To obtain insights on the role of ATF6α in DC homeostasis, we generated conditional knock-out animals lacking ATF6α in CD11c-expressing cells. To this end, we crossed the Itgax-Cre mice line with Atf6 ^fl/fl mice (referred to as “ATF6α^ΔDC mice”) (Engin et al., 2013). These animals delete exons 8-9 of Atf6 in CD11c-expressing cells (which fully targets cDC1s and cDC2s, while partially targeting pDCs and monocyte/macrophage subsets (Abram et al., 2014)). ATF6α^ΔDC mice are compared to control animals (Atf6 ^fl/fl littermates with no expression of Cre, scheme depicted in Figure 1A). Spleen cDCs were isolated from ATF6α^ΔDC mice and control animals and expression of ATF6α was quantified by qPCR (Figure 1B). As expected, cDC1s and cDC2s from ATF6α^ΔDC mice do not express Atf6 mRNA and maintain normal expression of Atf6b, validating the model of study. Next, we analyzed the composition of cDC1s and cDC2s and observed that ATF6α^ΔDC mice display unaltered frequencies of these subtypes (Figures 1C, D). Furthermore, cellular composition analysis of ATF6α^ΔDC mice in spleen revealed that these animals have normal composition of immune cell types (Figure 1E, gating analysis in Supplementary Figure S1).

To assess whether ATF6α deficient cDCs undergo normal differentiation/activation, we quantified expression of the costimulatory molecules PD-L1, CD86 and CD40, which are surface immunoregulatory molecules that allow cDCs to restrain or activate T cells, respectively (Kapsenberg, 2003; Hubo et al., 2013). Data in Figure 1F show that ATF6α deficiency does not alter surface expression of these proteins in cDCs. In conclusion, ATF6α does not regulate the differentiation/activation program of steady state cDCs in spleen.

Considering that cDCs in lymphoid organs are not equivalent to counterparts exposed to inflammatory stimuli in non-lymphoid tissues, we analyzed cDCs from the small intestine lamina propria (SiLP) and liver of ATF6α^ΔDC mice (Figure 2, gating analysis Supplementary Figures S1A, B). We verified that archetypical immune cell types were present in normal frequencies in the SiLP of ATF6α^ΔDC mice (Figure 2A). In the SiLP, bona-fide cDCs are divided in cDC1s (defined as CD103⁺CD11b⁻) and two subsets of cDC2s (CD103⁺CD11b⁺ and CD103⁻CD11b⁺) (Sun et al., 2020). Analysis of ATF6α^ΔDC mice show normal frequencies of the three cDC subtypes at the SiLP (Figures 2B–C). Liver tissue analysis show similar results, normal immune cell composition (Figure 2D, gating analysis in Supplementary Figure S2B) and comparable cDC frequencies between ATF6α deficient and control counterparts (Figures 2E, F). Altogether, these data indicate that ATF6α loss does not impair cDC homeostasis in tissues.

The Xbp1 gene is a transcriptional ATF6α target that contains an ER stress response element (ERSE) consensus sequence on its promoter region (Yoshida et al., 2000). Phenotypically, XBP1s deficiency in DCs leads to a marked reduction in cDC1 frequencies in the lung (Tavernier et al., 2017). The interplay between ATF6α and XBP1s led us to hypothesize that ATF6α^ΔDC mice may recapitulate the loss of lung cDC1s observed in XBP1 conditional deficient mice. To test this hypothesis, we generated conditional knock-out animals lacking XBP1s in CD11c-expressing cells by crossing the Itgax-Cre mice line with Xbp1 ^fl/fl mice (Lee et al., 2008) (referred to as “XBP1^ΔDC mice”). Lung cDCs from XBP1^ΔDC mice, ATF6α^ΔDC mice and control littermates were analyzed by flow cytometry (Figures 3A–D). We observed that ATF6α^ΔDC mice display comparable lung cDC1 percentages with control counterparts (Figures 3A, B). However, these observations were not recapitulated in XBP1^ΔDC mice, which revealed an evident loss of cDC1s compared to control littermates, confirming previous findings (Tavernier et al., 2017) (Figures 3C, D). These data indicate that despite reported evidence demonstrating transcriptional regulation of Xbp1 by ATF6α (Yoshida et al., 2001), the in vivo functional outcomes of these transcription factors in tissue DCs do not overlap. To explain these results, we investigated if ATF6α deficient DCs express altered levels of Xbp1 mRNA. PCR analysis of sorted splenic cDC subsets revealed that ATF6α-deficient cDC1s and cDC2s express normal levels of Xbp1u and Xbp1s (Figure 3E), indicating that ATF6α deficiency does not alter XBP1 expression in tissue cDCs.

