Temporal Nasal Epithelial Gene Expression Patterns in Healthy Individuals Infected With Rhinovirus‐16, and Its Modulation by Carrot Rhamnogalacturonan‐I

ABSTRACT

Innate immune and interferon‐induced responses to in vivo nasal infection with rhinovirus (RV)16 in healthy subjects are accelerated by low‐dose dietary supplementation with carrot‐derived pectic polysaccharide rhamnogalacturonan‐I (cRG‐I), together with reduced duration and severity of symptoms. We aimed to further identify temporal mRNA responses by nasal epithelial cells (NEC) after RV16 infection, and to assess the effect of cRG‐I supplementation. NECs were obtained prior to (day(d)−55) and after 8‐weeks (d‐1) of supplementation (0, 0.3, 1.5 g/d), and on d3, d6, d9 and d13 after exposure to 100 TCID50 RV16. Transcriptome data were generated and analysed with the R2: Genomics Analysis and Visualization platform (https://r2.amc.nl). RV16 infection reduced expression of genes related to oxidative phosphorylation (d3), induced gene expression by interferon (d6‐9), and reduced expression of cilia‐related genes (d13). cRG‐I changed these responses. At low‐dose, gene expression of important transcription factors and effector molecules (IRF4, IRF8, RFX3, IL‐1B, CASP1) was enhanced markedly earlier (d3‐6). At high‐dose, cRG‐I induced expression of inflammasome‐related genes already after 8‐weeks supplementation. cRG‐I, in a dose‐dependent manner, significantly affected the sequence and intensity of genes that regulate pathways involved in anti‐viral responses and epithelial repair. This may underlie the reduced duration and severity of symptoms.

1. Introduction

Respiratory viral infections occur frequently and either remain subclinical or cause mild to severe symptoms. Whether a virus causes symptoms is determined by many aspects, including the viral species (e.g., infectivity, replicative phase) and the health status of the host (e.g., responsiveness of innate and adaptive immune system, viral clearance, comorbidities) [1, 2]. Once respiratory viruses are in the upper airways, they primarily infect nasal epithelial cells (NECs), which are the most abundant target cells present, but they can also infect macrophages [3]. The initial anti‐viral response involves the production of interferons and is triggered predominantly by replicating viruses in infected cells [4]. This response serves to limit viral replication, the spread of the virus, and to trigger subsequent immune responses, which should ultimately lead to clearance of the virus and recovery from the infection [5].

In addition to lowering the risk of infection, for example, by avoiding contact with infected individuals, there are several ways to reduce the impact of a viral infection, ranging from vaccination to interventions that modulate the anti‐viral response [6]. Whereas virus‐specific vaccination tends to attenuate symptoms of subsequent infections with the same or closely related viruses, interventions to modulate broader innate anti‐viral responses can potentially reduce the impact of a broader range of infections. Thus far, however, studies have yielded variable results [7]. We have recently shown that dietary intake of a low‐dose (0.3 g/day) carrot‐derived pectic polysaccharide rhamnogalacturonan‐I (cRG‐I) soluble fiber for 8 weeks, compared to a no‐dose (0 g/d) and a high‐dose (1.5 g/d) supplementation, significantly reduced the duration and severity of clinical symptoms after an in vivo challenge of healthy individuals with a common cold virus (rhinovirus 16, RV16) [8, 9]. The low dose of cRG‐I accelerated the expression of several interferon‐induced response genes in nasal epithelial cells and uniquely enhanced the expression of EIF2AK2, which encodes the crucial anti‐viral protein, protein kinase R, as well as secretion of the interferon‐induced mediator CXCL‐10 in nasal lavage [8]. Compared to the no‐dose group, both the low‐ and high‐dose groups cRG‐I accelerated and increased other innate immune responses.

Respiratory viral infections of NEC impact various cellular metabolic activities and their functioning, such as loss of cilia that affects mucociliary transport [10, 11]. An overall view of changes that occur over time in response to a respiratory viral infection, however, is still missing. Let alone for NEC that were infected in vivo, whereby NEC also integrates responses from other cells. We hypothesized that a more extensive analysis of sequential gene expression in nasal epithelial cells infected in vivo by RV16 could clarify the sequence of diverse responses triggered by a viral infection. And, further, we hypothesized that comparing the sequential gene expression between the different dosage groups could clarify why the low‐dose group had a favourable clinical outcome compared to the no‐ and high‐dose groups.

