Regarding the paradoxical actions of TGFβ in brain, where the factor appears to be either protective against neuronal degeneration, as reported in this study, or deleterious, promoting inflammation, hydrocephalus, and vascular fibrosis and amyloidosis, (Wyss-Coray et al., 1995; 1997; 2000a), the following aspects should be taken into consideration:

1. The importance of the amount of TGFβ released. At physiological amounts the factor may be anti-inflammatory and neurotrophic, while when released in excess or in the absence of counter-regulatory elements, TGFbeta may turn to be proinflammatory and cause severe vascular abnormalities. There are other instances where the chronic dysregulated production of angiogenic factors, e.g., VEFG, have deleterious consequences (Detmar et al., 1998).

2. Although the studies describing protective and detrimental effects of TGFβ have been performed on apparently the same lines of TGFβ overexpressing mice, different animal batches were used. The present study by Brionne et al. does not disclose if, in the same animals where TGFβ protected...  Read more

Regarding the paradoxical actions of TGFβ in brain, where the factor appears to be either protective against neuronal degeneration, as reported in this study, or deleterious, promoting inflammation, hydrocephalus, and vascular fibrosis and amyloidosis, (Wyss-Coray et al., 1995; 1997; 2000a), the following aspects should be taken into consideration:

1. The importance of the amount of TGFβ released. At physiological amounts the factor may be anti-inflammatory and neurotrophic, while when released in excess or in the absence of counter-regulatory elements, TGFbeta may turn to be proinflammatory and cause severe vascular abnormalities. There are other instances where the chronic dysregulated production of angiogenic factors, e.g., VEFG, have deleterious consequences (Detmar et al., 1998).

2. Although the studies describing protective and detrimental effects of TGFβ have been performed on apparently the same lines of TGFβ overexpressing mice, different animal batches were used. The present study by Brionne et al. does not disclose if, in the same animals where TGFβ protected against kainic-acid induced degeneration, there were signs of vascular alteration (e.g., fibrosis and deposition of thioflavin-positive material) or glia activation, that were reported in the same lines previously (Wyss-Coray and col., 1995, 1997, 2000a). Whether the deleterious actions on vessels coexist with the protective actions on neurons is an intriguing question. If they don’t, a comparison between brain TGFβ contents in the animals used in the different studies would have helped to clarify if the opposite actions of TGFβ correlate with different degrees of production of the cytokine.

3. Studies in our lab using TGFβ-overexpressing mice have confirmed the pathological actions of the cytokine, including severe astrocytosis and microglia activation, hydrocephalus, tissue damage and robust deposition of thioflavin-positive material in vessels and meninges as reported by Wyss-Coray et al., 2000b. However, in contradiction with previous reports (Wyss-Coray et al., 1997; Lesne et al., 2003), we have found no evidence that there is upregulation of APP in these animals, nor that the thioflavin-labeled deposit contains amyloid-β, or any other amyloid for that matter (Galea et al., 2002). The notion that TGFβ-induced inflammation can cause Alzheimer disease-like vascular amyloid angiopathy should be thus thoroughly revised.

4. In conclusion, work in mice demonstrates the potential of TGFβ to perform “good” and “bad” actions in brain. The questions remains as to: i) what exactly determines that TGFβ be neuroprotective or detrimental; ii) the mechanisms underlying both actions; iii) the amyloidogenic role of TGFβ in vivo; and iv) whether the TGFβ overexpression detected in human brains in Alzheimer disease (Wyss-Coray et al., 1997) contributes to the pathology, or is a protective reaction. The evidence in this regard is at the moment just correlative.

View all comments by Elena Galea

Q&A with Tony Wyss-Coray. Questions by Tom Fagan.

Q: In your recent paper, you show that TGF-β1 may offer protection against excitotoxic injury to neurons. In previous papers, you had seen evidence that the cytokine may be toxic. Do the present observations take precedence?
A: We reported previously that TGF-β1 has detrimental effects on the cerebrovasculature in old TGF-β1 transgenic mice. This was not due to a toxic effect but more likely due to an inhibition of regenerative activities in blood vessels. From studies in peripheral organs and cell culture, it is evident that TGF-βs are produced by, and modulate, almost any cell type in the body. It is increasingly clear that TGF-βs can often exert positive and negative effects on a given biological process based on TGF-β concentration and receptor composition. For example, low levels of TGF-β1 appear to promote angiogenesis and vascular cell proliferation, but high levels inhibit cell growth and promote differentiation.

Consistent with these effects in the periphery, overexpression of TGF-β1...  Read more

Q&A with Tony Wyss-Coray. Questions by Tom Fagan.

