Posted 26 August 2006

E-mail discussion Printable version

Return to Discussion Text

Mike D'Andrea led this live discussion on 4 December 2003.

Participants: Mike D'Andrea, Johnson & Johnson Pharmaceutical Research Division; Dara Dickstein, University of British Columbia, Vancouver, Canada; March D. Ard, University of Mississippi Medical Center; Keith Crutcher, University of Cincinnati, Ohio; Monica Carson, The Scripps Research Institute; Bob Nagele, University of Medicine and Dentistry of New Jersey; Amanda McRae, University of the West Indies in Trinidad; Greg Howes, Bloomington, Illinois; Mark Smith, Case Western Reserve University, Cleveland, Ohio; Walker McGraw, McGraw Intelligent Design Research, Atlanta, Georgia; Bernard Viau, Coordinator for the Alzheimer Home in Saint-Jerome, Quebec; Gabrielle Strobel, Alzheimer Research Forum.

Note: The transcript has been edited for clarity and accuracy.

Gabrielle Strobel
Let's get started. I am Gabrielle Strobel, managing editor of Alzforum, and pleased to moderate today. Mike, would you like to start? I believe you may have some lines prepared?

Mike D'Andrea
Thank you for joining us, and thank you to the staff of the Alzheimer Research Forum for the invitation, support, and recognition of this work. As you may have read already, the exciting aspects of this work are not that neuronal-specific autoantibodies exist in normal and AD patient serum, and not that the blood-brain barrier (BBB) becomes dysfunctional with increasing age and in AD patients, or that AD is associated with inflammation, but what appears to be novel is:

1) there are unique populations of neurons as per Ig immunoreactivity;
2) Ig-positive neurons appear apoptotic-like;
3) more parenchymal Ig immunoreactivity was detected in AD than in age-matched control brain tissues; and
4) there are more Ig-positive neurons in AD than in age-matched controls.

We have sketched out a few questions from the background information. There is no order, so perhaps we can start off by asking if there are potential animal models available to test the hypothesis. Any thoughts?

Keith Crutcher
Mike, I am not completely sure I understand the hypothesis, so I would have a hard time proposing a suitable animal model. What is the precipitating event in your model? Loss of BBB function?

Mike D'Andrea
I guess one part is to determine if anti-neuronal antibodies (if allowed to enter the brain) can injure neurons.

Gabrielle Strobel
AD mouse models generally do not have much neuronal loss. Are there exceptions, or is the small amount of loss later on in their lives sufficient to test your hypothesis?

Mike D'Andrea
Certainly, if the antigen becomes characterized (and we're trying), the pursuit of animal models will become a bit easier.

Gabrielle Strobel
Mike, how could one isolate these antibodies?

Mike D'Andrea
There are methods to break down BBB function, i.e., Pertussis toxin, but we'll still be handicapped studying the models as regards the antigen. As for now, since it is somewhat known that anti-neuronal antibodies exist in non-AD people, and for that matter, perhaps us, certainly the BBB becomes a focal point.

Amanda McRae
Hello, Mike. As you know, we have found antibodies directed against microglia cells in the serum of AD patients. So I am very happy with your results. A very interesting paper, indeed.

Monica Carson
I would not advocate using Pertussis toxin (PT), which has effects on more than the BBB. For example, it induces T cells to leave the lymph nodes and infiltrate multiple tissues, including the brain. We find that without PT, 1) T cells can readily enter the CNS after homeostatic (antigen-independent) proliferation, and 2) interestingly, in a normal, healthy mouse the CNS autoreactive T cells are specifically prevented from proliferating. PT changes that, and allows the CNS autoreactive T cells to become CNS destructive. Some of the changes in microglia function seen in murine transgenic models of AD should cause changes in the ability of microglia to present antigen to T cells, and thus, changes in the ability of the CNS to regulate the proliferation/activation of CNS-autoreactive T cells.

