Toll‐like receptor 4 signalling regulates antibody response to adenoviral vector‐based vaccines by imprinting germinal centre quality

Summary

Adenoviral vectors (AdV) are considered promising candidates for vaccine applications. A prominent group of Toll‐like receptors (TLRs) participate in the adenovirus‐induced adaptive immune response, yet there is little information regarding the role of TLR4 in AdV‐induced immune responses in recent literature. We investigated the function of TLR4 in both adaptive and innate immune responses to an AdV‐based anthrax vaccine. By immunizing wild‐type and TLR4 knockout (TLR4‐KO) mice, we revealed the requirement of TLR4 in AdV‐induced innate responses. We also showed that TLR4 functions are required for germinal centre responses in immunized mice, as expression of the apoptosis‐related marker Fas was down‐regulated on germinal centre B cells from TLR4‐KO mice. Likewise, decreased expression of inducible costimulator on follicular T helper cells was observed in immunized TLR4‐KO mice. Moreover, a potent protective antigen‐specific humoral immune response was mimicked using an adjuvant system containing the TLR4 agonist monophosphoryl lipid A. Overall, our findings showed that very rapid antigen‐specific antibody production is correlated with the TLR4‐imprinted germinal centre response to AdV‐based vaccine. These results provide additional evidence for the use of the AdV and a TLR agonist to induce humoral responses. Our findings offer new insights into rational vaccine design.

Abbreviations

Ad

adenovirus

AdV

adenoviral vector

APC

antigen‐presenting cells

GC

germinal centre

HMGB1

high mobility group box 1

mDC

myeloid dendritic cell

MyD88

myeloid differentiation primary responses protein 88

NK

natural killer cells

PA

protective antigen

PRR

pattern recognition receptors

Tfh

T follicular helper cell

TLR

Toll‐like receptor

TRIF

Toll/interleukin‐1 receptor domain‐containing adapter inducing interferon‐β

Introduction

Vaccines that can provide rapid protection against exposure to pathogens through a single injection represent an important need in disease prevention. To date, replication‐defective adenovirus serotype 5 (Ad5) is the most widely used adenoviral vector (AdV) for both gene therapy and vaccines. For Ad5‐based vaccines, transgene expression is usually rapid and robust, accompanied by a burst of antibody production. A significant advantage of AdVs is that a relatively high titre of antigen‐specific antibodies is produced in the bloodstream in a relatively short time.1 An increasing number of studies have focused on the mechanisms behind the innate signal pathways involved in adaptive immunity to AdVs.1, 2, 3, 4

Upon vaccination, antigens are recognized by antigen‐presenting cells (APCs) via several receptors, and the processed antigen is transported to secondary lymphoid organs for immune cell activation. Maturation in classic APCs, such as dendritic cells (DCs) and macrophages, is triggered by recognition of receptors, such as Toll‐like receptors (TLRs).1, 2 B cells, which express both antigen‐specific B‐cell receptors and TLRs, can also act as APCs, especially for particulate antigens. In B cells, the up‐regulation of activation markers and antibody secretion is also mediated by TLRs. Activated B cells can then develop into germinal centres (GCs) with the assistance of CD4⁺ T cells.4, 5 GCs are the essential sites of the humoral immune response, where B cells undergo a series of changes including clonal expansion and affinity maturation. Well‐established GCs could serve as the main source of antibody output.5 Hence, both APC activation and GC formation could affect the rapid production of antibodies.

The TLRs are a group of evolutionarily conserved pattern recognition receptors. Members of this superfamily use myeloid differentiation primary response protein 88 (MyD88) and Toll/interleukin‐1 receptor domain‐containing adapter inducing interferon‐β (TRIF) to induce activation of transcription factors and to secret specific cytokines.3, 4, 6 In particular, TLR4 is the only receptor that can use the MyD88 as well as the TRIF adaptor to initiate downstream signals.6 TLR4 is widely recognized as the receptor for bacterial lipopolysaccharide (LPS). Recently, it has been shown that TLR4 is involved in the innate immune response to respiratory syncytial virus infection and has the ability to augment B‐cell migration in the direction of the GC.2, 7, 8 In addition, TLR4 ligands can augment the magnitude and functionality of antigen‐specific cellular immune responses.9 However, the precise molecular associations of TLR4 involved in the immune response to AdV‐based vaccines are not well understood. Given the importance of TLR4 in the recognition of viruses and the stimulation of different arms of the immune system, we sought to investigate the role of TLR4 in the humoral immune response to AdV‐based vaccines, using anthrax protein antigen (PA) as the model antigen.

Materials and methods

Materials

We previously purified and preserved anthrax protective protein (PA) in our laboratory.10 TLR4 agonist monophosphoryl lipid A (MPLA), Al(OH)₃‐based alum adjuvant (Alhydrogel) and squalene oil‐in‐water (o/w) emulsion (AddaVax) were obtained from InvivoGen (San Diego, CA). Hexon was purchased from Fitzgerald Industries International (Acton, MA), fibre was supplied by AtaGenix (Wuhan, China), and anti‐high mobility group box 1 (HMGB1) antibody was obtained from BioLegend (San Diego, CA).

Animals

C57BL/6, C57BL/10 and TLR4^−/− C57BL/10 mice were purchased from Model Animal Research Centre, Nanjing University. The 6‐ to 8‐week‐old female mice were housed in a specific‐pathogen‐free facility. The entire study was carried out in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All experimental protocols were approved by the Institutional Animal Care and Use committee at Beijing Institute of Biotechnology.

