Differential anti-viral response to respiratory syncytial virus A in preterm and term infants

Summary

Background

Preterm infants are more likely to experience severe respiratory syncytial virus (RSV) disease compared to term infants. The reasons for this are multi-factorial, however their immature immune system is believed to be a major contributing factor.

Methods

We collected cord blood from 25 preterm (gestational age 30.4–34.1 weeks) and 25 term infants (gestation age 37–40 weeks) and compared the response of cord blood mononuclear cells (CBMCs) to RSVA and RSVB stimulation using neutralising assays, high-dimensional flow cytometry, multiplex cytokine assays and RNA-sequencing.

Findings

We found that preterm and term infants had similar maternally derived neutralising antibody titres to RSVA and RSVB. Preterm infants had significantly higher myeloid dendritic cells (mDC) RSV infection compared to term infants. Differential gene expression analysis of RSVA stimulated CBMCs revealed enrichment of genes involved in cytokine production and immune regulatory pathways involving IL-10, IL-36γ, CXCL1, CXCL2, SOCS1 and SOCS3 in term infants, while differentially expressed genes (DEGs) in preterm infants were related to cell cycle (CDK1, TTK, ESCO2, KNL1, CDC25A, MAD2L1) without associated expression of immune response genes. Furthermore, enriched genes in term infants were highly correlated suggesting an increased co-ordination of their immune response to RSVA. When comparing DEGs in preterm and term infants following RSVB stimulation, no differences in immune response genes were identified.

Interpretation

Overall, our data suggests that preterm infants have a more restricted immunological response to RSVA compared with term infants. While further studies are required, these findings may help to explain why preterm infants are more susceptible to severe RSV disease and identify potential therapeutic targets to protect these vulnerable infants.

Funding

Murdoch Children's Research Institute Infection and Immunity theme grant.

Research in context.

Evidence before this study

Differences in the immune response to respiratory syncytial virus (RSV) in infants born preterm may predispose them to severe disease outcomes. This has been demonstrated in studies finding enhanced type 2 innate lymphoid cell related IL-4 secretion and infection of neonatal B regulatory cells by RSV which dampen the Th1 response in preterm compared to term infants.

Added value of this study

This study identifies a reduced ability of preterm infants to elicit anti-viral and immune regulatory responses to respiratory syncytial virus through decreased IL-36γ and IL-10 mRNA. Additionally, we identified trends that suggest RSV infection of myeloid dendritic cells may further reduce immune regulation during early infection.

Implications of all the available evidence

Evidence suggests that enhancing the capacity of preterm infants to elicit effective anti-viral and Th1 driven responses would be critical to reducing severe disease outcomes. These are areas that may be targeted by therapeutics to protect these highly vulnerable infants.

Introduction

Respiratory syncytial virus (RSV) is the leading viral pathogen causing acute lower respiratory tract infections in infants worldwide.¹ Globally, it is estimated to lead to 6.6 million infections and 13,300 in-hospital deaths each year in infants aged 0–6 months, however, the majority of infant deaths occur outside of the hospital.² Recently, the maternal respiratory syncytial virus vaccine (bivalent, recombinant) (Abrysvo, Pfizer) has been approved for use by the U.S Food and Drug Administration (FDA).^3,4 Further, a long-acting monoclonal antibody, nirsevimab™ (Beyfortus, Sanofi, Astra Zeneca), has also been approved for the prevention of RSV disease in infants.5, 6, 7 However, it is likely to be accessible only in high-income countries due to high cost despite the highest burden of RSV disease being in low-middle income countries.²

In addition to the acute effect, severe (hospitalised) RSV disease is also associated with the development of chronic respiratory conditions such as recurrent wheeze, decreased lung function and asthma.8, 9, 10 A major risk factor for the development of severe RSV disease is preterm birth, with preterms over-represented among hospitalised cases.^1,9 Infants born prematurely, <37 weeks gestational age (wGA), can be categorised into very-extreme preterm (<30 wGA), moderate preterm (30–34 wGA) and late preterm (35–36 wGA).¹¹ Preterm infants across all gestational ages are 2-fold more likely to develop severe RSV disease.^9,12 As moderate-late preterms contribute a large proportion of preterm birth (∼70–80%), they represent an important target group for intervention.¹³

A major contributor to the susceptibility of preterm infants to severe RSV disease is thought to be an immature innate immune system, characterised by reduced pathogen recognition, antigen presentation and production of pro-inflammatory cytokines.14, 15, 16, 17 Critically, anti-viral responses such as the secretion of IFN-α have shown to be reduced in preterm infants following viral ligand stimulation.¹⁶ While still debated, this is thought to result in a reduced T-helper type 1 (Th1) response and heightened Th2 response that may lead to severe RSV disease.^18,19 Therefore, distinct characteristics of the innate immune system in preterm infants may critically shape their response to RSV infections and therefore susceptibility to severe RSV disease, although limited data is currently available.

We present a comprehensive analysis of the immune response to RSV in preterm and term infants cord blood mononuclear cells (CBMCs). We undertook immune profiling in 25 preterm and 25 term infants using high-dimensional flow cytometry, multiplex cytokine assays, neutralising assays, and RNA sequencing (RNA-seq) to identify a unique preterm RSV-specific immune signature that will aid in explaining their enhanced susceptibility to severe disease.

Methods

Sample collection

Cord blood was obtained from a cohort of 25 healthy preterm (30.4–34.1 wGA) and 25 term (37–40 wGA) infants from Tu Du Hospital, Ho Chi Minh City, Vietnam. Details of this cohort have been published previously.²⁰ Mothers were not asked about medication use during their pregnancy. Cord blood was not collected from infants who had:

• •Major or suspected major malformations, including congenital heart disease, genetic syndromes,
• •Clinical evidence of chorioamnionitis,
• •Rupture of membranes for more than 24 h,
• •Suspected or confirmed early onset sepsis.

