This study was conducted with blood and biopsy samples obtained from patients with RRP, blood from control individuals without RRP, and with control neonatal foreskin and adult bariatric abdominal skin from surgical discard tissues obtained at the Long Island Jewish Medical Center, Northwell Health. The studies were approved by the Northwell Health Institutional Review Board.

The demographics of the patients with RRP used in these experiments are shown in Supplementary Table 1. Recurrent respiratory papillomatosis is a rare disease with an estimated prevalence in the United States of 4.3/10⁵ in children younger than 14 years, and of 1.8/10⁵ in adults (29, 30). The severity of RRP can be divided into two categories, severe and mild–moderate, based on the extent of disease at the time of surgery and the frequency of recurrence. At each surgery, the number of disease sites, the anatomic surface area of disease, and the extent of luminal obstruction are documented to yield a composite score as described previously (8, 31). This composite score is divided by the number of days that had elapsed since the previous surgery to yield a growth rate, which is a measurement of disease severity. The mean growth rate from multiple surgeries is used to define the overall severity score for an individual patient. An overall disease severity score of ≥0.06, or the presence of tracheal extension, is defined as severe disease. An overall severity score of <0.06 and the absence of tracheal extension are defined as mild–moderate disease.

Whole blood was collected in 10 mL heparin-containing Vacutainer tubes and centrifuged at 1,200 g for 10 min at room temperature to separate cells from the plasma. Aliquots of plasma were collected and immediately frozen at −70°C until use. A forward sequential competitive binding enzyme-linked immunosorbent assay (ELISA) (PKGE004B; R&D Systems, Minneapolis, MN, USA) was used as per the manufacturer's specifications. The cross reaction of anti-PGE₂ antibodies in this kit with other prostenoids is <5% as per the manufacturer. All samples were run in triplicate, and mean plasma levels ± SD calculated.

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized blood using Ficoll-Hypaque density gradient centrifugation (GE Healthcare, Fairfield, CT, USA). Monocytes were negatively selected using a commercially available magnetic isolation kit (Pan Monocyte Isolation Kit; Miltenyi Biotec, Bergisch Gladbach, Germany), which consistently yielded >90% purity. To generate iLCs, monocytes were cultured for 7 days in complete medium, which consisted of RPMI 1640 medium (Gibco/Life Technologies [Thermo Fisher Scientific Inc., Walthem, MA, USA]), 2 mM l-glutamine, 10 mg of streptomycin, 10,000 U/mL of penicillin G, and 10% fetal bovine serum (FBS) (Atlanta Biologicals, Flowery Branch, GA, USA), supplemented with 100 ng/mL granulocyte-macrophage colony-stimulating factor (GMCSF), 20 ng/mL IL-4, and 10 ng/mL transforming growth factor β1 (TGFβ1) (R&D Systems) (27). Cells were supplemented with complete medium + GMCSF and TGFβ1, but without IL-4, on days 2 and 4. On day 7, cells were collected, and iLCs were positively selected using CD1a MicroBeads (Miltenyi Biotec). Immature LCs were stained for viability (Live/Dead Fixable Aqua Dead Cell Stain Kit (Invitrogen/Life Technologies, Thermo Fisher Scientific Inc.) and then surface stained with fluorochrome-labeled antibodies specific for CD1a–fluorescein isothiocyanate (FITC) (BD Biosciences, San Jose, CA, USA) and E-cadherin-APC (Bio Legend, San Diego, CA, USA) and analyzed on a BD FACS Canto II (BD Biosciences). To determine the effect of PGE₂ on generation of monocyte-derived iLCs, 250 nM PGE₂ was added for all 7 days, and the iLCs were further selected for expression of CD207 (langerin) (anti–CD207-PE).

