Peripheral blood was collected from healthy donors at the University of Iowa and used as the source of peripheral blood mononuclear cells (PBMCs) for these studies, as approved by the University of Iowa Institutional Review Board (Biomedical IRB-01). Donors were between the age of 18–75 and both sexes were represented, with no preference.

PBMCs were isolated from peripheral blood by standard density gradient centrifugation using Histopaque-1077 (Sigma-Aldrich) and removal of red blood cells by RBC lysis buffer (Biolegend). For culture, collected PBMCs were suspended in complete RPMI 1640 media supplemented with 10% FBS, 1X GlutaMAX, 1mM Sodium Pyruvate, 10mM HEPES (Thermo Fisher Scientific), 50µM 2-Mercaptoethanol (Sigma-Aldrich) and 50 μg/ml Gentamycin (IBI Scientific), added to 96-well U-bottom culture plates at a fixed amount (2 x 10⁵ in 200μL per well) and treated with the outlined reagents for 2 or 20 hours at 37 °C in a humidified atmosphere with 5% CO₂.

Vidu and recombinant human anti-Qβ antibody were produced and provided by Checkmate Pharmaceuticals (now Regeneron). Fluorescently labeled Vidu was generated using the Invitrogen Alexa Fluor 647 Protein Labeling Kit (Thermo Fisher Scientific). Vidu (AF647 labeled or unlabeled) was used at a final protein concentration of 5 µg/mL and recombinant human anti-Qβ was used at a final concentration of 1, 5, 20 or 80 µg/ml, as noted in figure legends. Monoclonal REAfinity^® recombinant human antibodies targeting pDC surface antigens BDCA2 (Miltenyi, cat. #130-124-317) and BDCA4 (Miltenyi, cat. #130-124-318) were evaluated for their ability to inhibit binding and/or uptake of anti-Qβ-coated Vidu. FcγRs were blocked using polyclonal goat antibodies against human CD16 (R&D, cat. #AF1597), CD32 (R&D, cat. #AF1330) or CD64 (R&D, cat. #AF1257) (16, 33). Miltenyi REAfinity^® recombinant antibodies were selected because they are engineered to reduce undesirable binding to FcγRs. Goat antibodies were selected because their Fc region has low affinity for human FcγRs (34), thus limiting off-target effects. All blocking antibodies were used at a final concentration of 1 μg/ml. Species- and vendor-matched isotype control (IC) antibodies were included in blocking experiments. Prior to the addition of anti-Qβ and Vidu, antibodies against surface receptors were pre-incubated with cells for a minimum of 15 minutes and a maximum of 30 minutes at 37 °C, based on previously reported BDCA2 and CD32 blocking studies performed with PBMCs (35, 36).

PBMC (2 x 10⁵ per well) were cultured in 96-well U-bottom plates with treatments for 2 or 20 hours. Cells were collected and stained with Zombie Violet Fixable Viability Dye (Biolegend, cat. #423114) according to manufacturer’s protocol to distinguish live from dead cells. Following this, cells were surface stained with antibodies against human CD3 (Biolegend, cat. #300336), CD4 (Biolegend, cat. #344656), CD8 (Biolegend, cat. #301038), CD11b (Biolegend, cat. #301322), CD11c (Thermo Fisher, cat. #46-0116-42), CD14 (BD, cat. #563561), CD15 (Biolegend, cat. #323033), CD16 (Biolegend, cat. #302032), CD19 (BD, cat. #749173), CD32 (Biolegend, cat. #303215), CD45 (Biolegend, cat. #368542), CD56 (Biolegend, cat. #318351), CD64 (Biolegend, cat. #305027), CD80 (BD, cat. #751732), CD123 (Biolegend, cat. #306054), BDCA2 (Biolegend, cat. #354207), BDCA4 (Biolegend, cat. #354504), HLA-DR (Biolegend, cat. #307640) and PD-L1 (Biolegend, cat. #329736). Human TruStain FcX Fc Receptor Blocking Solution (Biolegend, cat. #422302) and True-Stain Monocyte Blocker (Biolegend, cat. #426103) were included during all surface staining steps. Stained samples were acquired on a Cytek Aurora CS Spectral Flow Cytometer (University of Iowa Flow Cytometry Facility) using SpectroFlo software. Data was analyzed using FlowJo v10 software (BD Biosciences). Dead (Zombie Violet positive), CD45 negative, and aggregated cells were excluded from analysis. Positive and negative staining gates were determined with FMO controls. pDCs were identified by the following staining pattern: 1) negative for CD11b, CD11c, CD14, CD15, CD19, CD3, CD56 and 2) positive for BDCA4, CD123, and BDCA2 (when not being blocked with antibody).

