16⁺ ILC precursors and their BM-counterparts (Fig. S4D) (Harly et al., 2018). Next, we confirmed the expression of proteins encoded by differentially abundant genes. Confirming the findings of the transcriptional analysis, we found differences in expression of IGF1R and CXCR6 on pulmonary ZBTB16⁺ ILC precursors compared to their BM-counterparts. Expression of transcription factors essential for the maintenance of ZBTB16⁺ ILC precursors for instance BCL11B and TCF7, were not different in pulmonary ZBTB16⁺ ILC precursors compared to their BM-counterparts (Fig. S4E). These data suggested that while pulmonary ZBTB16⁺ ILC and BM ZBTB16⁺ ILC share a common transcriptome associated with ILC development, unique differences related to growth factor signaling and chemotaxis mark central vs. pulmonary ILC precursors.

Platelet-derived growth factor receptor (PDGFR)α⁺ alveolar fibroblasts created a microenvironment essential for proliferation of pulmonary ILC3 precursors.

In the BM, locally restricted production of cytokines and growth factors in specialized microenvironments or niches direct the development and proliferation of hematopoietic precursors. We hypothesized that the production of growth factors and activation of growth factor signaling pathway in similar microenvironments might be necessary for the development of lung resident ILC precursors. Similar to primitive macrophages, ZBTB16⁺ ILC precursors localized to the lung parenchyma (Fig. 4A). We evaluated the spatial relationship between the ZBTB16⁺ ILC precursors and the alveolar epithelial cells (identified by NKX2.1) or fibroblast (identified by PDGFRα) with spatial distance modeling (Kunisaki et al., 2013; Short et al., 2013). ZBTB16⁺ ILC precursors were preferentially located closest to PDGFRα⁺ cells (alveolar fibroblast) as compared to NKX2.1⁺ cells (alveolar epithelial cells) in the newborn lungs (PN3–21) (Fig. 4B–F, Supplementary Video 1–4). Mesenchyme-derived fibroblasts are essential for survival and expansion of immature T cells, in secondary lymphoid tissue such as the lymph nodes and spleen (Link et al., 2007; Scandella et al., 2008). In the fetal liver, a population of periportal fibroblasts is critical for maintenance and expansion of hematopoietic stem cells (Khan et al., 2016) and ILC precursors (Koga et al., 2018). More recently, a population of IL-33 producing adventitial stromal cells was shown to support pulmonary ILC2 expansion (Dahlgren et al., 2019). Given the close physical association between the alveolar fibroblast and ZBTB16⁺ ILC precursors (Fig. 4B–F), we hypothesized that alveolar fibroblasts were essential for the expansion and maturation of pulmonary ZBTB16⁺ILC precursors.

Figure 4: ZBTB16⁺ ILC precursors populated a niche defined by platelet derived growth factor (PDGFR) α expressing fibroblast in the newborn lungs.

A) Representative low or high magnification images showing spatial relationship between alveolar epithelial (NKX2.1⁺) cells (green arrowhead - nuclear) (stained with anti-mouse NKX2.1 antibody [clone 1231], or fibroblast (PDGFRα⁺) (white arrowheads - cell surface) (stained with mouse anti-mouse PDGFRα antibody [clone APA5] or ILC (ZBTB16⁺) precursors (yellow arrowheads) (stained with anti-mouse GFP antibody) in age-defined Zbtb16^EGFP newborn (PN1, 7, and 14) or adult (PN28) mice. The immunostained lung tissues were imaged by Nikon Eclipse confocal microscope in 1-μm optical sections.

B) Spatial Distance Modeling. Reconstructed z-stack showing distance vectors between ZBTB16⁺ cell nuclei and PDGFRα⁺ cell nuclei or NKX2.1⁺ cell nuclei in Zbtb16^EGFP newborn (PN7) mice. Scale bars represent 100 micro M.

C) Distance vectors between ZBTB16⁺ cell nuclei and PDGFRα⁺ cell nuclei or D) NKX2.1⁺ cell nuclei plotted within two micro M bin sizes.

E) Distance between ZBTB16⁺ cells (ILC precursors) and PDGFRα⁺ cells (alveolar fibroblast) or NKX2.1⁺ cells (alveolar epithelial cells) and F) relative proportion of PDGFRα⁺ cells (alveolar fibroblast) or NKX2.1⁺ cells (alveolar epithelial cells) found within ten micro M of ZBTB16⁺ cells (ILC precursor). (mean, interquartile range, minimum and maximum value for defined age is displayed).

Data is representative of three independent experiments. Results shown as the means ± s.e.m (unpaired two-tailed Student’s t-test [Fig. 2E, 2F], *P ≤ 0.05; **P ≤ 0.01).

A three-dimensional aggregate assay involving alveolar epithelial cells (CD45⁻CD131⁺CD146-CD310⁻) and fibroblasts (CD45⁻CD131⁻CD146⁺CD310⁻) was used previously to model the alveolar stem cell niche ex vivo (Bertoncello and McQualter, 2011; Hegab et al., 2015; McQualter et al., 2010). We used a similar lung aggregate assay to mimic pulmonary ZBTB16⁺ ILC precursor niche. Pulmonary ZBTB16⁺ ILC precursors were co-incubated with alveolar aggregates in serum-free medium supplemented with IL-7, stem cell factor (SCF), and BrdU for ten days using a protocol described before (Yu et al., 2016). Pulmonary ZBTB16⁺ ILC precursors co-incubated with OP9-DL1 stromal cells served as a control (Fig. 5A). The progeny of pulmonary ZBTB16⁺ ILC precursors co-incubated with lung aggregate included all three ILC lineages (Fig. 5B,C). Furthermore, supplementing the culture medium with IL-23 or IL-33, cytokines which promote the development of specific ILC lineage, further expanded the number of ILC3s and ILC2s respectively (Fig. S5A,B). These results, which were consistent with previous reports (Constantinides et al., 2014; Walker et al., 2019; Xu et al., 2019), confirmed the multilineage potential of ZBTB16⁺ ILC precursors.

Figure 5: Insulin-like growth factor signaling was critical for the proliferation of ILC3 precursors in the pulmonary niche.

A) Pulmonary ZBTB16⁺ ILC precursors (sorted as CD45⁺Lineage [CD3ε, CD5, CD8, TCRβ, CD11b, CD11c, CD19, B220 and Ly6G]⁻, CD25⁻CD127⁺α4β7⁺NKp46⁻Nk1.1⁻KLRG1⁻CCR6-ZBTB16(GFP)⁺ cells) from newborn Zbtb16^GFPCre mice (PN3) were co-incubated with either OP9-DL1 stromal cells or B) three-dimensional aggregate of alveolar epithelial cells (CD45⁻CD326⁺CD140⁻CD31⁻) and alveolar fibroblast (CD45⁻CD326⁻CD140⁺CD31⁻) from age-matched newborn mice in the presence of BrdU.

C) Representative bivariate plots of ILC1s (defined as CD45⁺Lin⁻GATA3-RORγt⁺NKp46⁺NK1.1⁺Eomes⁺ cells) or ILC2s (defined as CD45⁺ Lin⁻CD127⁺GATA3⁺RORγt⁻) or ILC3s (defined as CD45⁺ Lin⁻CD127⁺GATA3⁻RORγt⁺ cells) in the indicated experimental groups.

D) Fold increase in the number of pulmonary ILC3s, in the indicated experimental groups.

E) Frequency of pulmonary ZBTB16⁺ ILC precursors incorporating BrdU in the indicated experimental groups.

F) Pulmonary ZBTB16⁺ ILC precursors from newborn Zbtb16^GFPCre mice (PN3) mice were co-incubated with either three-dimensional aggregate of alveolar epithelial cells with or without alveolar fibroblast isolated from same age-matched newborn mice in presence or absence of 13C6 L-lysine-2HCl.

G) Differentially abundant proteins in the conditioned medium quantified by Mass Spectrometry (MS) are displayed.

H and I) Gene Ontology terms overrepresented in differentially abundant proteins.

J) Pulmonary ZBTB16⁺ ILC precursors from newborn Zbtb16^GFPCre mice (PN3) were co-incubated with three-dimensional aggregate from age-matched newborn mice in presence of either anti-insulin like growth factor (IGF) 1 or anti-IgG antibody. Relative proportion of pulmonary ZBTB16⁺ ILC precursors incorporating BrdU or

K) Increase in the number of mature pulmonary ILC3s.

L) Pulmonary ZBTB16⁺ ILC precursors from newborn mice (PN3) were co-incubated with three-dimensional aggregate in presence of either anti-IL33 receptor (IL33R) or anti-IgG antibody. I) Increase in the absolute number of mature ILC3s quantified by flow cytometry. Data is representative of three independent experiments. Results shown as the means ± s.e.m (ANOVA [Fig. 6D, 6E, 6J, 6K, 6L], *P ≤ 0.05; **P ≤ 0. number of individual animals [n] are indicated in the figures). See also Fig. S5.

Next, we quantified the ability of pulmonary ZBTB16⁺ ILC precursors to expand and generate mature ILC3s when co-incubated with lung aggregates lacking either alveolar fibroblast (Fig. 5C, middle panel) or alveolar epithelial cells (Fig. 5C, lower panel). The number of mature ILC3s were > 100-fold lower (Fig. 5C, D) and the frequency (Fig. 5E) of BrdU incorporating ZBTB16⁺ ILC precursors were 4-fold lower when ZBTB16⁺ ILC precursors were co-incubated with aggregates lacking alveolar fibroblast compared to aggregates lacking alveolar epithelial cells. These data suggested that alveolar fibroblasts were critical for the expansion and maturation of pulmonary ZBTB16⁺ILC precursors.

Insulin-like growth factor signaling was critical for the proliferation of ILC3 precursors in the pulmonary niche.

We sought to identify the paracrine factors produced by the alveolar fibroblast that supported the expansion and maturation of pulmonary ZBTB16⁺ ILC precursors. We used stable isotope-labeled amino acids followed by mass spectrometry (MS) (Fig. 5F) to quantitatively asses the proteins secreted in the conditioned medium (CM) when alveolar aggregates were co-incubated with ZBTB16⁺ ILC precursors. An unsupervised analysis demonstrated consistent differences in the CM from alveolar aggregates with alveolar fibroblast, compared to alveolar aggregates lacking fibroblast (Fig. 5G). Differentially abundant proteins in CM were enriched for biological processes including, activation of IGF1 signaling, lymphocyte proliferation and lung development (Fig. 5 G–I). We hypothesized that IGF1 might play a role in the proliferation of ZBTB16⁺ILC precursors. The role of IGF1 in the expansion of the ILC precursors was interrogated in an alveolar aggregate assay wherein ZBTB16⁺ ILC precursors were co-incubated in presence or absence of anti-IGF1 neutralizing antibody (DiGirolamo et al., 2007). In the presence of anti-IGF1 antibody, the frequency of BrdU incorporating ZBTB16⁺ ILC precursors was decreased 5-fold (Fig. 5J), and mature ILC3s were reduced by approximately 50-fold (Fig. 5K). A recent report implicated adventitial stromal cell-derived IL-33 in the expansion of pulmonary ILC2 (Dahlgren et al., 2019). However, this study did not evaluate pulmonary ILC precursors, raising the possibility that IL-33 could be involved in the expansion and maturation of pulmonary ZBTB16⁺ ILC precursors. We co-incubated the alveolar aggregate and ZBTB16⁺ ILC precursors in the presence or absence of anti-IL33 receptor (IL33R) neutralizing antibody as done previously (Schmitz et al., 2005). In the presence of anti-IL-33R antibody, the number of mature ILC3s were reduced minimally, suggesting that IL-33 may have a limited role in the expansion of pulmonary ILC3s (Fig. 5L). This was not surprising considering the differential requirement of cytokines, for example, IL-25 (Huang et al., 2018) and IL-33 (de Kleer et al., 2016; Ricardo-Gonzalez et al., 2018) in development of ILC2s, but not ILC3s.

