LBSL (leukoencephalopathy with brain stem and spinal cord involvement and lactate elevation) is an autosomal recessive white matter disorder with slowly progressive cerebellar ataxia, spasticity and dorsal column dysfunction. Magnetic resonance imaging shows characteristic abnormalities in the cerebral white matter and specific brain stem and spinal cord tracts. LBSL is caused by mutations in the gene DARS2, which encodes mtAspRS (mitochondrial aspartyl-tRNA synthetase). The selective involvement of specific white matter tracts in LBSL is striking since this protein is ubiquitously expressed. Almost all LBSL patients have one mutation in intron 2 of DARS2, affecting the splicing of the third exon. Using a splicing reporter construct, we find cell-type-specific differences in the sensitivity to these mutations: the mutations have a larger effect on exon 3 exclusion in neural cell lines, especially neuronal cell lines, than in non-neural cell lines. Furthermore, correct inclusion of exon 3 in the normal mtAspRS mRNA occurs less efficiently in neural cells than in other cell types, and this effect is again most pronounced in neuronal cells. The combined result of these two effects may explain the selective vulnerability of specific white matter tracts in LBSL patients.
The autosomal recessive white matter disorder LBSL (leukoencephalopathy with brain stem and spinal cord involvement and lactate elevation) is caused by mutations in the gene DARS2 [1,2], which encodes mtAspRS (mitochondrial aspartyl-tRNA synthetase). LBSL is a neurological disorder, clinically characterized by a slowly progressive pyramidal, cerebellar and dorsal column dysfunction . Peripheral neuropathy may also occur . Deterioration of motor skills usually starts in childhood or adolescence , but can even start in adulthood [4,5]. The MRI (magnetic resonance imaging) pattern comprises signal abnormalities in the cerebral white matter and specific brain stem and spinal cord tracts. MRS (magnetic resonance spectroscopy) shows increased lactate in the abnormal white matter in almost all affected individuals . Involvement of other organs than the nervous system has never been reported.
Different types of mutations have been found in DARS2, i.e. deletions, nonsense and splice site mutations leading to major alterations at the protein level, and missense mutations. For several of these missense mutations a reduced catalytic activity of the encoded mtAspRS has been demonstrated . In patient-derived lymphoblasts, fibroblasts or muscle tissue, however, no abnormalities in mitochondrial function could be detected [1,2]. Mitochondria are present and necessary in almost all cells of the human body. Mitochondrial diseases in general preferentially affect tissues that have a high energy demand, such as muscle, heart and brain. The cause of the selective vulnerability of specific white matter tracts in LBSL is not understood. The fact that long tracts are affected suggests that the disease is due to a primary neuronal or axonal defect and not to a primary defect of myelin, oligodendrocytes or astrocytes.
Strikingly, no LBSL patients are known with homozygous mutations in DARS2 and almost all patients with LBSL have one mutation at the 3′ end of the second intron of DARS2 [2,3]. These mutations occur in a stretch of T- and C-residues upstream of exon 3. The most common mutation is 228–20_−21delTTinsC. The mutations in this region affect the splicing of the third exon of mtAspRS mRNA, leading to skipping of the third exon, a frameshift and a premature stop. This splicing defect is probably ‘leaky’, allowing expression of a reduced amount of the wild-type protein from this allele. The degree of loss of expression of mtAspRS caused by these splicing defects has not been addressed. The mutation on the other allele appears to lead to a severe or complete loss of activity of the encoded variant .
Mutations in another mitochondrial aminoacyl-tRNA synthetase gene, RARS2, were found to cause a form of pontocerebellar hypoplasia . The patients all had an intronic mutation affecting the splicing of the second exon of RARS2. The authors suggested that splicing mutations in mitochondrial aminoacyl-tRNA synthetase genes preferentially affect the brain because of a tissue-specific difference in the efficiency of the splicing machinery. It appears that the splice site at the end of intron 2 of DARS2 is a rather weak splice site even in the absence of disease-causing mutations. The acceptor site of exon 3 is not recognized by splice-site prediction software such as NNSplice or GeneSplicer.
