Previously, I've described endogenous retroviruses (ERVs) as sleeping dragons, even if they sometimes exhibit surprising activity. These genetic sequences, and their poor relations, retrotransposons, are found very widely, not just in humans, or mammals, or even animals. They exhibit similar characteristics in very different genomes. This prompts questions about why these pathogenic insertions are tolerated in the genomes of living things. They must not be entirely deleterious, or they would be quickly eliminated.
I'm going to propose that they serve a function comparable to guard dogs (canis lupus familiaris), which started out as wolves. I won't reference a list of papers because I don't want to give the impression that others are responsible for my own speculations. If I inadvertently raise issues that others have proposed I will be happy to relinquish any claim to priority, if people will only tell me.
In the idealized replication cycle of a retrovirus the virion is unpacked to expose the RNA strands it carries, and the reverse transcription enzymes it needs to insert DNA, in the vicinity of host-cell chromosomes. Reverse transcription then converts one RNA strand into DNA, which is actually inserted into a chromosome.
At some later time this inserted provirus is transcribed from DNA to messenger RNA by host-cell machinery. Some of this RNA is further transcribed into proteins which form various parts of the virion, and package newly-transcribed RNA inside it. This can then go on to leave the cell where it was formed and infect other cells or hosts.
Two features of this process turn out to have major implications. First, a major determinant of when the provirus is transcribed is the long terminal repeat (LTR) found at either end of a sequence. These typically include receptor elements which interact with host hormones or short nucleic acid fragments. Increasing the number of repetitions can increase the probability a proviral sequence will be transcribed in appropriate circumstances for replication. Second, retroviruses typically carry two RNA strands even though only one strand is reverse transcribed at a time. (This can be called diploidy, or pseudodiploidy, depending on how much of a purist you are.) In some respects this mirrors the behavior of DNA in sexual reproduction.
What determines which RNA strands are packaged together in a single virion? One thing is that they must be transcribed from host chromosomes at the same time. A second characteristic is that homologous strands tend to pair up. There are even subsequences which mark the appropriate strands for enzymes which carry out the packaging. The result is that RNA strands from the retrovirus are much more likely to be packaged together in a virion than most of the RNA strands found in the soup of biochemicals inside the host cell. In the case of a retrovirus invading a cell without related endogenous sequences this is a clear advantage for the invader.
If closely-related retroviruses have already infected the cell, and inserted provirus, things are not so simple. Now, there may be more than one kind of RNA transcribed under the same conditions and packaged in a single virion. This leads us to a further twist in the story. The enzyme which directs reverse transcription only reverse transcribes one strand of RNA at a time, but it can switch strands in the middle of the process. If the two strands are identical this will not be noticed. Should they be distinct the switch will result in a recombination event. For some known examples the probability of a switch is about 2% per kilobase. For an 8-kilobase genome this would mean there is about one chance in six of a recombination, assuming the two strands are not identical. Particular characteristic RNA structures can enhance or reduce this probability. These are not long odds, assuming you start with a replication-competent retrovirus.
The result is that recombination favors the most numerous sequences transcribed from DNA at the same time, and sufficiently similar to pair up with strands that will be packaged in virions. This means that a recently infected cell will be more likely to produce recombinations with the strands that have already been successful at infecting that cell. The infection evolves within a single individual host.
Once a retrovirus becomes fully endogenized its survival is tied to the survival of the host and/or offspring of that host. Killing the host would now be a serious strategic mistake.
Now consider what happens if the host cell genome is not pristine, but has numerous ERV sequences. Those sequences with the same LTR will be actively transcribed under the same conditions as a newly-inserted provirus of similar type. The transcribed RNA strands will pair with similar strands, and be packaged together.
If the ERV sequences are defective a high percentage of resulting virions will be defective. Even if one strand is replication-competent there is only a 50-50 chance it will be transcribed. Even then, there is a significant probability of a switch during transcription which will result in a recombination with a defective provirus. All these things reduce the infectivity of the newly-arrived retrovirus.