Q: In your recent paper, you show that TGF-β1 may offer protection against excitotoxic injury to neurons. In previous papers, you had seen evidence that the cytokine may be toxic. Do the present observations take precedence?
A: We reported previously that TGF-β1 has detrimental effects on the cerebrovasculature in old TGF-β1 transgenic mice. This was not due to a toxic effect but more likely due to an inhibition of regenerative activities in blood vessels. From studies in peripheral organs and cell culture, it is evident that TGF-βs are produced by, and modulate, almost any cell type in the body. It is increasingly clear that TGF-βs can often exert positive and negative effects on a given biological process based on TGF-β concentration and receptor composition. For example, low levels of TGF-β1 appear to promote angiogenesis and vascular cell proliferation, but high levels inhibit cell growth and promote differentiation.

Consistent with these effects in the periphery, overexpression of TGF-β1 from astrocytes at intermediate to high levels resulted in vascular fibrosis and amyloidosis (defined as thioflavin S-positive deposits) in our model. In aged mice, the vascular changes were more prominent and accompanied by vascular cell abnormalities. Overexpression at very high levels resulted in hydrocephalus, likely due to excessive production of extracellular matrix proteins at the sites of CSF resorption. We have not investigated the effects of low levels or acute production of TGF-β1 in the CNS.

Our latest experiments show that TGF-β1 has trophic effects on neurons that cannot be substituted by other factors. These effects are likely to be independent of the vasculature, since survival of isolated TGF-β1-deficient neurons is impaired. It will be important to dissect the molecular mechanisms underlying this dependence and determine whether these effects are direct or involve glial cells.

Finally, we observed and reported prominent effects of TGF-β1 on glial cells in TGF-β1 transgenic and knockout mice. TGF-β1 overexpression results in a prominent astrocytosis, particularly around cerebral blood vessels, and we have speculated that this activation is a response to TGF-β1’s effects on endothelial cells (these cells express unique TGF-β receptors, called endoglin and ALK1, which are most likely not expressed on other CNS cells). TGF-β1 overexpression resulted also in increased microglial activation in the hippocampus of aged mice but, curiously, lack of TGF-β1 expression resulted in an even more dramatic activation of these cells (but not in astrocytosis).

Taken together, TGF-β1 appears to modulate activation states and behavior of all cell types we have analyzed. Obviously, we are just at the beginning of trying to understand what the consequences of TGF-β1’s actions are on these cells.

Q: Has your view of the role of TGF-1 changed over the past six years?
A: Not in the sense that we found contradictory effects, but that I did not anticipate its importance. The work with TGF-β1 transgenic mice has been interesting and rewarding. I had never expected that we would see—and keep seeing—so many different effects of this factor in the brain.

Q: In your recent paper you expressed TGF-β1 in astrocytes. Is it expressed in neurons, too? Is its neuronal expression physiologically relevant?
A: TGF-β1 appears to be expressed by neurons, but data are sketchy. Clearly, neuroblastoma cells express TGF-β1. More importantly, however, neurons express functional TGF-β receptors, and primary hippocampal neurons can be stimulated by TGF-β1. It is too early to say at this moment what the physiological role of TGF-β signaling in neurons is. Our experiments, together with studies by many others, show that it might have a protective function in neurons; others have suggested that it modulates synaptic facilitation, or promotes neuronal cell death during embryogenesis. Again, these functions are not exclusive.

Q: In her comment, Elena Galea mentioned some factors, such as the level of expression, the type of animals used, etc., that may explain the apparent differences in response to TGF-β. How do you explain these differences?
A: I agree with most comments made by Dr. Galea, but I am not sure how she arrives at the conclusion: “The notion that TGF-β1-induced inflammation can cause Alzheimer’s disease-like vascular amyloid angiopathy should be thus thoroughly revised.” Our studies provide clear evidence that TGF-β1 overexpression from astrocytes in our model results in cerebrovascular amyloidosis (thioflavin S-positive deposits) (Wyss-Coray et al., 1997) and degeneration in old age ( Wyss-Coray et al., 2000). When these mice were crossed with hAPP mice, we observed a dramatic change in distribution of human Aβ: Most Aβ accumulated in cerebral blood vessels in hAPP/TGF-β1 mice (Wyss-Coray et al., 1997). Interestingly, there was a three- to fourfold reduction in Aβ in the brains of these mice. We concluded that TGF-β1 promotes amyloidogenesis in the vasculature while reducing Aβ deposition overall (Wyss-Coray et al., 2001). We were also able to show that, in human AD cases, TGF-β1 mRNA levels correlated positively with amyloid deposition in blood vessels and that TGF-β1 immunoreactivity was increased in blood vessels with amyloid deposits. Moreover, amyloid deposition in the vasculature correlated inversely with deposition in parenchymal plaques, a finding supported by several other groups, most recently by Tian et al., 2003.

