Neonatal T Follicular Helper Cells Are Lodged in a Pre-T Follicular Helper Stage Favoring Innate Over Adaptive Germinal Center Responses

T follicular helper (Tfh) cells have emerged as a critical limiting factor for controlling the magnitude of neonatal germinal center (GC) reactions and primary vaccine antibody responses. We compared the functional attributes of neonatal and adult Tfh cells at the transcriptomic level and demonstrated that the Tfh cell program is well-initiated in neonates although the Tfh gene-expression pattern (i.e., CXCR5, IL-21, BCL6, TBK1, STAT4, ASCL2, and c-MAF) is largely underrepresented as compared to adult Tfh cells. Importantly, we identified a TH2-bias of neonatal Tfh cells, with preferential differentiation toward short-lived pre-Tfh effector cells. Remarkably, adjuvantation with CpG-ODNs redirect neonatal pre-Tfh cells toward committed GC-Tfh cells, as illustrated by increased expression of Tfh signature genes and reduced expression of TH2-related genes.


INTRODUCTION
Neonates and young infants share a high vulnerability to infectious diseases. Inducing efficient and sustained B-cell responses remains challenging in this age group (1,2). Numerous factors concur to limit primary antibody responses, including delayed follicular dendritic cell maturation (3), the limited development and expansion of T follicular helper (T fh ) cells and, as a result, that of germinal center (GC) B cells and plasma cells (4,5).
Over the past years, the specific role of T fh cells as the main providers of B cell help has been unveiled, highlighting a critical role of T fh cells in vaccine elicited immune responses. We previously demonstrated that T fh cell development limits early life GC reactions and resulting primary vaccine antibody responses (4). Notably, adjuvantation of a vaccine with CpG-ODNs was sufficient to partially enhance neonatal antibody responses (4). In this study, we compared the transcriptional profile of neonatal and adult T fh cells and demonstrated that the preferential neonatal polarization toward TH2 is also observed among T fh cells, with increased expression of IL-13 and other TH2-related factors, which may represent an additional negative regulation checkpoint confining activated neonatal CD4 + T cells at a pre-T fh stage. Importantly, we showed that adjuvantation with CpG-ODNs reduced the expression of IL-13 and other TH2related genes and sufficiently strengthened the levels of T fh cellassociated signature molecules to drive the full completion of GC-T fh differentiation.

MATERIALS AND METHODS
Mice C57BL/6 mice were purchased from Charles River (L'Arbresle, France), bred, and kept in pathogen-free animal facilities in accordance with local guidelines. Mice were used at 1 week (neonates) or 6-8 weeks (adults) of age. All animal experiments were approved by the Geneva Veterinary Office and conducted under relevant Swiss and European guidelines.