Adenoviral vector production and characterization

Ad5‐based anthrax vaccine (Ad5‐PA) was previously purified and preserved in our laboratory.10 Briefly, the anthrax PA cassette was cloned into a shuttle vector and then into E1/E3‐deleted adenoviral molecular clones. The vectors were then propagated in HEK 293 cells and purified by gradient centrifugation. The infectious units were determined using an Adeno‐X Rapid Titer Kit (Takara, Kusatsu, Japan). All vectors were rendered free of endotoxin using a ToxinEraser™ Endotoxin Removal Kit (Genescript Biotech, Piscataway, NJ). All the above procedures were completed in a biosafety cabinet.

ELISA

The PA‐specific antibodies were detected by ELISA, as previously described, with some modifications.10 Briefly, ELISA plates (Corning, Corning ,NY) were coated with 2 μg/ml of recombinant PA and incubated at 4° overnight. After washing four times with 0·1% phosphate‐buffered saline (PBS)/Tween‐20, the plates were blocked with 5% skim milk (BD Biosciences, Franklin Lakes, NJ) in 0·1% PBS/Tween‐20 for 1 hr at room temperature. Plates were then washed and incubated for another 1 hr at room temperature with individual sample sera, after which plates were washed and incubated for 1 hr at room temperature with secondary horseradish peroxidase‐conjugated anti‐mouse total IgG or different IgG subclasses (IgG1 or IgG3; Abcam, Cambridge, UK). Plates were washed four times and developed with 100 μl of 3,3?,5,5?‐Tetramethylbenzidine colorimetric substrate (Solarbio, Beijing, China) for 5 min and then stopped with 1 M H₂SO₄. Optical density was determined using a Bio‐Rad iMark microplate reader (Bio‐Rad, Hercules, CA) at 450 nm.

Flow cytometry

Spleens from immunized mice were harvested (5 × 10⁷ cells/sample) and stored on ice in RPMI‐1640 medium (Gibco, Gaithersburg, MD) before preparation of a single‐cell suspension. Differential identification of cell subtypes was conducted, as described previously, with some modifications.11 For cell‐staining experiments, cells were blocked with anti‐CD16/anti‐CD32 (BioLegend) in PBS with 2% fetal bovine serum. Samples were then analysed using antibodies to CD4 (IM7), B220 (RA3‐6B2), GL7 (GL7), CXCR5 (L138D7), CD95 (Fas) (SA367H8), IgD (11‐26c.2a), PD‐1 (29F.1A12), CD86 (GL‐1), CD69 (H1.2F3), CD11b (M1/70), MHC class II (M5/114.15.2), CD11c (HL3) (BioLegend), and CD11c (HL3) (BD Biosciences). Intracellular BCL6 staining was performed using the Foxp3 fix/perm buffer set (BioLegend). Single‐stained controls were used to set compensation parameters and the indicated isotype‐matched antibody controls were used to set analysis gates.

Serum samples were collected from mice 24 hr following intramuscular (i.m.) immunization with 1 × 10⁷ plaque‐forming units (pfu) Ad5‐PA. Cytokines interleukin‐6 (IL‐6) and tumour necrosis factor‐α (TNF‐α) were quantified using standard reference curves according to the protocol provided in the BD Cytometric Bead Array (BD Biosciences). Samples were acquired on a BD FACSCanto II‐plus flow cytometer (BD Biosciences) and analysed using flowjo software (Tree Star Inc., Ashland, OR).

CD4⁺ T‐cell proliferation assay

One day after immunization with Ad5‐PA, splenic DCs were isolated from immunized mice using CD11c MACs beads (Miltenyi Biotec, Bergisch Gladbach, Germany), CD4⁺ T cells were isolated according to the instructions described in a CD4⁺ T‐Cell Isolation Kit II (Miltenyi Biotec). Then, CD4⁺ T cells were added to DCs at a 1 : 1 ratio. To measure T‐cell proliferation, cells were incubated with bromodeoxyuridine (BrdU) (Invitrogen, Carlsbad, CA, USA) for 12 hr. Three days after incubation, T‐cell proliferation was determined by flow cytometry using anti‐BrdU antibody (BD Biosciences).

ELISpot

ELISpot plates were coated with 10 μg/ml recombinant PA and incubated overnight at 4°. Antibody‐coated plates were washed three times with PBS and blocked with 1% bovine serum albumin + PBS for 2 hr at 37°. Then 100 μl of a splenocyte suspension (8 × 10⁵ cells/ml) from immunized wild‐type (WT) or TLR4 knockout (TLR4‐KO) mice was added to each well. The plates were incubated for 18–20 hr at 37°. After incubation, wells were washed at least three times with PBST (PBS containing 0·1% Tween‐20). After washing the plates, secreted antibodies were detected using anti‐mouse IgG (Abcam) and developed with AEC substrate (BD Biosciences) before analysis on an ELISpot counter.

