Tumor antigen-dependent and tumor antigen-independent activation of antitumor activity in T cells by a bispecific antibody-modified tumor vaccine.

Abstract

New approaches of therapeutic cancer vaccination are needed to improve the antitumor activity of T cells from cancer patients. We studied over the last years the activation of human T cells for tumor attack. To this end, we combined the personalized therapeutic tumor vaccine ATV-NDV--which is obtained by isolation, short in vitro culture, irradiation, and infection of patient's tumor cells by Newcastle Disease Virus (NDV)--with bispecific antibodies (bsAbs) binding to this vaccine and introducing anti-CD3 (signal 1) and anti-CD28 (signal 2) antibody activities. This vaccine called ATV-NDV/bsAb showed the unique ability to reactivate a preexisting potentially anergized antitumor memory T cell repertoire. But it also activated naive T cells to have antitumor properties in vitro and in vivo. This innovative concept of direct activation of cancer patients' T cells via cognate and noncognate interactions provides potential for inducing strong antitumor activities aiming at overriding T cell anergy and tumor immune escape mechanisms.

Full text

Page 1

Hindawi Publishing Corporation
Clinicaland Developmental Immunology
Volume 2010, Article ID 423781, 12 pages
doi:10.1155/2010/423781
Review Article
Tumor Antigen-DependentandTumor Antigen-Independent
ActivationofAntitumorActivityin
TCells bya Bispecific Antibody-Modified TumorVaccine
PhilippeFournier1andVolkerSchirrmacher1,2,3
1German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
2Tumor Immunology Program, DKFZ, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
3Medical Centre for Immunology and Oncology (IOZK), 50674 Cologne, Germany
Correspondence should be addressed to Volker Schirrmacher, v.schirrmacher@dkfz.de
Received 1 July 2010; Accepted 14 December 2010
Academic Editor: Stuart Mannering
Copyright © 2010 P. Fournier and V. Schirrmacher. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
New approaches oftherapeutic cancervaccinationareneeded toimprovetheantitumoractivity of Tcells fromcancerpatients. We
studied over the last years the activation of human T cells for tumor attack. To this end, we combined the personalized therapeutic
tumor vaccine ATV-NDV—which is obtained by isolation, short in vitro culture, irradiation, and infection of patient’s tumor
cells by Newcastle Disease Virus (NDV)—with bispecific antibodies (bsAbs) binding to this vaccine and introducing anti-CD3
(signal1) andanti-CD28 (signal2) antibody activities. This vaccine called ATV-NDV/bsAb showed the unique abilityto reactivate
a preexisting potentially anergized antitumor memory T cell repertoire. But it also activated naive T cells to have antitumor
properties in vitro and in vivo. This innovative concept of direct activation of cancer patients’ T cells via cognate and noncognate
interactions provides potential for inducing strong antitumor activities aiming at overriding T cell anergy and tumor immune
escape mechanisms.
1.Introduction
For decades, treatment of cancer has focused primarily on
surgery, chemotherapy, and radiation. Despite significant
advances by the introduction of new chemotherapeutic
agentsandalsorecentlybytheclinicalintroductionofmono-
clonal antibodies, major limitations of such treatments keep
the inability to eliminate the last tumor cell. The offspring
of those tumor cells that were not destroyed by the first-line
treatment may have a selective advantage, leaving the patient
witharecurrenceofcancerthatisoftenwidespreadandresis-
tant to further chemotherapy or radiotherapy. Therefore,
more effective therapies are needed. Immunotherapy based
on antitumor immune memory is a new modality for cancer
treatment. It holds great promise for affecting in a positive
way cancer patients’ survival with minimal toxicity.
For over 200 years, active immunotherapy approaches
have been used to successfully prevent numerous infectious
diseases such as smallpox. These active immunotherapy
concepts are now being applied to develop therapeutic
cancer vaccines with the intention of treating existing
tumors and/or preventing tumor recurrence. In principle,
anticancer vaccination (e.g., with autologous tumor cells,
peptide vaccines, dendritic cells, idiotypic antibodies, and
virus-based vaccines) is a meaningful additional approach
for treatment of cancer [1].
Tumor antigens (TAs) of patients’ tumor cannot be
recognized directly by the patients’ T cells. They need first
to be processed and to be properly presented by specialized
cells that are known as professional antigen-presenting cells
(APCs) such as dendritic cells (DCs). TA-presenting DCs
then migrate to lymph nodes, where they induce immunity
in TA-specific naive T cells. This results in the differentiation
into effector T cells—mainly CD8+ cytolytic T cells (CTLs),
which are capable of destroying tumor cells expressing the
tumor antigen. The response also leads to the generation