Alveolar fibroblast-derived IGF1 was critical for the proliferation of ILC3 precursors in the pulmonary niche in the newborn lungs.

Immunostaining with anti-IGF1 antibody demonstrated that the majority of IGF1⁺ cells in the newborn lungs (PN3, 7 and 21) were alveolar fibroblast (PDGFRα⁺) cells (Fig. 6A, B). We hypothesized that alveolar fibroblasts were the critical source of IGF1 which directed the expansion and maturation of ZBTB16⁺ ILC precursors in vivo. Mice lacking IGF1, either globally or in alveolar fibroblasts, exhibit profound defects in the lung development and in utero growth retardation (Baker et al., 1993; Galvis et al., 2015). Such transgenic mice die within the first week of life (Baker et al., 1993; Galvis et al., 2015) and therefore cannot be used to interrogate the role of alveolar fibroblast-derived IGF1 in postnatal expansion and maturation of pulmonary ZBTB16⁺ ILC precursors.

Figure 6: Alveolar fibroblast-derived IGF1 was critical for the proliferation of ILC3 precursors in the pulmonary niche in the newborn lungs.

A) Representative low or high magnification images showing insulin-like growth factor (IGF1) expressing cells in age-defined B6 newborn (PN3 and 7) or adult (PN 28) mice. IGF1⁺ cells (yellow arrowheads) co-expressed PDGFRα. Scale bars represent 100 micro M and 20 micro M in lower and higher magnification images respectively.

B) Percentage of IGF1⁺ cells in lungs of age-defined B6 newborn or adult mice.

C) T-distributed Stochastic Neighbor Embedding (tSNE) plot of single cell RNAseq of lung cells from newborn (PN1) mice demonstrating relative expression of Igf1 and cell type marker genes Glioma like protein [Gli1] or Platelet-derived growth factor [Pdgfra] (fibroblast) or Endomucin [Emcn] (endothelial cells) or Nkx2.1 (alveolar epithelial cells) or Protein tyrosine phosphatase, receptor type, C [Ptprc] (immune cells).

D) Generation of Gli1^CreER;Igf1^fl/fl newborn mice. Newborn Gli1^CreER or Gli1^CreER;Igf1^fl/fl mice (PN1) were treated with intraperitoneal (i.p.) tamoxifen. Two days later (PN3), the lungs were examined for number of IGF1⁺ cells by immunohistochemistry.

E) Newborn Gli1^CreER or Gli1^CreER;Igf1^fl/fl mice (PN1) were treated with i.p. tamoxifen. The number of pulmonary ZBTB16⁺ ILC precursors (defined as CD45⁺Lin⁻CD25⁻ICOSCD127⁺α4β7⁺CD135⁻Id2⁺PD1⁺GATA3⁻RORγt⁻ZBTB16⁺ cells) incorporating BrdU or F) mature pulmonary ILC3s (defined as CD45⁺ Lin⁻CD127⁺GATA3⁻RORγt⁺ cells) were quantified.

G) Newborn Gli1^CreER or Gli1^CreER;Igf1^fl/fl mice (PN1) treated with intraperitoneal tamoxifen were examined for susceptibility after intratracheal inoculation with S. pneumoniae on PN5. Some newborn mice in both experimental groups (Gli1^CreER or Gli1^CreER;Igf1^fl/fl mice ± tamoxifen treatment) received adoptive transfer of either WT ILC3s (sorted as Live CD45⁺Lin-CD127⁺(RORγt)GFP⁺ KLRG⁻ NK1.1⁻ cells from RORC^GFP/+ mice) or sham on PN3 before intratracheal (i.t.) inoculation with S. pneumoniae on PN5.

H) Generation of Zbtb16^ΔIgf1r newborn mice. Zbtb16^ΔIgf1r or Zbtb16^Cre newborn mice (PN5) were examined for pulmonary ZBTB16⁺ ILC precursors incorporating BrdU or I) mature pulmonary ILC3s.

J) Competitive in vivo expansion assay. Equal number of wild-type (CD45.1⁺) and Zbtb16^ΔIgf1r (CD45.2⁺) or Zbtb16^Cre (CD45.2⁺) common lymphoid precursors (CLP) [CD45⁺Lin⁻CD25⁻ICOSCD117⁺Sca-1⁻CD127⁺α4β7⁻CD135⁺] cells were co-transplanted into sublethally irradiated alymphoid newborn Rag2^−/−;Il2rg^−/− mice on PN1. ZBTB16⁺ ILC precursors or ILC3s expressing respective congenic markers were examined 2 weeks later in the lungs or small intestine.

K) Representative bivariate density plots. Numbers represent the frequency of WT (CD45.1⁺) CLP-derived ZBTB16⁺ ILC precursors (left panel) or ILC3s (right panel). L) Ratios of ZBTB16⁺ ILC precursors or mature ILC3s derived from wild-type (CD45.1⁺) and Zbtb16^ΔIgf1r or Zbtb16^Cre (CD45.2⁺) donor cells in indicated organs.

M) Zbtb16^ΔIgf1r or Zbtb16^Cre newborn were examined for susceptibility after i.t. inoculation with S. pneumoniae on PN5. Some newborn mice in both experimental groups received adoptive transfer of either WT ILC3s or sham on PN3 before i.t. inoculation with S. pneumoniae on PN5. Data is representative of three independent experiments. Results shown as the means ± s.e.m (ANOVA [Fig. 6E, F, H] or Kaplan-Meier log-rank test [Fig. 6G,M], *P ≤ 0.05; **P ≤ 0.01 and number of individual animals [n] are indicated in the figures).

We therefore sought to develop a model to delete IGF1 in alveolar fibroblast in the postnatal period in an inducible fashion. Analysis of single-cell RNA-seq datasets generated from cells dissociated from newborn lungs (PN1) (Guo et al., 2015) demonstrated that IGF1 was highly expressed in the alveolar fibroblasts that co-expressed PDGFRA and Glioma-associated protein (GLI1) 1 (Fig. 6C). Neither endothelial (endomucin [EMCN]-expressing), alveolar epithelial (NKX2.1-expressing) nor immune (Protein tyrosine phosphatase receptor type C [PTPRC]-expressing) cells expressed IGF1, consistent with immunohistochemical analysis. Gli1^CreER transgenic mice expressing tamoxifen-inducible Cre recombinase allowed inducible deletion of genes in Gli1 expressing cells. Gli1^CreER mice were bred with Igf1^fl/fl mice to generate Gli1^CreER; Igf1^fl/fl mice (Fig. 6D). Gli1^CreER; Igf1^fl/fl mice were treated with tamoxifen on PN1 via intraperitoneal route as described before (Ruiz-Camp et al., 2017). Tamoxifen administration decreased the number of IGF1⁺ cells in the postnatal period (Fig. 6D), reduced the proportion of ZBTB16⁺ ILC precursors incorporating BrdU (Fig. 6E), and significantly diminished the number of mature ILC3s in tamoxifen-treated Gli1^CreER; Igf1^fl/fl newborn mice compared to sham-treated Gli1^CreER; Igf1^fl/fl mice (Fig. 6F). Tamoxifen administration also rendered the Gli1^CreER; Igf1^fl/fl newborn mice susceptible after challenge with S. pneumoniae (Fig. 6G). Adoptive transfer of pulmonary ILC3s from WT newborn mice via the intratracheal route restored host resistance to pneumonia in age-matched newborn Gli1^CreER; Igf1^fl/fl mice treated with tamoxifen (Fig. 6G).

We sought to recapitulate these findings by quantifying the ability of alveolar fibroblast lacking IGF1 to support the proliferation of pulmonary ZBTB16⁺ ILC precursors in an ex vivo assay (Fig. S5C). The frequency of BrdU incorporating ZBTB16⁺ ILC precursors (Fig. S5D) and number of mature ILC3s (Fig. S5E) were severely diminished when WT ZBTB16⁺ ILC precursors were co-incubated with aggregates from tamoxifen-treated newborn Gli1^CreER; Igf1^fl/fl mice compared to the sham-treated newborn Gli1^CreER; Igf1^fl/fl mice or Igf1^fl/fl mice. These data, taken together, suggested that loss of alveolar fibroblast derived IGF1 impaired the expansion and maturation of pulmonary ZBTB16⁺ ILC precursors into ILC3s.

IGF1 signaling was critical for expansion and maturation of ILC precursors.

IGF1R is expressed in ZBTB16⁺ ILC precursors and mature ILC3s. We therefore sought to determine the relative contribution of IGF1 signaling in the development of ZBTB16⁺ ILC precursors and homeostasis and function of differentiated ILC3s. Zbtb16^EGFPCre mice were bred with Igf1r^fl/fl mice, to generate Zbtb16^IGF1R mice (Fig. 6H). ZBTB16 is expressed by hematopoietic cells, in addition to ILC developmental intermediates (Walker et al., 2019). However, we found no difference in the number of CLPs in the bone marrow of newborn Zbtb16^ΔIgf1r mice compared to Igf1r^fl/fl or Zbtb16^EGFPCre mice (Fig. S5F). Furthermore, the expression of IGF1R on significant innate lymphocyte populations, for instance, NK cells and γδ T cells, ILC2s and ILC3s in the newborn lung (Fig. S5G) or the SI (Fig. S5H) was unaffected, confirming the preferential deletion of IGF1R on ZBTB16⁺ ILC precursors in newborn Zbtb16^ΔIgf1r vs Zbtb16^EGFPCre mice.

The frequency of BrdU incorporating ZBTB16⁺ ILC precursors was decreased 5-fold, and number of ILC3s were reduced 60-fold in lungs of Zbtb16^ΔIgf1r newborn mice (Fig. 6H, I). The number of mature ILC3s were also diminished in the SI (Fig. S5I). However, the magnitude of change was lower in the SI compared to the lungs.

Zbtb16^ΔIgf1r mice lack an essential feature present in healthy animals: developmental competition between ZBTB16⁺ ILC precursors lacking or expressing IGF1, respectively. Competition between individual cells controls their access to limited microenvironments (Cyster et al., 1994). Such competition can impact the fitness of ILC precursors if the microenvironment, for example, the pulmonary ILC niche, provides signals necessary for their continued development. Therefore, we used an in vivo competitive expansion assay, wherein ILC precursors lacking IGF1 competed for expansion and maturation in the newborn lung. Equal number of wild-type (CD45.1⁺) and Zbtb16^ΔIgf1r (CD45.2⁺) or Zbtb16^EGFPCre (CD45.2⁺) common lymphoid progenitors (CLP) (sorted as CD45⁺Lin⁻CD25⁻ICOS⁻CD117⁺Sca-1⁻CD127⁺α4β7⁻CD135⁺ cells) were co-transplanted into sublethally irradiated alymphoid newborn Rag2^−/−;Il2rg^−/− mice, which lack all ILC subsets (Song et al., 2010) by intrahepatic injection on PN1 (Fig. 6J). Two weeks later, we found a significantly decreased number of CD45.2⁺ ZBTB16⁺ ILC precursors and severely reduced number of CD45.2⁺ ILC3s in the lung and SI. Zbtb16^ΔIgf1r CLP generated fewer ZBTB16⁺ ILC precursors or mature ILC3s in the lung, compared to Zbtb16^EGFPCre CLP (Fig. 6K,L). Finally, impaired expansion and maturation of ZBTB16⁺ ILC precursors rendered newborn Zbtb16^ΔIgf1r mice susceptible to intratracheal challenge with S. pneumoniae (Fig. 6M). Adoptive transfer of WT lung ILC3s as previously described (Gray et al., 2017) restored host resistance to pneumonia in newborn mice (Fig. 6M). These data, taken together, demonstrated that loss of IGF1 signaling impaired the expansion of ZBTB16⁺ ILC precursors and postnatal accumulation of ILC3s in the lungs.