Or if their mothers had:

• •Autoimmune disease or immunodeficiency syndrome, or immunosuppressant or immunomodifying treatment for more than 3 months.
• •Infection: human immunodeficiency virus, hepatitis B, hepatitis C, primary herpes simplex virus infection during current pregnancy.
• •Physical, psychiatric, or complex social situation where the mother and baby may not be able to fully participate such as maternal alcohol or substance dependency, issues about child protection.

Cord blood samples were transported to Pasteur Institute of Ho Chi Minh City, Vietnam for CBMC isolation and serum separation within 4 h of collection.²¹ CBMCs and sera were stored in liquid nitrogen and −80 °C respectively, until shipment to the Murdoch Children's Research Institute, Melbourne, Australia.

Ethics

This study was approved by the Pasteur Institute Ho Chi Minh City Ethics Committee (Ethics approval: 213/QD-PAS) and RCH Human Research Ethics Committee (HREC; 56904). Parents/guardians provided oral informed consent for the collection of cord blood from all infants.

Study reagents

RPMI-1640, fetal bovine serum (FBS), l-glutamine and penicillin-streptomycin were purchased from Sigma–Aldrich, St. Louis, USA. Live RSVA strain, live RSVB strain and A549 cell line was purchased from American Type Culture Collection (ATCC; Virginia, USA). At purchase, A549 cells were mycoplasma free, and aliquots were stored down and used for this study. Anti-mouse compensation beads were purchased from BD Bioscience, San Diego, CA, USA. All flow cytometry antibodies used, and their suppliers are indicated in Supplementary Tables S1 and S2

RSV preparation

For RSV stock preparation, RSVA and RSVB strains were grown in A549 cells and purified by centrifugation through 30% sucrose layer as described previously.²² The harvested virus was collected in DMEM culture medium containing 20% sucrose and aliquoted, then snap-frozen and stored at −80 °C until subsequent experiments. The titre of purified RSVA and RSVB stocks were determined by plaque assay using a previously published method.²³

Neutralising antibody assay

Maternally derived neutralising antibody (NAb) titres to RSVA and RSVB were measured in cord blood serum using a previously established protocol.²³ Briefly, a 96-well flat bottom plate was seeded to establish 100% confluency with A549 cells overnight. Serum samples were serially diluted, 1:2, and mixed with 200 plaque forming units of RSVA or RSVB and incubated for 1.5 h at 37 °C, 5% CO₂. Following this, media was aspirated, and agar was added prior to incubation at 37 °C, 5% CO₂ for 3 days. After, cells were fixed with 80% acetone and stained with a primary RSV F-protein monoclonal antibody (Merck, Darmstadt, Germany) and secondary Alexa Fluor-conjugated, anti-goat IgG (Gibco Life Technologies, Carlsbad, USA) antibody. Plates were read with a fluorescence ELISPOT reader (AID, Strasburg, Germany) using FITC (490 nm) to identify plaques. The titre was defined as the reciprocal of the dilution that results in 50% inhibition of RSV infection.

RSV stimulation assay and flow cytometry

CBMCs were thawed at 37 °C and resuspended with 10 ml R₁₀ media (RPMI-1640 medium supplemented with 10% FBS, 2 mM l-glutamine, 1000 IU penicillin-streptomycin) and then centrifuged at 500×g for 5 min. Following this, cells were resuspended in 10 ml of R₁₀ and counted on a haemocytometer (Marienfeld, Lauda- Königshofen, Germany) using trypan blue dye exclusion method to measure viability. For stimulation assays, 1 × 10⁶/ml CBMCs were left unstimulated (control) or stimulated with RSVA or RSVB at a multiplicity of infection (MOI) of 1 for 24 h and incubated at 37 °C, 5% CO₂. To account for the 20% sucrose used in RSV storage, a similar amount of sucrose was added to the control condition. The chosen MOI was based off our previous publications.^24,25 Supernatants were harvested and stored at −30 °C until use. The CBMC cell pellets from stimulated and unstimulated conditions were stored in 1 ml RNAlater (Sigma–Aldrich, NSW, Australia) and frozen at −80 °C until use for RNA extraction.

Following stimulation, CBMCs were washed with 1 ml PBS and centrifuged at 400×g for 5 min. CBMCs were then blocked (50 μl of 1% human FC-block and 10% normal rat serum in PBS) for 20 min on ice. CBMCs were then washed with 1 ml FACS buffer (PBS supplemented with 2% FBS and 2 mM EDTA) and stained with 50 μl of antibody cocktail (Supplementary Table S1) for 20 min on ice. CBMCs were then washed twice and resuspended in 100 μl FACS buffer for acquisition using the Cytek Aurora. Compensation was performed at the time of acquisition using compensation beads. Data was analysed using Flowjo v10.7.1 software. For flow cytometric data, all cells analysed were live, singlet leukocytes. Gating strategies are shown in Supplementary Figure S1.

Validation of FITC labelled RSV antibody staining and intracellular RSV infection assay

In this study, a fluorescein isothiocyanate (FITC) labelled RSV antibody (Abbexa Cambridge, UK) was used in flow cytometry staining to determine which immune cells were infected with RSVA. To validate the activity and specificity of the FITC labelled RSV antibody to RSVA, a plaque assay was performed in parallel with our established method (Supplementary Figure S2).²³