Peripheral blood mononuclear cells were isolated and monocytes purified by negative selection as stated above. Monocytes were washed once in phosphate-buffered saline (PBS) and immunostained with anti–CD14-PE and anti–CD16-Pacific Blue (BD Biosciences). Doublets and dead cells were excluded; monocytes were selected based on their light scatter and CD14 and CD16 expression and sorted into CD14⁺⁺ CD16⁻ (classical) CD14⁺⁺ CD16⁺ (intermediate) and CD14⁻ CD16⁺⁺ (non-classical) monocytes using a BD FACSAria sorter (BD Biosciences), and the percentage of monocytes in each subpopulation calculated. All three subpopulations of monocytes were cultured for 7 days in complete RPMI medium supplemented with 100 ng/mL GMCSF, 20 ng/mL IL-4, and 10 ng/mL TGFβ1 as above. On day 7, cells were harvested, surface stained with fluorochrome-labeled antibodies CD1a-FITC and E-cadherin-APC, and then analyzed as above on a BD FACS Canto II (BD Biosciences).

Purified iLCs, generated as above and selected with CD1a⁺ microbeads, were cultured for 48 h in complete RPMI medium with either 10 ng/mL of recombinant active IL-36γ (a.a.18–169) (R&D Systems), 250 nM PGE₂ + 10 ng/mL IL-36γ, or 100 ng/mL lipopolysaccharide (LPS) (Sigma-Aldrich, St. Louis, MO, USA). Cells were then harvested and surface stained for CD83-APC (BD Biosciences), CD1a-FITC, and E-cadherin-APC. Cells were fixed in 1% formaldehyde and analyzed on a BD FACS Canto II (BD Biosciences).

Deidentified surgical discards of abdominal skin or foreskin, or biopsies of respiratory papilloma tissue, were obtained within an hour of surgery. All tissues were washed in PBS containing antibiotics (100 μg/mL streptomycin and 100 U/mL of penicillin). For both abdominal skin and foreskin, adipose tissue was removed with forceps and a scalpel; small slices of the skin were placed in complete RPMI media plus 2.5 U/mL dispase (STEMCELL Technologies, Cambridge, MA, USA) with the epidermis facing up, and incubated overnight at 4°C. The next day, the epidermis was separated from the dermis using fine forceps and washed with PBS. To obtain single-cell preparations, abdominal and foreskin epidermis and papilloma biopsies were minced with a scalpel and scissors and then incubated with 0.05% trypsin (Gibco, Grand Island, NY, USA) and 100 U/mL DNAse I (Invitrogen, Thermo Fisher Scientific Inc.) at 37°C for 30 min. Complete RPMI media supplemented with 10% FBS was added to inactivate trypsin; cells were passed through a 70-μm cell strainer (Fisher) to obtain a single-cell suspension, collected by centrifugation at 600 g for 10 min, resuspended in PBS, and counted (Beckman Coulter, Indianapolis, IN, USA).

For the experiments shown in Figure 4, monocyte-derived iLCs and papilloma-derived LCs were isolated using Miltenyi beads for CD1a⁺, or CD207⁺ cells as described above, resuspended in medium without cytokines, and then stimulated for 4 h at 37°C with 10 ng/mL of recombinant active IL-36γ [amino acids (aa) 18–169] (R&D Systems), or with recombinant inactive full-length IL-36γ (aa 1–169) as a control.

For the experiments comparing LCs isolated from all three tissues, the LCs were stained with CD1a-FITC and CD207-PE, and sorted on a BD FACSAria as above. An aliquot of isolated LCs was immediately frozen for RNA isolation, and the remainder was cultured overnight in complete RPMI media supplemented with 100 ng/mL GMCSF and 10 ng/mL TGFβ1. The tissue-derived iLCs were then resuspended in fresh medium and stimulated for 4 h at 37°C with 1,500 ng/mL of poly(I:C) or 25 ng/mL of tumor necrosis factor α (TNFα) added to the medium. Addition of PBS was used as a control.

