Discordant rearrangement of primary and anamnestic CD8+ T cell responses to influenza A viral epitopes upon exposure to bacterial superantigens: Implications for prophylactic vaccination, heterosubtypic immunity and superinfections

Infection with (SAg)-producing bacteria may precede or follow infection with or vaccination against influenza A viruses (IAVs). However, how SAgs alter the breadth of IAV-specific CD8+ T cell (TCD8) responses is unknown. Moreover, whether recall responses mediating heterosubtypic immunity to IAVs are manipulated by SAgs remains unexplored. We employed wild-type (WT) and mutant bacterial SAgs, SAg-sufficient/deficient Staphylococcus aureus strains, and WT, mouse-adapted and reassortant IAV strains in multiple in vivo settings to address the above questions. Contrary to the popular view that SAgs delete or anergize T cells, systemic administration of staphylococcal enterotoxin B (SEB) or Mycoplasma arthritidis mitogen before intraperitoneal IAV immunization enlarged the clonal size of ‘select’ IAV-specific TCD8 and reshuffled the hierarchical pattern of primary TCD8 responses. This was mechanistically linked to the TCR Vβ makeup of the impacted clones rather than their immunodominance status. Importantly, SAg-expanded TCD8 retained their IFN-γ production and cognate cytolytic capacities. The enhancing effect of SEB on immunodominant TCD8 was also evident in primary responses to vaccination with heat-inactivated and live attenuated IAV strains administered intramuscularly and intranasally, respectively. Interestingly, in prime-boost immunization settings, the outcome of SEB administration depended strictly upon the time point at which this SAg was introduced. Accordingly, SEB injection before priming raised CD127highKLRG1low memory precursor frequencies and augmented the anamnestic responses of SEB-binding TCD8. By comparison, introducing SEB before boosting diminished recall responses to IAV-derived epitopes drastically and indiscriminately. This was accompanied by lower Ki67 and higher Fas, LAG-3 and PD-1 levels consistent with a pro-apoptotic and/or exhausted phenotype. Therefore, SAgs can have contrasting impacts on anti-IAV immunity depending on the naïve/memory status and the TCR composition of exposed TCD8. Finally, local administration of SEB or infection with SEB-producing S. aureus enhanced pulmonary TCD8 responses to IAV. Our findings have clear implications for superinfections and prophylactic vaccination.

We previously demonstrated that Gram-positive bacterial SAgs can also activate invariant natural killer T (iNKT) and mucosa-associated invariant T (MAIT) cells [11,12]. This triggers a rapid and heavy-handed inflammatory response, typified by a 'cytokine storm', with rebound immunosuppression and susceptibility to potentially fatal secondary infections. Severe IAV infections can on their own cause a cytokine storm in a strain-dependent manner. This creates a complex scenario involving numerous inflammatory mediators with pathogenic or protective roles in host defense against IAV and secondary bacterial infections [13].
T CD8 play a pivotal part in clearing IAV infections. Naïve IAV-specific T CD8 are primed by professional antigen (Ag)-presenting cells (APCs) displaying cognate peptide:MHC I complexes along with costimulatory signals [14]. They then proliferate and differentiate into fully armed cytotoxic T lymphocytes (CTLs) that utilize cytolytic effector molecules (e.g., perforin and granzymes) to destroy infected cells as well as inflammatory cytokines that limit viral replication and transmission [15].
How bacterial SAgs manipulate IAV infection-and vaccination-elicited T CD8 responses is poorly understood. Exposure to SAgs can create physical and/or functional holes in mouse T cell repertoires by eliminating or anergizing many T cells [16][17][18]. This is assumed to put a fraction of T CD8 with antiviral specificities out of commission, at least temporarily. Paradoxically, we demonstrated that SAgs activate and expand preexisting IAV-specific memory T CD8 in mice and humans [19], and not the opposite as conventionally predicted.
IAV has been a workhorse for studies of antiviral T CD8 immunodominance, the marked tendency of T cells to respond to viral peptides in a predictable hierarchy based on the MHC class I allomorphs present in an individual. T CD8 immunodominance hierarchies are controlled by many factors, including the type of APCs participating in Ag processing [20], the expression kinetics of viral proteins and the efficiency of their degradation in the cytosol [21,22], the rough selectivity of TAP for the resulting peptides [23], the affinity of MHC I molecules for transported peptides [23,24], and of course the frequencies of naïve peptide-specific T CD8 in one's repertoire [25] that may indeed be altered by SAgs. We previously demonstrated that the breadth of T CD8 responses can also be influenced by the immunomodulatory functions of CD4 + CD25 + regulatory T (Treg) cells [26], intracellular indoleamine 2,3-dioxygenase (IDO) [27] and the cell surface checkpoint molecule programmed cell death-1 (PD-1) [28], and by the enzymatic actions of terminal deoxynucleotidyl transferase (TdT) [29,30] and the mammalian target of rapamycin (mTOR) [31].
Immunodominance is an intriguing phenomenon from a basic biological standpoint and is also considered an obstacle to successful vaccination [32]. Bona fide T CD8 responses must be both sufficient in vigor and broad in scope to optimally target multiple peptide Ags. An immunodominant status does not necessarily confer a protective property upon any given T cell clone(s) [33] as subdominant T CD8 responses can also contribute to anti-pathogen immunity.
The impact of bacterial SAgs on T CD8 immunodominance is unknown. This represents a gap in our understanding of host responses to vaccination or infection with IAVs and also to other viruses. Furthermore, how recall T CD8 responses to IAV may be manipulated by SAgs is unclear. Finally, whether the time point at which a SAg is encountered may change the immunization outcome is essentially unexplored. We have addressed the above questions by using multiple SAgs or SAg-sufficient/deficient bacteria in several in vivo settings in which T CD8 responses could be monitored against wild-type (WT) and reassortant IAV strains, a live attenuated nasal flu vaccine, and vaccinia virus (VacV). The implications of our findings for sequential infections, superinfections, anti-IAV vaccination and heterosubtypic immunity will be discussed.

Ethics statement
All animal experiments were carried out using age-matched mice under protocols approved by Animal Care and Veterinary Services at Western University (AUPs 2010-239, 2010-241 and 2018-093) and following the Canadian Council on Animal Care guidelines.

Mice
Adult female BALB/c mice were purchased from Charles River Canada (Saint-Constant, QC) and cared for in our institutional barrier facility.
The Western Reserve strain of VacV was propagated in the thymidine kinase-deficient osteosarcoma cell line 143B.

WT and mutant Bacteria
The methicillin-resistant Staphylococcus aureus COL was used in superinfection experiments. A mutant strain of this bacterium with a deletion of seb, the gene encoding staphylococcal enterotoxin B (SEB), was previously described [37]. These strains will be simply referred to as COL and COL Δseb, respectively. For bacterial growth curve analysis, cells were picked from a tryptic soy agar (TSA) plate, grown overnight in Brain Heart Infusion (BHI) broth, and subcultured in half-BHI medium (BHI broth diluted 1:2 in sterile Milli-Q water). OD 600 was adjusted to 0.05 before 200 μL/well of cultures were placed in quadruplicate in a 96-well flat-bottom Microfluor 2 White plate (Thermo Fisher Scientific). Bacterial growth was monitored at 37˚C with shaking using a Biotek Synergy H4 plate reader, and OD 600 readings were recorded at one-hour intervals.

