For isolation of lung parenchymal cells, the collected lung tissue was cut into small pieces. After washing in 20 ml of complete RPMI-1640, digestion was performed by resuspending the mashed material in RPMI-1640 containing 1% P/S, 0.125 U/ml Collagenase D (Sigma-Aldrich), 0.180 mg/ml DNaseI (Sigma-Aldrich), 0.125 mg/ml Liberase DH (Sigma-Aldrich), and 0.125 mg/ml Liberase TM (Sigma-Aldrich). The material was incubated at 37°C in a shaking water bath (250 rpm) for 2 h. After filtering (70-μm cell strainer) and washing with ice-cold HBSS (PAN-Biotech GmbH), erythrocyte lysis was performed for 5 min at room temperature. After a second filtering step (40-μm cell strainer), cells were washed with complete RPMI-1640, and the cell pellet was resuspended in complete IMDM (PAN-Biotech GmbH) supplemented with 10% FCS (PAN-Biotech GmbH), 100 U/ml Penicillin and 100 μg/ml Streptomycin (PAN-Biotech GmbH).

For detection of cytokine-producing Th cells, isolated lung parenchymal cells were rested overnight at 37°C in complete IMDM (PAN-Biotech GmbH) supplemented with 10% FCS (PAN-Biotech GmbH), 100 U/ml Penicillin, and 100 µg/ml Streptomycin (PAN-Biotech GmbH) and stimulated on the following day with PMA (20 ng/ml, Sigma-Aldrich) plus ionomycin (1 µg/ml, Sigma-Aldrich) for 4 h and in the presence of BrefeldinA (3 µg/ml, eBioscience) during the last 3.5 h of restimulation.

Cells were stained with different combinations with fluorochrome-conjugated antibodies for flow cytometry analyses (BD FACS Canto II, BD FACSAriaIII, BD FACS Diva software, FlowJo v10 software by Tree Star) according to agreed standards (Cossarizza et al., 2019, EJI). The following porcine-specific antibodies were used: anti-CD3ϵ-PerCPCy5.5 (clone BB23-8E6-8C8, IgG2a, BD Biosciences), anti-TCR1δ-unlab (clone PGBL22A, IgG1, Kingfisher Biotech), anti-CD4α-AlexaFluor^® 647 or -PE-Cy7 or -PerCP-Cy5.5 (clone 4-12-4, IgG2b, BD Biosciences), anti-CD8α-FITC or -AlexaFluor^® 647 (clone 76-2-11, IgG2a, BD Biosciences) and anti-IFN-γ-PE or -AlexaFluor^® 647 (clone P2G10, IgG1, BD Biosciences), anti-CD163-FITC (clone 2A10/11, IgG1, Bio-Rad Laboratories, Inc.), anti-SLAII-unlab or -FITC (clone 2E9/13, IgG2b, Bio-Rad Laboratories, Inc.), anti-CD172a-PE (clone 74-22-15A, IgG2b, BD Biosciences), and anti-CD172a-Biotin (clone 74-22-15, IgG1, SouthernBiotech). In addition, several cross-reactive antibodies or secondary antibodies were used: anti-human CD14-Viogreen (clone Tük4, IgG2a, Miltenyi Biotec), anti-mouse/rat Foxp3-eFluor^® 450 (clone FJK-16s, IgG2a, Thermo Fisher Scientific), anti-human/mouse GATA3-PE (clone TWAJ, IgG2b, Thermo Fisher Scientific), anti-human/mouse Tbet-PE-Cy7 or -PE (clone 4B10, mouse IgG1, Thermo Fisher Scientific), anti-human TNF-α-Pacific Blue (clone Mab11, IgG1, BioLegend), anti-human IL-4-PE-Cy7 (clone MP4-25D2, IgG1, BioLegend), anti-human/mouse CD11b-APC-Cy7 (clone M1/70, IgG2b, BioLegend), anti-mouse IgG-Brilliant Violet 421™ (clone Poly4053, polyclonal IgG, BioLegend), anti-mouse IgG1-APC-Cy7 (clone RMG1-1, IgG, BioLegend), anti-mouse IgG1-FITC (clone M1-14D12, IgG, Thermo Fisher Scientific), and Streptavidin-PE/Dazzle™ 594 (Biolegend). Fixable viability dyes in eFluor^® 506 and eFluor^® 780 were used to exclude dead cells (purchased from Thermo Fisher Scientific). Intracellular/intranuclear antigens were stained after fixation and permeabilization of cells (Invitrogen™ eBioscience™ Foxp3/Transcription Factor Staining Buffer Set, Thermo Fisher Scientific).