Page 2

2 Clinical and Developmental Immunology
of TA-specific memory T cells which provide immune
protection against tumor recurrence.
Despite promising results of cancer vaccination obtained
in animal tumor models, results of published vaccine trials
reveal only a weak clinical response rate with less than 1%
for active specific immunization procedures in colorectal
cancer patients [2]. At the Surgery Branch of the National
Cancer Institute (Bethesda, Maryland, USA), an objective
response rate of only 2.6% was reported [3] with various
cancer vaccines, even though about 50% of the vaccinated
patients had developed CTL killer cells able to specifically
recognize and kill tumor cells in vitro.
Difficulties met by vaccination approaches to cancer
treatment have been attributed to tumor immune avoidance
mechanisms [4]. Tumors employ many escape strategies
in order to evade immune attack. These strategies include
downregulation of MHC molecules in order to hide from
immunerecognition [5],expression ofinhibitory factorsand
immunosuppressive cytokines [6–8], including TGF-β [9,
10], IL-10 [11], and recruitment of regulatory immune cells
CD4+CD25+FoxP3+ Tregs [12], Tr1 cells [13], tolerogenic
DCs, and myeloid suppressor cells, including immature
macrophages, granulocytes, DCs, and other myeloid cells
at earlier stages of differentiation [14, 15]. These immune
avoidance mechanisms employed by tumors render the
immune system tolerant. This may be responsible for tumor
immune evasion as many of the tolerance mechanisms that
prevent autoimmunity are the same as employed by tumors
to prevent immune destruction [16, 17].
In order to develop an effective immunotherapy strategy
for metastatic cancer, new approaches are required that not
only can create and enhance tumor-specific immunity but
canalso counteracttheabilityofthetumortoevadeimmune
destruction.To thisend,T cells ofthecancerpatientsneed to
beeducatedto attacktumorcells. NaiveCD8+T cellsrequire
two distinct signals for activation: signal 1 is provided by
engagement of the TCR with its cognate ligand, and signal
2 is provided by interaction of costimulatory receptors with
theirrespective ligands on theAPCs[18, 19].Memory CD8+
T cells, which have been primed to TA, are often anergic and
need to be properly reactivated in order to be able to destroy
the tumor cells.
The design of an efficient antitumor vaccine may be
influenced by an important paradigm shift in the field of
immunology regarding the regulation of immunity. A new
concept has emerged that proposes that the regulation of
immunity and tolerance is not only determined by the
specificity of immune T cells as previously thought but
also by the context in which the antigens are presented
to the immune system [20, 21]. The implications are that,
in the absence of appropriate inflammatory reactions, the
self- (tumor) antigens presented by APCs will not lead
to T cell activation. Since tumors can also produce anti-
inflammatory cytokines, they are capable of influencing the
immune response by preventing an inflammatory response.
Therefore, successful antitumor immunity will develop
only in situations where DCs are processing TAs in the
presence of an inflammatory microenvironment (“danger
signals”) which is potent enough to also downregulate
tumor-mediated immunosuppressive cytokine production.
The magnitude and duration of the immune response
will be dependent on the extent and quality of the local
inflammatory response and will be contained by a variety of
existing tolerogenic mechanisms.
Previous attempts at developing therapeutic cancer vac-
cines have demonstrated that it is possible to elicit specific
immunity against self-tumor antigens [2, 3]. Recent insights
on how immunity and tolerance are regulated indicate that
the failure ofthese vaccines in the clinicmay be related to the
absence of sufficient danger and T cell costimulation signals
at the time when tumor antigens are processed by DCs.
In this paper, we highlight some in vitro and in vivo
observations made during the evaluation of a tumor vaccine
that we developed in our laboratory. The tumor vaccine of
the second generation, modified with bsAb, will be shown
to be capable to reactivate memory T cells and to activate
nonspecifically naive T cells against the tumor.
2.The AutologousNDV-Based TumorVaccine
Over the last 10 years, we have developed and evaluated
an autologous tumor vaccine which is first modified by
virus infection and which later was modified further by
attachment of bispecific antibodies (see Figure 1). The aim
was to activate with such a vaccine potentially anergized
TA-specific memory T cells and to activate in addition
nonspecifically naive T cells to overcome tumor escape
variants that may lack TA expression. For virus infection,
we chose the avian paramyxovirus Newcastle Disease Virus
(NDV) [22]. NDV is one of five species of viruses that
are under clinical evaluation [23]. It is a negative strand
RNA virus with interesting antineoplastic and immune-
stimulating properties [23, 24]. Most remarkable is its
capacity to induce strong type I interferon responses by viral
protein [24] and RNA [25]. Detection of foreign RNA in
the cytoplasm by RIG-I induces an innate antiviral program
that initiates the transcription of RNA-responsive genes.
The responses involve a multimodal machinery of gene
regulation by the Interferon Regulatory Factor (IRF) family
of transcription factors [26] and link innate and adaptive
immunity [27]. There are 2 generations of NDV-based
tumor vaccine: the ATV-NDV and ATV-NDV/bsAb.
2.1. First-Generation Vaccine: ATV-NDV. The virus-mod-
ified tumour vaccine developed by us for human applica-
tion consists of virus-infected intact viable and irradiated
autologous tumor cells (see Figure 1). The tumor cell
infection by NDV is designed to provide the necessary
danger signals to elicit antitumor immunity. This strategy is
based on preclinical studies in metastatic animal tumours.
Antimetastatic effectswere observedafterlocal postoperative
vaccination with NDV-infected autologous tumour cells
[28]. The vaccination activated a tumour-line specific T cell-
mediated immune response, which also protected against a
second challenge with the same tumour line [29].
Tumor cell infection by NDV was found also in
humans to be an efficient and safe way to produce an
autologous tumor vaccine (first-generation ATV-NDV) with

