Characterization of immune cells in psoriatic adipose tissue

Subcutaneous gluteal adipose tissue biopsies were performed in psoriasis patients (n = 30) aged 18 to 70 years in a consecutive sample from the Psoriasis Atherosclerosis and Cardiometabolic Disease Initiative (PACI; NCT01778569). A dermatologist confirmed the diagnosis of plaque psoriasis, and performed body surface area (BSA) and Psoriasis Area and Severity Index (PASI) assessments. Psoriatic arthritis was confirmed by a rheumatologist. Exclusion criteria included history of another systemic inflammatory illness, myocardial infarction, stroke, or chronic infectious disease. Study approval was obtained from the National Heart Lung and Blood Institute (NHLBI) institutional review board in accordance with the Declaration of Helsinki. All study participants provided written informed consent.

Fasting levels of total, HDL, and LDL cholesterol, triglycerides, glucose, erythrocyte sedimentation rate (ESR) and high-sensitivity C-reactive protein (hsCRP) were measured in a clinical laboratory.

Adipose tissue specimens were collected in RPMI medium and minced into fine pieces using scissors after careful removal of infiltrating blood vessels with forceps and scissors. Samples were digested in 5 mg/mL Type IV collagenase (Life Technologies, Grand Island, NY) for 30 minutes at 37°C in an Eppendorf thermomixer (Sigma-Aldrich, St. Louis, MO) at 1300 RPM. Tissue fragments were passed through a 40 μM nylon filter (BD Falcon, Beford, MA), washed twice with 1X PBS, and the floating adipocyte fraction was removed by vacuum aspiration. Blood contamination was determined by the visible presence of large amounts of erythrocytes in the cell pellet after digestion and centrifugation. Further, blood contaminated samples demonstrated abundant granulocyte populations by flow cytometry that are not typically present in adipose tissue. Three adipose samples were excluded from the analyses due to blood contamination.

Adipose immune cells were stained with various cocktails (up to 15-parameter) of surface antibodies in staining buffer [1X PBS without calcium or magnesium containing 5% heat-inactivated fetal bovine serum (Atlas, Fort Collins, CO), 0.25 mM EDTA (Sigma-Aldrich), and 0.09% sodium azide (Sigma-Aldrich)] for 30 minutes at 4°C. A 1:500 solution of LIVE/DEAD Aqua (Life Technologies) was added during the last 10 minutes of staining. For samples requiring intracellular staining, cells were fixed and permeabilized using BD Cytofix/Cytoperm kits (BD Biosciences, San Jose, CA) according to the manufacturer’s instructions, followed by intracellular staining for 30 minutes at 4°C. T cell IFN-γ and TNF-α production was elicited by incubation with 10 ng/mL PMA (BD Biosciences) and 1 μg/mL ionomycin (BD Biosciences) in the presence of 10 μg/mL brefeldin A (BD Biosciences) for 3 hours prior to intracellular cytokine staining. FoxP3, granzyme B, IL-1α , and IL-8 were assessed in unstimulated cells without using a protein transport inhibitor. Foxp3 staining was conducted similarly to other intracellular antigens, except that Foxp3 staining kits (eBioscience, San Diego, CA) were used. Positive staining for each antibody-fluorochrome combination was determined using fluorescence minus one (FMO) controls. Additional file 1: Table S3 lists the antibodies utilized for conventional and imaging flow cytometry. Samples were acquired on a BD Biosciences LSR II flow cytometer equipped with 405 nm, 488 nm, 532 nm, and 633 nm excitation lasers using DIVA software (BD Biosciences). Compensation was performed with single color controls prepared using BD Biosciences CompBeads. Compensation matrices were calculated automatically followed by manual adjustment and sample analysis using FlowJo software (Tree Star, Ashland, OR).

Peripheral blood mononuclear cells (PBMC) were prepared over a Ficoll gradient using standard methods. CD14+ monocytes were positively selected from PBMC or adipose biopsy specimens utilizing CD14 MicroBeads (Miltenyi Biotech, San Diego, CA), followed by passage over MACS LS columns (Miltenyi Biotech) according to the manufacturer’s instructions. Enrichment of CD14+ ATM was ~6-fold compared to ~20-fold enrichment of CD14+ monocytes from PBMC. Phagocytosis assays were performed on CD14+ monocytes, CD14+ enriched adipose specimens, and whole adipose specimens using the pHrodo (Life Technologies) system. Cells were incubated (≥5 x 10⁵ per well) in 96-well plates containing complete DMEM medium (Life Technologies) for 1.5 hours at 37°C. Culture supernatants were removed by aspiration, and opsonized pHrodo S. aureus bioparticles (Life Technologies) were added to the cells for 1.5 hours at either 37°C or 4°C (negative control). Cells were washed in staining buffer and stained for surface antigens prior to flow cytometric analysis.

