Adenovirus-α-Defensin Complexes Induce NLRP3-Associated Maturation of Human Phagocytes via Toll-Like Receptor 4 Engagement

ABSTRACT

Intramuscular delivery of human adenovirus (HAdV)-based vaccines leads to rapid recruitment of neutrophils, which then release antimicrobial peptides/proteins (AMPs). How these AMPs influence vaccine efficacy over the subsequent 24 h is poorly understood. In this study, we asked if human neutrophil protein 1 (HNP-1), an α-defensin that influences direct and indirect innate immune responses to a range of pathogens, impacts the response of human phagocytes to three HAdV species/types (HAdV-C5, -D26, -B35). We show that HNP-1 binds to the capsids and redirects HAdV-C5, -D26, and -B35 to Toll-like receptor 4 (TLR4), which leads to internalization, an NLRP3-mediated inflammasome response, and interleukin 1 beta (IL-1β) release. Surprisingly, IL-1β release was not associated with notable disruption of plasma membrane integrity. These data further our understanding of HAdV vaccine immunogenicity and may provide pathways to extend the efficacy.

IMPORTANCE This study examines the interactions between danger-associated molecular patterns and human adenoviruses, and their impact on vaccines. HAdVs and HNP-1 can interact, and these interactions will modify the response of antigen-presenting cells, which will influence vaccine efficacy.

INTRODUCTION

The immunogenicity of adenoviruses (AdVs) in the context of infection of immunosuppressed individuals, therapeutic gene transfer, and vaccines is a pressing issue. With the occurrence of SARS-CoV-2, AdV-based (human and simian) vaccines are being administered on a global scale (1). Human AdVs (HAdVs) are ∼950 Å-diameter, nonenveloped proteinaceous particles containing an ∼36,000 (± 9,000) bp, double-stranded linear DNA genome. There are currently 7 HAdV species (A through G) and >100 types based on serology and phylogeny (http://hadvwg.gmu.edu). HAdVs typically cause self-limiting disease in the conjunctiva, or respiratory and/or gastrointestinal tracts in all populations regardless of health standards, but infections can be lethal in immunocompromised individuals (2). Due to their endemic nature at military training facilities, live HAdV types 4 and 7 have been used as vaccines since the 1950s to prevent severe respiratory illness in recruits (3). AdV-based vaccines have also been approved in a prime/boost regimen against Ebola virus in the EU and tested in more than 200 clinical trials (clinicaltrials.gov, May 2021).

In the context of natural infections, gene transfer, or vaccination, neutrophils and professional antigen-presenting cells (APCs) such as monocytes and dendritic cells (DCs) extravasate to the site of injection/infection and detect danger-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs). APCs differentiate between pathogens via pattern recognition receptors (PRRs) on their surfaces, in their endosomal compartments, or in the cytosol. PRRs include Toll-like receptors (TLRs) (of which the most thoroughly studied is TLR4), nucleotide-binding oligomerization domain (NOD)-leucine-rich repeats (LRR)-containing receptors (NLR) (of which NLRP3 is the most thoroughly studied to induce the inflammasome), retinoic acid-inducible gene 1-like receptors (RIG-1), and C-type lectin receptors. PRRs detect different PAMPs and DAMPs and hence, differentially affect the environment that is made up of proinflammatory cytokines, chemokines, and host defense proteins/peptides (HDPs). In some inflammatory environments, monocytes mature into DCs as they migrate to lymphoid tissues to initiate antigen-specific T- and B-cell responses.

Antimicrobial peptides (AMPs) are effector molecules of the innate immune system, and act via antibiotic-like properties against a broad array of infectious agents (4, 5), or by promoting the activation and maturation of some APCs. Neutrophils are a rich source of AMPs: ∼20% of their cytoplasmic content can be AMPs (6). The rapid release of HDPs acts as part of the initial response to a breach in tissue homeostasis (7). Lactoferrin, α-defensin, and cathelicidin (LL-37) are alarmins that also impact innate and adaptive immune responses by activating PRRs in APCs (4, 5). Additionally, in epithelial cells, α-defensins impair HAdV-C5, -A12, and -B35 infection by stabilizing an intrinsically disordered region of the capsid vertex, thus preventing the disassembly of the metastable HAdV capsid (8–10).

We recently reported how lactoferrin retargets HAdV-C5, -D26, and -B35 to TLR4 and initiates inflammasome assembly and interleukin 1 beta (IL-1β) release without loss of membrane integrity (11). The inflammasome is a multiprotein platform consisting of a pattern recognition receptor that promotes aggregation of ASC (apoptosis-associated speck-like protein containing a CARD), which recruits and auto-activates pro-caspase-1. Pro-caspase-1 auto-activation can be followed by removal of the N terminus of gasdermin D, which initiates the loss of plasma membrane integrity via pore formation (12). Canonical and noncanonical NLRP3 inflammasome formation is preceded by a transcriptional priming event, which is needed to produce inflammasome components and cytokines (13). In primary human monocytes, LPS-TLR4 engagement can induce an alternative inflammasome activation characterized by IL-1β release in the absence of a transcriptional priming event, inflammasome aggregation, and pyroptotic cell death (14).

Here, we expanded our understanding of how an α-defensin impacts innate immune sensing of HAdV-C5, -D26 and -B35, which were chosen based on their development as vaccine vectors (15–22). Mechanistically, HNP-1 binds HAdV-C5, -D26, and -B35 with nanomolar range affinities and increases HAdV uptake via a retargeting to TLR4 complexes which, in turn, induces NLRP3 inflammasome formation and release of IL-1β. This mechanism is strikingly similar to that of lactoferrin-bound HAdV capsids, which suggests a polyvalent function of TLR4-mediated detection of viruses.