IGF1 regulates multiple aspects of T-cell, B-cell, T reg function (Bilbao et al., 2014; Landreth et al., 1992), monocyte and macrophage effector function (Spadaro et al., 2017) through its interactions with IGF1R, raising the possibility that IGF1 signaling could also be important in homeostasis and function of ILC3s. We therefore bred Rorc^Cre mice with Igf1r^fl/fl mice to generate Rorc^ΔIgf1r mice (Fig. S5J), which lacked IGF1R in all ILC3 subsets. There was no difference in the frequency of CLP in the BM of Rorc^ΔIgf1r newborn mice compared to either Igf1r^fl/fl or Rorc^Cre newborn mice (PN3) (Fig. S5K). Similarly, the absolute number of ILC3s and frequency of different ILC subsets remained unchanged in the newborn lung (Fig. S5L,M) of Rorc^ΔIgf1r mice compared to either Igf1r^fl/fl or Rorc^Cre newborn mice (PN3). Likewise, the expression of IL-22 was not significantly different in ILC3s from newborn (PN3) Rorc^ΔIgf1r mice compared to Igf1r^fl/fl or Rorc^Cre mice (Fig. S5N). These data suggested that while IGF1 signaling was essential for the development and expansion of ZBTB16⁺ ILC precursors, it may have been redundant for ILC3 homeostasis and function.

Expansion and maturation of pulmonary ILC precursors was linked to postnatal lung development.

Colonization by commensal microbes influences the early innate immune environment at mucosal sites (Gensollen et al., 2016). We therefore investigated if commensal microbes were necessary for seeding and expansion of pulmonary ZBTB16⁺ ILC precursors. The BM CLPs were not different in newborn (PN3) germ-free (GF) mice, which lacked commensal microbes, compared to conventionalized (CNV) mice (Fig. S6A). Furthermore, the absolute number of pulmonary ZBTB16⁺ ILC precursors, the frequency of pulmonary ZBTB16⁺ ILC precursors incorporating BrdU (Fig. S6 B,C) were not significantly different in newborn GF mice, compared to CNV mice. In addition, the number of IGF1⁺ cells remained unchanged in the lung of newborn GF mice (PN3), compared to CNV mice (Fig. S6D,E). These data coupled with previous observations (Gray et al., 2017; Gury-BenAri et al., 2016) suggested that commensal microbes, while essential for activation of pulmonary ILC3s, were redundant in the expansion and maturation of pulmonary ZBTB16⁺ ILC precursors.

The temporal pattern of IGF1 mRNA expression in sorted murine (Fig. 7A), or human alveolar fibroblast (Fig. S6F) and postnatal increase in the number of pulmonary IGF1⁺ cells (Fig. 7B, C) paralleled the pattern of proliferation of pulmonary ZBTB16⁺ ILC precursors (Fig. 7D) and numerical expansion of pulmonary ILC3s (Fig. 7E). These findings coupled with the observation that conditioned medium from the alveolar aggregate assay was enriched for genes promoting lymphocyte proliferation and lung development (Supplementary Table 1,2), informed the hypothesis that the expansion of pulmonary ZBTB16⁺ ILC precursors is instructed by IGF-1 coinciding with lung development. Pulmonary ZBTB16⁺ ILC precursors from adult mice (PN28) co-incubated with alveolar aggregates from age-matched adult mice (PN28) (Fig. 7F) demonstrated decreased proliferative ability ex vivo, compared to pulmonary ZBTB16⁺ ILC precursors from newborn mice (PN5) co-incubated with alveolar aggregates from age-matched (PN5) mice (Fig. 7G–I). The reduced proliferative ability of pulmonary ZBTB16⁺ ILC precursors from adult animals could be reversed when co-incubated with alveolar aggregates from newborn mice (Fig. 7G–I).

Figure 7: Expansion and maturation of pulmonary ILC precursors was linked to postnatal lung development.

A) IGF1 mRNA expression in sorted pulmonary fibroblast (CD45⁻CD326⁻CD140⁺CD31⁻) isolated from age-defined B6 mice.

B) Representative high magnification images showing Insulin-like growth factor (IGF)1⁺ cells (yellow arrowheads) in age-defined newborn B6 or adult (PN28) mice. Scale bars represent 20 micro M. C) IGF1⁺ cells per high power field in lungs of age defined newborn B6 or adult mice (mean, interquartile range, minimum and maximum value for defined age is displayed).

D) The relative proportions of pulmonary ZBTB16⁺ ILC precursors (defined as CD45⁺Lin⁻CD25-ICOS⁻CD127⁺α4β7⁺CD135⁻Id2⁺PD1⁺GATA3⁻RORγt⁻ZBTB16 (GFP)⁺ cells) incorporating BrdU were quantified by flow cytometry in age-defined B6 newborn or adult mice.

E) Correlation between number of IGF1⁺ cells, the relative proportion of BrdU incorporating ZBTB16⁺ ILC precursors and absolute number of mature ILC3 (defined as CD45⁺Lin-CD127⁺GATA3⁻RORγt⁺ cells) in lungs of age-defined newborn or adult mice.

F) Pulmonary ZBTB16⁺ ILC precursors from newborn (PN5) or adult (PN28) ZBTB16^GFP mice were co-incubated with three-dimensional aggregate of alveolar epithelial cells (CD45⁻CD326⁺CD140⁻CD31⁻), and alveolar fibroblast (CD45⁻CD326⁺CD140⁻CD31⁻) from either age-matched newborn mice or adult mice as indicated.

G) Numbers represent the frequency of BrdU⁺ ILC precursors. Representative histograms. H) The relative proportion of pulmonary ZBTB16⁺ ILC precursors incorporating BrdU or I) fold increase in mature pulmonary ILC3s quantified by flow cytometry.

J) Equal number of newborn (PN1) or adult (P28) (both CD45.2⁺) common lymphoid precursors (CLP) (sorted as CD45⁺Lin⁻CD25⁻ICOS⁻CD117⁺Sca-1⁻CD127⁺α4β7⁻CD135⁺ cells) were co-transplanted into sublethally irradiated alymphoid newborn (PN1) or adult (PN28) Rag2^−/−;Il2rg^−/− mice. ZBTB16 ILC precursors or mature ILC3s were examined 2 weeks later in the lungs by flow cytometry. Numbers represent the frequency of donor (CD45.2⁺) CLP-derived ZBTB16⁺ ILC precursors (left panel) or ILC3s (right panel).

K) Newborn B6 mice were continuously exposed to hyperoxic environment (hyperoxia) or room air from PN1 for seven days. Representative low or high magnification images showing Insulin-like growth factor (IGF)1⁺ cells (yellow arrowheads) in lungs of newborn B6 mice (PN7) exposed to hyperoxic environment (hyperoxia) or room air. Scale bars represent 100 micro M and 20 micro M in lower and higher magnification images respectively.

L) IGF1⁺ cells per high power field in lungs in lungs of newborn B6 mice (PN7) exposed to hyperoxic environment (hyperoxia) or room air (mean, interquartile range, minimum and maximum value for defined age is displayed).

M) Relative proportions of pulmonary ZBTB16⁺ ILC precursors incorporating BrdU in newborn B6 mice continuously exposed to hyperoxic environment (hyperoxia) or room air from PN1 for seven days or N) absolute number of mature ILC3s in the lung was quantified by flow cytometry.

O) Newborn B6 mice (PN7) exposed to hyperoxic environment (hyperoxia) or room air were examined for susceptibility after intratracheal inoculation with S. pneumoniae on PN8. Some newborn mice in the hyperoxia experimental group received adoptive transfer of WT ILC3s (sorted as Live CD45⁺Lin-CD127⁺(RORγt)GFP⁺ KLRG⁻ NK1.1⁻ cells from RORC^GFP/+ mice) on PN7 before intratracheal inoculation with S. pneumoniae on PN8.

P) Bronchial lavage fluid (BLF) from infants with BPD or no BPD (Control) were examined for IGF1 protein levels by ELISA or

Q) relative abundance of ILC3s (defined as live CD45⁺Lin⁻CD127⁺RORγt⁺GATA3⁻ cells) was quantified by flow cytometry (mean, interquartile range, minimum and maximum value for each group is displayed).

Data is representative of three independent experiments. Results shown as the means ± s.e.m (Student’s t-test [Fig. 7A,C,D, L,M,N,P,Q] or ANOVA [Fig. 7H,I] or Kaplan-Meier log-rank test [7O] *P ≤ 0.05; **P ≤ 0.01 and number of individual animals [n] are indicated in the figures). See also Fig. S6.

We then sought to confirm if reduced proliferative ability of pulmonary ZBTB16⁺ ILC precursors from adult mice compared to newborn mice was cell-extrinsic. CLP from newborn (PN1) CD45.1⁺ or adult (P28) CD45.2⁺ mice were co-transferred into sublethally irradiated Rag2^−/− Il2rg^−/− newborn mice (PN1) or adult mice (PN28) (Fig. 7J). The relative proportions of CD45.2⁺ or CD45.1⁺ ZBTB16⁺ ILC precursors and mature ILC3s in the lungs of either newborn or adult mice were similar (Fig. 7J) when examined two weeks later. These data lead us to invoke the ‘bad soil’ hypothesis that microenvironment in the lungs of adult mice is less supportive for expansion and maturation of ZBTB16⁺ ILC precursors compared to newborn mice. This could potentially explain the decreased number of pulmonary ILC3s in adult mice.

Bronchopulmonary dysplasia, a disorder of interrupted postnatal lung development was marked by disrupted ILC3 homeostasis.

Reciprocally, disrupted lung development during the critical developmental window of newborn period can potentially interrupt the development of pulmonary mucosal defenses and result in an increased risk of respiratory infections and inflammatory disorders like increased airway hyperreactivity in infants. This is exemplified by bronchopulmonary dysplasia (BPD), a disorder marked by interrupted lung development which is common in prematurely born infants (Jobe and Bancalari, 2001). Infants with BPD have an increased likelihood of morbidity and death due to respiratory infections (Islam et al., 2015). In a preclinical model of BPD, newborn mice challenged in a hyperoxic environment (Warner et al., 1998) demonstrate alveolar simplification, decreased alveolar density and reduced lung volumes (Cox et al., 2017), hallmarks of interrupted lung development (Whitsett and Weaver, 2015). We found that newborn mice (PN1) exposed to the hyperoxic environment had decreased number of IGF1⁺ cells in the lungs (Fig. 7K,L), reduced frequency of BrdU incorporating ZBTB16⁺ ILC precursors (Fig. 7M), demonstrated 4-fold decrease in the number of pulmonary ILC3s (Fig. 7N) and exhibited susceptibility after intratracheal challenge with S. pneumoniae (Fig. 7O). Adoptive transfer of pulmonary ILC3 from age-matched control mice via intratracheal route restored the resistance to pneumonia in newborn mice (PN3) exposed to hyperoxic environment (Fig. 7O). We recapitulated this finding in human neonates. Infants with BPD had decreased IGF1 transcripts (Fig. S6G), more than a six-fold reduced concentration of IGF-1 (Fig. 7P) and lower relative abundance of ILC3s (defined as CD45⁺Lineage [CD3, CD5,CD11b, CD11c, CD14, FcεR1]⁻CD127⁺RORγt⁺GATA3⁻ cells) (Fig. 7Q, S6G) in the bronchial lavage fluid compared to infants without BPD (Supplementary Table 3). These data, taken together, suggested that the developmental program in the newborn lung directed the pulmonary ILC3 development and regulated pulmonary defense against bacterial pneumonia. More importantly, these observations potentially explained the increased morbidity due to pneumonia in human neonates with BPD.

