Background: Mass cytometry, or CyTOF (Cytometry by Time-of-Flight), permits the simultaneous detection of over 40 phenotypic and functional immune markers in individual cells without the issues of spectral overlap seen in tradi- tional flow cytometry. Methods: In this study, we applied CyTOF to comprehensively characterize the circulating immune cell populations in elderly individuals both before and after administration of an investigational adjuvanted protein vaccine against respiratory syncytial virus (RSV ) in a Phase 1a trial. Antigen-specific T cell responses to RSV by IFNγ ELISPOT had been observed in most but not all recipients in the highest dose cohort in this trial. Here, CyTOF was used to characterize the cellular response profile of ELISPOT responders and non-responders in this vaccine dose cohort. + + Results: Both CD4 and CD8 T cell antigen-specific IFNγ responses were observed. Principal components analysis revealed baseline differences between responders and non-responders, including differences in activated (HLA-DR ) + + CD4 and CD8 T cells, which were higher in non-responders versus responders. Using viSNE to analyze RSV-respon- + + sive CD4 and CD8 T cells, we also found increased expression of HLA-DR, CCR7, CD127 and CD69 in non-respond- ers versus responders. Conclusions: High parameter CyTOF can help profile immune components associated with differential vaccine responsiveness. Keywords: RSV, Vaccine, CyTOF, Immune profile Background inflammation, decreased antigen presenting cell function, Aged adults have decreased immune responses compared a higher threshold of T cell activation, decreased naïve to younger adults and are more prone to acute infections T cell numbers, a loss of T cell receptor diversity, a loss as well as reactivation of latent viruses. Waning adap- of central memory CD8 T cells and reduced CD8 T cell tive immunity can be seen in adults as young as 50 years priming are all mechanisms identified that impact T cell old . Extensive research on immune senescence in the responses in older adults [5–7]. elderly has identified multiple pathways by which aging T cells in combination with neutralizing antibodies mechanisms adversely affect immune responses, par - may have a key role in controlling respiratory viruses such as influenza and respiratory syncytial virus (RSV) ticularly T cell responses [2–4]. Increased background that can cause more acute infections in the elderly versus healthy young adults. In an elderly adult pop *Correspondence: email@example.com ulation, T cell IFNγ responses to influenza could dis- Christine M. D. Lingblom and Sangeeta Kowli contributed equally to this tinguish between those protected by vaccination and work Department of Immunity, Transplant, Infection, Stanford University, those who subsequently developed influenza illness Stanford, USA . Pre-existing influenza-specific CD4 T cells were Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Lingblom et al. J Transl Med (2018) 16:153 Page 2 of 12 associated with decreased illness severity following Methods influenza challenge of healthy volunteers lacking neu- Human samples tralizing antibodies . Though neutralizing antibody Heparinized whole blood was collected from 20 healthy titers to RSV are similar between elderly and young older adults (60 years and older) with informed consent adult populations , the elderly have decreased under an institutional review board-approved, rand- RSV-specific T cell responses compared to young omized phase 1a study of MEDI7510 (NCT02115815). adults [11, 12]. Demographics for this cohort are presented in Table 1. Vaccines have been only partially successful in Clinical endpoints, F-specific antibody, and F-specific reversing declining immune responses in older adults. IFNgamma ELISPOT responses have been previously Those meant for elderly subjects may benefit from reported . Samples were taken pre-vaccination, D8 inclusion of an adjuvant [13–15] or an increased anti- and D29 post-vaccination from those subjects dosed with gen dose [16, 17]. There is currently no approved vac- 80 μg of soluble RSV fusion protein sF adjuvanted with cine for RSV though the incidence of RSV illness in Glucopyranosyl Lipid A in 2% stable emulsion (GLA-SE) older adults is on par with that of influenza illness (Immune Design Corporation, USA). Peripheral blood . It has been proposed that a successful vaccine for mononuclear cells (PBMC) were separated by Histo- the elderly would need to induce both protective neu- paque Ficoll within 6 h of blood draw and cryopreserved tralizing antibodies and virus clearing T cells [19, 20]. in serum-free CTL Cryo ABC (CTL, USA), then trans- An investigational adjuvanted RSV vaccine that ferred to a central lab for LN storage until testing. aimed to induce both neutralizing antibodies and virus-specific T cells was evaluated in dose-finding IFNgamma ELISPOT Phase 1 trials in > 60 year old subjects . This vac- Cryopreserved PBMC from study subjects were batched cine consists of RSV soluble fusion protein formulated by subject, thawed in CTL Wash medium (Cellular without or with the adjuvant Glucopyranosyl Lipid Technology, USA) with benzonase nuclease (Novagen- A in 2% stable emulsion (GLA-SE). Humoral and cel- Millipore USA), washed and resuspended in CTL Test lular responses were measured in > 60 year olds fol- medium (Cellular Technology, USA), and tested by lowing vaccine dosing. At the 80 ug RSV sF + 2.5 μg F-specific IFNgamma ELISPOT as previously described GLA-SE dose tested, 100% of recipients demonstrated . Briefly, viable cells were plated at 300,000 cells/well a > threefold rise in humoral responses and 74% dem- in quadruplicate in human IFNgamma ELISPOT plates onstrated a > threefold rise in cellular responses . A (Mabtech, USA) and stimulated with either medium con- higher dose of 120 μg RSV sF + 5 μg GLA-SE induced taining 0.1% dimethyl sulfoxide (DMSO) (mock) or 2 μg/ similar rates of humoral and slightly higher rates of mL of overlapping peptide pools of RSV F (JPT GmbH, cellular immune responses . The goal of the work Berlin, Germany). 