Activated in cortex of mouse models of mucopolysaccharidoses I and IIIB

Kazuhiro Ohmi*†, David S. Greenberg*†, Kavitha S. Rajavel*, Sergey Ryazantsev*, Hong Hua Li*, and Elizabeth F. Neufeld*‡§¶

*Department of Biological Chemistry and ‡Brain Research Institute, David Geffen School of Medicine, and §Molecular Biology Institute, University of California, Los Angeles, CA 90095

Contributed by Elizabeth F. Neufeld, December 20, 2002 ␣-N-Acetylglucosaminidase deficiency (mucopolysaccharidosis IIIB, contain storage vacuoles and that other cells in the brain are also MPS IIIB) and ␣-L-iduronidase deficiency (MPS I) are heritable affected. There is prominent storage in microglia (4), as well as a lysosomal storage diseases; neurodegeneration is prominent in marked increase of reactive astrocytes in the MPS IIIB mice (5). MPS IIIB and in severe cases of MPS I. We have obtained morpho- Microglia (cells of the monocyte͞macrophage lineage in the logic and molecular evidence for the involvement of microglia in brain) have been implicated in the pathogenesis of a number of brain pathology of mouse models of the two diseases. In the neurodegenerative conditions, including Alzheimer’s disease, cortex, a subset of microglia (sometimes perineuronal) consists of HIV dementia, and (6). As part of the innate cells that are probably phagocytic; they have large storage vacu- immune defense mechanism, microglia can defend the central oles, react with MOMA-2 ( against macro- nervous system against damage, but they can also produce phages) and Griffonia simplicifolia isolectin IB4, and stain intensely neurotoxic substances (6–10). Involvement of microglia in the for the lysosomal Lamp-1, Lamp-2, and cathepsin D as well pathogenesis of a lysosomal storage disease was shown in the as for GM3 ganglioside. MOMA-2-positive cells appear at 1 and 6 mouse model of Sandhoff disease (11). The appearance of months in MPS IIIB and MPS I mice, respectively, but though their activated microglia was found to precede the massive apoptosis number increases with age, they remain sparse. However, a profusion that occurs in these mice. Bone marrow transplantation, and the of cells carrying the CD68͞macrosialin antigen appear in resulting migration of healthy donor microglia to the Sandhoff the cortex of both mouse models at 1 month. mRNA encoding mouse brain, led to suppression of activated microglia and to a ͞ CD68 macrosialin also increases at that time, as shown by microarray delay of neuronal death, even though neuronal GM2 ganglioside and Northern blot analyses. Ten other transcripts elevated in both storage was not decreased. It was suggested that the damage to mouse models are associated with macrophage functions, including neurons caused by the primary defect (lysosomal storage) was complement C4, the three subunits of complement C1q, lysozyme M, exacerbated by the inflammatory response of microglia. Involve- cathepsins S and Z, cytochrome b558 small subunit, macrophage- ment of microglia has also been suggested in mouse models of specific 1, and DAP12. An increase in IFN-␥ and IFN-␥ receptor two other lysosomal storage diseases: metachromatic leukodys- was observed by . These functional increases trophy (12) and Niemann–Pick disease type C (13). may represent activation of resident microglia, an influx and activa- In view of the implications of microglial involvement for the tion of blood , or both. They show an inflammatory pathogenesis as well as the potential treatment of MPS, we have component of brain disease in the two MPS, as is known for many undertaken a study of these cells in the brain of mouse models neurodegenerative disorders. of MPS I and MPS IIIB.