RESULTS

HNP-1 binds to HAdV-C5, -D26, and -B35, and increases infection of monocyte-derived DCs.

HNP-1 binds the HAdV-C5 capsid through electrostatic and hydrophobic interactions at the juxtaposition of the penton base and fiber, and on hexon (8, 23). These interactions can reduce HAdV-C5 and HAdV-B35 infection of A549 cells (transformed human lung epithelial cells) (8–10). To broaden our understanding of these interactions, we quantified the affinity of human HNP-1 to HAdV-C5, -D26, and -B35 using surface plasmon resonance (SPR) analyses. We show that HNP-1 binds to HAdV-C5, -D26, and -B35 with dissociation constants (K_D) ranging from 37 to 76 nM (Fig. 1A and B) using a two-state association-dissociation reaction (Fig. 1C).

FIG 1.

Analyses of HNP-1 binding to HAdV-C5, -D26 and -B35 by SPR. (A) Representative sensorgrams of HNP-1-HAdV interactions: HAdV-C5 (red), -D26 (black), and -B35 (blue) were coupled to the sensor chip. (B) HAdV-C5, -D26, and -B35 were covalently coupled to a CM5 sensor chip and escalating doses of HNP-1 (6.25 to 200 nM) for K_D determination. Overlaid sensorgrams are shown. (C) K_D of HNP-1 for HAdV-C5, -D26, and -B35. RU = resonance units.

HNP-1 increases infection of human monocytes and monocyte-derived DCs.

The receptors for HAdV-C5, -D26, and -B35 partially overlap in human DCs (24): HAdV-C5 predominantly uses DC-SIGN (CD209) (25, 26), HAdV-B35 uses CD46 (27), and HAdV-D26 uses sialic acid-bearing glycans (28) and engages CD46 through a nonconventional interaction involving the hexon (29). We therefore tested the impact of HNP-1 on HAdV capsid uptake using replication-defective (ΔE1) vectors containing reporter genes. Transgene expression was used as a proxy for endocytosis, cytoplasmic transport, delivery of the capsid to the nuclear pore, and transgene detection. Because ratios of infectious to physical particles may vary between HAdV stocks, and receptor density and affinity affect the percentage of cells expressing a transgene, we selected doses to have ∼30% of DCs expressing the vector-encoded transgene when possible. Consistent with earlier reports (25), we found that HNP-1 increased (P ≤ 0.02) HAdV-C5 uptake by monocyte-derived DCs (Fig. 2A). Monocyte-derived DCs incubated with HAdV-D26- and -B35-HNP-1 complexes also more readily expressed the transgene (P ≤ 0.001) than did each HAdV alone (Fig. 2A, upper panel). The changes in median fluorescence intensity (MFI) and geometric mean of fluorescence intensity (gMFI) corresponded to a 1.6-to 4.1-fold and a 1.6- to 3.6-fold increase, respectively, for HAdV-C5, -D26, and -B35 (Fig. 2A, lower panel). HAdV-D26 showed the lowest changes in MFI and gMFI, and HAdV-B35 showed the highest. Because the kinetics of virus uptake occur over several hours, we also tested whether it was necessary to preincubate HNP-1 with the HAdVs. We found that preincubating HAdVs with HNP-1, adding HNP-1 to the cell medium before HAdV, and adding HNP-1 to the medium after HAdV all increased virus uptake (Fig. 2B).

FIG 2.

HNP-1 interacts with HAdVs and increases infection of monocyte-derived DCs. (A) Upper panel: representative flow cytometry profiles of monocyte-derived DCs incubated with HAdV-C5, -D26, and -B35 replication-defective vectors containing fluorescent protein expression cassette. Mock-treated DCs (gray), or DCs incubated with HAdV-C5, -D26, or -B35 in the absence (red) or presence (blue) of HNP-1. Lower panel: MFI and gMFI for the FP+ and FP- cells for different capsids ± HNP-1 and fold changes in the presence of HNP-1 for a representative experiment are shown. (B) Representative flow cytometry profiles of cells incubated with HAdVs ± HNP-1. DCs were mock-treated (gray) or incubated with HAdV-C5, -D26, or -B35 alone (red); HNP-1 complexed with HAdV (blue); HAdV for 30 min and then HNP-1 (green); or HNP-1 for 30 min and then the HAdV (purple). Fluorescence was analyzed 24 h postincubation. (C) Cumulative data using HAdVs alone or HAdVs complexed with HNP-1, IVIg, or both. Two-tailed paired t-tests were used for comparison of HAdV versus HAdV + HNP-1 (n = 25, gray). Samples in black (n = 3) were used for analyses between HAdV + HNP-1 versus HAdV + HNP-1 + IVIg. (D to E) Monocytes and monocyte-derived LCs were incubated with HAdVs ± HNP-1 and transgene expression was analyzed 24 h postincubation (n = 3, statistical analyses by two-tailed paired t tests).

IVIg, which contains a high level of anti-HAdV-C5 antibodies, does not prevent HAdV-C5 infection of monocyte-derived DCs; however, like HNP-1 and lactoferrin, it does increase infection (30, 31). To benchmark the effects induced by HNP-1, we compared HAdVs complexed with either IVIg or HNP-1. Generally, the impact of HNP-1 was similar to that of IVIg (Fig. 2C). We also combined HNP-1 and IVIg with each type to determine whether a synergistic effect occurred. However, IVIg did not increase further HNP-1-enhanced transgene expression by HAdV-C5, -D26, or -B35.

