Research ArticlePARASITIC INFECTIONS

Helminth infections drive heterogeneity in human type 2 and regulatory cells

See allHide authors and affiliations

Science Translational Medicine  01 Jan 2020:
Vol. 12, Issue 524, eaaw3703
DOI: 10.1126/scitranslmed.aaw3703

Parasite perturbation of immunity

Helminths infect billions of people and are known to modulate host immune responses to promote their survival. De Ruiter et al. used mass cytometry to gain a better understanding of which cells are affected by helminth infection. They analyzed samples from rural Indonesians before and after deworming treatment. These were compared to samples from Europeans and urban Indonesians, neither of which had been exposed to helminths. Helminths expanded specific type 2 and regulatory cells; many immune alterations reverted upon deworming. These details on host-pathogen interaction could inform future targeted therapies.

Abstract

Helminth infections induce strong type 2 and regulatory responses, but the degree of heterogeneity of such cells is not well characterized. Using mass cytometry, we profiled these cells in Europeans and Indonesians not exposed to helminths and in Indonesians residing in rural areas infected with soil-transmitted helminths. To assign immune alteration to helminth infection, the profiling was performed before and 1 year after deworming. Very distinct signatures were found in Europeans and Indonesians, showing expanded frequencies of T helper 2 cells, particularly CD161+ cells and ILC2s in helminth-infected Indonesians, which was confirmed functionally through analysis of cytokine-producing cells. Besides ILC2s and CD4+ T cells, CD8+ T cells and γδ T cells in Indonesians produced type 2 cytokines. Regulatory T cells were also expanded in Indonesians, but only those expressing CTLA-4, and some coexpressed CD38, HLA-DR, ICOS, or CD161. CD11c+ B cells were found to be the main IL-10 producers among B cells in Indonesians, a subset that was almost absent in Europeans. A number of the distinct immune profiles were driven by helminths as the profiles reverted after clearance of helminth infections. Moreover, Indonesians with no helminth infections residing in an urban area showed immune profiles that resembled Europeans rather than rural Indonesians, which excludes a major role for ethnicity. Detailed insight into the human type 2 and regulatory networks could provide opportunities to target these cells for more precise interventions.

INTRODUCTION

There are considerable population differences in immune profiles (1, 2), which, although in part can be explained by genetic factors, seems to be largely driven by environmental exposures (3, 4). One such exposure is to helminths, which are ubiquitous in many parts of the world (5). Such parasites are known as the strongest natural inducers of type 2 immune responses (6) characterized by CD4+ T helper 2 (TH2) cells secreting the hallmark cytokines interleukin-4 (IL-4), IL-5, and IL-13, as well as group 2 innate lymphoid cells (ILC2s), which are the predominant innate source of type 2 cytokines. Although ILC2s have been studied extensively in mice (79), it has been challenging to study these cells in humans because of their low frequency in peripheral blood (10). Together, these type 2 immune responses lead to eosinophilia, expansion of basophils and mast cells, goblet cell hyperplasia, and the production of immunoglobulin E (IgE) (11). There is evidence for a role of type 2 immune responses in controlling helminth parasites through killing or expulsion, and in inducing tissue repair, necessary to protect against damage caused by tissue-migrating helminths (12). However, there is increasing realization that type 2 cells might participate in maintaining physiological homeostasis, e.g., glucose metabolism or thermoregulation (13). It is also known that these cells can play a pathological role, e.g., in allergic diseases, such as asthma (14).

Besides polarizing the immune system toward a type 2 immune response, helminths can induce a strong regulatory network (1518), which can enhance their survival within their host but at the same time affect responses to bystander antigens, such as allergens, autoantigens, or vaccines (19). Regulatory T cells (Tregs), expressing FOXP3, are an important component of such a network and mediate their effects through suppressive cytokines [e.g., IL-10 and transforming growth factor–β (TGF-β)] and/or via the expression of suppressor molecules such as cytotoxic T lymphocyte antigen–4 (CTLA-4) (20). Although longitudinal studies assessing the effect of deworming on Tregs are rare, we recently found that not the Treg frequencies but the expression of CTLA-4 on CD4+ T cells significantly declined in anthelmintic-treated individuals (21).

Given the spectrum of immunomodulatory effects that helminths exhibit, we believe that by understanding type 2 and regulatory responses in depth, it will be possible to devise interventions that could help vaccine responses or curtail inflammation that damages tissues and organs in a more targeted manner.

With the advent of mass cytometry, it has become possible to capture the complex profile of immune cells (22). To better understand the helminth-modulated responses and identify cell populations that might be closely linked to this process, we performed immune profiling by mass cytometry with an emphasis on type 2 and regulatory responses using simultaneous measurement of 37 cellular markers at the single-cell level to allow high-resolution dissection of the cellular composition. A well-established technique for mass cytometry data analysis is t-distributed stochastic neighbor embedding (t-SNE) (23), which allows visualization of all concurrent marker expression profiles of cells on a two-dimensional plot in an unbiased fashion. However, t-SNE does not scale well to large amounts of data. Recently introduced techniques, such as Uniform Manifold Approximation and Projection (UMAP) (24) and hierarchical SNE (HSNE) (25) promise to overcome this scaling problem. Here, we applied HSNE, as implemented in Cytosplore (26), to visualize and cluster the data. These methods were applied to study type 2 and regulatory responses ex vivo, in addition to determining function in terms of cytokine production in Europeans and helminth-infected Indonesians, before and one year after 3-monthly anthelmintic treatment, when they became free of helminth infection.

