Professional Documents
Culture Documents
Vaccine
journal homepage: www.elsevier.com/locate/vaccine
Short communication
a r t i c l e i n f o a b s t r a c t
Article history: Background: Varicella-zoster virus (VZV) infection during pregnancy is associated with serious fetal
Received 2 July 2019 anomalies. The live-attenuated VZV vaccine was approved in 1995, so many vaccinated women are
Received in revised form 24 December 2019 now of childbearing age. The question of long-term immunity to varicella is critical because breakthrough
Accepted 31 December 2019
chickenpox can occur after vaccination.
Available online xxxx
Objective: To compare humoral and T cell immunity between women of childbearing age who were
immunized by vaccination or chickenpox disease.
Keywords:
Study design: Non-pregnant females between 18 and 36 years old with a history of VZV immunization
Varicella-zoster virus
Chickenpox
(n = 20) or prior chickenpox disease (n = 20) were recruited. IgG antibody titers and T cell responses were
Pregnancy measured by flow cytometry-based methods in serum and peripheral blood, respectively.
Fetal anomalies Results: There were no significant differences in median antibody titers between vaccinated and chicken-
Vaccine pox groups (p = 0.34). The chickenpox group had significantly higher levels of VZV antigen-specific CD4 T
Flow cytometry cells (p = 0.004).
Humoral immunity Conclusion: Natural infection induced higher VZV-specific T cell immune responses than vaccination.
Cellular immunity Ó 2020 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.vaccine.2019.12.067
0264-410X/Ó 2020 Elsevier Ltd. All rights reserved.
Please cite this article as: E. Tourtelot, S. Quataert, J. C. Glantz et al., Women who received varicella vaccine versus natural infection have different long-
term T cell immunity but similar antibody levels, Vaccine, https://doi.org/10.1016/j.vaccine.2019.12.067
2 E. Tourtelot et al. / Vaccine xxx (xxxx) xxx
2. Materials and methods 2090KZ San Jose, CA) for 30 min at 4 °C; stained with a cocktail
of antibodies for intracellular markers at optimal concentrations
Two cohorts of 20 non-pregnant adult female subjects between in 1xBD permeabilization buffer (BD Biosciences,Cat#51-2091KZ,
18 and 36 years of age either vaccinated for varicella or having a San Jose, CA) for 80 min at 4 °C; and fixed with 2% formalde-
history of chickenpox were enrolled between March 2015 and June hyde/PBS prior to running on a BD LSRII [11].
2016. Patients presenting to the University of Rochester Medical Flow data were manually gated using FlowJoTM and also ana-
Center (URMC) outpatient setting for gynecology care were identi- lyzed using an automated high dimensional clustering algorithm,
fied by the study coordinator and asked to participate. The study SWIFT [12]. Manual and automated definitions of T cell popula-
was advertised in the waiting room using flyers and on the univer- tions are described in Supplementary Data (Figs. 1 and 2).
sity Health Research Website. Vaccination records were obtained Operators performing primary analysis of antibody and CMI
on all subjects (except those who had chickenpox prior to 1995) responses were blinded to the experimental groups. Wilcoxon rank
to ensure no subjects had both chickenpox and vaccination. An sum and chi-square tests were used to compare age; years since
accurate self-reported year of infection was required in the chick- exposure to VZV; and race between cohorts. Antibody and CMI
enpox cohort. Subjects were given an honorarium. Exclusion crite- responses were compared by two-sided Wilcoxon rank sum tests.
ria included male sex, age <18 or >36, pregnancy, inability to
provide vaccination records or date of chickenpox infection, a his-
tory of shingles, HIV, immunosuppression, and/or use of illegal 3. Results
drugs excluding marijuana. Health care workers having direct
patient contact were excluded to eliminate potential increased Demographics for age, race and time since exposure to chicken-
exposure to wild-type virus. All subjects in both groups were from pox or vaccine are shown in Table 1. Linear regression and ANOVA
the US, and most were from the North East particularly the Roche- showed no association between these three factors and median
ster community. The study was approved by the institutional antibody levels (all p > 0.05). Adjusting for these three variables
review board (#00055668) at the URMC. Informed consent was using multivariable linear regression and ANOVA had no effect
obtained from all subjects. on the median antibody levels before or after log transformation
Approximately 75 ml of whole blood were collected from each of continuous variables such that their distribution became Gaus-
subject. Plasma was isolated and stored at 80 C. Peripheral blood sian (all p > 0.05).
mononuclear cells (PBMC) were isolated by Ficoll Hypaque gradi- All subjects had positive levels of IgG antibody at the 1:4 dilu-
ent centrifugation and cryopreserved in 90% FBS/10%DSMO and tion cut off, normally associated with protection. Although the
stored in liquid nitrogen. PBMC were 90% viable after thawing median antibody levels were slightly higher in the chickenpox
as assessed by trypan blue exclusion. One sample of PBMC from group, the two groups did not differ significantly (p = 0.34) (Fig. 1).
the vaccine group was lost during processing due to operator tech- Unexpectedly, the proportions of total CD4 T cells were higher
nical error unrelated to participant characteristics. One vaccinated in naturally-infected than vaccinated subjects both by manual
subject received only one dose of VZV vaccine and was excluded (p = 0.0002) and automated (p = 0.0009) analysis (Fig. 2). This
from all analyses. was due mainly to an increase in naïve (CD45RAhigh) cells.
