III. Other non-Mendelian phenomena
There are many other biological phenomena that result in apparent deviation from Mendel's laws. Many of them look like they could be due to organelle inheritance if you look at only one generation. Looking at additional generations often reveals more complex biology and genetics. We'll only have time to touch on some of the better-studied systems. Your reference list includes a review for each,  if you would like more information.  < style="font-family: arial narrow;">

A. Maternal effect mutations

A maternal effect it NOT the same as maternal inheritance. Maternal effects result because the maternal parent produces the egg. The production of eggs is, of course, controlled by genes. Therefore maternal gene products may be present in the egg and regulate the development of the new offspring after fertilization.  In these cases, the genotype of the mother determines the developmental outcome or the phenotype of the progeny. This is in contrast to our usual expectation that the phenotype of an individual is determined by the genotype of that individual.

1.  Maternal effect genes in drosophila development

Some of the best examples of maternal effect genes come from the study of drosophila developmental genetics - a topic to be covered in the next section of this course.  The drosophila egg develops connected to nurse cells that contribute macromolecules to the developing egg. Because the nurse cells are located at one end of the egg, the egg develops in an asymmetrical environment, and macromolecules contributed by the nurse cells can have an asymmetrical distribution within the egg. One critical material is the mRNA for a gene known as bicoid. In situ hybridization shows the bicoid RNA is localized at one end of the egg, the end destined to be the anterior of the larva. If the maternal parent is bicoid -/-, no bicoid RNA is present in the egg, and the larvae fail to develop a normal anterior end. Instead the larvae develop two posterior ends. Transplanting cytoplasm from the egg of a bicoid +/- female into the anterior end of the egg from a bicoid -/- female will partially rescue the developmental defect. Bicoid is just one of many maternal-effect gene products present in the drosophila egg. Note, the genotype of the bicoid -/- mother is seen in the abnormal (ultimately lethal) development of all her offspring, regardless of their genotypes at the bicoid locus.

 2.  Shell coiling in Limneae peregra

< style="font-family: arial narrow;">To observe some of the unusual inheritance patterns produced by maternal effect genes, it's useful to have mutations in genes that don't result in death of the progeny. The classic example is the direction of shell coiling in limneae snails. Snails, by the way, are hermaphrodites, so you can mate a snail with itself. On the powerpopint slide, you can see that genotype of the mother determines the direction of shell coiling of the progeny snails. For example,  all progeny of a +/+ maternal parent have dextral (upward - right) shell coiling, whereas all progeny of a s/s maternal parent have sinestral (upward-left) shell coiling.  This can lead to apparent differences in reciprocal crosses, as seen in the F1 generation of the snail example. The + allele is dominant, so all progeny of +/s maternal parents have dextral coiling. In the F2 generation of the snail  example, you can see that sibiling progeny having different genotypes all have the same phenotype, conditoned by the maternal genotype.

3.  Maternal effects in other organisms < style="font-family: arial narrow;">Although many maternal effect genes have been identified in drosophila, relatively few have been identified in mammals and plants. This probably reflects differences in the mechanisms used to establish polarity and early developmental patterning in these organisms (topics we will take up in the next section of the course). Recently, an unudual maternal effect gene was identified in mouse (Tong et al. Nature Genet. 26:267). Female mice homozygous for a mutation at the mater locus develop normally, and produce normal eggs. Following fertilization, eggs of mater -/- females develop normally through the two cell stage but fail to develop further. (Mater stands for maternal antigen that embryos require). Mater is a single-copy gene expressed only in oocytes. Wild-type MATER protein is detected in developing oocytes and persists through the multi-cellular blastocyst stage of development. Its precise role in normal development is as yet unknown.

In plants, the short integument (sin1) mutation appears to conform to genetic expectations of a classical maternal effect gene. The sin1 mutation affects the apical-basal axis formation in the embryo (Ray et al. Devel. Biol. 180:365). The sin1/sin1 or sin1/+ embryos borne on a sin1/sin1 seed  (female) parent  plant show defects in the development of apical and basal organs, cotyledons and roots. In contrast, sin1/sin1 embryos borne on a sin1/+ seed parent develop normal axes.  These sin1/sin1 progeny do have some later developmental effects, such as delayed flowering time, which may result from other action of the sin1 gene later in plant development.   

If you read French, a nice recent review on maternal effect mutations is by  Christians, Medicine/Sciences 19:459.

