| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1,
,§,,¶
* Institut du Cancer de Montreal,
Centre de Recherche du CHUM 1560 Sherbrooke E., Montréal, Qc, H2L 4M1, Canada;
Départment de Médecine, Université de Montréal, and Departments of
§ Surgery and
¶ Oncology, McGill University, Montréal, Qc, H3G 1Y6, Canada
1Correspondence: Montreal General Hospital Research Institute, Rm. L12–132, 1650 Cedar Ave., Montréal, Canada H3G 1A4. E-mail: mbem{at}musica.mcgill.ca
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: lung cancer • retinoic acid receptor ß • tumor suppression • cDNA array
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and animal (2)
studies to have tumor
suppressive effects. These effects are mediated by two families of
nuclear RA receptors (RARs), each comprising three genes (RAR
, ß,
and 
as well as RXR , ß, and ; for review, see ref
3
). These receptors are transcription factors that
specifically bind RA-responsive elements (RAREs) in the promoters of
genes whose expression they control. Each gene codes for at least two
functional isoforms (3)
; of these receptors, it is RARß,
and more specifically the isoform ß2, that has been implicated most
frequently in suppression of epithelial cancers. For example, it has
been shown that a majority of lung, breast, and other tumor-derived
cell lines as well as the tumors themselves no longer express RARß
whereas expression is detected in the corresponding normal tissues
(4
5
6
7
8
9
10)
. In addition, its forced re-expression in
RARß- lines reduces or eliminates various
aspects of the tumor phenotype, including anchorage independence, focus
formation and growth rate in vitro, and tumor formation in
nude mice (reviewed in ref 11
). Transgenic mice expressing
antisense or other constructs that down-regulate RARß2 develop lung
and breast cancer (12
, 13)
. RAR appears to mediate
growth rate suppression in some breast tumor lines by modulating RARß
expression (14)
, but otherwise there is little indication
that the other RA receptors contribute to epithelial tumor suppression.
To understand the mechanism of RARß2-mediated tumor suppression, we
have searched for genes regulated either directly or indirectly by
RARß2 that could be implicated in this suppression. We considered it
important to compare lines that were as similar as possible, so we made
use of a panel of RAR-expressing derivatives of the Calu-1 cell line,
in which RARß2 tumor suppression has been well characterized
(15)
. We then compared patterns of gene expression using
the Atlas human cDNA array I and confirmed the differences for six of
the cDNAs by an independent assay. In addition to RARß itself, the
expression levels of 27 cDNAs were reproducibly affected and 3 other
genes were up-regulated by treatment with RA in a non-RARß-specific
fashion. A striking characteristic of the results is the concerted
nature of the effect the alterations are predicted to have on tumor
behavior. Thus, for example, we found increased expression levels of
several genes that regulate the immunogenicity of the tumor cell, most
notably intercellular adhesion molecule 1 (ICAM-1), major
histocompatibility complex (MHC) class I, and interleukin 1ß
(IL-1ß). Since the consequences of down-regulation of these would be
the well-documented crippling of the immune system’s capacity to
detect and kill the nascent tumor, we hypothesized a role for RARß in
this function. To test this, we first confirmed that cell surface
expression of the proteins was increased, and then demonstrated that
the consequence of these changes was increased alloantigen-specific CTL
response in both the induction and effector phases. These results have
implications for understanding the nature of tumor evasion of immune
surveillance.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The
RARß1-transfected line (Cß10) was obtained by following exactly the
same protocol, using the pSVL vector carrying the RARß1 cDNA
sequences (A. Toulouse et al., unpublished results). The
RARß2+ derivative of SK-MES, named C102, was
obtained by transfection with the RARß2 expression vector as
described previously (15)
. Where indicated, cells were
treated with pharmacological doses (100 nM) of all-trans
retinoic acid (Sigma, St. Louis, Mo.) prior to collection.
RNA preparation
All cells were cultured to 70% confluency, pelleted, and stored
at -80°C for subsequent RNA extraction using the LiCl-urea method
(16)
. Poly A+ mRNA was prepared using Qiagen’s Oligotex
mRNA midi kit (Mississauga, Canada) and its quality was assessed
following the manufacturer’s instructions.
