Dr Freeman. I'd like to thank the Western Thoracic Surgical Association, our moderators, and the members and guests. Today I'm going to talk with you about spinal cord IR injury.
As we know, paraplegia continues to be a devastating complication in thoracoabdominal aortic interventions, particularly in high-risk patients. We know that spinal cord ischemia then triggers resultant metabolic derangements that lead to a reperfusion injury. The alpha-2 agonist dexmedetomidine, which we know better Precedex (Hospira, Lake Forest, Ill), has been shown to attenuate IR injury in a variety of organ systems. In an in vivo mouse model, we previously showed that pretreatment with dexmedetomidine preserved functional outcome after the spinal cord IR injury. However, as with any model, this in vivo model had its limitations, particularly with having to use the whole spinal cord. We were unable to separate the effects that the dexmedetomidine had on the specific cellular components of the spinal cord. So the mechanism has yet to be fully elucidated.
Our original aims were to first maintain spinal cord cell cultures of astrocytes and neurons, and then subject the cells to an in vitro model of IR injury, called “OGD.” Our model of cell culture begins by harvesting cells from a spinal cord in block from postnatal mice. You may wonder why we have to use postnatal mice versus adult mice, and it's simply because adult mice neurons do not grow in culture. Then you follow by a digestion of the tissue and select it for a specific cell type with OptiPrep (Sigma-Aldrich, St Louis, Mo). What we are able to do further is select for neurons by the way you plate them and the type of media they are placed in. After they are in culture for 7 to 10 days, they are considered a mature neuron culture.
Astrocytes are likewise plated and grown in a media, basically the same as a neuron media but with addition of serum. This is a light microscope image of our neurons, and these are again considered mature. These are approximately 7 days in vitro, and you can see all the projections between each neuron. These are astrocyte cultures, and you can see by their morphology that they are different from the neurons.
Our model of OGD for IR injury begins by placing the oxygen glucose–deprived media onto the cells and then placing those cells into an anoxic chamber to simulate ischemia. Then we return the cells to a normal incubator with a normal neuron media to simulate a reperfusion.
What we found is basically an ischemic dose response with the OGD of neurons. At 30 minutes of OGD, neurons had a decrease in their viability. However, this decrease wasn't statistically significant. At 1 hour, we went down to a viability of neurons at approximately 60%, and this was statistically significant. At a further 2 and 4 hours, we had a statistical decrease in viability. We chose the 60% viable, the 1-hour OGD, for further studies for 2 reasons. There are enough cells remaining that you can potentially do further studies on them, and at that time point we had a statistical decrease in viability.
What we wanted to do after we were able to culture the cells, and then after we were able to kill the cells, is to find a way to preserve the cells. One way that we were interested in looking at doing this is the astrocytes around the neurons. In vitro neurons are by themselves; however, in the spinal cord, they are surrounded by astrocytes. Astrocytes are the most abundant cell type in the nervous system. They comprise approximately 85% of the cells. They organize the architecture, provide communication pathways, control ion homeostasis, and participate in the blood–brain barrier. Also, failure of these support systems can be detrimental to neuron survival in general, but specifically in an injured state.
Astrocytes are thought to provide specific neuroprotection by release of neurotrophic factors, and astrocytes are the major source of GDNF in the entire body. This GDNF has been shown to be a potent neurotrophic factor, specifically for motor neurons to modulate neuronal death, and to support neuronal survival and regeneration. Also, production of GDNF has been specifically linked to alpha-2 agonism in cerebral models.
Our next hypothesis moving forward was that treatment of astrocytes with dexmedetomidine, the alpha-2 agonist, would increase GDNF. We found that astrocytes in general did not produce GDNF on their own. They had to be made to produce it. With dexmedetomidine at 0.1 μmol/L, there was a slight increase at 24 hours, although it was not statistically significant. With dexmedetomidine treatment of 1 μmol/L and 10 μmol/L, there was a significant increase from all time points compared with control, but specifically 24 hours after treatment with dexmedetomidine, there was a significant increase in GDNF production. By looking further at the 1 μmol/L of dexmedetomidine treatment, we found that at 4, 8, and 16 hours, there was an increasing amount of GDNF. Specifically, at 24 hours we had a higher increase of GDNF and then a drop-off at 36 hours. What this led us to believe is that at 24 hours after dexmedetomidine treatment with 1 μmol/L, we were seeing a large increase in GDNF concentration in the astrocyte media. This gave us the idea to look at giving this astrocyte media to the neurons during reperfusion to see if we could elucidate preservation and viability of the neurons. Alpha-2 agonism with dexmedetomidine increases the production of GDNF in both a time- and dose-dependent fashion. Again, we wanted to further investigate what this GDNF would provide as far as neuroprotection.
Our next hypothesis moving forward was that neurons subjected to OGD that are given dexmedetomidine-treated astrocyte media instead of their normal reperfusion media would have a preservation in their viability. Again, this is showing the same part of a graph I have shown before, but just to reiterate, with no OGD we have 100% viable cells. Down with 100, or with 1 hour of OGD, we are down to 60% viable cells. When we added the dexmedetomidine-treated astrocyte media during reperfusion, there was significant preservation and viability up to approximately 80%. This was statistically significant preservation. Again, with 1 hour of OGD with controlled reperfusion media, we have 60% viable cells. When we use dexmedetomidine-treated astrocyte media instead of reperfusion media, we have up to approximately 80% viable cells, which is significant preservation.