Native and recombinant SAgs
Native Mycoplasma arthritidis mitogen (MAM) was purified as described elsewhere [38]. Briefly, M. arthritidis strain PG6 (ATCC 19611) was grown to senescence in autoclaved modified Edward-Hayflick medium, which was then subjected to (NH 4 ) 2 SO 4 fractionation. This was followed by sequential rounds of gel filtration and cation-exchange chromatography prior to the final purification of MAM by fast protein liquid chromatography.
Recombinant staphylococcal enterotoxins A and B (SEA and SEB) were cloned from S. aureus strains Newman and COL, respectively, expressed in Escherichia coli BL21 (DE3), and then purified by nickel column chromatography as we previously described [39]. We also generated an SEB variant carrying an N ! A point mutation at position 23 (SEB N23A ). N23 is critical for binding to mouse TCR Vβ8.2 and Vβ8.3 [11,40,41], prototypic targets of intact SEB.

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses

SAg administration, immunization and infection protocols
A bolus dose of SEA, SEB or SEB N23A (50 μg in 200 μL PBS) was administered intraperitoneally (i.p.) at indicated time points. In a limited number of experiments, mice were briefly anesthetized using 20% isoflurane before they were injected intranasally (i.n.) with 50 ng SEB in 25 μL sterile PBS. Control mice received an equal volume of PBS. MAM was injected i.p. at 50 ng in 200 μL of PBS containing 0.2% BSA.
In our primary IAV immunization protocols, infectious allantoic fluid containing indicated viruses was diluted in sterile PBS before~600 hemagglutinating units were injected i.p. into each mouse. In several experiments, PR8 was incubated in a 55˚C water bath for 30 minutes [44]. TCID 50 assays confirmed the loss of infectivity by heat-inactivated PR8, which was then injected intramuscularly (i.m.), 50 μL into each thigh muscle, using a 25-gauge needle.
Indicated cohorts of mice were vaccinated i.n. using 20 μL of FluMist under light isoflurane anesthesia.
To study recall T CD8 responses to IAV, mice were first primed with PR8 followed, 30 days later, by a boosting i.p. injection of SEQ12 as we previously reported [29].
To induce active PR8 infection, a 25-μL inoculum of infectious allantoic fluid approximating 0.3 MLD 50 (50% mouse lethal dose) was instilled into the nares of anesthetized mice. Animals were monitored for signs of morbidity, including lethargy, piloerection and weight loss. On day 4 post-PR8 infection [45], the abundance of replicative viral particles was evaluated in lung homogenate specimens and in bronchoalveolar lavage (BAL) fluid and reported as 50% tissue culture infectious dose (TCID 50 ) using Madin-Darby Canine Kidney (MDCK) cells (ATCC CCL-34). We employed a previously published protocol in our TCID 50 assays [46] but utilized chicken red blood cell hemagglutination, as opposed to cytopathic effects, as a readout [47].
For superinfection experiments, COL and COL Δseb colonies were picked from TSA plates, grown overnight in tryptic soy broth (TSB), and subcultured 2% (V/V) in TSB to yield an OD 600 of 3.0-3.5. Bacterial pellets were washed and adjusted to a concentration of 4 × 10 9 colony-forming units (CFUs)/mL in Hank's Balanced Salt Solution. Under isoflurane anesthesia, mice were instilled i.n. with 25 μL of PBS containing 1 × 10 8 CFUs of COL or COL Δseb. This was followed, 3 days later, by i.n. PR8 inoculation. Separate cohorts of mice were sacrificed after 4 and 10 days to assay for infectious viral titres and IAV-specific T CD8 responses, respectively. Animals were monitored for morbidity throughout superinfection experiments.

Detection of IAV-specific T CD8 by peptide:MHC I multimers
At indicated time points after IAV inoculation, mice were sacrificed by cervical dislocation for their spleen, which was manually homogenized and subjected to treatment with ammoniumchloride-potassium (ACK) lysis buffer to eliminate erythrocytes. Cells were washed, filtered through a strainer with 70-μm pores, and incubated on ice for 20 minutes with a 20-μL aliquot of 2.4G2 hybridoma supernatant to block Fc γ receptors (FcγR) II and III. Alexa Fluor 488-conjugated H-2K d tetramers loaded with synthetic peptides corresponding to NP 147-155 or HA 518-526 (NIH Tetramer Core Facility, Atlanta, GA) were then added to the cells at a 1:200 dilution in PBS containing 5% fetal bovine serum (FBS). NP 147-155 and HA 518-526 will be simply referred to as NP 147 and HA 518 . After 30 minutes on ice, an allophycocyanin-conjugated anti-CD8α mAb (clone 53-6.7) was added to the cells for an additional 30 minutes. In several experiments, cells were co-stained with mAbs to surface CD3ε (clone 145-2C11), CD127 (clone A7R34), Fas (clone 15A7), killer cell lectin-like receptor G1 (KLRG1) (clone 2F1), lymphocyte activation gene 3 (LAG-3) (clone C9B7W) or programmed cell death-1 (PD-1) (clone J43) or with a mAb detecting intracellular Ki67 (clone SolA15). All fluorochromelabeled mAbs and isotype controls were purchased from Thermo Fisher Scientific.
In kinetic experiments, tetramer staining was also conducted on serial peripheral blood mononuclear cells (PBMCs) isolated from each animal. Up to 200 μL of peripheral blood was collected from the lateral saphenous vein, immediately mixed with 20 μL of heparin sodium (Sandoz Canada Inc.), diluted 1:2 in PBS, and overlaid on 500 μL of low-endotoxin Ficoll-Paque PLUS (GE Healthcare Life Sciences). Cells were spun at 400 × g for 30 minutes at room temperature, and PBMCs located at the plasma-Ficoll interface were harvested and washed to remove platelets before they were stained.
Where indicated, we employed H-2K d :NP 147 and H-2K d :HA 518 dextramers generously provided by Dr. Stephen Haley (Immudex, Copenhagen, Denmark) to detect NP 147 -and HA 518specific T CD8 , respectively. Up to 3 × 10 6 splenocytes were incubated first with a 2.4G2 supernatant aliquot and then with 10 μL of dextramers in dark for 10 minutes at room temperature. This step was followed by staining with anti-CD8α mAb (clone 53-6.7) as described above.
Cells were washed to remove excess reagents and immediately interrogated using a BD FACS-Canto II flow cytometer. FlowJo software (Tree Star, Ashland, OR) was used for data analysis.

