Functional analysis of mutations in UDP-galactose-4- epimerase (GALE) associated with galactosemia in Korean patients using mammalian GALE-null cells You-Lim Bang 1, *, Trang T. T. Nguyen 1, *, Tram T. B. Trinh 1, *, Yun J. Kim 1 , Junghan Song 2 and Young-Han Song 1 1 Ilsong Institute of Life Science, Hallym University, Anyang, Korea 2 Department of Laboratory Medicine, Seoul National University Bundang Hospital, Gyeonggi-do, Korea Galactose is metabolized by three enzymes: galactokin- ase, galactose 1-phosphate uridyltransferase and UDP- galactose-4-epimerase (GALE; EC 5.1.3.2) [1]. Defects in any of these enzymes lead to the autosomal reces- sive disorder, galactosemia [2]. The majority of patients with GALE-deficiency galactosemia (MIM 230350) are identified as having a clinically benign ‘peripheral’ condition. These patients show defective GALE activity only in circulating red and white blood cells, and the condition is relatively com- mon in African Americans [3] and in Japanese [4]. The clinically severe ‘generalized’ form of GALE-deficiency, galactosemia, is extremely rare and enzyme activity is defective in all tissues examined. These patients show growth retardation and developmental delay [5]. ‘Inter- mediate GALE deficiency’ has also been applied to patients who exhibit partial impairment of GALE activity in nonperipheral cells, such as lymphoblasts [6]. The molecular mechanisms that distinguish these subtypes of GALE-deficiency galactosemia have not been identified. GALE is a member of the short-chain dehydroge- nase ⁄ reductase superfamily. It is widely distributed in nature and has been isolated and characterized in several species [7–9]. The 3D structure of Escherichia coli [10], Trypanosoma brucei [11], Saccharomyces Keywords galactosemia; mutation; protein aggregation; UDP-galactose-4-epimerase; unstable protein Correspondence Y H. Song, Ilsong Institute of Life Science, Hallym University, 1605-4 Gwanyang-dong, Dongan-gu, Anyang, Gyeonggi-do 431 060, Korea Fax: +82 31 388 3427 Tel: +82 31 380 1897 E-mail: ysong@hallym.ac.kr *These authors contributed equally to this work (Received 15 December 2008, revised 19 January 2009, accepted 21 January 2009) doi:10.1111/j.1742-4658.2009.06922.x Galactosemia is caused by defects in the galactose metabolic pathway, which consists of three enzymes, including UDP-galactose-4-epimerase (GALE). We previously reported nine mutations in Korean patients with epimerase-deficiency galactosemia. In order to determine the functional consequences of these mutations, we expressed wild-type and mutant GALE proteins in 293T cells. GALE E165K and GALE W336X proteins were unstable, had reduced half-life, formed aggregates and were partly degraded by the proteasome complex. When expressed in GALE-null ldlD cells GALE E165K , GALE R239W , GALE G302D and GALE W336X had no detectable enzyme activity, although substantial amounts of protein were detected in western blots. The relative activities of other mutants were lower than that of wild-type. In addition, unlike wild-type, GALE R239W and GALE G302D were not able to rescue galactose-sensitive cell prolifera- tion when stably expressed in ldlD cells. The four inactive mutant proteins did not show defects in dimerization or affect the activity of other mutant alleles identified in patients. Our observations show that altered protein stability is due to misfolding and that loss or reduction of enzyme activity is responsible for the molecular defects underlying GALE-deficiency galactosemia. Abbreviations BrdU, bromodeoxyuridine; CHO, Chinese hamster ovary; EGFP, enhanced green fluorescent protein; GALE, UDP-galactose 4¢-epimerase; UDP-galNAc, UDP-N-acetylgalactosamine; UDP-glcNAc, UDP-N-acetylglucosamine. 1952 FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS cerevisiae [12] and human [13] GALE has been determined by high-resolution X-ray crystallography. GALE from all species examined to date, with the exception of the bacterium, Aeromonas hydrophila functions as a dimer [14]. Since the cloning of human GALE in 1995 [15], 22 mutations have been identified in the GALE coding sequences of GALE- deficiency galactosemia patients (Fig. S1A) [5,6,16– 21]. To understand the functional consequences of the mutations, biochemical studies have been carried out by expressing wild-type and mutant human GALE in bacteria or in yeast null for the GALE homologue, gal10p. Among the 12 mutants charac- terized to date, the most severe defects in GALE protein were observed with mutants G90E, V94M and L183P. Mutants G90E and V94M show a reduc- tion in turnover number by factors of 800 and 