- Email: [email protected]
Generation of a Mouse with Low Galactocerebrosidase Activity by Gene Targeting: A New Model of Globoid Cell Leukodystrophy (Krabbe Disease)
Molecular Genetics and Metabolism 73, 211–223 (2001) doi:10.1006/mgme.2001.3194, available online at http://www.idealibrary.com on
Generation of a Mouse with Low Galactocerebrosidase Activity by Gene Targeting: A New Model of Globoid Cell Leukodystrophy (Krabbe Disease) Paola Luzi,* Mohammad A. Rafi,* Mariam Zaka,* Mark Curtis,† Marie T. Vanier,‡ and David A. Wenger* ,1 *Department of Neurology, †Department of Pathology, Jefferson Medical College, Philadelphia, Pennsylvania; and ‡INSERM U 189, Lyon-Sud Medical School and Fondation Gillet-Merieux, Lyon-Sud Hospital, Pierre Benite, France Received March 27, 2001; published online June 27, 2001
changes less severe than twi mice in the central and peripheral nervous systems, and live about 15 days longer than twi mice. They have large litters and will play a role in therapy trials using new procedures currently under development. © 2001
Globoid cell leukodystrophy (Krabbe disease) is a severe leukodystrophy caused by mutations in the galactocerebrosidase (GALC) gene leading to extremely low (less than 5% of normal activity) GALC activity. Human patients include primarily severely affected infants as well as patients with a later onset of symptoms. The infants usually die before 2 years of age, but it is difficult to predict the clinical course in older patients. In addition to these patients, additional individuals identified in this laboratory have 10 –20% of normal GALC activity measured in accessible tissues. These individuals have a wide range of clinical presentations involving neurological degeneration. On molecular analysis of the GALC gene they all have three or more mutations considered to be normal polymorphisms resulting in amino acid changes in the two copies of the GALC gene. In order to investigate the role these amino acid changes may play on clinical, biochemical, and pathological findings, a new transgenic mouse was generated by homologous recombination. After preliminary studies determined what effect each amino acid change had on mouse GALC activity in transient transfection experiments, mice containing a cysteine residue at codon 168 instead of histidine (H168C) were produced. These mice developed symptoms, but they were delayed by 10 –15 days from the well-characterized twitcher (twi) mouse. They accumulated psychosine slightly slower than twi mice, showed pathological
Academic Press
Key Words: Krabbe disease; globoid cell leukodystrophy; white matter disease; transgenic mouse; galactocerebrosidase; myelin.
Globoid cell leukodystrophy (GLD) or Krabbe disease is an autosomal recessively inherited disorder resulting from very low galactocerebrosidase (GALC) activity (see 1,2, for review). The GALC cDNA was first cloned in 1993 (3) and subsequently the gene organization was determined (4 – 6). Most human patients present with symptoms before 6 months of age and die by 2 years of age. Older patients, including adults, are also diagnosed with GLD. The clinical course in older patients is highly variable even among siblings; however, many of the juvenile and adult onset patients present with weakness, tremor, and vision loss followed by varying degrees of intellectual impairment (7–11). The diagnosis is made by measuring very low GALC activity in available tissue samples, such as leukocytes and cultured skin fibroblasts; however, the residual GALC activity measured does not predict the clinical course a patient will take. While many patients with GLD have at least one copy of the common 30-kb deletion first identified in this laboratory (12,13), over 60 additional disease-causing and poly-
1 To whom correspondence should be addressed at Jefferson Medical College, 1020 Locust St., Room 394, Philadelphia, PA 19107. Fax: (215) 955-9554. E-mail: [email protected].
211 1096-7192/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.
212
LUZI ET AL.
morphic mutations have been identified (reviewed in 2,14). Within certain ethnic groups one common mutation has been found (15). One mutation, G809A (G270D), is found in a relatively high frequency in late-onset GLD patients. Carrier identification by measuring GALC activity is difficult due to the wide range of “normal” and carrier values. This is primarily due to polymorphisms in the coding regions of the gene that result in amino acid changes in the two processed subunits of GALC. Expression studies have demonstrated that the presence of these amino acids lowers the measured GALC activity from that measured with the most common amino acids (14). In addition to the human patients with GLD, several animal models, including twitcher (twi) mice, Cairn and Westhighland White terriers, and rhesus monkeys, have been identified and characterized (reviewed in 2,16,17). All of these models have very low GALC activity, and other biochemical and pathological findings are similar to those of human patients. Their respective GALC genes have been cloned and the mutations causing the disease have been identified (17). The twi mouse model has been used for a large number of studies since its identification in 1980 (18). These mice appear to be normal until 20 days of age when they start to have tremors. Weight gain slows, and they continue to deteriorate with most dying by 40 days of age. Twi mice given hematopoietic stem cell transplantation (HSCT) by 10 days of age can live until about 100 days (19 –22). During the enzyme-based screening of human patients for neurodegenerative disorders, GALC activity in leukocytes is often measured. Although true GLD patients have very low (⬍5% of our normal mean) GALC activity and are easy to identify, there is a wide “normal” range. However, a significant number of individuals (more than 50) were found to have GALC activity between 10 and 20% of our normal mean. Most of these people had clinical and other evidence for white matter disease suggestive of a leukodystrophy, but were not considered to have GLD. Subsequent mutation analysis for the most common disease-causing mutations and polymorphisms in the GALC gene revealed that almost all of these individuals have three or more nucleotide changes considered to be polymorphisms. No other changes were found upon sequencing all of the coding regions and exon-intron boundaries of the GALC gene in four patients. These changes are considered to be normal polymorphisms since homozygosity for one of them does not appear to have clinical implications. The clinical spectrum of these individuals is
TABLE 1 Polymorphisms in the Human GALC Gene and the Generated Amino Acid Changes in Mouse GALC Human polymorphisms
Mouse
Nucleotide change a
Amino acid change b
Allele frequency c
Amino acid change
C502T T1637C C502T/T1637C e
R168C I546T R168C/I546T
4–5% 30–40% ⬍2%
H168C V545T d H168C/V545T
a cDNA nucleotide number counting from the A of the initiation codon. b Codon number. c Frequency of the minor allele in the general population. d Due to the loss of codon 508 in the mouse GALC gene, human and mouse differ in nucleotide number at codon 546/545. e Present on the same allele.
broad so it was considered they might have white matter disease due to the low GALC activity if another genetic or environmental factor were present. In order to gain experimental evidence for the hypothesis that low (10 –20% of normal), but not deficient, GALC activity could cause white matter disease under certain conditions, we generated a mouse model by homologous recombination that contained a polymorphic change found in humans. This resulted in the generation of a mouse that had low GALC activity and symptoms delayed from those observed in the twi mouse model. Clinical, biochemical, and pathological studies on the homozygous mice are presented. While we had hoped to generate a mouse without symptoms or only very mild symptoms until stressed by environmental factors, we obtained a mouse with a longer life span than twi mice that will find use in treatment studies under development. In addition, these mice have large litters resulting in more affected mice for study. MATERIALS AND METHODS Expression of Mutations in COS-1 Cells Two common polymorphisms have been found in the Caucasian population alone and together on one allele (Table 1). Expression of GALC activity was measured using human GALC cDNA with and without the polymorphic changes following cloning into pcDNA3 as previously described (12). Mouse GALC cDNA was also cloned into the pcDNA3 vector. In order to change the amino acids at codon 168 from his to cys and codon 545 from val
TRANSGENIC MOUSE MODEL OF GLD
to thr, two nucleotide changes at cDNA positions 502 and 503 (CA to TG) and at positions 1633 and 1634 (GT to AC), respectively, were introduced using PCR-mediated mutagenesis. Constructs containing these changes alone or together were prepared. The constructs (normal and mutated) were transiently transfected into COS-1 cells using Lipofectamine (Gibco, Rockville, MD), and 72 h later the cells were harvested and GALC activity was measured. Measurement of GALC Activity Cultured cells were harvested with trypsinEDTA, washed with phosphate-buffered saline, and homogenized in deionized water. The total homogenate was used for enzyme analysis. Brain, liver, spleen, kidney, and heart from normal, heterozygous, and homozygous mice were removed immediately after death and homogenized in 10 vol of 0.05 M phosphate buffer (pH 6.5). Aliquots were diluted 1–10 with deionized water for determination of the protein concentration using the method of Lowry and for measurement of GALC activity using [ 3H]galactose-labeled galactosylceramide as previously described (23). Targeting Vector Construction Based on the results of the expression studies, cys at codon 168 was introduced into the mouse GALC gene. Genomic clones from a 129SVJ mouse genomic library (Stratagene, La Jolla, CA) were isolated using a GALC cDNA mouse probe. These clones were used, together with the amplification products obtained by XL-PCR (Perkin Elmer, Foster City, CA), to determine the genomic organization of the mouse GALC gene. Intron 4, spanning 2.0 kb, was completely sequenced, intron 5, spanning 5.6 kb, and intron 3, spanning 2.3 kb, were partially sequenced, and restriction analysis was performed. A 2.8-kb genomic fragment was obtained by XL-PCR using a sense primer located in intron 4 and an antisense primer located in intron 5. This fragment spanned 1200 bp of intron 4, exon 5 (about 150 bp), and 1400 bp of intron 5. The fragment was cloned into a modified pBluescript plasmid (Stratagene) in which the EcoRV and KpnI sites were removed. The 2 nt changes at cDNA positions 502 and 503 (exon 5) were introduced by directed mutagenesis using sequential PCR steps. The mutated fragment was cloned into TA vector (Invitrogen, Carlsbad, CA) and sequenced. It was then excised from the TA construct by digestion with NcoI and KpnI enzymes and
213
cloned into the modified pBluescript to replace the homologous fragment without the mutation. This fragment containing the desired mutation was cut out from pBluescript by EcoRV (restriction site present in intron 4) and BamHI (from pBluescript cloning site) digestion. The 2.0-kb band obtained was purified, BamHI linkers were added and it was ligated into the BamHI site of the pNTK plasmid. A second 4kb-genomic fragment, spanning 2.3 kb of intron 3, exon 4, and 1.6 kb of intron 4 (up to the EcoRV site) was obtained by XL-PCR. This fragment was cloned into the ClaI site of the pNTK vector. Using this cloning strategy the neomycin phosphotransferase (neo) cassette was located in the middle of intron 4 in the opposite orientation to the GALC gene. The thymidine kinase (TK) gene was located downstream and in the opposite orientation with respect to the GALC gene. The resulting targeting vector, designated pNTK-GALC contains a 5⬘ arm of 4 kb and a 3⬘ arm of 2 kb homologous with the endogenous GALC gene (Fig. 1). Electroporation and ES Cells Selection D3 embryonic stem (ES) cells (provided by Dr. Jaspal Khillan) were grown on ␥-irradiated G418resistant feeder cells on gelatinized plates following standard methods. Confluent 10-cm plates (about 2.5 ⫻ 10 7 cells) were trypsinized and resuspended in 0.9 ml of PBS. Forty micrograms of the pNTK-GALC construct was linearized by NotI digestion, precipitated, resuspended in TE buffer, and incubated with the cells for 5 min at room temperature in a Bio-Rad (Hercules, CA) electroporation cuvette. Cells were electroporated with a Bio-Rad Gene Pulser at 200 V and 500 F. After 10 min the ES cells were plated in six 10-cm plates containing feeder layers. G418 (Gibco) was added to the medium 24 h after electroporation at a concentration of 200 g/ml. Five of the six plates were also treated with 0.2 M 2⬘-fluoro-2⬘-deoxy-5-iodouracil--D-arabinofuranoside (FIAU) (Moravek, Brea, CA). Media were changed every 24 h. After 6 to 9 days single colonies were picked under a dissecting microscope with a Pasteur pipette. Identification of Recombinant Clones The colonies were grown in a 24-well plate. Once they became confluent half of the cells were frozen and half were used to extract DNA. The DNA was then analyzed by XL-PCR to identify the positive colonies using a sense primer in the neo gene (5⬘-
214
LUZI ET AL.
