A group of researchers operating out of Emory University, in Atlanta, and the Scripps Research Institute and University of Miami, in Florida, recently reported on a novel gene that regulates fear both in mice and humans with Post-traumatic Stress Disorder (PTSD). Their work was published in the upstart, cutting-edge journal Science Translational Medicine last month.
Following now-popular trend in the sciences, the researchers conducted a full-scale, animal-to-human translational research program to demonstrate the potential importance of this new gene to human “fear-related” disorders like PTSD. Given the recent push in the sciences to make basic research results “translate” to the clinic, this trend is at once both exciting and a little disconcerting. For example, while the breadth of this report is admirable, it does give me the sense that the researchers stitched together two distinct research papers into a single, over-extended compilation.
Nonetheless, this paper presents some potentially interesting work and promises to open up some avenues for treating PTSD. This is a pretty major area of interest for me, so I think it’s certainly worth a look.
For the sake of clarity, I will attack each of the “two” papers separately and discuss the results as I go along. I’ve marked out “overview” sections if you want to jump right to my summaries of some of the key data as well as principle criticisms.
(To save space in this post, I briefly discuss Post-traumatic stress disorder and its potential causes, elsewhere.)
An animal model of “trauma”
It should come as no surprise that animal models of PTSD are pretty controversial. While influential “biological” models of PTSD symptoms assume that common factors underlie both the general mammalian “fear” or “threat” response and PTSD, it is widely accepted that most animal research paradigms cannot match the severity of traumatic experiences that precede the development of PTSD.
The present paper attempts to mimic some aspects of human traumatic experience by exposing rodents to a single traumatic event using the “immobilization to wooden board” (IMO) paradigm. This protocol works exactly as it sounds:
“[A mouse was] immobilized by gently restraining its four limbs in a prone position to metal arms attached to a wooden board for 2 hours.”
A control group underwent “compensatory handling” for the same duration of time. In order to test that this protocol produced a long-term effect on animal behaviors, the researchers waited 6 days, before having all animals perform a number of common memory and “anxiety” tasks.
After the 6-day waiting period, all animals were trained on the ever-popular Morris Water Maze, a spatial memory task where animals must locate a hidden platform within a large “water tank” by using landmarks placed on the wall of the room that are located outside of the tank (see left). Animals were trained on this task once a day over a period of 4 days. After each day of training, animals were tested on how much time they spent in the correct quadrant of the water tank (which holds the hidden platform).
Both groups performed similarly on all training and memory trials, except for on a final “delayed” memory test, which occurred 24 hours after the final training day, and 11 days after animals had experienced the original IMO “trauma”. On this final trial, animals which had been exposed to IMO spent moderately less time in the correct location (~35% relative to ~52% for the control group, on average). No follow-up trials were conducted to determine if this performance difference was maintained.
Animals also completed a “radial arm maze” (5 mins total) and an “open field maze” (30 mins total), on a single occasion, also 6 days post-IMO. Animals exposed to IMO traveled into the “unsafe” regions of each maze on average about 40% less than the control animals, suggesting a higher level of rodent “anxiety” as a result of the IMO exposure.
Together these results suggest that the IMO stressor moderately impaired aspects of rodent behavior, although the relationship of these changes to human PTSD symptoms is unknown.
Effects of “trauma” on subsequent “fear conditioning”
Given this evidence that the IMO stress produces measurable effects on rodent behavior, the researchers next sought to assess how prior IMO stress affects memory for subsequent “fear/threat conditioning”. A typical “fear” conditioning protocol (now, often referred to as “threat-conditioning”, as conditioned rodents may not actually experience fear) relies upon Pavlovian learning principles: animals are trained to associate a sound or “tone” with a light foot-shock, applied to the floor of a specialized cage. Over time, the tone presented in the absence of a shock will continue to initiate a stereotypical protective behavior, previously only present in response to the shock (in rodents, this behavior is usually a “freezing” reaction, where the animal stands very still to avoid detection by predators). This defensive freezing behavior is well-known to heavily rely upon functions of the brain’s amygdalae, along with the activities of other brain regions.
