TLR4-dependent activation of dendritic cells by an HMGB1-derived peptide adjuvant

High mobility group box protein 1 (HMGB1) acts as an endogenous danger molecule that is released from necrotic cells and activated macrophages. We have previously shown that peptide Hp91, whose sequence corresponds to an area within the B-Box domain of HMGB1, activates dendritic cells (DCs) and acts as an adjuvant in vivo. Here we investigated the underlying mechanisms of Hp91-mediated DC activation. Hp91-induced secretion of IL-6 was dependent on clathrin- and dynamin-driven endocytosis of Hp91 and mediated through a MyD88- and TLR4-dependent pathway involving p38 MAPK and NFκB. Endosomal TLR4 has been shown to activate the MyD88-independent interferon pathway. Hp91-induced activation of pIRF3 and IL-6 secretion was reduced in IFNαβR knockout DCs, suggesting an amplification loop via the IFNαβR. These findings elucidate the mechanisms by which Hp91 acts as immunostimulatory peptide and may serve as a guide for the future development of synthetic Th1-type peptide adjuvants for vaccines. Electronic supplementary material The online version of this article (doi:10.1186/1479-5876-12-211) contains supplementary material, which is available to authorized users.


Introduction
The adaptive immune response is commonly initiated via pathogen-associated molecular patterns (PAMPs) that are recognized by Toll-like receptors (TLRs) [1]. Dendritic cells (DC) are central for the initiation of adaptive immune responses and are activated by exogenous PAMPs such as lipopolysaccharides (LPS), CpG, or poly(I:C) [2] as well as endogenous signals of tissue and cell damage, sometimes referred to as alarmins or danger signals [3,4]. Alarmins can take the form of inflammatory cytokines secreted by cells proximal to the site of injury [5] or internal components of damaged cells. Evidence for the latter includes reports that necrotic cell lysates, more specifically heat shock proteins (HSPs) and high mobility group box protein 1 (HMGB1) in the lysates, can induce DC maturation [6][7][8].
Structure-function studies have revealed that the proinflammatory domain in HMGB1 maps to the B-box domain, which recapitulates the cytokine activity of full-length HMGB1 [30,31]. We have previously shown that a B-box domain derived peptide, named Hp91 acted as a potent maturation stimulus for DCs and induced a cytokine profile typical of a Th1-type response [32]. We recently showed that Hp91 potentiates antigen-specific humoral and cellular immune responses in vivo [33]. This study explored the mechanism by which Hp91 activates antigen presenting cells by investigating cellular uptake, receptor dependence, and signaling pathways. We found that Hp91-induced secretion of IL-6 was mediated through a MyD88/TLR4dependent pathway involving p38MAPK and NFκB.

Animals
Female C57BL/6 mice 8-12 weeks of age were used for experiments. C57BL/6 mice were purchased from Charles River Laboratories (Boston, MA, USA). TLR4−/− and IL1R−/− mice were purchased from The Jackson Laboratories (Bar Harbor, ME, USA). IFNαβR−/− mice were purchased from B&K Universal (England, UK). MyD88−/− and TLR7−/− mice were a gift from S. Akira (Osaka University, Osaka, Japan) and backcrossed for 10 generations onto the C57BL/6 background. Mice were bred and maintained at the Moores UCSD Cancer Center animal facility and all animal studies were approved by the Institutional Animal Care and Use Committee of UCSD and were performed in accordance with the institutional guidelines.

Confocal microscopy
1x10 5 immature human DCs were precooled on ice and subsequently incubated for 30 min on ice with biotinylated-Hp91 or Hp121 to allow peptide binding. Cells were washed and then incubated for the indicated time at 37°C. Cells were cytospun (Shandon Cytospin 2 centrifuge) onto glass slides, fixed, permeabilized with acetone, and stained with Streptavidin-Alexa 488 (Invitrogen) to visualize biotinylated peptides and Hoechst 33258 (Invitrogen) to visualize DNA. Cells were imaged on a Zeiss LSM confocal microscope.

Binding/uptake studies
For most experiments, iDCs or mouse BM-DCs were precooled on ice for 30 min, as indicated in the figure legends. Cells were subsequently incubated for the indicated times and temperatures in culture medium with biotinylated peptides. Cells were washed, permeabilized with Cytofix/Cytoperm (BD Biosciences, Franklin Lakes, NJ), stained with Streptavidin-Alexa 488 (Invitrogen), and analyzed by flow cytometry. For experiments with endocytosis inhibitors, cells were pre-treated for 30 min with the indicated inhibitors or controls prior to incubation with the biotinylated peptides. For experiments with J774 mouse macrophages, cells were pre-cooled on ice, pre-treated with 30 min with the indicated inhibitors, and subsequently incubated for 30 min with fluorescently-labeled Hp91 (Cp488-Hp91). Cells were immediately analyzed by flow cytometry using the FACSCalibur (Beckon Dickinson, Franklin Lakes, NJ). Data were analyzed using the FlowJo software (Tree Star, Inc., Ashland, OR).

