TLR1/2 orchestrate human plasmacytoid predendritic cell response to gram+ bacteria

Gram+ infections are worldwide life-threatening diseases in which the pathological role of type I interferon (IFN) has been highlighted. Plasmacytoid predendritic cells (pDCs) produce high amounts of type I IFN following viral sensing. Despite studies suggesting that pDCs respond to bacteria, the mechanisms underlying bacterial sensing in pDCs are unknown. We show here that human primary pDCs express toll-like receptor 1 (TLR1) and 2 (TLR2) and respond to bacterial lipoproteins. We demonstrated that pDCs differentially respond to gram+ bacteria through the TLR1/2 pathway. Notably, up-regulation of costimulatory molecules and pro-inflammatory cytokines was TLR1 dependent, whereas type I IFN secretion was TLR2 dependent. Mechanistically, we demonstrated that these differences relied on diverse signaling pathways activated by TLR1/2. MAPK and NF-κB pathways were engaged by TLR1, whereas the Phosphoinositide 3-kinase (PI3K) pathway was activated by TLR2. This dichotomy was reflected in a different role of TLR2 and TLR1 in pDC priming of naïve cluster of differentiation 4+ (CD4+) T cells, and T helper (Th) cell differentiation. This work provides the rationale to explore and target pDCs in bacterial infection.


Introduction
Tuberculosis (TB) and multidrug-resistant bacteria are a major concern for worldwide health [1]. In TB and gram+ infection, type I interferon (IFN) has been shown to play a pathological role [2,3]. Plasmacytoid pre-dendritic cells (pDCs) are known to produce high amounts of type I IFN in response to viral sensing [4]. It is reported that pDCs are able to respond to gram+ bacteria [5], can be recruited at the site of the infection, and are enriched in TB lymph nodes [6].
Gram+ bacteria express lipoproteins on their surface membrane, which play an important role in their survival and pathogenicity [7]. Bacterial lipoproteins are recognized by toll-like receptor (TLR)1/2 and induce activation and maturation in dendritic cells (DCs) [8]. TLR2 knockout mice are more susceptible to mycobacterial infection, but Mycobacterium tuberculosis is able to hijack TLR2 signaling to enhance its survival in the host [9]. TLR1  expression in human pDCs has not been reported [10], leading to the conclusion that TLR 1 and 2 do not have a functional role in pDCs. Human pDCs express mainly TLR7 and TLR9, localized in the endosomes and capable of sensing nucleic acids [10][11][12]. pDCs also express a range of cytosolic sensors, either at steady state, such as the helicases DEAH box protein 9 (DHX9) and DHX36 [13], or following innate activation, such as retinoic acid inducible gene 1 (RIG-I) [14]. However, how pDCs sense gram+ bacteria is still debated, and their role in gram+ infections is still poorly investigated [15].
Here, using human primary cells, we provide definite evidence that pDCs sense gram+ bacteria through TLR1 and TLR2.

Human pDCs respond to bacterial lipoproteins through TLR1 and TLR2
In order to investigate how pDCs sense gram+ bacteria, we screened steady-state blood pDCs for TLR mRNA expression. In addition to the known expression of TLR7 and TLR9, we detected low levels of TLR1, TLR2, TLR6, and TLR10 (S1A Fig). Among the TLRs expressed by pDCs, TLR1 and TLR2 mediate bacterial sensing by binding lipoproteins [8].
We measured pDC TLR1 and TLR2 mRNA expression on freshly isolated blood pDC and following stimulation with PAM3CSK4 (PAM3), a bacterial lipoprotein used as a prototypical TLR1/2 ligand. HeLa cells were used as negative control and CD11c + DCs as positive control for the expression of TLR1/2. pDCs maintained a stable TLR1 mRNA expression following stimulation. PAM3 activation increased TLR2 expression as compared with ex vivo pDCs ( Fig 1A).
We further investigated whether pDCs express TLR1 and TLR2 at the protein level. Using flow cytometry, we confirmed in freshly isolated peripheral blood mononuclear cells (PBMCs) that pDCs expressed TLR1 and TLR2 at their surface, as compared to isotype control ( Fig 1B  and quantification in S1B Fig).
