CLINICAL AND VACCINE IMMUNOLOGY, July 2009, p. 1025–1032
Copyright © 2009, American Society for Microbiology. All Rights Reserved.
Vol. 16, No. 7
Development of a Murine Mycobacterial Growth Inhibition Assay for
Evaluating Vaccines against Mycobacterium tuberculosis?†
Marcela Parra,1Amy L. Yang,1JaeHyun Lim,1Kristopher Kolibab,1Steven Derrick,1
Nathalie Cadieux,1Liyanage P. Perera,2William R. Jacobs,3
Michael Brennan,1,4and Sheldon L. Morris1*
Center for Biologics Evaluation and Research, United States Food and Drug Administration, Bethesda, Maryland1; Metabolism Branch,
Center for Cancer Research, National Cancer Institute, Bethesda, Maryland2; Howard Hughes Medical Institute, Department of
Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, New York3; and Aeras Global TB Vaccine Foundation,
Received 6 February 2009/Returned for modification 7 March 2009/Accepted 11 May 2009
The development and characterization of new tuberculosis (TB) vaccines has been impeded by the lack of
reproducible and reliable in vitro assays for measuring vaccine activity. In this study, we developed a murine
in vitro mycobacterial growth inhibition assay for evaluating TB vaccines that directly assesses the capacity of
immune splenocytes to control the growth of Mycobacterium tuberculosis within infected macrophages. Using
this in vitro assay, protective immune responses induced by immunization with five different types of TB
vaccine preparations (Mycobacterium bovis BCG, an attenuated M. tuberculosis mutant strain, a DNA vaccine,
a modified vaccinia virus strain Ankara [MVA] construct expressing four TB antigens, and a TB fusion protein
formulated in adjuvant) can be detected. Importantly, the levels of vaccine-induced mycobacterial growth-
inhibitory responses seen in vitro after 1 week of coculture correlated with the protective immune responses
detected in vivo at 28 days postchallenge in a mouse model of pulmonary tuberculosis. In addition, similar
patterns of cytokine expression were evoked at day 7 of the in vitro culture by immune splenocytes taken from
animals immunized with the different TB vaccines. Among the consistently upregulated cytokines detected in
the immune cocultures are gamma interferon, growth differentiation factor 15, interleukin-21 (IL-21), IL-27,
and tumor necrosis factor alpha. Overall, we have developed an in vitro functional assay that may be useful for
screening and comparing new TB vaccine preparations, investigating vaccine-induced protective mechanisms,
and assessing manufacturing issues, including product potency and stability.
The tuberculosis (TB) epidemic is a global public health
tragedy that is being fueled by the spread of human immu-
nodeficiency virus/AIDS and the increasing incidence of
multiple-drug-resistant Mycobacterium tuberculosis strains.
Annually, about 2 million people worldwide die from tuber-
culosis and 8 to 9 million new cases of this disease are
reported (34). Although the current TB vaccine, Mycobac-
terium bovis BCG, has been widely used for decades, its
effectiveness has been shown to be highly variable in well-
controlled clinical trials (5). While immunization with BCG
is effective against severe childhood disease, BCG does not
adequately protect against the most prevalent form of the
disease, adult pulmonary tuberculosis (13). Vaccinated in-
dividuals who become infected with TB are susceptible to
disease progression when the BCG-induced immune re-
sponses are suppressed or wane with time (32). Clearly, to
curb the global TB epidemic, more effective immunization
strategies must be generated and evaluated.
The development of new vaccines against TB has been hin-
dered by our limited understanding of the mechanisms of pro-
tective immunity against M. tuberculosis. While it is known that
acquired cellular immune responses are critical for controlling
tuberculosis infections, the cell subsets that confer antituber-
culosis protective immunity have not been adequately defined
(14). In addition, the immune mechanisms that are responsible
for inhibiting the intracellular growth of M. tuberculosis have
not been fully delineated and the surrogate biomarkers of this
growth inhibition remain unknown. Because it is difficult to
study the multiple components of the immune system and their
numerous interactions in vivo, the development of an in vitro
system which models the in vivo immune responses should
facilitate the identification of antituberculosis protective im-
mune mechanisms. The availability of a relevant in vitro
assay should allow a more direct study of the mediators of
protective immunity against M. tuberculosis in a controlled
system. Although in vitro mycobacterial growth inhibition
assays for human cells have been developed and are being
characterized for their capacity to detect vaccine-induced
immunogenicity in human clinical trials, the development
and assessment of preclinical assays to measure vaccine-
induced activity against M. tuberculosis has thus far been
limited (4, 6, 16, 18, 30, 35).
To accelerate TB vaccine development and investigations of
protective immune mechanisms, we initiated studies aimed at
developing a murine in vitro functional assay for evaluating the
protective activity of TB vaccines. For this assay, antitubercu-
losis protection was evaluated by targeting an important end
point, the control of M. tuberculosis growth within its primary
host cell, the macrophage. By assessing the immune-mediated
* Corresponding author. Mailing address: FDA/CBER, Building 29/
Room 502, 29 Lincoln Drive, Bethesda, MD 20892. Phone: (301)
496-5978. Fax: (301) 435-5675. E-mail: email@example.com.
† Supplemental material for this article may be found at http://cvi
?Published ahead of print on 20 May 2009.
inhibition of mycobacterial growth, we hypothesized that our
results would correlate more directly with in vivo protection
than the measurement of other immune responses, including
cytokine expression. In addition to assessing cellular immune
mechanisms, a relevant in vivo assay could be useful for screen-
ing and comparing new TB vaccine candidates. From a man-
ufacturing viewpoint, a standardized in vitro functional assay
could also be adapted to measure vaccine potency, lot-to-lot
production consistency, and vaccine stability.
Here we describe our initial results from the characteriza-
tion of a murine in vitro functional assay for assessing the
activity of TB vaccines. We show that vaccine-induced protec-
tion seen in vitro for five different TB vaccines correlates with
the antituberculosis protective immunity detected in a mouse
model of pulmonary TB. Also, we establish an in vitro profile
of cytokine expression which is associated with the activity of
BCG vaccine and demonstrate that similar in vitro cytokine
responses were detected for the four other types of TB vac-
cines that were tested in this study.