The defective sequences compete for biochemical resources when they are produced and packaged. The resulting virions, even if defective, can compete for host-cell receptors. A virion which blocks a receptor can protect that cell from infection by another virion. This also makes the retroviral infection less dangerous to the host.
For those sequences which arrived via inheritance from germ-line cells we have an additional assurance: the inherited sequence would not be there if every individual in the lineage back to the original infected host had not been able to survive infection and reproduce. In this context recombination takes on a new significance, host cells are converting a problem they don't know how to solve into one they have already solved -- just like any experienced mathematician.
Defense in depth against retroviruses involves many things: complement reactions, antibodies, apoptosis, hypermutation. There is also the effect of activating many defective sequences at the same time as a new retrovirus, so they become incorporated into its virions, thus reducing its infectivity. The important point to remember is that HERVs are well adapted to the host, and outnumber the new retrovirus. The following is about the role of HERVs in the last ditch defense against a retrovirus which succeeds in bypassing all those defenses.
From the standpoint of the invading retrovirus things must seem downright spooky. This has prompted me to write a horror story which might be considered a parable for young retroviruses about the dangers of infecting a host with many similar ERVs.
---
Imagine a burglar who succeeds in evading police, disabling alarms, and breaking into an empty house. He checks around and finds no evidence the house has been recently occupied, though there is no question it was occupied long ago. When he sees this he decides to settle in himself -- in effect stealing the entire house. This is much more convenient than carrying all the contents away. (He wouldn't be a burglar if he liked hard work.)
After a time there he notices one day that his left hand now has a ring he never put on. Checking further, he discovers a tattoo on his left arm is missing. Fingerprints on his left hand no longer match police records, (a real advantage in his line of business.) He decides this miraculous change is all for the best, he has gained a ring and lost incriminating evidence.
Later, his smoker's cough goes away. His lungs have been replaced. Still, this seems to be an improvement, so he stays. Finally, his brain is replaced. At this point he has become the missing occupant. As such he locks the doors, checks around for prowlers, and goes back to sleep to await the next intruder.
I'm going to propose that they serve a function comparable to guard dogs (canis lupus familiaris), which started out as wolves. I won't reference a list of papers because I don't want to give the impression that others are responsible for my own speculations. If I inadvertently raise issues that others have proposed I will be happy to relinquish any claim to priority, if people will only tell me.
In the idealized replication cycle of a retrovirus the virion is unpacked to expose the RNA strands it carries, and the reverse transcription enzymes it needs to insert DNA, in the vicinity of host-cell chromosomes. Reverse transcription then converts one RNA strand into DNA, which is actually inserted into a chromosome.
At some later time this inserted provirus is transcribed from DNA to messenger RNA by host-cell machinery. Some of this RNA is further transcribed into proteins which form various parts of the virion, and package newly-transcribed RNA inside it. This can then go on to leave the cell where it was formed and infect other cells or hosts.
Two features of this process turn out to have major implications. First, a major determinant of when the provirus is transcribed is the long terminal repeat (LTR) found at either end of a sequence. These typically include receptor elements which interact with host hormones or short nucleic acid fragments. Increasing the number of repetitions can increase the probability a proviral sequence will be transcribed in appropriate circumstances for replication. Second, retroviruses typically carry two RNA strands even though only one strand is reverse transcribed at a time. (This can be called diploidy, or pseudodiploidy, depending on how much of a purist you are.) In some respects this mirrors the behavior of DNA in sexual reproduction.
What determines which RNA strands are packaged together in a single virion? One thing is that they must be transcribed from host chromosomes at the same time. A second characteristic is that homologous strands tend to pair up. There are even subsequences which mark the appropriate strands for enzymes which carry out the packaging. The result is that RNA strands from the retrovirus are much more likely to be packaged together in a virion than most of the RNA strands found in the soup of biochemicals inside the host cell. In the case of a retrovirus invading a cell without related endogenous sequences this is a clear advantage for the invader.