Atlas human cDNA expression array
The Atlas human cDNA array kit was purchased from Clontech
Laboratories (Palo Alto, Calif.). All procedures for labeling and
purifying the probes were accomplished by following manufacturer’s
recommendations. Complex -32P-dCTP-labeled
cDNA probes were generated by reverse transcription of mRNA from
untreated Calu-1 cells as well as C30 (neor), C24
(RARß2+), and C64
(RARß2+) treated with RA for 48 h. The
probes were purified by column chromatography (ChromaSpin) and met or
exceeded the manufacturer’s recommendation for specific activity. The
membranes were hybridized in ExpressHyb solution overnight at 68°C,
washed twice (20 min with 0.1x SSC and 0.5% sodium dodecyl sulfate),
and exposed for varying periods of time on Kodak XAR autoradiographic
films.
RNA analysis
Northern blot analysis was performed according to standard
protocols (17)
. The blot was probed using the insert of an
IL-1ß construct generated by amplification of a cDNA fragment
(primers: 5'-GCTGCTCTGGGATTCTCTTC-3', 5'-AGCACAGGACTCTCTGGGTA-3'),
which was digested with HindIII and AccI
prior to being cloned into pGEM-3Z. RNase protection assays were
performed using Ambion’s RPA II kit (Austin, Tex.) and the riboprobes
were synthesized following standard protocols (17)
. The
actin probe used as control was described elsewhere (15)
.
The ID-3 probe was prepared by amplification of a cDNA fragment
overlapping nucleotides 342 to 751 (primers: 5'-GCACCTCTGGACTCACTC-3',
5'-TGGAGGTGTCAGGACACG-3'). The polymerase chain reaction (PCR) product
was digested with SmaI; a 409 bp fragment was isolated and
cloned into the SmaI site of pGEM-3Z (Promega, Madison,
Wis.). The plasmid was linearized using HindIII and the
probe was synthesized using T7 RNA polymerase. The GADD-45 probe was
obtained from Dr. P. A. Dion.
Immunological reagents
Lympholite H and phycoerythrin-conjugated goat anti-mouse IgG1
were purchased from Cedarlane (Hornby, Ontario, Canada).
FITC-conjugated mAb (clone W6/32) against the conserved region of the
human HLA class I antigen was obtained from Sigma (Oakville, Ontario,
Canada), and mAb against ICAM-1 (clone 8.4A6) was from BioSource
International (Camarillo, Calif.). Chromium-51 was purchased from ICN
(Costa Mesa, Calif.).
Flow cytometry
One-color flow cytometry analysis was performed as follows.
Where indicated, the cells were treated with RA at 100 nM for 5 or 7
days. To detect MHC class I proteins, 106 cells
were incubated with 2 µg of FITC-conjugated monoclonal antibody W6/32
against human MHC class I antigen for 30 min on ice. The cells were
then washed with phosphate-buffered saline (PBS) and fixed in 1%
paraformaldehyde in PBS. To detect ICAM-1 protein expression,
106 cells were incubated with 1 µg of
monoclonal antibody 8.4A6 for 30 min on ice and washed with PBS. These
cells were then incubated with 600 ng of a goat anti-mouse IgG-1
antibody conjugated to R-phycoerythrin. The samples were read on a
Profile I flow cytometer (Coulter, Burlington, Ontario, Canada).
Two-color cytometry was performed in the same way except that labeling
with the two antibodies conjugated with the respective fluorochromes
was done simultaneously. All cytometric determinations were performed
at least twice.
Cytotoxic lymphocytes (CTL) assay
Peripheral blood mononuclear cells (PBMC) were prepared as
described previously (18)
. The Calu-1 derivatives C30 and
C64 were cultured in either the presence or absence of RA (100 nM) for
5 days, then treated with mitomycin C. These mitomycin C-treated cells
were used as stimulator cells for PBMC in a 5 day alloreactive culture
in the absence of exogenous RA (18)
. The culture was
performed in 24-well plates with 2 x 106
PBMC/ml and 2 x 106 stimulators/ml in a
final volume of 2 ml/well. IL-2 (10 U/ml) was added at the beginning of
the culture. After 5 days, the cells were washed and recounted. Fresh
C30 and C64 cells that had been treated with 100 nM RA for 5 days prior
to the assay were labeled with 51Cr and used as
targets in a standard 4 h 51Cr release assay
as described in a previous publication (19)
. The ratios of
effector/target cells were set at 30:1, 10:1, 3:1, and 1:1. A fixed
number of 15,000 target cells/well was used for all the determinations.