But what we were wondering was, you know we looked at GDNF. That would be wonderful if that's what was causing the preservation and viability, but how do you know? There could be lots of things in the media that are causing the neurons to be preserved. So what we did was use a GDNF neutralizing antibody, and we found that use of this antibody actually removed the preservation and viability that the astrocyte media gave with the neurons. There was a significant decrease in preservation and viability when we used a neutralizing antibody. What this told us was that specifically, GDNF is causing preservation.
Isolated spinal cord neurons subjected to OGD mimics IR injury. The alpha-2 agonist dexmedetomidine increases GDNF, and dexmedetomidine-treated astrocyte media provides protection from neuronal IR injury via specific GDNF stimulation. Spinal cord cultures can be a powerful tool for future study in spinal cord IR injury, and treatment with the alpha-2 agonist dexmedetomidine has clinical potential in aortic surgery as a way to prepare the spinal cord for the insult of IR injury.
Dr Ikonimidis (Charleston, SC). Your neutralizing antibody work is compelling with regard to normalization of cell death in this model. I have a few questions related to it. The first is a conceptual one. You acknowledge the fact that you are not able to culture human nerve cells, and so you have to use pediatric cells, pediatric neurons for culture, and clearly this is a disease of the adult spinal cord and not the infant spinal cord. Could you speculate on the potential confounding with regard to the age of the cells?
Dr Freeman. As you know with any in vitro model, you have limitations specific to what works and what does not in vitro, as in certain medias only available, and particularly to neurons. Adult cells just can't be cultured. So in a mouse model, what people do to grow neurons, they use a prenatal mouse, as in cut a pregnant mouse open and use prenatal neurons, or they use postnatal neurons. Now when you talk about low-hanging fruit, culturing prenatal mice neurons is easy. They never die, they always live. Even postnatal mice that are 2 to 3 days old survive approximately 60% of the time in culture. In adult neurons, although the literature says that it does not really happen, it can't really be done, I tried it anyway, and I got them to stay alive only 2 to 3 days in culture, at best, which is not really long enough for them to settle in.
Dr Ikonimidis. True. Question your ability to separate out cell types. So you are now able to separate out astrocytes from neurons, but the neuronal component is made up of both motor and sensory neurons, and I guess because we are conceptually speaking, we are interested in protecting the motor neurons. Do you have any data on what cell death you are observing in your cultures because the model as you have it has approximately 40% cell death, which is reduced to 20% with GDNF. Which neurons are you protecting? Are you protecting motor, sensory, a mix of those? Do you have any comments on that?
Dr Freeman. In general, when you are culturing the spinal cord, you know you can't really separate out. It can't be done with the way that we separate out neurons, and I haven't found a way to separate out them more specifically. If there was a way, that would be great. It would give you even more information on how to preserve the neurons. The only way that we would have knowledge of it would be by morphology, and even then in vitro you can't tell the different cell types. However, there's potential in the future to look toward a way that you can culture out specifically the motor neurons versus nonmotor neurons to get you even closer to identifying what we can do to preserve these.
The GDNF specifically has been shown to be a potent neurotrophic factor for motor neurons, specifically, Parkinson's and amyotrophic lateral sclerosis research has been booming in GDNF. Unfortunately, the biggest issue with GDNF is not whether it's going to help the neurons, but how to deliver it. A lot of the amyotrophic lateral sclerosis literature has been focusing on how to get GDNF to the brain or the cord, and they are having trouble with crossing the blood–brain barrier, which is why dexmedetomidine (Precedex) has such potential because it can elicit this response without having to worry about how to get GDNF across the blood–brain barrier.
Dr Ikonimidis. The astrocyte-conditioned media was delivered to your neurons after 1 hour in your oxygen glucose–deprived state. Clinically, that would equate to a patient developing paraplegia, and then we give the drug. Is that how you envision dexmedetomidine (Precedex) being used clinically? Could you comment on that?
Dr Freeman. I think what we envision for an actual clinical model would be patients admitted the day before, maybe receiving a lumbar drain, things like that, and maybe starting dexmedetomidine at that time. Patients generally tolerate dexmedetomidine (Precedex) very well. We don't know what the dosing would have to be to get the response we see in vitro, but of course that would be sorted out in the future, but giving dexmedetomidine (Precedex) as a pretreatment, I think instead of waiting to see if the patient is one who develops paralysis, that everybody potentially could get it, because the side effect profile is so low.
Dr Ikonimidis. My point being that you might want to consider an experimental design.
Dr Freeman. Absolutely.
Dr Ikonimidis. That includes pretreatment. You're talking about this hypoxia glucose deprivation model as a model of IR, and it probably isn't. Because the definition of ischemia is really a low-flow state, not a low-oxygen state. I think you could give consideration to how you frame your description of that. It might be more of a hypoxia reperfusion model than an IR model. It's semantic, but to basic scientists who do this sort of stuff with oxygen-deprived cultures, it's an issue. If you have a lot of volume, you won't develop toxic metabolite accumulation that could contribute to cell death. It's just something to consider.
Dr Freeman. Right. Sometimes we talk about things and it's just semantics, but word choice really is everything when you're describing a model. I think it's the best model we have right now, but there are always ways to improve on what we have.
Published online: November 21, 2014
© 2015 The American Association for Thoracic Surgery. Published by Elsevier Inc.