2. Methods

2.1. Study Subjects and Design

NEC transcriptome data were obtained from a randomly selected subgroup of participants (16 participants per dose group) from a trial in 177 healthy individuals studying the effect of cRG‐I on a standardized challenge with RV16 (n = 16 for each dose of cRG‐I (0 g/d (no‐dose), 0.3 g/d (low‐dose), 1.5 g/d (high‐dose); registered at https://www.onderzoekmetmensen.nl/en/trial/29458 under NTR6773 [8, 9]. In short, participants were confirmed RV16 sero‐negative at enrolment and, double‐blinded, took their daily dose of cRG‐I in a total of 3.5 g powdered supplement from day (d)−55 till the last visit on d13 after the challenge. After the initial 8 weeks of supplementation, during which participants did not contract a respiratory viral infection, they were inoculated with a low dose RV16 (100 TICD₅₀) in the nasal cavity, according to established safety recommendations [8]. NECs were collected from nasal brushes taken from these participants on set consecutive days, before and after viral exposure, as shown in Supporting Information S3: Figure S1. The RNA isolated from these brushes, all with RIN scores between 7 and 10 and thus passing QC for RNA seq, was used for cDNA preparation and Illumina sequencing (see supplement of reference [8]). Detailed information on the trial, participants, and procedures was reported earlier [8].

2.2. Data Processing and Analysis

The gene expression data used here were generated previously [8]. In short, the reads were aligned against the human reference genome GRCh37.75′, which was done using a short reader aligner based on Burrows–Wheeler transform. A default mismatch rate of 2% (three mismatches in a read of 150 bases) was used. The frequency of the reads mapped on the transcript was determined as counts, which is used as input for downstream analysis. Additionally, RPKM and FPKM (reads, respectively, fragments per kilobase of exon per million reads mapped) values were calculated. The RNA sequencing counts were loaded into the R2: Genomics Analysis and Visualization platform (https://r2.amc.nl).

Downstream analysis was performed using a variance‐stabilizing transformation, which was applied to normalize the matrix of counts. The DESeq.2 package within the R2 platform was used to determine the differentially expressed genes (DEGs), with pair‐wise comparisons for every time point respective to their baseline gene expression. A false discovery rate correction was applied for multiple testing, and an adjusted p value of < 0.05 was considered to be statistically significant.

Identified DEGs were analysed using the Enrichr platform [12] to look at associations and enrichment of specific gene sets and pathways from various database libraries. Enrichment was based on overlap between a gene set and the list of DEGs, and a false discovery rate adjusted p value < 0.05 was considered to be significantly enriched. Among the libraries present in Enrichr, was the Molecular Signature DataBase (MSigDB) Hallmark collection, which is a select number of gene sets that each describe a specific biological state or process with coherent expression [13]. Additionally, this database also has the gene ontology (GO) database, where gene sets are formed based on shared functionality. Enrichment of gene sets within this platform was based on overlap with the list of DEGs. Aiming to validate and complement the enriched gene sets, a pre‐ranked Gene Set Enrichment Analysis (GSEA; reference [14]) was performed on the log2FC gene expression values for all genes. The results of GSEA were visualized in R Studio (version 2026.01.0+392) using the ggplot2 package [15].

Gene signatures were based on the genes in the enriched pathways, identified by Enrichr analysis of the DEGs found in this study. Signature scores were based on the weighted average of z‐scores for all genes in the pathway. The z‐score for each gene is calculated by subtracting the mean interferon response gene expression of all individuals from the gene expression of the individual and divided by the standard deviation.

A General Estimating Equations (GEE) model was used to detect differences in the change in weighted z‐scores within the dose range and time interval as applied in the present study, as described before [8]. In short, the model was used in a stepwise fashion in an individual‐repeated (nested) design, using the change in z‐score as outcome as dependent parameter and as independent parameters various confounding factors (e.g., gender, BMI, infected (yes vs. no), alcohol consumption, sporting activities and vegetarian diet and start value of the respective dependent parameter, but most importantly time (applying the full interval of the current experiment (in days), dose (0–0.3–1.5 g/d) and interaction terms between time and dose, applying non‐linear time and dose relationships as well). Subsequently, two by two comparisons (between doses) regarding the change in weighted z‐scores were investigated within the GEE model too, applying a dummy approach with respect to dosing.

3. Results

3.1. Sequential Gene Expression in NECs Upon RV16 Challenge in Healthy Subjects

To reveal the genes transcribed in response to RV16 infection in healthy individuals, transcriptomes of NEC obtained on d‐1 (i.e., one day before RV16 challenge) were compared with those obtained on d3, d6, d9, and d13 post‐infection (Supporting Information S3: Figure S1). DEGs were determined for each time point after infection by comparing to d‐1 for each dose (Table 1). The no‐dose group reflects a normal response to a RV16 infection, and the peak for DEGs in this group is on d9, while the peak for the low‐dose cRG‐I group is earlier (d6), and for the high‐dose cRG‐I group later (d13). The majority of DEGs in the no‐dose group are upregulated after d3, whereas in the low‐ and high‐dose cRG‐I groups, the majority of DEGs appear to be downregulated after d3.