The fluorescent antibody to membrane antigen assay (FAMA)
was used to titrate antibody binding to live unfixed VZV infected
human embryonic lung fibroblast (MRC-5) cells with a positive
IgG cutoff titer of 1:4 [9]. Plasma centrifuged at 14,000 RCF for
5 min at 4 °C to remove particulates and dilutions were incubated
with VZV-infected cells for 30 min on ice. After washing twice in
assay buffer (0.05 M PBS/1% BSA), the cells were stained with
1:1000 dilution of goat anti-human IgG-AlexaFluor 647 (Southern
Biotech, AL, US) for 30 min at 4 °C in the dark, washed again, fixed
with 4% formaldehyde in PBS and analyzed on a BD LSRII flow
cytometer. The median fluorescent intensity for VZV-infected cells
at a 1:400 plasma dilution (minus background with no plasma) in
the linear part of the titration curve was used to compare antibody
levels in all subjects.
Anti-VZV cytokine responses in T cells were assessed by in-vitro
stimulation of PBMC with 15-mer VZV IE-62 peptides with 11
amino acid overlap (325 peptides, PepMix VZVIE62, JPT Peptide
Technologies), or a lysate of Oka VZV-infected MRC cells, followed
by intracellular cytokine staining (ICS) and flow cytometry. This
method provides more detailed information on the cytokine pat-
terns of VZV-responsive T cells than Elispot or RCF assays [10].
1.5 million viable PBMC were thawed, rested overnight in RPMI
1640 medium with 10% FBS and 1 mg/mL DNAse, stimulated for
2 h at 37 °C with either 1 mg/mL peptide in medium with 0.32%
DMSO, medium with 0.32% DMSO only, 2x104 cell equivalents/mL
Fig. 1. Relative Anti-VZV antibody levels in VZV vaccinated and naturally-
of VZV-infected cell lysate or control uninfected lysate. Cells were infected subjects. The whisker plot shows the median fluorescent intensity of the
then incubated for an additional 8 h in the presence of 1:1000 Gol- binding of anti-VZV antibody in plasma to VZV-infected MRC5 cells at a dilution of
giPlugTM (BD Biosciences, Cat#51-2301KZ, San Jose, CA) plus 2mi- 1:400 as measured by flow cytometry FAMA (N = 20 Natural Infection, N = 19 VZV
croM monensin (Sigma-Aldrich, Cat#M5273, US). Cells were Vaccine groups) and the 25 and 75% interquartile range with the median (line) and
mean (X) values for each group. The MFI (2530 ± 284, N = 9) of VZV-infected cells
stained with a cocktail of antibodies at optimal dilution for surface
with anti-IgG/AF647 alone was subtracted from the MFI at 1:400. MFI for unstained
markers and 7-AAD for dead cells for 80 min at 4 °C; permeabilized VZV infected cells was 2472 ± 287 (N = 9). No staining was seen at 1:40 plasma
and fixed with 1xBD Cytofix/CytopermTM (BD Biosciences, Cat#51- dilutions on non-infected MRC-5 cells, data not shown).
Please cite this article as: E. Tourtelot, S. Quataert, J. C. Glantz et al., Women who received varicella vaccine versus natural infection have different long-
term T cell immunity but similar antibody levels, Vaccine, https://doi.org/10.1016/j.vaccine.2019.12.067
E. Tourtelot et al. / Vaccine xxx (xxxx) xxx 3
Fig. 2. Bulk T cell populations in VZV vaccinated and naturally-infected subjects. T cell sub-populations were assessed in unstimulated PBMC samples by antibody
staining and flow cytometry. Data analysis was performed by manual gating in FlowJo (bottom), and also by SWIFT clustering analysis (top). All cell populations were
normalized to one million live T cells (CD3 + 7-AAD- lymphocytes). T = CD3+7-AAD- lymphocytes; CD4 = CD3+7-AAD-CD4+ T cells; CD4mem = CD3+7-AAD-CD4+CD45RA- T
cells; CD4naive = CD3+7-AAD-CD4+CD45RA+ T cells; CD8 = CD3+7-AAD-CD8+ T cells; CD8mem = CD3+7-AAD-CD8+CD45RA- T cells; CD8naive = CD3+7-AAD-CD8+CD45RA+ T
cells.
Table 1
Baseline characteristics of the study population.
Subjects With: Varicella Vaccination N = 19 Varicella Vaccination (minus lost sample) N = 18 Chicken Pox Diagnosis N = 20 q-value*
Mean Age (range) 21 (18–28) 21 (18–28) 27 (19–33) q < 0.05
Mean Years since exposure (range) à
17 (12–20) 17 (12–20) 22 (12–27) q < 0.05
Race**
White or Caucasian 9 (47.3) 8 (45) 12 (60) q < 0.05
Black or African American 8 (42.1) 8 (45) 2 (10)
Asian 0 0 (0) 2 (10)
Other or no comment 2 (10.5) 2 (10) 4 (20)
*
q-values given between vaccinated and chickenpox disease groups calculated by Chi-square and Wilcoxon Rank Sum analysis.