A feature of pDCs is high type I IFN secretion. The ability of PAM3 to induce type I IFN secretion in human pDCs has been questioned [17,18]. Here, highly pure (99%) pDCs responded to 1 and 10 μg/ml of PAM3 by secreting type I IFN (Fig 1D). In addition, PAM3 induced the secretion of pro-inflammatory cytokines (interleukin [IL]-6, tumor necrosis factor [TNF]-α), and chemokines (IL-8, IP-10), although to a lower extent than with FLU ( Fig 1D). Furthermore, pDCs secreted Granzyme B (GZMB) in response to bacterial lipoprotein stimulation ( Fig 1D). Both 1 and 10 μg/ml of PAM3-induced pDCs the expression of costimulatory molecules and cytokine secretion at comparable levels ( Fig 1C and 1D).
We used PAM3 to stimulate pDCs purified from tonsils, a site of frequent encounter with gram+ bacteria. Tonsillar pDCs up-regulated surface costimulatory molecules (CD86, CD80, RT-PCR quantification of TLR1 and TLR2 expression from total mRNA of sorted human blood pDCs before and after 1-hour activation with PAM3 as compared to CD11c + DCs and HeLa cells. Results were normalized on 3 housekeeping genes. Results include 5 donors. (B) pDCs and CD11c + DCs were stained in freshly isolated PBMCs with anti-TLR1 and anti-TLR2 antibody (dark gray), respective cognate isotype (light gray). (C-D) Sorted human pDCs were cultured during 24 hours with medium (Ø), 0.1 μg/mL LPS, 1 and 10 μg/mL PAM3, 100 ng/mL GM, or 82 HA/ml FLU. (C) Specific MFI for surface expression of costimulatory or coinhibitory molecules from activated pDCs by FACS. Results include the mean of 9 donors. (D) Cytokine secretion by pDCs. Results include the mean of 17 donors. Each dot represents a donor. � p < 0.05; �� p < 0.01; ��� p < 0.001 (Wilcoxon test). Underlying data for this CD40, and PDL1) (S1E Fig) and MHC-II complex in line with our data on blood pDCs (S1D Fig).
These data suggest that pDCs from both blood and from physiological bacterial interfaces functionally respond to bacterial lipoproteins.

TLR1/2 pathway is necessary for pDC response to gram+ bacteria
We next questioned whether, in addition to purified lipoproteins, pDCs could respond to whole gram+ bacteria. Although pDC activation by Staphylococcus aureus was reported [5], whether pDCs can respond to M. tuberculosis is still debated [6]. Sorted blood pDCs were stimulated with 3 different heat-killed gram+ bacteria relevant to human infections: M. tuberculosis, S. aureus, and Listeria monocytogenes. We observed up-regulation of CD80 and CD86 following pDC culture with heat-killed bacteria (Fig 2A). To establish the role of TLR1/2, we took advantage of a chemical antagonist for both TLR1 and 2, CU-CPT22 [19]. CU-CPT22 did not affect unstimulated pDCs, nor did it impact costimulatory molecule expression (CD80, CD86) or type I IFN secretion in FLU-activated pDCs (S2A- S2C Fig). On the contrary, CU-CPT22 treatment strongly decreased bacteria-induced CD80 and CD86 expression by pDCs (Fig 2A and S2D Fig). Furthermore, gram+-stimulated pDCs secrete high amount of type I IFN ( Fig 2B) thus indicating full activation of pDCs by bacteria ( Fig 2B). TLR1/2 blocking by CU-CPT22 almost completely abrogated type I IFN production ( Fig 2B). Therefore, pDCs responded to whole gram+ bacteria in a TLR1/2-dependent manner.