We also quantified the composition of additional lung immune cells in ATF6α^ΔDC mice. Interestingly, analysis revealed a significant reduction in the frequencies of lung monocytes compared to control mice (Figures 3F, G). Notably, this reduction was not observed in XBP1^ΔDC mice (Figures 3H, I), suggesting that ATF6α and XBP1s regulate the fate of myeloid cells by independent mechanisms. These findings prompted us to investigate whether wild-type lung monocytes show signs of ATF6α transcriptional activity in steady state. To this end, we quantified expression of Atf6 mRNA and the ATF6α targets Hspa5 (BiP) and Grp94 by qPCR in isolated lung monocytes from control animals, and transcript levels were compared with those measured in lung T cells (Figure 3M, grey bars). Data indicated that lung monocytes tend to express higher levels of Atf6, Hspa5 and Grp94 than T cells isolated from the same tissue, suggesting that the former cell type show signs of ATF6α transcriptional activity in the steady state lung.

Finally, to evaluate if the remaining population of lung monocytes from ATF6α^ΔDC mice display signs of cellular dysregulation, we measured expression of canonical surface molecules and ATF6α targets by flow cytometry and qPCR, respectively. These cells express normal levels CD11c, MHC-II, CD64 and F4/80 (Figure 3J–L), and were CD11c^-/int MHC-II^-/lo/+ CD64^lo F4/80^lo, in line with the definition of Ly6C⁺ pulmonary monocytes (Gibbings et al., 2017). In addition, lung monocytes from ATF6α^ΔDC mice expressed similar levels of Atf6, Hspa5 and Grp94 than control animals, suggesting that the remanent monocyte population from ATF6α^ΔDC mice are not targeted by Cre-mediated recombination (Figure 3M). To sum up, these data suggest that lung monocytes show signs of basal ATF6α transcriptional activity and that ATF6α loss in CD11c⁺ cells result in partial reduction of lung monocyte frequencies.

Finally, given that ATF6α does not control expression of Xbp1u/s in cDCs (Figure 3E), we sought to investigate whether ATF6α could regulate expression of additional UPR components in these cells. To this end, target genes of the ATF6α, PERK and IRE1 (XBP1s and RIDD targets) branches were quantified by qPCR in cDC1s and cDC2s isolated from spleen of ATF6α^ΔDC and ATF6α^WT mice (Figure 4). Data show that ATF6α deficient cDC1s express a trend towards reduced Hspa5 expression, which did not reach statistical significance (Figure 4A). Furthermore, ATF6α deficient cDCs express unaltered levels of the XBP1s targets Erdj4 and Edem1 (Figure 4B), the RIDD substrates Cd18 and Bloc1s1 (Figure 4C) and the PERK targets Chop, Atf4 and Gadd34 (Figure 4D). Notably, cDCs from ATF6α^ΔDC mice also expressed normal levels of the reported ATF6 targets HerpUD and Grp94 (Figure 4E). These data show that cDCs do not constitutively activate the ATF6α branch in steady state. Furthermore, these observations indicate that in absence of canonical ER stress, ATF6α does not regulate expression of UPR components in cDCs in vivo.