**
Data shown as number of subjects and (%).
à
Either first vaccine dose or date of chickenpox disease.
Conversely, CD8 T cell populations were lower in naturally infected ducing TNFa, IFNc and IL-2, or only TNFa and IL-2. The chickenpox
subjects. group had higher levels of VZV-specific CD4 T cells than vaccinated
After antigen stimulation, strong CD4 T cell responses to VZV IE- subjects, as assessed by either traditional manual gating or by the
62 peptides were detected in most subjects, with most cells pro- automated SWIFT algorithm (Fig. 3A, B). The responses of the
Please cite this article as: E. Tourtelot, S. Quataert, J. C. Glantz et al., Women who received varicella vaccine versus natural infection have different long-
term T cell immunity but similar antibody levels, Vaccine, https://doi.org/10.1016/j.vaccine.2019.12.067
4 E. Tourtelot et al. / Vaccine xxx (xxxx) xxx
Fig. 3. Anti-VZV CD4 T cell responses in VZV vaccinated and naturally-infected subjects. PBMC were stimulated with VZV peptides; DMSO solvent controls; VZV-infected
cell lysate; or uninfected control cell lysate, then analyzed by ICS and flow cytometry. The resulting FCS files were analyzed by SWIFT clustering (A, C, D), or manually (B).
DMSO control values were subtracted from peptide responses (A, B, C) and uninfected lysate values were subtracted from VZV-infected lyate results (D). Cell numbers were
normalized to total live lymphocytes (A, B, D) or total CD4 T cells (C). After log transformation of the dependent variable and adjustment for age, years since exposure, and
race using backward linear regressions, TNFa + IL-2 + IFNc- anti-VZV peptide responses (A) were no longer significant.
chickenpox group were higher whether the results for each subject VZV is unusual in that infection results in latent infections that
were normalized to total live events (p = 0.004) or total CD4 T cells are hypothesized to undergo silent reactivations that boost CMI
(p = 0.008) (Fig. 3A, C). Responses to VZV-infected cell lysate were [15]. The Oka vaccine strain also induces latent infection [16]. Both
generally higher, possibly due to the presence of additional protein groups may also receive boosting from wild-type virus in the com-
antigens. Importantly, the chickenpox group also had significantly munity. This may explain why the highest CMI occurred in
higher responses to this antigen than the vaccine group (Fig. 3D). naturally-infected subjects in their mid-thirties, although there
CD8 T cell responses were not detected, possibly because: CD8 was considerable variation [17]. As our vaccinated cohort is on
anti-VZV responses require two antigen stimulations in vitro [13]; average six years younger and has five years less time since initial
the 15-mer peptides were sub-optimal for CD8 responses as prote- infection/vaccination by VZV, the vaccinated cohort may have
olytic cleavage would be required to produce the shorter peptides lower CMI responses because there has been less time for these
needed for MHC class I presentation; and the VZV lysate antigens two potential mechanisms of boosting. Alternatively, CMI
would not be presented efficiently on MHC class I. responses may be lower because the wild-type strains may
undergo more silent reactivations than the Oka strain. Thirdly,
4. Discussion the vaccine strain may induce lower initial clonal expansion of
VZV-specific CD4 T cells, relative to wild-type.
A large, recent study of 10,173 air force recruits (2571 females) The lower CMI response raises the possibility that vaccinated
showed that vaccinated subjects were 24% less likely to be women may be at increased risk of breakthrough infection. Break-
seropositive than naturally-infected subjects [14] using the Bio through infection is usually, but not always mild (possibly due to
Rad Bio Plex 220 MMRV IgG multiplex assay. Our study used the residual CMI responses that may be high enough to contain the
more sensitive FAMA assay for IgG antibody levels, and detected infection). The risk of breakthrough disease complications would
anti-VZV antibodies in all subjects. Consistent with the air force be particularly important for a pregnant woman, because the fetus
study, the anti-VZV antibody titers in our naturally-infected sub- would also be at risk. If breakthrough infection occurs around the
jects were slightly greater than in the vaccinated subjects, but this time of delivery, the neonate would also be at risk.
did not reach statistical significance, possibly due to our smaller It is not clear why the total CD4 T cell numbers (not VZV-
number of subjects. specific T cells) are higher in our naturally-infected subjects - this
Please cite this article as: E. Tourtelot, S. Quataert, J. C. Glantz et al., Women who received varicella vaccine versus natural infection have different long-
term T cell immunity but similar antibody levels, Vaccine, https://doi.org/10.1016/j.vaccine.2019.12.067
E. Tourtelot et al. / Vaccine xxx (xxxx) xxx 5
Please cite this article as: E. Tourtelot, S. Quataert, J. C. Glantz et al., Women who received varicella vaccine versus natural infection have different long-
term T cell immunity but similar antibody levels, Vaccine, https://doi.org/10.1016/j.vaccine.2019.12.067