Dynamics and Transcriptomics of Skin Dendritic Cells and Macrophages in an Imiquimod-Induced, Biphasic Mouse Model of Psoriasis

Long-term application of IMQ on mouse skin recapitulates the early and late phases of human psoriasis

Topical treatment of mouse skin with Aldara cream containing IMQ triggers a psoriasis-like inflammation (21). In most models, 6.25 mg Aldara cream is applied on recently shaved back skin. Depilation induces LC activation (17), so we selected the ear to apply Aldara cream because it has only sparse hair follicles that preclude prior depilation. By using 7 mg Aldara cream (0.35 mg IMQ) per ear side, we induced the full flare of psoriatic skin inflammation. In contrast to most previous conditions in which treatment was limited to 5–7 d, we applied Aldara cream for 14 consecutive d. As documented below, this regimen allowed the study of the early acute phase and late chronic phase of the resulting psoriasis-like inflammation. Using an adapted human psoriasis activity and severity score, clinical parameters such as skin thickness, erythema, and scaling continuously increased over the first 7 d of application, denoted as the early phase, and reached a plateau by day 7. During the late phase (days 8–14), those clinical parameters remained stable (Fig. 1A).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Long-term application of IMQ recapitulates the early and late phases of human psoriasis. (A) IMQ was applied on mouse ear skin for 14 consecutive d, and ear thickness, erythema, and scaling were measured on a daily basis. Erythema and scaling were evaluated according to a score of 1–4 (see Materials and Methods). One representative experiment out of three is depicted (mean ± SEM). (B) Representative H&E staining of histological sections of ears treated for 0, 3, 5, 7, 11, and 14 d with IMQ (n ≥ 6 ears/group). The higher magnification provided for day 5 shows that microabscesses consist of an accumulation of granulocytes with lobed and segmented nuclei. A thickening of the stratum corneum (hyperkeratosis) is also noted as compared with day 0. Within the stratum corneum, the flattened nuclei that remain in the corneocytes (white arrow) illustrate the phenomenon of parakeratosis. Histopathological features resembling the human hallmarks of psoriasis are pointed out on representative micrographies. Black arrow, acanthosis; red arrow, epidermal neutrophil microabscess; white arrow, para- and hyperkeratosis; and white square, dense dermal infiltrate and dilatation of blood vessels. Scale bars, 50 μm. D, dermis; E, epidermis.

Histopathological examination of the psoriasis-like lesions that developed over the course of IMQ treatment further supported the existence of distinguishable early and late phases (Fig. 1B). The first changes appearing during the early phase corresponded to dense dermal infiltrate and dilatation of blood vessels. The plaques that developed during the early phase were associated with neutrophilic pustules in the epidermis. Moreover, the thickening epidermis showed the typical histopathological hallmarks of human psoriatic disease that is acanthosis, parakeratosis, and hyperkeratosis. Therefore, the clinical and histopathological signs developing during the first 7 d of IMQ treatment resembled those present in the early acute phase of human disease in which the first papules are being built. Most histopathological parameters plateaued during the late phase, indicating that the inflammation was fully installed and resembled the stable psoriasis plaques observed in humans (Fig. 1A, 1B). At days 11 and 14 of the late phase, neutrophilic abscesses decreased, whereas epidermal changes stayed pronounced. Time points corresponding to days 5 and 11 are representative of the early and late phases, respectively, and were primarily used in subsequent studies.

Distinct types of myeloid cells contribute to the early and late phases

We recently established a multicolor phenotyping flow cytometry key enabling to identify LCs, cDC subsets, neutrophils, monocytes, moDCs, and macrophages among the CD45⁺Lin⁻ cells of the skin (8, 9). Neutrophils can be identified as Ly-6G⁺CD11b⁺ cells (Supplemental Fig. 1A) (22). MHC II⁺ skin cells are composed of LCs, CD11b⁻CD24⁺ cDCs (also known as CD103⁺ or XCR1⁺ DCs), CD11b⁺ cDCs (CD24^lowCD11b⁺), and double-negative cDCs (CD11b⁻CD24⁻) (Supplemental Fig. 1B). As shown in Supplemental Fig. 1C, CD11b⁺ cDCs can be distinguished from dermal CD11b⁺ non-cDCs due to their Ly-6C⁻CD64⁻ phenotype. Dermal CD11b⁺ non-cDCs can be further resolved on the basis of CCR2 and CD64 expression into CCR2⁺CD64^low cells that comprise dermal monocytes and Ly-6C^high and Ly-6C^low moDCs and CCR2^lowCD64⁺ cells that comprise MHC II^low and MHC II^high dermal macrophages (Supplemental Fig. 1C). This flow cytometry key kept its discriminatory power when applied to the inflamed skin resulting from IMQ treatment (Supplemental Fig. 1D, 1E).