MATERIALS AND METHODS
Vaccines. The BCG Pasteur vaccine preparation was derived from the myco-
bacterial culture collection of the Trudeau Institute. Heat-killed BCG was pre-
pared by autoclaving 108CFU of BCG Pasteur at 121°C for 20 min. No myco-
bacterial growth was seen when aliquots of the heat-killed preparation were
incubated for 3 to 4 weeks on Lowenstein-Jensen slants. The SD1 DNA vaccine
was generated by cloning a fusion of the ESAT6 and antigen 85B genes into the
pVAX DNA vector (Invitrogen, San Diego, CA) as described previously (12).
The E6-85B protein is an ESAT6-antigen 85B M. tuberculosis fusion protein
which was purified by nickel affinity chromatography after cloning and expressing
the ESAT6-antigen 85B fusion gene in the pET23b vector system (Novagen, San
Diego CA). The protein-adjuvant formulation was prepared by mixing the fusion
protein (50 ?g/ml) with dimethyldioctadecylammonium bromide (DDA; 150
?g/ml; Kodak) and monophosphoryl lipid (MPL; 250 ?g/ml; Avanti Polar Lipids,
Alabaster, AL). The ?secA2 gene deletion mutant was isolated by electroporat-
ing the pMB179 suicide vector containing a ?secA2 allele and a sacB marker into
M. tuberculosis H37Rv and then counterselecting on Middlebrook 7H11 plates
containing 38 mM (NH4)2SO4and 3% sucrose (3). The MVA-4TB vaccine was
generated by cloning four M. tuberculosis genes (antigen 85A, antigen 85B,
ESAT6, and HSP65) as well as the interleukin-15 (IL-15) gene into a modified
vaccinia virus Ankara (MVA) vector (27).
Immunization schedules. In these in vivo studies, five C57BL/6 mice per group
were utilized. For the live BCG and ?secA2 vaccines, 106CFU were given once
subcutaneously. A dose equivalent to 106CFU of heat-killed BCG was also
injected once by the subcutaneous route. Five micrograms of the E6-85B protein
in the DDA (15 ?g)–MPL (25 ?g) adjuvant was administered three times, 2
weeks apart, while an identical dose and schedule of the adjuvant was given as a
control. For the DNA immunization, 200 ?g of the SD1 DNA vaccine or the
pVAX vector control was injected three times, 3 weeks apart, by the intramus-
cular route. Finally, two doses of 5 ? 107PFU of the MVA-4TB construct or the
MVA vector were given subcutaneously 1 month apart.
In vitro coculture assay. The coculture assay for evaluating TB vaccines was
based on procedures described earlier by Elkins and coworkers (1, 6, 7). The
primary modification to the published methods included the preparation and use
of the target cells for the assay, bone marrow macrophages (BMM?). These
procedural changes included using 30% fewer and more purified (without red
blood cells) macrophages, incubation after the M. tuberculosis macrophage in-
fection without antibiotic, and infecting with the WHO standard M. tuberculosis
Erdman strain. In our procedures, BMM? were flushed through the femurs of
C57BL/6 mice with Dulbecco’s modified Eagle’s medium (DMEM). The red
blood cells were then lysed in ACK buffer solution for 3 min. After washing the
cells and preparing a single-cell suspension, 7 ? 105cells/ml were suspended in
DMEM containing 10% fetal bovine serum, 10% L929a conditioned medium,
and 1% of the following reagents: L-glutamine, modified Eagle’s medium non-
essential amino acids, HEPES buffer solution, and sodium pyruvate. The cells
were then placed in each well of a 24-well plate and incubated for 7 days at 37°C
in 5% CO2. The medium was replaced every 2 to 3 days during the 7-day
incubation. After the 7-day culture, the concentration of BMM? was about 107
cells per well. For the mycobacterial infections, M. tuberculosis Erdman was
added to each well at a multiplicity of infection of 1:100 (bacteria to BMM?) for
2 hours and then the wells were washed five times with phosphate-buffered saline
(PBS). To determine the extent of bacterial uptake, a fraction of the macrophage
cultures was immediately lysed with 0.1% saponin and the resulting cell lysates
were diluted in PBS–0.04% Tween 80 and plated on Middlebrook 7H11 plates
supplemented with 10% oleic acid, albumin, dextrose, catalase (OADC) enrich-
ment medium (Becton Dickinson, Sparks MD). Typically, these plates were
counted after 14 to 17 days of incubation at 37°C. Growth of the M. tuberculosis
infection within BMMO was further monitored by lysing cultures at 4, 7, and 10
to 11 days after culture initiation and plating lysates on Middlebrook 7H11 plates
with 10% OADC.
The activity of the test vaccines was evaluated by coculturing splenocytes from
immunized mice with the M. tuberculosis-infected BMM?. For the live vaccines,
the in vitro studies were initiated 6 weeks after the immunization. For the other
vaccine preparations, the in vitro assays were begun 1 month following the final
immunization. To harvest the splenocytes, the spleens were aseptically removed
from three immunized and naïve C57BL/6 mice and disrupted to prepare a
single-cell suspension. The erythrocytes were then lysed with ACK buffer for 4
min and the remaining spleen cells were washed with cold DMEM. To remove
adherent splenic macrophages, spleen cells were added to culture flasks for 2 h
at 37°C and nonadherent cells were recovered by gentle pipetting. Finally, 5 ?
106of the nonadherent splenocytes were overlaid on 107M. tuberculosis-infected
macrophages and incubated at 37°C with 5% CO2. At the specified time, the
adherent cells were lysed with 0.1% saponin and diluted cell lysates were plated
on Middlebrook 7H11–10% OADC plates for enumeration of mycobacterial
CFU as described above. To test whether macrophage lysis contributed to the
reduction in mycobacterial CFU during the course of the coculture assay, culture
supernatants from selected samples at each time point were also plated on
Middlebrook 7H11 plates. Since the number of mycobacteria detected in the
supernatants at every time point from all vaccines and controls was at least
100-fold less than the bacteria present in adherent cells, intracellular killing (and
not macrophage lysis) is likely the primary mechanism of bacterial reduction in
the coculture system.