Page 3

Clinical and Developmental Immunology3
vaccine of the 2nd generation has
been initiated in colorectal
cancer patients and the clinical
results are pending .
NDV
1st generation vaccine (ATV-NDV)
NDV :
universal anchor molecule
(iii) High safety profile for
humans
Prolongation of survival
among cancer patients for
various tumor entities [31-37]
2nd generation vaccine (ATV-NDV /bsAb)
vaccine for cross-linking of CD3 and
CD28 molecules at the surface of T
CD3 complex (
(iv) bsHN-CD28 binds to HN and to
1)
(signal
signal
2)
Advantages:
attachment to the vaccine
(v) Broad applicability (any
cancer cell which is
susceptible to NDV infection)
(vi) Controlled dosage and
delivery of T cell activating
signals
attachment of the bsAbs
(viii) Preclinical studies with human
cells in vitro and in
xenotransplanted mice in vivo
(ix) Clinical studies planned for the
with the ATV-NDV tumor
bsHN-CD3
Anti-HN
Anti-HNAnti-CD3
Anti-CD28
bsHN-CD28
HN
TA
cells
(iii) bsHN-CD3 binds to HN of NDV
and interacts with TCR-associated
the costimulatory molecule CD28
future: a clinical phase I/II trial
Bispecific Abs :
expressed at the surface of the vaccine
(i) Tumor selective replication
(ii) Introduction of viral HN as
(i) Studies in mouse models
[28, 29]
(ii) Preclinical studies
(iii) Clinical studies
(i) Production as recombinant proteins
(ii) Attachment of antibodies to the
(i) Easy production
(ii) Safety
(iii) Monovalency
(iv) BsAb active only upon
(vii) Easy protocol for the specific
[30]
Tumor cell
(107cells)
Figure 1: Principles of the NDV-based tumor vaccine of the first and second generation and status of the art (for more details, see the main
text).
pleiotropic immuno-stimulatory properties [30]. Promising
results based on the prolongation of survival among cancer
patients with various tumor entities have already been
reported from several clinical phase II trials, including breast
cancer [31, 32], colon carcinoma [33, 34], Head and Neck
SquamousCell Carcinoma (HNSCC)[35], and glioblastoma
[36]. The antitumor clinical efficacy has been shown also
in a randomized study performed among colon carcinoma
patients operated for liver metastasis [37].
2.2. Second-Generation Vaccine: ATV-NDV/bsAb. The use
of NDV during the generation of the ATV-NDV tumor
vaccine presents the advantage that, upon infection, the viral
hemagglutinin-neuraminidase (HN) protein is expressed on
the vaccine cells and can serve as universal anchor molecule
for the binding of new ligand proteins. The ATV-NDV can
then be combined easily with bispecific antibodies (bsAbs)
binding to the HN protein for introducing anti-CD3 and
anti-CD28 antibodies at the surface of the tumor vaccine
to obtain the NDV-based tumor vaccine of the second
generation (see Figure 1). This novel strategy is designed to
intensify T cellactivation via agonisticanti-CD3 and/oranti-
CD28 single-chain antibody reagents (scFv). When subop-
timal amounts of anti-CD3 are employed, the combination
of TA, anti-CD3, and anti-CD28 should help to intensify
TA-induced signal 1 and HN-induced costimulatory signal
2 [38, 39]. Because of virus infection, the ATV-NDV/bsAb
vaccine provides a highly inflammatory environment for T
cells in the presence of tumor cells. The presence of cell
surface bsAbs binding to CD3 and CD28 receptor molecules
serves among other effects for augmenting signal strength in
T cells to override anergic states of TA-specific T cells. Signal
intensity and duration (strength) of TCR stimulation has an
impact on setting the balance between adaptive responses
and immunopathology [40] and influences induction of T
cell activation or anergy [41]. Signal strength, timing, and
tuning are also important for T cell costimulation (signal
2). Combining optimal signals 1 and 2 at the surface of
the tumor vaccine is expected to generate strong antitumor
adults. We showed over the last years that this vaccine acts
on TA-specific memory T cells but also on naive T cells via
TA-dependent and TA-indepentent pathways.
3.Reactivation of TA-Specific Memory T Cells
fromCancerPatientsuponOptimal
Combination of bsHN-CD3 and bsHN-CD28
withthe VaccineATV-NDV
The ATV-NDV tumor vaccine of the second generation
was capable of reactivating anergic T cells from tumor-
draining lymph nodes of cancer patients. This was revealed