Surface staining was performed as described above. Cells were washed with 1X PBS buffer containing 0.5 mM EDTA and 0.2% BSA at pH 7.2, suspended at a concentration of 1–5 × 10⁶/mL, and then incubated in 0.1 mM Hoechst (Life Technologies) at 37°C for 30 minutes. Positive staining for each antibody-fluorochrome combination was determined using FMO controls. Samples were acquired on an Amnis ImageStream X Mark II instrument equipped with 405 nM, 488 nM, 561 nM, and 640 nM lasers utilizing INSPIRE software (Amnis, Seattle, WA). Automatic compensation was performed with single color controls (BD Comp Beads), followed by manual adjustment and analysis using IDEAS 6.0 software (Amnis).

Spearman correlations were performed between adipose NK Cell frequencies and BMI, and multivariate linear regression was used to adjust for CMD risk factors (age, sex, diabetes, and tobacco use) and for treatment with oral corticosteroids, disease-modifying anti-rheumatic drugs (DMARDs), and/or biologic agents. No significant effects of treatment were identified. Thus, we report results from multivariate linear regression modeling after adjustment for CMD risk factors. Kruskall-Wallis testing with post-hoc Dunn’s multiple comparisons testing was performed to compare MFI values for surface markers among ATM populations. Adipose cell populations and cytokine expression were compared between psoriasis and control patients using Mann–Whitney U tests. Significance was considered at p < 0.05. Statistical tests were performed using Graphpad Prism (LaJolla, CA) and STATA (College Station, TX) software.

Patient characteristics (n = 30) and laboratory measurements are presented in Table 1. Our study population had a median age of 54 years [interquartile range (IQR) 41–61], was 54% male, had a median BMI of 29 (IQR 25.9-32.3), had moderate psoriasis (mean BSA 9.2 ± 16, mean PASI score 7.8 ± 9.3), and 38% had psoriatic arthritis (Table 1). Medication usage and CMD were also assessed. Topical steroid use was common (37%) and 3 patients received phototherapy (Table 1). Biologic therapy (39%) was more common than DMARD (9%) treatment (Table 1). Hypertension (32%), dyslipidemia (68%), diabetes (11%), and tobacco use (9% active, 28% former) were prevalent in our study population (Table 1), as was treatment for hypertension (19%), dyslipidemia (37%), and diabetes (6%).

Multi-parameter flow cytometry was applied to psoriatic adipose tissue, and gating strategies for cell identification (Figures 1, 2) and immune cell frequencies (Additional file 2: Table S1) are presented. Together, ATM (CD3-CD14+CD15-CD16-CD19-CD56-, Figure 1) were the most numerous immune cell type, comprising ~10% of viable cells. ATM made the prototypical cytokines IL-1β and IL-8 (Figure 1). Three populations of CD14+ ATM were identified based on HLADRII and CD206 expression (Figure 1). The HLADRII+CD206- subset was more abundant than the HLADRII-CD206- and HLADRII+CD206+ subsets. Frequencies of IL-1β and IL-8 expressing cells were not statistically different among the ATM subsets, except that the percentage of HLADRII+CD206-IL-8+ cells was significantly greater (p < 0.05) than HLADRII-CD206-IL-8+ cells (Additional file 3: Table S2). T cells (CD3+CD14-CD15-CD16-CD19-CD56-, Figure 2) collectively made up ~4.5% of viable cells, and were primarily αβ T cells with a predominance of CD4+ T cells. αβ T cells were largely either effector or memory cells (CD45RA-, Figure 2) and made the T cell cytokines IFN-γ and TNF-α (Figure 2). CD8+ T cells, FoxP3+ Tregs (predominately CD4+, Additional file 4: Figure S1), γδ T cells, and NKT cells were also identifiable (Figure 2). Confirmation of NKT cell (CD3+CD14-CD15-CD16-CD19-CD56+SSC^lo, Figure 2) phenotype was attempted using alpha-gal-cer loaded CD1d tetramers, but the scarcity of this population precluded definitive characterization. NK cells (CD3-CD14-CD15-CD16+/−CD19-CD56^Hi/LoSSC^lo, Figures 1, 2) comprised ~1.5% of viable cells, were largely CD16+CD56^Lo versus CD16-CD56^Hi, and expressed the characteristic protease granzyme B (Figure 2). B cells (CD3-CD14-CD15-CD16-CD19+CD56-PD-1+, Figure 2, Additional file 5: Figure S2) made up <1% of viable cells. Neutrophils (CD3-CD14-CD15+CD16 + CD19-CD56-SSC^Hi, Figures 1, 2) were also infrequent and expressed the characteristic marker granzyme B (Figure 2).