Mike D'Andrea
Monica, thanks. What are the actions of those T cells once in the brain?

Gabrielle Strobel
Monica and Mike, this is so interesting. You probably know the Martino study of Aβ vaccination with Pertussis toxin. It generated encephalitis. Monica, which autoreactive T cells are you referring to? What are they reacting against in brain?

March D. Ard
Animal models may be lacking exactly this feature of AD—the autoantibodies. Are mice comparable to humans in their partial loss of immune regulation with aging, allowing expression of numerous autoimmune diseases?

Gabrielle Strobel
How much of a barrier is the BBB really, even normally? I am confused about that.

Keith Crutcher
Mike, is it possible that the BBB dysfunction is a secondary consequence of something else going wrong in the brain and then allowing these antibodies to get in?

Bob Nagele
Keith, a great example is stroke-induced AD onset, and head trauma-associated AD.

Mike D'Andrea
Keith, great point; in fact, couldn't it be possible that the BBB dysfunction is related to increased amyloid?

Keith Crutcher
Sure, if you think amyloid has anything to do with AD!

Amanda McRae
March, if you would like, we can check to see if autoantibodies exist in the serum of mouse models.

March D. Ard
Amanda, that would really be a useful piece of information.

Amanda McRae
March, send me your e-mail address and we can discuss this further.

Mike D'Andrea
Gabrielle, what struck me about the vaccine story was that the Igs were getting into the brain.

Gabrielle Strobel
Mike, I think the numbers there are very unclear. Some labs say a fraction of a percent gets in. The Koistinaho lab transplanted GFP bone marrow of APP transgenic mice into irradiated animals and their brains became full of green microglia, so these cells came in (see ARF related news story). This was in old mice.

Dara Dickstein
When you irradiate a mouse, don't you compromise the integrity of the BBB? This would explain the GFP-positive microglia.

Gabrielle Strobel
Dara, I did not know that. Does irradiation compromise the BBB?

Monica Carson
Yes, irradiation does harm the BBB. Greg Howes
Blood vessels are very sensitive to the effects of irradiation.

Mike D'Andrea
...which brings us back to the integrity of the BBB as a function of how much Igs can enter the brain.

Bob Nagele
Mike, the pronounced memory and cognitive loss seen in patients with systemic lupus are a point in favor of this whole idea.

Keith Crutcher
Mike, did you look at any regions other than the entorhinal cortex and hippocampus?

Mike D'Andrea
How about methods to find the autoantigen? Any thoughts?

Gabrielle Strobel
Mike, good question. It has taken a huge effort to identify the offending autoantigen in multiple sclerosis and I am not sure it is definitively proven.

Mike D'Andrea
We are trying several modalities, such as using laser capture to characterize the Ig-positive neurons vs. the Ig-negative neurons, like those in Figure 3 of the paper (see D'Andrea, 2003).

Monica Carson
How important is it to find the autoantigen? There really hasn't been much progress on MS or autoimmune diabetes on these fronts. Due to the genetic polymorphisms in the human population, multiple antigens could be a problem.

Amanda McRae
Hello, Mike. As we have discussed, how do your results relate to Down's syndrome?

Mike D'Andrea
Amanda, how is the cardiovascular system in Down's? Are there BBB issues?

Bob Nagele
Amanda, Down's patients have many cardiovascular problems, which most likely extend to the level of defective vessels, especially in the elderly.

Monica Carson
In answer to Mike's earlier question: We have placed a transgene antigen in the CNS (not localized anywhere else in the mouse), and then have followed T cells specific for this antigen. In our case, they reacted to astrocytes.

Bob Nagele
Monica, we may be looking for more than one antigen.

Monica Carson
Bob, I agree with you.

Mike D'Andrea
Perhaps we might be able to understand the receptor type that's expressed on the immunoreactive neuron, and presume that the autoantigen could be a unique receptor, ion channel, etc?