ucopolysaccharidoses (MPS) type I and type IIIB are Methods Mautosomal recessive disorders caused by mutations in the Animals. The mouse model of MPS IIIB, in which the Naglu gene genes encoding ␣-L-iduronidase and ␣-N-acetylglucosamini- is disrupted in exon 6, has been described (4). The mouse model dase, respectively. Both enzymes are required for degradation of of MPS I was generated by disruption of the Idua gene by glycosaminoglycans (GAG): ␣-L-iduronidase for dermatan sul- insertion of the neomycin resistance gene on a tk (herpes simplex fate and heparan sulfate, and ␣-N-acetylglucosaminidase for virus thymidine kinase) promoter into the unique BstEII site heparan sulfate. In the absence of either of these enzymes, the (blunt-ended) in exon 6, in the opposite orientation. Both undegraded or partially degraded GAG accumulate in lyso- mutant genes were placed onto a C57BL͞6 inbred background somes, giving rise to cell and tissue pathology and to devastating by repeated back-crossing. The phenotype of our MPS I mice is clinical consequences (1). The clinical manifestations of MPS similar, though not identical, to that described earlier (2, 3), IIIB (Sanfilippo syndrome type B) are seen primarily in the perhaps because of the greater inbreeding. The animal studies central nervous system, with profound mental retardation, be- were approved by the University of California, Los Angeles, havioral disturbances, and neurodegeneration; death usually Animal Research Committee. occurs in the second decade. The manifestations of MPS I (Hurler, Hurler–Scheie, and Scheie syndromes) are more varied Tissue Preparation for Light Microscopy. Before dissecting the brain, and affect many organs, including skeleton, joints, eyes, heart, mice were deeply anesthetized with 100 mg͞kg pentobarbital, liver, and spleen in addition to the central nervous system. then perfused through the left ventricle with PBS, pH 7.4. For Profound mental retardation and death in childhood are seen frozen sections, the brains were submerged in Tissue-Tek only in the most severe form of MPS I (Hurler syndrome). The Optimal Temperature Cutting Compound (Sakura Finetek, less severely affected MPS I patients have normal intelligence Torrance, CA) for freezing and kept at Ϫ80°C before sectioning and a lifespan ranging from two decades to near normal (Hurler– in a cryostat. For vibratome sectioning, the brains were post- Scheie and Scheie). fixed in 4% formaldehyde overnight and stored at 4°C in 70% The cellular and molecular mechanisms by which lysosomal storage of GAG can lead to neurodegeneration are not under- stood. The availability of mouse models of both MPS makes it Abbreviations: MPS, mucopolysaccharidosis; GAG, glycosaminoglycan. possible to investigate such questions. Earlier studies have shown †K.O. and D.S.G. contributed equally to this work. that neurons of mouse models of MPS I (2, 3) and MPS IIIB (4) ¶To whom correspondence should be addressed. E-mail: [email protected].

1902–1907 ͉ PNAS ͉ February 18, 2003 ͉ vol. 100 ͉ no. 4 www.pnas.org͞cgi͞doi͞10.1073͞pnas.252784899 Downloaded by guest on October 1, 2021 ethanol. Frozen and vibratome sections were 7 and 40 ␮m thick, respectively.

Electron Microscopy. Mice anesthetized as above were perfused with PBS, followed by PBS containing 4% formaldehyde and 0.1% glutaraldehyde. The brain was rapidly removed, and 500 ␮m-thick slices in the coronal plane were fixed overnight at 4°C in 4% paraformaldehyde͞2% glutaraldehyde in PBS, and then for2honicein1%OsO4. The slices were then processed, embedded in Spurr resin by standard protocols, and cut to 30–40 nm. Sections were stained with uranyl acetate and lead citrate by standard procedures and analyzed in a JEM-1200EX electron microscope (JEOL) at ϫ3–6,000 magnification.