Immunofluorescence

Mouse spleens were harvested, immersed in optimum cutting temperature compound (Sakura, Torrance, CA), and immediately frozen on dry ice. Then 8–10 μm sections were blocked with 5% goat serum in PBS for 30 min at room temperature. Slides were stained with anti‐GL7 Pacific Blue and then anti‐IgD Alexa Fluor 594 antibodies (BioLegend) for 1 hr. After each step, slides were washed with 0·1% Tween‐20 in PBS at least four times. Slides were then visualized using a Zeiss LSM 880 confocal microscope. Images were acquired using zen software (Zeiss, Oberkochen, Germany).

Statistical analysis

All values were expressed as the mean ± SEM. prism 6 software (GraphPad, San Diego, CA) was used to produce the figures and for statistical analyses. Statistical significance was calculated using a Student's t‐test or one‐way analysis of variance with Bonferroni post hoc tests for pairwise comparison of individual columns. Statistical significance was set at P < 0·05.

Results

Ad5‐PA can induce potent antibody responses and elevate TLR4 expression in CD11c⁺ cells

An ideal vaccine would generate a rapid humoral response to the pathogen. C57BL/6 mice received i.m. injections of different formulations of vaccine containing PA. Here, we used PA in alum as the control to mimic the particulate system. Fourteen days later, blood samples were collected and anti‐PA antibody titres were assessed by ELISA, as shown in Fig. 1(a). Intramuscular immunization with Ad5‐PA induced a prompt primary antibody response whereas mice immunized with PA and PA in alum showed a significantly lower IgG response; anti‐PA antibody titres were barely detected in the PBS immunization group.

Figure 1.

Adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA) can induce potent antibody responses and elevate Toll‐like receptor 4 (TLR4) expression on CD11c⁺ cells. (a) C57BL/6 mice (n = 6 to n = 8) injected with 1 × 10⁷ pfu/mouse of Ad5‐PA, 5 μg PA with or without 100 μg alum and phosphate‐buffered saline,(PBS), respectively. At 14 days post immunization, PA‐specific antibody titre was determined by ELISA. (b) Mice injected with 5 μg PA, 1 × 10⁷ pfu/mouse Ad5, 1 × 10⁷ pfu/mouse Ad5‐PA, 1 × 10⁷ pfu/mouse Ad5 plus 5 μg PA, and PBS. Histograms show expression of TLR4 on CD11c⁺ DCs sorted from different groups of immunized mice at 24 hr post immunization, represented as mean fluorescence intensity (MFI). Data shown are representative of three replicate experiments. Error bars represent mean ± SEM. Analysis of variance with Bonferroni post hoc tests for pairwise comparison of individual columns were used. *P < 0·05, **P < 0·01, ***P < 0·001.

Innate immune responses are key factors in the establishment of adaptive immune responses and among the different innate immune pathways, AdVs have been shown to activate TLR signalling.1, 2, 3 Previous studies have unravelled the capacity of AdVs to trigger different TLRs, such as TLR2 and TLR9.1, 2 TLR4 has been demonstrated to be involved in multiple aspects of antiviral immunity.2, 6, 7, 9 To test whether TLR4 signalling is involved in AdV‐induced humoral response, we evaluated the possibility of TLR4 participation in Ad5‐PA immunization.

Adjuvants and viral vectors initiate the humoral immune response by increasing antigen uptake by APCs. Because CD11c⁺ cells are well‐known APCs and predominant in TLR4 expression,2, 6 we hypothesized that TLR4 signalling might correlate with the rapid production of antibodies owing to the effective activation of APCs. Therefore, we examined the expression of TLR4 on CD11c⁺ cells. We immunized mice with different vaccine formulations: PA at a dose of 5 μg, Ad5 at a dose of 1 × 10⁷ pfu, Ad5‐PA (AdV expressing PA) at a dose of 1 × 10⁷ pfu, and Ad5 + PA (5 μg PA protein mixed with 1 × 10⁷ pfu non‐coding AdV). Upon immunization, TLR4 expression on sorted CD11c⁺ cells was evaluated at 24 hr, as shown in Fig. 1(b), TLR4 expression was highly elevated in the Ad5‐PA and Ad5 + PA groups. Although TLR4 expression was elevated in the PA and Ad5 groups, it remained significantly lower than that in the Ad5‐PA and Ad5 + PA groups, which shows both PA and Ad5 contribute to the TLR4 elevation in vaccination. In addition, there was no significant difference between Ad5‐PA and Ad5 + PA, indicating that the spatial link between Ad5 and PA is unrelated to TLR4 expression on CD11c⁺ cells.

TLR4 signalling is involved in induction of cytokine responses and anti‐PA antibody production

To evaluate the biological consequences of TLR4 signalling in humoral responses to Ad5‐PA, we quantified PA‐specific antibodies following Ad5‐PA administration in both WT and TLR4‐KO mice. The results indicated that the lack of TLR4 signalling resulted in a moderate yet statistically significant decrease in total IgG levels against PA at day 14 after i.m. injection (Fig. 2a).

Figure 2.

Anti‐PA antibody production and cytokine responses to adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA) are mediated by Toll‐like receptor 4 (TLR4) signalling. (a) At 14 day post immunization, protective antigen (PA) ‐specific antibody titre determined by ELISA. (b, c) Samples were isolated 24 hr post immunization, cytokines interleukin 6 (IL‐6) and tumour necrosis factor‐α (TNF‐α) analysed. Data shown are representative of three replicate experiments. Error bars represent mean ± SEM. Unpaired two‐tailed Student's t‐test was used: *P < 0·05, **P < 0·01.