CBMCs were thawed at 37 °C and resuspended with 10 ml R₁₀ and then centrifuged at 500×g for 5 min. Following this, cells were resuspended in 10 ml of R₁₀ and counted on a haemocytometer using trypan blue dye exclusion method to measure viability. For stimulation assays, 1 × 10⁶/ml CBMCs were left unstimulated (control) or stimulated with RSVA at an MOI = 1 and incubated at 37 °C, 5% CO₂. After 6 h, CBMCs were blocked with 1x brefeldin A and 1x monensin for a further 18 h. Following the 24 h total stimulation, CBMCs were centrifuged at 400×g for 5 min, supernatants were removed and CBMCs were washed again in 1 ml FACS buffer by centrifuging at 400×g for 5 min. After washing, CBMCs were blocked for 20 min on ice, washed again and stained with 50 μl of antibody cocktail (Supplementary Table S2) for 20 min on ice. CBMCs were than washed twice and fixed with 100 μl of fixation buffer using the Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences, San Diego, USA) for 20 min on ice. Following fixation, CBMCs were washed twice with 1 ml of permeabilisation buffer at 400×g for 5 min then stained with 50 μl of the intracellular antibody cocktail and incubated on ice for 20 min. After staining, CBMCs were washed with 1 ml of permeabilisation buffer followed by another wash with 1 ml of FACS buffer. CBMCs were then resuspended in 100 μl of FACS buffer for acquisition on the Cytek Aurora. Compensation was performed at the time of acquisition using compensation beads. Data was analysed using Flowjo v10.7.1 software. Gating strategies are shown in Supplementary Figure S3. For visualisation of RSVA-infected immune cells, a Uniform Manifold Approximation and Projection (UMAP) was used through the Spectre package.²⁶

Multiplex cytokine/chemokine assay

A commercial multiplex bead array kit (27-plex human cytokine assay; Bio-Rad, New South Wales, Australia) was used to measure IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 (p70), IL-13, IL-15, IL-17 A, eotaxin, FGF-basic, G-CSF, GM-CSF, MCP-1, IFN-γ, TNF-α, IP-10, RANTES, MIP-1α, MIP-1β, PDGF and VEGF from supernatants according to manufacturer's instructions. Results were analysed on a Luminex 200 instrument (Luminex, Texas, USA) fitted with the Bio-Plex Manager Version 6 software and results were reported in pg/ml.

For IFN-α measurement, a commercial Enzyme-linked Immunosorbent Assay (ELISA) kit was used (R&D systems, Minnesota, United States) according to the manufacturer's instruction. The plate was read immediately after the assay at 450 nm (reference wavelength 630 nm) using a microplate reader and values were reported in pg/mL.

RNA extraction and library preparation for RNA-sequencing

RNA extractions were performed for CBMCs following 24 h stimulation with RSV by using RNEasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. As each sample had 3 conditions (unstimulated, RSVA and RSVB), a total of 150 samples were extracted for RNA. Extracted RNA was stored at −80 °C until use for RNA sequencing. RNA yields and quality were measured by Qubit RNA high sensitivity assay kit (Invitrogen, USA) for quality control (QC) prior to library preparation. Samples with >100 ng total RNA were selected for the cDNA RNA-seq library preparation. Samples were processed at Victorian Clinical Genetics Services (VCGS) using the Illumina TruSeq stranded mRNA Library Prep kit as per manufacturer's instructions. Sequencing was performed on NovaSeq 6000 with 20 million reads of 2 × 150 bp read length per sample. One term unstimulated sample did not meet these QC criteria and was excluded from RNA-seq analysis.

RNA-seq pre-processing and quality control

Raw RNA-seq data was received as fastq files which were quality checked with FastQC. Bowtie2 was used to align reads to the human transcriptome (GrCh37v70) and gene counts were derived using HTSEQ.^27,28 A Q20 cut-off was applied, and all samples were of high quality. RNA-seq data are available on the GEO repository, accession code GSE196134.

Identification of differentially expressed genes (DEGs)

The edgeR package was used to normalise total counts data (Trimmed Means of M-values approach) and construct the DGE list, and variation between samples was limited.²⁹ Non-protein coding genes and genes with an average counts per million (CPM) of <1.5 across all samples were excluded from analysis, leaving expression data 13,936 genes. The limma-voom package was used to identify differentially expressed genes (DEGs) comparing expression of genes in un-stimulated samples vs stimulated samples and DEG lists were annotated using the biomaRt package.^30,31 Genes that exhibited an FDR-adjusted (Benjamini–Hochberg) p-value <0.05 and fold change >1.5 were considered differentially expressed between groups (RSVA stimulated preterm vs RSVA stimulated term, unstimulated vs RSVA stimulated for both preterm and term groups).³² When comparing unstimulated and stimulated samples from the same group a paired analysis was ran to identify DEGs. For differences between preterm and term an unpaired analysis was ran. The design matrix for all analyses included sex as a co-variate to ensure differences observed were not due to sex. Functional enrichment analysis to identify biological pathways involved from identified DEGs was performed using the KEGG (Kyoto Encyclopedia of Genes and Genomics) database through the DAVID tool, and Reactome Pathway databases. Up- and down-regulated genes were analysed in isolation through DAVID, only pathways with an adjusted p < 0.05 (measured by a modified Fisher's exact test with a Benjamini–Hochberg adjustment) were presented.33, 34, 35

Statistics

The sample size for this study was based on plasmacytoid dendritic cells (pDC) IFN-α responses as this is a major contributor to innate antiviral immunity. Published data has shown a mean IFN-α level of 1500 pg/ml for term infants, with a standard deviation of 200 pg/ml.¹⁶ To detect a 35% decrease in IFN-α in children born preterm compared with those born at term, with 80% power, and a two tailed α of 0.05, a minimum sample size of 15 children born preterm and 15 term is required.

For neutralising antibody data, a Mann–Whitney U test was used to compare preterm and term infants. Neutralising antibody, flow cytometry and cytokine data were presented as boxplots (median, Q1–Q3, min–max whiskers) with dots outside the whiskers representing outliers (values greater or less than 1.5∗interquartile range). Principal component analysis (PCA), which allows us to observe trends in the data, was used through the ggplot2 package to graphically represent differences in cytokine response to RSV stimulation in preterm and term infant. To generate PCA plots, all raw data for each cytokine for each individual was included. PC1 and PC2 represent the primary and secondary trends of the data. A Friedman test was used to compare unstimulated and stimulated cellular and cytokine data within preterm and term infants. A Kruskal–Wallis test as used to compare frequency of cells and cytokine concentrations between preterm and term groups. All post-hoc analyses for multiple comparison testing were adjusted with a Benjamini–Hochberg test. Differences in the median values for cytokines, cellular and gene data were presented as median and 95% confidence interval (CI). For correlation matrices between DEGs in the RSVA stimulated condition between preterm and term infants, a Pearson's correlation coefficient was used. The data was graphically represented and statistically analysed using R version 3.6.1 and Graphpad Prism v8. All tests performed were two-tailed and a p-value <0.05 was considered significant.