Reagents for Staining. A rat monoclonal antibody against mouse macrophages͞monocytes (MOMA-2) and a rat monoclonal an- tibody against mouse CD68 (macrosialin) were purchased from Serotec. Goat polyclonal antibodies against Lamp-1 and Lamp-2 were purchased from Santa Cruz Biotechnology. A mouse monoclonal antibody against GM3 ganglioside was purchased from Seikagaku America (Falmouth, MA). Rabbit anti-mouse IFN-␥ and rabbit anti-mouse IFN-␥ receptor chain 1 were from PBL Biomedical Laboratories (New Brunswick, NJ). These reagents were used at the concentration recommended by the Ј supplier. For secondary antibodies, biotin-labeled F(ab )2 for Ј staining and FITC and TRITC-labeled F(ab )2 for fluorescence were obtained from Jackson ImmunoResearch. Goat Fab anti- mouse IgG and goat Fab anti-mouse IgM were also obtained MEDICAL SCIENCES from Jackson ImmunoResearch. FL (bodipy)-pepstatin A, a Fig. 1. A microglial cell with large inclusions, juxtaposed to a neuron in the fluorescent probe for cathepsin D, was purchased from Molec- cortex of a 6-month-old MPS III B mouse. M and N identify the nucleus of the ular Probes. Griffonia (Bandeireae) simplicifolia isolectin IB4,a microglia and of the neuron, respectively. Note the large, nearly empty probe for ␣-galactosyl groups and a macrophage marker, was inclusions in the microglia (**) and the much smaller and denser inclusions in purchased from Sigma. the neuron (*). Arrowheads trace the membrane of the neuron. (Scale bar ϭ 5 ␮m.) Immunostaining. Sections were preincubated sequentially for 30 min with PBS, 5 min with PBS containing 0.1% H2O2, and 5 min with PBS. To minimize nonspecific staining, the sections were sized by in vitro transcription of the cDNA with the Enzo incubated for 30 min with a mixture of goat Fab anti-mouse IgG BioArray High Yield RNA transcript labeling (Enzo Diag- and anti-mouse IgM in PBS, followed by 30 min in blocking nostics), fragmented, and then hybridized to duplicate Affime- trix Murine Genome U74v2, GeneChip A. The chip contains solution of PBS containing 2% BSA for anti-GM3, or 2% BSA plus 0.1% Triton X-100 for other antibodies. The sections were probe sets for 12,000 full-length mouse genes and EST clusters. then reacted overnight with primary antibody in blocking solu- Hybridization and scanning were done by the University of Cali- tion and washed for 30 min with PBS. For visualization by fornia, Los Angeles, Microarray Core facility, with the Affimetrix microscopy, the sections were reacted for 30 min with secondary Fluidic Station 400 and Gene Array scanner. Analysis was per- formed by using the Affimetrix Microarray Suite version 4.0. Only antibody conjugated to biotin. Color was developed by the Ͼ avidin-biotin-peroxidase method (ABC Elite kit, Vector Labo- genes with a 1.7-fold increase and flagged as Present were ratories) with diaminobenzidine (DAB) or VIP (Vector Labo- considered significantly increased over the wild type. ratories) as chromogen. For visualization by fluorescence, the ␮ sections were reacted with secondary antibody conjugated to Northern Blot Analysis. Aliquots (5 g) of total RNA (obtained fluorescent dye, FITC or tetramethylrhodamine B isothiocya- from the cortex of individual mice) were used for Northern blot nate (sequentially for double staining). Staining was observed by analysis. The RNA was subjected to electrophoresis on a 1% fluorescence microcopy (DIA PHOTO, Nikon Instruments) or agarose-formaldehyde gel and blotted onto a nylon membrane. by confocal laser scanning microscopy (Leica). Some sections Templates for the synthesis of riboprobes, 250–300 bp long, were were counterstained with Hoechst dye 33342 (Molecular generated by PCR with oligonucleotide primers from the coding Probes). Staining for G. simplicifolia isolectin IB4 and for region of each gene. Reverse primers contained a T7 promoter cathepsin D was performed as per instructions of the suppliers. sequence. 32P-labeled riboprobes were prepared with Maxiscript T7 labeling kit (Ambion, Austin, TX). Radioactivity was visu- Microarray Analysis. After death, the brain was removed, and alized and quantitated with a PhophorImager (Molecular cortex was dissected. The cortexes of three ϩ͞ϩ or of three Ϫ͞Ϫ Dynamics) and normalized to ␤-actin. mice, age 3 months, were pooled. Total RNA was isolated with Trizol Reagent (Invitrogen) by the supplier’s instructions. Syn- Results thesis of cDNA and biotin-labeled cRNA, fragmentation, and Morphological Evidence of Microglia. Electron microscopic exam- hybridization were performed according to the Affimetrix Ge- ination of the cortex of an MPS IIIB mouse shows microglia that nechip Expression Analysis Technical Manual (Affimetrix, are sometimes apposed to neurons (Fig. 1). The microglial cell, Santa Clara, CA). Briefly, 20 ␮g of total RNA was used for identifiable in the electron micrograph by the shape of the cDNA synthesis using the Superscript II cDNA synthesis kit nucleus, has enormous vacuoles that appear empty or contain (Invitrogen); the cDNA was then cleaned by Phase Lock Gel some flocculent material and have the characteristic appearance Centrifugation (Eppendorf). Biotin-labeled cRNA was synthe- of lysosomes engorged with GAG. The adjoining neuron, on the