Furthermore, as IgG1 and IgG3 are best known for their antiviral abilities,12 levels of IgG1 and IgG3 at day 14 post injection were then assayed. Again, significantly lower levels of PA‐specific IgG1 and IgG3 were observed in TLR4‐KO mice (see Supplementary material, Fig. S1A). Similarly, TLR4‐KO mice also demonstrated decreased production of plasma cells (PCs) at day 14 after immunization (see Supplementary material, Fig. S1B).

Cytokines are the key mediators for humoral immunity, and DC maturation can be achieved in response to inflammatory cytokines.12 Tumour necrosis factor‐α is widely recognized as a mediator of the host response to infection, and IL‐6 has been associated with the immune response following AdV vaccination.1, 2, 12 As for TLR signalling, cytokine secretion can be different when using different adaptors, MyD88 or TRIF. The MyD88 dependency was demonstrated in the early phase of TLR4 activation, whereas in the late phase, secretion of IL‐6 and TNF‐α was correlated with the TRIF‐dependent signalling pathway.1, 6, 12 Because TLR4 is the only TLR that can use both MyD88 and TRIF, we sought to identify the relative roles of TLR4 in the induction of cytokines IL‐6 and TNF‐α. At 24 hr following i.m. injection of Ad5‐PA, there were significantly lower circulating levels of IL‐6 and TNF‐α in immunized TLR4‐KO mice (Fig. 2b,c). This indicates a regulatory role of TLR4 in the inflammatory response after Ad5‐PA immunization.

TLR4 deficiency impairs B‐cell activation and APC function during Ad5‐PA immunization

Both DCs and B cells are major APCs that can link the innate recognition of viruses to adaptive immune responses. B‐cell receptors primarily recognize intact proteins, such as AdV particles.4, 13 To undergo the functional changes, APCs need to acquire maturation signals and fully competent antigen‐presenting capacity. Based on the gating strategies illustrated by Kachura et al.,11 we examined the expression of activation marker CD69 and maturation marker CD86 (Fig. 3a–c, and see Supplementary material, Fig. S2). Compared with other formulations containing PA, we observed a significantly higher expression of CD69 on B cells and natural killer cells in Ad5‐PA group (see Supplementary material, Fig. S2A–C). To determine whether early activation of B cells is regulated by TLR4, we examined the expression of CD69 on B cells at 24 hr post immunization. As shown in Fig. 3(b), the B cells from TLR4‐KO mice showed significantly lower expression of CD69 than those from WT mice, indicating that activation of B cells after immunization with Ad5‐PA is regulated by TLR4.

Figure 3.

Toll‐like receptor 4 (TLR4) deficiency impairs B‐cell activation and antigen‐presenting cell function during adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA) immunization. (a) Representative flow cytometry plots exhibiting gating scheme for B cells (CD3⁻ CD19⁺) and myeloid dendritic cells (mDCs) (CD11c CD11b⁺ MHCII ⁺). Spleen cells were prepared as described. Following light‐scatter gating, dead cells and debris were excluded from total splenocytes using a dump channel; all indicated cell types were determined using the isotype control. (b, c) Expression of activation marker CD69 and maturation marker CD86. (d) Expression of CD11b on mDCs from mice immunized with Ad5‐PA, measured by flow cytometry and represented as mean fluorescence intensity (MFI). (e) DCs immunized with Ad5‐PA from wild‐type (WT) and TLR4 knockout (TLR4‐KO) co‐cultured with naive CD4⁺ T cells. T‐cell proliferation was determined by measuring BrdU incorporation. Data pooled from three independent experiments (total four to six mice per group). Error bars represent mean ± SEM. Unpaired two‐tailed Student's t‐test was used. *P < 0·05, **P < 0·01.

To investigate the effects of TLR4 signalling on maturation signals after Ad5‐PA immunization, we assessed CD86 expression on major APCs. We found that up‐regulation of CD86 on macrophages and myeloid dendritic cells (mDCs) occurs in Ad5‐PA‐immunized mice (see Supplementary material, Fig. S2D, E). However, expression of CD86 on mDCs from TLR4‐KO mice was significantly decreased (Fig. 3c, d), and we also found that the expression of integrin CD11b on mDCs decreased significantly, which correlates with the regulatory role of TLR4 signalling on mDCs.14

Maturation of DCs is required for the induction of efficient T‐cell responses. To test whether the antigen‐presenting function requires TLR4 on DCs, we treated WT and TLR4‐KO mice with 1 × 10⁷ pfu Ad5‐PA and isolated splenic DCs 24 hr after immunization, assessing the function of splenic DCs by incubation with naive CD4⁺ T cells. The DCs isolated from TLR4‐KO mice induced only a very weak proliferative response in CD4⁺ T cells, whereas DCs from WT mice could induce a relatively high T‐cell proliferation rate (Fig. 3e).

Collectively, these results proved that B‐cell activation and APC function after Ad5‐PA immunization are partially affected by TLR4 deficiency.

Attenuation of TLR4 function results in impaired GC development

Activated B cells can develop into GCs, which are the essential sites of the humoral immune response. Both B‐cell activation and GC response are essential for antibody responses. To address the role of TLR4 in GC reactions, the GC response was visualized using immunofluorescent staining of mouse spleen sections collected at day 14 after immunization with Ad5‐PA. Weaker induction of GCs was observed in TLR4‐KO mice, as the relative size of GCs had shrunk in these mice (Fig. 4a).