Samples were in triplicate. The percentage of lysis was calculated as
 |
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Nine additional genes that are thought to be constitutively expressed
are displayed on the bottom row, G. Three negative controls include
lambda, M13, and plasmid DNA, also in row G, and genomic DNA is arrayed
along the right-hand and bottom borders to allow assessment of
uniformity of hybridization. The cDNAs present on the membrane have
been shown to be expressed in a regulated fashion and are involved in
key cellular processes. Hybridizations were performed on four separate
membranes, with probes generated from Calu-1 (a RARß-deficient cell
line) cells grown in the absence of added RA and three transfected
derivatives grown in the presence of 100 nM RA for 48 h: C24, C64
(both RARß2-expressing), and C30 (a
neor-derivative), which we hereafter call C24-RA,
C64-RA, and C30-RA. The latter was chosen as the RARß-deficient line
to be grown in the presence of RA because we have established that the
parental Calu-1 line occasionally expresses trace levels of RARß when
grown in RA-supplemented medium, whereas C30 does not (data not shown).
Previous work (11)
had shown that the RNA fingerprints of
these four lines, as well as two other lines derived from Calu-1 by
transfection of RARß1, varied from one line to another in less than
0.5% of their transcripts as determined by arbitrarily primed reverse
transcription PCR (20)
, so the probes were comparable.
|
Uniformity of hybridization was demonstrable by equality of intensity
of hybridization of the genomic DNA spots on the extreme right border
and the bottom row of section G on the original films. The negative
controls built into the membrane in row G were all completely negative
on all membranes (Fig. 1)
. To accommodate differences in specific
activities of the probes used, the membranes were exposed for varying
lengths of time so that meaningful comparisons could be made among all
the membranes. A total of 71% of the cDNAs (428/597) were detectable
after a 4 day exposure of the C64-RA-hybridized membrane (results
available on request from W. E. C. Bradley), with somewhat
lower numbers for the other membranes due to lower specific activity of
the probes.
To evaluate the array results, spot densitometries were performed using
the NIH Image software on the membranes probed with Calu-1, C30-RA, and
C64-RA. Since the probe derived from C24-RA was weaker, densitometry of
the corresponding array was less informative, and this array was used
primarily to confirm by visual inspection results from the other
arrays. Densitometric values for each spot were normalized to the mean
value of the genomic DNA control spots on the membrane. Relative ratios
were then calculated between the membranes (C64-RA/Calu-1 and
C64-RA/C30-RA). Spots that were consistently changed in both ratios are
reported in Table 1
. Some of the changes are of the order of 1.5 to 2.5; these may be of
marginal physiological significance, but are included for completeness
(see Discussion).
|
A measure of the confidence one can place in these results is the
comparison of their consistency between the two comparisons of
RARß+ vs. RARß-derived probes. A total of 28 cDNAs
(including RARß) were found to be altered in a consistent manner,
with 21 being specifically up-regulated and 7 down-regulated in the
presence of RARß2 (Table 1)
. Only three cDNAs varied in intensity in
a way, suggesting regulation by RA rather than specifically by RARß2
(most probably by other RARs present in the cells), namely, TNF
receptor 1 (spot C1m), FAS ligand (spot C4m), and ID-2 (spot D1g)
(Table 1)
. The ratios for these were close to 1.0 in C64-RA/C30-RA
while being two- to threefold higher in the C64-RA/Calu-1 comparison,
which suggests up-regulation in the RA-treated cells. A limited number
of spots varied in what we interpret to be a clonal fashion, the
intensities varying on the three membranes but not in a way suggestive
of RA or RARß2 regulation (Table 1)
. In addition, the results for 15
of the spots (in general, the more intense ones; identified in Table 1
)
were confirmed using the fourth membrane probed with RA-treated C24
cDNA. In no case was a discordance seen between this set of results and
the others.