FIG. 1. Schematic representation of the point mutations introduced in the GALC gene by homologous recombination. (A) Restriction map of the wild-type GALC gene in the region homologous to the targeting vector. (B) Restriction map of the targeting construct linearized by NotI digestion. The asterisk in exon 5 indicates the location of the two point mutations (CA to TG) introduced at position 502 and 503. A neo cassette was inserted in intron 4 and a TK cassette was added 3⬘ of the region of homology. (C) Predicted structure of the mutated allele after homologous recombination. The restriction pattern obtained after BamHI digestion is shown for the wild type, targeting vector, and product of homologous recombination. Southern blot of BamHI-digested DNA from positive colonies was used to identify homologous recombination events.
GAGGCCACTTGTGTAGCGCC-3⬘) and an antisense primer 3⬘ to the region of homology contained in the targeting vector (5⬘-CACTGCAAACTCCCTGAGCC-3⬘). PCR products were resolved in a 0.8% agarose gel and those showing the expected 2.0-kb band were considered positive. The DNA from these clones was analyzed by another XL-PCR to confirm that homologous recombination occurred using a sense primer located 5⬘ of the region of homology contained in the targeting construct (5⬘-CCCGGCTTCTAG TAAATTACCC-3⬘) and an antisense primer located in the neo gene (5⬘-GCAGCCTCTGTTCCACATACAC-3⬘). A 4-kb band indicated a positive clone. Southern blotting was performed to confirm the presence of the correct construct in the positive clones. Ten micrograms of genomic DNA from the positive clones was digested overnight with BamHI, and the digest was run on a 0.7% agarose gel, trans-
ferred to a Hybond-N ⫹ (Amersham, Piscataway, NY) membrane, and hybridized with a 32P-labeled random primed probe using a 1.5-kb GALC gene fragment (spanning exon 5 and part of intron 5). The presence of the mutation in positive clones was confirmed by sequencing. Karyotyping was also done to verify the euploid condition of this clone. Production of Chimeric Mice The positive clone was expanded to prepare chimeric animals. The FVB/N females were superovulated and crossed with FVB/N males. Morulas were isolated after 2.5 days of mating. Chimeric animals were prepared by a one-step aggregation method developed by Khillan and Bao (24). The ES cells were transferred to a petri dish with microwells, and about 15–20 cells settled down on the bottom. The morulas without the zona pellucida were trans-
215
TRANSGENIC MOUSE MODEL OF GLD
ferred on top of the ES cells and were allowed to develop into blastocysts overnight. The following day, blastocysts were washed in medium and transferred to 2.5-day pseudopregnant females. The pups were born after 17 days. Chimeric animals were mated to wild-type animals. Heterozygous animals were identified and mated to produce mice homozygous for the desired mutation.
corresponding to 2–2.5 mg of protein using a HPLC procedure. The extraction procedure and general conditions have been previously described (25,26). The analysis of the orthophtalaldehyde derivatives was conducted on a 20-cm Spherisorb 5-m ODS2 C18 column with a 5-cm Spherisorb ODS2 C18 guard column. The mobile phase was methanol–5 mM sodium phosphate buffer, pH 7.0 (89/11) with 50 mg/L sodium octylsulfate as an ion-pairing agent.
Northern Blot and RT-PCR Liver, brain, spleen, kidney, and heart from normal, heterozygous, and homozygous animals were removed immediately after death and frozen in liquid nitrogen. Total RNA was extracted following the RNeasy Midi protocol (Qiagen) for animal tissue. Twenty micrograms of RNA was subjected to electrophoresis on a 1% agarose-formaldehyde gel and transferred to a nylon membrane. Northern blots were hybridized with a labeled probe prepared using a 700-bp mouse GALC cDNA fragment, at 68°C in ExpressHyb Hybridization solution (Clontech, Palo Alto, CA). The Northern blots were then washed in 0.1⫻ SSC– 0.1% SDS at 50°C and exposed over several days. RT-PCR was performed following the OneStep RT-PCR protocol from Qiagen using 1 g of RNA, and the PCR products were separated by electrophoresis on a 1% agarose gel. Clinical Evaluation of the Homozygous Mice Mice were genotyped between 2 and 4 days of age. DNA was isolated from clipped toes using the QIAamp tissue kit (Qiagen). The mutated allele was identified using the XL-PCR method used to detect the positive colonies. The nonmutated allele was detected by PCR using a sense primer located 5⬘ of the neo gene (5⬘-CTGATCTTGGTTGCCTGACC-3⬘) and the same antisense primer, located 3⬘ of the neo gene, used above in the PCR amplification for the mutated allele. Mice were examined every day for tremor and they were weighed every other day beginning at Day 10. When it was noted that the mice had weakness in the hind limbs and were having trouble reaching and chewing the standard mouse chow pellets, powdered food (Pico Vac, Test Diet, Richmond, IN) and water were placed on the floor of the cages. Measurement of Psychosine Levels Brain psychosine (galactosylsphingosine) was determined on aliquots of 20% tissue homogenates
Pathology For histological analysis tissues were fixed in 10% buffered formalin and embedded in paraffin. After slicing the sections were stained with hematoxylin and eosin, luxol fast blue/periodic acid Schiff (PAS), and Bielchowsky silver stain using standard procedures and examined by light microscopy. For ultrastructural analysis the tissues were fixed in 2% buffered (0.1 M Sorensen phosphate buffer, pH 7.4) glutaraldehyde and embedded in standard low viscosity embedding medium (10 g vinyl cyclohexane– dioxide, 6 g diglycidyl ether polypropylene glycol, 26 g nonenyl succinic anhydride, and 0.6 g dimethylaminoethanol). One-micrometer sections were cut and stained with toluidine blue. One hundred-nanometer-thick sections were cut on a Reichert Ultracut S microtome. The 100-nm sections were stained with uranyl acetate and lead citrate. Ultrastructural evaluations were performed on a Jeol 100 CX electron microscope. RESULTS Expression of Mutations in COS-1 Cells Several polymorphic changes have been found in human GALC (2); however, C502T (R168C) and T1639C (I546T) were encountered with high frequency in our patient population (Table 1). Expression of these polymorphisms alone and together in human GALC cDNA in COS-1 cells was measured after the desired nucleotide change in pcDNA3 was introduced as previously described (14). Before we tried to create a mouse model with low but not totally deficient GALC activity the same amino acid changes were made in mouse GALC, and the effects they had on GALC activity in vitro were determined. To produce a cysteine at position 168 the CAT codon for the unconserved histidine in mouse was changed to TGT to code for cysteine by PCR-mediated mutagenesis. A threonine was created at mouse GALC codon position 545 (one less codon than human be-
216
LUZI ET AL.
FIG. 2. Expression of GALC activity in COS-1 cells transfected with human or mouse GALC cDNA containing the most common sequence and polymorphic changes. Cells were transfected with pcDNA3 containing the constructs prepared as described under Materials and Methods, harvested 3 days later, and assayed for GALC activity.
cause of the loss of three nucleotides at codon 508) by replacing GTG sequence for valine to ACG for threonine. Both amino acid changes were made in one construct to mimic the human allele with both changes. Expression of these amino acid changes in human and mouse GALC was compared (Fig. 2). The creation of a cysteine residue at codon 168 in human GALC decreased the activity only 20%, while creating a cysteine residue in that position in mouse GALC resulted in a decrease of 80 –90% of normal activity. In contrast to the expression found in the human GALC, the decrease in activity due to the 168 cys change (H168C) was more significant than that obtained with the substitution of a threonine at position 545 (V545T) and with both changes in the same allele (H168C ⫹ V545T) (Fig. 2). Therefore, to generate a mouse model with low (10 –20% of normal) GALC activity we decided to produce a mouse with a cysteine residue at codon 168 by homologous recombination. Preparation of the Targeting Construct and Identification of Recombinant Clones The targeting construct was prepared by cloning genomic fragments of the mouse GALC gene, obtained by XL-PCR, into the pNTK vector. A 4.0-kb fragment containing intron 3, exon 4, and part of
intron 4 was cloned 5⬘ of the neo gene. A second 2.0-kb fragment composed of the rest of intron 4, exon 5, and part of intron 5 was cloned 3⬘ of the neo gene and 5⬘ of the TK gene. The desired two point mutations at cDNA positions 502 and 503 in intron 5 were introduced before cloning the fragment into the pNTK vector, using PCR-based methodology. After electroporation of the linearized targeting vector into ES cells, the transformants were selected with G418 or with G418 ⫹ FIAU. A total of 96 colonies was picked: 34 from a plate subjected to single selection and 62 from plates that underwent double selection. Homologous recombinant clones were detected by XL-PCR using primers in the neo gene coupled with primers located outside of the region of homology contained in the targeting construct. Three positive clones were identified: one was a colony selected with G418 only and the other two were obtained by double selection. PCR-positive clones were confirmed by Southern blot analysis of genomic DNA digested with BamHI and hybridized to a probe spanning exon 5 and part of intron 5. Two of the three positive clones identified by PCR showed the expected restriction pattern: a 6.1-kb band corresponding to the normal allele as well as the predicted 3.7-kb band created following homologous recombination (data not shown). Direct sequencing of
TRANSGENIC MOUSE MODEL OF GLD
217
FIG. 3. GALC activity in organs from noncarrier (NC), carrier (C), and affected (A) mice for H168C. Brain, liver, spleen, heart, and kidney were homogenized, and GALC was measured as described under Materials and Methods.
