The Journal of Neuroscience, April 11, 2012 • 32(15):5017–5023 • 5017

Disease Focus

Editor’s Note: Disease Focus articles provide brief overviews of a neural disease or syndrome, emphasizing potential links to basic neural mechanisms. They are presented in the hope of helping researchers identify clinical implications of their research. For more information, see http://www.jneurosci.org/misc/ifa_minireviews.dtl.

Alexander Disease

Albee Messing,1 Michael Brenner,2 Mel B. Feany,3 Maiken Nedergaard,4 and James E. Goldman5 1Waisman Center and Department of Comparative Biosciences, University of Wisconsin Madison, Madison, Wisconsin 53705, 2Department of Neurobiology, Evelyn F. McKnight Brain Institute, Center for Glial Biology in Med, University of Alabama Birmingham, Birmingham, Alabama 35294, 3Department of Pathology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115, 4Center for Translational Neuromedicine, Departments of and Neurosurgery, University of Rochester Medical Center, Rochester, New York 14642, and 5Department of Pathology and Cell Biology, Columbia University, New York, New York 10032

Introduction science audience to the unique research op- cation divides patients into three catego- 1The past decade has witnessed a burst of portunities posed by this disease. ries based on age of onset, infantile (0–2 speculation and data about how The first recognized patient was a 16- years), juvenile (2–12 years), and adult dysfunction contributes to the phenotypes month-old boy who died after a progres- (Ͼ12 years) (Russo et al., 1976). More re- of the well known neurodegenerative dis- sive course that included , cently, a different classification has been eases such as Alzheimer’s, Parkinson’s, , and psychomotor delays proposed, with only two categories of type Huntington’s, and amyotrophic lateral scle- (Alexander, 1949). Pathology revealed I and type II, hinging more on distribution rosis, as well as other types of disorders such abundant astrocytic accumulations of eo- of lesions and clinical presentation rather as epilepsies and (Rempe sinophilic cytoplasmic inclusions, recog- than age of onset (all type I cases being early and Nedergaard, 2010). However, these are nized by neuropathologists as Rosenthal onset, and type II cases occurring at all ages) complex syndromes that likely represent fibers (after the 19th century German (Prust et al., 2011). Early-onset patients pre- combined abnormalities of neurons, glia, pathologist who first described them in dominate in the literature, but this likely re- and immune cells. The clearest example of the context of an old astrocyte scar; flects ascertainment bias as the adult-onset acting as the primary culprit in Rosenthal, 1898) (for review, see Wippold et patients in particular are frequently misdi- disease is Alexander disease, which is caused al., 2006) (Fig. 1). During the subsequent 15 agnosed with other conditions. The early- by dominant gain-of-function in years, additional individuals with similar onset patients typically present with the glial fibrillary acidic protein (GFAP) pathology were reported, and in 1964 Friede , , or developmental delays, (for an extensive review, see Brenner et suggested that these all represented a single whereas the later-onset patients more often al., 2009). Although this disorder is quite disease, and recommended they be named have signs of hindbrain dysfunction such as rare, the extent to which we can understand after Alexander. Although the initial finding , palatal myoclonus, and or how astrocyte function is impaired in Alex- of prominent aggregates in astrocytes dysphonia (Russo et al., 1976). Diagnosis is ander disease, and the strategies we can de- prompted Alexander himself to suggest that usually suspected based on characteristic vise to restore astrocyte function, will have this might represent a primary disorder of appearances on MRI, with a frontal leuko- significant implications for how we deal astrocytes, it was the discovery of the genetic dystrophy common in the younger patients with the many more common neurological basis for the disease that established Alexan- and a hindbrain predominance of lesions, diseases that confront us. The purpose of der disease as a primary disorder of this ma- sometimes with atrophy of the medulla ob- this review is to introduce the wider neuro- jor CNS cell type (Brenner et al., 2001). longata and cervical spinal cord, in the later- Approximately 95% of patients harbor onset patients (Fig. 2) (van der Knaap et al., mutations in the GFAP gene, and no other 2001, 2005, 2006; Namekawa et al., 2002). Received Oct. 25, 2011; revised Jan. 19, 2012; accepted Feb. 21, 2012. genetic causes are known (Brenner et al., Lifespan is related to age of onset; type I pa- This work has been supported by grants from the NIH (NS-22475, NS- 42803, NS-060120, and HD-03352), the Juanma Fund, the Jack Palamaro 2009). Only one population-based survey tients have a median survival of 14 years, Fund, and the Jelte Rijkaart Fund. We are extremely grateful to the pa- has been conducted, arriving at an inci- and type II patients a median survival of 25 tients, families, and clinicians who have participated in these studies over dence of ϳ1:2.7 million (Yoshida et al., years (Prust et al., 2011). Most of the initially the years. 2011), although this figure is likely an un- discovered mutations occurred de novo, but Correspondence should be addressed to Albee Messing, 1500 Highland Av- enue, Room 713, Madison, WI 53705. E-mail: [email protected]. derestimate. The age of onset is quite vari- with sharpening the tools for diagnoses of DOI:10.1523/JNEUROSCI.5384-11.2012 able, ranging from prenatal through the the more difficult patients to discern the Copyright © 2012 the authors 0270-6474/12/325017-07$15.00/0 sixth decade. The most common classifi- later-onset form, increasing numbers of fa- 5018 • J. Neurosci., April 11, 2012 • 32(15):5017–5023 Messing et al. • Alexander Disease