Evaluation of vaccine-induced protective immunity in a mouse model of pul-
monary tuberculosis. Six weeks after vaccination with the live attenuated vac-
cines and 4 weeks following the final vaccinations with subunit, viral-vectored,
and DNA vaccines, the mice were aerogenically challenged with M. tuberculosis
Erdman suspended in PBS with 0.04% Tween 80 at a concentration known to
deliver about 200 CFU in the lungs over a 30-min exposure in a Middlebrook
chamber (Glas Col, Terre Haute, IN). To determine the infection dose and the
postinfection bacterial burden, mice were sacrificed at 4 h and 28 days postchal-
lenge and then the lungs and spleens were removed aseptically and homogenized
separately in PBS–0.04% Tween 80 using a Seward Stomacher 80 blender (Tek-
mar, Cincinnati OH). After serial dilutions in PBS-Tween 80, the lung and spleen
homogenates were plated on Middlebrook 7H11 plates containing 10% OADC,
10 mg/ml ampicillin, 50 mg/ml cycloheximide, and 2 mg/ml 2-thiophenecarboxylic
acid hydride (Sigma). The addition of 2-thiophenecarboxylic acid hydride to the
growth medium inhibits BCG growth but not the growth of M. tuberculosis.
Again, the plates were incubated for 14 to 17 days at 37°C before counting
Evaluation of cytokine responses induced in the coculture assay. At the
specified time period, nonadherent splenic cells were recovered from super-
natants of the culture wells and stored in RNAlater (Qiagen, Valencia CA).
Total RNA was isolated from these cellular suspensions using the RNAeasy
minikit protocol (Qiagen). Equivalent amounts of RNA from these samples
were reverse transcribed to cDNA using the SuperScript first-strand synthesis
kit (Invitrogen, San Diego CA). The effectiveness of the DNA synthesis for
individual samples was assessed by analyzing the PCR products generated
with primers for the glyceraldehyde 3-phosphate dehydrogenase (GAPDH)
housekeeping gene. To quantitate the cytokine transcriptional responses in
cells recovered from the in vitro system, the cDNA was evaluated using RT2
profile cytokine PCR arrays (SAB Biosciences, Frederick, MD) and an ABI
Prism 7000 sequence detection system (Applied Biosystems, Foster City,
CA). For the cytokine PCR assay, the expression of 84 cytokine-like genes
was evaluated (21). The mRNA expression levels for each gene were then
normalized to the expression of the GAPDH gene using the following equa-
tion: relative mRNA expression ? 2–(Ct of cytokine – Ct of GAPDH), where Ct is the
threshold cycle. To determine whether the relative levels of gene expression were
significantly different than the expression levels in naïve mice, the PCR array results
were compared using the Wilcoxon matched pairs test (GraphPad Prism software,
test (GraphPad Prism, version 4). Finally, the relative gene expression values in
1026 PARRA ET AL.CLIN. VACCINE IMMUNOL.
immune cell cultures were determined by dividing the gene expression levels in
represents the mean increase (or decrease) of RNA expression relative to the naïve
controls for 12 BCG vaccine experiments and three to five studies with the other
To assess cytokine protein concentrations by enzyme-linked immunosorbent
assay (ELISA), culture supernatants were centrifuged to remove nonadherent
cells. The levels of gamma interferon (IFN-?), tumor necrosis factor alpha
(TNF-?), IL-27, and IL-10 were then evaluated using cytokine ELISA kits as
described by the manufacturers. The levels of IFN-?, TNF-?, and IL-10 were
measured using BD OptiEIA kits (Becton Dickinson), while IL-27 Quantikine
ELISA kits were purchased from R&D Systems (Minneapolis, MN).
Statistics. The data from these experiments were analyzed using the Graph-
Pad Prism 4 program. The in vivo and in vitro protection results were evaluated
by t tests and the Spearman correlation test (GraphPad Prism, version 4). The
cytokine expression data were evaluated using t tests and the nonparametric
Wilcoxon matched pairs test.
Evaluation of five different vaccines against M. tuberculosis.
To investigate whether vaccine-induced antituberculosis im-
munity could be detected in vitro for the various TB vaccine
preparations, mice were immunized with either BCG vaccine,
an M. tuberculosis attenuated mutant with a deletion in the
secA2 gene, a TB DNA vaccine, a TB fusion protein, or a
viral-vectored vaccine expressing four TB antigens (Table 1).
These preparations included five types of TB vaccines which
express different antigens and have distinct immunogenic ca-
pacities. The BCG vaccine preparation is derived from a BCG
Pasteur stock culture. The ?secA2 strain is a highly protective
proapoptotic deletion mutant of M. tuberculosis (17). The SD1
DNA vaccine construct expresses an ESAT6-antigen 85B fu-
sion protein, which has been shown to boost BCG-induced
immune responses and to protect against primary M. tubercu-
losis infections (12). The ESAT6-antigen 85B protein is a tu-
berculosis fusion antigen which induces substantial protective
immunity when formulated in DDA/MPL adjuvant. Finally,
the MVA-4 TB vaccine is a modified vaccinia Ankara construct
that expresses four TB antigens, antigen 85A, antigen 85B,
ESAT6, and Hsp 65 (27). As controls, mice were injected with
either heat-killed BCG, the DNA vaccine vector, the MVA
vector, or the DDA/MPL adjuvant.
Characterization of in vitro mycobacterial growth inhibition
for evaluating vaccines against tuberculosis. To assess whether
vaccine activity could be evaluated using in vitro assays, the
growth of M. tuberculosis in bone marrow macrophages that
were cocultured with immune splenocytes was monitored over
an 11-day period. Representative M. tuberculosis growth curves
for a coculture assay which tested two vaccines (BCG and the
SD1 protein-adjuvant formulation) are shown in Fig. 1. When
M. tuberculosis-infected macrophages were incubated without
splenocytes, logarithmic increases in the numbers of tubercle
bacilli were observed. Interestingly, significant decreases in the
in vitro mycobacterial growth were detected when naïve spleen
cells were cultured with macrophages infected with M. tuber-
culosis. Typically, a 0.5-log10reduction in the proliferation of
M. tuberculosis organisms (relative to the infected macro-
phages alone) was seen in naïve spleen cell cultures, presum-
ably due to innate antimycobacterial mechanisms (29). How-
ever, greater decreases in the in vitro growth of M. tuberculosis
were detected when splenocytes from mice immunized with the
ESAT6-antigen 85B fusion protein-adjuvant formulation or
BCG vaccine were cultured with M. tuberculosis-infected mac-
rophages. For the experiment shown in Fig. 1, relative de-
creases in bacterial numbers of 0.8 and 0.9 log10(compared to
naïve controls) were seen for the cultures containing the TB
fusion protein and BCG immune splenocytes, respectively, at
day 7 of the assay. In contrast, splenocytes from mice injected
with an inactive control, the vector DNA, did not control the
intramacrophage tubercular growth better than cells from
To evaluate the reproducibility of this assay, four experi-
ments were run using naïve and BCG immune splenocytes
from four different groups of mice. In these studies, the tem-
poral growth profiles and the mean protective responses for
the BCG immune cells were consistent. The mean BCG-in-
duced protection seen at day 7 (i.e., the difference between the
naïve and experimental log10CFU) at day 7 of the coculture
assays was 0.82 ? 0.1 (mean ? standard error of the mean
[SEM]). Furthermore, similar mycobacterial growth patterns
were seen when M. tuberculosis clinical isolates were substi-
tuted for the Erdman laboratory strain. For example, when
naïve and BCG-immune splenocytes were cocultured with
macrophages infected with the CDC1551 clinical isolate, the
mycobacterial growth profile was essentially equivalent to the
growth pattern seen in the M. tuberculosis Erdman infections
(data not shown).