Discussion:

The origins of ILCs from lymphoid precursors in the fetal liver during newborn period and later in BM is well characterized (Klose et al., 2014). It is hypothesized that ILC precursors infiltrate peripheral tissues during development and subsequently differentiate and repopulate those peripheral sites as needed. Such precursors can be detected in multiple secondary lymphoid organs (Scoville et al., 2016) and blood in humans (Lim et al., 2017) and within the SI in mice (Bando et al., 2015). A recent study demonstrates that ILC2s appear in the fetal lungs during late gestation and the early newborn period. The bulk of peripheral ILC2s are generated de novo from expansion of the mature ILC2 in postnatal period (Schneider et al., 2019). Nevertheless, the extent to which the development and functional programming of pulmonary ILC3 occurs centrally, in the BM, versus peripherally, in the lung, is unresolved. Such an understanding is vital as disruption of seeding and expansion of tissue-resident precursors during the critical developmental window of the newborn period can have enduring consequences for tissue ILC3 pool and by extension, to the health of the host animal.

Using developmentally appropriate models, we demonstrated that loss of ILC precursors in early life had long-lasting effects on the pulmonary ILC3 pool. Increased susceptibility to pneumonia after an early-life loss of pulmonary ZBTB16⁺ ILC precursors prevailed beyond the neonatal period, into adulthood. Restoring pulmonary ZBTB16⁺ ILC precursors in early life replenished the pulmonary ILC3 pool and restored resistance to pneumonia. Our data suggested that the pulmonary ILC3 pool was primarily endowed by seeding in the newborn period, highlighting the significance of the neonatal period as a critical developmental window. These data could perhaps explain the observations from epidemiological studies which show that perturbations incurred during development, including premature births, are associated with increased risk of respiratory infections (Vrijlandt et al., 2013) and chronic inflammatory disorders, such as asthma, well into adolescence (Gessner and Chimonas, 2007). Whether the lungs are continuously repopulated by ILC precursors from the BM throughout the life, or whether they depend on the initial pool of ILC precursors which were seeded during the newborn period, is not clear. Such studies remain an area of active investigation in several laboratories (Kotas and Locksley, 2018).

Why growth factor signaling is differentially required for ILC development in the lungs is not yet clear. Tissue-resident ILC precursors are thought to depend upon transcription factor networks and cytokine signals for development similarly to their BM counterparts. Nevertheless, there remain important exceptions. For example, T cell-specific transcription factor (TCF)7 is critical for the development of BM ILC precursors (Yang et al., 2013). Yet, only mild defects are evident in ILCs at peripheral sites in Tcf7^−/− mice (Mielke et al., 2013). Similarly, mice deficient in transcription factor (TOX) (Tox^−/−) lack central ILC precursors (Aliahmad et al., 2010). However, the frequencies of mature ILC3 appear unchanged in the intestine of Tox^−/− mice (Seehus et al., 2015). Similarly, cytokine IL-7 is important for central ILC development. However, IL7R-independent ILC2 and ILC3 subsets are present in SI lamina propria and are fully functional as they can confer protection to IL7Rα^−/− mice against Citrobacter rodentium infection (Robinette et al., 2017), implying differential requirements in the maturation of central vs. tissue-resident precursors. Our data suggested that while pulmonary ZBTB16⁺ ILC and BM ZBTB16⁺ ILC shared a common transcriptome associated with ILC development, differences related to growth factor signaling and chemotaxis marked central vs. pulmonary ILC precursors. We posit that pulmonary ILC development requires distinct mechanisms for cell maintenance compared to the bone marrow.

Despite substantial progress, experimental perturbation of ILC precursors remains challenging (Kotas and Locksley, 2018) due to the restricted repertoire of cell surface molecules and the identification of transitional ILC developmental intermediates that require the simultaneous expression of key transcription factors, for instance, Id2, Bcl11b, and ZBTB16. (Walker et al., 2019; Xu et al., 2019). Our use of Zbtb16^iDTR mice highlighted the inadequacy of current tools for experimental manipulation of ILC precursors in the lungs. For instance, intratracheal DT treatment disrupted the homeostasis of other ILC subtypes, including ILC2s, raising the possibility that increased susceptibility to pneumonia in Zbtb16^iDTR newborn mice could be due to lack of ILC2s. A recent study demonstrated that ILC2s enforce type 2 immune environment in the newborn lung (Saluzzo et al., 2017). However, this immune quiescence contributes to a delayed response to pneumonia. We found that adoptively transferred ILC2s failed to reverse the expanded susceptibility to pneumonia in DT-treated Zbtb16^iDTR newborn mice, which suggested that ILC2s are redundant for host defense in a murine model of bacterial pneumonia.

In the BM, locally restricted production of growth factors in specialized microenvironments or niches directs the development and proliferation of hematopoietic precursors. Lymphoid tissue organizer cells, a group of mesenchyme-derived fibroblasts are essential in the development of ILC3s in SI and secondary lymphoid tissues (Cupedo et al., 2004; Hoorweg et al., 2015). More recently, an adventitial stromal cell niche was implicated in the expansion of pulmonary ILC2s (Dahlgren et al., 2019). Our data unveiled a role for IGF1 in the maturation and expansion of ILC precursors in the lungs. This was consistent with the prior reports implicating IGF1 in self-renewal, and proliferation of myeloid and erythroid progenitor cells (Soon et al., 1999) and mesenchymal stem cells (Bendall et al., 2007). IGF1 is critical for the development of T and B cells from pluripotent precursors (Murphy et al., 1992) and promotes the development of innate lymphocytes like NK cells (Ni et al., 2013). We identified a population of PDGFRα⁺ alveolar fibroblast that were a critical organizing component of the ILC3 niche in the newborn lungs (Fig. S6H). Our data, coupled with contemporaneous reports (Mahlakoiv et al., 2019; Rana et al., 2019) suggested that fibroblast-like stromal cells are crucial organizing components for tissue macrophages, lymphocytes and DCs in both health and disease.

Finally, we showed that developmentally programmed expression of IGF1 in alveolar fibroblasts, linked postnatal lung growth with expansion and maturation of pulmonary ILC3 precursors, thus coordinating the concurrent development of lung and pulmonary mucosal defenses. Why would such coordination be necessary? Since the postnatal lung is subjected to processes of growth and remodeling lasting for a few weeks in mice to years in humans (Herriges and Morrisey, 2014), contemporaneous development can allow the developing lung to ‘instruct’ rapidly maturing pulmonary mucosal defenses (Lloyd and Marsland, 2017). This convergence in the timing of lung growth and immune maturation could ensure robust protection against respiratory pathogens while restricting damage to the organ’s critical gas exchange machinery. Our data potentially explain why perturbations incurred during development, for instance, premature birth and bronchopulmonary dysplasia are associated with increased risk of respiratory infections and chronic inflammatory disorders, for example, asthma, which persist beyond infancy into adolescence. Understanding how organ-specific cues instruct the ILC3 development may provide the framework to develop enduring lung protective strategies for infants and children.

STAR Methods:

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Hitesh Deshmukh (hitesh.deshmukh@cchmc.org). Il22^fl/fl used this study were obtained under an MTA agreement with mutant mouse resource and research center (MMRRC) that does not allow for distribution of lines generated. We will provide the Il22^fl/fl mice to requesting labs that have received approval for the use of these mice via an MTA with MMRRC.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice:

Institutional Animal Care and Use Committee at Cincinnati Children’s Hospital Medical Center (CCHMC) approved all the animal studies (IACUC2014–0055), which were carried out in accordance with NIH’s Guide for the Care and Use of Laboratory Animals. We obtained Zbtb6^CreGFP (Stock # 024529) (Constantinides et al., 2014), Rorc^CreGFP (Stock # 022971) (Eberl and Littman, 2004), Rosa26^iDTR (Stock #007900)(Buch et al., 2005), Gli1^CreER (Stock # 007913)(Ahn and Joyner, 2004), Igf1^fl/fl (Stock # 016831)(Liu et al., 1998), Igf1r^fl/fl (Stock # 012251)(Dietrich et al., 2000), Rag2^−/−;Il2rg^−/−(Stock # 014593)(Song et al., 2010) and B6 Cd45.1, Pep/BoyJ (Stock# 002014) from Jackson Laboratory. We obtained Il22^fl/fl (Stock # 032402) from MMRRC. We bred Zbtb16^Cre or Rorc^Cre mice with Rosa26^iDTR mice to generate Zbtb16^iDTR or Rorc^iDTR mice respectively. We bred Rorc^Cre mice with Il22^fl/fl mice to generate Rorc^ΔIl22 mice. We bred Gli1^CreER mice with Igf1^fl/fl mice to generate Gli1^CreER; Igf1^fl/fl mice. We bred Zbtb16^Cre or Rorc^Cre mice with Igf1r^fl/fl mice to generate Zbtb16^ΔIgfr or Rorc^ΔIgfr1mice, respectively. We backcrossed Zbtb16^Cre or Rorc^GFP mice with Pep/BoyJ mice to generate Zbtb16^Cre or Rorc^Cre on CD45.1 background. Zbtb16^Cre or Rorc^GFP mice on CD45.1 background were used to isolate ILC precursors, ILC2s or ILC3s for adoptive transfers. We maintained B6, Rosa26^iDTR, Zbtb16^Cre, Zbtb16^iDTR, Gli1^CreER, Igf1^fl/fl, Igf1r^fl/fl, Il22^fl/fl, Rag2^−/−;Il2rg^−/−mice, B6 Cd45.1, Pep/BoyJ mice and Zbtb16^Cre or Rorc^GFP on CD45.1 background at specific pathogen free (SPF) animal facility at CCHMC. We maintained the germ-free (GF) B6 mice in plastic isolator cages with autoclaved feed and water at CCHMC Germ-Free Core facility. We used male and female B6, Rosa26^iDTR, Zbtb16^CreGFP, Zbtb16^DTR, Gli1^CreER, Igf1^l/fl, Gli1^CreER; Igf1r^fl/fl, Zbtb16^ΔIgfr1, B6 CD45.1, Pep/BoyJ or GF B6 mice between ages PN1 and PN28 and appropriate, age- and genetic strain-matched controls to account for any variations in data. We determined group sizes necessary for adequate statistical power analysis using preliminary data sets. The mice were selected at random from the cohort but not formally randomized. We did not exclude any samples. The investigators were not blinded to group allocation during the collection and analysis of the data, due to the nature of the experiments.

Human samples:

The Institutional review board at CCHMC approved all the human studies. For the bronchopulmonary dysplasia (BPD) group, the biological samples were collected from prematurely born infants diagnosed with bronchopulmonary dysplasia (defined as need for supplemental oxygen or respiratory support beyond 28 days of life). These infants underwent bronchoscopy as part of routine clinical evaluation for ventilator dependence and respiratory failure. For control group, biological samples were collected from infants who underwent clinical evaluation of either upper airway obstruction or trachea bronchomalacia, common anomalies of the airways in infants. The samples were collected after obtaining informed consent from the parent (s) (IRB approval #2013–3309). Characteristics of the subjects, including the gestational age, sex and comorbidities are provided in Supplementary Table 3.