frank cavitation, there is no obvious de- (Tang et al., 2010) that may in fact result crease in astrocyte numbers. However, in from a slowing of the normal polymeriza- without cavitation there is a tion rate. In addition, astrocytes respond marked loss of and a variable (but in a self-destructive fashion by increasing unquantitated) loss of oligodendrocyte their expression of GFAP (Jany et al., 2011). lineage cells and axons. There is also a loss The resultant elevation in GFAP monomers of neurons, particularly in the hippocam- available for polymerization could explain pus and striatum (our unpublished obser- why astrocytes are able to form normal- vations). The oligodendrocyte (myelin) appearing filaments, whereas intermediate and neuronal deficits that produce the filament proteins with homologous muta- clinical signs of Alexander disease are tions in other cell types do not. What stim- likely consequences of astrocyte dysfunc- ulates this astrocyte response is not clear, but tion rather than astrocyte loss. This dys- could involve an intercellular circuit in function could involve the loss of a which other cell types are damaged by the normal, supportive role (e.g., glutamate dysfunctional astrocytes and then signal uptake), the gain of a new or increased back to the astrocytes via any of several path- toxic activity (e.g., TNF␣ release), or a ways (Buffo et al., 2010), or could occur en- combination of both. tirely within the affected astrocyte by mechanisms not yet known. This increase in Disease mechanisms expression, coupled with decreased degra- Alexander disease as a gain-of-function dation (see The concept of GFAP toxicity, disorder of GFAP below), results in a positive feedback loop Alexander disease is considered a gain-of- for disease progression. The presence of the Figure 1. Morphological features of Rosenthal fibers. A, function disorder in the sense that the feedback loop could also be the reason that Rosenthal fibers concentrated in the astrocytic endfeet sur- GFAP mutations produce consequences only astrocytes, and not other GFAP- rounding a blood vessel (V) in the brainstem of a 1-year-old that differ dramatically from those caused expressing cells, form Rosenthal fibers. child with Alexander disease. Hematoxylin and eosin stain, by the absence of GFAP (for review, see While the above model provides an ex- paraffin section. B, Rosenthal fibers (arrow) surrounded by Brenner et al., 2009). There are three ob- planation for why normal-appearing fila- intermediate filaments (arrowhead) in an astrocyte cell body servations that indicate that the primary ments are seen in Alexander disease but froma17-month-oldchildwithAlexanderdisease,viewedby effect of the GFAP mutations is a gain of not in other disor- transmission electron microscopy [reprinted from Eng et al. function: (1) Gfap-null mice have a mini- ders, it does raise the possibility that these (1998), their Fig. 5B, with permission of Wiley-Liss, a subsid- mal phenotype that bears little resem- other disorders may also have gain-of- iary of John Wiley & Sons]. blance to Alexander disease; (2) astrocytes function components in addition to their of Alexander disease patients display loss-of-function phenotypes. Indeed, it milial cases are being detected. When passed abundant levels of normal-appearing, 10 has been observed that disease due to ker- to progeny, the mutations have typical nm GFAP filaments; and (3) all GFAP atin point mutations, in which aggregates autosomal-dominant inheritance with mutations discovered in Alexander dis- form, may be more severe than that pro- nearly 100% penetrance (for further discus- ease patients allow production of essen- duced by its absence (Fuchs and Cleve- sion on the topic of penetrance, see Stumpf tially full-length mutant protein. Neither land, 1998). The extent to which the toxic et al., 2003; Messing et al., 2012). There is no nonsense mutations nor major deletions mechanisms are similar in these different predilection for any ethnic population, or have been found, and the only frameshifts intermediate filament disorders is yet to any gender bias. are at the extreme C-terminal end of the be explored. It is also interesting, but per- protein (Fig. 3). The few internal deletions haps reflective of our lack of understand- Concept of a primary astrocyte disease and insertions that have been reported are ing of intermediate filament structures, that The assignment of Alexander disease as an in-frame. In these respects, Alexander dis- mutations throughout the rod and tail do- astrogliopathy—a primary disease of as- ease differs from diseases due to muta- mains of GFAP cause the same disease. trocytes—rests principally on the astro- tions in other cytoplasmic intermediate cyte specificity of expression of the GFAP filament , such as keratins, which The concept of GFAP toxicity gene. Multiple other cell types do express have a strong loss-of-function component That GFAP accumulation beyond a toxic GFAP, such as Schwann cells, but at much (Omary, 2009). Nevertheless, these disor- threshold is itself pathogenic came