Gabrielle Strobel
Mike, regarding the ion channel: The disease limbic encephalitis can, in rare cases, be caused by autoantibodies against potassium channels. British researchers reported this at the last ANA meeting this October. Patients who were treated with plasma exchange, intravenous immunoglobulins, and steroids reportedly improved. Has anyone heard about this?

Mike D'Andrea
It's interesting to note that Rasmussen's encephalitis (RE) has one known autoantigen, but I agree with Bob that it could be several, not to mention an antigen that mimics another.

Gabrielle Strobel
If finding the autoantigen(s) is not the priority, what should the next experiments be to advance the hypothesis?

Mike D'Andrea
Good question, Gabrielle. We also have efforts to characterize serums of all sorts (AD, non-AD, young) and see if there are pattern differences in the labeling, titer differences, etc.

Bob Nagele
All, it would be interesting to see if the autoantibodies are cell type-specific.

Monica Carson
If the BBB is generally leaky, then why are selected neurons affected in AD? Should we model specific neuron autoimmune responses against specific neurons (and not specific autoantigens) to see the physiological effects?

Mike D'Andrea
Monica, I think you hit it on the head. Why are only specific neurons affected?

Dara Dickstein
Monica, some recent papers state that the leakiness of the BBB in AD may be transient and/or focal. Maybe this explains why certain neurons are affected.

Keith Crutcher
Dara, your suggestion might account for regional variations, but I am not sure I see how you would get neighboring neurons with such differences.

March D. Ard
It would be interesting to see if antibodies from AD serum can really cause neuronal apoptosis.

Monica Carson
I like March's experiment!

Mike D'Andrea
March, your idea may not be too difficult to perform...in vitro on neuronal cultures. What do you think?

March D. Ard
Mike, yes, I think it would be reasonably easy.

Bob Nagele
Monica, I would like it better if there were more evidence that apoptosis really played a role in early stages of AD.

Walker McGraw
Could it be that distressed neurons (i.e., neurons performing overactive autophagy) are spewing out antigenic components?

Monica Carson
I think it would be interesting to test if AD serum specifically binds certain brain regions (i.e., in AD, do we have specific immune responses against specific neurons or susceptibility of specific neurons to a global immune insult?).

Mike D'Andrea
Walker, certainly, but as I mentioned in the paper, greater than 99 percent of apoptotic cells were Ig-positive.

Gabrielle Strobel
Mike, would that not suggest that the autoantigen is a specific receptor? Or perhaps the affected neurons are already "vulnerable" due to specifics of their metabolism (this seems to play a role in PD) or oxidative stress, or proximity to amyloid.

Mike D'Andrea
I think we're back to neuronal-specific antigens, right, Keith?

Keith Crutcher
Mike, yes, and the fact that next-door neighbors appear so differently labeled.

Mike D'Andrea
Keith, I think that what appeared most interesting was seeing some very strongly labeled neurons right next to completely negative neurons.

Keith Crutcher
Yes, that is what I meant. It would seem that any local opening of the BBB would affect all of the neurons in the neighborhood, not just a subset, unless there is something else going on.

Monica Carson
Has anyone tested if antibodies from AD patients bind certain brain regions (AD-affected regions) of non-AD human autopsy brains? This would indicate a specific cellular autoimmune target, not necessarily one single autoantigen.

Mike D'Andrea
Of course, it may also be that some positive neurons next to some negative ones could be explained by the projection of these neurons. If the entry of the Ig is through the synapse, and if neighboring neurons have their synapses in different areas of the brain, then Ig-positive neurons could lie within microns of an Ig-negative neuron. So then, if the BBB injury (a focal event) does not incur total Ig labeling to neighboring neurons, then the projections may provide a better clue.

Keith Crutcher
Mike, that is an interesting possibility. But then you might expect some correlation with soma size. Did you see that?