Table 1.

Differentially expressed genes after RV16 challenge. The total number of DEGs is listed for every time point, with the number of upregulated genes between brackets. Full overview of DEGs is listed in Appendix 1.

	Response to viral challenge (vs. day ‐1)
cRG‐I	Day 3	Day 6	Day 9	Day 13
No‐dose	12 (0)	173 (162)	257 (237)	8 (7)
Low‐dose	40 (39)	676 (257)	8 (6)	91 (55)
High‐dose	3 (0)	9 (7)	245 (38)	763 (311)

3.2. NEC Transcriptome Profile After Rhinovirus Infection in the No‐Dose Group

In the no‐dose group on d3, 12 DEGs were identified that were all downregulated. In particular, we found multiple genes encoded by mitochondrial DNA that are part of the oxidative phosphorylation system complex 1 (MT‐ND1, MT‐ND2, MT‐ND4) and complex V (MT‐CO1, MT‐CO2, MT‐CO3). A pre‐ranked GSEA of the changes in gene expression revealed downregulation of the Hallmark oxidative phosphorylation gene set (Supporting Information S3: Figure S2). Additionally, despite the absence of upregulated genes, various immune gene sets were upregulated.

The number of DEGs increased on d6 (n = 173) and peaked on d9 (n = 257). Analyzing these genes with the Hallmark gene set from mSigDb, showed that most upregulated genes were associated with immune‐ and antiviral gene sets (Figure 1), with the interferon gamma response set being most prominent. Notable upregulated DEGs found on d6 and d9 included the main docking site for rhinovirus, ICAM1, immune transcription factors IRF4 and IRF8, and the chemokines CXCL9 and CXCL10. On d13, eight DEGs were identified, with most of the upregulated genes being associated with the antiviral interferon response, similar to d6 and d9.

Figure 1.

Hallmark gene set enrichment in the no‐dose group after nasal rhinovirus challenge. Enrichment was based on the nasal epithelial cell (NEC) gene expression data obtained at Day 6 (A) and Day 9 (B) after viral infection. Shown are the top 10 most enriched gene sets based on overlap with the DEGs found at this timepoint (left panel). These top 10 enriched terms for the input gene set are displayed based on the −log10(p value), with the actual p value shown next to each term. Gene sets with a p value < 0.05 were considered to be significantly enriched and colored blue. An asterisk (*) next to a p value indicates the term also has a significant adjusted p‐value (< 0.05; Benjamini–Hochberg method for multiple testing). For the significantly enriched gene sets, the number of overlapping DEGs is listed. Additionally, a pre‐ranked Gene Set Enrichment Analysis (GSEA) was performed on the log2FC gene expression values for all genes, and the top 10 most significantly up‐ or down‐regulated gene sets are displayed (right panel). Significance was determined with a false discovery rate‐adjusted q value < 0.05.

3.3. Low‐Dose cRG‐I Alters NEC Transcriptome Profiles (Interferon Response, Cilia and Mitochondrial Encoded Genes) in Response to Rhinovirus Infection

In earlier work, we showed that 8‐weeks of supplementation with low‐dose cRG‐I led to an accelerated anti‐viral response in healthy subjects after a challenge with RV16 [8, 9]. Analyzing the 40 DEGs, 39 of which were upregulated (Table 1), on d3 with the Hallmark gene set library showed that these were mostly associated with interferon responses (Supporting Information S3: Figure S3A). In particular, there was upregulation of genes involved in recognition of viral RNA (RIGI, MDA5) and the innate immune response to viral infection (STAT2, PKR, OAS1, OAS2, OAS3). Most of these genes were also differentially expressed (upregulated) in the no‐dose group, but at a later time point (d9). The downregulated mitochondrial‐encoded genes found in the no‐dose group on d3 were absent in the low‐dose group.

On d6, we found the highest number of DEGs in the low‐dose group (n = 676) with the majority of genes downregulated (Table 1). Hallmark gene set analysis showed upregulation of interferon and antiviral gene sets, but only the bile acid metabolism gene set was significantly downregulated (Supporting Information S3: Figure S3B). Expanding the GSEA to the Gene Ontology Biological Process database revealed downregulation of various gene sets associated with cilia structure and function (Supporting Information S3: Figure S3C). Cilia are organelles located in the apical cellular membrane, and ciliary beating by the airway epithelial cells is an essential function for mucociliary transport [16]. Therefore, we looked in more detail at the crucial regulatory genes for cilia in relation to viral infection [17].