Role of funders

The funders had no role in study design, data collection, data analyses, interpretation, or writing of the report.

Results

Sample cohort

Cord blood samples were collected from 25 preterm and 25 term infants. The gestational age for preterm infants ranged from 30.4 to 34.1 wGA (median 32.6, IQR 32.1–33.45) and from 37 to 40 wGA in term infants (median 39.1, IQR 38.05–39.75). In our preterm and term cohort, 13 (52%) and 7 (28%) were female, respectively. All infants were born via normal vaginal delivery and no infants developed early onset infection (Fig. 1a). Maternally derived neutralising antibodies were similar between preterm and term infants for both RSVA (median 6.76, IQR 6.02–7.24 and median 6.80, IQR 6.35–8.45 respectively) and RSVB (median 7.45, IQR 6.29–8.15 and median 8.1, IQR 7.16–8.96 respectively) (Fig. 1b).

Fig. 1.

Description of the study cohort. a) Schematic of sample collection, sample processing, data acquisition and cohort demographics. b) Maternally derived neutralising antibodies to RSVA and RSVB in infants born preterm or term. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Significance between preterms and terms was determined by a two-tailed Mann Whitney U-test.

Reduced pro-inflammatory cytokine production in preterm infants

Innate responses are critical to provide early control of RSV infection. To examine the innate immune response to RSV in preterm and term infants, we stimulated CBMCs with RSVA and RSVB (MOI = 1) for 24 h. Following stimulation, a large variance in cytokine, chemokine, and growth factor production for both preterm and term infants in response to RSVA and RSVB was observed by unsupervised PCA analysis using two principal components (PC1 and PC2) (Fig. 2a). The most variance was explained by PC1 (75.4% and 74.4% for preterm and term, respectively) and little variance was explained by PC2 (8.6% and 11.1% for preterm and term, respectively). No distinct differences were observed when preterm and term immune responses to RSVA and RSVB were plotted together. When identifying factors that contributed to the PC1 variance, preterm and term infants showed a similar profile of inflammatory cytokine, chemokine, and growth factor response following RSVA and RSVB stimulation, with important innate cytokines IP-10, IFN-α, IL-6, IL-1β and TNF-α contributing the most to this variance (Fig. 2b). While many cytokines and chemokines were elevated in term infants (IL-6, IL-10, IL-12, IL-15, eotaxin, GM-CSF, PDGF and VEGF) compared to preterm infants following RSVA stimulation, many of these differences were also observed at baseline in the unstimulated samples, indicating their different immunological status (Supplementary Figure S4).

Fig. 2.

Innate inflammatory and cellular response to RSV. CBMCs from 25 preterm and 25 term infants were left unstimulated (US; red) or stimulated with RSVA (green) or RSVB (blue) at an MOI = 1 for 24 h. a) PCA of the response to RSVA and RSVB stimulation in preterm and term infants using all raw cytokine data. Individuals with missing values for cytokines in any condition were excluded from PCA analysis b) Contribution of each cytokine, chemokine, or growth factor to the PC1 variance in preterm (red) and term (blue) infants. c) Comparison of innate immune cell subsets in response to RSVA and RSVB stimulation between preterm and term infants. IFN-α was measured by ELISA and all other factors by multiplex. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Significance between preterms and terms was determined by a Kruskal–Wallis test and adjusted for multiple comparisons using a Benjamini–Hochberg post-hoc test.

Previously, we have reported the cellular phenotypes of this cohort at baseline.²⁰ To determine whether innate immune cells from preterm and term infants differ in their response to RSV stimulation, we conducted high-dimensional flow cytometry on innate immune cell subsets. An increased frequency of mDCs were observed in terms infants compared with preterm infants for all conditions (unstimulated, RSVA-stimulation and RSVB-stimulation) (p < 0.0001; Kruskal–Wallis). For all other populations, similar frequencies were observed in preterm and term infants (Fig. 2c). This was consistent when the differences in median and 95% CI data was considered, with non-overlapping CIs for mDCs between preterm and term infants (Supplementary Table S4).

Transcriptomic analysis reveals differences in immune pathways between preterm and term infants in response to RSV

Stimulation with RSVA led to extensive transcriptional changes in preterm (5299 genes, 2874 upregulated and 2425 downregulated; red (Fig. 3a and b) and term infants (3977 genes, 2307 upregulated and 1670 downregulated; blue (Fig. 3a and b).

Fig. 3.

Differential gene expression to RSVA stimulation. CBMCs from 25 preterm and 24 term infants were stimulated with RSVA for 24 h prior to RNA extraction and RNA sequencing. a) Volcano plots (log10 (p-value) vs logFC) of DEGs comparing RSVA stimulation to unstimulated conditions and between RSVA stimulation for preterm and term infants. b) Venn-diagram of total DEGs between comparisons. Overlapping genes highlighted in yellow, pink, and purple underwent DEG analysis. c) Enriched pathways associated with upregulated DEGs. DEGs were determined by a as FC > 1.5 and an adjusted p-value of <0.05. An adjusted p < 0.05 measured by Fisher's exact test (Benjamini–Hochberg post-hoc test) was used to identify associated KEGG pathways.

There were 2846 genes that overlapped between the preterm and term infant response to RSVA (Fig. 3B, in black circle). Functional enrichment analysis of these overlapping genes showed enrichment for KEGG pathways associated with cytokine–cytokine receptor interactions, RIG-1 like receptor signalling and TLR signalling, indicating a strong conserved antiviral response to RSVA in both preterm and term infants (Supplementary Table S3). The 2156 DEGs in response to RSVA stimulation in preterm infants (large red circle) and the 937 DEGs in term infants (large blue circle) were not associated with any immunological pathways (Fig. 3b; Supplementary Tables S5 and S6).