30, respectively [22], and are unable to restore galactose- sensitive growth when expressed in epimerase-null yeast [18]. Even though the turnover number of mutant L183P is lower than that of wild-type GALE by a factor of three [22], the relative abundance of this mutant protein was greatly reduced in bacteria [22] and yeast [23], thereby decreasing the relative enzymatic activity to 4% of wild-type protein [23]. Mutants N34S, P293L and G319E show the mildest defects in turnover number (reduced by factors of 1.1, 1.8 and 1.2, respectively) [22]. The remainder of the mutant alleles characterized to date do not show substantial defects in the restoration of galactose- sensitive growth in yeast [21,24] and show a mild reduction in turnover number (decrease by a factor of 2.4–7.2) [22]. None of the mutant human GALE tested to date show defects in dimerization [21,22] and mutant alleles N34S and L183P are dominant negative with respect to each other [23]. These results suggest that not all mutant alleles found in patients cause functional defects and underscore the importance of functional studies on the mutant proteins. We previously described nine GALE mutations in Korean patients with GALE-deficient galactosemia [20]. These GALE mutations were novel with the exception of R335H, which had been shown earlier to cause mild defects in enzyme activity [19,22]. In this study we generated expression plasmids containing wild-type and mutant GALE cDNA from Korean galactosemia patients and expressed the constructs in 293T or GALE null ldlD cells [25] to examine the functional consequences of the mutations. Our study showed that GALE E165K , GALE R239W , GALE G302D and GALE W336X exhibited no detectable enzyme activ- ity, and GALE E165K and GALE W336X formed protein aggregates and showed greatly reduced stability. More- over, unlike wild-type proteins, the two stable inactive mutants, GALE R239W and GALE G302D , were unable to rescue defects in galactose-sensitive cell prolifera- tion, confirming that these mutants are not functional in vivo either. Results Protein stability To investigate the functional properties of the mutant GALE proteins, full-length wild-type or mutant GALE cDNAs were cloned into expression plasmids containing N-terminal myc or FLAG epitopes. When transiently expressed in 293T cells, steady-state levels of myc– GALE E165K and myc–GALE W336X proteins were signifi- cantly reduced compared with wild-type, even though the transfection efficiencies, based on the expression level of cotransfected enhanced green fluorescent protein (EGFP), were comparable (Fig. 1A). The steady-state levels of FLAG-tagged GALE E165K and GALE W336X were similarly reduced (data not shown). The steady-state protein level can be decreased by mutations because of reductions in either RNA or pro- tein stability. In order to distinguish between the two possibilities, we performed semi-quantitative RT-PCR using GALE transgene-specific primers and EGFP as both an internal control and control for transfection efficiency. The amount of PCR product increased upon addition of threefold more cDNA template, confirming that a plateau was not reached under these PCR con- ditions. The levels of GALE E165K and GALE W336X transcripts were not significantly different from that of wild-type, suggesting that these mutations did not alter RNA stability (Fig. 1B). We tested the stability of the mutant proteins by determining protein half-life using cycloheximide. Wild-type GALE produced stable protein with a half- life of > 8 h. Unlike that of wild-type, the half-life of GALE E165K and GALE W336X proteins decreased to 2.0 and 2.1 h, respectively (Fig. 1C). In order to determine whether the mutant proteins are degraded by the proteasome complex, we investigated the effect of the proteasome inhibitor, MG132 on protein half-life. In the presence of MG132, the half-lives of GALE E165K and GALE W336X were increased signi- ficantly up to 8.7 and 6.7 h, respectively, but MG132 treatment does not appear to restore the steady-state protein level to that of wild-type (Fig. 1D). These results suggest that GALE E165K and GALE W336X are unstable proteins which are degraded to some extent by the proteasome complex. Y L. Bang et al. Functional analysis of GALE mutation in ldlD cells FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS 1953 Subcellular localization We performed indirect immunofluorescence to deter- mine whether the subcellular localization of GALE is altered by the mutations. Wild-type GALE showed a pattern of diffuse cytoplasmic immunoreactivity when expressed in 293T cells (Fig. 2A). Mutant