To compare the in vitro mycobacterial growth inhibition
responses induced by immunization with the different vaccine
FIG. 1. Inhibition of intramacrophage growth of M. tuberculosis by
splenocytes recovered from immunized mice. Murine bone marrow
macrophages that were infected with M. tuberculosis were cocultured
with splenocytes taken from mice immunized with BCG or the ESAT6-
antigen 85B protein suspended in DDA-MPL adjuvant and the growth
of M. tuberculosis was monitored over an 11-day culture period. As
controls, M. tuberculosis-infected bone marrow macrophages were cul-
tured with splenocytes from mice injected with vector DNA, naïve
splenocytes, or in the absence of spleen cells (I-BMMO).
TABLE 1. Tuberculosis vaccine preparations
BCG........................................M. bovis BCG Pasteur
?secA2 mutant.......................M. tuberculosis proapoptotic attenuated
strain with ?secA2 gene deletion
SD1 DNA vaccine.................DNA vaccine expressing ESAT6 antigen
85B fusion protein
E6–85B protein .....................ESAT6 antigen 85B fusion protein
formulated in DDA-MPL adjuvant
MVA-4 TB.............................Modified vaccinia virus Ankara strain
expressing four TB antigens, ESAT6,
antigen 85A, antigen 85B, and HSP65
VOL. 16, 2009IN VITRO ASSAY FOR EVALUATING TB VACCINES1027
preparations, at least three coculture assays were completed
using splenocytes removed from mice vaccinated with each of
the vaccines and the controls. Table 2 shows the mean in vitro
growth inhibition responses for these vaccine preparations af-
ter 7 days of culture. A 7-day period was chosen because in
most experiments the maximal differences in intramacrophage
growth between naïve and immune cell cultures were usually
seen at this time. Statistical analysis of these data indicated
that immunization with each of the TB vaccines induced sig-
nificantly elevated in vitro antituberculosis activity compared
to the naïve controls (P ? 0.05). Moreover, the in vitro growth-
inhibitory responses evoked by each vaccine were significantly
increased relative to the corresponding control (e.g., protein-
adjuvant versus adjuvant alone). Overall, the inhibitory re-
sponses elicited by the vaccine preparations could be separated
into two groups: high and moderate in vitro activity. The highly
active vaccines (BCG, ?secA2, and the ESAT6-antigen 85B
fusion protein-adjuvant formulation) induced substantial
0.87 to 0.93 log10CFU in vitro growth-inhibitory responses,
while the moderately effective preparations, heat-killed
BCG, MVA-4 TB, and the SD1 DNA vaccine, evoked less
antituberculosis immunity (0.40 to 0.57 log10CFU). In fact, the
in vitro activities detected in cocultures of BCG and ?secA2
immune splenocytes were significantly greater (P ? 0.05) than
that induced in the moderately active group. Importantly, in-
jection of the vector and adjuvant controls did not induce
antituberculosis immunity that exceeded the growth-inhibitory
responses seen in naïve controls.
Comparison of in vivo protection with in vitro activity. The
in vivo activities of these vaccine preparations were measured
in standard vaccination/challenge experiments as described
previously (11). Mice were vaccinated as described in Materials
and Methods, challenged by the aerosol route with 200 CFU of
M. tuberculosis Erdman, and sacrificed 28 days later to deter-
mine relative organ bacterial burdens. The in vivo activity was
calculated by determining the mean protective responses
(naïve control organ CFU – vaccinated organ CFU) for two to
four experiments (Table 2). Again, the vaccine-induced pro-
tection could be separated into the same two groups. The
highly active vaccines, BCG, ?secA2 mutant, and the E6-85B
fusion protein, induced substantial antituberculosis protective
immunity. For these vaccines, greater than a 1-log10reduction
in mycobacterial burden in the lung, relative to naïve controls,
was seen at 28 days postchallenge. Although the moderately
active vaccine preparations (heat-killed BCG, TB MVA, and
the SD1 DNA vaccine) evoked modest protection in the lungs
(0.57 to 0.79 log10CFU compared to naïve controls) at the
4-week postchallenge time point, these protective responses
were significantly greater than the immune responses detected
in naïve animals. Similar to the in vitro studies, injection of the
vector and adjuvant controls did not evoke elevated in vivo
antituberculosis responses relative to naïve mice. Finally, we
compared the in vivo and in vitro data to assess the relevance
of the vaccine activity seen in the coculture assay. Importantly,
Spearman analysis showed that the correlation between in
vitro vaccine-mediated activity and in vivo vaccine-induced
protection in the lungs and spleens was highly significant (P ?
0.001) for these five different vaccines and the control prepa-
Identification of immune biomarkers associated with vac-
cine-induced protective responses. To determine whether im-
mune biomolecules were differentially regulated in cocultured
naïve and BCG immune splenocytes, RNA expression of 84
cytokines in nonadherent cells was assessed using PCR arrays.
The extent of expression was determined by normalizing the
real-time PCR values to the expression of the GAPDH house-
keeping gene and then by comparing the GAPDH-adjusted
results to the level of expression in naïve controls. In these
studies, vaccine-induced differential regulation was defined as
significantly different levels of expression in naïve and immune
cell cultures by the Wilcoxon matched pairs test and expression
levels at least twofold higher (or lower) than the naïve controls.
