Content uploaded by Herbert B Slade
Author content
All content in this area was uploaded by Herbert B Slade on Sep 23, 2014
Content may be subject to copyright.
Available via license: CC BY 2.0
Content may be subject to copyright.
Genome Biology 2007, 8:R8
comment reviews reports deposited research refereed research interactions information
Open Access
2007Panelliet al.Volume 8, Issue 1, Article R8
Research
Sequential gene profiling of basal cell carcinomas treated with
imiquimod in a placebo-controlled study defines the requirements
for tissue rejection
Monica C Panelli
*
, Mitchell E Stashower
†
, Herbert B Slade
‡
, Kina Smith
*
,
Christopher Norwood
§
, Andrea Abati
¶
, Patricia Fetsch
¶
, Armando Filie
¶
,
Shelley-Ann Walters
‡
, Calvin Astry
‡
, Eleonora Aricó
*
, Yingdong Zhao
¥
,
Silvia Selleri
*#
, Ena Wang
*
and Francesco M Marincola
*
Addresses:
*
Immunogenetics Section, Department of Transfusion Medicine, Clinical Center National Institutes of Health, Bethesda, MD
20892, USA.
†
The Clinical Skin Center of Northern Virginia, Fairfax, VA 22033, USA.
‡
3M Pharmaceuticals, St Paul, MN 55144-1000, USA.
§
Department of Dermatology, National Naval Medical Center, Bethesda, MD 20889, USA.
¶
Laboratory of Pathology, National Cancer Institute,
Bethesda, MD 20892, USA.
¥
Biometric Research Branch, Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda,
MD 20892, USA.
#
Universita' degli Studi di Milano, Department of Human Morphology, via Mangiagalli, 20133 Milan, Italy.
Correspondence: Francesco M Marincola. Email: Fmarincola@mail.cc.nih.gov
© 2007 Panelli et al.; licensee BioMed Central Ltd.
This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Imiquimod response profiling<p>An analysis of basal cell carcinoma subjected to local application of imiquimod revealed that most transcripts stimulated by imiquimod involve the activation of cellular innate and adaptive immune-effector mechanisms.</p>
Abstract
Background: Imiquimod is a Toll-like receptor-7 agonist capable of inducing complete clearance of basal cell
carcinoma (BCC) and other cutaneous malignancies. We hypothesized that the characterization of the early
transcriptional events induced by imiquimod may provide insights about immunological events preceding acute
tissue and/or tumor rejection.
Results: We report a paired analysis of adjacent punch biopsies obtained pre- and post-treatment from 36
patients with BCC subjected to local application of imiquimod (n = 22) or vehicle cream (n = 14) in a blinded,
randomized protocol. Four treatments were assessed (q12 applications for 2 or 4 days, or q24 hours for 4 or 8
days). RNA was amplified and hybridized to 17.5 K cDNA arrays. All treatment schedules similarly affected the
transcriptional profile of BCC; however, the q12 × 4 days regimen, associated with highest effectiveness, induced
the most changes, with 637 genes unequivocally stimulated by imiquimod. A minority of transcripts (98 genes)
confirmed previous reports of interferon-α involvement. The remaining 539 genes portrayed additional
immunological functions predominantly involving the activation of cellular innate and adaptive immune-effector
mechanisms. Importantly, these effector signatures recapitulate previous observations of tissue rejection in the
context of cancer immunotherapy, acute allograft rejection and autoimmunity.
Conclusion: This study, based on a powerful and reproducible model of cancer eradication by innate immune
mechanisms, provides the first insights in humans into the early transcriptional events associated with immune
rejection. This model is likely representative of constant immunological pathways through which innate and
adaptive immune responses combine to induce tissue destruction.
Published: 15 January 2007
Genome Biology 2007, 8:R8 (doi:10.1186/gb-2007-8-1-r8)
Received: 15 August 2006
Revised: 6 October 2006
Accepted: 12 January 2007
The electronic version of this article is the complete one and can be
found online at http://genomebiology.com/2007/8/1/R8
R8.2 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
Background
In 2004, Aldara™ (imiquimod 5% cream, 3M Pharmaceuti-
cal, St Paul, MN, USA) labeling was extended by the Food and
Drug Administration to include treatment of superficial basal
cell carcinoma (BCC) based upon randomized controlled tri-
als demonstrating complete histological clearance in 78% to
87% of superficial BCC treated topically 5 days per week for 6
weeks [1,2]. Pilot-scale and investigator initiated trials had
shown 90% to 100% clearance with q12 hours (twice per day)
dosing [3].
Imiquimod belongs to a family of synthetic small nucleotide-
like molecules with potent immuno-modulatory activity
mediated through Toll-like receptor (TLR)-7 (and 8) signal-
ing. When applied topically, these compounds display
immune-mediated anti-tumoral activity without damaging
normal tissues [1,3-7] Imiquimod targets predominantly
TLR-7 expressing plasmacytoid dendritic cells (pDCs) with
secondary recruitment and activation of other DC and macro-
phage subtypes and induction of T helper
1
responses within
three to five days of treatment [4]. Stimulation of pDCs
through TLR-7/myeloid differentiation response gene 88
(My-D88)/IRF-7 signaling induces expression of interferon
(IFN)-α, which appears to act upon natural killer (NK) cells
and conventional dendritic cells (DCs) to stimulate IFN-γ,
tumor necrosis factor (TNF)-α, monocyte chemoattractant
proteins (MCPs) and other cytokines [5,8,9] This immuno-
logical cascade leads within two weeks to apoptotic death of
cancer cells and their substitution by a mononuclear cell infil-
trate [3-5,8]
Although imiquimod function seems particularly associated
with IFN-α-stimulated genes (ISGs) [10], it remains unclear
whether this pathway is solely responsible for all the down-
stream effects ultimately resulting in tumor clearance.
Indeed, a comprehensive and conclusive characterization of
the events leading to tumor rejection based on a prospectively
controlled study has never been reported. We previously
characterized ISGs in vitro [11] and in vivo (Belardelli F and
Arico' E, manuscript in preparation), compiling a road map
for the interpretation of transcriptional surveys of biological
conditions affecting the tumor microenvironment (Addi-
tional data file 1).
Here, we report a paired analysis of adjacent punch biopsies
obtained pre- and post-treatment from 36 patients with BCC
subjected to local application of imiquimod or a control
cream in a blinded, randomized protocol.
Results
A total of 65 subjects were screened, but 27 were ineligible
due to their pre-enrollment biopsy excluding BCC and 2 were
ineligible for other reasons. A total of 36 subjects were eligible
for the study and started treatment with either imiquimod (n
= 22) or vehicle cream (n = 14) (Table 1). After unblinding,
treatment groups were color-coded to facilitate the discus-
sion. Out of the subjects, 61% had nodular BCC, 17% superfi-
cial BCC, and 22% unspecified BCC. Of note is that all 4
subjects randomized to the imiquimod q12 hours × 4 days
group had nodular BCC. Post-treatment biopsies were taken
<12 hours after last dose for 17% of subjects, >36 hours after
the last dose date for another 17%, and between 18 and 30
hours after last dose for 33%. This variability was uncontrol-
lable and due to patient compliance. The locations of the
tumors were: 41% on the face; 25% on the extremities; 22% on
the trunk; and 11% on either the neck or scalp. Furthermore,
patient (P) 23 and P28 did not complete treatment, missing
two placebo and one imiquimod dose, respectively. The
imbalance in the distribution of the elapsed time between last
treatment dose and post-treatment biopsy did not signifi-
cantly affect the results except, possibly, for the q24 × 8 (pink)
cohort. Interestingly, at this early time point, already 9 of 22
imiquimod-treated BCCs were found to be clear of tumor
cells, particularly among patients treated with the most
intense schedule.
Quantitative PCR
At this early stage of treatment, no changes were observed in
TNF-α and MCP-1 expression, in contrast with others' find-
ings at later stages [5,8,9] IFN-γ 2
-ΔΔCT
from baseline to end of
treatment (EOT) was significantly increased compared to
dose-matched controls at all but the earliest time point (q12 ×
2, orange group; Figure 1a). IFN-α followed a similar pattern
but significance was observed only with the most intense reg-
imen (q12 × 4, blue group; Figure 1b).