30,000 viable cells/well were stimu- described here was to use a targeted multiparameter lated with Staphylococcus aureus enterotoxin B (SEB) as evaluation of the RSV F-specific T cell response to fur- a positive control. At 20–24 h plates were developed and ther characterize the cellular response to RSV in these counted on the ImmunoSpot Analyzer (Cellular Technol- vaccinated subjects. ogy, USA). Data was expressed as the spot forming cells To more comprehensively characterize T cell (SFC) per million PBMC after background subtraction responses to RSV, we used CyTOF mass cytometry, of mock wells, with a lower limit of detection of 33 SFC/ a highly multiparametric version of flow cytometry million PBMC (Additional file 1). that uses heavy metal ion labels and mass spectrom- etry as the readout in lieu of fluorochromes and light Extracellular and intracellular staining and CyTOF analysis detection. This methodology has the dual benefit of PBMC from study subjects as well as a healthy con- allowing many more specificities to be probed in par- trol sample for each batch of samples were thawed in allel in the same samples, while dramatically reducing warm CTL 10× wash medium (CTL, USA) diluted 1:10 spillover between detector channels, which is a major in RPMI (Gibco-Life Technologies, USA) containing issue in fluorescence flow cytometry . Using stim- ulation with RSV F antigen peptides, or with Phor- bol 12-myristate 13-acetate (PMA) + ionomycin, we Table 1 Demographics of study cohort were able to read out antigen-specific as well as global Demographic parameter Non-responder (5) Responder (14) immune parameters using CyTOF, and to relate these Age (year; average± SD) 70 ± 7 70 ± 7 to vaccine response as measured by enzyme-linked Age (year; range) 62–79 61–82 immune spot (ELISPOT). Female [number, (%)] 4, (80%) 6, (43%) Lingblom et al. J Transl Med (2018) 16:153 Page 3 of 12 l -Glutamine (Gibco-Life Technologies, USA) and Ben- Table 2 Antibody panel for CyTOF ICS, with metal labels, clones and source zonase nuclease (Novagen-Millipore, USA), washed twice then resuspended in CTL test medium (CTL, No. Specificity Metal label Clone Source USA) containing l -Glutamine, and viable cells were 1 Live/dead In115 – In house counted by Vicell (Merck Millipore, USA). Cells were 2 CD49d Pr141 9F10 Fluidgm added to a V-bottom microtiter polystyrene plate at 1 3 CD19 Nd142 HIB19 Fluidgm million viable cells/well, for each sample one well was 4 ICOS Nd143 DX29 Fluidgm kept as unstimulated, one for the RSV F peptide pool 5 CD69 Nd144 FN50 Fluidgm stimulation (overlapping 15-mers custom-produced by 6 CD4 Nd145 RPA-T4 Fluidgm JPT, final concentration 5 μg/mL, ) for 6 h and one 7 CD8 Nd146 SK1 Fluidgm for PMA/Ionomycin (Sigma-Aldrich, USA, final con - 8 CD20 Sm147 2H7 Fluidgm centration 10 ng/mL and 1 μg/mL, respectively) for 4 h 9 CD57 Nd148 HCD57 In house at 37 °C, in a CO incubator. Simultaneously, activation 10 CD54 Sm149 HA58 In house reagent, Brefeldin A (Sigma-Aldrich, USA), and secre- 11 CD134 (OX-40) Nd150 ACT35 Fluidgm tion inhibitor Monensin (Sigma-Aldrich, USA) was 12 CD107a Eu151 H4A3 Fluidgm added to all the wells. PMA, Ionomycin, Brefeldin A 13 TNFα Sm152 Mab11 Fluidgm and Monensin was diluted in CyPBS (10× PBS without 14 CD45RA Eu153 HI100 Fluidgm heavy metal contaminants diluted 1:10 in MilliQ water, 15 CD3 Sm154 UCHT1 Fluidgm ROCKLAND, USA). Final DMSO and ethanol concen- 16 CD28 Gd155 L283 In house tration from all sources (peptides, brefeldin A, mon- 17 CD38 Gd156 HB-7 In house ensin) did not exceed 0.5%. At the end of stimulation, 18 HLA-DR Gd157 G46-6 In house 0.8 μL 5 M EDTA was added to the wells, to a final con - 19 CD33 Gd158 WM53 Fluidgm centration of 2 mM and incubated for 15 min at room 20 CD11c Td159 Bu15 Fluidgm temperature. The cells were washed three times with 21 CD14 Gd160 M5E2 Fluidgm CyFACS (CyPBS with 2 mM EDTA and 0.05% sodium 22 IFNγ Dy161 4S.B4 In house azide) followed by extracellular staining for 45 min on 23 CD80 Dy162 2D10.4 Fluidgm ice with 70 μL of the antibody cocktail (Table 2). All 24 IL-4 Dy163 MP4-25D2 Fluidgm antibodies were either from purified unconjugated, 25 IL-17 Dy164 N49-653 Fluidgm carrier-protein-free stocks from eBiosciences, Biole- gend, or R&D Systems that we conjugated with metal 26 CD127 Ho165 A019D5 Fluidgm isotopes ourselves or they were conjugated with metal 27 IL-2 Er166 MQ1-17H12 Fluidgm isotopes from Fluidigm. The cells were washed three 28 CD27 Er167 L128 Fluidgm times with CyFACS buffer and then resuspended in 29 CD40L Er168 24–31 Fluidgm 100 μL CyPBS of 1:3000 diluted 5 mg/mL Live-Dead 30 CCR7 Tm169 15053 In house (1,4,7,10-tetraazac yclododecane-1,4,7,10-tetraacetic 31 PD1 Er170 EH12.1 In house