Equations that Spell Disaster
Researchers are pinpointing the factors that combine to produce complex diseases.
When Hurricane Katrina made landfall on the Gulf Coast in August 2005, the Category 5 storm caused such severe and lingering damage because it encountered the perfect combination of vulnerabilities: weak infrastructure, poor lines of communication, and a dysfunctional emergency rescue system. These conditions coalesced to produce one of the worst human catastrophes in recent US history. In a similar way, complex diseases result from a series of events that may not amount to much when considered one by one, but together, coalesce into a perfect storm that spells disaster for a particular organ or system.
1 He and his colleagues had managed to express a variant of a Crohn’s susceptibility gene, known as ATG16L1, in mice. This gene had initially been identified in a subset of human patients. Like human carriers, mice bearing this variant display abnormalities in a group of cells that line the small intestine and are responsible for creating a healthy mucosal layer. Known as Paneth cells, they normally carry out the crucial function of releasing antimicrobial proteins into the layer of mucus that covers the epithelium of the small intestine. These proteins discourage all but a few species of bacteria from living in the mucus and are thought to play a role in shaping the composition of the surrounding microbial community. But, for reasons yet unknown, in the presence of the Atg16L1 variant, the Paneth cells cannot package and secrete these antimicrobial products—flagging the sugar-rich mucosal layer as prime real estate for commensal bacteria to colonize, despite its being a privileged location that is, for the most part, microbe-free. It’s “potentially a setup for inflammation,” Stappenbeck says. However, the story didn’t end there. They made a serendipitous observation when they began raising the mice in a new supersterile facility: even with the Atg16L1 allele, the mice showed no abnormalities in their Paneth cells. Something in the less sterile environment had to have been working with the allele against the Paneth cells, and that something turned out to be a mouse RNA virus belonging to an extremely common family known as noroviruses. In humans, noroviruses are responsible for nearly all nonbacterial gastrointestinal infections. In Stappenbeck’s mice, only one viral strain seemed to induce the abnormalities in Paneth cells. Dubbed CR6, the strain wasn’t immediately cleared by the immune system and caused a persistent infection. While viruses have been implicated in triggering Crohn’s for a long time, this was the first study to provide direct evidence for the link, Sartor says.
But even the susceptibility gene plus the virus still weren’t enough to induce widespread inflammation and damage to the gut. The researchers were able to recreate the condition only after feeding the mice the oral toxin dextran sodium sulfate (DSS). Once this toxin damages the lining of the gut, which is already aggravated by commensal overgrowth in the mucosal layer, the immune system is unleashed. Cytokines such as TNF-α and IFN-γ permeate the entire wall of the gut, causing massive numbers of immune cells to attack surrounding tissue. Ulcers and lymphoid aggregates appear, and the muscle walls thicken—all symptoms of Crohn’s disease in humans. “It’s the perfect storm,” Stappenbeck says.
Although this is an animal model of the disease, Sartor believes it’s possible that some norovirus infections in humans could jump-start a similar pathway in people carrying the ATG16L1 allele. But he cautions: “That could be one of literally thousands of triggers.” There may indeed be other genes besides ATG16L1 that affect Paneth cell function and lead to the same outcome, some of which may not require a viral trigger. And there may be cases with a completely different etiology altogether. After all, ATG16L1 is only one of some 30 susceptibility genes that have already been associated with the disease to date.
These two autoimmune diseases are, in some ways, two versions of the same story. One’s immune system is coaxed into damaging a specific type of its own cells. In Type 1 diabetes, it’s a group of insulin-producing cells in the pancreas, while in multiple sclerosis it’s the myelin sheath surrounding the axons of certain neurons.
Both diseases also share susceptibility genes, of which the most important seem to be ones associated with the major histocompatibility complex (MHC), which presents both self and foreign antigens to immune cells.
“Presumably, the immune system is not grossly defective,” says Hartmut Wekerle, director of the Max Planck Institute of Neurobiology; it is just more vulnerable to turning against the host given the right triggers.
Myelin-specific T cells + unknown species of gut bacteria = multiple sclerosis
Researchers first suspected the role of infectious agents in multiple sclerosis (MS) when they observed that cases of the disease cluster geographically and can appear as outbreaks in populations. Attempts to conclusively pinpoint a single infectious agent responsible for MS, however, have been unsuccessful, even though a growing list of human pathogens, such as the human herpesvirus 6 (HHV-6) and Chlamydia pneumoniae, have been detected in the brains of MS patients.
Figuring out how infection can induce the host’s immune system to attack and destroy the myelin sheaths that envelop neuronal axons, resulting in MS, has been a major research focus and has led many researchers, such as Stephen Miller from Northwestern University, to recreate virus-triggered demyelinating diseases in mice. Miller studies the Theiler’s murine encephalomyelitis virus (TMEV), commonly used to induce a mouse model of MS.2 Theiler’s virus infects the neurons of mice and is not known to cross over to humans. Susceptible mouse strains, such as the SJL strain, fail at clearing the viral infection quickly, thereby enticing lymphocytes specific to the virus to enter and remain in the central nervous system and cause damage to cells in the area. In the process, a subset of the immune cells is primed against components of myelin that are present in the cellular debris—a process known as epitope spreading. Once the virus is cleared, the myelin sheath becomes the focus of the attack. Resistant strains of mice, however, clear the virus early and don’t allow the infection to become persistent.
