Diet, Epigenetic Events, and Cancer Prevention Symposium

Diet, Epigenetic Events, and Cancer Prevention Symposium
September 26-27, 2007
Gaithersburg Marriott Washingtonian Center


Introductory Remarks
Sharon Ross, Ph.D., Program Director
Nutritional Science Research Group
Division of Cancer Prevention
National Cancer Institute

The Division of Cancer Prevention (DCP), National Cancer Institute (NCI), and the Office of
Dietary Supplements (ODS), Office of the Director, NIH, sponsored the symposium Diet,
Epigenetic Events, and Cancer Prevention, held September 26-27, 2007. This symposium
represented a continuation of a previous trans-Department of Health and Human Services
workshop on diet, methylation, and health, held in August, 2001. This earlier workshop gave
rise to a number of publications as well as Requests for Applications (RFAs) that led to the
funding of 10 projects in collaboration with ODS. Active funding opportunities are currently
active in the areas of diet, epigenetic events, and cancer prevention.

The goal of this meeting was to discuss and critically evaluate the evidence for the impact of diet
and bioactive food components (BFCs) on epigenetic processes (including DNA methylation,
histone modifications, chromatin remodeling factors, and noncoding regulatory RNA),
implications for cancer prevention, as well as next steps for advancing diet, epigenetic events,
and cancer prevention research.

Peter Greenwald, M.D., Dr. P.H., Director
Division of Cancer Prevention
National Cancer Institute

The field of epigenetics and cancer prevention is an emerging one, although the idea of
epigenetic aspects in cancer development is generally accepted. There is solid evidence that
histone modification and other epigenetic processes affect gene expression. There also is
evidence that food and dietary supplements influence gene expression as well as an individual‟s
risk of developing cancer. Further understanding of the interaction between diet, genetics, and
critical times for exposure during development and throughout the entire lifespan is needed.
Animal models suggest that epigenetic effects during pregnancy can have an impact on the
offsprings‟ cancer risk later in life. Bioactive food components (BFCs) are known to exert their
effects over an individual‟s lifespan, with implications for effective use of these compounds for
cancer prevention. The information presented at this meeting will help improve progress in
basic, translational, and clinical research related to the use of BFCs in cancer prevention, and
will provide information on new tools to facilitate epigenetic research.





1

---

Paul Coates, Ph.D., Director
Office of Dietary Supplements
Office of the Director
National Institutes of Health

Dietary supplements are usually versions of BFCs found in food, so similar questions arise
concerning the effects of supplements on cancer risk. Information concerning the potential
impacts of BFCs, both positive and negative, on health is needed because many people in the
United States use significant amounts of dietary supplements. Accurate information is needed to
fully inform consumers, healthcare providers, and researchers on the effects of these
supplements. NIH funding of dietary supplements research is growing and now includes projects
on epigenetics, nutrigenomics, and other relevant technologies and research areas. Since ODS
exists within the Office of the Director of the National Institutes of Health (NIH), rather than
within a distinct institute, it lacks direct funding authority. However, ODS partners with the NCI
and DCP and other groups to support new research initiatives and co-fund grants relevant to
research on dietary supplements.

John Milner, Ph.D., Chief
Nutritional Science Research Group
Division of Cancer Prevention
National Cancer Institute

Today, funding of epigenetics research has increased from approximately $500,000 to more than
$13 million. Nonetheless, publications describing the relationship between diet, epigenetics, and
cancer are few. Understanding epigenetics will help to define this relationship and explain an
individual‟s response to diet, particularly in terms of cancer prevention. The effects of folate on
methylation and epigenetics have been described, and other compounds likely modify the
epigenome as well. Researchers in this field were asked to identify priority research areas and
tools necessary to advance the study of BFCs and their effects, epigenetic or otherwise, on
cancer prevention. Areas to address to clarify the effects of diet on cancer prevention include
identification of relevant compounds, issues related to dose and timing of exposure, and
nutrigenomics. Whether the effects of BFCs are independent or dependent on genetic
polymorphisms remains to be determined. Epigenetic research also may help to identify
individuals likely to develop cancer and those likely to benefit from nutritional cancer prevention
strategies.

