From the Department of Psychiatry, The University of Texas Southwestern Medical Center at Dallas.

Corresponding author: Gregory G. Dimijian, MD, 3277 Brincrest Drive, Dallas, Texas 75234 (e-mail: dimijian@home.com).

Editor's note: See interview of the author on p. 391.

BACTERIAL ANCESTRY OF CELLULAR ORGANELLES

Among the most intimate of symbioses are those between eukaryotic cells and their mitochondria and chloroplasts. Mitochondria are the powerhouses of eukaryotic cells, the little organelles that manufacture adenosine triphosphate (ATP), the universal energy currency in eukaryotes. Chloroplasts are organelles of plant cells that carry out photosynthesis, converting light energy from the sun to chemical energy and “fixing” the carbon of carbon dioxide in organic compounds.

Both organelles carry abundant evidence that their ancestors were free-living bacteria. They are the size of bacteria and have a remnant genome of their own, consisting of a loop of DNA. They replicate on their own in the cytoplasm, independently of the host cell. The cell has only 2 copies of most nuclear genes, but in most cells there are hundreds to thousands of mitochondria. Mitochondrial DNA (mtDNA) is so much more abundant than nuclear DNA that it is often the only DNA that can be recovered from fossils. A significant part of our genome thus lies outside of the 46 chromosomes, a fact often forgotten in descriptions of human genome sequencing (Figure 1).

Mitochondria and chloroplasts, relict bacteria that moved into precursors of eukaryotic cells, became subcellular organelles bound inextricably to their host. The symbiotic union became so strong that many genes migrated to the cell's nucleus, where they code for proteins bearing signals ensuring that they are targeted back to the organelles.

The endosymbiont hypothesis that posits this origin of mitochondria and chloroplasts was originally championed by the biologist Lynn Margulis in the middle decades of the 20th century, long before it became widely accepted. She now writes: “The branches on the tree of life do not always diverge. One branch can merge with another, and from these unions new limbs can grow, unlike anything seen before” (2).

Mitochondrial genome sequences point to a single ancestral genome closely related to the rickettsial subdivision of the a-Proteobacteria, a group of obligate intracellular bacteria that include Rickettsia and Ehrlichia, the closest known living relatives of mitochondria. About 2 billion years have elapsed since free-living a-Proteobacteria became permanent passengers inside their host cells. As cells diversified, their mitochondria also became more diverse, with different mitochondrial lineages retaining different traces of their free-living past (3).

Chloroplast genomes, in contrast, are most like those of cyanobacteria, photosynthetic bacteria that were once called “blue-green algae.” Chloroplasts are the photosynthetic specialists of a family of membrane-bound plant cell organelles called plastids, all of which are believed to have evolved from cyanobacteria.

Lynn Margulis elaborates on the evidence of ancestry:

Bacteria, merging in symbiosis, leave us hints of their former independence. Both mitochondria and plastids are bacterial in size and shape. Most importantly, these organelles reproduce so that many are present at one time in the cytoplasm but never inside the nucleus. Both types of organelles . . . not only proliferate inside cells but reproduce differently and at different times from the rest of the cell in which they reside. Both types, probably 1000 million years after their initial merger, retain their own depleted stores of DNA. The ribosomal DNA genes of mitochondria still strikingly resemble those of oxygen-respiring bacteria living on their own today. The ribosomal genes of plastids are very much like those of cyanobacteria (4).

Margulis concludes with a jolting insight: we have Lamarckianism after all--the inheritance of acquired genomes. This is not exactly the scenario proposed by Lamarck, the inheritance of environmentally acquired traits, but is even more remarkable--the acquisition of another organism and its genes.

Lewis Thomas, who died in 1993, wrote eloquently on mitochondria and chloroplasts:

There they are, moving about in my cytoplasm, breathing for my own flesh, but strangers. They are much less closely related to me than to each other and to the free-living bacteria out under the hill. . . . The usual way of looking at them is as enslaved creatures, captured to supply ATP for cells unable to respire on their own, or to provide carbohydrate and oxygen for cells unequipped for photosynthesis. This master-slave arrangement is the common view of full-grown biologists, eukaryotes all. But there is the other side. From their own standpoint, the organelles might be viewed as having learned early how to have the best of possible worlds, with least effort and risk to themselves and their progeny. Instead of evolving as we have done . . . they elected to stay small and stick to one line of work. To accomplish this, and to assure themselves the longest possible run, they got themselves inside all the rest of us.

It is a good thing for the entire enterprise that mitochondria and chloroplasts have remained small, conservative and stable, since these two organelles are, in a fundamental sense, the most important living things on earth. Between them they produce the oxygen and arrange for its use. In effect, they run the place (5).

Mammalian mtDNA has a mutation rate 10 times that of nuclear DNA. What can explain the difference? Mitochondria are highly exposed to free oxygen radicals generated by their intense metabolic activity, and they have no effective DNA repair system. These handicaps may have favored transfer of most mitochondrial genes to the nucleus. A high mutation rate may also have resulted from the rarity of introns (noncoding sequences) in mtDNA; a mutation is therefore more likely to strike a coding sequence. There is evidence that mitochondrial mutations accumulate over a person's lifetime and contribute to a decline in energy production in many tissues (6).

In keeping with these findings, we are harshly reminded of mitochondrial diseases, highly variable syndromes characterized by degenerative changes in muscle, eye, and central nervous system. Mitochondrial diseases are maternally inherited because most human mitochondria are transmitted only in the cytoplasm of the egg. There is preliminary evidence of limited paternal transmission of mtDNA, and nuclear mitochondrial genes are transmitted in both egg and sperm chromosomes, but there is no clinical evidence of paternal transmission of mitochondrial diseases (7).

