Research ArticleDengue

Increased adaptive immune responses and proper feedback regulation protect against clinical dengue

See allHide authors and affiliations

Science Translational Medicine  30 Aug 2017:
Vol. 9, Issue 405, eaal5088
DOI: 10.1126/scitranslmed.aal5088

Distinguishing dengue presentation

Although dengue hemorrhagic fever can be life-threatening, not all dengue virus infections even present with symptoms. To determine what may be driving the differences in clinical and asymptomatic infections, Simon-Lorière et al. examined the serum and immune gene transcripts of Cambodian children infected with dengue virus serotype 1. They saw that relative to those with clinical infections, the small subset of asymptomatic children had increased signs of antigen presentation, T cell activation, and T cell apoptosis; plasmablast differentiation and anti-dengue antibodies seemed relatively lower. These results provide clues for pathways that may drive pathologic responses in severe dengue virus infections.

Abstract

Clinical symptoms of dengue virus (DENV) infection, the most prevalent arthropod-borne viral disease, range from classical mild dengue fever to severe, life-threatening dengue shock syndrome. However, most DENV infections cause few or no symptoms. Asymptomatic DENV-infected patients provide a unique opportunity to decipher the host immune responses leading to virus elimination without negative impact on an individual’s health. We used an integrated approach of transcriptional profiling and immunological analysis to compare a Cambodian population of strictly asymptomatic viremic individuals with clinical dengue patients. Whereas inflammatory pathways and innate immune response pathways were similar between asymptomatic individuals and clinical dengue patients, expression of proteins related to antigen presentation and subsequent T cell and B cell activation pathways was differentially regulated, independent of viral load and previous DENV infection history. Feedback mechanisms controlled the immune response in asymptomatic viremic individuals, as demonstrated by increased activation of T cell apoptosis–related pathways and FcγRIIB (Fcγ receptor IIB) signaling associated with decreased anti-DENV–specific antibody concentrations. Together, our data illustrate that symptom-free DENV infection in children is associated with increased activation of the adaptive immune compartment and proper control mechanisms, leading to elimination of viral infection without excessive immune activation, with implications for novel vaccine development strategies.

INTRODUCTION

Dengue is the most prevalent arthropod-borne viral disease. Every year, dengue virus (DENV) is estimated to cause at least 50 million infections, 500,000 hospitalizations, and 12,500 deaths (14). Fifty percent of the world’s population is considered at risk of infection (2). DENV belongs to the genus Flavivirus, is transmitted by Aedes spp. mosquitoes, and consists of four antigenically distinct serotypes (DENV-1 to DENV-4). Each of the four DENV serotypes can cause dengue fever (5, 6), occasionally progressing to severe dengue, a life-threatening condition characterized by a cytokine storm, vascular leakage, and shock (79). Despite more than 50 years of research on dengue pathophysiology, the mechanisms leading to a severe clinical outcome upon DENV infection remain elusive and likely involve complex interactions between viral, immunological, and human genetic factors (10). Increased risk of severe dengue during secondary infection could be due to antibody-dependent enhancement, where low-affinity, serotype–cross-reactive antibodies increase viral infection of antigen-presenting cells (APCs), or due to cross-reactive cytotoxic T cells (1113). However, it remains to be investigated to what extent these mechanisms contribute to human pathogenesis.

A large fraction of overall DENV infections remain subclinical, resulting in insufficient discomfort to disrupt an individual’s daily routine (1, 14, 15). Although having no direct effect on health, subclinical infections nevertheless have major epidemiological consequences including viral dissemination and maintenance, leading to the immune priming of naïve populations, which may affect the clinical outcome of subsequent infections (16, 17). Finding subclinical cases during routine surveillance studies is highly challenging, and a clear definition of subclinical dengue is lacking. This phenomenon therefore remains poorly documented and understood. “Subclinical infection” most often refers to DENV infection without major symptoms requiring medical attention, whereas “asymptomatic infection” is a confirmed DENV infection in the complete absence of any reported or detected symptoms (1720). Epidemiological risk factors, such as age, time interval between consecutive DENV infections, previous DENV infecting serotype, and the concentrations of preexisting heterotypic neutralizing antibodies, have been associated with subclinical or asymptomatic outcome after DENV infection (15, 2123). However, the molecular and immunological mechanisms underlying control of DENV infection without disease manifestation remain unknown.

To identify host mechanisms involved in the control of DENV infection without clinical symptoms, we used an integrated approach of transcriptional profiling and immunological analysis of Cambodian children who were DENV-infected and strictly asymptomatic despite viremia compared to viremic patients with clinical signs of dengue. Our results show distinct transcriptional profiles in asymptomatic individuals, with an increased activation of the adaptive immune compartment, and proper regulatory mechanisms leading to the control of viral infection without excessive immune activation.