RESULTS

Distinct immune signatures between Europeans and rural Indonesians

The first study cohort consisted of age- and sex-matched Europeans without prior exposure to helminths and Indonesians from a rural area who were infected with soil-transmitted helminths before and after 1 year of deworming (tables S1 and S2). The 37-metal isotope–tagged monoclonal antibody panel (table S3) allowed the identification of six major immune lineages [CD4+, CD8+, γδ T cells, B cells, myeloid cells, and innate lymphoid cells (ILCs; CD3CD7+)] and major subpopulations and detailed cell clusters within. By using the t-SNE–based HSNE (26), we could explore the full mass cytometry dataset containing 20.3 million live CD45+ cells at the single-cell level without the need for downsampling. At the overview level, landmarks (representative cells) depict the global composition of the entire immune system and distinguish the major immune lineages, which were annotated based on lineage marker expression overlays (Fig. 1, A and B). Quantification of cell frequencies revealed a significantly higher frequency of B cells in Indonesians (P < 0.05), but no other differences were found between Europeans and Indonesians at the lineage level (Fig. 1C and fig. S1). Next, CD4+ T cell landmarks, representing 6.3 million cells, were selected at the overview level, and a new higher-resolution embedding was generated at the second level of the hierarchy (Fig. 1D), revealing subpopulations within the CD4+ T cell lineage (Fig. 1D). Stratification by origin of the samples revealed a notably different distribution of the CD4+ T cells between Europeans and Indonesians, visualized by cell density plots (Fig. 1E). Different cellular distributions were also observed within other lineages (CD8+, γδ T cells, B cells, myeloid cells, and ILCs) (Fig. 1E), suggesting very distinct immune signatures between Europeans and rural Indonesians in both the innate and adaptive immune compartments.

Fig. 1 Distinct immune signatures between Europeans and rural Indonesians.

(A) First HSNE level embedding of 20.3 million peripheral immune cells from Europeans (n = 10) and rural Indonesians infected with helminths (n = 10) before and after 1 year of deworming. In all figures, color represents arcsin5-transformed marker expression as indicated. Size of the landmarks represents area of influence (AoI). (B) Major immune lineages annotated on the basis of lineage marker expression. (C) Comparison of lineage proportions relative to total cells between Europeans (EU) and rural Indonesians (ID). Differences between EU and ID were tested with Student’s t test. *P < 0.05. (D) Second HSNE level embedding of the CD4+ landmarks selected from the overview level of total PBMCs, as indicated by the red circle. Both landmarks (left) and the density features of the CD4+ T cells (right) are shown. Density is indicated by color. (E) Density plots per lineage stratified by sample origin. TCR, T cell receptor.

A CD161+ subpopulation of TH2 cells is expanded in rural Indonesians and decreases after anthelmintic treatment

Within the CD4+ T cells, a distinct population of TH2 cells was found that expressed GATA3, CD25, CD127, CD45RO, and chemoattractant receptor-homologous molecule expressed on TH2 (CRTH2), the latter being the most reliable marker to identify human TH2 cells (Fig. 2A) (27). The frequency of total TH2 cells was higher in rural Indonesians compared with Europeans, and deworming resulted in a decrease (Fig. 2B). This is in line with the observation that the proportion of circulating eosinophils and serum levels of total IgE, both markers of the type 2 response, significantly decreased in the helminth-infected Indonesians after treatment (P < 0.01; fig. S2).

Fig. 2 A CD161+ subpopulation of TH2 cells is expanded in rural Indonesians and decreases after anthelmintic treatment.

(A) Fourth HSNE level embedding of the CRTH2+ landmarks (TH2 cells) selected from the second HSNE level embedding of 6.3 million CD4+ T cells, as indicated by the black dashed circle. All cells in the third level were selected for the fourth level. (B) Frequency of TH2 cells relative to CD4+ T cells. Differences between EU (n = 10) and rural ID pretreatment (ID Pre, n = 10) were tested with Student’s t test, whereas differences between ID Pre and ID Post (n = 10) were assessed using paired t tests. **P < 0.01. (C) Cluster partitions of TH2 cells using density-based GMS clustering. The black dashed circle indicates three subpopulations. (D) Marker expression of CD27 and CD161 on TH2 cells. (E) Density features of TH2 cells. (F) Frequency of three TH2 subpopulations (CD161+CD27, CD161CD27, and CD161CD27+) relative to total TH2 cells. *P < 0.05. (G) A heat map summary of median expression values (same color coding as for the embeddings) of cell markers expressed by CRTH2+ clusters identified in (C) and hierarchical clustering thereof. To compare cluster abundance between EU and ID Pre, Student’s t test was used, whereas paired t test was used to compare ID Pre and ID Post. Colored symbols below the clusters indicate statistical significance.

Further analysis of TH2 cells revealed 22 phenotypically distinct clusters using the Gaussian mean shift (GMS) clustering (Fig. 2C). TH2 cells showed considerable heterogeneity, and based on the expression of the lectin-like receptor CD161 and CD27, the latter which is lost on highly differentiated memory CD4+ T cells (28, 29), three subpopulations could be identified (CD161+CD27, CD161CD27, and CD161CD27+ TH2 cells; Fig. 2, C and D). Visualization of TH2 cells by density plots (Fig. 2E) showed that the proportion of CD161+ TH2 cells may have been expanded in rural Indonesians compared with Europeans, although the difference fell short of statistical significance, but a decrease was observed after anthelmintic treatment (Fig. 2F). Deeper analysis at the cluster level revealed three clusters within the CD161+ TH2 population that substantially declined upon deworming (Fig. 2G). These clusters were not only reciprocally expressing CD7 and KLRG1 (CD161+CD7KLRG1+ and CD161+CD7+KLRG1) but also contained cells that did not express CD7 or KLRG1 (Fig. 2G), suggesting that helminth infection expanded not only effector cells in a more terminally differentiated state (30) but also CD161+ cells in transition. In contrast, after 1 year of treatment, we observed a notable increase of three CD7+ TH2 clusters that only weakly expressed GATA3 and CRTH2 (Fig. 2G), suggesting an expansion of cells with lower type 2 cytokine production (31, 32).

The proportion of poorly differentiated CD27+ TH2 cells was higher in Europeans (Fig. 2F), and this difference was specifically in the CCR7 cluster within the CD27+ TH2 cell subpopulation (Fig. 2G). This finding indicates that in contrast to rural Indonesians, the frequency of highly differentiated memory effector TH2 cells in the immune system of Europeans is low.