Mike D'Andrea
Keith, another good point. No, I did not adjust for size, but could take that into consideration. It's not always easy to get the proper orientation.

Keith Crutcher
Mike, I asked a question early on that I think got overlooked. Have you looked at any other brain regions, i.e., areas that are not primarily implicated in AD?

Mike D'Andrea
Good question. I did look at cerebellum and noted Ig labeling, but not to the point where those cells appeared apoptotic.

Greg Howes
The BBB theory has been argued before with leaking amyloid proteins, causing start-up amyloid plaques. However, if the BBB is leaking, it would be a more general pattern, I would think.

Bob Nagele
Not if the leaks were focal and caused by local pathological changes in blood vessels.

Mark Smith
Sorry if this has been asked, but do you see anything in diseases with a high vascular component, e.g., "pure" CAA? Also, do you see correlation between Ig neurons and CAA in vessels?

Mike D'Andrea
Ah, I did do some earlier work on Aβ and Ig; in fact, it's what got me to this point. I wanted to validate that there are unique plaque types in the AD brain, such as vascular vs. neuronal, so I figured leaky proteins such as Ig could be the fingerprint, but not all plaques lined up. I'm still pursuing this.

Monica Carson
Thomas Krucker at The Scripps Research Institute is studying the APP23 murine model and has argued that changes in the vasculature of the young APP23 mouse appear earlier than any other changes. See Beckmann et al, 2003.

Gabrielle Strobel
Mike, may I ask why you did this study at home?

Mike D'Andrea
Peace and no distractions allow intense microscopy. Is that what you ask?

Gabrielle Strobel
I was just curious because it is so unusual. Having done this “home alone,” were you able to blind your samples? Or was the difference in Ig binding between controls and AD very large?

Mike D'Andrea
Gabrielle, all were blinded, but if anyone tells you otherwise, image analysis really doesn't care which is which. I think the BBB could be the first hit as per AD, but is that attributed to oxidative damage, head trauma, amyloid?

Bob Nagele
Gabrielle, I have also observed what Mike is reporting here—but in my lab!

Mike D'Andrea
Thanks, Bob. Did you also do that at your home?

Bob Nagele
Mike, You know me—at home—at the shore—it's easy to carry a slide box.

Gabrielle Strobel
Bob, thanks for the confirmation—always a plus in science.

Mark Smith
I may have missed it in the paper; did you isotype Ig?

Mike D'Andrea
Yes, IgG, IgM, and IgA were all present. We tested our antibodies and we also have specific neurons that light up.

Dara Dickstein
We have recently published in Microcirculation (Ujiie et al., 2003) that in the Tg2576 mouse, the BBB appears to be leaky as early as four months, whereas plaques do not appear until around 10 months.

Mark Smith
Dara, so you find similar neuronal Ig in Tg mice?

Dara Dickstein
We didn't look at specific neuronal Ig.

Mark Smith
Surely you will now?

Bob Nagele
Dara, could that be because it may take a while for this stuff to accumulate in neurons before plaques appear?

Mike D'Andrea
Dara, can I ask how you stain for plaques? Formic acid? Are they around vessels?

Dara Dickstein
Mike, we and others have looked for plaques using both Congo red staining and using a monoclonal against human Aβ. Not sure about the vessels in particular; however, Aβ is known to be present in vascular cells.

Mike D'Andrea
Dara, I ask because of our (Bob Nagele's, as well) notion that there are different plaque types. Can we obtain some of these brains? I can do some detection of murine Igs in the brain as a function of BBB leakiness? And as for neuronal death, did you see it in these animals, and at what stage? Do you typically use formic acid for your Aβ detection?

Gabrielle Strobel
It looks as if a lot of attention these days is shifting away from neuronal loss and toward synaptic dysfunction, synaptic loss, and then a gradual dying back of the projections toward the cell body rather than an initial insult at the cell body that causes apoptosis. Do you think this emphasis is misguided? Or would the two processes fit together?