proteins showed a distribution pattern similar to wild-type (data not shown), with the exception of GALE E165K and GALE W336X , which formed perinuclear A B C D Fig. 1. GALE E165K and GALE W336X proteins have shorter half-lives and are degraded by proteasome complex. (A, B) Empty vector ()), wild- type (WT) or mutant myc–GALE expression plasmids were co-transfected into 293T cells with EGFP expression construct as a control for transfection efficiency. Forty hours after transfection, cell lysates or total RNA were prepared and western analysis (A) or RT-PCR (B), respectively, were carried out. (A) Myc–GALE and EGFP were detected using antibodies against myc and GFP by western analysis. (B) Equal amount of total RNA prepared from transfected cells was reverse transcribed and 1 or 3 lL of cDNA was used to amplify EGFP and myc–GALE. (C) In order to determine the half-life of myc–GALE wt , myc–GALE E165K and myc–GALE W336X proteins, 293T cells transiently transfected with each myc–GALE expression plasmid were equally divided among the wells of a six-well plate and treated with 50 lgÆmL )1 cycloheximide (CHX). At the indicated time intervals, the cells were harvested and analyzed by western blotting. The amount of extract from cells expressing mutant GALE was increased to reach a detectable level of protein. Tubulin serves as a loading control. The relative amount of mutant protein was calculated based on the ratio of the band intensities of myc-GALE to tubulin in the western blot and indicated below the blot with 100 set as the zero time point. The relative levels of myc–GALE protein at each time point were plotted using a complete linear least squares curve-fit algorithm, and the time point at which GALE levels decreased to 50% of their original value was determined and reported as the half-life. (D) In parallel experiments, cells were treated with 25 l M of the proteosome inhibitor, MG132, 1 h prior to and for the duration of cycloheximide treatment. Functional analysis of GALE mutation in ldlD cells Y L. Bang et al. 1954 FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS aggregates. The appearance of the aggregates varied and presented as a single, large perinuclear sphere, sev- eral smaller spheres or perinuclear aggregates with additional, multiple foci throughout the cytoplasm (Fig. 2A). In some cells, the mutant proteins were also observed in the cytoplasm, as seen for the wild-type protein. Similar staining patterns were observed in other human cell lines, including hepatoma Hep3B cells and fibroblast 2fTGH cells (data not shown), sug- gesting that mislocalization of the mutant proteins was not cell-type-specific. Aggresomes are structures formed in the presence of misfolded proteins and con- tain components of the protein degradation and refold- ing machinery, including proteasome complexes and members of the Hsc70 and Hsp70 families of proteins [26,27]. Because the morphological appearance of aggresomes is similar to the aggregates seen in cells expressing mutant GALE, we investigated whether Hsc70 is recruited to the GALE E165K and GALE W336X aggregates. Unlike cells expressing wild-type GALE, cells transfected with GALE E165K and GALE W336X showed Hsc70 redistribution and Hsc70 colocalized with the GALE proteins, suggesting that these mutant proteins indeed become incorporated into aggresomes. GALE activity towards UDP-galactose To determine the functional consequences of the muta- tions, we expressed wild-type and mutant GALE in the GALE-null cell line, ldlD, derived from Chinese hamster ovary (CHO) cells [25,28]. As shown previ- ously, ldlD cells did not exhibit any GALE activity (data not shown), whereas cells expressing myc- or FLAG-tagged wild-type GALE were able to catalyze the conversion of UDP-galactose into UDP-glucose. Enzyme activities of myc-tagged GALE E165K , GALE R239W , GALE G302D and GALE W336X were unde- tectable even though substantial amounts of the proteins were detected (Fig. 3). Results were the same for FLAG-tagged constructs (data not shown). It is interesting to note that the steady-state levels of GALE E165K and GALE W336X proteins in ldlD cell wt W336XW336X E165KE165KE165K Hsc70 Merge GALE B A wt E165K W336X Fig. 2. GALE E165K and GALE W336X proteins form aggresomes in the cell. (A) Wild-type and mutant myc–GALE proteins were expressed in 293T cell and immunohistochemistry was carried out to visualize GALE proteins using mouse anti-myc mAb and rhoda- mine-conjugated anti-mouse IgG (red). Nuclei are stained with DAPI (blue). (B) 293T cells were transfected with FLAG–GALE and stained for FLAG–GALE (red) and Hsc70 (green). n = 3 80 100 120 40 60 0 20 wt A25V R40C D69E ∗∗∗∗ Relative enzyme activity (%) E165K R239W G302D W336X R169W R335H myc-GALE Tubulin Fig. 3. Relative enzymatic activities of wild-type and mutant GALE proteins expressed in ldlD cells. Cell lysates prepared from ldlD cells transiently transfected with wild-type and mutant myc–GALE were analyzed for enzyme activity using UDP-galactose as sub- strate. Enzyme activity was normalized against the relative protein abundance calculated from the intensities of myc–GALE and tubulin bands in the western blot. A representative blot is shown. Values represent the mean and standard deviation of relative enzyme activities, with wild-type activity set as 100%. At least three trans- fections were carried out for each point. Asterisks signify no detectable enzyme activity. Y L. Bang et al. Functional analysis of GALE mutation in ldlD cells FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS 1955 lysates appear similar to that of wild-type, unlike what we observed in 293T cells (compare Figs 1A and 3). This is likely to be caused by differences in the time of harvest and the methods used to extract proteins (see Materials and methods for detail), because mutant protein levels were dramatically decreased when ldlD cell lysates were prepared similar to 293T cells (data not shown). The relative enzyme activities of the other mutants were also affected and decreased to 24–71% of the wild-type level (Fig. 3). Thus, the enzyme activi- ties of the E165K, R239W, G302D and W336X mutant proteins were not detectable, whereas the remainder of the mutants showed only mild defects. Restoration of galactose-sensitive proliferation defects of ldlD cells GALE-null ldlD cells show defects in cell proliferation when grown in the presence of galactose [28]. This sen- sitivity to galactose can be rescued by expressing wild- type GALE and the degree of rescue is dependent on the level of GALE enzyme activity. In order to test whether the four inactive mutants can restore galactose sensitivity, we generated stable cell lines expressing myc-tagged wild-type and mutant proteins. Although a similar number of colonies was analyzed for each mutant, we were not able to generate ldlD cell lines that stably expressed GALE E165K or GALE W336X . All stable cell lines used in this assay showed proliferation rates in the absence of galactose similar to that of ldlD cells. A bromodeoxyuridine (BrdU) incorporation assay was performed in the presence and absence of 0.25 mm galactose. The level of BrdU incorporated into each cell line treated with galactose was normal- ized relative to BrdU incorporation in the absence of galactose, which was set to 100% (Fig. 4). Similar to the previous report, wild-type CHO and ldlD cells incorporated 119% and 21% of the level of BrdU in the presence of 0.25 mm galactose. In the ldlD(myc– GALE) wild-type lines, BrdU incorporation was restored to 116% (wt #3), 88% (wt #11) and 52% (wt #8); and the level restored was dependent on the relative enzyme activity (29.0, 5.2 and 0.9, respectively) and protein level (17.9, 1.0 and 0.2, respectively). Because the relative enzyme activity of wt #11 was 5.2-fold higher than that of CHO cells and the level of BrdU incorporation was restored to 88%, we selected two independent stable lines for GALE R239W and GALE G302D , which expressed a similar or higher (0.8- to 2.0-fold) level of mutant GALE expression com- pared with wt #11. There was no detectable enzyme activity in extracts of cells stably expressing either of the two mutants and BrdU incorporation upon galac- tose treatment decreased to 21–33% of control values, confirming that the GALE R239W and GALE G302D proteins are non-functional in vivo as well. Dimerization and the dominant-negative effect In order to test whether the loss of enzyme activity of the four mutants is due to defects in dimer formation, we performed an immunoprecipitation assay. We transfected FLAG- and myc-tagged wild-type GALE alone or together into ldlD cells and immunoprecipitat- ed the protein with myc antibody. When increasing amounts of lysate from cells transfected only with FLAG–GALE were immunoprecipitated, no GALE bands were detected on western blots probed with anti- body against FLAG or myc (Fig. 5A). By contrast, FLAG–GALE was detected in the anti-myc immuno- precipitates of cell lysates expressing both myc–GALE and FLAG–GALE in a dose-dependent manner (Fig. 5A), confirming that dimerization of the GALE proteins had taken place. Similarly, GALE E165K , GALE R239W , GALE G302D and GALE W336X were able 160.0 n = 3 80.0 100.0 120.0 140.0 0.0 20.0 40.0 