Overall frequency of ILC2s is expanded in rural Indonesians but does not decrease after anthelmintic treatment

The study of human helper-like ILCs has been difficult because of their low frequency in peripheral blood. However, using mass cytometry, it was possible to analyze these cells at a detailed level as part of 2.8 million ILCs (CD3CD7+), which include natural killer cells as well (Fig. 3A). The embedding of a CD25+CD127+CD161+ subpopulation revealed two distinct clusters that were phenotyped as ILC2s and ILC3s based on the expression of CRTH2 and c-Kit, respectively (Fig. 3A). It was found that other markers such as GATA3, KLRG1, CD45RA, and CCR6 could also be expressed by ILC2s and/or ILC3s (Fig. 3A). Consistent with previous work (33), we observed that whereas GATA3, the TH2 master transcription factor, was highly expressed in CRTH2+ ILC2s, it was also expressed in CRTH2 ILCs and can therefore not exclusively be used to define ILC2s. In contrast to ILC3s, the proportion of ILC2s was higher in rural Indonesians compared with Europeans but did not decrease after anthelmintic treatment (Fig. 3B). Further characterization of ILC2s resulted in eight phenotypically distinct clusters that could be distinguished by the expression of CD45RA, KLRG1, and CCR6 (Fig. 3C), showing more heterogeneity than what has been described so far based on the varying expressions of KLRG1 and CCR6 on ILC2s (34). Moreover, the expression of CCR6, involved in gut homing (35), has been seen on peripheral blood ILC2s in patients with inflamed tissues (36). We observe a lower frequency of KLRG1+CCR6+ ILC2s (cluster 5) in Indonesians infected with helminths residing in the gastrointestinal tract, compared with Europeans (Fig. 3D).

Fig. 3 Overall frequency of ILC2s is expanded in rural Indonesians but does not decrease after anthelmintic treatment.

(A) The right panel shows the third HSNE level embedding of the CD25+CD161+CD127+ landmarks selected from the second HSNE embedding of 2.8 million ILCs (CD3CD7+) shown in the left and indicated by the black dashed circle. (B) Frequency of ILC2s relative to total CD45+ cells. Differences between EU (n = 10) and ID Pre (n = 10) were tested with Student’s t test, whereas differences between ID Pre and ID Post (n = 10) were assessed using paired t tests. **P < 0.01. (C) Data level embedding of ILC2s. The top left panel shows the cluster partitions using GMS clustering, whereas the other panels show the expression of KRLG1, CCR6, and CD45RA. (D) A heat map summary of median expression values (same color coding as for the embeddings) of cell markers expressed by ILC2 cell clusters identified in (C) and hierarchical clustering thereof. To compare cluster abundance between EU and ID Pre, Student’s t test was used, whereas paired t test was used to compare ID Pre and ID Post. Colored symbols below the clusters indicate statistical significance.

Helminth infections increase the expression of inhibitory molecules on Tregs

Tregs characterized as FOXP3+CD25highCD127low cells appeared as a distinct subpopulation within CD4+ T cells (Fig. 4A). The frequency of Tregs, when assessed as a whole, did not differ between Europeans and Indonesians nor changed after deworming (Fig. 4B). However, the immune profile of rural Indonesians consisted of more effector Tregs expressing CD45RO (Fig. 4C), whereas in Europeans, a clear population of Tregs was positioned in the naïve, CD45RA+, compartment (Fig. 4D). In line with previous studies describing Tregs as phenotypically and functionally heterogeneous (20), we found several subpopulations within the effector Treg compartment expressing CTLA-4, HLA-DR, CD38, and/or inducible costimulator (ICOS) (Fig. 4E). A substantial proportion of the effector Tregs expressed CTLA-4, a marker that is associated with the suppressive function of Tregs (37). In concordance with this, the proportion of CTLA-4+ Tregs was higher in rural Indonesians compared with Europeans and decreased after anthelmintic treatment (Fig. 4F).

Fig. 4 Helminth infections increase the expression of inhibitory molecules on Tregs.

(A) Fourth HSNE level embedding of the FOXP3+ landmarks (Tregs) selected from the second HSNE embedding of 6.3 million CD4+ T cells, as indicated by the black dashed circle. All cells from the third level were selected. (B) Frequency of Tregs relative to CD4+ T cells. Differences between EU (n = 10) and ID Pre (n = 10) were tested with Student’s t test, whereas differences between ID Pre and ID Post (n = 10) were assessed using paired t tests. (C) Frequency of CD45RO+ effector Tregs relative to total Tregs. *P < 0.05. (D) Density features of Tregs. (E) Marker expression of CTLA-4, ICOS, CD38, HLA-DR, and CD161 by Tregs. (F) Frequency of CTLA-4+ Tregs relative to total Tregs. **P < 0.01. (G) Treg cluster partitions using GMS clustering. (H) A heat map summary of median expression values (same color coding as for the embeddings) of cell markers expressed by FOXP3+ Treg clusters identified in (G) and hierarchical clustering thereof. To compare cluster abundance between EU and ID Pre, Student’s t test was used, whereas paired t test was used to compare ID Pre and ID Post. Colored symbols below the clusters indicate statistical significance. (I) Frequency of CD161+ Tregs relative to total Tregs. **P < 0.01.

Next, we identified 27 phenotypically distinct clusters within Tregs (Fig. 4, G and H). Analysis at the cluster level revealed that the frequencies of 6 of 27 clusters were substantially higher in rural Indonesians compared with Europeans, distinguished by the expression of only CTLA-4 or coexpression of CD38 and/or HLA-DR and/or ICOS (Fig. 4H). One of these clusters (cluster 7) expressing CTLA-4, HLA-DR, CD38, and ICOS, decreased upon deworming, suggesting that this population of effector Tregs is particularly important in the immune response induced by helminths. Of note, among the CTLA-4+ Tregs, we identified a cluster expressing CD161 (cluster 25; Fig. 4H). CD161+ Tregs were recently described as a highly suppressive, cytokine-producing population of Tregs that was enriched in the intestinal mucosa, particularly in inflammatory bowel disease, where they can enhance wound repair (38). The proportion of CD161+ Tregs was higher in rural Indonesians compared with Europeans but did not change after deworming (Fig. 4I), suggesting that factors other than helminth infections might be associated with the presence of such cells.