ini- lower levels (Su et al., 2004), and the only ders share highly homologous hotspots tially from studies of transgenic mice en- consistent symptoms are referable to the for mutations that interfere with filament gineered to overexpress wild-type human CNS (Brenner et al., 2009). Probably be- assembly. GFAP constitutively. Mice in the highest cause of a requirement for a threshold A rationalization of this apparent par- expressing lines died within a few weeks level of GFAP, only astrocytes have been adox has been discussed in detail by Li et after birth, and their brains contained found to display the Rosenthal fibers that al. (2002), who proposed a model in copious quantities of Rosenthal fibers are the defining feature of this disorder. which the effect of the is to slow (Messing et al., 1998). Other mouse models This is also consistent with Rosenthal fi- the rate of normal polymer formation have since been developed by homologous bers being particularly abundant in sub- rather than to block it completely. More recombination of the endogenous mouse pial and white matter CNS regions, which recent findings indicate that mutant gene (the genes are highly conserved be- are areas of high GFAP content, compared GFAP monomers form abnormally large, tween human and mouse, with 92% identity with gray matter areas such as neocortex. soluble oligomers (as discussed below in at the amino acid level) to produce the Apart from lesions that have progressed to the context of effects on the proteasome) equivalent of the common and particularly Messing et al. • Alexander Disease J. Neurosci., April 11, 2012 • 32(15):5017–5023 • 5019

is activation of stress kinase pathways such as JNK and p38, the latter promoting the transcription of ␣B-crystallin and promot- ing autophagy (Tang et al., 2006, 2008). The increased autophagy may reflect an attempt, albeit unsuccessful, to degrade the accumu- lated GFAP and associated proteins. Pro- teasome inhibition and the consequent accumulation of short-lived proteins such as the transcription factor Nrf2 (known to be elevated in several neurodegenerative diseases associated with oxidative stress) may help to explain another sign of cell stress, the widespread activation of antiox- idant genes found in the GFAP- Figure 2. Typical MRIs for Alexander disease. The two images on the left are from a juvenile-onset patient, illustrating the overexpressing and R236H point mutant characteristic abnormalities of frontal white matter including increased signal on T2-weighted images and decreased signal on mice (Hagemann et al., 2005, 2006). This T1-weightedimages(courtesyofDr.M.S.vanderKnaap,VUUniversityMedicalCenter,Amsterdam,theNetherlands).Theimage coordinated response is mediated through on the right is a T2-weighted midline sagittal section from an adult-onset patient, illustrating the characteristic atrophy and the binding of Nrf2 to a common antioxi- signal abnormalities in the (arrow) and atrophy of the cervical spinal cord [reprinted from Pareyson et al. dant response element contained within the (2008), their Fig. 1, by permission of Oxford University Press]. promoters of these genes. deleterious human R79H and R239H muta- Alexander disease show profound altera- Does GFAP toxicity cause loss of function tions (R76H and R236H in the mouse). The tions in cell shape (Fig. 4A,B) and in func- in other pathways? mice expressing the point mutant forms of tion. One dramatic effect is the induction Although we present the view that Alex- GFAP also form Rosenthal fibers and spon- of cell stress, initially prompted by the ander disease reflects a gain-of-function taneously increase their levels of GFAP, finding that Alexander disease astrocytes disorder of GFAP, there could be down- although they have normal lifespans contain massive amounts of the small heat stream effects that are more properly (Hagemann et al., 2006). However, when shock protein, ␣B-crystallin (Iwaki et al., viewed as the loss of function of other cel- certain lines are crossed together to raise 1989). Subsequently, we have found up- lular components. An important consid- GFAP levels to a far higher degree, the regulation of MAPK pathways, constitu- eration is that large accumulations of double-transgenics consistently die by 3–5 tive activation of JNK and p38 kinases, GFAP, as well as the Rosenthal fibers weeks after birth, most likely due to seizures upregulation of small heat shock protein themselves, could function as substrates (Hagemann et al., 2006). Curiously, none of genes, and an increase in autophagy (Tang to which other molecules bind. In some the mouse models made to date exhibit any et al., 2006, 2008, 2010). The mouse mod- cases, these molecules may be taken away apparent abnormalities of white matter els described above also exhibit upregula- from their normal cytosolic functions. For (Hagemann et al., 2006; Tanaka et al., 2007), tion of ␣B-crystallin (Hagemann et al., instance, several other proteins appear to though detailed analysis of the murine white 2005). The downstream effects of cell be associated with GFAP and Rosenthal matter remains to be done. Hypomyelina- stress may be many fold, but one that is fibers, including the intermediate fila- tion and/or demyelination are severe in the particularly relevant to Alexander disease ment— linker protein, (Tian early-onset form of the human disease, and and the accompanying accumulation of et al., 2006), and the 20S proteasome sub- milder or even undetectable in the adult- GFAP is impairment of proteasomal activity unit (Tang et al., 2010). Normally, plectin onset patients (Barkovich and Messing, (Tang et al., 2006, 2010; Cho and Messing, promotes a spread intermediate filament 2006). 