Mark Smith
Gabrielle, processes certainly fit together; emphasis on one or another likely has to do with shifts in direction of wind and, of course, which model you are trying to validate.

Bob Nagele
Gabrielle, you are right—this may be reflecting the recognition that most of what we know about AD comes from brains that are pretty far gone, with lots of secondary inflammation.

Mike D'Andrea
Gabrielle, no, not misguided. In fact, synaptic decline precedes plaques (at least from what we have learned thus far), and synaptic decline appears to correlate with amyloid loading, as well. Can you comment, Bob?

Walker McGraw
Mike, I know in five-micrometer sections that it is difficult to see, but did you see any signs of T cell recruitment either in or adjacent to the vessels?

Mike D'Andrea
Good question. I do see monocytes in vessels and some appear to be marginating, but not to the point where I see many around these affected neurons. What's your thought?

Walker McGraw
Just thinking about how loss of BBB and endothelial damage can lead to recruitment.

Mark Smith
Mike, do we know what these antibodies are recognizing or do you think they are random selections from blood?

Mike D'Andrea
Mark, no, not yet, but at first glance they appear to be recognizing something very specific.

Monica Carson
Mike, what kind of specificity?

Mike D'Andrea
Specificity from the point to labeling. As there is plenty of Ig labeling in the parenchyma (Figure 1C), most appear not to be specific to neurons.

Mark Smith
I would agree. Do you think only the specific ones get through or all get through and only the specific ones stay?

Mike D'Andrea
I agree. Sometimes I notice intense Aβ labeling around small vessels, yet the endothelium appears to be just fine. I'm not a BBB person—can you have leakiness with an intact endothelium? Glial cell issues? I sometimes think it's dumb luck that these vascular-derived Igs find their way in, and then bind.

Monica Carson
Mike, how significant (for disease) is the non-neuronal Ig labeling of non-neurons?

Mike D'Andrea
Monica, I think it's pretty benign, as the astrocytes appear to have internalized the Igs and they look fine. Is that what you ask?

Monica Carson
I might suggest that glial binding of Ig may not be benign. In vitro, it isn't. It can lead to glial activation (such as cytokine production).

Mike D'Andrea
Okay, but as for neuropathology, they "look" okay.

Monica Carson
Mike, while glial morphology changes are clearly indicative of changed function, I think it's important to note that glial activation and glial changes in gene expression do not always correlate with overt changes in morphology. Thanks for the discussion!

Walker McGraw
Yes, there is a great deal of work with vascular permeability enhancement (directed and active), especially with bradykinnin and proteinases. That is one reason I asked about monocyte recruitment.

Mike D'Andrea
Walker, do you see such things?

Walker McGraw
I am mostly a wet biochemist, but some of the proteinases I have been working on have a very similar staining pattern with neighboring cells that are just as "clean."

March D. Ard
Mike, Gabrielle, and all, thanks; this has been fun.

Gabrielle Strobel
Mike, would it make sense to compare the APP23 mice to other transgenics, since the former have lots of CAA? Perhaps that would provide indirect hints at the role of the BBB?

Keith Crutcher
Very interesting idea, Mike. I think the biggest questions for me are: Is this a cause or consequence of the disease? Is there regional specificity that correlates with other pathology? And how would this hypothesis relate to known genetic risk factors? Thanks for a stimulating session!

Mike D'Andrea
Keith, you ask the silver bullet ones. I guess at this point, we can't afford to miss out.

Walker McGraw
No, the Ig-positive neurons are dystrophic, etc., but there is a tight correlation to vascular location, i.e., halo effects.

Mike D'Andrea
I agree!

Gabrielle Strobel
As we have reached the end of the hour, let me thank you all for coming. Thank you to Mike and all for your time and interest.

Walker McGraw
Thanks for the interesting discussion.

Mike D'Andrea
Thanks to all. Please e-mail me for follow-up comments and questions.