Figure 4.

Toll‐like receptor 4 (TLR4) signalling is required for germinal centre GC response to adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA). (a) At 14 days after immunization, spleens from wild‐type (WT) and TLR4 knockout (TLR4‐KO) mice harvested, and frozen sections labelled with anti‐GL7 (blue) and anti‐IgD (red) to delimit GCs and follicles, respectively. Images acquired with a 10× lens. Boxed area indicates image to the right. Each group shows proportion of the follicular cell area occupied by GCs. Individual points denote individual spleen sections. (b) Total number of GC B cells (defined as B220⁺ CD4⁻ IgD⁻ GL7⁺) and (c) Total number of Bcl6⁺ T follicular helper Tfh (B220⁻ CD4⁺ CXCR5⁺ PD‐1⁺ Bcl6⁺) cells measured by flow cytometry 14 days after immunization. (d) Expression of ICOS on Tfh cells and (e) expression of Fas on GC B cells, represented as mean fluorescence intensity (MFI). Data pooled from three independent experiments (total five or six mice per group). Error bars represent mean ± SEM. Unpaired, two‐tailed Student's t‐test was used: *P < 0·05, **P < 0·01.

As GC B cells are the main source of antibody‐secreting cells (ASCs), to test whether TLR4 signalling contributes to the characteristics of GC cells, we first examined the number of GC B cells, and observed decreased cell numbers in immunized mice owing to the deficiency of TLR4, as assessed by flow cytometry (Fig. 4b). In addition, the GC response is highly dependent on T follicular helper (Tfh) cells that localize to the GC and provide selection signals for decisions about GC B‐cell fate.4, 5 As B‐cell lymphoma 6 (Bcl6) expression in T cells is essential for both the formation of Tfh cells and GC B cells,15 we next examined the number of Bcl6⁺ Tfh cells in both WT and TLR4‐KO mice after immunization with Ad5‐PA (Fig. 4c), and observed a significant decline in the number of Bcl6⁺ Tfh cells in GCs when TLR4 was deleted. This deficiency of Bcl6⁺ Tfh cells in immunized TLR4‐KO mice may alter the phenotype of Tfh cells to a lesser extent when differentiating into Tfh.

Certain co‐stimulatory signals, like inducible co‐stimulator (ICOS), provided by Tfh cells are important for GC B cells and required for GC reaction, by directly promoting follicular recruitment of activated Tfh cells.5, 15 To test whether TLR4 signalling shapes the characteristics of Tfh cells, we evaluated ICOS expression on Tfh cells. Ad5‐PA immunization generated marked effects on the expression of ICOS on Tfh cells (see Supplementary material, Fig. S3B). TLR4‐KO mice exhibited down‐regulation of ICOS expression on Tfh cells at day 14 after immunization (Fig. 4d). This may correlate with the loss of Bcl6⁺ Tfh cells, because ICOS co‐stimulation is thought to specifically induce Bcl6, although evidence for a direct signalling connection is lacking.15

Fas is highly expressed on GC B cells and can induce lymphocyte apoptosis, helping to avoid the formation of ‘rogue’ GC B cells that escape balanced regulatory controls; Fas thereby imprints the GC response in an efficient way.5, 16 In parallel with ICOS, Fas expression on GC B cells from Ad5‐PA‐immunized mice was higher than on those from mice immunized with soluble PA (see Supplementary material, Fig. S3C). TLR4‐KO mice exhibited down‐regulation of Fas expression on GC B cells at day 14 after immunization (Fig. 4e). Together with the decrease in ICOS expression, this finding indicates that TLR4 signalling contributes to the GC cells in qualitative ways.

Enhancement of the GC response contributes to high‐quality antibody production.4, 6 Those changes in either transcription factors or co‐stimulators can alter the phenotypes of GC cells, leading to the unbalanced GC responses. Together, these data suggest that TLR4 signalling participated in the Ad5‐PA‐induced humoral immunity, potentially by affecting the early GC response.

Long‐lived plasma cells in bone marrow after Ad5‐PA immunization are positively regulated by TLR4 signalling

A stable GC structure can induce differentiation of PCs and memory B cells. Both cells are essential for host defence against virus invasion.4, 5, 17, 18 However, persistent antibody production is maintained by a combination of short‐ and long‐lived PCs, and disrupted GC responses can induce unbalanced differentiation of PCs. Our group has previously shown that Ad5‐PA can generate rapid antibody production and sustain it for a relatively long period,10 which is largely dependent on antibody secreted by long‐lived PCs (LLPCs) in bone marrow (BM). To test whether TLR4 signalling is involved in the differentiation of PCs in BM, we used flow cytometry and ELISpot to examine the number of LLPCs and PA‐specific ASCs in BM. As shown in Fig. 5(a), Ad5‐PA induced a decline in the number of LLPCs (defined as CD138^hi CD43⁺ B220⁻) in TLR4‐KO mice, indicating that TLR4 exerts positive control in regulating the PCs in BM.

Figure 5.

Long‐lived plasma cells (LLPC) in bone marrow after adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA) immunization positively regulated by Toll‐like receptor 4 (TLR4) signalling. (a) Identification of LLPCs (defined as CD138^hi CD43⁺ B220⁻) in bone marrow by flow cytometry at day 90 after immunization with Ad5‐PA. (b) ELISpot assay of PA‐specific IgG‐secreting cells at day 90 after immunization without further culture. Data pooled from three independent experiments (total five or six mice per group). Error bars represent mean ± SEM. Unpaired two‐tailed Student's t‐test was used: **P < 0·01, ***P < 0·001.