Expression of FOXJ1, an important regulator of cilia genes, showed a reduction in the low‐dose group, but this was not significant (log2FC = −0.68). However, the transcriptional co‐activator of FOXJ1‐mediated cilia expression, RFX3, was significantly reduced (Day 6 vs. Day 1, FDR‐adjusted p value = 0.03026, Log2FC = −1,097768279; also see Appendix 1). In line herewith, some genes known to be regulated by these two regulators, CETN2, SPAG6, DNALI1, and TUBA1A, were all significantly (see Appendix 1) downregulated. An accelerated decrease in expression of these cilia genes in the low‐dose group was also observed across the dosage group in comparison with the no‐dose group (Figure 2A).

Figure 2.

Low‐dose cRG‐I supplementation reduces cilia gene expression in response to RV challenge. Shown are the z‐scores based on the expression of cilia genes throughout this study. Additionally, the gene signature score (weighted z‐score for all genes) is shown. The order of study subjects was the same for each timepoint (A). The change in weighted z‐values for cilia gene expression in NEC in response to RV16 challenge was shown. Each symbol represents the mean +/− 1x standard error of the mean of at least 15 individuals (B). * denotes a significant difference (p < 0.05) between the observations of the complete time interval of the low‐ and the high‐dose.

To examine the changes in cilia gene expression, a signature score was created (Appendix 2). Compared to d‐1, the signature scores were significantly lower in the low‐dose group on d6, d9, and d13, whereas in the no‐dose group this was only on d13 (Supporting Information S3: Figure S4). No significant changes were found in the high‐dose group.

Further assessment of the changes in the scores after infection, GEE analysis showed that there was a decrease in weighted z‐values for cilia‐related genes when all doses were taken together (p < 0.001) (Figure 2B). Additionally, this decrease was larger for the low‐dose group than for the high‐dose group (p < 0.05). Interestingly, when observing the changes in weighted z‐values for cilia gene expression, the values of the low‐dose group return to baseline on d13, while those in the no‐dose group continue to decrease, suggesting a faster recovery in the former than in the latter (Figure 2B).

3.4. High‐Dose cRG‐I and the NEC Transcriptome Profiles in Response to Rhinovirus Infection

On d3 in the high‐dose group, three genes were downregulated, none of which were related to the innate or antiviral response (Table 1). On d6, 9 DEGs were identified, these included markers for the innate immune response (OAS2), as well as genes related to the anti‐viral interferon response (IFIT1, CXCL10, CXCL11, ISG15), that were all upregulated. However, the number of DEGs related to the immune response and the total number of DEGs were substantially lower than for the no‐dose and low‐dose groups at this time point. The number of DEGs increased on d9 and peaked on d13 in the high‐dose group (Table 1). At both time points, the upregulated genes associated with the interferon response gene sets, were similar to the two other groups (Supporting Information S3: Figure S5). However, we also observed that the majority of genes at these time points were downregulated and associated with various immune and inflammatory pathways.

3.5. DEGs by Comparing the Groups

DEGs between groups for each time point are shown in Table 2.

Table 2.

Differentially expressed genes between groups in this study after RV16 challenge. The total number of DEGs is listed for every time point, with the number of upregulated genes between brackets.

	Timepoint in response to viral challenge
	Day 3	Day 6	Day 9	Day 13
Low‐ vs. no‐dose	1116 (682)	4 (0)	106 (103)	0
High‐ vs. no‐dose	412 (357)	1 (1)	4 (3)	1 (1)
High‐ vs. low‐dose	0	1 (0)	2 (2)	0

The most DEGs in the within‐group comparisons were observed in the high‐dose group on d9 and d13 (Table 1), while the between‐group comparisons show the largest differences in the comparison no‐dose versus low‐dose cRG‐I, on d3 and d9. Most of the DEGs in response to cRG‐I are primarily seen for the d3 comparisons.

3.6. Comparison Between Low‐ and High‐Dose Groups Versus No‐Dose

We identified 1116 DEGs between the low‐dose and no‐dose group on d3 (Table 2). The upregulated genes in the low‐dose group were strongly associated with inflammation and immune‐related pathways (Supporting Information S3: Figure S6). These included both innate markers, such as CXCL8 and IL1B, and their regulatory pathways, and antiviral interferon response genes, like IFIH1 and IFIT2. Moreover, IL1B, the main effector of the NAIP/NLRC4 inflammasome and NLRC4, was also upregulated in the low‐dose group when compared to the no‐dose group.

The downregulated genes on d3 in the low‐dose group showed significant associations with various pathways related to cellular energy metabolism, with aerobic glycolysis most strongly associated (Supporting Information S3: Figure S6). No significant enrichment of genes of glycolysis, however, was observed in GSEA. Still, we feel this is an important finding as it is known that rhinovirus infection induces glucose uptake and glycolysis in infected cells for optimal viral replication [18]. Indeed, the association did include DEGs that are part of the glycolysis pathway and are responsible for importing glucose into the cell, SLC2A1, and subsequent conversion for energy metabolism, such as HK1, GPI, and ALDOA. SLC2A1 and HK1 were also downregulated DEGs in the high‐dose group when compared to the no‐dose group at this point in time.