Analysis of the absolute numbers of each myeloid cell types found in the skin over 14 d of continuous IMQ application showed that neutrophils, monocytes, and Ly-6C^hi moDCs increased during the early phase and peaked around day 5 (Fig. 2A, 2B). CD11b⁻CD24⁺, CD11b⁻CD24⁻, and CD11b⁺ dermal cDCs decreased during the first days of treatment, likely due to their enhanced migration to draining LN. Baseline numbers of dermal cDCs were restored at the end of the treatment period, with CD11b⁺ dermal cDCs showing a more rapid rebound (Fig. 2C). Prior to analyzing the dynamics of skin pDCs following IMQ treatment, we noted that PDCA-1 (also known as BST-2 or CD317), a marker commonly used to identify pDCs, was also expressed at intermediate levels on dermal mast cells (data not shown). After excluding dermal mast cells from CD45⁺Lin⁻ skin cells on the basis of their CD117 expression, we failed to detect any pDCs, defined on the basis of their expected PDCA-1⁺B220⁺Ly-6C⁺CD11c^lowMHC⁻ to ^lowCD117⁻CD11b⁻ phenotype, in the remaining cells in steady-state skin and during the whole course of IMQ treatment (data not shown). In contrast, using the same phenotyping strategy, pDCs were readily detectable in skin-draining auricular LNs over the whole course of IMQ treatment. These observations are consistent with the fact that pDC ablation had no effect on the magnitude of IMQ-induced psoriatic plaque formation (7, 14).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Distinct skin-infiltrating myeloid cell types characterize the early and late phases of IMQ-induced psoriasis-like inflammation. Single-cell suspensions from mouse ears were analyzed by flow cytometry (see Supplemental Fig. 1), leading to the identification and determination of the absolute numbers of neutrophils (A), dermal monocytes (B), Ly-6C^high and Ly-6C^low moDCs (B), CD11b⁻CD24⁺, CD11b⁻CD24⁻, and CD11b⁺ dermal cDCs (C), MHC II^low and MHC II^high dermal macrophages, and LCs (D). Mean ± SEM is depicted. (E) Pie chart representation of myeloid cell types found in skin at steady state and after 5 and 11 d of IMQ application. The surface of each chart is proportional to the absolute numbers of myeloid cell types per ear. Data are representative of at least three experiments with ≥3 animals per condition.

Dermal macrophages started to increase during the late phase, their numbers plateauing around day 11 (Fig. 2D), whereas LC numbers increased continuously until they peaked at day 11 (Fig. 2D). LC numbers decreased thereafter and reached steady-state levels 20 d after the initiation of continuous IMQ application (data not shown). Therefore, the early and late phases of IMQ-induced psoriatic-like inflammation correlated with the presence of distinct types of myeloid cells, with neutrophils, monocytes, and moDCs dominating the early phase and LCs and macrophages transiently increasing during the late phase (Fig. 2E). This is reminiscent of psoriatic patients, in whom early lesions show an increase in neutrophils, whereas macrophages and cells from the adaptive immune system predominate in stable plaques (23, 24).

Gene-expression profiling of skin myeloid cell types during the early and late phases of IMQ-induced psoriasis-like inflammation