In total, 464 genes (240 upregulated and 224 downregulated) were significantly different between preterm and term infants following RSVA stimulation (green; Fig. 3a and b). However, when considering genes that were different between preterm and term infants and in response to RSVA, we identified 297 genes in preterm infants that responded uniquely to RSVA stimulation (142 upregulated and 155 downregulated; small circle yellow, Fig. 3b), while in term infants, 194 genes responded uniquely (100 upregulated and 94 downregulated; small circle purple, Fig. 3b). Pathways associated with genes in these lists were identified through KEGG and reactome databases. In term infants, the upregulated genes (100 genes) were associated with cytokine–cytokine receptor interactions, IL-17 signalling, TNF signalling and viral protein interaction with cytokine and cytokine receptors (Fig. 3C; Supplementary Table S7). Similarly, pathways identified through the reactome database showed enrichment for cytokine signalling, particularly through the IL-10 pathway in term but not preterm infants (Supplementary Table S8). No immune response pathways were identified among the upregulated genes in preterm infants; these were related to cell cycle (Supplementary Tables S9 and S10).

In contrast, for RSVB stimulation, although stimulation with RSVB also led to several transcriptional changes in both preterm and term infants, there were negligible differences DEGs between preterm and term infants. In response to RSVB, there were 883 unique DEGs in preterm infants and 1157 unique DEGs in term infants, but these genes were not associated with immunological pathways (Supplementary Tables S11 and S12). Therefore, we did not explore further for RSVB (Supplementary Figure S5).

Reduced anti-viral gene expression following RSVA stimulation in preterm infants

To validate the differences in the 100 upregulated genes associated with the immune pathways identified by KEGG (Fig. 3c), these were compared between preterm and term infants. This showed increased expression for all genes (i.e., MMP1, CXCL1, IL19, IL36γ, CSF3, CXCL6, IL-1β, CSF2, CXCL2, IL-10, CD14, IL36RN, MMP14, EREG, XCL1, METRNL, TGF-α, IL1R1, MET, GDF15, TNFSF12, SOCS3, FGFR1, SOCS1) in term infants (Fig. 4a). These genes were then correlated to each other to identify whether immune co-ordination was similar in preterm and term infants. In term infants, two clusters representing strong immune gene co-ordination was identified, whereas only a weak immune co-ordination was observed in preterm infants (Fig. 4b). The expression of genes such as IL-10, XCL1, CSF2, CXCL1, CXCL2, IL36γ, SOCS1 and SOCS3 are critical for immune regulation, chemotaxis, and viral protection and these were significantly reduced in preterm infants compared with term infants following adjustment for multiple comparisons (Fig. 4c). This effect was only seen following stimulation and not in the unstimulated conditions (Supplementary Figure S6). For all genes, there was no overlap between the 95% CI in preterm and term infants strongly supporting the data observed (Supplementary Table S13).

Fig. 4.

Comparison of differentially expressed genes. a) Heatmap representation of immune-related DEGs identified through KEGG to RSVA stimulation between preterm and term infants (gene expression values shown as normalised Z-score). b) Correlation matrix of the DEGs presented in the heatmap in preterm and term infants. c) Comparison of the expression of selected DEGs between preterm and term infants due to their involvement in chemotaxis, immune regulation, and anti-viral activity. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. A Pearson's correlation coefficient was used for the correlation matrices. Darker red implies a stronger positive correlation and darker blue implies a stronger negative correlation. Differences between preterm and term infants was determined by an FDR-adjusted (Benjamini–Hochberg) test.

Monocytes and mDCs from preterm infants are susceptible to RSV infection

Given the different innate immune profiles observed between preterm and terms infants, we next aimed to identify which specific innate immune cells are susceptible to RSV infection and therefore contribute to the impaired immune responses observed in preterm infants. We stimulated CBMCs with RSVA for 24 h and identified RSVA infected cells from FITC labelling. We found that mDCs and monocytes were the primary cell types infected (Fig. 5a; Supplementary Figure S5). Preterm infant mDCs had a higher level of RSVA infection compared with term infants (p < 0.01, Kruskal–Wallis; Fig. 5b). RSVA infection of mDCs (RSV+) produced a greater cytokine response than RSV uninfected mDCs (RSV-) for all cytokines measured in both preterm and term infants (Fig. 5c). Similar levels of mDC-specific cytokine production was observed in preterm and term infants. This was further supported by the overlap in 95% CIs in the mDC specific cytokine response for preterm and term infants with the exceptions of IL-10 and IL-1β (Supplementary Table S14). A similar production of total (RSV+ and RSV- combined) mDC-derived IL-1β, IL-6, IL-8, and TNF-α in both preterm and term infants was observed (Supplementary Figure S7a). However, total mDC IL-10 was increased in term compared to preterm infants in both the unstimulated and RSVA stimulated conditions (Supplementary Figure S7a). No differences were observed in pDC IFN-α production between preterm and term infants, although the level of RSV infection in pDCs was low overall (<5%) (Supplementary Figure S7b). For monocytes, no differences in cytokine response were observed between RSV- and RSV+ cells from preterm and term infants, although all cytokines were elevated following RSV stimulation (Fig. 5d and Supplementary Figure S7c). Similarly there was overlap in 95% CIs for monocyte specific cytokine responses in preterm and term infants (Supplementary Table S14). Similar to our observations in the mDC population, total IL-10 in monocytes was higher in term compared to preterm infants in both the unstimulated and RSVA stimulated conditions (Supplementary Figure S7c).

Fig. 5.

RSVA infection of immune cells and the intracellular cytokine response. CBMCs from 25 preterm and 25 term infants were left unstimulated or stimulated with RSVA at an MOI = 1 for 24 h a) UMAP representation of immune cell populations with RSVA median fluorescence intensity (MFI) overlayed showing those populations with RSVA infection. This was performed on a subset of samples, n = 6 combining preterm and term samples. b) Comparison of immune cell populations that were infected by RSVA in preterm and term infants c) Intracellular cytokine response in RSV+ and RSV- mDCs. d) Intracellular cytokine response in RSV+ and RSV- monocytes. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Significance between RSV+ and RSV- conditions was calculated by a Friedman test. Significance between preterms and terms was determined by a Kruskal–Wallis test. All analyses were adjusted for multiple comparisons using a Benjamini–Hochberg post-hoc test.