To examine the effect of cRG‐I supplementation on aerobic glycolysis after viral infection, we created a signature score for genes related to this pathway (Appendix 2) and compared the different dose groups on d3 after the rhinovirus challenge (Figure 3). Both the low‐ and high‐dose groups had significantly lower weighted expression z‐values of the aerobic glycolysis genes, when compared to the no‐dose group. This decreased expression was strongest in the low‐dose group.

Figure 3.

Differences in aerobic glycolysis gene expression between the different cRG‐I doses on Day 3. Shown are the signature scores for each individual subject at this point in time, coloured for their expression of the HK1 gene. Dosage group means were compared to the no‐dose group with Welch's ANOVA (*p < 0.05, ***p < 0.001).

3.7. Transcriptional Expression Over Supplementation Period Prior to the Challenge With Rhinovirus

To determine whether 8 weeks of dietary supplementation with cRG‐I affected the NEC transcriptome before the rhinovirus infection, we compared transcriptomes from d‐55 with those on d‐1, for the no‐, low‐ and high‐dose groups. The number of DEGs, within and between group comparisons, is shown in Table 3.

Table 3.

Differentially expressed genes (DEGs) as a function of 8 weeks of supplementation with cRG‐I (before infection). Shown are the DEGs within groups (A) and between groups (B). The number of upregulated genes is shown in brackets.

A Supplementation effect within groups (Day‐1 vs. Day 55)
No‐dose	1 (1)
Low‐dose	2 (1)
High‐dose	90 (73)
B Supplementation effect between groups for Day‐1
Low‐dose vs. no‐dose	1 (0)
High‐dose vs. no‐dose	841 (789)
High‐dose vs. low‐dose	0

DEGs were not expected for the no‐dose group; however, MT‐TT, which encodes mitochondrial threonyl tRNA synthetase (TARS2), was increased on d‐1 compared to d‐55. In the low‐dose group, KRT14 (encoding (cyto)keratin 14) was downregulated, and the non‐coding RNA SMYD3‐IT1 (SET and MYND domain‐containing 3 intronic transcript 1) was upregulated after 8 weeks of intervention. In marked contrast, 90 DEGs were found in the high‐dose group after 8 weeks intervention, 73 of which were upregulated and 17 downregulated. These included genes involved in specific immune responses (Supporting Information S3: Table S1). Association of the 91 DEGs with the Gene Ontology (GO) Biological Process database also showed an enrichment of various immune response‐related gene sets (Supporting Information S3: Table S2). Multiple of these associated gene sets showed overlap with the upregulated CASP1, NAIP, and NLRC4 genes. These genes are all part of the NAIP/NLRC4 inflammasome, an important cascade in the innate host defense system [19].

When comparing expression between the different groups at d‐1, a single DEG was found in the low‐dose group compared to the no‐dose group, the HLA‐DRB5 gene. Interestingly, 841 DEGs were identified between the high‐ and no‐dose groups; these were associated with various immune pathways (Supporting Information S3: Figure S7). This was strongest for pathways centered around TYROBP and chemokine signalling, with a strong association and upregulation of the innate immune mediator IL‐8 (CXCL8) and its receptors CXCR1 and CXCR2. Both these pathways were significantly enriched and upregulated (Supporting Information S3: Figure S7).

In order to examine the effect of the cRG‐I supplementation on these pathways, TYROBP (based on WikiPathways WP3937 and WP3945), and IL‐8 (based on the PID_IL8_CXCR2_PATHWAY) gene signatures were created (Appendix 2) and assessed for each individual. When comparing the different dose groups, a subset of individuals (n = 8) in the high‐dose group shows increased expression of these gene signatures after 8 weeks of dietary intervention, but not before infection with rhinovirus (Figure 4A).

Figure 4.

Individual dose‐dependent effects of cRG‐I on inflammation regulatory pathways. Gene signatures were created and plotted for TYROBP and CXCL8 (IL‐8) signaling on d‐55 and d‐1 in all dose groups (A). Similarly, the gene expression values for the inflammasome effectors CASP1 and IL1B were plotted (B), and the inflammasome components NLRP3 and NLRP6 (C). The same subset of 8 individuals in the high‐dose group is marked in each panel by a black outline.

