The umbilical cord blood of term-born infants (n = 4, 38-40 weeks of gestation) was collected aseptically into heparin tubes after cesarean section. Infants with congenital malformations and congenital infections and infants whose mothers suffered from autoimmune disease, or inborn or acquired immune deficiencies were not considered eligible for study inclusion. All procedures were approved by the ethics committee of the Medical University of Vienna (ref. 1923/2012) and informed consent was obtained from the mothers.

Ficoll-Paque density gradient centrifugation was used to enrich peripheral blood mononuclear cells (PBMCs). CD14⁺ monocytes were subsequently isolated from the PBMC-fraction using magnetic bead separation (anti-CD14 MicroBeads, Miltenyi Biotec, Bergisch-Gladbach, Germany). Consequently, 0.75 × 10⁶ cells/mL were transferred into a 6-well flat-bottomed plate (Greiner Bio-One, Kremsmünster, Austria), cultured in RPMI-1640 medium supplemented with 10% fetal calf serum (FCS) (GIBCO, Carlsbad, CA, USA) and differentiated using macrophage colony-stimulating factor (M-CSF) (100 ng/ml; PeproTech, Rocky Hill, NJ, USA) in the absence or presence of hLF (500 µg/ml; Sigma Aldrich, St Louis, MO, USA). Cells were kept in an incubator for 7 days at 37°C, 5% CO₂, and 95% humidity and either stimulated with ultrapure lipopolysaccharide (LPS, 1 ng/mL, E. coli O111:B4, InvivoGen, San Diego, CA, USA) 24h prior to RNA-extraction or left unstimulated.

Total RNA was isolated from 1.5 x 10⁶ differentiated macrophages using the RNeasy Plus Micro Kit (Qiagen, Venlo, the Netherlands) according to the manufacturer’s instructions. RNA was then quantified using a NanoDrop 8000 device (Thermo Scientific, Waltham, MA, USA) and RNA integrity number (RIN) was determined employing the RNA 6000 Nano Kit (Agilent Technologies, Santa Clara, CA, USA). To prepare samples for microarray analysis, fragmented and biotin-labeled sense-stranded cDNA was prepared from 100 ng of total RNA (20 ng/µl) using the GeneChip Whole Transcript (WT) PLUS Reagent Kit (Affymetrix, Santa Clara, CA, USA). cDNA was then hybridized to the GeneChip Human Gene 2.1 ST Array Plate (Affymetrix, Santa Clara, CA, USA) and automatically processed and analyzed using the GeneTitan instrument (Affymetrix, Santa Clara, CA, USA).

RNA-microarray data were processed employing the maEndToEnd Affymetrix differential gene expression workflow (16). Briefly, data were background-corrected, normalized, and summarized using the Robust Multi-array Average (RMA) algorithm with quantile normalization before intensity-based filtering. Afterwards, transcript clusters were annotated using the hugene21sttranscriptcluster.db package (17) and differential gene expression analyzed using limma (18). Visualizations were performed employing the R packages ggplot2 (19), VennDiagram (20), networkD3 (21), GoPlot (22), and ComplexHeatmap (23).

Highly abundant in breast milk, human lactoferrin exerts a diverse set of immunological and developmental functions. Next to direct bactericidal effects, hLF might also act as an immunological primer of macrophages, a cell subset capable of propagating inflammatory damage in the evolving human gut. Thus, we established an in vitro culture system to differentiate neonatal monocytes into macrophages in the presence or absence of hLF. Additionally, we used LPS-treatment to mimic pro-inflammatory stimuli encountered in the intestine upon bacterial colonialization ( Figure 1A ). Thus, following our previous work on hLF-mediated attenuation of pro-inflammatory macrophage polarization, we analyzed transcriptomic changes elicited by hLF to decipher biological patterns that underlie its manifold of immunological functions (15). Transcriptomic signatures of macrophages differentiated with M-CSF in the absence (uMϕ) or presence (hlfMϕ) of hLF, either stimulated with LPS (uMϕ+LPS, hlfMϕ+LPS) or left untreated (uMϕ, hlfMϕ), were investigated using RNA microarray technology. Differential gene expression analysis revealed 406 differentially expressed genes (DEG) between hlfMϕ and uMϕ. Moreover, stimulation of uMϕ with LPS induced an increase in transcriptional activity (971 DEGs) that vastly differed from that observed in hlfMϕ+LPS (823 DEGs). This was of particular interest, as the comparison between hlfMϕ and hlfMϕ+LPS revealed no DEG ( Figure 1B ). To completely exclude the existence of DEG between hlfMϕ and hlfMϕ+LPS we re-analyzed the comparison without employing logFC cut-offs. Even after this adaptation, no DEG were found between the two conditions ( Supplementary Figure 1 ). Correspondingly, hierarchical clustering of DEG deriving from all conditions revealed a distinct transcriptomic profile of hLF-treated cells when compared to uMϕ and uMϕ+LPS ( Figure 1C ). This finding is also in line with the high expressional overlap between stimulated and unstimulated hlfMϕ observable in the Venn diagram ( Figure 1D ) and the clear separation between uMϕ, uMϕ+LPS and hLF-treated cells shown by principal component analysis ( Figure 1E ).