acid (DOTA)-maleimide containing natural-abundance 32 Granzyme B Yb171 GB11 Fluidgm indium 115, Macrocyclics, USA) and incubated 30 min 33 NKG2C Yb172 134591 In house on ice. The cells were washed three times with CyPBS 34 CD25 Yb173 M-A251 In house and then resuspended in 100 μL 2% para-formaldehyde 35 CD16 Yb174 3G8 In house (PFA) in CyPBS and placed at 4 °C overnight. The next 36 Perforin Lu175 B-D48 Fluidgm day the cells were washed three times with eBiosci- 37 CD56 Yb176 NCAM16.2 Fluidgm ence permeabilization buffer (1× in MilliQ water) fol- 38 DNA1 Ir191 – Fluidgm lowed by intracellular staining for 45 min on ice with 39 DNA2 Ir193 – Fluidgm 70 μL of the antibody cocktail (Table 2) before wash- 40 CD11b Bi209 ICRF44 Fluidgm ing three times with CyPBS. The cells were resuspended in 100 μL iridium-containing DNA intercalator (1:2000 dilution in 2% PFA in CyPBS; Fluidigm) and incubated v0.2 MCR  (Additional file 2). Data analysis was at room temperature for 20 min. The cells were washed performed using FlowJo v10 by gating on intact cells three times in CyPBS and three times in MilliQ water. based on the iridium isotopes from the DNA intercala- The cells were diluted in a total volume of 700 μL in tor, then on singlets by DNA intercalator versus event MilliQ water before injection into the CyTOF Helios length, then on live cells which is the Indium-Live- (Fluidigm). The data were normalized using Normalizer Dead negative population, followed by cell subset gat- ing (Additional file 2 , Fig. 2a). Lingblom et al. J Transl Med (2018) 16:153 Page 4 of 12 Statistical analyses with the lowest events. After importing the concatenated Multivariate analyses files into cytobank, viSNE was run using default cyto - Multivariate analyses of pattern recognition “orthogo- bank parameters (iterations = 1000, perplexity = 30 and nal projections to latent structures by means of partial theta = 0.5). In each figure, all samples were derived from least squares discriminant analysis” (OPLS-DA) were the same viSNE run. viSNE maps show median marker performed using the SIMCA-P (version 14.1) statistical expression for each population. Scales on the maps are package (MKS Data Analytics Solutions, Malmö, Swe- individually generated for each marker with the intensity den). OPLS-DA is a development of principal component levels from low (blue) to high (red) expression. analysis (PCA), in which Y variables are introduced and their relationship to X variables examined. In our case, Univariate analyses multivariate models were created where study patients Unpaired t test was used to determine statistical signifi - were set as Y variables (Group Y1 for non-respond- cance between the study subjects on day 0 and day 8 as ers and group Y2 for responders) and 29 components well as on day 0 and day 29. GraphPad Prism 7.0 was (cytokine responses and phenotype markers) were set as used to plot graphs (GraphPad, San Diego, CA, USA). X variables. In the figures the 13 variables with biggest impact on the models are shown. The two-component Results models (PC1 and PC2) is defined by a value for explana - T cell responses by IFNγ ELISPOT tory power or goodness of fit, R2, which estimates the As previously reported, pre- and post-vaccination PBMC amount of variance in Y that is explained by the X-varia- obtained from the clinical trial participants were tested bles. A high value indicates that the selected X-variables for T cell responses by a qualified F-specific IFNγ ELIS - have generated a model that can explain differences that POT assay, with a peak response at Day 8 post vaccina- exist between the studied groups. A model is also given tion . A minimum threefold change in F-specific a value for stability, Q2, which describes the validity of responses at Day 8 versus pre-vaccination was used to the model. This is determined with cross validation, a designate responders by this assay. Among the 20 sub- procedure where one study subject is removed and the jects in the treatment cohort, one subject was dropped capacity of the remaining subjects to predict the sepa- as the PBMC viability was low and the sample failed the ration between the groups is assessed. This procedure is acceptance quality criteria for the F-specific IFNγ ELIS - repeated for all the subjects; a high value indicates that POT assay. Of the 19 subjects with reportable data for the model is stable no matter which subject is excluded. both the prevaccination and Day 8 timepoints, 14 sub- A number between 0 and 1 is given or 0–100%. 0 being jects demonstrated a ≥ threefold rise in responses, with worst and 1 being best. pq1 is a value that explains the responses ranging from 4.3- to 32.2-fold over baseline impact that the X variables has on the model. The pro - (Fig. 1). 