In the present study, half of the animals from each group (IMO and control) were trained on a single occasion (Day 1, Threat conditioning: 5 brief “tone”-shock pairs, each 5 minutes apart), whereas the remaining animals were presented with the tone absent any shock (No conditioning). There were no measurable differences between the groups of unconditioned animals (IMO vs control). However, among conditioned animals, those previously exposed to IMO were moderately more likely to freeze during the 5 minutes between shocks than were control animals. Nonetheless, throughout the entire training period, both groups spent the same amount of time freezing, so it’s unclear there was much of a difference here.
On Day 2, when conditioned animals were exposed to the “tone” without a shock present (15 trials of 30s, 1.5 minutes apart), the IMO animals now spent more time freezing overall across the 15 trials (on average, ~45-60% of time for IMO animals vs ~30-50% of the time for controls). This effect was significant and pretty clearly suggested that IMO exposure had an effect on the second phase of the threat conditioning procedure.
In summary, these results demonstrate that prior IMO exposure produces a modest effect on freezing during the training phase of fear conditioning (Day 1) and a moderate effect on freezing behaviors in a second phase, without any shock applied (tone only, Day 2).
Identification of a “trauma” gene
Finally, the researchers extracted amygdala tissue from all of the “conditioned” animals 2 hours after the final experiment and used this tissue to perform a microarray analysis of mRNA expression. They identified potential genes of interest as follows:
i) Include differentially expressed genes (IMO animals vs control): 4.5% of total data
ii) Refine selection: only include genes that demonstrate >1.3-fold increase (IMO tissue) or <0.7-fold decrease (control tissue) in expression, relative to “home cage” (HC) animals that had never been exposed to experimental stimuli
iii) Select genes highly expressed in amygdala relative to other brain regions
Apparently, this strategy only identified a single gene: opioid receptor-like 1 (OPRL1). In replication studies, the authors use an alternative method (qPCR) to confirm that OPRL1 mRNA expression was down-regulated in control animals (0.7-fold), following either phase of fear conditioning, but was unchanged in all IMO-exposed animals (1.0-fold).
The OPRL1 gene produces a product called the “nociceptin” receptor or NOP, which is named for its role in analgesia, or pain reduction. In order to test the importance of the NOP receptor in threat-conditioning, the authors applied a novel drug that specifically activates this receptor (but doesn’t active “opioid” receptors, which have similar structures), to the brains of animals during the first phase of threat conditioning (Day 1: tone-shock pairing). In these animals, applying the drug before or after threat conditioning reduces freezing to the tone during the subsequent test phase by up to 50%. The authors find this result both in control animals and animals which had previously been exposed to IMO, suggesting that NOP has a general purpose influence on the development of threat conditioning.
Animal results overview
Overall, these animal results are fairly interesting. The authors demonstrate that an animal “trauma” model behaves differently after fear conditioning than control animals and that this behavior change is associated with altered gene expression. They further confirm that selectively activating the product of this gene abolishes the behavioral effect (reduces freezing by quite a bit). I also like how they actually perform multiple replications (I know, a shock!), even if their comparisons are probably under-powered, as is typical for the animal research literature (also see my summary).
There are still some clear issues with this result, which question its relevance to PTSD.
One of the more technical points is that the authors don’t investigate why OPRL1 gene expression changes only in the control animals (but not IMO) during threat conditioning. One possible mechanism is that the NOP receptor was activated in controls and negative feedback mechanisms reduced expression of the gene, whereas in IMO-exposed animals the receptor was never activated (so, no compensation). This was an obvious question to ask, but the authors never do. Thus, the basis for this molecular change is a little patchy.
A second, more obvious point, is that we don’t know if this animal model is relevant to PTSD at all. For one, this animal model only moderately affects rodent behavior (in fact, it’s hard to imagine what a “large behavior change” would look like) and does so over a short time period (2 weeks), whereas PTSD is often very robust and long-lasting (starts 1 month post-trauma, lasts potentially for years).
