Differential evolution of peripheral cytokine levels in symptomatic and asymptomatic responses to experimental influenza virus challenge

Viral challenge

For the human viral challenge study, in collaboration with hVivo Ltd (formerly Retroscreen Virology; London, UK), we inoculated intranasally 17 healthy volunteers aged 22–41 years (average age 27 years) with influenza A/Wisconsin/67/2005 (H3N2), as described in detail previously 13. All volunteers provided informed consent and underwent extensive pre‐enrolment health screening, and inclusion required negative baseline haemagglutination inhibition titres to the specific strain of influenza utilized in the study. After 24 h in quarantine, we instilled 10⁶ 50% tissue culture infective dose (TCID₅₀) influenza A into bilateral nares of subjects using standard methods 14, 15. At predetermined intervals (every 8 h for the first 5 days following inoculation), we collected blood, serum and plasma. We obtained nasal lavage samples from each subject daily for qualitative viral culture and and/or quantitative influenza reverse transcription–polymerase chain reaction (RT–PCR) to assess viral shedding. Blood and nasal lavage collection continued throughout the duration of the quarantine. Symptom scores were tabulated twice daily throughout the study 13, 16, where subjects ranked upper respiratory and systemic symptoms (runny nose, sinus stuffiness, sneezing, sore throat, earache, cough, shortness of breath, malaise, myalgias, fever) on a scale of 0–3 of ‘no symptoms’, ‘just noticeable’, ‘bothersome but can still do activities’ and ‘bothersome and cannot do daily activities’. All subjects were negative by rapid antigen detection (BinaxNow Rapid Influenza Antigen; Inverness Medical Innovations Inc./Alere Inc., Waltham, MA, USA) at time of discharge, and no additional post‐challenge adverse events were reported.

Cytokine quantification

Serum chemokine and cytokine levels were evaluated using the Invitrogen human cytokine 25‐plex assay (Carlsbad, CA, USA), as per the manufacturer's instructions. Briefly, beads are conjugated to cytokine‐specific capture antibodies and added along with the sample of interest into the wells of a filter‐bottomed microplate. After washing the beads, biotinylated detector antibodies are added followed by addition of streptavidin‐R‐phycoerythrin (RPE). The streptavidin‐RPE binds to the biotinylated detector antibodies associated with the immune complexes on the beads, forming a four‐member solid‐phase sandwich. After washing to remove unbound streptavidin‐RPE, the beads are analysed on a Luminex detection system.

RNA purification and microarray analysis

For each challenge, we collected peripheral blood at 24 h prior to inoculation with virus (baseline), immediately prior to inoculation (prechallenge) and at set intervals following challenge as described 13. RNA was extracted at Expression Analysis (Durham, NC, USA) from whole blood using the PAXgene™ 96 Blood RNA Kit (PreAnalytiX, Valencia, CA, USA) employing the manufacturer's recommended protocol. Hybridization and microarray data collection was performed at Expression Analysis using the GeneChip^® Human Genome U133A 2·0 Array (Affymetrix, Santa Clara, CA, USA).

Viral titration

Confluent monolayers of Madin–Darby canine kidney (MDCK) cells were inoculated with replicate (n = 4) serial 10‐fold dilutions of virus stocks or clinical samples (nasopharyngeal washes) in 96‐well microtitre format. After 90 min at 37°C in a humidified 5% CO₂ incubator, the inocula were removed and the cells were washed with Iscove's medium (IM) and cultured for 6 days in a humidified incubator at 37°C and 5% CO₂, followed by enumeration and calculation of TCID₅₀ by standard practice 17.

Statistical analysis

Temporal development of the host response to viral challenge

In our previously completed H3N2 (A/Wisconsin/67/2005) challenge trial, we inoculated 17 volunteers (mean age 27 years; range 22–41 years), of whom nine (53%) developed symptomatic influenza infection 13. For the nine symptomatic‐infected subjects, symptom onset began at an average of 45–49 h after inoculation, and they experienced maximal symptoms 85–93 h, on average, after inoculation followed by a slow steady improvement. Viral shedding increased rapidly, with an early peak at 48 h post‐inoculation followed by a comparatively more rapid decline compared to symptom resolution (Fig. 1). For the current study, we assayed 25 cytokines in peripheral blood at each of 17 different time‐points ranging from pre‐inoculation to the entire course of clinical disease and eventual symptom resolution. Levels of the various cytokines in peripheral blood exhibited significant variability among individuals and across time. This variability in the cytokine responses among individuals manifests quantitatively and, to a lesser degree, temporally. However, the character, direction, relative magnitude and shape of the temporal responses seen for each cytokine are remarkably conserved among symptomatic individuals. This can be visualized by examining a composite curve for all individuals which illustrates this conserved motif for each significantly altered cytokine (Supporting information, Fig. S1). The majority of this variation is quantitative in nature and can be dealt with by calculating fold change compared to baseline for each individual, cytokine and time‐point studied (Fig. 2), which reveals the overall conserved character of the host response.