the PCR fragments derived from the two positive clones identified by Southern blotting and PCR was performed to verify the presence of the expected mutations at positions 502 and 503. The two desired mutations were present only in one of the clones, with the other showing the wild-type sequence. This clone was one obtained by double selection. Characterization of Mice with H168C Mutation The ES cells positively targeted were expanded and used to prepare chimeric animals. Overall 97 embryos were transferred to 9 foster mothers. Only 4 of them became pregnant and a total of 9 pups were born 4 out of which were chimeras. Two of the chimeras died soon after birth and the other two (both males) were mated with wild-type females to produce positive progeny. These males produced seven litters with a total of 36 pups, all of which were positives for ES cell recombination. The het-
erozygous mice were intercrossed to obtain mice homozygous for the desired change. GALC activity was measured in brain, liver, spleen, heart, and kidney samples from noncarrier, heterozygotes, and homozygous affected mice (Fig. 3). Heterozygous mice show GALC activities that are about 50% of normal, while homozygous mice show low values. It is not possible to state that the low GALC activities measured in the tissues from the transgenic mice are more than that measured in twi mice. For the first 25–30 days the homozygous mice feed normally and grow like heterozygote and homozygous normal littermates. However at about 25–30 days of age the homozygous mice start to show signs of tremor and hind leg weakness. This compares to about 20 days in twi mice. The tremor is most obvious when the mice are picked up by their tails. This strain of mice tends to be larger than twi mice, but eventually they start to lose weight. The mutant
218
LUZI ET AL.
(Fig. 6). This slower rate of increase could explain the protracted course in the transgenic mice. Pathological Studies in the Transgenic Mice
FIG. 4. Total body weights of normal mice of the FVB/N strain, affected transgenic mice, and affected twi mice at different ages. The number of mice at each time point varies from 1 to 10. If more than three, the weight is expressed as mean ⫾ standard deviation.
mice gain weight until around 35– 40 days of age, they then maintain their weight for few days, and then they begin to loose weight and generally die 10 –15 days later (Fig. 4). At the terminal stage the mutant animals generally become paralyzed in the hind legs having a different appearance from twi mice. Severe wasting of the hind limbs is clearly noted. This is seen in Fig. 5 where we compare a 50-day-old terminal transgenic mouse (right panel) to a 38-day-old terminal twitcher mouse (brown mouse in left panel). Skeletal muscle of both the transgenic and twi mice show atrophy consistent with a neurogenic origin. While the average life span of these animals is about 50 days, there is a high degree of variability. Some animals lived up to 60 – 63 days while others survived only for 40 – 42 days. Psychosine Values in Brain As psychosine has been implicated in the pathogenesis of oligodendrocytes and its concentration increases in brain with age in humans and animals with GLD, the psychosine concentration was measured in brains of these mice. As with the twi mouse model psychosine increases with age. However, the rate of increase in affected transgenic mice appears to be slightly slower than that observed in twi mice
In these pathological studies a transgenic mouse of 49 days is compared to a twitcher mouse of 36 days. Demyelination with infiltration of macrophages was identified in the cerebellar white matter, brainstem, cerebral white matter, and cerebral gray matter of a 36-day-old twitcher mouse as previously described by Suzuki and Suzuki (27). Luxol fast blue/PAS staining of cerebellar white matter from a twi mouse is shown in Fig. 7C. In contrast to the twitcher mouse, the 49-day-old transgenic mouse showed less macrophage infiltration in both white matter (Fig. 7B) and gray matter compared to the twi mouse (not shown). As expected, no PAS-positive macrophages were identified in the brains of control mice (Fig. 7A). Luxol fast/PAS-stained peripheral nerve from the control mouse showed well myelinated intact nerve (Fig. 8A). There was a marked loss of myelin and abundant infiltration of macrophages in the peripheral nerve from the 36-day-old twitcher mouse (Fig. 8C). The peripheral nerve from the 49-day-old transgenic mouse also showed a loss of myelin; however, the macrophage infiltration was less than that observed in the 36-day-old twitcher (Fig. 8B). Thick sections stained with toluidine blue show normal appearing myelinated axons in the control mouse (Fig. 8D). A moderate loss of myelin is observed in the peripheral nerve of the 49-day transgenic mouse (Fig. 8F). There were some scattered macrophages. In contrast, only rare myelinated axons remain in the peripheral nerve from the 36-dayold twitcher mouse and numerous endoneurial lipid laden macrophages are present (Fig. 8E). Ultrastructural analysis of the peripheral nerve showed similar changes in the mice with low GALC activity with good preservation of normal appearing myelin in the control mouse (Fig. 8G). Myelin loss and macrophage phagocytosis of myelin debris are identified in the peripheral nerve from both the 36day-old twitcher mouse (Fig. 8I) and the 49-day-old transgenic mouse (Fig. 8H). The loss of myelin and macrophage infiltration is greater in the twitcher mouse compared to the transgenic mouse. The ultrastructural appearance of the intact myelin that remains in the transgenic mouse is normal.
TRANSGENIC MOUSE MODEL OF GLD
FIG. 5. Photographs of a 53-day-old transgenic mouse (white) and 38-day-old twi mouse (left panel) and a 50-day-old transgenic mouse at the terminal stage (right panel). Note the pronounced wasting of the hind limbs in the 50-day-old transgenic mouse compared to the severely affected twi mouse and the difference in appearance between the two transgenic mice of similar ages.
Characterization of GALC mRNA in the Transgenic Mouse Because the mice were more severely affected than expected from the insertion of a missense mutation in the 50-kDa subunit, the mRNA produced was examined. To verify the production of GALC message carrying the desired mutation, RT-PCR and Northern blot analysis were performed using RNA extracted from different organs of the mutant mice. RNA from brain, liver, and kidney of noncarrier and homozygous animals was amplified using primers spanning the GALC cDNA region from nucleotides 150 to 600. Two main products were obtained in the affected mice: one of the expected size of 450 bp and the other of about 900 bp that was especially represented in the brain (Fig. 9). Sequencing of the two bands revealed that the 450 bp is derived from the correct message containing the 2 bp changes and the 900 bp is produced by an incorrect splicing of intron 4. In the homologous recombinant mice the GALC gene contains the neo gene in intron 4, located about 400 bp from the 3⬘ end of the intron. The incorrect splicing may be due to the presence of this extra piece of DNA. Even with some incorrect splicing, a significant amount of correctly spliced GALC containing the 2 nt change should be produced. Northern analysis revealed that level and size of the GALC mRNA in mutant animals are similar to wild type (data not shown). No larger GALC mRNA in brain was detected.
219
present. While the lower than normal GALC activity can be explained by the presence of multiple copies of normal polymorphisms, it is difficult to prove that the variable clinical features observed in these individuals are due to the lower than normal GALC activity. Other factors may be involved. If some agent or event resulted in focal or generalized demyelination, attempts to repair this damage would require the activation of oligodendrocytes in the central nervous system and Schwann cells in the peripheral nervous system to initiate remyelination. This would involve the synthesis of all proteins required for production of stable myelin including integral membrane proteins and enzymes required for the synthesis of myelin components such as galactosphingolipids. This remodeling of damaged myelin may require production of precise amounts of certain components, such as key lipids, and this could require adequate amounts of both biosynthetic and degradative enzymes. While partially deficient GALC activity may not have serious consequences in the initial synthesis of myelin to be wrapped around naked axons in an infant, it could have a clinical effect during remyelination when less than optimum conditions are present. While the H168C replacement created a mouse with low GALC activity, it presented with symptoms earlier than expected. The problem of generating an exact model of a human disease by replacing an amino acid in a protein from a mouse or other species is difficult. Clearly transfection studies are needed as an initial experiment to determine if a proposed change will have an effect on the enzy-
DISCUSSION The original reason for creating a mouse with low (10 –20% of normal) GALC activity was to determine if such an activity could cause white matter disease if certain environmental and/or genetic factors were
FIG. 6. Psychosine concentrations in brains from transgenic and twi mice of different ages. In normal mouse brain psychosine is almost undetectable at 15 days of age and it increases slowly to about 20 pmol/mg protein by 40 days of age.
220
LUZI ET AL.
FIG. 7. Luxol fast blue/PAS staining of cerebellar white matter from a 50-day-old control mouse (A), a 49-day-old transgenic mouse (B), and a 36-day-old twitcher mouse (C). (640⫻, original magnification).
matic activity. Preliminary transient expression studies replacing the more common amino acids in mouse GALC with the polymorphic amino acids found in human GALC lead to the first unexpected results. While in humans R168C had little effect,
H168C in mouse GALC had a pronounced effect on activity (Fig. 2). V545T had less of an effect on mouse GALC activity than H168C. This relatively common polymorphic change (I546T) in humans is the major cause for the wide range of GALC values
FIG. 8. Light and electron microscopy of peripheral nerves from a control mouse (A, D, G), 49-day-old transgenic mouse (B, E, H), and 36-day-old twitcher mouse (C, F, I). Panels A, B, and C show Luxol fast blue staining (640⫻, original magnification). Toluidine blue staining is shown in panels D, E, and F (640⫻, original magnification). Electron microscopy of peripheral nerves is shown in panels G, H, and I (7250⫻, original magnification).