2009; Chen et al., 2011). Accumulation of system in cells by linking filaments to Rosenthal fibers also appear in many abnormal oligomers of mutant GFAP ap- other cytoskeletal elements and to the cell non-Alexander disease conditions, such pears to be the cause of proteasomal dys- periphery (Wiche, 1998). Binding to as old gliotic scars in infarcts, multiple function. In addition, these abnormal Rosenthal fibers, which presumably con- sclerosis, trauma, and in pilocytic astrocy- oligomers inhibit the degradation of artifi- tain tightly knit filaments, would remove tomas (Chin and Goldman, 1996; Wip- cial proteasomal substrates, suggesting that plectin from its normal locations. We pold et al., 2006), and are replicated by they may also prevent other proteins from have found that overexpressing plectin in mice that simply overexpress wild-type being degraded. One prediction from these astrocytes that express an Alexander mu- human GFAP (Messing et al., 1998). This results is that protein degradation and turn- tation will convert the GFAP coils and ag- strongly suggests that high enough levels over are slowed, a hypothesis that is cur- gregates that form in these cells into a of wild-type GFAP can produce the astro- rently being tested in the mouse models. normal, spread filament system (Tian et cyte phenotype of Alexander disease. How- Stress pathways and the machinery for al., 2006). Therefore, the ratio of plectin to ever, astrocytes in cultures that express protein degradation are linked in several filaments may be critical in promoting a mutant GFAP may accumulate GFAP more ways that may contribute to the positive spread organization. Since some of the rapidly and to higher levels than astrocytes feedback loops that result in an ever- proteasome subunits in cells normally as- that express wild-type GFAP (Tang et al., increasing accumulation of GFAP. For sociate with intermediate filaments (Tang 2006, 2010). instance, inhibiting proteasome activity et al., 2010), it is not clear whether the 20S Although much attention has been fo- results in increased GFAP accumulation, binding to Rosenthal fibers would inhibit cused on Rosenthal fibers, since they are which itself presumably causes further proteasome function. Since we have the most striking pathology, astrocytes in proteasome inhibition. In addition, there found that the mutant GFAP will inhibit 5020 • J. Neurosci., April 11, 2012 • 32(15):5017–5023 Messing et al. • Alexander Disease

Figure 3. Locations of GFAP mutations in Alexander disease in relation to the protein structure. The four open rectangular boxes represent the helical coiled-coil rod domains of GFAP; these structural motifs are highly conserved among most intermediate filament proteins. The solid lines joining these segments are nonhelical linker regions, and the solid lines at either end are the nonconserved, random coil, N-terminal and C-terminal regions. The gray box just before segment 1A is a nonconserved prehelical sequence important for initiation of rod formation at the start of 1A; the gray box at the end of 2B represents a highly conserved sequence that includes the end of the coiled coil 2B segment. The wild-type amino acid is indicated next to the structure, and amino acid replacements within symbols on either side. Early-onset cases (first symptom before the age of 2 years) are on the left, shown as blue circles, and late-onset cases (first symptom after the age of 2 years) are on the right, shown as red circles. Each symbol represents a single patient, except that familial cases, including identical twins, are represented by a single symbol coded for the onset type of the proband. Adapted with permission from Brenner et al. (2009), their Figure 24.4. proteasome function (see The concept of affected by the accumulation and aggrega- the shift in solubility noted above may GFAP toxicity, above) the binding to ab- tion taking place in Alexander disease. In produce a deficiency in the soluble pool normal filaments may well participate in normal astrocytes, ␣B-crystallin exists that is critical for astrocyte function. One the proteasome inhibition. Activated JNK primarily in a soluble pool of cytosolic cannot blame the full Alexander pheno- also associates with Rosenthal fibers (Tang et proteins, whereas in cells expressing mu- type on ␣B-crystallin deficiency, however, al., 2006), perhaps binding to GFAP, as it tant GFAP it shifts its localization to a cy- since mouse knockouts and humans with has been shown to do with (He et toskeletal fraction, presumably because so nearly complete deletions of this gene al., 2002). Its association with keratin 8 may much of it is bound to the increased GFAP have predominantly skeletal muscle phe- promote the phosphorylation of the keratin, (Tang et al., 2006; Perng et al., 2008). In notypes with no evidence yet of neurolog- but also inhibit JNK-dependent phosphor- one of the mouse models of Alexander ical effects (Brady et al., 2001; Del Bigio et ylation of other substrates, such as c-Jun disease, crosses to ␣B-crystallin nulls al., 2011). Even more striking is the find- (He et al., 2002). revealed a dose-dependent increase in ing that forcing overexpression of ␣B- Some of GFAP’s normal binding part- mortality resulting from ␣B-crystallin de- crystallin above the natural induction ners, such as ␣B-crystallin, are also ficiency (Hagemann et al., 2009). Hence, caused by the disease results in complete Messing et al. • Alexander Disease J. Neurosci., April 11, 2012 • 32(15):5017–5023 • 5021