Epigenetic Events and Cancer Prevention: Changes in Early Neoplasia
James G. Herman, M.D.
The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins
Baltimore, MD

Changes in methylation, particularly in promoter regions, are commonly observed in a number of
cancers. Transcriptional silencing associated with promoter methylation or histone deacetylation
can have the same effect as coding mutations on crucial tumor suppressor genes. The enzymes
that mediate hyper- or hypomethylation of DNA and modification of histone proteins are
potential targets for cancer prevention by BFCs and dietary supplements.

2

---

A recent analysis of the Nurses‟ Health Study suggests that the BFC folate affects cancer
development. In this study, a small decrease in colorectal cancer incidence was observed for
women with increased dietary or supplemental folate intake. The duration of exposure was
proportional to the degree of protection. Data from the Netherlands Cohort Study, which
examined the effects of dietary folate and alcohol intake on colorectal cancer incidence, found
slightly increased methylation of six genes known to be involved in colorectal cancer in the
tumors of participants with low methyl donor intake.

In lung cancer, an increasing number of genes have been found to be hypermethylated in their
promoter regions, leading to gene silencing. The gene p16, which regulates RB and the cell
cycle, shows increased methylation associated with progressive histologic changes leading to
squamous cell lung cancer. A distinct form of lung cancer, adenocarcinoma of the lung, is
increasing in frequency but is less associated with smoking than is squamous cell carcinoma of
the lung. Similar to squamous cell lung cancer, increased methylation of the promoter regions of
17 genes, such as p16, MGMT, APC, and RUNX3, was observed to be associated with
progression from normal tissue to atypical adenomatous hyperplasia (AAH), believed to be a
precursor to adenocarcinoma, to adenocarcinoma of the lung. The frequency of methylation was
proportional to the degree of malignancy. Multifocal AAHs from the same patients showed
distinct patterns of promoter hypermethylation, indicating a divergent epigenetic field effect.

Many non-squamous cell lung carcinoma cell lines have evidence of active Wnt signaling.
Antagonists of Wnt activity such as APC, DKK1, and RUNX3 were silenced, which correlated
with promoter hypermethylation. Promoter hypermethylation increased with degree of
malignancy, and thus could be predictive for lung cancer progression. A similar analysis showed
that methylation of the p16 and APC loci appear to be predictive for progression of Barrett‟s
esophagus to esophageal cancer.

Promoter methylation is a common event in cancer and is related to tumor progression. Genes
involved in regulation of the cell cycle, apoptosis, DNA repair, relevant signaling pathways, and
invasion are differentially methylated during progression from normal tissue to carcinoma. The
frequency and timing of the methylation changes suggest that they may be useful for early
detection or for risk assessment and may also be related to dietary or environmental exposure.

Discussion

Dr. Herman explained that in the lung cancer studies, differences in methylation between lung
adenocarcinoma samples from smokers versus nonsmokers could not be determined because this
work used archived samples. However, previous studies have shown that p16 is more highly
methylated in smokers and is increasingly methylated in squamous cell lung cancer.

Dr. Asad Umar (DCP, NCI) asked why p16 and APC but not MGMT show methylation changes
in Barrett‟s esophagus, given that dietary and environmental factors likely affect its
development. Dr. Herman answered that MGMT shows early methylation changes that are not
related to progression. Methylation changes at the MGMT locus may permit mutations to occur
in other genes, such as K-ras or p53. MGMT methylation is not predictive because this event

3

---

occurs early and events secondary to MGMT methylation may be needed for cancer to develop.
The mechanism by which this occurs is currently unknown.