Discussion

We present a comprehensive analysis of the immune response to RSV in preterm and term infant CBMCs using a combination of high dimensional immune profiling, multiplex cytokine assays and transcriptomic analysis. We found genes involved in cytokine–cytokine receptor interactions, IL-17 signalling, TNF signalling, viral protein interaction with cytokine and cytokine receptors, and the IL-10 pathway were upregulated in term infants compared to preterm infants. Additionally, we identified that genes associated with these pathways are highly co-ordinated in term infants compared to preterm infants. Furthermore, total IL-10 in mDCs and monocytes from preterm infants was reduced compared to term infants and mDCs were more susceptible to RSVA infection. Our data provides important insights that may help to explain why preterm infants are more susceptible to severe RSV disease.

Preterm infants are at higher risk for severe RSV disease than term infants. Previous studies have compared immune responses to a range of different pathogens, however, very few studies have focused specifically on RSV. We focused on innate immune responses as these are critical in the immune defence against primary RSV infection. The innate immune system is considered relatively immature in preterm compared to term infants.¹⁷ RSV infection is usually most severe during the first 6 months of life, suggesting that maternal NAb from natural infection alone may be insufficient to confer protection.³⁶ However, it is important to consider that palivizumab, nirsevimab and the recently approved bivalent, recombinant respiratory syncytial virus vaccine provide high NAb titres leading to effective prevention of severe RSV disease during early life.^3,4 We found that maternal NAb to RSVA and RSVB (in cord blood) were similar in preterm and term infants, suggesting that protective antibodies are not the only factor to consider in determining disease susceptibility among preterm infants, supporting a role for the dysregulation of innate immunity.

In vitro studies have demonstrated a reduced capacity of preterm infants to produce pro-inflammatory cytokines and chemokines (IL-1β, IL-6, IL-8, IL-12, TNF-α, IFN-α) in response to TLR2, TLR4, TLR7, TLR8 and TLR9 ligands.^15,37,38 Similarly, preterm infants had a reduced response for many cytokines following RSVA stimulation compared to term infants. We observed a reduced frequency of mDC in preterm infants similar to our previously published data.²⁰ During RSV infection, mDCs can be pathogenic, promoting Th2 responses leading to sustained inflammation and increased epithelial tissue damage. However, this is related to the expression of IL-4Rα and production of TGF-β which we found no differences in at the gene level. Therefore, the difference in frequency of mDC may have little consequence in preterm infants.^39,40

Term infants upregulated genes associated with anti-viral activity, neutrophil chemotaxis, and immune regulation (IL-10, IL-36γ, CXCL1, CXCL2, SOCS1 and SOCS3) which was not observed in preterm infants.41, 42, 43 The enhanced co-ordination between these genes in term infants suggests that preterm infants may have a reduced capacity to co-ordinate inflammatory responses, which is vital to promote viral clearance and reduce immunopathology. We found an increased gene expression of IL-10 in term infants compared to preterm infants. The role of IL-10 during severe RSV disease is controversial. In mice, IL-10 upregulation in response to RSV dampens harmful inflammatory responses and reduces cellular infiltration to limit pulmonary pathology.⁴⁴ Similarly, in mice, the presence of IL-10 during the acute response is required to regulate inflammation, while late induction of IL-10 is associated with increased disease severity.⁴⁵ Additionally, in human neonates IL-10 secreted from RSV infected B-regulatory cells was associated with an increased disease severity.⁴⁶ Therefore, the increased gene expression of IL-10 we observed suggests an increased capacity to regulate acute inflammation which would be beneficial during RSV disease.

Similar to other studies, we showed high infectivity of mDCs, monocytes, and to a lesser extent pDCs following RSV stimulation.^47,48 The higher infectivity of mDCs in preterm infants compared to term infants in our study may be detrimental as RSV-infected DCs are less effective at activating naïve CD4 T-cells leading to reduced proliferation and cytokine production.⁴⁹ Additionally, the non-structural proteins in RSV directly inhibit the maturation of DCs and impair the production of type I IFN secretion promoting Th2 polarisation.⁵⁰ Higher levels of total IL-10 in mDCs and monocytes from term infants were observed which is consistent with our IL-10 mRNA data. Early production of IL-10 is critical to regulating the inflammatory response to RSV, leading us to speculate that the reduced IL-10 in preterm infants may enhance lung pathology due to uncontrolled inflammation, although further studies are needed to confirm this.

Another interesting observation from this study was the reduced IL36γ in preterm infants. IL-36γ is reported to be protective against influenza infection in mice by improving immune regulation, promoting an M1 alveolar macrophage phenotype and promoting their survival by reducing apoptosis.⁵¹ M1 alveolar macrophages are critical in controlling infection and reducing pulmonary pathology though type I IFN production. Therefore, reduced IL-36γ in preterm (compared to term) infants may lead to enhanced susceptibility to severe RSV disease.⁵² Increased levels of CXCL1 and CXCL2 suggests a higher capacity for neutrophil chemotaxis in term infants but not preterm infants. This is interesting as preterm infants have fewer neutrophils at birth compared to term infants.⁵³ Moreover, our data found that these chemokine responses appear to be highly co-ordinated during the response to RSVA among term infants but not preterm infants. Whether the reduced chemotaxis in preterm infants is harmful or beneficial is uncertain as infants <6 months have previously shown no association between neutrophil gene overexpression and disease severity.⁵⁴ However, in adults, mucosal neutrophilic infiltration has been associated with disease susceptibility.⁵⁵ As neutrophils promote key antiviral responses by NK cells and CD8+ T-cells, a reduction in neutrophil chemotaxis in preterm infants could impair antiviral responses during RSV infection.¹⁷ We identified an increased expression of inflammatory genes related to cytokine suppression (SOCS1 and SOCS3) in term infants. While they are necessary to suppress over-active inflammatory responses and leukocyte egression, it is important to note that these genes can also be harmful during RSV infection.^54,56