We then investigated whether PA‐specific ASCs in BM were affected by the loss of TLR4. We verified the specificity of ASCs by ELISpot without further culture ex vivo. As shown in Fig. 5(b), PA‐specific ASCs in BM were largely decreased in TLR4‐KO mice, indicating a role of TLR4 in regulating the PA‐specific ASCs in BM, so reflecting an output from GC responses.

In fact, several factors contribute to the survival of LLPCs, such as IL‐17A.19, 20 To test whether the level of IL‐17A was affected by the TLR4 deficiency, we measured the secretion of IL‐17A using flow cytometry. However, we did not observe a significant difference between WT and TLR4‐KO mice with respect to IL‐17A secretion at day 56 or day 90 (data not shown), indicating that the production of LLPCs after immunization with Ad5‐PA regulated by TLR4 signalling does not depend on IL‐17A.

Elevated expression of TLR4 is mediated by hexon and ablated by anti‐HMGB1 antibody

To confirm the role of TLR4 in humoral immunity, the TLR4 agonist MPLA was used to mimic rapid antibody production. Because particulate delivery systems are usually designed to enhance activation of APC immunopotentiation, and recently they are also proven to have the ability to enhance the GC responses,4, 13, 21 we co‐administered PA with either alum or AddaVax plus MPLA. As expected, enhancement of PA‐specific antibodies was seen at day 14 post immunization in both groups, corresponding to Ad5‐PA (Fig. 6a).

Figure 6.

Elevated expression of Toll‐like receptor 4 (TLR4) is mediated by hexon and ablated by anti‐HMGB1 antibody. (a) C57BL/6 mice (n = 6) injected with 1 × 10⁷ pfu/mouse adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA), 5 μg PA plus 1 × 10⁷ pfu/mouse Ad5, or 10 μg MPLA, co‐administered with either 200 μg alum or AddaVax (1 : 1, v/v). PA‐specific IgG antibody measured 14 days after immunization. (b) Total number of germinal centre (GC) B cells per 10⁵ live splenocytes measured by flow cytometry 14 days after immunization with different formulations as described in Fig. 1(b). (c) Different groups of mice immunized with PA (5 μg), Ad5 (1 × 10⁷ pfu), and Ad5‐PA (1 × 10⁷ pfu). At 12 hr after immunization, mice were injected with control IgG (50 μg/mouse) and anti‐HMGB1 (50 μg/mouse). At 24 hr after immunization, spleens were harvested and single‐cell suspensions were prepared. Expression of TLR4 on CD11c⁺ cells was evaluated by flow cytometry and represented as mean fluorescence intensity (MFI). (d) Expression of TLR4 on CD11c cells 24 hr after immunization with hexon (10 μg/mouse), fibre (15 μg/mouse), Ad5 (10⁷ pfu/mouse), and Ad5‐PA (10⁷ pfu/mouse). Data were pooled from three independent experiments (total 5–6 mice per group). Error bars represent mean ± SEM. Analysis of variance with Bonferroni post tests for pairwise comparison of individual columns were used: *P < 0·05, **P < 0·01, ***P < 0·001.

As described earlier, the expression of TLR4 on CD11c⁺ DCs sorted from immunized mice was comparable between Ad5‐PA and Ad5 + PA (Fig. 1b). Nevertheless, how this elevation reflected in GC responses remains unclear. We then examined the number of GC B cells and found no significant difference between Ad5‐PA and Ad5 + PA (Fig. 6b) similar to the expression of TLR4 in those groups.

Hence, we wanted to know the ligand for TLR4 during Ad5‐PA immunization. HMGB1 is an endogenous ligand for TLR4.22 We sought to use anti‐HMGB1 antibody to test whether the up‐regulation of TLR4 expression could be comprised by it. Interestingly, such elevated expression of TLR4 can be suppressed by intraperitoneal injection of anti‐HMGB1 antibody (Fig. 6c), which means that HMGB1‐related‐apoptosis/necrosis occurs and HMGB1 may act as the TLR4 ligand during immunization.

To determine whether the Ad components can also act as the ligand for TLR4 during vaccination, mice were immunized with hexon and fibre, two Ad structural proteins that have been reported to trigger interferon‐γ and act as the potent adjuvants.1, 23 Compared with fibre, hexon induced a much higher level of TLR4 expression on DCs, which was comparable to that of Ad5 (Fig. 6d). This finding indicates that TLR4 up‐regulation on DCs after Ad5‐PA immunization may correlate with the hexon.

Collectively, these results confirmed a vital role for TLR4 in humoral immunity to AdV and showed that elevated expression of TLR4 after immunization with Ad5‐PA is mediated by hexon and ablated by anti‐HMGB1 antibody.

Discussion

In this study, attenuated antibody production, cytokine secretion and antigen‐presenting function were observed in immunized TLR4‐KO mice. TLR4 deficiency impaired the expression of ICOS on Tfh cells and Fas on GC B cells, influencing the quality of follicular CD4⁺ T cells and GC B cells. In addition, robust antigen‐specific antibody production was seen when using an adjuvant system containing a TLR4 agonist. We also showed that TLR4 expression upon immunization with AdV was mediated by adenoviral components and HMGB1.