TRANSGENIC MOUSE MODEL OF GLD
FIG. 9. RT-PCR of RNA from brain, liver, and kidney of a noncarrier (N) and two homozygous-affected transgenic mice (A1 and A2). The primers used span the GALC cDNA from nucleotides 150 to 600.
measured in diagnostic samples. Making both amino acid changes in the same construct had a pronounced effect on human GALC activity but less than either change alone in the mouse. Based on these findings it was decided to make a mouse with H168C. These studies show the difficulty in trying to make a mouse model of a human disease by generating an amino acid change found in human patients and assuming the effect will be the same. As the mouse GALC protein is only 84% identical to the human protein (29), the effects of changing one or more amino acid have unpredictable effects on activity and stability. While this study did not answer the initial question, it does provide us with an additional model for studying the pathophysiology and treatment of GLD. Generation of the model with a cysteine residue at codon 168 by homologous recombination was successful. Examination of organs from carriers and homozygous-affected mice showed the expected levels of GALC activity (Fig. 3). While we expected the affected mice to have some residual GALC activity, we could not conclude that it was higher than that found in twi mice that have a nonsense mutation at codon 339 (29). It may not be possible to accurately determine 10 –15% of normal activity in homogenates of organs. We did not take advantage of the Cre/loxP method for creating point mutations; however, we examined organs for the expression of normal size mRNA. Brain, liver, and kidney from affected mice were examined by RT-PCR and Northern blotting. While products of the expected size were found, some larger size amplification products were seen, especially in brain, apparently from aberrant splicing. Therefore, less than the predicted amounts of the GALC activity could have been produced. However, since previous studies (26) have demonstrated that low, even undetectable, levels of GALC activity have an ability to slow the course of
221
the disease in twi mice, it was predicted that even 5% of normal activity would be enough to delay or prevent symptoms. It appears that the small amount of activity present is enough to slow the course of the disease only slightly. The transgenic mice appear to be normal until approximately 25–30 days of age when fine tremor is noted. They continue to gain weight until about 40 days, when most twi mice have died, and then live an additional 10 –20 days. As shown in Fig. 5 there are some phenotypic differences between the transgenic mice and twi mice, as well as the variability between transgenic mice. The 50-day-old transgenic mouse shown on the right panel is more affected than the White 53-dayold transgenic mouse shown on the left panel. However, soon after this photo was taken the White transgenic mouse started to have more tremors and rapidly deteriorated and died at 60 days of age. What causes the significant clinical differences between these mice is not known at this time. Overall these mice appear to have a milder form of GLD than twi mice. An additional advantage of these mice is that they have large litters and are easy to handle. They accumulate psychosine at a slightly slower rate than twi mice, although they eventually reach levels similar to those of twi mice. Pathological studies in the central and peripheral nervous systems comparing a 36-day-old twi mouse with a 49-day-old transgenic mouse also confirm that the disease is progressing at a slower rate in the transgenic mouse. However, eventually the deficient GALC activity results in enough demyelination to cause death. Previous studies of bone marrow transplantation in twi mice resulted in an extension of their lives from 40 days until about 100 days (19 – 22). At that point psychosine levels and pathology were similar to those seen in untreated twi mice at 40 days. It appears from these and other studies that twi mice and this new transgenic mouse model initially make healthy looking myelin; however, at some point it starts to degenerate leaving bare axons. It has been shown that psychosine induces apoptosis of oligodendrocytes (30) and this probably initiates the demyelination process. Therefore therapy must be aimed at preventing psychosine accumulation either by preventing its synthesis or by supplying a source of GALC activity to stimulate its degradation before toxic levels are reached. Recent studies in vitro have demonstrated that GALC activity supplied to oligodendrocytes via viral vectors or via uptake from neighboring cells corrects their phenotype and prevents their premature demise
222
LUZI ET AL.
(31). This new transgenic mouse model of GLD will be useful as new methods for supplying GALC activity are developed.
13.
Luzi P, Rafi MA, Wenger DA. Characterization of the large deletion in the GALC gene found in patients with Krabbe disease. Hum Molec Genet 4:2335–2338, 1995.
14.
Wenger DA, Rafi MA, Luzi P. Molecular genetics of Krabbe disease (globoid cell leukodystrophy): Diagnostic and clinical implications. Hum Mutat 10:268 –279, 1997.
15.
Rafi MA, Luzi P, Zlotogora J, Wenger DA. Two different mutations are responsible for Krabbe disease in the Druze and Moslem Arab populations in Israel. Hum Genet 97:304 – 308, 1996.
16.
Suzuki K, Suzuki K. Genetic galactosylceramidase deficiency (globoid cell leukodystrophy, Krabbe disease) in different mammalian species. Neurochem Pathol 3:53– 68, 1985.
17.
Wenger DA. Murine, canine and non-human primate models of Krabbe disease. Molec Med Today 6:449 – 451, 2000.
18.
Kobayashi T, Yamanaka T, Jacobs JM, Teixera F, Suzuki K. The twitcher mouse: An enzymatically authentic murine model of human globoid cell leukodystrophy (Krabbe disease). Brain Res 202:479 – 483, 1980.
19.
Yeager AM, Brennan S, Tiffany C, Moser HW, Santos GW. Prolonged survival and remyelination after hematopoietic cell transplantation in the twitcher mouse. Science 225: 1053–1054, 1984.
20.
Ichioka T, Kishimoto Y, Brennan S, Santos GW, Yeager AM. Hematopoietic cell transplantation in murine globoid cell leukodystrophy (the twitcher mouse): Effects on levels of galactosylceramidase, psychosine, and galactocerebrosides. Proc Natl Acad Sci USA 84:4259 – 4263, 1987.
21.
Hoogerbrugge PM, Poorthuis BJ, Romme ARE, van de Kamp JJ, Wagemaker G, van Bekkum DW. Effect of bone marrow transplantation on enzyme levels and clinical course in the neurologically affected twitcher mouse. J Clin Invest 81:1790 –1794, 1988.
22.
Hoogerbrugge PM, Suzuki K, Suzuki K, Poorthuis BJHM, Kobayashi T, Wagemaker G, Bekkum DWV. Donor derived cells in the central nervous system of twitcher mice after bone marrow transplantation. Science 239:1035–1038, 1988.
23.
Wenger DA, Williams C. Screening for lysosomal disorders. In Techniques in Diagnostic Human Biochemical Genetics. A Laboratory Manual (Hommes FA, Ed.). New York: WileyLiss, pp 587– 617, 1991.
ACKNOWLEDGMENTS This research was supported in part by grants from the National Institutes of Health (DK38795) and from the European Leukodystrophy Association (M.T.V.). The excellent technical assistance of Han Zhi Rao and H. Xian Shen is acknowledged.
REFERENCES 1.
Wenger DA, Rafi MA, Luzi P, Datto J, Costantino-Ceccarini E. Krabbe disease: Genetic aspects and progress toward therapy. Molec Genet Metab 70:1–9, 2000.
2.
Wenger DA, Suzuki K, Suzuki Y, Suzuki K. Galactosylceramide Lipidosis. Globoid cell leukodystrophy (Krabbe disease). In The Metabolic and Molecular Bases of Inherited Disease, 8th ed. (Scriver CR, Beaudet AL, Valle D, Sly WS, Eds.). New York: McGraw-Hill, pp 3669 –3694, 2001.
3.
Chen YQ, Rafi MA, de Gala G, Wenger DA. Cloning and expression of cDNA encoding human galactocerebrosidase, the enzyme deficient in globoid cell leukodystrophy. Hum Molec Genet 2:1841–1845, 1993.
4.
Luzi P, Rafi MA, Wenger DA. Structure and organization of the human galactocerebrosidase (GALC) gene. Genomics 26: 407– 409, 1995.
5.
Luzi P, Victoria T, Rafi MA, Wenger DA. Analysis of the 5⬘ flanking region of the human galactocerebrosidase (GALC) gene. Biochem Molec Med 62:159 –164, 1997.
6.
Sakai N, Fukushima H, Inui K, Fu L, Nishigaki T, Yanagihara I, Tatsumi N, Ozono K, Okada S. Human galactocerebrosidase gene: Promoter analysis of the 5⬘-flanking region and structural organization. Biochim Biophys Acta 1395:62– 67, 1998.
7.
Thomas PK, Halpern J-P, King RHM, Patrick P. Galactosylceramide lipidosis: Novel presentation as a slowly progressive spinocerebellar degeneration. Ann Neurol 16:618 – 620, 1984.
8.
Vanier M. Symptomatology of late onset Krabbe’s leukodystrophy: The European experience. Dev Neurosci 13:240 –244, 1991.
24.
9.
Kolodny EH, Raghavan S, Krivit W. Late-onset Krabbe disease (globoid cell leukodystrophy): Clinical and biochemical features of 15 cases. Dev Neurosci 13:232–239, 1991.
Khillan JS, Bao Y. Preparation of high degree chimeric animals by one-step coculture of embryonic stem cells and preimplantation embryos. Biotechniques 22:544 –549, 1997.
25.
10.
Luzi P, Rafi MA, Wenger DA. Multiple mutations in the GALC gene in a patient with adult-onset Krabbe disease. Ann Neurol 40:116 –119, 1996.
Rodriguez-Lafrasse C, Rousson R, Pentchev PG, Louisot P, Vanier MT. Free sphingoid bases in tissues from patients with type C Niemann-Pick disease and other lysosomal storage disorders. Biochim Biophys Acta 1226:138 –144, 1994.
11.
Jardim LB, Giugliani R, Pires RF, Haussen S, Burin MG, Rafi MA, Wenger DA. Protracted course of Krabbe disease in an adult patient bearing a novel mutation. Arch Neurol 56:1014 –1017, 1999.
26.
12.
Rafi MA, Luzi P, Chen YQ, Wenger DA. A large deletion together with a point mutation in the GALC gene is a common mutant allele in patients with infantile Krabbe disease. Hum Molec Genet 4:1285–1289, 1995.
Matsumoto A, Vanier MT, Oya Y, Kelly D, Popko B, Wenger DA, Suzuki K, Suzuki K. Transgenic introduction of human galactosylceramidase into twitcher mouse: Significant phenotype improvement with a minimal expression. Dev Brain Dysfunct 10:142–154, 1997.
27.
Suzuki K, Suzuki K. The twitcher mouse: A model for Krabbe disease and for experimental therapies. Brain Pathol 5:249 –258, 1995.
TRANSGENIC MOUSE MODEL OF GLD 28.
Luzi P, Rafi MA, Wenger DA. Krabbe disease: Correlations between mutations in the GALC gene, GALC activity clinical presentation. Am J Hum Genet 57:A245, 1995. 29. Sakai N, Inui K, Tatsumi N, Fukushima H, Nishigaki T, Taniike M, Nishimoto J, Tsukamoto H, Yanagihara, Ozono K, Okada S. Molecular cloning and expression of cDNA for murine galactocerebrosidase and mutation analysis of the twitcher mouse, a model of Krabbe’s disease. J Neurochem 66:1118 –1124, 1996.
30.
223
Taniike M, Mohri I, Eguchi N, Irikura D, Urade Y, Okada S, Suzuki K. An apoptotic depletion of oligodendrocytes in the twitcher, a murine model of globoid cell leukodystrophy. J Neuropathol Exp Neurol 58:644 – 653, 1999. 31. Luddi A, Volterrani M, Strazza M, Smorlesi A, Rafi MA, Datto J, Wenger DA, Costantino-Ceccarini E. Retrovirusmediated gene transfer and galactocerebrosidase uptake into twitcher glial cells results in appropriate localization and phenotype correction. Neurobiol Dis, in press.
