Poly(I:C) is an adjuvant used for antitumor treatment and vaccines because of its prominent effects on CD8 T cells and NK cells. Poly(I:C) binds TLR3 and this interaction is thought to be central for driving cell-mediated immune responses. We investigated the importance of TLR3 in poly(I:C)-mediated endogenous CD8 T cell responses using the pathogenic T cell stimulant Staphylococcus aureus enterotoxin A. While the responsive CD8 T cells expanded comparably in both wild-type and TLR3−/− mice, differentiation of effector CD8 T cells was enhanced by poly(I:C) in the TLR3−/− mice. A higher percentage of Ag-specific CD8 T cells became IFN-γ and TNF-α producers in the absence of TLR3 signaling. Consistent with this boosted response was the observation that TLR3-deficient cells synthesized less IL-10 compared with TLR3-sufficient cells in response to poly(I:C). Ultimately, however, the fundamental mechanism of CD8 effector T cell differentiation through the TLR3-independent pathway was shown to be completely IFN-α/β-dependent. Administration of IFN-α/β-neutralizing Abs abolished the poly(I:C) effects in TLR3−/− mice. These findings reveal specific roles of how dsRNA receptors shape CD8 T cell responses, which should be considered as poly(I:C) is authenticated as a therapeutic adjuvant used in vaccines.
Double-stranded RNA is regarded as a pathogen-associated molecular pattern (PAMP),3 not merely because some viruses exist in the form of dsRNA, but also because dsRNA intermediates are often produced during viral replication. In view of its physiological importance, the cellular and molecular mechanisms by which dsRNA activate the immune system have been actively investigated. Poly(I:C) is a synthetic dsRNA that mimics the effects of naturally occurring dsRNA and is frequently used as an adjuvant. The adjuvant properties are likely a result of enhanced maturation of APCs, induction of type I IFNs (IFN-α/β), proinflammatory cytokines (IL-6, IL-12, and TNF-α), and the survival cytokine IL-15 (1, 2, 3, 4). The majority of poly(I:C)’s pleiotropic effects have been attributed to recognition by TLR3 and the signaling adaptor Toll/IL-1 receptor (TIR) domain-containing adaptor IFN-β (TRIF)/TIR domain-containing adaptor molecule-1 (TICAM1) (3). TLR3 functions as a dimer and is localized to the endosome with the ectodomains facing the lumen (5, 6). Although TLR3 is not positioned where it can directly detect extracellular dsRNA, it is thought that receptor-ligand recognition is facilitated by cell surface CD14 (7).
In addition to the discovery of TLR3, other dsRNA receptors have been uncovered. Retinoic acid-induced gene I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) are cytosolic RNA helicases that play an important role in antiviral immunity. Their cellular location suggests that these are the prime sensors of viral dsRNA as viral replication occurs in the cytoplasm. Both RIG-I and MDA5 depend on CARDIF (also known as IPS-1/MAVS/VISA), a mitochondria-associated protein, for downstream activation of NF-κB and IFN regulatory factor (8, 9). LGP2, another dsRNA receptor, is also located in the cytoplasm. LGP2 binds dsRNA but unlike RIG-I or MDA5, it does not interact with CARDIF because it lacks a CARD domain that is required for receptor-adaptor interaction. Hence, ligand binding to LGP2 fails to induce type I IFN production (10), but instead LGP2 could be a negative regulator for other cytosolic dsRNA receptors.
The existence of multiple dsRNA receptors in the cell may reflect attempts of an evolving immune system to cope with the diversity of viruses. TLR3−/−, RIG-I−/−, and MDA5−/− mice are each susceptible or resistant to specific viruses (11, 12, 13). The location of TLR3 indicates that it detects dsRNA from an extracellular source. However, most viruses do not exist in dsRNA form but they do produce dsRNA intermediates as they replicate inside the cell. Hence, cytosolic receptors may be more suited for detecting viral infection. Nevertheless, TLR3 has a determining role on the outcome of certain virus infections. During West Nile virus, vaccinia virus, and phlebovirus infection, less severe inflammatory reactions were documented in TLR3−/− mice (14, 15, 16). It is thought that this dampened inflammatory response improved survival of TLR3−/− mice due to reduced pathological organ damage. On the other hand, TLR3−/− mice were shown to be more susceptible to encephalomyocarditis virus and murine cytomegalovirus infections (17, 18). In these models, higher viral load in the TLR3−/− mice correlated with decreased production of proinflammatory cytokines. Nevertheless, the effects of TLR3 deficiency on adaptive immune responses have rarely been examined. This is a critical issue since there have been limited attempts to dissect the requirement of TLR3 vs other dsRNA receptors during the T cell adjuvant effects of poly(I:C).
