|
Saturday June 5/Evening
EPIGENETIC REGULATION OF GENE EXPRESSION
Matthew Vaughn
The Saturday evening
session was chaired by Vicki Chandler (University of Arizona) and
covered a wide range of topics. Plants were the model organism for the
first three talks, in which Arabidopsis flowering time,
organogenesis, and seed development were discussed, while the next talk
centered on human epigenomics. After the coffee break, specification of
cell identity, heterochromatin targeting in Drosophila, and
Neurospora DNA methylation were discussed.
Caroline Dean (John
Innes Centre) began by providing insights into the mechanism of the
vernalization process in plants, in which flowering time is synchronized
with amenable temperatures and photoperiods. Exposure to cold reduces
expression of the FLC gene, which is known to promote flowering.
Cellular memory of the cold treatment remains, as flowering remains
repressed even if plants are moved back to warmer temperatures. FLC
expression is de-repressed in vrn mutants. VRN2 is homologous to
the Su(z)12 polycomb protein and VRN1 is a plant-specific DNA binding
protein shown to associate with all five chromosomes. Vernalization
causes increased H3K9 and H3K27 dimethylation at the FLC promoter. The
H3mK27 is lost in vrn2 , while H3mK9 is lost in both vrn1
and vrn2, indicating that these genes cooperate to repress FLC
expression. Additional vrn- alleles were identified and found to
be members of a PHD (Plant Homeodomain) and FNIII (fibronectin)
domain-containing family. The VRN genes exhibit differential expression
patterns, indicating that they have diverse functions. These expression
patterns in combination with Y2H experiments suggest a temporal
progression for silencing the FLC gene where VIN3/VRN7 and VRN5 act
first to deacetylate the FLC promoter followed by facilitation of H3K9
and K27 dimethylation by VRN2 and VRN1. This, in turn, allows
recruitment of the Arabidopsis HP1 ortholog LHP1 to reinforce
repression of the FLC gene.
Next, Marja Timmermans
(Cold Spring Harbor Laboratory) described the deep evolutionary
conservation among mechanisms by which plants specify organ polarity.
The model system for these studies is the maize rolled leaf 1 (rld1)
gene, which is a class III homeodomain/leucine zipper transcription
factor (HD-ZIPIII) that is homologous to the Arabidopsis organ
polarity genes PHABULOSA, PHAVOLUTA and REVOLUTA.
Plants defective in rld1 have tightly rolled and dorsalized
leaves. The Arabidopsis genes have binding sites for microRNAs
miR166 and miR165 and this miRNA binding site is conserved in maize
rld1. Interestingly, dominant alleles of rld1 have point
mutations in this conserved domain. In Rld1 mutants, rld1
is misexpressed on the ventral side of leaf primordia. Sequence
comparisons among Arabidopsis and rice genomes facilitated
cloning of the maize pre-miR166 gene. In situ hybridizations
showed that miR166 is found initially below the developing leaf but soon
expands in a complementary pattern to rld1 expression. This
indicates that miR166 is a developmental signal that is responsible for
specifying the dorsoventral axis in leaves. The initial accumulation of
miR166 below incipient leaves suggests that it is released from a
specific signaling center to act as a mobile morphological determinant.
Consistent with this idea is an observation that expression of viral
TBG1, which affects RNA trafficking, yields dorsoventral patterning
defects reminiscent of rolled leaf 1.
Ueli Grossniklaus
(University of Zürich) talked about Arabidopsis medea.
This mutant in a polycomb-like protein displays gametophytic maternal
effect lethality, wherein seeds derived from the mutant female
gametophyte abort regardless of parental contribution. The medea
gene is imprinted, with only the maternally-inherited allele expressed
after fertilization. A search for modifiers of medea uncovered
other members of the FERTILIZATION INDEPENDENT SEED (FIS) class
of genes, including fis2 and fie1, which are known to
restrict cell proliferation. The fis2 gene is orthologous to
Suv12, while mea is similar to Enhancer of Zeste. The
MEA,FIE1, and also MIS1 proteins were found to be in a mammalian-like
600 kDa polycomb complex, as opposed to a C. elegans-like
MES2/MES3/MES6 255 kDa complex. The MSI1 protein was found to interact
directly with FIE1. Furthermore msi1 mutants exhibit
fie-like phenotypes, including 50% seed abortion and
fertilization-independent endosperm development. These observations
raised the question of whether the effect of FIE-class genes on G1->S
transition genes is direct and mediated by recruitment of the
FIE1/MEA/MSI1 complex, or indirect and mediated by via other unknown
proteins. Microarray analysis identified two genes that are consistently
upregulated in mea plants, pheres (phe) and
meidos (meo). The phe gene is a direct target of the
MEA/FIE complex, as ChIP was used to demonstrate that it bound the
phe promoter. Because phe is not a gene known to mediate the
G1->S, transition, an indirect model for regulation of cell
proliferation is favored.