At this point, we still don’t know what the nature of the thioflavin S-positive deposits is in TGF-β1 single transgenic mice. Like many other pathologists, we called the deposits amyloid because of their specific color-imparting properties in binding thioflavin S. We believe that a biochemical analysis of the deposits will be necessary to identify their main protein component, but this experiment is very challenging to do, given the small quantities that can be obtained from a mouse brain (this is what we concluded in our initial report in Nature, 1997). Several groups have shown that TGF-β1 induces APP expression in cell culture, but we were not able to confirm this in vivo and never said so (this was misquoted by Dr. Galea). Unpublished observations from our lab showed a shift from shorter to KPI-containing splice forms of mouse APP at the mRNA level in TGF-β1 transgenic mice, but overall, there was no increase in mouse APP mRNA. Whatever the deposits are, they bind thioflavin S, are electron dense, and most importantly, they colocalize with human Aβ in hAPP/TGF-β1 mice. We demonstrated that TGF-β1 overproduction results in excessive accumulation of basement membrane protein in the vascular wall. We hypothesize that this results in the trapping of Aβ while it is being cleared, at least in part, via interstitial drainage channels along blood vessels or is transported across the blood-brain barrier into the plasma. There are likely small amounts of mouse Aβ present in these deposits, but this is a moot point given the prominent effect of TGF-β1 on human Aβ deposition.

Q: Do you think the elevated levels of TGF-β in AD brains are a cause or an effect of AD?
A: I don’t think increased TGF-β1 expression underlies AD. Rather, any form of cellular injury, as well as aging, results in increased production of TGF-β1 and activation of TGF-β signaling. There is a good possibility that this TGF-β1 modulates AD pathogenesis, similar to what we can observe in our mouse models.

Q: What needs to be done next to nail down the role of TGF-β?
A: Most studies on TGF-β focus on its peripheral effects. We need much more research focusing on its physiological function in the CNS. We recently generated a mouse that harbors a TGF-β response element in all cells of the body. The brain, in particular the hippocampus, showed several-fold higher baseline activity for this TGF-β response element than all other major organs (see ARF related news story). This further underlines the potential importance of this signaling pathway in the CNS. We will need to generate mouse models that directly inhibit or stimulate TGF-β signaling in a given cell type in vivo (using dominant negative or active TGF-β receptor transgenes or conditional knockouts), or use viral constructs to achieve this. Such models will be needed to further dissect the pathophysiological role of TGF-β in neurodegeneration and other CNS diseases, but they will also be helpful for electrophysiological and behavioral studies. Cell culture experiments using primary CNS cells will be needed to study the molecular mechanisms of TGF-β signaling. Again, we are just at the beginning.

View all comments by Tony Wyss-Coray

I wanted to thank Serge Rivest, Mathias Jucker, Tony Wyss-Coray, Joseph El Khoury, and Pritam Das for their helpful and thought-provoking comments, and to address some of their questions. I find it terribly interesting that the recent report by Richard, Rivest, and colleagues showed spontaneously increased TGF-β expression in immune cells near plaques of Tg APP/TLR2-/- mice. I agree that these striking findings are in line with the interpretation that increased TGF-β1 levels in AD patient brains, as shown by Wyss-Coray, Masliah, Mucke, and colleagues, likely serve the maladaptive role of maintaining an “immune privileged” brain milieu in AD patients and in these transgenic mouse models of the disease. We believe that overcoming this non-productive immune state will likely be key in targeting beneficial immune-mediated clearance of cerebral amyloid—and what better immune cell to target than the blood-borne macrophage (Greek etymology—“big eater”)? We also agree with Joseph El Khoury that a key aspect of this therapeutic modality will be promoting the Aβ phagocytosis response while...  Read more