Identification of treatment (imiquimod)-specific genes
Unsupervised analysis applying various filtering parameters
failed to segregate samples according to treatment, suggest-
ing that imiquimod affects an insufficient number of genes to
alter the global transcript of BCC. A paired t-test (cut-off p
2
value < 0.05) was applied to identify genes differentially
expressed by identical lesions before and after treatment
within each cohort. For instance, the q12 × 4 (blue) cohort dif-
ferentially expressed 1,578 genes at EOT compared to paired
pre-treatment samples. Reclustering of these genes demon-
strated that most were similarly expressed by post-treatment
samples treated with placebo, reflecting changes due to vehi-
cle alone or the tissue repair induced by the adjacent pre-
treatment biopsy. A node, however, contained 263 genes
exclusively upregulated in all EOT imiquimod-treated sam-
ples (Figure 1c (part b), vertical blue bar). This cohort-based
training/prediction analysis was repeated with the other
three treatment regimens, providing independently similar
results. In all cases, nodes were identified inclusive of genes
uniquely expressed in EOT imiquimod-treated samples (Fig-
ure 1c (parts a and d); Additional data file 4). The number of
imiquimod-induced genes varied among cohorts, however,
with the largest amount in the q12 × 4 (blue) cohort, in line
with the higher clinical effectiveness of this intense dosing
regimen [3]. There was extensive overlap among the genes
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.3
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
Table 1
Composition of study cohorts
Patient ID Cohort Doses received EOT → B× time lapse (hours) Histology ΔCD8 ΔCD56 Tumor at EOT
P5 Imiq q12 × 2 days 4 13 Nodular 0 -1 +
P6 Imiq q12 × 2 days 4 14 Undetermined 0 0 +
P17 Imiq q12 × 2 days 4 36 Undetermined NE NE -
P18 Imiq q12 × 2 days 4 33 Nodular +1 0 +
P30 Imiq q12 × 2 days 4 16 Nodular 0 0 -
P38 Imiq q12 × 2 days 4 17 Nodular 0 0 +
P231 Imiq q12 × 2 days 4 22 Undetermined +1 0 +
P10 Vehic q12 × 2 days 4 12 Nodular 0 0 +
P23 Vehic q12 × 2 days 2 15 Nodular +2 0 +
P26 Vehic q12 × 2 days 4 45 Nodular 0 0 +
Mean ± SD = 22 ± 11.5
P1 Imiq q12 × 4 days 8 8 Nodular 0 0 +
P21 Imiq q12 × 4 days 8 41 Nodular 0 +1 +
P22 Imiq q12 × 4 days 8 11 Nodular +1 0 -
P40 Imiq q12 × 4 days 8 17 Nodular +1 +1 -
P42 Imiq q12 × 4 days 8 3 Undetermined +3 0 -
P129 Imiq q12 × 4 days 8 19 Nodular +1 0 +
P135 Imiq q12 × 4 days 8 21 Superficial +1 +1 -
P41 Vehic q12 × 4 days 8 28 Nodular +1 0 +
P134 Vehic q12 × 4 days 8 19 Nodular +2 0 +
P8 Vehic q12 × 4 days 8 20 Nodular 0 0 +
P20 Vehic q12 × 4 days 8 16 Superficial 0 0 +
Mean ± SD = 18 ± 10.2
P11 Imiq q24 × 4 days 4 26 Nodular 0 +1 +
P28 Imiq q24 × 4 days 3 20 Nodular NE NE +
P112 Imiq q24 × 4 days 4 44 Nodular 0 0 +
P214 Imiq q24 × 4 days 4 51 Nodular +2 +1 -
P4 Vehic q24 × 4 days 4 16 Superficial NE -1 -
P13 Vehic q24 × 4 days 4 30 Nodular 0 0 +
P36 Vehic q24 × 4 days 4 25 Superficial 0 0 +
Mean ± SD = 30 ± 12.8
P233 Imiq q24 × 8 days 8 32 Undetermined 0 +1 +
P132 Imiq q24 × 8 days 8 159 Undetermined 0 +2 -
P24 Imiq q24 × 8 days 8 48 Superficial +1 0 -
P3 Imiq q24 × 8 days 8 12 Undetermined -1 0 +
P2 Vehic q24 × 8 days 8 6 Undetermined 0 -1 +
P15 Vehic q24 × 8 days 8 21 Nodular 0 0 +
P27 Vehic q24 × 8 days 8 26 Nodular NE NE +
P137 Vehic q24 × 8 days 8 11 Superficial 0 0 +
Mean ± SD = 39 ± 50.3
Punch biopsies are labeled according to patient number (P1 to P42) and timing of excision: PB0, pre-enrollment; PB1 and PB2, pre-treatment; PB3
and PB4, post-treatment. Biopsies from patients replacing drop-outs were labeled one digit to the serial number (that is, P101 to P142 or P201 to
P242. PB1 and PB3 were collected for total RNA isolation; PB2 and PB4 for IHC. Undetermined refers to a BCC histology in-between superficial and
nodular. ΔCD8 and ΔCD56 scores differences in infiltrate between EOT and pre-treatment samples (see Materials and methods). Tumor at EOT:
identifiable (+) or not identifiable (-) tumor cells in the hematoxylin eosin stained EOT biopsy. Imiq, imiquimod; NE, not evaluated; Vehic, vehicle.
R8.4 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
identified by the various comparisons (Figure 1c (part e)); 41
(63%) of 65, 40 (71%) of 56 and 16 (70%) of 23 genes differ-
entially expressed in the orange, green and pink groups,
respectively, were included among those identified as differ-
entially expressed in the blue group. Reclustering of experi-
mental samples based on imiquimod-specific signatures from
each cohort suggested their independent predictive value in
sorting imiquimod-treated BCC from pre-treatment and con-
trol samples as exemplified by the blue cohort signature,
which clumped together not only the samples from the blue
group, which served as a basis to select the genes used for
clustering, but also 9 of the other 15 imiquimod-treated sam-
ples compared with only 3 of 14 vehicle-treated samples
(Fisher p
2
value = 0.04). Four of the five samples that did not
cluster together with the blue group samples belonged to the
orange group (Figure 1d).
Thus, different dosing schedules differed quantitatively but
not qualitatively, with the same genes being induced among
them. The striking difference in number of genes induced
between the q12 × 2 (orange) and the q12 × 4 (blue) cohorts
strongly emphasizes the importance of the number of doses;
however, the q24 × 8 (pink) group, which received the same
number of imiquimod applications as the blue group in twice
the amount of time, displayed similar but dampened
transcriptional changes, emphasizing the importance of
administration to sustain the pro-inflammatory stimulus
associated with the higher efficacy of the q12 schedule.
This analysis supports the specificity of our findings but also
simultaneously emphasized the need to discriminate imiqui-
mod-specific effects from those due to vehicle cream applica-
tion and/or tissue repair induced by the adjacent pre-
treatment biopsy. Because q12 dose scheduling had been
observed previously to produce the highest rates of clearance
[3], we adopted this cohort as the basis for further analysis.
This selection offered the additional advantage of allowing
the largest number of temporally matched placebo-treated
samples (q12 × 4 and q24 × 4 cohorts). At EOT, 1,578 genes
were significantly altered in expression in the q12 × 4 (blue)
cohort compared to pre-treatment (paired t-test cut-off p
value < 0.05; Figure 2a). To eliminate placebo and/or surgical
bias, an unpaired t-test (cutoff p value < 0.05) was applied to
this gene pool, identifying transcripts differentially expressed
between imiquimod-treated EOT samples and vehicle cream-
treated samples. This analysis left 637 genes unequivocally
modulated by imiquimod (Figure 2b,c; Additional data file 3).
A global test was applied to this gene set to test the likelihood
of getting this proportion of significant genes by chance (at
the 0.05 level) if there were no real differences between the
two classes. Such likelihood was negligible, with a permuta-
tion p value of 0.001. The false discovery rates (FDRs) of the
differentially expressed genes are less than 11.9%. To estimate
the specificity/accuracy of the 637 'imiquimod-induced'
genes, we considered as a training set the samples utilized for
their identification (q12 × 4 days treatment group and the q12
× 4 and q24 × 4 days vehicle groups; Figure 2b). The trained
predictors were then used to segregate post-imiquimod treat-
ment samples from pre-treatment or vehicle treated samples
belonging to the other groups. This analysis was performed
using the Support Vector Machines (a supervised learning
algorithm that classifies data by finding optimal fit between
different statistical classes); this analysis yielded a sensitivity
of 60%, specificity of 92% and an overall accuracy of 82.4%.