CD11c + DCs are known to express TLR1/2 and to be able to induce Th cell differentiation. We investigated the differences in naïve CD4 + T-cell priming by PAM3-activated CD11c + DCs and pDCs (S2E Fig). T cells primed with PAM3-activated CD11c + DC or pDCs showed a comparable state of activation. However, pDCs induced a prominent Th2-like profile compared with CD11c + DCs (higher secretion of IL-4, IL-5, and IL-10), suggesting different contributions to immune regulation in the context of bacterial infection (S2E Fig).
To establish whether TLR1/2-activated pDCs were able to induce cytokine production by memory T cells, we cultured PAM3-activated pDCs with allogeneic memory CD4 + T cells from healthy donor peripheral blood. Memory CD4 + T cells secreted significant amounts of IFN-γ, IL-10, IL-3, IL-4, and IL-9 when cocultured with PAM3-activated pDCs compared with memory CD4 T cells cocultured with untreated pDCs (S3A Fig). The amounts of these cytokines were comparable to FLU condition and much higher than the negative control LPS. Moreover, PAM3-activated pDCs were the only ones capable of inducing the production of both IL-17A figure can be found in S1 Data. AU, arbritrary unit; CD, cluster of differentiation; DC, dendritric cell; FACS, fluorescence-activated cell sorting; FLU, influenza virus; GM, GM-CSF; Gm-CSF, granulocyte-macrophage colony-stimulating factor; GZMB, Granzyme B; ICOSL, inducible T cell costimulator ligand; HA, hemagglutinin; IFN, interferon; IL, interleukin; IP-10, Interferon gamma-induced protein 10; LPS, lipopolysaccharide; PBMC, peripheral blood mononuclear cell; PAM3, PAM3CSK4; pDC, Plasmacytoid predendritric cell; PDL1, programmed cell death ligand 1; RT, real time; TLR, toll-like receptor; TNF, tumor necrosis factor.  and IL-17F from memory CD4 + T cells as compared with untreated pDCs and FLU-pDCs. This shows that PAM3-activated pDCs are capable of inducing effector cytokine production by memory CD4 + T cells, including IL-17A and F, important in epithelial immunity.
Recent results demonstrated the existence of a rare DC subset defined as DC5 or AXL + SI-GLEC6 + (AS-DC) [21]. This subset is characterized by the expression of the surface markers CD2, CD5, and AXL receptor tyrosine kinase (AXL) but also shares some markers with pDCs, leading to potential contamination of the pDC population. In order to determine whether pure pDCs (DC5-depleted pDCs) were able to induce T-cell expansion and Th polarization to the same extent as LIN -CD4 + CD11c -pDCs, we cell sorted pure pDCs following the presented gating strategy (S3B Fig). CD2 -CD5 -AXL -pDCs were activated for 24-hours with PAM3, FLU, LPS, or GM-CSF and cocultured with allogeneic naïve CD4 + T cells from healthy peripheral blood. We found that TLR1-activated pure pDCs were capable of inducing CD4 Tcell expansion and Th cell differentiation (S3C Fig), with increased production of IFN-γ, IL-10, IL-3, IL-4, IL-9, and GM-CSF as compared with nontreated pDCs. These results show that CD4 + T-cell expansion and Th cell differentiation induced by TLR1-activated pDCs is not due to contamination with DC5.

TLR1 and TLR2 play a differential role in the pDCs response to bacterial lipoproteins
In order to investigate the differential contribution of TLR1 and TLR2 in mediating pDC response to bacterial lipoproteins, we separately blocked the 2 receptors with specific antibodies, as compared with matched isotype controls [22,23]. TLR1 functional blocking significantly reduced costimulatory molecule expression (CD80, CD86, and ICOSL), whereas TLR2 blockade did not (Fig 3A and S4A Fig). TLR1 blocking almost completely abolished secretion of the pro-inflammatory cytokines IL-6 and TNF-α ( Fig 3B). Conversely, TLR2 blocking inhibited type I IFN secretion, which was not impacted by TLR1 blocking (Fig 3B). Combined TLR1 and TLR2 blockade, as well as the TLR1/2 competitive antagonist CU-CPT22, inhibited both costimulatory molecule expression and cytokine release (Fig 3A and 3B). These results suggest a differential control of pDC functions by TLR1 and TLR2.