Enumeration of interferon (IFN)-γ-producing virus-specific T CD8 by intracellular cytokine staining (ICS)
Following immunization or infection with indicated IAVs or VacV, mice were euthanized at indicated time points and erythrocyte-depleted splenocytes or peritoneal exudate cells (PECs) were prepared. In the case of i.n. IAV infection, BAL fluid and lungs were also aseptically collected. BAL was conducted using 1 mL of cold PBS. This procedure was repeated thrice to recover the maximal number of cells. Lungs were then perfused with cold PBS to remove contaminating peripheral blood cells and subsequently homogenized to obtain non-parenchymal cells.
Cells were washed and seeded at 0.3-2.0 × 10 6 cells/well of a U-bottom microplate containing RPMI 1640 medium supplemented with 10% heat-inactivated FBS, 2 mM GlutaMAX, 0.1 mM MEM nonessential amino acids, 1 mM sodium pyruvate, 10 mM HEPES, 100 U/mL penicillin and 100 μg/mL streptomycin. Cells were left untreated, exposed to an irrelevant control peptide (NP 118-126 or simply NP 118 ), or stimulated with synthetic peptides corresponding to IAV-or VacV-derived T CD8 epitopes (Table 1). NP 118 is an immunodominant peptide epitope of lymphocytic choriomeningitis virus (LCMV) [48] (Table 1) to which our animals had never been exposed. Each peptide was used at a final concentration of 500 nM. After 2 hours at 37˚C, 10 μg/mL of brefeldin A was added to ensure the retention of IFN-γ in the endoplasmic reticulum (ER) of reactivated T CD8 , and cultures were continued for another 3 hours. Cells were subsequently washed, briefly incubated on ice with 2.4G2 supernatant, and stained for surface CD8α. In several experiments, we additionally used fluorescein isothiocyanate (FITC)-conjugated anti-Vβ mAbs from a Mouse Vβ TCR Screening Panel kit (BD Pharmingen). Next, cells were washed, fixed with 1% paraformaldehyde, permeabilized with 0.1% saponin and stained with an anti-IFN-γ mAb (clone XMG1.2 from Thermo Fisher). The percentage of IFN-γ + cells was determined after live gating on CD8 + events, which was also used to calculate virus-specific T CD8 numbers per organ/sample. Background noise from untreated cells was always negligible and deducted from values obtained from peptide re-stimulation cultures.
immunodominance [26,28,29,49]. Briefly, to prepare control and cognate target cells, erythrocyte-depleted syngeneic splenic cell suspensions were pulsed in separate tubes with 500 nM of NP 118 or indicated IAV-derived peptides. After 45 minutes at 37˚C, cells in each tube were washed, resuspended in PBS and labeled for 15 minutes at 37˚C with a specific CFSE concentration. Typically, target cells displaying NP 118 , HA 518 and NP 147 were labeled with 0.025, 0.25 and 2 μM of CFSE, respectively. The CFSE reaction was stopped using cold FBS, and target cells were washed and mixed in equal numbers for intravenous (i.v.) tail vein injections into naïve and IAV-primed recipients. Where indicated, target cells were injected into animals that had been given SEB (or PBS), but not PR8, as an additional control. Each mouse received a total of 1 × 10 7 cells in 200 μL of PBS. Exactly 1, 2 or 4 hours later, mice were sacrificed for their spleens, which were quickly homogenized and immediately transferred onto ice. Two thousand CFSE low events were acquired on a cytometer, and specific lysis of target cells was determined by the following formula: % specific cytotoxicity = {1 -[(CFSE int/high event number in IAV-primed mouse � CFSE low event number in IAV-primed mouse) � (CFSE int/high event number in naive mouse � CFSE low event number in naive mouse)]} × 100.

IAV-specific IgG2b detection
Three weeks after i.n. infection with PR8, mice were terminally bled via cardiac puncture. Blood specimens were left to clot and then centrifuged at 17,000 × g for 40 minutes at 4˚C. IAV-specific IgG2b was detected in serum samples using a previously established protocol [50] and an Invitrogen eBioscience Ready-SET-Go! Kit. MDCK cells were infected with PR8 at a multiplicity of infection (MOI) of 1, harvested 36 hours later, washed and resuspended in PBS containing 1 mM phenylmethylsulfonyl fluoride protease inhibitor (Thermo Fisher), and subjected to three freeze-thaw cycles. The resulting cell lysate was diluted 1:400 in a coating buffer, transferred to wells of an ELISA plate and incubated overnight at 4˚C. The wells were subsequently blocked with PBS containing 2% FBS before they were washed using PBS plus 0.05% Tween 20. Serially diluted serum samples were then added for 1 hour at room temperature. After removing unbound serum, a mAb against mouse IgG2b was used to capture bound IgG2b. Next, the plate was washed, and a horseradish peroxidase (HRP)-conjugated antimouse IgG polyclonal Ab was added to each well for 3 hours at room temperature. The plate was washed again and received 100 μL/well of a tetramethylbenzidine substrate solution. The enzymatic reaction was stopped using 1 M phosphoric acid, and OD 450 values were determined using a Synergy H4 Hybrid Microplate Reader.

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses

Statistical analyses
Statistical comparisons were made using GraphPad Prism 8 software. Student's t-tests or oneway Analysis of Variance (ANOVA) with Tukey's post-hoc tests were employed as appropriate. Differences with p<0.05 were considered statistically significant.

Exposure to SEB before intraperitoneal PR8 immunization amplifies the expansion of 'select' IAV-specific T CD8 clones that retain their IFN-γ production and cognate cytolytic capacities
To address how bacterial SAgs alter the magnitude and breadth of primary anti-IAV T CD8 responses, we injected cohorts of BALB/c mice i.p. with PBS or with 50 μg of SEB, a prototypic and clinically relevant staphylococcal SAg that is well tolerated in this mouse strain. Three days later, a time point at which SEB-induced T cell proliferation reaches its maximum [51], we immunized mice with PR8, a mouse-adapted IAV strain. We then enumerated cognate T CD8 by MHC I multimers or by ICS for IFN-γ at several time points, including on day 7 post-PR8 priming when primary T CD8 responses to IAV epitopes peak [29,52] ( Fig 1A).
The above model provides multiple advantages. First, unlike in several other laboratory strains, MHC molecules expressed in BALB/c mice have a sufficient binding avidity for SEB [53]. Second, i.p. inoculation of mice with PR8 does not lead to viral propagation, which is due, at least in part, to the modest ability of this virus to infect macrophages and to activate or recruit peritoneal leukocytes [54]. However, it can prime T cells adequately [29,52]. This simulates many seasonal flu vaccination protocols in which inactivated viruses are introduced to the immune system via an unnatural route. Third, the peptide epitopes of PR8 are well-defined in BALB/c mice (Table 1) and give rise to T CD8 clones consistently arranged in a reproducible hierarchical order [29,52]. This enables examination of the breadth of IAV-specific T CD8 responses. Fourth, since i.p. inoculation of PR8 does not cause active viral infection, cognate T CD8 responses could be monitored in the absence of severe complications otherwise ensuing in moribund animals.
We found a single i.p. injection of SEB prior to IAV immunization to raise the frequency of splenic NP 147 -specific T CD8 by dextramer staining (Fig 1B). This could be translated, in terms of absolute numbers, to an approximately 3-fold increase over control values ( Fig 1C). The observed rise in NP 147 -specific T CD8 was also clearly evident in the peripheral blood of SAgpretreated animals ( Fig 1D).
MHC I multimers identify Ag-specific T CD8 based solely on their TCR specificity while providing little information about their functional competence. Since SAgs may anergize T cells [17,18], it was important to study T cell responses using functional readouts. Furthermore, although NP 147 is immunodominant in our system [55], it is not the only epitope recognized in PR8-primed BALB/c mice. Therefore, in the next series of experiments, we tracked functional T CD8 recognizing all known peptide epitopes of PR8 by ICS for IFN-γ. Upon incubation with an irrelevant peptide (NP 118 ), splenocytes from PR8-inoculated mice did not synthesize any IFN-γ above the background level. In contrast, NP 147 -specific T CD8 exhibited a high IFN-γ production capacity (Fig 1E and 1F). Interestingly, SEB administration before PR8 inoculation also 'selectively' increased the frequencies and absolute numbers of T CD8 recognizing HA 518 and HA 462 without affecting the clonal size of other subdominant T CD8 (Fig 1E and 1F). In fact, the observed increase in the case of the HA 518 -specific clone resulted in its 'rank promotion', thus placing it in the second position in the immunodominance hierarchy and only behind NP 147 -specific T CD8 . We also confirmed the enlarged clonal size of the HA 518 -specific population by dextramer staining (S1A and S1B Fig).

