Molecular profiles of CXCR5+ CD4 Memory T cells Associated with Flu Vaccine Responses

Lesley de Armas 1, Nicola Cotugno 1,2,3, Suresh Pallikkuth 1, Paolo Palma 2, Biju Issac 4, Alberto Cagigi 2, Paolo Rossi 2,3, and Savita Pahwa 1

Background: HIV-infected patients of all ages consistently underperform in flu vaccine efficacy. Peripheral T follicular helper (pTfh) cells, a subset of CD4 T cells which provide help to B cells for developing into Ab secreting cells, are characterized by (CXC) motif chemokine R5 (CXCR5) expression. We previously demonstrated functional impairment of this subset in adult HIV+ flu vaccine NR (Pallikkuth Blood 2012). Microarray data from a pediatric HIV-infected cohort receiving the 2009/10 pandemic influenza vaccine revealed upregulation of CXCR5 in whole blood of vaccine responders (R) compared to non-responders (NR) 4 weeks post-vaccination. We hypothesized that pTfh play a role in antibody responses to the flu vaccine and investigated whether defects in pTfh function could be identified prior to vaccination. Methods: Fluidigm BioMark RT-PCR was performed on pTfh cells from a cohort of virally suppressed, vertically infected HIV+ children who received the 2012 seasonal flu vaccine. PBMC from baseline (T0) were stimulated with H1N1 antigen (16h) and then sort-purified as 500 cell pools directly into PCR buffer. This methodology allowed us to study a panel of 96 genes related to Tfh function, immune activation and inflammation, TCR signaling, and co-stimulation/inhibition simultaneously in pTfh from R (n=7) and NR (n=9), compared to healthy controls (HC, n=9). The patients in this cohort have been immunized annually and had sero-protective antibody titers at baseline; vaccine R were defined as having hemagluttination inhibition titers > 1:40, and > 4-fold increase above baseline 4 wks post-vaccination. Results: There were no observed differences in the frequencies of pTfh between the groups at baseline or after in vitro stimulation (p=0.9, ANOVA). However, transcriptomic analysis of pTfh from HC and R demonstrated a response to H1N1 stimulation which was marked by increased or stable expression of Tfh-related genes (MAF, IL21, BCL6, and ICOS) and TCR signaling genes (CD3D, MAPK3, FYN, PKC A), while in NR they were significantly downregulated after stimulation. Interestingly, pTfh from NR displayed higher levels of the inhibitory receptor CTLA4 (p=0.01, student’s t test) compared to R suggesting a possible mechanism for reduced responses to vaccination. Conclusions: Targeted molecular profiling of pTfh in previously vaccinated HIV-infected children, before influenza vaccination revealed antigen-driven favorable gene expression selectively in R. These results suggest that quality rather than quantity of pTfh in the periphery is controlling responses to vaccination.

Abstract

Experimental Strategy

+/- H1N1 Ag 16 hours

Pre-Vaccination PBMC

IN VITRO CULTURE SURFACE STAINING AND FLOW SORTING FLUIDIGM AND BIOINFORMATICS

500 cells sorted directly into PCR Buffer containing primers for 96 genes

18 cycles PCR to pre-amplify specific transcripts

96.96 dynamic array chip Fluidigm BioMark and IFC controller

Data Ouput: Ct values Bioinformatics analysis and Systems Biology

1 Department of Microbiology and Immunology, University of Miami, FL, 2 Children Hospital Bambino Gesú and 3 University of Rome “Tor Vergata” Rome, Italy, 4 Sylvester Comprehensive Cancer Center, University of Miami, FL

#947

Results

Figure 1. Principal Component Analysis (PCA) shows clustering of sorted populations. Based on gene expression of the panel (above) total CD4 and pTFH cells overlap, while PBMC (unsorted) and Non-pTFH exhibited distinct patterns of gene expression. Example shown from unstimulated cells of HIV neg. controls but HIV+ patients exhibited a similar pattern.

Results

Gene Panel

Results

Summary and Conclusions

Figure 2. Differentially expressed genes (DEGs) in T cell subsets at pre-vaccination timepoint. Venn diagram showing DEGs (ANOVA p<0.05) between HIV+ Flu vaccine responders and non-responders without in vitro stimulation. The DEGs identified in PBMC did not overlap with CD4, pTFH and non-pTFH highlighting the need to perform transcriptomic studies in cell subsets.