Taken together, the differentiation of neonatal monocytes in the presence of hLF induced a distinct transcriptomic profile and severely impaired the responsiveness of derived macrophages to LPS stimulation.

To understand the cellular responses hLF evokes in neonatal monocyte-derived macrophages, a gene set enrichment analysis (GSEA) was performed using the Reactome database of curated cellular pathways ( Figure 2A ). Biological theme comparison revealed cell-cycle specific effects of hLF, as displayed by marked downregulation of cell cycle progression, DNA replication, and deposition of CENP-A nucleosomes needed for maintaining centromere identity. Moreover, hLF-treated cells presented with an abrogation of LPS-induced effects on cytokine signaling, including engagement in interferon-alpha/beta and interferon-gamma pathways, as exemplified by diametrically opposed enrichment results, whilst concomitantly maintaining cell-cycle specific properties ( Figure 2A ). To gain a deeper understanding of the drivers of hLF-function and its immunosuppressive properties, chord diagrams displaying the most disparate expression profiles for respective Reactome pathways were generated ( Figures 2B, C ). In this context, genes of several histones, proteins (NDC80, CENPF, SKA1, TPX2), cyclins (CCNA2, CCNB1, CCNB2), and cyclin-dependent kinases (CDK1) were downregulated in hlfMϕ when compared to uMϕ - all of which relevant to chromosome segregation, kinetochore function, or cell cycle regulation. In addition, hLF seemed to promote interactions with extracellular matrix proteins via upregulation of transmembrane receptors ITGB3 and ITGA2 as well as FN1 and SCD4. It also triggered downstream effects comparable with IL-4- and IL-13-induced signaling by increasing the expression of JAK3, FOS, SOCS3, ALOX5, CCL22, and PIM1 ( Figure 2B ). Additionally, the upregulation of anti-inflammatory signaling molecules FOS, EGR1, and JUN, as well as increased expression of decoy receptor IL1R2 and anti-inflammatory cytokine IL-10 seem to stand in stark contrast to an LPS-induced proinflammatory phenotype ( Figure 2C ). In line with this finding, hLF also potently blocked LPS-induced upregulation of cytokines and chemokines CXCL8, CXCL10, CCL5, EBI3, and CCL20 as well as interferon-stimulated genes IFITM1, ISG15, IFIT1, and OAS2.

Overall, functional analyses revealed that hLF-induces a ‘quiescent’ state in macrophages due to modification of cell cycle regulation, an important mechanism recently described to drive macrophages into an M2-like phenotype (36).

In order to understand the gene regulatory programs that underlie hLF-induced quiescence, we performed a gene regulator enrichment analysis using RegEnrich (27). This algorithm identifies genetic regulators based on differential gene expression analysis and regulator-target network inference, and provides a metric for their regulatory importance. Hierarchical clustering of the top-25 regulators per condition revealed prominent changes in the mean expression of gene regulatory molecules in hLF-treated samples, irrespective of LPS stimulation ( Figure 3A ). These gene regulatory changes were driven by the downregulation of cell cycle regulatory proteins FOXM1, E2F8, HMGB2, and BRIP1 and the upregulation of immunomodulatory protein IVNS1ABP. Additionally, downregulation of pro-inflammatory regulators IRF7, EZH2, HDGF, and EIF2AK2 was observed when comparing hlfMϕ+LPS to uMϕ+LPS, highlighting that hLF induces transcriptional unresponsiveness towards pro-inflammatory LPS-stimuli ( Figure 3B ). To better understand the functional implications of hLF-mediated transcriptional regulation, the top-five regulators per pathway found to be enriched in hlfMϕ were retrieved. This analysis revealed the importance of BRIP1, BRCA2, BARD1, ASH2L, and BCLAF1 as functional regulators of cell-cycle progression and CBX5, ATF6B, BHLHE40, EAF1, and ABTB1 as mediators of IL-4/IL-13-like effects ( Figure 3C ). Because RegEnrich – per default – infers a regulatory network on all significantly expressed genes, irrespective of their logarithmic fold change value (logFC), a protein-protein interaction network (PPI) for genetic regulators of hLF-induced functions and differentially expressed genes that displayed a log2-fold change of 0.57 or higher was constructed. Although functional regulators interacted with only 27 proteins directly, network analysis revealed high connectivity with all differentially expressed genes identified in our experimental model ( Figure 3D ).