Tracing mtDNA into the past is “cleaner” than tracing nuclear genes, because little or no recombination of mitochondrial genes occurs between mitochondria of males and females. “Mitochondrial Eve” was thus proposed, the hypothetical most recent female ancestor of contemporary humans. If we ever succeed in finding a reliable path to a Mitochondrial Eve through the jungle of mtDNA branchings, she would represent the point in the past at which all contemporary human mtDNA originates. But there are flaws in the argument that she is our true ancestor. She need not have contributed any other genes to contemporary humans. Although every gene must have a common ancestor, the ancestors of different genes could be different individuals living at different times. Herein lies the danger in relying on a single genetic locus to reconstruct human history or the history of any species. With these limitations in mind, mitochondrial gene trees can nevertheless be useful in phylogenetic and population studies, because they provide higher resolution than nuclear DNA and offer unique information on where humans originated and how their populations spread.

In summary, mitochondria and chloroplasts are remnants of bacteria that have become semi-independent organelles of eukaryotic cells. The endosymbiotic theory is a milestone of 20th-century biology, showing that symbiosis can be a creator of new worlds of organisms, indeed, the very world we know.

Other cellular organelles

Is it possible that still other microorganisms have become incorporated into cells as symbiotic organelles? Margulis thinks so and argues that flagella and cilia were once free-living bacteria. The absence of genes in these organelles makes this hypothesis difficult to test. But other researchers are finding tantalizing evidence of such unions. An unusual self-replicating organelle, the apicoplast, has been identified inside the cells of 3 important human parasites: Toxoplasma gondii, Plasmodium falciparum, and Cryptosporidium (8). The apicoplast has its own maternally inherited DNA and resembles chloroplasts but has no photosynthetic function. Apicoplasts appear to have been acquired by endosymbiosis of a green alga, and they replicate by fission within the cytoplasm of the host cell. Their genomes are abnormally small; have they lost genes to the nucleus? Apicoplast DNA encodes bacteria-like proteins, suggesting that we may be able to derive antiparasitic drugs from agents that act on the apicoplasts. This may explain the susceptibility of the malaria parasite to antibiotics such as doxycycline and azithromycin, which interfere with bacterial protein synthesis. Ciprofloxacin inhibits replication of the apicoplast in T. gondii, and this inhibition blocks parasite replication (9).

The engulfing of cells and co-opting of cell parts may be more common than we think. Marine ciliates can be seen today phagocytizing other single-celled photosynthetic organisms and harvesting their chloroplasts, maintaining the chloroplasts as functioning sugar-producing organelles in their own cells. The smoking gun in this endosymbiotic interpretation is the finding of 4 membranes around the organelle--2 from the green alga and 2 from the alga's chloroplast. The alga and its chloroplast seem to have been engulfed intact. A majority of ciliates and flagellates in the anoxic bottom waters of the Santa Barbara Basin have been found to harbor bacterial symbionts, and the ecosystem has been called a “symbiosis oasis” (10). A vast menagerie of protists, including amoebae, ciliates, flagellates, and other single-celled organisms, awaits an understanding of their complex anatomy, including bizarre organelles that may have originated by endosymbiosis (11).

Changing views of the “tree of life”

What do these new findings suggest about the tree of life? Until recently we imagined a solid trunk representing the earliest life forms, spreading out into an ever-branching crown. The newly revised view is that early organisms constituted a soup of changing genetic entities with promiscuous horizontal (lateral) gene exchange (12). With such beginnings it would be difficult to trace family or gene lineages into the distant past. Carl Woese, who pioneered our understanding of the 3 domains of life (Archaea, Bacteria, Eucarya), puts it this way:

Initially . . . lateral gene transfer . . . was pandemic and pervasive to the extent that it, not vertical inheritance, defined the evolutionary dynamic. As increasingly complex and precise biological structures and processes evolved, . . . the evolutionary dynamic gradually became that characteristic of modern cells. The various subsystems of the cell “crystallized,” i.e., became refractory to lateral gene transfer. . . . Organismal lineages, and so organisms as we know them, did not exist at these early stages. The universal phylogenetic tree, therefore, is not an organismal tree at its base but gradually becomes one as its peripheral branchings emerge. The universal ancestor is not a discrete entity. It is, rather, a diverse community of cells that survives and evolves as a biological unit. This communal ancestor has a physical history but not a genealogical one (13).

If for now we can only postulate that lateral gene transfer characterized the early soup of organisms, we have no doubt that it occurs today between bacteria, enabling genes for antibiotic resistance and virulence factors to spread quickly and easily. Bacteria need not delay the spread of such genes until they reproduce; they can merely spread the instructions around to contemporary cells. Viruses achieve similar genetic exchange during replication, when they trade gene segments among strains; this “antigenic shift” accounts in part for new influenza A strains every year (with pig and avian viral segments reassorting). Viruses also serve as vectors for bacterial genes and for genes we artificially insert into their small genomes. There is a trace of the soup still with us in microbial gene exchange.

VIRAL REMNANTS IN THE GENOME

Vertebrate genomes are cluttered with ghosts of past infections in the form of endogenous retrovirus sequences. These proviruses, as they are called, may be cause for concern when animal organs are transplanted to humans, enabling proviruses to cross the “species barrier.” Matt Ridley writes:

There are several thousand nearly complete viral genomes integrated into the human genome; these HERVs, or human endogenous retroviruses, account for 1.3% of the entire genome. That may not sound like much, but “proper” genes account for only 3%. If you think being descended from apes is bad for your self-esteem, then get used to the idea that you are also descended from viruses (14).