As expected, the expressed housekeeping controls in row G are
also of similar relative intensity from one cell line to the other,
with the striking exception of MHC class I (G14, see below), which is
up-regulated by at least threefold in the RARß2-positive line
(confirmed with the C24-RA-probed array). To further assess the
dependability of the array results, some genes listed in Table 1
have
been tested by RNase protection or Northern blot for relative
expression levels in a panel of RARß1-positive (Cß10),
RARß2-positive (C24 and C64), and RARß-deficient (C30) derivatives
of Calu-1 cells. RARß itself was detected in the C64-RA and C24-RA
probed arrays (albeit at a low intensity) but was not visible in the
two arrays probed with cDNA from the RARß-deficient cells (Table 1)
,
and this is concordant with RNase protection and PCR assays (ref
15
and data not shown). Probes were also generated for
several other genes that gave differences on both sets of arrays
(GADD-45, ID-3, ICAM-1, and IL1-ß, corresponding to C7f, D1d, E5h,
and F5m, respectively). The results of RNase protection and Northern
analyses of message levels of these genes (Table 2
and results not shown) confirmed their differential expression. Only
slight increases in the levels of expression were seen upon RA
treatment of RARß2-deficient cell lines (Calu-1, C30, and Cß10).
However, in the case of IL-1ß and ID-3, transfection by RARß2
resulted in a substantial RA response of about threefold or more. For
GADD-45, an increase was seen in both RARß2-transfected lines in
comparison to Calu-1 and C30 in absence of retinoic acid, but no
further increase was seen upon addition of exogenous RA. It may be that
the physiological levels of RA present in the cells were sufficient to
mediate the activation of this gene, but we cannot confirm this yet.
|
The modulation of the MHC class I and ICAM-1 genes may be of importance
if we regard it in light of the theory of immune surveillance. The
products of these genes are involved in the presentation of foreign or
tumor antigens to cytotoxic cell precursors and in enhancing T cell and
antigen-presenting cell interaction (21
, 22)
. If
down-regulation of these molecules occurs in tumor cells, they will
have a better chance of evading the immune system since the signaling
strength of any tumor-specific antigen at the cell surface would
necessarily be compromised.
Since ICAM-1 and MHC class I can play these central roles in eliciting
an immune response only if they are displayed on the cell surface, it
was important to determine whether the RARß2-induced increases in
mRNAs were reflected at this level. We therefore assayed surface ICAM-1
and MHC class I by flow cytometry in two lung tumor lines and
their RARß+ transfectants. C30
(neor-transfectant,
RARß-) and C64
(RARß2+) cells were labeled as described in
Materials and Methods and analyzed by flow cytometry. The results in
Table 3
show that the mean fluorescence intensity (MFI) of ICAM-1 protein is
~threefold higher on the cell surface of the RARß2-transfected C64
cells than on the C30 cells. Treating the cells with 100 nM RA for 7
days increased the expression in both cell lines but the increase was
greater in C64 cells than in C30 cells, as demonstrated in Table 3
(2.7-fold vs. 1.7-fold). Table 3
also shows results obtained with
antibody W6/32 against MHC class I proteins. As seen for ICAM-1
protein, the RARß2-transfected C64 cells show slightly higher MHC
class I MFI than C30 cells prior to RA treatment. The 7 day incubation
in the presence of retinoic acid resulted in the induction of MHC class
I molecules in C64 cells, but no augmentation was seen with C30 cells
after the treatment (Table 3)
.