RESULTS

Strict follow up of acute DENV-infected children and inclusion of 9 asymptomatic individuals and 76 clinical dengue patients

A total of 85 DENV-infected Cambodian children were included in the present study. Dengue infection was confirmed in all individuals by detection of viral RNA in serum as measured by viral RNA copies/ml (Table 1, total cohort). During the dengue transmission seasons of 2012 and 2013, clinico-epidemiological investigations were conducted around households to identify DENV-infected individuals without symptoms (17). Nine individuals remained strictly asymptomatic at the time of inclusion and during the 10-day follow-up period. A rigorous definition of asymptomatic infection was applied, involving complete absence of reported or detected clinical symptoms (including but not limited to retro-orbital pain, headache, rash, fever, and abdominal pain), and DENV infection was confirmed by DENV-positive quantitative real-time polymerase chain reaction (qRT-PCR) (17). For all clinical dengue patients, blood collection was performed 3 ± 2 days after fever onset. Clinical dengue patients included were classified according to the World Health Organization (WHO) 1997 classification criteria (24): 37% were dengue fever cases, 38% had dengue hemorrhagic fever, and 25% had dengue shock syndrome (Table 1, total cohort).

Table 1. Demographic data and clinical parameters of the studied populations.

Patients are characterized according to the WHO 1997 criteria. DENV serotype and viral load were determined by qRT-PCR. Primary or secondary infection was determined by an HI test on acute and convalescent samples. N/A, not applicable; CD, clinical dengue; ASD, asymptomatic dengue-infected individuals; IQR, interquartile range; DF, dengue fever; DHF, dengue hemorrhagic fever; DSS, dengue shock syndrome.

View this table:

We performed gene expression analysis on purified peripheral blood mononuclear cells (PBMCs), antibody and cytokine measurements in serum, and immunophenotyping. For all analyses, samples were selected on the basis of availability and quality of samples. It is important to note that when comparing gene expression profiles of clinical dengue patients and asymptomatic dengue-infected individuals, we only studied patients infected by DENV serotype 1 (DENV-1), the main circulating viral serotype in Cambodia at the time of the study, with comparable viral loads as measured by viral RNA copies between asymptomatic cases and clinical dengue patients, to minimize confounding factors (Fig. 1A and Table 1, analysis 1). Eight strictly asymptomatic viremic individuals and 25 dengue patients experiencing symptoms were included. Of these, secondary infections as determined by hemagglutination inhibition (HI) testing according to the WHO criteria (2) were identified in 50% of the asymptomatic cases and in 100% of the clinical dengue cases (Table 1). Serum cytokine and antibody measurements were performed on 8 asymptomatic dengue samples and 58 symptomatic individuals, which included all patients analyzed for differential gene expression (Table 1, analysis 2). We performed phenotypic analysis on purified PBMCs for 6 of the 9 asymptomatic dengue cases, which included the one DENV-4–infected case, and 18 additional clinical dengue cases that were not included in the gene expression or serum cytokine analysis but had readily available dimethyl sulfoxide (DMSO)–frozen PBMCs of sufficient quality. The demographic characteristics of these patients were similar to those of other clinical dengue patients (Table 1, analysis 3). To investigate correlations between gene expression and viral load, we analyzed gene expression profiles in 36 clinical dengue patients and 8 asymptomatic individuals (Table 1, analysis 4).

Fig. 1. Characteristics of asymptomatic individuals and clinical dengue patients.

(A) Viral load as measured by qRT-PCR (as viral RNA copies/ml plasma) and days of fever for all patients included in transcriptome analysis. (B to D) Percentages of cells subsets and ratio as determined by cell surface marker expression (ASD, n = 6; CD, n = 18). Bar represents median with IQR. P values were obtained with Mann-Whitney test.

Cell population composition was similar between asymptomatic and clinical dengue subjects, with similar percentages of innate and adaptive immune cells (CD14+ monocytes, Lin-CD11c+ dendritic cells, CD19+ B cells, and CD335+ natural killer cells) (Fig. 1B and fig. S1), with the exception of T cells: We observed an increase in the percentage of CD4+ T cells and a decrease in CD8+ T cells in the lymphocyte gate in asymptomatic viremic individuals. Correspondingly, the CD4/CD8 ratio was inversed in clinical dengue patients (median, 1.2; IQR, 1.0 to 2.2 versus median, 0.8; IQR: 0.4 to 1.5; P < 0.01) (Fig. 1, C and D), suggesting deregulated T cell response in children with clinical dengue infection as shown previously (25).

Discrimination between asymptomatic and clinical dengue patients using transcriptional signatures

To gain insight into the mechanisms contributing to asymptomatic outcome of dengue infection, we performed differential gene expression analysis on purified PBMCs of 8 asymptomatic and 25 clinical dengue patients. A total of 1663 genes were differentially expressed, of which 1045 genes were up-regulated and 618 genes were repressed in asymptomatic viremic individuals versus clinical dengue patients (table S1). A hierarchical cluster analysis was carried out on the differentially expressed genes. All asymptomatic patients clustered together, indicating that they exhibit similar gene expression patterns as shown by a heat map (Fig. 2A). We performed gene ontology (GO) enrichment analysis as implemented in GOrilla to identify biological processes that diverge the most between asymptomatic and clinical dengue individuals (26). The most significantly (P < 1 × 10−7) differentially regulated processes were related to immune processes, with 12 of the 21 most significant processes related to immune response and immune activation (Fig. 2B and table S2).

Fig. 2. Transcriptional signatures discriminate between asymptomatic and clinical dengue patients.