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses
The ultimate function of T CD8 is to sense the presence of and destroy target cells displaying cognate peptide:MHC I complexes. We employed a CFSE-based in vivo killing assay to test whether SAg-expanded IAV-specific T CD8 could fulfill their CTL function. Syngeneic naïve splenocytes were separately pulsed with peptides corresponding to NP 147 , HA 518 or NP 118 (irrelevant peptide). Control and cognate target cells thus prepared were washed, mixed in equal ratios and injected i.v. into naïve or PR8-vaccinated mice that had received PBS or SEB earlier according to the schematic timeline depicted in Fig 1A. Four hours later, target cells were tracked based on their differential CFSE intensity. While all three target populations remained intact in naïve mice, NP 147 -coated targets were almost completely absent from the spleens of PR8-primed mice regardless of their prior treatment with PBS or SEB (Fig 1G and  1H). Importantly, SEB pretreatment dramatically augmented HA 518 -specific cytotoxicity ( Fig  1G and 1H). As expected, NP 118 -coated target cells were not removed from the spleens. We also confirmed the observed cytotoxicity to be IAV-specific and not due, rather fortuitously, to in vivo exposure to a SAg. Following an identical timeline, mice were given SEB or PBS i.p. and left undisturbed for another 10 days in their cages before they were injected i.v. with a mixture of NP 118 -, NP 147 -and HA 518 -pulsed target cells. As anticipated, in the absence of PR8 inoculation, each of the three target cell populations was still present in the spleens of SEBand PBS-treated animals, almost in equal numbers. To be precise, four hours after the injection of target cells, the mean remaining event number (per corresponding gate) ± SEM in SEBvs. PBS-treated mice (n = 3 per group) was 1,825 ± 35 vs. 1,758 ± 37 (NP 118 ), 1,524 ± 77 vs. 1,383 ± 28 (NP 147 ), and 1,434 ± 70 vs. 1,419 ± 27 (HA 518 ), respectively.
The above results demonstrate that i.p. exposure to SEB expands, rather than deletes, several PR8-specific T CD8 clones that exhibit an impeccable IFN-γ production capacity and cytolytic function at the peak of their primary response.

Systemic SEB administration before intramuscular immunization with heat-inactivated PR8 or intranasal instillation of FluMist enlarges the clonal size of NP 147 -specific/cross-reactive T CD8
Although i.p. injection of mice with PR8 is commonly used in murine immunization studies, we sought to extend our work to two additional models of seasonal flu vaccination. First, we administered SEB (or PBS) i.p. to mice that subsequently received an i.m. injection of heatinactivated PR8 (S2A Fig). Seven days later, a tiny but clearly detectable population of NP 147specific T CD8 was found in the spleen of SEB-treated mice (S2B and S2C Fig). In a second model, i.p. administration of SEB (or PBS) preceded i.n. instillation of FluMist. Interestingly, once administered to BALB/c mice, the A/California/7/2009 component of this human vaccine yields a PR8-cross-reactive NP 147 -specific response [56]. Ten days after FluMist instillation, NP 147 tetramer + cells were more frequent among T CD8 in the lungs of SEB-pretreated mice compared with PBS-injected controls [2.2 ± 0.41% (n = 5) vs. 0.63 ± 0.08% (n = 5), p = 0.011]. Similarly, more IFN-γ-producing NP 147 -specific T CD8 were present in the spleens of SEB-pretreated animals [66,980 ± 10,921 (n = 5) vs. 26,070 ± 4,325 (n = 5), p = 0.015]. Therefore, pretreatment with SEB amplifies flu-specific T CD8 responses in multiple vaccination models.

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses

SAg-induced clonal expansion of PR8-specific T CD8 is dictated by the TCR Vβ families they utilize as opposed to their hierarchical status
The observation that SEB expanded only certain PR8-specific clones (Fig 1E and 1F) suggested to us that this was not a global effect caused by non-specific bystander activation of T cells. We hypothesized that SEB's reactivity with select TCR Vβ families was instead required. To test this hypothesis, we took several approaches.
First, we determined the TCR Vβ usage of the three top-ranked clones in the hierarchy (NP 147 , PB2 289 and HA 518 ). These clones were chosen also because they were impacted differently by SEB (Fig 1E and 1F). Examination of 15 different TCR Vβ segments revealed that NP 147 -and HA 518 -specific T CD8 , but not PB2 289 -specific cells, preferentially utilized TCR Vβ families that are known to be 'SEB-responsive'. These Vβs are depicted in Fig 2A using various patterns against a red background. The most predominantly utilized Vβs in NP 147 -and HA 518 -specific clones were Vβ8.1/2 and Vβ8.3, respectively. This was apparent in both PBSand SEB-pretreated animals, but more so in the latter cohort (Fig 2A).
To shed further mechanistic light on this phenomenon, we employed SEB N23A in parallel with SEB in our in vivo experiments. SEB N23A is a mutant SAg incapable of binding to mouse TCR Vβ8.2 [11,41]. N23 (Asn 23 ) should also mediate the reactivity of SEB with mouse Vβ8.3. This notion is supported by a previous report that several SEB variants carrying point mutations at this position (SEB N23S , SEB N23I , SEB N23Y and SEB N23K ) induce only minimal interleukin (IL)-2 production by a Vβ8.3 + hybridoma [40]. To understand how SEB targets and activates mouse T cells expressing Vβ8.3, we superimposed the β-chain from the crystal structure of mouse Vβ8.3/Vα3.3 [57] onto the structure of the homologous mouse Vβ8.2 chain in complex with SEB [58], and then built MHC class II into the complex from the SEB-HLA-DR1 crystal structure [59]. Based on this model (Fig 2B), SEB forms a ternary complex whereby the TCR is 'wedged' apart from the antigenic peptide displayed by MHC II, and the sidechain of SEB N23 engages three residues within CDR2β (Asn 54 , Leu 55 and Gln 56 ), and likely forms an intermolecular hydrogen bond with the side chain of Vβ8.3 Asn 54 . Although these three residues differ between Vβ8.2 and Vβ8.3, critical contacts between SEB and the TCR β-chain are believed to occur through backbone atoms of the β-chain, rather than sidechains, indicating that Vβ recognition by SEB is based mostly on the conformation of the CDR2 loop [60,61]. Therefore, similar to SEB N23S , SEB N23I , SEB N23Y and SEB N23K [40], our N23 mutant (SEB N23A ) should be unable to bind Vβ8.3 ( Fig 2B). Indeed, SEB, but not SEB N23A , enhanced the responsiveness of NP 147 -and HA 518 -specific T CD8 (Fig 2C), which are biased towards using Vβ8.1/2 and Vβ8.3, respectively (Fig 2A). Moreover, SEB was significantly more potent than SEB N23A in inducing HA 518 -and HA 462 -specific responses, and a similar trend was observed in the case of NP 147 , which approached but did not reach statistical significance.
In the next series of experiments, we tested two additional SAgs, namely MAM and SEA, in our model. MAM is produced by Mycoplasma arthritidis, which is fundamentally different from S. aureus in terms of structural and genomic complexity, tissue tropism, and pathogenicity [62]. This SAg offers two additional advantages. First, it has a relatively high affinity for mouse MHC molecules [63]. Second, it binds to Vβ8.1-3 in BALB/c mice [63], the main Vβ families utilized by NP 147 -and HA 518 -specific cells (Fig 2A). These features provided us with a unique opportunity not only to extend our work beyond testing a single SAg (i.e., SEB) but also to solidify our conclusion that SAgs expand IAV-specific T CD8 in a TCR Vβ-specific fashion. As hypothesized, pre-exposure to MAM selectively enhanced NP 147 -and HA 518 -specific responses (Fig 2D), which mirrored our findings in SEB-pretreated mice (Fig 1F).
SEA, on the other hand, binds to Vβ1, 3, 10, 11 and 17 [64]. Of these, Vβ3 and Vβ17 are responsive to both SEA and SEB. However, they are scarce or absent in BALB/c mice, due to