Microarray and Analysis
One week-old (16 mice/group) and adult C57BL/6 mice (5 mice/group) were immunized i.m. as described above. Ten days post-vaccination, inguinal draining LNs (dLNs) were pooled per mouse and per group to have a sufficient number of cells. T fh cell populations were isolated by flow-cytometry cell sorting using a MoFlo R Astrios TM flow cytometer (Beckman Coulter). Six independent experiments have been performed to obtain three independent samples per age group. Total RNA was labeled and hybridized on Agilent Whole Mouse Genome Oligo Microarrays 8 × 60 K at Miltenyi Biotec (Germany) and according to the manufacturer's protocol. Arrays were scanned with the Agilent microarray scanner and raw intensities were extracted with Feature Extraction v10.6 software. Raw intensities were integrated, background corrected and log transformed, following the quantile normalization between arrays. Intensities with detection p-values <0.01 were arbitrarily discarded. Differentially expressed genes (DEGs) were identified by the ANOVA with Tukey post-hoc test considering adjusted p-value ≤ 0.05 and fold-change (FC) ≥ 2. Protein-protein interaction networks were built with DEGs using the NetworkAnalyst program (42) and the InnateDB PPIs as database (43). Enrichment analyses were performed with the program Gene Set Enrichment Analysis (GSEA) (44), using customs gene sets of upregulated genes from CD4 + T fh effector cells [GSE43863 (45)] and from Bcl6 + T fh cells or Bcl6 − T fh cells [GSE40068 (8)]. First, raw expression data from GSE43863 and GSE40068 studies were normalized by RMA using the affy R/Bioconductor package (46), and submitted to quality control with the arrayQualityMetrics R/Bioconductor package (47). For both studies, the up-regulated genes were identified using the R/Bioconductor LIMMA package (48). The T fh effector signature (GSE43863) was generated by comparing CD4 + T fh effector cells compared with naïve and TH1 CD4 + T cells (adjusted p-value < 0.005 and FC ≥ 1.5), while the T fh Bcl6 + and Bcl6 − signatures (GSE40068) were generated by comparing CD4 + CXCR5 + Bcl6 + and CD4 + CXCR5 + Bcl6 − T cells with CD4 + CXCR5 − T cells (adjusted p-value < 0.05 and FC ≥ 2). Co-expression modules were identified with the CEMiTool R/Bioconductor package (49) using variance filter p-value < 0.05 and ORA p-value < 0.2. CEMiTool package is available at Bioconductor (https://bioconductor.org/ packages/release/bioc/html/CEMiTool.html) (49). This package unifies the discovery and the analysis of coexpression gene modules, evaluating whether modules contain genes that are over-represented by specific pathways or that are altered in a specific sample group. Biological and functional enrichment analyses were also performed with the program GSEA using the REACTOME gene sets (50). Finally, unsupervised hierarchical clustering of the samples was carried out via multiscale bootstrap resampling with the PVCLUST R package (51).

Transcriptional Profile of Neonatal T fh Cells
We (4, 52) and others (5) have shown that neonatal T fh cells elicited by aluminum (AlOH)-based adjuvanted vaccines are few and functionally altered compared to adult cells. We therefore investigated the functional attributes of neonatal and adult CD4 + CXCR5 high PD-1 high T fh cells at the transcriptomic level. CD4 + CXCR5 high PD-1 high T fh and CD4 + CXCR5 − PD-1 − T (non-T fh ) cells were FACS sorted from the draining lymph nodes (LNs) at the previously identified peak (day 10) of the primary germinal center (GC) reaction induced by TT/AlOH (4) for comparative transcription profile analysis ( Figure 1A). To visualize the global gene expression patterns of the various subsets, we first performed a principal component analysis (PCA), retaining the top 2,000 genes that contributed most to the total variance ( Figure 1B). The projection of the data variance onto the principal components plane efficiently discriminated T fh cells from non-T fh cells in both age groups (Figure 1B), while clustering adult and neonatal T fh cells together. This was confirmed by unsupervised hierarchical analysis, which grouped T fh cells from both age groups ( Figure S1). Thus, when successful the T fh differentiation process essentially follows a similar path in early as in adult life.
Nevertheless, the gene expression profiles of neonatal and adult T fh samples differed, revealing functionally differently programmed T fh cells ( Figure 1C). Comparing T fh cells from neonatal and adult mice with the corresponding age-matched non-T fh cells identified 2,301 and 3,549 differentially expressed genes, respectively. Overlap comparison showed that 1,710 genes were differentially expressed in T fh cells of both neonatal and adult immunized mice, 591 genes were exclusively differentially expressed in neonatal T fh cells, and 1,839 genes were exclusively differentially expressed in adult T fh cells ( Figure 1C).
To get more insight into the key genes leading to functionally differently programmed T fh cells in early or adult life, PPI networks were generated from the differentially up-regulated genes between neonatal and adult T fh cells ( Figure 1D). The network derived from the genes differentially expressed in adult T fh cells showed that most of the up-regulated proteins have an established role in T fh biology and function (Bcl6, Ascl2, Pou6f1, IL-21, and Cxcr5). In accordance with the preferential TH2 polarization of early life responses, IL-13 was strongly enriched in neonates vs. adults ( Figure 1D). Unexpectedly, three cancer related-pathways genes (Tal1, PPAR-γ, and RXRA) were identified as hub genes in neonates ( Figure 1D). Tal1 is expressed early, in hematopoietic stem cells and progenitor cells (53,54), and subsequently silenced during T-cell development [reviewed in (55)]. It forms a large transcriptional complex with E proteins, LMO family proteins, LDB1, GATA2, and GATA3 (56)(57)(58). TAL1, GATA3, and RUNX1 coordinately regulate the expression of downstream target genes. PPAR-γ is a member of the peroxisome proliferator-activated receptor family and forms heterodimer with RXRs to promote their downstream effects, i.e., suppress the transcription of target genes (59). Remarkably, in adults both PPAR-γ (41) and RXRA (60) negatively regulate T cell activation to prevent T fh cell formation. These hub genes may thus play an essential role to functionally alter neonatal T fh cell differentiation.