5 subjects with a < threefold rise in responses gram is set to mean centering and unit variance scaling were designated as non-responders. These ELISPOT to give all variables an equal chance of providing model responses were used to categorize the vaccine subjects leverage independently of data scale and distribution. for subsequent multiparameter intracellular cytokine staining by CyTOF analysis. Cytomegalovirus (CMV) sta- tus is unknown for the study subjects; CMV may affect T viSNE (visual high‑dimensional single‑cell data analysis cell response rates to other antigens. based on the t‑Distributed Stochastic Neighbor Embedding (t‑SNE) algorithm) CyTOF analysis and detection of RSV-specific T cell viSNE is a dimensionality reduction algorithm that per- responses mits visualization of multi-dimensional data as a two A representative manual gating schema used to identify dimensional scatter plot. We performed viSNE analysis major peripheral blood subsets including B cells, C D4 in cytobank (Cytobank, Santa Clara, CA). Boolean ‘OR’ and CD8 T cells, NK cells and monocytes is shown in + + gates for CD107a, IFNγ, TNFα and IL-4 from responders Fig. 2a. While the percentage of C D3 CD4 T cells was (n = 14) and non-responders (n = 5) were concatenated slightly higher in the non-responders compared to the + + in FlowJo v10.1 for both CD4 and CD8 T cells after responders (Additional file 3A), no significant differences + + + + RSV (F) peptide stimulation at day 0 and day 8. For com- were found in either the CD3 CD4 or CD3 CD8 T parative analysis, samples were down sampled and viSNE cell percentages between responders and non-respond- maps were generated from a mixture of equal-sized sam- ers, pre-vaccine or on day 8 or 29 post-vaccine (Addi- + + ples (CD4 T cells = 30,205 events per sample; CD8 T tional file 3A, B). Figure 2b is a representative dot plot + + + + cells = 10,437 events per sample). The event count for showing the induction of CD4 IFNγ and CD4 TNFα each T cell population was determined by the sample after stimulation with RSV s(F) pp on D8. Lingblom et al. J Transl Med (2018) 16:153 Page 5 of 12 Fig. 1 IFNγ ELISPOT responses, Day 8 vs Day 1. A threefold rise was declared as a responder + + + Th1 versus Th2 responses dual IFNγ TNFα CD4 T cells is markedly upregu- To determine the effects of RSV s(F) pp on Th1 and Th2 lated (Fig. 4a, top panel). In subject o, the proportion + + responses between responders and non-responders, of IL-17 CD4 T cells is dominant pre- and post-vac- levels of IFNγ and IL-4 were independently assessed cination. We also find that the expression of TNFα + + by manual gating in FlowJo. RSV s(F) pp stimulation and IFNγ CD4 T cells reduces slightly post-vaccina- + + resulted in both CD4 and CD8 IFNγ responses, and tion on day 8 compared to pre-vaccination (day 0) but CD4 IL-4 responses (Fig. 3a–c). An upward trend was increases post-vaccination on day 29 to levels compa- + + observed in the responders for both CD4 and C D8 rable to pre-vaccination (day 0) (Fig. 4a, middle panel). + + + IFNγ and CD4 IL-4 at D8 and D29 compared to D0. In subject r, pre-vaccination (day 0) the proportion of + + Furthermore, we observed significantly higher baseline IL-4 CD4 T cells is most dominant. However, post- levels of CD8 IFNγ in the non-responders compared vaccination on day 8 and day 29 we observe a switch + + + + + to the responders at D0 (before vaccination) (Fig. 3b). from IL-4 CD4 T cells to TNFα and IFNγ CD4 T A gradual increase in the frequency of IL-4 producing cells (Fig. 4a, bottom panel). In CD8 T cells, post-vac- CD8 T cells post-stimulation with RSV s(F) pp was also cination in subject l, the cytokine production switches + + seen in the responders (Fig. 3d). from IL-17 (day 0) primarily to IFNγ at day 8 and day 29 (Fig. 4b, top panel). In subject o, IL-17 produc- tion is dominant both pre-(day 0) and post-vaccination Longitudinal patterns of cytokine responses (day 8 and day 29) (Fig. 4b, middle panel). In subject + + + + To evaluate the multifunctionality of C D4 and C D8 r, post-vaccination (day 29), induced TNFα CD8 T T cells in response to RSV, we assessed the expres- cells as well as dual- and tri-cytokine positive CD8 + + + + + sion of TNFα , IFNγ , IL-2 , IL-4 , IL-17 as well as T cells (Fig. 4b, bottom panel). These results suggest dual- and tri-cytokine positive combinations pre-(day that first, there is substantial heterogeneity in cytokine 0) and post-vaccination (day 8 and day 29). Shown here production among the subjects. Second, pre- and post- are the top three responders. The average of day 8 and vaccination, single cytokine production is dominant + + + day 29 boolean ‘OR’ gates for IFNγ , TNFα , IL-2 , compared to the contribution from dual- or tri-posi- + + + + IL-4 , IL-17 expression was used to determine the tive cytokines both in C D4 and CD8 T cells. In gen- top three responders. In subject l, IFNγ expression is eral, the proportion of the single cytokine population dominant at baseline. Post-vaccination on day 8 and prevalent for any given subject pre-vaccination (day 0) day 29, TNFα expression as well as the proportion of Lingblom et al. J Transl Med (2018) 16:153 Page 6 of 12 + + Fig. 2 CyTOF gating hierarchy of peripheral blood mononuclear cells (PBMCs). Major immune cell subsets including monocytes, CD4 and CD8 T cells, B cells and natural killer cells were identified with a manual gating strategy post-normalization using FlowJo v10.1 software (a). Representative + + + dot plots of CD4 T cell cytokine ( TNFα and IFNγ ) responses at day 0 and day 8 (b) does not dramatically change post-vaccination (day 8 We also found that non-responders had higher levels of + + + + and day 29). CD4 CD69, CD8 IFNγ and non-B-cells at baseline − + and responders had higher levels of C D3 CD19 and B Principal components analysis of RSV-specific T cell cells (Fig. 5b). The generated model had a stability of 44% responses (Q2Y = 0.44) and explanatory power of 66% (a goodness The multivariate method of OPLS-DA was used to see of fit R2Y = 0.66). if it was possible to predict the outcome of the vaccine based on the immune profile of the patients at baseline. viSNE analysis of RSV-specific T cell responses We divided the patients into two groups, responders (R) To determine differences in antigen-specific CD4 and non-responders (NR), based on the results from the and CD8 T cells between responders and non- IFNγ ELISPOT (Fig. 1). The cytokine responses as well responders, we performed viSNE analysis. For each as surface marker levels after RSV stimulation at base- participant, boolean ‘OR’ gates on RSV specific + + + + + + line (day 0) were set as study variables (X) and the two CD107a IFNγ TNFα IL4 CD4 and CD8 T cells pre- study groups were set as outcome variables (Y). The and post-vaccination (day 0 and day 8, respectively) were two study groups formed two distinct clusters, a tight generated in flowjo v10.1. The individual fcs files for the one composed of responders and a scattered one com- boolean gates were then concatenated into single stand- posed of non-responders (Fig. 5a). When we constructed ard fcs files for responders and non-responders, resulting a loading plot to see which variables that contributed in a total of 4 concatenated files for each T cell population to the separation, interestingly, the non-responders (Additional file 4). Using Cytobank software, the viSNE + + − were positively associated with CD4 HLA-DR CD38 algorithm analysed ungated cell populations for equal + + − and CD8 HLA-DR CD38 and the responders with number of events per time point for each T cell popula- + − − + − − CD4 HLA-DR CD38 and CD8 HLA-DR CD38 . tion (as described above in MM). viSNE plots are shown Lingblom et al. J Transl Med (2018) 16:153 Page 7 of 12 + + + + Fig. 3 IFNγ expression in CD4 and CD8 T cells. Responders develop CD4 T cell IFNγ response over time after stimulation with RSV (F) peptides. + + IFNγ responses in CD8 T cells shows significant differences (p < 0.05) between responders (n = 14) and non-responders (n = 5) at day 0 after + + stimulation with RSV (F) peptides (a, b). Responders develop CD4 T cell and CD8 T cell IL-4 responses over time (c, d) as two-dimensional scatter plots with the x- and y-axes was different for all the markers. Again, the findings from identified by tSNE1 and tSNE2. Each dot on the plot rep - PCA corroborate with some of the viSNE results, + + + + + + resents a single cell positioned according to similarity in with CD4 HLA-DR, CD4 CD69, CD8 HLA-DR , + + the high-dimensional space. For comparative purposes, CD4 CCR7 being positively co-related with the non- viSNE on pre- and post-vaccinated days (Day 0 and Day responders (Fig. 5b). 8) between responders and non-responders for each T cell population was performed in the same run. For both Discussion + + CD4 (Fig. 6a) and CD8 (Fig. 6b) T cell populations, In this study, we successfully derived baseline cellular viSNE analysis showed an increased expression of HLA- features from our CyTOF analyses that were associated + + + + DR, CD127, CCR7 and CD69 cells in responders with ELISPOT response to an experimental RSV vac- and non-responders post-vaccination (day 8) compared cine. Notable among these features were high baseline + + to baseline (day 0). Furthermore, the expression of CD4 levels of IFNγ-producing RSV-specific CD8 T cells, + + + + + and CD8 HLA-DR, CD127, CCR7 and CD69 which were associated with non-responder status. In cells was higher in the non-responders compared to the the setting of influenza vaccination, a high baseline titer responders both pre- and post-vaccination (day 0 and day of hemagglutinin-inhibiting (HI) antibodies has been + + 8, respectively). Interestingly, for both C D4 and C D8 T linked to lower fold-change in HI antibodies post-vac- cells, viSNE identified the same population of cells. The cination . A similar relationship has been found for + − pattern of expression was similar but the intensity levels CD4 IFNγ producing T cells specific to influenza . Lingblom et al. J Transl Med (2018) 16:153 Page 8 of 12 + + + + Fig. 4 Multifunctional analysis of CD4 and CD8 T cell cytokine responses to RSV. Day 8 and day 29 boolean ‘OR’ gates for CD4 and CD8 T + + + + + cell IFNγ , TNFα , IL-2 , IL-4 and IL-17 expression in response to RSV (F) peptide stimulation, were generated in Flowjo 10.1. The average of the boolean ‘OR’ gates was used to determine the top three responders. Multifunctional analysis revealed heterogeneity in cytokine production among + + the subjects as well as dominance of single cytokine production both pre- and post-vaccination compared to dual- or tri-CD4 (a) and CD8 (b) positive cytokines Our findings with this experimental RSV vaccine may homogenous and “healthy” phenotype before vaccination be related, in that it may be more difficult to boost a pre- whereas non-responders do not. The different profiles existing immune response with vaccination. As Falloon explain why we see 9 variables associated with non- et al. concluded, a correlate of protection from RSV is responders and only 4 variables associated with respond- yet unknown, and while both antibodies and T cells were ers in the loading plot. Consequently, if a patient has induced by vaccination these did not provide protection increased levels of any of these 9 variables the outcome . of the vaccine against RSV may be poor. A number of fea- We also saw an association of activated (HLA-DR ) tures from the PCA-model displayed excellent correlation + + + + CD4 and CD8 T cells with non-responder status. with the viSNE analysis. High levels of CD4 HLA-DR , + + + + + + While the short term in vivo exposure to antigen should CD4 CD69, CD8 HLA-DR, CD4 CCR7 at baseline be too short to induce HLA-DR expression, these cells were associated with non-responders using both statisti- might indicate the level of chronic inflammation in these cal methods. These variables were among the nine vari - elderly subjects. Chronic inflammation in the elderly is ables with the highest discriminatory power in the PCA considered a cause of immunologic aging . Higher model. levels of inflammatory response transcripts have been Surprisingly, our univariate analysis revealed that + + linked with hypo responsiveness to hepatitis B vaccina-only CD8 IFNγ were statistically significant between tion . responders and non-responders at baseline, although, a The PCA-model revealed two study groups that formed trend was observed at day 8. We could also see a minor two distinct clusters, a tight one composed of respond- increase for responders when comparing day 0 and day + + + + + + ers and a scattered one composed of non-respond- 29 for C D4 IFNγ, CD4 IL4 and CD8 IL4 . Although ers. The results indicates that responders have a more Lingblom et al. J Transl Med (2018) 16:153 Page 9 of 12 Fig. 5 PCA analysis on gated subsets. a Multivariate analysis of cytokine responses and phenotype marker levels in the study groups at day 0, before vaccination. “Orthogonal partial least square-discriminant analysis” (OPLS-DA) was done to see if the cytokine responses and markers (X-variables) could segregate the two study groups (Y-variables, e.g. non-responders (n = 5) and responders (n = 14) before vaccine against RSV was given. The generated model had a stability of 44% (Q2Y = 0.44) and explanatory power of 66% (a goodness of fit R2Y = 0.66). b Column graph of the OPLS-DA was done to see which variables had the largest impact on the separation of the two groups (responders and non-responders at day 0). Variables closest to the subject group are positively associated this did not reach statistically significance, this was not dominated before and after vaccination. The time point seen for non-responders. post vaccination were not optimal for studying innate Among subjects with the highest responses (shown immune cells, therefor we directed our attention to adap- in the pie charts), a single cytokine seemed to dominate tive immune cells. while the proportions were not greatly changed by vac- Possibly, the already activated state in non-respond- cination, with the exception of IL-4 in the majority of pies ers makes it harder for their immune system to react and IL-17 as well as IFNγ in some pies. Interestingly, we again, as we previously discussed, and would ben- could see a wide heterogeneity among the donors; dif- efit from a higher dose. If one could produce a more ferent subjects have different cytokines that dominate in effective vaccine based on the immune profile of the their immune profile, and in general, the same cytokine subject perhaps immunosenescence would no longer Lingblom et al. J Transl Med (2018) 16:153 Page 10 of 12 Fig. 6 viSNE on antigen-specific cells finds differential markers. Boolean ‘OR’ gates for CD107a, IFNγ, TNFα and IL-4 from responders (n = 14) and non-responders (n = 5) were concatenated in Flowjo v10.1 for both CD4 and CD8 T cells after RSV (F) peptide stimulation at day 0 and day 8, and visualized in t-SNE space with viSNE software in Cytobank. viSNE analysis on antigen-specific T cells showed high expression of HLA-DR, CCR7, CD69 and CD127 in non-responders compared to responders at day 0 and day 8 in CD4 T cells (a). The same trend is seen in CD8 T cells (b) using this approach. Cells are colored by median intensity levels from high (red) to low (blue) be a determining factor for vaccine response. In addi- methods are getting more available in vaccine research, tion, the huge economic cost for the society associ- we will be able to get a clearer picture and enhanced ated with elderly not responding to vaccines, resulting understanding of both immunosenescence and vaccine in increased medical care, increased use of medicines response. and lost time at work would benefit from vaccines that are designed after the immune profile of the subject at Conclusion baseline. The elderly population  is increasing and Our results show the impact of individual immune pro- finding vaccines and other pharmaceuticals suited for files on successful RSV-vaccine immune response. As our that expanding group is becoming substantially more knowledge of the immune system increases, we believe important. this is a factor that needs to be accounted for when In this study, we demonstrate the advantage of com- designing vaccines in the future. Taken together, our bining vaccine studies with mass cytometry. It enables findings demonstrate the potential of CyTOF as a pow - an understanding of the complex immune response erful technology that permits comprehensive profiling that is difficult to obtain with other methods. Here, it of immune components, thereby enabling prediction of specifically revealed how elderly responded to vaccines responses to vaccines. based on their baseline cellular features. With this high- Additional files dimensional