Introduction

Krabbe disease, also known as globoid cell leukodystrophy is one of the autosomal-recessive lysosomal storage disorders (LSDs) resulting from a deficiency of galactocerebrosidase (GALC) activity [1]. The GALC protein precursor (immature GALC) is an 80-kd protein comprising 669 amino acids, which is processed into the N-terminal fragment (50 kD) and C-terminal fragment (30 kD) [2,3,4], as mature GALC, presumably in the lysosome. We previously found that pathogenic missense mutations result in conformational changes in the GALC precursor, thereby disrupting its trafficking to lysosomes [5]. Such mutant proteins are localized in the endoplasmic reticulum (ER) or Golgi apparatus [6, 7]. where GALC hydrolyzes both galactocerebroside and psychosine (galactosylsphingosine), which are specifically cytotoxic to oligodendrocytes and Schwann cells, respectively [8, 9]. Therefore, Krabbe disease is caused by demyelination in both the central and peripheral nervous systems, with the accumulation of mainly psychosine [10]. Phenotype–genotype correlations can be seen with some mutations, such as c.635–646delinsCTC, p.P302A, p. D528N, and p.T652P for infantile onset and p.G270D and p.L618S for adult onset Krabbe disease. These findings indicate that Krabbe disease may involve the accumulation of GALC proteins with different conformational changes depending on the mutation, and in various organelles, in addition to the cytotoxicity of psychosine. Proteins that mature with folding or intra/intermolecular disulfide bond formation in the ER, are subsequently transported to their destinations. However, if proteins fail to adopt their proper conformation, they accumulate in the ER lumen and disrupt ER homeostasis, which leads to a condition referred to as ER stress. To maintain homeostasis in the ER, a quality control system exists to determine whether newly synthesized proteins are folded properly. If abnormal proteins are detected within the ER, cells activate the unfolded protein response (UPR) via three ER transmembrane proteins, namely, activating transcription factor 6 (ATF6), inositol-requiring kinase 1 (IRE1), and protein kinase regulated by RNA-like ER kinase (PERK) the main sensors of the UPR in eukaryotes [11].

In the steady-state, IRE1 and PERK are inactivated by their binding with 78 kDa glucose-regulated protein (GRP78) (also known as Binding immunoglobulin protein), which acts as an ER chaperone. The accumulation of unfolded/misfolded proteins evokes GRP78 dissociation from these proteins, followed by the subsequent activation of the three receptors, which transmit downstream signals. ATF6 translocates to the Golgi upon conditions of ER stress, where it is cleaved by site 1 and site 2 proteases to release the soluble N-terminal fragment. Translocation of this fragment to the nucleus induces the expression of a spectrum of UPR mediators, including ER chaperones, as typified by GRP78, and enhances the proper folding of proteins in the ER. IRE1 dimerizes after dissociation of GRP78 followed by autophosphorylation of the kinase domain and activation of the RNase domain. The activated RNase domain of IRE1 catalyzes the removal of a small intron from the mRNA of the gene encoding X-box-binding protein (XBP1), resulting in the production of spliced XBP1 proteins (XBP1-s), which indicates the indirect activation of the IRE1 pathway [12, 13] and subsequently inducing apoptosis. PERK-induced phosphorylation of eukaryotic initiation factor 2α (eIF2α) leads to translational attenuation [14, 15]. PERK also dimerizes or oligomerizes after the dissociation of GRP78 followed by autophosphorylation of the kinase domain. Activated PERK specifically phosphorylates eIF2α, which inhibits DNA transcription, resulting in a reduced level of new protein synthesis in the ER. C/EBP (CCAAT/enhancer binding protein) homologous protein (CHOP), which is a common downstream factor of the PERK-ATF4 and ATF6 pathway, activates proteins, such as DNA damage-inducible gene 34 and Bcl-2 interacting mediator of cell death. In addition to these pathways, ER chaperones, such as GRP78 or Homocysteine-responsive endoplasmic reticulum-resident ubiquitin-like domain member 1 (HERPUD1), which bind unfolded/misfolded proteins or retrotranslocate them to the cytoplasm, also play a role in the UPR.