Thus, the set of 637 genes identified by this study represent a
highly specific functional signature of imiquimod-induced
changes during the early stages of therapy in lesions whose
transcriptional profiles were sufficiently activated. The rela-
tively low sensitivity of the gene set as predictors most likely
reflects the exclusion of lesions in the earliest cohort (orange
group) that were not exposed sufficiently to imiquimod.
Of the 637 genes, 65 were also significantly altered in expres-
Differential expression of IFN-γ and IFN-α in EOT compared to pre-treatment samples in all cohorts; hierarchical clustering based on genes differentially expressed at EOT compared to pre-treatment samples in each treatment cohort and dendrogram showing the degree of relatedness of samples based on imiquimod-induced genes in the blue groupFigure 1 (see following page)
Differential expression of IFN-γ and IFN-α in EOT compared to pre-treatment samples in all cohorts; hierarchical clustering based on genes differentially
expressed at EOT compared to pre-treatment samples in each treatment cohort and dendrogram showing the degree of relatedness of samples based on
imiquimod-induced genes in the blue group. The 2
-ΔΔCT
describes (a) IFN-γ and (b) IFN-α gene expression fold change at EOT relative to baseline after
normalization according to the endogenous reference cyclophilin G. C
T
equals the mean cycle times of duplicate wells and ΔΔC
T
= (C
T
, Target-C
T
,
cyclophilin) EOT - (C
T
, Target-C
T
, cyclophilin) baseline. The fold-change data were transformed using logarithm
10
. The box and whisker style box plot
gives the median and interquartile range (box), 1.5 of the inter-quartile range (whiskers), points outside the whiskers (square symbols) and the mean (cross
symbol). Statistics: p values refer to 2-sample t-tests between treatment and control groups. (c) Based on a paired t-test cut-off p
2
value < 0.05, 1,311
genes were differentially expressed between the pre-treatment and EOT samples in the q12 × 2 (orange) cohort. Reclustering of these genes identified a
node of 65 genes uniquely upregulated in the imiquimod-treated EOT samples (part i). Similar analyses were performed for the other imiquimod-treated
cohorts; 1,578 genes were differentially expressed in the q12 × 4 (blue) cohort, including an imiquimod-specific node of 263 genes (part ii and the vertical
blue bar in adjacent complete data set); 650 genes were differentially expressed in the q24 × 4 (green) cohort, including an imiquimod-specific node of 58
genes (part iii); and 495 genes were differentially expressed in the q24 × 8 (pink) cohort, including an imiquimod-specific node of 23 genes (part iv). A Venn
diagram displays the extent of overlap among genes differentially expressed in the three most informative orange, blue and green groups (part v). (d)
Reclustering of all BCC samples based on the imiquimod-specific 263-gene signature identified in the q12 × 4 (blue) cohort. Straight lines identify
imiquimod-treated EOT samples color coded according to treatment regimen; dashed lines identify vehicle cream-treated EOT samples and unlabeled are
the all pre-treatment samples. A diagram illustrating the strategy used to prepare Figure 1c,d is available as Additional data file 4.
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.5
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
Figure 1 (see legend on previous page)
After treatment
(Px-PB3)
IFN-γ
IFN-α
Before treatment
(Px-PB1)
Vehicle
Log
10
2 -ΔΔCT (post-pre)
Imiquimod
65 genes
q12,2D q12,4D q24,4D q24,8D
263 genes
56 genes
23 genes
263
65
23
23
18
1
22
200
15
56
iiiiiiiv
v
(a)
(b)
(c)
(d)
R8.6 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
sion in the q12 × 2 (orange) cohort; we refer, therefore, to
these as 'primary' responders to imiquimod and to the rest as
'secondary'. Finally, the 637 genes were matched to our data-
base of IFN-α-related signatures consisting of 426 genes
identified using the same cDNA platform and reference sys-
tem in monocytes stimulated with various IFN-α subtypes in
vitro [11] and/or induced in vivo by systemic IFN-α
2b
ther-
apy. Only 98 (22 included among the primary) genes matched
the database and were considered bona fide ISGs. The
primary ISGs included STAT-1, MX1, MX2 and IFITM1. By
four days, secondary ISGs had broadened to STAT2, IRF-2
and IRF7, JAK-2 and JAK-3 and N-myc interactor (NMI).
Moreover, CXCL10/IP-10 was significantly upregulated;
CXCL10 is a monocyte and T lymphocyte chemoattractant
interacting with the chemokine receptor CD183 (CXCR3) and
T-cell CD26. The remaining 539 genes were induced through
IFN-α-independent pathways, suggesting that only a small
proportion of the effector activity of imiquimod is mediated
by IFN-α.
Primary non-IFN-
α
-stimulated genes
By the second day of q12 imiquimod treatment, 65 primary
non-ISGs were identified, echoing predominantly innate
immune effector functions (Figure 3a). CXCR3, a ligand for
IP-10 and monokine induced by IFN-γ (MIG/CXCL9) was the
earliest upregulated cytokine receptor, suggesting its early
involvement in the crosstalk leading to migration and activa-
tion of monocytes and lymphocytes. Also induced by IFN-γ
were several HLA class I and class II transcripts, including
HLA-B and HLA-DRβ1. Transcripts critical for the activation
of innate immune effector cells, such as NK cells and mono-
nuclear phagocytes, were highly expressed; for example,
TYROBP, a killer-cell immunoglobulin-like receptor family
member and cytochrome β-245, a component of phagocytes'
lytic function. Activation of macrophages was also strongly
supported by the upregulation of CD68, and the modulation
of complement component 1 qα (C1QA) and MY-D88 [12].
The induction of CD37 represented an early sign of the tran-
sition from an innate to an adaptive immune response as
CD37 regulates T cell proliferation through TCR signaling
[13]. Finally, Caspase 10 upregulation suggests an early initi-
ation of apoptotic mechanisms.
Secondary non-IFN-
α
-stimulated genes
The vast majority of transcriptional effects were observed
four days after q12 treatment (Figure 3b), when the inflam-
matory process is amplified by the induction of cytokines,
their receptors and genes related to their interactions, such as
dual specificity phosphatase 5 (DUSP-5) and the gene encod-
ing the anti-apoptotic BCL2. The induction of pro-inflamma-
tory molecules was strongly reminiscent of the broad
transcriptional changes induced by the in vitro stimulation of
peripheral blood mononuclear cells (PBMCs) by interleukin
(IL)-2 [14]. In particular, the upregulation of cytokines and
corresponding receptors within the common γ chain receptor
family (particularly IL-15 and the IL-15 receptor α-chain, the
IL-2/IL-15 receptor β-chain and the common γ chain itself;
Figure 3b) suggest early activation within the tumor microen-
vironment of CD8 T and NK cells [15,16]. This notion is also
supported by the modulation of downstream transcription
factors of IL-2/IL-15 receptor triggering, such as Jak kinases,
STAT-1, STAT-3 and STAT-5, and the upregulation of T cell
receptor subunits, cytotoxic granules and NK-activation
receptors (Figure 3b). The increased expression of the chem-
okine (C-C motif) receptor 7 (CCR-7) also supports a potent
activation of pro-inflammatory signals; CCR7 is expressed by
activated B and T lymphocytes and NK cells and controls their
migration to inflamed tissues [17]. MIG is a chemoattractant
for CXCR3-bearing immune cells that may contribute,
together with IP-10, to the intensification of the acute inflam-
matory process. Monocyte inflammatory protein (MIP)-1α
(CCL3), MIP-1β (CCL4) and MCP-3 (CCL7) were also induced
at this point. Among them, MCP-3 has been shown to aug-
ment monocyte anti-tumor activity while CCL3/MIP-1α and
MIP-1β represent potent pro-inflammatory factors with
chemotactic properties for neutrophils and DC and NK cells.
Interestingly, CD64 and the low-affinity IgG Fc receptor II-B
(FCGR2B), which were also upregulated among the second-
ary non-ISGs (Figure 3c), have been shown to stimulate MIP-
1α and MIP-1β release [18].
Cytotoxic T and NK cell signatures
The most striking effects of imiquimod were on cytotoxic
mechanisms, with the induction of NK cell gene-5 (NKG-5),
NK cell protein-4 (NK4)/IL-32 granzyme-B, -A and -K, per-
forin and lymphotoxin-β receptor [19,20]. (Figure 3b,c).