This study was designed to elucidate the role of TLR4 in the generation of humoral immune responses against AdV‐based vaccines. Importantly, the route, dose and timing of immunization can affect the immune response to AdV. Intravenous immunization of AdV has been used extensively in gene therapy. However, our study focused on the changes in humoral responses after i.m. immunization. Moreover, TLR4 is a well‐known receptor for LPS from bacteria. The Ad5‐PA used here was purified from a eukaryotic expression system, HEK 293 cells, and rendered free of LPS using a commercially available endotoxin removal kit (the final endotoxin level was < 25 EU/ml). In addition, we excluded other possible involvement of LPS to confirm that there was no LPS contamination during our experiments. Hence, our findings suggest that, independent of LPS, Ad5‐PA could activate TLR4 signalling to induce the humoral response.

The immunogenicity of viral vectors is partly determined by viral ligands for specific host cell surface receptors. TLRs can provide the first level of defence and contribute to programmed antibody responses,6 so forming a bridge between innate and adaptive immunity. During the writing of our manuscript, Anchim et al.3 reported that humoral responses elicited by Ad5‐based vaccines are shaped by the TLR/MyD88 pathway. The discrepancies between their study and ours can be related to differences in mouse strains, virus dose, and the antigen insertion site or the route of immunization. Importantly, we both agree that the difference in the requirement for production of antibodies against the transgene product might be linked to the intrinsic nature of the antigen. Both studies concur that TLR signalling is key to the Ad5‐based vaccine; however, the precise TLR involved in MyD88 activation was not investigated by Anchim et al.3

Innate immune responses are key factors in the establishment of adaptive immune responses. We know that TLR4 can interact with viral proteins like VSV‐G,2, 7 yet little is known about the role of TLR4 signalling in altering AdV‐induced antibody responses. The finding of our study further indicated that TLR4 signalling is involved in Ad5‐PA immunization. The study presented here explored the TLR4 signalling requirement for rapid generation of antibodies by i.m. immunization with Ad5‐PA, and our findings confirmed a role of TLR4 in activating B cells and the regulation of mDCs after Ad5‐PA immunization. Integrins CD11b/CD18 on mDCs interact with TLR4 to deliver optimal signalling for early activation of immune responses,14 suggesting that TLR4 signalling could be mediated through its binding to CD11b. Further, we have shown that the antigen‐presenting function is influenced by the loss of TLR4, by examining the T‐cell proliferation rate after incubation of DCs isolated from WT and TLR4‐KO mice. Our data suggest that TLR4 deficiency causes a diminished capacity of DCs to activate naive T cells, supporting a critical role for this pathway in the efficiency of AdV.

Previous studies have found that TLR4 signalling plays an important role in GC responses, because it can enhance B‐cell migration8, 24 and modulate follicular dendritic cell activation.21 In this study, we observed a shrunk GC area and decreased GC B‐cell number in TLR4‐KO mice, indicating a weaker GC reaction at that time. In a GC response, both the GC B‐cell numbers and the quality of Tfh cells are important to the antibody production. Initial activation by APCs and subsequent interaction with B cells could induce T cells to acquire a complete GC Tfh phenotype. As for our study, a decreased number of Bcl6⁺ Tfh cells was seen in immunized TLR4‐KO mice. ICOS and Fas expression were also partially affected by TLR4 deficiency, therefore, ICOS has been shown to determine the size of the Tfh cell population,15 and this impact of TLR4 signalling on expression of the co‐stimulatory factor ICOS likely affected Tfh cell properties. Fas expression can help to prevent emergence of ‘rogue’ GC B cells,16 and can induce pro‐inflammatory cytokine production. Therefore, our data showed that TLR4 signalling plays a vital role in GC development after immunization with Ad5‐PA. This may explain the impaired antibody production in TLR4‐KO mice after immunization. However, we examined the expression levels of these co‐stimulatory factors in multicellular interactions at a given moment, so further studies using conditional knockout mice may provide a valuable opportunity in fine tuning the regulation of TLR4 signalling in Ad5‐PA‐ induced GC responses.

A hallmark of adaptive immune response is the generation of LLPCs and memory B cells. Both types of cell are thought to primarily derive from the GC.17, 18 Several studies have proven that TLRs are involved in GC regulation and TLR4 is crucial for the in vivo generation of LLPCs.2, 4, 25 As we have shown, TLR4 signalling can positively regulate the GC responses after Ad5‐PA immunization; we also observed a regulatory role of TLR4 in LLPC and PA‐specific ASC production in BM. The notion that LLPCs derive from the GC is well established.17 This may indicate that the participation of TLR4 in GC responses alters the nature of output from GC B cells to LLPCs as the response progresses. We examined the IL‐17A secretion in serum after immunization, because IL‐17A can assist with the survival of LLPCs.19, 20 However, we did not observe a significant difference between immunized WT and TLR4‐KO mice. Hence, the requirement of IL‐17A or other cytokines for the generation of LLPCs after immunization with Ad5‐based vaccine requires further study.