In LSDs, including GM1-gangliosidosis, neural ceroid lipofuscinosis, Tay-Sachs disease, and Gaucher disease, the UPR pathway is thought to play a role [16,17,18,19,20]. However, the role of UPR in Krabbe disease has not been investigated previously. Therefore, in the present study, we used COS-7 cells expressing pathogenic mutations in which phenotype–genotype correlations are apparent, and used cultured SFs derived from patients with Krabbe disease, to investigate whether UPR activation occurs in Krabbe disease.

Materials and methods

Cloning and construction of expression vectors of human GALC

Wild-type human GALC cDNA, containing a fragment from 19-bp upstream from the first ATG translation initiation site [21] to the AseI site (178-bp downstream from the stop codon) was constructed in a pSVL vector (GE healthcare, Amersham, UK), and six mutations including c.635–646delinsCTC, p.G270D, p.P302A, p.D528N, p.L618S, and p.T652P were introduced by polymerase chain reaction (PCR), and the sequence of all constructs were confirmed by DNA sequencing, as previously described [5].

Transfection and cell culture

COS-7 cells were prepared on the day prior to transfection. Briefly, both 2.7 μL of TransFectin™ Lipid Reagent (Bio-Rad Laboratories K.K., Tokyo, Japan) and 0.7 μg of plasmids, including the empty vector (mock), wild-type GALC cDNA, and GALC mutant constructs were diluted using Dulbecco’s Modified Eagle Medium (Life Technologies Co., Carlsbad, CA) with 10% fetal bovine serum (referred to as regular medium) to a final volume of 200 μL according to the manufacturer’s instructions, and then mixed together following incubation at room temperature for 20 min. For the cotransfection of two types of GALC mutant constructs, 0.35 μg of each plasmid was added. Then, 0.2–0.25 × 106 cells were added to the incubated solution to a final volume of 1000 μL with regular medium to be seeded in 12-well plates (Greiner Bio-one, Kremsmünster, Austria). Transfection reactions were performed for 48 h in a 5% CO2 incubator at 37 °C, and finally cells were collected for RNA isolation.

Patients

This study included four Japanese patients. Their diagnosis was based on clinical symptoms, neurological findings, laboratory confirmation of low GALC activity in lymphocytes, and pathological mutations. The mutations found in three of the patients had been identified previously [5], and one was a newly confirmed mutation (Table 1). GALC mutations in the present report are described following the nomenclature reported previously [22, 23].

Table 1 Expression of each UPR gene in patient SFs, and patient phenotypes

Genomic DNA analysis by direct sequencing

DNA was prepared by standard methods from patient lymphocytes. PCR reactions and direct sequencing were conducted as previously described [23].

Patient SFs

SFs were established from all four patients. The normal human dermal fibroblast cell line was purchased from Gibco (Thermo Fisher Scientific, MA, USA). SFs were cultured in regular medium containing 1% antibiotic-antimycotic solution in a 5% CO2 incubator at 37 °C for 48 h prior to cDNA extraction. The number of passages of all SFs used was <10 times.

Measurement of GALC activity

Enzyme activity was measured using 6-hexadecanoylamino-4-methylumbelliferyl-β-D-galactopyranoside (HM-gal) (Slater and Frit Ltd, Norwich, UK) as an artificial fluorescence substrate following the Wiederschain protocol [24].. Briefly, collected cells were sonicated and incubated with the substrate in citrate-phosphate (CP) buffer (pH 4.2) at 37 °C for 4 h and fluorescence (excitation at 385 nm/emission at 450 nm) was measured using a microplate reader. Enzyme activity was calculated as nmol/h/mg protein.

RNA extraction and DNA preparation

Total RNA was extracted from cells using NucleoSpin® RNA (Takara, Kusatsu, Japan) and subsequently cDNA was extracted using ReverTra Ace qPCR RT Master Mix with gDNA Remover (TOYOBO, Osaka, Japan) according to the manufacturer’s instructions. This kit contained a DNase step. cDNA products were frozen at −20 °C.