Moreover, the concomitant transcription of several caspases
indicate active cytotoxicity [20] combined with granule-
mediated apoptosis suggested by the upregulation of prote-
oglycan 1 secretory granule (PRG1) [21].
Identification of treatment (imiquimod)-specific transcripts in the most intensive schedule (q12 × 4 (q12,4d), blue cohort)Figure 2 (see following page)
Identification of treatment (imiquimod)-specific transcripts in the most intensive schedule (q12 × 4 (q12,4d), blue cohort). (a) A pairwise t-test (p value <
0.05) was applied to identify genes differentially expressed between pre-treatment and EOT biopsies from the same BCC belonging to the q12 × 4 (blue)
cohort. The 1,578 genes identified were then tested for treatment specificity by identifying those differentially expressed between the blue group treated
with imiquimod (TX) compared with temporally matched, vehicle control-treated EOT biopsies (combined blue and green groups (b) The remaining 637
treatment-specific genes were classified based on their significant expression also in the earlier q12 × 2 (orange) group as primary (65 genes) while the
other ones were considered secondary. Finally, the same genes were also compared to a database of IFN-α-associated transcripts as described in the
Materials and methods. In the same panel the 637 genes are shown in a supervised-sample hierarchical clustering of the genes. (c) Legend of samples,
dashed and solid bars identify vehicle control or imiquimod-treated samples, respectively.
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.7
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
Figure 2 (see legend on previous page)
(a)
(b)
(c)
17k genes DATASET
65 genes
Pre-treatment
Post-treatment (EOT)
637 genes
572 genes
1578 genes q12,4D DATASET
ttest p-value < 0.05
ttest p-value < 0.05
PRE
POST TX
POST TX
n=7 n=7
n=7 n=7
1578 genes
637 genes
POST vehicle
R8.8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
Several T cell receptor signaling and amplification-associated
genes were also upregulated, including those encoding TCR-
α, -β and -γ chains, ζ-chain (ZAP70), CD3Z, T cell immune-
regulator 1 and related co-receptor CD5 [22,23]. Moreover,
CD2/LFA-2 mediates T and NK cell activation through inter-
actions with CD59, which is also upregulated at this time
point [24,25]. Similarly, the overexpression of CD69 marks
the activation of T and NK cells and it has been correlated by
Posselt et al. [26] with acute renal allograft rejection.
Several transcripts suggest a primary involvement of NK cells
in the process, such as the NKG2 family of genes, which
encode receptors that are expressed on most NK cells [27]:
killer cell lectin-like receptor subfamily C, member 2
(KLRC2/NKG2C), member 3 (KLRC3/NKG2E), and member
4 (KLRC4/NKG2F). Moreover, all NK receptor adapter pro-
teins containing an immune-receptor tyrosine based activa-
tion motif (ITAM) were found to be upregulated (FCERIg),
CD3z and TYROBP/DAP12. The upregulation of KLRC2/
NKG2C, TYROBP/DAP12 and FCER1G suggests the
occurrence of NK and T cell activation, which would lead to
release of pre-made cytotoxic granules and secretion of
cytokines [27]. Another NK cell-related gene is that encoding
Cathepsin w, a cysteine proteinase associated with the mem-
brane and the endoplasmic reticulum of NK and T cells and
regulation of their cytolytic activities [28]. Finally, the minor
histocompatibility antigen HA-1 may be one of the immuno-
dominant stimulators of graft-versus-host and graft-versus-
malignancy effects through increasing cytotoxic mechanisms
[29].
Markers of immune infiltrates
Transcriptional analysis portrayed a predominant enhance-
ment of immune infiltrates associated with T and NK cells.
Because 9 of 22 imiquimod-treated BCCs were cleared of
tumor cells at EOT it was impossible to further analyze
whether the identified changes were occurring in specific his-
tological areas as sharply defined in pre-treatment lesions. In
such cases, changes in immune infiltrates were calculated
comparing EOT results with pre-treatment peri-tumoral
infiltrates. With all four imiquimod treatment groups pooled
together, significant increases were noted in CD56 (NK cells),
CD4 and CD8 T cells, with CD56 (NK cells) showing signifi-
cant difference relative to the pooled vehicle group (Table 2,
Figure 4). Moreover, BCL-2 expression was selectively
enhanced in immune but not cancer cells. Importantly,
enhancement of CD8 expression was strongly dependent
upon treatment schedule, with 5 of 7 subjects treated in the
q12 × 4 (blue) cohort experiencing increases in the number of
CD8 T cells (p value < 0.05). Other markers did not reach sta-
tistical significance, including those associated with cytotoxic
activity, such as granzymes and perforin, suggesting that the
differences identified at the transcript level may precede
changes detectable as protein expression, as we recently
observed studying transcript to protein relationships in IL-2-
stimulated PBMCs [14]. These data confirm the transcrip-
tional observation that imiquimod primarily induces
recruitment and activation of T and NK cells within the BCC
microenvironment.
Discussion
This is the first prospectively controlled study conducted to
identify the early biological events associated with the eradi-
cation of BCC through an immune-mediated mechanism. By
protocol design, tumor regression did not represent an end-
point and tumors were removed at the end of the study. Thus,
the association between the molecular/genetic findings and
tumor clearance is presumptive, based on the historical 80%
to 90% clearance rates recognized by the Food and Drug
Administration for the release of imiquimod for clinical use
[2]. However, it is interesting to note that 9 of 22 (41%) imiq-
uimod-treated BCCs were devoid of cancer cells by EOT (2 to
8 days from beginning of treatment) while only 1 of 14 (7%)
control-treated BCCs had no identifiable tumor cells (Fisher
test p value = 0.05), suggesting that artifacts due to vehicle
administration or surgical trauma were not responsible for
the early tumor clearance.
As indicated by qPCR, IFN-γ transcription was more preva-
lent than IFN-α transcription. This is in line with the evidence
of predominant NK, CD8 and CD4 T cell activity in this study.
Sullivan et al. [30] had indeed previously observed similar
cellular infiltrates (particularly CD4 and CD56 expressing
cells) in a smaller, open-label, matched controlled, non-rand-
omized study in which six patients with BCC treated with imi-
quimod at daily intervals for a total of ten administrations
were compared with six patients receiving comparable vehi-
cle cream treatment. The predominance of IFN-γ transcrip-
tion suggests that pDCs trigger other immune functions
through the production of IFN-α, which in turn activates res-
ident T and NK cells, selective producers of IFN-γ [31]. We
hypothesize that these secondary immune effector mecha-
nisms induce destruction of target cells, providing antigen to
professional antigen presenting cells for priming of naive T-
cells in draining lymph nodes [31,32]. Indeed, several of the
Visual display of selected treatment (imiquimod)-specific transcripts (complete database available on line)Figure 3 (see following page)
Visual display of selected treatment (imiquimod)-specific transcripts (complete database available on line). (a) Display of selected primary treatment-
specific genes identified as per Figure 2. (b) Secondary treatment-specific genes related to effector functions with primary focus on cytokines, cytokine
receptors and lytic enzymes. (c) Secondary treatment-specific genes representative of cell surface markers, receptors and associated molecules. In red are
genes whose expression was found to be associated with acute renal allograft rejection [37]. Treatment cohorts are described by the bars on top of each
cluster.
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.9
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
Figure 3 (see legend on previous page)
(a)
- CXCL10/IP-10
- CXCL7/MCP-3
- CXCL9/Mig
-----------------------------
- JAK-2
- CD68
---------------------- Macrophage stimulating 1
- CD64
-------------------------- HLA-G
Caspase 8
Caspase 1
--------------------------------------------- CD2
----------------- KLRC3
- CD4
- ZAP 70
----------------------- Allograft inflammatory factor 1
- IL-15
- CCR7
- IL-2/IL-4/IL-7/IL-9/IL-15 Rg
- PRG-1
- Natural killer cell gene -5
---------------------------------------------
Granzyme K
Perforin
CCL4/MIP-1b
- IL15 Ra
- Granzyme B
- Caspase 5
- IL-6
----------------------- STAT-1
------------------------ Interferon-stimulated factor 3
- Granzyme A
- Natural killer-cell transcript 4/IL-32
- IL-2/IL-15 Rb
------------------------ Lymphotoxin receptor precursor
- T cell immune-regulator 1
- CD8
- insulin-like growth factor 1 receptor
- T-cell receptor
------------------------------------------- CD5
-------------------------- CD62L
- CD3 Zeta
- Minor histocompatibility antigen HA-1
- Cathepsin W
- CD69
- CD59
------------------------- TNF receptor
(b)
(c)
- CD37
-------------- CD68
- CXCR3
- TYROBP
-------------- HLA-DM a
- C1QA
- Caspase 10
-------------- MYD88
-------------- HLA-DRb1
- HLA-B
R8.10 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
transcripts associated with imiquimod treatment show acti-
vation of T and NK cells and induction of IFN-γ stimulated
genes (Figure 3). The cytotoxic T and NK cell signatures iden-
tified here (granzymes, perforin and other NK cell-related
genes) have recently been described in a mouse model of IFN-
α and IFN-γ-producing killer DCs (IKDCs) [33], which simul-
taneously display cytotoxic and pro-inflammatory functions.