Ohmi et al. PNAS ͉ February 18, 2003 ͉ vol. 100 ͉ no. 4 ͉ 1903 Downloaded by guest on October 1, 2021 Fig. 2. Histochemical and immunohistochemical characterization of microglia in cortex. (A) G. simplicifolia isolectin IB4-positive microglia apposed to neuron; MPS IIIB, 7 months (ϫ1,750). (B) MOMA-2-positive microglia apposed to neuron; MPS IIIB, 3 months (ϫ1,150). (C and D) MOMA-2-positive cells seen at low power; MPS IIIB 3 months and MPS I, 14 months, respectively (ϫ140). (E–G) Confocal images of sections stained with antibody against Lamp-1 (green), MOMA-2 (red), and merged; MPS IIIB, 6 months (ϫ350). (H–J) Fluorescent images using antibody against ganglioside GM3 (green), MOMA-2 (red), and merged; MPS IIIB, 6 months (ϫ260). (K–M) Images of sections stained with antibody against CD68͞macrosialin, MPS IIIB, MPS I, and control, respectively, 3 months (ϫ65). (N and O) Images of sections stained with antibody against IFN-␥; MPS IIIB, and control, respectively, 3 months (ϫ250). (P and Q) Images of sections stained with antibody against IFN-␥ receptor; MPS IIIB and control, respectively, 7 months (ϫ250).

other hand, has very small vacuoles filled with denser material. The identification of the perineuronal cells as microglia was Unattached microglia have similar morphology. Similar images confirmed by staining with G. simplicifolia isolectin IB4 (Fig. 2A) were seen in the cortex of MPS I mice (data not shown). or MOMA-2 (Fig. 2B), reagents that specifically recognize cells

1904 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.252784899 Ohmi et al. Downloaded by guest on October 1, 2021 Table 1. Transcripts increased in cortex of mouse models of MPS IIIB and MPS I, determined by microarray analysis GenBank accession no. Fold* Transcripts