Type 2 cytokine–producing cells in rural Indonesians (ILC2s, CD4+, CD8+, and γδ T cells) and their alteration after deworming

After the phenotypic characterization of type 2 cells, the cytokine production of phorbol 12-myristate 13-acetate (PMA)/ionomycin-stimulated cells was analyzed through a second CyTOF panel (table S4), which included interferon-γ (IFN-γ), tumor necrosis factor–α (TNFα), IL-2, IL-17, IL-10, and IL-4/IL-5/IL-13 (simultaneously assessed and hereafter referred to as “type 2 cytokine”), to assess their functional capacity. Type 2 cytokine–producing CD45+ cells were analyzed in Cytosplore, and clustering on surface markers revealed notable differences between Europeans and helminth-infected Indonesians (Fig. 5, A to C). Overall, four distinct type 2 cytokine–producing subpopulations were identified (Fig. 5D and fig. S3), of which CD4+ T cells and ILC2s showed the greatest per-cell type 2 cytokine expression (Fig. 5E). Whereas the CD25 proportion of CD4+ and CD8+ T cells and all γδ T cells coexpressed IFN-γ with the type 2 cytokines, ILC2s did not (Fig. 5, E and F). Helminth-infected rural Indonesians exhibited a higher frequency of total type 2 cytokine–producing cells compared with Europeans, and this trend was consistently observed for the individual subpopulations (Fig. 5, G and H), suggesting that helminths expand the cellular sources for these cytokines. The proportion of type 2 cytokine–producing cells declined after deworming (Fig. 5G), and this was attributed to a decrease in type 2 cytokine–producing CD4+ and CD8+ T cells and ILC2s (Fig. 5H).

Fig. 5 Type 2 cytokine–producing cells (ILC2s, CD4+, CD8+, and γδ T cells) and their alteration after deworming in rural Indonesians.

(A) IL-4/5/13+ cells were manually gated using FlowJo. Data from one representative individual are shown. (B) First HSNE level embedding of total IL-4/5/13+ cells clustered on surface markers. (C) Density features of IL-4/5/13+ cells. (D) Major immune cell subpopulations producing type 2 cytokines annotated on the basis of lineage marker expression (see fig. S3). (E) Histogram showing the median signal intensity (MSI) distribution of IL-4/5/13 and IFN-γ for the subpopulations identified in Fig. 5D. (F) Marker expression of CD25 and IFN-γ by IL-4/5/13+ cells. (G) Frequency of total IL-4/5/13–producing cells relative to total CD45+ cells. Differences between EU (n = 10) and ID Pre (n = 10) were tested with Student’s t test, whereas differences between ID Pre and ID Post (n = 10) were assessed using paired t tests. **P < 0.01. (H) Frequency of IL-4/5/13–producing clusters identified in (D) relative to total CD4+ T, ILC2, γδ T, or CD8+ T cells. *P < 0.05 and **P < 0.01.

The proportion of type 2 cytokine–producing CD8+ T cells correlated strongly with a subset previously defined as type 2 cytotoxic T (Tc2) cells (r = 0.87, P < 0.001) (39, 40), which were phenotypically identified as a GATA3+ cluster within a subpopulation of CD45RO+CCR7CD161CD56 CD8+ T cells (2.1% of CD8+ T cells), expressing CRTH2, CD25, CD127, and CD7 (Fig. 6). However, it should be noted that the frequency of Tc2 cells varied considerably, from 0.09 to 16.8% of CD8+ T cells, in different individuals. Although there was a decrease in type 2 cytokine–producing CD8+ cells, there was no substantial decrease in Tc2 cells after deworming, which might indicate that factors other than helminths can induce these cells, or historical rather than current exposure to helminths is more decisive in driving the expansion of these cells.

Fig. 6 Tc2 cells are the source of type 2 cytokines produced by CD8+ T cells.

(A) CCR7CD161CD56 landmarks were selected from the second-level HSNE embedding of CD8+ T cells, as indicated by the black dashed circle. (B) From the next-level embedding, GATA3+ landmarks were selected, and their marker expression is shown at the fourth-level embedding. (C) Correlation of type 2 cytotoxic T (Tc2) cells identified in (A) and (B) and IL-4/5/13–producing CD8+ T cells identified in (Fig. 5D). Colors indicate paired samples (before and after treatment) from Indonesian individuals (n = 10), and black dots represent European individuals (n = 10). Spearman’s rank correlation was used for statistical analysis.

IL-10–producing B and CD4+ T cells revealed by mass cytometry

Similarly, we sought cells that produced the suppressor cytokine IL-10 to further understand the regulatory network. The IL-10–producing CD45+ cells were analyzed in Cytosplore (Fig. 7, A and B), and clustering on surface markers revealed a major cluster of CD4+ T cells and minor clusters of CD8+ T cells, CD4CD8 T cells, and B cells (Fig. 7, C and D, and fig. S4). Unlike what was seen with CTLA-4 expression, no differences in frequencies of total IL-10–producing cells (relative to CD45+ cells) were observed among Europeans and rural Indonesians [Europeans (EU), 0.5% (median); Indonesian pretreatment (ID Pre), 0.5% of total CD45+ cells; P = 0.86], and these did not change after anthelmintic treatment (ID Pre, 0.4% of total CD45+ cells; P = 0.40).

Fig. 7 IL-10–producing B and CD4+ T cells revealed by mass cytometry.

(A) IL-10+ cells were manually gated using FlowJo. Data from one representative individual is shown. (B) First HSNE level embedding of total IL-10+ cells clustered on surface markers. (C) Density features of IL-10+. (D) The major immune cell subsets producing IL-10, annotated on the basis of lineage marker expression (see fig. S4). (E) Data level embedding of IL-10+ B cells and the cluster partitions using GMS clustering (bottom right). (F) Relative composition of IL-10+ B cells, comparing EU (n = 10) and ID Pre (n = 10). Below, density features of IL-10+ B cells are shown. (G) First level HSNE embedding of IL-10+CD4+ landmarks selected in (D) and clustered on cytokines. Four IL-4/5/13+ clusters were identified as indicated in the right panel showing the cluster partitions. (H) Histograms showing the MSI distribution of cytokines for the four clusters identified in (G). (I) Frequency of cluster 1, coexpressing IL-4/5/13 and IL-10 [identified in (G)], relative to total IL-10–producing CD4+ T cells. Differences between EU and ID Pre were tested with Student’s t test, whereas differences between ID Pre and ID Post (n = 10) were assessed using paired t tests. *P < 0.05 and **P < 0.01.