|
To determine whether the RARß-induced increase in expression of these
proteins could be generalized to other lines, SK-MES, which is stably
RARß deficient (5)
, was transfected as described
(15)
and a derivative, C102, was characterized and shown
to express RARß2 (results not shown). SK-MES and C102 were grown in
the presence or absence of RA (as above for Calu-1 derivatives) for 5
days and the proteins were assayed simultaneously by flow cytometry
(Table 3)
. Again, an increase of ~twofold in ICAM-1 was seen in
untreated C102 and the increase in MHC class I was more than threefold
compared to that in SK-MES cells. RA treatment increased levels of both
proteins to a similar extent (~30%) in both the parental and the
transfected cell lines. This increase is less pronounced than seen in
the Calu-1 RARß2+ cells but, as mentioned above for GADD-45, it is
possible that the basal level of bioavailable RA is sufficient in
SK-MES to stimulate expression of these genes. In any event, the
transfection of RARß2 had a similar effect on cell surface expression
of the two proteins in both cell lines. Furthermore, a similar effect
has been observed on transfection of RARß2 into breast and colon
cancer cell lines (J. Pappas and W. E. C. Bradley,
unpublished results), which supports the observations reported here and
suggests that this RARß2 effect applies to other cancers in addition
to lung.
To show that this increase in potentially immunogenic proteins on the
surface of the RARß2+ cells actually had a
physiological effect, we performed a CTL assay on these cells
(23)
. We first isolated PBMC from heterologous donors and
incubated them with either C30 or C64 stimulator cells that had been
treated with RA for 5 days prior to incubation with the PBMC (see
Materials and Methods). The PBMC were then incubated with
51Cr-labeled target cells (RA-pretreated C64
cells or, for some combinations, RA-pretreated C30 cells) and the
extent of lysis of these targets was determined. The C64
(RARß2-expressing) line was 2.5- to 3-fold more efficient than the
control line (C30) in inducing CTL effector cells (one experiment shown
in Fig. 2
) against a C64 target, as judged by chromium release. In addition, when
used as target cells, the RARß2+ line was 1.3-
to 2-fold more susceptible to lysis by C64 RA-stimulated CTL cells than
was the C30 RARß- line (Fig. 3
).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 14 had intensities that differed by at least
2.5-fold in one of the comparisons; of these, however, only one, VRP,
was down-regulated in the RARß-expressing lines. We also found that
three of the genes on the array were up-regulated by RA without
specific dependence on the RARß receptor, most probably through the
other RA receptors expressed in these cells. We also tested six of the
genes by an independent assay for RNA or protein level (Tables 2
and 3
and data not shown) and found up-regulation to levels similar to those
shown by the array experiments. Based on this high degree of
concordance, we are confident that the array results reflect genuine
regulatory effects of RARß2, at least in Calu-1 cells. The subsequent
demonstration of up-regulation at the protein level of two of the genes
in another cell line, SK-MES, increases our confidence that our results
can be generalized.
Consistent with the high level of reliability of the array, several of
the genes listed in Table 1
have previously been reported to be
regulated by RA or other retinoids—for example, GADD-45
(24)
, ICAM-1 (25)
, IL-1ß
(26)
—and the results presented here allow us to propose
RARß as the specific receptor mediating this regulation. Some other
genes have been found by others to be controlled specifically by RARß
(MHC class I and recently IGF-BP3) and our data confirm the involvement
of RARß2 as a regulator of these genes (27
, 28)
.
Of particular interest in the work presented here is the identification
of genes that may be the downstream effectors of the RARß2-mediated
tumor suppression. Table 1
shows the functions of the regulated genes,
and it is noteworthy that many of them are consistent with this role
for RARß2. These results highlight several potentially synergistic
pathways regulated by RARß2 that are involved in the control of cell
behavior. In addition to the two major group discussed below, the
expression of genes involved in repair (GADD-45), IGF signaling
(IGF-BP3), as well as certain cell cycle regulators (p14) would be
expected to contribute in a concerted fashion to suppress the
development of tumor cells (29)
. The genes identified in
this study as well as others yet to be identified will therefore
provide a better understanding of the mechanisms by which RARß2
promotes tumor suppression.
Of particular interest is that about a third (9/31) of the genes influence decision-making in the commitment to apoptosis, and another one-quarter are involved in provocation of an immune response. It is possible that the observed clustering of regulated genes into two groups reflects the selection of genes presented on the array; however, in light of the tumor suppressive properties of RARß2 and the modification of the CTL response (discussed below), we believe they represent some of the mechanisms involved in RARß2-mediated tumor suppression.