Cells were collected by centrifugation, and total mRNA isolated following DNase-1 treatment as per manufacturer's instructions (Qiagen, Valencia, CA, USA). The integrity of all mRNAs was determined by Bioanalyzer (RIN > 7.0) (Agilent Technologies, Santa Clara, CA, USA). To quantitate expression of the proinflammatory or anti-inflammatory cytokines/chemokines IL-1β, CCL-1, CCL-20, and TNFα, reverse transcription–polymerase chain reaction (RT-PCR) was performed using gene-specific intron-spanning primers, either purchased from Applied Biosystems/Life Technologies (Thermo Fisher Scientific, Waltham, MA, USA) or individually designed with a free bioinformatics program available on the Fisher website (Fisher Scientific [Thermo Fisher Scientific Inc., Pittsburg, PA, USA]), using the Human Universal Probe Library (Roche, Branford, CT, USA) for detection. Quantitative RT-PCR (qRT-PCR) was carried out on an Applied Biosystems 7900 HT (Applied Biosystems/Life Technologies [Thermo Fisher Scientific, Waltham, MA, USA]) following amplification with the iScript One-Step RT-PCR kit (Bio-Rad, Hercules, CA, USA). Total mRNA was reverse transcribed at 50°C for 10 min, followed by 5 min at 95°C to inactivate the reverse transcriptase and activate the Taq polymerase (hot-start), followed by 40 cycles of two-step PCR at 95°C for 15 s and 60°C for 1 min. Individual mRNA samples were run in duplicate, and the average Ct values for the gene of interest normalized to the average Ct values for the housekeeping gene (GAPDH) to calculate δ Ct. Results are expressed as level of expression relative to GAPDH, in order to compare relative expression of multiple cytokines and chemokines across different cells. Expression levels below that of GAPDH are shown as a decimal, and those above GAPDH expression are shown as whole numbers.

All results are expressed as mean ± S.D. Comparisons between two groups were done with a two-tailed, unpaired t-test. Comparisons between multiple groups were done with either a one-way analysis of variance (ANOVA) with Tukey-Kramer multiple-comparisons test or a nonparametric Kruskal-Wallis test with Dunn multiple-comparisons test when the standard deviations were markedly different. Significance was set at p < 0.05. For comparisons of tissue-derived iLCs, mean δ Cts ± SDs were determined, statistical significance was calculated by one-way ANOVA with Bonferroni multiple-comparisons correction, and results are expressed as numerical values relative to the level of GAPDH expression. For samples with no detectable mRNA, analysis was done, setting the δCt value to 10. Thus, the biological differences are likely to be underestimates.

Langerhans cells can be generated from peripheral blood monocytes by incubating monocytes with IL-4, TGFβ1, and GMCSF (27). Peripheral blood mononuclear cell–derived monocytes can be separated into classical, intermediate, and non-classical subpopulations by gating on CD14 and CD16 expression (Figure 1A). While the number of PBMCs/mL of peripheral blood and the percentage of monocytes in the PBMC were comparable between patients and controls (data not shown), the relative percentage of monocyte subpopulations differed between patients and controls (Figure 1B). The percentage of classical monocytes (CD14^bright, CD16^neg) from patients was significantly lower than that from controls (88.96 ± 5.69 vs. 78.45 ± 10.01, p < 0.05). In contrast, the percentage of intermediate (CD14^bright, CD16^dim cells) monocytes (5.28 ± 2.57 vs. 7.99 ± 3.37) and alternative (CD14^dim, CD16^bright) monocytes from controls and patients (5.5 ± 4.54 vs. 12.5 ± 7.92) were similar.

We then asked the potential of each monocyte subpopulation to differentiate into CD1a⁺, E-cadherin⁺ iLCs (Figure 1C). Almost all of the classical monocytes from controls differentiated into iLCs (96.5 ± 0.65%), fewer of the intermediate monocytes differentiated into iLCs (9.25 ± 2.21%), and very few of the alternative monocytes generated iLCs (2.75 ± 0.62%). While the patterns were very similar, significantly fewer of the classical monocytes from patients generated iLCs as compared to controls (61.6 ± 12.98%, p < 0.05), as did the intermediate monocytes (3.8 ± 0.8%, p < 0.05), while the difference with alternate monocytes was not significant due to the very small number of iLCs generated (3.0 ± 1.52%).

Previously, we showed that COX-2 is constitutively overexpressed in respiratory papillomas and the adjacent upper airway epithelium of RRP patients, leading to an increased synthesis of PGE₂ in these tissues (24). We have now found that the mean concentration of PGE₂ in the plasma of patients (690 ± 109 pg/mL) is also significantly higher than in plasma obtained from controls (372 ± 63 pg/mL) (p < 0.015) (Figure 2A). In addition, there was greater variability in plasma PGE₂ levels in patients compared with controls. Of note, the plasma PGE₂ levels in all patients (N = 6) studied in Figure 1C were above the mean value for both patients and controls. In addition, the patient with the lowest number of classical monocytes had the highest plasma level of PGE₂. Prostaglandin E₂ levels did not correlate with the severity of disease, calculated at the time of surgery for the RRP patients.