PBMC (2 x 10⁵ per well) were cultured in 96-well U-bottom plates with treatments for 20 hours. Supernatants were collected and IFNα was measured with the VeriKine Human IFNα ELISA Kit using the extended standard range (PBL Assay Science, cat. #41100-2) according to the manufacturer’s protocol. ELISA plates were read on a ThermoMax Precision microplate reader using SoftMax^® microplate data acquisition software (Molecular Devices).

PBMC (2 x 10⁵ per well) were cultured in 96-well U-bottom plates with treatments for 20 hours. All samples were collected and stained with Zombie Violet Fixable Viability Dye (Biolegend, cat. #423114) according to manufacturers’ protocol to distinguish live from dead cells, followed by staining to analyze the pDC population for Vidu-AF647 uptake or internalization of BDCA2.

To analyze pDC uptake of Vidu-AF647, cells were surface stained with antibodies against human CD3 (Biolegend, cat. #300336), CD11b (Biolegend, cat. #101229), CD11c (Biolegend, cat. #337215), CD19 (Biolegend, cat. #302241), CD45 (Biolegend, cat. #368542), BDCA4 (Biolegend, cat. #354504) and BDCA2 (Biolegend, cat. #354207). Cells were left unfixed and analyzed immediately to preserve the integrity and magnitude of the Vidu-AF647 signal associated with cells.

To analyze internalization of BDCA2 in pDCs, cells were first surface stained with antibodies against human CD3 (Biolegend, cat. #300336), CD11b (Biolegend, cat. #101229), CD11c (Biolegend, cat. #337215), CD19 (Biolegend, cat. #302241), CD45 (Biolegend, cat. #368542) and BDCA4 (Biolegend, cat. #354504). This was followed by fixation and permeabilization with the FIX & PERM Cell Fixation and Cell Permeabilization Kit (Thermo Fisher Scientific, cat. #GAS003), with the staining antibody against BDCA2 (Biolegend, cat. #354207) added during the permeabilization step, as outlined in the manufacturer’s protocol.

Human TruStain FcX Fc Receptor Blocking Solution (Biolegend, cat. #422302) and True-Stain Monocyte Blocker (Biolegend, cat. #426103) were included during all surface staining steps. Samples were acquired on a Cytek Amnis ImageStream MkII Imaging Flow Cytometer (University of Iowa Flow Cytometry Facility) using INSPIRE software version ISX. Samples were run at a low flow rate with 40X or 60X magnification, as noted in figure legends, and single-color controls were collected to calculate a compensation matrix. Data was analyzed using IDEAS software version 6.2 (Cytek). Events that were out of focus, aggregated, CD45 negative or dead (Zombie Violet positive) were excluded from analysis. pDCs were identified as: 1) negative for CD11b, CD11c, CD3 and CD19 and 2) positive for BDCA2 and/or BDCA4. Internalized Vidu-AF647 signal was quantified with the IDEAS software using an internalization mask defined by 75% erosion from the brightfield (BF) cell membrane outline to stringently detect particle uptake and exclude background noise. Internalized BDCA2 was quantified with the IDEAS software using the Internalization Wizard with “Cell Image” defined by CD45 and the “Internalizing Probe” defined as BDCA2.