Baselinne Characteristics HIV NR HIV R HIV neg

Age years, mean (SEM) 15.16 (2.173) 13.72 (2.35) 14.3 (3.39)

n (female) 12 (7) 11 (5) 10 (5)

Lymphocytes/mm3 2494 (278.9) 2961 (363.1) 3063 (427.8)

%CD4+ T cells, mean (SEM) 37.97 (4.92) 32.49 (6.09) 29.79 (6.25)

HIV RNA, <50cp/mL, n 11/12 10/11 NA

SPR to H1N1 at T0, % 92 91 60

SPR H3N2 at T0, % 100 100 40

SPR influenza B strain at T0, % 16 9 30

Patient Cohort

Background

CD4+ T pTFH Non-pTFH

Study Objectives

• HIV-infected children may not fully restore immune function despite treatment with cART, leading to suboptimal vaccine responses. There is a gap in the knowledge of the predictive transcriptomic signatures that influence immune responses to vaccines in virologically suppressed HIV-infected children.

• Microarray data from peripheral blood of HIV-infected children responding to influenza vaccination showed upregulation of CXCR5, a marker of peripheral T follicular helper cells (pTFH), post-vaccination (1).

• pTFH are a subset of circulating CD4+ memory T cells which express CXCR5 and share functional characteristics with TFH residing in germinal centers. Our group has shown previously in HIV-infected adults that a functional impairment of pTFH is associated with deficient Ab responses to influenza vaccination (2).

• ‘Systems biology’ approaches have been employed to identify specific signatures of disease and vaccine response. However, most often such studies have been performed in PBMC samples of healthy subjects.

• To investigate whether a predictive transcriptional signature in pTFH could be defined that correlates with influenza vaccine induced Ab responses in HIV-infected children.

• To compare pre-vaccination gene expression profiles in purified T cell subsets and PBMC in response to in vitro stimulation with the influenza vaccine antigen (H1N1) to identify correlates of vaccine responses.

Table 1. Pediatric patients (vertically infected HIV+ and HIV negative) in Childrens Hospital Bambino Gesú, Rome, Italy were given a single dose of trivalent inactivated influenza vaccine (TIV) during 2012-2013 Flu season. Peripheral blood was collected before (T0) and 21 days post-vaccination (T1); PBMC were cryopreserved. Numbers of patients with Seroprotection (SPR, Ab titer >1:40) at T0, Vaccine Responders at T1 (R, >4-fold increase in Ab titer to H1N1 antigen and >80 H1N1 specific spots/106 by ELISpot assay at T1], and Vaccine Non-responders (NR) are depicted.

Table 2. Gene panel used for Fluidigm BioMark experiments, divided into functional groups. All primers and probes were human Taqman gene expression assays. Genes were chosen according to basic biological queries, literature, and microarray data generated by our lab.

Functional Groups Genes

TCR-signaling CD4, CD3D, CD2, FYN, SYK, ZAP70, CAMK4, SELL, NFATC1, PKCA, MAPK3, PIK3C2B, PLCG, DUSP4, DUSP6, DOCK8, CAV1, STAT3, STAT4, STAT5A, MTOR

Immune Activation CD74, CD38, CD69, FAS, LAG3

Inhibitory Receptors LILRB1, CTLA4, PDL1, HAVCR2, KLRG1, TIGIT

Cytokines/Chemokines IL2, IL10, IL6, IL1B, IFNG

Cytokine/Chemokine Receptors CCR2, CXCR3, IL6ST, IL6RA, CCR6, CCR7, IL7R, CCR5, CXCR4, IL2RA, IL21R

IFN-induced IFIT2, IFNAR2, OAS1, MX1, STAT1, NFKB1

Inflammation CXCL10, PTEN, ICAM1, IL12RB2, ADAM17, ABCB1, ITCH, NOD2, PTX3

TFH-associated BCL6, CXCR5, ICOS, IL21, MAF, IRF4, PDCD1

Transcription Factors TBX21, PRDM1, GATA3, RORA, BATF, EOMES, SOCS1, FOXO3, FOXP3, RUNX3, RORC, ID2, ID3, TFAM, TGIF1

Viral restriction Factors SAMHD1, TRIM5, BST2, APOBEC3G

TNF Family TNFSF13 (APRIL), TNFRSF13 (LIGHT), TNFRSF4 (OX40), CD40L, TNF

Figure 5. pTFH from Non-responders fail to induce TFH and TCR-related gene expression in response to H1N1 stimulation. Colored squares in panels A and B represent the log10 (Fold Change) of the mean ET for each group. Fold change= H1N1 stimulated/Unstimulated conditions. Heatmaps were generated using ‘pheatmap’ package in R.

Figure 3. Network analysis of the DEGs between responders and non-responders in PBMC. Graphic was produced using Genemania online software. Overall, Responders exhibited lower expression of the identified DEGs compared to Non-responders suggesting that immune activation and inflammation at the time of vaccination can lead to impaired responses even for ‘sero-protected’ HIV-infected children.

Cell-specific DEGs in Responders vs Non-responders before Vaccination

PBMC from Responders and Non-responders exhibit differential expression in Immune activation and inflammation pathways.