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses experiments) were injected i.p. with PBS, 50 μg of SEB, or 50 μg of SEB N23A , followed 3 days later by PR8 immunization i.p. After 7 days, T CD8 responses to indicated epitopes were quantified by ICS for IFN-γ. (D) Mice were injected i.p. with PBS or with 50 ng of MAM before they received, 3 days later, an i.p. inoculum of PR8. On day 7 post-PR8 immunization, epitope-specific T CD8 were enumerated. Data pooled from two independent experiments are shown. (E) Three days before PR8 inoculation, mice were injected with PBS or with 50 μg of SEA i.p. Seven days after PR8 immunization, epitope-specific T CD8 were enumerated. Error bars represent SEM. � and �� denote significant differences with p<0.05 and p<0.01, respectively, using unpaired, two-tailed Student's t-tests (D) or one-way ANOVA followed by Tukey's post-hoc tests (C). https://doi.org/10.1371/journal.ppat.1008393.g002

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses the presence of certain endogenous retroviruses in this strain [65,66]. Since the remaining SEA-reactive Vβ families are not expressed by NP 147 -and HA 518 -specific T CD8 (Fig 2A), SEA should not alter the response magnitude of these clones, which is exactly what we observed (Fig 2E).
Taken together, the above results demonstrate that SAg-triggered expansion of IAV-specific T CD8 is mechanistically linked to their TCR Vβ makeup.

SEB invigorates select epitope-specific T CD8 responses to WT and reassortant IAV strains
Our experiments thus far involved only PR8, an H1N1 strain suitable for mouse studies and for propagation in chicken eggs. It was therefore pertinent to explore whether pre-exposure to SEB could similarly enhance T CD8 responses to other IAV strains, including H3N2 viruses that are included in seasonal flu vaccine formulations. To this end, we tested several WT and recombinant viruses. These included WT NT60 and HK strains whose HA and neuraminidase (NA) together comprise an H3N2 coat that is antigenically dissimilar to the H1N1 coat of PR8 ( Fig 3A). Another cohort of mice was vaccinated with X31, a reassortant virus that contains a PR8 core and an HK coat [35]. Finally, we used the reassortant H3N1 strain J-1, which contains seven PR8 genes along with an HK-derived HA [36]. As illustrated in Fig 3B-3E, a sizeable response to NP 147 was evident following immunization with NT60, HK, X31 or J-1, which all expressed a nucleoprotein (NP) containing this immunodominant peptide. Importantly, pretreatment with SEB enhanced this response in all cases (Fig 3B-3E). In contrast, NP 39 -and NP 218 -specific responses were not augmented. This was consistent with the findings from our PR8 vaccination experiments (Fig 1E and 1F). Finally, none of the H1-derived epitopes elicited a measurable response in PBS-or SEB-treated mice (Fig 3B-3E). This was expected since NT60, HK, X31 and J-1 strains all express H3, not H1 (Fig 3A). The above results demonstrate that cognate T CD8 responses to multiple IAV strains can be intensified by SEB.

SAgs can augment or attenuate recall anti-IAV T CD8 responses depending on when they are encountered
Infection with or vaccination against one IAV serotype may induce cross-protection against another, which is commonly referred to as heterosubtypic immunity.
In our primary vaccination model, introducing SEB three days before PR8 led to augmented responses against NP 147 and HA 518 (Fig 1A-1H). Based on our kinetic analyses, NP 147 -and HA 518 -specific clones remained significantly enlarged in the spleen of SEB-pretreated animals after day 7 (Fig 4A and 4B). Therefore, we asked whether pre-exposure to SEB affects the progression of IAV-specific T CD8 to a memory state. We found that at the peak of their primary response, both NP 147 -and HA 518 -specific T CD8 pools contained more CD127 high KLRG1 low cells and less CD127 low KLRG1 high cells in SEB-treated mice (Fig 4C). These populations are considered memory precursors and short-lived terminal effectors, respectively, in antiviral defense [67].
The above findings prompted us to hypothesize that SEB administration before IAV inoculation should increase the precursor frequency of memory T CD8 . We used a prime-boost strategy to test this hypothesis. Accordingly, mice were injected with SEB, primed with PR8 and boosted 30 days later with SEQ12 ( Fig 4D). SEQ12 was initially generated through sequential passages of PR8 in the presence of anti-HA mAbs [34]. Therefore, it cannot be efficiently neutralized by anti-HA Abs produced by mice during the priming phase. This prevents the removal of the virions before they can induce an anamnestic T CD8 response. It is also noteworthy that SEQ12 has accumulated mutations within its HA's globular domain while retaining

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses the structure of this membrane glycoprotein and identical sequences in and around HA 518 , one of our epitopes of interest. Therefore, the response to HA 518 should be preserved as we previously reported [29].
As hypothesized, heterologous prime-boost vaccination with PR8 and SEQ12 resulted in a more rigorous secondary response to both NP 147 and HA 518 compared with the primary response to PR8 (Fig 4E and 4F and Fig 1E and 1F). Furthermore, seven days after SEQ12 inoculation, SEB-pretreated mice had significantly higher proportions and absolute numbers of NP 147 -and HA 518 -specific cells in their splenic T cell pools (Fig 4E and 4F). Together, the above results demonstrate that when a SAg is introduced before a primary response to an IAV strain is mounted, subsequent responses to unrelated IAV serotypes harboring identical core Systemic pre-exposure to SEB augments the primary NP 147 -specific response following immunization with multiple IAV strains. Mice were injected i.p. with PBS or SEB followed, 3 days later, by i.p. immunization with WT or recombinant IAV strains expressing the same NP (A). These strains were NT60 (B), HK (C), X31 (D) and J-1 (E). Seven days after IAV inoculation, the absolute numbers of IFN-γ-producing epitope-specific T CD8 were calculated. Error bars represent SEM. � and �� denote significant differences with p<0.05 and p<0.01, respectively, using unpaired Student's t-tests. https://doi.org/10.1371/journal.ppat.1008393.g003