MPS III B mice M21050 6.0 Lysozyme M M64086 5.8 Mouse spi2 proteinase inhibitor (spi2͞eb4) X02801 4.1 Glial fibrillary acidic protein X06454 3.8 Complement C4 Fig. 3. Increase of MOMA-2-positive microglia with age. (A) Comparison of AW123191 3.5 Similar to interferon-inducible protein the number of MOMA-2-positive cells in cortex of MPS IIIB (Ϫ) and wild-type AF024637 3.3 DAP 12 control (ϩ) mice at 0.5, 1, 3, and 6 months. Each bar represents the number of AV00854 3.0 Glyceraldehyde-3-phosphate dehydrogenase MOMA-2-positive cells in a low-power field of the cortex of one mouse. (B) X68273 2.9 CD68͞macrosialin Comparison of density of MOMA-2-positive cells in a low-power field in the M22531 2.5 Complement C1q B chain cortex of MPS I and MPS IIIB mice of different ages; each bar represents one X61800 2.5 CCAAT͞enhancer binding protein (C͞EBP) ␦ mouse, with SD derived from counting cells in three low-power fields. U22033 2.3 Large multifunctional protease 7 AJ242663 2.3 Cathepsin Z precursor L20315 2.3 Macrophage-specific gene 1 of the macrophage͞monocyte lineage (14–16). MOMA-2- U74683 2.2 Cathepsin C positive microglia appeared in the cortex of MPS IIIB mice X60980 2.2 Thymidine kinase between 2 weeks and 1 month, and increased over time (Fig. 3A); AB007599 2.2 MD-1 they were not seen in the cortex of wild-type mice at any time X53929 2.2 Decorin (Fig. 3A) or of heterozygous mice up to 3 months, the only ages X00496 2.2 Mouse Ia-associated invariant chain (Ii) examined (not shown). MOMA-2-positive cells were also seen in AF010254 2.1 Complement component 1 inhibitor other parts of the brain, but in lesser number (not shown). The X58861 2.1 Complement C1q, A chain MOMA-2 cells were also present in the brain of MPS I mice, but AJ223208 2.1 Cathepsin S in lower number and with later appearance (Fig. 3B). However, AI430879 2.1 Mus musculus cDNA MEDICAL SCIENCES they remained sparse in both mouse models (Figs. 2 C and D AW124151 2.1 Mus musculus cDNA and 3B). U06119 2.0 Cathepsin H X66295 2.0 Complement C1q, C chain The MOMA-2-positive cells in both mouse models stained AV356071 2.0 Lysosomal transmembrane protein intensely for lysosomal markers. Staining by MOMA-2 coincided AF053943 2.0 Aortic carboxypeptidase-like protein ACLP with immunohistochemical staining of Lamp-1 (Fig. 2 E–G)as U29539 2.0 Retinoic acid-inducible E3 protein well as of Lamp-2 (data not shown), both lysosomal membrane X98113 1.9 -activation gene 3 proteins. The colocalization observed by confocal microscopy X57796 1.9 Tumor necrosis factor receptor superfamily, suggests that the MOMA-2 antigen and Lamp-1 are present in member 1a the same intracellular organelles. The MOMA-2-positive cells D16195 1.8 Granulin also stained with pepstatin-A, an inhibitory peptide that can be U14419 1.8 GABA-benzodiazepine receptor ␤-2 used as a probe for the soluble lysosomal enzyme cathepsin D subunit (data not shown). Such strong immunostaining of lysosomal AV242495 1.8 Mus musculus cDNA membrane and soluble proteins is in agreement with the pres- AF069051 1.8 Pituitary tumor transforming gene protein ence of the very large storage vacuoles (lysosomes) seen in the U60020 1.8 Transporter 1, ABC microglia (Fig. 1). AF013486 1.8 Type I interferon receptor soluble isoform The MOMA-2-positive cells also showed immunostaining for M31775 1.7 Cytochrome ␤-558, small subunit AI642048 1.7 Mus musculus cDNA GM3 ganglioside (Fig. 2 H–J). The colocalization was not abso- MPS I mice lute in that not all MOMA-2 cells stained for GM3, and not all M21050 4.4 Lysozyme M cells staining for G were microglia. Some neurons were also M3 X02801 4.4 Glial fibrillary acid protein stained by the antibody (not shown). R75193 3.9 Mus musculus cDNA Although MOMA-2 staining was seen in relatively few cells of L20315 3.2 Macrophage-specific gene 1 younger mice, particularly of the MPS I model, immunostaining X68273 3.2 CD68͞macrosialin of macrosialin, the murine homolog of CD68 and a marker of AF024637 3.1 DAP 12 cells of the macrophage lineage, showed a profusion of positive M22531 2.4 Complement C1q, B chain cells in mutant mice of both genotypes at 3 months (Fig. 2 K and X58861 2.3 Complement C1q, A chain L) and even as early as 3 weeks (data not shown). No CD68͞ AJ223208 2.3 Cathepsin S macrosialin-positive cells were seen in the parenchyma of the X06454 2.2 Complement C4 cortex of normal mice (Fig. 2M). However, CD68͞macrosialin- X66295 2.1 Complement C1q, C chain positive staining was seen in both normal and mutant mice in X87142 2.0 Ca͞calmodulin-dependent protein kinase cells lining the capillaries and in the corpus callosum (not II ␣ shown). AJ242663 1.9 Cathepsin Z precursor Preliminary experiments showed that immunostaining for D16195 1.8 Granulin IFN-␥ was increased in the cortex of MPS IIIB mice relative to AA815845 1.8 Mus musculus cDNA control mice (Fig. 2 N and O), as was immunostaining for IFN-␥ M31775 1.8 Cytochrome ␤-558, small subunit receptor 1 (Fig. 2 P and Q). The staining was present in microglia AJ007909 1.7 Erythroid differentiation regulator and perhaps also in neurons, but because of the close apposition (activin A) of microglia and neurons, the identification of the cells is not The transcripts shown in bold are those increased in both mouse models. definitive. *Average fold increase over normal.