Miller’s mouse model is one of the best in describing how epitope spreading can be triggered by a viral infection, says Kristen Drescher, an immunologist at Creighton University School of Medicine in Omaha, Nebraska. But while they provide powerful proof-of-concept scenarios, Hartmut Wekerle believes Theiler’s virus-induced MS models may not be “as close to human MS,” because the virus does not infect people.
Wekerle works with transgenic mice that develop demyelinating disease spontaneously. All the mice need is a population of T cells that can recognize a myelin antigen. Immune cells that can recognize components of the body’s own cells (referred to as self-reactive immune cells) are necessary components of the healthy immune system, Wekerle explains. “What happens in MS, we assume, is that these self-reactive immune cells are activated and turn from self-reactive to auto-aggressive.” He has recently discovered an unlikely ally in that process—gut microbiota. Working with these transgenic mice, he noticed that animals that normally develop autoimmunity did not do so when reared in a supersterile environment. Although the mice were engineered to express lots of T cells specific to myelin, the T cells were not attacking the central nervous system. The missing factor was their gut microbiota. “Now the big question is,” Wekerle says, “which bacteria in the gut are responsible for activation and triggering of the disease? Could it also be that in people with MS, the initial step has been in the gut?”
LCMV protein + LCMV Armstrong strain + Pichinde virus = rapid progression of diabetes
In Type 1 diabetes, immune cells attack the insulin-producing β cells of the pancreas, inducing widespread damage to the organ. “No one really knows what the specific antigen that the immune system reacts against to destroy the β cells is,” says Michael Oldstone of the Scripps Research Institute in La Jolla, California. But once again, viruses are prime suspects as triggers of the autoimmune process, and they have been shown to induce diabetes in mice through the epitope-spreading mechanism described above. Oldstone was the first to demonstrate that a virus can induce diabetes in mice through another mechanism, also implicated in MS, called molecular mimicry. This mechanism requires the virus to express a peptide that closely resembles a molecule made by the host itself. When the host mounts an immune response against the virus, some of the immune cells specifically directed against the viral peptide can also recognize the self-molecule, thus mounting an attack against both the virus and the host cells bearing the similar antigen. In Oldstone’s mice, which had been engineered to express a viral protein from a strain of lymphocytic choriomeningitis virus (LCMV) in their β-cells, the disease developed only after the mice were exposed to the virus.
The MHC background of susceptible people may be such that antigen-presenting cells, when priming T cells, display parts of pathogenic antigens that are very similar to self-antigens, Stephen Miller explains. “I might get infected with the same virus as you, and I may not have any consequences because my genetics won’t present the same portion of the virus to my immune system.” More than fifteen pathogens have been shown to have molecular sequences similar enough to human sequences that molecular mimicry may be induced, and at least one known instance of its occurring naturally has been observed. Following infection with C. pneumoniae, some people develop antibodies that recognize elements in cardiac tissue and induce widespread inflammation that results in atherosclerosis.
Oldstone later used the same mouse model to demonstrate that subsequent infection with another member of the arenavirus family, one not closely related to LCMV, can alter the course of the disease.3 Mice infected with a strain of Pichinde virus after infection with LCMV developed Type 1 diabetes twice as fast as mice that had only been exposed to LCMV. This occurs, presumably, because newly-primed T cells specific to the second virus are able to cross-react with an antigen from the first virus—essentially expanding the army of attacking T cells. It’s a two-hit phenomenon that requires a virus that can cause disease and one that, although it cannot cause the disease by itself, is capable of accelerating the process once it has been set in motion. “The implication is that when a person gets diabetes and you look for what the cause is, the first virus, the one that started it, has already been cleared and you would never find it,” Oldstone says.
Steven Tracy, a virologist at the University of Nebraska Medical Center in Omaha, has implicated yet another family of RNA viruses in the etiology of Type 1 diabetes: enteroviruses. These viruses—distant cousins of the Theiler’s virus used in MS mouse models—infect millions of people worldwide and are responsible for diseases such as polio. Working with mice that develop diabetes spontaneously, Tracy found that when he inoculated young pups with a strain of enterovirus, most were guarded against developing the disease later in life.4 Since then, another research group has shown that in these mice, enterovirus infection induces a rise in protective regulatory T cells, thus keeping many mice from developing diabetes despite the strong propensity to destroy their own islets. But older mice displaying early-stage inflammation of the pancreas worsened when infected with the enterovirus.