Lung tissue from the caudal left lung lobe was sampled for histology and fixed in a formalin solution (Roti-Histofix 10%, Carl Roth GmbH + Co. KG) for 6 h at RT and then stored at 4°C. Fixed lung tissue was embedded in paraffin. Paraffin sections of 1–2 µm thickness were cut, dewaxed, and stained histochemically with hematoxylin and eosin (HE) for histomorphometric analysis, and grading of interstitial broadening and infiltration was performed. For immunohistochemical evaluation, lung sections were dewaxed and subjected to a heat-induced epitope retrieval step prior to incubation with anti-Ki67 (clone MIB1, Agilent Technologies) followed by incubation with biotinylated secondary antibody (donkey anti-mouse, Dianova). Biotin was detected by streptavidin coupled with alkaline phosphatase (AP) and RED as chromogen [both Dako REAL™ Detection System, Alkaline Phosphatase/RED, Mouse (Agilent Technologies)]. After chromogen development, proteins in Ki67 stained sections were inactivated followed by incubation with a cross-reactive anti-human CD3 (GA503; polyclonal rabbit, Agilent Technologies). For detection, EnVision+ System-HRP Labelled Polymer Anti-Rabbit (Agilent Technologies) was used. HRP was visualized with diaminobenzidine (Agilent Technologies) as chromogen. Nuclei were counterstained with hematoxylin (Merck Millipore) and slides coverslipped with glycerol gelatine (Merck). Negative controls were performed by omitting the primary antibody. Myeloperoxidase-positive (MPO⁺) cells were counted in 10 high power fields in histological cross sections of the lung applying a cross-reactive polyclonal Ab against human MPO (A0398; polyclonal rabbit, Agilent Technologies) and the labeled streptavidin biotin detection method. Eosinophils were detected with modified Sirius Red staining as described before (25). All slides were coverslipped with glycerol gelatine (Merck). Cells were quantified in 10 high power fields (hpf) (0.237 mm²) per section. Images were acquired using the AxioImager Z1 microscope (Carl Zeiss MicroImaging, Inc.). All evaluations were performed in a blinded manner.

Total RNA was extracted from lung tissue using innuPREP RNA Mini Kit 2.0 (Analytik Jena) and reverse transcribed using High Capacity RNA-to-cDNA Kit (Applied Biosystems). LightCycler^® 480 SYBR Green I Master Mix (Roche) was used for amplification of mRNA transcripts of CCL2 (26), IL33 (27) and RPL19 (26). Efficiencies for each primer pair were determined by generating a standard curve, and mRNA expression was normalized to the housekeeping gene ribosomal protein L19 (RPL19). The results for the relative gene expression were calculated using efficiency-corrected 2^-ΔΔCT method. All amplification reactions were performed on Light-Cycler^® 480 II system (Roche).