Real-time reverse transcription-PCR (RT–PCR)

Real-time quantitative RT–PCR (qPCR) was performed using SYBR Green PCR Master Mix (TOYOBO) and ABI PRISM 7900HT with cDNA (equivalent to 30 ng of total RNA for the analysis of CHOP, ATF6, GRP78, and HERPUD1 and 10 ng for the analysis of GAPDH. The primer concentration was 0.45 nM in a reaction volume of 15 µL. Each reaction was performed in triplicate or twice in triplicate. The thermal cycling parameters were as follows: 95 °C for 1 min, 40 cycles of 95 °C for 15 s and 60 °C for 45 s, and a final cycle at 95 °C for 15 s, 60 °C for 15 s and 95 °C for 15 s. Relative expression of mRNA for the target genes was performed by the comparative CT method using the GAPDH gene as a control. Data are presented as means ± SE. P-values were calculated using analysis of variance and a p-value of less than 0.05 was considered to indicate a statistically significant difference between two groups. Sense and antisense primer sequences for RT–PCR amplification are shown in Supplementary Table.

XBP-1 splicing assay

After first-strand cDNA synthesis, PCR was carried out using KOD Plus (TOYOBO) for 94 °C (two min), 35 cycles at 98 °C (10 s), 48 °C (30 s), and 68 °C (30 s) for XBP-1 and 94 °C (2 min), 30 cycles at 98 °C (10 s), 55 °C (30 s), and 68 °C (16 s) for GAPDH containing 20 ng of each cDNA obtained from both COS-7 cells and SFs. PCR reactions were subjected to electrophoresis through a 2% agarose gel to distinguish the unspliced (XBP-1u, 473 bp) from the spliced band (XBP-1s, 447 bp). Splicing rate was calculated by the ratio of XBP-1s to XBP-1u band intensities using Image Lab software (Bio-Rad Laboratories K.K., Tokyo, Japan). Sense and antisense primers used were 5′-ccttgtagttgagaaccagg−3′ and 5′-ggggcttggtatatatgtgg−3′.

Induction of UPR compounds in transfected COS-7 cells by tunicamycin and thapsigargin treatment

Cytotoxicity assays were performed using tunicamycin (TN) and thapsigargin (Tg). In each experiment, COS-7 cells transfected with empty vector (mock), wild-type GALC cDNA, and GALC mutant constructs (0.2–0.25 × 106 cells/well/1000 μL regular medium) were plated at time zero in 12-well plates. After 48 h, 1 μL of TN or Tg dissolved in deionized-distilled H2O (dd H2O) was added to different wells at final concentrations of 1 μg/mL for TN and 1 μM for Tg. In all the experiments, 1 μL ddH2O was added to the control wells. Cells were then incubated for 6 h in a humidified atmosphere containing 5% CO2 at 37 °C. After this incubation, cells were collected for RNA isolation and subsequent RT–PCR analysis.

Results

GALC enzyme activities in COS-7 cells and patient SFs

We prepared lysates of COS-7 cells transfected with each expression construct and of patient SFs, and measured their GALC enzyme activities. GALC enzyme activities of cells expressing GALC with infantile onset mutations were similar to mock-transfected cells, whereas cells expressing GALC with adult onset mutations showed higher enzyme activity, and cells expressing the G270D mutant showed almost normal enzyme activity (Fig. 2a). On the other hand, unlike COS-7 cells, no apparent association was observed between enzyme activity and phenotype in patient SFs (Fig. 3a).

Fig. 1
figure 1

Effect of the expression of each UPR protein in COS-7 cells. a mRNA levels of CHOP, ATF6, GRP78, and HERPUD1 were compared upon treatment with a lipid agent and mock transfection. b Splicing rate of XBP-1 under each transfection condition was calculated by the ratio of the band intensities of XBP-1s to XBP-1u (XBP-1s / [XBP-1s + XBP-1u]). The DNA templates used were the same as those used in a. c mRNA levels of CHOP, ATF6, GRP78, and HERPUD1 with TN or Tg treatment were compared to no treatment with transfection. (d) Splicing rate of XBP-1 under each treatment condition was calculated by the ratio of the band intensities of XBP-1s/[XBP-1s + XBP-1u] (e) Electrophoresis of XBP-1u and XBP-1s under different treatment conditions of lipid agents, vehicle, and TN/Tg. Splicing ratios were calculated using this figure. Data are shown as the mean ± SE (n = 3) *p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.001 vs. no lipid reagent, no vehicle, or no treatment group. TN tunicamycin, Tg thapsigargin, NC negative control