A distinct population of IL-10+ B cells (0.5% of total B cells) was identified, and further analysis showed that it consisted of three clusters, namely, CD11c+CD38, CD11cCD38+, and CD11cCD38 cells (Fig. 7E). Whereas the composition of IL-10+ B cells did not change after deworming, IL-10+ B cells from rural Indonesians clearly contained more CD11c+CD38 cells compared with Europeans, who had relatively more CD11cCD38+ IL-10+ B cells (Fig. 7F). These results indicate a different phenotype of IL-10+ B cells in the two populations, which is in line with significantly more CD11c+ B cells, representing chronically stimulated cells, in rural Indonesians compared with Europeans [EU, 5.0% (median); ID Pre, 13% of total B cells; P < 0.01].

When IL-10–producing CD4+ T cells were clustered on cytokines, four clusters were identified, which all expressed type 2 cytokines but varied in terms of coexpression of other cytokines (IFN-γ, TNFα, and/or IL-2) (Fig. 7, G and H). When considering all CD4+ T cells producing IL-10 and type 2 cytokines, irrespective of other cytokines, their frequency was higher in rural Indonesians compared with Europeans. However, only the IL-10 and type 2 cytokine–producing cluster that was negative for IFN-γ, TNFα, and IL-2 (cluster 1) decreased notably after deworming (Fig. 7I).

The immune profile of urban Indonesians resembles that of Europeans rather than rural Indonesians

We next sought to assess whether ethnic differences were driving the observed differences between Europeans and rural Indonesians. Therefore, we analyzed a second cohort of eight Europeans and eight Indonesians from urban centers, such as central Jakarta, where helminth infections are not found (urban Indonesians) (41), and eight Indonesians from the rural setting with current helminth infection (rural Indonesians) (tables S1 and S2). The same CyTOF panels and analysis strategy were used for the 14.3 million unstimulated and 12.2 million stimulated cells measured in this independent cohort (Fig. 8 and figs. S5 and S6).

Fig. 8 The immune profile of urban Indonesians is more similar to Europeans than rural Indonesians.

(A) Frequency of TH2 cells relative to CD4+ T cells. Differences between EU (n = 8), Indonesians from urban settings without helminth infections (ID urban, n = 8) and rural Indonesians infected with helminths (ID rural, n = 8) were tested with ANOVA followed by Tukey’s posttest. **P < 0.01 and ***P < 0.001. (B) Third HSNE level embedding of the TH2 cells and their density features. (C) Frequency of CD161+CD27 cells within TH2 cells. *P < 0.05 and **P < 0.01. (D) Data level embedding of ILC2s. The top right panel shows the cluster partitions using GMS clustering, whereas the other panels show the expression of KRLG1, CCR6, and CD45RA. Right panel shows the frequency of CCR6KLRG1+ cells within ILC2s. *P < 0.05 and **P < 0.01. (E) Frequency of IL-4/5/13–producing cells within CD4+ T cells. **P < 0.01 and ***P < 0.001. (F) Frequency of IL-4/5/13–producing cells within ILC2s. *P < 0.05. (G) Frequency of CD161+ (left) and CTLA-4+ (right) cells within Tregs. *P < 0.05, **P < 0.01, and ***P < 0.001. (H) Data level embedding of IL-10+ B cells and the cluster partitions using GMS clustering (bottom right), whereas the other right panels show the expression of CD11c and CD38 and the density features in left panels. (I) Frequency of CD11c+ cells within IL-10–producing B cells. **P < 0.01.

Increased frequencies of TH2 cells were present in rural Indonesians compared with both European and urban Indonesians (Fig. 8A). Within TH2 cells, rural Indonesians had increased levels of CD161+ cells compared with Europeans and urban Indonesians (Fig. 8, C and D). Although in this independent cohort total ILC2 frequencies were not increased in rural Indonesians based on surface markers (fig. S5D), ILC2 subsets differed between populations (Fig. 8D). Similar to the initial cohort, rural Indonesians and also urban Indonesians had increased CCR6KLRG1+ ILC2s compared with Europeans. Functionally, ILC2 and TH2 cells were also increased in rural Indonesians as increased type 2 cytokines were produced compared with Europeans, whereas urban Indonesians resembled Europeans (Fig. 8, E and F).

As observed in the original cohort, rural Indonesians had increased proportions of CD161+ and CTLA-4+ Tregs compared with Europeans and urban Indonesians (Fig. 8, G and H). Total Tregs were also increased in rural Indonesians in this cohort compared with Europeans (fig. S5E). IL-10+ B cells of rural Indonesians in the independent cohort were characterized by increased expression of CD11c compared with Europeans and urban Indonesians as in the original cohort (Fig. 8, H and I).

Together, we found urban Indonesians resembled Europeans more closely than rural Indonesians, demonstrating that ethnic differences were not driving the altered immune profiles. Moreover, we were able to validate the main findings of altered type 2 and regulatory immune cell populations in an independent cohort. A summary of the heterogeneity of the type 2 and regulatory cell phenotypes, and the type 2 cytokine– and IL-10–producing cells that align with Europeans, rural helminth-infected Indonesians (before and after anthelmintic treatment), and urban Indonesians is given in fig. S6.

DISCUSSION

Our study used mass cytometry to analyze type 2 and regulatory immune cells of healthy Europeans, as well as of helminth-infected Indonesians before and 1 year after deworming, to start to understand the heterogeneity in the immune response of populations living in different geographical areas. We have identified and shown the expansion of not only TH2 but also the rare ILC2 cells, both sources of type 2 cytokines, in helminth-infected Indonesians compared with Europeans. When considering the cells by their phenotype, TH2 cell frequencies substantially decreased after deworming, but the frequencies of ILC2s remained unchanged. However, the frequency of not only TH2 but also ILC2s producing type 2 cytokines declined after anthelmintic treatment, indicating decreased functional activity. There is very little known about the role of ILC2s in human helminth infections (42, 43). Although the identification of CRTH2 as a marker of human ILC2s has been a great advantage (44), only one study analyzed ILC2s using CRTH2 in the context of human helminth infection (43). This study found lower frequencies of ILC2s in Schistosoma haematobium–infected African children, but in line with the trend in our study, the proportion of ILC2s increased 6 weeks after anthelmintic treatment (43).