T fh Cell Differentiation Is Initiated in Neonates but T fh Cells Remain Lodged in a pre-T fh Stage
We then selected the T fh signature genes from published data sets to perform gene set-enrichment analyses (GSEA) with our data. This confirmed that both neonatal and adult cells were enriched for the T fh lineage gene set (GEO accession code GSE43863) (Figure 2A), indicating that the T fh cell differentiation program may succeed in neonates. However, neonatal T fh Bcl6 + cells exhibited reduced gene expression signatures compared to adults, while using another GSEA (accession code GSE40068) indicated increased T fh Bcl6 − signatures (Figure 2A) To further analyze the transcriptional differences among early life and adult T fh cells, we examined the expression of a set of genes described as up-or down-regulated in T fh cells compared to non-T fh CD4 + helper T cells (8,12,13,61,62). The heatmap of the differentially expressed genes (DEGs) ( Figure 2B) confirmed that the gene expression of neonatal and adult T fh samples differed from the control CD4 + CXCR5 − PD-1 − samples. Although the overall T fh signature was again present in neonatal T fh cells, the expression levels of Cxcr5, Batf, c-maf, and Il-21 were lower than in adult T fh cells, strengthening our previous observations (4). Compared to adult T fh cells, neonatal T fh cells exhibited much lower expression of Bcl6 (8,29,63,64) and achaete-scute homolog 2 (Ascl2) (9), which controls CXCR5 expression and thus the follicular positioning of pre-T fh cells (9). The TBK1 kinase, which controls the maintenance of Bcl6 expression and is thus required for the commitment to the GC-T fh program (27), was also less strongly expressed in neonatal T fh cells.
Bcl6 can bind to promoters and enhancers of genes that encode proteins that control T cell-migration, promoting nonfollicular positioning of T cells (7). IL-7R is one of the most repressed T fh -relevant genes by Bcl6 and its suppression is critical in T fh cell differentiation (65). Although IL-7R is down-regulated in neonatal T fh cells, its inhibition is almost two times weaker than in adult T fh cells (expression level change of 1.89-fold greater in neonates compared to adults). Liu et al. (65) recently demonstrated that IL-7R expression was inversely correlated with T fh commitment, more precisely with the expression of classical T fh markers: PD-1, CXCR5, and Bcl6. A limited follicular positioning of neonatal pre-T fh cells is also supported by the decreased expression (expression level change of −1.92-fold in neonates compared to adults) of S1pr2, known to suppress CXC12/CXCL13-mediated migration, thus restricting premature egress of T fh cells out of GC (66). Thus, numerous transcription factors contribute to prevent the follicular positioning of neonatal T fh cells, depriving them from interacting with follicular dendritic cells (FDCs) and germinal center B (GC B) cells.
The expression of the Pou2af1 and Pou6f1 transcription factors was also reduced. Although their role in T fh cells remains to be fully investigated, Pou6f1 is expressed in early fate committed T fh cells (6) and Pou2af1 is highly expressed in early stage GC-T fh cells (10,67). A recent report by Stauss et al. (68) established that the Pou2af1 gene promotes Bcl6 expression and T fh cell development. A general reduction in CXCR5 expression was observed on Pou2af1 −/− CD4 + T cells as well as fewer GL7 + T fh cells in Pou2af1 −/− mice (68). Therefore, the POU family transcription factors seems to fine-tune T fh cell development and their reduced expression in neonates may contribute to the limited expression of CXCR5 and GL7 in neonatal T fh cells (4).
The lower expression of STAT4 (expression level change of −2.22-fold in neonates compared to adults) in neonates cements that the specific early life environment prevents the differentiation of pre-T fh cells toward committed GC-T fh cells: the IL-12-STAT4 pathway indeed contributes to the expression of key T fh -associated molecules, such as IL-21, CXCR5, and ICOS as well as multiple important transcription factors involved in T fh -cell generation, such as Bcl6, c-Maf, and Batf (69,70).
How does the TH2-like preferential polarization of neonatal effector T cells persist in T fh cells? Although, TH2 signature genes, including GATA3, IL-4, and IL-5, were not differentially expressed in neonatal T fh cells (Table S1), we observed significant changes in IL-13 and PPAR-γ. Nobs et al. recently showed that PPAR-γ expression in T cells controls the development of type-2 immunity (71). Therefore, the increased expression of PPAR-γ and of additional genes associated with TH2 polarization, including RXRA, ccr2 (72), il17rb (73)(74)(75), and cntnap1 (75), may all play a role in maintaining the default TH2-bias of neonatal T fh cells.
Semiquantitative RT-PCR analyses confirmed both the reduced transcript abundance of T fh -cell-associated signature genes in neonates (black bars) compared to adults (open bars) ( Figure 2C and Figure S2), and their preferential bias toward TH2, as shown by the higher levels of IL-13, PPAR-γ, RXRA, and CCR2 ( Figure 2D).