Evolution is opportunistic. One of these genomic “fossils” appears to have been appropriated by mammalian hosts for a useful purpose. One feature of the pathology of retroviruses is their ability to fuse host cells together into a syncytium. Provirus genes are known to be expressed in the placenta, where the fetal-maternal interface, or syncytiotrophoblast, is a thin layer of fused fetal trophoblast cells (15).

Andean mummies 1500 years old have been found to harbor proviral DNA from the human T-cell lymphotropic virus type I (16). An ancient infection thus left its traces in fossil human genomes, evidence that human T-cell lymphoma virus-1 was carried with ancient Mongoloids to the Andes before the Colonial era. In contemporary human genomes there are endogenous retroviruses dating back at least 30 million years which encode a factor analogous to an HIV-1 protein, suggesting that an immunodeficiency virus genome infected primate germ cells long before hominids evolved (17).

Even more common than endogenous retroviruses are retrotransposons, believed to be relict fragments of viral genomes that have acquired the ability to multiply and reinsert new copies of themselves anywhere in the genome. Retrotransposons represent an enormous fraction, possibly 15%, of the human genome. They can act as insertional mutagens by inserting copies of themselves in the middle of genes, causing mutations.

Most eukaryotic genomes are littered with these seamlessly integrated parasitic elements. The term “symbiosis” is not usually used to describe these mergers, but they are clearly examples of ancestral genomes that fused with those of their original hosts.

SYMBIOTIC HUMAN PATHOGENS

Surprising numbers of pathogenic viruses and bacteria have become symbiotic with host organisms for most or all of the host's lifetime, only occasionally causing serious illness (Table). A challenging question is: Why do these symbionts ever produce illness? If pathogen and host have achieved a kind of truce through coevolution, why is the relationship not 100% friendly?

Herpesviruses

Consider the extraordinary achievement of latency, characteristic of herpesviruses. Herpes simplex virus type 1 enters the sensory processes of neurons on the lips and moves into neuronal cell bodies in the trigeminal ganglion, where it may reside for the lifetime of the host. The virus has achieved the ideal state of harmlessly infecting a healthy, active host, which can broadcast it far and wide. At times the virus is truly latent and is not being replicated by host cells; at other times it is replicated and moves down axon terminals onto mucosal surfaces where it is spread to other hosts. Usually it does not cause illness, but occasionally, even in immunocompetent hosts, it causes life-threatening illness, from encephalitis to neonatal herpes. Latency may be terminated if the host is stressed, when viral replication resumes and lesions are produced on the face and in the mouth.

During latency a single viral gene--appropriately named LAT--is abundantly transcribed and appears to block cellular apoptosis. Infected ganglion cells thus cannot initiate their own self-destruction by using the major histocompatibility complex to display viral proteins on the cell surface and are preserved, serving as periodic virus factories (18).

Humans are not unique hosts to symbiotic herpes simplex viruses. Monkeys stressed by capture and transport break out with gingivostomatitis from their own latent herpesvirus strains.

Other herpesviruses display the trait of latency. Varicella-zoster virus causes chickenpox early in life and then enters latency in the dorsal root ganglia. It can later reactivate as herpes zoster (shingles). Studies have demonstrated the identity of the viruses causing both diseases. Human cytomegalovirus, another herpesvirus, causes a latent infection with periodic reactivation for the life of its human host; incidence of host infection reaches 70% in cities of the USA and nearly 100% in parts of Africa. Human herpesvirus 4 (or Epstein-Barr virus) produces a latent B-cell infection for the life of the host, hiding in the very cells that make antibodies. Human herpesviruses 6 and 7 are believed to initiate a lifelong infection in nearly all humans in early childhood (19).

Bacteria

Not to be outdone by viruses, pathogenic bacteria boast many species that have become symbiotic. Neisseria meningitidis is a major cause of bacterial meningitis and septicemia, often spreading on college campuses, with a case fatality rate of 10% to 12%. In spite of its reputation as an aggressive pathogen, it is more often found as a harmless commensal in the human nose and pharynx. It has been estimated that at any given time between 5% and 10% of humans carry the organism. It causes sporadic cases of disease at the much lower incidence of 1 to 5 per 100,000 a year. The meningococcus lives only in humans, passing from 1 person to another in aerosols; there are no other animal or environmental niches that it is known to colonize. It is adapted to living peacefully in humans and has evolved a variety of mechanisms to evade our defenses, including the ability to produce variable cell-surface proteins that provide an ever-changing target for the immune system (20).

Haemophilus influenzae, another symbiotic bacterial pathogen of humans, has no other known natural host. It is among the normal bacterial flora of the pharynx and, to a lesser extent, the conjunctivae and genital tract. Up to 80% of healthy persons are carriers of the bacterium, which only occasionally causes serious illness such as meningitis.

Helicobacter pylori is a spiral bacterium uniquely adapted to the low pH of the stomach. Its ecology is revealed in its unusually small genome, which lacks genes found in other bacteria adapted to more variable environments. H. pylori is spread by vomitus and feces and possibly by flies and is especially common in developing countries. By age 10, >50% of children worldwide carry the symbiotic bacterium. An untreated infection is believed to last the lifetime of the host.

Fungi

Fungi of the genus Candida are normal commensals of humans, found on the skin, in sputum, in the female genital tract, and throughout the entire gastrointestinal tract. They are usually found as single cells without the branching mycelia seen in bread mold fungi. Serious infections, uncommon in healthy persons, include vaginitis, oral thrush, esophagitis, pneumonia, endocarditis, and meningitis.

Pathogens and disease

Many, if not most, microbial pathogens are not symbiotic with hosts. Influenza viruses, for example, cause a time-limited infection and are eliminated by the immune system if the host survives. Yet the closer we look, the more often we discover long-term carrier states and low-grade, chronic, long-lasting infections. Mycobacterium tuberculosis, Plasmodium vivax, and HIV-1 often colonize a host for the better part of its lifetime. Should we consider these pathogens symbiotic? Whatever our preference, we should address the extraordinary fact that many pathogens are adapted to colonize our bodies for the better part of our lifetime. We are normal animals in this respect.