technique, co-expression of cytokines can be studied in several cell types and subgroups in a sin- Additional file 1. Mock subtracted IFNγ responses at Day 1, at Day 8 gle sample. As such, it helps us to paint a wider picture undepleted, CD4 depleted and CD8 depleted. of the immune system. A challenging task is to process Additional file 2. CyTOF workflow. the large amounts of generated data and interpret it + + + + Additional file 3. Percentage of RSV specific CD3 CD4 and CD3 CD8 correctly. Therefore, it is essential to use multivariate responses. analysis, e.g., PCA and viSNE, as we do in this study. As Additional file 4. Workflow to perform viSNE analysis on antigen-specific mass cytometry as well as more advanced multivariate cells. Lingblom et al. J Transl Med (2018) 16:153 Page 11 of 12 Abbreviations 5. Britanova OV, Putintseva EV, Shugay M, Merzlyak EM, Turchaninova MA, CyTOF: Cytometry by Time-of-Flight; RSV: respiratory syncytial virus; ELISPOT: Staroverov DB, et al. Age-related decrease in TCR repertoire diversity The Enzyme-Linked ImmunoSpot; GLA-SE: Glucopyranosyl Lipid A in 2% measured with deep and normalized sequence profiling. J Immunol. stable emulsion; PBMC: peripheral blood mononuclear cells; SFC: spot form- 2014;192:2689–98. https ://doi.org/10.4049/jimmu nol.13020 64. ing cells; DMSO: dimethyl sulfoxide; SEB: Staphylococcus aureus enterotoxin 6. Yan J, Greer JM, Hull R, O’Sullivan JD, Henderson RD, Read SJ, B; PMA: Phorbol 12-myristate 13-acetate; PFA: paraformaldehyde; CMV: et al. The effect of ageing on human lymphocyte subsets: com- cytomegalovirus; DOTA: 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic parison of males and females. Immun Ageing. 2010;7:4. https ://doi. acid; PCA: principal components analysis; OPLS-DA: orthogonal partial least org/10.1186/1742-4933-7-4. square-discriminant analysis; viSNE: visualization tool for high-dimensional 7. Briceño O, Lissina A, Wanke K, Afonso G, von Braun A, Ragon K, et al. single-cell data based on the t-Distributed Stochastic Neighbor Embedding Reduced naïve CD8(+) T-cell priming efficacy in elderly adults. Aging Cell. (t-SNE) algorithm; HI: hemagglutin-inhibiting. 2016;15:14–21. https ://doi.org/10.1111/acel.12384 . 8. McElhaney JE, Xie D, Hager WD, Barry MB, Wang Y, Kleppinger A, et al. T Authors’ contributions cell responses are better correlates of vaccine protection in the elderly. J CL performed the experimental portions, CL and SK performed statistical Immunol (Baltimore, Md : 1950). 2006;176:6333–9. analysis, CL and NS performed studies to establish stimulation conditions, 9. Wilkinson TM, Li CKF, Chui CSC, Huang AKY, Perkins M, Liebner JC, et al. SL conceived of and designed the study, all authors participated in study Preexisting influenza-specific CD4 T cells correlate with disease protec- revisions, data interpretation, drafting and revising the article for important tion against influenza challenge in humans. Nat Med. 2012;18:274–80. intellectual content. All authors read and approved the final manuscript.https ://doi.org/10.1038/nm.2612. 10. Falsey AR, Walsh EE, Looney RJ, Kolassa JE, Formica MA, Crid- Author details dle MC, et al. Comparison of respiratory syncytial virus humoral Department of Immunity, Transplant, Infection, Stanford University, Stanford, immunity and response to infection in young and elderly USA. Translational Sciences, MedImmune, Mountain View, CA, USA. adults. J Med Virol. 1999;59:221–6. https://doi.org/10.1002/ (SICI)1096-9071(199910)59:2<221:AID-JMV16>3.0.CO;2-H. Acknowledgements 11. Cherukuri A, Patton K, Gasser RA, Zuo F, Woo J, Esser MT, et al. Adults Authors wish to acknowledge the Ph1a subjects, Ph1a clinical sites (Miami 65 years old and older have reduced numbers of functional memory T Research Associates, Compass Research, and Accelovance) and key MedIm- cells to respiratory syncytial virus fusion protein. Clin Vaccine Immunol. mune individuals involved in Ph1a study design and operation: Judy Falloon, 2013;20:239–47. https ://doi.org/10.1128/CVI.00580 -12. Kristin Kopacz, Mark Esser, and Tonya Villafana. 12. Cusi MG, Martorelli B, Di Genova G, Terrosi C, Campoccia G, Correale P. Age related changes in T cell mediated immune response and effector Competing interests memory to Respiratory Syncytial Virus (RSV ) in healthy subjects. Immun The authors declare that they have no competing interests. Ageing. 2010;7:14. https ://doi.org/10.1186/1742-4933-7-14. 13. van Essen GA, Beran J, Devaster J-M, Durand C, Duval X, Esen M, et al. Availability of data and materials Influenza symptoms and their impact on elderly adults: randomised trial The datasets used and analyzed during the current study are available from of AS03-adjuvanted or non-adjuvanted inactivated trivalent seasonal the corresponding author on reasonable request. influenza vaccines. Influenza Other Respir Viruses. 2014;8:452–62. https :// doi.org/10.1111/irv.12245 . Consent for publication 14. Couch RB, Bayas JM, Caso C, Mbawuike IN, López CN, Claeys C, Not applicable. et al. Superior antigen-specific CD4 T-cell response with AS03- adjuvantation of a trivalent influenza vaccine in a randomised trial Ethics approval and consent to participate of adults aged 65 and older. BMC Infect Dis. 2014;14:425. https ://doi. The clinical protocol was approved by relevant institutional review boards or org/10.1186/1471-2334-14-425. ethics committees. All subjects provided written informed consent. This study 15. Parodi V, de Florentiis D, Martini M, Ansaldi F. Inactivated influenza was conducted in accordance with the Declaration of Helsinki and good clini- vaccines: recent progress and implications for the elderly. Drugs Aging. cal practice guidelines. This study did not involve the use of any animal. 