We identified eight ILC2 clusters based on the heterogeneous expression of KLRG1, CD45RA, and CCR6, paving the way for future studies to investigate their specific function. We identified CD45RA+c-Kit+ ILC3s that did not express CCR6, a marker previously described to be expressed by ILC3s (33, 45), and hypothesize that this subpopulation might consist of the recently described ILC precursors, as these cells lack CCR6 expression and are CD45RA+ (46).

There is increasing evidence for the heterogeneity of TH2 cells, e.g., pathogenic effector TH2 (peTH2) cells have recently been found in patients with allergic eosinophilic inflammatory diseases, which have enhanced effector function as assessed by cytokine production (47, 48). It is thought that chronic antigen exposure drives peTH2 cells, characterized as CD161+hPGDS+CD27, to differentiate from conventional TH2 cells; however, it is not known whether peTH2 cells are induced by human helminth infections (48). Here, we describe the increase of a peTH2-like CD27CD161+ subset of TH2 cells in helminth-infected individuals, which substantially decreased after deworming. Given the general lack of severe allergic diseases in rural areas where helminths are endemic (49), it would be interesting to assess whether in individuals chronically infected with helminths, these cells have a more regulated function and, thus, less pathogenic activity.

Tregs have been described as a heterogeneous population with differing expression of HLA-DR (50), ICOS (51), and CD38 (52) within the FOXP3+ Treg population. Functionally, Tregs expressing HLA-DR or CD38 have been shown to be highly suppressive (50, 52), and whereas the expression of ICOS has been suggested to define a Treg subset that has the capacity to produce large amounts of IL-10, ICOS Tregs seem to produce mainly TGF-β (51). Moreover, a recent study has reported an interesting subset of Tregs expressing CD161 (38). By using mass cytometry, we could not only confirm the presence of these distinct Treg phenotypes in our study population but also visualize the marker distribution, which revealed that all HLA-DR+, ICOS+, CD38+, and CD161+ cells were found within the CTLA-4+ Treg subset.

Although total Treg frequencies were not expanded in helminth-infected individuals and no treatment-related change was observed, when considering them at the subset level, the proportion of CTLA-4+ Tregs was substantially higher in helminth-infected individuals and declined after treatment, consistent with previous work in children from the same study area (18, 21). Analysis at the cluster level revealed that the CTLA-4+ clusters in particular, often coexpressing ICOS and/or HLA-DR and/or CD38, were expanded in rural Indonesians compared with Europeans, indicating that helminths induce a particular Treg phenotype that could be represented by cells with increased regulatory capacity (21). We also found a substantially higher proportion of CD161+ Tregs in rural Indonesians compared with Europeans, and given that this population was shown to accelerate epithelial barrier healing in the gut (38), they might play a role in healing of wounds caused by soil-transmitted helminths, either during their migration throughout the body or upon their residence in the intestine.

Besides TH2 cells and ILC2s being a source of type 2 cytokines, we identified Tc2 cells [type 2 cytokine–secreting CD8+ T cells (39)] and a cluster of γδ T cells, which produced type 2 cytokines. Although Tc2 cells have recently been found to be enriched in patients with eosinophilic asthma (40), this current study characterizes these cells in the context of human helminth infections. Like Tc2 cells, not much is known about type 2 cytokine–producing γδ T cells, although their existence has been described before (5357). Inagaki-Ohara et al. (54) showed a protective role of γδ T cells against infection with Nippostrongylus brasiliensis in mice, which was associated with the production of type 2 cytokines, particularly IL-13. However, to our knowledge, the presence of type 2 cytokine–producing γδ T cells in human helminth infection has not been previously reported. With the increasing recognition of the role γδ T cells play in shaping adaptive responses in infectious diseases (58, 59), it would be interesting to delineate their possible participation in the development of TH2 responses.

IL-10–producing B cells, known as regulatory B cells (Bregs), represent a relatively rare cell type that can suppress inflammatory responses (60). The frequency of IL-10–producing B cells was similar in Europeans and Indonesians, and deworming did not affect their frequency. However, the phenotype of these cells was markedly different with relatively more CD11c+CD38 and less CD11cCD38+ IL-10+ B cells present in Indonesians. There is increasing evidence that CD11c+ B cells are a distinct population of memory B cells, and they have been shown to expand in settings of chronic infections such as HIV, malaria, and tuberculosis, as well as in several autoimmune diseases (61, 62). Our finding of a substantially expanded population of CD11c+Tbet+ B cells in Indonesians with chronic helminth infection, of which a small fraction appeared capable of producing IL-10, indicates that through the use of mass cytometry, it becomes possible to identify previously unknown cell types that would need to be investigated for their properties including suppressor functions.

Here, we demonstrate the advantage of using mass cytometry combined with HSNE, which allowed the identification of rare cell populations, such as ILC2s, Tc2 cells, and Bregs, and provided an opportunity to analyze the heterogeneity within TH2 cells and Tregs. A comparison of urban and rural Indonesians and Europeans demonstrates that the alterations in type 2 and regulatory immune cells were not because of ethnic differences, as immune profiles of Indonesians resembled that of Europeans more than rural Indonesians. The strength of this study is that we validated these findings in an independent cohort, demonstrating the robustness of these findings. Limitations of our study include analyzing peripheral immune cells from a relatively small number of individuals. In addition, there is a lack of placebo-treated individuals and the possibility that other infections are affected by deworming. However, this study is a stepping stone to verify the findings in larger studies that allow detailed delineation of their function not only in helminth infections but also in other disease settings.

MATERIALS AND METHODS

Study design

The objective of this study was to profile type 2 and regulatory immune responses in Europeans and in helminth-infected Indonesians before and 1 year after deworming by applying mass cytometry on peripheral blood samples. The Indonesian samples were part of the SugarSPIN trial, a household-based cluster-randomized double-blind trial that was conducted in three rural villages in Nangapanda, Ende district of Flores Island (East Nusa Tenggara), Indonesia (63). The trial was approved by the ethics committee of the Faculty of Medicine, Universitas Indonesia (FKUI) (reference no. 549/H2·F1/ETIK/2013) and filed by the ethics committee of Leiden University Medical Center (LUMC). The trial is registered as a clinical trial (reference no. ISRCTN75636394). Written informed consent was obtained from participants before the study.