Is disease sometimes the sorry expression of a residual degree of virulence maintained by natural selection because of competition among strains or species of pathogens? If virulence often means a higher rate of replication, a strain of pathogen with a higher virulence might outcompete another strain and be the one we see today. Coevolution with hosts might provide an upper limit to virulence in pathogens, as premature host death would limit spread. Disease would then occur in a minority of hosts but be inevitable because of competition among pathogens.

There is also another cause of disease, of increasing importance in today's world. Crossing the species barrier exposes a pathogen to a new host with which it has had no chance to coevolve, and either the pathogen or the host may be no match for the other. In today's globally disrupted ecosystems there are rampant opportunities for species barriers to be crossed for the first time. The situation is analogous to a perforated colon, in which previously harmless bacteria end up where they don't belong.

SYMBIOTIC COLONISTS IN INVERTEBRATES AND PLANTS: WOLBACHIA AND AGROBACTERIUM

The tiny bacterium Wolbachia spends its entire life within the cells of the ovaries and testes of many arthropod species, not causing illness but nevertheless playing havoc with the sex ratio of its host's progeny. It is passed on to the host's offspring only in eggs, not in sperm. Thus the bacterium's strategy for passing on its genes: skew the sex ratio markedly in favor of females, or spare no guns and do away with males altogether. In either case the strategy works.

Wolbachia achieves this devious end by altering host chromosome behavior, reducing male numbers or completely eliminating males. In the latter case the host becomes asexual; the whole infected population becomes parthenogenetic for as many generations as the bacterium lives in the host's gonads. Many cases of parthenogenesis, once thought to be genetic and permanent, are being found to be infectious in origin. Asexuality in some wasps, for example, can be reversed by treating the wasps with antibiotics or subjecting them to heat in the laboratory, which kills the bacteria. Two genders magically reappear among the offspring. Antibiotics cure asexuality!

How could natural selection bring about this behavior in Wolbachia? One possibility is through competitive exclusion. Once a microbe became capable of invading the ovary and living there, a competing strain that discovered a way to skew the host's sex ratio would multiply more rapidly than the other strain and would be the one we see today.

Wolbachia infection is amazingly common, occurring in members of >90% of known arthropod species, including 5 orders of insects. In most cases the result is a skewed sex ratio, not asexual reproduction.

New strategies of disease prevention sometimes come from the most unlikely places. Studies are under way to use Wolbachia to disrupt transmission of pathogens carried by arthropod vectors. Genetically modified Wolbachia could spread a second maternally inherited gene into the vector population very rapidly, just as it spreads its own genes; the inserted gene could disrupt the pathogen's life cycle (21).

Another symbiotic bacterial species has been harnessed as a gene vector. Agrobacterium tumefaciens, itself a natural genetic engineer, infects a wounded plant and inserts genes from its plasmid into the genome of the host plant. Some of the transferred genes code for the synthesis of hormones that transform plant cells to tumor cells, forming a large “crown gall” tumor that houses the bacteria. Others direct the synthesis of chemicals called opines that cannot be metabolized by the host and that provide nitrogen-rich and carbon-rich compounds to the bacteria. The transferred genes thus arrange for room and board for their original owner, which provides no known service in return.

Human genetic engineers use Agrobacterium as a vector for “artificial” genes to be inserted into plant cells. In the laboratory these genes are integrated into the large bacterial plasmid, converting Agrobacterium into a shuttle for inserting the genes into plants. Some of these genes code for antimicrobial compounds, including the enzyme replicase, which disrupts viruses. Disease-resistant varieties of potatoes have been created in this way. Another gene, removed from the bacterium Bacillus thuringiensis (“Bt”) and inserted into the Agrobacterium plasmid, directs the synthesis of an insecticidal toxin. Inserted by Agrobacterium into the corn genome, it continues to code for toxin in its new host, “Bt corn,” which is resistant to lepidopteran pests.

REMAINING APART: ANT-PLANT MUTUALISMS

Some of the most remarkable symbiotic associations occur between organisms that remain physically separate, with neither colonizing the body of the other. Both remain separate, but their lives and activities are so interdependent that they evolve almost as 1 organism.

Almost half of all insect species are herbivores, which defoliate plants. If evolution is opportunistic, plants should not have failed to notice the power of ants to defend against herbivores, from mammals to invertebrates, including other ants. Because ants live in colonies, they can collectively defend against herbivores with the striking force of a legion of workers, all willing to sacrifice themselves at a moment's notice.

Plants have indeed seduced ants into a dazzling variety of cooperative exchanges. In return for shelter, nectar, or food bodies, a colony of ants can defend every square inch of the plant. Nest cavities are provided in the form of swollen thorns (Figure 2), hollow stems, leaf pouches, and even hollow roots. Nectar or food bodies are provided strategically on growing shoot tips or young leaves, where plant tissues are most vulnerable to browsing. Ants defend against herbivorous insects and vertebrates and even prevent vines from climbing up a trunk and taking over a tree.

In the Peruvian Amazon a light tap on the trunk of a Triplaris tree will bring a rain of stinging ants down onto one's head and shoulders from the foliage overhead. In Costa Rica an Inga plant provides extrafloral nectaries for ants to drink from and receives protection in return. In East Africa hollow thorns develop on whistling thorn acacias even if the plants are grown from seed in a nursery; the thorns provide nest cavities for ants. Two of the ant species that colonize whistling thorns are so militant that within seconds of touching a tree an intruder is attacked by a horde of biting ants.