LCs, CD11b⁺ dermal cDCs, monocytes, Ly-6C^high moDCs ,and MHC II^low and MHC II^high macrophages were sorted from the skin prior to and after 5 or 11 d of IMQ treatment and subjected to global gene-expression profiling (Supplemental Table I). Two striking observations can be made from hierarchical clustering analysis. First, irrespective of IMQ treatment, LCs clustered together and apart from the other analyzed myeloid cell types (Fig. 3A). Such a feature likely reflects the prenatal origin of LCs, whereas other skin myeloid cell types are generated via adult hematopoiesis (10, 11). Moreover, LCs are the sole type to reside in the epidermis, and this might contribute to their unique pattern of gene expression (25–27). Second, all of the non-LC cell types isolated from IMQ-treated mice clustered together and apart from their counterparts from IMQ-untreated mice. This suggests that a convergent transcriptional reprogramming occurs among all of the non-LC cell types during IMQ treatment. Interestingly, for most cell types, the samples isolated at days 5 and 11 of IMQ treatment clustered separately, supporting the existence of two distinguishable phases of inflammation development in our long-term IMQ application model.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Gene-expression profiling of skin myeloid cell types during the course of IMQ-induced psoriasis-like inflammation. (A) Hierarchical clustering performed on the cell types specified in the key on the top left side prior to (D0) or after 5 (D5) or 11 d (D11) of IMQ treatment. Numbers above nodes indicate their robustness (see Materials and Methods). The smaller frequency of Ly-6C^low moDCs and CD11b⁻CD24⁺ and CD11b⁻CD24⁻ cDCs found in IMQ-treated skin prevented their analysis, whereas technical issues prevented the analysis of MHC II^highMF.Dermis.IMQ.D5 and Monocytes.Dermis.IMQ.D11 samples. (B) PCA of gene expression by the six cell types prior to and after IMQ treatment (see Key).

PCA confirmed and extended the observations deduced from hierarchical clustering. The PC1 axis separated LCs from all of the other non-LC cell types, irrespective of IMQ treatment (Fig. 3B), a finding consistent with the two primary branches observed in hierarchical clustering. The PC2 axis separated untreated from IMQ-treated cells, showing that IMQ affected all cell types, including LCs, in a convergent manner. A result confirmed by the analysis of the individual expression patterns of genes contributing the most to the PC2 axis (Supplemental Table I). This convergence is illustrated in Supplemental Fig. 2A and 2B for a few selected genes, among which stand out the S100a8 and S100a9 genes, which code for microbicidal and chemotactic alarmins (28). Elevated S100A8 and S100A9 protein levels are a hallmark of human psoriasis, contributing to sterility of the psoriatic plaques and to noxious proinflammatory effects (28, 29). Akin to PC2, the PC3 axis separated untreated from IMQ-treated cells, and, in the case of LCs and CD11b⁺ cDCs, it also separated those isolated after 5 and 11 d of IMQ treatment. Indeed, several of the genes contributing the most to the PC3 axis were differentially regulated at day 5 or 11 of IMQ treatment in at least one of the cell types examined (Supplemental Fig. 2C, 2D, Supplemental Table I). For instance, after 5 d of IMQ treatment, LCs selectively and transiently expressed three genes coding for endopeptidase inhibitors (Stfa2l1, Staf2, and Serpinb9 in Supplemental Fig. 2E). The protein encoded by Serpinb9 belongs to serpins and may prevent lysis by cytotoxic T cells (30), whereas Stfa2l1 and Staf2 code for stefins A that are intracellular inhibitors of cysteine cathepsins involved in Ag processing (31).

Psoriatic inflammation is associated with increased expression of canonical IFN-I–responsive genes

IFN-I plays an important role in human psoriasis (4). Consistent with that view, several IFN-stimulated genes (ISGs), such as Ifitm1, Cd274, and Plac8, were expressed at higher levels after 5 d of IMQ treatment in several of the analyzed cell types (Supplemental Fig. 2C). To determine the effect of prolonged IMQ treatment on ISG expression in skin myeloid cells, we performed GeneSet Enrichment Analyses (GSEA) using an ISG gene set that encompasses all known ISGs (20) and a “KC_IFNα_UP” gene set that corresponds to genes reported to be significantly induced by IFN-I in cultured keratinocytes (4). We also used the C2-curated gene set that contains 4722 molecular signatures (18). After 5 d of IMQ treatment, LCs were the sole to be significantly enriched for the ISG gene set, whereas all other analyzed cell types were enriched for the KC_IFNα_UP gene set at day 5, and even at day 11 for some of them (Supplemental Fig. 2F). The higher expression of ISGs noted for LCs after 5 d of IMQ treatment turned out to be due to the fact that at steady state, they expressed much lower level of ISGs than the other cell types examined. Accordingly, the rather minor increase in ISG expression that occurred in LCs after 5 d of IMQ treatment translated in substantial fold changes over steady-state conditions and thus biased the GSEA. Therefore, LCs expressed most ISGs at low levels prior to and after IMQ treatment, a feature that likely contributes to maintain epidermal homeostasis. However, after IMQ treatment, LCs and to a lesser extent CD11b⁺ dermal cDCs selectively and strongly expressed the Ifi205, Mreg, and Fscn1 ISGs (Supplemental Fig. 2G). Fscn1 expression has been associated with DC maturation (32). In contrast to the situation observed for LCs, continuous IMQ treatment induced a substantial increase in the expression of large numbers of canonical ISGs in all of the non-LC cell types examined (Supplemental Fig. 2G, Supplemental Table I), suggesting that IFN-I play an important role in the long-term IMQ application model.