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses

Fig 4. Pre-exposure to SEB raises the frequency of memory T CD8 precursors and invigorates recall responses of SEB-binding IAV-specific T CD8 . (A-B)
Mice were injected i.p. with PBS or with 50 μg of SEB before they received i.p. immunization with PR8. On indicated days, members of each cohort were sacrificed for their spleens in which the frequencies of NP 147 -specific (A) and HA 518 -specific (B) T CD8 were determined by ICS for IFN-γ. Each symbol represents an individual mouse, and data are pooled from two independent experiments yielding similar results. (C) On day 7 post-PR8 immunization, the frequencies of KLRG1 + CD127 -T CD8 (short-lived effectors) and KLRG1 -CD127 + T CD8 (memory precursors) were determined after surface staining of splenic cells with mAbs to CD3 and CD8α along with MHC I tetramers detecting NP 147 -or HA 518 -specific cells. (D) Mice were injected with PBS or SEB three days before they were primed with PR8 i.p. Thirty days later, boosting immunization with SEQ12 was conducted using the same injection route. On day 7 post-challenge with SEQ12, the frequencies (E) and absolute numbers (F) of IFN-γ-producing epitope-specific T CD8 in the spleens were determined. The results depicted in E-F are pooled from two independent experiments. Error bars (A-C and E-F) represent SEM. � and �� denote significant differences with p<0.05 and p<0.01, respectively, using unpaired Student's t-tests. https://doi.org/10.1371/journal.ppat.1008393.g004

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses protein epitopes can be enhanced, provided that T CD8 specific for such epitopes utilize SAgresponsive TCR Vβ families.
Next, we explored the effect of SEB on recall responses when this SAg is 'seen' before a secondary anti-IAV response gets underway. Mice were primed with PR8, given SEB, and then boosted with SEQ12 ( Fig 5A). Remarkably, seven days after boosting, the frequencies and absolute numbers of IAV-specific IFN-γ-producing T CD8 were drastically reduced, a finding that was manifest across all tested epitopes regardless of their hierarchical position or TCR Vβ usage preferences (Fig 5B and 5C). To rule out the possibility that injection with SEB shortly (3 days) before SEQ12 may unusually weaken the responsiveness of T CD8 to this particular strain irrespective of their naïve versus memory status, we examined the primary response to SEQ12. In this control experiment, naïve mice received SEB or PBS three days before they were inoculated with SEQ12 (S3A-S3C Fig). Similar to primary immunizations with PR8, NT60, HK, X31 and J-1 (Fig 1 and Fig 2), select cognate responses to SEQ12 were augmented. Therefore, the results shown in Fig 5B and 5C reflect a unique feature of anamnestic T CD8 responses launched shortly after exposure to SEB.
To dissect this hyporesponsive phenotype, we examined the expression levels of several proliferation, death and exhaustion markers among detectable T CD8 . Cognate tetramer + T CD8 from SEB-treated animals showed diminished expression of Ki67 but elevated levels of Fas and LAG-3 on a per-cell basis (Fig 5D). In addition, there was a trend towards increased PD-1, especially in HA 518 -specific cells. Therefore, exposure to SEB may impede the proliferative capacity of anti-IAV memory T CD8 and render them prone to death or exhaustion when animals encounter another IAV serotype with conserved internal epitopes.
The above findings indicate that naïve and memory T CD8 are qualitatively different in their susceptibility to a 'double hit' with a SAg and IAV within a short time span.

Exposure to SEB after IAV inoculation raises the frequency of functional antiviral T CD8
In certain scenarios, exposure to SAgs may succeed a recent encounter with a new IAV strain, due for instance to vaccination. To determine how T CD8 react to IAV in this case, we inoculated mice with PR8 followed, three days later, by an SEB injection. Seven days later, PR8-specific T CD8 were enumerated by ICS, and their in vivo cytolytic function was also assessed ( Fig  6A). We found a dramatic increase in both the frequencies and the absolute numbers of IFNγ-producing T CD8 recognizing multiple PR8 epitopes (Fig 6B and 6C). Furthermore, NP 147displaying target cells were much more readily eliminated in SEB-treated mice (Fig 6D and  6E). This was evident at several time points after cognate target cells were injected into PR8-primed animals (Fig 6E). Therefore, functionally competent IAV-specific CTL pools are expanded following an in vivo exposure to SEB.

Local but not systemic exposure to SEB augments pulmonary T CD8 responses to active IAV infection
We next sought to determine whether exposure to SEB affects T CD8 responses to IAV infection as opposed to vaccination. Separate cohorts of mice were left uninfected or injected i.p. with PBS or SEB before they received a sublethal i.n. inoculum of PR8. As anticipated, while uninfected mice showed a slight weight gain in the course of these experiments, infection resulted in a gradual weight loss, which was worst at around day 7 post-PR8 inoculation and which was partially reversed afterwards (Fig 7A). A control cohort receiving only SEB exhibited a minimal and insignificant weight loss. In addition, the extent and kinetics of weight loss and recovery were comparable between infected mice that had received PBS or SEB (Fig 7A). Finally,

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses systemic pre-exposure to SEB did not alter the viral titres within the lungs and BAL of PR8-infected animals (Fig 7B).
On day 10 post-PR8 infection, which corresponds to the peak of the anti-IAV response in the lungs [45], IFN-γ-producing epitope-specific T CD8 numbers were nearly equal in the lungs and within the BAL fluid of PBS-and SEB-treated mice (Fig 7C). Interestingly however, SEB administration almost doubled splenic NP 147 -specific T cell numbers (Fig 7C). In this setting, only a trend towards an increased response to HA 518 was noticeable in SEB-primed animals. These data indicate that i.p. pretreatment with SEB can enhance the systemic, but not the local, immunodominant T CD8 response in the aftermath of a respiratory IAV infection.
Intraperitoneally injected SEB can reportedly reach the lungs and increase vascular permeability and leukocyte accumulation in mice [68]. However, it failed to alter PR8-specific T CD8 numbers in the lungs or in BAL in our model as indicated above, which was curious. We posited that pulmonary IAV-specific T CD8 responses may be influenced more forcibly by local SEB administration. Injection of SEB via the i.n. route has been used in the past to model exposure to this SAg due to bacterial colonization or through inhalation of aerosolized SEB Mice were primed with PR8, injected with either PBS or 50 μg of SEB, and then boosted with SEQ12 through the i.p. route according to the schematically depicted schedule. Seven days after SEQ12 inoculation, animals were sacrificed for their spleens in which the frequencies (B) and the absolute numbers (C) of IFN-γ-producing epitope-specific T CD8 were determined by ICS. (D) NP 147 -and HA 518 -specific T CD8 , which were identified by MHC I tetramers, were co-stained with mAbs against intracellular Ki67 and surface Fas, LAG-3 and PD-1. Error bars represent SEM (n = 4 per group). � , �� and ��� denote statistically significant differences with p<0.05, p<0.01 and p<0.001, respectively, using unpaired Student's t-tests. https://doi.org/10.1371/journal.ppat.1008393.g005