Total genomic DNA from lung mucosal swabs, fecal, and environmental samples was extracted using commercial kits following the instructions of the manufacturer (for fecal samples: Qiagen Stool kit, Qiagen, Hilden, Germany; for lung mucosal and environmental swabs: Macherey Nagel NucleoSpin Tissue kit, Düren, Germany, including NucleoSpin^® Forensic Filters for DNA recovery from swabs). Sample preparation, including amplification of the V3 to V4 regions of the 16S rRNA gene using primer set 341F-785R, equimolar mixing, sample clean-up and sequencing by Illumina MiSeq, was performed by LGC Genomics (Berlin, Germany). Demultiplexed and primer-clipped sequence data were uploaded to the MG-RAST Server (https://www.mg-rast.org/; study ID ‘formula_lung’, ‘formula_feces’, ‘formula_penfloor’, respectively) and processed by its SEED software tool. The phylogenetic profile of each sample was computed using the Green Gene reference data bank for identification. Bacterial taxa with five or less identical sequence reads per sample were removed from further analysis. Similarly, sequence reads occurring in one sample only were ignored. The remaining sequence reads were used to calculate the relative contribution of specifically assigned sequences to total sequence reads in a sample, calculation of ecological indices (Shannon diversity index, Evenness), and further statistical analyses for identified putative taxa.

Graphics were created and statistics were performed with GraphPad PRISM software 8.4.2 (GraphPad Software, USA). Normality was tested with the Shapiro–Wilk test. Normally distributed data were analyzed by unpaired t-test or one-way ANOVA followed by Dunnett’s multiple comparisons test, Tukey’s multiple comparisons test or Holm–Sidak’s multiple comparisons test, and by Brown–Forsythe ANOVA test followed by Dunnett’s T3 multiple comparisons test. Nonnormally distributed data were analyzed by Mann–Whitney test or Kruskal–Wallis test followed by Dunn’s multiple comparisons test. For multiple comparisons tests values with P <.05 (*), P <.01 (**), and P <.001 (***) were considered to be significant. Correlation analyses were assessed by parametric Pearson correlation analysis or non-parametric Spearman correlation analysis procedure. For correlation analysis, values with P <.1234 (*), P <.0332 (**), P <.0021 (***) were considered to be significant.

A partial least square discriminant analysis (PLS-DA) model was used to determine the bacterial taxa that contributed to discrimination between experimental groups in Exp 1, followed by variable importance in projection (VIP) using SIMCA-P+ (Version 12.0; Umetrics, Umea, Sweden). Putative taxa with the VIP values >1 were considered most discriminatory (28). Venn diagrams illustrating unique or shared individual bacterial taxa of neonatal piglets created using Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/).

We first characterized postnatal lung maturation over time (neonatal phase, pre-weaning, and post-weaning) in piglets normally reared with the sow and suckled until weaning ( Figure 1A ). Histological analyses revealed distinct changes in the morphological structure of the porcine lung after birth ( Figure 1B ) with a significant increase in the number of alveoli at postnatal day (PND) 14 ( Figures 1C , S1A ) and a moderate but non-significant increase in alveolar size over time ( Figures 1C , S1A ). Immunohistochemical co-staining of Ki67 and CD3 showed significantly higher numbers of proliferating lung parenchymal cells in neonatal piglets than in weaned piglets ( Figures 1D , S1A ). As a characteristic of lung growth and alveolarization, the number of alveoli and the number of cell-cycle active (Ki67⁺) lung parenchymal cells correlated significantly (r = 0.3881, P < 0.05) during postnatal lung development ( Figure 1E ). Lung-tissue infiltrating neutrophils showed initially high numbers which dropped significantly at the end of the neonatal phase ( Figures 1F , S1B ). In contrast, infiltrating eosinophils demonstrated a moderate but non-significant early increase at PND7 and a late postweaning increase at PND42 ( Figures 1F , S1B ). Interestingly, the frequency of pulmonary SLAII expressing CD172a⁺ myeloid cells being able to present antigens to T cells ( Figure 1G ) peaked on PND14 and decreased significantly after weaning ( Figures 1H, I ). Our data thus indicated an increased activation of the mononuclear phagocyte system during the neonatal phase, which was changed into a more anti-inflammatory state as the porcine lung develops. So far, ILC2s have not been described in pigs. In mice and humans ILC2s are characterized as negative for lineage marker typically expressed on T cells, B cells, or dendritic cells and highly express the transcription factor GATA3 (29). Based on the exclusion of the porcine T cell lineage markers CD3, CD4, and CD8α, we analyzed live GATA3^high γδTCR1⁻ non-T cells ( Figure 1J ), since high GATA-3 expression levels were also found in porcine extra-thymic γδ T cells, declining with age (30). Remarkably, pulmonary GATA3^high γδTCR1⁻ non-T cells increased significantly until weaning ( Figures 1K, L ). This is consistent with findings in adult mice in which most ILC2s in the lung developed as tissue-resident cells in the neonatal period (31) and only very few ILCs were circulating (32). During the neonatal period, infiltration with ‘ILC2-like’ cells in the porcine lung was clearly associated with alveoli formation, characterized by a significant positive correlation of pulmonary GATA3^high γδTCR1⁻ non-T cells and alveoli counts (r = .6047, **P <.0332) ( Figure 1M ).

