Megabats, microbats and the most interesting gene in the genome

Science writers typically have a few favourite stories: scientific detective stories or debates that intrigue at first encounter, then exert a siren call on the psyche for years, even decades, to come.

The story of Professor Emeritus Jack Pettigrew's quest to confirm that fruit bats (Megachiroptera) are flying cousins of primates has been a personal favourite since I first wrote about it in The Age the early 1990s. Both a detective story and a debate, it has twisted and turned for nearly 18 years, always intriguing, but without final resolution.

Pettigrew's story, which featured in the January/February edition of Australian Life Scientist, clearly appealed to our readers as well - it attracted the most comment of almost any article in recent times.

Just after ALS went to press, Pettigrew drew my attention to a new study, published in the open-access on-line journal, Public Library of Science 1 (PLoSOne), titled 'Accelerated FOXP2 Evolution in Echolocating Bats'.

The letters 'FOXP2' in the title leapt out of the page. The forkhead transcription factor gene FOXP2, known to some as the Chomsky Gene, has strong claims to the title of Most Interesting Gene in the Human Genome.

FOXP2 is active during early embryonic development in the developing gut, heart, lungs and, most significantly, in the brain. One of most highly conserved genes in vertebrates, it is essentially invariant among terrestrial vertebrates including reptiles, birds and mammals.

So a decade ago, when researchers learned that evolution had co-opted FOXP2 to a unique and remarkable role in the human brain, I was instantly intrigued - along with all those evolutionary geneticists sifting the human genome in search of the key genetic changes that made us so different from our higher primate cousins, chimpanzees, bonobos and gorillas.

Like most mammals, mice have the archetypal FOXP2, but in humans, three mutations have occurred in a particular exon of FOXP2 over the 80 million-odd years since the ancestors of mice and men parted evolutionary company.

My first encounter with the then-unidentifed FOXP2 gene was at a research conference at the Menzies Hotel in Melbourne in 1997. Professor Anthony Monaco, a UK Wellcome Trust researcher, gave a talk on his search for a mutant gene at a locus on chromosome 7, that he was calling Speech7. The mutation caused a severe speech disorder in three generations of the KE family.

In 1996 Monaco was hunting for genes involved in autism when clinicians at the Institute of Child Health in London drew his attention to a family in which three generations suffered from a severe inherited speech disorder.

The dominant mutation - it affected around 50 per cent of KE family members - disrupted the finely coordinated movements of the muscles, tongue and lower face required to speak intelligibly.

Further studies showed the mutation also affected areas of the brain involved in comprehending and producing grammatical speech - the "language engine" that US linguist Noam Chomsky theorised is hard-wired into the neural circuitry of the human brain.

Affected KE family members lack simple grammatical rules for spoken and written language, such as creating plurals from singular nouns or knowing that a person who builds is a "builder". They also perform poorly at language tasks, such as writing down a list of words beginning with the letter 'b'.

At the International Congress of Genetics in Melbourne, in 2003, US geneticist Molly Przeworska, from Svante Paabo's team at the Max Planck Institute for Evolutionary Anthropology in Germany, described her research to determine which human genes have been under the strongest selection pressure in the relatively recent past.

When a mutant allele arises that confers a major advantage in survival and reproduction, it will eventually dominate in a freely interbreeding population, at the expense of all other alleles. The phenomenon is called a selective sweep.

When a selective sweep occurs, genes flanking the new allele are also swept up and passed to the individual's descendants en bloc. In a relatively short period of time, the mutant allele and its time-travelling hangers-on dominate, driving rival linked alleles to extinction. A selective sweep effectively bleaches the locus of variation, so all individuals are invariant at that locus.

Over evolutionary time, the locus accumulates new, random mutations at a predictable rate, that allows the approximate timing of the selective sweep to be estimated.

Of the 112 candidate genes that Molly Przeworska compared for evidence of relatively recent selective sweeps, one gene on chromosome 7 emerged as almost white hot: FOXP2.

---PB--- White hot Fox

The first of the three mutations, presumably in an ancestral primate, occurred too long ago to estimate its age. The second and third mutations are quite recent, and exclusive to humans: their absence from chimpanzees, bonobos and gorillas means they occurred after the human and chimp lineages diverged around seven million years ago.