We were interested in individuals who were helminth infected and successfully treated with albendazole (n = 56). The treatment consisted of a single tablet of albendazole (400 mg; PT Indopharma Pharmaceutical, Bandung, Indonesia) for three consecutive days, and this regimen was given every 3 months for a total of four rounds (maximum of 12 tablets in total) between May 2014 and February 2015. Before the start of drug administration and 6 weeks after the last round of drug administration, blood and stool samples were collected as previously described (63). To ensure that the individuals who were studied were unlikely to have become reinfected (subpatently), we selected individuals in whom eosinophils and IgE were decreased after treatment (n = 10). Last, age- and sex-matched samples of healthy volunteers from The Netherlands (Caucasians) or from urban centers in Indonesia, such as central Jakarta, with no helminth infections were included in the study (tables S1 and S2). Primary data are reported in data file S1.

Parasitology

Aliquots of fresh stool samples were frozen at −20°C in the field study center and subsequently at −80°C at the Department of Parasitology of FKUI and LUMC for DNA extraction. Stool DNA isolation and real-time polymerase chain reaction (PCR) were performed pairwise (baseline and follow-up). DNA isolation from stool was performed as described elsewhere (64). Multiplex real-time PCR was performed to simultaneously detect the presence of hookworms (Ancylostoma duodenale and Necator americanus), Ascaris lumbricoides, Trichuris trichiura, and Strongyloides stercoralis using a method described previously (64). Stool samples were considered positive by PCR when cycle threshold (Ct) values were <50.

Eosinophil count and total IgE

A Giemsa-stained peripheral thin blood smear was read to assess the differential white blood cell count, resulting in a relative percentage of basophils, eosinophils, neutrophils, lymphocytes, and monocytes. Total IgE was measured in serum as described previously (65).

Peripheral blood mononuclear cell cryopreservation

After diluting heparinized venous blood twice with Hanks’ balanced salt solution (HBSS), peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll density gradient centrifugation within 12 hours after blood collection. The HBSS contained penicillin G sodium (100 U/ml) and streptomycin (100 μg/ml). After washing twice with HBSS, the PBMCs were cryopreserved in RPMI 1640 containing 20% of heat-inactivated fetal calf serum (FCS; Bodinco) and 10% dimethyl sulfoxide (DMSO). The RPMI medium contained 1 mM pyruvate, 2 mM l-glutamine, penicillin G, and streptomycin. Cryovials containing the cell suspension were transferred to a Nalgene Mr. Frosty Freezing Container (Thermo Fisher Scientific), which was placed at a −80°C freezer for a minimum of 4 hours. Subsequently, vials were stored in liquid nitrogen until analysis. The cryopreserved PBMCs collected in the field were shipped in a liquid nitrogen dry vapor shipper from Jakarta, Indonesia, to Leiden, The Netherlands for analysis.

Mass cytometry antibody staining

Two antibody panels were designed to (i) phenotype immune cells ex vivo and (ii) assess cytokine production after 6 hours of stimulation with PMA and ionomycin. Details on antibodies used are listed in tables S3 and S4. Antibody-metal conjugates were either purchased or conjugated using a total of 100 μg of purified antibody combined with the Maxpar X8 Antibody Labelling Kit (Fluidigm) according to the manufacturer’s protocol V7. The conjugated antibody was stored in 200 μl of Antibody Stabilizer PBS (CANDOR Bioscience GmbH) at 4°C. All antibodies were titrated on study samples.

On the day of the staining, cryopreserved PBMCs were thawed with 50% FCS/RPMI medium at 37°C and washed twice with 10% FCS/RPMI medium. Next, 3 × 106 cells per sample, used for phenotyping (panel 1), were stored on ice temporarily, whereas another 3 × 106 cells per sample were transferred to 5-ml round-bottom Falcon tubes (BD Biosciences) for 6 hours of incubation in 10% FCS/RPMI with PMA (100 ng/ml; Sigma) and ionomycin (1 μg/ml; Sigma). After 2 hours of incubation at 37°C, brefeldin A (10 μg/ml; Sigma) was added, after which the cells were incubated for four more hours. Subsequently, the cells were washed with phosphate-buffered saline and resuspended in Maxpar staining buffer (Fluidigm) before continuing with the antibody staining (panel 2).

For phenotyping (panel 1), the staining was based on Maxpar Nuclear Antigen Staining Protocol V2 (Fluidigm). First, cells were washed with Maxpar staining buffer and centrifuged for 5 min at 300g in 5-ml Eppendorf tubes. Then, the cells were incubated with 1 ml of 500× diluted 500 μM Cell-ID Intercalator-103Rh (Fluidigm) in staining buffer at room temperature for 15 min to identify dead cells. After washing with staining buffer, cells were incubated with 5 μl of Human TruStain FcX Fc receptor blocking solution (BioLegend) and 40 μl of staining buffer at room temperature for 10 min. Then, 50 μl of freshly prepared surface antibody cocktail was added and incubated at room temperature for another 45 min. Subsequently, cells were washed twice with staining buffer and fixed and permeabilized using the eBioscience Foxp3/Transcription Factor Staining Buffer Set (eBioscience). After cells were incubated with 1 ml of the freshly prepared Fix/Perm working solution (prepared according to the manufacturer’s instructions) for 45 min, cells were washed twice with 1× permeabilization buffer at 800g for 5 min. Next, 50 μl of the intranuclear antibody cocktail was added to 50 μl of cells resuspended in 1x permeabilization buffer and incubated for 30 min at room temperature. After the incubation, cells were washed once with 1× permeabilization buffer and twice with staining buffer before being stained with 1 ml of 1000× diluted 125 μM Cell-ID Intercalator-Ir (Fluidigm) in Maxpar Fix and Perm Buffer (Fluidigm) at 4°C overnight to stain all cells. After three washes with staining buffer and centrifugation at 800g, cells were stored as a pellet at 4°C and measured within 2 days.