In addition to these gene sets, we found a similar association with the nucleotide‐binding oligomerization domain (NOD) pathway. The association of the DEGs and the NOD pathway was based on various genes related to inflammasome activation and subsequent signalling. The inflammasome comprises various innate immune receptors and their activation cascades, that share the effectors caspase 1 and IL‐1β [20]. To assess the effect of the cRG‐I supplementation on inflammasome activation, the gene expression values for these effectors was compared between groups (Figure 4B). Similar to the TYROBP and CXCL8 signature scores, the same subset of individuals from the high‐dose group showed higher expression of these genes.

As IL1B and CASP1 are the main effectors of inflammasomes and were differentially expressed before and after RV16 infection in subjects that received cRG‐I, we examined this pathway in more detail. Inflammasomes are comprised of various innate immune sensors and signaling pathways that can be activated by infectious, microbiome, or host signals [21]. Upon activation, the inflammasome sensors will recruit pro‐caspases, and activation of these caspases results in inflammatory, pyroptotic, or apoptotic signaling. We found multiple inflammasome sensors to be upregulated in the high‐dose compared to the no‐dose group at d‐1, including NLRP3. In general, activation of the inflammasome consists of two steps, as it needs to be primed before it can be activated [22]. In case of the NLRP3 inflammasome, priming leads to the expression of key inflammasome components NLRP3, pro‐caspase 1, and pro‐IL1B. All three of these corresponding genes were upregulated prior to infection in the high‐dose group when compared to the no‐dose group, as well as the non‐canonical caspases 4 and 8.

Upon activation by RNA viruses, caspase 1 will be activated by the NLRP3‐inflammasome complex and subsequently cleaves and activates IL‐1β and IL‐18, resulting in inflammatory signaling [21]. Furthermore, activation of caspase 4 leads to pyroptotic signaling, as well as activation of the NLRP3 inflammasome complex, potentially creating a positive feedback loop. We also found upregulation of the inflammasome sensors NLRP6 and NLRP12. NLRP6 functions similarly to NLRP3, as it is upregulated in the presence of external signals and upon activation by microbe or damage‐related stimuli, will activate caspase 1 and caspase 4/5, resulting in the previously described signaling. Lastly, NLRP12 can regulate the same caspase 1/IL‐1β activation cascade as NLRP3, NLRP6, and NLRC4, but has also been described as a negative regulator of antiviral interferon responses [23]. Interestingly, the same subset of individuals that showed increased expression of the other discussed inflammatory components showed enhanced expression of NLRP3 and NLRP6 (Figure 4C).

Despite the difference in number of DEGs found between the high‐ and no‐dose group (841 DEGs), and the low‐ and no‐dose group (1 DEG), no DEGs were found comparing the high‐ with the low‐dose group. The apparent reason for failing to find DEGs between the high‐ and low‐dose groups is due to the heterogenous expression of genes, particularly in the high‐dose group (for example Figure 3, and Supporting Information S3: Figure S4). This heterogeneity appeared to be strongest in the high‐dose group, suggesting a dose‐dependent effect of cRG‐I intervention.

4. Discussion

We performed an in‐depth analysis of transcriptomic profiles over time in NEC from healthy volunteers in response to an in vivo challenge with RV16, and assessed the dose‐dependent effect of cRG‐I on these profiles.

In the no‐dose group, RV16 infection resulted in early changes in gene expression of a number of crucial response pathways. First, there was suppression of genes (d3) related to the mitochondrial respiratory chain and, therefore, likely suppressing oxidative phosphorylation. This has not been shown before for rhinoviruses in vivo in humans, but was shown for other positive single stranded RNA viruses such as coronaviruses, a.o. SARS CoV2 [24, 25]. The reason for this suppression is unclear, but it is known that rhinovirus infections can reprogram metabolic pathways to increase viral replication [26, 27]. As rhinoviruses and coronaviruses replicate in de novo‐synthesized replicating organelles, the metabolic change may assist the formation of these lipid‐containing organelles.

The enhanced expression of genes related to inflammasomes (d6‐d9) is in line with its essential role in the innate immune response to viral infection, leading to IL‐1β and IL‐18 production as well as pyroptosis [28]. And, so is the induction of anti‐viral interferon response genes (d6–d9), which may be driven by autocrine innate interferons and/or, for example, by interferon from macrophages/monocytes.

The reduction in cilia‐related genes (d13) may be a direct effect of RV16 infection, possibly preventing infection of the (other) ciliated cells [29]. Alternatively, ciliated cells are the major target cells of rhinoviruses, and their replication may reduce the number of ciliated cells due to cytopathic effects. Rhinovirus infection, however, causes no marked cytopathic effects [29].