In the present study, we have investigated the splicing defect owing to DARS2 mutations upstream of exon 3 in detail. Using patient-derived lymphoblast cell lines, we investigated whether the splicing defect of DARS2 exon 3 is a partial defect. We then tested two hypotheses that may explain the selective vulnerability of nervous tissue as compared with other tissues and long tracts as compared with other regions of the nervous system: (i) the splicing of exon 3 in the wild-type mtAspRS mRNA may be less efficient in neural cells than in other cell types, and within the nervous system less efficient in neuronal cells than glial cell types; and (ii) the intron 2 mutations may have a cell-type-specific effect on the splicing of exon 3 and could be less favourable in neural cells than cells from other organs, and within the nervous system less favourable in neuronal cells than glial cell types. Both hypotheses could be true and the effect of one would enhance the effect of the other, explaining the selective vulnerability of nervous system tracts in LBSL.
MATERIALS AND METHODS
Patient cell lines
Informed consent for use of patients' lymphoblasts was obtained from the patients or their parents. The study was approved by the Institutional Review Board of the VU Medical Center. The patients' lymphoblasts that were used in the present study and the mutations that have been found in these patients are listed in Table 1.
Reporter constructs containing exon 2, intron 2, exon 3, intron 3 and a part of exon 4 of DARS2 were created by PCR amplification of genomic DNA from control and two different patients (LBSL2 and LBSL33 ). An initiation codon was created in exon 2 (Primer BglII-F, for primers see Table 2). Products were cloned into pEYFP-N1 [Clontech; EYFP is enhanced YFP (yellow fluorescent protein)] and sequenced. The initiation codon of YFP was removed by site-directed mutagenesis using QuikChange® (Stratagene) according to the manufacturer's protocol.
|For expression studies|
|For expression studies|
Control and patient-derived lymphoblasts were obtained and cultured under the same conditions as described previously . Mitochondria were isolated from lymphoblasts using the Mitochondrial Isolation Kit (Miltenyi Biotec) according to the manufacturer's protocol.
HEK (human embryonic kidney)-293T, HeLa, HOG and NTera-2/D1 (NT2) cells were grown in DMEM (Dulbecco's modified Eagle's medium) with GlutaMAX™ and 10% FBS (fetal bovine serum). MO3.13 cells were grown in RPMI with GlutaMAX™ and 10% FBS. SH-SY5Y and BE(2)-M17 cells were grown in OptiMEM (Invitrogen) with 10% FBS. Cells were cultured at 37°C and 5% CO2.
Differentiation of NT2 and SH-SY5Y cells
NT2 cells were differentiated into neurons and astrocytes as described previously . In brief, NT2 cells were plated at a density of 2.5×106 cells per T-75 flask in DMEM supplemented with 10% FBS. Cells were incubated at 37°C in 5% CO2 until they reached approximately 80% confluence prior to initiating differentiation. During the differentiation period, cells were cultured in DMEM containing 10 μM RA (retinoic acid). Medium was replaced twice a week for a total of 4 weeks. After 4 weeks, cells were split 1:6 and cultured for 2 weeks. Neurons (NT2/N) were separated from the underlying astrocytes (NT2/A) by mechanically dislodging neurons by gently striking the flask.
SH-SY5Y cells were differentiated in OptiMEM containing 0.5% FBS and 0.1 μM RA for 6 days. Medium was replaced every 2 days.
Transfections were performed in six-well plates with cultured cells at 80% confluency. For BE(2)-M17, SH-SY5Y, HOG and MO3.13 cell lines Lipofectamine™ 2000 reagent (Invitrogen) was used, and for HeLa and HEK-293T cells FuGENE™ reagent (Roche Applied Sciences) was used, according to the manufacturers' protocols. Fluorescence was confirmed on a fluorescence microscope (Nikon) after 24 h.
Cell lysates were obtained 44 h after transfection by harvesting in lysis buffer [20 mM Hepes (pH 7.4), 100 mM KCl, 10% glycerol, 1% Triton X-100 and 0.1 mM EDTA] with PRIC (protease inhibitor cocktail, Roche Applied Sciences). After lysis, nuclei and cell debris were removed by centrifugation (13000 g for 1 min at 4°C) and protein concentrations of the remaining lysates were determined using the Bradford method (Bio-Rad).