Statistical analyses were performed in GraphPad Prism (v10.2.2). Data are presented as a mean with individual donor/cell data points shown. Paired sample t-tests were used to compare two groups with matched donors. Repeated measures one-way ANOVA with Dunnett’s multiple-comparisons test or two-way ANOVA with Sidak’s multiple comparisons test were used for comparisons of more than two groups, as noted in figure legends. Data were significant when p values ≤ 0.05 (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001). Data were not significant (ns) when p values > 0.05.

We previously reported that anti-Qβ is required for Vidu to activate pDCs and induce an immune response in both murine and human models (16). To further evaluate how anti-Qβ mediates this response, we used flow cytometry to assess how anti-Qβ impacts on the association of various human immune cells with Vidu. The term “association” is used for these initial results because flow cytometry is not able to distinguish between adherence of Vidu to the surface of cells and uptake of Vidu by cells. Fluorescently tagged Vidu (Vidu-AF647) was added to human PBMCs, with or without recombinant human IgG anti-Qβ, and the association of the Vidu-AF647 with various immune cell types determined by flow cytometry (Figure 1). After a 2-hour culture, all CD45⁺ cell subsets exhibited a relatively low level of detectable Vidu-AF647 association in the absence of anti-Qβ that significantly increased in the presence of anti-Qβ (Figures 1A, B). Association of Vidu-AF647 as demonstrated by percent positivity was most pronounced with monocytes (CD14⁺ Mono) while association of Vidu with T cells (CD3⁺) was limited (Figure 1C). Anti-Qβ co-incubation with Vidu significantly increased the frequency of Vidu-AF647⁺ granulocytes (CD15⁺ Gran) and NK cells (CD56⁺), with B cells (CD19⁺) from some donors also exhibiting an increase (Figure 1C). Median fluorescent intensity (MdFI) of Vidu-AF647 signal was most significantly affected by the addition of anti-Qβ in the monocyte and granulocyte cell population (Figure 1D). Incubation for 2 hours at 4 °C was assessed to determine the impact of internalization on anti-Qβ-coated Vidu-AF647 signal (Supplementary Figure 1). Vidu-AF647 signal was similar at 4 °C and 37 °C, demonstrating minimal effect of internalization on overall VLP binding. Thus, the low level of Vidu-AF647 signal associated with monocytes in the absence of anti-Qβ (Figures 1C, D) likely reflects surface-binding. These findings are consistent with prior research showing that monocytes can interact directly with bacteriophages (37).

Given the inconsistent data regarding FcγR expression on human pDCs, we characterized surface expression levels of CD32 and BDCA2 across immune cell types from multiple healthy donors by flow cytometry. As expected, CD32 was highly expressed on monocytes, granulocytes and B cells, and was present at very low levels on NK cells, T cells and pDCs (Supplementary Figure 2A). In contrast, surface BDCA2 expression at significant levels was limited to pDCs (Supplementary Figure 2B).

Given the vital role pDCs play in the response to Vidu, particular focus was paid to the impact of anti-Qβ on the association of Vidu with pDCs. Anti-Qβ markedly enhanced association of Vidu with pDCs, in terms of both frequency (%) and magnitude (MdFI) across multiple healthy donors after a 2-hour incubation (Figures 2A-C). Anti-CD32 induced little change in the frequency of anti-Qβ-coated Vidu-AF647⁺ pDCs (~2% reduction in positive cells, Figure 2D), although there was a trend towards anti-CD32 reducing Vidu-AF647 signal per pDC indicated by MdFI (Figure 2E). After a longer incubation (20 hours), anti-CD32 induced a more consistent and significant drop in the frequency of Vidu-AF647⁺ pDCs (~15% reduction, Figure 2F), but did not have a significant impact on Vidu-AF647⁺ pDCs MdFI (Figure 2G). One possible explanation for the difference in magnitude of effect between 2 hours and 20 hours is that CD32^hi cells (non-pDCs) are present in greater numbers and are more rapidly saturated with antibody, while CD32^lo pDCs present in a lower number are more slowly saturated and blocked (Supplementary Figure 2A). Antibodies to CD16 (FcγRIII) or CD64 (FcγRI) had no clear effect on the interactions between anti-Qβ-coated Vidu-AF647 and pDCs (0% reduction, Supplementary Figures 3A, B). Anti-CD16, but not anti-CD64, reduced the magnitude (MdFI) of anti-Qβ-coated Vidu association with other CD45⁺ cell types (Supplementary Figures 3C-E).