Fig. 2
figure 2

a GALC enzyme activities measured using HM-gal substrate in COS-7 cells expressing wild-type or mutant GALC. be Effect of the expression of GALC mutants on the levels of various UPR genes in COS-7 cells. (mock, n = 6; other mutations, n = 3) (b) mRNA levels of CHOP, ATF6, GRP78, and HERPUD1 upon the expression of each mutant were calculated as a ratio to mock transfection, which was set as 1 (c) Splicing rate of XBP-1 upon the expression of each mutant was calculated by the ratio of the band intensities of XBP-1s to XBP-1u, as in Fig. 1d. d, e mRNA levels of CHOP, ATF6, GRP78, and HERPUD1 with TN (c) or Tg (d) treatment were compared with mock transfection. The mRNA level was calculated as a ratio with the mock transfection with treatment group considered as 1. *p < 0.05, **p < 0.01, and ***p < 0.005 vs. the mock transfection group (a, b) or mock transfection with treatment (c, d)

Effect of transfection procedures on the expression of UPR target genes in COS-7 cells

As some transfection procedures (e.g., sonoporation) induce the UPR [25], we analyzed the effects of lipid transfection reagents on the expression of UPR target genes. No UPR target genes were increased or decreased by the transfection reagent without vector compared with no treatment; however, CHOP, ATF6, and HERPUD1 were activated by transfection reagents containing the mock vector. Both ATF6 mRNA and the level of XBP-1 mRNA splicing did not show any changes in all conditions. (Fig. 1a, b, e).

Fig. 3
figure 3

a GALC enzyme activities measured using HM-gal substrate in patient SFs. bd Levels of UPR genes in patient SFs. b mRNA levels of CHOP, ATF6, GRP78, and HERPUD1 were compared with normal SFs as a standard. c Splicing rate of XBP-1 in the SFs of each patient calculated in Fig. 1d. Statistical analysis was calculated using normal SFs as a standard. The same DNA templates as in Fig. 3b were used. d Electrophoresis of XBP-1u and XBP-1s. Splicing ratios were calculated using this figure. (n = 3) *p < 0.05, **p < 0.01, and ****p < 0.001 vs. normal SF

Effect of TN and Tg on the expression of UPR target genes in COS-7 cells

TN and Tg, which are UPR-inducing compounds, inhibit protein glycosylation in the ER and ER calcium ATPase activity, respectively. These UPR compounds induced a significant increase in the mRNA levels of GRP78 and HERPUD1, XBP-1 mRNA splicing levels, as well as a mild increase in CHOP mRNA compared with no treatment in mock-transfected COS-7 cells (Fig. 1c, d, e).

Effect of mutant GALC expression on UPR target gene expression

Expression of the T652P GALC mutant showed a significant increase in the mRNA levels of CHOP, GRP78, and HERPUD1, as well as the level of XBP-1 mRNA splicing (Fig. 2b, c, Table 2), whereas the expression of L618S showed a significant decrease in the mRNA level of CHOP as well as the level of XBP-1 mRNA splicing. The expression of c.635–646delinsCTC/T652P and P302A/L618S showed a significant increase in XBP-1 mRNA splicing (Fig. 2c). Wild G270D, and D528N inhibited the mRNA levels of CHOP.

Table 2 Effect of the expression of various GALC mutants on UPR gene expression in COS-7 cells

Effect of tunicamycin and thapsigargin on mutant GALC transfection

Under TN treatment, compared with mock transfection, P302A/L618S activated the response of CHOP, ATF6, and HERPUD1, and inhibited the response of GRP78, whereas T652P and c.635–646delinsCTC/T652P activated the response of CHOP and inhibited the response of GRP78 and HERPUD1, and P302A inhibited the response of GRP78 and HERPUD1, and L618S inhibited all of these four UPR factors (Fig. 2d). On the other hand, under Tg treatment, c.635–646delinsCTC/T652P and P302A/L618S activated the response of GRP78 and HERPUD1, T652P inhibited the response of GRP78 and HERPUD1, and P302A inhibited the response of GRP78 but not HERPUD1, which was different from TN treatment (Fig. 2e). These changes were all statistically significant, although most of them changed by less than two-fold.