While many DEGs in preterm and term infants were observed following RSVA stimulation, no differences were found with RSVB stimulation. This suggests that preterm infants may not exhibit altered gene responses to all pathogens.²¹ There is a 50% difference in the G-protein and a 10% difference in the F-protein between RSV A and RSV B genome, which are the two important immunogenic viral proteins.⁵⁷ It is still not clear whether RSV severity differs between RSVA and RSVB infections.⁹ Our unexpected result of RSVB stimulation showing no immune response related DEGs between preterm and term infants cannot be explained by the current epidemiological data and ongoing studies are aiming to explore this in more detail.⁵⁷ Similarly, as RSVA and RSVB were produced in the exact same manner and stimulated with the same MOI it is unclear why such differences were observed at a gene expression level beyond any strain-related differences.

Our study showed reduced gene expression of IL-10 and reduced total IL-10 in mDCs and monocytes from preterm infants. These results have important implications for strategies to protect preterm infants against severe RSV disease. For example, vaccine adjuvants able to enhance IL-10 secretion may be beneficial in preterm infants. In mice, an AS01-adjuvanted RSV prefusion F-protein vaccine led to increased IL-10 secretion, activation of TLR4 and promotion of Th1 immunity, suggesting that increased IL-10 could improve antiviral immunity.^58,59 Another study in mice using an RSV fusion protein vaccine with CpG-ODN (TLR9 agonist) and innate defence regulator (IDR) peptide adjuvant was found to be safe and was associated with enhanced viral clearance due to increased neutralising antibodies and a Th1-based response.⁶⁰ Furthermore, manipulation of the microbiome of preterm infants has been shown to protect against severe RSV disease. The decreased Bifidobacteria observed in preterm infants is associated with a reduced mDC derived IL-10 compared to term infants which supports our data.^61,62 Further studies on strategies that promote IL-10 production are needed. Mouse studies of exogenous IL-36γ delivery have shown that it improves bacterial and viral clearance and leads to reduced morbidity and mortality by improving pro-inflammatory and anti-viral responses.⁶³ Validating our findings of increased term infant IL-36γ and IL-10 in a clinical cohort would be critical to understanding disease susceptibility in preterm infants. This can be achieved by combining paediatric peripheral blood samples with mucosal samples from RSV infected infants to provide a more complete snapshot of the immune response to RSV and then comparing these factors between those with severe and moderate disease.

As moderate preterms comprise a large proportion of hospitalisations due to RSV infection it is crucial to understand why this group are more susceptible to severe RSV disease. This understanding will contribute to reducing the burden of disease. The sample size of this study and the broad range of immunological analyses undertaken including neutralising antibodies, cytokine and cellular responses, gene expression and cell-specific responses are strengths of this study. While we adjusted for sex as a known confounder in our analysis, the fact that in our cohort, both preterm and term groups had a similar mode of delivery, no clinical chorioamnionitis and were free from early onset neonatal infection, suggests that the role of other known confounders on our data would be minimal.²⁰ The lack of DEGs between preterm and term infants following RSVB stimulation is an area that needs to be explored in further detail for future studies. A limitation of this study is that we did not include an inactivated RSVA control for infection assays, thus, the percentage of infected cells may not entirely reflect productive infection. Whilst RSV stimulation of cord blood is likely to partly reflect real-world immune responses to primary RSV infection, it is a limitation that there may be some differences and thus our data should be treated with caution until verified in RSV infected infant cohort studies. Therefore, future studies characterising the immune response to RSV by using peripheral blood and mucosal samples from RSV-infected infants will be valuable to validate these findings. Additionally, birth cohort studies relating immune profiles in cord blood to RSV disease outcomes would be highly valuable.

In conclusion, we identified several important differences in preterm compared to term infant immune responses to RSVA stimulation as measured in cord blood. These involve reduced enrichment for genes involved in anti-viral activity, neutrophil chemotaxis, and immune regulation in preterm immune responses. These findings help to explain the immune contribution to preterm infants increased susceptibility to severe RSV during early life. While supporting data is available from other studies, our results need to be confirmed in clinical cohorts of hospitalised infants with RSV disease and will inform the development of interventions to improve protection against and inform future therapies for RSV in this vulnerable group.

Contributors

J.A performed the experiments, the analyses, and wrote the original draft; S.I, Y.Y.N, T.T.W and S.A contributed to data analysis, representation and critically revised the manuscript. C.M.T and L.Q.T, coordinated the clinical aspects of this study of samples, contributed to study design and revised the manuscript; V.T.T.D and P.V.T coordinated samples and performed isolation of CBMCs and revised the manuscript; B.T.H.N, D.N.X.T, P.T.P.T and L.T.B coordinated the study, enrolled patients, collected clinical data and cord blood samples and revised the manuscript; N.T.V and N.T.T contributed to the study design and revised the manuscript; B.N, M.L.K.T, D.W, K.M and D.G.P critically revised the manuscript and contributed to the interpretation of data. L.A.H.D and P.V.L conceived the study, developed the study protocol, and provided major input into the manuscript revision and analysis. J.A and S.I had access to and verified all data. J.A, L.A.H.D, and P.V.L made the decision to submit the manuscript. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Data sharing statement

RNA-seq data are available on the GEO repository, accession code GSE196134.