In the next talk,
Andrew Chess (MIT and the Whitehead Institute) spoke on the seemingly
unrelated topics of regulation of genes that specify cell identity and
monoallelic gene expression. Cell identity is implemented by either
expressing combinations of many related genes, such as odorant receptors
or interleukins, or by expressing isoforms of a single gene, such as
some immunoglobulins, the T-cell receptors, or, the focus of much of
Chess’ talk, DSCAM (Down Syndrome Cell Adhesion Molecule). Around 28,000
isoforms of DSCAM are generated by alternative splicing in Drosophila.
It has a role in axon guidance and appears to specify cellular adhesion
properties. Single-cell isoform microarray analysis of individual cells
showed that, while tissues and populations expressed the same general
populations of isoforms, each cell expressed a unique complement of
DSCAM molecules. Mammalian DSCAM does not have alternative splicing, but
Dr. Chess speculated that molecules such as neurexins and protocadherins
could fulfill this role.
Continuing on to the
next topic, the audience was reminded that, in addition to genes
silenced by X-inactivation, there are several examples of monoallelic
expression of autosomal genes. What may tie this together with the DSCAM
cell identity story is that many of these are genes that unique to
specific cells, such as odorant receptors, cytokines, and
immunoglobulins. This inactivation takes place on a chromosome-wide
basis around the same time that X-inactivation occurs, suggesting some
degree of shared mechanism. Genes expressed in monoallelic fashion are
either early- or late-replicating. Using a FISH-based assay for altered
replication timing, Chess and colleagues have screened 80 genes and have
found that 10% of them exhibit the asynchronous replication that is
diagnostic of monoallelic expression, indicating that this could be a
widespread phenomenon in mammals. Dr. Chess suggested that random
monoallelic expression might enhance evolvability of some loci, as it
could serve to decrease time to fixation for advantageous alleles.
Sarah Elgin
(Washington University) followed Dr. Chess with a talk about targeting
heterochromatin assembly in Drosophila, and how it relies upon
the piwi, aubergine, and homeless RNAi genes. Eric
Selker (University of Oregon) concluded this evening’s session by
discussing how DNA methylation in Neurospora crassa is controlled
by histone methylation. The Neurospora genome is protected from
repeat sequences such as transposons via a mechanism called Repeat
Induced Point mutation (RIP), in which duplicated DNA sequences are
detected and modified by making G:C to A:T mutations. Cytosines in a
RIP-affected area are typically methylated. The resulting mutated
sequences are capable of signaling de novo methylation even if
the methylation is stripped from the sequences before they are
transformed back into Neurospora. Experiments with various
combinations of A/T in a 25-nucleotide oligo showed that (TAAA)n is the
best signal for initiating methylation. This suggested that the
methylation machinery senses structure rather than sequence, perhaps via
an AT-hook protein. Consistent with this idea is the observation that
distamycin, a chemical homolog of the AT-hook domain, inhibits DNA
methylation. Several mutants that are defective in methylation
(dim) have been identified by Dr. Selker’s laboratory. The
dim2 mutation is within a DNA methyltransferase responsible for
vegetative cell DNA methylation but that is not involved in RIP, while
the dim5 mutation identified a SET domain protein responsible for
H3K9 methylation. Mutants in dim5 as well as hp1 have been
shown to lose DNA methylation. As in other organisms, Neurospora HP1 has
been shown to localize to heterochromatin. In dim5 mjutants, this
localization is lost. Dr. Selker mentioned that mutations in RNAi
components had no effect on DNA methylation or HP1 localization. Taken
together, this suggested that HP1 might read trimethyl H3K9 to signal
DNA methylation.
Other
Dispatches
Symposium
69 Live
Symposia
Past (a bit of history and photographs from previous Symposia)
Online
Symposium Volumes (searchable database of past Symposia volumes and
currently received manuscripts) |
Matthew Vaughn
(Martienssen lab)
2
0
0
4 |