Generation of a Mouse with Low Galactocerebrosidase Activity by Gene Targeting: A New Model of Globoid Cell Leukodystrophy (Krabbe Disease) Paola Luzi,* Mohammad A. Rafi,* Mariam Zaka,* Mark Curtis,† Marie T. Vanier,‡ and David A. Wenger* ,1 *Department of Neurology, †Department of Pathology, Jefferson Medical College, Philadelphia, Pennsylvania; and ‡INSERM U 189, Lyon-Sud Medical School and Fondation Gillet-Merieux, Lyon-Sud Hospital, Pierre Benite, France Received March 27, 2001; published online June 27, 2001
changes less severe than twi mice in the central and peripheral nervous systems, and live about 15 days longer than twi mice. They have large litters and will play a role in therapy trials using new procedures currently under development. © 2001
Globoid cell leukodystrophy (Krabbe disease) is a severe leukodystrophy caused by mutations in the galactocerebrosidase (GALC) gene leading to extremely low (less than 5% of normal activity) GALC activity. Human patients include primarily severely affected infants as well as patients with a later onset of symptoms. The infants usually die before 2 years of age, but it is difficult to predict the clinical course in older patients. In addition to these patients, additional individuals identified in this laboratory have 10 –20% of normal GALC activity measured in accessible tissues. These individuals have a wide range of clinical presentations involving neurological degeneration. On molecular analysis of the GALC gene they all have three or more mutations considered to be normal polymorphisms resulting in amino acid changes in the two copies of the GALC gene. In order to investigate the role these amino acid changes may play on clinical, biochemical, and pathological findings, a new transgenic mouse was generated by homologous recombination. After preliminary studies determined what effect each amino acid change had on mouse GALC activity in transient transfection experiments, mice containing a cysteine residue at codon 168 instead of histidine (H168C) were produced. These mice developed symptoms, but they were delayed by 10 –15 days from the well-characterized twitcher (twi) mouse. They accumulated psychosine slightly slower than twi mice, showed pathological
Academic Press
Key Words: Krabbe disease; globoid cell leukodystrophy; white matter disease; transgenic mouse; galactocerebrosidase; myelin.
Globoid cell leukodystrophy (GLD) or Krabbe disease is an autosomal recessively inherited disorder resulting from very low galactocerebrosidase (GALC) activity (see 1,2, for review). The GALC cDNA was first cloned in 1993 (3) and subsequently the gene organization was determined (4 – 6). Most human patients present with symptoms before 6 months of age and die by 2 years of age. Older patients, including adults, are also diagnosed with GLD. The clinical course in older patients is highly variable even among siblings; however, many of the juvenile and adult onset patients present with weakness, tremor, and vision loss followed by varying degrees of intellectual impairment (7–11). The diagnosis is made by measuring very low GALC activity in available tissue samples, such as leukocytes and cultured skin fibroblasts; however, the residual GALC activity measured does not predict the clinical course a patient will take. While many patients with GLD have at least one copy of the common 30-kb deletion first identified in this laboratory (12,13), over 60 additional disease-causing and poly-
1 To whom correspondence should be addressed at Jefferson Medical College, 1020 Locust St., Room 394, Philadelphia, PA 19107. Fax: (215) 955-9554. E-mail: [email protected].
211 1096-7192/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.
212
LUZI ET AL.
morphic mutations have been identified (reviewed in 2,14). Within certain ethnic groups one common mutation has been found (15). One mutation, G809A (G270D), is found in a relatively high frequency in late-onset GLD patients. Carrier identification by measuring GALC activity is difficult due to the wide range of “normal” and carrier values. This is primarily due to polymorphisms in the coding regions of the gene that result in amino acid changes in the two processed subunits of GALC. Expression studies have demonstrated that the presence of these amino acids lowers the measured GALC activity from that measured with the most common amino acids (14). In addition to the human patients with GLD, several animal models, including twitcher (twi) mice, Cairn and Westhighland White terriers, and rhesus monkeys, have been identified and characterized (reviewed in 2,16,17). All of these models have very low GALC activity, and other biochemical and pathological findings are similar to those of human patients. Their respective GALC genes have been cloned and the mutations causing the disease have been identified (17). The twi mouse model has been used for a large number of studies since its identification in 1980 (18). These mice appear to be normal until 20 days of age when they start to have tremors. Weight gain slows, and they continue to deteriorate with most dying by 40 days of age. Twi mice given hematopoietic stem cell transplantation (HSCT) by 10 days of age can live until about 100 days (19 –22). During the enzyme-based screening of human patients for neurodegenerative disorders, GALC activity in leukocytes is often measured. Although true GLD patients have very low (⬍5% of our normal mean) GALC activity and are easy to identify, there is a wide “normal” range. However, a significant number of individuals (more than 50) were found to have GALC activity between 10 and 20% of our normal mean. Most of these people had clinical and other evidence for white matter disease suggestive of a leukodystrophy, but were not considered to have GLD. Subsequent mutation analysis for the most common disease-causing mutations and polymorphisms in the GALC gene revealed that almost all of these individuals have three or more nucleotide changes considered to be polymorphisms. No other changes were found upon sequencing all of the coding regions and exon-intron boundaries of the GALC gene in four patients. These changes are considered to be normal polymorphisms since homozygosity for one of them does not appear to have clinical implications. The clinical spectrum of these individuals is
TABLE 1 Polymorphisms in the Human GALC Gene and the Generated Amino Acid Changes in Mouse GALC Human polymorphisms
Mouse
Nucleotide change a
Amino acid change b
Allele frequency c
Amino acid change
C502T T1637C C502T/T1637C e
R168C I546T R168C/I546T
4–5% 30–40% ⬍2%
H168C V545T d H168C/V545T
a cDNA nucleotide number counting from the A of the initiation codon. b Codon number. c Frequency of the minor allele in the general population. d Due to the loss of codon 508 in the mouse GALC gene, human and mouse differ in nucleotide number at codon 546/545. e Present on the same allele.
broad so it was considered they might have white matter disease due to the low GALC activity if another genetic or environmental factor were present. In order to gain experimental evidence for the hypothesis that low (10 –20% of normal), but not deficient, GALC activity could cause white matter disease under certain conditions, we generated a mouse model by homologous recombination that contained a polymorphic change found in humans. This resulted in the generation of a mouse that had low GALC activity and symptoms delayed from those observed in the twi mouse model. Clinical, biochemical, and pathological studies on the homozygous mice are presented. While we had hoped to generate a mouse without symptoms or only very mild symptoms until stressed by environmental factors, we obtained a mouse with a longer life span than twi mice that will find use in treatment studies under development. In addition, these mice have large litters resulting in more affected mice for study. MATERIALS AND METHODS Expression of Mutations in COS-1 Cells Two common polymorphisms have been found in the Caucasian population alone and together on one allele (Table 1). Expression of GALC activity was measured using human GALC cDNA with and without the polymorphic changes following cloning into pcDNA3 as previously described (12). Mouse GALC cDNA was also cloned into the pcDNA3 vector. In order to change the amino acids at codon 168 from his to cys and codon 545 from val
TRANSGENIC MOUSE MODEL OF GLD
to thr, two nucleotide changes at cDNA positions 502 and 503 (CA to TG) and at positions 1633 and 1634 (GT to AC), respectively, were introduced using PCR-mediated mutagenesis. Constructs containing these changes alone or together were prepared. The constructs (normal and mutated) were transiently transfected into COS-1 cells using Lipofectamine (Gibco, Rockville, MD), and 72 h later the cells were harvested and GALC activity was measured. Measurement of GALC Activity Cultured cells were harvested with trypsinEDTA, washed with phosphate-buffered saline, and homogenized in deionized water. The total homogenate was used for enzyme analysis. Brain, liver, spleen, kidney, and heart from normal, heterozygous, and homozygous mice were removed immediately after death and homogenized in 10 vol of 0.05 M phosphate buffer (pH 6.5). Aliquots were diluted 1–10 with deionized water for determination of the protein concentration using the method of Lowry and for measurement of GALC activity using [ 3H]galactose-labeled galactosylceramide as previously described (23). Targeting Vector Construction Based on the results of the expression studies, cys at codon 168 was introduced into the mouse GALC gene. Genomic clones from a 129SVJ mouse genomic library (Stratagene, La Jolla, CA) were isolated using a GALC cDNA mouse probe. These clones were used, together with the amplification products obtained by XL-PCR (Perkin Elmer, Foster City, CA), to determine the genomic organization of the mouse GALC gene. Intron 4, spanning 2.0 kb, was completely sequenced, intron 5, spanning 5.6 kb, and intron 3, spanning 2.3 kb, were partially sequenced, and restriction analysis was performed. A 2.8-kb genomic fragment was obtained by XL-PCR using a sense primer located in intron 4 and an antisense primer located in intron 5. This fragment spanned 1200 bp of intron 4, exon 5 (about 150 bp), and 1400 bp of intron 5. The fragment was cloned into a modified pBluescript plasmid (Stratagene) in which the EcoRV and KpnI sites were removed. The 2 nt changes at cDNA positions 502 and 503 (exon 5) were introduced by directed mutagenesis using sequential PCR steps. The mutated fragment was cloned into TA vector (Invitrogen, Carlsbad, CA) and sequenced. It was then excised from the TA construct by digestion with NcoI and KpnI enzymes and
213
cloned into the modified pBluescript to replace the homologous fragment without the mutation. This fragment containing the desired mutation was cut out from pBluescript by EcoRV (restriction site present in intron 4) and BamHI (from pBluescript cloning site) digestion. The 2.0-kb band obtained was purified, BamHI linkers were added and it was ligated into the BamHI site of the pNTK plasmid. A second 4kb-genomic fragment, spanning 2.3 kb of intron 3, exon 4, and 1.6 kb of intron 4 (up to the EcoRV site) was obtained by XL-PCR. This fragment was cloned into the ClaI site of the pNTK vector. Using this cloning strategy the neomycin phosphotransferase (neo) cassette was located in the middle of intron 4 in the opposite orientation to the GALC gene. The thymidine kinase (TK) gene was located downstream and in the opposite orientation with respect to the GALC gene. The resulting targeting vector, designated pNTK-GALC contains a 5⬘ arm of 4 kb and a 3⬘ arm of 2 kb homologous with the endogenous GALC gene (Fig. 1). Electroporation and ES Cells Selection D3 embryonic stem (ES) cells (provided by Dr. Jaspal Khillan) were grown on ␥-irradiated G418resistant feeder cells on gelatinized plates following standard methods. Confluent 10-cm plates (about 2.5 ⫻ 10 7 cells) were trypsinized and resuspended in 0.9 ml of PBS. Forty micrograms of the pNTK-GALC construct was linearized by NotI digestion, precipitated, resuspended in TE buffer, and incubated with the cells for 5 min at room temperature in a Bio-Rad (Hercules, CA) electroporation cuvette. Cells were electroporated with a Bio-Rad Gene Pulser at 200 V and 500 F. After 10 min the ES cells were plated in six 10-cm plates containing feeder layers. G418 (Gibco) was added to the medium 24 h after electroporation at a concentration of 200 g/ml. Five of the six plates were also treated with 0.2 M 2⬘-fluoro-2⬘-deoxy-5-iodouracil--D-arabinofuranoside (FIAU) (Moravek, Brea, CA). Media were changed every 24 h. After 6 to 9 days single colonies were picked under a dissecting microscope with a Pasteur pipette. Identification of Recombinant Clones The colonies were grown in a 24-well plate. Once they became confluent half of the cells were frozen and half were used to extract DNA. The DNA was then analyzed by XL-PCR to identify the positive colonies using a sense primer in the neo gene (5⬘-
214
LUZI ET AL.