Page 4

4 Clinical and Developmental Immunology
via a modified short-term IFN-γ ELISpot assay which we
established for reactivation of cancer-reactive memory T
cells. As shown in Figure 2(A), from four different head and
neck squamous cell carcinoma (HNSCC) patients, primary
tumor cells were expanded in vitro, and, from each tumor
line, autologous tumor vaccines were prepared as a vaccine
(through irradiation, infection by NDV, and loading with
the bsAbs). These vaccines were finally combined with T
cells isolated from tumor-draining lymph nodes from the
corresponding HNSCC patients. The data from Figure 2(B)
show a strong IFN-γ response only in that group where
T cells were stimulated with autologous TAs together with
anti-CD3 and anti-CD28 signals. These responses required
HNSCC-derived TAs. Another tumor line (the heterologous
promonocytic U937 cell line) modified following the same
waycouldnotreactivateashort-termmemoryresponsefrom
the patients’ T cells (Figure 2 (B), lower part).
We conclude
(i) that tumor-reactive memory T cells from draining
lymph nodes of HNSCC patients could not be
activatedwiththefirst-generationvaccineATV-NDV,
(ii) that the cells, however, could become efficiently
reactivated using the same vaccine together with
optimized signals 1 and 2, and
(iii) that a similarly modified vaccine from an unrelated
tumor cell line was not capable to elicit such a
memory T cell response.
The fact that the ATV-NDV vaccine without bsAbs was
not capable to reactivate a response might be explained by
an anergic state of the tumor-draining lymph node-derived
T cells.
4.ActivationofAntitumorActivityfromNaiveT
Cellsby theSecond-GenerationVaccine
The tumor vaccine ATV-NDV/bsAb was shown to have
also an effect on naive T cells which were obtained from
total T cells of normal healthy donors by removal of
the CD45RO+ cells. This was observed by analyzing the
activation of thenaive T cellstowards tumorcellsafter 6days
ofincubationwithvariousvaccines.Figure 3(A)showsCD25
and CD45RO expression on naive T cells upon coincubation
with various NDV-based tumor vaccines. 1 × 105naive T
cells from a normal healthy donor were labeled with CFSE.
They were then coincubated with 1 × 104irradiated MCF-
7 cells which were modified with 100 HU NDV Ulster and
with suboptimal bsHN-CD3 (500pg/well) and 84.4U/well
bsHN-CD28 (d). Naive T cells, which came in contact only
with the vaccine (a) or with the vaccine loaded with each of
the constructs separately (b and c) served as controls. After
six days, the cells were harvested, blocked with Endobulin,
and then stained with anti-CD25-APC (a) or anti-CD45RO-
PE (b) before being analyzed on the FACS. Dead cells
were excluded by propidium iodide staining. The red boxes
indicate actively dividing (CFSE low) cells expressing CD25
or CD45RO. The frequencies are indicated as numbers in %
(adapted from [42]). We observed a strong activation only
when both bsAbs(bsHN-CD3and bsHN-CD28)were added
to the tumor vaccine (see Figure 3(A), (d)). Figure 3(B)
shows that tumor growth inhibition induced in naive T cells
uponcoincubationwith variousNDV-basedtumorvaccines.
In order to test the effectiveness of the new vaccine strategy
to activate in vitro naive T cells, an in vitro assay that we
called tumor neutralization assay (TNA) was developed. It
consists of an adherent tumor cell monolayer (here human
MCF-7 breast cancer cells) in a cell culture plate in which
γ-irradiated NDV-modified vaccine cells, fusion proteins,
and naive T cells of a healthy donor as effector cells are
brought into contact.During the incubation period of five to
seven days, the vaccine cells activate the effector cells which
increase their cytotoxic potential. During the effector phase,
the “bystander” tumor cells are lysed, or their growth is
inhibited. This results in a decrease in the number of live
tumor cells in monolayer (when compared to the controls).
After removal of the nonadhering remaining effector cells,
the number of surviving tumor cells can be quantified with
MTS as dye reagent for measuring the amount of viable
tumor cells per well. We observed that the tumor vaccine
with the 2 bound bsAbs bsHN-CD3 and bsHN-CD28was by
farthemostefficientinthisassayamongallthecombinations
of tumor vaccine and bsAbs tested (Figure 3(B)). We also
observed that the extent of T cell-mediated tumor growth
inhibitioninvitrodependedonanoptimalamountofbsHN-
CD3 and bsHN-CD28 present at the surface of the tumor
vaccine (data not shown, see [42]). Titration curves revealed
for each of the recombinant proteins, upon attachment to
the vaccine, a low dose optimum, of T cell stimulating or
costimulating activity [42].
An important aspect of T cell effector activity relates to
its duration. To test this, we activated purified T cells once
by coculture with the second-generation vaccine, performed
the TNA, and then repeatedly transferred the T cells from
the suspension above the destroyed monolayer onto fresh
tumor monolayers and followed their destruction thereafter.
In these ways, live adherent MCF-7 monolayers were coin-
cubated in 96-well plates with MCF-7-NDV vaccine cells,
purified T cells, a suboptimal dose of bsHN-CD3 (1μg
per 107vaccine cells; signal 1), and one of the following
costimulatory fusion proteins: bsHN-CD28 (signal 2a) or
tsHN-IL-2-CD28 (signal 2a-2b), each at a concentration
of 84U per 107vaccine cells (passage 0). T cells activated
in the presence of suboptimal bsHNCD3 alone (negative
controls) showed no tumor growth inhibition in this assay
(data not shown). For serial passages, the same conditions
were performed in parallel using a 6-well TNA format with
identical protein concentrations (TNA passage plates). After
7 days, the TNA test plates were developed using MTS as
reagent to obtain the value of tumor growth inhibition as
described in [43]. The cells from the TNA passage plates
were harvested, washed, and transferred for another 3 days
onto fresh MCF-7 monolayers, either in 96-well TNA test
plates(passage 1)orin6-well TNApassage platesforanother
round (passage 2). The results(Figure 3(C))revealeda much
longer duration of bystander antitumor activity in T cells
activated by the vaccine bound bsHN-CD3 and bsHN-CD28
(group b) than in T cells activated by the vaccine alone