Table 3 shows the cytokine genes that were differentially reg-
ulated at days 5 and 7 of the coculture assay using BCG-
immune cells. Cytokine expression was evaluated at days 5 and
7 of the coculture because significant inhibition of mycobacte-
rial growth was detected on these days. At day 5, the expression
levels of 5 of 84 cytokine-related genes were consistently up-
regulated and 2 were downregulated in the in vitro cultures. At
day 7, the expression levels of nine cytokine-related genes were
differentially regulated in BCG-immune cultures. At both time
TABLE 2. Comparison of in vitro antituberculosis activity with in
vivo protection resultsa
SD1 DNA vaccine
0.87 ? 0.05*
0.93 ? 0.03*
0.91 ? 0.18*
0.57 ? 0.03*
0.47 ? 0.07*
0.40 ? 0.13*
0.07 ? 0.02
0.02 ? 0.01
0.05 ? 0.01
1.04 ? 0.12*
1.19 ? 0.13*
1.20 ? 0.21*
0.57 ? 0.22*
0.79 ? 0.05*
0.70 ? 0.15*
0.20 ? 0.08
0.18 ? 0.07
0.10 ? 0.05
1.06 ? 0.11*
1.15 ? 0.23*
0.85 ? 0.10*
0.38 ? 0.10*
0.67 ? 0.22*
0.58 ? 0.04*
0.11 ? 0.05
0.08 ? 0.05
0.10 ? 0.06
a*, significantly different than naïve controls (P ? 0.05).
bIn vitro activity is the difference between the naïve and experimental CFU
(log10) at day 7 of the in vitro coculture assay (mean ? SEM).
cProtection is the difference between the naïve and experimental CFU (log10)
in mouse lungs (or spleens) at 28 days after an aerosol challenge with M.
tuberculosis (mean ? SEM).
TABLE 3. Normalized cytokine mRNA expression at days 5 and 7
of coculture with BCG-immune splenocytes
Mean (?SEM) normalized expressionaon:
Day 5Day 7
3.43 ? 0.76*
17.11 ? 3.65*18.61 ? 4.81*
3.10 ? 0.65*
2.05 ? 0.33*
7.34 ? 1.61*
3.68 ? 0.72*
3.08 ? 0.41*
0.20 ? 0.05#
0.22 ? 0.05#
0.39 ? 0.08#
7.42 ? 2.02*
4.72 ? 1.01*
2.98 ? 0.42*
0.41 ? 0.12#
0.30 ? 0.05#
aMean normalized expression values represent mean fold differences com-
pared to expression by naïve controls, which were assigned a value of 1. *,
upregulated cytokine expression in BCG-immune cell cultures or cytokine ex-
pression in cultures using naïve splenocytes; #, downregulated cytokine expres-
sion in BCG-immune cell cultures relative to expression in naïve control cultures.
1028PARRA ET AL.CLIN. VACCINE IMMUNOL.
points, the expression levels of two cytokines known to be
critical for conferring antituberculosis activity, IFN-? and
TNF-?, were significantly upregulated. Furthermore, the ex-
pression levels of two other genes (IL-21 and IL-27) were
upregulated and two were consistently downregulated (Bmp1
and IL-1) at 5 and 7 days after the initiation of a coculture of
TB-infected macrophages with BCG-immune splenocytes. Im-
portantly, the levels of mRNA for many cytokines were not
differentially regulated during these experiments. These cyto-
kines included IFN-?, IL-2, IL-3, Il-4, Il-7, Il-12, and IL-15 (see
Table S1 in the supplemental material).
To verify that the gene regulation values reflected cytokine
protein expression levels for cytokines in which ELISA test kits
were available, specific cytokine concentrations in superna-
tants from cocultures containing either BCG-immune or naïve
splenocytes were assessed. For three cytokines which had ele-
vated gene expression levels, IFN-?, TNF-?, and IL-27, cyto-
kine ELISA responses were also significantly elevated in su-
pernatants from BCG-immune cocultures compared to the
naïve controls (Fig. 2). In contrast, the protein levels of IL-10,
a molecule whose expression is not consistently differentially
regulated in BCG-immune cultures, were increased less than
1.5-fold in the cocultures of BCG-immune cells compared to
naïve controls (data not shown).
For the validation of immune biomarkers against M. tuber-
culosis, it is important to define correlates of antituberculosis
protective immunity which are applicable to all new TB vac-
cines as well as the BCG vaccine. To identify general antitu-
berculosis biomarkers, cytokine gene expression analysis was
extended at day 7 of cocultures to splenocytes recovered from
mice immunized with the different TB vaccines. As shown in
Table 4, the expression levels of five cytokine genes (GDF15,
IFN-?, IL-21, IL-27, and TNF-?) were often upregulated
and three were frequently downregulated (Bmp1, IL-1, and
Tnfsf14) in nonadherent cells recovered from coculture assays.
It should be emphasized that elevated levels for IFN-?, IL-21,
and TNF-? mRNA and decreased levels of IL-1 mRNA were
seen in cocultures of all vaccines tested. Among the controls,
only upregulation of IL-21 or downregulation of BMP1 was
detected in the in vitro assays using splenocytes from mice
injected with the DNA vector or the adjuvant, respectively. It
is of interest that the only consistent differences in the cytokine
profiles between the highly active vaccines (BCG, ?secA2 mu-
tant, and ESAT6-antigen 85B fusion protein) and the moder-
ately active preparations (heat-killed BCG and the SD1 DNA
vaccine) were the higher IFN-? levels (11.2, 17. 1, and 23.7
versus 3.8 and 7.1, respectively) detected in cocultures using
spleen cells from mice immunized with the highly active vac-
cines. In addition, as seen in earlier in vivo studies, the cytokine
responses in the cocultures containing BCG and ?secA2 im-
mune cells were nearly identical (21). No substantive differ-
ences in the levels of cytokine regulation were detected in
cocultures containing BCG- or ?secA2-immune cells. Surpris-
ingly, despite the significantly higher levels of growth inhibition
FIG. 2. Protein expression levels of IFN-?, TNF-?, and IL-27 were higher in BCG coculture supernatants than controls. Bone marrow
macrophages infected with M. tuberculosis were cocultured with BCG-immune splenocytes (BCG), naïve splenocytes, or in the absence of spleen
cells (BMM). At day 7, the cytokine protein levels in culture supernatants were determined using cytokine ELISA protocols.*, P ? 0.05.