Dr. Mukesh Verma (Division of Cancer Control and Population Sciences [DCCPS], NCI) asked
why reports on hypomethylation are not common. Dr. Herman explained that global loss of
methylation at repetitive elements could lead to aberrant expression of retroviral genes, but it can
be difficult to determine the impact of global hypo- or hypermethylation on cancer. Genome-
wide approaches will help link hypomethylation events to cancer.

Dr. Steven Belinsky (Lovelace Respiratory Research Institute) noted that recent epidemiological
studies have shown that adenocarcinoma incidence is increasing and the link to smoking is
growing stronger. Despite observing methylation changes at 50 genes, no “magic target”
indicative of cancer development in non-smokers has been identified. More information about
the etiology and risk factors for people who have never smoked is needed.

Dr. Roderick Dashwood (Oregon State University) asked Dr. Herman to comment on the
apparent ability of folate to enhance growth of established tumors, yet function protectively in a
primary prevention manner, in the absence of precancerous changes. This suggests opposing
beneficial vs deleterious effects for folate in primary vs secondary prevention. Dr. Herman
speculated that the impact of folate on cancer development may depend on which hypo- or
hyper-methylation events have already occurred. If carcinogenesis has progressed sufficiently,
folate may not be protective.

A participant asked for information on the proportion of cells in a tumor that are
hypermethylated. Dr. Herman answered that most lesions are not pure but instead contain
normal tissue, from which most of the signal is derived. Methylation is associated with gene
silencing; clonal selection may contribute to heterogeneity in tumors.

Dr. Milner (DCP) asked whether the cells surrounding the tumor represented a second target for
cancer prevention. Dr. Herman acknowledged that field defects were common in lung cancer
and that second primary cancers are the leading cause of death for lung cancer patients. To
develop a tissue-specific surrogate marker, different samples (i.e., tissues, fluids, etc.) likely will
be needed to identify field defect changes or other events.


Session I: Are There Critical Times of DNA Methylation and Potential Modification by
Dietary Factors?
Moderator: Johanna Dwyer, D.Sc. (ODS, NCI)

Maternal Nutrient Supplementation Counteracts Bisphenol A-Induced DNA
Hypomethylation in Early Development
Dana Dolinoy, Ph.D.
Department of Radiation Oncology
Duke University Medical Center
Durham, NC


4

---

Epigenetic mechanisms can mediate gene-environment interactions that allow environmental
factors to affect health and disease. Epigenetic variability, genetics, and environment likely
interact to have an impact on disease susceptibility in complex disorders such as asthma,
cardiovascular disease, and schizophrenia. Environmental factors such as diet can cause
epigenetic changes. Determining critical windows of vulnerability to relevant environmental
exposures is crucial to understanding the full impact of these factors on health.

The viable yellow agouti (Avy) mouse model can serve as an epigenetic biosensor to determine
how maternal exposures affect the fetal epigenome. Coat color variation in these mice is related
to epigenetic marks established early in development. Yellow mice are completely unmethylated
at the Avy locus, which allows ectopic expression of the agouti locus throughout life, giving rise
to a variable yellow coat color and increased rates of diabetes and tumors. If completely
methylated, ectopic expression of Avy does not occur and the mice are brown in color with
decreased risk of obesity, diabetes, and cancer. Avy is a metastable epiallele, which is an allele
that is variably expressed in genetically identical individuals due to epigenetic modifications that
are established early during development. Metastable alleles show potential for
transgenerational inheritance in addition to mitotic inheritance and are targets for
environmentally induced epigenetic modifications of the fetal epigenome.