The particulate emulsion MF59 is a potent adjuvant that can facilitate antigen uptake by immune cells at the injection site.26 In this study, MPLA, co‐delivered with either alum or AddaVax (a squalene‐based oil‐in‐water emulsion with a formulation similar to MF59), yielded a comparable production of PA‐specific antibodies with Ad5‐PA. MPLA is responsible for activating TLR4 signalling whereas alum or AddaVax particles (similar to the AdV particles) appear to prolong the MPLA response at the injection site and directly use B cells as APCs. The adjuvant MPLA is a TRIF‐biased agonist of TLR4, indicating that Ad5‐PA may possess a TRIF‐dependent signalling‐biased adjuvant property.

We additionally observed that besides Ad5‐PA, soluble PA and Ad5 particles could boost TLR4 expression on DCs and that this elevation was suppressed by anti‐HMGB1 antibody. By using two different ligations of Ad5 and PA (Ad5‐PA and Ad5 + PA), we found that TLR4 elevation and antibody production have comparable expression, and GC B‐cell number. This may indicate that the spatial link between Ad5 and PA has less effect on the humoral immunity induced by AdV. Among all the adenoviral components, hexon is critical for virus infection and blood factor binding.23, 27 One study reported that the epitopes displayed on Ad5 hexon or fibre can overcome the limits induced by pre‐existing immunity.3 We observed a potent elevation of TLR4 expression after injection of Hexon rather than fibre. Together, this indicates that Ad5‐PA immunization may cause the release of HMGB1, a danger‐associated molecular pattern molecule, from dying or damaged cells, thereby leading to the activation of TLR4.22 In this way, infectious or mature virus‐particles in AdV‐based vaccines could invade the cells and activate TLR4 signalling by releasing danger‐associated molecular patterns. However, the nature of the antigen and the spatial link between hexon and antigen should also be taken into consideration, and other lipidic components in AdV may interact with TLR4.

Whereas several works have investigated the role of TLR2 and TLR9 signalling in Ad5‐induced cytokine release,1 no studies have investigated the molecular mechanisms related to TLR4 signalling in antibody responses after Ad5‐PA immunization. Our results revealed a role of TLR4 in both innate and humoral immune responses to Ad5‐PA by showing that rapid antigen‐specific antibody production is correlated with the TLR4‐imprinted GC responses. The Ad5‐PA vaccine, particularly its hexon structural protein, up‐regulated surface TLR4 on DCs in a manner dependent on HMGB1, a ligand for TLR4,22 thereby inducing GC reactions and antibody responses. In addition, TLR4 deficiency impaired the number of GC cells as well as the expression of ICOS on Tfh cells and Fas on GC B cells, indicating that TLR signalling acts in multiple complementary ways to mediate both the quantity and quality of GC reactions. Therefore, activated TLR4 signalling can further enhance vaccine potency (see Fig. 7 for a graphic representation). This study provides a paradigm for future studies on the relationship between innate and adaptive responses to AdVs that will allow for safer and more effective use of this important vaccine/therapy platform. In addition, our findings may serve to further motivate efforts to incorporate TLR ligands in the design of rational vaccines used to combat outbreaks of infectious diseases.

Figure 7.

Involvement of Toll‐like receptor 4 (TLR4) in multiple elements in adenovirus vector (AdV) ‐based vaccine. Adenovirus serotype 5‐based anthrax vaccine (Ad5‐PA), particularly its hexon component, up‐regulates surface TLR4 on dendritic cells (DCs), dependent on HMGB1, inducing germinal centre (GC) reactions and antibody responses. Activated myeloid DCs (mDCs) participate in innate responses to AdV‐based vaccine; activated B cells can further migrate to follicles to form the GC, together with CD4⁺ helper T cells. Co‐stimulatory factor ICOS, transcription factor Bcl6, and apoptosis‐related marker Fas all participate in development of the GC response to enhance antibody response.

Disclosures

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.

Author contributions

RL, JL, JX and WC conceived and designed the experiments; RL, YL and QY performed the experiments; YL, QL, XZ and RL analysed the data; XZ, YL, LH and SW were responsible for the reagents; and RL and JX wrote the paper. All authors read and approved the final manuscript.

Supporting information

Figure S1. (A) Serum was pooled from immunized wild‐type (WT) or Toll‐like receptor 4 knockout (TLR4‐KO) mice and was measured by ELISA at 14 days after immunization with Ad5‐PA. IgG1 and IgG3 were measured, respectively. (B) Representative flow cytometry plots exhibiting gating scheme for plasma cells, total numbers of plasma cells (B220^midCD138⁺) were determined by flow cytometry at 14 days after immunization. Data are pooled from three independent experiments with a total of four to six mice per group.

Figure S2. (A–C) Expression of activation marker CD69 on B cells (CD3^– CD19⁺) and natural killer cells (CD3^– CD19^– DX5⁺) and (D, E) maturation marker CD86 on macrophages (CD11c^– CD11b⁺ MHC‐II⁺ F4/80⁺) and myeloid dendritic cells (CD11c⁺ CD11b⁺ MHC‐II⁺) in the spleen at 24 hours after immunization were assessed by flow cytometry and represented as median fluorescence intensity (MFI). Histogram data shown are representative of three independent experiments. Data are pooled from three independent experiments with a total of four to six mice per group.

Figure S3. (A) Representative flow cytometry plots exhibiting gating scheme for Tfh (B220^– CD4⁺ CXCR5⁺ PD‐1⁺) and germinal centre B (B220⁺ CD4^– IgD^– GL7⁺) cells.