For the responses to the viral challenge in the low‐dose group, we confirmed and extended the accelerated expression (d3) of interferon‐induced genes as compared to the no‐dose (and high‐dose) group. Of particular interest are the increased expression of IRF4 and IRF8, encoding transcription factors that regulate type 1 interferon production and inflammatory responses, respectively [30]. Inflammasome‐related genes showed an early peak at d3 and declined subsequently. Also, gene expression related to cilia was downregulated by d6 and, in particular, by d9, followed by an apparent rebound at d13. There was also a remarkable metabolic shift in that genes related to oxidative phosphorylation were reduced at d3 in the no‐dose group, while the genes related to aerobic glycolysis were reduced in the low‐dose group. The apparent lack of a reduced expression of mitochondrial genes may already have occurred before d3, when the first sample was collected after the viral challenge. cRG‐I, however, does seem to cause a genuine metabolic shift in response to RV16 infection by reducing the expression of genes related to aerobic glycolysis. Inhibition of aerobic glycolysis has been shown to impair replication of rhinoviruses [10, 18].

The high‐dose group showed a delayed interferon response gene profile compared to the low‐ and no‐dose groups. Expression of the inflammasome‐related genes in the high‐dose group, including distinct NLRP3 inflammasome‐related genes, however, was higher at baseline (d‐1) and remained high during RV16 infection, although there was an apparent reduction by d13. Together with our earlier reported data (figure S2 in reference [8] and reference [9]) this suggests that the high‐dose also leads to increased immune readiness or priming (e.g., enhanced expression of inflammasome‐related genes, slightly higher levels of CXCL8 and CXCL10, low influx of neutrophils), which may result in a faster response to a pro‐inflammatory stimulus. Similarly, the NLRP3 inflammasome requires a priming step to be effectively activated [31]. This priming process is generally initiated by microbial stimuli, however, this priming can also be triggered in a sterile environment [32]. As transcriptomic data are unable to detect the phosphorylation required for downstream activation of the inflammasome, it is unclear whether this potential priming leads to activation. The absence of inflammatory symptoms in these subjects prior to infection, as supported by the absence of neutrophil activation, that is, no increase in released myeloperoxidase (MPO), indicates that the inflammasome is not activated (figure S2 in reference [8]). Whether this priming of innate immune responses during the initial supplementation period is related to the delayed interferon response gene profile in the high‐dose group is unknown.

Of note, 8 out of 16 of the individuals in the high‐dose group of this nested study showed high gene responses to this dietary supplementation (Figure 4), which suggests that this effect is not coincidental. As the list of DEGs involves multiple markers of neutrophils, we also reanalysed the cell differentials in nasal lavage. Only 3 of 8 individuals with high and 3 of 8 individuals with low gene responses, had sufficient cell numbers in the nasal lavage, collected at d‐1, for a reliable cell differentiation. Those with high gene responses showed a significantly (p = 0.005) higher neutrophil count (n = 3; 552 ± 152 neutrophils per mL) versus those with low gene responses (n = 3; for example 17 ± 11). None of the demographic parameters distinguished these eight subjects from the rest of the high‐dose group, and we also excluded that this effect may have been due to the presence of common respiratory viruses [8]. And, as the WURSS‐21 questionnaire only revealed a slightly higher score for “feeling tired” in some subjects across all groups on d‐1, this indicates that there were no underlying airway infections. We were unable to determine whether the presence of bacterial commensals may underlie this responsiveness. It is worth noticing that a similar dichotomy in gene responses was not seen in the low‐dose group.

The cilia‐related gene expression in the high‐dose group did not decrease as much as in the low‐ and no‐dose groups, although we noticed considerable variation between individuals in this group. This variation may underlie the very few DEGs in the high‐ versus low‐dose group comparisons. In the low‐dose group, KRT14, encoding cytokeratin 14 and expressed predominantly by basal cells, decreased during the 8‐week supplementation, whereas the non‐coding RNA (SMYD3‐IT1) was increased. When basal cells differentiate, they can lose cytokeratin 14. In theory, this may have led to an overrepresentation of ciliated cells, and might explain the early and prolonged decline of cilia‐related genes in the low‐dose group upon viral infection (Figure 2). Since DEG comparisons of d‐55 with d‐1, and comparisons on d‐1 between the groups did not reveal any cilia‐related DEGs, we consider the presence of more ciliated cells an unlikely explanation. Also, as MKI67 encoding Ki‐67, a marker of proliferation, was not affected by cRG‐I. It is, as yet, unknown what this early and prolonged decline of cilia‐related genes truly reflects. It might reflect a more pronounced anti‐viral response and effective clearance of infected ciliated epithelial cells by, for example, professional phagocytes in the low‐dose group, possibly due to an initial, more extensive viral replication in these cells.

The role of stable intronic transcripts is unclear, but that it involves an intron of SMYD3 is interesting, as SMYD3 has methyltransferase activity, which may epigenetically regulate gene expression. Therefore, although speculative, the low‐dose cRG‐I may affect the differentiation of NEC. Further studies, with sampling at earlier time points (e.g., d‐20, d‐40), may be required to assess the effect of cRG‐I on differentiation of NEC.