Page 5

Clinical and Developmental Immunology5
HNO patients
Primary tumor
Tumor vaccine
PBMC
Coculture for
40 hours
T cells
Tumor cell culture
CD28
CD3
(A)
CD28
CD28
CD28
CD3
CD28
CD3
CD3
CD3
3587 39116
(a)(b) (c)(d)
(e)(f)
Mean =
(g)(h)
T cells
Vaccine (autologous with TA for memory T cells of patient (red), U937 without right TA (green))
TA
Patient no. 1
Patient no. 1
Patient no. 2
Patient no. 3
Patient no. 4
(B)
Figure 2: Reactivation of tumor-reactive memory T cells of cancer patients (here HNO cancer patients) by the ATV-NDV tumor vaccine
of the 2nd generation. (A) Protocol for the ex vivo stimulation of T cells from cancer patients by the NDV-based tumor vaccine of the
second generation in an autologous setting. Purified T cells isolated from tumor-draining lymph nodes (LN) of HNSCC patients were
tested for their capacity to be restimulated and to produce IFN-γ upon contact with various combination of the ATV-NDV tumor vaccine
(generated from the autologous tumor) and bsAbs molecules. For that, they were coincubated for 40 hours with the indicated autologous
vaccineformulations(tumor)(n = 4).Forspecificity control,theunrelated humanpromonocytictumorcell lineU937(U937)wasmodified
identically and used to reactivate the patients’ T cells. (B) IFN-γ ELISPOT results from A. Mean: mean number from triplicates of 4 patients
spot forming T cells per 1 million cells. Each patient’s T cells were stimulated with four different formulationsof either autologous (a–d) or
heterologous (e–h) vaccine. T cells of patientsno. 2, no. 3,andno. 4 alsodidnotreact to the heterologous vaccine(data notshown)(adapted
from [42]).