Given that antibody to CD32 did not fully inhibit binding of anti-Qβ-coated Vidu to pDCs, we evaluated whether other receptors were involved. The pDC-specific receptor BDCA2 was of particular interest because it is known to bind to IgG via glycosylation moieties found in the constant region (28). Blocking BDCA2 had a time-dependent effect on anti-Qβ-coated Vidu association with pDCs. Anti-BDCA2 had a modest effect on the fraction of pDCs associated with anti-Qβ-coated Vidu association at 2 hours (~6% reduction). This impact increased to 16% by 20 hours. The magnitude (MdFI) of anti-Qβ-coated Vidu uptake by those pDCs that were positive was not significantly altered by anti-BDCA2 (Figures 2H–K). This suggests that early interactions of anti-Qβ-coated Vidu with pDCs are relatively BDCA2-independent, whereas the receptor significantly influences cumulative binding and internalization over time. The combination of both anti-CD32 and anti-BDCA2 reduced interaction between anti-Qβ-coated Vidu and pDCs to a greater degree than either antibody alone, further supporting a role for both receptors in binding of anti-Qβ-coated Vidu to pDCs (Figures 2L,M). We also tested blocking with an antibody against an unrelated pDC-specific receptor, BDCA4, and found anti-BDCA4 had no significant impact on the anti-Qβ-coated Vidu interaction with pDCs (Supplementary Figures 3F, G).

We then evaluated the functional impact of blocking CD32 and BDCA2 on anti-Qβ and Vidu activation of pDCs. Because we are using Vidu labeled with AF647, we confirmed that the fluorescent label did not alter the IFNα response from pDCs when treated with anti-Qβ-coated Vidu (Supplementary Figure 4). After a 20-hour culture, anti-BDCA2 was more effective than anti-CD32 at blunting IFNα production in response to anti-Qβ-opsonized Vidu-AF647 (Figures 3A, B). Next, we characterized surface expression of PD-L1 and CD80 on pDCs to determine how each blocking antibody affected pDC differentiation into previously defined subsets (38). Anti-BDCA2 was consistently more effective than anti-CD32 at blunting pDC differentiation from the immature P0 subset (PDL1^-CD80^-) to the type I IFN-producing P1 (PDL1⁺CD80^-) subset (Figures 3C, D).

To determine if treating cells with receptor-blocking antibody might also affect internalization of Vidu, cellular location (intracellular versus extracellular) within the pDC population was assessed using multispectral imaging flow cytometry. PBMCs (n=4 matched donors) were treated with IC, anti-CD32 or anti-BDCA2 prior to adding anti-Qβ and Vidu-AF647 for 20hrs. Vidu⁺ pDCs were identified by a specific gating strategy (Supplementary Figure 5), visualized to see morphology (brightfield-BF), surface expression of CD45, BDCA2 and BDCA4 (Figure 4A) and gated based on Vidu signal (Figure 4B). Vidu-AF647⁺ pDCs were analyzed using the IDEAS software to calculate internalization scores (ranging from -8 to 7.5) which correspond to how much of the Vidu signal is extracellular versus intracellular within each cell (Figure 4C). While both anti-CD32 and anti-BDCA2 influenced internalization, anti-BDCA2 was more effective and significant (mean internalization = 1.117, 45% reduction) than anti-CD32 (mean internalization = 1.454, 27% reduction) at reducing pDC internalization of anti-Qβ-coated Vidu when compared to their species-matched IC (mean internalization = ~2) (Figures 4D, E). Further assessment of the cumulative effect of combining blocking antibodies are needed and planned for future mechanistic studies.