Together, these data showed that postnatal lung development in newborn piglets was highly dynamic and characterized by strong cell proliferation and marked tissue remodeling during alveolarization, corresponding to the induction of innate type 2 immunity and increased activation of the pulmonary mononuclear phagocyte system.

To determine the time course of T cell recruitment and T helper cell polarization during postnatal lung maturation in the pig, their frequencies and phenotypes in lung tissue lymphocytes were investigated ( Figure 2A ). CD marker expression in pigs has some peculiarities compared to humans, such as the expression of CD8αα-homodimers and SLAII (MHCII) molecules on activated/memory CD4⁺ Th cells (29). Within the first weeks of the life of a pig, T cells infiltrate into the lung ( Figure 1D , see also Figure S2A ), showing a highly activated phenotype characterized by increased frequencies and expression levels of SLAII ( Figure S2B ). The differentiation/activation state of porcine Th cells during postnatal development was examined by the analysis of CD8α co-expression on CD4⁺ T cells, and the frequency of pulmonary CD4⁺CD8α⁺ T cells (effector/memory Th cells) was found significantly increased after weaning ( Figure 2B ). For the discrimination of lung T cell differentiation we used transcription factor-based detection of functional subpopulations, which we previously established for the pig (33). The frequency of Tbet expressing CD4⁺ T (Th1) cells in the lung increased significantly during the first two weeks of life, indicating an early initiation of Th1 differentiation at mucosal sites ( Figures 2C, D ). Accordingly, the Tbet protein expression level per pulmonary CD4⁺ T cell (Tbet GMFI) increased significantly until the post-weaning phase ( Figure S2C ). The initial high frequencies of GATA3 expressing CD4⁺ T (Th2) cells (mean 5% of CD4⁺, PND3) in the lung dropped significantly after weaning ( Figures 2C, E ), consistent with the postulation of a strong Th2-biased immunological environment in the newborn lung (34). Accordingly, T helper cell differentiation during postnatal lung maturation was also reflected in the Th1/Th2 ratio, which increased significantly after weaning ( Figure 2F ). Regulatory T (Treg) cells are essential for the establishment of self-tolerance and immune homeostasis (35), and interestingly, they are highly abundant in neonatal murine lungs during alveolarization (9). Similarly, we found high frequencies of Foxp3 expressing CD4⁺ Treg cells during early postnatal lung maturation in the piglets, which decreased significantly after weaning ( Figure 2G ). Consequently, the ratio of effector/memory Th cells and Treg cells increased significantly after weaning at PND42 ( Figure S2D ), and an inverse correlation was observed between effector/memory Th cell and Treg cell frequencies (r = −.5415, ** P <.0332) ( Figure 2H ). To obtain temporal resolution of postnatal Th1 cytokine production in the pig, we examined IFNγ and TNFα expression in pulmonary Th cells after stimulation with PMA/ionomycin ( Figure 2I ). The frequency of IFNγ ⁺TNFα⁺ Th cells ( Figures 2J, K ) and TNFα expression levels ( Figure S2F ) in the lungs increased significantly during the late neonatal period at PND14 and before weaning at PND28. Accordingly, we demonstrated a significant positive correlation of the frequencies of Th1 cells with IFNγ ⁺TNFα⁺ Th cells (r = .6143, *** P <.0021) during postnatal lung development ( Figure S2E ). IL-4⁺TNFα⁺-producing Th cells were detectable only in low frequency, and these significantly decreased after weaning ( Figures S2G, H ).