Genistein is a plant phytoestrogen found in soy and soy products that has mixed pro- and anti-
estrogenic effects depending on tissue, dose, and timing of exposure. The putative protective
effective of soy against cancer in Asian populations is thought to be due to the isoflavones,
including genistein. To determine whether genistein exerts protective effects by modifying the
fetal genome, non-agouti female mice were fed a genistein-supplemented diet 2 weeks prior to
mating with heterozygous agouti males. Offspring of female mice fed the genistein-
supplemented diet showed a shift toward the brown coat color. Site-specific methylation
analysis showed an increase in methylation at six CpG sites at the Avy locus. Methylation at
CpG site 4 was principally responsible for the effects of genistein supplementation on coat color.
DNA methylation profiles across a number of tissue types and times of development showed that
the effects of genistein occurred before germ layer differentiation. Analysis of body weight
found that the offspring of the unsupplemented females were more likely to be obese.

The effects of maternal exposure to bisphenol A (BPA), an endocrine active compound
commonly found in polycarbonate plastics, also was analyzed using the Avy model. Similar to
the genistein study, female mice were fed a diet containing high amounts of BPA prior to mating.
Offspring of mothers fed the BPA diet showed a shift toward yellow coat color. Mice exposed to
BPA in utero also had decreased methylation at the Avy locus. Methylation rates were highly
correlated across germ layers, indicating that BPA exerts its effect early in development.
Maternal exposure to BPA also decreased methylation at another metastable epiallele, the CDK5
activator protein CabpIAP, indicating that BPA promotes hypomethylation at multiple metastable
epialleles. Analysis of methylation changes at other metastable epialleles is ongoing.

To determine whether exposure to methyl donors such as folate could counteract the effects of
BPA, mice were fed a methyl donor-supplemented diet along with BPA. There was no
difference in coat color distribution or DNA methylation between control animals and those fed
BPA plus methyl donors, indicating that methyl donor supplementation can counteract the

5

---

effects of BPA. Feeding mice a genistein-supplemented diet together with BPA also resulted in
a coat color distribution and DNA methylation pattern similar to the control animals. Parental
nutritional supplementation thus could be considered as a preventive intervention approach for
counteracting deleterious environmental influences on the fetal epigenome. The Avy model also
might be useful as a biosensor for other environmental exposures, including other BFCs, low-
dose radiation, environmental tobacco smoke, or pharmaceuticals and anesthetic agents.

Discussion

A participant asked about the timing of genistein supplementation. Dr. Dolinoy clarified that the
mice were fed the supplemented diet from 2 weeks before mating until after weaning. Thus, the
critical days of exposure are now known, but because methylation changes are observed across
all germ layers, the effect likely occurs early in development.

Another participant asked whether BPA had dose-response effects. Dr. Dolinoy explained that
these experiments used only a single dose of BPA. The dose used was significantly higher than
that at which low-dose BPA effects are observed; the experiment will be repeated to determine
the effects of lower doses of BPA.

Dr. Robert Waterland (Baylor College of Medicine) asked Dr. Dolinoy to describe approaches
for finding metastable epialleles in humans. Dr. Dolinoy explained that a genomic microarray
approach is underway to identify additional metastable epialleles in mice. Genes with agouti
expression profiles but high variability across individuals have been identified. It will be
difficult to identify metastable epialleles in humans, although there appear to be many candidates
that may show imprinting or parent-of-origin effects.

Dr. Cornelia Ulrich (Fred Hutchinson Cancer Research Center) asked Dr. Dolinoy to describe
potential molecular mechanisms that could explain the effects of genistein or BPA on
methylation. Dr. Dolinoy answered that the link between BPA exposure and changes in
methylation is currently unknown. Non-phytoestrogen estrogens appear to cause DNA
hypomethylation, in contrast to genistein which has estrogenic activity. The effects may occur
through scavenging of free radicals rather than through the estrogen pathways.

A participant asked whether color changes could be observed in yellow mice fed a methyl-
deficient diet. Dr. Dolinoy answered that changes in coat color would not be observed but
changes in obesity and cancer rates could occur. Brown mice fed a methyl-deficient diet could
be examined for changes in methylation at the Avy locus or for changes in bodyweight.