GFAP mutations than those in gray mat- ter. Glutamate uptake is not a major role for white matter astrocytes, where gluta- matergic transmission is likely consider- ably lower (Ziskin et al., 2007), and the maximal glutamate transporter current recorded from these fibrous astrocytes represents Ͻ10% of that recorded from cortical protoplasmic astrocytes (Regan et al., 2007). On the other hand, white mat- ter astrocytes provide critical metabolic support to oligodendrocytes, which are dependent upon astroglial lactate for their homeostatic maintenance, and astrocytic exchangers for interstitial ion homeostasis (Nedergaard et al., 2003). Potassium buff- ering by white matter astrocytes may be critically important, as axonal conduc- tance is linked to large-amplitude changes in interstitial ion concentrations (Ransom et al., 2000). Analysis from fibrous astro- cytes derived from the spinal white matter suggests that these cells aggressively buffer extracellular potassium, via influx through

potassium channels (Kir4.1) and concurrent active transport by the Na,K ATPase. On Figure 4. Astrocytes in Alexander disease. A, B, Enlarged, mis-shapen astrocytes in hemispheric white matter (arrows), some this basis, we speculate that an additional of which contain Rosenthal fibers (eosinophilic inclusions) in cell bodies (A). Other Rosenthal fibers lie in astrocyte processes. Multinucleated astrocytes are common (B). Scale bar, 20 ␮m. C, D, Astrocytes in the hippocampal CA1 stratum moleculare in downstream effect of GFAP toxicity may be wild-type (C) and knock-in (R236H) (D) mice. Immunofluorescence for GFAP (green) and GLT-1 (red) shows enlarged, irregular dysfunction of astrocytic potassium buffer- astrocytes in the knock-in and an appreciable loss of GLT-1 [reprinted from Tian et al. (2010), their Fig. 2A]. Scale bar, 20 ␮m. ing, which can lead to a failure in both myelin formation and maintenance. Con- sistent with a supportive, normal astrocyte rescue of the lethal phenotype observed tivity in the mouse models initially appears function for oligodendrocytes, two other when the knock-in point mutant is com- in a mosaic pattern. As the disease pro- have also recently been as- bined with the transgenic overexpressing gresses, more astrocytes lose GLT-1 immu- cribed to genetic mutations affecting astro- wild-type GFAP (Hagemann et al., 2009). noreactivity and concurrently lose their cytes. Vanishing white matter disease is One mechanism for the rescue effects of protoplasmic shape (A. Sosunov, E. Guil- caused by mutations in the eukaryotic trans- ␣B-crystallin is its ability to reduce the foyle, G. McKhann, J. E. Goldman, lation initiation factor 2B, and is character- levels of toxic GFAP oligomers to produce unpublished observations), and patch- ized by dysmorphic astrocytes with a largely monomeric population, which clamping of the abnormal astrocytes shows abnormal intermediate filament architec- can then be degraded by the proteasome a marked loss of glutamate transporter cur- ture (Bugiani et al., 2011). A phenotypically (Tang et al., 2010). Since ␣B-crystallin rent (on average, a 50% loss; X. Wu, E. Guil- similar but genotypically distinct disorder, overexpression appears to have no obvi- foyle, G. McKhann, J. E. Goldman, megalencephalic leukoencephalopathy with ous deleterious effects on astrocytes, this unpublished observations). Indeed, the loss subcortical cysts, is caused by mutations in stress protein now becomes an interesting of GLT-1 could put neurons at high risk for MLC1, which codes for a protein that is prospect as a therapeutic target. excitotoxic death, a phenomenon that has almost exclusively present in astrocytic Physiological consequences of GFAP been observed in astrocyte-hippocampal vascular endfeet (Blattner et al., 2003). toxicity—gray matter versus white matter neuron cocultures (Tian et al., 2010). A Both of these diseases are similar to Alex- Seizures are one of the most common man- diminution of GLT-1 in the hippocampus ander disease in that they are character- ifestations of Alexander disease among the of the R236H mice (Tian et al., 2010) could ized by childhood abnormalities of white type I patients. Thus, identifying abnormal- explain the increased severity of kainic acid matter. Thus, convergent evidence points ities in physiological properties of gray mat- seizures that result in pyramidal neuron to astrocytic dysfunction as a common ter astrocytes is an obvious area of interest. death (Hagemann et al., 2006) (Fig. 3C,D). antecedent to white matter loss in multi- One of the most notable changes in the Al- A loss of glutamate buffering could also play ple leukodystrophies. exander astrocyte is the diminution of a role in oligodendrocyte toxicity and loss of GLT-1, the major glutamate transporter in myelin, since oligodendrocyte precursors New model systems astrocytes (Tian et al., 2010). Whole hip- are sensitive to glutamate toxicity (Deng et To address the many important questions pocampal lysates of the R236H mice reveal al., 2006). Whether the baseline glutamate about astrocyte biology and pathology Ͼ75% reduction in GLT-1 as assayed by concentration in this tissue has actually in- posed by Alexander disease, there is a crit- immunoblots, and astrocytes in human creased is not yet known. ical need for new model systems. Re- hippocampal CA1 show variable to