Stimulation of DCs
At days 5-7 of culture, DCs were either left untreated or were stimulated with indicated doses of peptide. For inhibition experiments, immature human DCs were pretreated with the indicated doses of SB203580, PD98059, N-tosyl-L-phenylalanine chloromethyl ketone (TPCK), or DMSO control for 30 min prior to stimulation. For experiments with human DCs, supernatants were collected 48 h after stimulation and the level of IL-6 analyzed by IL-6 ELISA (eBioscience, Inc. San Diego, CA). For experiments with mouse BM-DCs, supernatants were analyzed by ELISA (eBioscience), 24 h after stimulation.

Cytokine Release Assay
Splenocytes were cultured overnight with 2.5 μg/ml OVA-I peptide, 5 μg/ml concanavalin A positive control (Sigma-Aldrich), or left unstimulated (media only). After 18 h, cell culture supernatants were collected and analyzed for the presence of IL-2 by ELISA (eBioscience).
In some experiments, cells were pre-treated with the endocytosis inhibitor Dynasore.

Qualitative real-time PCR
Qualitative real-time PCR (qPCR) was performed in a Stratagene Mx3005P (Agilent, Santa Clara, CA) for mouse IFN-α2 and GAPDH. GAPDH was used as an endogenous standard for normalization of the IFN-α2 gene. Briefly, 1.25 × 10 5 J774 macrophages/well were serum-starved overnight in a 96-well flat-bottom plate and stimulated in duplicate with LPS (10 ng/ml), acetylated Hp91 (200 μg/ml), or left unstimulated (media) for 6 hours. Cells were harvested and RNA was isolated using TRIzol as follows: the J774 cell pellets were lysed in approximately 1 ml of TRIzol Reagent (Invitrogen) by repetitive pipetting. The cleared homogenate solution was incubated for 5 min at RT, 200 μl of chloroform was added and samples were shaken for 15 seconds and incubated at RT for an additional 2-3 min. Samples were centrifuged at 12000 × g for 15 minutes at 4°C. Pellets were washed with 1 ml 75% RNAse-free ethanol, centrifuged for 7000 × g for 5 min at 4°C, and the RNA pellets were air dried. DNase was removed from samples using a Turbo DNAfree DNase treatment (Applied Biosystems/Ambion, Austin, TX). cDNA was synthesized using Superscript III-RT polymerase (Invitrogen) and related reagents as per the manufacture's instructions. qPCR samples were setup using Brilliant II SYBR Green QPCR Master Mix (Invitrogen) and the following Q-primers: IFN-α2 (For. 5'-ACTCTGTGCTT TCCTCGTGATGCT-3'; Rev. 5'-ATCCAAAGTCCTGC CTGTCCTTCA-3') and GAPDH (For. 5'-TCACCACC ATGGAGAAGGC-3'; Rev. 5'-GCTAACCAGTTGGTG GTGCA-3'). Primers were purchased from IDT. qPCR was performed on duplicate samples in a Stratagene Mx3005P. Amplification product lengths were confirmed on a DNA gel. Values are normalized against GAPDH controls.

Statistical analysis
Data were analyzed for statistical significance using unpaired or paired Student's t-test or the Log Rank test. Statistical analysis was performed using GraphPad software version 5.01 for Windows (GraphPad Software, San Diego, CA, USA). A p value <0.05 was considered statistically significant.

Results
Hp91 amino acid sequence is critical for uptake and activation of DCs To gain insight into the mechanism of action of the DC stimulatory peptide Hp91, we investigated its physical interaction with DCs. Hp91 was taken up in a dose dependent manner ( Figure 1A), which plateaued between 10 and 30 minutes ( Figure 1B). The control peptide, Hp121, which also corresponds to a sequence present in HMGB1 and has the same length, a similar charge, and isoelectric point as Hp91 was not taken up by DCs and uptake of a scrambled version of Hp91, was significantly lowered ( Figure 1C, 1D and 1E), suggesting that uptake of Hp91 is sequence specific and not due to overall charge. Confocal microscopy was used to distinguish between cellular binding and uptake and showed localization of Hp91 inside DCs. The control peptide Hp121 was not detected inside the cells, even after 30 minutes incubation at 37°C ( Figure 1C). Since the scrambled version of Hp91 interacted with DCs, although to a much lesser extent than Hp91 ( Figure 1D and 1E), we examined if it had similar immunogenic properties as Hp91, which as we have previously described potentiates antigen-specific CD8+ T cell immune responses in vivo [33]. The scrambled version of Hp91 did not augment OVA-specific CD8+ CTL responses or IL-2 release ( Figure 1F and 1G), suggesting that Hp91 functional activity is sequence specific.