(A) Unsupervised hierarchically clustered heat map of the genes differentially expressed between asymptomatic dengue (n = 8) and clinical dengue patients (n = 25). (B) Top immunity-related molecular processes found as a result of GO enrichment analysis using GOrilla. The significance of the observed enrichment for GO molecular processes was estimated by the P value plotted on a log10 scale. The enrichment score reflects the strength of association between input gene expression and enriched molecular processes. (C) Immunity-related canonical pathways found as a result of pathway enrichment analysis using the IPA software. The significance of the association between gene expression and canonical pathways was estimated by the P value plotted on a log10 scale, and the ratio value reflects its strength. The z score reflects the activation state of the canonical pathway (activated in asymptomatic dengue: z score > 0; inhibited in asymptomatic dengue: z score < 0. (D) Venn diagram showing the number of overlapping significantly different genes between asymptomatic individuals versus clinical dengue patients, healthy controls versus clinical dengue patients, and viral load response.

We further explored pathways significantly enriched in the list of differentially expressed genes between asymptomatic viremic individuals and clinical dengue patients using Ingenuity Pathway Analysis (IPA) (27). From this, 379 canonical pathways were found to be significantly different as a result of pathway enrichment analysis for the differentially expressed genes between asymptomatic viremic and clinical dengue patients (Fig. 2C and table S3). Immune processes were mainly activated in asymptomatic viremic individuals.

Bias may occur due to the difference of immune status between groups, because 50% of the asymptomatic DENV infections were primary infections, whereas all 25 clinical dengue patients were undergoing a secondary infection. To examine this more closely, we performed an analysis excluding the four primary asymptomatic DENV-infected individuals. Despite lower numbers of differentially expressed genes and associated pathways, similar results were obtained, although the significance threshold was not reached in all comparisons (tables S1 to S3).

To verify that genes differentially expressed in asymptomatic viremic individuals compared to clinical dengue patients did not simply correspond to differentially expressed genes between healthy, uninfected individuals and clinical dengue patients, we analyzed a previously published data set from Nicaragua involving 8 healthy children and 41 clinical dengue cases (28) and calculated the list of differentially expressed genes between the two groups. Of the 1385 genes differentially expressed between asymptomatic viremic children and clinical dengue patients, only 38 genes were overlapping with the 204 differentially expressed genes found between healthy controls and clinical dengue patients (Fig. 2D and table S1). Combining these results, transcriptional profiling, where most of the gene expression changes are related to immune responses, can differentiate between DENV-1–infected individuals with or without clinical symptoms irrespective of viral load and immune status.

Correlation between viral load and differentially regulated key genes

We used linear regression models to correlate viral load variation and disease state. Investigating the cohort of patients used for analysis 1 [n = 8 (ASD) and 25 (CD)], the model showed an average viral load of 103.83 in the asymptomatic group and a mean difference of 100.72 between the two groups. The model showed no significant association between viral load and disease state. However, including all 36 clinical dengue patients (Table 1, analysis 4), the model showed a higher mean difference of 101.68 and a significant association between viral load and disease state (fig. S2).

To investigate the effect of genes, controlling for viral load, on the asymptomatic versus clinical outcome, we performed another linear regression analysis between gene expression and viral load while taking into account disease status, using LIMMA (Linear Models for Microarray Data) (29). We identified 31 genes that showed correlation with viral load as measured by viral RNA copies. None of these genes were significantly differentially expressed between asymptomatic viremic individuals and clinical dengue patients (Fig. 2D and table S1). These genes are involved in the negative regulation of the viral life cycle, viral processing, and viral genome replication (tables S2 and S3).

No major differences in innate immune response between asymptomatic and clinical dengue patients

Examination of immune processes differentially regulated between DENV-infected individuals with or without clinical symptoms revealed no major differences in activation of innate immune response. IPA pathways, such as regulation of innate immunity, antiviral innate immunity, activation of pattern recognition receptors, interleukin 8 (IL-8) signaling, and type I or II interferon (IFN)–regulated pathways, did not differentiate between clinical dengue and asymptomatic infection (P > 0.05). Consistent with the transcriptomic results, signature cytokines produced by innate immune cells, such as IL-8, IL-15, CCL3, and CCL4, displayed comparable serum concentrations between groups (Fig. 3), and serum concentrations of key cytokines regulating innate immune functions and activation were similar between both groups (fig. S2). IFN-γ, a cytokine critical for both innate and adaptive immunity against viral infections, was increased in asymptomatic viremic individuals (fig. S3). In addition, serum concentrations of inflammatory cytokines such as tumor necrosis factor–α (TNFα) and IL-6 were not different between asymptomatic viremic and clinical dengue patients (Fig. 3). Although serum concentrations of IL-1β were below the detection limit, a similar expression of IL-1–regulated genes was observed irrespective of clinical outcome. Furthermore, in asymptomatic viremic individuals, we observed a down-regulation of IL-1 receptor–associated kinase 2, a serine/threonine kinase associated with IL-1 receptor upon stimulation (table S1). Comparing the cytokine profiles of the four asymptomatic cases undergoing secondary infection with the clinical dengue patients undergoing secondary infection yielded similar results (fig. S4). Together, these data suggest that activation of innate immune pathways is similarly regulated in asymptomatic and symptomatic viremic dengue-infected children.