In addition to HNP-1-enhanced uptake by monocyte-derived DCs, we found that HNP-1 increased HAdV-C5 and -D26 uptake by monocytes but decreased HAdV-B35 uptake (Fig. 2D). By contrast, HNP-1 had no additive effect on HAdV-C5- or D26-mediated transgene expression by monocyte-derived Langerhans cells (LCs), whereas it increased HAdV-B35-mediated transgene expression in LCs, but not in monocytes (Fig. 2E). Together, these results demonstrate that HNP-1 binds to three HAdV species/types and that its presence tends to increase HAdV uptake by primary human APCs.

HAdV-HNP-1 complexes induce cytokine secretion and activation of phagocytes.

Individually, HAdVs and HNPs can promote the activation of DCs by directly engaging PRRs (4, 5, 7). To determine whether there is a synergy between HAdVs and HNP-1, we used a multiplex cytokine array to screen the response by human phagocytes. Since all components of the complexes can individually influence DC maturation, we provide individual baselines to identify the impact of the complex. In addition, we used lipoplysaccharide (LPS) as a TLR4 inducer to show the relative range of cytokines produced. We found that HAdV-D26 and -B35 induced a greater cytokine response than either HAdV-C5 or mock-treated monocyte-derived DCs (Fig. 3A and 4, columns on the left). HAdV-HNP-1 complexes increased the release of IL-1α and IL-1β (Fig. 3A, 2nd group of 4 columns) compared to HNP-1-treated monocyte-derived DCs. When comparing HAdV with the HAdV-HNP-1 complexes, the addition of HNP-1 induced increases in most of the cytokines tested (Fig. 3A, the 3 pairs of columns on the right). The marked effect of HNP-1 on HAdV-C5 may be due to the lower impact of HAdV-C5 on monocyte-derived DCs.

FIG 3.

HAdV-HNP-1 induced cytokine secretion and monocyte-derived DC maturation. (A) Monocyte-derived DCs were incubated with HAdV-C5, -D26, or -B35 ± HNP-1, and cytokine secretion in supernatants was assessed by Luminex. A blanked reference is located to the left of each set of columns. For the 4 HAdV columns on the left, the reference is mock-treated cells; for the middle HNP-1, columns the reference is HNP-1-treated cells; for the “paired” columns on the right (-C5, -D26, and -B35) the reference is HAdV-infected cells compared to HAdV-HNP-1 infected cells. LPS was used as a positive control for TLR4 activation. (B) IL-1β release induced by HAdV-C5, -D26, and -B35 ± HNP-1 at 4 h (red circles) and 24 h (black dots) postincubation (n = 5). As above, cells were treated with LPS and nigericin as controls. (C) HNP-1-mediated HAdV infection of monocytes was analyzed 24 h postincubation by flow cytometry (n = 3). (D) DCs were incubated with HAdV-C5, -D26, or -B35 ± HNP-1, IVIg, or HNP-1 + IVIg. IL-1β release was quantified at 24 h postincubation (n = 3). (E) DCs were incubated with HAdV-HNP-1 complexes for 24 h. Phenotypic maturation was assessed by CD86 surface levels using flow cytometry. (F) DCs were incubated at 4°C or 37°C for 30 min with 1 mg/mL TRITC-labeled dextran, washed with PBS, fixed with 4% PFA, and analyzed by flow cytometry (lower fluorescence = lower phagocytosis = greater maturation) (n = 3). Statistical analyses were performed by two-tailed paired t tests unless otherwise noted.

FIG 4.

HAdV-HNP-1 complexes engage TLR4 in monocyte-derived DCs. (A) DCs were incubated with HAdV-HNP-1 complexes ± TAK-242 and analyzed by flow cytometry for vector-mediated transgene expression (n = 13). (B) IL-1β release from HAdV-HNP-1 complex-challenged monocyte-derived DCs, ± TAK-242 treatment, was quantified by ELISA (n = 3, Student's t test). (C) TNF secretion by monocyte-derived DCs following challenge with HAdV-HNP-1 complexes ± TAK-242 (n = 3). (D) Percentage of monocyte-derived DCs expressing the transgene following challenge with HAdV-HNP-1 complexes ± oxPAPC or Pepinh-TRIF (n = 6). (E) IL-1β release by monocyte-derived DCs following challenge by HAdV-HNP-1 ± oxPAPC or Pepinh-TRIF (n = 3). (F) HAdV-HNP-1 complexes were created with recombinant TLR4, TLR4/MD-2, MD-2 proteins, or anti-CD14 antibody. Recombinant protein/HAdV ± HNP-1 was then added to the DCs. Vector-mediated transgene expression was analyzed 24 h postincubation (n = 5). Statistical analyses were performed by two-tailed paired t tests unless otherwise noted.

We then quantified IL-1β release. The HNP-1-HAdV complexes induced greater IL-1β release than each HAdV alone, increasing from 4 to 24 h poststimulation (P ≤ 0.017 and P ≤ 0.013, respectively) (Fig. 3B). Monocytes, which typically generate low levels of IL-1β, released 2- to 3-fold more IL-1β when challenged with HAdV-HNP-1 compared to HNP-1 alone, but this increase was not statistically significant (P > 0.05) (Fig. 3C). To characterize the impact of HNP-1 on monocyte-derived DCs, we compared its effect to that induced by IVIg-complexed HAdVs. We found that IVIg-complexed HAdV induced modestly more IL-1β release than HNP-1-complexed HAdVs (Fig. 3D). When IVIg and HNP-1 were combined with HAdVs, a minor increase was also seen (Fig. 3D).