Bacteria:

S. pneumoniae serotype 19A was obtained from (ATCC 700674). We prepared large number of master stocks to limit frequent passage. Virulence was periodically validated to ensure the consistency of the results.

Cell lines:

Delta like 1 over expressing OP9 cell line was a gift from Dr. J. C. Zúñiga-Pflücker. Identity of specific sex from which these cells were derived is unknown.

METHOD DETAILS

Murine tamoxifen treatments:

Male or female newborn (PN1) Gli1^CreER or Igf1^fl/fl or Gli1^CreER; Igf1^fl/fl mice were treated with tamoxifen (200 microg g⁻¹) (Sigma) or sham via the intraperitoneal.

Intratracheal delivery of diphtheria toxin:

Male or female newborn (PN3) Rorc^iDTR or Zbtb16^DTR or Rosa26^iDTR or Zbtb16^Cre or Rorc^Cre mice were treated with diphtheria toxin (DT) (10 ng g⁻¹) (Sigma) or sham via intratracheal route as follows. We first anesthetized the newborn mice with isoflurane (3% isoflurane/oxygen v/v mixture, 90–120 seconds) and confirmed the plane of anesthesia by toe pinching. Next, we aspirated 50 μl of air into a 100 μl gas-tight precision syringe (Hamilton Company, Reno, NV) by withdrawing the Teflon plunger of the syringe. Next, 50 μl of diphtheria toxin was aspirated into the syringe, by advancing the Teflon plunger from 50 μl mark to the 100 μl mark on the syringe body. Thus, when the sample was injected into an intubated newborn mouse, the 50 μl DT suspension was delivered first, followed by a 50 μl air cushion which distributed the DT suspension throughout the lung and ensured minimal dispersion into the esophagus. The syringe was then fitted with 24 or 26-gauge, 19 mm long intravascular catheter (BD Insyte). The anesthetized newborn mouse was secured on intubation platform with Velcro strips and the platform was inclined to 45°. This position resulted in slightly hyperextension of the neck and allowed optimal visualization of vocal cords. Then, we retracted the tongue with a rolling motion, using a micro cotton applicator tip with the dominant hand. An operating otoscope (Welch Allyn, #21700) fitted with 2 mm speculum (Welch Allyn, #22002) was gently introduced into oropharynx using the nondominant hand while maintaining tongue retraction. When introduced appropriately, the sides of the 2 mm speculum make a seal with the tracheal inlet. The glottis and the vocal cords were directly visualized under the magnification through the operating otoscope. The vocal cords appear shiny white and move with each breath. Extending the neck slightly further improved the visualization of vocal cords. The gas-tight precision syringe preloaded with DT and fitted with 24-gauge intravascular catheter was then gently introduced through the speculum with the dominant hand. The 24-gauge intravascular catheter was advanced through the vocal cords under direct visualization using the side of the speculum as a guide. We injected the liquid/air bolus into the trachea in one fluid motion by depressing the plunger with dominant hand and immediately withdrew the catheter and the speculum. The delivery of the liquid was verified by rhonchorous sounds made by the newborn mouse when breathing until the fluid was distributed fully in the lungs. The time taken for the procedure was around 3 minutes with 90–120 seconds to induce anesthesia with isoflurane and 30–45 seconds to identify and intubate the airways. The procedure was done by operators fluent in intubation and iv cannulation of extremely premature infants. To confirm preferential delivery of contents to the lungs we used 50 μl trypan blue solution in place of DT solution.

Fifteen minutes post injection, the animals were euthanized and presence or absence of trypan blue in stomach and the lung was confirmed by direct visualization (Supplementary Fig. 3A–H, Supplementary Video 5).

For infectious studies, we grew S. pneumoniae serotype 19A (ATCC 700674) with gentle aeration (37°C, 200 rpm) in tryptic soy (TS) broth to log-phase growth. Male or female mice were inoculated with S. pneumoniae (10³ CFU g⁻¹) or sham via the intratracheal route on PN5 or PN28. The animals were examined every six hours (h) for signs of distress and were euthanized 120 h later or earlier if moribund. For in vivo proliferation studies, age-defined newborn mice were treated with BrdU (Sigma) (500 micro g g⁻¹) via the intraperitoneal route.

Murine model of BPD:

For hyperoxia studies, male or female newborn B6 mice were continuously exposed to oxygen concentration ≥90% O₂ or 21% (room air) from birth up to PN7 as done previously (Warner et al., 1998). Nursing dams were rotated between groups every 48 hr to prevent oxygen toxicity to the dams. The newborn mice were allowed to recover in room air for 1 day and inoculated S. pneumoniae (10³ CFU g⁻¹) via the intratracheal route on PN8.

Preparation of single cell suspensions:

Male or female age-defined mice were anesthetized with ketamine/xylazine and 1 ml of PBS (Sigma-Aldrich) was injected into the right ventricle to perfuse the lung tissue. To isolate lymphocytes from the lungs, lung tissues from 4–6 newborn mice were pooled together, diced and incubated (37 °C, 30 min) with shaking (150 r.p.m.) in digestion buffer (RPMI 1640 with 10% FBS, 15 mM HEPES, 1% penicillin/streptomycin (wt/vol) and 300 U ml⁻¹ collagenase VIII), then pressed through a 100-μm nylon strainer and centrifuged (4°C, 400 × g, 5 min). The cells were resuspended in RPMI 1640 with 10% FCS. The pooled preparations or single cells constituted a single data point in our analysis. To isolate SI lymphocytes, the SI from 4 newborn mice were pooled together, flushed with PBS, diced and incubated (37 °C, 30 min) with shaking (150 r.p.m.) in digestion buffer (RPMI 1640 with 10% FBS, 1 mM EDTA, 1 mM DTT, 1% penicillin/streptomycin (wt/vol), 0.2 mg ml⁻¹ DNase I and 300 U ml⁻¹ collagenase VIII) and pressed through a 100-μm nylon strainer and centrifuged (4°C, 400 × g, 5 min). Freshly isolated livers or femurs and tibias from 4 newborn mice were pooled together. Liver tissue was gently dissociated by mechanical agitation and bone marrow was released by gently crushing femurs and tibias. The resulting single cell suspension was pressed through 100 μm nylon strainer. The pooled preparations of liver or bone marrow cells constituted a single data point in our analysis.

Flow cytometry:

The cells were pooled from male or female six PN3 newborn mice. Cells (1 × 10⁷) were incubated (4°C, 30min) with anti-mouse CD16/CD32 (93) to block Fc receptors. The cells were re-incubated (4 °C, 30 min) with anti-mouse CD3ε antibody (145–2C11), anti-mouse CD5 antibody (53–7.3), anti–mouse CD8 antibody (53–5.8), anti–mouse CD11b antibody (M1/70), anti-mouse CD11c (N418), anti–mouse CD19 antibody (1D3), anti-mouse CD25 antibody (PC61.5), anti-mouse CD45.1 (A20), anti-mouse CD45.2 (104), anti-mouse CD117 antibody (ACK2), anti-mouse Ly6G antibody (1A8), anti-mouse F4/80 antibody (BM8), anti-mouse FLT3 antibody (A2F10), anti-mouse α4β7 antibody (DATK32), anti-mouse NKp46 antibody (29A1.4), anti-mouse CCR6 antibody (29–2L17), anti-mouse KLRG1 antibody (2FI) (all diluted 1:100), anti-mouse NK1.1 (PK136) (diluted 1:50), anti-mouse PD1 antibody (J43) and anti–mouse CD127 antibody (A7R34) (diluted 1:50). For intracellular staining, we washed and fixed (4 °C, 60 min) the surface-stained cells in 1X Cytofix/Cytoperm buffer (BD Biosciences) and permeabilized them (4 °C, overnight) using 1X Permeabilization Buffer (BD Biosciences) according to manufacturer instructions. We stained the cells intracellularly with anti–mouse RORγt antibody (Q31–378), anti-mouse T-bet antibody (4B10), anti-mouse ZBTB16 antibody (R17–809), anti-mouse TCF7 antibody (812145), anti-mouse GATA3 antibody (16E10A23), anti-mouse Id2 antibody (17-9475-82) anti-mouse Eomes antibody (TBR2) (all diluted 1:50) and anti-mouse BrdU antibody (Bu20a) (diluted 1:100) and then washed (2×) and resuspended them in flow cytometry buffer. We collected the data with LSRII (BD Biosciences) and analyzed the data with FlowJo (Treestar). Antibodies were obtained from either eBioscience or Biolegend or Bioxcell. CLP were identified as CD45⁺Lin⁻CD25⁻ICOS⁻CD117⁺Sca-1-CD127⁺α4β7⁻CD135⁺Id2⁻PD1⁻RORγt⁻GATA3⁻ cells. ZBTB16⁺ ILC precursors (ILCP) were identified as CD45⁺Lin⁻CD25⁻ICOS⁻CD127⁺α4β7⁺CD135⁻CD25⁻Id2⁺PD1⁺GATA3⁻RORγt-ZBTB16⁺ cells. ILC1 were identified as CD45⁺Lin⁻CD127⁺RORγt⁻GATA3⁻NK1.1⁺NKp46⁺ cells. ILC2 were identified as CD45⁺Lin⁻CD127⁺ RORγt⁻GATA3⁺ cells. ILC3s were identified as CD45⁺Lin⁻CD127⁺ RORγt⁺GATA3⁻ cells. Lineage panel included CD3ε, CD5, F4/80, CD11b, CD19 and Ly6G.

Isolation of pulmonary ZBTB16⁺ ILC precursors:

We pooled lungs from 10–12 (P1) newborn ZBTB16^CreGFP mice (on CD45.1 background). Single cells were stained with 7-AAD Viability staining solution (Biolegend), anti-mouse CD3ε antibody (145–2C11), anti–mouse CD4 antibody (GK1.5), anti-mouse CD8 antibody (53–5.8), anti–mouse CD11b antibody (M1/70), anti-mouse CD11c (N418), anti–mouse CD19 antibody (1D3), anti-mouse B220 (2FI), anti-mouse LY6G antibody (1A8), anti-mouse NKp46 antibody (29A1.4), anti-mouse α4β7 antibody (DATK32), anti-mouse CCR6 antibody (29–2L17), anti–mouse CD127 antibody (A7R34) and anti-mouse KLRG1 antibody (2FI) (all diluted 1:100, Biolegend). The stained cells were sorted with Sony SH800S. ZBTB16⁺ ILC precursors were sorted as Live CD45⁺Lineage (CD3ε, CD4, CD8, CD11b, CD11c, CD19, B220 and Ly6G), CD127⁺α4β7⁺NKp46⁻KLRG1⁻CCR6⁻CD25⁻PD1⁺GFP⁺ cells. Using this protocol, we isolated an enriched population of 800–1000 ZBTB16⁺ ILC precursors. We confirmed the purity of the sorted population by intracellular staining with anti-mouse RORγt antibody (Q31–378), anti-mouse T-bet antibody (4B10), anti-mouse PD1 antibody (J43), anti-mouse Id2 antibody (17-9475-82) and anti-mouse GATA3 antibody (16E10A23) (all diluted 1:50) (Supplementary Fig. 5 A, B). The cells were resuspended in normal saline to final concentration of 1×10³ cells/100 μl. We adoptively transferred 1×10³ ZBTB16⁺ ILC precursors via ntratracheal route.