Presymptomatic times (0–45 h post‐inoculation)

There were no significant differences in baseline, pre‐inoculation cytokines between those who would go on to become symptomatic or asymptomatic. In symptomatic individuals, however, some cytokines commonly increased well before the average time of symptom onset (45 h, Fig. 2). The primary cytokines which experienced the earliest spikes in symptomatic individuals are IL‐6, which shows increased circulating levels of as much as 70% by as early as 5 h post‐inoculation, IL‐12 (21 h), MCP‐1 and eotaxin (29 h), IP‐10 (36 h) and IL‐15 (45 h post‐inoculation, Fig. 2).

Symptomatic times (45–91 h post‐inoculation)

Between the time of symptom onset and symptom maximum, spikes in the circulating levels (minimum 50% increase compared to baseline) of IL‐4 and IL‐7 (small, 60 h), IL‐10 (69 h), IFN‐α (69 h) and tumour necrosis factor (TNF)‐α (60 h) were seen. A slightly larger increase is seen in IL‐1RA at 53 h, a threefold increase. Additionally, further evolution is seen in of the elevation of the most active cytokines as they reach their peak values – IP‐10 (peak 53–69 h, an almost 10‐fold increase), and IL‐6 (peak 60 h, sixfold increase), MCP‐1 (peak 53 h at 4.3‐fold) and IL‐15 (peak 60 h at 4.3‐fold increase).

Interestingly, starting at around the time of maximal symptoms and cytokine expression (84–96 h) and extending even later (132–168 h), symptomatic subjects exhibit not only a gradual reduction in the levels of previously elevated cytokines, but also show specific down‐regulation (relative to baseline) of a number of important mediators, including IL‐1β, IL‐13, IL‐17, granulocyte–macrophage colony‐stimulating factor (GM‐CSF), IL‐2, IL‐7 and IL‐4.

Cytokine changes relative to symptoms

Individual cytokine expression levels did not correlate directly with any individual symptom, nor were there any specific cytokines that distinguished ‘upper respiratory’ (runny nose, sinus stuffiness, sneezing, sore throat, earache, cough, shortness of breath) or ‘systemic’ (malaise, myalgias, fever) symptom subgroups. Even the best‐performing individual cytokine (IP‐10) showed only a moderate correlation with overall symptom score (correlation coefficient 0·64). However, peripheral cytokine levels tended to broadly increase prior to the time of symptom onset and peaked just prior to the time of maximal symptoms, and it is possible to derive a model utilizing the combined score of differentially weighted individual cytokine levels which achieves much closer correlation with overall symptom scores (correlation coefficient 0·84, Fig. 3). There were no significant correlations between cytokine levels achieved and d28 influenza HAI titres (data not shown).

Cytokine changes relative to viral load

On average, viral loads peaked at approximately 48 h post‐inoculation, around the mean time of symptom onset but, interestingly, showed no significant correlation at the individual level with the degree of clinical disease (as determined by symptom scoring). However, when divided into the individuals with the highest level of viral shedding and those with the least (top/bottom thirds, as determined by quantitative culture) there were several key differences noted in peripheral cytokine levels (Fig. 4). Those individuals with the highest levels of viral shedding exhibited higher circulating levels of many of the cytokines tested. The largest increases were seen with IL‐6 (threefold higher in high viral shedders), IL‐1RA (threefold higher), IL‐10 (twofold higher) and IP‐10 (threefold higher). The relatively higher levels of peripheral cytokine expression seen in those with the highest viral loads is most prominent around the time of peak viraemia (36–53 h, Figs 1 and 4), further supporting a direct relationship between these variables.