To assess the cytokine production of PBMCs (panel 2), the staining was based on Maxpar Cytoplasmic/Secreted Antigen Staining Protocol V3. While the surface staining was performed exactly as described above, cells were afterward fixed by incubating them with 1 ml of freshly prepared 1× Maxpar Fix I Buffer (Fluidigm) for 20 min at room temperature. Next, cells were washed three times with Maxpar Perm-S Buffer (Fluidigm), and 50 μl of cytokine antibody cocktail was added to 50 μl of cell suspension and incubated for 40 min at room temperature. Then, cells were washed three times with staining buffer and stained with Cell-ID Intercalator-Ir as described above.

Mass cytometry data acquisition

Measurement of samples was randomized per individual to avoid bias, but samples belonging to the same individual were stained and measured together. Samples were measured with a Helios mass cytometer (Fluidigm), which was automatically tuned according to Fluidigm’s recommendations. Before measuring, cells were counted, washed with Milli-Q water, passed over a cell strainer, and brought to a concentration of 1.0 × 106 cells/ml with 10% EQ Four Element Calibration Beads (Fluidigm) in Milli-Q water. Mass cytometry data were acquired and analyzed on-the-fly using dual-count mode and noise-reduction on. Next to channels, to detect antibodies, channels for intercalators (103Rh, 191Ir, and 193Ir), calibration beads (140Ce, 151Eu, 153Eu, 165Ho, and 175Lu), and background/contamination (133Cs, 138Ba, and 206Pb) were acquired. After data acquisition, the mass bead signal was used to normalize the short-term signal fluctuations with the reference EQ passport P13H2302 during the course of each experiment. When applicable, normalized FCS files were concatenated using Helios software without removing the beads.

Mass cytometry data analysis

FlowJo V10 for Mac (FlowJo LLC) was used to gate out beads, and we discriminated live, single CD45+ immune cells with DNA stain and event length. The selected cells were exported as FCS files and analyzed using HSNE (25), as implemented in Cytosplore (26, 66). HSNE constructs a hierarchy of nonlinear similarities that can be interactively explored with a stepwise increase in detail up to the single-cell level (26). Briefly, the data exploration starts with the visualization of the embedding at the highest level, the overview level, where the layout of the landmarks (representative cells) indicates similarity in the high-dimensional space. The landmark size reflects its area of influence (AoI), containing cells that are well represented by the landmark, and the similarity of two landmarks is defined as the overlap of their respective AoIs. Coloring of the landmarks is used to represent marker expressions. At the overview level, a group of landmarks (also referred to as cluster) can be selected by manual gating based on visual cues such as marker expression or by performing unsupervised GMS clustering (66, 67) of the landmarks based on the density representation of the embedding. Next, we can zoom into this selection by means of a more detailed embedding and visualize all selected landmarks in a next level. Although this process can be repeated until the data level is reached where each dot represents a single cell, this is not imperative, and we often clustered at an intermediate level without reaching the data level. The number of hierarchical levels depended on the input-data size.

Before HSNE was applied, data were transformed using a hyperbolic arcsinh with a cofactor of 5. Furthermore, within Cytosplore, an extra channel called “SampleTag” was added to the FCS files to be able to identify from which sample an event originated after HSNE.

Clusters produced in Cytosplore were analyzed using R software (R x64 version 3.5.1, R Foundation for Statistical Computing) and RStudio (Rstudio Inc.). The package “cytofast” was used to produce heat maps, scatterplots showing subset abundance, and histograms showing the median signal intensity distribution of markers (68).

Statistical analysis

Statistical analyses were performed using R software. To compare subpopulation and cluster abundance between Europeans and Indonesians before treatment, unpaired t tests were used. Paired t tests were applied to compare pre- and posttreatment samples of Indonesians. Total IgE and eosinophil counts were log transformed for analysis, and paired t tests were used in GraphPad Prism (GraphPad Software). Spearman’s correlation was used to assess the relationship between the frequencies of Tc2 cells and type 2 cytokine-producing CD8+ T cells. P values <0.05 were considered statistically significant. For comparison of three groups in the second cohort, analysis of variance (ANOVA) followed by Tukey’s posttest were used with one-sided testing based on the differences identified in the first cohort.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/12/524/eaaw3703/DC1

Fig. S1. Lineage frequencies.

Fig. S2. Eosinophil counts and total IgE in rural Indonesians.

Fig. S3. Type 2 cytokine–producing cells.

Fig. S4. IL-10–producing cells.

Fig. S5. Immune composition of Europeans and urban and rural Indonesians.

Fig. S6. Heat map summary.

Table S1. Characteristics of the study cohorts.

Table S2. Helminth infection details of rural Indonesians.

Table S3. Antibody panel 1 (phenotyping).

Table S4. Antibody panel 2 (cytokine production).

Data file S1. Primary data.

REFERENCES AND NOTES

Acknowledgments: We would like to thank the volunteers who participated in this study. We would also like to thank all the field workers from Universitas Indonesia and Nangapanda, and Y. Kruize, S. de Jong, and A. Voskamp for help with the study. Funding: This work was supported by the Royal Netherlands Academy of Arts and Science (KNAW) Ref 57-SPIN3-JRP awarded to J.W.A.S., Universitas Indonesia Research Grant BOPTN 2742/H2.R12/HKP.05.00/2013 awarded to T.S., and the European Union’s Horizon 2020 Research and Innovation Programme under the Marie Skłodowska-Curie grant agreement no. 707404 awarded to S.P.J. The opinions expressed in this document reflect only the author’s view. The European Commission is not responsible for any use that may be made of the information it contains. Author contributions: K.d.R. was in charge of the field study in Indonesia, designed and performed the experiments, analyzed the data, and wrote the manuscript. S.P.J. performed experiments, analyzed the data, and reviewed the manuscript. D.L.T. was in charge of the field study in Indonesia and involved in setting up the study, treatment, and clinical care. K.A.S. provided input for the data analysis and developed the cytofast package in R. M.K. and S.L. performed experiments. V.v.U. developed the Cytosplore application and reviewed the manuscript. T.H. developed the Cytosplore application. M.M. provided expertise on rural urban comparisons. B.P.F.L. developed the Cytosplore application. F.K. provided consultation with mass cytometry and reviewed the manuscript. E.S. supervised and coordinated the field study. J.W.A.S. Designed the field study and reviewed the manuscript. T.S. supervised and coordinated the field study and reviewed the manuscript. M.Y. conceived and designed the study, and supervised the data analysis and writing of the manuscript. All authors revised the final manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article