The unfolding of psoriatic inflammation is associated with enhanced LC proliferation

The most impressive enrichment observed in GSEA corresponded to gene sets associated with cell proliferation and denoted as “CCNB2” and “Cell cycle” (20) (Supplemental Fig. 2F). LCs and MHC II^low macrophages found in the skin after 5 and 11 d of IMQ-treatment showed the highest enrichment in these gene sets. Mitosis-associated genes were overexpressed in LCs both after 5 and 11 d of IMQ treatment, whereas their expression peaked after 5 d of IMQ treatment in MHC II^low dermal macrophages (Supplemental Fig. 2H, Supplemental Table I).

To corroborate the presence of proliferating LCs in psoriatic lesions, we analyzed whether LCs expressed Ki67, a nuclear protein that characterizes actively cycling cells. Confocal microscopy of skin sections from LangEGFP mice that express a GFP under the control of the gene coding for langerin (CD207) allowed the visualization of epidermal LCs on the basis of EGFP expression and epidermal localization (17). Among the EGFP⁺ cells present in the skin of LangEGFP mice that had been treated with IMQ for 11 d, 22.6 ± 2.4% were Ki67⁺, whereas none of those found in untreated control LangEGFP mice were Ki67⁺ (Fig. 4A). Quantification by flow cytometry of the increase in Ki67⁺ LCs over the course of IMQ application showed a peak at day 9 when ∼30% of the LCs were Ki67⁺ (Fig. 4B). Congruent with those results, when mice were exposed to BrdU for 4 d prior to the end of IMQ treatment, ∼15% of the LCs had incorporated BrdU as compared with <5% in untreated control mice (Fig. 4C). Importantly, analysis of separated epidermal and dermal sheets demonstrated that the increase in the total numbers of LCs was exclusively due to their accumulation in the epidermis (Fig. 4D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

LCs proliferate during the late phase of psoriatic inflammation. Mice were left untreated, or their ears were treated for the specified period with IMQ. (A) Histological sections of ears from LangEGFP mice were stained for Ki67 (red) and EGFP (green). Arrows indicate double-positive cells. Representative examples of ≥6 ears per condition are shown. A higher magnification highlighting four Ki67⁺ LCs is provided. Scale bars, 80 μm. (B) Ears of wild-type mice were left untreated (0) or treated for 8, 9, 10, or 11 d with IMQ, and the percentage of Ki67⁺ cells among LCs was assessed by flow cytometry. (C) Mice were treated for 11 d with IMQ and received BrdU for the last 4 d of IMQ treatment (11 d IMQ) or solely received BrdU for 4 d (untreated). The percentage of BrdU⁺ cells among LCs was determined. (D) The epidermis and dermis of the ear skin were separated, and LCs were quantified as in Fig. 2D and Supplemental Fig. 1B. Error bars represent means ± SEM. (E) Absolute numbers of TRITC⁺ CD103⁻CD24⁺ migratory cDCs extracted from skin-draining LNs. (F) Absolute numbers of radioresistant migratory LCs extracted from the skin-draining LNs of B6 (CD45.1) → B6 (CD45.1 × CD45.2) BM chimeras. In (B), (C), (E), and (F), each dot corresponds to a mouse, and the mean is indicated. One out of two to three representative experiments is depicted. *p < 0.05, **p < 0.01, ***p < 0.005. D, dermis; E, epidermis.