BDCA2 crosslinking is known to induce BDCA2 internalization. This then results in a signaling cascade that blocks the pDC type I IFN response to TLR stimulation (27, 35, 39). Consistent with these reports, multispectral imaging flow cytometry demonstrated anti-BDCA2 antibody induces significant BDCA2 internalization by pDCs compared to media or IC (Figures 5A, B). The published literature suggests BDCA2 is an inhibitory receptor and that BDCA2 engagement and internalization impedes the ability of pDCs to respond to TLR9 stimulation (27, 35, 39–41). We therefore tracked the level of BDCA2 internalization and IFNα production by pDCs in response to different doses of anti-BDCA2 delivered with G10 (a soluble CpG-A ODN TLR9 agonist). BDCA2 internalization was not observed to any significant degree after treatment with G10 alone but did visibly increase in a dose-dependent manner as the anti-BDCA2 concentration increased (Figure 5C). Significantly higher internalization of BDCA2 was seen across 4 donors at higher anti-BDCA2 antibody concentrations (Figure 5D). BDCA2 internalization correlated with a decrease in IFNα levels detected in cell culture supernatants after 20 hours of stimulation with G10 (Figure 5D). These results confirm a link between BDCA2 internalization and inhibition of the IFNα response to a TLR9 agonist.

Given the data indicating a role for BDCA2 as a receptor for anti-Qβ-coated Vidu and that this results in activation (not suppression) of an IFNα response, we evaluated this apparent paradox further by testing how anti-Qβ-coated Vidu affected BDCA2 internalization in pDCs. PBMCs were incubated with a constant amount of Vidu and increasing concentrations of anti-Qβ for 20 hours prior to analysis of BDCA2 internalization on pDCs. In pilot studies, signal from labeled Vidu (Vidu-AF647) was diminished after the fixation/permeabilization needed to detect intracellular BDCA2, thus unlabeled Vidu was used for these experiments. We previously reported that Vidu delivered alone (0 μg/ml anti-Qβ) does not induce an IFNα response (16, 42). Thus, we began testing the effect of anti-Qβ at a sub-optimal level (1 μg/ml) based on IFNα dose-response curves we have previously tested (42). The degree of BDCA2 internalization steadily increased as anti-Qβ levels increased from 1 μg/ml to 80 μg/ml, with the highest levels of internalization occurring at 20 μg/ml and 80 μg/ml (Figures 5E, F). Consistent with our prior observations, 1 μg/ml anti-Qβ delivered with Vidu stimulated minimal IFNα response and also low BDCA2 internalization. The IFNα response peaked at an anti-Qβ concentration of 5 μg/ml, when BDCA2 internalization was modest. At higher anti-Qβ concentrations (20 μg/ml to 80 μg/ml), there was a stepwise decrease in the IFNα response. These higher doses of anti-Qβ were those that induced a greater degree of BDCA2 internalization. It is important to note that even at 20-80 μg/ml of anti-Qβ, the degree of BDCA2 internalization did not reach that seen with 1-10 μg/ml of anti-BDCA2 which can explain why the IFNα response was not fully suppressed even at high anti-Qβ concentrations. These data suggest that the amount of anti-Qβ coating Vidu is a critical determinant of the IFNα response. Thus, there appears to be a “Goldilocks Effect” to explain the “bell-shaped” relationship observed between anti-Qβ concentration, BDCA2 internalization and the IFNα response (Figure 6). If anti-Qβ is too low, then optimal uptake of Vidu and stimulation of pDCs by the TLR9 agonist is not achieved (Figure 6A). When anti-Qβ is too high, TLR9 agonist is delivered, but BDCA2 crosslinking inhibits TLR9 activation (Figure 6C). It is only at intermediate concentrations of anti-Qβ that there is adequate uptake of the TLR9 agonist without crosslinking and downstream inhibitory signaling mediated by BDCA2 (Figure 6B).