Fig 2: PF4 influences binding Ad infectivity and cell docking with serum and cell type dependency(A–C) For infectivity assays (A–C), VPs were incubated for 10 min at 37°C in optiMEM in the presence or absence of 10 μg/mL PF4, 10% fetal bovine serum (FBS), or human serum seronegative for Ad5, then allowed to infect cultured cells. (A) Fold change in Ad infectivity in A549 cells following Ad incubation with PF4. After infection using 20 vpc, internalized Ad genomes were titrated by qPCR 3 h post-infection (hpi). For each Ad type and each metric, pairwise comparisons were conducted between the samples with and without PF4 using two-sided Mann-Whitney U tests. N ≥ 6, two to five independent repeats. Data are represented as mean ± standard deviation. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001. (B) Primary peripheral blood mononuclear cells were infected with 2000 vpc of Ad5 vector from the Ad-GLN collection. Ad-expressed GFP fluorescence was quantified 48 hpi, and the proportions of GFP-positive cells were normalized to the average of the “no PF4, FBS” condition for each cell type. N ≥ 3, one or two independent repeats. Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001. (C) Primary human nasal epithelium cells were infected with 20 vpc of Ad5 vector from the Ad-GLN collection. Ad-expressed luciferase luminescence was quantified 24 hpi and normalized to the average of the “no PF4, no serum” condition. N ≥ 7. Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.(D) Principle of the erythrocyte pull-down technique. VPs aggregated or docking on erythrocytes, are separated from free VPs by low speed centrifugation and titrated by qPCR. Figure created with BioRender.(E) Erythrocyte pull-down of a fiber-modified Ad5 with ablated CAR tropism (Ad5-ΔCAR) in the absence or presence of PF4. A two-way ANOVA test indicated that both the erythrocytes (p = 0.00162) and PF4 (p = 1.15E-8) factors were significant. Pairwise comparisons were performed by two-sided Mann-Whitney U tests. N = 8. Data are represented as mean ± standard deviation. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.

Fig 3: Immunogenicity of PF4 non-binding COVID-19 vaccine candidates in mice(A) Experimental design. Five 10–12 weeks old C57BL/6JCrl male mice per group were immunized intravenously with 5E8 VP of E1-deleted or E1/E3-deleted vectors expressing the S1 domain of the SARS-CoV-2 spike protein (Hu-1 strain). Blood was drawn at 14 and 28 days after immunization (dai), and mice were sacrificed at 30 dai for splenocyte collection.(B–D) (B) Representative FACS plots show the frequencies of peripheral blood S1-epitope specific CD8+ T cells in the different groups (C, D). Percentages of S1-epitope specific CD8+ T cells in blood (C) and numbers in spleen (D) as determined by MHC tetramer staining. Time-course analysis using a mixed model two-way ANOVA was performed in (C). Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.(E) Representative FACS plots of stimulated or unstimulated IFN-γ- and TNF-α-secreting CD8+ and CD4+ T cells.(F) Total counts of S1-specific IFN-γ- and TNFα-secreting CD8+ (left) and CD4+ (right) T cells in spleen upon peptide stimulation. Numbers of IFNγ and TNFα secreting cells in non-stimulated control samples were subtracted. Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.(G) SARS-CoV-2 neutralizing antibody titer (NT50) in mouse serum. Data are represented as mean ± standard deviation.(H) Vector particle binding antibodies in the serum of immunized mice were determined by ELISA. Each serum was tested against the vector used for the immunization of the respective group. Symbols represent individual mice. LOD: limit of detection. Data are represented as mean ± standard deviation.