We then tested whether cytokine release was associated with phenotypic or functional maturation of DCs. Using CD86 surface level and changes in phagocytosis as readouts, we found that HNP-1 increased DC maturation when combined with HAdV-C5 and -D26 (Fig. 3E and F). By contrast, HAdV-B35 induced DC maturation in the absence of HNP-1 to similar levels as LPS, and HNP-1 addition did not further augment DC maturation (Fig. 3E–F). Together, these data demonstrate that HNP-1 increased the innate immune response to HAdV-C5 and -D26 and led to functional, phenotypic maturation of monocyte-derived DCs. By contrast, the response to HNP-1 + HAdV-B35 was conspicuously lower, and may provide lines to explore the surprising lack of HAdV-B35 antibodies found in all cohorts tested (32). Finally, these data also mirror the increased efficacy of HAdV-associated transgene expression induced by HNP-1.

HAdV-HNP-1 complexes induce monocyte-derived DC maturation through TLR4 engagement.

As α-defensins are structurally similar to β-defensins, which can activate DCs in a TLR4-dependent manner (33), we explored the possibility that the impact of HNP-1 on the HAdVs was associated with TLR4 engagement. We therefore used TAK-242, a cell-permeable cyclohexene-carboxylate, to disrupt TLR4-TIRAP and -TRAM interactions, which influence NF-κB activation (34–36). After TLR4 engagement, its intracellular TIR domain recruits TIRAP and MyD88. This interaction then activates NF-κB, AP-1, and IRF via IRAK and MAPK family members. NF-κB, AP-1, and IRF initiate the transcription of genes coding for proinflammatory cytokines and inflammasome components. We found that TAK-242 reduced (P ≤ 0.019) HNP-1-enhanced uptake of HAdV-C5, -D26, and -B35 (Fig. 4A). The influence of TAK-242 on IL-1β release (Fig. 4B) reflected the infection assay: we found a reduction (P ≤ 0.045) postincubation with HAdV-C5- and -B35 + HNP-1 challenge, but this reduction was less prominent for HAdV-D26. Of note, preincubation with TAK-242 reduced extracellular IL-1β levels to background levels. Consistent with other reports, TAK-242 also reduced (P ≤ 0.05) tumor necrosis factor (TNF) secretion induced by HNP-1-HAdV complexes and LPS (Fig. 4C).

To evaluate the role of the extracellular TLR4 domain in HNP-1-enhanced uptake, we used oxPAPC, which disrupts TLR4-MD2 interactions. oxPAPC reduced infection by HAdV-C5- and -D26-HNP-1 complexes (P < 0.043), but not infection by -B35 (Fig. 4D). We then used Pepinh-TRIF to inhibit cytoplasmic TLR4-TRIF interactions and found that Pepinh-TRIF had no effect on HAdV-C5 and -D26, but surprisingly increased (P < 0.0007) HNP-1-HAdV-B35-mediated transgene expression (Fig. 4D). Additionally, oxPAPC and Pepinh-TRIF decreased (P ≤ 0.037) IL-1β release induced by HNP-1-HAdV complexes (Fig. 4E). Combined, these data suggest that TLR4 signaling is involved in HNP-1-enhanced infection and IL-1β and TNF release.

We then addressed TLR4 engagement using another approach. MD-2 can form a complex with TLR4, with the former acting as a coreceptor for exogenous and endogenous ligands (37), or with CD14 for ligand internalization (38). We therefore tested whether TLR4 engagement could be influenced by recombinant TLR4, MD-2, or TLR4/MD-2 dimers, or by a CD14-blocking antibody. Using this approach, only recombinant MD-2 notably modified HNP-1-enhanced infection by increasing (P ≤ 0.006) HAdV-C5 infection. Recombinant TLR4 trended toward an increase in HAdV-D26 infection (Fig. 4F); however, the increase was not significant (P > 0.05). In these assays, we found no significant differences on HAdV-B35 + HNP-1 uptake by DCs.

Together, these data are consistent with TLR4 complex engagement and intracellular TLR4 signaling during the sensing of HNP-1-complexed HAdV-C5, -D26, and -B35.

HAdV-HNP-1 complexes induce NLRP3 inflammasome formation.

Canonically, there is a 2-step pathway in classic and alternative inflammasome activation. The initial step induces the transcriptional and translational upregulation of inflammasome components, and the second step is inflammasome formation. We therefore used reverse transcriptase quantitative PCR (RT-qPCR) to determine whether engagement of TLR4 induced the transcription of NLRP3, CASP1, and IL1B mRNAs. HNP-1-complexed HAdVs induced higher (P ≥ 0.012) levels of all three mRNAs compared to HAdVs alone. However, the increase was only higher than HNP-1 alone in the case of NLRP3 by HNP-1-HAdV-B35 (Fig. 5A–C), suggesting that HNP-1-induced levels were near saturation.

FIG 5.