In summary, our data demonstrate that Th2 and regulatory immune responses dominate in the porcine neonatal lung shortly after birth, with a shift in favor of effector/memory and Th1 immune responses after weaning (PND42).

Studies investigating early nutritional and environmental influences on postnatal lung development at the end of the neonatal period are currently lacking. Here we assessed whether isolated rearing from the sow and an altered diet impact postnatal lung maturation. One-day-old, colostrum-fed piglets were allocated into artificial rearing units and fed either sow milk or a bovine milk-based formula. In addition, conventionally sow-reared suckling piglets were fed a bovine milk-based formula. Postnatal lung development within the first 14 days of life of differently housed and fed newborn piglets were compared with normally developing sow-reared suckling piglets at PND7 and PND14 ( Figures 3A , S8A–G ). Isolator-rearing significantly reduced lung growth ( Figures 3B, C , S3A ) and body weight ( Figure 3D ) in formula-fed piglets at PND14. Accordingly, we showed a positive correlation between relative lung weight and body weight gain (r = .5470, *** P <.0021) during the neonatal period ( Figure 3E ). Deficits in neonatal lung growth were also reflected by histomorphological analysis ( Figure 3F ) which revealed diminished interstitial broadening and cellular infiltrates ( Figure 3G and see also Figure S3B ) and reduced alveoli counts ( Figure 3H ) in HE stained lung tissue of 14 days old piglets reared isolated from the sow. Immunohistochemical staining of Ki67 revealed significantly higher numbers of proliferating lung parenchymal cells in sow-reared suckling piglets compared to abnormally reared piglets at PND14 ( Figures 3F, I ). To provide further evidence of delayed alveolarization due to altered rearing conditions, we examined IL33 gene expression in lung tissue that was decreased at PND14 in formula-fed piglets kept isolated from the sow ( Figure 3J and see also Figure S3C ). Interestingly, we detected a significant increase in IL33 gene expression at PND14 in sow-reared piglets receiving an additional formula diet. Indicative of an association with the progress of postnatal lung growth, lung weight correlated significantly with IL33 gene expression at PND14 (r = .7539, **** P <.0001) ( Figure 3K ).

Taken together, these data suggest that abnormal nutritional and rearing conditions during the neonatal period affect postnatal lung growth and that feeding breastmilk in an isolated environment mitigates the delay in postnatal lung maturation.

The lung mononuclear phagocyte system is predominantly active in the neonatal period because, as part of the innate immune system, it does not require specific antigen experience (36). We therefore asked whether abnormal rearing conditions such as formula feeding and isolation from the mother affect maturation of the pulmonary mononuclear phagocyte system and whether breast milk feeding reverses adverse effects of isolation of formula-fed neonates. Identification of porcine phagocyte mononuclear subsets within lung parenchymal cells by flow cytometry analysis was adapted from Maisonnasse et al. (20) ( Figure 4A ). The frequency of total myeloid cells (CD3⁻CD172a⁺) in the lungs was not affected by different rearing and feeding conditions ( Figure 4B ). However, the expression levels of SLAII ( Figures 4C and S4A ) and the frequency of SLAII⁺ myeloid cells ( Figure 4D ) were significantly altered between experimental groups at PND14, with no differences at PND7 ( Figure S4B ). Similarly, the frequency of CD11b⁺ myeloid cells decreased significantly at PND14 in the lungs of piglets kept isolated from the sow ( Figure 4E and see also Figure S3C ). Analysis of the different lung mononuclear phagocyte subsets revealed significant changes in the frequencies of cDC2 and moDC, which were most pronounced on PND14 ( Figures 4F–H and see also Figures S4D–F ). Compared to sow-reared suckling piglets, formula feeding of piglets together with isolation resulted in a significant decrease in moDC ( Figure 4G ) and cDC2 frequencies ( Figure 4H ) at PND14. Interestingly, moDC and cDC2 lung frequencies were not affected in piglets receiving sow milk when kept isolated from the sow ( Figures 4G, H ).