Ohmi et al. PNAS ͉ February 18, 2003 ͉ vol. 100 ͉ no. 4 ͉ 1905 Downloaded by guest on October 1, 2021 Table 2. Increase in transcripts in cortex of mice at different ganelles as Lamp-1, a lysosomal membrane protein. Taken ages determined by Northern blot analysis together, the data suggest that MOMA-2 reactive cells in the MPS III B, mo. MPS I, mo. mouse models represent a subset of microglia that are phagocytic and have a strongly developed lysosomal system, and that the 13913unknown antigen of MOMA-2 is probably a lysosomal protein. Complement C1q A chain 2.3 2.0 1.9 2.5 2.2 The phagocytic nature of the MOMA-2 microglia may explain CD68͞macrosialin 1.7 2.1 2.0 2.2 2.6 the enrichment of GM3 ganglioside in these cells. Ganglioside Lysozyme M 13 15 – 3.0 9.0 GM3, and to a lesser extent ganglioside GM2, are present in excess Macrophage-specific gene 1 3.6 1.9 2.4 3.3 3.2 amount in the brain of animal models (3, 4, 21) and of humans DAP12 1.7 2.3 1.7 1.7 2.0 patients affected by MPS I and MPS III B (22, 23). Although GM3 Cathepsin S 2.2 2.2 1.8 ––ganglioside is ubiquitously made, it is both a precursor and a Glial fibrillary acid protein 2.0 2.9 3.5 ––degradation product of the more complex gangliosides (such as GM2, GM1, and the di- and tri-sialylated gangliosides) that are The numbers represent the fold-increase of the transcripts, determined by especially abundant on neuronal membranes (24). The microglia Northern blot quantitation as described in Methods. may have acquired the excess ganglioside GM3 by phagocytosis of neurons that had undergone normal apoptosis during devel- opment or of neurons that had been damaged by the disease. Molecular Evidence for Microglial Involvement. Microarray analysis Although MOMA-2-positive cells are limited in number, showed that a number of transcripts were present at a level increase slowly over the lifetime of the mouse, and are more significantly higher in the cortex of MPS IIIB and MPS I mice numerous in MPS IIIB than in MPS I mice, CD68͞macrosialin- than in the cortex of control mice (Table 1). Of the 39 transcripts positive cells appear in large number at 1 month or even earlier increased in MPS IIIB mice and 17 transcripts increased in MPS (data not shown), and in roughly equal numbers in the two I mice, 13 were increased in both mouse models (Table 1, shown mouse models. Macrosialin, the murine homolog of CD68, is in bold). One of the transcripts increased in both is glial fibrillary related to the lysosome-associated membrane acid protein, a marker of activated astrocytes, whereas another, (Lamp) family of endosomal-lysosomal proteins, and is re- granulin, is of unknown cellular origin. The others are transcripts stricted to cells of the macrophage lineage (25). The immuno- ͞ produced by, or enriched in, cells of the macrophage histochemical appearance of CD68͞macrosialin corresponds lineage. temporally with a doubling of its transcript, as seen on micro- Northern blot analyses of selected transcripts confirmed the array and Northern blots. increase observed by microarray and showed that the mRNA Coinciding with an increase in the transcripts for CD68͞ levels were significantly elevated in the cortex of MPS I and MPS macrosialin, there is an increase in both mouse models of other IIIB mice as young as 1 month of age (Table 2). The elevated transcripts associated with cells of the macrophage͞monocyte level persisted for up to 9 months for those transcripts that were lineage. Cathepsin S is preferentially expressed in such cells (26), examined in older mice. as is lysozyme M (27) and macrophage-specific gene 1 (28). Cytochrome b558 is part of the NADPH oxidase complex, which Discussion is responsible for the oxidative burst of , including Our interest in microglia in the brain of the mouse models of microglia (29). DAP12, a tyrosine protein kinase binding pro- MPS I and IIIB was prompted by the remarkable appearance of tein, was originally described in natural killer cells (30) but later perineuronal microglia in electron micrographs. Similar pictures was also found in monocytes and microglia (31). Cathepsin Z has had been shown for the brain of a canine model of MPS I (17) been found in phagosomes of (32). Complement and of a mouse model of MPS IIIA, a deficiency of heparan C1q is specifically and dramatically increased in microglia in N-sulfatase (18). The enormous and nearly empty inclusions in transient global cerebral ischemia (33). C1q and complement C4 the microglial cell, characteristic of lysosomes loaded with GAG, are produced by microglia (34) but also by other brain cells in contrast with the small inclusions in the neuron to which it is Alzheimer’s disease (35). It is noteworthy that the increase in apposed. Many microglia not apposed to neurons have similarly transcripts observed in the cortex of MPS I and MPS IIIB mice large inclusions. The basis of the difference in storage between is an early event, seen by 1 month of age. It precedes the neurons and microglia is not known, but we surmise that the behavioral abnormality in an open field test, which was observed large amount of undegraded GAG stored in microglia is not just in MPS IIIB mice at 4.5 months but not at 2.5 months (4). Finally, of endogenous origin, but is also acquired by scavenging from the the preliminary evidence for increase in IFN-␥ and its receptor environment or by phagocytosis of dead or damaged neurons. indicate an inflammatory response in the brain; IFN-␥ is pro- Identification of the perineuronal cell as a microglia was duced by microglia and astrocytes (36) as well by activated confirmed by its reactivity with G. simplicifolia isolectin IB4 and macrophages (37). with MOMA-2. The isolectin has long been used to identify Our histological and molecular results do not discriminate macrophages (14) and microglia (15), and is greatly increased on between an expansion and activation of microglia already activation of these cells. It reacts with ␣-galactoside residues on present in the brain, an influx and activation of monocytes from an unidentified membrane glycoprotein(s). MOMA-2 is a mono- blood, or a combination thereof. However, the distinction bears clonal antibody that had been selected to react with mononuclear primarily on the mechanism underlying the appearance of phagocytes (16). The original study showed that it reacted with antigens and the increase in certain transcripts. Regardless of monocytes and most tissue macrophages (although not micro- whether the mechanism involves activation of preexisting mi- glia) of normal mice, and that it was localized to the same cells croglia and up-regulation of gene expression in these cells as acid phosphatase. It was later shown to stain only a subset of and͞or an admixture and activation of blood monocytes display- macrophages (19). The only non-macrophage cells reported to ing the antigens and transcripts, the effect is the same. The react with MOMA-2 are 3T3- cells during their phagocytic cortexes of the MPS IIIB and the MPS I mice show a marked preadipocyte stage (20). In the brains of MPS IIIB and MPS I increase in functions associated with cells of the macrophage͞ mice, MOMA-2 reactivity was found on a relatively small monocyte lineage. An important question, relevant to develop- number of cells that also stained strongly for lysosomal proteins ment of therapy, is whether this increase results in an exacer- (Lamp-1, Lamp-2, and cathepsin D). Confocal microscopy in- bation of the disease process. At the very least, microglia should dicated localization of MOMA-2 in the same subcellular or- be considered a target cell in therapeutic strategies.