SDS/PAGE and Western blot analysis
Mitochondrial or cytoplasmic lysates were run on SDS/PAGE (12.5% gels) and blotted on to PVDF membranes (Millipore). Membranes were blocked in PBS-T (PBS containing 0.1% Tween 20) and 5% non-fat skimmed milk. Proteins of interest were detected by incubating the membranes with the appropriate antibodies in PBS-T containing 0.5% non-fat skimmed milk for either 2 h at room temperature (21°C) or overnight at 4°C. Antibodies used were mouse anti-GFP (green fluorescent protein; 1:1000, Sigma), rabbit anti-eIF2α (eukaryotic initiation factor 2α; 1:1000, Santa Cruz Biotechnology), mouse anti-mtAspRS (1:500, Abcam), rabbit anti-mtIsoRS (mitochondrial isoleucyl-tRNA synthetase; 1:500, Abcam), rabbit anti-mtValRS (mitochondrial valyl-tRNA synthetase; 1:3000, a gift from Dr R. Lightowler, Mitochondrial Research Group, University of Newcastle, Newcastle, U.K.), rabbit anti-LysRS (lysyl-tRNA synthetase; 1:2000, Abcam) and mouse anti-COXIV (cytochrome c oxidase IV; 1:1000, Abcam). After washing with PBS-T, alkaline-phosphatase-conjugated secondary antibodies (1:5000, Sigma) were added for 1 h and the immunoreactive bands were detected using ECF substrate (Amersham) on a FLA-5000 image reader (Fujifilm). Western blots were quantified using the AIDA Image Analyzer software package.
A tailed antisense oligonucleotide was designed  and obtained from Integrated DNA Technologies with the following sequence: 5′-rA*rG*rG*rA*rG*rGrArCrGrGrArGrGrArCrGrGrArGrGrArCrAmCmAmA*mG*mA*mA*mU*mGmUmGmU*mU*mU*mU*mG-3′, in which ‘r’, ‘m’ and the asterisk refer to RNA bases, 2′O-methyl and phosphorothioate chemical groups respectively. The oligonucleotides were co-transfected with the reporter constructs into SH-SY5Y cells.
RNA isolation, RT (reverse-transcription)–PCR and qPCR (quantitative PCR)
Transfected cells were harvested 48 h after transfection. NT2 cells were harvested as undifferentiated cells or after the differentiation protocol as described above. Where indicated in the Figure legends cycloheximide (250 μg/ml) was added to the culture medium 4 h prior to RNA extraction. Total RNA was extracted using TRIzol® (Invitrogen). First-strand cDNA synthesis was carried out with Superscript III RT (Invitrogen). PCR was done on cDNA with Platinum Taq according to the manufacturer's protocol (Invitrogen). qPCR was performed using SYBR Green (Roche) on a LightCycler 480 (Roche). The primers used are listed in Table 2.
Levels of endogenous proteins or mRNA levels, or YFP levels in transfected cells were calculated as means±S.D.; error bars indicate the S.D. We used Student's t test to assess the statistical significance of differences in these levels between different cell lines as indicated in the text.
Reduced mtAspRS expression in patient lymphoblasts
To address the effect of the mutations upstream of exon 3 on the expression of mtAspRS we determined the expression of mtAspRS in lymphoblasts from LBSL patients with different types of mutations (Table 1). Except for LBSL50, all had a mutation in intron 2 of DARS2 on one allele. Some of these cell lines cannot express full-length protein from the second allele owing to a deletion or a nonsense mutation (Table 1). As a consequence, the full-length mtAspRS that is expressed in these cells must come from the allele with the intron 2 mutation. Western blotting was performed on isolated mitochondria from lymphoblasts using an antibody directed against mtAspRS (Figure 1). Comparing the expression in control cells with the expression in patient-derived cells showed that the expression of mtAspRS in all patient lymphoblasts was substantially lower than in control lymphoblasts (Figure 1A). The expression levels were reduced to 15–35% of the levels in controls. To study the expression levels of other mitochondrial tRNA synthetases the experiment was repeated with more cells as the starting material (Figure 1B). Western blotting using antibodies against the mitochondrial tRNA synthetases for lysine, valine and isoleucine showed similar levels of expression of these enzymes in patients and control cell lines when corrected for loading differences (Figure 1B). Reduced levels were only seen for mtAspRS. Quantification of the blots showed a significant reduction in the expression of mtAspRS in the patients' cells compared with control cells (Figure 1C). The expression of the individual cell lines are indicated in Figure 1(D).