A participant asked whether multiple pathways are affected by genistein in this model. Dr.
Dolinoy answered that changes in methylation at the Avy locus result in changes in gene
expression. Expression of other genes probably is affected, but genome-wide approaches will be
needed to determine this.


6

---

Effects of methyl donors on the germline epigenetic stability of the Avy allele
David I.K. Martin
Chair, Center for Genetics
Children’s Hospital Oakland Research Institute
Oakland, CA

Epigenetics mediates interactions between genes and the environment; naked DNA does not act
on its own. Although methylation at CpG islands is a common focus of epigenetic studies
because it is a covalent modification of DNA and thus convenient to analyze, this is only one of
a number of epigenetic modifications that work in concert with modification of histones and
other chromatin proteins. Changes in the structure of chromatin have profound affects on gene
expression and a number of elaborate modifications ensure that transcription initiation is
normally suppressed, except at certain sites. CpG methylation recruits histone deacetylase
(HDAC) activity, which is dependent on methylation of histone H3 lysine 9 and is strongly
associated with suppression of transcription initiation.

Controlling elements are transposable elements that are subject to epigenetic effects and can
disrupt normal gene regulation in the vicinity of their integration sites. The epigenetic state of a
controlling element determines its ability to disrupt regulation of a nearby gene; usually the
active state produces disruption. Disruption may involve either inappropriate activation or
suppression of the controlled gene. Because the controlling element is subject to epigenetic
silencing, these elements can act in mosaic, tissue-specific, heritable, or inducible patterns.
Mammalian genomes contain many potential controlling elements, such as Avy, which may affect
phenotypic variation. The variation in coat color observed in otherwise isogenic Avy mice is due
to variations in methylation at this locus and is an example of a “disease risk” that is entirely
epigenetic. The epigenetic state of Avy is highly unstable in the germline, but there is weak
retention of the epigenetic state in the female germline.

Avy serves as a model of epigenetic involvement in fetal programming. When pregnant mice are
fed a diet supplemented with methyl donors such as folate, choline, betaine, and vitamin B12,
beginning 2 weeks before mating and continuing throughout pregnancy and lactation, their
offspring are more likely to carry a silent, methylated Avy allele; mice from these litters are more
likely to be brown in color and less prone to obesity and diabetes. The epigenetic effect of the
supplemented diet (silencing of the allele) was observed only when the Avy allele was transmitted
through the father. To further investigate the heritability of methyl donor supplementation, a
pseudoagouti female offspring from this experiment was bred and supplemented with methyl
donors only during mid-gestation to target only the primordial germ cells that would give rise to
the F2 generation, without affecting the F1 generation. Surprisingly, the F1 generation was
affected by mid-gestation supplementation; a shift toward brown or pseudoagouti coat color was
observed. Again, the epigenetic effect of supplementation was observed only when the Avy allele
was transmitted through the father. This indicates that an environmentally induced effect on
epigenotype can survive the epigenetic resetting events that occur between generations and that
the mother‟s diet may influence the phenotype of several succeeding generations.

To determine the cumulative effects of maternal methyl donor supplementation over generations
and to determine if the inheritance of the methyl donor effect lasts for more than one generation,

7

---

pseudoagouti males were bred to produce subsequent generations, continuing until the pattern of
inheritance of the Avy allele stabilizes. Results from the F0 to F4 generations show a trend for
decline in the numbers of mice with yellow coats, raising the possibility that stable extinction of
the Avy phenotype could occur.

To explore mechanisms by which methyl donors silence the Avy locus, bisulphate allelic
sequencing of this locus found that CpG methylation of this region was incomplete, even in mice
in which Avy was transcriptionally silent. In utero exposure to methyl donors also did not
increase the density of CpG methylation in either the F1 or F2 generations. This suggests that
methyl donor supplementation does not directly increase methylation of Avy; instead, the locus is
more likely to become methylated and may shift to a silent state in early embryogenesis.