To address what causes differential cell infiltration we investigated expression of the chemoattractant CCL2. CCL2 expression increased over time in the lungs of sow-reared suckling piglets but was significantly reduced at PND14 in formula-fed piglets isolated from the sow ( Figure 4I ). Statistical analysis revealed a significant positive correlation between CCL2 expression levels and cDC2 frequencies in the developing lung (r = .5759, **P <.0332) ( Figure 4J ).

Taken together, these data indicate that deficits in postnatal maturation of the pulmonary mononuclear phagocyte system of newborns can be caused by formula feeding and isolation from the mother and that these deficits can be partially compensated by feeding with breast milk.

To investigate whether altered early life conditions affect postnatal pulmonary Th1 differentiation (as shown in Figure 2 ) and the regulatory T cell response, Tbet and Foxp3 expressing Th cells were analyzed by flow cytometry in the lungs of differently kept and fed neonatal piglets. At PND14, overall T cell frequencies in the lungs were slightly reduced in formula-fed piglets isolated from sows compared to sow-reared suckling piglets ( Figure 5A ). Formula feeding in isolated piglets resulted in significantly decreased Th1 cell frequencies at PND14 ( Figures 5B, C ) and slightly decreased Tbet expression levels ( Figure S5A ), whereas the frequency of Th1 cytokine-expressing Th cells did not change significantly ( Figure S5B ). Importantly, a significant correlation was demonstrated between Th1 cell frequencies and lung weight (r = .5146, ***P <.0021), SLAII expression levels in myeloid cells (r = .4661, ***P <.0021), and cDC2 frequencies in lungs (r = .6524, ****P <.0001) at PND7 and PND14 ( Figure 5D ). Furthermore, the number of proliferating lung parenchymal cells correlated significantly with pulmonary Th1 cell frequencies (r = .7288, ***P <.0021) at PND14 ( Figure S5C ). Pulmonary Treg cell frequencies were not significantly affected by different early life conditions ( Figures 5B, E ).

These results indicate that formula feeding in an isolated environment delays postnatal pulmonary Th1 differentiation, which is also reflected in correlated impairment of lung growth and immaturity of the pulmonary mononuclear phagocyte system. Breast milk feeding attenuates these effects, which may indicate a possible mucosal cross-talk between the gut and lungs.

We used our pig model to investigate whether isolated rearing and formula feeding of newborn piglets lead to changes in the developing bacterial communities of the lungs. Mucosal swabs were taken immediately post mortem from the secondary bronchi to avoid contamination by species from the oral cavity and upper respiratory tract ( Figure 6A ). Rearing and feeding conditions clearly influenced the relative abundance of the major phyla ( Figure 6B ) and the 17 most abundant genera and species richness ( Figure 6C ) in the lungs of 7- and 14-day-old piglets. On PND7, Actinobacteria and Bacteroidetes predominated in piglets isolated from the sow, whereas Fusobacteria almost disappeared on PND14 ( Figure 6B ). At the genus level, Lactobacillus abundance was reduced in piglets reared in isolation from the sow at PND7, and Propionibacterium and Actinobacillus dominated at PND14 ( Figure 6C ). Importantly, the environment had a significant effect on species richness, Shannon diversity, and species evenness, which were reduced in sow contact-deprived piglets ( Figure 6D ). In addition, our partial least squares-discriminant analysis revealed that microbiota variations were closely clustered between normally suckled piglets and isolated but sow milk-fed piglets compared with formula-fed groups on PND14 ( Figure S6A and Tables S1 and S2 ). Our data indicated a significant correlation between Shannon diversity and relative lung weight (r = .3364, **P <.0332) or pulmonary Th1 cell frequency (r = .3977, **P <.0332) ( Figure 6E ). Because the inhaled or aspirated environment shapes the lung microbiota (10) and, in contrast to humans, the coprophagy of the pig contributes to microbiota transfer (37), we analyzed the shared bacterial species between different habitats (feces vs. environment vs. lung) with the majority of species present in the lungs ( Figure S6B and Table S3 ) also detectable in the feces and environment ( Figure 6F and Table S4 ). Accordingly, we next studied the impact of material from the maternal environment on lung immunity of sow-deprived formula-fed newborn piglets ( Figure 6G ). Th1 cell frequency ( Figures 6H, I ) and Th1/Th2 ratio ( Figures 6J , S6C ) were significantly increased in the group receiving material from the maternal environment, indicating improved postnatal lung development by microbial transfer.