We therefore asked whether the higher PGE₂ plasma levels in patients with RRP might contribute to the reduced differentiation of their monocytes into iLCs, determining the effect of added PGE₂ on differentiation (Figure 2B). Monocytes from eight additional controls were compared to monocytes from three additional patients with high PGE₂ plasma levels (2,640, 1,560, and 1,090 pg/mL). Prostaglandin E₂ added to control monocytes suppressed their ability to differentiate into iLCs from 31.1 ± 4.52% to 17.9 ± 2.62% (p < 0.007). The patients' monocytes showed lower levels of iLC differentiation in the absence of added PGE₂ (Figure 2B), and differentiation did not decrease further when exogenous PGE₂ was added (16.4 ± 2.76% vs. 14.3 ± 1.76%, p < 0.76), suggesting that they were already maximally repressed.

We previously showed that day 7 monocyte cultures enriched with iLCs from controls expressed moderately high levels of CCL1 mRNA at baseline, which was further increased following IL-36γ stimulation (27). This inflammatory chemokine selectively binds CCR8 present on monocytes, macrophages, neutrophils, T_H2 T cells, Tregs, and other cells (32, 33). In contrast, the iLCs from RRP patients showed decreased levels of baseline CCL1 mRNA expression, which correlated with RRP disease severity. In contrast, the iLCs from patients showed increased baseline expression of the more anti-inflammatory T_H2-like chemokine CCL20 (27). We therefore asked whether PGE₂ altered the baseline mRNA expression of CCL1 and/or CCL20 by monocyte-derived iLCs. When control iLCs (n = 7) were pre-incubated with PGE₂, there was a trend toward a reduction in CCL1 mRNA expression and an increase in CCL20 mRNA expression. Although the change in CCL1 or CCL20 did not reach significance alone, the CCL1/CCL20 mRNA ratio for the control iLCs exposed to PGE₂ was significantly reduced (p < 0.02) (Supplementary Figure 1A). In contrast, the patients' CCL1/CCL20 expression ratio, which was low because CCL1 mRNA expression is poorly expressed by RRP patients' iLCs, did not change significantly in response to the addition of PGE₂ (Supplementary Figure 1B). This lack of response to PGE₂ is consistent with the observation that added PGE₂ did not suppress differentiation of monocytes from RRP patients to iLCs (Figure 2).

To determine whether PGE₂ also affected iLC maturation, identified by CD83 expression (34), we exposed monocyte-derived iLCs from patients and controls to PGE₂, IL-36γ (a proinflammatory cytokine constitutively expressed by papilloma cells) (28), a combination of these mediators, or LPS as a positive control. A representative flow analysis is shown in Figure 3A. Prostaglandin E₂ and IL-36γ both induced modest numbers of iLCs to express CD83. However, the combination of IL-36γ and PGE₂ strongly induced almost all of the iLCs to express CD83, similar to the effect of LPS. There was no difference between patients and controls in iLC maturation in response to any of these mediators (Figure 3B), but there was significant variation in CD83 expression by maturing iLCs from patient-derived LCs compared with controls under all culture conditions. Thus, unlike the suppressive effects of PGE₂ on control monocyte iLC differentiation, PGE₂, IL-36γ, and their combination did not alter iLC maturation once differentiation was completed.

We then asked whether expression of CCL1 and CCL20 mRNA by purified monocyte-derived iLCs from controls and patients differed from iLCs isolated from papilloma tissues (Figure 4). As previously reported (15), relative baseline CCL1 mRNA expression by monocyte-derived iLCs from controls (50.7 ± 27.7) was significantly higher than that from patients (6.31 ± 1.95), and both types of cells increased expression after IL-36γ treatment (controls 1,164 ± 319, patients 348 ± 97, p < 0.01) (Figure 4A). In contrast, iLCs from papillomas showed little, if any, CCL1 mRNA expression at baseline and very little increase after IL-36γ treatment. Baseline expression of CCL20 mRNA (Figure 4B) by monocyte-derived iLCs from controls (24.17 ± 15) and patients (68.9 ± 19.1) was significantly increased after IL-36γ treatment (controls 731 ± 206; patients (519 ± 92) (controls p < 0.01; patients p < 0.01). Unlike CCL1, papilloma-derived iLCs expressed very high levels of CCL20 mRNA at baseline that did not further increase following IL-36γ stimulation (410 ± 65 and 422 ± 80, respectively).