Whereas this in vivo approach strengthens the relevance of the findings, there are some weaknesses to be considered. Transcriptomic changes do not imply that the actual processes, such as protein expression and activity, at the cellular level are impacted. Still, we did observe effects on the course of the RV16 infection, for example, CXCL10 production, and on symptom severity and duration, making it likely that expression of proteins and their activities were affected. In addition, the involvement of multiple genes in the same pathway strengthens the potential contribution of these genes. Our results are also in line with previous in vitro studies, assessing transcriptomic profiles over time after rhinovirus infection of human polarized NECs [33, 34]. With this in vitro model, gene expression of viral replication sensors (RIG‐I, MDA5), specific chemokine genes (CXCL10, CXCL11), and interferon‐stimulating genes (ISGs; OASL, RSAD2, ISG15) were upregulated at Day 3, and another set of ISGs (IFI44L, IFIT1, IFI6) was upregulated at Day 14 post‐infection. These genes were also upregulated in the no‐dose and high‐dose group at Day 6, 9, and later, and earlier in the low‐dose group [8]. The apparent delay in increased gene expression for the no‐dose group, as compared to the in vitro data, is likely due to a delay of significant viral replication in the in vivo setting. This delay may also explain why we have not seen the second wave of ISGs. In several other studies [33, 34], as in this study, significant effects on FOXJ1 gene expression, a major regulator of ciliogenesis, were not seen upon rhinovirus infection, although we did notice a trendwise reduction. Interestingly, infection with a more pathogenic strain RV‐C15, but not with RV16 (RV‐A16), led to a marked reduction of FOXJ1 gene expression [35]. Contrary to this result, the authors [33, 34] did show loss and subsequent regrowth of cilia upon rhinovirus infection. This is in line with the RV16‐induced reduced expression of cilia‐related genes in both the no‐ and low‐dose groups. In the low‐dose group, we even noticed a significant initial reduction of RFX3, the transcriptional co‐activator of FOXJ1‐mediated cilia expression, and the reduced expression CETN2, SPAG6, DNALI1, TUBA1A, which are regulated by FOXJ1 and RFX3. In line with the regrowth of cilia, in the low‐dose group, we noticed an increase in cilia‐related gene expression by Day 13. For all of this, it is unclear whether the observed effects apply to the infected cells only or also to neighbouring non‐infected cells. And finally, whereas it is clear that cRG‐I affects the transcriptome of NEC, the underlying mechanism by which cRG‐I affects NEC is still uncertain. We have previously postulated that dietary supplementation with cRG‐I exerts its beneficial effect via a dual mode of action: (i) by directly modulating responsiveness of immune cells via pattern recognition receptors leading to priming/training of innate immune cells that migrate from the gut mucosa to more distant tissues, for example, the airways; and (ii) indirectly via modulation of the gut microbiota leading to production of active metabolites like short chain fatty acids that can influence immune responsiveness and barrier function [36, 37, 38].

In conclusion, low‐dose cRG‐I changes the responsiveness of NEC to a respiratory viral infection. Both the initiation of the anti‐viral IFN response as well as the recovery phase were markedly accelerated in the low‐dose group, minimizing the impact of RV16 infection on symptom severity and duration as well as on quality of life.

Author Contributions

Jasper Mol, Lara Ravanetti, Ruud Albers and René Lutter conceived the study, Lara Ravanetti provided initial analyses, Jasper Mol analyzed the data in detail, Wim Calame performed statistical support and analyses, and Jasper Mol and René Lutter wrote the manuscript, with critical comments by Sue McKay, Ruud Albers, and Jan Koster. Abilash Ravi provided and commented on the data, Richard Volckmann and Jan Koster developed methods to analyze sequential transcriptomic data. All authors had full access to all the data in the study and had final responsibility for the decision to submit the study for publication.

Ethics Statement

The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Medical Ethics Committee of AMC, the Netherlands (protocol NL62623.018.17 and clinical trial registry number: NL6559).

Consent

Informed consent was obtained from all subjects involved in the study.

Conflicts of Interest

R.L. received funds from NutriLeads to perform the current study, received funding from ZonMw for studies outside the submitted work, and has received a fee for a webinar from Pfizer. R.A. is the founder of NutriLeads and owns shares. S.M. is an employee of NutriLeads B.V. W.C. is the founder and employee of StatistiCal. B.V. J.M, R.V., A.R., L.R., and J.K. declare no conflicts of interest.

Supporting information

Supporting File 1:

Supporting File 2:

Supporting File 3:

Data Availability Statement

The data presented in this study are available on any reasonable request from the corresponding author.