If closely-related retroviruses have already infected the cell, and inserted provirus, things are not so simple. Now, there may be more than one kind of RNA transcribed under the same conditions and packaged in a single virion. This leads us to a further twist in the story. The enzyme which directs reverse transcription only reverse transcribes one strand of RNA at a time, but it can switch strands in the middle of the process. If the two strands are identical this will not be noticed. Should they be distinct the switch will result in a recombination event. For some known examples the probability of a switch is about 2% per kilobase. For an 8-kilobase genome this would mean there is about one chance in six of a recombination, assuming the two strands are not identical. Particular characteristic RNA structures can enhance or reduce this probability. These are not long odds, assuming you start with a replication-competent retrovirus.
The result is that recombination favors the most numerous sequences transcribed from DNA at the same time, and sufficiently similar to pair up with strands that will be packaged in virions. This means that a recently infected cell will be more likely to produce recombinations with the strands that have already been successful at infecting that cell. The infection evolves within a single individual host.
Once a retrovirus becomes fully endogenized its survival is tied to the survival of the host and/or offspring of that host. Killing the host would now be a serious strategic mistake.
Now consider what happens if the host cell genome is not pristine, but has numerous ERV sequences. Those sequences with the same LTR will be actively transcribed under the same conditions as a newly-inserted provirus of similar type. The transcribed RNA strands will pair with similar strands, and be packaged together.
If the ERV sequences are defective a high percentage of resulting virions will be defective. Even if one strand is replication-competent there is only a 50-50 chance it will be transcribed. Even then, there is a significant probability of a switch during transcription which will result in a recombination with a defective provirus. All these things reduce the infectivity of the newly-arrived retrovirus.
The defective sequences compete for biochemical resources when they are produced and packaged. The resulting virions, even if defective, can compete for host-cell receptors. A virion which blocks a receptor can protect that cell from infection by another virion. This also makes the retroviral infection less dangerous to the host.
For those sequences which arrived via inheritance from germ-line cells we have an additional assurance: the inherited sequence would not be there if every individual in the lineage back to the original infected host had not been able to survive infection and reproduce. In this context recombination takes on a new significance, host cells are converting a problem they don't know how to solve into one they have already solved -- just like any experienced mathematician.
Defense in depth against retroviruses involves many things: complement reactions, antibodies, apoptosis, hypermutation. There is also the effect of activating many defective sequences at the same time as a new retrovirus, so they become incorporated into its virions, thus reducing its infectivity. The important point to remember is that HERVs are well adapted to the host, and outnumber the new retrovirus. The following is about the role of HERVs in the last ditch defense against a retrovirus which succeeds in bypassing all those defenses.
From the standpoint of the invading retrovirus things must seem downright spooky. This has prompted me to write a horror story which might be considered a parable for young retroviruses about the dangers of infecting a host with many similar ERVs.
---
Imagine a burglar who succeeds in evading police, disabling alarms, and breaking into an empty house. He checks around and finds no evidence the house has been recently occupied, though there is no question it was occupied long ago. When he sees this he decides to settle in himself -- in effect stealing the entire house. This is much more convenient than carrying all the contents away. (He wouldn't be a burglar if he liked hard work.)
After a time there he notices one day that his left hand now has a ring he never put on. Checking further, he discovers a tattoo on his left arm is missing. Fingerprints on his left hand no longer match police records, (a real advantage in his line of business.) He decides this miraculous change is all for the best, he has gained a ring and lost incriminating evidence.
Later, his smoker's cough goes away. His lungs have been replaced. Still, this seems to be an improvement, so he stays. Finally, his brain is replaced. At this point he has become the missing occupant. As such he locks the doors, checks around for prowlers, and goes back to sleep to await the next intruder.