In this study, staphylococcal enterotoxin A (SEA) was used as a powerful pathogenic T cell stimulant. SEA is one of several superantigens produced by Staphylococcus aureus and is associated with food poisoning outbreaks, toxic shock, and recently respiratory diseases (19, 20). SEA crosslinks MHC II on APCs with the TCR Vβ1, Vβ3, Vβ10, Vβ11, or Vβ17 chains on T cells (21). Thus, following immunization with SEA, endogenous CD8 and CD4 T cell expansion and effector differentiation are incredibly robust.
Herein, it is demonstrated that poly(I:C) preferentially induced CD8 Vβ3 T cell expansion over CD4. Second, although TLR3 pathway deficiency did not significantly alter the magnitude of CD8 T cell expansion, effector differentiation was actually enhanced in the absence of TLR3. To better understand this counterintuitive result, cells from TLR3−/− mice were analyzed against wild-type mice for cytokine output in response to PAMPs. The TLR3-independent pathway induced high amounts of the immunosuppressive cytokine IL-10 in response to CpG but not in response to poly(I:C), while wild-type cells responded well to each PAMP. Although IL-10 may suppress effector differentiation (22), we postulated that IFN-γ and cell killing potential was fundamentally dependent on the presence of innate cytokines, not the absence of immunosuppressive ones. Thus, we demonstrated that CD8 effector differentiation was completely dependent on TLR3-independent production of IFN-α/β. These data suggest that efficacious therapeutic use of poly(I:C) requires careful consideration in targeting the desired dsRNA receptor pathway.
Materials and Methods
Mice and reagents
C57BL/6 mice were purchased from The Jackson Laboratory and the National Cancer Institute-Frederick (Frederick, MD). TRIF-deficient mice on the C57BL/6 background were purchased from The Jackson Laboratory (strain name: C57BL/6J-TicamLps2/J, stock no. 005037). TLR3−/− mice on the C57BL/6 background were a kind gift from Dr. Richard Flavell (Yale University, New Haven, CT), and preliminary experiments using B6 × 129 TLR3−/− mice were purchased from The Jackson Laboratory. All mice were housed and bred in the Center for Laboratory Animal Care of the University of Connecticut Health Center and were handled according to the National Institutes of Health federal guidelines. SEA was purchased from Toxin Technology. Poly(I:C) was purchased from InvivoGen and Alexis Biochemicals (Axxora). CpG was purchased from Midland Certified Reagent. LPS, derived from Salmonella typhimurium, was purchased from Sigma-Aldrich. IFN-α/β-neutralizing serum was generated by one of the authors (M.G.T.) in sheep along with a control Ig. Quantity of reagents used for i.p. injection was 1 μg SEA, 40 μg poly(I:C), or 50 μg CpG. Reagents were premixed in balanced salt solution (BSS) supplemented with HEPES, l-glutamine, penicillin, streptomycin, and gentamicin sulfate in a total volume of 200 μl before injection. IL-6, IL-10, IFN-γ, and TNF-α ELISA kits were purchased from BD Biosciences.
Cell processing and flow cytometry
Spleens were crushed through cell strainers (BD Falcon) and were treated with Gey’s solution to lyse RBC before staining or in vitro culture. For experiments involving liver and lung lymphocytes, animals were first perfused with PBS containing heparin (Sigma-Aldrich) at 75 U/ml. Livers were crushed through cell strainers and the cell suspension was partitioned on 35% Percoll (Sigma-Aldrich) to obtain lymphocytes. Remaining RBC in the samples were lysed with Gey’s solution. Lungs were cut into smaller pieces, incubated in BSS containing 1.3 mM EDTA (pH 7) at 37°C for 30 min, followed by digestion with collagenase (Clostridium histolyticum, type IV) (Sigma-Aldrich) at 37°C for 1 h. Subsequently, digested lung samples were crushed through cell strainers and partitioned on 44% and 67% Percoll (Amersham Biosciences) gradient to obtain lymphocytes at the interphase.
For surface staining, cells were incubated with fluorochrome-conjugated Abs in BSS supplemented with 3% FBS and 0.1% sodium azide on ice for 30 min. For intracellular staining (granzyme B, IFN-γ, and TNF-α), cells were fixed with 2% paraformaldehyde, permeabilized with 0.25% saponin before staining. All samples were analyzed using a FACSCalibur instrument (BD Biosciences) and data analysis was performed using FlowJo software (Tree Star).