Moreover, the “symptoms” presented by this animal model may or may not be at all relevant to human PTSD symptoms, and so the relative importance of this new mechanism is ambiguous. Threat conditioning protocols are widely used as correlates of human and animal “fear”, but does curing a “fear/threat learning deficit” affect the underlying PTSD symptoms? The answer to this isn’t clearly explained by the current literature.
A final point, less important point is that this study doesn’t actually identify a new mechanism: previous research already identified the NOP receptor as a potential target (here, here, and here). Nonetheless, this is a good replication effort, and I am glad to see it published in a quality journal.
In general, the broadest conclusions we can reach from these results are that “IMO exposure is potentially a model of human trauma, which it alters animal behavior and threat conditioning responses” and that “activating the NOP receptor also alters threat conditioning responses” (but, not behavior). But, is the NOP receptor a target that eliminates the underlying effects of IMO exposure? This isn’t tested. The authors don’t even tell us if it corrects the moderate differences in water maze performance or “anxious” behaviors I discussed at the outset of this article. So even if IMO were equivalent to a “PTSD-causing trauma” (which is fairly dubious), we don’t know for sure that the NOP receptor is a treatment target for either.
Let’s address this latter question in more detail below.
Association between Oprl1, childhood trauma and PTSD symptoms
With this in mind, I want to jump over to the human research results. This is the part of the study which is intended to provide a “bridge” between this animal research and human relevance and the inclusion of these data almost certainly explain the publication of this paper in this relatively high impact journal.
Keep in mind that if these additional results demonstrate an important role of the OPRL1 gene in human PTSD, then some of the criticisms I note above go right out the window.
The authors first checked the Allen Brain Atlas to confirm that OPRL1 is highly expressed in the human amygdala (follow the link and scroll over the left-side of the “heat-graph” data, which is hot red over top of the amygdala; you can also visualize the data using Brain Explorer).
In order to test this hypothesis further, the researchers used data from a large cohort of urban, heavily trauma-exposed African American subjects living in Atlanta, Georgia. Past studies by this same group have compared the presence of genetic markers called single-nucleotide polymorphisms (SNPs), which reside nearby individual genes, to the incidence of childhood and adulthood traumatic experiences and current PTSD symptoms.
For the present study, the researchers focus on only two factors (childhood abuse, SNP markers) and attempt to determine the combined effect of these factors on current PTSD symptoms. Their analysis controls for age, sex and substance abuse (previous studies have identified a relationship between OPRL1 and substance abuse, which is also very high among this cohort).
When statistically controlling for multiple selected markers near the OPRL1 gene, the authors do find a single region of DNA that demonstrates a significant interaction between both childhood trauma and the specific version of the gene that participants have (Note: there are two alleles for every gene, so there are three possible combinations of alleles, in this case: GG, GC and CC). A secondary statistical test (main effect of G allele) demonstrates that participants carrying a copy of the G allele (genotype: GG or GC) had a higher level of PTSD symptoms than non-carriers, if they had previously experienced childhood abuse (see Figure 2, left).
The researchers also selected a subset of these participants to undergo a human threat-processing task, where images were paired with an aversive blast of air to the eye (image-blast pairing). This “air blast” produces with a robust “fear-potentiated” startle (FPS) response, which can be measured using electrodes applied to the skin. The authors report that individuals carrying the a copy of the G allele produced large startle responses, both to conditioned images (which were paired with a “harmful” airblast) and unconditioned images (no airblast), whereas C allele carriers clearly produced a much larger startle to the conditioned images, suggesting that the former were unable to appropriately discriminate “dangerous” stimuli.
Finally, a second subset of the sample (n=29, women only) underwent functional imaging while viewing “fearful” and “neutral” faces. There was no difference in amygdala activation between G carriers and non-carriers while viewing fearful relative to neutral faces, although there was still a significant difference in connectivity between the amygdala and inferior insula activity for the same (fearful-neutral) contrast.
Human data overview
Consistent with the animal research results, this human data suggests that genetic variation at the OPRL1 gene may increase PTSD symptoms in individuals who have been exposed to high levels of childhood abuse. This is really a very big step because most traditional animal research results are limited by a lack of clarity about their relevance to human neurobiology. In this context, the combined animal and human data suggest that OPRL1 might be a novel drug target for reducing certain PTSD symptoms in humans.