Discussion

Dr. Jean-Pierre Issa (MD Anderson Cancer Center) asked whether transmission of the Avy allele
had been analyzed in the absence of methyl donor supplementation. Dr. Martin answered that
this had not been done, because the paternal effect was not initially recognized. A current goal is
to derive an Avy line with an extinguished phenotype, without supplementation. It is possible,
although doubtful, that Avy silencing is not mediated by methyl donor supplementation. Dr. Issa
asked about the potential effect of strain background on Avy silencing. Dr. Martin answered that
strain background does have an effect, because some strains are more likely to have agouti
offspring. The strain used for these experiments is well-defined and relevant strain-specific
modifiers of the somatic state are known.

Dr. Belinksy asked whether this work supported the idea of methylation changing expression of
tumor suppressor genes, but not as a rate-limiting effect. Dr. Martin answered that the
methylation patterns observed at the Avy allele are unusual, and he suspects they are a reflection
of the underlying epigenetic states that may be transmitted through the germline.

Dr. Waterland described a similar study in which transgenerational exposure was studied via
transmission of the Avy allele through the female germline. In this case, there was no evidence of
inheritance of effects. Dr. Martin answered that only pseudoagouti mice were bred in this study
because these mice likely carried a silenced Avy allele in their germ cells. Dr. Waterland‟s study
involved mainly breeding of yellow mice; because the inheritance and methyl donor effects are
weak, increasing the number of mice used in this study likely would reveal a trend toward brown
coat color. Dr. Martin said that it is difficult to demonstrate that methylation specifically is the
inherited mark, but efforts are underway to create embryonic stem cell lines to find
hypermethylated alleles.

A participant commented that offspring receive differential nutrition based on their position in
the uterus and that this could affect distribution of the Avy allele. Natural selection also may
have a role when pseudoagouti mice are maintained on a high methyl donor diet and then re-
bred. Dr. Martin noted that genetic selection does not occur because the strain is isogenic.
Additionally, breeding females are drawn from the general mouse colony, not from the
experimental groups.


8

---

Modulation of Colorectal Carcinogenesis by Dietary Folate: Other One-Carbon
Micronutrients and Age as Essential Co-Determinants
Joel Mason, M.D.
Human Nutrition Research Center on Aging
Tufts University
Boston, MA

Low folate intake is associated with increased risk for cancer, particularly colorectal cancer. The
impact of folate on cancer risk is affected by family history; data from the Nurses‟ Health Study
showed that folate had a robust protective effect primarily on women with primary relatives who
had colorectal cancer. Thus, folate inadequacy is not in itself carcinogenic, but rather enhances
the likelihood of developing colorectal cancer if there is an underlying predisposition. Folate
may modulate carcinogenesis by altering methylation of nucleic acids and/or other
macromolecules, resulting in alterations in the stability and integrity of DNA. Folate also is a
cofactor for synthesis of purines and thymidines needed for DNA synthesis. Low folate levels
can lead to misincorporation of uracil rather than thymidine into DNA, leading to mutagenic
effects and DNA strand breaks during DNA repair.

Folate and vitamins B2, B6, and B12 participate in one-carbon metabolism and are
biochemically interdependent. A number of epidemiological studies have suggested that these
one-carbon nutrients all impact cancer risk. Several surveys also have shown that aging is
associated with marginal levels of these nutrients, as well as increased cancer risk. To determine
whether multiple, mild one-carbon nutrient deficiencies work synergistically to create a pro-
carcinogenic milieu in the colon, C57Bl6 mice were fed diets mildly depleted in folate alone, or
depleted in folate and one or all of vitamins B2, B6, and B12. None of the single or double
depletion states had an effect on global genomic DNA methylation in colonic mucosa. However,
in multiply vitamin deficient mice, a greater than 50 percent decrease in methylation was
observed.