Isolation of pulmonary ILC2 and ILC3:

We pooled lungs from 8–10 (P1) newborn RORC^GFP/+ mice (on CD45.1 background). We used a sorting strategy described before (Chu et al., 2018). Single cells were stained with 7-AAD Viability staining solution (Biolegend), anti-mouse CD3ε antibody (145–2C11), anti–mouse CD5 antibody (53–7.3), anti-mouse FcεR1 antibody (Mar-1), anti–mouse B220 antibody (RA3–6B2), anti–mouse CD11b antibody (M1/70), anti-mouse CD11c (N418), anti–mouse CD19 antibody (1D3), anti–mouse CD127 antibody (A7R34), anti-mouse KLRG1 (2FI), anti-mouse Nk1.1 antibody (PK136) (all diluted 1:100). The stained cells were sorted with Sony SH800S. ILC2 were sorted as Live CD45⁺Lineage (CD3ε, CD5, CD11b, CD11c, FcεR1 and B220) CD127⁺(RORγt)GFP⁻KLRG⁺ NK1.1⁻cells. Using this protocol, we isolated approximately 20,000 ILC2s. ILC3s were sorted as Live CD45⁺Lineage (CD3ε, CD5, CD11b, CD11c, FcεR1 and B220) CD127⁺(RORγt)GFP⁺ KLRG⁻ NK1.1⁻ cells. Using this protocol, we isolated approximately 8000–10,000 ILC3s. We confirmed the purity of the sorted population by intracellular staining with anti-mouse RORγt antibody (Q31–378) and anti-mouse GATA3 antibody (16E10A23) (all diluted 1:50) (Supplementary Fig. 4A). The cells were resuspended in normal saline to final concentration of 1×10⁴ cells/100 μl. We adoptively transferred 5×10³ ILC3s or 10⁴ ILC2s via intratracheal route.

Alveolar aggregate assay:

We generated alveolar aggregate as described previously(Frank et al., 2016) with modifications. We pooled and cut the freshly isolated lungs from 3–4 age-defined mice (P3 or p28) mice and incubated (37 °C, 30 min) the cut tissues with shaking (150 r.p.m.) in digestion buffer (RPMI 1640 with 10% FBS, 15 mM HEPES, 1% penicillin/streptomycin (wt/vol) and 300 U ml⁻¹ collagenase VIII) and pressed through a 100-μm nylon strainer to obtain single-cell suspension. We incubated (25°C, 1 h) the cells (1 × 10⁷) with anti-CD131(clone REA193, 1:10), anti-CD45.2 (V450, 1:100), anti-CD140 (APA5, 1:10), anti-CD326 antibody (caa7–9G8, 1:10) (all from BD biosciences) and anti- CD31 (MEC13.3, 1:100, BD Biosciences). The cells were washed, resuspended in FACS buffer supplemented with 5% FBS and sorted using FACSAria II (BD Biosciences) cell sorter into either epithelial cells (defined as CD45⁻CD131⁺CD146⁻CD310⁻) or fibroblast (CD45⁻CD131⁻CD146⁺CD310⁻). To isolate ZBTB16⁺ ILC precursors, we pooled and cut the freshly isolated lungs from newborn (P3) ZBTB16^CreGFP mice and incubated (37 °C, 30 min) the cut tissues with shaking (150 r.p.m.) in digestion buffer (RPMI 1640 with 10% FBS, 15 mM HEPES, 1% penicillin/streptomycin (wt/vol) and 300 U ml⁻¹ collagenase VIII) and pressed through a 100-μm nylon strainer to obtain single-cell suspension. The cells were washed, then stained as described earlier. ZBTB16⁺ ILC precursors were sorted as Live CD45⁺Lineage (CD3ε, CD4, CD8, CD11b, CD11c, CD19, B220 and Ly6G), CD127⁺α4β7⁺NKp46⁻KLRG1⁻CCR6⁻CD25⁻PD1⁺GFP⁺ cells.

Acellular matrix composed of Matrigel and serum-free growth medium was prepared as described before (Frank et al., 2016). One hundred micro L of 1:1 mixture of Matrigel (growth factor reduced, phenol-red free) (Corning) and serum-free epithelial cell growth medium [DMEM:F12 supplemented with murine EGF (25 ng ml⁻¹), bovine pituitary extract (30 micro g ml⁻¹), insulin (5 micro g ml⁻¹), transferrin (5 micro g ml⁻¹), retinoic acid (15 ng ml⁻¹) and 0.5% Amphotericin (wt/vol)] was pipetted in 6.5 mm transwell with 3.0 μm pore polycarbonate membrane insert (Corning). After solidification, the cellular matrix consisting of epithelial cells (10⁵) and fibroblasts (10⁵) suspended in epithelial cell growth medium was layered on top of the acellular matrix and the insert was placed in 24 well transwell plate (Corning). Serum-free epithelial cell growth medium [DMEM:F12 with EGF, bovine pituitary extract, insulin, transferrin, retinoic acid and Amphotericin)] supplemented with Rho-associated, coiled-coil containing protein kinase (ROCK) inhibitor, Y-27632 (10 uM) (ThermoFischer) was added to the bottom of the well and cultures were incubated (37°C, 5% CO²) for 24 hr to stabilize the matrix. ZBTB16⁺ ILC precursors (10²) suspended in DMEM:F12 were layered on top of the matrix. The medium in the bottom of the well was replaced with serum free epithelial growth medium [DMEM:F12 with EGF, bovine pituitary extract, insulin, transferrin, retinoic acid and Amphotericin)] containing murine IL-7 (25 ng ml⁻¹) (R&D) and stem cell factor (25 ng ml⁻¹) (BioLegend) with or without IL-23 or IL-33 (both 20 ng ml⁻¹) (R&D). The aggregates were incubated 37°C, 5% CO²) for 96 hrs. BrdU (10 micro M) was added to the culture medium 2 hr prior to harvest. The aggregates were incubated with 1X Matrisperse Cell recovery solution (4 °C, 60 min) (BD Biosciences) to liberate cells. The cells were washed, resuspended in FACS buffer and number of ZBTB16⁺ ILC precursors and mature ILC3s were quantified by flow cytometry. Conditioned medium was collected from the bottom of the well and snap frozen for further analysis.

In complementation studies, the cellular matrix consisting of epithelial cells (10⁵) alone (without any fibroblast) was layered on top of the acellular matrix. Twenty-four hours later ZBTB16⁺ ILC precursors (10²) suspended in DMEM:F12 were layered on top and the aggregates were cultured as described in the above section. Conditioned medium was collected from the bottom of the well and snap frozen for further analysis and number of ZBTB16⁺ ILC precursors and mature ILC3s were quantified by flow cytometry as before.

In studies involving neutralizing antibodies, alveolar aggregates were incubated with either anti-IGF1 antibody (0.5 micro g ml⁻¹) (Santa Cruz, SC 461) or anti-IL33R antibody (1 micro g ml⁻¹) (R&D Biosystems, AF 1004) or anti-IgG antibody (0.5 micro g ml⁻¹). The above reagents were added at the time of the first media change (after removal of ROCK inhibitor). For OP9-DL1 cultures, OP9-DL1 stromal cells were incubated (37°C, 5% CO²) with Opti-MEM (supplemented with 5% FBS, 1 % penicillin/streptomycin (wt/vol) and 60 micro M 2-mercaptoethanol) to 70% confluence. ZBTB16+ ILC precursors (10²) were co-incubated (37°C, 5% CO², 96 hrs) with OP9-DL1 cells in presence of murine IL-7 (25 ng ml⁻¹) (R&D), stem cell factor (25 ng ml⁻¹) (BioLegend) with or without IL-23 or IL-33 (both 20 ng ml⁻¹) (R&D) as described before (Constantinides et al., 2014). BrdU (10 micro M) was added to the culture medium 2 hr prior to harvest. The cells were incubated with 1X cell recovery solution (4 °C, 60 min) (BD Biosciences) followed by incubation (4 °C, 60 min) with 0.25% Trypsin-EDTA to liberate cells.

The cells were washed, resuspended in FACS buffer and number of ZBTB16+ ILC precursors and mature ILC3s were quantified by flow cytometry.

Competitive in vivo expansion assay:

To compare the proliferative potential of ILC precursors, we isolated CLP from the newborn liver by negative and positive selection as described before (Sitnicka et al., 2002). Briefly, pooled liver tissue > 10 newborn mice was gently dissociated by mechanical agitation and strained through a 100-μm nylon filter to obtain single-cell suspension. Cells (1 × 10¹⁰) were incubated (4°C, 30 min) with anti-mouse CD16/CD32 (93) to block Fc receptors, then re-incubated with lineage cocktail [anti-mouse CD3ε antibody (145–2C11), anti–mouse CD4 antibody (GK1.5), anti–mouse CD8 antibody (53–5.8), anti–mouse CD11b antibody (M1/70), anti-mouse CD11c (N418), anti–mouse CD19 antibody (1D3), anti-mouse Ly6G antibody (1A8), anti-mouse F4/80 antibody (BM8)], anti-mouse CD117 antibody (ACK2), anti–mouse CD127 antibody (A7R34), anti-mouse FLT3 antibody (A2F10), anti-mouse α4β7 antibody (DATK32) and anti-mouse Sca1 (LY-6A/E) (D7) (all diluted 1:100, Biolegend). The cells were washed, resuspended in FACS buffer supplemented with 5% FBS and sorted using FACSAria II cell sorter into CLP, defined as CD45⁺Lin⁻CD25⁻ICOS⁻CD117⁺Sca-1⁻CD127⁺α4β7-CD135⁺Id2⁻PD1⁻RORγt⁻GATA3⁻ cells. This protocol yielded approximately 3×10 CLP⁵ cells. CLP (10⁵ cells) isolated from either newborn (P1) ZBTB16^ΔIgfr1 or ZBTB16^Cre (CD45.2⁺) mice or CD45.1⁺ mice were mixed in ratio of 1:1 and injected via intrahepatic injection into age-matched newborn (P1) Rag2^−/−; Il2rg^−/− mice that were sublethally irradiated (4.2 Gy) on the same day as done previously (Lopez-Lastra et al., 2017). Two weeks post-reconstitution, single cell suspensions were isolated from various organs (lungs, SI, liver and BM) of the mixed chimeric animals. Number of CLPs, ZBTB16⁺ ILC precursors and mature ILC3s were quantified by flow cytometry and normalized to the reconstitution ratio of each chimera, as determined by the CD45.2⁺/CD45.1⁺ ratio of hepatic or BM lymphocytes.

To compare the proliferative potential of ILC precursors from newborn (P1) or adult mice, CLP (10⁵ cells) isolated from either newborn (P1) (CD45.2⁺) mice or adult CD45.1⁺ mice were mixed in ratio of 1:1 and injected via intrahepatic injection into either newborn (P1) or adult Rag2^−/−; Il2rg^−/− mice that were sublethally irradiated (4.2 Gy) on the same day. Two weeks post-reconstitution, number of CLP, ZBTB16⁺ ILC precursors and mature ILC3s were quantified by flow cytometry.