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses maliciously employed as a bioweapon [69,70]. In fact, SEB is regarded as a 'category B priority bioterrorism agent' by the Centers for Disease Control and Prevention [71]. In our system, i.n. injection of SEB did not change the weight loss pattern of PR8-infected mice (Fig 8A and 8B) but now increased NP 147 -specific T CD8 numbers in the lungs consistent with an augmented local response (Fig 8C).
Finally, we examined how SEB may affect anti-PR8 T cell responses in the context of bacterial colonization/superinfection. To do so, we injected B6 mice i.n. with either WT S. aureus COL or an SEB deletion mutant of this bacterium (COL Δseb) that we previously generated [37]. These strains exhibited identical growth rates when cultured in vitro (Fig 9A). We found that the lungs and BAL of mice that were infected i.n. with PR8 following infection with either Exposure to SEB after PR8 immunization augments IAV-specific T CD8 responses. (A) Mice were inoculated with PR8 and then injected with PBS or SEB (50 μg) i.p. Seven days after the SEB challenge (or after PBS injection), animals were euthanized, and PR8-specific T CD8 were enumerated by ICS for IFN-γ in each spleen. Data pooled from two independent experiments yielding similar results are shown in B-C. In separate experiments, CFSE low naïve splenocytes displaying NP 118 (control target cells) and CFSE high naïve splenocytes pulsed with NP 147 (cognate target cells) were mixed in equal numbers and injected i.v. into PR8-inoculated mice that had subsequently received PBS or SEB. Mice were sacrificed after 1, 2 or 4 hours for their spleens in which target cells were identified based on their differential CFSE intensities. (D) Representative histograms from a 1-hour in vivo killing assay are depicted. (E) In vivo cytotoxicity at indicated time points is averaged and shown for 3 mice/group/time point. Error bars represent SEM. � , �� , ��� and ���� denote differences with p<0.05, p<0.01, p<0.001 and p<0.0001, respectively, using unpaired Student's t-tests. https://doi.org/10.1371/journal.ppat.1008393.g006

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses COL or COL Δseb contained comparable infectious viral titres (Fig 9B). Nevertheless, our tetramer staining and ICS assays revealed either a statistically significant decline or a trend towards a decrease in pulmonary and splenic NP 147 -specific T CD8 in COL Δseb-infected mice ( Fig 9C).
All together, the above results indicate that pulmonary PR8-specific T CD8 responses can be impacted by SEB if this SAg is introduced locally as a stand-alone toxin or after it is released in the airway by colonizing or infectious S. aureus bacteria.

Discussion
How bacterial SAgs modify bulk or Ag-specific T cell responses has been a subject of several investigations in the past. Anergy, TCR downregulation and apoptotic death have often been reported as outcomes of exposure to these potent exotoxins, which may contribute to either a transient state of unresponsiveness or longer-term immunosuppression [12,[16][17][18]72,73]. Paradoxically, Ag-specific T cell growth and activation have also been reported in the face of a challenge with SAgs. For instance, Coppola and Blackman reported that splenic CD8 + CD62L lo CD44 hi bulk memory cells from X31-infected mice kill IAV-infected cells more efficiently when they are exposed to SEB in vitro [74]. We recently demonstrated that 'preexisting' antiviral memory T CD8 can be expanded and activated by bacterial SAgs in mice and among human PBMCs in the absence of a recall challenge [19]. How SAgs affect the magnitude and breadth of vaccine-and infection-elicited in vivo recall responses, which mediated

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses heterosubtypic immunity and potentially cross-protection, has been unclear and a subject of our current work.
Another important question is whether SAgs may manipulate or reshape the immunodominance hierarchies of Ag-specific T CD8 , including IAV-specific cells, which in turn determines the narrowness or breadth of primary and recall anti-IAV responses. Although immunodominance has been often investigated in experimental settings using mouse models, increasing evidence suggests that it is also relevant to human anti-IAV responses [75,76,77].

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses Of note, IAVs are highly mutating viruses whose HA and NA constantly change, in a process called antigenic drift, to avoid detection by neutralizing Abs. In contrast, internal proteins,

Fig 9. Respiratory infection with an SEB deletion mutant of S. aureus before PR8 results in weaker NP 147 -specific T CD8 responses. (A) COL and COL Δseb
were cultured in half-BHI medium, and their growth rates at 37˚C were monitored. OD 600 readings were used to generate a growth curve for each strain. Mice were then given PBS, 1 × 10 8 CFUs of COL or 1 × 10 8 CFUs of COL Δseb i.n. three days before they were infected sublethally with PR8. On day 4 post-PR8 infection, mice (n = 3/group) were sacrificed for their lungs and BAL fluid in which infectious PR8 titres were determined by TCID 50 assays. Each symbol represents an individual mouse (B). Separate cohorts of mice were monitored for weight loss until day 10 post-PR8 infection (C), at which point animals were sacrificed for their spleens and lungs in which NP 147 -specific T CD8 were enumerated by MHC I tetramer staining or ICS for IFN-γ (C). Mean ± SEM values are depicted for 11 mice per group pooled from three independent experiments yielding similar results. Statistical comparisons between COL-and COL Δseb-infected mice were carried out using an unpaired Student's t-test, and � denotes a significant difference with p<0.05. https://doi.org/10.1371/journal.ppat.1008393.g009

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses such as NP, are highly conserved. However, this does not necessarily mean that they never mutate. In fact, mutations in HLA-B27-and HLA-B35-restricted NP-derived peptides to facilitate escape from CTL recognition have been reported [78,79]. In this investigation, we found pre-exposure to SEB and MAM to selectively enhance primary responses to NP 147 and HA 518 , which promoted the hierarchical rank of the latter. Importantly, which T CD8 clones could be expanded by the above SAgs was mechanistically linked to the TCR Vβ composition of T CD8 as opposed either to their default hierarchical status or to the propensity of the protein source of the epitope they detect, namely NP and HA. It will be interesting to assess whether SAgexpanded IAV-specific T CD8 will be equally likely, less likely or more likely to undergo mutations, which may potentially affect their significance in the context of influenza following exposure to the toxins or bacterial pathogens that release them.
To investigate how SAgs influence T CD8 responses in heterosubtypic immunity to IAV, we administered SEB at two different but critical time points, which yielded opposite results. Exposure to SEB before priming augmented recall responses to NP 147 and HA 518 . This was consistent with increased frequencies of T CD8 responding to these epitopes at and after their primary response peak and with a differentiation program favoring the generation of KLRG1 -CD127 + memory precursors. On the contrary, SEB administration prior to boosting immunization drastically diminished IAV-specific T CD8 responses, which was accompanied by lowered Ki67 expression and increased levels of Fas and LAG-3. The observed contrast may highlight differences between naïve and memory T CD8 in terms of their quality and threshold of activation by bacterial SAgs and before they are exposed or re-exposed to cognate IAV Ags, respectively. Therefore, whether the immune system sees an entirely new IAV strain or IAVs harboring similar Ags to those previously seen may determine, at least partially, how T CD8 will behave when they encounter a SAg. This has obvious implications for pandemic and seasonal flu with complications arising from sepsis, non-menstrual toxic shock, food poisoning and other SAg-mediated illnesses and conditions.
Huang et al. previously reported that exposure to SEB seven days after a primary infection with X31 does not change the frequency of TCR Vβ8.3 + cells among NP 366-374 -specific T CD8 (an immunodominant population in C57BL/6 mice) upon a PR8 challenge [80]. In contrast with our recall system, Huang et al. examined NP 366-374 -specific T CD8 within the BAL of intranasally primed-boosted mice. MHC II molecules in the C57BL/6 strain have a notoriously low affinity for SEB, which may explain why a larger dose of SEB (100 μg) had to be administered. Therefore, even though NP 366-374 -specific T CD8 preferentially use Vβ8.3, they may not establish strong and/or long enough contacts with SEB and MHC II + APCs.
In our initial PR8 infection model (Fig 7), SEB was injected i.p. first to mimic systemic exposure to this SAg. SEB administered through this injection route can reportedly find its way into the lungs [68]. We found the absolute numbers of anti-IAV T CD8 to remain unaltered in the lungs and within the BAL. In contrast, there was a significant rise in the splenic NP 147specific response. According to a mathematical model, the spleen contributes the lion's share of the overall IAV-specific T CD8 response in the lungs, especially at later time points post-IAV infection [81]. However, it was pertinent to find out whether intranasal co-administration of SEB and PR8 may affect the local PR8-specific response. Indeed, we observed an increased NP 147 -specific response in the lungs of SEB-pretreated mice in this setting (Fig 8), a finding that could be recapitulated in a superinfection model (Fig 9).
The bacterial culprits of SAg-mediated illnesses typically express more than one toxin. How multiple SAgs encoded by a single bacterial species may cross-regulate T cell responses to each other is unknown. Furthermore, we previously demonstrated that SAg-provoked T cell activation is modulated by Toll-like receptor 2 (TLR2) ligands embedded within the cell wall peptidoglycan of S. aureus [39]. Therefore, the effects of SAgs on anti-IAV immunity may also vary