FIG. 1. Schematic representation of the point mutations introduced in the GALC gene by homologous recombination. (A) Restriction map of the wild-type GALC gene in the region homologous to the targeting vector. (B) Restriction map of the targeting construct linearized by NotI digestion. The asterisk in exon 5 indicates the location of the two point mutations (CA to TG) introduced at position 502 and 503. A neo cassette was inserted in intron 4 and a TK cassette was added 3⬘ of the region of homology. (C) Predicted structure of the mutated allele after homologous recombination. The restriction pattern obtained after BamHI digestion is shown for the wild type, targeting vector, and product of homologous recombination. Southern blot of BamHI-digested DNA from positive colonies was used to identify homologous recombination events.
GAGGCCACTTGTGTAGCGCC-3⬘) and an antisense primer 3⬘ to the region of homology contained in the targeting vector (5⬘-CACTGCAAACTCCCTGAGCC-3⬘). PCR products were resolved in a 0.8% agarose gel and those showing the expected 2.0-kb band were considered positive. The DNA from these clones was analyzed by another XL-PCR to confirm that homologous recombination occurred using a sense primer located 5⬘ of the region of homology contained in the targeting construct (5⬘-CCCGGCTTCTAG TAAATTACCC-3⬘) and an antisense primer located in the neo gene (5⬘-GCAGCCTCTGTTCCACATACAC-3⬘). A 4-kb band indicated a positive clone. Southern blotting was performed to confirm the presence of the correct construct in the positive clones. Ten micrograms of genomic DNA from the positive clones was digested overnight with BamHI, and the digest was run on a 0.7% agarose gel, trans-
ferred to a Hybond-N ⫹ (Amersham, Piscataway, NY) membrane, and hybridized with a 32P-labeled random primed probe using a 1.5-kb GALC gene fragment (spanning exon 5 and part of intron 5). The presence of the mutation in positive clones was confirmed by sequencing. Karyotyping was also done to verify the euploid condition of this clone. Production of Chimeric Mice The positive clone was expanded to prepare chimeric animals. The FVB/N females were superovulated and crossed with FVB/N males. Morulas were isolated after 2.5 days of mating. Chimeric animals were prepared by a one-step aggregation method developed by Khillan and Bao (24). The ES cells were transferred to a petri dish with microwells, and about 15–20 cells settled down on the bottom. The morulas without the zona pellucida were trans-
215
TRANSGENIC MOUSE MODEL OF GLD
ferred on top of the ES cells and were allowed to develop into blastocysts overnight. The following day, blastocysts were washed in medium and transferred to 2.5-day pseudopregnant females. The pups were born after 17 days. Chimeric animals were mated to wild-type animals. Heterozygous animals were identified and mated to produce mice homozygous for the desired mutation.
corresponding to 2–2.5 mg of protein using a HPLC procedure. The extraction procedure and general conditions have been previously described (25,26). The analysis of the orthophtalaldehyde derivatives was conducted on a 20-cm Spherisorb 5-m ODS2 C18 column with a 5-cm Spherisorb ODS2 C18 guard column. The mobile phase was methanol–5 mM sodium phosphate buffer, pH 7.0 (89/11) with 50 mg/L sodium octylsulfate as an ion-pairing agent.
Northern Blot and RT-PCR Liver, brain, spleen, kidney, and heart from normal, heterozygous, and homozygous animals were removed immediately after death and frozen in liquid nitrogen. Total RNA was extracted following the RNeasy Midi protocol (Qiagen) for animal tissue. Twenty micrograms of RNA was subjected to electrophoresis on a 1% agarose-formaldehyde gel and transferred to a nylon membrane. Northern blots were hybridized with a labeled probe prepared using a 700-bp mouse GALC cDNA fragment, at 68°C in ExpressHyb Hybridization solution (Clontech, Palo Alto, CA). The Northern blots were then washed in 0.1⫻ SSC– 0.1% SDS at 50°C and exposed over several days. RT-PCR was performed following the OneStep RT-PCR protocol from Qiagen using 1 g of RNA, and the PCR products were separated by electrophoresis on a 1% agarose gel. Clinical Evaluation of the Homozygous Mice Mice were genotyped between 2 and 4 days of age. DNA was isolated from clipped toes using the QIAamp tissue kit (Qiagen). The mutated allele was identified using the XL-PCR method used to detect the positive colonies. The nonmutated allele was detected by PCR using a sense primer located 5⬘ of the neo gene (5⬘-CTGATCTTGGTTGCCTGACC-3⬘) and the same antisense primer, located 3⬘ of the neo gene, used above in the PCR amplification for the mutated allele. Mice were examined every day for tremor and they were weighed every other day beginning at Day 10. When it was noted that the mice had weakness in the hind limbs and were having trouble reaching and chewing the standard mouse chow pellets, powdered food (Pico Vac, Test Diet, Richmond, IN) and water were placed on the floor of the cages. Measurement of Psychosine Levels Brain psychosine (galactosylsphingosine) was determined on aliquots of 20% tissue homogenates
Pathology For histological analysis tissues were fixed in 10% buffered formalin and embedded in paraffin. After slicing the sections were stained with hematoxylin and eosin, luxol fast blue/periodic acid Schiff (PAS), and Bielchowsky silver stain using standard procedures and examined by light microscopy. For ultrastructural analysis the tissues were fixed in 2% buffered (0.1 M Sorensen phosphate buffer, pH 7.4) glutaraldehyde and embedded in standard low viscosity embedding medium (10 g vinyl cyclohexane– dioxide, 6 g diglycidyl ether polypropylene glycol, 26 g nonenyl succinic anhydride, and 0.6 g dimethylaminoethanol). One-micrometer sections were cut and stained with toluidine blue. One hundred-nanometer-thick sections were cut on a Reichert Ultracut S microtome. The 100-nm sections were stained with uranyl acetate and lead citrate. Ultrastructural evaluations were performed on a Jeol 100 CX electron microscope. RESULTS Expression of Mutations in COS-1 Cells Several polymorphic changes have been found in human GALC (2); however, C502T (R168C) and T1639C (I546T) were encountered with high frequency in our patient population (Table 1). Expression of these polymorphisms alone and together in human GALC cDNA in COS-1 cells was measured after the desired nucleotide change in pcDNA3 was introduced as previously described (14). Before we tried to create a mouse model with low but not totally deficient GALC activity the same amino acid changes were made in mouse GALC, and the effects they had on GALC activity in vitro were determined. To produce a cysteine at position 168 the CAT codon for the unconserved histidine in mouse was changed to TGT to code for cysteine by PCR-mediated mutagenesis. A threonine was created at mouse GALC codon position 545 (one less codon than human be-
216
LUZI ET AL.
FIG. 2. Expression of GALC activity in COS-1 cells transfected with human or mouse GALC cDNA containing the most common sequence and polymorphic changes. Cells were transfected with pcDNA3 containing the constructs prepared as described under Materials and Methods, harvested 3 days later, and assayed for GALC activity.
cause of the loss of three nucleotides at codon 508) by replacing GTG sequence for valine to ACG for threonine. Both amino acid changes were made in one construct to mimic the human allele with both changes. Expression of these amino acid changes in human and mouse GALC was compared (Fig. 2). The creation of a cysteine residue at codon 168 in human GALC decreased the activity only 20%, while creating a cysteine residue in that position in mouse GALC resulted in a decrease of 80 –90% of normal activity. In contrast to the expression found in the human GALC, the decrease in activity due to the 168 cys change (H168C) was more significant than that obtained with the substitution of a threonine at position 545 (V545T) and with both changes in the same allele (H168C ⫹ V545T) (Fig. 2). Therefore, to generate a mouse model with low (10 –20% of normal) GALC activity we decided to produce a mouse with a cysteine residue at codon 168 by homologous recombination. Preparation of the Targeting Construct and Identification of Recombinant Clones The targeting construct was prepared by cloning genomic fragments of the mouse GALC gene, obtained by XL-PCR, into the pNTK vector. A 4.0-kb fragment containing intron 3, exon 4, and part of
intron 4 was cloned 5⬘ of the neo gene. A second 2.0-kb fragment composed of the rest of intron 4, exon 5, and part of intron 5 was cloned 3⬘ of the neo gene and 5⬘ of the TK gene. The desired two point mutations at cDNA positions 502 and 503 in intron 5 were introduced before cloning the fragment into the pNTK vector, using PCR-based methodology. After electroporation of the linearized targeting vector into ES cells, the transformants were selected with G418 or with G418 ⫹ FIAU. A total of 96 colonies was picked: 34 from a plate subjected to single selection and 62 from plates that underwent double selection. Homologous recombinant clones were detected by XL-PCR using primers in the neo gene coupled with primers located outside of the region of homology contained in the targeting construct. Three positive clones were identified: one was a colony selected with G418 only and the other two were obtained by double selection. PCR-positive clones were confirmed by Southern blot analysis of genomic DNA digested with BamHI and hybridized to a probe spanning exon 5 and part of intron 5. Two of the three positive clones identified by PCR showed the expected restriction pattern: a 6.1-kb band corresponding to the normal allele as well as the predicted 3.7-kb band created following homologous recombination (data not shown). Direct sequencing of
TRANSGENIC MOUSE MODEL OF GLD
217
FIG. 3. GALC activity in organs from noncarrier (NC), carrier (C), and affected (A) mice for H168C. Brain, liver, spleen, heart, and kidney were homogenized, and GALC was measured as described under Materials and Methods.