Hp91 enters dendritic cells via clathrin-mediated endocytosis
Hp91 was taken up at 37°C, but not at 16 and 4°C (Additional file 1: Figure S1), indicating that the uptake occurred via an energy dependent process, i.e. endocytosis. This was further supported by a significant inhibition of uptake of a fluorescently labeled version of Hp91 (Cp488) by sodium azide, which inhibits energy-dependent endocytosis (data not shown). The endocytic pathways include phagocytosis, macropinocytosis, clathrin-mediated endocytosis, and lipid-raft/caveolin-mediated endocytosis. The mechanism involved in the Hp91 uptake was determined using specific inhibitors of these endocytosis pathways. Latrunculin B (LatB), a specific phagocytosis inhibitor, did not reduce uptake of Hp91 (Figure 2A), whereas it completely abrogated uptake of the control Dextran FITC (data not shown). Phenylarsine oxide and chlorpromazine, inhibitors of clathrin-mediated endocytosis, significantly reduced the uptake of Hp91, suggesting that uptake occurs via clathrinmediated endocytosis ( Figure 2B). In contrast, no effects were seen on the uptake of Hp91 for the lipid-raft/ caveolin-mediated endocytosis inhibitor, nystatin, or the macropinocytosis inhibitor, amiloride ( Figure 2B), indicating that the uptake was not occurring via neither lipid rafts nor macropinocytosis. These data show that the major mechanism for Hp91 uptake into DCs occurs via a clathrin coated pit dependent manner.

Hp91-mediated activation of mouse antigen presenting cells is clathrin-and dynamin-dependent
Since ligand engaged TLR4 is endocytosed in a dynamindependent manner for downstream signaling in macrophages exposed to LPS [35], we evaluated if Hp91 signals through a similar TLR4 mechanism as dynamin is part of (See figure on previous page.) Figure 1 Hp91 uptake by DCs is dose, time, and sequence dependent. (A) Immature human DCs were pre-incubated on ice for 30 min, then incubated with biotinylated-Hp91 (0, 10, or 100 μg/ml) for 30 min at 37°C. Cells were permeabilized with Cytofix/Cytoperm, stained with Streptavidin-Alexa 488, and analyzed by flow cytometry. The results shown is one representative experiment (N = 6). (B) Immature human DCs were incubated with 100 μg/ml biotinylated Hp91 for 10 or 30 minutes, permeabilized, stained, and analyzed as above. Results shown are mean (±SEM) of N = 4. (C) Pre-cooled immature human DCs (iDCs) were incubated on ice for 30 min with 200 μg/ml biotinylated Hp91 or Hp121 to allow peptide binding, washed, then incubated for 15, or 30 additional min at 37°C. Cells were cytospun, fixed, permeabilized, and stained with Streptavidin-Alexa 488 to visualize biotinylated peptides (Green) and Hoechst DNA stain (Blue). Cells were imaged on a Zeiss LSM confocal microscope. Data shown is one representative experiment of N = 3. (D-E) Immature human DCs were pre-cooled on ice for 30 min, then incubated with media only, biotinylated-Hp91, biotinylated-Hp121, or biotinylated-scrambled Hp91 ("Scramble") at 200 μg/ml for 30 min at 37°C. Cells were permeabilized, stained, and analyzed by flow cytometry as above. (D) is mean (±SEM) for N = 3 and (E) is one representative experiment. *p < 0.05 compared to medium; Student's t-test. (F, G) Mice were immunized with OVA-I peptide in PBS with Hp91 or scrambled Hp91 peptide (250 μg). (F) Freshly isolated splenocytes from the immunized mice were examined in an OVA IFN-γ ELISpot assay. (G) Supernatants were collected and analyzed for IL-2 secretion by ELISA. The data shown is mean (±SEM) for 5-10 mice/group. *p < 0.05 between groups; Student's t-test. the machinery for clathrin coated pit endocytosis [36]. The dynamin-dependent endocytosis inhibitor, dynasore, significantly inhibited uptake of Hp91 by macrophages ( Figure 3C), which is in agreement with the finding reported for LPS [35]. The clathrin-mediated endocytosis inhibitor phenylarsine oxide also inhibited uptake of Hp91 by macrophages ( Figure 3C). In addition, blocking uptake of Hp91 via dynamin or clathrin mediated endocytosis significantly reduced IL-6 secretion ( Figure 3D). Since a reduction in Hp91-mediated IL-6 secretion was observed in human DCs following pretreatment with the p38MAPK and NF-κB inhibitors (Figure 4), we evaluated if the same was true for mouse macrophages. As was seen in human DCs, inhibition of p38MAPK or NF-κB signaling cascades significantly lowered the amount of Hp91-stimulated IL-6 secreted from mouse macrophages ( Figure 3D), further confirming the involvement of p38MAPK and NF-κB in Hp91 signaling.