Histology, cell counts and spatial distance quantification:

Age-defined Zbtb16^CreGFP mice or B6 mice were anesthetized with ketamine/xylazine and 1 ml of PBS was injected into the right ventricle to perfuse the lung tissue. Lungs were inflation fixed with 2% paraformaldehyde, then dehydrated, paraffin embedded and sectioned into 20 micro M thick slices. Sections were subjected to heat mediated antigen retrieval with use of citrate buffer in microwave. Following antigen retrieval, the sections were incubated (4°C, overnight) with chicken anti-mouse GFP (1:500) (ThermoFischer, A10262), rabbit anti-mouse NKX2.1 (1:1000) (Seven Hills Bioreagents, 1231), mouse anti-mouse PDGFRα (1:100) (ThermoFischer, APA5), hamster anti-mouse IGF1 (1:100) (Millipore, Sm1.2) and DAPI. The immunostained lung tissues were imaged by Nikon Eclipse confocal microscope in 1-μm optical sections. Z-stacks were obtained from at least five random areas of each lung tissue and from a minimum of n = 3 mice. Number of IGF1⁺ cells were determined with Cell Counter plug-in for ImageJ. Cells were counted in at least three different areas for each mouse, to obtain a total count of >500 cells at each postnatal time point (P1, P3, P7, P14, P21 and P28). Spatial distance was quantified using Imaris software (Bitplane). Distance vectors, calculated in Imaris software, were drawn from the ZBTB16⁺ cell to the nearest NKX2.1⁺ cell or PDGFRα⁺ cell with the Xtension tool for >500 cell pairs at each postnatal time point (P1, P3, P7, P14, P21 and P28) by a blinded researcher as described previously. For image presentation, images obtained with automatic deconvolution algorithms in Nikon Elements software are presented.

Stable isotope labeling of amino acids and quantitative assessment of proteins in the conditioned medium:

For quantitative assessment of secreted proteins in the conditioned medium, the alveolar aggregate assay performed with or without fibroblast in presence of serum free epithelial growth medium with DMEM:F12 containing either heavy ¹³C₆ L-Lysine-2HCl or non-heavy ¹²C₆ L-Lysine-2HCl (ThermoFischer). Incorporation of heavy ¹³C₆ L-Lysine in alveolar aggregates at time of harvest > 85%. The conditioned medium was centrifuged (4°C, 1500 × g, 10 min) to eliminate cellular debris and proteins were precipitated with acetone (−20°C, 12 hr), centrifuged (4°C, 10000 × g, 1 hr). The pellet was air dried (25°C, 30 min) and suspended in RIPA buffer. The proteins were separated by 4–12% Bis-Tris gels with MOPS running buffer and stained with coomassie blue. After destaining, each gel lane was cut into 20 bands, dehydrated and reduced (56°C, 30 min) with acetonitrile and dithiothreitol (10 mM) respectively and then incubated (25°C, overnight) with 1% trypsin in 10 mm ammonium bicarbonate and 10% (vol/vol) acetonitrile. Digested peptides were centrifuged (4°C, 10000 × g, 1 hr) and resuspended in 0.1% (vol/vol) formic acid for LC-MS/MS analysis. Digested peptides were analyzed with LC–nanospray-MS/MS with LTQ-OrbiTrap XL linear ion trap mass spectrometer (ThermoFischer). Peptide samples were desalted on RP trap columns (Zorbax 300 SB C18, Agilent Technologies, and separated on an RP column (150 μm, 150 mm length, New Objective). The mobile phase A was 0.1% formic acid in HPLC-grade water, and the mobile phase B was 0.1% formic acid in acetonitrile. The peptides were loaded on the columns to separate with a flow rate of 2 μl/min by using a linear gradient of 4%–50% B for 110 min, 50%–100% B from 110 min to 115 min and sustained at 100% B for 5 min. Data-dependent MS/MS spectra were obtained simultaneously. Each scan cycle consisted of one full MS scan in profile mode followed by seven MS/MS scans in centroid mode.

MaxQuant (version 1.5.1.2) was used for protein identification and quantitation(Tyanova et al., 2016a). Mass spectrometry data were searched against Ensembl version GRCm38_71 (released April, 2013) using Andromeda(Cox et al., 2011). Carbamidomethyl (C) was selected as a fixed modification, and oxidation (M), acetyl (protein N-term) and ¹³C₆-Lys labeling were set as the variable modification. Initial peptide mass tolerance was set to 7 ppm and fragment mass tolerance was 0.5 Da. At least 1 unique peptide per protein group and minimum peptide length of 7 amino acids was required for identification. The maximum false discovery rate (FDR) cutoff of 0.01 (1%) was set for both the peptide spectra matches and the protein group levels. Abundance estimation of the identified proteins was performed using intensity-based absolute quantification (iBAQ) values with Perseus (version1.5.8.5) as described before (Tyanova et al., 2016b). 271 proteins were identified in CM from alveolar aggregates with fibroblast, of which 185 were quantified by ¹³C₆-Lys labeling. 192 proteins were identified in CM from alveolar aggregates without fibroblast, of which 123 were quantified by ¹³C₆-Lys labeling. A total of 92 proteins were both uniquely identified and quantified in CM from alveolar aggregates with fibroblast and alveolar aggregates without fibroblast. A two-tailed unpaired t-test was used to identify differentially abundant proteins between two experimental conditions (alveolar aggregates with or without fibroblast). Of 92, 34 proteins were differentially abundant (defined as > log₂ fold change and p value <0.05) in CM from alveolar aggregates with fibroblast, compared to CM from alveolar aggregates without fibroblast. Gene Ontology (GO) enrichment analysis was performed using the ToppGene Suite (http://toppgene.cchmc.org/).

Human studies:

The bronchial lavage (BAL) fluid samples were centrifuged (4°C, 10 min, 400 g). The resultant supernatant was frozen (−80 °C) and the cells were cryopreserved (−150 °C) in 90% FBS and 10% DMSO. We used the frozen supernatant and cryopreserved cells in our analysis (IRB approval #2015–7983). Characteristics of the subjects are provided in Table S3. After thawing, we incubated (4°C, 30 min) the cells (0.5 × 10⁵) with anti-human CD3 (UCHT1), anti-human CD5 (UCHT2), anti-human CD11c (S-HCL-3), anti-human CD11b (M1/70), anti-human CD19 (HIB19), anti-human FcεRI (AER-37), anti-human CD34 (581), anti-human CD14 (HCD14), anti-human CD45 (HI30), anti-human CD127 (A019D5) and isotype-matched control antibodies (MOPC-173) (all diluted 1:100). The cells were then treated with 10 ng/ml phorbol 12-myristate 13-acteate (PMA) and 500 ng/ml calcium ionophore A23187 (both from Sigma-Aldrich) (4 °C, 60 min). The cells were then washed and fixed (4 °C, 60 min) in 1X Cytofix/Cytoperm buffer (BD Biosciences) and permeabilized (4 °C, overnight) using 1X Permeabilization Buffer (BD Biosciences) according to manufacturer instructions. The permeabilized cells were incubated with anti-human GATA-3 (TWAJ), anti-human RORγT (AFKJS-9) and anti-human IL-22 antibody (22URTI) (all 1:100, all eBiosciences) as described before (De Grove et al., 2016).

Transcriptome analysis:

The full data sets for murine single-cell RNA-seq experiments are available in the NHLBI LungMAP (Ardini-Poleske et al., 2017) web resource (https://lungmap.net/breath-download-page/) or can be downloaded from NCBI Sequence Read Archive (Accession # GSE122332). For single cell transcriptome analysis, lungs from newborn mice were dissociated by mechanical agitation. The resulting single-cell suspension was used to perform Drop-seq as described previously(Macosko et al., 2015). Drop-seq library was sequenced using Illumina HiSeq. The paired-end reads were aligned to mouse genome build GRCm38/mm10 using HISAT. Drop-seq tools (Version 1.12) (http://mccarrolllab.com/download/922/version 1.12) was used to generate digital expression matrix as done previously(Shekhar et al., 2016). The expression matrix was processed with Seurat (version 2.2) to identify major cell types and perform t-Distributed Stochastic Neighbor Embedding (tSNE) analysis as described before(Butler et al., 2018). t-SNE gene expression overlays were carried out using the FeaturePlot function in Seurat. These data can be accessed publicly at Gene Expression Omibus Database (GSE122332)

Transcriptome analysis of lung vs. bone marrow ZBTB16⁺ ILC precursors:

We pooled lungs or bone marrow from 10–12 (P3) newborn ZBTB16^CreGFP mice. Single cells were stained with 7-AAD Viability staining solution (Biolegend), anti-mouse CD3ε antibody (145–2C11), anti–mouse CD4 antibody (GK1.5), anti-mouse CD8 antibody (53–5.8), anti–mouse CD11b antibody (M1/70), anti-mouse CD11c (N418), anti–mouse CD19 antibody (1D3), anti-mouse B220 (2FI), anti-mouse LY6G antibody (1A8), anti-mouse NKp46 antibody (29A1.4), anti-mouse α4β7 antibody (DATK32), anti-mouse CCR6 antibody (29–2L17), anti–mouse CD127 antibody (A7R34) and anti-mouse KLRG1 antibody (2FI) (all diluted 1:100, Biolegend). The stained cells were sorted with Sony SH800S. ZBTB16⁺ ILC precursors were sorted as Live CD45⁺Lineage (CD3ε, CD4, CD8, CD11b, CD11c, CD19, B220 and Ly6G), CD127⁺α4β7⁺NKp46⁻KLRG1⁻CCR6⁻CD25⁻PD1⁺GFP⁺ cells. Using this protocol, we isolated an enriched population of 800–1000 ZBTB16⁺ ILC precursors. Individually indexed libraries were generated from approximately 5000 ZBTB16⁺ ILC precursors from respective tissue using the Illumina TruSeq Stranded Total RNA kit. The libraries were sequenced with Illumina HiSeq lanes, yielding over 250 million 100bp single-end reads. Reads were mapped to the mm10 genome using HISAT2 with default parameters, and gene counts were obtained with Seqmonk (version 1.45.1). Putative genes, pseudo-genes, allosomal genes, and genes with fewer than 5 counts per million (CPM) were excluded from analyses. Gene counts were normalized using edgeR (Robinson and Oshlack, 2010) and then transformed to log2-CPM and modeled as a function of genotype using the limma package to determine differential expression across conditions(Ritchie et al., 2015). Results were corrected for multiple testing using the Benjamini and Hochberg method to produce FDR-adjusted p values. These data can be accessed publicly at NCBI Short Read Archive (Accession # PRJNA599444).

For analysis of Igf1 gene expression, we used previously published RNAseq datasets (Endale et al., 2017a),(Endale et al., 2017b). In this study, high-quality RNA isolated from 10⁵ sorted PDGFRα⁺ alveolar fibroblast from fetal (E18.5), newborn (P7, P14) or adult (P21) mice was sequenced using Illumina HiSeq 2500. The raw sequencing reads were aligned to mouse genome build GRCm38/mm10 using HISAT. Seqmonk (version 1.41.0) was used to summarize and normalize counts, which were expressed as reads per kilobase per million mapped reads.

For analysis of IGF1 gene expression in lungs of human infants, we used previously published RNAseq datasets (Du et al., 2017). The study does not qualify as “human subjects” research, as confirmed by the CCHMC IRB. Lungs from brain-dead (deceased) neonates (age <28 days) (n=4), children (age > 1 year and <10 years) (n=6) and adult (age>20 years) (n=1) obtained from Lung MAP consortium were dissociated by enzymatic digestion to obtain single cell suspension and then enriched by flow cytometry. High-quality RNA isolated from sorted epithelial (CD45-CD326⁺CD31⁻CD144⁻), endothelial (CD45⁻CD326⁻CD31⁺CD144⁺), immune (CD45⁺CD326-CD31⁻CD144⁻), and mesenchymal cells (CD45⁻CD326⁻CD31⁻CD144⁻) was sequenced with Illumina2500. Sequences were aligned to GRCh38p2 using HISAT. Seqmonk (version 1.41.0) was used to summarize and normalize counts. The full data sets for murine or human RNA-seq experiments are available in the NHLBI LungMAP(Ardini-Poleske et al., 2017) web resource (https://lungmap.net/breath-download-page/).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics:

Each data point represents a pool of 5–6 newborn mice that were pooled before the isolation of lymphocytes from the indicated tissue. All data meet the assumptions of the statistical tests used. We compared differences between groups by the either unpaired two-tailed Student’s t-test or ANOVA or Wilcoxon signed-rank test. We used Pearson correlation coefficient to measure correlation between different variables. We used the Kaplan-Meier log-rank test to compare survival between groups. (All in GraphPad Prism 6.0). P-values are indicated as follows: * p ≤ 0.05 or ** p ≤ 0.01.