2011;28:93–106. https ://doi.org/10.2165/11586 770-00000 0000-00000 . 16. Falsey AR, Treanor JJ, Tornieporth N, Capellan J, Gorse GJ. Randomized, Funding double-blind controlled phase 3 trial comparing the immunogenicity of Funding was provided by MedImmune, a subsidiary of AstraZeneca. high-dose and standard-dose influenza vaccine in adults 65 years of age and older. J Infect Dis. 2009;200:172–80. https ://doi.org/10.1086/59979 0. 17. Levin MJ, Smith JG, Kaufhold RM, Barber D, Hayward AR, Chan CY, et al. Publisher’s Note Decline in varicella-zoster virus ( VZV )-specific cell-mediated immunity Springer Nature remains neutral with regard to jurisdictional claims in pub- with increasing age and boosting with a high-dose VZV vaccine. J Infect lished maps and institutional affiliations. Dis. 2003;188:1336–44. https ://doi.org/10.1086/37904 8. 18. Fleming DM, Taylor RJ, Lustig RL, Schuck-Paim C, Haguinet F, Webb DJ, Received: 12 February 2018 Accepted: 28 May 2018 et al. Modelling estimates of the burden of respiratory syncytial virus infection in adults and the elderly in the United Kingdom. BMC Infect Dis. 2015;15:443. https ://doi.org/10.1186/s1287 9-015-1218-z. 19. Kurzweil V, Tang R, Galinski M, Wang K, Zuo F, Cherukuri A, et al. Trans- lational sciences approach to RSV vaccine development. Expert Rev References Vaccines. 2013;12:1047–60. https ://doi.org/10.1586/14760 584.2013.82470 1. Goronzy JJ, Weyand CM. Understanding immunosenescence to improve responses to vaccines. Nat Immunol. 2013;14:428–36. https ://doi. 20. McElhaney JE, Xie D, Hager WD, Barry MB, Wang Y, Kleppinger A, et al. T org/10.1038/ni.2588. cell responses are better correlates of vaccine protection in the elderly. J 2. Goronzy JJ, Weyand CM. Successful and maladaptive T cell aging. Immu- Immunol. 2006;176:6333–9. nity. 2017;46:364–78. https ://doi.org/10.1016/j.immun i.2017.03.010. 21. Falloon J, Ji F, Curtis C, Bart S, Sheldon E, Krieger D, et al. A phase 1a, 3. Wertheimer AM, Bennett MS, Park B, Uhrlaub JL, Martinez C, Pulko V, et al. first-in-human, randomized study of a respiratory syncytial virus F protein Aging and cytomegalovirus infection differentially and jointly affect dis- vaccine with and without a toll-like receptor-4 agonist and stable emul- tinct circulating T cell subsets in humans. J Immunol. 2014;192:2143–55. sion adjuvant. Vaccine. 2016;34:2847–54. https ://doi.org/10.1016/j.vacci https ://doi.org/10.4049/jimmu nol.13017 21. ne.2016.04.002. 4. Palmer DB. The effect of age on thymic function. Front Immunol. 22. Falloon J, Talbot HK, Curtis C, Ervin J, Krieger D, Dubovsky F, et al. Dose 2013;4:316. https ://doi.org/10.3389/fimmu .2013.00316 . selection for an adjuvanted respiratory syncytial virus F protein vaccine Lingblom et al. J Transl Med (2018) 16:153 Page 12 of 12 for older adults based on humoral and cellular immune responses. Clin vaccine in adults 50–80 years of age. Hum Vaccines Immunother. Vaccine Immunol. 2017;24:e00157. https ://doi.org/10.1128/cvi.00157 -17. 2014;10:1195–203. https ://doi.org/10.4161/hv.28313 . 23. Tanner SD, Bandura DR, Ornatsky O, Baranov VI, Nitz M, Winnik MA. Flow 27. He X-S, Holmes TH, Sasaki S, Jaimes MC, Kemble GW, Dekker CL, et al. cytometer with mass spectrometer detection for massively multiplexed Baseline levels of influenza-specific CD4 memory T-cells affect T-cell single-cell biomarker assay. Pure Appl Chem. 2008;80:2627–41. https :// responses to influenza vaccines. PLoS ONE. 2008;3:e2574. https ://doi. doi.org/10.1351/pac20 08801 22627 .org/10.1371/journ al.pone.00025 74. 24. Patton K, Aslam S, Shambaugh C, Lin R, Heeke D, Frantz C, et al. Enhanced 28. O’Connor J-E, Herrera G, Martínez-Romero A, de Oyanguren FS, Díaz immunogenicity of a respiratory syncytial virus (RSV ) F subunit vaccine L, Gomes A, et al. Systems Biology and immune aging. Immunol Lett. formulated with the adjuvant GLA-SE in cynomolgus macaques. Vaccine. 2014;162(1 Pt B):334–45. https ://doi.org/10.1016/j.imlet .2014.09.009. 2015;33:4472–8. https ://doi.org/10.1016/j.vacci ne.2015.07.025. 29. Fourati S, Cristescu R, Loboda A, Talla A, Filali A, Railkar R, et al. Pre-vacci- 25. Finck R, Simonds EF, Jager A, Krishnaswamy S, Sachs K, Fantl W, et al. nation inflammation and B-cell signalling predict age-related hypore - Normalization of mass cytometry data with bead standards. Cytometry A. sponse to hepatitis B vaccination. Nat Commun. 2016;7:10369. https :// 2013;83:483–94. https ://doi.org/10.1002/cyto.a.22271 .doi.org/10.1038/ncomm s1036 9. 26. Ross TM, Lin CJ, Nowalk MP, Huang H-H, Spencer SM, Shay DK, et al. Influ- 30. Mather M, Jacobsen LA, Pollard KM. Aging in the United States. Popul Bull. ence of pre-existing hemagglutination inhibition titers against historical 2015;70:1–8. influenza strains on antibody response to inactivated trivalent influenza Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your ﬁeld rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions
Journal of Translational Medicine – Springer Journals
Published: Jun 5, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera