20 years ago, scientists got hooked on a single transcription factor that responds to a number of drugs of abuse. Will their work lead to treatments?
By Kerry Grens
1 When the drug was removed, the enzyme's activity spiked, which the authors interpret as a cellular withdrawal from dependency: "This phenomenon can be likened to the abstinence syndrome in animals."
|"You're profoundly altering the nature of nerve cells." -Eric Nestler|
It wasn't until more than a decade later, in the early 1990s, when Nestler, then at Yale University, and his group replicated the results in vivo and moved two steps downstream from adenylyl cyclase to the activation of CREB. They showed that a dose of morphine impairs the phosphorylation of CREB (a marker of CREB activation), but that activity returns to normal after a longer exposure to the drug.2 "Around the same time," Nestler recalls, "we were asking: The locus ceruleus is just a model system for the opiate system, but do other neurons respond?" He turned to the nucleus accumbens, a group of neurons that receive dopaminergic inputs from the ventral tegmental area, and which are involved in the brain's reward system. There, Nestler found similar results: Chronic use of morphine increases the activity of CREB.
The CREB-deficient mice in Schütz's lab presented an opportunity to measure whether CREB was necessary in the addiction process. With Rafael Maldonado, who was then at the University of Paris, Blendy showed in 1996 that her mutant mice lacked the symptoms of morphine withdrawal that normal animals exhibit.3 "By definition, dependence means the presence of a withdrawal syndrome when the drug is removed," Blendy says. "The question is, were the animals never dependent on the drug in the first place?" Blendy concluded that CREB was important in initiating addiction. But such a simple explanation was too good to be true.
Bill Carlezon, now an associate professor of psychiatry at Harvard's McLean Hospital in Belmont, Mass., was a postdoc in Nestler's lab in the mid-1990s, studying cocaine in the nucleus accumbens. At the time, there was no good way to target CREB directly, so Nestler's group developed a virus vector with a mutant form of CREB that competes with endogenous CREB and blocks its activity. When mutant animals were given cocaine they showed a heightened preference for the drug, whereas when CREB was overexpressed in animals they showed an aversion to it.4
Blendy found similar results after she moved to the University of Pennsylvania in 1997 with the CREB-deficient mouse line. When these animals were given low doses of cocaine (doses small enough to make them indistinguishable from saline to wild-type animals) the animals showed a strong place-preference for the side of the box where they received cocaine.5 "CREB-deficient animals show an enhancement in cocaine reward," Blendy says.
Though Blendy's data agreed with Nestler and Carlezon's results, they were findings that seemed out of line with her results for morphine. While CREB deficiency seemed to get animals more interested in cocaine, it acted oppositely with morphine. Blendy suspects the discrepancy is related to the different brain regions each drug works on. Though all drugs of abuse end up increasing dopamine in the nucleus accumbens, they act through different mechanisms: Cocaine blocks dopamine transporters in the nucleus accumbens, whereas morphine disinhibits dopamine cells in the ventral tegmental area.
The complexity of molecular changes that are involved in drug addiction doesn't stop there. Alcohol and nicotine act differently on CREB than cocaine and morphine do. Moreover, another transcription factor, DFosB, is upregulated in similar fashion to CREB, though with opposite effects on behavior.6 The DFosB response to drugs may be just as important to addiction as CREB, particularly regarding long-term changes. According to Nestler, CREB essentially delivers negative feedback from drugs, andDFosB promotes the rewarding effects of drugs. "DFosB can be seen in many ways as the converse of CREB," Nestler says. Yet with bewildering complexity, both transcription factors can be upregulated in the same cell. "Some cells show CREB activation, some induce DFosB, and some overlap," Nestler explains. "It's a very complex process that needs to be worked out."
Despite these gaps in knowledge, by the 1990s, scientists felt sure that CREB was important in regulating the effects of drugs of abuse. They had also realized, however, that it was apparently not a one-size-fits-all marker for all drugs in all regions of the brain, let alone a treatment. "CREB will never be a therapeutic target. It's too important, it's too ubiquitous," Blendy says.
She decided to turn downstream. Her group is tracking the expression of a number of CREB target genes during the process of drug addiction, from initial exposure to dependency to withdrawal. "The hope is that some of the target [genes] that it's responsible for will be ideal."
One downstream target gene that has shown the most dramatic results in differential expression through these phases is corticotropin-releasing factor (CRF, also referred to as corticotrophin-releasing hormone, CRH). CRF is important in mediating stress responses, but it is also involved in Koob's dark side of addiction. "What we're finding," says Koob, "and this is particularly true with alcohol and opiates and nicotine and maybe a little less so for cocaine - when animals take a lot of drug, the CRF system becomes engaged and contributes to excessive drug-taking."