Thus, IKDCs could summarize in a cellular unit our findings
of ISG activation combined with broader cytotoxic and pro-
inflammatory properties. At present, IKDCs have not been
characterized in humans, nor it is known whether they
express TLR-7; future studies should address their role as
putative mediators of immune rejection.
Imiquimod treatment stands as a unique opportunity to study
the mechanisms of immune-mediated rejection directly in
human tissues. This TLR-7 agonist links multiple immune
pathways. Of these, IFN-α plays a consistent but not exclusive
role. Previous transcriptional surveys have provided a broad
view of the biological processes associated with immune-
mediated tissue destruction, identifying convergent
characteristics. Neoplastic inflammation approaches the
unresolving process of chronic hepatitis C virus (HCV) infec-
tion where the presence of antigen-specific immune
responses do not lead to clearance of the pathogen in the
majority of cases [34,35]. Both diseases are characterized by
the expression of ISGs that do not seem sufficient to clear the
pathogenic procress. Similar signatures can be identified in
IHC staining for CD56 and CD8 in BCC from (a) P40 (imiquimod treated) and (b) P8 (vehicle-control)Figure 4
IHC staining for CD56 and CD8 in BCC from (a) P40 (imiquimod treated) and (b) P8 (vehicle-control). Lesions were graded blindly by two pathologists
(AA and AF) and graded before and at EOT for peri-tumoral and intra-tumoral immune cells infiltrate. Cancer cells were evaluated separately for each
marker. When BCC was absent at EOT as in P40 the immune infiltrate was compared to the peri-tumoral pre-treatment infiltrate. NE, not evaluable
because no tumor cells were left at EOT.
(a)
(b)
EOT
Pre-treatment
Tumor
ΔCD56=+1
65DCE&
H
CD8
EOT
Pre-treatment
ΔCD8=+1
Peri-tumoral Peri-tumoral
65D
C
E&
H
CD8
Tumor
Tumor
Peri-tumoral Peri-tumoral
ΔCD56= 0 ΔCD8 = 0
No tumor cells
P40
P8
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.11
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
liver biopsies from patients with chronic HCV infection [36]
and in chronic allograft rejection controlled with standard
immune suppression [37]. ISGs are also consistently
expressed in melanoma metastases following the systemic
administration of IL-2 independent of clinical outcome [38].
Thus, it appears that ISGs are part of immunological proc-
esses associated with chronic inflammation insufficient to
clear its cause. On the contrary, several non-ISGs identified
by this study delineate potent inflammatory (CCL7/MCP-3,
CCL4/MIP-β, and so on) and cytotoxic (granzymes, perforin,
NKG-5, and so on) functions rarely observed in chronically
inflamed tissues but described in the acute inflammation
associated with destruction of a tumor [38] or allograft [37],
liver damage in HCV-induced cirrhosis [36] or gut dysfunc-
tion during flares of Crohn's disease [39]. This study corrob-
orates the impression that immune-mediated tissue
destruction comprises at least two components: a baseline
cluster of ISGs that may be necessary but insufficient to
induce tissue rejection and a less common activation of broad
cytotoxic and other potent pro-inflammatory innate immune
effector functions that are more tightly associated with rejec-
tion. Our findings are supported by the recent description of
clearance of established cancers by the adoptive transfer of
innate immune effector cells in the powerful model of sponta-
neous regression/complete resistance mice [40].
Sarwal M et al. [37] reported strikingly similar results evalu-
ating the transcriptional behavior of renal cell allograft dur-
ing acute rejection, basing the analysis on a similar array
platform and utilizing the same RNA amplification method
[41] (Figure 3, transcripts labeled in red). In spite of these
similarities, they also reported a B cell signature character-
ized by enhanced expression of CD20 and several immu-
noglobulins that we did not identify in our study. This
discrepancy could be explained by a specific role that B cell-
mediated immunity may play in the context of allo-recogni-
tion. In the case of BCC, the strong pro-inflammatory stimu-
lus induced by imiquimod through TRL-7 signaling might
bypass the requirement for an endogenous, tissue specific
insult responsible for the secondary triggering of the cellular
immune effector mechanisms identified by both studies. The
signatures identified by both studies also match the anecdotal
identification of the same genes in a melanoma metastasis
that underwent regression following systemic IL-2 therapy
[38].
Among the genes mutually reported by the previous three
studies, NK4/IL-32 was recently recognized as a central
mediator of Crohn's disease [42] and associated with liver
damage during HCV infection [36]. NK4/IL-32 is a potent
inducer of pro-inflammatory cytokines and it is selectively
expressed by immune cells stimulated with IFN-γ IL-2 or the
combination of IL-12 and IL-18 [14,39]. Indeed, we found
NK4/IL-32, together with other genes associated with cyto-
toxic function, to be constitutively expressed by NK cells but
only by activated CD8+ T cells [43]. Moreover, we recently
observed NK4/IL-32 to be preferentially expressed in
metastatic melanoma compared with other less immune
responsive cancers [44]. It is possible that NK4/IL-32 may
play a central role during imiquimod treatment by amplifying
inflammatory stimuli through the induction of a cytokine cas-
cade. Thus, this novel cytokine emerges as a central player in
immune rejection or autoimmunity.
Dermatologists have long used imiquimod to treat BCC
[4,45,46] Imiquimod mimics the action of single-stranded
viral RNA [31], activating a pro-inflammatory cascade as a
chemical prototype of the danger model of immune activation
[47]. Meanwhile, tumor immunologists have struggled to
Table 2
Scoring of immune infiltrate by immuno histochemistry
Pooled treatment groups Δ Score post - pre-treatment Within group p value
-3 -2 -1 0 1 2 3
CD56
Imiquimod 0 0 1 12 6 1 0 0.03
Vehicle 0 0 2 11 0 0 0 0.17
CD8
Imiquimod 0 0 1 10 7 1 1 0.01
Vehicle 0 0 0 9 1 2 0 0.22
CD4
Imiquimod 0 0 1 12 7 0 0 0.03
Vehicle 0 0 2 5 4 1 0 0.22
BCL-2
Imiquimod 0 0 1 11 6 2 0 0.02
Vehicle 0 0 2 9 1 1 0 0.72
P values associated with the paired t-test for within group shifts relative to baseline. Δ Score refers to differences in infiltrate between EOT and pre-
treatment samples using the scoring scale described in Materials and methods (IHC section).
R8.12 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
explain the paradoxical co-existence of tumor antigen-spe-
cific T cells induced by vaccination with growing tumor tis-
sues. Indirect evidence suggests that vaccine-induced T cells
reach the tumor site [48] and recognize tumor cells producing
IFN- However, this is not sufficient for tumor rejection
since other effector mechanisms are not simultaneously acti-
vated [49] because cancers do not provide the danger signal
necessary for full implementation of the immune responses
[50]. Thus, immunization successfully affects the afferent
loop of the immune response by eliciting TA-specific T cells
but cannot affect T cell activation at the receiving end [51,52].
The cancer specificity of TLR agonists consists of the prefer-
ential attraction of TLR-7 expressing pDCs to chronically
inflamed tissues and their enhanced recruitment [53]. Simi-
lar conclusions were recently reached by Torres et al. [54],
who followed the biological events induced by imiquimod
when administered to patients with actinic keratosis. Thus,
TLR agonists exemplify how the gap between the induction of
TA-specific T cells by immunization and their activation at
the receiving end could be closed. It is thus conceivable that
preparations of TLR agonists suitable for systemic adminis-
tration may be used in the future as single agent therapy for
other tumor types (trials are currently ongoing in Europe for
melanoma) or as adjuvants to enhance the effectiveness of
active-specific immunization approaches [55-57].