TABLE 4. Normalized cytokine mRNA expression at day 7 of coculture for candidate TB vaccines
Normalized mRNA expression
?secA2 E6-85B kBCGa
3.4 ? 0.8
17.1 ? 3.6
7.3 ? 1.6
3.7 ? 0.7
3.1 ? 0.4
0.20 ? 0.1
0.22 ? 0.1
0.39 ? 0.1
3.1 ? 0.3
23.7 ? 5.4
4.8 ? 0.9
2.4 ? 0.6
7.1 ? 3.0
0.5 ? 0.4
0.3 ? 0.1
0.3 ? 0.2
4.9 ? 1.9
7.1 ? 1.1
2.3 ? 1.1
4.8 ? 0.9
2.0 ? 0.4
0.4 ? 0.1
0.40 ? 0.1
2.0 ? 0.3
3.8 ? 1.0
5.7 ? 0.3
11.2 ? 1.5
14.6 ? 1.12.8 ? 0.9
6.4 ? 2.5
0.2 ? 0.1
0.2 ? 0.04
0.25 ? 0.1
2.2 ? 0.7
0.2 ? 0.01
0.5 ? 0.2
akBCG, heat-killed BCG.
bDNAVC, vector DNA.
VOL. 16, 2009 IN VITRO ASSAY FOR EVALUATING TB VACCINES1029
activity detected in BCG-immune cultures relative to heat-
killed BCG controls, only modest differences were seen when
the cytokine profiles of cultures using BCG-immune or heat-
killed BCG-immune splenocytes were compared. Only ele-
vated IFN-? and IL-21 expression levels (2.4- and 3.2-fold
increases, respectively) were detected in the BCG-immune cell
cultures in comparison to cultures containing splenocytes re-
covered from mice injected with heat-killed BCG.
To facilitate the characterization of vaccines against tuber-
culosis, we have developed a murine in vitro functional assay
for evaluating vaccine-induced antituberculosis activity. We
reasoned that inhibition of intramacrophage growth of M. tu-
berculosis was a direct and relevant end point for assessing the
potency of TB vaccines. With this functional assay, multiple
protective mechanisms likely contribute to the in vitro reduc-
tion in mycobacterial proliferation, including both innate and
adaptive immune responses. Importantly, we demonstrated
that the vaccine-induced in vitro growth inhibition activity de-
tected in our coculture assay significantly correlates with the
protective immunity seen postvaccination in our mouse model
of pulmonary tuberculosis and is relevant for estimating the in
vivo potency of TB vaccines. In addition, we have shown that
the protective responses evoked by different types of vaccines
can be assessed using this assay. In the human in vitro func-
tional studies reported thus far, only BCG-induced activity has
been evaluated (4, 18, 35). Our results suggest that the antitu-
berculosis protective immunity evoked by subunit, viral vector,
or DNA vaccines should be detectable in similar assays de-
signed to measure vaccine-induced immune responses in hu-
man clinical trials. An unexpected finding from our in vivo and
in vitro protection studies was the moderate antituberculosis
activity detected with the heat-killed BCG vaccine preparation.
Although this result was surprising, it is consistent with earlier
data reported by Opie and Freund, who showed that injection
of heat-inactivated tubercle bacilli was nearly as effective as
live BCG immunization in protecting rabbits against infection
with M. tuberculosis (23).
A major goal of tuberculosis vaccine research during the
past 2 decades has been to identify the correlates of protective
immunity against M. tuberculosis. The identification of protec-
tive correlates of immunity would clearly facilitate the evalua-
tion and characterization of new TB vaccines both in preclin-
ical studies and clinical trials. In vaccine studies, molecules that
are upregulated (or downregulated) after the immunization
are candidates as protective correlates. In our studies, using
the in vitro coculture assay, cytokine profiles associated with
antituberculosis protective activity were identified. Impor-
tantly, the patterns of cytokines that are up or downregulated
after immunization with active TB vaccines were similar. For
most of the vaccines, GDF15, IFN-?, IL-21, IL-27, and TNF-?
expression levels were upregulated while BMP1, IL-1 and
Tnfsf14 expression levels were downregulated (relative to na-
ives) in cocultures of infected BMM? and immune spleno-
cytes. In contrast, the expression of these same cytokines was
usually not differentially regulated for in vitro assays using
control splenocytes. This result suggests that different types of
TB vaccines evoke similar patterns of protective immune me-
diators in a mouse model of tuberculosis. Interestingly, the
BCG-induced cytokine patterns detected with the in vitro assay
resembled the cytokine responses seen in the lungs of BCG-
vaccinated mice at 10 days after an aerogenic challenge with M.
tuberculosis (21). In both the in vitro and in vivo experiments,
IFN-?, IL-21, IL-27, and TNF-? were upregulated (after ex-
posure to a M. tuberculosis infection) in the cells of animals
immunized with BCG. The IFN-? and TNF-? results are not
surprising since these cytokines have been shown to be critical
components of protective immunity against M. tuberculosis (14,
15, 22, 25). However, the roles of IL-21 and IL-27 during a
mycobacterial infection are less certain. IL-21 is a type I cyto-
kine which is produced largely by antigen-activated T cells.
Although its major functions are to activate CD8 T cells and
NK cells and to stimulate B-cell immunoglobulin production,
IL-21 can also suppress dendritic cell activity (2, 20). Based on
this immunosuppressive activity of IL-21, it has been suggested
that this cytokine plays a pivotal role in the regulation of
pathogen-induced immune responses. It has recently been
shown that the combination of IL-21 and TGF-? induces
proinflammatory Th17 cells (36). It would be of interest to
determine whether TH17 cell-promoting activity is also de-
tected when IL-21 is combined with GDF15, another TGF-?
family member that was shown to be upregulated in this study.