60.0 Relative BrdU incorporation (%) CHO ldlD wt #3 wt #1 1 wt #8 R239W #15 R239W #5 G302D #10 G302D #13 Tubulin myc-GALE 1.0 0.2 1.7 0.8 2.0 17.9 1.6 RP A REA (CHO:1) 0.9 29.0 5.2 0 0 0 0 Fig. 4. GALE R239W and GALE G302D are not able to restore galac- tose-sensitive cell proliferation when stably expressed in ldlD cells. Stable ldlD cells expressing varying amounts of myc–GALE wt , myc–GALE R239W and myc–GALE G302D were generated. Relative proliferation of these cells in the presence and absence of 0.25 m M galactose was determined by measuring BrdU incorporation. The level of BrdU incorporation into each cell line cultured in the absence of galactose was set to 100% and the corresponding results from galactose treated cells were calculated accordingly. Relative protein abundance (RPA) was calculated based on the ratio of intensity of the myc–GALE to tubulin bands in the western blot and normalized by setting the amount of GALE in ldlD(myc–GALE) wt #11 as 1.0. The western blot for determining the protein level in extracts derived from each cell line is shown. The relative enzyme activity (REA) was determined and normalized by setting the enzyme activity of CHO cell lysates as 1.0. Functional analysis of GALE mutation in ldlD cells Y L. Bang et al. 1956 FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS to form homodimers (Fig. 5B), suggesting that these mutant proteins exist as dimers, even though their enzymatic activity is severely compromised. The four mutant proteins were able to form heterodimers with the alleles found in the patient (Fig. 5C). Because GALE N34S and GALE L183P are dominant-negative with respect to each other [23], we investigated whether the four inactive mutants have a similar effect. When tested in ldlD cells, we could not observe such effect (Fig. S2). Discussion Previously, we identified nine mutations in Korean patients with GALE-deficiency galactosemia [20]. Our results on the functional consequences of these muta- tions reveal that they could be categorized into three groups. The first group includes E165K and W336X, which had no detectable GALE activity, formed protein aggregates and were degraded by proteasome complex. The second group, consisting of R239W and G302D, also exhibited no detectable enzymatic activ- ity, although the stability of the proteins was unaf- fected. The remaining mutants showed mild defects in enzyme activity with apparently wild-type protein stability. In this study, we utilized GALE null ldlD cells to test enzyme activity and in vivo function of GALE mutants. The ability to rescue defects in cell proliferation of galactose-treated ldlD cells correlated with the defects in enzyme activity. Interestingly, stable cells expressing mutants showing mild defects in enzyme activity behaved in a manner similar to the original CHO cells (data not shown), suggesting that the mild defects did not affect this phenotype. This is consistent with previous results in yeast showing a steep threshold relationship between growth rate in galactose and GALE activity [29]. W336X is the only nonsense mutation that has been identified in GALE-deficiency galactosemia patients. It is well known that mRNAs harboring premature ter- mination codons are rapidly degraded by the RNA surveillance mechanism known as nonsense-mediated mRNA decay [30]. In mammalian cells, premature stop codons generally need to be at least 55 nucleotides upstream of the last intron in order to trigger mRNA decay [30]. Because the W336X mutation generates a premature stop codon in the last exon, it is not likely to induce nonsense-mediated mRNA decay. In addi- tion, we could not detect changes in the mRNA stabil- ity caused by this mutation when the mutant protein was expressed from cDNA (Fig. 1B). When newly synthesized proteins are misfolded for a number of reasons, including missense mutation or lack of oligomeric assembly partners, they aggregate to form an aggresomal particle and are targeted for degradation by the ubiquitin–proteasome system [26]. The overall A B C Fig. 5. Capacity of wild-type and four mutant GALE proteins with undetectable enzyme activity to dimerize. (A) FLAG- and myc-tagged wild- type GALE expression plasmids were transfected individually or in combination into ldlD cells and immunoprecipitation was carried out with myc antibody using detergent-soluble cell lysates containing the amount of protein indicated. Immunoprecipitates (IP) were analyzed by wes- tern blotting (IB) with FLAG antibody, stripped, and reprobed with myc antibody. (B) In order to determine if the four mutant proteins are able to form homo-dimers, FLAG- and