In concert, these results highlight the tight connection between cell-cycle regulators and functional heterogeneity in hlfMϕ and underline their importance in the development of a quiescent state that remains transcriptionally irresponsive to immunogenic stimuli.

As our initial analyses pointed towards induction of a quiescent M2-like state, we aimed to explore whether this homeostatic phenotype also extends to the interactome of hLF-treated macrophages. Harnessing ImmuneDBs data on immunologically relevant genes, Immports cytokine registry data, and information on subcellular protein localization available from Human Protein Atlas, membrane protein, and cytokine gene abundances were computed for each experimental condition. Global changes in cytokine and membrane protein gene expressions were evident for all immunologically implicated enriched Reactome pathways ( Figure 4A ). Upon closer examination, hlfMϕ displayed modified expression of PLAUR, ICAM1, ITGB3, and ITGA3, arguing for changes in cell adhesion and matrix interaction properties. Moreover, together with the altered presence of CD180, Ly9, LAT, and PLSCR1, these changes suggested modifications in phagocytotic and efferocytotic functions. Complementing this homeostatic interactome are the increased expression of SIGLEC7, SDC4, S100A9, and TNFRSF9, proteins that have been ascribed anti-inflammatory properties ( Figure 4B ). Regarding soluble interactions, upregulation of homeostatic and gut protective cytokines IL-10, TNFSF14, and CXCL8 was seen in macrophages treated with hLF. Moreover, hlfMϕ displayed an increased expression of cytokines involved in regulating the proliferation and activity of mucosal innate lymphoid cells. Contradictory at first glance, hLF also mediated the expression of pro-inflammatory molecules OSM, CCL7, CXCL5, IL1B, and CCL3. Yet globally, hlfMϕ+LPS displayed substantial decreases in the expression of pro-inflammatory molecules when compared to uMϕ+LPS ( Figure 4C ). To additionally validate our findings, we employed ANOVA with post-hoc Tukey-correction to exclude a solely TLR4 dependent reduction of LPS-signaling in hlfMϕ and statistically support the increases observed in IL-10, CCL22 and TNFSF15 production – cytokines detrimental to intestinal homeostasis ( Figure 4D ).

Collectively, these results argue for the induction of a quiescent, tissue-homeostatic interactome in hLF-treated neonatal macrophages. Moreover, the near complete abrogation of LPS-induced effects suggests that hLF efficiently drives the cellular communication of neonatal macrophages towards tolerance.

Due to the marked changes hlfMϕ displayed in membrane protein and cytokine expression, we next evaluated their global propensity to interact with other immune cells. Thus, we combined the previously computed membrane protein and cytokine expression data with Innate DB’s manually curated annotation of immunologically relevant physical protein-protein interactions, cytokine-receptor interactions stored in the KEGG pathway database, and normalized gene expression values available from the Monaco immune cell gene data provided by the Human Protein Atlas.

hlfMϕ displayed a markedly increased probability for physical cell-cell interactions when compared to uMϕ and uMϕ+LPS. These physical interactions were biased towards granulocytes, regulatory T-cells, and myeloid dendritic cells ( Figure 5A ) – although to a markedly lesser extent than soluble interactions ( Figure 5B ). Indeed, treatment with hLF strongly increased the capability for soluble interaction with neutrophilic granulocytes when compared to uMϕ, arguing for extensive crosstalk between these cell types upon hLF exposure. Moreover, large differences between uMϕ+LPS and hlfMϕ+LPS were observable for both physical and soluble interaction propensities. Detailed analysis of interactome differences between these two conditions revealed differential expression of CD9, IL7R, ICAM1, TGFBR1, and CD300LB on macrophages as drivers of hLF-induced alterations in physical cell-cell interactions ( Figure 5C ). Regarding cytokines, downregulation of chemoattractants CCL8, CXCL10, CCL19, CCL7, and CXCL9 – amongst others – mediated the decreased probability for soluble interaction observable in hlfMϕ+LPS, when compared to uMϕ+LPS ( Figure 5D ).

Highlighting our previous findings, upregulation of the propensity for physical cell-cell interaction and concomitant downregulation of LPS-mediated soluble interactions argue for hLF-induced polarization of neonatal macrophages into a tissue-homeostatic phenotype.