the PCR fragments derived from the two positive clones identified by Southern blotting and PCR was performed to verify the presence of the expected mutations at positions 502 and 503. The two desired mutations were present only in one of the clones, with the other showing the wild-type sequence. This clone was one obtained by double selection. Characterization of Mice with H168C Mutation The ES cells positively targeted were expanded and used to prepare chimeric animals. Overall 97 embryos were transferred to 9 foster mothers. Only 4 of them became pregnant and a total of 9 pups were born 4 out of which were chimeras. Two of the chimeras died soon after birth and the other two (both males) were mated with wild-type females to produce positive progeny. These males produced seven litters with a total of 36 pups, all of which were positives for ES cell recombination. The het-
erozygous mice were intercrossed to obtain mice homozygous for the desired change. GALC activity was measured in brain, liver, spleen, heart, and kidney samples from noncarrier, heterozygotes, and homozygous affected mice (Fig. 3). Heterozygous mice show GALC activities that are about 50% of normal, while homozygous mice show low values. It is not possible to state that the low GALC activities measured in the tissues from the transgenic mice are more than that measured in twi mice. For the first 25–30 days the homozygous mice feed normally and grow like heterozygote and homozygous normal littermates. However at about 25–30 days of age the homozygous mice start to show signs of tremor and hind leg weakness. This compares to about 20 days in twi mice. The tremor is most obvious when the mice are picked up by their tails. This strain of mice tends to be larger than twi mice, but eventually they start to lose weight. The mutant
218
LUZI ET AL.
(Fig. 6). This slower rate of increase could explain the protracted course in the transgenic mice. Pathological Studies in the Transgenic Mice
FIG. 4. Total body weights of normal mice of the FVB/N strain, affected transgenic mice, and affected twi mice at different ages. The number of mice at each time point varies from 1 to 10. If more than three, the weight is expressed as mean ⫾ standard deviation.
mice gain weight until around 35– 40 days of age, they then maintain their weight for few days, and then they begin to loose weight and generally die 10 –15 days later (Fig. 4). At the terminal stage the mutant animals generally become paralyzed in the hind legs having a different appearance from twi mice. Severe wasting of the hind limbs is clearly noted. This is seen in Fig. 5 where we compare a 50-day-old terminal transgenic mouse (right panel) to a 38-day-old terminal twitcher mouse (brown mouse in left panel). Skeletal muscle of both the transgenic and twi mice show atrophy consistent with a neurogenic origin. While the average life span of these animals is about 50 days, there is a high degree of variability. Some animals lived up to 60 – 63 days while others survived only for 40 – 42 days. Psychosine Values in Brain As psychosine has been implicated in the pathogenesis of oligodendrocytes and its concentration increases in brain with age in humans and animals with GLD, the psychosine concentration was measured in brains of these mice. As with the twi mouse model psychosine increases with age. However, the rate of increase in affected transgenic mice appears to be slightly slower than that observed in twi mice
In these pathological studies a transgenic mouse of 49 days is compared to a twitcher mouse of 36 days. Demyelination with infiltration of macrophages was identified in the cerebellar white matter, brainstem, cerebral white matter, and cerebral gray matter of a 36-day-old twitcher mouse as previously described by Suzuki and Suzuki (27). Luxol fast blue/PAS staining of cerebellar white matter from a twi mouse is shown in Fig. 7C. In contrast to the twitcher mouse, the 49-day-old transgenic mouse showed less macrophage infiltration in both white matter (Fig. 7B) and gray matter compared to the twi mouse (not shown). As expected, no PAS-positive macrophages were identified in the brains of control mice (Fig. 7A). Luxol fast/PAS-stained peripheral nerve from the control mouse showed well myelinated intact nerve (Fig. 8A). There was a marked loss of myelin and abundant infiltration of macrophages in the peripheral nerve from the 36-day-old twitcher mouse (Fig. 8C). The peripheral nerve from the 49-day-old transgenic mouse also showed a loss of myelin; however, the macrophage infiltration was less than that observed in the 36-day-old twitcher (Fig. 8B). Thick sections stained with toluidine blue show normal appearing myelinated axons in the control mouse (Fig. 8D). A moderate loss of myelin is observed in the peripheral nerve of the 49-day transgenic mouse (Fig. 8F). There were some scattered macrophages. In contrast, only rare myelinated axons remain in the peripheral nerve from the 36-dayold twitcher mouse and numerous endoneurial lipid laden macrophages are present (Fig. 8E). Ultrastructural analysis of the peripheral nerve showed similar changes in the mice with low GALC activity with good preservation of normal appearing myelin in the control mouse (Fig. 8G). Myelin loss and macrophage phagocytosis of myelin debris are identified in the peripheral nerve from both the 36day-old twitcher mouse (Fig. 8I) and the 49-day-old transgenic mouse (Fig. 8H). The loss of myelin and macrophage infiltration is greater in the twitcher mouse compared to the transgenic mouse. The ultrastructural appearance of the intact myelin that remains in the transgenic mouse is normal.
TRANSGENIC MOUSE MODEL OF GLD
FIG. 5. Photographs of a 53-day-old transgenic mouse (white) and 38-day-old twi mouse (left panel) and a 50-day-old transgenic mouse at the terminal stage (right panel). Note the pronounced wasting of the hind limbs in the 50-day-old transgenic mouse compared to the severely affected twi mouse and the difference in appearance between the two transgenic mice of similar ages.
Characterization of GALC mRNA in the Transgenic Mouse Because the mice were more severely affected than expected from the insertion of a missense mutation in the 50-kDa subunit, the mRNA produced was examined. To verify the production of GALC message carrying the desired mutation, RT-PCR and Northern blot analysis were performed using RNA extracted from different organs of the mutant mice. RNA from brain, liver, and kidney of noncarrier and homozygous animals was amplified using primers spanning the GALC cDNA region from nucleotides 150 to 600. Two main products were obtained in the affected mice: one of the expected size of 450 bp and the other of about 900 bp that was especially represented in the brain (Fig. 9). Sequencing of the two bands revealed that the 450 bp is derived from the correct message containing the 2 bp changes and the 900 bp is produced by an incorrect splicing of intron 4. In the homologous recombinant mice the GALC gene contains the neo gene in intron 4, located about 400 bp from the 3⬘ end of the intron. The incorrect splicing may be due to the presence of this extra piece of DNA. Even with some incorrect splicing, a significant amount of correctly spliced GALC containing the 2 nt change should be produced. Northern analysis revealed that level and size of the GALC mRNA in mutant animals are similar to wild type (data not shown). No larger GALC mRNA in brain was detected.
219
present. While the lower than normal GALC activity can be explained by the presence of multiple copies of normal polymorphisms, it is difficult to prove that the variable clinical features observed in these individuals are due to the lower than normal GALC activity. Other factors may be involved. If some agent or event resulted in focal or generalized demyelination, attempts to repair this damage would require the activation of oligodendrocytes in the central nervous system and Schwann cells in the peripheral nervous system to initiate remyelination. This would involve the synthesis of all proteins required for production of stable myelin including integral membrane proteins and enzymes required for the synthesis of myelin components such as galactosphingolipids. This remodeling of damaged myelin may require production of precise amounts of certain components, such as key lipids, and this could require adequate amounts of both biosynthetic and degradative enzymes. While partially deficient GALC activity may not have serious consequences in the initial synthesis of myelin to be wrapped around naked axons in an infant, it could have a clinical effect during remyelination when less than optimum conditions are present. While the H168C replacement created a mouse with low GALC activity, it presented with symptoms earlier than expected. The problem of generating an exact model of a human disease by replacing an amino acid in a protein from a mouse or other species is difficult. Clearly transfection studies are needed as an initial experiment to determine if a proposed change will have an effect on the enzy-
DISCUSSION The original reason for creating a mouse with low (10 –20% of normal) GALC activity was to determine if such an activity could cause white matter disease if certain environmental and/or genetic factors were
FIG. 6. Psychosine concentrations in brains from transgenic and twi mice of different ages. In normal mouse brain psychosine is almost undetectable at 15 days of age and it increases slowly to about 20 pmol/mg protein by 40 days of age.
220
LUZI ET AL.
FIG. 7. Luxol fast blue/PAS staining of cerebellar white matter from a 50-day-old control mouse (A), a 49-day-old transgenic mouse (B), and a 36-day-old twitcher mouse (C). (640⫻, original magnification).
matic activity. Preliminary transient expression studies replacing the more common amino acids in mouse GALC with the polymorphic amino acids found in human GALC lead to the first unexpected results. While in humans R168C had little effect,
H168C in mouse GALC had a pronounced effect on activity (Fig. 2). V545T had less of an effect on mouse GALC activity than H168C. This relatively common polymorphic change (I546T) in humans is the major cause for the wide range of GALC values
FIG. 8. Light and electron microscopy of peripheral nerves from a control mouse (A, D, G), 49-day-old transgenic mouse (B, E, H), and 36-day-old twitcher mouse (C, F, I). Panels A, B, and C show Luxol fast blue staining (640⫻, original magnification). Toluidine blue staining is shown in panels D, E, and F (640⫻, original magnification). Electron microscopy of peripheral nerves is shown in panels G, H, and I (7250⫻, original magnification).