HNP-1 induces transcription of inflammasome components. (A to C) mRNA levels of selected inflammasome components (encoded by NLRP3, CASP1, and IL1B) were analyzed using qRT-PCR 4 h postincubation. Total RNA was isolated from monocyte-derived DCs incubated with the indicated HAdV ± HNP-1; cDNAs were generated and quantified in triplicate. GAPDH mRNA levels were used to standardize samples. (D) Prior to challenge with HAdV-HNP-1 complexes, monocyte-derived DCs were treated with KCl, NAC, or MDL (n = 6). IL-1β release was quantified at 24 h postincubation. (E) Percentage of monocyte-derived DCs expressing vector-encoded transgene following KCl, NAC, or MDL (n = 3) pretreatment and subsequent challenge with HAdVs ± HNP-1. Statistical analyses were by two-tailed paired t tests unless otherwise noted.

To identify the HAdV-HNP-1-associated trigger(s) of inflammasome induction, monocyte-derived DCs were pretreated with KCl (to prevent K⁺ efflux), N-acetyl-l-cysteine (NAC, a reactive oxygen species [ROS] scavenger), and MDL (to inhibit endosomal protease-mediated activation of the NLRP3 inflammasome after endosome lysis) (12). LPS, in the presence or absence of the NLRP3 inducer nigericin, was used as a control. We found that only the addition of extracellular MDL significantly (P < 0.05) decreased the release of IL-1β for HAdV-C5, -D26, and -B35 (Fig. 5D). Adding K⁺ or NAC increased the number of DCs that expressed the transgene for HAdV-C5 and -D26 (Fig. 5E). Of note, following receptor-mediated endocytosis by epithelial-like cells, HAdVs escape from early endosomal compartments via the endosomolytic activity of protein VI (39). This rapid escape before the capsid can be degraded shields the viral genome from detection by innate sensors. However, HAdV particles that are covered with anti-HAdV IgGs are more efficiently taken up by phagocytes via FcγRs (30, 31). In phagocytes, the acidification of endosomes and degradation of internalized cargo is slower, allowing the endocytic vesicle to fuse with lysosome-like vesicles (40). The massive HAdV capsid uptake, followed by the activity of protein VI, leads in turn to the rupture of endosomal compartments and the release of proteases. Therefore, the data showing that MDL reduces IL-1β release are consistent with the idea that endosomal rupture caused by HNP-1-enhanced uptake and cathepsin B release trigger inflammasome formation.

We then used small molecule inhibitors to address the crosstalk between TLR4-induced signaling and NLRP3 induction in the presence of HNP-1-HAdV complexes. DCs were preincubated with R406 (blocking MyD88-Syk interactions and interrupting the signaling between TLR4 engagement and NLRP3 inflammasome regulation), or Bay11-7082 (inhibitor of NLRP3 [41]). We found that IL-1β release was reduced (P ≤ 0.038) in response to HNP-1 complexes in all cases, with the exception of HNP-1-HAdV-D26 (Fig. 6A). Bay11-7082 and R406 did not significantly impact HNP-1-enhanced uptake (except for HAdV-B35) (Fig. 6B). Importantly, MCC950, an NLRP3 inhibitor, also significantly (P ≤ 0.004) reduced IL-1β release induced by HNP-1 + HAdV-D26 and -B35, but not that induced by HNP-1 + HAdV-C5 (Fig. 6C).

FIG 6.

TLR4-induced signaling and NLRP3 induction in monocyte-derived DCs. (A) DCs were treated with R406 or Bay11-7082 prior to challenge with HAdVs ± HNP-1. IL-1β release was quantified 24 h postchallenge by ELISA. LPS/nigericin was used as a control for DC activation. (B) Monocyte-derived DCs were treated with R406 or Bay11-7082 prior to challenge with HAdVs ± HNP-1. The percentage of DCs expressing the reporter gene was quantified by flow cytometry. Cumulative data from 5 donors are shown. (C) Monocyte-derived DCs were treated with MCC-950 prior to challenge with HAdVs ± HNP-1. IL-1β release was quantified 24 h postchallenge by ELISA. Statistical analyses were by two-tailed paired t tests unless otherwise noted.

Upstream of IL-1β release is the activation of selected caspases, in particular caspase-4 and -5 following TLR4 engagement (12–14) and caspase-8 for gasdermin D cleavage. We therefore inhibited selected caspases using VX765 (caspase-1 and -4), Z-IETD (capsase-8), WEHD (caspase-1, -4, -5, and -8), and Z-YVAD-FMK (caspase-1, -4, and -5). We found that all of these inhibitors reduced IL-1β release to some extent (Fig. 7A to C). Moreover, Z-YVAD-FMK also decreased TNF levels induced by HAdV-HNP-1 complexes (Fig. 7D).

FIG 7.

NLRP3 inflammasome is downstream of TLR4 engagement. To identify the pathway of IL-1β production, we inhibited selected caspases using (A) WEHD, (B) YVAD and VX765, and (C) Z-IEDT. IL-1β release was quantified 24 h postchallenge by ELISA. (D) DCs were preincubated with inhibitors of caspases 1/4/5 (YVAD and VX765) and then challenged with HAdV ± HNP-1. Extracellular TNF levels were quantified 24 postchallenge. (E to F) Transgene expression and IL-1β release were quantified in response to monocyte-derived DCs pretreated with RIPK1-RIPK3 inhibitors (GSK963, necrosulfonamide, or GSK872) and then incubated with HAdV-HNP-1 complexes (n ≥ 3). (G) Combined analyses of IL-1β release broken down by each HAdV type. Statistical analyses were by two-tailed paired t-tests unless otherwise noted.