Przeworska estimated that Mutation No 2 occurred soon after the human-chimp divergence, hinting that FOXP2 may have had an influential role in the split. A newly acquired capacity for complex vocalisation would have been invaluable as our ancestors abandoned their aboreal niche in the African rainforest to exploit a new resource-rich but hazardous savannah-woodland environment.

Przeworska's estimate involved some rubbery assumptions about the size of human populations, reproductive age and reproductive success. It indicated that the third and most recent FOXP2 mutation occurred only 100,000 to 200,000 years ago - around the time that humans are thought to have developed the capacity for complex language.

It was too recent by half: last October, Svante Paabo's group achieved the remarkable feat of sequencing the first nuclear gene from a Neanderthal. They succeeded in recovering and sequencing FOXP2 from a Neanderthal skeleton from a cave in northern Spain. It proved to be identical to that of modern humans.

The timing of the divergence of pre-modern humans and Neanderthals is uncertain: it could have been 365,000 years ago, or as long ago as 780,000 years. Fossils from Portugal and Spain complicate the issue, by providing clear evidence of interbreeding between modern humans and Neanderthals in Europe less than 40,000 years ago.

But if Paabo's Spanish fossil is the real McCoy, the clear implication is that Neanderthals possessed the neuromuscular control mechanisms to produce complex language at least 300,000 years ago.

This is approximately the time that early H. sapiens began making more sophisticated stone tools, after more than a million years of technological stasis. Something interesting seems to have happened to human brain function during the Middle Paleolithic Transition. With complex language, humans could progressively accumulate and transmit complex cultural knowledge to each succeeding generation.

With the PloSOne paper last September, the FOXP2 and bat evolution stories - two of my favourite science stories - intersected, in a most unexpected and intriguing way. The gene that made humans human seems also to have lit the evolutionary fuse for the development of ultrasonic vocalisation and echolocation in microbats.

---PB--- Theories bearing fruit

After rodents, microbats are the most speciose group of mammals, comprising 25 per cent of all mammal species on the planet today. But as the most recent issue of ALS described, it now seems very likely that fruit bats are not true bats, but airborne, upside-down primates with leathery wings.

In the late 1980s, Jack Pettigrew asked how his opponents could claim that DNA hybridization evidence supported the monophyly of fruit bats and microbats, when it could not even dissect the deep lineages within the microbat clade. In contrast, multiple, independent lines of evidence supported an evolutionary link between fruit bats and primates.

Why, for example, had fruit bats dispensed with ultrasonic sonar, the microbat superweapon? Why is the fruit bat wing supported by greatly elongated fingers, a primate characteristic, when the microbat wing is braced by elongated metacarpals - the bones between the wrist and the fingers?

And why does the fruit bat eye, and the wiring of the optic nerve to the brain's visual centres, conform to the bizarre Heath-Robinsonian primate model, not the microbat model?

These were just three of 50-odd differences listed by Pettigrew, a list recently extended by the discovery that fruit bats have a menstrual cycle - another almost exclusively primate trait.

Not much more than a disputed comparative DNA dataset now argues against a large, interlaced body of evidence that two mammalian lineages - one sharing a 70-million year old ancestor with primates- independently evolved flapping flight.

In the early 1990s, Pettigrew nominated colugos, a strange gliding mammal from south-east Asia's rainforests, as the likely ancestors or a sister group of fruit bats.

The relationship of colugos to other eutherian mammals was a taxonomic mystery - like primates, they comprised an orphan order, with no known sister order. Pettigrew's suggestion was remarkably prescient. A recent comparative DNA analysis shows colugos are the sister group to primates. They are our closest non-primate relatives.

Unfortunately, the analysis did not include fruit bats, because the authors did not explore the possibility that they might be related to colugos.

Until late last year, DNA-based approaches were not up to the task of dissecting the fine detail of evolutionary relationships within the Microchiroptera. In September, a British-Chinese research collaboration found a scalpel in FOXP2.