DC activation by Hp91 peptide involves signaling via a p38 MAPK and NF-κB dependent pathway
Our previous findings have shown involvement of the p38MAPK pathway in the HMGB1 B box-induced secretion of IL-6 by human DCs [37]. Pretreatment of human DCs with the p38MAPK inhibitor SB203580, and the NF-κB inhibitor N-p-Tosyl-L-phenylalanine chloromethyl ketone (TPCK), significantly reduced the Hp91-induced IL-6 production in DCs (Figure 4). In contrast, PD98059, a MEK1 inhibitor, failed to significantly decrease the IL-6 production ( Figure 4). The p38MAPK pathway was further confirmed by immunoblotting for phosphorylated p38MAPK (p-p38). p-p38 was up-regulated in mouse BM-DCs after 40-60 min of Hp91 stimulation ( Figure 3B). In most experiments, up-regulation of p-p38 was observed as early as 20 minutes after Hp91 stimulation. These results suggest that Hp91 induced the DC activation in a p38MAPK and NF-κB pathway dependent manner.
Hp91 activates IRF3 and subsequent IFNα production in antigen presenting cells TLR4 can signal through MyD88 dependent and independent pathways, whereas other TLRs are MyD88dependent. Our data indicate that both pathways might be involved in Hp91 signaling since a decrease in Hp91induced IL-6 secretion was observed for both MyD88−/− and IFNαβR−/− knockout BM-DCs ( Figure 3A). We investigated if Hp91 activated the MyD88 independen signaling pathway by investigating phosphorylation of the downstream interferon regulatory factor 3 (IRF3) transcription factor. The phosphorylation of IRF3 (pIRF3) was up-regulated after Hp91 stimulation as early as 20 min ( Figure 3E) and at a maximum around 40 min after stimulation. In addition blocking Hp91 endocytosis with dynasore reduced the phosphorylation of IRF3 at 20, 40, and 60 min ( Figure 3E). To further investigate involvement of the MyD88-independent signaling pathway, we evaluated IFN-α mRNA expression in Hp91-stimulated mouse macrophages as this factor should be up-regulated by pIRF3. Although lower than the increase induced by LPS, Hp91 induced a 99-fold increase in IFN-α mRNA levels ( Figure 3F). These findings suggest that Hp91 uptake is required for MyD88-independent activation of the IRF3 pathway and production of IFN-α.

MyD88 signaling is not required for Hp91-elicited CTL responses in vivo
Since Hp91 activated both the MyD88 dependent as well as independent pathways ( Figure 3A and 3E), we tested if MyD88 was required for Hp91-induced CD8+ CTL responses in vivo. We immunized age-matched wild type and MyD88−/− knockout mice with OVA-I (SIINFEKL) peptide co-injected with PBS or Hp91. Both wild type and MyD88−/− knockout mice showed similar antigen-specific cellular immune responses, as demonstrated by increased number of OVA-specific IFN-γ secreting cells ( Figure 3G) suggesting that induction of this antigen specific CD8+ CTL response did not require MyD88 signaling. The lack of IL-6 secretion in MyD88−/− DCs after Hp91 exposure indicated that Hp91 can signal through MyD88, whereas the adjuvant effect elicited by Hp91 in vivo is MyD88 independent.