com- Astrocytes in white matter, where cently, we described one such model in plete loss of immunostaining (Tian et al., GFAP expression is higher than in gray Drosophila (Wang et al., 2011). When 2010). Reduction of GLT-1 immunoreac- matter, may be affected differently by wild-type or disease-linked mutant ver- 5022 • J. Neurosci., April 11, 2012 • 32(15):5017–5023 Messing et al. • Alexander Disease sions of GFAP are expressed in fly glia, the GFAP to Drosophila glial cells are conserved epilepsy, neurodegenerative diseases, or resultant animals show many phenotypic with vertebrate systems. Importantly, excel- hypoxia/ischemia. features of authentic Alexander disease. lent vertebrate cell culture and transgenic Seizures, a common feature of affected in- mouse models are available in which to test References fants, are increased in young transgenic the relevance of results derived from genetic Alexander WS (1949) Progressive fibrinoid de- flies. Neuropathologically, there is dysfunc- screens in flies. The observation of non-cell- generation of fibrillary astrocytes associated tion and some apoptotic death of glia, which autonomous neurodegeneration in the with mental retardation in a hydrocephalic in- is accompanied by non-cell-autonomous Drosophila model, as is seen in the human fant. Brain 72:373–381. degeneration of neurons. Perhaps most disease (recall that neither neurons nor oli- Barkovich AJ, Messing A (2006) Alexander dis- ease: not just a anymore. strikingly, transgenic GFAP aggregates into godendrocytes express GFAP), suggests that Neurology 66:468–469. eosinophilic, beaded, and elongated inclu- genetic definition of the molecules and Blattner R, Von Moers A, Leegwater PA, Hanefeld sion bodies in fly glia. These inclusions bear pathways involved in astrocyte–neuron FA, Van Der Knaap MS, Ko¨hler W (2003) strong similarity to authentic Rosenthal fi- communication may be possible. If so, there Clinical and genetic heterogeneity in megalen- bers at both the light and electron micro- may be broader implications of the work in cephalic leukoencephalopathy with subcortical scopic levels. Disease-associated mutant Alexander disease model flies for a variety of cysts (MLC). Neuropediatrics 34:215–218. Brady JP, Garland DL, Green DE, Tamm ER, Gib- forms of GFAP, including R79H, R239C, functions and dysfunctions of the nervous lin FJ, Wawrousek EF (2001) ␣B-crystallin R239H, L352P, and A364P, are significantly system involving cross talk between glia and in lens development and muscle integrity: a more toxic than wild-type human GFAP, neurons. gene knockout approach. Invest Ophthalmol consistent with the Drosophila model, re- Vis Sci 42:2924–2934. flecting toxicity relevant to the human Strategies for therapy Brenner M, Johnson AB, Boespflug-Tanguy O, disease. Despite the many gaps in our understand- Rodriguez D, Goldman JE, Messing A (2001) Similarities to vertebrate systems are also ing of the mechanisms and impact of as- Mutations in GFAP, encoding glial fibrillary acidic protein, are associated with Alexander apparent at the cell biological level. As in trocyte dysfunction in Alexander disease, disease. Nat Genet 27:117–120. mouse models of Alexander disease, overex- several strategies for therapy have been Brenner M, Goldman JE, Quinlan RA, Messing A pression of the fly homolog of ␣B-crystallin suggested (Messing et al., 2010). The most (2009) Alexander disease: a genetic disorder strongly ameliorates the behavioral and obvious approach is to reduce the expres- of astrocytes. In: Astrocytes in (patho)phys- neuropathological defects of GFAP trans- sion or accumulation of GFAP, so as to iology of the nervous system (Parpura V, genic flies. Additional evidence for a critical avoid the initial insult that drives the en- Haydon PG, eds), pp 591–648. New York: Springer. role for protein misfolding in toxicity is pro- tire process. One drug screen has already Buffo A, Rolando C, Ceruti S (2010) Astrocytes vided by suppression by multiple heat shock been completed using wild-type astro- in the damaged brain: molecular and cellular proteins, including HSP26, HSP27, and cytes in primary culture (Cho et al., 2010), insights into their reactive response and heal- HSP70. Quantitative analysis of levels of ag- and several of these compounds are now ing potential. Biochem Pharmacol 79:77–89. gregated GFAP supports a role for the heat being investigated using the in vivo mod- Bugiani M, Boor I, van Kollenburg B, Postma N, shock proteins acting via amelioration of els. A second approach is to target down- Polder E, van Berkel C, van Kesteren RE, Win- drem MS, Hol EM, Scheper GC, Goldman SA, abnormal protein aggregation. Other path- stream effects of GFAP toxicity, such as van der Knaap MS (2011) Defective glial ways previously implicated in Alexander the deficit in GLT-1 expression discussed maturation in vanishing white matter disease. disease or vertebrate models of the disorder, above. Ceftriaxone, an antibiotic previ- J Neuropathol Exp Neurol 70:69–82. including oxidative stress, autophagy, JNK ously identified through a drug screen for Chen YS, Lim SC, Chen MH, Quinlan RA, Perng signaling, and glial glutamate transport, also enhancers of GLT-1 expression, is already MD (2011) Alexander disease causing muta- appear to play an important role in the Dro- in clinical trials for ALS and may be a good tions in the C-terminal domain of GFAP are sophila model. The facile genetics of the fly candidate for such an approach (Roth- deleterious both to assembly and network