In summary, these results suggest that isolation of neonates from the mother modifies the developing lung microbiota, with decreased richness of bacterial species correlating with reduced lung growth and Th1 differentiation. The decreased Th1 differentiation could be reversed by transferring material from the maternal environment.

We next asked whether the length of interfering with natural bonding and nutrition differently impacts postnatal lung maturation and the pulmonary immune system, and thus returning piglets following such an intervention to the mothers after 3 or 7 days converges the lung phenotype to normally developed piglets ( Figure 7A ). For that purpose, we investigated differences in lung growth, ILC2, and myeloid cells as well as the lung Th1 differentiation program in experimentally reared and fed groups of n = 5–6 piglets. Returning piglets to their mothers at both 3 and 7 days postpartum resulted in significant increases in lung growth ( Figures 7B, C ) and body weight ( Figure 7C ) compared with sow contact-deprived piglets that received formula during the entire experiment. ‘ILC2-like’ cell frequencies in the lungs were highest in piglets returned to sows 3 days after birth, with significantly elevated levels compared to sow-deprived piglets that received formula throughout the experiment ( Figure 7D ). Interestingly, lung weight and the frequency of pulmonary ‘ILC2-like’ cells were significantly correlated (r = .4882, *P <.1234) ( Figure 7E ), indicating a relationship with the progress of lung maturation. The frequency of SLAII⁺ myeloid cells was significantly reduced only in sow-deprived piglets fed formula for the entire experiment ( Figures 7F, G ). Returning piglets to their mothers on day 3 postpartum caused a significant increase in pulmonary Th1 cell frequencies compared to sow-deprived piglets fed formula throughout the experiment, although pulmonary Th2 cell frequencies did not change by the different conditions ( Figures 7H, I ). Compared to sow-deprived piglets fed formula throughout the experiment and piglets returned to their mothers at day 7 postpartum, returning piglets to their mothers at day 3 resulted in a significantly increased Th1/Th2 ratio comparable to sow-reared piglets ( Figure 7J ). Tbet expression level was significantly reduced only in sow-deprived piglets fed formula for the entire experiment, with expression levels in piglets returned to their mothers on postpartum days 3 and 7 approaching those of sow-reared piglets ( Figures S7A, B ). Treg cell frequencies in the lung were not affected significantly by different early life conditions ( Figure S7C ). The frequency of IFNγ ⁺TNFα⁺ Th cells was significantly reduced only in sow-deprived piglets fed formula for the entire experiment, and the frequency in piglets returned to their mothers on the third postpartum day approached that of sow-reared piglets ( Figures 7K, L ). Accordingly, we demonstrated a significant positive correlation of the frequencies of Th1 cells with IFNγ ⁺TNFα⁺ Th cells (r = .7464, ***P <.0021) in the lungs of differently reared newborn piglets, again demonstrating the gradual differences in Th1 differentiation ( Figure 7M ).

In conclusion, these results suggest that delayed postnatal maturation of the pulmonary immune system caused by early confounding factors, such as altered diet and separation from the mother, can be reversed within a critical window of time after birth.