 B. Infectious cytoplasmic agents

Wolbachia are intracellular parasites, symbionts or mutualists found in a vast array of invertebrate organisms. Recently interest in wolbachia has exploded. (See Knight, Nature 412:12  and Bandi et al. Trends Parasitol. 17:88 for reviews on the fascinating biology of these organisms.) Since wolbachia live in the cytoplasm, they are transmitted from generation to generation through the female. Therefore, male hosts genetic dead ends for the wolbachia parasites, and wolbachia have evolved a number of clever ways to increase the percentage of female progeny at the expense of male progeny.

1.  Sex converters

The wasp Nasonia vitripennis has a haplo-diplo sex determination system. Fertilized eggs (2N) develop as females. Unfertilized eggs (1N) develop as males. The wolbachia parasites causes chromosome duplication in unfertilized eggs. Therefore, unfertilized eggs that would normally develop as males develop as females and enhance the transmission of the wolbachia to the future generations.

In woodlice, wolbachia convert male offspring to female offspring by somehow suppressing male hormones.

2.  Maternally inherited abnormal sex-ratio (son killers)

An example is the sex ratio distorter (SR) found in wild populations of  D. willistoni:

        D. willistoni   female  x    D.  melanogaster   >   primarily female progeny (occasional males)
        F1  female  x D. melanogaster   >   primarily female progeny
        D. melanogaster  x  rare F1 male  >  1:1 (male:female) progeny

In the first two crosses, the male embryos fail to develop due to poorly characterized mechanisms.  The trait can be cured by antibiotics and transmitted by injection of ooplasm from a distorting female into the larvae of normal females. Son-killers are found in many different insects. This strategy can result in insect populations with very few males. One question under active investigation is whether wolbachia can drive host to extinction or limit the range of the host to areas where environmental conditions (such as warm temperatures) limit the transfer of wolbachia to the next generation.

3.  Cytoplasmic incompatiblity (CI) systems

CI systems result from wolbachia infections of many different insect hosts

Uninfected female x infected male  >  zygote lethality
Infected female x infected male  > normal hatching
Infected female x uninfected male > normal hatching
Zygote lethality is due to delay of paternal chromosome participation in the first mitotic division following fertilization. Loss of paternal chromosomes results in embryonic death. In eggs of infected females, paternal chromosomes are rescued by an unknown mechanisms and viable embryos result. Models involving epigenetic modification of sperm chromosomes and chromatin remodeling following fertilitzation are currently favored to explain these observations (Harris and Braig, Biochem. Cell Biol. 81: 229). The result of all this is that infected females have a reproductive advantage and the wolbachia, which are transmitted only through the female, can spread rapidly through a population.

cytoplasmic microbes are obviously of interest from an evolutionary standpoint. They also potentially provide biocontrol agents for insects and other invertebrates. Finally, it is  interesting that the effects of these microbes appear to be mediated through abnormalities in chromosome structure-function. This will be a recurring theme in non-Mendelian inheritance of nuclear genes.

C. Meiotic drive systems

Meiotic drive is defined as any alteration of meiosis or subsequent production of gametes that results in the biased transmission of a particular genotype. Meiotic drive has been observed in insects, mammals, plants and fungi. Each example is unique in terms of  genetic locus and mechanism. In mammals, most examples seem to involve biased transmission through the female parent, whereas in insects, bias occurs most commonly through the sperm (reviewed by Pennisi, Science 301:1837). Gametophytic effects in plants (discussed below) can also be considered as meiotic drive, since the haploid gametophyte includes the gametes.  Gametophytic effects can result in transmission bias through the egg or the sperm.

 1. The segregation distorter (Sd) locus in Drosophila melanogaster (Ganetzky, Amer. Scientist 88:128; Kusano et al., Bioessays 25:108) is a well-studied meiotic drive system.

The transmission bias of Sd over Sd+ is observed only through the male parent. The reason for this became obvious when sperm from Sd/Sd+ males were examined. 50% (the Sd+ genotypes) were not viable because the chromosomes do not condense properly in the developing Sd+ spermatids. The effect of Sd on Sd+ chromosomes is mediated through a second locus on the same chromosome - the responder (Rsp) locus.  Sd chromosomes carry an insensitive responder allele (Rsp-i). Sd+ chromosomes carry a susceptible Rsp allele (Rsp-s).