DATA AND CODE AVAILABILITY

The accession numbers for the bulk RNA-Seq and scRNA-Seq data reported in this paper are Gene Expression Omnibus GSE122332 and PRJNA599444.

Raw data related to the figures reported in this paper are available at Mendeley Data and can be accessed at http://dx.doi.org/10.17632/zkj6v7s8cp.1

Supplementary Material

Video 5

Supplementary Video 5: Intratracheal delivery of diphtheria toxin in newborn mice. (Refers to Fig. 3, supplementary Fig. 2)

Video 1

Supplementary Video 1: ZBTB16⁺ ILC precursors populated a niche defined by platelet derived growth factor (PDGFR) α expressing fibroblast in the newborn lungs. (Refers to Figure 4)

Video 2

Supplementary Video 2: ZBTB16⁺ ILC precursors populated a niche defined by platelet derived growth factor (PDGFR) α expressing fibroblast in the newborn lungs. (Refers to Figure 4)

Video 3

Supplementary Video 3: ZBTB16⁺ ILC precursors populated a niche defined by platelet derived growth factor (PDGFR) α expressing fibroblast in the newborn lungs. (Refers to Figure 4)

Video 4

Supplementary Video 4: ZBTB16⁺ ILC precursors populated a niche defined by platelet derived growth factor (PDGFR) α expressing fibroblast in the newborn lungs. (Refers to Figure 4)

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-mouse CD3ε antibody	BioLegend	Clone: 145–2C11
anti-mouse CD5 antibody	BioLegend	Clone: 53–7.3
anti–mouse CD8 antibody	BioLegend	Clone: 53–5.8
anti–mouse CD11b antibody	BioLegend	Clone: M1/70
anti-mouse CD11c	BioLegend	Clone: N418
anti–mouse CD19 antibody	BioLegend	Clone: 1D3
anti-mouse CD25 antibody	BioLegend	Clone: PC61.5
anti-mouse CD45.1	BioLegend	Clone: A20
anti-mouse CD45.2	BioLegend	Clone: 104
anti-mouse CD117 antibody	BioLegend	Clone: ACK2
anti-mouse Ly6G antibody	BioLegend	Clone: 1A8
anti-mouse F4/80 antibody	BioLegend	Clone: BM8
anti-mouse FLT3 antibody	BioLegend	Clone: A2F10
anti-mouse α4β7 antibody	BioLegend	Clone: DATK32
anti-mouse NKp46 antibody	BioLegend	Clone: 29A1.4
anti-mouse CCR6 antibody	BioLegend	Clone: 29–2L17
anti-mouse KLRG1 antibody	BioLegend	Clone: 2FI
anti-mouse NK1.1	BioLegend	Clone: PK136
anti-mouse PD1 antibody	Bioxcell	Clone: J43
anti–mouse CD127 antibody	BioLegend	Clone: A7R34
anti–mouse RORγt antibody	BD Biosciences	Clone: Q31–378
anti-mouse T-bet antibody	BioLegend	Clone: 4B10
anti-mouse ZBTB16 antibody	BD Biosciences	Clone: R17–809
anti-mouse TCF7 antibody	BioLegend	Clone: 812145
anti-mouse GATA3 antibody	BioLegend	Clone: 16E10A23
anti-mouse Id2 antibody	ThermoFisher	Clone: 17-9475-82
anti-mouse Eomes antibody	ThermoFisher	Clone: TBR2
anti-mouse BrdU antibody	BioLegend	Clone: Bu20a
anti–mouse CD4 antibody	BioLegend	Clone: GK1.5
anti-mouse B220	BioLegend	Clone: 2FI
anti-mouse FcεR1 blocking antibody	BD Biosciences	Clone: 93
anti-CD131	BD Biosciences	Clone: REA193
anti-CD45.2	BD Biosciences	Clone: V450
anti-CD140	BD Biosciences	Clone: APA5
anti-CD326 antibody	BD Biosciences	Clone: G8.8
anti- CD31	BD Biosciences	Clone: MEC13.3
anti-IGF1	Santa Cruz	Clone: SC 461
anti-IL33R	R&D Biosystems	Clone: AF1004
anti-IgG	BioLegend	Clone: RMG1–1
anti-mouse Sca1	BioLegend	Clone: D7
chicken anti-mouse GFP	ThermoFisher	Catalog: A10262
rabbit anti-mouse NKX2.1	Seven Hills Bioreagents	Clone: 1231
mouse anti-mouse PDGFRα	ThermoFisher	Clone: APA5
hamster anti-mouse IGF1	Millipore	Clone: Sm1.2
anti-human CD3	BioLegend	Clone: UCHT1
anti-human CD5	BioLegend	Clone: UCHT2
ant--human CD11c	BioLegend	Clone: S-HCL-3
anti-human CD11b	BioLegend	Clone: M1/70
anti-human CD19	BioLegend	Clone: HIB19
anti-human FcεRI	BioLegend	Clone: AER-37
anti-human CD34	BioLegend	Clone: 581
ant-human CD14	BioLegend	Clone: HCD14
anti-human CD45	BioLegend	Clone: HI30
anti-human CD127	BioLegend	Clone: A019D5
isotype-matched control antibodies	BioLegend	Clone: MOPC-173
anti-human GATA-3	eBiosciences	Clone: TWAJ
anti-human RORγT	eBiosciences	Clone: AFKJS-9
anti-human IL-22 antibody	eBiosciences	Clone: 22URTI
Bacterial and Virus Strains
S. pneumoniae serotype 19A	ATCC	Catalog: 700674
Biological Samples
Preterm human infants diagnosed with bronchopulmonary dysplasia or term healthy infants.	CD45+ cells in Bronchial lavage fluid	NA
Chemicals, Peptides, and Recombinant Proteins
True Nuclear Transcription Factor Buffer	BioLegend	Catalog: 424401
Zombie NIR fixable viability marker	BioLegend	Catalog: 423105
Matrigel (growth factor reduced, phenol-red free)	Corning	Catalog: 356238
Protein kinase (ROCK) inhibitor, Y-27632	ThermoFisher	BDB562822
murine IL-7	R&D	Catalog: 407-ML-005
stem cell factor	BioLegend	Catalog: 579702
1X Matrisperse Cell recovery solution	BD Biosciences	Catalog: 354253
phorbol 12-myristate 13-acteate (PMA)	Sigma Aldrich	Catalog: P8139
calcium ionophore A23187	Sigma Aldrich	Catalog: C7522
DMEM:F12 containing heavy and non-heavy 13C6 L-Lysine-2HCl	ThermoFisher	Catalog: 88439
Critical Commercial Assays
Illumina TruSeq Stranded Total RNA kit	Illumina	RS-122–2201
Deposited Data
Single-cell and time-course RNA-seq experiments from newborn lungs	NCBI SRA	GEO accession # GSE122332
RNA-seq of sorted ZBTB16+ ILC precursors from newborn lungs and bone marrow	NCBI SRA	GEO accession # PRJNA599444
Raw Data	Mendeley Data	http://dx.doi.org/10.17632/zkj6v7s8cp.1
Experimental Models: Cell Lines
OP9-DL1	Dr. J. C. Zúñiga-Pflücker	N/A
3D Alveolar Aggregates	This manuscript	N/A
Experimental Models: Organisms/Strains
Zbtb16CreGFP	Jackson Laboratory	Stock # 024529
RorcCreGFP	Jackson Laboratory	Stock # 022971
Rosa26iDTR	Jackson Laboratory	Stock #007900
Gli1CreER	Jackson Laboratory	Stock # 007913
Igf1fl/fl	Jackson Laboratory	Stock # 016831
Igf1rfl/fl	Jackson Laboratory	Stock # 012251
Rag2−/−;Il2rg−/−	Jackson Laboratory	Stock # 014593
B6 Cd45.1, Pep/BoyJ	Jackson Laboratory	Stock# 002014
Il22fl/fl	mutant mouse resource and research center (MMRRC	Stock # 032402
Software and Algorithms
FlowJo	Treestar	https://www.flowjo.com/
ImageJ	ImageJ	https://imagej.nih.gov/ij/
Imaris software	Bitplane	http://www.bitplane.com/ProductDetail.aspx?ProductID=57&SeriesId=55&Page=camera
Nikon Elements software	Nikon	https://www.microscope.healthcare.nikon.com/products/software/nis-elements
MaxQuant	https://www.maxquant.org/	version 1.5.1.2
Ensembl	https://useast.ensembl.org/info/website/archives/index.html	version GRCm38_71 (released April, 2013)
Perseus	https://omictools.com/perseus-tool	version1.5.8.5
ToppGene Suite	https://toppgene.cchmc.org/https://toppgene.cchmc.org/	https://toppgene.cchmc.org/
NHLBI LungMAP	https://www.lungmap.net/about/lungmap-team/nhlbi/	https://www.lungmap.net/about/lungmap-team/nhlbi/
Illumina HiSeq 2500	https://www.illumina.com/systems/sequencing-platforms/hiseq-2500.html	https://www.illumina.com/systems/sequencing-platforms/hiseq-2500.html
HISAT	http://www.ccb.jhu.edu/software/hisat/index.shtml	http://www.ccb.jhu.edu/software/hisat/index.shtml
Seurat	https://www.rdocumentation.org/packages/Seurat/versions/2.2.0	version 2.2
HISAT2	https://ccb.jhu.edu/software/hisat2/index.shtml	https://ccb.jhu.edu/software/hisat2/index.shtml
Seqmonk	https://www.bioinformatics.babraham.ac.uk/projects/seqmonk/	version 1.45.1
Seqmonk	https://www.bioinformatics.babraham.ac.uk/projects/seqmonk/	version 1.41.0
Prism 6	Graphpad	https://www.graphpad.com/scientific-software/prism/
Other
LSRII	BD Biosciences	https://www.bd.com/resource.aspx?IDX=17868
FACSAria II	BD Biosciences	https://www.bdbiosciences.com/documents/BD_FACSAria_II_cell_sorter_brochure.pdf
Nikon Eclipse confocal microscope	Nikon	https://www.microscope.healthcare.nikon.com/products/inverted-microscopes/eclipse-ti2-series
LTQ-OrbiTrap XL linear ion trap mass spectrometer	ThermoFisher	IQLAAEGAAVFACZMAIK
Sony SH800S	Sony Biotechnology	https://www.sonybiotechnology.com/us/instruments/sh800s-cell-sorter/

Highlights:

• Pulmonary ILC3s descend from ILC precursors that populate a spatially distinct niche

• Alveolar fibroblasts-derived IGF1 instructs postnatal lung ILC precursor biogenesis

• Alveolar fibroblast signals link postnatal development with ILC3 biogenesis in the lung.

• Premature infants with bronchopulmonary dysplasia have reduced IGF1 and pulmonary ILC3s