Conclusion
This study stands as a proof of principle that, when tissues are
easily accessible, mechanistic observation about the effects of
a treatment can be easily performed in humans by combining
minimally invasive techniques (fine needle aspirates, through
cut or punch biopsies) with high-fidelity mRNA amplifica-
tion; such approaches are fundamental to refresh scientific
hypotheses through direct human observation. Second, it
provides insights into the early events leading to tumor
rejection in a most powerful human model. Finally, it sug-
gests that immune-mediated tumor rejection is only one
aspect of tissue-specific destruction, which follows a constant
immunological pathway shared by other anti-cancer immu-
notherapies, acute allograft rejection, autoimmune disease
and tissue damage during chronic pathogen infections.
Materials and methods
Detailed methods are available as Additional data file 2.
Study design and patient information
This double-blind, placebo-controlled, randomized, parallel
group clinical trial sponsored by 3M Pharmaceuticals and
registered before patient enrollment (3M/NNMC study
#1454-IMIQ) was designed to evaluate the early transcrip-
tional events induced by topical imiquimod administration.
The trial was conducted at the National Naval Medical Center
(Bethesda, MD, USA) in compliance with the Code of Federal
Regulations and the guidelines for Good Clinical Practice.
Imiquimod (5%, 12.5 mg) or vehicle cream were supplied in
single-use 250 mg sachets. Following biopsy confirmation
and time for healing, subjects applied a sufficient quantity of
cream to cover the entire BCC and an area approximately 2
cm around. Each dose was left on the skin for eight hours. For
the study, 48 subjects were supposed to be randomized in a
2:1 ratio to either imiquimod or vehicle within each of 4 dos-
ing regimens (q12 hours for 2 or 4 days or q24 hours for 4 or
8 days). Subjects were randomized at the time of screening
when the pre-enrollment biopsy was taken. Once eligibility
was determined based on the biopsy result, the investigator
contacted the subject, who either started treatment on a date
instructed by the investigator or returned the study drug.
Replacement subjects were identified for all subjects with a
biopsy result negative for BCC or who discontinued prior to
EOT procedures. BCCs were to be a least 7 mm diameter and
were to be located on the scalp, face, trunk or proximal
extremities. Punch biopsies (PB; 2 mm diameter) were
obtained pre-enrollment to verify the diagnosis of BCC, pre-
treatment (PB1 and PB2) and at EOT (PB3 and PB4), approx-
imately 24 hours after the last dose taken. PB1 and PB3 were
transferred immediately at the bedside into cryovials with 2
μl Rnalater (Ambion, Austin, TX, USA), frozen in liquid nitro-
gen and stored at -80°C for total RNA isolation. PB2 and PB4
were placed in a cryomold, filled with OCT compound (Tis-
sue-Tek, Elkhart, IN, USA), frozen in liquid nitrogen and
stored at -80°C for immunohistochemistry (IHC).
RNA isolation and amplification and cDNA arrays
Total RNA was isolated with RNeasy minikits (Qiagen, Ger-
mantown, MD, USA) and amplified into anti-sense RNA as
previously described [41,58,59] with the following modifica-
tions to minimize RNA degradation by abundant skin
RNAases. Samples were homogenized in disposable tissue
grinders (Fisher Scientific, Lafayette, CO, USA). Proteins
potentially interfering with RNA isolation were removed by
incubating the homogenate in 590 μl distilled water and 10 μl
PROTEINASE K solution (Qiagen) at 55°C for 10 minutes
then centrifuged at ambient temperature for 3 minutes.
Supernatants were combined with 0.5 volumes of ethanol
(96% to 100%) into a Rnase-Dnase free tube and RNA was
isolated through a RNeasy mini column. First strand cDNA
synthesis was accomplished in 1 μl SUPERase•In (Ambion)
and ThermoScript RT (Gibco BRL, Gaithersburg, MD, USA)
in 2 μg bovine serum albumin. RNA quality was verified by
Agilent technologies (Palo Alto, CA, USA). Anti-sense RNA
was used for probe preparation or quantitative real-time PCR
(qPCR). For microarray analysis, test samples were labeled
with Cy5-dUTP (Amersham, Piscataway, NJ, USA) and co-
hybridized with reference pooled normal donor PBMCs
labeled with Cy3-dUTP to custom made 7 K-cDNA microar-
rays [60]. Arrays were scanned on a GenePix 4000 (Axon
Instruments, Union City, CA, USA) and analyzed using Clus-
γ
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.13
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
ter and Tree View software [61]. Gene ratios are presented
according to the central method for display [62]. Gene anno-
tations were mined using web-based tools such as DAVID
[63], GeneCards [64], COPE [65] and Bioinformatic Har-
vester [66].
Quantitative PCR
QPCR was applied to detect the expression of IFN-α, IFN-γ,
TNF-α and MCP-1 using an ABI Prism 7900 HT (Applied Bio-
systems, Foster City, CA, USA). Primers and probes were cus-
tom-designed to span exon-intron junctions and generate
<150 base-pair amplicons (Biosource, Camarillo, CA, USA).
Taqman probes were labeled at the 5' and 3' ends with the
reporter FAM (6-carboxyfluorescein; emission λ
max
= 518
nm) and the quencher TAMRA (6-carboxytetramethylrhod-
amine; emission λ
max
= 582 nm), respectively. Standard
curves were based on amplicons generated from human leu-
kocyte antigen (HLA)-A*0201 expressing lymphocytes stim-
ulated with IL-2 (300 IU/ml) and Flu M1:58-66 peptide; copy
numbers were estimated with Oligo Calculator [67]. Linear
regression R
2
-values pertinent to all standard curves were ≥
0.98. QPCR reactions were conducted in a 20 μl volume,
including 1 μl cDNA, 1× Taqman Master MIX (Applied Bio-
systems), 2 μl of 20 μM primer and 1 μl of 12.5 μM probe.
Thermal cycler parameters included 2 minutes at 50°C, 10
minutes at 95°C and 40 cycles involving denaturation at 95°C
for 15 s, annealing-extension at 60°C for 1 minute. The 2
-ΔΔCT
method was utilized to compute fold change in gene expres-
sion at EOT relative to baseline after normalization according
to cyclophilin G expression [68].
Immunohistochemistry
After confirming the presence of epidermis, dermis and
tumor using hematoxylin and eosin, IHC was performed by
staining 7 mm consecutive acetone-fixed sections for the
expression of CD4, CD8, CD56, CD95, FasL, granzyme A and
B, perforin, BCL-2, TRAIL, caspase 3 and PARP. Secondary
staining consisted of biotinylated goat-anti-mouse IgG fol-
lowed by avidin-biotin-peroxidase. A semi-quantitative esti-
mation was conducted to separate histological entities as:
tumor cells; intra-tumoral immune infiltrate; and peri-
tumoral immune infiltrate. Scoring was assigned independ-
ently by two blinded pathologists (AA and AF) as: 0 (none), 1+
(few), 2+ (moderate), 3+ (numerous). Data are presented as
shift in scores at EOT compared to baseline.
Statistics
Significance testing was based on paired or 2-sample two-
tailed Student t-test as appropriate. P values < 0.05 were con-
sidered statistically significant. No adjustment was made for
multiple comparisons. Fisher exact test was used to test the
level of significance comparing the frequency of events
between treatment groups. All analyses related to class com-
parison and class prediction was done using the BRB-Array
Tools [69] developed by Simon [70]. Microarray raw data
were curated according to GEO (series # GSE5121) [71].
Additional data files
The following additional data are available with the online
version of this paper. Additional data file 1 provides a list of
genes previously shown to be associated with the stimulation
of various cell types with IFN-α. Additional data file 2 is an
extended version of the Materials and methods, providing full
disclosure of the methodology used. Additional data file 3
provides a complete list of the 637 genes specifically induced
by imiquimod treatment based on the statistical approach
presented in the text. Additional data file 4 is a diagram
illustrating the mining strategy that was implemented for the
preparation of Figure 1c,d.