Intracellular cytokine staining
One million splenocytes from immunized mice were cultured in the presence of 5 μg/ml brefeldin A and 1 μg/ml SEA in a total volume of 200 μl CTM (MEM supplemented with l-glutamine, FCS, tumor cocktail). Cells were incubated at 37°C in 5% CO2 for 5 h. IFN-γ and TNF-α staining in activated Ag-specific CD8 T cells was determined by gating on CD4-negative Vβ3 cells due to technical difficulty on CD8 staining after fixation.
CD107a and CD107b analysis
Five hundred thousand hepatic lymphocytes were cultured in the presence of 1 μg/ml SEA in a total volume of 200 μl CTM. Cells were incubated at 37°C in 5% CO2 for 2 h. At the end of 2 h, cells were immediately stained with anti-CD107a, anti-CD107b, anti-CD8, and anti-Vβ3 Abs and analyzed.
In vitro culture
One million splenocytes from naive wild-type or TLR3−/− mice were cultured in a final volume of 200 μl CTM. Variable concentrations of poly(I:C), CpG, or LPS were added and cells were incubated at 37°C in 5% CO2. A day later, supernatant was collected and 100 μl was used for IL-6, IL-10, and IFN-γ ELISAs according to the manufacturer’s protocol.
Type I IFN neutralization
The generation and purification of anti-IFN-α/β Abs were previously described in detail (23). To neutralize IFN-α/β in vivo, TLR3−/− mice were injected i.p. with 200 μl sheep anti-mouse IFN-α/β Abs or control Abs at 12 h before immunization and 24 h after immunization. Neutralization was confirmed by staining peripheral blood CD4 and CD8 T cells for Ly6A/E surface marker at 24 h postimmunization. During the course of four experiments encompassing nine mice treated with anti-IFN-α/β, one mouse failed to neutralize IFN-α/β, as exemplified by control staining showing surface up-regulation of Ly6A/E on peripheral lymphocytes. Data obtained from this mouse were eliminated based on the control staining, and hence were excluded in the analysis.
NK cell depletion
Depletion of NK1.1+ cells was accomplished by using PK136 Ab and corresponding isotype control Ab (24). Twenty-five microliters of PK136 was i.p. injected into each mouse at 18 h before immunization with SEA. Depletion was confirmed by staining peripheral blood with Abs against CD94 and Dx5 markers before immunization.
Three million splenocytes from naive or poly(I:C)-treated animals were collected for RNA extraction using RNeasy mini kit (Qiagen) according to the manufacturer’s protocol. RNA samples were further treated with RNase-free DNase I (Qiagen) to eliminate DNA contamination. Reverse transcription was performed with Moloney murine leukemia virus reverse transcriptase (Qiagen) using random primers. Recombinant TaqDNA polymerase (Invitrogen) was used for RT-PCR, and PCR products were resolved on 6% polyacylamide gel. Primer sequences for amplifying type I IFN cDNAs were as follows: IFN-α forward, CCTGATGGTCTTGGTGGTGAT, IFN-α reverse, CAGTTCCTTCATCCCGACCAG; IFN-β forward, AGCTCCAGCTCCAAGAAAGGACGAACAT, IFN-β reverse, GCCCTGTAGGTGAGGTTGATCT; β-actin forward, GTGGGCCGCTCTAGGCACCA, β-actin reverse, CTCTTTGATGTCACGCACGA.
Unpaired, two-tailed, and equal variance Student’s t tests were performed for all data shown. Error bars indicate SE of mean.
Poly(I:C) enhances T cell expansion in vivo in a TLR3- and TRIF-independent manner
SEA is a well-characterized pathogenic protein that we utilized to study endogenous T cell expansion in TLR3−/− mice. SEA activates endogenous T cells that express Vβ3 TCR but not those that express Vβ6. In wild-type mice, poly(I:C) increased the frequency of Vβ3+ T cells within the CD8 population by ∼3-fold compared with SEA immunization alone (Fig. 1,A). The dose of poly(I:C) was based on titration studies (data not shown). We hypothesized that since poly(I:C) was administered in a soluble form but not in complex with any transfecting reagent, it would be detected by endocytosis. Consequently, its adjuvant effects should be mediated through the TLR3 pathway in the endosomal compartment. We predicted that poly(I:C) would fail to enhance CD8 T cell expansion in TLR3−/− and TRIF-deficient mice; however, the expansion of CD8+Vβ3+ T cells was not impaired in response to poly(I:C) (Fig. 1,A). Likewise, total numbers of CD8+Vβ3+ cells in the spleen of knockout mice were increased by poly(I:C) immunization (data not shown). The frequency of Vβ6+ control T cells within the CD8 population was not increased by poly(I:C) treatment (Fig. 1,A, left), showing that in this model the effects of poly(I:C) can only be detected with SEA-activated T cells. In comparison to CD8 T cells, poly(I:C) had a less dramatic but nonetheless significant effect on CD4 T cell expansion. After poly(I:C) treatment, the frequency of Vβ3+ within the CD4 population was increased by ∼40% in wild-type mice, showed no significant increase in TLR3−/− mice, and was increased by ∼80% in TRIF-deficient mice (Fig. 1,B). Interestingly, we found that TRIF-deficient mice showed greater T cell expansion when immunized with poly(I:C) over wild-type mice (Fig. 1 A), but this was not observed to the same extent in the TLR3−/− mice. Thus, the TRIF pathway may have a suppressive effect on T cell expansion that is not related to TLR3 signaling. At present the basis of this hyperresponsiveness to poly(I:C) is unknown. Data presented here suggest that poly(I:C) could preferentially influence CD8 T cell expansion in vivo in a TLR3/TRIF-independent manner.