The study involved zoologists and evolutionary geneticists from East China Normal University in Shanghai, the Institute of Zoology and Graduate University of the Chinese Academy of Science, the University of London, and the University of Bristol.

It shows that FOXP2 is uniquely variable in microbats, among all mammals, and that the variation almost certainly reflects the evolution of multiple modes of echolocation. It also shows that distantly related microbat lineages have independently converged on very similar echolocation techniques.

According to Li et al, microchiropteran echolocation divides broadly between two modes: predominantly low-duty cycle, in which vocalisation is switched on for around 20 per cent of the time, and is frequency modulated (FM) - i.e. it sweeps up and down through a range of frequencies; and predominantly high duty cycle, in which vocalisation is switched on more than 30 per cent of the time, at a constant frequency.

---PB--- Yin and yang

There are six distinct and presumably ancient microbat lineages, each of which has evolved its own version of sonar, based on ultrasonic vocalisation and echolocation. Fruit bats, with the singular exception of members of the genus Rousettus, do not echolocate.

Rousettus uses echolocation to find its way around in the dark caves where it lives but it employs audible tongue clicks, not ultrasonic squeaks. Pettigrew argues that Rousettus evolved echolocation relatively recently, and independently of microbats.

On comparative DNA evidence, molecular taxonomists have proposed a yin-yang scheme that groups all bats within two suborders: Yinpterochiroptera, which includes all fruit bats and members of the rhinolophoid superfamily of microbats, and the Yangochiroptera, which includes all other echolocating microbat lineages.

Ultrasonic-echolocating Yinpterochiroptera - excluding fruit bats - are mainly constant-frequency, nasal emitters that can compensate for Doppler-shifted echoes as their flight speed varies.

In contrast, most Yangochiroptera employ orally emitted, ultrasonic, frequency modulated vocalisation.

But there are many exceptions. Sonar adaptations like nasal emission and whispering echolocation, compensation for Doppler shifted echoes, and passive listening to localise prey, have clearly evolved multiple times in phylogenetically distant groups.

The ancestor of microbats - a glider - diverged around 80 million years ago from other major mammalian lineages in the super-order Laurasiatheria, which includes hedgehogs, moles, shrews, cetaceans (whales, porpoises and dolphins), ungulates like pigs, hippopotamus, camels, horses and ruminants, pangolins and carnivores.

The major microbat lineages arose during a burst of evolutionary radiation in the Eocene, which may have been driven by the development of ultrasonic echolocation.

The announcement in February of the discovery of a primitive fossil microbat in 52-million year old early Eocene rocks in Wyoming's Green River Formation allows the evolutionary radiation of echolocating microbats to be accurately dated.

Onychonycteris finneyi bears a strong physical resemblance to latter-day microbats, except that it has claws on all five fingers, compared with only one or two in modern bats. It also has longer hind legs, and shorter forearms, suggesting it was formerly a climber that dangled beneath branches, like sloths.

But O. finneyi lacked the ear anatomy to detect ultrasound -- it may have been a daytime hunter, that relied on eyesight to hunt insects on the wing. Icaronycteris, the earliest microbat previously known from the fossil record, came from the same formation. Just 2 million years after O. finneyi, it was hunting with ultrasonic sonar.

Li et al observe that echolocation places extreme demands on bats' sensory and motor systems. A hunting microbat can emit up to 200 ultrasonic chirps per second, interpret the returning echoes, and make appropriate motor responses, including changes in flight direction and speed, all within a few milliseconds.

These demands have placed extreme selection pressure on the gene that coordinates it all - FOXP2 - that is apparent in the highest level of sequence variation in any mammalian order. Interestingly, Li et al observe that bats are among only few groups of vertebrates that exhibit vocal learning, a likely precursor to language.

The two most variable exons are exon 7 - the three human mutations also occur within this exon - and exon 17.

Only one lineage of bats has no variation in exon 17: Professor Pettigrew's "flying primates", the fruit bats. Like primates, they share the archetypal mammalian version of exon 17.

---PB--- Monophyly v paraphyly

When Li et al reconstructed the ancestral FoxP2 amino acid sequences for representative lineages of yinpterochiropterans and yangochiropterans it proved to be identical to the archetypal sequence found in most mammals - after excluding fruit bats.