In the alternative inflammasome pathway, activation of the NLRP3 inflammasome depends on signaling through TLR4 and RIPK1 (receptor interacting serine/threonine kinase 1), whereas the involvement of RIPK3 (receptor interacting serine/threonine kinase 3) and MLKL (mixed lineage kinase domain-like) downstream of RIPK1 are not required (14). Consequently, we used GSK963 (RIPK1 inhibitor), GSK872 (RIPK3 inhibitor), and necrosulfonamide (MLKL inhibitor) to investigate whether the alternative inflammasome pathway was involved. Inhibition of the three kinases had no significant effect on infection (Fig. 7E). By contrast, GSK963 reduced (P ≤ 0.016) IL-1β release following challenge with HAdV-C5 and -B35 + HNP-1, while necrosulfonamide reduced the impact of HNP-1 on HAdV-D26 and -B35 uptake (P ≤ 0.007) (Fig. 7F and G).

Globally, the HAdV-HNP-1 complexes induced NLRP3 inflammasome formation and IL-1β release via the activation of the TLR4 pathway. However, there appeared to be mechanistic differences between the HAdV types with respect to the pathways engaged: e.g., HAdV-C5 signaled through RIPK1, while HAdV-D26 and -B35 did not. By contrast, inhibition of the MLKL pathway reduced IL-1β release induced by HNP-1 + HAdV-D26 and -B35 uptake, but not for -C5.

IL-1β release and membrane integrity.

These data incited us to investigate the activation of the alternative inflammasome pathway (14). To determine whether HAdV-HNP-1 complexes induced the release of cytoplasmic proteins, we quantified extracellular activity of l-lactate dehydrogenase (LDH) at 4 h postincubation. We found no notable (P > 0.05) increase in any condition (Fig. 8A). To determine whether HAdV-HNP-1 complexes impacted membrane integrity, we added 7-AAD and quantified its uptake by DCs. We found that the percentage of 7-AAD⁺ cells induced by HAdV-HNP-1 complexes was greater (P ≤ 0.044) than HNP-1- or HAdV-challenged cells (Fig. 8B). Additionally, when HNP-1 was added to HAdV-IVIg complexes, a modest increase in the percentage of 7-AAD⁺ cells was detected (Fig. 8C). Therefore, within the time frame of our assays, holes large enough to allow small molecule entry were created but LDH was not released into the medium, suggesting that pyroptosis was not efficiently induced.

FIG 8.

IL-1β release without loss of monocyte-derived DC membrane integrity. (A) DCs were incubated with HAdV-C5, -D26, or -B35 ± HNP-1, or with LPS/nigericin. Extracellular LDH activity was quantified 4 h postincubation (n = 6). (B) 7-AAD uptake by monocyte-derived DCs, after challenge with HAdV-HNP-1 complexes, 24 h postincubation (n = 9). (C) Monocyte-derived DCs were incubated with HAdV-C5, -D26, or -B35 ± HNP-1, IVIg, or HNP-1 + IVIg. 7-AAD uptake was quantified 24 h postincubation (n = 3). Statistical analyses were by two-tailed paired t tests unless otherwise noted.

DISCUSSION

We used biochemical, pharmaceutical, cellular, and molecular approaches to demonstrate that HNP-1 binds to the HAdV capsids and enhances the infection of primary human phagocytes through the engagement of TLR4 to induce proinflammatory cytokines via the NLRP3 inflammasome.

The most striking features of our study using HNP-1 are the similarity to the impact of lactoferrin on HAdVs (11) and the opposing effect that HNP-1 has on HAdV uptake by APCs versus epithelial-like cells. Both HNP-1 and lactoferrin induced a retargeting of three HAdV types to TLR4 complexes. Given this modus operandi, we would not be surprised to find that high mobility group box 1, a DNA-binding nuclear protein that can act also as an alarmin (42), also redirects HAdV to TLR4. By acting as a direct and indirect pattern recognition receptor, TLR4 has polyvalent roles which may be central to understanding innate immune responses to numerous viruses and HAdV-based vaccines. Like lactoferrin, HNP-1-enhanced HAdV uptake, which induced a response similar to the alternative NLRP3 inflammasome pathway (11). The alternative pathway is characterized by the activation of TLR4-TRIF-RIPK1-FADD-CASP8 cascade and the absence of transcriptional priming, K⁺ efflux, and pyroptosome formation (14). While some of these key features are well aligned with our data, it is notable that the inhibition of RIPK1 only partially abrogated IL-1β secretion, and that MLKL kinase may be involved in the response to HAdV-D26 and -B35. These differences may be due, in part, to the different cell types used in the studies (14), which express different levels of CD14 that may affect TLR4 stoichiometry and signaling. Equally important, LPS and HAdVs substantially differ in size, molecular composition, and functional factors that affect intracellular processing and cytosolic PAMP detection. TLR4-mediated endocytosis of LPS involves sequential interaction of CD14 and MD2 with LPS and depends on the homodimerization of TLR4. The ∼950-Å icosahedral HAdV capsid allows multiple HNP-1 molecules to bind the capsid and induce TLR4 dimerization/multimerization directly.