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses depending on the presence, absence or physical proximity of the very bacteria that secrete these toxins. In our superinfection model, infection with an SEB-deficient mutant of S. aureus COL resulted in a weaker T CD8 response to NP 147 (Fig 9). Therefore, SEB production by COL enhanced the subsequent response to this IAV epitope, an effect that could have been potentially more pronounced in the absence of the inhibitory effects of TLR2 agonists. The implications of this finding may be clinically relevant as far as secondary viral infections in the aftermath of toxic shock and sepsis are concerned.
Another important question is whether exposure to bacterial SAgs similarly modulates cognate T CD8 responses to other viruses, especially those used in vaccination modalities. We addressed this question in a VacV immunization model in a pilot study. We chose to work with VacV because: i) it is very different from IAV structurally and genetically and also in terms of tissue tropism; ii) it is used in prophylactic vaccination against smallpox; iii) poxviral vectors have been employed in vaccine formulations to protect against other infectious diseases, including influenza [82,83,84]; iv) several VacV-derived epitopes have been identified in BALB/c mice [85], which enables studies on T CD8 immunodominance.
Intriguingly, SEB administration before immunization with VacV failed to increase T CD8 responses to this virus (S4 Fig). In fact, both local (peritoneal) and systemic (splenic) responses against F2L 26 , the immunodominant peptide epitope of VacV in BALB/c mice [85] were dramatically reduced. By comparison, T CD8 responses to two subdominant determinants (A52R 75 and E3L 140 ), were preserved in SEB-pretreated animals (S4A-S4D Fig). Why the immunodominant T CD8 response to VacV is depressed, rather than elevated, is unknown at this point and a subject of our ongoing investigation. Unlike IAV, VacV readily propagates in the peritoneal cavity to establish an active infection. However, the ability of a viral intruder to propagate locally should not account, at least largely, for the observed differences. Our finding that SEB could enhance the splenic NP 147 -specific response following both i.p. and i.n. IAV inoculations (Fig 1 and Fig 7C) lends support to the above notion. This is because unlike the i.p. injection of IAV, its i.n. administration results in viral propagation and active infection. At this point, we favor the possibility that distinct requirements for T CD4 help, APC types and/or T CD8 priming pathways may determine the susceptibility of antiviral T CD8 to SAgs. In our IAV infection experiments, SEB reduced virus-specific IgG2b titres in the serum (S5 Fig), which can be viewed as an indirect measure of T CD4 help in Ab class-switching. However, despite the apparent SEB-induced anergy in the T CD4 compartment, the splenic (systemic) NP 147 -specific T CD8 frequency was raised (Fig 7C). In contrast, primary anti-VacV T CD8 responses are known to depend on T CD4 help when the virions are injected via the i.p. route [86]. How various SAgs may impact APC functions, for instance after binding to MHC II, and potentially promote one priming pathway versus another in different organs remains an open question.
In summary, we have investigated how bacterial SAgs, including but not limited to SEB, alter both the strength and the breadth of T CD8 responses to multiple IAV strains. In doing so, we employed several in vivo models of primary and recall anti-IAV immunization and active viral infection. We propose that the nature and the TCR makeup of T cell populations participating in IAV-specific responses, the phase at which a SAg (or multiple SAgs) is/are encountered, and the means and modes of IAV Ag processing and presentation are among the most important determinants of protection versus immunopathogenesis in the context of vaccination and superinfections with IAV and Gram-positive bacterial pathogens.
Supporting information S1 Fig. SEB administration before PR8 immunization increases the clonal size of HA 518specific T CD8 . BALB/c mice were injected i.p. with PBS or with 50 μg of SEB three days before

PLOS PATHOGENS
SAgs modify flu-specific T CD8 responses they were immunized i.p. with the PR8 strain of IAV. Seven days later, HA 518 -specfic T CD8 were identified through surface staining with an anti-CD8α mAb and MHC I dextramers. Representative dot plots after live gating on CD3 + events are demonstrated in panel A. In addition, the absolute numbers of splenic HA 518 -specific cells were calculated (B). Data are shown as mean ± SEM for 4 mice per group. �� denotes a statistically significant difference with p<0.01 using an unpaired Student's t-test. (TIF) S2 Fig. Pretreatment with SEB prior to intramuscular immunization with heat-inactivated PR8 invigorate NP 147 -specific T CD8 . (A) Mice were injected i.p. with PBS or with 50 μg SEB followed, three days later, by i.m. vaccination with heat-inactivated PR8. Seven days after vaccination, the frequency (B,C) and the absolute number (C) of NP 147 -specific T CD8 was determined in each spleen by ICS for IFN-γ. Representative dot plots (B) and summary data (C) are depicted. Error bars represent SEM, and �� and ��� denote statistically significant differences with p<0.01 and p<0.001, respectively, using an unpaired Student's t-test. with PBS or with 50 μg SEB three days before they were immunized i.p. with SEQ12. At the peak of the primary response (i.e., on day 7 post-immunization), T CD8 recognizing the indicated epitopes were enumerated by MHC I tetramer staining (B) and by ICS for IFN-γ (C). Error bars represent SEM. � and �� denote statistically significant differences with p<0.05 and p<0.01, respectively, which were determined using an unpaired Student's t-test (TIF) with PBS or with 50 μg SEB three days before they received VacV i.p. On day 6 post-immunization, a time point at which VacV-specific T CD8 responses reach their peak, peritoneal (B,D) and splenic (C,E) responses to indicated peptides were quantified by ICS for IFN-γ. Representative zebra plots in B and C illustrate the frequencies of peritoneal (local) and splenic (systemic) T CD8 responses to F2L 26 , respectively. Summary data from PBS-and SEB-treated mice (n = 5/group pooled from two independent experiments) also depict subdominant responses to A52R 75 and E3L 140 (D,E). Error bars represent SEM, and � and �� denote significant differences with p<0.05 and p<0.01, respectively, using an unpaired Student's t-test. (TIF)

S5 Fig. SEB administration prior to a sublethal PR8 infection results in reduced antiviral
IgG2b levels. BALB/c mice were infected i.n. with 0.3 MLD 50 of PR8 three days after they received an i.p. injection of PBS or SEB (50 μg). Three weeks after PR8 infection, mice were terminally bled, and the presence of PR8-specific IgG2b was evaluated in serially diluted serum samples as described in Materials and Methods. Error bars represent SEM. � , �� and ��� denote statistically significant differences with p<0. 05