In 1872 an extraordinary naturalist, Thomas Belt, first proposed that ants protect trees. Belt spent 4 years in the Nicaraguan rain forest, observing plants and animals and keeping careful notes. In his classic account, The Naturalist in Nicaragua, he writes:

These ants form a most efficient standing army for the plant, which prevents not only the mammalia from browsing on the leaves, but delivers it from the attacks of a much more dangerous enemy--the leaf-cutting ants. For these services the ants are not only securely housed by the plant, but are provided with a bountiful supply of food, and to secure their attendance at the right time and place, the food is so arranged and distributed as to effect that object with wonderful perfection. . . . At the end of each of the small divisions of the compound leaflet there is, when the leaf first unfolds, a little yellow fruit-like body united by a point at its base to the end of the pinnule. Examined through a microscope, this little appendage looks like a golden pear. When the leaf first unfolds, the little pears are not quite ripe, and the ants are continually employed going from one to another, examining them. When an ant finds one sufficiently advanced . . . it breaks it off and bears it away in triumph to the nest. All the fruit-like bodies do not ripen at once, but successively. . . . Thus the young leaf is always guarded by the ants; and no caterpillar or larger animal could attempt to injure them without being attacked by the little warriors (22).

Today we call the small food bodies on the acacia leaves Beltian bodies. Belt's hypothesis was dismissed by leading biologists of the time, who retorted that considering ants beneficial to plants was like considering fleas beneficial to dogs. It took experiments by Janzen in the 1960s to resolve this issue and show that when the ants were removed, herbivores decimated the acacias (23). Since Janzen's experiments, ants have been shown repeatedly to increase plant fitness by foraging actively in large numbers over plant surfaces, removing arthropods that consume plant parts.

Caterpillars, like plants, are opportunists in exploiting ants. By providing ants with nectar from glands on their bodies, many caterpillars have gained protection by ants in the same way that plants have. Some of these caterpillars can feed with impunity on plants that otherwise would not tolerate them (poisoning them with toxins synthesized in the leaves). Instead the plants allow them to browse on the leaves, in exchange for the protection provided by the ants they attract (Figure 3). Some caterpillars can even summon ants from nests at the base of the plant by releasing chemicals as signals or by producing vibrations that the ants can feel.

Ant societies have invited other species into their midst repeatedly through evolutionary history. Many ants adopt aphids, mealybugs, or other arthropods that provide a steady diet of honeydew and provide protection in return. Some of these complex associations appear to be 3-way mutualisms, with plant, caterpillar, and ant benefiting in complex and varying ways.

There are pitfalls, however, in making these interpretations. The ants we see on 1 plant may not be the same species that coevolved with the other 2 members of the mutualism, but may instead be interlopers, even cheaters. In addition, the mutualisms may not always be specific but may be more general, with plants and caterpillars sharing ants of different species. The ants may also protect other insects and plants at the same site. Pitfalls include the assumption that ants and plants coevolve; it is more difficult to measure the reliance of ants on plants than vice versa. Under the stress of herbivory, plants have evolved services used by ants, but have the ants also evolved in concert or are they just taking the offerings and behaving like ants?

CHEATERS: PARASITES OF MUTUALISMS

Acacias in Central America are protected by obligate acacia-ants but after a prolonged drought or other disturbance may be temporarily deprived of the resident ant colony. During this vulnerable period a cheater ant species may occupy the plant, feeding on the food bodies and enjoying the shelter of the nest cavities but not protecting the plant. The temporary cheater is a parasite of the ant-acacia mutualism, benefiting from the services of the plant but providing no service in return (24).

A cheater species has evolved in the savanna of East Africa, where whistling thorn acacias offer services to ants. This parasitic ant species (Crematogaster nigriceps) occupies an individual acacia and enjoys shelter and food without providing protection. Its parasitic behavior does not stop there; the ants proceed to prune the tree, removing the growing shoots, and sterilize the tree by chewing off its flower buds. This odd behavior was an enigma to researchers until it became apparent that the ants were preventing growth toward other trees that were colonized by competitively superior ant species, which would stream over to their tree if the trees were in contact (25).

A beetle in Costa Rican rain forests can stimulate Piper ant-plants to produce food bodies, as if ants were present. The beetles take over the plant, exploiting nest sites usually occupied by ants and consuming the food produced for the ants but providing no protection in return. The beetles even prey on any ant brood present on the plant (26).

Parasites of mutualisms constitute a cost to both parties. There is often little defense, and cheaters thrive and diversify.

The biologist Judith L. Bronstein has divided cheaters into several useful categories. Some species cheat exclusively, other species consist of some individuals that cheat and others that cooperate, and still other species consist of individuals that cheat some of the time and not at other times. Mutualist and cheater species are often close relatives from sister taxonomic groups, consistent with the hypothesis that a lineage has escaped the cost of mutualistic investment. The cheater species may even be derived from the mutualist species (JL Bronstein, personal communication).

Cheating may thus be analogous to pathogenicity, both occurring along an unbroken continuum in which a pathogen or cheater evolved from a cooperative species or vice versa. This insight may be revolutionary in exploring disease, as the evolutionary dynamics of cheating vs cooperating in mutualisms may be virtually identical to the dynamics of host-pathogen relations, as both involve selfish evolutionary trade-offs (Figure 4).

LEAFCUTTER ANTS

Among the few organisms that cultivate their own food are leafcutter ants, which maintain fungal gardens in their underground nests. Leafcutters are the dominant plant-eaters of the New World tropics; some 15% of leaves of tropical forests in Central and South America disappear down their nests (Figure 5). There the smaller workers chop up and clean the fragments and mix them with the fungi. The fungi break down the cellulose walls of the plant cells, make the protein available, use some of it themselves, and detoxify some of the plant alkaloids. The ants then feed on the fungi (special nutrient-filled hyphae are provided by fungi cultivated by more recently evolved leafcutters). In return, the fungus is assured a regular food supply of leaves and a sheltered nest and is guaranteed transportation to new nests by the founding queen (Figure 6).