Next, we performed coculture experiments with PAM3-treated pDCs and naive CD4 + T cells, with and without TLR1 or TLR2 blocking antibodies. TLR1 blocking during PAM3 activation reduced T-cell expansion and proliferation (S4B and S4C Fig). Following polyclonal restimulation, we did not detect any difference in the Th1 prototypical cytokine IFN-γ ( Fig 3C  and S4E Fig). However, TLR1 blocking in pDCs decreased prototypical Th2 cytokines (IL-13, IL-4, IL-5) (Fig 3C and S4E Fig). TLR1 blocking also diminished IL-10 production by Th cells, suggesting a decrease in Treg generation (Fig 3C). We found that TLR1 blocking reduced IL-9 secretion by Th cells (Fig 3C). After 4 days of pDCs-T cell coculture, we performed intracellular staining for Th master regulator transcription factors to better characterize the Th subsets induced. TLR1/TLR2 blocking did not reduce Tbet induction (Fig 3D and S4D Fig), in line with our observation on IFN-γ production. However, TLR1 blocking diminished GATA3 and FOXP3 expression (Fig 3D and S4D Fig), in line with its impact on Th2 and Treg polarization. TLR1 blocking also reduced BCL-6 expression (Fig 3D and S4D Fig), involved in T-follicular helper (Tfh) generation [24].

TLR1 and TLR2 activate different signaling pathways in response to bacterial lipoproteins
In pDCs, MAPK and NF-κB pathway activation leads to costimulatory expression and proinflammatory cytokine release, whereas PI3K signaling controls Type I IFN induction [25]. In Results include the mean of 10 independent donors. Each dot is an independent donor. (C-D) The 24-hour-stimulated pDCs were cocultured with the case of TLR7 and TLR9, these 2 signaling pathways are activated in early and late endosomes, respectively [26]. We performed phospho-fluorescence-activated cell sorting (phospho-FACS) to investigate which pathways were activated by bacterial lipoproteins in pDCs. Stimulation with PAM3 (1 and 10 μg/mL) led to p38, p65, and AKT serine/threonine kinase AKT phosphorylation as compared with untreated pDCs (Fig 4A). pDC stimulation with FLU virus was used as positive control (Fig 4A). These results suggested that MAPK, NF-κB, and PI3K were activated following bacterial lipoproteins activation.
These data suggest that the mechanism behind the differences observed in pDCs innate versus adaptive responses following TLR1 and TLR2 blocking is related to different signaling pathways controlled by the 2 receptors.

Discussion
pDCs are known to express a narrow TLR pattern that is restricted to TLR7 and TLR9 [10]. Accordingly, TLR1 and TLR2 expression was considered a prototypical feature of myeloid cells and absent from pDCs [10]. The low expression level of TLR1/2 on pDCs as compared with TLR7 and 9 may have previously suggested that it is not functionally relevant. However, peripheral blood pDCs are considered the major source of type I IFN following S. aureus stimulation [15]. We found that pDCs express TLR1 at steady state and TLR2 in a stimulationdependent manner, and that those 2 TLRs are functional for PAM3 sensing.
Commensal bacteria have an immunomodulatory impact in the gut. Some of them, such as Bacteroides fragilis and Clostridia, are gram+ [27,28]. Here, we show that pDCs respond to the lipoprotein characteristic of gram+ bacteria and that lipoprotein-activated pDCs induced IL-10 and FOXP3 expression in CD4 + T cells. pDCs are present in the human gut at steady state [29]. However, other groups report that pDCs can participate in sustaining inflammation in acute colitis [30]. Our study suggests that pDCs, following bacterial sensing, could instruct CD4 + T cells in the gut and promote a mixed Th cell cytokine profile-including a regulatory phenotype-but also cytokines prototypical of Th1, Th2, and Th17 inflammation. Therefore, a detailed investigation of pDC role in the gut is warranted. Our results provide a strong basis for a functional link between pDCs and gram+ bacteria in various physiopathological contexts.