Reduced expression of mtAspRS in patient-derived cells
qPCR was performed on the same lymphoblast lines with primers crossing the exon 2–3 boundary. This showed a reduction of total mtAspRS mRNA expression of approximately 50% in patients LBSL7 and LBSL152, but no reduction in the other patients as compared with control levels (Table 3).
|Lymphoblasts||Total (−cycloheximide)||Without exon 3 (−cycloheximide)||Without exon 3 (+cycloheximide)||Without exon 3 (percentage of total)|
|Lymphoblasts||Total (−cycloheximide)||Without exon 3 (−cycloheximide)||Without exon 3 (+cycloheximide)||Without exon 3 (percentage of total)|
The expression level in cell line C1 was set at 100%.
Using a primer that spans the exon 2–4 boundary we analysed the extent of aberrant splicing (Table 3). Compared with control cells, the level of mtAspRS mRNA lacking exon 3 increased by 2–5-fold in patient-derived cells with an intron 2 mutation. In the presence of cycloheximide to inhibit nonsense-mediated decay, these numbers showed a 4–9-fold increase. In these cells the amount of the variant lacking exon 3 represented approximately 10% of the total mtAspRS mRNA.
Expression of reporter constructs in SH-SY5Y cells
Figure 1 shows that DARS2 mutations decrease the expression of mtAspRS in patient-derived lymphoblasts. To explore the role of the mutations in intron 2 on exon 3 splicing in other cell types, we designed reporter constructs containing exons 2 and 3 and part of exon 4 interspersed by introns 2 and 3 fused to YFP (see Figures 2A and 2B). When exon 3 is included in the mRNA from these minigene constructs, the YFP coding region is out-of-frame with the AUG start codon that is newly created in exon 2. When exon 3 is skipped, exon 2 is spliced on to exon 4 and the YFP coding region is in-frame with the start codon in exon 2. Therefore these reporters result in a gain of YFP signal when exon 3 is excluded from the final transcript. These constructs were made with the wild-type sequence and with two different mutations in intron 2 that were found in LBSL patients (Figure 2C). The reporter constructs were transfected into SH-SY5Y cells, showing increased YFP expression with the mutant constructs, indicating an increase in exon 3 skipping (Figure 2D).
Reporter construct to measure exclusion of exon 3
Intron 2 mutations affect exon 3 splicing in a cell-type-specific manner
The three reporter constructs were transiently transfected into different cell lines, i.e. the non-neural cell lines HEK-293T  and HeLa (cervical carcinoma cells)  and different neural cell lines, including BE(2)-M17 (neuroblastoma cells) , SH-SY5Y (neuroblastoma cells) , HOG (oligodendrocytoma cells) , MO3.13 (immortalized hybrid oligodendrocyte cells)  and U373 (astrocytoma cells) .
The efficiency of exon 3 splicing was evaluated by means of Western blotting using an antibody against YFP (Figure 3). Bands were detected at the expected size of 31 kDa. Figure 3 shows higher YFP expression with the mutant constructs than with the wild-type constructs, but the degree differed between cell lines. To determine the effects of the mutations of exon 3 splicing efficiency, the bands on the Western blots were quantified (Figure 3B). The differences in YFP expression between the wild-type and the mutant constructs were most pronounced in U373, HOG, BE(2)-M17 and SH-SY5Y cells in comparison with non-neural cell types (HEK-293T and HeLa), with the largest effect in the neuronal cell line SH-SY5Y. The effect of both mutations was significantly larger in SH-SY5Y cells than in HEK-293T, HeLa and MO3.13 cells (P<0.05). Also the effect in BE(2)-M17 cells was significantly larger than in HeLa cells for both mutations, and only for mutation A in HEK-293T, HOG and MO3.13 cells (P<0.05). All P values for differences in YFP expression between the different cell types are given in Table 4.