Finally, we asked whether papilloma-resident iLCs also differed from monocyte-derived iLC in their expression of other cytokines or chemokines in response to IL-36γ. However, repeated tries were unable to induce any measurable responses to either IL-36γ or to the related proinflammatory cytokine IL-1β. We therefore considered the possibility that iLCs were refractory to stimulation because of the immunosuppressive environment within papilloma tissues. To test this, we removed the iLCs from the papillomas and incubated them overnight in supporting medium containing GMCSF and TGFβ1. Then, we stimulated them with two strong iLC stimuli, the TLR3 agonist poly (I:C) or TNFα (28), and measured expression of several cytokines and chemokines, including IL-36γ mRNA, because we had previously shown that IL-36γ stimulation upregulates its own mRNA expression through a positive feedback loop in monocyte-derived iLCs and keratinocytes (27, 28). Other mucosal and skin-derived iLCs are of an embryonic origin, not monocytes (35–38), so we used iLCs isolated from surgical discards of abdominal skin and foreskin for comparison to the papilloma iLCs.

Immature LCs from papillomas and foreskin did not express detectable CCL1 mRNA expression at baseline, and abdominal skin iLCs expressed extremely low levels of CCL1 (Figure 5A). Overnight culture alone had no significant effect on expression by abdominal or foreskin iLC, but papilloma iLCs markedly upregulated CCL1 mRNA, approaching the expression levels of GAPDH (>200-fold increase compared to baseline levels, p < 0.001). These results suggest that iLCs from the papillomas appear “primed” to express CCL1 but are suppressed when in the immunosuppressive papilloma microenvironment (15, 27). Subsequent stimulation of papilloma iLCs with poly(I:C) or TNFα did not further increase their high level of CCL1 mRNA expression. Foreskin-derived iLCs expressed significant CCL1 after poly (I:C) and TNFα stimulation [97.9-fold (p < 0.01) and 17.2-fold (p < 0.05) respectively], whereas increases in CCL1 expression by abdominal skin iLCs after stimulation were not significant.

Immature LCs from papillomas expressed detectable levels of IL-36γ mRNA at the time of isolation from the tissues (Figure 5B), which was significantly different from the baseline expression by either abdomen or foreskin iLCs (p < 0.05) and might reflect the fact that they were exposed to IL-36γ peptide in vivo. Expression was further increased after overnight culture outside of papilloma tissue, although this increase was not significant because of the small number of samples. Immature LCs from the abdomen and foreskin expressed no IL-36γ mRNA at baseline, but did also upregulate expression following culture in the absence of their normal tissue microenvironments (14.1- and 114-fold, p < 0.05 and p < 0.01, respectively). None of the cells showed further significant increases with poly (I:C) or TNFα treatment.

The pattern of expression of the other cytokines/chemokines analyzed was quite similar for iLCs from all three tissues and generally quite high under all conditions (Supplementary Figure 2). Immature LCs from the papillomas and foreskin expressed high levels of CCL20 mRNA that were close to GAPDH levels at baseline, with no significant change after removal from their microenvironments or following stimulation with either poly(I:C) or TNFα (Supplementary Figure 2A). Abdominal iLCs made less CCL20 mRNA at baseline, but it was upregulated 27-fold (p < 0.01) to the levels of the other iLCs after removal from the tissue, with no further significant increase after stimulation. Immature LCs from the three different tissues also expressed high levels of IL-1β mRNA and TNFα under all conditions (Supplementary Figures 2B,C). It was notable that iLCs from papillomas expressed significantly more IL-1β at baseline than the abdomen and foreskin iLCs (52- and 8-fold, respectively, p < 0.05), but this dropped back down to the levels seen with the other cells after overnight culture (Supplementary Figure 2B). This high level of IL-1β message may not be of functional significance because posttranslational processing is required for the production of active IL-1β (39, 40).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.