Fig. 3. Serum cytokines related to innate immune response and inflammation are not associated with the clinical outcome of dengue infection.

Serum concentrations of various cytokines measured by Luminex in asymptomatic dengue-infected individuals (n = 8) and clinical dengue patients (n = 58). Line represents median. P values were obtained with Mann-Whitney test.

Differentially regulated pathways related to antigen presentation asymptomatic viremic individuals

The most significantly activated pathway in asymptomatic individuals was “nuclear factor of activated T cells (NFAT) mediated regulation of immune response” (Fig. 2C and table S3). NFAT are major regulators of the adaptive immune response and are expressed after antigenic stimulation of lymphocytes (30). Therefore, we first investigated the regulation of antigen presentation in asymptomatic viremic individuals. Pathway analysis revealed a differential regulation of the antigen-presentation pathway and dendritic cell maturation in asymptomatic viremic individuals (Figs. 2C and 4A and table S3). Genes up-regulated twofold or more in asymptomatic viremic individuals included CIITA, CD74, and various human leukocyte antigen (HLA) genes. In stark contrast, the CD86 costimulatory molecule was significantly down-regulated (table S1).

Fig. 4. Asymptomatic viremic individuals show differentially regulated pathways and molecules related to antigen presentation.

(A) Antigen-presentation pathway showing differentially expressed genes, adapted from IPA. Red, up-regulated in ASD; green, down-regulated in ASD; white, no change. (B) Representative histograms of HLA-DR and CD86 expression on CD14+ monocytes and LinCD11c+ dendritic cells of asymptomatic dengue-infected individuals (gray) and clinical dengue patients (white). Data are summarized on the right, where lines represent median and IQR (ASD, n = 6; CD, n = 18). MFI, mean fluorescence intensity. (C) Serum concentrations of IL-12 and IL-23 (pg/ml) measured by Luminex in asymptomatic dengue-infected individuals (n = 8) and clinical dengue patients (n = 58). Line represents median. P values were obtained with Mann-Whitney test.

These data were confirmed by ex vivo phenotypic analysis of PBMCs collected from both groups. HLA-DR expression was increased on CD14+ monocytes of asymptomatic viremic individuals and could be induced by the observed increase in serum IFN-γ concentrations. (Fig. 4B and fig. S3). In contrast, CD86 expression was decreased in both CD14+ monocytes and LinCD11c+ dendritic cells (Fig. 4B). Serum concentrations of IL-12 and IL-23, both secreted by APCs and indicative of their activation, were increased in asymptomatic viremic individuals compared to clinical dengue patients (Fig. 4C). Avoiding bias due to immune status, we compared secondary cases of dengue infection only. Despite the low number of samples, significant differences were observed between asymptomatic viremic individuals and symptomatic patients for most parameters (figs. S4 and S5). Together, these transcriptional and protein expression data suggest that activation of APC is differentially regulated in asymptomatic viremic children, possibly including feedback regulation through decreased CD86 expression on APC (31, 32).

Increased T cell activation and T cell apoptosis in asymptomatic viremic individuals

In accord with the observed up-regulation of the antigen-presentation pathway, protein kinase Cθ (PKCθ) signaling in T lymphocytes was highly activated in asymptomatic viremic individuals. PKCθ is an essential component of the T cell supramolecular activation cluster and mediates several crucial functions in T cell receptor (TCR) signaling. Genes up-regulated twofold or more in asymptomatic viremic individuals included AKT3, SOS1, PAK1, and SLAMF6 (Fig. 5A and table S1). Furthermore, several T cell costimulatory pathways were up-regulated in asymptomatic viremic individuals, such as ICOS-ICOSL signaling in T helper cells and CD28 and CTLA4 signaling in cytotoxic T lymphocytes (Fig. 2C and table S3). In addition, IL-2 serum concentration was increased, and the IL-2 signaling pathway was up-regulated in asymptomatic viremic individuals (Fig. 5B). CD69, an early activation marker expressed on many cell types, was found to be significantly up-regulated twofold or more (table S1). CD69 expression was significantly higher in both CD4+ and CD8+ T cell populations in asymptomatic viremic individuals (Fig. 5C). Comparing only secondary cases of DENV-infected children yielded similar up-regulated genes and pathways in asymptomatic individuals (tables S1 and S3). In addition, IL-2 serum concentrations and CD69 expression remained elevated, although the sample size was small (fig. S6).

Fig. 5. Increased T cell activation in asymptomatic viremic individuals.

(A) PKCθ signaling in T lymphocytes, adapted from IPA. Red, up-regulated in ASD; green, down-regulated in ASD; white, no change. (B) Serum concentrations of IL-2 measured by Luminex in asymptomatic dengue-infected individuals (n = 8) and clinical dengue patients (n = 58). Line represents the median. (C) Representative dot plots of CD69 expression on both CD4+ and CD8+ T cells (ASD, n = 6; CD, n = 18). Data are summarized on the right, where lines represent the median and IQR. P values were obtained with Mann-Whitney test.