Adjuvantation With CpG 1826 Bypasses the Neonatal TH2-Bias of pre-T fh Cells and Supports Terminal GC-T fh Cell Differentiation
We and others have shown that administration of TT/AlOH supplemented with TLR9 agonist CpG 1826 enhanced neonatal antibody responses through the induction of higher T fh and GC B cell numbers (4), as observed in adult mice (76,77). We thus asked whether neonatal CpG 1826 adjuvantation induced transcriptional changes in the genes/factors identified as differing between neonatal and adult T fh cells. Semiquantitative RT-PCR on FACS-sorted CD4 + CXCR5 high PD-1 high T fh cells isolated 10 days after TT/AlOH + CpG 1826 immunization showed that CpG adjuvantation increased the transcriptional abundance of T fhcell specific signature genes in neonatal T fh cells ( Figure 2C and Figure S2). Flow cytometry analyses confirmed significantly lower CXCR5 expression by neonatal T fh cells (4) (Figure S3A). Remarkably, CpG adjuvantation significantly enhanced the expression of CXCR5 on neonatal T fh cells (Figure S3A), although not to adult like levels. The results support a follicular positioning of neonatal T fh cells, facilitating the T fh -GC B cell crosstalk required to provide B cell help during the GC reaction. In contrast, the expression of PD-1 was significantly decreased in 1 week-old and adult mice immunized with CpG-ODNs ( Figure S3B).
Notably, TH2-related genes (i.e., PPAR-γ, RXRA, IL-13, and CCR2) were significantly reduced in neonatal T fh cells, reaching similarly low levels as in adult mice immunized without CpG-ODNs ( Figure 2D). However, similar to our previous observation (4), IL-4 mRNA transcripts were significantly lower in neonatal Tfh cells and were not affected by CpG adjuvantation (Figure S2B), suggesting that IL-17 transcription would also remain unaffected as previously demonstrated by Debock et al. (5).
Thus, CpG adjuvantation may (1) abrogate the regulation of early life T fh cell differentiation exerted by the TH2-related genes PPAR-γ and RXRA, (2) facilitate the follicular positioning of neonatal T fh cells, as mirrored by increased levels of S1pr2 and CXCR5 and by reduced IL-7R, and (3) support neonatal T fh cell differentiation toward committed GC-T fh cells. This explains our previous observations demonstrating the increase of T fh cell numbers, of GL7 expression by neonatal T fh cells, of GC reactions and thus of Ab titers following neonatal CpG-ODNs adjuvantation (4) (Figures S3C,D).