Comparison to peripheral gene expression

We have published previously on the temporal development of gene expression signatures in circulating white blood cells, which are capable of diagnosing acute respiratory viral infection both at the time of clinical presentation 13, 14 as well as much earlier in the presymptomatic state 13. Specifically, in this cohort of H3N2 challenge patients, a gene expression signature capable of detecting infected subjects accurately was discernible in symptomatic subjects as early as 29 h after inoculation, but absent from those who would remain asymptomatic 13. This gene signature included IFN‐stimulated pathways such as those including radical S‐adenosyl methionine domain‐containing 2 (RSAD2), IFN regulatory factor 7 (IRF7), myxovirus resistance 1 (MX1), 2'‐5'‐oligoadenylate synthetase 3 (OAS3), melanoma differentiation‐associated protein‐5 (MDA‐5), retinoic acid‐inducible I (RIG‐I) and others which are thought to drive both innate and, to a lesser degree, adaptive immune responses to viral infection. In order to examine the hypothesis that increases in some circulating peripheral serum cytokines are driven by production in circulating immune cells (while others are not), we examined differential gene expression levels for those peripheral cytokines which were elevated most significantly in infected subjects. In a subset of individuals (five asymptomatic and seven symptomatic) both cytokine and transcriptomic data were available from the same blood draws, and these were utilized for the analysis depicted in Fig. 5. At the time of symptom onset (45–49 h post‐inoculation) there are already marked increases in the levels of IP‐10, MCP‐1, IL‐15 and IL‐6 (Fig. 2). Interestingly, however, of these cytokines only IP‐10, MCP‐1 (four to fivefold) and to a lesser degree IL‐15 (1.9‐fold) have undergone similar increases in gene expression in PBMCs themselves to that point (Fig. 5). Genes driving production of IL‐6 are not up‐regulated in the peripheral cells themselves. Also, while there are modest changes in the concentration of circulating IFN‐α and IFN‐γ (in the order of 50–60% increases in each), their gene expression remains completely unchanged in the PBMCs (Fig. 5). However, IFN‐inducible genes (such as IFN‐induced protein with tetratricopeptide repeats (IFIT)1, IFIT3, IFI44L and others) are among the most strongly up‐regulated genes in these peripheral cells 13. Later, at the time of maximal symptoms, the levels of some of these peripheral cytokines have decreased substantially (MCP‐1, IL‐6, IL‐8), while the level of gene expression of these cytokines in peripheral cells is only mildly lower. However, at these late times the level of expression of IFN‐inducible (IFI) genes is reaching its highest level, and these IFI genes dominate the influenza‐specific host gene signature at those times. Together, these suggest that while some of the circulating cytokines seen probably originate in circulating peripheral blood leucocytes, others seem more likely to be produced at a distal site (such as locally in the upper respiratory tract).

Key cytokine changes in the asymptomatic state

Despite extensive prescreening and identical exposures, eight of the 17 individuals in this study (47%) exhibited no symptoms or viral shedding following inoculation, which is similar to the rate reported in previous challenge studies 11, 12, 19. However, we have demonstrated previously that the peripheral gene expression profile of these individuals indicates that there is still a significant host response in these asymptomatic individuals 2. The asymptomatic but exposed state is not passive, but involves a significant genomic response in peripheral white blood cells – different from that seen in symptomatic subjects – yet these individuals do not exhibit clinical signs of influenza infection. Concordantly, in the current work we also see temporal changes in their peripheral cytokine expression patterns which are different from those seen in symptomatic subjects. There is some minor and variable up‐regulation of some cytokines starting immediately post‐inoculation, and this is seen most prominently in the same cytokines which dominate the symptomatic response, although at significantly lower levels (Fig. 2). However, the asymptomatic cytokine response is dominated by a significant early down‐regulation of many of the cytokines tested. There was a generalized down‐regulation of IL‐7, IL‐5, IFN‐γ (all essentially immediately following inoculation) and, to a lesser degree, IL‐17, IL‐15, IL‐4 and MIP‐1a. While there is also down‐regulation of most of these cytokines in symptomatic subjects, it does not occur until late (84–96 h) in the process when symptoms are high and viral load is decreasing. In symptomatic subjects this time corresponds to a natural correction of the immune response towards regulation and a return to homeostatic levels; however, in asymptomatic subjects this down‐regulation occurs almost immediately and persists throughout the study period. Interestingly, despite a lack of detectable viral shedding three asymptomatic individuals seroconverted to challenge virus by day 28. No significant differences were noted in the cytokine profiles of these three relative to the other asymptomatic subjects.