Quintessential TLR4 agonists do not have complex intracellular processing. However, HAdVs traffic through endosomal compartments until the pH drops sufficiently to allow capsid escape into the cytosol. When a HAdV capsid is covered with neutralizing antibodies, lysosomal degradation is more efficient, the viral genome is partially exposed, and the DNA is then detectable via TLR9 (30, 43). Crosstalk between TLRs and synergistic effects on IL-1β secretion has been previously reported in human DC in vitro (44) and in mice (45, 46). Furthermore, hierarchical dependency of cell surface receptors (TLR4) and intracellular PRRs in phagocytes can affect the intracellular processing of the “ligand” (47). Notably, TLR4 engagement channels its ligands, in a TRIF-dependent manner, into specialized intracellular compartments (48), and intracellular trafficking of a PAMP through endosomal and lysosomal compartments and the cytosol can lead to differential signaling and the production of type I IFN, TNF, and IL-1β (49, 50). In addition, HNP-1 increases the sensitivity of human plasmacytoid DCs to TLR9 ligands, and thus promotes the induction of NF-κB and IRF1 response genes (51). Our assays indicate that TLR4 signaling through MyD88 is required for HNP-1-enhanced HAdV uptake, whereas signaling through MyD88 and TRIF is required for the induction of the NLRP3 inflammasome. While we did not investigate how HAdV processing occurs downstream of TLR4-mediated uptake, it is possible that the HNP-1-dependent increase in transgene expression coincides with increased amounts of intracellular viral DNA and disrupted intracellular vesicles. It is possible that the higher levels of viral DNA stimulate TLR9 or other PRRs and account for the differences seen between the alternative inflammasome pathway and our data. Of note, our data also indicate that cathepsin B release from endosomal compartments, concomitant with HAdV endosomal escape, induces NLRP3 induction and IL-1β secretion, which may further add to the differences seen here versus those reported by Gaidt et al. (14).

Based on their efficacy as vaccines and differences in receptor use, it was not surprising to observe similarities and differences between HAdV types. Notably, HAdV-B35 also had the strongest inflammatory response even though it was used at the lowest dose (1,000 pp/cell), which may be attributed to the engagement of CD46 and the TLR4-NLRP3 axis described here. It is worth noting that CD46 engagement can induce the formation of NLRP3 inflammasome in human CD4⁺ cells (52). We speculate that the individual attachment receptors for HAdV-C5, HAdV-D26, and HAdV-B35, and their stoichiometry, are involved in the differences found here. Notably, CAR is not expressed on most human phagocytes, and crosstalk between TLR4 signaling with DC-SIGN or CD46, respectively, to induce proinflammatory responses has been described and likely accounts for the differences between the HAdV types and phagocytes (53–55).

Nemerow and colleagues showed that α- and β-defensins inhibit HAdV infection of epithelial cells (8), while we showed that HNP-1 enhances the infection of HAdV by primary human monocytes, DCs, and to some extent LCs. At this time, the most parsimonious explanation is that, initially, the higher levels of cell surface TLR4 on phagocytes bring in more HAdV particles, some of which escape HNP-1 neutralization. Additionally, receptor-mediated uptake and endosomal trafficking, acidification, and fusion with lysosomes differ significantly between epithelial cells and phagocytes (40, 56). While in epithelial cells, species C HAdVs need to escape quickly from early endosomes to circumvent degradation, this is unlikely to be the case for phagocytes when they are swamped by HAdV particles entering via many pathways. Moreover, as mentioned earlier, acidification and fusions of endosomes, and degradation of cargo in phagocytes are slower to allow cross-presentation. This trafficking may therefore circumvent transient HNP-1 neutralization of capsid disassembly.

Greater uptake by phagocytes may indicate that defensins redirect HAdVs from infection of epithelial cells to innate immune cells to initiate an antiviral response. Because HNP-1 can stimulate monocyte, DC, and T-cell recruitment (33, 57–59), this synergistic effect likely plays a role in the downstream adaptive response. By further exploring these early events, we address the mechanisms by which an α-defensin modifies innate and adaptive immune responses to HAdVs. In our model, HAdV-HNP-1 induced signals are likely received immediately prior to NLRP3 engagement. We hypothesize that the coordination of priming and de-ubiquitination of NLRP3 (60), as well as TNF survival signals (61), favor adaptive cellular and humoral immune responses. Recently, moderate NLRP3 activation in human conventional type 2 DCs induced a hyperactivated phenotype characterized by secretion of IL-1β and IL-12 family cytokines in the absence of pyroptosis, and was followed by strong induction of Th1 and Th17 responses (62). Of note, HNP-1 also enhanced HAdV-induced IL-12-p40 secretion (Fig. 3A). We speculate that the involvement of the HAdV-HNP-1-TLR4-NLRP3 axis may help drive the T-cell responses toward a Th1 phenotype. Clinical trials indicate that Ad26.COV2.S, an HAdV-D26-based vaccine, induces potent Th1 responses against the SARS-CoV-2 spike (63, 64). It is conceivable that HNP-1 acts as a natural adjuvant, increasing the innate response and ultimately the breadth of the adaptive response to intracellular pathogens. In addition, the control of this inflammatory response may be due to the production of IL-1α, which promotes cell survival (65).

In conclusion, we examined the multilayered innate immune response of DCs/phagocytes in vitro, which may have implications for HAdV-based vaccines and pathogenesis (66).

MATERIALS AND METHODS

Cells and culture conditions.

Blood samples were purchased from the regional blood bank (EFS, Montpellier, France). An internal review board approved their use. DCs were generated from freshly isolated or frozen CD14⁺ monocytes (30). DC stimulation was performed 6 d postisolation. Monocyte-derived LCs were generated using 200 ng/mL granulocyte-macrophage colony-stimulating factor and 10 ng/mL TGF-β. 911, and 293 E4-pIX cells were grown in Dulbecco’s modified Eagle medium, minimum essential medium with Earle’s salts, and l-glutamine supplemented with 10% fetal bovine serum (FBS).