For the first of these major groups, genes that both promote and
inhibit apoptosis are up-regulated, but close inspection of the arrays
indicates that most of the latter are expressed at a very low level, so
the predominant effect of RARß2 may be to enhance the probability
that the cell will commit to apoptosis if other requirements are
fulfilled. Of particular interest within this group is the ID family of
inhibitors of differentiation and DNA binding proteins. All three
members on the array are up-regulated by RA, and in two of the
cases—ID-1 and ID-3—specifically by RARß2. In addition, the latter
two are expressed at substantial levels. These genes promote passage
from G1 to S-phase and have been shown to be potent inducers of
apoptosis (30)
. Their functions and expression patterns
overlap (31)
, suggestive of functional redundancy, so the
conclusion that RARß2 up-regulates them, if confirmed on analysis of
other lines, may point to an important mechanism of defense against
tumorigenesis.
The second group of genes that have similar function is that involved
in immune response. As documented above, two genes essential for
eliciting an effective cytotoxic immune response, MHC class I (G14) and
ICAM-1 (E5h), are up-regulated by RARß2. The results presented in
Table 3
and those from Pappas and Bradley (unpublished results) also
confirm that the effect of RARß on levels of the two surface proteins
is not restricted to one cell line. In addition, increased levels of
several inflammatory cytokines that enhance this response (IL-1ß,
MIP2, and IL-8) were seen. These are potent attractants and
activators of macrophages, granulocytes, and mast cells, and
consequently of the immune system’s ability to eliminate any lesions
(reviewed in ref 32
). In this context a reduced level of
expression of RARß would be hypothesized to weaken the overall immune
response by compromising several of its components.
When this was tested in a heterologous CTL assay, the results (Figs. 2
and 3)
showed that, at least as measured by this assay, the immune
response would indeed be weakened against RARß-deficient tumor cells.
RARß2-transfected cells were better stimulators and targets of the
CTLs, establishing a role for RARß2 in maintaining an effective
immune response. The potential implications of these results are of
major importance as they establish a physiological role for some of the
genes whose expression is influenced by RARß2. This physiological
role may be exploitable for therapeutic benefits if RARß2 expression
can be up-regulated by gene therapy or by simple administration of
retinoids, as has been shown in patients at risk (33
, 34)
.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
B (p50–p65) and retinoic acid receptor ß-retinoid X receptor ß heterodimers. Mol. Cell. Biol. 13,6157-6169This article has been cited by other articles:
|
|
 |
|
  Z. D. Liang, S. M. Lippman, T.-T. Wu, R. Lotan, and X.-C. Xu RRIG1 Mediates Effects of Retinoic Acid Receptor {beta}2 on Tumor Cell Growth and Gene Expression through Binding to and Inhibition of RhoA. Cancer Res., July 15, 2006; 66(14): 7111 - 7118. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. P. Simmons, M. L. Peach, J. R. Friedman, M. M.B. Green, M. C. Nicklaus, and L. M. De Luca Evidence that sequence homologous region in LRAT-like proteins possesses anti-proliferative activity and DNA binding properties: translational implications and mechanism of action Carcinogenesis, April 1, 2006; 27(4): 693 - 707. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Mocellin, E. Wang, M. Panelli, P. Pilati, and F. M. Marincola DNA Array-Based Gene Profiling in Tumor Immunology Clin. Cancer Res., July 15, 2004; 10(14): 4597 - 4606. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Zhuang, T. N. Faria, P. Chambon, and L. J. Gudas Identification and Characterization of Retinoic Acid Receptor {beta}2 Target Genes in F9 Teratocarcinoma Cells Mol. Cancer Res., June 1, 2003; 1(8): 619 - 630. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Sirchia, M. Ren, R. Pili, E. Sironi, G. Somenzi, R. Ghidoni, S. Toma, G. Nicolo, and N. Sacchi Endogenous Reactivation of the RAR{beta}2 Tumor Suppressor Gene Epigenetically Silenced in Breast Cancer Cancer Res., May 1, 2002; 62(9): 2455 - 2461. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. K. Blanchard, J. B. Moore, C. L. Green, and R. J. Cousins Inaugural Article: Modulation of intestinal gene expression by dietary zinc status: Effectiveness of cDNA arrays for expression profiling of a single nutrient deficiency PNAS, November 20, 2001; 98(24): 13507 - 13513. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