“These two observations occurred with a typical, common enterovirus (the coxsackie B virus) humans see a lot of in life around the world,” Tracy says, “one that is spread by a fecal-oral route.” He adds that before the 20th century, Type 1 diabetes was rare, as was polio. But as knowledge about the importance of hygiene led to a rise in the availability of sewers and indoor plumbing, polio outbreaks began to occur annually. Then, after the 1950s, the incidence of diabetes began to rise. Considering the evidence from mouse models, Tracy and other proponents of what is known as “the hygiene theory” suggest that a cleaner lifestyle, particularly during childhood, has deprived a subset of the population at risk for developing autoimmunity of its exposure to pathogens that act as natural vaccines.
Short allele of serotonin transporter gene + stressful event in early life = depression Long allele of serotonin transporter gene + poor socioeconomic conditions = psychopathic traits
Dissecting the etiology of complex diseases becomes exceedingly difficult when entering the realm of behavioral biology. Though it may seem obvious that different people react differently to the same events, or that people with similar backgrounds show varying predispositions to certain risky behaviors, unraveling the gene-environment interactions that modulate behavior has been a rocky endeavor. To date, the identification of susceptibility genes that confer risk for specific psychiatric disorders is still speculative, and studies linking genes to particular behavioral phenotypes have been plagued by a lack of independent validation.
Of these genes, the one that codes for the serotonin transporter, 5-HTT (also known as SLC6A4), has been hogging the spotlight. This gene is responsible for fine-tuning the transmission of serotonin, a neurotransmitter known to modulate mood, emotion, sleep, and appetite.
“It is probably the most studied gene in psychiatry,” says Robert Philibert, a professor of psychiatry at the University of Iowa. The serotonin transporter protein is the target for the widely used class of antidepressants known as selective serotonin-reuptake inhibitors (SSRIs), which includes Prozac, Paxil, and Zoloft. “Alterations in serotonergic function are very important in how we form relationships,” Philibert adds.
In 2003, Science published a hotly debated paper that tied a particular polymorphism in the promoter region of 5-HTT, known as the short allele, to an increased risk of depression.5 This polymorphism results in a diminished transcription of the serotonin transporter gene. People with one or two copies of the short allele show increased activity in the amygdala (a component of the brain’s limbic system known to be involved in processing emotional responses) in response to stressful stimuli and are more susceptible to startling, more fearful, and more averse to risk taking. In this study, led by Avshalom Caspi, now a professor of psychology and neuroscience at Duke University, carriers of the short allele were found to be much more likely to become depressed if exposed to a stressful life event early in life. It seems these people “are not able to pick themselves up by the bootstraps” after this initial stress, Philibert says.
Carriers of a different 5-HTT polymorphism, known as the long allele, on the other hand, seem to be buffered against depression, despite exposure to similar stressful events. This has led to the hypothesis that 5-HTT variation modulates the capacity to cope with stress.
But while the long allele seems to offer a better set of cards to its carriers, subsequent research has suggested that, given the right variables, having this allele may predispose people to a disorder at the opposite end of the spectrum from depression: psychopathy. A preliminary small-scale study carried out by Edelyn Verona and colleagues at the University of Illinois at Urbana-Champaign examined whether carriers of the long allele who grew up in poor socioeconomic conditions exhibited enhanced expression of psychopathic traits such as detachment, callousness, and narcissism, when compared to those who grew up in better socioeconomic environments.6 “When you’re less susceptible to feeling rejected or distressed,” she explains, “in a good environment that may breed bravery and courageousness, but in a bad environment may make you more callous toward others.” She found youths homozygous for the long allele who grew up in a poor socioeconomic environment were more likely to display certain traits associated with psychopathy, such as unemotionality. “It is a very interesting, but small, study whose results need to be validated in larger populations,” Philibert says. Nonetheless, Verona’s study adds a layer of complexity to the interplay between genes and the environment in manifesting psychiatric disorders, as alleles can have both positive and detrimental effects depending on the environments to which the host is exposed.
Philibert is currently trying to figure out how exactly stressful events combine with genetic susceptibility at a molecular level to result in these psychiatric disorders. In looking for epigenetic changes, such as methylation of DNA and histone modification, he has obtained some early evidence suggesting that DNA methylation in the promoter region of the serotonin transporter gene is greater in people with a history of child abuse.7 As seen in carriers of the short allele, increased DNA methylation also leads to diminished transcription of 5-HTT. “The amount of methylation is a function of exposure to abusive environments,” he adds. But Philibert warns not to lose sight of the overwhelming genetic complexity of these diseases. Ultimately, very significant genes such as 5-HTT may only account for 0.1 percent of the disease’s genetic variance.
Of course, adding X, mutating Y, and knocking down Z in mice do not a human disease make. Most of these animal models of complex diseases are still just proof-of-concept studies, meant only to guide researchers in their search for similar mechanisms in the human afflictions. For Hartmut Wekerle, studying his multiple sclerosis mouse models is just the very first step. “The question of course is: can we translate what we have found in these mice to humans?”