myc-tagged GALE constructs were co-transfected and analyzed as in (A). (C) FLAG–GALE was trans- fected into stable ldlD lines expressing myc-GALE as indicated and immunoprecipitation was performed and analyzed as in (A). Y L. Bang et al. Functional analysis of GALE mutation in ldlD cells FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS 1957 structure of aggresomes appears to vary depending on the nature of the aggregating protein and the cell type: a single sphere of diameter 1–3 lm or an extended ribbon- shaped structure [26]. It has been shown that the machinery for protein refolding and degradation, including such components as Hsc70, Hsp70, Hsp40 family of proteins and proteasomes are recruited to the aggresome to eliminate the aggregated material [31]. Aggregated proteins that can not be refolded by the chaperone system or degraded by the proteasomes are efficiently removed by autophagy [26]. The subcellular distribution, co-localization with Hsc70 and proteo- some-dependent degradation of GALE E165K and GALE W336X proteins support the theory that these mutant proteins are misfolded, form aggresomes and are degraded by the proteasome complex. Substitution of the negatively charged amino acid, glutamate, with the positively charged lysine (E165K) and deletion of the 13 C-terminal amino acids (W336X) probably dis- rupted normal protein folding. Because previous reports indicate that some protein aggregates have a toxic ‘gain of function’ activity, it will be interesting to explore whether aggregation of these GALE mutant proteins results in pathological symptoms that are related to tox- icity and independent of the symptoms caused by the absence of GALE activity. Consistent with this possibil- ity, we were not able to generate stable cell lines express- ing E165K and W336X mutant proteins, even though the possibility of very low steady-state levels of protein cannot be ruled out. Further studies are required to pursue this question. Unlike enzymes isolated from E. coli and yeast, mammalian GALE enzymes can convert UDP-N-acet- ylgalactosamine (UDP-galNAc) to UDP-N-acetylglu- cosamine (UDP-glcNAc), and are known to require exogenous NAD + [9]. We did not test the GALE mutations for enzyme activity on UDP-galNAc. Previ- ous reports show that GALE G90E and GALE V94M are defective on both substrates [18] suggesting the possi- bility that the above four mutants might have impaired activity on UDP-galNAc. It has been reported that purified GALE N34S requires more than a fourfold higher concentration of NAD + to achieve half-maxi- mal activity compared with the wild-type enzyme [23]. This result explains the observation that GALE N34S present in crude yeast extracts demonstrated only 70% of wild-type activity when assayed at 4 mm NAD + , whereas less-pronounced differences were observed when a higher concentration of NAD + was used [23]. We used 5 mm NAD + in the enzyme activity assay and it remains to be determined if any of the muta- tions found in Korean galactosemia patients also show increased dependence on exogenous NAD + . The results from this and previous reports identified seven amino acid residues with severe defects in enzyme activity and ⁄ or protein stability (Fig. S1A). When GALE sequences from eight species were aligned, approximately one third of the amino acids were strictly conserved in all species (Fig. S1B). Of 20 amino acids that have been characterized, none of the mutations with mild defects are associated with these conserved amino acid residues, with the exception of N34S. By contrast, all seven GALE mutants showing severe defects were identical in amino acid sequence among species (Fig. S1B), underscoring the importance of these amino acids in proper folding and catalytic activity of the protein. The importance of Arg239 and Gly302 is further supported by existing evidence based on X-ray crystallography data that these amino acids occur in a region which undergoes major conforma- tional change upon substrate binding [13]. Moreover, Arg239 is one of the key amino acid residues that forms electrostatic interactions with the phosphoryl oxygens of the UDP-glucose [13]. In this report, we studied the functional conse- quences of nine mutations in Korean patients with epimerase-deficiency galactosemia. We found that mutants E165K and W336X produced proteins with greatly reduced protein stability, probably due to misfolding, and they formed aggregates that were degraded by proteasomes. In addition, these proteins also lacked enzyme activity. By contrast, mutants R239W and G302D generated proteins with normal protein stability but lacking