One of the most significantly activated pathways in asymptomatic viremic individuals, calcium-induced T lymphocyte apoptosis (Fig. 2C), is implicated in TCR-mediated apoptosis, which might correspond to regulative measures against the proliferative response that follows TCR stimulation (33). These data suggest that an asymptomatic outcome of DENV infection appears to be associated with increased T cell activation and apoptosis.

Correlation of the up-regulation of gene expression pathways leading to plasma cell development and the secretion of anti-DENV antibodies with the development of clinical dengue

We further investigated the association between clinical outcome and B cell responses after DENV infection. IPA indicated an activation of the B cell receptor (BCR) signaling pathway in asymptomatic viremic individuals with genes such as BANK1 and MS4A1 (CD20) significantly up-regulated twofold or more (Fig. 6, A and B). However, PI3K signaling in B cells, a pathway activated within seconds of BCR stimulation, was inhibited in asymptomatic individuals (Fig. 6A). This can be explained by the twofold or more up-regulation of molecules involved in inhibition of B cell activation and differentiation such as CD22, FCRL1, and FCRL6 (Fig. 6B) (34). In addition, FcγRIIB (Fcγ receptor IIB) signaling, mediating inhibition of BCR signaling after antigenic stimulation, was activated in asymptomatic viremic individuals (Fig. 6A).

Fig. 6. Increased plasmablast differentiation in clinical dengue patients.

(A) Ingenuity canonical pathways associated with B cell biology. Ratio indicates the strength of association between the pathway and the list of differentially expressed genes. The z score reflects the activation state of the canonical pathway (activated in asymptomatic dengue: z score > 0; inhibited in asymptomatic dengue: z score < 0). (B) Summary of genes associated with B cell biology. The adjusted P value was calculated using the Benjamini-Hochberg (BH) procedure. Log fold change (FC) indicates the log2 FC for the gene between asymptomatic dengue and clinical dengue infection. (C) Serum concentrations of IL-10 and IL-21 measured by Luminex in asymptomatic dengue-infected individuals (n = 8) and clinical dengue patients (n = 58). Line represents the median. (D) Left: Serum anti-DENV IgM as measured by MAC-ELISA, where optical density (OD) is reported. Right: HI test where the log2 of the maximum dilution preventing agglutination is shown. Patients are stratified according to immune status. Asymptomatic dengue-infected individuals (n = 8) and clinical dengue patients (n = 58). Sera were analyzed 8 ± 2 days after inclusion in the study. Line represents median. P values were obtained with Mann-Whitney test.

Moreover, key transcription factors for plasma cell differentiation, such as PRDM1 (PR domain zinc finger protein 1) (BLIMP-1) and IRF4 (IFN regulatory factor 4), were down-regulated at least twofold in asymptomatic infections compared to clinical dengue patients (Fig. 6B). These gene expression data suggest that although B cells are activated in asymptomatic individuals, inhibitory mechanisms preventing the differentiation to plasma cells are also activated. Conversely, differentiation to plasma cells seems to be increased in clinical dengue patients. We investigated the serum concentrations of IL-4, IL-10, and IL-21 cytokines important for growth and differentiation of human B cells. We detected high IL-10 serum concentrations [1.7 pg/ml (0.0 to 5.3) versus 24.2 pg/ml (9.5 to 38.9); P < 0.0001] and overexpression of IL-10 transcripts in clinical dengue patients (Fig. 6, B and C), supporting the observation of increased plasma cell differentiation in clinical dengue patients. This is specific to IL-10, because IL-21 and IL-4 serum concentrations did not differ between patients and clinical dengue patients (Fig. 6C).

Last, to explore the hypothesis that B cell responses and plasmablast development are inhibited in asymptomatic dengue-infected individuals, as suggested by the gene expression data, we investigated the anti-DENV antibody response 8 ± 2 days after inclusion in the study. By stratifying samples according to primary or secondary infection, we observed decreased concentrations of immunoglobulin M (IgM) as measured by IgM antibody capture enzyme-linked immunosorbent assay (ELISA) and a lower HI titer in asymptomatic individuals (Fig. 6D). Hence, gene expression pointing toward a decreased inhibition of B cell activation and increased plasma cell differentiation, combined with increased serum concentrations of anti-DENV antibodies, is associated with clinical dengue.

DISCUSSION

Here, we interrogated the host immune response during viremic, strictly asymptomatic DENV infection in children. We report here the identification of differential adaptive immune responses associated with the clinical outcome of DENV infection. Asymptomatic viremic children showed differential antigen presentation, increased T cell activation and apoptosis, and decreased B cell activation and plasmablast differentiation.

We controlled for confounding factors, such as DENV serotype, viral load as measured by viral RNA copies, age, and sex, in our transcriptomic analysis. Four of the eight asymptomatic viremic individuals underwent a primary dengue infection, whereas all 25 clinical dengue patients experienced a secondary dengue infection. Analyzing only secondary asymptomatic viremic individuals yielded similar, but less significant differences, due to the low number of individuals. Because our results suggest higher activation of adaptive immunity in asymptomatic viremic individuals, this result is unlikely to be biased by primary infection in the asymptomatic viremic group. Because Cambodia is highly endemic for dengue, our study cohort consists of young children. Immune responses associated with protection from clinical dengue reported in this paper could be different in adults. Furthermore, it is impossible to assess the exact timing of infection in asymptomatic dengue infection, which could influence gene expression and T cell activation; however, both asymptomatic and clinical dengue-infected patients had detectable viral loads corresponding to the acute phase of disease.