Finally, we sought to evaluate if ATF6α regulates the acquisition of immunogenic features in contexts of DC activation. To this end, we studied bone-marrow derived DCs cultured in presence of the differentiation factor GM-CSF (referred to as ‘GM-DCs’). We corroborate that GM-DC cultures of ATF6α^ΔDC mice generate normal proportion of DCs (Figure 5A) and display Atf6 ablation without interfering with Atf6b expression (Figure 5B). To determine ATF6α transcriptional activity, we treated GM-DCs from ATF6α^ΔDC and control mice with the pharmacological ER stressor tunicamycin (Figure 5C). As expected, tunicamycin treatment induces activation of the ATF6α targets HerpUD and Grp94, and expression of these transcripts are reduced in GM-DCs deficient for ATF6α (Figure 5C), validating the model of study. Next, we interrogated if ATF6α contributes to the acquisition of immunogenic features during DC activation. For this purpose, we studied two types of stimuli; R848 (Resiquimod, an imidazoquinoline agonist of toll-like receptor 7 -TLR7 that possesses antiviral activity), and R848 combined with palmitic acid (R848/PA), which is a saturated fatty acid reported to induce activation of XBP1s, ATF4 and CHOP in activated DCs, potentiating their immunogenic function (Mogilenko et al., 2019). Our data indicate that both R848 and R848/PA elicit competent GM-DC activation by means of CD86 expression (Figure 5D). However, only R848/PA induces persistent BiP induction in GM-DCs (Figure 5E), confirming that the mixed stimuli trigger sustained UPR activation in these cells. As reported (Mogilenko et al., 2019), R848/PA also induced efficient activation of XBP1s y PERK branches (Supplementary Figure 4A), so we sought to investigate if R848/PA was also competent to trigger ATF6α transcriptional activity. Data depicted in Figure 5F shows that R848/PA efficiently induce HerpUD and Grp94 expression in GM-DCs in an ATF6α dependent manner, confirming that the mixed stimuli activate the transcription factor in DCs. To connect these findings with a functional role, we investigated if ATF6α regulates cytokine production in activated DCs. GM-DCs from ATF6α^ΔDC and control mice were stimulated with R848, R848/PA or vehicle and the production of proinflammatory cytokines was determined on mRNA and protein level. As reported (Mogilenko et al., 2019), PA treatment markedly augments the expression of Il-23p19 mRNA in R848-stimulated GM-DCs, although we found that expression of the cytokine is not regulated by ATF6α (Supplementary Figure 4B). IL-23 belongs to the IL-12 family of cytokines, in which IL-12 is a broadly studied factor involved in the generation of T helper 1 and natural killer responses, among others (Gee et al., 2009). The bioactive IL-12 form (termed IL-12p70) is comprised of the IL-12p35 and IL-12p40 subunits, in which the latter component is shared with IL-23 (IL-23p19/IL-12p40) (Gee et al., 2009). We observed that R848/PA stimulation also led to a significant increase in Il-12p35 mRNA expression compared to R848 alone (Figure 5G). Interestingly, compared to control GM-DCs, ATF6α deficient cells show decreased Il-12p35 mRNA expression upon R848/PA stimulation (Figure 5G). To extend these findings to protein level, we quantified IL-12p70 secreted in the supernatants of activated GM-DCs. Data depicted in Figure 5H shows that ATF6α-deficient GM-DCs stimulated with R848/PA secrete lower levels of IL-12p70 compared to control counterparts. These data indicate that ATF6α regulates IL-12 production by activated DCs. Next, we analyzed expression of additional proinflammatory cytokines and found that IL-6 was also reduced in the supernatants from ATF6α knock-out GM-DCs stimulated with R848 or R848/PA (p-value = 0.06) (Figure 5I). Notably, we found no regulation of Il-6 mRNA transcript levels by ATF6α (Supplementary Figure 4C), suggesting regulation of the cytokine on translational/posttranslational level. Furthermore, the regulation of ATF6α on cytokine production was not extended to all proinflammatory cytokines as TNF, another factor produced by activated GM-DCs was not regulated by the transcription factor (Figure 5J; Supplementary Figure 4D). Altogether, these findings demonstrate that ATF6α selectively contributes to the production of IL-12p70 and IL-6 in DCs activated with inflammatory triggers.

Statistical analysis was conducted using GraphPad Prism software (v9.1.2). Results are presented as mean ± SEM. Two groups were compared using non-parametric two-tailed Mann-Whitney test as indicated in figure legends. A p-value < 0.05 was considered statistically significant.

All animal procedures were approved and performed in accordance with institutional guidelines for animal care of the Fundación Ciencia y Vida and the Faculty of Medicine, University of Chile, and were approved by the local ethics committee.