for- mation with the potential to both activate model has allowed us to investigate the rela- stein et al., 2005). Manipulation of stress caspase 3 and decrease cell viability. Exp Cell tionships among these pathways. We find pathways such as those involving ␣B- Res 317:2252–2266. that both protein aggregration and oxida- crystallin or Nrf2 may also prove effective. Chin SS, Goldman JE (1996) Glial inclusions in tive stress act upstream of, and promote, au- The latter is under investigation for appli- CNS degenerative diseases. J Neuropathol Exp tophagy in glia expressing mutant human cation in a wide spectrum of neurological Neurol 55:499–508. GFAP. In contrast, dysregulation of glial disorders (Vargas and Johnson, 2010). Cho W, Messing A (2009) Properties of astro- cytes cultured from GFAP over-expressing glutamate transport appears to be a down- It should be noted that astrocytes up- and GFAP mutant mice. Exp Cell Res stream consequence of altered proteostasis regulate GFAP in response to nearly all in- 315:1260–1272. and oxidative metabolism, and is critical for juries and diseases of the CNS. Thus, Cho W, Brenner M, Peters N, Messing A (2010) non-cell-autonomous neurodegeneration. Alexander disease could be viewed in part as Drug screening to identify suppressors of The similarities between the vertebrate a highly exaggerated form of gliosis, or as- GFAP expression. Hum Mol Genet and invertebrate systems motivate further trocyte scarring. Indeed, Alexander astro- 19:3169–3178. Del Bigio MR, Chudley AE, Sarnat HB, Campbell use of the fly model to delineate mecha- cytes share characteristics with astrocytes in C, Goobie S, Chodirker BN, Selcen D (2011) nisms of disease pathobiology. Large-scale, gliotic responses, including increases in en- Infantile muscular dystrophy in Canadian ab- unbiased forward genetic screens have long dothelin B receptor (Hagemann et al., originals is an ␣B-crystallinopathy. Ann Neu- been a major strength of invertebrate model 2006), loss of GLT-1 (Sheldon and Robin- rol 69:866–871. systems. GFAP transgenic flies display phe- son, 2007), and in some situations even for- Deng W, Yue Q, Rosenberg PA, Volpe JJ, Jensen notypes amenable to such screens, includ- mation of Rosenthal fibers (Wippold et al., FE (2006) Oligodendrocyte excitotoxicity deter- ing increased frequency. As the 2006). Our knowledge of astrocyte dysfunc- mined by local glutamate accumulation and mitochondrial function. J Neurochem results of these genetic screens emerge, we tion in Alexander disease will likely give us 98:213–222. will be able to determine the extent to which important insights into how astrocyte func- Eng LF, Lee YL, Kwan H, Brenner M, Messing A mechanisms controlling toxicity of human tions change in such diverse pathologies as (1998) Astrocytes cultured from transgenic Messing et al. • Alexander Disease J. Neurosci., April 11, 2012 • 32(15):5017–5023 • 5023

mice carrying the added human glial fibrillary riero MR, Farina L, Ceccherini I, Savoiardo M way on glial fibrillary acidic protein (GFAP) acidic protein gene contain Rosenthal fibers. (2008) Adult-onset Alexander disease: a series accumulation in Alexander disease. J Biol Chem J Neurosci Res 53:353–360. of eleven unrelated cases with review of the liter- 281:38634–38643. Friede RL (1964) Alexander’s disease. Arch ature. Brain 131:2321–2331. Tang G, Yue Z, Talloczy Z, Hagemann T, Cho W, Neurol 11:414–422. Perng MD, Wen SF, Gibbon T, Middeldorp J, Messing A, Sulzer DL, Goldman JE (2008) Fuchs E, Cleveland DW (1998) A structural scaf- Sluijs J, Hol EM, Quinlan RA (2008) Glial Autophagy induced by Alexander disease- folding of intermediate filaments in health fibrillary acidic protein filaments can tolerate mutant GFAP accumulation is regulated by and disease. Science 279:514–519. the incorporation of assembly-compromised p38/MAPK and mTOR signaling pathways. Hagemann TL, Gaeta SA, Smith MA, Johnson GFAP-d, but with consequences for filament Hum Mol Genet 17:1540–1555. DA, Johnson JA, Messing A (2005) Gene ex- organization and ␣B-crystallin association. Tang G, Perng MD, Wilk S, Quinlan R, Goldman JE pression analysis in mice with elevated glial Mol Biol Cell 19:4521–4533. (2010) Oligomers of mutant glial fibrillary fibrillary acidic protein and Rosenthal fibers Prust M, Wang J, Morizono H, Messing A, acidic protein (GFAP) inhibit the proteasome reveals a stress response followed by glial acti- Brenner M, Gordon E, Hartka T, Sokohl A, system in Alexander disease astrocytes, and the vation and neuronal dysfunction. Hum Mol Schiffmann R, Gordish-Dressman H, Albin R, small heat shock protein, ␣B-crystallin, reverses Genet 14:2443–2458. Amartino H, Brockman K, Dinopoulos A, the inhibition. J Biol Chem 285:10527–10537. Hagemann TL, Connor JX, Messing A (2006) Al- Dotti MT, Fain D, Fernandez R, Ferreira J, Tian R, Gregor M, Wiche G, Goldman JE (2006) exander disease-associated glial fibrillary acidic Fleming J, Gill D, et al (2011) GFAP muta- Plectin regulates the organization of glial protein mutations in mice induce Rosenthal fi- tions, age of onset, and clinical sub-types in fibrillary acidic protein in Alexander disease. ber formation and a white matter stress re- Alexander disease. Neurology 77:1287–1294. Am J Pathol 168:888–897. sponse. J Neurosci 26:11162–11173. Ransom CB, Ransom BR, Sontheimer H (2000) ϩ Tian R, Wu X, Hagemann TL, Sosunov AA, Mess- Hagemann TL, Boelens WC, Wawrousek EF, Mess- Activity-dependent extracellular K accumu- ing A, McKhann GM, Goldman JE (2010) ing A (2009) Suppression