Similar to IL-21, IL-27 may have an important immune regu-
latory capacity since it has both proinflammatory and antiin-
flammatory properties. While IL-27 has been shown to pro-
mote inflammation, Th1 responses, and IFN-? production, it
can also inhibit inflammatory responses (19, 33). Surprisingly,
animals deficient in the IL-27 receptor were able to limit M.
tuberculosis infections more effectively than controls and neu-
tralization of IL-27 in vitro led to enhanced antituberculosis
activity in human monocytes (26, 28). Clearly, further studies
will be needed to define the precise roles of IL-21 and IL-27 in
mediating protective immunity against M. tuberculosis. Besides
allowing the identification of contrasting immune responses to
different vaccines, these data permit comparisons of the cyto-
kine responses between highly active and less active vaccine
formulations in the mouse model. It is surprising that the only
consistent difference we observed in the cytokine profiles for
the highly active and moderately effective vaccines was the
increased IFN-? levels seen for the most potent vaccines. We
are currently investigating whether the elevated IFN-? levels
are associated with greater protection because of the increased
expression of IFN-?-inducible genes such as the CXCL9 and
In addition to identifying cytokine patterns that correlate
with vaccine-induced protection, this murine in vitro system
should be useful for defining cell subsets and cellular immune
mechanisms which contribute to antituberculosis protective
immunity. Using a similar murine coculture assay, Cowley and
Elkins have demonstrated that double-negative T cells, mem-
brane-bound TNF, and IFN-?-independent processes partially
mediate the anti-Francisella protective immunity induced by
the live vaccine strain of Francisella tularensis (7–10). In pre-
liminary experiments, we found that purified splenic T cells
from BCG-vaccinated mice inhibited mycobacterial growth in
vitro and that the depletion of CD4 cells but not CD8 cells
largely abrogated the protective effect of BCG vaccine (K.
Kolibab and S. Derrick, unpublished results). These data are
1030PARRA ET AL.CLIN. VACCINE IMMUNOL.
consistent with earlier in vivo and in vitro results which indi-
cated that the BCG vaccine induces a strong CD4 antituber-
culosis protective response (24, 31). It should be emphasized
that cells from other relevant organs can be utilized in this
assay. For example, we have shown that cells from the lymph
nodes of BCG-vaccinated mice also inhibit the intramacro-
phage growth of M. tuberculosis in the coculture assay (M.
Parra, unpublished results). Future experiments with the co-
culture assay will further examine the relative importance of
T-cell subsets from the different relevant organs in mediating
the protective immunity induced by the various types of TB
In summary, we have described the development of a murine
in vitro coculture assay to characterize the protective activity
induced by TB vaccines. We established the relevance of the
assay by showing that in vivo and in vitro vaccine-induced
protection results and cytokine patterns were similar. Given
this correlation between in vivo and in vitro activity, we antic-
ipate that the coculture assay will be useful for screening and
comparing new TB vaccine preparations and for elucidating
antituberculosis protective immune mechanisms. Moreover, as
TB vaccines progress through clinical trials, this assay could be
adapted to evaluate manufacturing consistency and vaccine
stability and to potentially bridge preclinical data to human
clinical trial results.
This project has been funded in part with Federal funds from the
National Institute of Allergy and Infectious Diseases, National Insti-
tutes of Health, Department of Health and Human Services, under
1. Bosio, C. M., and K. L. Elkins. 2001. Susceptibility to secondary Francisella
tularensis live vaccine strain infection in B-cell-deficient mice is associated
with neutrophilia but not defects in specific T-cell-mediated immunity. In-
fect. Immun. 69:194–203.
2. Brandt, K., P. B. Singh, S. Bulfone-Paus, and R. Ruckert. 2007. Interleukin-
21: a new modulator of immunity, infection, and cancer. Cytokine Growth
Factor Rev. 18:223–232.
3. Braunstein, M., B. J. Espinosa, J. Chan, J. T. Belisle, and W. R. Jacobs, Jr.
2003. SecA2 functions in the secretion of superoxide dismutase A and in the
virulence of Mycobacterium tuberculosis. Mol. Microbiol. 48:453–464.
4. Cheon, S. H., B. Kampmann, A. G. Hise, M. Phillips, H. Y. Song, K. Landen,
Q. Li, R. Larkin, J. J. Ellner, R. F. Silver, D. F. Hoft, and R. S. Wallis. 2002.
Bactericidal activity in whole blood as a potential surrogate marker of im-
munity after vaccination against tuberculosis. Clin. Diagn. Lab. Immunol.
5. Colditz, G. A., T. F. Brewer, C. S. Berkey, M. E. Wilson, E. Burdick, H. V.
Fineberg, and F. Mosteller. 1994. Efficacy of BCG vaccine in the prevention
of tuberculosis. Meta-analysis of the published literature. JAMA 271:698–
6. Cowley, S. C., and K. L. Elkins. 2003. CD4?T cells mediate IFN-gamma-
independent control of Mycobacterium tuberculosis infection both in vitro
and in vivo. J. Immunol. 171:4689–4699.
7. Cowley, S. C., and K. L. Elkins. 2003. Multiple T cell subsets control Fran-
cisella tularensis LVS intracellular growth without stimulation through mac-
rophage interferon gamma receptors. J. Exp. Med. 198:379–389.
8. Cowley, S. C., M. F. Goldberg, J. A. Ho, and K. L. Elkins. 2008. The
membrane form of tumor necrosis factor is sufficient to mediate partial
innate immunity to Francisella tularensis live vaccine strain. J. Infect. Dis.
9. Cowley, S. C., E. Hamilton, J. A. Frelinger, J. Su, J. Forman, and K. L.
Elkins. 2005. CD4-CD8-T cells control intracellular bacterial infections both
in vitro and in vivo. J. Exp. Med. 202:309–319.
10. Cowley, S. C., J. D. Sedgwick, and K. L. Elkins. 2007. Differential require-
ments by CD4?and CD8?T cells for soluble and membrane TNF in control
of Francisella tularensis live vaccine strain intramacrophage growth. J. Im-
11. Delogu, G., A. Li, C. Repique, F. Collins, and S. L. Morris. 2002. DNA
vaccine combinations expressing either tissue plasminogen activator signal
sequence fusion proteins or ubiquitin-conjugated antigens induce sustained
protective immunity in a mouse model of pulmonary tuberculosis. Infect.
12. Derrick, S. C., A. L. Yang, and S. L. Morris. 2004. A polyvalent DNA vaccine
expressing an ESAT6-Ag85B fusion protein protects mice against a primary
infection with Mycobacterium tuberculosis and boosts BCG-induced protec-
tive immunity. Vaccine 23:780–788.
13. Fine, P. E. 2001. BCG: the challenge continues. Scand. J. Infect. Dis. 33:
14. Flynn, J. L. 2004. Immunology of tuberculosis and implications in vaccine
development. Tuberculosis (Edinburgh) 84:93–101.
15. Flynn, J. L., J. Chan, K. J. Triebold, D. K. Dalton, T. A. Stewart, and B. R.
Bloom. 1993. An essential role for interferon gamma in resistance to Myco-
bacterium tuberculosis infection. J. Exp. Med. 178:2249–2254.