Both the Sd and Rsp loci have been cloned. The Rsp locus is a noncoding repeat of 120 nucleotide pairs located in a heterochromatic region of the chromosome. Rsp-i alleles have about 50 copies of the repeat whereas Rsp-s alleles have several hundred copies. The Sd allele encodes a truncated version of a RanGTPase activating  protein (RanGAP).  RanGAPs regulate the functions of RanGTPases. These functions include nuclear-cytoplasmic transport and cell cycle regulation. The Sd RanGAP retains enzymatic activity but is mis-localized, remaining in the nucleus rather than the cytosol. How this mis-localization results in failure of Sd+; Rsp-s chromosomes to condense during sperm development is a mystery that remains to be solved.

Another well-studied meiotic drive system is the t locus of the mouse (Schimenti, Trends Genet. 16:240). The genetics of the t locus are similar to that of the Sd locus in drosophila. Both distorter and responder loci are required for the effect. Molecular mechanisms of segregation distortion, may however, be completely different. In this system the sperm carrying the wild-type chromosome fail to function due to motility, rather than chromosomal, defects.
 
2.    Gametophytic effects in plants

Gametophytic effects can result in the biased transmission of one allele over another in plants. The development of the 1N plant gametophytes (the pollen grain and embryo sac) are, of course, under genetic control. Mutant alleles disrupting development of the embryo sac will not transmit through the female, but may transmit normally through pollen. Mutant alleles disrupting develpment of pollen will not transmit through the male, but may transmit through the female (reviewed by Chaudhury and Berger, Cell & Develop. Biol. 12:381; Yang and Sundaresan, Curr. Opin. Plant Biol. 3:53). Self-pollination of a plant heterozygous for a female gametophyte mutation will produce 50% aborted seeds.  Alleles or markers tightly linked in cis to a female gametophytic mutation will not transmit through the seed.   A plant heterozygous for a male gametophytic mutation will produce 50% normal and 50% non-functional pollen.  In some, but not all cases, the mutant pollen can be recognized by morphological abnormalitites such as collapse, delayed development,  or failure to germinate. Alleles or markers tightly linked in cis to a male gametophytic mutation will not transmit through the pollen.

D.    Epigenetics

Epigenetics is defined as heritable changes in gene expression that do not involve changes in gene sequence. Pennisi (Science 293:1064) and Riddihough and Pennisi (Science 293:1063)  present a nice discussion of several epigenetic phenomena, and point out the commercial and medical importance of epigenetics.

If the DNA sequence does not change, how do changes in gene expression occur?  It is now clear that many epigenetic phenomena occur largely via changes in chromatin structure (reviewed by Felsenfeld and Groudine, 2003. Nature 421:448-453). In general, methylation of DNA is associated with turning off gene expression, but we will see some examples where the opposite is true.  Some organisms that clearly exhibit epigenetic effects (such as drosophila) do not have extensive DNA methlyation. Modifications of histones and non-histone chromosomal proteins have also been implicated in epigenetics. Acetylation of histones facilitates transcription. Enzymes that remove acetyl groups from histones often work with DNA methylases to silence genes. Histone proteins themselves can be methylated in a manner that blocks acetylation and favors an inactive gene state. How DNA regions are chosen to become active or silenced is still an open question, but it seems that chromosomal regions to be regulated differently are separated by "boundary elements" containing "islands" of repeated CpG.

1.    Paramutation

Paramutation is an example of epigenetic modifications that can be maintained through one or more meiotic generations.  It was discovered in maize by R.A. Brink (Brink et al. Science 159:161-170). Brink defined this phenomenon as the change in expression of an allele (paramutable) brought about by association with another allele (paramutagenic). Once altered, the paramutable allele is termed a paramutant allele and can often then function as a paramutagenic allele. If an allele is neither paramutagenic nor paramutable, it is termed a neutral allele.

The best-studied examples of paramutation were observed at loci ( r, b, p and pl) encoding transcription factors that regulate the anthocyanin pigmentation of the plant and the aleurone (outer layer of the triploid endosperm).  If paramutable and paramutagenic alleles are brought together in an F1, and then segregated by crossing with a neutral tester, both alleles are transmitted to the progeny, but the paramutable allele now has reduced expression; it has become a paramutant allele. Reduced expression is associated with reduced accumulation of gene transcripts. Hence it is the expression of a gene, not the transmission, that is abnormal or unexpected.  This altered gene activity is commonly transmitted to the next generation. Furthermore, the paramutant allele is itself paramutagenic.  (For recent reviews see Stam and Scheid, Trends in Plant Sci. 10:283; Della Vedova and Cone, Plant Cell 16:1358-1364.) There are some differences among the three well-studied examples.  Paramutable alleles of b and pl can undergo spontaneous paramutation (in the absence of a paramutagenic allele) whereas paramutable alleles at r cannot.  Paramutant alleles at r and pl can revert back to their original states after several sexual generations, but paramutant alleles at b have never been observed to revert.  