of GFAP toxicity by lation in rat optic nerve: the role of glial and Alexander disease mutant GFAP compro- ␣ ϩ B-crystallin in mouse models of Alexander dis- axonal Na pumps. J Physiol 522:427–442. mises glutamate transport in astrocytes. ease. Hum Mol Genet 18:1190–1199. Regan MR, Huang YH, Kim YS, Dykes-Hoberg MI, J Neuropathol Exp Neurol 69:335–345. He T, Stepulak A, Holmstro¨m TH, Omary MB, Jin L, Watkins AM, Bergles DE, Rothstein JD van der Knaap MS, Naidu S, Breiter SN, Blaser S, Eriksson JE (2002) The intermediate fila- (2007) Variations in promoter activity reveal a Stroink H, Springer S, Begeer JC, van Coster ment protein keratin 8 is a novel cytoplasmic differential expression and physiology of gluta- R, Barth PG, Thomas NH, Valk J, Powers JM substrate for c-Jun N-terminal kinase. J Biol mate transporters by glia in the developing and (2001) Alexander disease: diagnosis with MR Chem 277:10767–10774. mature CNS. J Neurosci 27:6607–6619. imaging. Am J Neuroradiol 22:541–552. Iwaki T, Kume-Iwaki A, Liem RK, Goldman JE Rempe DA, Nedergaard M (2010) Targeting glia van der Knaap MS, Salomons GS, Li R, Franzoni E, (1989) ␣B-Crystallin is expressed in non- for treatment of neurological disease. Neuro- Gutie´rrez-Solana LG, Smit LM, Robinson R, lenticular tissues and accumulates in Alexan- therapeutics 7:335–337. Ferrie CD, Cree B, Reddy A, Thomas N, Banwell der’s disease brain. Cell 57:71–78. Rosenthal W (1898) U¨ ber eine eigenthu¨mliche, B, Barkhof F, Jakobs C, Johnson A, Messing A, Jany P, Hagemann TL, Messing A (2011) Bio- mit syringomyelie complicirte geschwulst des Brenner M (2005) Unusual variants of Alex- markers of CNS injury in mouse models of ru¨ckenmarks. Bietr Pathol Anat 23:111–143. ander disease. Ann Neurol 57:327–338. Alexander disease. Soc Neurosci Abstr Rothstein JD, Patel S, Regan MR, Haenggeli C, van der Knaap MS, Ramesh V, Schiffmann R, Bla- 37:875.20/13. Huang YH, Bergles DE, Jin L, Dykes Hoberg ser S, Kyllerman M, Gholkar A, Ellison DW, Li R, Messing A, Goldman JE, Brenner M (2002) M, Vidensky S, Chung DS, Toan SV, Bruijn LI, van der Voorn JP, van Dooren SJ, Jakobs C, GFAP mutations in Alexander disease. Int J Su ZZ, Gupta P, Fisher PB (2005) ␤-Lactam Dev Neurosci 20:259–268. antibiotics offer neuroprotection by increas- Barkhof F, Salomons GS (2006) Alexander Messing A, Head MW, Galles K, Galbreath EJ, Gold- ing glutamate transporter expression. Nature disease: ventricular garlands and abnormali- man JE, Brenner M (1998) Fatal encephalopa- 433:73–77. ties of the medulla and spinal cord. Neurology thy with astrocyte inclusions in GFAP transgenic Russo LS Jr, Aron A, Anderson PJ (1976) Alex- 66:494–498. mice. Am J Pathol 152:391–398. ander’s disease: a report and reappraisal. Neu- Vargas MR, Johnson JA (2010) in Messing A, Daniels CM, Hagemann TL (2010) rology 26:607–614. amyotrophic lateral sclerosis: role and therapeu- Strategies for treatment in Alexander disease. Sheldon AL, Robinson MB (2007) The role of tic potential of astrocytes. Neurotherapeutics Neurotherapeutics 7:507–515. glutamate transporters in neurodegenerative 7:471–481. Messing A, Li R, Naidu S, Taylor JP, Silverman L, diseases and potential opportunities for inter- Wang LQ, Colodner KJ, Feany MB (2011) Protein Flint D, van der Knaap MS, Brenner M vention. Neurochem Int 51:333–355. misfolding and oxidative stress promote glial- (2012) Archetypal and new families with Al- Stumpf E, Masson H, Duquette A, Berthelet F, mediated neurodegeneration in an Alexander exander disease and novel mutations in GFAP. McNabb J, Lortie A, Lesage J, Montplaisir J, disease model. J Neurosci 31:2868–2877. Arch Neurol 69:208–214. Brais B, Cossette P (2003) Adult Alexander Wiche G (1998) Role of plectin in Namekawa M, Takiyama Y, Aoki Y, Takayashiki disease with autosomal dominant transmis- organization and dynamics. J Cell Sci 111: N, Sakoe K, Shimazaki H, Taguchi T, Tanaka sion: a distinct entity caused by mutation in 2477–2486. Y, Nishizawa M, Saito K, Matsubara Y, Na- the glial fibrillary acid protein gene. Arch Wippold FJ 2nd, Perry A, Lennerz J (2006) Neuro- kano I (2002) Identification of GFAP gene Neurol 60:1307–1312. pathology for the neuroradiologist: Rosenthal fi- mutation in hereditary adult-onset Alexan- Su M, Hu H, Lee Y, d’Azzo A, Messing A, Brenner bers. Am J Neuroradiol 27:958–961. der’s disease. Ann Neurol 52:779–785. M (2004) Expression specificity of GFAP Yoshida T, Sasaki M, Yoshida M, Namekawa M, Nedergaard M, Ransom B, Goldman SA (2003) transgenes. Neurochem Res 29:2075–2093. Okamoto Y, Tsujino S, Sasayama H, Mizuta I, New roles for astrocytes: Redefining the func- Tanaka KF, Takebayashi H, Yamazaki Y, Ono K, Nakagawa M (2011) Nationwide survey of tional architecture of the brain. Trends Neu- Naruse M, Iwasato T, Itohara S, Kato H, Ike- Alexander disease in Japan and proposed rosci 26:523–530. naka K (2007) The murine model of Alexan- new guidelines for diagnosis. J Neurol 258: Omary MB (2009) “IF-pathies”: a broad spec- der disease: analysis of GFAP aggregate 1998–2008. trum of intermediate filament-associated dis- formation and its pathological significance. Ziskin JL, Nishiyama A, Rubio M, Fukaya M, eases. J Clin Invest 119:1756–1762. Glia 55:617–631. Bergles DE (2007) Vesicular release of gluta- Pareyson D, Fancellu R, Mariotti C, Romano S, Sal- Tang G, Xu Z, Goldman JE (2006) Synergistic ef- mate from unmyelinated axons in white mat- maggi A, Carella F, Girotti F, Gattellaro G, Car- fects of the SAPK/JNK and the proteasome path- ter. Nat Neurosci 10:321–330.