Additional data file 1List of genes previously shown to be associated with the stimulation of various cell types with IFN-αList of genes previously shown to be associated with the stimulation of various cell types with IFN-α.Click here for fileAdditional data file 2Extended version of the Materials and methods, providing full dis-closure of the methodology usedExtended version of the Materials and methods, providing full dis-closure of the methodology used.Click here for fileAdditional data file 3Complete list of the 637 genes specifically induced by imiquimod treatment based on the statistical approach presented in the textComplete list of the 637 genes specifically induced by imiquimod treatment based on the statistical approach presented in the text.Click here for fileAdditional data file 4The mining strategy that was implemented for the preparation of Figure 1c,dThe mining strategy that was implemented for the preparation of Figure 1c,d.Click here for file
References
1. Schulze HJ, Cribier B, Requena L, Reifenberger J, Ferrandiz C, Garcia
Diez A, Tebbs V, McRae S: Imiquimod 5% cream for the treat-
ment of superficial basal cell carcinoma: results from a ran-
domized vehicle-controlled phase III study in Europe. Br J
Dermatol 2005, 152:939-947.
2. National Cancer Institute: Drug Information Summaries
[http://www.cancer.gov/clinicaltrials/developments/newly-approved-
treatments]
3. Stockfleth E, Trefzer U, Garcia-Bartels C, Wegner T, Schmook T,
Sterry W: The use of Toll-like receptor-7 agonist in the treat-
ment of basal cell carcinoma: an overview. Br J Dermatol 2003,
149(Suppl 66):53-56.
4. Urosevic M, Maier T, Benninghoff B, Slade H, Burg G, Dummer R:
Mechanisms unerlying imiquimod-induced regression of
basal cell carcinoma in vivo. Arch Dermatol 2003, 139:1325-1332.
5. Stanley MA: Imiquimod and the imidazoquinolones: mecha-
nism of action and therapeutic potential. Clin Exp Dermatol
2002, 27:571-577.
6. Naylor MF, Crowson N, Kuwahara R, Teague K, Garcia C, Mackinnis
C, Haque R, Odom C, Jankey C, Cornelison RL: Treatment of len-
tigo maligna with topical imiquimod. Br J Dermatol 2003,
149(Suppl 66):66-70.
7. Ray CM, Kluk M, Grin CM, Grant-Kels JM: Successful treatment
of malignant melanoma in situ with topical 5% imiquimod
cream. Int J Dermatol 2005, 44:428-434.
8. Hurwitz DJ, Pincus L, Kupper TS: Imiquimod: a topically applied
link between innate and acquired immunity. Arch Dermatol
2003, 139:1347-1350.
9. Majewski S, Marczak M, Mlynarczyk B, Benninghoff B, Jablonska S:
Imiquimod is a strong inhibitor of tumor cell-induced
angiogenesis. Int J Dermatol 2005, 44:14-19.
10. KEGG PATHWAY Database [http://www.genome.ad.jp/kegg/
pathway.html]
11. Stroncek DF, Basil C, Nagorsen D, Deola S, Arico E, Smith K, Wang
E, Marincola FM, Panelli MC: Delayed polarization of mononu-
clear phagocyte transcriptional program by type I interferon
isoforms. J Transl Med 2005, 3:
24.
12. KEGG: Toll-like Receptor Signaling Pathways [http://
www.genome.jp/dbget-bin/www_bget?path:hsa04620]
13. van Spriel AB, Puls KL, Sofi M, Pouniotis D, Hochrein H, Orinska Z,
Knobeloch KP, Plebanski M, Wright MD: A regulatory role for
CD37 in T cell proliferation. J Immunol 2004, 172:2953-2961.
14. Jin P, Wang E, Provenzano M, Deola S, Selleri S, Voiculescu S, Stron-
cek DF, Panelli MC, Marincola FM: Molecular signatures induced
by interleukin-2 on peripheral blood mononuclear cells and
T cell subsets. J Transl Med 2006, 4:26.
15. Schluns KS, Stoklasek T, Lefrancois L: The roles of interleukin-15
receptor alpha: trans-presentation, receptor component, or
both? Int J Biochem Cell Biol 2005, 37:1567-1571.
16. Zambricki E, Shigeoka A, Kishimoto H, Sprent J, Burakoff S, Carpen-
ter C, Milford E, McKay D: Signaling T-cell survival and death by
IL-2 and IL-15. Am J Transplant 2005, 5:2623-2631.
17. Sanchez-Sanchez N, Riol-Blanco L, de la Rosa G, Puig-Kroger A, Gar-
cia-Bordas J, Martin D, Longo N, Cuadrado A, Cabanas C, Corbi AL,
et al.: Chemokine receptor CCR7 induces intracellular signal-
ing that inhibits apoptosis of mature dendritic cells. Blood
2004, 104:619-625.
R8.14 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. http://genomebiology.com/2007/8/1/R8
Genome Biology 2007, 8:R8
18. Fernandez N, Renedo M, Garcia-Rodriguez C, Sanchez Crespo M:
Activation of monocytic cells through Fc gamma receptors
induces the expression of macrophage-inflammatory pro-
tein (MIP)-1 alpha, MIP-1 beta, and RANTES. J Immunol 2002,
169:3321-3328.
19. Bade B, Boettcher HE, Lohrmann J, Hink-Schauer C, Bratke K, Jenne
DE, Virchow JC Jr, Luttmann W: Differential expression of the
granzymes A, K and M and perforin in human peripheral
blood lymphocytes. Int Immunol 2005, 17:1419-1428.
20. Adrain C, Murphy BM, Martin SJ: Molecular ordering of the cas-
pase activation cascade initiated by the cytotoxic T lym-
phocyte/natural killer (CTL/NK) protease granzyme B. J Biol
Chem 2005, 280:4663-4673.
21. Veugelers K, Motyka B, Frantz C, Shostak I, Sawchuk T, Bleackley RC:
The granzyme B-serglycin complex from cytotoxic granules
requires dynamin for endocytosis. Blood 2004, 103:3845-3853.
22. Brossard C, Semichon M, Trautmann A, Bismuth G: CD5 inhibits
signaling at the immunological synapse without impairing its
formation. J Immunol 2003, 170:4623-4629.
23. Choudhuri K, Kearney A, Bakker TR, van der Merwe PA: Immunol-
ogy: how do T cells recognize antigen? Curr Biol 2005,
15:R382-R385.
24. Hahn WC, Menu E, Bothwell AL, Sims PJ, Bierer BE: Overlapping
but nonidentical binding sites on CD2 for CD58 and a second
ligand CD59. Science 1992, 256:1805-1807.
25. Badour K, Zhang J, Siminovitch KA: The Wiskott-Aldrich syn-
drome protein: forging the link between actin and cell
activation. Immunol Rev 2003, 192:98-112.
26. Posselt AM, Vincenti F, Bedolli M, Lantz M, Roberts JP, Hirose R:
CD69 expression on peripheral CD8 T cells correlates with
acute rejection in renal transplant recipients. Transplantation
2003, 76:190-195.
27. Lanier LL:
NK cell recognition. Annu Rev Immunol 2005,
23:225-274.
28. Wex T, Buhling F, Wex H, Gunther D, Malfertheiner P, Weber E,
Bromme D: Human cathepsin W, a cysteine protease pre-
dominantly expressed in NK cells, is mainly localized in the
endoplasmic reticulum. J Immunol 2001, 167:2172-2178.
29. Fujii N, Hiraki A, Ikeda K, Ohmura Y, Nozaki I, Shinagawa K, Ishimaru
F, Kiura K, Shimizu N, Tanimoto M, Harada M: Expression of minor
histocompatibility antigen, HA-1, in solid tumor cells. Trans-
plantation 2002, 73:1137-1141.
30. Sullivan TP, Dearaujo T, Vincek V, Berman B: Evaluation of super-
ficial basal cell carcinomas after treatment with imiquimod
5% cream or vehicle for apoptosis and lymphocyte
phenotyping. Dermatol Surg 2003, 29:1181-1186.
31. Iwasaki A, Medzhitov R: Toll-like receptor control of the adap-
tive immune responses. Nat Immunol 2004, 5:987-995.
32. Hemmi H, Kaisho T, Takeuchi O, Sato S, Sanjo H, Hoshino K, Hori-
uchi T, Tomizawa H, Takeda K, Akira S: Small anti-viral com-
pounds activate immune cells via the TLR7 MyD88-
dependent signaling pathway. Nat Immunol 2002, 3:196-200.
33. Chan CW, Crafton E, Fan HN, Flook J, Yoshimura K, Skarica M,
Brockstedt D, Dubensky TW, Stins MF, Lanier LL, et al.: Interferon-
producing killer dendritic cells provide a link between innate
and adaptive immunity. Nat Med 2006, 12:207-213.