Although we were unable to assess white blood cell counts, frequencies of major immune cell populations were similar in all acute DENV-infected children. Asymptomatic viremic dengue patients clustered together in terms of transcriptome, serum cytokine concentrations, and cellular phenotype. This appears mainly due to immune process regulation, rather than to stress-mediated pathways associated with viral infection. In addition, differences do not appear to be linked to innate response pathways or inflammatory responses. Genes and proteins involved in antigen processing and presentation, and serum concentrations of both IL-12 and IL-23, were increased in asymptomatic viremic individuals, suggesting activation of APCs and increased antigen presentation. However, expression of CD86 (but not CD80) was decreased in both circulating monocytes and dendritic cells. Although we were unable to perform APC assays to confirm our observations, these data suggest that more controlled immune responses are taking place in asymptomatic cases (31, 32).

Our results also revealed increased T cell activation and apoptosis in asymptomatic viremic individuals, supporting a role for T cells in protection from clinical dengue, which has major implications for future vaccine development. A decreased CD4/CD8 ratio has previously been observed in clinical dengue and could be the result of specific expansion or apoptosis of responder subsets or, alternatively, might merely be due to the young age of our study cohort (25, 33, 35).

The involvement of T cells in the control of disease has been suggested in several human and mouse studies. The protective function of CD8+ T cells has been demonstrated via the susceptibility of CD8+-depleted IFNAR−/− (IFN-α/β receptor) mice to DENV infection and the observation that both serotype-specific and cross-reactive T cells confer protection after peptide vaccination to DENV infection (36, 37). In humans, an HLA-linked protective role of CD8+ and CD4+ T cell responses has been observed in a Sri Lankan population (38, 39). Furthermore, CD4+ cytotoxic T cells confer protection against dengue infection ex vivo (35). Last, increased frequency of DENV-specific CD4+ and CD8+ T cells was detected in Thai school children who subsequently experienced subclinical infection, compared with symptomatic secondary DENV infections (40).

Recently, CD4 and CD8 T cell epitopes of DENV have been mapped in different human populations (38, 4144). CD8+ T cell epitopes preferentially cluster in nonstructural proteins such as NS3, NS4B, and NS5, whereas CD4+ T cell epitopes are skewed toward envelope, capsid, and NS1 epitopes, which are also targeted by the B cell response. Most of these epitopes are lacking in the licensed dengue vaccine Dengvaxia, which might affect the generation of an adequate memory T cell response. Hence, our results, together with previous findings, could provide an explanation for the observed relatively low efficacy of protection against virologically confirmed dengue of all four serotypes obtained with Dengvaxia (4547). These data emphasize the need to reconsider the inclusion of T cell–specific epitopes in future vaccine design.

Our in-depth analysis of the immune response reveals that the clinical outcome of dengue infection seems to be determined by the defective control of B cell responses and increased plasmablast differentiation. This finding is consistent with previous studies, where massive expansion of antibody-producing plasmablasts was observed in the blood of severe dengue patients (4851). This expansion observed in clinical dengue cases may be driven by IL-10, as we found increased IL-10 transcripts in PBMCs and elevated IL-10 serum concentrations in these patients. In addition, IL-10 is required for plasmablast differentiation of B cells stimulated by DENV-infected monocytes (51). Notably, IL-10 has been associated with severe dengue disease and proposed as a marker predicting severity in a Venezuelan cohort (52).

Efforts to understand the antigenic specificity of the expanded plasmablasts have been undertaken, mainly through in vitro production of monoclonal antibodies derived from sorted plasmablasts during acute DENV infection (5355). However, the frequency of DENV-specific circulating plasmablasts, and their origin, remains unknown, as does the contribution of polyclonal bystander activation to disease pathogenesis. In accordance with our observations, a previous study on early/late convalescent samples showed lower serum concentrations of anti-DENV antibodies (against PrM and E) in asymptomatic DENV-infected individuals compared to clinical dengue patients (56). These observations favor the hypothesis that antibodies may play a pathogenic role in the risk of the development of clinical dengue, possibly through antibody-dependent enhancement, where low-affinity serotype–cross-reactive (that is, heterologous) antibodies would increase viral infection of APCs such as monocytes and dendritic cells (11, 12).

We have performed an integrated immunologic analysis of strictly asymptomatic dengue-infected individuals with confirmed DENV viral load. We investigated PBMCs during the acute phase of the disease, when circulating cells were in an activated state. However, we cannot delineate which of the observed phenomena can be attributed to antigen-specific responses. Therefore, future studies using tools to identify DENV-specific T cells and B cells in strictly asymptomatic individuals will be of great value (38, 53)

One previous report analyzing differential gene expression in asymptomatic DENV-infected individuals investigated patients in the convalescent phase of the disease, as defined by the absence of detectable viral load (57). Unsurprisingly, they found that most genes related to host defense mechanisms were down-regulated.