TLR3 or TRIF deficiency improved CD8 effector T cell differentiation
Naive CD8 T cells can differentiate into effector killer T cells that display an array of cellular functions. Cues provided by adjuvants not only augment the proliferation of primed naive CD8 T cells, but also enhance such effector functions. To investigate the importance of TLR3 signaling on poly(I:C)-induced CD8 effector T cell differentiation, we investigated the ability of in vivo primed CD8+Vβ3+ T cells to synthesize IFN-γ and TNF-α. Wild-type, TLR3−/−, and TRIF-deficient mice were immunized with SEA and poly(I:C) and 5 days later total splenocytes were isolated and restimulated with SEA for 5 h. IFN-γ and TNF-α production was measured by intracellular staining, and the frequency of CD8 effectors was obtained by gating on CD4−Vβ3+ cells. Typically, within the wild-type CD8+Vβ3+ population, 10% are effector cytokine producers with this vaccine dose. Surprisingly, this effector population was increased to 20–25% in the absence of TLR3 or TRIF signaling (Fig. 2 A). This demonstrated that although TLR3 signaling does not significantly alter the magnitude of CD8 expansion, it may hinder CD8 effector differentiation. One possible means in which TLR3 could dictate effector T cell activation is through fine-tuning the production of specific cytokines that impact the microenvironment.
The balance of proinflammatory and immunosuppressive cytokines early after immunization can be a major determinant of the development of adaptive responses. Some examples of these cytokines, such as TNF-α, IL-6, and IL-10, are known to be induced by poly(I:C) (25). Selective production or absence of these cytokines may alter effector differentiation in TLR3−/− mice, in particular, the immunosuppressive cytokine IL-10, which is known to suppress IFN-γ and TNF-α production (26, 27). Since we were not able to detect IL-10 in the serum of either wild-type or TLR3−/− mice at various time points (1, 2, 4, 12, 24 h) after poly(I:C) injection (data not shown), an alternative approach was taken. Splenocytes from naive wild-type or TLR3−/− mice were treated with various doses of poly(I:C) in vitro, and IL-10 production in the culture supernatant was measured by ELISA 24 h later. Poly(I:C) increased IL-10 production in a dose-dependent manner and this relied on the presence of TLR3 (Fig. 2,B). The inability of TLR3−/− splenocytes to produce IL-10 was not an inherent defect associated with the genetic deficiency because CpG DNA (a TLR9 ligand) was able to induce IL-10 from TLR3−/− splenocytes. TLR3−/− cells also failed to synthesize IFN-γ during the 24-h culture. Secretion of IL-10 and IFN-γ are likely due to activation of innate cell types upon poly(I:C) binding since there was no Ag added in the culture medium. In agreement with published observations, poly(I:C) induced the expression of proinflammatory cytokines TNF-α and IL-6. Both TNF-α and IL-6 peaked rapidly, at 2 h, in the serum of wild-type mice after immunization with poly(I:C) (Fig. 2 C). While TNF-α production is solely dependent on TLR3, IL-6 production could be achieved in a TLR3-independent manner, even though it was delayed and to a lower magnitude compared with wild-type mice. Thus, poly(I:C) failed to induce IL-10, IFN-γ, and TNF-α in the absence of TLR3 signaling.