Cell-type-specific differences in splicing efficiencies
Splicing of endogenous mtAspRS mRNA
To examine the possibility that splicing of wild-type mtAspRS pre-mRNA is regulated in a cell-type-specific manner, we determined the splicing efficiency of exon 3 of the endogenous wild-type mtAspRS mRNA in HEK-293T and SH-SY5Y cells by qPCR (Figures 4A and 4B). A significantly higher amount of mRNA product without exon 3 was seen in SH-SY5Y cells than in HEK-293T cells (P<0.05). Treatment of SH-SY5Y cells with RA to induce neuronal differentiation increased exon 3 skipping as compared with undifferentiated SH-SY5Y cells in some experiments, but the result was not consistent (results not shown). We then compared splicing of endogenous mtAspRS mRNA in undifferentiated and differentiated NT2 cells. In the undifferentiated state, NT2 cells have a lower expression level of the variant without exon 3 than SH-SY5Y cells (2.3% compared with 6.3%). NT2 cells can be differentiated into a neuronal (NT2/N) or astrocytic (NT2/A) phenotype  (Figure 4C). qPCR showed that the amount of mRNA product without exon 3 was higher in the NT2/N cells than in the undifferentiated (P<0.01) and astrocytic (P<0.1) cells (Figure 4D).
Increased exclusion of exon 3 in neuronal cells
Antisense oligonucleotides increase exon 3 inclusion
Prediction software indicates that the acceptor site at the end of intron 2 represents a weak splice site. To assess the question as to whether we can influence the efficiency of exon 3 splicing, we tested an antisense oligonucleotide that was designed to attract splicing factors to this region. The oligonucleotide was complementary to the first part of exon 3 and contained an additional non-complementary sequence that was shown previously to mimic exonic splicing enhancer sequences . The oligonucleotide was co-transfected with the splicing constructs in SH-SY5Y cells. The efficiency of exon 3 splicing was determined by RT–PCR (Figure 5). The results confirm the negative effect of LBSL mutations on correct splicing of exon 3 and show that the oligonucleotide promoted inclusion of exon 3.
Antisense oligonucleotides with splicing enhancer sequences increase correct splicing of exon 3
Mutations in the gene DARS2 cause LBSL, which is a disease exclusively affecting the central and peripheral nervous system. MtAspRS, the protein encoded by DARS2, is needed for the translation of mitochondrial mRNA and therefore necessary in almost all cells in the human body. To date, how mutations in this ubiquitously expressed gene cause a disorder in which tracts of the central and peripheral nervous system are selectively affected is unexplained. Tract-wise involvement suggests a neuronal or axonal defect rather than a defect of myelin, oligodendrocytes or astrocytes.
We have determined the impact of mutations that occur 12–20 nucleotides upstream of exon 3 on the expression of mtAspRS in patient-derived lymphoblasts. An increase in mRNA product without exon 3 is seen in patients with a mutation in intron 2. By analysing mtAspRS levels we have confirmed that this splicing defect is ‘leaky’. In patients with one nonsense, frameshift or another splice site mutation other than the one in intron 2, expression of normal mtAspRS is entirely dependent on the allele with the intron 2 mutation; in these patients expression of normal mtAspRS is reduced but still detectable. Despite this decrease in expression, respiratory chain activities in lymphoblasts from LBSL patients are not affected , indicating that this remaining mtAspRS activity is sufficient for normal mitochondrial activity.
qPCR repeatedly showed only a 50% reduction of mtAspRS mRNA in two patients (LBSL7 and LBSL152) and no reduction in the other three patients. Four out of the five patients investigated have an intron 2 mutation on one allele. On the other allele LBSL7 and LBSL152 have a nonsense mutation and a deletion leading to a frameshift and a premature stop respectively. These mutations are expected to lead to nonsense-mediated decay, which explains the 50% reduction in these patients. The other three cell lines have missense mutations or an in-frame deletion, which are not expected to affect mRNA stability. The intron 2 mutations do not seem to affect the transcript levels significantly in patients' lymphoblasts and the amount of aberrantly spliced mRNA, without exon 3, represents only a low percentage of the total. Despite the apparently small change in mRNA levels, the protein levels are substantially reduced. We speculate that the mtAspRS mRNA derived from the allele with the intron 2 mutations is relatively stable, but has a low translation competence. At this moment we have no data to explain the underlying mechanism.
With a minigene construct and different types of cultured cells Disset et al.  showed that a reduced number of T-residues in a polymorphic region at the end of intron 8 of the cystic fibrosis transmembrane conductance regulator pre-mRNA exerted a cell-type-specific effect on skipping of exon 9. Using a similar approach with a DARS2 minigene construct and various cultured cell lines, we confirm that intron 2 mutations, as observed in LBSL patients, increase exon 3 skipping. Interestingly, the mutations tend to have a larger impact in neural cell lines, with the largest effect found in the neuronal cell line SH-SY5Y.