To determine whether the increase in epidermal LCs observed during the course of IMQ treatment resulted from their impaired migration to skin draining LNs, we combined IMQ treatment with epicutaneous application of TRITC, a fluorescent dye that labels skin DCs and allows to track them during their migration to LN. Analysis of ear-draining LNs after 11 d of IMQ application revealed a strong increase of TRITC⁺ cells among CD103⁻CD24⁺ migratory cells as compared with control mice (Fig. 4E). Considering that CD103⁻CD24⁺ migratory cells comprise both LCs and CD103⁻CD24⁺ dermal cDCs, we relied on the fact that LCs are radioresistant to distinguish them from radiosensitive CD103⁻CD24⁺ dermal cDCs (12). Accordingly, lethally irradiated B6 (CD45.1 × CD45.2) mice were reconstituted with BM cells isolated from B6 (CD45.1) mice. Analysis of the LN draining the ear of B6 (CD45.1) → B6 (CD45.1 × CD45.2) BM chimeras after 11 d of IMQ treatment showed that the 10-fold increase in the absolute number of CD103⁻CD24⁺ migratory cells was fully accounted by radioresistant, host-derived migratory LCs (Fig. 4F). Therefore, the increase in LC numbers observed in the epidermis of IMQ-treated ear did not result from impaired migration of epidermal LCs to the LN. When considered together with a recent study demonstrating that BM-derived cells that infiltrate the inflamed skin during IMQ treatment do not differentiate into epidermal LCs (7), our results suggest that LC accumulation in the epidermis of IMQ-induced psoriatic skin lesions is primarily due to local proliferation of the LC pool and that those LCs are distinct from the BM-derived, short-lived LCs that develop in the skin following UV irradiation (33).

LC depletion led to increased numbers of neutrophils in the late phase of inflammation

The activation status of the LCs and cDCs found in the skin during the long-term IMQ application model was assessed by measuring the levels of expression of CD80, CD40, and MHC II molecules at their surface. Upon IMQ application, LCs showed an upregulation of all of the analyzed activation markers whereas CD11b⁻CD24⁺, CD11b⁻CD24⁻, and CD11b⁺ dermal cDCs showed no detectable changes (Supplemental Fig. 3 and data not shown). Considering that upon IMQ treatment, LCs initially increased in number and were the only one to show robust signs of activation, we next analyzed their contribution to the pathology that develops in our biphasic IMQ application model. We used LangDTREGFP knockin mice in which it is possible to ablate Langerin-positive DCs following treatment with DT (17). Both LCs and CD11b⁻CD24⁺ dermal cDCs express Langerin (34). However, after in vivo DT treatment, LCs and CD11b⁻CD24⁺ dermal cDCs repopulate the skin with different kinetics. CD11b⁻CD24⁺ cDCs are reconstituted at day 13 after DT administration, whereas LCs are undetectable for up to 30 d after DT administration (34). Therefore, IMQ treatment was started 13 d after the last DT injection, a time point at which LCs were absent, and the adventitious effect of DT treatment noted in some models were no longer significant (9), and pursued for 11 d. A similar IMQ treatment was applied to LangDTREGFP mice that received no DT. The lack of LCs had no detectable influence on measured clinical parameters, although there was a trend during the late phase toward stronger inflammation (Fig. 5A). Importantly, flow cytometry analysis of cell suspensions from ear skin revealed that in the absence of LCs, the late phase of inflammation was characterized by a significantly stronger influx of neutrophils, whereas the other myeloid cell types found in the skin showed no significant differences (Fig. 5B and not shown). We used MMP9, a tertiary granule protein that is formed in mature granulocytes with band-shaped nuclei (35), to determine whether the increase in neutrophil numbers observed in the absence of LCs occurred in the epidermis. Confocal microscopy of MMP9-stained ear skin section of IMQ-treated LangDTREGFP mice that lack LCs showed typical neutrophilic epidermal abscesses as compared with IMQ-treated LangDTREGFP mice that contain LCs (Fig. 5C), a result supported by H&E staining (Fig. 5D). Therefore, in the absence of LCs, a significant epidermal accumulation of neutrophils occurred during the late phase of the long-term IMQ application model of psoriasis-like inflammation.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

The lack of LCs during the whole course of the inflammation results in increased numbers of neutrophils during the late phase of psoriatic inflammation. IMQ application for 0, 5, or 11 d on ears of LangDTREGFP mice for which LCs have been depleted via injection of DT (+DT) or left untouched (-DT). (A) Ear thickness, erythema, and scaling were measured daily. Increase in ear thickness was calculated as percent of thickness at day 0. (B) Absolute numbers of neutrophils in the ear skin of LangDTREGFP mice. (A and B) One representative experiment is depicted out of three (mean ± SEM). (C) Ear sections were stained for MMP9 (red) and phalloidin (white). Scale bars, 50 μm. (D) Representative H&E staining of ear section of LangDTREGFP mice. Scale bars, 100 μm. (C and D) ≥6 ears were analyzed per conditions. *p < 0.05, **p < 0.01.