In their evolutionary past the ants captured the fungus, or--as E. O. Wilson reminds us--maybe it was the other way around. In either case they now own each other.

No one suspected that a third mutualist would be found in leafcutter colonies, but a remarkable study has just revealed one. It turns out to be a bacterium that protects the fungal garden from a fungal pathogen (27). Found on the bodies of the ants, the filamentous bacterium produces antibiotics specifically targeted to suppress the growth of the pathogenic fungus. The 3-party mutualism--ant, fungus, and bacterium (from 3 kingdoms--animals, fungi, and prokaryotes)--appears to be an ancient association, the partners having coevolved over the past 10 million years or more. A new queen carries some of the bacteria with her, along with the fungal inoculum, when she leaves the nest to mate and found a new colony.

It may not be a coincidence that the bacteria belong to the genus Streptomyces, from which over half of our antibiotics have been derived. Unlike our antibiotics, however, this one (with its evolving variations) has remained effective for 10 million years.

SLAVE-MAKING ANTS (SOCIAL PARASITISM)

Each day at about 4 PM in the pine-oak-juniper woodlands of the Chiricahua Mountains of southeastern Arizona, raiding columns of Polyergus ants scurry across and under the leaves on the forest floor, following their leader scouts to the underground nest of another ant species of the genus Formica. There the Polyergus queen fights to the death with the Formica queen, eventually killing her (Figure 7). She licks her victim and acquires the scent of the dying queen, after which the Formica workers treat her as their own, allowing her raiding workers to pick up the pupae and carry them off. Back over the forest floor they go, carrying the pupae back to their own nest (Figure 8). When the pupae emerge from their cocoons they will form social bonds with their slave-making nestmates through olfactory imprinting. A visitor to the Southwestern Research Station of the American Museum of Natural History may be privileged to witness this daily event, where it has been studied for over a decade.

The Polyergus slave makers have lost the ability to maintain their nest or care for their young. The queen cannot rear her brood without slave labor. The workers, which conduct the daily slave raids, do little else. They do not forage for food, feed their brood or queen, or even clean their own nest. All of these chores are performed with no complaints by the slaves, which forage for food and regurgitate food to colony members of both species. The slaves even send out scouts to locate a new nesting site when the population of the nest grows too large. When they all move, the slaves carry all eggs, larvae, pupae, and every host adult, even the host queen, to the new nest site (28).

The 2 ant species are symbiotic but their bodies remain separate, like ants and plants, ants and caterpillars, and ants and their fungal gardens. The relationship is both symbiotic and parasitic; the term “social parasitism” is appropriate, with 1 party duped into serving the other. The slave workers help the genes of another species achieve future representation.

This extraordinary scenario is not an aberration. Of the ~8800 known species of ants, some 200 species have evolved some degree of symbiotic relationship with other ants (29). At 1 end of the continuum are temporary parasites and at the other end, the “inquiline” ants. Inquilines are parasitic species that spend their entire lives in the nest of the host species; in the most extreme example the entire worker caste has been eliminated, as have the slave raids. The queen is absurdly tiny and rides on the host queen's back. Males and new queens produced by the diminutive queen copulate inside the host nest; the newly mated queens then locate other host colonies to parasitize, and the cycle is repeated, with complete dependence on the host. The parasitic ants bear the marks of extensive morphological degeneration correlated with loss of function; even the brain is diminished in size. Yet for the survival of their genes they must retain the skills of locating and parasitizing new hosts. Their condition is an evolutionary sink; a return to a nonparasitic mode of life is unlikely. What do they gain? The services of their host. What does the host gain? Nothing, as far as we know, but there is always a chance that we are wrong.

A bizarre feature of these ant symbioses has evaded explanation: parasitic ants generally originate from the closely related ants that are their hosts. This seems paradoxical; how can a species generate its own parasite? One hypothesis proposes that a population splits and becomes geographically isolated; if the 2 resulting populations reunite and cannot interbreed, one may specialize as a parasite of the other (30).

FLOWERS AND POLLINATION

The highly specialized and coevolved relationships between flowers and their pollinators--morphological, physiological, and behavioral--have fascinated biologists ever since they were first described. Pollination is one of the most widespread mutualisms in nature. Early radiations of flowering plants occurred concomitantly with radiations of insects that pollinated them; Cretaceous and Tertiary flower-visiting insects included an impressive variety of beetles, flies, moths, wasps, and bees. Butterflies joined them in the mid-Tertiary. Many flower parts and insect body structures are morphologically convergent, having apparently evolved together (31).

Two extraordinary pollination mutualisms--those of yucca plants with yucca moths, and fig trees with fig wasps--and 1 nonsymbiotic association merit description here.

Yucca plants

The female yucca moth is the sole pollinator of many species of yucca plants. She inserts her ovipositor into the ovary of the flowers and deposits her eggs, then walks up to the stigma of the flower and actively deposits a small amount of pollen. In doing so she makes it possible for seeds to develop; seeds will become the exclusive food of her progeny. The mutualism is so specific that other pollinators are excluded; this has created an absolute mutual dependence of the plant on the moth and vice versa. Neither can reproduce without the other.

This has been called the “perfect” mutualism, with the idea that both parties need and benefit each other. Closer scrutiny reveals conflicts of interest and self-defense; the yucca has developed subtle strategies to prevent exploitation. Fruits often fail to mature if they contain heavy egg loads or low pollen counts; this may be a way to select against moths laying many eggs or providing low-quality pollination. This is the less cooperative face of mutualism (32).