Conventional flow cytometry demonstrated various immune cell subsets in psoriatic adipose tissue using classic phenotypic markers. The ability of these markers to correctly identify adipose immune cells was confirmed by imaging flow cytometry. This technology combines features of brightfield microscopy, flow cytometry, and immunofluorescence microscopy in a single instrument, thus allowing for high-resolution characterization of cell populations in tissues [35]. Imaging flow cytometry verified ATM by positive CD14 staining, abundant cytoplasm, and round to U-shaped nuclei consistent with macrophage morphology (Figure 3, panels A-C). Three subsets of ATM were present based on HLADRII and CD206 expression (Figure 3), and ATM rarely expressed CD16 (mean 1.47 ± 1.77%, Additional file 6: Figure S3). Conventional flow cytometry was then performed to uncover ATM markers that may link metabolism and immunity in psoriasis (Figure 4). ATM expressed a wide array of surface receptors with roles in microbial pattern recognition (TLR2, TLR4), immune suppression (CD274), cell adhesion (CD11c), cholesterol trafficking (ABCA1), chemokine binding (CX3CR1), and macromolecule clearance (CD163, SR-B1, LOX-1, MSR1, RAGE, and CD36) (Figure 4A). Mean fluorescence intensity (MFI) values for these receptors were strikingly different among the macrophage subpopulations (Figure 4B). TLR2, CD11c, MSR1, and HLADRII expression were highest in HLADRII+CD206+, intermediate in HLADRII+CD206-, and lowest in HLADRII-CD206- ATM (Figure 4B). LOX-1 expression was lowest in the HLADRII+CD206- population, CD36 expression was lowest in the HLADRII+CD206+ subset, and RAGE expression was highest in HLADRII-CD206-, intermediate in HLADRII+CD206-, and lowest in HLADRII+CD206+ cells (Figure 4B). ATM phagocytic function was also assessed by flow cytometry using the pHrodo system, which utilizes bioparticles that fluoresce only upon acidification within phagocyte endosomes [36]. Phagocytosis was equivalent among the 3 ATM subpopulations (Figure 4C).

Imaging flow cytometry confirmed other immune cell populations in psoriatic adipose tissue. T cells were verified by positive CD3 staining, round nuclear morphology, and scant cytoplasm (Figure 3, panel D). To determine cell types that express CD16 in adipose tissue, CD3-CD14-CD16+ cells were manually categorized (hand tagged) as having mononuclear (circular, Figure 5, panel A) or polymorphic (polymorph, Figure 5, panel B) nuclei. The Feature Finder wizard in IDEAS software was then used to distinguish cells from the source population with similar morphologic characteristics. This strategy allowed segregation of cellular debris/unfocused cells (Figure 5, panel D), mononuclear cells (Figure 5, panel C), and polymorphonuclear cells (Figure 5, panel E) within the CD16+ population. CD16+ polymorphonuclear cells had multi-lobed (2–3 lobes) nuclei and dense granularity, characteristic of neutrophils. In contrast, CD16+ mononuclear cells had circular nuclei and variable granularity, confirming their identity as NK cells. Further, CD16+ NK cells did not express the macrophage markers HLADRII or CD206 (Additional file 6: Figure S3), nor did they express IL-1β, ACBA1, TLR2, TLR4, CD274, LOX-1, MSR1, CD36, RAGE, CD163, or SR-B1 (Additional file 7: Figure S4).

To better understand how adipose immune cells relate to obesity in psoriasis, we performed Spearman correlation analyses between individual immune cell subset frequencies and BMI. Both CD16-CD56^Hi (r −0.552, p = 0.008) and CD16+CD56^Lo NK cells (r −0.626, p = 0.002) were inversely correlated with BMI in unadjusted analyses (Figure 6). The inverse relationship between BMI (outcome variable) and adipose CD16+CD56^Lo NK cells remained significant after adjusting for the CMD risk factors age, sex, diabetes, and tobacco use (β -46.5, p = 0.04), and with additional adjustment for treatment with oral corticosteroids, DMARDs, and/or biologics (β -54.04, p = 0.026). In contrast, BMI and CD16-CD56^Hi NK cell frequencies were no longer associated after adjustment for CMD risk factors (β -1.89, p = 0.172). BMI was also unrelated to the other adipose immune cell types.

To determine whether adipose immune cell composition is affected by psoriasis, we performed a nested case°Control study (n = 6 non-diabetic control and psoriasis patients) matched for age, sex, BMI, smoking, and dyslipidemia. We observed that CD16-CD56^Hi NK cells, but not other immune cells, were statistically greater (0.42% live cells in psoriasis versus 0.06% live cells in controls, p = 0.014) in psoriatic compared to control adipose. While the frequencies of total ATM were similar in psoriasis and controls (10.2% live cells in psoriasis versus 12.9% live cells in controls versus, p = 1.0), IL-1β producing (46.1% of psoriatic ATM versus 29.3% of control ATM, p = 0.076) and IL-8 producing (66.1% of psoriatic ATM versus 51.0% of control ATM, p = 0.175) ATM were more abundant in psoriatic compared to control adipose tissue. However, only the HLADRII+CD206+ ATM subset demonstrated a statistically significant difference in IL-1β (40.3% in psoriasis versus 19.6% in controls, p = 0.047), but not IL-8 (62.5% in psoriasis versus 45.7% in controls, p = 0.117), production between psoriatic and control adipose tissue.