detectable enzyme activity. These proteins also failed to rescue galactose-sensitive cell proliferation of ldlD cells. We utilized ldlD cells and provided evidence that these cells will serve as an important system for future studies on the function of GALE. Further studies are warranted to test whether previously identified unstable GALE alleles could form protein aggregates and if these aggregations are toxic. Materials and methods Plasmids and cell culture Wild-type human GALE cDNA was kindly provided by J. L. Fridovich-Keil (Emory University, Atlanta, GA, USA). Mutant GALE cDNAs were generated using the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) and were cloned into pCDNA3 (Invi- trogen, Carlsbad, CA, USA) or pCMV-Tag2A (Stratagene) to produce GALE containing N-terminal myc or FLAG epitope, respectively. All expression constructs were con- firmed by sequencing the entire coding region. 293T, CHO- K1 cells (gift of M. Krieger, Massachusetts Institute of Functional analysis of GALE mutation in ldlD cells Y L. Bang et al. 1958 FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS Technology, Cambridge, MA, USA) and the ldlD cell line (ATCC, Manassas, VA, USA) were maintained as previ- ously described [28]. RT-PCR 293T cells grown in 60 mm dishes were transfected with 4 lg myc–GALE expression construct together with 0.3 lg EGFP expression plasmid, pEGFP-C1 (Clontech, Palo Alto, CA, USA) using the calcium phosphate method. Forty hours after transfection, total RNA was isolated and equal amount of RNA (4 lg) was reverse transcribed using SuperScriptÔ II reverse transcriptase (Invitrogen) and 0.5 mm oligo(dT) 12–18 as primer. One or three microliters of cDNA products were amplified with Taq DNA polymerase in the presence of primers specific for the GALE transgene and EGFP. Stability of GALE proteins Forty hours after transfection 293T cells were harvested in lysis buffer [50 mm Tris, pH 7.4, 0.5% Nonidet P-40, 150 mm NaCl, 1 mm EDTA and Complete Protease Inhibi- tor Cocktail Tablet (Roche Applied Science, Mannheim, Germany)]. After centrifugation, the supernatant was resolved by SDS ⁄ PAGE and myc–GALE and EGFP pro- teins were detected by western analysis using antibody against myc (9E10; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and antibody against GFP (Santa Cruz). To determine protein half-life, 293T cells grown in 100 mm dishes were transfected with 10 lg wild-type or mutant myc–GALE expression plasmid. On the following day, cells were evenly divided among the wells of a six-well plate and allowed to incubate overnight. Wells were treated with 50 lgÆmL )1 cycloheximide (Sigma, St Louis, MO, USA) for the durations indicated in the text. In parallel experiments, cells were treated with the 25 lm proteasome inhibitor, MG132 (Calbiochem, San Diego, CA, USA) 1 h prior to and for the duration of cycloheximide treatment. The myc– GALE protein levels were determined by western analysis and tubulin levels were determined as a loading control. Immunofluorescence microscopy 293T cells grown on coverslips were transfected with myc– GALE or FLAG–GALE, fixed with 4% paraformaldehyde in NaCl ⁄ P i for 20 min, and permeabilized with 1% Nonidet P-40 ⁄ 10 mm glycine for 5 min. Cells on coverslips were blocked in 3% BSA for 30 min. Myc–GALE-transfected cells were treated with anti-myc mAb followed by rhoda- mine-conjugated goat anti-mouse IgG (Jackson Immuno- Research Laboratories Inc, West Grove, PA, USA). GALE and Hsc70 were detected in FLAG–GALE-transfected cells using rabbit FLAG antibody and mouse Hsc70 antibody followed by rhodamine-conjugated goat anti-rabbit IgG and FITC-conjugated goat anti-mouse IgG. Nuclei were stained with 10 lgÆmL )1 DAPI and observed by confocal fluorescence microscopy using a Zeiss 510 laser-scanning microscopy (Carl Zeiss, Jena, Germany). GALE enzyme assay ldlD cells grown in 100 mm dishes were transfected with 5 lg GALE expression plasmid using 15 lL Lipofecta- min 2000 (Invitrogen) according to the manufacturer’s instructions and harvested 24 h later. Cell pellets resus- pended in distilled water were sonicated and centrifuged. The supernatant was used for the enzyme assay and wes- tern analysis to determine relative protein abundance. GALE enzyme activity was determined using UDP-galac- tose as previously described [20,32]. The relative protein abundance in each lysate was calculated by measuring the intensity of the individual GALE and tubulin bands in the western blot using Scion Image (Scion Corporation, Frederick, MD, USA). Generation of stable cell lines and BrdU incorporation assays In order to test the galactose