Paramutagenic / paramutant alleles exhibit reduced transcription. Current models (stam and Scheid, Trends in Plant Sci. 10:28) suggest that this transcriptional silencing is mediated by the RNA silencing pathway (described below), or by the transfer of chromatin remodeling proteins from paramutagenic to paramutable allele when the two alleles are paired.

Recent work on paramutation provides and elegant example of how genetics can be used to learn more about biological processes. Dorweiler et al. (Plant Cell 12:2101-2118) used genetics to find mutants that disrupted paramutation. They crossed B'B' (paramutagenic) x BIBI (paramutable). Because paramutation at b is rapid, efficient and stable, all F1 progeny were B'B' and light in color. They self-pollinated F1 plants and progeny families were screened to identify families where 1/4 of the seedlings had the BI pigmentation, indicating that capacity for paramutation had been lost due to a recessive nuclear mutation.  The single mutation they identified (mop for mediator of  paramutation) also prevented paramutation at the r and pl loci, indicating a shared mechanism of paramutation at the three loci. The mop mutation was recovered in a transposon-active background, and so is potentially tagged, although it has not yet been cloned. Homozygous mutation at mop does not affect "global" DNA methylation. The mop locus is hypothesized to encode some type of chromatin remodeling protein.

Stam et al. (Genetics 162:917) recently used genetic markers to recover recombinants between neutral allele (B-Peru)  and a paramutagenic allele (B'). Recombinants were examined to see if any recombination events created a paramutagenic version of  B-Peru or a neutral version of the B' allele.  Recombinants were found, enabling Stam et al. to identify the cis-adjacent regions governing paramutation. Surprisingly, the region is 93-106 kb upstream from the promoter at the b locus. Therefore, relatively long-distance interactions are implicated in the chromatin changes associated with paramutation. This cis element has seven tandem repeats of an 853 bp repeat. A decreased number of repeats was associated with decreased paramutagenicity, ie. neutrality. Hypermethylation of repeats and open chromatin structure are associated with active expression (transcription) of BI , whereas hypomethylation of repeats and closed chromatin structure associated with repressed transcription of B’. So here is an example where DNA methylation is associated with increased gene expression.

2.     Genomic imprinting

In genomic imprinting (reviewed by Walter and Paulsen, Seminars in Cell and Developmental Biology 14:101-110), genes are expressed differently depending on whether they are inherited through the maternal or paternal parent. For some genes, expression during development is from the paternal allele only (maternally silenced / paternally expressed); for others, expression is from the maternal allele only (paternally silenced / maternally expressed). Imprinting, like paramutation,  results from changes in gene expression rather than aberrant transmission of alleles. Imprinting was first discovered in plants, and subsequently in mammals.

Evidence that maternal and paternal genes are not equivalent in mammalian development is reviewed by Tycko et al. (J. Androl. 18:480-486).  "Isoparental" embryos in mice can be produced by removing maternal or paternal pronuclei and replacing with pronuclei from other fertilizations. Isoparental embryos do not develop normally. Gynogenotes (having two female genomes) ultimately fail due to underdeveloped extraembryonic placental tissue. Androgenotes (having two male genomes) result in an abnormal (often overgrown) embryo and overdeveloped extraembryonic placental tissue. "Uniparental disomies" (embryos having 2 chromosomes from 1 parent) can be produced by the mating of translocation stocks. For some chromosomes, development was normal. For other chromosomes, developmental abnormalities such as those seen in isoparental embryos, were observed.

Several imprinted loci have been studied in detail (Walter and Paulsen, Seminars in Cell and Developmental Biology 14:101-110). The regulatory details differ from locus to locus, but some similar features have become apparent. One well-studied example of imprinting is the human chromosome region 15q11-q13.  In Prader-Willi syndrome (PWS) a deletion of this region is inherited from the paternal parent. The affected individuals therefore carry only a maternal allele. The syndrome includes mild retardation, excessive weight gain,  and sluggish behavior. In Angelman syndrome (AS) a deletion of this region is inherited from the maternal parent. The affected individuals carry only a paternal allele. The individuals are mildly retarded, thin and hyperactive.