Targeted transcriptomic analysis in CD4 T cells

Differential Gene Expression in pTFH at Pre-vaccination

N R R2 2

2 4

2 6

2 8

3 0

3 2

IL 7R

log

(2)

Exp

ress

ion p = 0 .0 3 2

N R R1 0

1 5

2 0

2 5

3 0

ID 2

log

(2)

Ex

pre

ssio

n p = 0 .0 2 3

N R R1 6

1 8

2 0

2 2

2 4

2 6

2 8

IC O S

log

(2)

Ex

pre

ssio

n p = 0 .0 3 9

In vitro Stimulation with H1N1 antigen reveals ‘unresponsive’ pTFH in HIV+ Non-responders

H C N R R0

2 0

4 0

6 0

8 0

%p

TF

H(C

XC

R5

+ o

f T

CM

)

n s

p = 0 .0 5

Figure 4. Gene expression rather than cell frequencies are altered in pTFH before vaccination. A) NR and R exhibited marginally higher frequencies of pTFH compared to healthy controls (HC) and no significant differences were observed between NR and R. B) Log(2) expression (ET=40-Ct) values for each of the indicated genes are plotted for NR and R. ET =13 represents the lower limit of detection.

A B

A TFH-related B TCR-related

Integration of Fluidigm Gene Expression data with Clinical and Immunological data

Figure 6. Cluster Image Map (CIM) displays Pearson correlation values for data integration. Transcript-‘omic’ data and 92 individual clinical (age, viral load, etc…) and immunological parameters derived from flow cytometry, ELISpot, ELISA, and HAI assays were correlated using the ‘mixOmics’ package run in R statistical software (3).

Genes

Immunological Parameters

Data Integration shows Correlations between Responders and Immunological Marker of Vaccine Response

1 8 2 0 2 2 2 4 2 6 2 8 3 00

2 0 0

4 0 0

6 0 0

8 0 0B A T F

B A T F (lo g 2 E x p re s s io n )

H1

N1

sp

ots

/10

6 (

T1

)

N RR

R 2= 0 .4 1 4 7

2 2 2 4 2 6 2 8 3 0 3 2 3 40

2 0 0

4 0 0

6 0 0

8 0 0IL 7 R

IL 7 R (lo g 2 E x p re s s io n )

H1

N1

sp

ots

/10

6 (

T1

)

N RR

R 2 = 0 .6 0 1 5

Figure 7. Gene expression in pTFH cells following in vitro stimulation correlates with immunological data in Responders only. A) BATF, IL7R, and IL21 expression positively correlate with T1 ELISpot data (previously collected) in Responders, while no correlation is observed in vaccine Non-responders. Importantly for IL21, the transcript was detectable in 4/12 NR and 8/11 R patients. BATF R (p= 0.0325), IL7R R (p=0.005), IL21 R (p=0.06).

1 6 1 8 2 0 2 2 2 40

2 0 0

4 0 0

6 0 0

8 0 0IL 2 1

IL 2 1 (lo g 2 E x p re s s io n )

H1

N1

sp

ots

/10

6 (

T1

)

N RR

R 2= 0 .4 6 6 8

• We used Fluidigm technology to interrogate gene expression of a curated panel of 96 genes from several cell populations (PBMC, CD4+ T cells, pTFH, and non-pTFH) of HIV-infected and HIV negative children in samples collected before seasonal Flu vaccination in an attempt to identify predictive signatures of vaccine responses in patients who were classified as R and NR post vaccine.

• In vitro stimulation of PBMC from the pre-vaccination timepoint with H1N1 antigen revealed pTFH defects in gene induction related to TFH functions and TCR signaling, suggesting that quality of pTFH, rather than quantity, is associated with good responses to Flu vaccination.

• Statistical data integration (‘mixOmics’) program revealed a positive correlation between gene expression in Ag-stimulated pTFH from R and flu-specific antibody production post-vaccination.

• pTFH (and other cell types assayed) from HIV+ TIV R and NR exhibited significant differences in gene expression prior to vaccination. Work is ongoing to incorporate these results into a transcriptional signature of vaccine response.

References 1. de Armas, L et al. Transcriptomic analysis of response to the pandemic H1N1/09 influenza vaccine in HIV-infected children and adolescents. Manuscript in

preparation

2. Pallikkuth, S et al. 2012. Impaired peripheral blood T-follicular helper cell function in HIV-infected nonresponders to the 2009 H1N1/09 vaccine. Blood 120 985-93

3. Liquet et al. A novel approach for biomarker selection and the integration of repeated measures experiments from two assays. BMC Bioinformatics 2012 13:325

Acknowledgements This work was made possible by support from a pilot award (FY15) from the

Miami Center for AIDS Research (CFAR) at the University of Miami Miller School

of Medicine which is funded by a grant (P30AI073961) and (R01AI108472) to SP

from the National Institutes of Health (NIH). Also supported by grants obtained

from the Childrens Hospital Bambino Gesú, Rome, Italy.

Run Fluidigm BioMark