Neonatal T fh Cells Preferentially Give Rise to Short-Lived Effector Cells
To better understand the fate of the pre-T fh cells elicited in early life, we next ran the Co-Expression Molecules identification Tool (CEMiTool) (49) on our data set. CEMiTool is an R package that provides in an automated manner unsupervised gene filtering, automated parameter selection for identifying modules, enrichment and module functional analyses as well as integration with interactome data (49). This modular expression analysis identified 6 different co-expression modules ( Figure 3A and Data Sheet S1), of which only module M2 was significantly enriched in neonates (Figure 3B). This module was enriched for genes related to cell cycle (49) (Figure 3C), suggesting the capacity of neonatal T fh cells to enter the cell cycle more rapidly than their adult counterparts. CEMITool also integrates co-expression analysis with protein-protein interaction data. Expression of important genes associated with cell-cycle progression, including gene encoding E2F1 and TK1, were identified as hubs in module M2 (Figure 3D). This early life characteristic was previously observed (78)(79)(80): neonatal T and B lymphocytes have the capacity to enter the cell cycle more quickly and thus efficiently mobilize responses from an otherwise completely naive population-possibly to compensate for the limitations in immune cell function in early life (i.e., lack of immunological memory) (78)(79)(80). Yet, rapid cycle entry only gives rise to short-lived effector cells (78)(79)(80).
To complete our observations, we performed a pathway analysis which revealed cell-intrinsic differences between neonatal and adult T fh cells (Figure 4): neonatal T fh cells were enriched in pathways associated with cell proliferation, apoptosis and key metabolic reactions, such as glycolysis, considered to play an important role in T cell activation and differentiation, while adult cells were enriched in mitogenactivated protein kinase (MAPK)-signaling pathways, thus outlining age-associated differences in the maturity and basic function. Interestingly, enrichment of Hedgehog signaling, which predispose T cell differentiation toward the TH2 pathway (81), further supports the TH2-bias of neonatal T fh cells. Altogether, these transcriptional analyses of neonatal vs. adult T fh cells reveal the existence of multiple coordinated regulatory mechanisms resulting into the preferential differentiation of neonatal CD4 + T cells toward innate, short-lived pre-T fh effectors rather than adaptative (GC-derived) immunity defense mechanisms.

DISCUSSION
We previously identified the induction of T fh cells as limiting early life GC and Ab responses elicited by vaccines including aluminum-based adjuvants (4). We now demonstrate that the few T fh cells elicited in early life retain a preferential bias toward TH2, strongly expressing IL-13, and PPAR-γ and RXRA  which negatively regulate T fh cell differentiation (41,60), and that numerous transcription factors contribute to restrict activated neonatal CD4 + T cells at a pre-T fh cell stage of short-lived effectors favoring innate rather than GC-associated adaptive responses. Importantly, we show that this fate is not inevitable as adjuvantation with CpG-ODNs reduced the expression of TH2-related genes and sufficiently strengthened the T fh cell-associated signature molecules to drive the GC-T fh differentiation program to its completion and fine tune the GC reaction.
Following immune challenges, neonatal responses are often weak (82). This has been associated with a propensity of neonatal T cells to give rise to short-lived effector cells (78)(79)(80) and to produce elevated levels of TH2-type cytokines compared to adults (83)(84)(85). We show that these two key neonatal characteristics persist during T fh cell differentiation, lodging the cells in a pre-T fh stage characterized by a TH2 bias. Our results suggest that a delicate balance of several signals known to promote TH2 development may contribute in maintaining an optimal environment for the TH2-biased T fh cell differentiation in neonates, including increased expression of PPAR-γ (71,86), RXRA (87), ccr2 (72), il17rb (73)(74)(75), cntnap1 (75), Hedgehog signaling (81), and lower levels of c-maf mRNA transcripts. A critical role for c-maf in limiting TH2 responses and in driving T fh cell development was recently unveiled by Andris et al. (18). Further investigations are warranted to delineate whether Tal1 may also play a fundamental role in the generation and persistence of TH2-biased T fh cells in neonates.
A limitation of our study is that the very few T fh cells induced in early life (about 2 × 10 4 T fh cells from a pool of 8 neonates) and thus the small amount of recovered RNA precluded the analysis of all potentially interesting genes. As the microarray did not reveal significant changes in TH17-related genes, in GATA3 or in IL-5 expression in neonatal T fh vs. non-T fh cells (Table S1), we did not compare their mRNA transcript levels to those of adult cells. In our model, both IFN-γ and IL-4 mRNA transcript levels are significantly lower in neonatal T fh cellsand not affected by CpG adjuvantation (Figure S2B). The similar expression of Foxp3 in neonatal and adult T fh cells was confirmed by semiquantitative RT-PCR ( Figure S2D). Altogether, these results suggest that in our model, neonatal T fh cells do not exhibit a bias toward TH1, TH17 nor Treg cells. Our attempts to develop validated assays to reliably measure several proteins in the few recovered neonatal T fh cells did not succeed, and such proteins remained below detection levels when assessed in lymph nodes homogenates (not shown).
PPAR-γ and RXRA were identified as critical hub genes in neonates. Therefore, besides their role in maintaining the overall TH2 bias, PPAR-γ and RXRA may also negatively regulate T fh cell differentiation. PPAR-γ is known to (71,86) promote Tregs survival (88)(89)(90) and inhibit the formation of T fh cells and GC reactions via the regulation of Bcl6 and IL-21 (41). That inhibition of Bcl6 expression is illustrated in neonatal T fh cells by increased expression of the TH2-related gene, IL-13, previously identified as one of the most repressed Bcl6-target gene (65). A role for Tgif1-RXR interaction in the establishment or inhibition of a chronically elevated T fh cell population was recently computationally predicted and demonstrated by Leber et al. (60); an increase with Tgif1 was associated with an increase in the T fh response, while an increase in RXR was more closely correlated with the T fh decline phase (60). Small changes in RXRA, such as 10% change in expression were previously demonstrated to result in a 50% change in activity and significant alteration of downstream transcriptional targets (91). We conclude that the differential expression of TH2-related genes PPAR-γ and RXRA might be involved in the distinct genetic programming of neonatal and adult T fh cells.
Although the T fh cell program is well-initiated in neonates, the gene-expression pattern of neonatal T fh cells underrepresented that of adult T fh cells, suggesting that T-B interactions fail to elicit appropriate signals and provide efficient help to neonatal pre-T fh cells to further differentiate into committed GC-T fh cells. Indeed, T fh cells differentiation involves a multisignal process that includes expression of CXCR5, IL-21, Bcl6, TBK1, STAT4, Ascl2, and c-maf which were all expressed to lower levels as compared to adult T fh cells. Remarkably, adjuvantation with CpG-ODNs, skewed neonatal pre-T fh cells toward committed GC-T fh cells, as illustrated by increased expression of T fh -signature genes ( Figure 2C and Figure S2). In parallel, genes associated with follicular positioning of T fh cells were increased (i.e., s1pr2, Ascl2, and CXCR5), facilitating the cognate T fh -B cell interactions for completing T fh cell differentiation (92)(93)(94)(95)(96), with concomitant increase in Bcl6 and IL-21 expression (10,95). Ascl2 directly regulates the localization of T fh cells via CXCR5 expression and suppression of CCR7 and PSGL1 (9). CXCR5 allows T fh cells to migrate into the B cell follicles and form stable contacts with antigen-primed B cells (92,97). These results indicate that a combination of several T fh -specific signals, in addition to previously described environment factors and CD4 + T cells intrinsic determinants (4), maintain a favorable environment for TH2-biased T fh cell differentiation, restricting neonatal CD4 + T cells at a pre-T fh stage of short-lived effector cells. Adjuvantation with CpG-ODNs is sufficient to counteract the TH2-biased response of neonatal T fh cells, reducing TH2-related genes to adultlike levels, while T fh signature genes (i.e., Bcl6, CXCR5, IL-21, Ascl2, C-maf, Pou6f1, s1pr2, Batf, CXCR4, and TBK1) are progressively enhanced, resulting in differentiated and GC-committed T fh cells. As illustrated in Figures 2C,D, the switch from pre-T fh to mature T fh cells involves changes in the expression levels of several factors-such that "classical" mechanistic approaches including knockout/knock-in mice were not attempted.
We recently demonstrated that adjuvantation of a vaccine with a liposome including a C-type lectin receptor agonist was able to elicit potent GC reactions in neonates after a single dose (98). Altogether, these results show that immune deficiencies seen in early life can be overcame by providing the right signals, and are in accordance with the current understanding that the neonatal immune system is not deficient but tightly regulated to best adapt to the unique challenge of a rapidly required adaptation from a sterile to a microbial environment (99).
Further studies are necessary to investigate whether abrogating the TH2 bias of T fh cells in early life is critical for the full commitment of T fh cell differentiation and the subsequent GC B cell and antibody responses, resulting in effective responses to vaccination in early life.