These observations suggested the presence of both paternally and maternally imprinted genes within this region, because both paternal and maternal copies of the region are necessary for normal development. The region has now been characterized in detail. It contains several paternally expressed genes and one maternally expressed gene. The maternally expressed gene encodes a ubiquitin ligase (UBE3A) required for normal protein turnover. Interestingly, both maternal and paternal copies of the ube3A gene are expressed in most tissues. However, only the maternal copy is expressed in particular regions of the brain. Thus imprinting can be tissue-specific.

Although mechanisms of imprinting are not understood fully, many aspects are now known. At the PWS / AS locus, a bi-partite imprinting center  (IC) is located within the locus. The PWS region of the IC acts in cis on the paternal chromosome to promote demethylation and expression of paternal alleles. The AS region of the IC acts in cis on the maternal chromosome to methylate and repress action of the PWS-IC. Maternal alleles therefore remain methylated and repressed, except for the ube3a gene. In brain tissue, the PWS-IC on the paternal chromosome also promotes antisense  transcription of the paternal ube3a locus. The antisense transcript silences the paternal ube3a allele, hence only the maternal allele is expressed in this tissue. Study of other imprinted loci reveals common features including gene clusters containing oppositely imprinted genes, cis-acting imprinting centers, parent-specific DNA methylation patterns and in some cases, antisense transcripts of silenced paternal alleles.

Note that the imprints must be erased and "reset" in each generation of gametes for the proper "parent of origin" expression to occur in the next generation. In mammalian development, imprints are erased in primordial germcells and re-established in the gametes (reviewed by Reik et al. Science 293:1089). Genetic mutations within the IC regions have been observed. IC mutations can result in failure to reset imprints properly, "locking" the genes on the mutant chromosome into a maternal or paternal gene expression pattern. IC mutations were key to identification of the PWS and AS IC regions. Individuals with IC mutations can exhibit PSA or AS, even though they carry the full complement of structural genes in the region. For example, if a chromosome locked into a paternal expression pattern is inherited through a maternal parent, the offspring will effectively have no maternal-specific expression pattern. The effect is similar to inheriting a structural gene defect through the maternal parent, resulting in AS.

3.    Epigenetics and development

The development of multicellular organisms results from changes in gene expression highly regulated through space and time. If each of our cells derives from the same fertilized egg and contains the same DNA, how is it that the cells in different organs and tissues become specialized in form and function? A limited subset of the genetic information is expressed in differentiated somatic cells, with the rest of the information being silenced.

Changes in DNA methylation indicate that chromatin is extensively remodeled in the early embryo (reviewed by Reik et al. Science 293:1089). Chromatin is extensively de-methylated immediately after fertilization and remethlyated at the time of embryo implantation. This remodeling in early development presumably sets the stage for the proper program of developmental gene expresssion to unfold.  Embryonic stem cells can undergo many different types of developmental programs because the chromatin has not been configured for a specialized pattern of gene expression. In primordial germ cells (a small subset of early embryonic cells) parent-of-origin imprints are erased, and then re-set later during gamete development. Parent-of-origin imprints are generally very stable and only reset in primordial germ cells. With all this accomplished, the develomental program - a cascade of changes in gene regulation - unfolds normally.

Rideout et al. (2001. Science 293:1093-1098) consider the implications of developmental chromatin remodeling for nuclear cloning in mammals. In nuclear cloning, the nucleus of an embryonic stem cell or a somatic cell is put into the cytoplasm of an enucleated oocyte. Particularly in the case of a somatic nucleus, chromatin is configured for a highly specialized program of gene expression rather than for normal embryonic development.  Hence it must be correctly de-programmed and re-programmed in the oocyte. A high rate of failure is likely the result of abnormal gene expression through development. Most of the observed abnormalities are similar to those observed when imprinting is disrupted. Possibly imprinted genes, which are normally only reset in the germline are erroneously programmed in the massive de-program / re-program that occurs prior to development of a nuclear clone.

In the next section of this course, we will look at some specific examples of the cascades in gene expression occuring through the process of development.

4.    Gene silencing

Gene silencing, first revealed in transgenic plants, results from natural mechanisms controlling gene expression. In early observations, transgenic petunia plants engineered to overexpress enzymes for pigment production exhibited less and variable, rather than more, pigmentation. Experimental work in many different organisms including plants, fungi and C. elegans  uncovered two basic pathways of silencing  - transcriptional (DNA based, in the nucleus) and post-transcriptional (RNA-based, in the cytosol). We now know these pathways share fundamental mechanisms that involve the generation of short interfering RNAs (siRNA). (reviewed by Denli and Hannon, Trends Biochem Sci 28:196; Mello and Conte, Nature 431; 338). 