I wanted to thank Serge Rivest, Mathias Jucker, Tony Wyss-Coray, Joseph El Khoury, and Pritam Das for their helpful and thought-provoking comments, and to address some of their questions. I find it terribly interesting that the recent report by Richard, Rivest, and colleagues showed spontaneously increased TGF-β expression in immune cells near plaques of Tg APP/TLR2-/- mice. I agree that these striking findings are in line with the interpretation that increased TGF-β1 levels in AD patient brains, as shown by Wyss-Coray, Masliah, Mucke, and colleagues, likely serve the maladaptive role of maintaining an “immune privileged” brain milieu in AD patients and in these transgenic mouse models of the disease. We believe that overcoming this non-productive immune state will likely be key in targeting beneficial immune-mediated clearance of cerebral amyloid—and what better immune cell to target than the blood-borne macrophage (Greek etymology—“big eater”)? We also agree with Joseph El Khoury that a key aspect of this therapeutic modality will be promoting the Aβ phagocytosis response while opposing the proinflammatory response, both of which likely exist as a continuum of innate immune cell activation profiles (Town et al., 2005). But, if we can accomplish this, will amyloid-reducing therapies ultimately be successful AD therapeutics? As stated by Dave Morgan and others on this forum, the first test of the amyloid cascade hypothesis of AD in humans will likely be the Aβ vaccine. We anxiously await whether the hypothesis holds up and delivers an efficacious AD therapy. If it does, then the floodgates will open for a whole host of amyloid-targeted AD therapeutics—both immune and non-immune.

About the issue raised by Mathias Jucker and Tony Wyss-Coray of CD11c as a marker for blood-borne innate immune cells/macrophages versus microglia, I should mention that we initially thought that CD11c would be a microglial marker in the context of AD. However, after examining numerous brain sections from various ages of wild-type versus Tg2576 or mutant APP/PS1 doubly transgenic mice for CD11c expression, we concluded that while microglia in the parenchyma around Aβ deposits were CD11b, CD45, MHC II, F4/80 Ag, and CD68 positive, they were negative for CD11c. However, we did observe a small number of round, non-process bearing CD11c positive cells within the lumen of blood vessels in both Tg2576 and APP/PS1 mice, consistent with Stalder and colleagues’ report of invading hematopoietic cells in brains of aged Tg2576 mice. At the time that we were checking for CD11c expression in AD mice, Alon Monsonego and Harold Weiner published a review in Science where they mentioned (as data not shown) that plaque-associated microglia were CD11c positive. I called Alon and asked him about the methodological details. However, after trying various tissue handling techniques, antibodies, and confocal settings, I was unable to reproduce this despite getting microglia in day 20 MOG-EAE brain sections to light up like a Christmas tree with CD11c. I came away thinking that it is possible to acutely activate microglia with the necessary vigor to promote CD11c expression, for example, in the context of EAE. However, I believe that this form of activation does not occur in AD mice, where the profile more closely resembles a chronic, persistent, low-level inflammation.

I have recently read the paper by Bulloch and coworkers with great interest, which shows the presence of CD11c/EYFP “dendritic-like” mouse microglia in multiple stages of life. However, because the authors did not quantify their observations, it is unclear how prevalent these cells are in the brain, and/or whether these cells arose from the blood or were long-term CNS residents. Further, the authors had difficulty in co-staining these cells with CD11c antisera in tissue sections, raising a possibility that those who work with transgenics are all too aware of: expression of transgenes is often more promiscuous than expected. In our study, we demonstrated a seven- to eightfold increase in CD45+CD11b+CD11c+CD68+Ly-6C- cells (presumed “anti-inflammatory” macrophages initially immunophenotyped by Littman’s group in Geissmann et al., 2003) in our crossed mice, and immunohistochemical approaches revealed prominent vascular cuffing, where these cells appeared to be entering the brain via cerebrovessels. Regarding the questions from Joseph El Khoury and Pritam Das about the origin of these brain macrophages, we agree that the “acid test” of whether the macrophage-like cells that we see in and around cerebral vessels and β amyloid plaques arise from the periphery or from within the CNS would either be a chimeric approach or parabiosis. We moved away from the chimeric approach following recent reports in Nature Neuroscience (Ajami et al., 2007; Mildner et al., 2007) showing that the act of irradiating the mice leads to brain infiltration of monocytes/macrophages—the very dependent variable that we are interested in testing. However, we believe that 1) parabiosis of AD mice with GFP+CD11c-DNR mice or 2) chemical methods of ablating hematopoietic cells in AD mice followed by reconstitution with GFP+CD11c-DNR bone marrow containing or depleted of macrophages represent possible strategies that we are currently pursuing.

Finally, Pritam Das raises the interesting questions of the long-term consequences of inhibiting TGF-β signaling on peripheral macrophages and the effects on T cells. We did not observe increased peripheral numbers of innate immune cells (including macrophages and dendritic cells), CD4+ or CD8+ T cells, or B cells in CD11c-DNR mice alone or in Tg2576xCD11c-DNR crossed mice, suggesting that an autoimmune state was not generated and that the increased abundance of macrophages in the brains of our crossed mice was β amyloid-directed. We also quantified T cells in brains of our crossed mice versus singly transgenic animals, and detected that about 4-5 percent of brain hematopoietic cells were TcRαβ positive (presumed T cells), and they were divided about equally between CD4+ and CD8+ subsets—however, these numbers were similar amongst wild-type, CD11c-DNR, APP/PS1, and APP/PS1xCD11c-DNR mice, suggesting that neither the CD11c-DNR nor the APP/PS1 transgenes were able to modify brain entry of T cells. Finally, regarding the issue of assessing neurodegeneration, we are currently pursuing this line of investigation by quantitative synaptophysin immunohistochemistry and hope to answer this question in the near future.