16. Hanekom, W. A., H. M. Dockrell, T. H. Ottenhoff, T. M. Doherty, H.
Fletcher, H. McShane, F. F. Weichold, D. F. Hoft, S. K. Parida, and U. J.
Fruth. 2008. Immunological outcomes of new tuberculosis vaccine trials: W.
H. O. panel recommendations. PLoS Med. 5:e145.
17. Hinchey, J., S. Lee, B. Y. Jeon, R. J. Basaraba, M. M. Venkataswamy, B.
Chen, J. Chan, M. Braunstein, I. M. Orme, S. C. Derrick, S. L. Morris, W. R.
Jacobs, Jr., and S. A. Porcelli. 2007. Enhanced priming of adaptive immunity
by a proapoptotic mutant of Mycobacterium tuberculosis. J. Clin. Investig.
18. Hoft, D. F., S. Worku, B. Kampmann, C. C. Whalen, J. J. Ellner, C. S.
Hirsch, R. B. Brown, R. Larkin, Q. Li, H. Yun, and R. F. Silver. 2002.
Investigation of the relationships between immune-mediated inhibition of
mycobacterial growth and other potential surrogate markers of protective
Mycobacterium tuberculosis immunity. J. Infect. Dis. 186:1448–1457.
19. Hunter, C. A. 2005. New IL-12-family members: IL-23 and IL-27, cytokines
with divergent functions. Nat. Rev. Immunol. 5:521–531.
20. Leonard, W. J., and R. Spolski. 2005. Interleukin-21: a modulator of lym-
phoid proliferation, apoptosis and differentiation. Nat. Rev. Immunol.
21. Lim, J., S. C. Derrick, K. Kolibab, A. L. Yang, W. R. Jacobs, and S. L.
Morris. 2009. Post-infection characterization of early pulmonary cytokine
and chemokine responses in mice immunized with TB vaccines and chal-
lenged by the aerosol route with Mycobacterium tuberculosis. Clin. Vaccine
22. Lin, P. L., H. L. Plessner, N. N. Voitenok, and J. L. Flynn. 2007. Tumor
necrosis factor and tuberculosis. J. Investig. Dermatol. Symp. Proc. 12:22–25.
23. Opie, E. L., and J. Freund. 1937. An experimental study of protective
inoculation with heat killed tubercle bacilli. J. Exp. Med. 66:761–788.
24. Ordway, D., M. Henao-Tamayo, C. Shanley, E. E. Smith, G. Palanisamy, B.
Wang, R. J. Basaraba, and I. M. Orme. 2008. Influence of Mycobacterium
bovis BCG vaccination on cellular immune response of guinea pigs chal-
lenged with Mycobacterium tuberculosis. Clin. Vaccine Immunol. 15:1248–
25. Ottenhoff, T. H., D. Kumararatne, and J. L. Casanova. 1998. Novel human
immunodeficiencies reveal the essential role of type-I cytokines in immunity
to intracellular bacteria. Immunol. Today 19:491–494.
26. Pearl, J. E., S. A. Khader, A. Solache, L. Gilmartin, N. Ghilardi, F. deSauvage,
and A. M. Cooper. 2004. IL-27 signaling compromises control of bacterial
growth in mycobacteria-infected mice. J. Immunol. 173:7490–7496.
27. Perera, P. Y., S. C. Derrick, K. Kolibab, F. Momoi, M. Yamomato, S. L.
Morris, T. A. Waldmann, and L. P. Perera. 2009. A multi-valent vaccinia
virus-based tuberculosis vaccine molecularly adjuvanted with interleukin-15
induces robust immune responses in mice. Vaccine 27:2121–2127.
28. Robinson, C. M., and G. J. Nau. 2008. Interleukin-12 and interleukin-27
regulate macrophage control of Mycobacterium tuberculosis. J. Infect. Dis.
29. Sada-Ovalle, I., A. Chiba, A. Gonzales. M. B. Brenner, and S. M. Behar.
2008. Innate invariant NKT cells recognize Mycobacterium tuberculosis-in-
fected macrophages, produce interferon-gamma, and kill intracellular bac-
teria. PLOS Pathog. 4:e1000239.
30. Silver, R. F., Q. Li, W. H. Boom, and J. J. Ellner. 1998. Lymphocyte-
dependent inhibition of growth of virulent Mycobacterium tuberculosis
H37Rv within human monocytes: requirement for CD4?T cells in purified
protein derivative-positive, but not in purified protein derivative-negative
subjects. J. Immunol. 160:2408–2417.
31. Soares, A. P., T. J. Scriba, S. Joseph, R. Harbacheuski, R. A. Murray, S. J.
Gelderbloem, A. Hawkridge, G. D. Hussey, H. Maecker, G. Kaplan, and
W. A. Hanekom. 2008. Bacillus Calmette-Guerin vaccination of human new-
borns induces T cells with complex cytokine and phenotypic profiles. J. Im-
32. Sterne, J. A., L. C. Rodrigues, and I. N. Guedes. 1998. Does the efficacy of
BCG decline with time since vaccination? Int. J. Tuberc. Lung Dis. 2:200–
33. Stumhofer, J. S., A. Laurence, E. H. Wilson, E. Huang, C. M. Tato, L. M.
Johnson, A. V. Villarino, Q. Huang, A. Yoshimura, D. Sehy, C. J. Saris, J. J.
O’Shea, L. Hennighausen, M. Ernst, and C. A. Hunter. 2006. Interleukin 27
negatively regulates the development of interleukin 17-producing T helper
VOL. 16, 2009IN VITRO ASSAY FOR EVALUATING TB VACCINES1031
cells during chronic inflammation of the central nervous system. Nat. Im- Download full-text
34. World Health Organization. 2008. Global tuberculosis control: surveillance,
planning, financing WHO/HTM/TB/2008.393. WHO, Geneva, Switzerland.
35. Worku, S., and D. F. Hoft. 2000. In vitro measurement of protective myco-
bacterial immunity: antigen-specific expansion of T cells capable of inhibiting
intracellular growth of bacille Calmette-Guerin. Clin. Infect. Dis. 30(Suppl.
36. Yang, L., D. E. Anderson, C. Baecher-Allan, W. D. Hastings, E. Bettelli,
M. Oukka, V. K. Kuchroo, and D. A. Hafler. 2008. IL-21 and TGF-
beta are required for differentiation of human TH17 cells. Nature 454:
1032 PARRA ET AL.CLIN. VACCINE IMMUNOL.