In post-transcriptional silencing, double-stranded RNAs with similarity to a target transcript are cleaved to siRNA by RNase III-type enzymes (sometimes called Dicers). The siRNAs act as guide RNAs, guiding the RNA-induced silencing complex (RISC) to perform endonuclease cleavage of full-length transcripts.  In plants, some form of double-stranded RNA can also move systemically, to expand the silencing signal.

Some of the strongest evidence for this model comes from work with interfering RNA (RNAi) in C. elegans, drosophila and arabidopsis. RNAi results from the transcription of a transgenic region engineered to contain an inverted repeat. The transcript can then fold back upon itself to form a double-stranded RNA.  RNAi effectively silences endogenous and transgene loci. Silencing by RNAi is much more effective than silencing by a single gene copy in either the sense or antisense orientation. For example, in arabidopsis  an RNAi construct silenced the endogenous agamous gene in 99% of transgenic plants, whereas a single-copy antisense construct silenced the endogenous agamous gene in 2% of transgenic plants (Chuang and Meyerowitz, Proc. Natl. Acad. Sci. USA 97:4985).  Recently siRNAs were shown to be effective silencers in mammalian cells (Elbashir et al. Nature 411:494-498) and may, in the long run, prove effective in gene-specific therapies.

In cases of silencing that do not involve RNAi transgenes, RNA dependent RNA polymerases (RdRPs) encoded either by the cell or by infecting viruses, copy transcripts to create double-stranded RNA, which can then be processed into siRNA. Screens for mutations in genes necessary for silencing identified genes endoding RdRPs in arabidopsis, neurospora, chlamydomonas and C. elegans. An alternative source of double-stranded RNA that can be cleaved to siRNA is deliberate or aberrant antisense transcription.  In addition to mutations disrupting RdRPs (see above), mutations disrupting genes encoding RNase III - type ('Dicer') enzymes were identified in C. elegans. Mutations that disrupt RNA silencing implicate the 'Argonaut'  class of basic proteins as components of the RISC. Argonaut proteins potentially bind RNA and particpate in protein-protein interactions.

Post transcriptional gene silencing is likely a defense mechanism against viruses in plants. This would explain how many different transgenes derived from viruses are capable of protecting plants from virus infection. Not surprisingly, some plant viruses have evolved means of  suppressing silencing (reviewed by Vance and Vaucheret, Science 292:2277-2280).

More recently,  short RNAs encoded in eukaryotic genomes and termed microRNAs (miRNA) have been implicated in the cleavage of mRNAs by a mechanim similar to siRNA-guided RNA cleavage. Genomically encoded miRNAs in some cases regulate gene expression at the translation level, by binding to RNAs and . preventing translation, rather than promoting cleavage. Control of gene expression by miRNAs is a key aspect of normal growth and development (reviewed by Carrington and Ambros, Science 301: 336; Benfy, Nature 425:244-245).

There are mechanistic links between transcriptional and post-transcriptional gene silencing (reviewed by Gendrel and Colot, Curr. Opin. Plant Biol. 8:142; Mello and Conte, Nature 431:338). In transcriptional gene silencing, promoter regions become methylated and genes are not transcribed. It is becoming apparent that siRNAs in the nucleus function to guide the methylation of promoter sequences. Methylation, in turn, alters patterns of gene expression. Early evidence in support of this concept came from the observation that plant viroids (noncoding RNAs with considerable double-stranded structure) mediated methylation of viroid-homologous transgenic sequences in the tobacco nucleus (Wassenegger et al. 1994. Cell 76:567-576). Recently Volpe et al. (2003. Chromosome Res. 11:137-146) demonstrated that RNAi machinery is required for methlyation, silencing and heterochoromatin formation essential to the development of functional centromeres in Schizosaccharomyces pombe.

Transcriptional gene silencing may be a natural means of defense against transposable elements. In an elegant demonstration, Hirochika et al. (Plant Cell 12:357-368) deployed a tobacco retrotransposon (Tto1) in arabidopsis. After an initial increase in copy number, the Tto1 elements became methlyated and silenced. Introduction of the ddm1 mutation (which disrupts methylation) resulted in hypomethlyation and re-activation of the transposon.

Gene silencing is a problem when the desired effect is the over-expression of a gene product. However, it provides a very convenient means of "knocking out" gene expression for experimental or commercial purposes. For example, important strategies for engineering virus resistance involve the deployment of nuclear transgenes designed to silence the expression of viral genes in the cytoplasm.