DATA AVAILABILITY
The data has been deposited at the Gene Expression Omnibus repository-accession number is GSE126843.

ETHICS STATEMENT
This study was carried out in accordance with the recommendations of the Geneva Veterinary Office and conducted under relevant Swiss and European guidelines. The protocol was approved by the Geneva Veterinary Office.

AUTHOR CONTRIBUTIONS
BM-G, P-HL, and C-AS contributed to formulation of theory and prediction. BM-G, P-HL, and C-AS designed the research. BM-G and MV performed the experiments and analyzed and/or interpreted the data. BM-G, MV, and C-AS wrote the manuscript. PG-D, FF, LC, and HN performed the microarray analysis and critically revised the manuscript. All authors reviewed the manuscript.

ACKNOWLEDGMENTS
We thank Paola Fontannaz, Stéphane Grillet, and Chantal Tougne for their key contribution to the complex experimental work required by this study; Anthony Joubin for excellent assistance with animal care; Anne Rochat for running the realtime PCR; Mylène Docquier and the Genomics Platform for assistance with the real-time PCR; Jutta Kollet for assistance with the Microarray (Miltenyi Biotech GmbH Bioinformatics); Jean-Pierre Aubry-Lachainaye and Cécile Gameiro from the flow cytometry platform.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.01845/full#supplementary-material Figure S1 | Unsupervised hierarchical analysis groups together T fh cells from both age groups. Hierarchical clustering of CD4 + CXCR5 high PD-1 high T fh cells and respective controls CD4 + CXCR5 − PD-1 − T cell samples was obtained by pvclust R package. CD4 + CXCR5 high PD-1 high T fh cells from young (purple) and adult (orange) mice and their respective controls (light colors) formed distinct groups. Figure S2 | CpG adjuvantation is sufficient to increase Bcl6 targeted T fh specific genes, such as Atp1a3 as well as T fh cell-related genes, including BATF, TBK1, and CXCR4. One week-old and adult C57BL/6 mice (5-8 mice/group) were immunized i.m. with TT/AlOH. Ten days post-vaccination the draining LNs were collected to simultaneously isolate CD4 + CXCR5 high PD-1 high T fh cells and CD4 + CXCR5 − PD-1 − T (non-T fh ) cells by flow-cytometry cell sorting. The cells obtained from the two inguinal draining LNs of either 16 neonates/group or 5 adults/group per experiment were pooled before sorting to recover a sufficient number of cells for experimentation. Semi-quantitative RT-PCR analysis of selected T fh cell-related genes (A), IL-4 (B), TH1 cell-related genes (C), Foxp3 (D), or Bcl6 targeted T fh specific genes (E) in sorted cells, normalized to results obtained for the control genes (EEF1, GusB, RPS9). The graph display mean ± SEM. Cumulative data from adult [TT/AlOH (n = 10)] and 1 week-old [TT/AlOH (n = 64), TT/AlOH+ CpG 1826 (n = 32)] mice from at least two independent experiments. Fold changes are shown relative to 1 week-old mice immunized with TT/AlOH. Statistical analysis was performed with Prism software (Version 7, GraphPad), using unpaired t-test. * P < 0.05, * * P < 0.01, * * * P < 0.001, * * * * P < 0.0001.  Table S1 | Summary table of TH1, TH2, TH17, and Treg selected genes. Indicated are the p-values, Median intensity values of all samples, the flag counts and the ProbeID. Fold-change values (columns C-E) which did not pass the selection criteria (Anova P-value ≤ 0.05, Tukey P-value ≤ 0.05, fold-change ≥2 or ≤-2, and reliable detection of the signal (Flag counts) ≤1 in the group with higher expression) are in black color, while up-regulated ones are shown in red and down-regulated ones are shown in green.
Data Sheet s1 | CEMiTool output html file for all modules.