1906 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.252784899 Ohmi et al. Downloaded by guest on October 1, 2021 We thank Dr. Alexei Slesarev for preparing the constructs used in Institutes of Health Grants NS22376 and DK38857, a fellowship from the making the MPS I mouse model, and Hui-Zhi Zhao for excellent National MPS Society, Inc., and a grant from the Children’s Medical technical assistance. This work was supported in part by National Research Foundation.

1. Neufeld, E. F. & Muenzer, J. (2001) in The Metabolic and Molecular Bases of 20. Cousin, B., Munoz, O., Andre, M., Fontanilles, A. M., Dani, C., Cousin, J. L., Inherited Disease, eds. Scriver, C. R., Beaudet, A. L., Sly, W. S. & Valle, D. Laharrague, P., Casteilla, L. & Penicaud, L. (1999) FASEB J. 13, 305–312. (McGraw–Hill, New York), pp. 3421–3452. 21. Constantopoulos, G., Shull, R. M., Hastings, N. & Neufeld, E. F. (1985) 2. Clarke, L. A., Russell, C. S., Pownall, S., Warrington, C. L., Borowski, A., J. Neurochem. 45, 1213–1217. Dimmick, J. E., Toone, J. & Jirik, F. R. (1997) Hum. Mol. Genet. 6, 503–511. 22. Constantopoulos, G., Eiben, R. M. & Schafer, I. A. (1978) J. Neurochem. 31, 3. Russell, C., Hendson, G., Jevon, G., Matlock, T., Yu, J., Aklujkar, M., Ng, K. Y. 1215–1222. & Clarke, L. A. (1998) Clin. Genet. 53, 349–361. 23. Hara, A., Kitazawa, N. & Taketomi, T. (1984) J. Lipid Res. 25, 175–184. 4. Li, H. H., Yu, W. H., Rozengurt, N., Zhao, H. Z., Lyons, K. M., Anagnostaras, 24. Gravel, R. A., Kaback, M. M., Proia, R. L., Sandhoff, K., Suzuki, K. & Suzuki, S., Fanselow, M. S., Suzuki, K., Vanier, M. T. & Neufeld, E. F. (1999) Proc. K. (2001) in The Metabolic and Molecular Bases of Inherited Disease, eds. Natl. Acad. Sci. USA 96, 14505–14510. Scriver, C. R., Beaudet, A. L., Sly, W. S. & Valle, D. (McGraw–Hill, New York), 5. Li, H. H., Zhao, H. Z., Neufeld, E. F., Cai, Y. & Gomez-Pinilla, F. (2002) J. Neurosci. Res. 69, 30–38. pp. 3827–3876. 6. Gonzalez-Scarano, F. & Baltuch, G. (1999) Annu. Rev. Neurosci. 22, 219–240. 25. Holness, C. L., da Silva, R. P., Fawcett, J., Gordon, S. & Simmons, D. L. (1993) 7. Streit, W. J., Walter, S. A. & Pennell, N. A. (1999) Prog. Neurobiol. 57, 563–581. J. Biol. Chem. 268, 9661–9666. 8. Smits, H. A., Boven, L. A., Pereira, C. F., Verhoef, J. & Nottet, H. S. (2000) 26. Petanceska, S., Canoll, P. & Devi, L. A. (1996) J. Biol. Chem. 271, 4403–4409. Eur. J. Clin. Invest. 30, 526–535. 