We show that control of infection without the concurrent development of clinical symptoms is associated with strong and regulated adaptive immune response in a cohort of Cambodian children. Molecules involved in antigen presentation are differentially regulated, and T cell activation is increased in acute DENV-infected asymptomatic children, whereas up-regulation of gene expression pathways leading to plasmablast development, and the secretion of anti-DENV antibodies, correlates with the development of clinical dengue. These results contribute to our understanding of the development of symptomatic dengue and will lead to novel strategies for future vaccine development.

MATERIALS AND METHODS

Study design

The aim of the study is to identify pathways that protect against clinical dengue during DENV infection. Therefore, we collected biological specimen from individuals undergoing acute DENV infection without any clinical symptoms and patients who developed clinical dengue. Using a household investigation approach (19), we were able to detect 27% of the household members infected by DENV, 18% of which had viremia without any symptoms in Southeast Asia. We used this information to design a new cohort study to collect samples from asymptomatic viremic individuals and dengue index cases (DICs). Study design with identification of DICs and cluster participants was described previously in detail (17). Briefly, DICs were identified from patients presenting with acute dengue-like illness between June and October of 2012 and 2013 at Kampong Cham City Provincial Hospital, at two district hospitals in Kampong Cham province, and from village-based active fever surveillance. A cluster investigation was initiated, enrolling all family members in the household and people living within a 200-m radius of the home of the DICs. DICs and cluster participants were examined during sequential visits as described in (17). For laboratory analysis, RNA preparation, and microarray experiment, cytokine measurement and phenotyping researchers were blinded during the experiment and assessment of outcome for patient classification.

Patient recruitment and classification

Blood samples were taken from hospitalized patients at two time points: one at hospital admission and the other at hospital discharge, during the convalescent phase. Disease severity of clinical patients was assessed according to the WHO 1997 criteria (24). DENV-positive cluster investigation participants were assessed prospectively at day 0 (D0), D1, D2, D3, D4, D5, D6, D7, and D10, for occurrence of clinical symptoms and blood sampling. Only patients displaying no clinical symptoms during this follow-up period were considered asymptomatic and included in the present study. Serum was stored at −80°C for future analysis, and PBMCs were separated using Ficoll-Paque density gradient centrifugation and stored in RNAlater (Thermo Fisher Scientific) or DMSO until transcriptomic or cellular phenotypic analysis, respectively. The study was approved by the National Ethics Committee of Health Research of Cambodia, and written informed consent of all participants or legal representatives for participants under 16 years of age was obtained before inclusion in the study.

Laboratory diagnosis

DENV infection was confirmed on serum samples collected at admission of hospitalized patients or at inclusion in the cluster investigation by nested qRT-PCR at the Institut Pasteur in Cambodia, the National Reference Center for arboviral diseases in Cambodia (58). Serological tests were performed on the specimen collected during the acute and convalescent phase of the infection, in both symptomatic and asymptomatic groups, for detection of antibodies against DENV. Because of the potential cross-reactivity among flaviviruses, all specimens were tested for both anti-DENV and anti-Japanese encephalitis virus antibodies using an in-house IgM capture ELISA (MAC-ELISA) and HI assay, as previously described (17). Primary or secondary immune status of DENV infection was determined by the HI test according to the WHO criteria (2).

RNA preparation, microarray hybridization, and analysis

RNA was extracted with an RNeasy kit (Qiagen) and hybridized overnight with the probes contained in the Affymetrix Human Transcriptome Array 2.0. Microarray data were obtained with an Affymetrix GeneChip Scanner 3000. After quality control and data normalization, we performed differential gene expression analysis in R using the LIMMA data package (29). The log FC value (calculated with the empirical Bayes method) indicates the log2 FC for that gene between the subgroups of interest. An adjusted P value cutoff of 0.05 (calculated with the BH procedure) and an absolute value greater than 0.6 (corresponding to change in expression higher than 1.5) for log2 FC were considered statistically significant and biologically relevant.

Analyzing all patients included for gene expression analysis showed a significant difference in mean viral load between the two groups (fig. S2). Therefore, to differentiate the effect of genes controlling viral copies and effect of genes affecting asymptomatic versus clinical dengue outcome, two types of analyses were performed: (i) filter out clinical dengue patients with high viral titers above 106. The restriction resulted in a data set of 33 patients (8 asymptomatic individuals and 25 clinical dengue patients; Table 1, analysis 1) and (ii) correlate gene expression levels to viral load and identify key genes associated with viral load variation in the two groups (Table 1, analysis 4). Linear regression analysis was implemented using the LIMMA package to assess the gene expression response to viral load variation in asymptomatic individuals and clinical dengue patients. Disease status was controlled for by including an interaction term between disease status and viral load.

To compare our results to gene expression in healthy controls, we included a previously published transcriptomic data set from Nicaragua (28), involving both clinical dengue patients and healthy controls. Using the data from these individuals, a list of statistically significant differentially expressed genes was constructed (performed at the gene rather than transcript level due to different microarray platforms used) and compared to analysis 1 (Fig. 2D and table S1).