References:
Ajami B, Bennett JL, Krieger C, Tetzlaff W, Rossi FM. Local self-renewal can sustain CNS microglia maintenance and function throughout adult life. Nat Neurosci. 2007 Dec;10(12):1538-43. Abstract

Bulloch K, Miller MM, Gal-Toth J, Milner TA, Gottfried-Blackmore A, Waters EM, Kaunzner UW, Liu K, Lindquist R, Nussenzweig MC, Steinman RM, McEwen BS. CD11c/EYFP transgene illuminates a discrete network of dendritic cells within the embryonic, neonatal, adult, and injured mouse brain. J Comp Neurol. 2008 Jun 10;508(5):687-710. Abstract

Geissmann F, Jung S, Littman DR. Blood monocytes consist of two principal subsets with distinct migratory properties. Immunity. 2003 Jul;19(1):71-82. Abstract

Mildner A, Schmidt H, Nitsche M, Merkler D, Hanisch UK, Mack M, Heikenwalder M, Brück W, Priller J, Prinz M. Microglia in the adult brain arise from Ly-6C(hi)CCR2(+) monocytes only under defined host conditions. Nat Neurosci. 2007 Dec 1;10(12):1544-53. Abstract

Monsonego A, Weiner HL. Immunotherapeutic approaches to Alzheimer's disease. Science. 2003 Oct 31;302(5646):834-8. Abstract

Richard KL, Filali M, Préfontaine P, Rivest S. Toll-like receptor 2 acts as a natural innate immune receptor to clear amyloid beta 1-42 and delay the cognitive decline in a mouse model of Alzheimer's disease. J Neurosci. 2008 May 28;28(22):5784-93. Abstract

Stalder AK, Ermini F, Bondolfi L, Krenger W, Burbach GJ, Deller T, Coomaraswamy J, Staufenbiel M, Landmann R, Jucker M. Invasion of hematopoietic cells into the brain of amyloid precursor protein transgenic mice. J Neurosci. 2005 Nov 30;25(48):11125-32. Abstract

Town T, Nikolic V, Tan J. The microglial "activation" continuum: from innate to adaptive responses. J Neuroinflammation. 2005 Oct 31;2:24. Abstract

Wyss-Coray T, Masliah E, Mallory M, McConlogue L, Johnson-Wood K, Lin C, Mucke L. Amyloidogenic role of cytokine TGF-1 in transgenic mice and in Alzheimer's disease. Nature. 1997 Oct 9;389(6651):603-6. Abstract

View all comments by Terrence Town

I am glad that the researchers studying transgenic models are finally confirming our results published in 2002 (Fiala et al., 2002), which showed transmigration of macrophages across the brain vessel wall and clearance of plaques by these large macrophages.

The migrating macrophages broke through ZO-1 tight junction barrier and aggregated around brain vessels similarly as in HIV encephalitis. This has been followed by a recent publication in PNAS (Fiala et al., 2007). The animal studies cannot resolve the crucial question: are macrophages of patients with AD different from those of control subjects? The answers for interested readers are available in our PNAS article and more current work presented at ICAD. Not only macrophages penetrate across the blood-brain barrier but also clear oligomeric amyloid-β from neurons.

References:
Fiala M, Liu QN, Sayre J, Pop V, Brahmandam V, Graves MC, Vinters HV. Cyclooxygenase-2-positive macrophages infiltrate the Alzheimer's disease brain and damage the blood-brain barrier. Eur J Clin Invest. 2002 May;32(5):360-71. Abstract

Fiala M, Liu PT, Espinosa-Jeffrey A, Rosenthal MJ, Bernard G, Ringman JM, Sayre J, Zhang L, Zaghi J, Dejbakhsh S, Chiang B, Hui J, Mahanian M, Baghaee A, Hong P, Cashman J. Innate immunity and transcription of MGAT-III and Toll-like receptors in Alzheimer's disease patients are improved by bisdemethoxycurcumin. Proc Natl Acad Sci U S A. 2007 Jul 31;104(31):12849-54. Abstract

View all comments by Milan Fiala