Discussion
Endogenous TLR agonists and inflammatory mediators are attractive candidates as vaccine adjuvants, especially for subunit vaccines that many times are poorly immunogenic, and the mechanism through which these types of adjutants augment immune responses is via the innate immunity. There is a great need for safe and potent adjuvants seeing that and we have previously shown that the 18 amino acid long immunostimulatory peptide Hp91, derived from the B box of the HMGB1 protein, is a potent stimulus for human DCs with the ability to generate a Th1-type immune response in vitro [32]. In addition, Hp91 acted as adjuvant in vivo, inducing cellular immune responses to peptide and both cellular and humoral immune responses to protein antigen [33]. In this study we characterize the mechanism of action for this adjuvant and show here that Hp91 exerted its immunostimulatory effects on DCs by inducing cellular uptake and activating signaling cascades. The Hp91 peptide was taken up into cells very rapidly via clathrin-mediated endocytosis in a sequence specific manner. Scrambling the amino acid sequence of Hp91 resulted in a great reduction of binding/uptake by cells, indicating that it is neither the total charge nor total hydrophobicity that is important for uptake rather the unique amino acid sequence. Further characterization showed that Hp91 mediated activation of DCs, i.e. IL-6 production in vitro, was dependent on TLR4, MyD88, and IFNαβR, whereas MyD88 signaling in vivo was not required for activation of CD8+ CTL responses.
Multiple HMGB1 binding and signaling events mediate activation of innate immune responses. The binding and uptake of Hp91 by DCs was a rapid and sequence dependent event and we found that the internalization occurred via clathrin dependent endocytosis. Since clathrin-mediated endocytosis is a receptor dependent uptake, we explored possible receptors involved in the uptake of Hp91. Several receptors have been implicated, including RAGE, TLR4, TLR2, CD24/Siglec-10. HMGB1 has been shown to contribute to LPS-mediated DC maturation via RAGE [18]. TLR2 and TLR4 have been shown to be involved in HMGB1 signaling in vitro [19][20][21][22][23]. In vivo data have shown binding and signaling through TLR4 to be involved in HMGB1-induced cytokine release, i.e. inflammation leading to tissue damage in the body [20,38].
The C-terminal motif of HMGB1 (150-183 amino acids) is responsible for RAGE binding [38], whereas the C-terminal end contains the TLR4 binding site. The Hp91 peptide is located in the B box area of HMGB1 protein and contains the TLR4 binding domain and we found that the ability to bind TLR4 was still intact in the Hp91 peptide.
It has been shown that LPS, a TLR4 ligand, binds TLR4, and is subsequently endocytosed together with the receptor [35] and this seems also to be the case for Hp91 peptide. By evaluating IL-6 secretion from knockout mice, we show that Hp91-stimulated activation of DCs is dependent on TLR4 and its downstream adaptor protein, MyD88 and further downstream signaling via p38MAPK and NF-κB. We have previously shown the involvement of the p38MAPK pathway in induction of IL-6 secretion in human DCs by the HMGB1 subunit B box [37] and others have shown that this pathway is involved in HMGB-1 induced activation of neutrophils [39]. This indicates that both the HMGB1 derived peptide and the whole protein have the ability to activate the p38MAPK pathway and that the activating sequence seems to be located within the Hp91 peptide.
Hp91 acts as adjuvant in vivo; inducing cellular immune responses to peptide and both cellular and humoral immune responses to protein antigen [33]. Immunization of MyD88−/− mice, using Hp91 as adjuvant, induced cellular immune responses comparable to WT mice. This suggests that even though MyD88 was required in vitro for Hp91-mediated IL-6 secretion in DCs, the Hp91induced cellular immune responses in vivo are MyD88independent. Furthermore, active uptake of Hp91 was required for signaling through this MyD88-independent pathway. Hp91 induces the in vivo production of Th1type cytokines, such as IL-12, and IFN-γ [33]. The Th1 cellular immune response is highly characterized by IFN responses and we found that Hp91-induced a signaling cascade activating the type I IFN pathway via IRF3, leading to elevated expression of IFN-α mRNA. Blocking the type I IFN receptor dampened the immune response to another potential adjuvant, i.e. Poly I:C, indicating that type I IFN play an important role in the immune activation induced by this adjuvant [40] Our data suggest that the activation of DCs by Hp91 is dependent on an autocrine type I IFN feedback loop as cells derived from IFNαβR knockout mouse failed to respond. This combined with the result showing that MyD88 was not necessary in vivo for cellular immune responses, suggests that the MyD88-independent pathway plays a prominent role in Hp91 induction of cellular immune responses in vivo.
These new findings provide a better understanding of the cellular mechanisms by which the immunogenic peptide induces potent immune responses. This peptide activates myeloid DCs and should be suitable as an adjuvant in cancer immunotherapies as well as vaccines against infectious diseases. BTM and DM wrote the paper and participated in manuscript revision. All authors read and approved the final manuscript.