Replication-defective vectors.

HAdV-C5 contained a green flourescent protein (GFP) expression cassette (67). HAdV-D26 contained a GFP-luciferase fusion expression cassette (17). HAdV-B35 contained a yellow flourescent protein (YFP) expression cassette (8). Vectors were propagated in 911 or 293 E4-pIX cells and purified by density gradients (67). For infection assays we used HAdV-C5 (5,000 physical particles [pp]/cell), -D26 (20,000 pp/cell), or -B35 (1,000 pp/cell). The samples were collected 24 h later and prepared for flow cytometry, and 25,000 events were acquired/sample.

DC stimulation with HAdV-HNP-1 complexes.

DCs (4 × 10⁵ in 400 μL of complete medium) were incubated with HAdV-C5, -D26, or -B35 (0.1 to 2 × 10⁴ pp/cell). HAdV-HNP-1 complexes were generated by incubating HAdVs with 1.40 μg HNP-1 (Sigma-Aldrich) (68, 69) for 30 min at room temperature. When noted, cells were complexed with IVIg (human IgG pooled from between 5,000 and 50,000 donors/batch) (Baxter SAS). Cells were incubated with HAdV-HNP-1 for 4 h, then washed and incubated for 24 h. LPS (Sigma-Aldrich) and NLRP3 inflammasome inducer nigericin (InvivoGen) were used at 100 ng/mL and 10 μM, respectively. Inhibitor concentrations used are as follows: TAK-242 (Merck Millipore) at 1 μg/mL, oxPAPC (InvivoGen) at 30 μg/mL, TRIF inhibitory peptide (InvivoGen) at 25 μM, Syk inhibitor R406 (InvivoGen) at 5 μM, KCl (Sigma-Aldrich) at 40 mM, N-acetyl-l-cysteine (NAC) (Sigma-Aldrich) at 2 mM, MDL 28170 (Tocris Bioscience) at 0.1 mM, MCC-950/CP-456773 (Sigma-Aldrich) at 10 μM, Bay11-7082 (Sigma-Aldrich) at 10 μM, WEHD (Santa Cruz) and YVAD (InvivoGen) at 20 μM, VX765 (InvivoGen) at 10 μM, caspase-8 inhibitor Z-IEDT at 20 μM, GSK963 (Sigma-Aldrich) at 3 μM, GSK872 (Merck Millipore) at 3 μM, and necrosulfonamide (R&D systems) at 1 μM. TLR4/MD-2, TLR4 (R&D Systems), MD-2 (PeproTech) recombinant protein, and anti-CD14 antibody (Beckman) were used at 20 μg/mL. Inhibitors were added on cells, and recombinant proteins or antibody were added on HAdV-HNP-1 complexes, 1 h before stimulation.

Surface plasmon resonance analyses.

SPR analyses were performed on a BIAcore 3000 apparatus in HBS-EP buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.005% vol/vol polysorbate 20 [pH 7.4]). HAdV-C5, -D26, and -B35 diluted in acetate buffer (pH 4) were immobilized on three flow cells of a CM5 sensor chip by amine coupling. Immobilization levels were between 3,500 and 4,000 resonance units (RU). A blank flow cell was used as a control. HNP-1 was injected at 100 nM on the 4 flow cells simultaneously. To determine the K_D, 6.25 to 200 nM HNP-1 was injected at 30 μL/min during 180 s association and 600 s dissociation with running buffer. Regeneration was performed with Gly-HCl (pH 1.7) pulses. The kinetic constants were assessed from the sensorgram after double-blank subtraction with BIAevaluation software version 3.2 (GE Healthcare) using a 2-state fitting model. All experiments were repeated at least twice for each vector on freshly coated flow cells.

Flow cytometry.

GFP or YFP expression from the HAdV-C5, -B35, and -D26 vectors was assayed by flow cytometry. CD86 surface expression level was assessed with an anti-CD86 antibody (clone 2331, APC, BD Biosciences). For dextran uptake assays, we used 1 mg/mL for 30 min at 37°C, or 4°C for the negative control (TRITC-dextran; Sigma-Aldrich). Cell membrane integrity was assessed by collecting cells by centrifugation at 800 × g, and the cell pellets were resuspended in phosphate-buffered saline (PBS), 2% FBS, 1 mM EDTA, and 7-aminoactinomycin D (7-AAD) (Becton, Dickinson Pharmigen) and analyzed on a FACS Canto II (Becton, Dickinson Pharmigen) or NovoCyte (ACEA Biosciences) flow cytometer.

Cytokine secretion and LDH release.

Supernatants were collected 4 or 24 h postchallenge, and levels of TNF and mature IL-1β were quantified by ELISA using an OptEIA human TNF ELISA Set (BD Biosciences) and human IL-1β/IL-1F2 DuoSet ELISA (R&D Systems). Twenty-two cytokines were detected using a Bio-Plex human chemokine cytokine kit (Bio-Rad). LDH release was quantified using an LDH Cytotoxicity assay kit (Thermo Fisher Scientific) and as previously described (11).

Quantification of NLRP3, CASP1 and IL1B mRNAs.

mRNA levels were analyzed using qRT-PCR as previously described (11).

Data availability.

All data generated or analyzed during this study are included in the published article.