Declaration of interests

D.W receives funds from GSK to lead a study evaluating RSV testing in Australia. Payments are made to research funds. D.W also receives consultancy funds from Sanofi, MSD and Praxhub which are paid to research funds. D. W is a co-chair on TSANZ and a working group member of Therapeutic Guidelines Ltd, (Antibiotics, respiratory) Australian clinical guidelines. L.A.H.D receives grants from CEPI, PATH and BMGF where payments are made to the institution. P.V.L receives grants from CEPI where payments are made to the institution. M.L.K.T receives grants from NHMRC and Allergy and Immunology Foundation of Australasia. M.L.K.T receives royalties from the book “Food Allergies for Dummies”, and consultancy fees from the Abrocitinib advisory board of Pfizer. M.L.K.T has received flight support from APAAACI Congress and the CUHK Allergy conference. M.L.K.T has patents covering food allergy and pending patents from allergy treatments. M.L.K.T is the board director for AllergyPal, on the board of directors for APAAACI and an advisory board membership for AAA which are all unpaid. M.L.K.T has received probiotics from Metagenics for the PEAT study and is a CFAR3 investigator, associate editor for JACI global and on the editorial board of World Allergy Organisation Journal, Asia Pacific Allergy Journal and JACI In Practice. K.M received funding support from the BMGF, funding paid to the institute and is a member of the WHO SAGE committee. K.M is also an investigator on a observational study of adult pneumonia in Mongolia funded by Pfizer. K.M has received funding support to attend meetings by the WHO and Christian Medical College Vellore India. K.M has also participated on the data safety monitoring board of Novavax COVID-19 studies in an unpaid position.

Appendix A. Supplementary data

Supplementary Figure S1

Gating strategy for innate cells and B-cells from cord blood mononuclear cells. Innate cells were identified by CD3-CD19- expression on live single mononuclear cells. Innate cells were further categorised into NK cells (CD56+CD14-HLA-DR-) and HLA-DR+CD14+ cells. NK cells were further categorised into CD56^dim and CD56^bright. From CD56^dim NK cells CD16+ and CD16+CD57+ cells were gated. NKG2A+ expression was gated from total NK cells. From the HLA-DR+CD14+ fraction, cells were divided into monocytes (CD14+CD16-) and dendritic cells (HLA-DR+CD14-CD16-). Dendritic cells were further categorised into myeloid dendritic cells (CD11c+) and plasmacytoid dendritic cells (CD123+).

Supplementary Figure S2

Validation of RSV-FITC antibody for intracellular flow cytometric staining. A) Established plaque assay method using a primary antibody B) FITC labelled RSV antibody (1:100 dilution). C) Comparing PFU/ml of both assay procedures. The titre of RSV was calculated using the formula: PFU/ml = Average number of plaques/(Dilution factor × Volume of diluted virus added to the well). Data was presented as median and interquartile range.

Supplementary Figure S3

Gating strategy for RSVA infection of innate immune cells and the intracellular cytokine response. From live single cells, CD3-CD19- populations were gated to remove T (CD3+) and B (CD19+) cells. NK cells were gated as HLA-DR- CD14- CD56+ cells. HLA-DR+ cells were then divided into two cell populations based on CD14 and CD16 expression: monocytes (HLA-DR+CD14+) and dendritic cells (HLA-DR+CD14-). Dendritic cells were further classified into myeloid dendritic cells (CD123-CD11c+) and plasmacytoid dendritic cells (CD123+CD11c-). RSV+ and RSV- were gated on RSV-stimulated monocytes and mDCs based on RSV antibody expression. An example is shown in the mDCs. Cytokine expression was then gated on relevant cell populations and examples are shown in the mDCs.

Supplementary Figure S4

Comparison of cytokines in preterm and term infants following RSVA and RSVB stimulation. For IL-8 in preterms, n = 22 for RSVB. For IL-8 in terms, n = 22, n = 20 and n =19 for US, RSVA and RSVB respectively due to levels too high to detect. For MIP-1β in preterms, n = 24 and n = 14 for RSVA and RSVB respectively. In terms, n = 24, n = 23 and n = 17 for US, RSVA and RSVB respectively due to levels too high to detect. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Significance between preterms and terms was determined by a Kruskal-Wallis test and adjusted for multiple comparisons using a Benjamini-Hochberg post-hoc test.

Supplementary Figure S5

Differential gene expression to RSVB stimulation.a) Volcano plots of differentially expressed genes comparing RSVB stimulation to unstimulated conditions and between RSVB stimulation for preterm and term infants. b) Venn-diagram of total differentially expressed genes.

Supplementary Figure S6

Comparisons of unique DEGs in preterm and term infants at baseline (unstimulated). Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Differences between preterm and term infants was determined by an FDR-adjusted (Benjamini-Hochberg) test.

Supplementary Figure S7

Comparison of cell-specific cytokine production following RSVA infection in preterm and term infants.a). mDC specific cytokine production. b). pDC specific cytokine production. c). Monocyte specific cytokine production. Significance between RSV+ and RSV- conditions was calculated by a Friedman test. Data was presented as a boxplot (median and IQR), dots outside the whiskers represent outliers. Significance between preterms and terms was determined by a Kruskal-Wallis test. All analyses were adjusted with a Benjamini-Hochberg post-hoc test.

Supplementary Table S3

KEGG pathways for DEGs upregulated in both preterm and term infants in response to RSVA.

Supplementary Table S5

KEGG pathways for DEGs upregulated in preterm infants in response to RSVA but not different between the groups.

Supplementary Table S6

KEGG pathways for DEGs upregulated in term infants in response to RSVA but not different between the groups.

Supplementary Table S7

KEGG pathways associated with upregulated DEGs uniquely expressed by term infants.

Supplementary Table S8

Reactome pathways associated with upregulated DEGs uniquely expressed by term infants.

Supplementary Table S9

KEGG pathways associated with upregulated DEGs uniquely expressed by preterm infants.

Supplementary Table S10

Reactome pathways associated with upregulated DEGs uniquely expressed by preterm infants.

Supplementary Table S11

List of the 883 genes that were upregulated in response to RSVB in preterm infants but not different between the two groups. Data in the columns represent the logCPM for each gene for each individual in this study. M refers to unstimulated, A refers to RSVA stimulated, and B refers to RSVB stimulated samples.

Supplementary Table S12

List of the 1157 genes that were upregulated in response to RSVB in term infants but not different between the two groups. Data in the columns represent the logCPM for each gene for each individual in this study. M refers to unstimulated, A refers to RSVA stimulated, and B refers to RSVB stimulated samples.