Supporters of bat monophyly seem to have been deceived by one of nature's most spectacular examples of convergent evolution.

In FOXP2, molecular geneticists have finally acquired a scalpel to dissect out the deepest evolutionary relationships within the microchiropterans, and to test whether the supposed relationship between rhinolophoid microbats and fruit bats is real, or, as Pettigrew argues, an artefact of a mutational bias common to all species that fly, including insects, birds and mammals.

The former conclusion makes little sense: if the major lineages of microbats share a common ancestor, how can fruit bats be related to just one lineage - rhinolophoids - but not the others?

Pettigrew posted a commentary on the PLoSOne website, titled, FOXP2 Destroys the Microbat Paraphyly Hypothesis: a New Tool for Bat Phylogeny.

He said that the new study was weakened by its assumption that bats are a paraphyletic group - that is, a group in which all species descend from a common ancestor (monophyly), but one clade is separated from the core group.

Reptiles (snakes, lizards, turtles, dinosaurs) are a paraphyletic group, if birds - the descendants of theropod dinosaurs - are placed in Class Aves, but reptiles are a monophyletic group when birds are included.

Pettigrew again argues that microbat paraphyly is an artefact of a mutational bias that arises from the extreme energy demands of flight. The bias loads the genomes of flying creatures - whether mammal, bird or insect - with a surfeit of A-T base pairs.

In DNA hybridization studies, these A-T regions undergo complementary base-pairing purely by chance, and the extra "stickiness" makes fruit bats appear more closely related to microbats than they really are.

Fruit bats appear closely related to Rhinolophoid microbats, which includes the families Hipposideridae (Old World leaf-nosed bats), Rhinolophidae (horseshoe bats) and Megadermatidae (false vampire bats).

Pettigrew believes this apparently close relationship is due to the fact that Rhinilophoid bats, which include most of the larger microbat species in the world, have the highest A-T bias of any microbat group at - rivalling that of fruit bats, which have the highest A-T bias of any mammal (75 per cent).

Remove fruit bats from the Rhinolophoid clade, and take account of the fact that all microbat DNA is also A-T rich, and the major lineages of microbats will also appear to have diverged relatively recently.

If FOXP2 is one of the most highly conserved genes in mammals, Pettigrew says the extreme variation found in the gene in microbats is more likely to reflect an ancient radiation that gave rise to the major lineages of microbats.

Pettigrew notes that while megabats do vary from the standard mammalian FOXP2 "recipe", not one megabat amino-acid substitution is shared with any microbat - not even with their supposed cousins, rhinolophoids.

The microbat variation in FOXP2 is concentrated in two exons of the gene: 7 and 17. Li et al do not provide details of any differences in exon 7, but found that exon 17 is identical to the mammalian consensus sequence.

Unless mutation has run backwards and repaired the mutations, this means megabats cannot be descended from, or related to, rhinolophoid microbats, their supposed sister group within the Yinpterochiroptera. And it is even less likely that they are related in any way to the Yangochipteran microbats.

Pettigrew says the "impressive" divergence of the highly conservative FOXP2 gene in microbats is most likely explain by an ancient origin of microbats - much earlier than 50-million year old Icaronycteris, which had wings and cochlears like modern echolocating microbats.

He cites immunological studies of proteins that place the divergence of microbats around 100 million years ago, a figure that accords with the variation found in FOXP2.

And there is also the circumstantial evidence of a 75-million year old noctuid moth fossil that had already evolved a thoracic "ear" to detect ultrasonic bat sonar - mantises, which evolved in the early Cretaceous (~120mya) also have a thoracic ear to detect bat signals.

Pettigrew has his own comparative DNA study of microbats, fruit bats, colugos and primates in press. Until now, nobody had thought to test relationships between these groups, because bat monophyly rules, and Pettigrew's "flying primate" hypothesis was considered heretical.

When last year's comparative DNA study finally linked colugos to primates, the world in which bats evolved just once, wobbled perceptibly on its North American axis. Pettigrew's forthcoming study could overturn it.