A fairly technical question left unanswered by this study is how genetic variants at the OPRL1 gene might influence expression of the NOP receptor and how exactly childhood abuse alters the relationship between genetic variant and receptor expression or function. This will almost certainly be an area of follow-up research.
More broadly, the relative importance of this finding to PTSD suffers is still entirely unclear. While this study is the first to suggest that OPRL1 is associated with PTSD, we still don’t know if it has a very significant influence. The graph shown above shows a small differences in PTSD symptoms for people who have suffered high abuse, but all of these people have PTSD anyways, regardless of whether or not they carry the G allele. Among the participants with low levels of PTSD symptoms (either no diagnosis or not current PTSD), the G allele either has no effect or may even be protective.
The follow-up FPS and fMRI experiments don’t improve this issue. One concern I have about these data is that the authors select only a small subset of participants, but don’t report on any participant characteristics, including how they were selected or whether they had equivalent levels of current PTSD symptoms, substance abuse, etc.
Normally, we might assume that researchers would control for obvious factors like this, but a recent study by these same authors failed to do so this simple control, so I cannot be so sure. For the FPS data, the authors do statistically control for sex, age, degree of childhood trauma and PTSD symptoms, but, strangely don’t control for substance abuse. Differences in FPS responses between G carriers and non-carriers declined (but remained significant)–but might be absent if substance use was controlled for.
Similarly, the fMRI data doesn’t suggest very strongly that the G allele affects brain function, given that one of two hypothesized differences were not found (differences in amygdala activation) and the authors report only a modest effect for the other (amygdala-insula connectivity).
These technical considerations underscore one of the major challenges for assessing how genetic variants affect PTSD symptoms. Because the method for identifying allele carriers requires that they have higher PTSD symptoms, this means that symptom scores are confounded with the allele. If G allele carriers have higher levels of symptoms (including substance abuse and other negative outcomes), then wouldn’t we expect them to potentially have stronger FPS and fMRI responses? The alternative is no better: if G allele carriers have higher FPS and fMRI responses than non-carriers, but the same levels of symptoms, then it’s not clear that these small differences in physiological responsiveness among PTSD sufferers are relevant to the underlying condition.
Food for thought.
Overall, this study provides some evidence to suggest that the OPRL1 gene is affected by “traumatic” experience in mice and is associated with PTSD in individuals who have experienced significant amounts of childhood trauma. Previous studies had already linked OPRL1 to “fear” and “trauma” in mice, so this is not a new finding, but it nicely replicates past results and introduces a new, selective drug to target this site. In contrast, no previous studies have suggested the genetic variants for this gene might directly influence PTSD symptoms in humans. One prominent possibility is that if this gene is relevant to the development of PTSD symptoms, then it may serve as a useful target for post-trauma therapeutics.
I think the more important implication is that if this study is validated it may serve as a template for future studies to identify additional genetic variants that might influenced PTSD symptoms in concert.
However, as I note above, the relative importance of OPRL1 to human PTSD is unclear.
Thus, this study leaves us out on a (promising) limb: maybe OPRL1 is important in the development of PTSD, and maybe not. It might also be a potential drug target, but we don’t yet know if it will tackle the underlying symptoms of PTSD.
To sum up, here is the authors’ entirely valid conclusion, with my emphasis added:
“If this hypothesis is clinically validated and our studies are replicated, NOP-R agonists may be important candidates for the prevention of PTSD, particularly in an early intervention setting, shortly after exposure to traumatic experiences.”
Only time will tell.
Andero R, Brothers SP, Jovanovic T, Chen YT, Salah-Uddin H, Cameron M, Bannister TD, Almli L, Stevens JS, Bradley B, Binder EB, Wahlestedt C, & Ressler KJ (2013). Amygdala-Dependent Fear Is Regulated by Oprl1 in Mice and Humans with PTSD. Science translational medicine, 5 (188) PMID: 23740899