Another fly has been found in the idealistic ointment. When 2 kinds of moths depend on the same yucca, the door is opened for one to become a parasite of both the yucca and the other moth. This very scenario is taking place in Florida, where 2 moth species lay eggs in the same yucca. The cheater moth has split off from the other species, no longer wasting its energy on growing special mouth parts for gathering pollen. Both species inject their eggs under the surface of the fruit, but the cheater species does so later, after the fruit has begun maturing--thus avoiding the yucca's defense. These cheater moths are parasites of the mutualism, their success depending on the mutualism itself (33).

Some biologists would classify this mutualism as nonsymbiotic, because plant and pollinator live apart. It is included in this review because of the tightly interwoven life histories and obligatory association of both partners.

Fig trees

The second pollination mutualism, this one stranger than fiction, has evolved between fig trees and their only pollinators, tiny fig wasps. Each species of fig tree is pollinated only by its own species of wasp, which can breed only inside the fig, at a time when the fig is in flower. The female wasp, bearing pollen, enters a fig “fruit,” pollinates the flowers (which line the inside surface of the fig), and lays eggs, usually dying there. Each larva develops at the expense of a single tiny fig flower. When the new generation of wasps reaches adulthood (still inside the fig), they mate, and females leave, loaded with pollen. The figs then ripen and are eaten by frugivores, which disperse the seeds. As in the case of the yucca and yucca moth, this is a very tight mutualism, and neither fig tree nor wasp can reproduce without the other (23).

Orchids

Orchids are renowned for their intimate and intricate coevolutionary associations with insect pollinators. They provide a lesson, however, in when not to call an association mutualistic. The flowers of Ophrys orchids have evolved elaborate ways of mimicking the appearance, hairiness, and sex pheromones of female bees (34). This visual, tactile, and chemical mimicry entices male bees to attempt to copulate with the flower, and the bee may spend several hours in this activity, even ejaculating, but taking no food for himself. Instead, he flies away with pollen on his body, which he delivers to the next flower. The relationship is thus not mutualistic but parasitic, as only 1 partner is known to benefit. It is also not symbiotic. An evolutionary arms race is likely to be in progress, with male bees being selected to discriminate between real and floral females and the orchid selected for maintaining the ruse.

SHIFTING IDENTITIES

There may be a fine line between pathogenicity and mutualism--between 1 end of the continuum and the other.

There is a growing belief that many mutualistic symbioses originate as parasitic symbioses. A striking example of a transition from pathogen to mutualist has been documented in a fungus that infects plant cells; mutation at a single locus was found in the nonpathogenic strain, which otherwise was genetically identical to the pathogenic strain and retained the same host specificity (35).

Even Wolbachia bacteria may shift along the continuum. The filarial worm Onchocerca volvulus, a leading cause of blindness in Africa, appears to depend on endosymbiotic Wolbachia for its normal fertility. Treating the worms with tetracycline leads to elimination of the bacterial endosymbionts and sterility of adult worms. Worms not originally infected with Wolbachia are not harmed by the antibiotic (36).

For intracellular bacteria, the term “infection” applies to both pathogens and mutualists; in both cases cells are “infected.” In order to understand infection by a pathogen, it may help to understand infection by a mutualist.

Brucella abortus, a mammalian pathogen that causes brucellosis, and Rhizobium meliloti, a phylogenetically related plant mutualist, establish chronic infections in their respective hosts. The same gene product is needed to establish and maintain chronic intracellular infection in both bacterial species. Both bacteria are endocytosed by host cells and live for prolonged periods in intracellular compartments. A pathogen and a mutualist may thus be closely related and utilize similar molecular pathways for infecting the host (37).

CONCLUSIONS

Symbiotic associations between organisms appear to constitute the biological and social milieu of animals and plants in ways that we never expected even a half-century ago. Larger organisms are colonized by smaller ones, and organisms of similar size associate intimately without merging. Genomes have merged and coalesced since the early soup of lateral gene transfer.

Movement along the interactional continuum from antagonistic to cooperative is dynamic and changing, with any 1 symbiotic relationship having both antagonistic and cooperative elements active at the same time. Many mutualistic relationships may have begun as antagonistic ones. Microbes and hosts share overlapping life histories in the context of a dynamic standoff, each looking after its own genetic interests. The consequences range from fatal infections to mutualisms and everything in between. The complex ways in which cheaters exploit mutualisms may provide revolutionary insights into infectious and parasitic diseases, as many cheaters and pathogens are close relatives of species at the other end of the continuum.

Host disease occurs in at least 2 evolutionary settings: 1) a highly coevolved pathogen has titrated its pathogenicity to a level high enough to be competitive with other strains (and other species) and low enough to maximize host survival and spread to other hosts; and 2) a pathogen recently having crossed the “species barrier” may be more virulent than its best interests dictate, but it has not had time to coevolve with the host.

Models predict that long-term interactions are evolutionarily stable only when both interacting species possess mechanisms to prevent excessive exploitation. Evolutionary time is needed, as E. O. Wilson reminds us, for symbiotic bargains to be struck.

The understanding of symbiotic relationships promises new insights into biology and medicine, from the evolution of species to the dynamic relations between microorganisms and hosts.

Acknowledgments

Miriam Muallem, librarian at Medical City Dallas Hospital, provided invaluable expertise in acquiring reference material. Mary Beth Dimijian provided helpful editing advice, and Keith Johansen, MD, made the important recommendation that the symbiosis material be presented in 2 parts.