34. Marincola FM, Wang E, Herlyn M, Seliger B, Ferrone S: Tumors as
elusive targets of T cell-directed immunotherapy. Trends
Immunol 2003, 24:335-342.
35. Bowen DG, Walker CM: Adaptive immune responses in acute
and chronic hepatitis C virus infection. Nature 2005,
436:946-952.
36. Smith MW, Yue ZN, Korth MJ, Do HA, Boix L, Fausto N, Bruix J, Car-
ithers RL Jr, Katze MG: Hepatitis C virus and liver disease: glo-
bal transcriptional profiling and identification of potential
markers. Hepatology 2003, 38:
1458-1467.
37. Sarwal M, Chua MS, Kambham N, Hsieh SC, Satterwhite T, Masek M,
Slavatierra O Jr: Molecular heterogeneity in acute renal allo-
graft rejection identified by DNA microarray profiling. N Engl
J Med 2003, 349:125-138.
38. Panelli MC, Wang E, Phan G, Puhlman M, Miller L, Ohnmacht GA,
Klein H, Marincola FM: Genetic profiling of peripheral mononu-
clear cells and melanoma metastases in response to sys-
temic interleukin-2 administration. Genome Biol 2002,
3:RESEARCH0035.
39. Kim SH, Han SY, Azam T, Yoon DY, Dinarello CA: Interleukin-32:
a cytokine and inducer of TNFalpha. Immunity 2005,
22:131-142.
40. Hicks AM, Riedlinger G, Willingham MC, Alexander-Miller MA, von
Kap-Herr C, Pettenati MJ, Sanders AM, Weir HM, Du W, Kim J, et al.:
Transferable anticancer innate immunity in spontaneous
regression/complete resistance mice. Proc Natl Acad Sci USA
2006, 103:7753-7758.
41. Wang E, Miller LD, Ohnmacht GA, Liu ET, Marincola FM: High fidel-
ity mRNA amplification for gene profiling. Nat Biotechnol 2000,
18:457-459.
42. Netea MG, Azam T, Ferwerda G, Girardin SE, Walsh M, Park JS,
Abraham E, Kim JM, Yoon DY, Dinarello CA, Kim SH: IL-32 syner-
gizes with nucleotide oligomerization domain (NOD) 1 and
NOD2 ligands for IL-1beta and IL-6 production through a
caspase 1-dependent mechanism. Proc Natl Acad Sci USA 2005,
102:16309-16314.
43. Monsurro V, Wang E, Yamano Y, Migueles SA, Panelli MC, Smith K,
Nagorsen D, Connors M, Jacobson S, Marincola FM: Quiescent
phenotype of tumor-specific CD8+ T cells following
immunization. Blood 2004, 104:1970-1978.
44. Wang E, Panelli MC, Zavaglia K, Mandruzzato S, Hu N, Taylor PR,
Seliger B, Zanovello P, Freedman RS, Marincola FM: Melanoma-
restricted genes. J Transl Med 2004, 2:34.
45. Richwald GA: Imiquimod. Drugs Today (Barc) 1999, 35:497-511.
46. Dahl MV: Imiquimod: a cytokine inducer.
J Am Acad Dermatol
2002, 47:S205-S208.
47. Matzinger P: Introduction to the series. Danger model of
immunity. Scand J Immunol 2001, 54:2-3.
48. Panelli MC, Riker A, Kammula U, Wang E, Lee KH, Rosenberg SA,
Marincola FM: Expansion of tumor-T cell pairs from fine nee-
dle aspirates of melanoma metastases. J Immunol 2000,
164:495-504.
49. Kammula US, Marincola FM, Rosenberg SA: Real-time quantita-
tive polymerase chain reaction assessment of immune reac-
tivity in melanoma patients after tumor peptide vaccination.
J Natl Cancer Inst 2000, 92:1336-1344.
50. Fuchs EJ, Matzinger P: Is cancer dangerous to the immune
system? Semin Immunol 1996, 8:271-280.
51. Monsurro V, Wang E, Panelli MC, Nagorsen D, Jin P, Katia Z, Smith
K, Ngalame Y, Even J, Marincola FM: Active-specific immuniza-
tion against melanoma: is the problem at the receiving end?
Sem Cancer Biol 2003, 13:473-480.
52. Marincola FM: A balanced review of the status of T cell-based
therapy against cancer. J Transl Med 2005, 3:16.
53. Urosevic M, Dummer R, Conrad C, Beyeler M, Laine E, Burg G, Giliet
M: Disease-independent skin recruitment and activation of
plasmacytoid predendritic cells following imiquimod
treatment. J Natl Cancer Inst 2005, 97:1143-1153.
54. Torres A, Storey L, Anders M, Miller RL, Bulbulian BJ, Jin J, Raghavan
S, Lee J, Slade HB, Birmachu W: Immune-mediated changes in
actinic Keratosis following topical treatment with Imiqui-
mod 5% cream. J Transl Med 2007 in press.
55. Rechtsteiner G, Warger T, Osterloh P, Schild H, Radsak MP: Cutting
edge: priming of CTL by transcutaneous peptide immuniza-
tion with imiquimod. J Immunol 2005, 174:2476-2480.
56. Wenzel J, Uerlich M, Haller O, Bieber T, Tueting T: Enhanced type
I interferon signaling and recruitment of chemokine recep-
tor CXCR3-expressing lymphocytes into the skin following
treatment with the TLR7-agonist imiquimod.
J Cutan Pathol
2005, 32:257-262.
57. Craft N, Bruhn KW, Nguyen BD, Prins R, Lin JW, Liau LM, Miller JF:
The TLR7 agonist imiquimod enhances the anti-melanoma
effects of a recombinant Listeria monocytogenes vaccine. J
Immunol 2005, 175:1983-1990.
58. Wang E, Marincola FM: Amplification of small quantities of
mRNA for transcript analysis. In DNA arrays: A Molecular Cloning
Manual Edited by: Bowtell D, Sambrook J. Cold Springs Harbor, NY:
Cold Spring Harbor Laboratory Press; 2002:204-213.
59. Wang E: RNA amplification for successful gene profiling
analysis. J Transl Med 2005, 3:28.
60. National Cancer Institute, Center for Cancer Research:
mAdb GAL Files (Hs-CCDTM17 5 k-1px gal) [http://nci
array.nci.nih.gov/gal_files/gal_custom_current.shtml]
61. Eisen MB, Spellman PT, Brown PO, Botstein D: Cluster analysis
and display of genome-wide expression patterns. Proc Natl
Acad Sci USA 1998, 95:14863-14868.
62. Ross DT, Scherf U, Eisen MB, Perou CM, Rees CA, Spellman PT, Iyer
V, Jeffrey SS, van de Rijn M, Waltham M, et al.: Systematic variation
in gene expression patterns in human cancer cell lines.
Nature Genetics 2000, 24:227-235.
http://genomebiology.com/2007/8/1/R8 Genome Biology 2007, Volume 8, Issue 1, Article R8 Panelli et al. R8.15
comment reviews reports refereed researchdeposited research interactions information
Genome Biology 2007, 8:R8
63. DAVID Bioinformatics Resource 2006 [http://
david.abcc.ncifcrf.gov/]
64. GeneCards Database [http://www.genecards.org/index.shtml]
65. COPE: Cytokines and Cells Online Pathfinder Encyclopedia
[http://www.copewithcytokines.de]
66. Bioinformatic Harvester Database [http://harvester.embl.de/
harvester/P599/P59901.htm]
67. Biomedical Research Support Facilities, University of
Pittsburgh [http://www.pitt.edu/~rsup/OligoCalc.html]
68. Livak KJ, Schmittgen TD: Analysis of relative gene expression
data using real-time quantitative PCR and the 2(-Delta Delta
C(T)) Method. Methods 2001, 25:402-408.
69. National Cancer Institute, Biometric Research Branch: BRB
Array Tools [http://linus.nci.nih.gov/BRB-ArrayTools.html]
70. Simon R, Lam A, LI MC, Ngan M, Menenzes S, Zhao Y: Analysis of
gene expression data using BRB-Array Tools. Cancer
Informatics 2007 in press.
71. NCBI: Gene Expression Omnibus (GEO) [http://
www.ncbi.nlm.nih.gov/geo/]