TLR3-independent CD8 expansion and effector differentiation induced by poly(I:C) required type I IFNs
IFN-α and IFN-β are the two most studied type I IFNs, and they interact with the same receptor and influence T cell activation (28). Due to the compensatory nature of cytosolic dsRNA receptors, TLR3−/− mice were capable of producing type I IFN in response to poly(I:C) as shown by RT-PCR using total splenocytes from mice that were immunized with poly(I:C) at 1, 2, or 4 h earlier (Fig. 3,A). However, it is noteworthy that type I IFN transcripts in TLR3−/− mice appeared an hour later compared with wild-type mice. To determine whether type I IFN plays a major role in T cell expansion and effector function in TLR3−/− mice, type I IFN was neutralized in vivo. TLR3−/− mice were injected with two doses of IFN-α/β-neutralizing Abs before and after immunization. The blocking effects of IFN-α/β-neutralizing Abs were validated by examining Ly6A/E (ScaI) expression on peripheral blood CD8 T cells at 48 h postimmunization (Fig. 3 B). Ly6A/E is a target gene for IFNs whose transcription is initiated once cells sense IFN-α/β/γ (29). Immunization with SEA alone induced expression of Ly6A/E over that of naive cells, indicating that IFNs were produced. Treatment with poly(I:C) further increased Ly6A/E expression, indicating increased IFN-α/β synthesis induced by poly(I:C). IFN-α/β neutralization decreased Ly6A/E expression to a level lower than SEA immunization alone, but not to the level displayed by naive CD8 T cells. This suggests that the remaining Ly6A/E expression after type I IFN neutralization might be due to the production of type II IFN, and other cytokines resulted from TCR activation, such as IFN-γ and TNF (30).
In the peripheral blood of SEA- and poly(I:C)-treated mice, peak of expansion of CD8+Vβ3+ T cells occurred on days 2–3 postimmunization. At this point, nearly 20% of the CD8 T cells in the circulation were Vβ3+ cells. IFN-α/β blockade completely abolished this effect by rendering the magnitude of expansion back down to the level seen with SEA alone (Fig. 3,C). By day 5, enhanced accumulation of CD8+Vβ3+ T cells in both secondary lymphoid organ (spleen) and tissues (liver and lung) of poly(I:C)-immunized mice could also be observed. Similar to data obtained with peripheral blood, neutralization of IFN-α/β early in the response significantly reduced frequency and numbers of CD8+Vβ3+ T cells in poly(I:C)-treated mice (Fig. 3 D and data not shown).
In addition to CD8 T cell expansion, effector functions were also severely impaired in the presence of type I IFN neutralization. Granzyme B production, degranulation, and effector cytokine production by activated CD8 T cells were tested (Fig. 4). Poly(I:C) increased granzyme B production by 2.8-fold in liver CD8+Vβ3+ T cells and by 4.3-fold in lung CD8+Vβ3+ T cells over SEA immunization alone (Fig. 4,A and data not shown). To effectively kill target cells, granzyme B is released within close vicinity of target cells and this act of degranulation results in CD107a (LAMP1) or CD107b (LAMP2) expression on T cell surfaces (31). Under normal circumstances these markers are detected on intracellular vesicles, but when cells degranulate, CD107a and CD107b transiently appear on the cell surface and can be experimentally detected by flow cytometry. We restimulated day 5 hepatic lymphocytes in vitro with SEA for 2 h, an optimized time point, to study degranulation by CD8+Vβ3+ T cells derived from TLR3−/− mice. The ability of activated CD8+Vβ3+ T cells to degranulate paralleled that of intracellular granzyme B content (Fig. 4,B and data not shown), as well as the capacity of these T cells to rapidly synthesize IFN-γ and TNF-α (Fig. 4 C and data not shown). Consistent with CD8 T cell expansion, neutralization of type I IFNs simultaneously lowered multiple aspects of T cell effector functions in TLR3−/− mice that received poly(I:C). These data showed that TLR3 signaling was not absolutely required for poly(I:C)-induced T cell clonal expansion and effector differentiation. In contrast, type I IFN, which is a downstream product of poly(I:C) binding, played an essential role in developing the adaptive response.