Our data show that GZMB can be induced by bacterial lipoproteins. It has been shown that pDCs in TB patients' lymph nodes produce GZMB [6]. Our data suggest that bacterial sensing through TLR1/2 could induce GZMB in pathological conditions, such as TB infection.
In TB, a diversity of Th responses has been observed [31]. It has been proposed that Th1 is the protective response in TB and in many gram+ bacterial infections, whereas Th2 and Treg have been shown to promote the disease [31][32][33][34]. In atopic dermatitis, in which there is a strong link between disease flare and S. aureus skin infection [35], it has been shown that both Th1 and Th2 responses coexist [36]. In vitro, we showed that gram+ stimulation of pDCs induced a mixed Th1, Th2, and Treg cytokine profile, suggesting that they could contribute to the in vivo-observed Th diversity.
Our results showed that TLR1 and TLR2 play a different and complimentary role in the pDC response to bacterial lipoproteins. Although it is reported that TLR2 in inflammatory monocytes can be endocytosed and activate IRF7 in response to a viral ligand [37], our results are the first to link a type I IFN response and the TLR2 pathway in response to a bacterial ligand.
Furthermore, we observed that TLR1 and TLR2 blocking on pDCs had differential effects on Th cytokine secretion. TLR1 blocking on pDCs decreased T-cell polarization toward Th2, Treg, and Tfh but not Th1 cells. Conversely, TLR2 blocking showed a specific inhibition of Type I IFN secretion without impacting T-cell polarization. These data show that TLR1 activation could promote an adaptive response (costimulatory molecule expression, pro-inflammatory cytokine secretion, Th proliferation and polarization), whereas TLR2 activation induced type I IFN, which broadly functions in innate immunity. Our data suggest that the innate and/ or adaptive response of pDCs could be differentially targeted.
These data suggest that the mechanism behind the differences observed in pDCs' innate versus adaptive responses following TLR1 and TLR2 blocking is related to different signaling pathways controlled by the 2 receptors.
Our findings open broad perspectives on the possible role of pDCs in gram+ bacterial diseases. Here, we showed that M. tuberculosis, S. aureus, and L. monocytogenes induced high levels of type I IFN production by pDC and that this is abrogated by TLR1/2 antagonist (CU-CPT22). Type I IFN is highly expressed in TB, in which it has been proposed to dampen immune response [38]. Therefore, our data establish pDCs as a possible source of type I IFN in TB-infected tissues. Furthermore, TLR1 polymorphisms are associated with TB susceptibility [39,40]. Future studies are required to establish whether pDCs could represent a pharmacological target in TB. In the past few years, different attempts to develop a vaccine direct against of S. aureus have failed [41]. Subsequently, lipoproteins have been considered promising candidates [7]. Besides, vaccines in combination with TLR7 ligand show a boost in the protective immunity [42]. Our results suggest the possible role of pDCs in vaccine efficacy considering their capacity to respond to lipoproteins, high TLR7 expression, and capacity to prime T cells in response to gram+ bacteria.

Blood samples and cell isolation
PBMCs were isolated by Ficoll density gradient centrifugation (Ficoll-Paque, GE Healthcare, Chicago, IL). pDCs and CD11c + DCs were isolated by a first step of total DC enrichment (EasySep human Pan-DC Enrichment kit, Stemcell, Canada) followed by FACS sorting as Lineage − CD11c − CD4 + to a 99% purity [20]. Tonsil pDCs were isolated using the following protocol by Durand and Segura [43]. DC5 -pDCs was isolated by a first step of total DC enrichment (EasySep human Pan-DC Enrichment kit, Stemcell, Canada) followed by FACS sorting as Lineage − CD11c − CD4 + CD2 − CD5 − AXL − to 99% purity. Human naive CD4 + T cells were isolated from PBMCs by negative selection (naïve CD4 T-cell isolation kit, Miltenyi, Germany) to a >98% purity. Total Memory CD4 + T cells were isolated from PBMCs by negative selection (Memory CD4 + T Cell isolation Kit and LS columns, Miltenyi, Germany).