Fig 4: PF4 likely binds to Ad5 hexon HVR1 loop and partially protects Ad5 against neutralizing antibodies(A) Schematic representation of the HVR1 exchange performed to construct the Ad5H34 and Ad34H5 vectors. Figure created with BioRender.(B) ELISA-qPCR of the Ad5H34 and Ad34H5 variants for PF4 binding. Numbers of bound VPs in PF4 coated wells were normalized on the average number from the non-coated wells of the same experiment repeat. N = 8, three independent repeats. Significance levels on top of bars represent the comparison (Mann-Whitney U test) of bound titers with versus without PF4. Pairwise comparisons were also conducted between bound titer ratios of Ad5 versus Ad5H34, and of Ad5 versus Ad34H5. Data are represented as mean ± standard deviation. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.(C) Fraction of PF4 found at given surface-to-surface distance from adenovirus hexons, as sampled from Brownian dynamics (BD) simulations.(D) Distribution of computed surface electrostatic potential as well as integrated values across the whole hexons or hexon surface.(E) Surface potential of Ad particles measured by electrophoretic light scattering (ELS). Data are represented as mean ± standard deviation. N = 3.(F) BD simulations of popular regions for PF4 occupancy on hexons. Structural alignments of Ad34’s hexon and Ad5’s hexon suggest that Ad5’s HVR1 loops protrude more than those of Ad34, potentially enhancing their likelihood to interact with PF4 in the bulk solvent. Detailed molecular images mapping popular PF4 interacting residues to their molecular positions in either Ad34’s hexon or Ad5’s hexon are given in Figure S3B.(G) PF4 interference assay with Ad5 human serum neutralizing antibodies. Luciferase-expressing Ad5 vectors from the GLN collection were incubated with or without 10 μg/mL PF4 and 1/50 diluted human seronegative serum (“naive serum”) or pooled human intravenous immunoglobulins (“IVIG”) at varying dilutions. A549 cells were then infected with the suspensions at 500 VP/cell (vpc), and luciferase luminescence was measured at 24 hpi. The ratio of luminescence levels between samples with and without PF4 that received identical serum or immunoglobulin treatment is displayed. UT: untreated, without human serum or antibodies. N = 12, four independent repeats. ANOVA test of displayed results yielded p < 0.0001, and Dunnett post-hoc tests were conducted against the “UT” sample for all other samples. Pairwise comparisons that did not yield significant p-values are not displayed on the figure. Data are represented as mean ± standard deviation. The significance threshold was set at p < 0.05. Significance symbols: ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.

Fig 5: Identification of adenovirus and adeno-associated virus vectors lacking binding to PF4 by ELISA-qPCR and aggregate pull-down screening(A) Principle of the ELISA-qPCR technique. Adenovirus (Ad) vector particles (VPs) are allowed to interact with proteins, e.g., PF4, coated on an ELISA plate. After washing, the genomes of VPs that specifically interact with the proteins are released by heating and alkaline treatment and quantified by qPCR. Figure created with BioRender.(B) PF4 binding of vaccine-equivalent vectors. Ad5 was obtained from the Ad-GLN collection. N ≥ 6, two independent repeats. Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.(C and D) Screening of human and gorilla Ad (C) and AAV (D) collections for PF4 binding by ELISA-qPCR. For each experiment repeat, the PF4 binding index is computed as described in STAR Methods (Statistics), with positive values indicating significant binding to PF4 and negative values corresponding to overlap in the number of bound VPs in PF4-coated versus control samples. Averages and minimum/maximum range of the PF4 binding index from two to four (Ad5, Ad11, Ad34, and Ad80) independent repeats are displayed. Data are represented as mean ± standard deviation.(E) ELISA-qPCR of Ad5 hexon genetic and chemical variants for PF4 binding. Studied variants include: D151C and T273C point mutations; a 5 kDa polyethylene-glycol (PEG) polymer covalently linked to a cysteine residue, which prevents binding on part of the hexon surface by steric competition; deletion of the HVR1 loop; and the T425A substitution known to ablate the binding of factor X.12 The E1-deleted, GFP-expressing Ad5 vector was used as control (Ad5). HVR: hyper-variable region. PEG: polyethylene glycol. Schematics created with BioRender. N = 6, two independent repeats. Data are represented as mean ± standard deviation.(F) ELISA-qPCR of selected Ads for binding to mouse PF4. The binding index is computed as described in STAR Methods (Statistics). Measurements were performed once with technical triplicates. Data are represented as mean.(G) Principle of the aggregate pull-down technique. Aggregates forming upon interaction with PF4 are separated from free VPs by low speed centrifugation and titrated by qPCR. Figure created with BioRender.(H) Aggregate pull-down of selected Ads in the absence or presence of PF4. N = 8, two independent repeats. The aggregation rate was calculated based on the titration of Ad genomes both in the pellet (aggregates) and in the supernatant (free VPs). Data are represented as mean ± standard deviation. Two-sided Mann-Whitney U tests. The significance threshold was set at p < 0.05. Significance symbols: ns = non-significant, ∗ = p < 0.05, ∗∗ = p < 0.01, ∗∗∗ = p < 0.001.