1. Dimijian GG. Evolving together: the biology of symbiosis, part 1. BUMC Proceedings 2000;13:217-226.
2. Margulis L, McMenamin M. Marriage of convenience. The Sciences 1990;Sept/Oct:31-36.
3. Martin W. Perspectives: evolutionary biology. A powerhouse divided. Science 2000;287:1219.
4. Margulis L. Symbiotic Planet: A New View of Evolution. New York: Basic Books, 1998:39-40.
5. Thomas L. Notes of a biology-watcher: organelles as organisms. N Engl J Med 1972;287:294-295.
6. Wallace DC. Mitochondrial DNA in aging and disease. Sci Am 1977;277:40-47.
7. Morris AA, Lightowlers RN. Can paternal mtDNA be inherited? Lancet 2000;355:1290-1291.
8. Kohler S, Delwiche CF, Denny PW, Tilney LG, Webster P, Wilson RJ, Palmer JD, Roos DS. A plastid of probable green algal origin in Apicomplexan parasites. Science 1997;275:1485-1489.
9. Fichera ME, Roos DS. A plastid organelle as a drug target in apicomplexan parasites. Nature 1997;390:407-499.
10. Bernhard JM, Buck KR, Farmer MA, Bowser SS. The Santa Barbara Basin is a symbiosis oasis. Nature 2000;403:77-80.
11. Dyer BD, Obar RA. Tracing the History of Eukaryotic Cells: The Enigmatic Smile. New York: Columbia University Press, 1994.
12. Doolittle WF. Uprooting the tree of life. Sci Am 2000;282:90-95.
13. Woese C. The universal ancestor. Proc Natl Acad Sci U S A 1998;95:6854-6859.
14. Ridley M. Genome. New York: Harper-Collins, 1999:125-126.
15. Mi S, Lee X, Li X, Veldman GM, Finnerty H, Racie L, LaVallie E, Tang XY, Edouard P, Howes S, Keith JC Jr, McCoy JM. Syncytin is a captive retroviral envelope protein involved in human placental morphogenesis. Nature 2000;403:785-789.
16. Li HC, Fujiyoshi T, Lou H, Yashiki S, Sonoda S, Cartier L, Nunez L, Munoz I, Horai S, Tajima K. The presence of ancient human T-cell lymphotropic virus type I provirus DNA in an Andean mummy. Nat Med 1999;5:1428-1432.
17. Yang J, Bogerd HP, Peng S, Wiegand H, Truant R, Cullen BR. An ancient family of human endogenous retroviruses encodes a functional homolog of the HIV-1 Rev protein. Proc Natl Acad Sci U S A 1999;96:13404-13408.
18. Perng GC, Jones C, Ciacci-Zanella J, Stone M, Henderson G, Yukht A, Slanina SM, Hofman FM, Ghiasi H, Nesburn AB, Wechsler SL. Virus-induced neuronal apoptosis blocked by the herpes simplex virus latency-associated transcript. Science 2000;287:1500-1503.
19. Straus SE. Human herpesvirus types 6 and 7. In Mandell GL, Bennett JE, Dolin R. Mandell, Douglas, and Bennett's Principles and Practice of Infectious Diseases, 5th ed. Philadelphia: Churchill Livingstone, 1999:1613-1617.
20. Feavers IM. Bacterial genomics: ABC of meningococcal diversity. Nature 2000;404:451-452.
21. Beard CB, Durvasula RV, Richards FF. Bacterial symbiosis in arthropods and the control of disease transmission. Emerg Infect Dis 1998;4:581-591.
22. Belt T. The Naturalist in Nicaragua. Chicago: University of Chicago Press, 1985:219-220. (Originally published in 1874.)
23. Bronstein JL. Mutualisms. In Fox C, Fairbairn D, Roff D, eds. Evolutionary Ecology: Perspectives and Synthesis. New York: Oxford University Press. In press.
24. Janzen D. Pseudomyrmex nigropilosa: a parasite of a mutualism. Science 1975;188:936-937.
25. Stanton ML, Palmer TM, Young TP, Evans A, Turner ML. Sterilization and canopy modification of a swollen thorn acacia tree by a plant-ant. Nature 1999;401:578-581.
26. Letourneau DK. Code of ant-plant mutualism broken by parasite. Science 1990;248:215-217.
27. Currie CR, Scott JA, Summerbell RC, Malloch D. Fungus-growing ants use antibiotic-producing bacteria to control garden parasites. Nature 1999;398:701-704.
28. Topoff H. Slave-making queens. Sci Am 1999;281:84-90.
29. Wilson EO, Holldobler B. The Ants. Cambridge, Mass: Belknap Press of Harvard University Press, 1990:436-448.
30. Wilson EO, Holldobler B: 449.
31. Labandeira CC. How old is the flower and the fly? Science 1998;280:85-88.
32. Pellmyr O, Huth CJ. Evolutionary stability of mutualism between yuccas and yucca moths. Nature 1994;372:257-260.
33. Pellmyr O, Thompson JN, Brown JM, Harrison RG. Evolution of pollination and mutualism in the yucca moth lineage. American Naturalist 1996;148:827-847.
34. Schiestl FP, Ayasse M, Baulus HF, Lofstedt C, Hansson BS, Ibarra F, Francke W. Orchid pollination by sexual swindle. Nature 1999;399:421-422.
35. Freeman S, Rodriguez RJ. Genetic conversion of a fungal plant pathogen to a nonpathogenic, endophytic mutualist. Science 1993;260:75-78.
36. Hoerauf A, Volkmann L, Hamelmann C, Adjei O, Autenrieth IB, Fleischer B, Buttner DW. Endosymbiotic bacteria in worms as targets for a novel chemotherapy in filariasis. Lancet 2000;355:1242-1243.
37. LeVier K, Phillips RW, Grippe VK, Roop RM II, Walker GC. Similar requirements of a plant symbiont and a mammalian pathogen for prolonged intracellular survival. Science 2000;287:2492-2493.