The Wnt signaling pathway is involved in colorectal carcinogenesis. Under normal conditions,
-catenin is degraded. In colorectal cancer, defective APC prevents -catenin degradation and
permits translocation of -catenin to the nucleus, where it promotes transcription of pro-
transformation genes such as c-myc and cyclin D1. Folate depletion did not affect APC
promoter methylation. However, folate depletion was associated with a trend toward and
increases in DNA strand breaks in the APC mutation cluster region (MCR) that also was
associated with decreased levels of APC in the mucosa; these trends were more significant in the
multiply deficient mice. -catenin protein expression in colonic mucosa also was increased in
the multiply deficient mice, as was nuclear translocation of -catenin and cyclin D1 expression;
several components of the Wnt signaling pathway were increased in a pro-transformation
manner. Histologic examination of the colonic epithelium of these mice showed that mild
depletion of multiple B vitamins was associated with a decrease in apoptosis.

Age also may be a co-determinant of carcinogenesis associated with one-carbon nutrient status.
Uracil incorporation is more likely to occur in colonic DNA of older versus younger rats in the
presence of folate deficiency. Given identical levels of dietary folate, older colon has 50 percent
less folate than colon of younger animals. Supplementation with folate increases colonic folate

9

---

levels to that observed in younger animals, but plasma folate levels remain the same in old and
young rats at any level of dietary folate. One key difference is a lack of methyl THF in folate-
depleted old rats, but only a modest decrease in methyl THF in depleted younger rats. In mice,
increased dietary folate results in higher increases in promoter methylation at the colonic p16
gene promoter in older versus younger animals. Thus, the colonic mucosa of older animals is
more vulnerable to qualitative and quantitative changes in folate metabolism compared to
younger animals, which translates to an increased susceptibility to DNA base substitution and
changes in promoter methylation. In humans, 2 months of folate deprivation resulted in no
change in plasma folate, but a 55 percent decrease in colonic folate levels, indicating that a minor
degree of folate depletion, as determined by plasma folate levels, is magnified in the colon.
Folate depletion is magnified in the colonic mucosa of both animals and humans, regardless of
age; this could be attributed to the high proliferation rate of the colonic epithelium.

Discussion

Dr. Issa asked if a 50 percent decrease in genomic DNA methylation in colon was observed in
the multiple vitamin B deficient mice. Dr. Mason answered that on an absolute level, a decrease
in genomic methylation was observed. The ability to determine absolute genomic methylation is
affected by the technology used to detect methylation. Use of liquid chromatography/mass
spectrometry (LCMS) for detecting absolute genomic methylation is new; earlier studies relied
on the SssI method. Changes in colonic DNA methylation have been observed using this
method, but because it is semi-quantitative, it cannot be determined if the drop in absolute
methylation was 50 percent. A drop as dramatic as this probably would not be observed in
humans.

Dr. Ulrich asked Dr. Mason to comment on the similarities and discrepancies between
manipulation of one-carbon metabolites in mice versus humans. Dr. Mason answered that the
animals in this study had adequate levels of methionine. Because the human study involved only
folate depletion, the effects of multiple one-carbon nutrient depletion in humans is unknown.
However, enough similarities exist between rodents and humans that the results of the mouse
studies are somewhat translatable to humans; multiple B vitamin deficiencies in humans are
likely to have similar effects.

Dr. Milner noted that the colon appears to be the most vulnerable or responsive organ and asked
whether other tissues had been examined. Dr. Mason explained that the colon is vulnerable
because of the high proliferative rate of mucosal epithelium, which turns over every 72 to 96
hours. This tissue sustains high rates of DNA replication and thus is sensitive to any disruption
of DNA synthesis. It is likely that any highly proliferative tissue would be similarly sensitive to
B vitamin depletion. Plasma does not reflect changes in dietary B vitamins, and thus is
inadequate for a biomarker. Skin scrapings also may not be useful, because surface skin cells are
not highly proliferative and thus also would probably not reflect changes in B vitamin levels.


10

---