NK cell activation by poly(I:C) is TLR3-independent but type I IFN-dependent
Similar to CD8 T cells, NK cells can be activated by poly(I:C) (32), and it was commonly thought that this is related to TLR3 activation, although NK cells express all known activating dsRNA receptors (33). Wild-type or TLR3−/− mice were immunized with SEA with or without poly(I:C). A day later, lung lymphocyte populations were isolated and NK activation status was analyzed by examining CD69 staining on NK1.1+Dx5+ cells. We chose lung and liver because they contain a higher proportion of NK cells within their lymphocyte population (data not shown). Immunization with SEA alone induced activation of approximately one-fourth of lung NK cells, whereas addition of poly(I:C) converted nearly all NK cells into the activated phenotype (Fig. 5,A). The majority of both wild-type and TLR3−/− NK cells became CD69+ when mice were treated with poly(I:C), demonstrating that the absence of TLR3 signaling does not affect NK activation. Similar observations were made in peripheral blood, spleen, and liver (data not shown). Since TLR3 played a negligible role in poly(I:C)-induced NK activation, we next neutralized type I IFN in TLR3−/− mice to examine the requirement of these cytokines for NK activation. NK activation was followed by monitoring CD69 up-regulation on peripheral blood NK1.1+Dx5+ cells during the course of 4 days postimmunization. CD69 expression peaked at 24 h in SEA- and poly(I:C)-immunized mice, which then slowly diminished in the next 2 days (Fig. 5 B). Blocking type I IFNs prevented the up-regulation of CD69, implying that NK cells, similar to CD8 T cells, relied on induction of type I IFN for activation. However, unlike CD8 T cells, whose expansion depended on type I IFN, NK cell frequency in the peripheral blood was not altered by type I IFNs in TLR3−/− mice (data not shown).
It was reported that poly(I:C) enhanced CD8 T cell responses partially through NK cells (34). Therefore, lack of NK activation raised the notion that this might be the reason for weakened CD8 T cell responses after type I IFN neutralization. To address this possibility, we depleted NK cells in wild-type mice before immunization with SEA and poly(I:C) and analyzed CD8 T cell responses 5 days later. Depletion was accomplished by using PK136 Ab that targets NK1.1+ cells. Frequency of NK cells was subsequently determined by using Dx5 and CD94 markers, since PK136 Ab occupied the binding site of NK1.1. The NK1.1+ population was successfully depleted in spleen when examined on day 5 (Fig. 5,C), and even at 18 h postinjection NK1.1+ cells were already absent from the peripheral blood (data not shown). Poly(I:C) enhanced splenic expansion of CD8+Vβ3+ T cells by 3.7-fold even with depletion of NK1.1+ cells, similar to nondepleted mice (Fig. 5,D). In fact, the number of CD8+Vβ3+ T cells was 50% higher in the spleens of NK1.1+ cell-depleted mice (data not shown). Importantly, this effect translated to significant increases in percentages (up by 55.5%) and total numbers (up by 2.5-fold) of effector CD8 T cells (Fig. 5 E and data not shown). We conclude that although type I IFN neutralization inhibited both NK and CD8 populations, the effect on NK cells is unlikely to have any significant impact on CD8 activation in this model.
TLR3 deficiency resulted in reduced poly(I:C)-mediated CD8 T cell memory
Poly(I:C) has been shown to enhance both primary and memory CD8 T cell responses using the OT-I adoptive transfer system (35). Wild-type or TLR3−/− mice were immunized with SEA with or without poly(I:C), and CD8 T cell memory response was assessed 3 mo later. In a representative experiment, within the CD8 population, poly(I:C) treatment expanded CD11ahigh memory Vβ3+ T cells by >10-fold in liver and lung of wild-type mice (Fig. 6,A). The CD8+Vβ3+ memory population was reduced to less than half in the TLR3−/− mice compared with wild-type mice. On average, wild-type mice showed an 8-fold increase in the frequency and numbers of CD11ahighCD8+Vβ3+ memory T cells in the lungs (Fig. 6,B), and similar data were also seen in the liver (Fig. 6,C). In one experiment, however, we detected an aberrantly high number CD11ahighCD8+Vβ3+ memory T cells in a TLR3−/− mouse, which caused us to eliminate this mouse from our analysis since we reasoned it was not representative of the other data, and we have no explanation for this result. TLR3−/− mice had a 2.4-fold increase in the frequency of CD11ahigh memory Vβ3+ T cells in the lungs, but this increase was not statistically significant (Fig. 6,B), while in the liver the trend was still less in the TLR3−/− mice but was statistically significant (Fig. 6,C). It was possible that the overall decreased memory response in the TLR3−/− mice reflected an intrinsic weakness to support the maintenance of memory T cells, and therefore in five of the experiments we included a control group using a different adjuvant. Unlike poly(I:C), TLR3−/− mice exhibited a normal memory response to immunization with the TLR9 ligand CpG (Fig. 6 A). This finding suggests that TLR3 is required for complete memory response to poly(I:C) immunization.
Even though fewer memory CD8 T cells were recovered from TLR3−/− mice, they were not impaired in effector functions. We examined intracellular granzyme B staining of ex vivo CD11ahighVβ3+ T cells from lung and liver and found that granzyme B expression in TLR3−/− memory CD8 T cells was similar to their wild-type counterparts (data not shown). Thus, while TLR3 was unnecessary for effector cell differentiation in response to poly(I:C), it appeared to play an important role for memory T cell survival.