Furthermore, we show that the efficiency of exon 3 inclusion in wild-type mtAspRS mRNA is cell-type-dependent. The neuronal cell line SH-SY5Y has significantly higher levels of mtAspRS mRNA without exon 3 than the non-neural cell line HEK-293T. Differentiation of NT2 cells into neuronal cells results in a clear increase in the level of the endogenous mtAspRS transcript without exon 3, which excludes a possible effect of differences in genetic background of the cells. Differentiation into astrocytic cells also increases exon 3 exclusion, but to a lesser extent. These results indicate that the splicing of DARS2 exon 3 is regulated in a cell-type-specific manner: inclusion of exon 3 is less efficient in different neuronal cells than in other neural and in non-neural cells.
Mitochondrial defects tend to preferentially affect tissues with a high energy demand, the highest of which is found in the nervous system. The results of the present study show that two factors contribute to the selective vulnerability of tracts within the nervous system in LBSL. First, the mutations in intron 2 have a larger effect on exon 3 exclusion in neural cells and this effect is most pronounced in neuronal cells. Secondly, the correct inclusion of exon 3 in the normal mtAspRS mRNA occurs less efficiently in neural cells than in other cell types, with again the largest effect in neuronal cell types. The combined result of these two effects may explain the selective vulnerability of specific tracts seen in LBSL.
Most likely, these phenomena are related to each other and reflect differences in the abundance of splicing factors. Several brain-specific splicing factors have been identified, such as nPTB, NOVA1, NOVA2 and Hu/Elav proteins , that could account for splicing differences between non-neural and neural cells. In addition, in proliferating and post-mitotic mouse brain cells region- and cell-type-specific expression of many RNA-binding proteins has been observed . The selective vulnerability of specific white matter tracts found in LBSL may be explained by differences in expression of splicing factors between different brain regions and cell types.
A few LBSL patients have two missense mutations (four out of 100) . In these patients, the selective vulnerability of specific tracts could be explained by a general reduction of mtAspRS activity due to the missense mutations and a further reduction of the expression level in neuronal cells owing to less efficient splicing.
Cell-type-specific differences in a splicing defect have also been found in familial dysautonomia, which is caused by mutations in the gene IKBKAP. In this case a common mutation is found at the 5′ end of the intron: a reduced number of T-residues near the donor splice site of intron 20 of this gene. The relative wild-type over mutant IKBKAP RNA levels were highest in cultured patient lymphoblasts and lowest in post-mortem central and peripheral nervous tissues, suggesting that reduced production of full-length IKBKAP mRNA from the allele with the common mutation underlies the selective degeneration of sensory and autonomic neurons in familial dysautonomia .
Because almost all LBSL patients have a mutation just upstream of exon 3, increasing the splicing efficiency of this exon would be an interesting target for treatment of LBSL. We designed an antisense oligonucleotide using a method described previously , which is directed against exon 3 and has a tail that attracts splicing enhancers. The oligonucleotide was indeed able to increase the amount of product with exon 3, indicating that it is possible to increase the splicing efficiency of this exon. In patients this could lead to increased wild-type mtAspRS levels, which makes this splicing event an interesting target for treatment.
cytochrome c oxidase IV
Dulbecco's modified Eagle's medium
eukaryotic initiation factor 2α
enhanced yellow fluorescent protein
fetal bovine serum
green fluorescent protein
human embryonic kidney
leukoencephalopathy with brain stem and spinal cord involvement and lactate elevation
mitochondrial aspartyl-tRNA synthetase
astrocytic NT2 cell
neuronal NT2 cell
PBS containing 0.1% Tween 20
yellow fluorescent protein
Laura van Berge performed the experiments and wrote the paper. Stephanie Dooves, Carola van Berkel and Emiel Polder performed the experiments. Marjo van der Knaap contributed to conceptualization of the study and critical revision of the paper. Gert Scheper supervised and designed the study, and contributed to revision of the paper.
We thank the patients and their families. We also thank J. van Horssen (VU University Medical Center) for providing us with the HOG and MO3.13 cell lines.
This work was supported by the Prinses Beatrix Fonds [grant number WAR07/31].