27. Cross, M., Mangelsdorf, I., Wedel, A. & Renkawitz, R. (1988) Proc. Natl. Acad. 9. Gebicke-Haerter, P. J. (2001) Microsc. Res. Tech. 54, 47–58. Sci. USA 85, 6232–6236. 10. Hanisch, U. K. (2002) Glia 40, 140–155. 28. Spilsbury, K., O’Mara, M. A., Wu, W. M., Rowe, P. B., Symonds, G. & 11. Wada, R., Tifft, C. J. & Proia, R. L. (2000) Proc. Natl. Acad. Sci. USA 97, Takayama, Y. (1995) Blood 85, 1620–1629. 10954–10959. 29. Klegeris, A. & McGeer, P. L. (1994) J. Neuroimmunol. 53, 83–90. 12. Hess, B., Saftig, P., Hartmann, D., Coenen, R., Lullmann-Rauch, R., Goebel, 30. Lanier, L. L., Corliss, B. C., Wu, J., Leong, C. & Phillips, J. H. (1998) Nature H. H., Evers, M., von Figura, K., D’Hooge, R., Nagels, G., et al. (1996) Proc. 391, 703–707. Natl. Acad. Sci. USA 93, 14821–14826. 31. Bakker, A. B., Wu, J., Phillips, J. H. & Lanier, L. L. (2000) Hum. Immunol. 61, 13. German, D. C., Liang, C. L., Song, T., Yazdani, U., Xie, C. & Dietschy, J. M. 18–27. (2002) Neuroscience 109, 437–450. 32. Lennon-Dumenil, A. M., Bakker, A. H., Maehr, R., Fiebiger, E., Overkleeft, Proc. Natl. Acad. Sci. USA 14. Maddox, D. E., Shibata, S. & Goldstein, I. J. (1982) H. S., Rosemblatt, M., Ploegh, H. L. & Lagaudriere-Gesbert, C. (2002) J. Exp. 79, 166–170. Med. 196, 529–540. 15. Streit, W. J. & Kreutzberg, G. W. (1987) J. Neurocytol. 16, 249–260. 33. Schafer, M. K., Schwaeble, W. J., Post, C., Salvati, P., Calabresi, M., Sim, R. B., MEDICAL SCIENCES 16. Kraal, G., Rep, M. & Janse, M. (1987) Scand. J. Immunol. 26, 653–661. Petry, F., Loos, M. & Weihe, E. (2000) J. Immunol. 164, 5446–5452. 17. Shull, R. M., Breider, M. A. & Constantopoulos, G. C. (1988) Pediatr. Res. 24, 347–352. 34. Eikelenboom, P. & Veerhuis, R. (1996) Neurobiol. Aging 17, 673–680. 18. Bhaumik, M., Muller, V. J., Rozaklis, T., Johnson, L., Dobrenis, K., Bhatta- 35. Emmerling, M. R., Watson, M. D., Raby, C. A. & Spiegel, K. (2000) Biochim. charyya, R., Wurzelmann, S., Finamore, P., Hopwood, J. J., Walkley, S. U. & Biophys. Acta 1502, 158–171. Stanley, P. (1999) Glycobiology 9, 1389–1396. 36. De Simone, R., Levi, G. & Aloisi, F. (1998) Cytokine 10, 418–422. 19. Leenen, P. J., de Bruijn, M. F., Voerman, J. S., Campbell, P. A. & van Ewijk, 37. Munder, M., Mallo, M., Eichmann, K. & Modolell, M. (1998) J. Exp. Med. 187, W. (1994) J. Immunol. Methods 174, 5–19. 2103–2108.

Ohmi et al. PNAS ͉ February 18, 2003 ͉ vol. 100 ͉ no. 4 ͉ 1907 Downloaded by guest on October 1, 2021