The role of TLR3 for cytokine production has been extensively studied in the past using innate immune cell types (36, 37). Comparatively speaking, relatively little has been done to evaluate the significance of this receptor on adaptive response. Our data herein show that the TLR3 deficiency does not impede CD8 T cell expansion, but its absence may even be beneficial for CD8 effector differentiation in response to poly(I:C). This observation demonstrated that a relatively low dose of poly(I:C) mediated an adaptive response without binding to TLR3 in T cells, or to innate cell types. These are unexpected findings given the difference in cellular location between TLR3 and other dsRNA receptors.
The major mechanism poly(I:C) uses to induce CD8 T cell responses is through the action of type I IFN. Virtually every aspect of the CD8 T cell response we examined, from proliferation to production of effector molecules, requires the presence of type I IFN. In comparison to CD8, poly(I:C) showed minimal effects on CD4 expansion at the dosage used here, perhaps because CD4 T cells are less sensitive to type I IFNs. This notion is supported by two observations: first, poly(I:C) beyond 40 μg per mouse could augment CD4 expansion in a dose-dependent manner (data not shown), and second, during viral infection where type I IFNs are abundant, the CD4 response is affected by these cytokines (38). On the other hand, although it is becoming clear that soluble poly(I:C) could trigger MDA5 for type I IFN transcription (39), there is little understanding on how soluble poly(I:C) enters the cytosol, where MDA5 is located, since dsRNA is presumably cell-impermeable. One possibility is that poly(I:C) enters through cell surface transporters. One example of such dsRNA transporters was found in Caenorhabditis elegans (40), but there is not yet any evidence of a similar transporter on mammalian cells.
The action of type I IFNs is not restricted to promoting CD8 T cell responses alone. Reports from other laboratories (41, 42) and our data herein revealed that these cytokines also play an important role in the NK cell response. CD8 T cells and NK cells are important components of antitumor immunity, in which each cell type separately targets MHC-positive and MHC-negative tumors. Poly(I:C) as a type I IFN inducer may boost CD8 and NK cell activation, representing a promising therapeutic. These two cell types are remarkably similar in their TLR3 independence and requirement of type I IFN during poly(I:C)-induced activation. However, the activation of each cell type may be independent of one another since NK1.1+ cell depletion did not affect CD8 T cells (Fig. 5). Contrary to our expectation, the absence of NK1.1+ cells resulted in better CD8 effector differentiation. One possible explanation is that the elimination of NK1.1+ cells prevented the early release of IFN-γ, which can have a suppressive effect on CD8 T cells (43), thereby increasing Ag-specific CD8 T effectors in both frequency and numbers. Alternatively, PK136 Ab depleted additional NK1.1+ cell subsets that may be suppressive to CD8 effector differentiation, since other cell types, such as NK T cells, some TCRαβ T cells, and B220+ cells also express the NK1.1 marker (44, 45).
One novel finding from our study is the enhanced effector differentiation of CD8 T cells in the absence of TLR3 signaling. This phenomenon does not occur on the level of a single cell, but rather represents an increase in conversion of more activated CD8 T cells into effector cytokine producers. This may be related to a skewed cytokine milieu such as reduced levels of IFN-γ and IL-10 (Fig. 2, B and C), thereby enhancing better effector T cell differentiation in TLR3−/− mice. Thus, IFN-γ and IL-10 may create an immunosuppressive microenvironment in vivo. Although multiple studies showed that IFN-γ can negatively regulate CD8 T cell responses (43, 46), there are also reports indicating that IFN-γ is stimulatory for CD8 T cells during lymphocytic choriomeningitis virus infection (47). On the other hand, IL-10 is well known for its immunosuppressive effects as an inhibitor of proinflammatory cytokines (48). Excessive IL-10 provided in the form of recombinant protein or in transgenic mice increases susceptibility to many pathogens (49, 50), whereas deficiency in IL-10 results in resistance to various diseases (51) but also susceptibility to inflammatory bowel disease (52). Hence, an important role for IL-10 appears to be as a regulator of effective immunity vs injurious pathology.
