1. Early Work on Genetics
and Developmental Polymorphism in Rotifers
I did undergraduate research with Tracy Sonneborn and John Preer on the
genetics of Paramecium. For my thesis, I initiated studies
of the rotifer Asplanchna, which reproduces sexually or asexually
(via parthenogenesis), as a potential experimental system. I did some genetics
on both quantitative and single-factor traits, and studied the roles of
dietary vitamin E and pheromones in controlling the morphogenesis of the
female embryo and determining whether it will develop as a sexual or asexual
(parthenogenetic) female (reviewed in Birky and Gilbert 1971). This work
initiated my interest in the evolutionary advantages (and disadvantages)
of sexual reproduction.
2. Mechanisms of Organelle
Heredity
After serendipitously detecting mitochondrial DNA (mtDNA) synthesis in the
rotifers and the transfer of mtDNA from nurse cells to maturing oocytes,
I turned to yeast (Saccharomyces cerevisiae) as a better experimental
system for studying organelle genetics. I demonstrated that mitochondrial
genes show intracellular selection and random drift of gene frequencies
within single cells. Random drift was also demonstrated in chloroplast genes
in Chlamydomonas. From these studies I developed the general
concept that organelle genes differ from nuclear genes in replication and
partitioning at cell division are relaxed, with genomes selected randomly
for replication and partitioned randomly to daughter cells. I showed how
this difference accounts for the differences in heredity between nuclear
and organelle genes, and argued that it is a natural consequence of the
origin of eukaryotic cells (Birky 1983, 1994).
This concept is extended in the manuscript submitted to Nature,
in which I show that all genomes fall into two classes. Stringent
genomes, such as the nuclear genomes of eukaryotes and prokaryotes and the
kinetoplast of trypanosomes, are those in which each chromosome is replicated
once in every cell cycle, because a replication origin can not re-initiate
once it has replicated; then one copy of each chromosome is partitioned
to each daughter cell (e.g. by mitosis). Relaxed genomes, such
as those of mitochondria, chloroplasts, and plasmids, are replicated randomly
so that some copies may replicate more than others by chance; and are partitioned
randomly to daughter cells. This difference has numerous consequences. For
example: (i) stringent genomes maintain heterozygosity, while relaxed genomes
lose it (vegetative segregation, plasmid incompatibility); (ii) stringent
genomes can become larger than relaxed genomes because they can have multiple
replication origins on each chromosome; (iii) relaxed genomes are subject
to intracellular selection for small size and entire relaxed genomes can
be selfish; and (iv) sex is less important for relaxed genomes. I will write
several more papers on this topic, giving more details about the consequences
of relaxed vs. stringent behavior for cell and developmental biology, genetics,
and evolution.
3. Evolution and Population
Genetic Theory for Organelle Genes
The theoretical studies began with population genetics (Birky, Maruyama,
and Fuerst 1983; Birky, Fuerst, and Maruyama 1989). We showed how existing
theory for nuclear genes could be adapted to organelle genes by defining
new effective gene numbers and migration parameters that take into account
the effects of vegetative segregation and uniparental inheritance. Maruyama
and Birky (1991) obtained a simple equation showing the effects of hitchhiking
on gene diversity in organelle genomes and others with complete linkage.
4. Experimental Studies of Molecular Evolution and Population Genetics
Banks and Birky (1985) did the first large-scale measurement of organelle gene diversity in a natural plant population. Later my students and I began to study molecular evolution in Polytoma, an alga that is closely related to Chlamydomonas but lacks chlorophyll and the ability to do photosynthesis, but retains a plastid (leucoplast) with DNA and protein synthesis machinery. This organism offers a unique opportunity to compare the rates and mechanisms of molecular evolution in the presence and absence of selection, because the plastid genes for protein synthesis are functional and subject to selection, while the remnants of the photosynthetic genes no longer have a function and are not subject to selection. We used nuclear small-subunit rDNA sequences to construct a phylogenetic tree showing that Polytoma species arose at least twice from Chlamydomonas-like ancestors (Rumpf et al. 1996). Pam Mackowski extended this phylogenetic analysis to a related nonphotosynthetic genus, Polytomella (Birky and Mackowski 1999). This organism has four flagellae, a trait shared within the Chlamydomonadaceae only by the green genus Carteria, which is bi-phyletic. We found that three isolates of Polytomella form a clade well separated from both Carteria clades and from both Polytoma clades. It thus represents a third independent instance of the loss of photosynthesis, and a third indeplendent swittch from two flagellae to four flagellae. Graduate student Dawne Vernon analyzed two leucoplast genes, coding for small-subunit rDNA and elongation factor Tu, from several strains of Polytoma. Her data show that these genes are still functional (Vernon et al. 1999a); nevertheless their evolution has been accelerated in a nonphotosynthetic lineage. The acceleration is due to relaxed selection on on both nonsynonymous and synonymous substitutions, but the effect is greater on synonymous substitutions. This acceleration is accompanied by a switch from highly biased codon usage in the green Chlamydomonas to a nearly unbiased codon usage (Vernon and Birky 1999b). The reduced codon bias is probably a consequence of the fact that leucoplasts translate many fewer proteins than do chloroplasts, so that fewer molecules of elongation factor Tu (thetufA gene product) are required. This gene has thus switched from being highly expressed to weakly expressed and synonymous substitutions have changed from detrimental to neutral. The additional mutations seen in the rrn16 may also be ones that reduce the rate of transcription. Both genes may also tolerate mutations that make them less effective or less accurate in translation.
5. Evolutionary Consequences of Asexual Reproduction
Birky and Walsh (1989) used a simple mathematical treatment and computer simulation to show the rate of molecular evolution at a site is affected by selection at linked or unlinked sites in the background genome; neutral substitutions are unaffected, while the substitution rate is increased for detrimental mutations and decreased for advantageous mutations. Subsequently we simulated realistically large populations and small mutation rates on a Cray; the results (unpublished) showed that the classic diffusion equation for the fixation probability of a new mutant gene seriously overestimates the fixation probability for advantageous genes, and underestimates it for detrimental genes. Maruyama and Birky (1991) obtained a simple equation showing the effects of hitchhiking on gene diversity in organelle genomes and others with complete linkage.
An important consequence of long-term asexual reproduction has gone largely unnoticed: in a diploid (or polyploid) organism, the two (or more) alleles in an individual or clone will accumulate different neutral mutations and gradually diverge from each other. This allelic sequence divergence can be limited by mitotic recombination and ploidy changes. Limited or not, it is likely to confound the phylogenetic analysis of sequences (Birky 1996 Genetics). Moreover, homologous chromosomes will diverge in structure. Bob Rumpf showed that there are at least 11 different alleles in each of two species of the highly polyploid and putatively asexual Acanthamoeba. However, sequence divergence was well within the range found within species in sexual organisms (Birky and Rumpf, in preparation).
A review of uniparental inheritance of organelle genes (Birky 1995 Proc. Nat. Acad. Sci. USA) includes the first phylogenetic analysis of this trait. Uniparental vs. biparental inheritance show a frequent character reversals and homoplasy, suggesting that selective pressures on the trait vary greatly among lineages. Although biparental inheritance and recombination, i.e. organelle sex, has advantages for organisms and species, these advantages are usually small because of the small size of organelle genomes. Organelle sex is thus easily lost as a consequence of selection for reproductive traits such as oogamy that cause uniparental inheritance, or selection against the spread of cytoplasmic parasites and selfish organelle genes or genomes.
Current Research and Future Plans are summarized on the Birky Lab home page.