Markus Heilig, the clinical director at the National Institute on Alcohol Abuse and Alcoholism, says the upregulation of the CRF system appears to be permanent. "In the last year a series of papers nailed down in rodents that if you have a long history of dependence on alcohol, and cycles of intoxication and withdrawal, it will induce remarkably long-lasting changes in the endogenous [CRF] system," Heilig says.
Koob has shown most recently that animals dependent on alcohol drink less during withdrawal when they've been given an antagonist to the CRF1 receptor.7 "One would predict [a CRF1 receptor antagonist] could be a great drug for withdrawal," Nestler says. Finding an antagonist safe for use in humans, however, has been difficult. Clinical trials on one drug, NBI 30775, were discontinued several years ago after two patients developed elevated liver enzyme levels.8 Koob and Heilig, among others, are working to find a CRF1 receptor antagonist that can be used safely to treat addiction.
In March of this year, Heilig published promising results in alcohol-dependent rats, showing that the drug stopped dependent animals from seeking alcohol when they were exposed to a stressor.9 If a CRF1 receptor antagonist could be developed into a drug, says Koob, the most likely application would be during acute withdrawal.
|"CREB will never be a therapeutic target. It's too important, it's too ubiquitous."
About 300 miles north of Blendy's Penn lab, a similarly gowned, gloved, and bootied scientist at Harvard's McLean Hospital wheels a cart of animals down a hallway. These are Carlezon's experimental rats, each outfitted with an antenna-like electrode surgically mounted to the top of its head. The electrode feeds pleasurable stimulation, in the form of square-wave pulses of electrical current, into the brain's reward center. In Carlezon's office he plays a video of a mouse incessantly spinning a wheel to receive the pleasure-giving current. "It probably feels like the best thing ever. Animals will choose it over drugs, food, sex - anything," he says.
At a certain point, the mouse stops spinning the wheel when the current dips below a frequency where the reward is diminished, Carlezon says. When animals go through withdrawal from cocaine, the minimum amount of current they are willing to work for - their "reward threshold" - goes up.10 "Because the stimulation isn't as rewarding during withdrawal, the mice require higher frequencies to get them to work [for the stimulation]." Carlezon's conclusion: "We think they're making more dynorphin."
Dynorphin is an endogenous opioid that acts at k-opioid receptors and is a downstream target gene of CREB. Carlezon showed that when CREB or dynorphin function is elevated in the nucleus accumbens, cocaine is less rewarding, and sometimes even aversive. But when k-receptors are blocked, the effect goes away.4 Carlezon is experimenting now with giving k-opioid receptor antagonists to rats experiencing cocaine withdrawal and measuring whether these drugs can hold reward thresholds steady. If the antagonists work, Carlezon says they could be candidates for treating the symptoms of drug withdrawal.
Koob and Brendan Walker, also at Scripps, have used a k-opioid receptor antagonist, nor-binaltorphimine, in ethanol dependent rats and found they drank less. In animals who were not dependent, they're drinking behaviors didn't change. Walker says it supports the idea that the dynorphin-k-opioid system is involved in the "dark side" of addiction11. "In general," Walker says, "that's the hypothesis-when animals are dependent and you remove alcohol, the heightened negative affect makes them want to get more alcohol. It looks like if we can block that system then we can, in a sense, block the motivation of the aninal to excessively consume alcohol."
Charles O'Brien, the vice chair of psychiatry at the University of Pennsylvania and director of the Center for Studies in Addiction, says research on dynorphin and CRF1 receptor antagonists appears promising, but to truly treat addiction the target has to be the brain's memory center, which may or may not involve CREB.
"Addiction is compulsive drug-seeking behavior," O'Brien says. "It's not the changes produced by alcohol, cocaine, or heroin itself. It's the fact that after the drug is gone, there's a learning, a memory trace, that produces craving, that produces drug-seeking and relapse." Nestler says those memories are life long. "As a kid we touched a hot stove and remembered it's never worth experimenting with a hot stove.... Drug memories may be as powerful as those memories."
Whether the learning of addiction involves CREB is uncertain. Nevertheless, Nestler says that manipulating CREB's target genes to ease withdrawal symptoms can aid other addiction treatments, such as cognitive-behavioral therapy, which might influence memories. "In my opinion, the sooner that you get these [potential drugs] into human subjects, the better," says O'Brien. "The animal models can point the way, but sooner or later you have to get them into human beings."