More recently, macrophages and dendritic cells were shown to be a source of IL-10 when stimulated with TLR ligands (25). Similarly, TLR ligand-activated B cells antagonized IFN-γ production by CD4 T cells through an IL-10-dependent mechanism (53). In addition to its antiinflammatory properties and suppression on CD4 T cells, IL-10 negatively impacts CD8 responses through the induction of regulatory T cells (54, 55) and, in turn, regulatory T cells suppress IFN-γ production by effector CD8 T cells through IL-10 (56). Our hypothesis is that TLR3 triggers release of IL-10 and prevents poly(I:C) from attaining its full potential in the activation of the adaptive response during the priming stage. This does not imply that cells will permanently lose their capacity to produce IL-10. IL-10 transcripts are present in the cells at a basal level and their expression is also regulated at a posttranscriptional level (57). It is not clear how TLR triggering leads to further IL-10 transcription. One possibility is through inducible factors related to TLR activation because several NF-κB-like recognition sites and IL-6-responsive element are detected on the noncoding region of the murine IL-10 gene (58). Second, Stat3 was shown to be involved in LPS-induced IL-10 transcription in B cells (59). Nevertheless, these findings point out the value of blocking TLR3 signaling while using poly(I:C) as an adjuvant to prevent IL-10 production as an approach to optimize CD8 T cell responses. Alternatively, TLR3 signaling can be bypassed by transfecting poly(I:C) or by designing small chemical compounds that can permeate through cell membranes and specifically target MDA5. This is especially crucial in situations where poly(I:C) is to be used for reactivation of existing CD8 T effectors at sites in which established tolerance frequently involved the presence of IL-10 (60), such as tissues harboring tumors.
Eliciting a robust effector T cell response and generating a functional memory population are equally important criteria to consider when choosing an adjuvant for vaccination. Recently, it has been suggested that the fate of activated CD8 T cells to become effector or memory T cells depends on the level of proinflammatory cytokines such as IL-12 (61). We were not able to detect IL-12 in serum or in vitro at various time points after poly(I:C) treatment, but other investigators demonstrated that IL-12 production can be TLR3-dependent (2). Another proinflammatory cytokine that is differently regulated in wild-type and TLR3−/− mice is TNF-α. TNF-α was shown to limit the size of the CD8 memory population by promoting T cell apoptosis during the effector phase of lymphocytic choriomeningitis virus infection (62). The absence of these inflammatory cytokines in TLR3−/− mice seemingly did not explain the reduction of CD8+Vβ3+ memory T cells in this study. These data suggest that another cytokine(s) that is essential for CD8 survival may have been overlooked, such as IL-2, IL-15, or IFN-α/β. IL-15 sustains survival of CD8 T cells with an activated phenotype (CD44high) and is thought to act predominantly in a membrane-associated form (63, 64). As poly(I:C) induces IL-15 expression through type I IFN (4), the survival effects on CD8 T cells by type I IFNs may be partially dependent on IL-15. However, there is also evidence indicating that CD8 T cell survival requires direct action of type I IFNs (28, 65). In TLR3−/− mice, IFN-α/β transcription is delayed and transcripts were produced in lower quantity after poly(I:C) treatment (Fig. 3 A). The reduced level of IFN-α/β may be sufficient for sustaining short-term CD8 T cell expansion and activation but jeopardized long-term survival. Increasing dosage of poly(I:C) may perhaps salvage the induction of type I IFNs as more poly(I:C) in uncomplexed form will be able to reach the intracellular MDA5 receptor. Whether this could rescue memory CD8 T cell development remains to be tested. However, clinically speaking, high dosage of poly(I:C) is not recommended, as toxic reactions were recorded during earlier clinical trials for cancer treatment (66, 67).
Finally, as poly(I:C) is being modified into a less toxic derivative (poly(I:C)12U) and developed as a therapeutic for the treatment of AIDS, various tumors, and chronic fatigue syndrome (68, 69), an in-depth understanding of the cellular and molecular mechanism of how this molecule initiates immunological responses would help to direct its clinical usefulness. A recent report suggested that poly(I:C)12U protects against Punta Toro virus infection and this effect is mediated through recognition by TLR3 (70). Since TLR3 activation triggers IL-10 production, as shown by our study here, it remains to be investigated whether neutralization of immunosuppressive cytokines will synergize with such therapy or whether selective stimulation of MDA5 would be more efficacious. In conclusion, we found that the synthetic dsRNA poly(I:C) enhances CD8 adaptive response in a TLR3-independent but type I IFNs-dependent manner.
We thank Dr. Leonardo Aguila (University of Connecticut Health Center, Farmington, CT) for providing the PK136 Ab and for helpful advice, and Dr. Adam Adler for helpful advice and for carefully reading our manuscript.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants RO1 AI428585 and AI52108.
Abbreviations used in this paper: PAMP, pathogen-associated molecular patterns; BSS, balanced salt solution; CTM, MEM supplemented with l-glutamine, FCS, tumor cocktail; MDA5, melanoma differentiation-associated protein 5; RIG-I, retinoic acid-inducible gene I; SEA, staphylococcal enterotoxin A; TRIF, Toll/IL-1R domain-containing protein inducing IFN-β.