TRANSGENIC MOUSE MODEL OF GLD
FIG. 9. RT-PCR of RNA from brain, liver, and kidney of a noncarrier (N) and two homozygous-affected transgenic mice (A1 and A2). The primers used span the GALC cDNA from nucleotides 150 to 600.
measured in diagnostic samples. Making both amino acid changes in the same construct had a pronounced effect on human GALC activity but less than either change alone in the mouse. Based on these findings it was decided to make a mouse with H168C. These studies show the difficulty in trying to make a mouse model of a human disease by generating an amino acid change found in human patients and assuming the effect will be the same. As the mouse GALC protein is only 84% identical to the human protein (29), the effects of changing one or more amino acid have unpredictable effects on activity and stability. While this study did not answer the initial question, it does provide us with an additional model for studying the pathophysiology and treatment of GLD. Generation of the model with a cysteine residue at codon 168 by homologous recombination was successful. Examination of organs from carriers and homozygous-affected mice showed the expected levels of GALC activity (Fig. 3). While we expected the affected mice to have some residual GALC activity, we could not conclude that it was higher than that found in twi mice that have a nonsense mutation at codon 339 (29). It may not be possible to accurately determine 10 –15% of normal activity in homogenates of organs. We did not take advantage of the Cre/loxP method for creating point mutations; however, we examined organs for the expression of normal size mRNA. Brain, liver, and kidney from affected mice were examined by RT-PCR and Northern blotting. While products of the expected size were found, some larger size amplification products were seen, especially in brain, apparently from aberrant splicing. Therefore, less than the predicted amounts of the GALC activity could have been produced. However, since previous studies (26) have demonstrated that low, even undetectable, levels of GALC activity have an ability to slow the course of
221
the disease in twi mice, it was predicted that even 5% of normal activity would be enough to delay or prevent symptoms. It appears that the small amount of activity present is enough to slow the course of the disease only slightly. The transgenic mice appear to be normal until approximately 25–30 days of age when fine tremor is noted. They continue to gain weight until about 40 days, when most twi mice have died, and then live an additional 10 –20 days. As shown in Fig. 5 there are some phenotypic differences between the transgenic mice and twi mice, as well as the variability between transgenic mice. The 50-day-old transgenic mouse shown on the right panel is more affected than the White 53-dayold transgenic mouse shown on the left panel. However, soon after this photo was taken the White transgenic mouse started to have more tremors and rapidly deteriorated and died at 60 days of age. What causes the significant clinical differences between these mice is not known at this time. Overall these mice appear to have a milder form of GLD than twi mice. An additional advantage of these mice is that they have large litters and are easy to handle. They accumulate psychosine at a slightly slower rate than twi mice, although they eventually reach levels similar to those of twi mice. Pathological studies in the central and peripheral nervous systems comparing a 36-day-old twi mouse with a 49-day-old transgenic mouse also confirm that the disease is progressing at a slower rate in the transgenic mouse. However, eventually the deficient GALC activity results in enough demyelination to cause death. Previous studies of bone marrow transplantation in twi mice resulted in an extension of their lives from 40 days until about 100 days (19 – 22). At that point psychosine levels and pathology were similar to those seen in untreated twi mice at 40 days. It appears from these and other studies that twi mice and this new transgenic mouse model initially make healthy looking myelin; however, at some point it starts to degenerate leaving bare axons. It has been shown that psychosine induces apoptosis of oligodendrocytes (30) and this probably initiates the demyelination process. Therefore therapy must be aimed at preventing psychosine accumulation either by preventing its synthesis or by supplying a source of GALC activity to stimulate its degradation before toxic levels are reached. Recent studies in vitro have demonstrated that GALC activity supplied to oligodendrocytes via viral vectors or via uptake from neighboring cells corrects their phenotype and prevents their premature demise
222
LUZI ET AL.
(31). This new transgenic mouse model of GLD will be useful as new methods for supplying GALC activity are developed.
13.
Luzi P, Rafi MA, Wenger DA. Characterization of the large deletion in the GALC gene found in patients with Krabbe disease. Hum Molec Genet 4:2335–2338, 1995.
14.
Wenger DA, Rafi MA, Luzi P. Molecular genetics of Krabbe disease (globoid cell leukodystrophy): Diagnostic and clinical implications. Hum Mutat 10:268 –279, 1997.
15.
Rafi MA, Luzi P, Zlotogora J, Wenger DA. Two different mutations are responsible for Krabbe disease in the Druze and Moslem Arab populations in Israel. Hum Genet 97:304 – 308, 1996.
16.
Suzuki K, Suzuki K. Genetic galactosylceramidase deficiency (globoid cell leukodystrophy, Krabbe disease) in different mammalian species. Neurochem Pathol 3:53– 68, 1985.
17.
Wenger DA. Murine, canine and non-human primate models of Krabbe disease. Molec Med Today 6:449 – 451, 2000.
18.
Kobayashi T, Yamanaka T, Jacobs JM, Teixera F, Suzuki K. The twitcher mouse: An enzymatically authentic murine model of human globoid cell leukodystrophy (Krabbe disease). Brain Res 202:479 – 483, 1980.
19.
Yeager AM, Brennan S, Tiffany C, Moser HW, Santos GW. Prolonged survival and remyelination after hematopoietic cell transplantation in the twitcher mouse. Science 225: 1053–1054, 1984.
20.
Ichioka T, Kishimoto Y, Brennan S, Santos GW, Yeager AM. Hematopoietic cell transplantation in murine globoid cell leukodystrophy (the twitcher mouse): Effects on levels of galactosylceramidase, psychosine, and galactocerebrosides. Proc Natl Acad Sci USA 84:4259 – 4263, 1987.
21.
Hoogerbrugge PM, Poorthuis BJ, Romme ARE, van de Kamp JJ, Wagemaker G, van Bekkum DW. Effect of bone marrow transplantation on enzyme levels and clinical course in the neurologically affected twitcher mouse. J Clin Invest 81:1790 –1794, 1988.
22.
Hoogerbrugge PM, Suzuki K, Suzuki K, Poorthuis BJHM, Kobayashi T, Wagemaker G, Bekkum DWV. Donor derived cells in the central nervous system of twitcher mice after bone marrow transplantation. Science 239:1035–1038, 1988.
23.
Wenger DA, Williams C. Screening for lysosomal disorders. In Techniques in Diagnostic Human Biochemical Genetics. A Laboratory Manual (Hommes FA, Ed.). New York: WileyLiss, pp 587– 617, 1991.
ACKNOWLEDGMENTS This research was supported in part by grants from the National Institutes of Health (DK38795) and from the European Leukodystrophy Association (M.T.V.). The excellent technical assistance of Han Zhi Rao and H. Xian Shen is acknowledged.
REFERENCES 1.
Wenger DA, Rafi MA, Luzi P, Datto J, Costantino-Ceccarini E. Krabbe disease: Genetic aspects and progress toward therapy. Molec Genet Metab 70:1–9, 2000.
2.
Wenger DA, Suzuki K, Suzuki Y, Suzuki K. Galactosylceramide Lipidosis. Globoid cell leukodystrophy (Krabbe disease). In The Metabolic and Molecular Bases of Inherited Disease, 8th ed. (Scriver CR, Beaudet AL, Valle D, Sly WS, Eds.). New York: McGraw-Hill, pp 3669 –3694, 2001.
3.
Chen YQ, Rafi MA, de Gala G, Wenger DA. Cloning and expression of cDNA encoding human galactocerebrosidase, the enzyme deficient in globoid cell leukodystrophy. Hum Molec Genet 2:1841–1845, 1993.
4.
Luzi P, Rafi MA, Wenger DA. Structure and organization of the human galactocerebrosidase (GALC) gene. Genomics 26: 407– 409, 1995.
5.
Luzi P, Victoria T, Rafi MA, Wenger DA. Analysis of the 5⬘ flanking region of the human galactocerebrosidase (GALC) gene. Biochem Molec Med 62:159 –164, 1997.
6.
Sakai N, Fukushima H, Inui K, Fu L, Nishigaki T, Yanagihara I, Tatsumi N, Ozono K, Okada S. Human galactocerebrosidase gene: Promoter analysis of the 5⬘-flanking region and structural organization. Biochim Biophys Acta 1395:62– 67, 1998.
7.
Thomas PK, Halpern J-P, King RHM, Patrick P. Galactosylceramide lipidosis: Novel presentation as a slowly progressive spinocerebellar degeneration. Ann Neurol 16:618 – 620, 1984.
8.
Vanier M. Symptomatology of late onset Krabbe’s leukodystrophy: The European experience. Dev Neurosci 13:240 –244, 1991.
24.
9.
Kolodny EH, Raghavan S, Krivit W. Late-onset Krabbe disease (globoid cell leukodystrophy): Clinical and biochemical features of 15 cases. Dev Neurosci 13:232–239, 1991.
Khillan JS, Bao Y. Preparation of high degree chimeric animals by one-step coculture of embryonic stem cells and preimplantation embryos. Biotechniques 22:544 –549, 1997.
25.
10.
Luzi P, Rafi MA, Wenger DA. Multiple mutations in the GALC gene in a patient with adult-onset Krabbe disease. Ann Neurol 40:116 –119, 1996.
Rodriguez-Lafrasse C, Rousson R, Pentchev PG, Louisot P, Vanier MT. Free sphingoid bases in tissues from patients with type C Niemann-Pick disease and other lysosomal storage disorders. Biochim Biophys Acta 1226:138 –144, 1994.
11.
Jardim LB, Giugliani R, Pires RF, Haussen S, Burin MG, Rafi MA, Wenger DA. Protracted course of Krabbe disease in an adult patient bearing a novel mutation. Arch Neurol 56:1014 –1017, 1999.
26.
12.
Rafi MA, Luzi P, Chen YQ, Wenger DA. A large deletion together with a point mutation in the GALC gene is a common mutant allele in patients with infantile Krabbe disease. Hum Molec Genet 4:1285–1289, 1995.
Matsumoto A, Vanier MT, Oya Y, Kelly D, Popko B, Wenger DA, Suzuki K, Suzuki K. Transgenic introduction of human galactosylceramidase into twitcher mouse: Significant phenotype improvement with a minimal expression. Dev Brain Dysfunct 10:142–154, 1997.
27.
Suzuki K, Suzuki K. The twitcher mouse: A model for Krabbe disease and for experimental therapies. Brain Pathol 5:249 –258, 1995.
TRANSGENIC MOUSE MODEL OF GLD 28.
Luzi P, Rafi MA, Wenger DA. Krabbe disease: Correlations between mutations in the GALC gene, GALC activity clinical presentation. Am J Hum Genet 57:A245, 1995. 29. Sakai N, Inui K, Tatsumi N, Fukushima H, Nishigaki T, Taniike M, Nishimoto J, Tsukamoto H, Yanagihara, Ozono K, Okada S. Molecular cloning and expression of cDNA for murine galactocerebrosidase and mutation analysis of the twitcher mouse, a model of Krabbe’s disease. J Neurochem 66:1118 –1124, 1996.
30.
223
Taniike M, Mohri I, Eguchi N, Irikura D, Urade Y, Okada S, Suzuki K. An apoptotic depletion of oligodendrocytes in the twitcher, a murine model of globoid cell leukodystrophy. J Neuropathol Exp Neurol 58:644 – 653, 1999. 31. Luddi A, Volterrani M, Strazza M, Smorlesi A, Rafi MA, Datto J, Wenger DA, Costantino-Ceccarini E. Retrovirusmediated gene transfer and galactocerebrosidase uptake into twitcher glial cells results in appropriate localization and phenotype correction. Neurobiol Dis, in press.