Differential gene expression enrichment

GO enrichment analysis using GOrilla. We used GOrilla, a tool for GO vocabulary enrichment analysis, to identify sets of biological processes that are significantly overrepresented in the list of significant differentially expressed genes between asymptomatic and clinical dengue patients (26). Part of the GOrilla analysis involves discovering GO terms in a target set of genes (differentially expressed genes) versus a background set of genes (gene expression data scanned with Affymetrix GeneChip). The discovery of enriched biological processes is accomplished using a hypergeometric model that computes an enrichment P value for each overrepresented process (59). We selected 10−3 as an enrichment P value threshold to identify significantly overrepresented processes. In addition to the P value, GOrilla outputs the false discovery rate q value as an adjusted P value for multiple testing using the BH procedure and the list of genes associated with the processes that appear in the top 20 in the list.

Pathway enrichment analysis using IPA software. We used IPA (Qiagen) (www.qiagen.com/ingenuity) to identify the immunity-related canonical pathways significantly enriched in the list of differentially expressed genes between asymptomatic viremic and clinical dengue patients. IPA calculates significance values for canonical pathways using the right-tailed Fisher’s exact test (27). The significance value indicates the probability of association of the given genes with the canonical pathway by chance. IPA considers a canonical pathway to be significant and nonrandomly associated with the given genes if the P value is below a threshold of 0.05. In addition to the P value, IPA outputs other statistical measures for each canonical pathway: The z score gives the pathway SD and can be used to predict the pathway activation state, and the ratio value indicates the strength of association between the pathway and the list of differentially expressed genes.

Serum cytokine measurements

Cytokine serum concentrations were measured with the Bio-Plex Pro Human Cytokine 27-plex and Bio-Plex Pro Human Th17 Cytokine Panel 15-plex assay (Bio-Rad) and analyzed on a Luminex Magpix system (Millipore).

PBMC phenotyping

PBMCs were thawed, washed and counted in phosphate-buffered saline/bovine serum albumin, stained with surface markers [CD11c phycoerythrin (PE), CD3 PerCP/Cy5.5, CD335 PerCP/Cy5.5, HLA-DR fluorescein isothiocyanate, CD86 BV421, CD8 PE-Cy7, CD4 PerCP/Cy5.5, CD69 BV421, and CD19 APC-Cy7 (all purchased from BioLegend)], and analyzed on a FACSCanto II (BD). Data analysis was performed with FlowJo software.

Statistics

Statistical analysis was performed using GraphPad Prism (version 5.0; GraphPad) for the analysis of IgM and HI data, serum cytokine data, and PBMC phenotyping data. Differences between groups of research subjects were analyzed for statistical significance with Mann-Whitney test. A P < 0.05 was considered significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/405/eaal5088/DC1

Fig. S1. Gating strategy identifying cell subsets.

Fig. S2. Viral load in asymptomatic and clinical dengue patients.

Fig. S3. Serum cytokines related to innate immune response and inflammation are not associated with the clinical outcome of DENV infection.

Fig. S4. Serum cytokines related to innate immune responses and inflammation are not associated with the clinical outcome of dengue infection in individuals undergoing secondary DENV infection.

Fig. S5. Asymptomatic viremic individuals undergoing secondary DENV infection show differentially regulated pathways and molecules related to antigen presentation.

Fig. S6. Increased T cell activation in asymptomatic viremic individuals undergoing secondary DENV infection.

Table S1. Differentially expressed genes obtained by LIMMA.

Table S2. GO enrichment analysis using GOrilla.

Table S3. Ingenuity canonical pathways analysis of pathways significantly enriched in asymptomatic and clinical dengue infection.

REFERENCES AND NOTES

  1. Acknowledgments: We thank all the patients who accepted to participate in the study. We acknowledge all the Virology and Epidemiology Units’ staff at Institut Pasteur in Cambodia for their contribution. We thank the doctors and nurses of the three hospitals in Kampong Cham province for patient enrollment and sample collection and H. Rekol and the team from the National Dengue Control Program. We also thank A. Zhukova at Institut Pasteur in Paris for helping with the healthy control data set. Funding: This work was supported by the European Union Seventh Framework Programme (FP7/2007-2011) under grant agreement 282 378 and the PTR373 funding of Institut Pasteur International Network. Author contributions: E.S.-L. selected the samples, performed the experiments, and performed data analysis. V.D. selected the samples, conducted experiments, and performed the data analysis. A.T. performed the data analysis and prepared the manuscript. S.U. conducted the experiments and performed the data analysis. S.L. performed the fieldwork. I.C. and M.P. performed the experiments. N.C. performed the data analysis. P.B. conceived the project and study design, included the patients, performed the experiments, and revised the manuscript. K.B. performed the healthy control data set analysis. A.T. included the patients, coordinated the fieldwork, and revised the manuscript. P.D. included the patients, interpreted the data, and wrote the manuscript. T.C. designed the study, analyzed and interpreted the data, and wrote the manuscript. A.S. conceived the project, designed the study, analyzed and interpreted data, and wrote the manuscript. Competing interests: P.B. is currently an employee of GlaxoSmithKline vaccines. All other authors declare that they have no competing interests. Data and materials availability: The data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE100299.
View Abstract

Navigate This Article