UPR levels in patient SFs

c.635–646delinsCTC/c.635–646delinsCTC showed an increase in the mRNA levels of CHOP, ATF6, and HERPUD1, and c.635–646delinsCTC/T652P and P302A/L618S showed an increase in the mRNA levels of CHOP, ATF6, and HERPUD1, as well as XBP-1 splicing, and L618S/L618S showed an increase in CHOP and the inhibition of GRP78 (Fig. 3b, c, d, Table 1). An increasing ratio of ATF6 showed statistical significance but was not obviously different compared with CHOP and HERPUD1.

Discussion

The major finding of the current study is that UPR activation in Krabbe disease has selectivity depending on the mutation.

In COS-7 cells, T652P activated multiple UPR factors, such as CHOP, GRP78, HERPUD1, and XBP1s. This indicates activation of the ER chaperones and each downstream pathway of IRE1 and PERK. The PERK pathway works as a controller of DNA transcription and protein synthesis, and the IRE1 pathway works as an inducer of ER-associated degradation by increasing XBP1 mRNA splicing. Therefore, it is suggested that T652P inhibits the synthesis of mutated proteins and translocates them from the ER, thereby inducing apoptosis. With the other mutations, only the apoptosis-associated pathway is involved, mainly as shown for the activation of XBP1.

However, we speculated that activation of the UPR pathway may partially contribute to the phenotype, because the cytotoxicity of psychosine may be the main factor determining the severity of Krabbe disease. Furthermore, the UPR is known to be composed of several branches or arms, involving the three ER stress sensors ATF6, IRE1, and PERK; however, several downstream factors are common to the branches and different branches activate similar targets but stimulate alternative effects, including apoptosis, protein folding, and anti-oxidant activity, etc. Therefore, different activation in different branches might cause similar cellular effects. SFs expressing the L618S mutant of GALC activated only CHOP, which suggested that the pathway of translational attenuation via eIF2α and the induction of apoptosis were mainly activated. On the other hand, SFs expressing the c.635–646delinsCTC mutant of GALC, which contains in-frame insertions/deletions, activated HERPUD1, CHOP, and ATF6, which suggested that the mechanism of cytoplasmic translocation of the mutated GALC proteins were activated in addition to translational attenuation and apoptosis induction. Most GALC proteins with the L618S mutation are trafficked to the lysosomes [26, 27], whereas those with the c.635–646delinsCTC mutation are trafficked ineffectively; therefore, HERPUD1 may be activated depending on the amount of mutated GALC in the ER, whereas CHOP is activated independent of the mutations. The activation pathways observed in the compound-heterozygous GALC mutant may be a combination of that observed for each mutant. Mutated GALC proteins are also partially trafficked to lysosomes and are subsequently processed to have some enzyme activity depending on the mutations. Therefore, an enhancement of the translocation of mutated GALC proteins to the cytoplasm may lead to a decrease in GALC trafficked to the lysosomes, and therefore also a decrease in residual enzyme activity, which may result in more severe phenotypes. It is speculated that the ER chaperon pathway in addition to the ATF6, PERK, and IRE1 pathways may be activated by the expression of GALC with the missense mutation with low residual enzyme activity. In addition the ATF6, PERK, and IRE1 pathways without the ER chaperon pathway may be activated by the expression of GALC with the missense mutation with high residual GALC enzyme activity. In addition, none of the pathways may be activated by expression of GALC with the nonsense mutation owing to nonsense mRNA decay [28].

There are three main limitations in this study; the cell types used, analysis of the polymorphism in patient SFs and in vivo model animals. First, as UPR activation differs in each cell type [29], the results of transfected COS-7 cells may not apply to clinical conditions. The ATF6, IRE1, and PERK pathways are the main pathways of the UPR and are expressed ubiquitously. On the other hand, old astrocyte-specifically induced substance (OASIS) family proteins, which are ER stress sensors, including LUMAN/CREB3, OASIS/CREB3L1, BBF2H7/CREB3L2, CREBH/CREB3L3, and CREB4/AIbZIP/CREB3L4 demonstrate tissue-specific expression [30]. Although we did not analyze these factors in the current study, we speculate that they may have different effects on UPR activation in COS-7 cells and SFs. Second, we did not coexpress proteins with polymorphisms in COS-7 cells. Some polymorphisms change the trafficking efficiency of proteins to the lysosome [7, 26]. Third, we did not analyze in vivo model animals. Second and third limitations require further investigation in the future [31, 32].