sensitive proliferation of the cells, stable ldlD cell lines were generated and BrdU incor- poration assays were performed as described previously [28]. Dimerization To determine whether the mutant GALE proteins are able to form homo- and heterodimers, we transfected sta- ble cell lines expressing wild-type or mutant myc–GALE allele with the FLAG–GALE expression constructs. In the case of mutant myc–GALE, in which the expression level was low in the stable lines, we co-transfected both myc- and FLAG-tagged hGALE into ldlD cells. Twenty- four hours after transfection, ldlD cells were harvested with lysis buffer (50 mm Tris, pH 7.4, 0.5% Nonidet P-40, 150 mm NaCl, 1 mm EDTA and Complete Protease Inhibitor Cocktail Tablet) and centrifuged. The superna- tant was used for immunoprecipitation with myc antibody and precipitated protein was analyzed by western blotting. Membranes were probed with anti-FLAG mAb, stripped, and re-probed with horseradish peroxidase-conjugated myc antibody. Alternatively, the membranes were probed with rabbit FLAG antibody, stripped, and re-probed with anti-myc mAb. Acknowledgements We wish to thank J. L. Fridovich-Keil of Emory University for providing the hGALE cDNA con- Y L. Bang et al. Functional analysis of GALE mutation in ldlD cells FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS 1959 structs. This study was supported by a grant from the Seoul National University Research Fund (JS). References 1 Holden HM, Rayment I & Thoden JB (2003) Structure and function of enzymes of the Leloir pathway for galactose metabolism. J Biol Chem 278, 43885–43888. 2 Fridovich-Keil JL (2006) Galactosemia: the good, the bad, and the unknown. J Cell Physiol 209, 701–705. 3 Ng WG, Xu Y-K, Cowan TM, Blitzer MG, Allen RJ, Bock H-GO, Kruckeberg WC & Levy HL (1993) Eryth- rocyte uridine diphosphate galactose-4-epimerase defi- ciency identified by newborn screening for galactosemia in the United States. Screening 2, 179–186. 4 Misumi H, Wada H, Kawakami M, Ninomiya H, Suei- shi T, Ichiba Y & Shohmori T (1981) Detection of UDP-galactose-4-epimerase deficiency in a galactosemia screening program. Clin Chim Acta 116, 101–105. 5 Walter JH, Roberts RE, Besley GT, Wraith JE, Cleary MA, Holton JB & MacFaul R (1999) Generalised uri- dine diphosphate galactose-4-epimerase deficiency. Arch Dis Child 80, 374–376. 6 Openo KK, Schulz JM, Vargas CA, Orton CS, Epstein MP, Schnur RE, Scaglia F, Berry GT, Gottesman GS, Ficicioglu C et al. (2006) Epimerase-deficiency galacto- semia is not a binary condition. Am J Hum Genet 78, 89–102. 7 Darrow RA & Rodstrom R (1968) Purification and properties of uridine diphosphate galactose 4-epimerase from yeast. Biochemistry 7, 1645–1654. 8 Wilson DB & Hogness DS (1969) The enzymes of the galactose operon in Escherichia coli. II. The subunits of uridine diphosphogalactose 4-epimerase. J Biol Chem 244, 2132–2136. 9 Tsai CM, Holmberg N & Ebner KE (1970) Purification, stabilization, and properties of bovine mammary UDP-galactose 4-epimerase. Arch Biochem Biophys 136, 233–244. 10 Thoden JB, Frey PA & Holden HM (1996) Crystal structures of the oxidized and reduced forms of UDP- galactose 4-epimerase isolated from Escherichia coli. Biochemistry 35, 2557–2566. 11 Alphey MS, Burton A, Urbaniak MD, Boons GJ, Ferguson MA & Hunter WN (2006) Trypanosoma brucei UDP-galactose-4¢-epimerase in ternary complex with NAD+ and the substrate analogue UDP-4-deoxy- 4-fluoro-alpha-d-galactose. Acta Crystallogr Sect F Struct Biol Cryst Commun 62, 829–834. 12 Thoden JB & Holden HM (2005) The molecular archi- tecture of galactose mutarotase ⁄ UDP-galactose 4-epim- erase from Saccharomyces cerevisiae. J Biol Chem 280, 21900–21907. 13 Thoden JB, Wohlers TM, Fridovich-Keil JL & Holden HM (2000) Crystallographic evidence for Tyr 157 func- tioning as the active site base in human UDP-galactose 4-epimerase. Biochemistry 39, 5691–5701. 14 Agarwal S, Gopal K, Upadhyaya T & Dixit A (2007) Biochemical and functional characterization of UDP- galactose 4-epimerase from Aeromonas hydrophila . Biochim Biophys Acta 1774, 828–837. 15 Daude N, Gallaher TK, Zeschnigk M, Starzinski-Po- witz A, Petry KG, Haworth IS & Reichardt JK (1995) Molecular cloning, characterization, and mapping of a full-length cDNA encoding human UDP-galactose 4¢-epimerase. Biochem Mol Med 56 , 1–7. 16 Maceratesi P, Daude N, Dallapiccola B, Novelli G, Allen R, Okano Y & Reichardt J (1998) Human UDP- galactose 4¢ epimerase (GALE) gene and identification of five missense mutations in patients with epimerase- deficiency galactosemia. Mol Genet Metab 63, 26–30. 17 Alano A, Almashanu S, Chinsky JM, Costeas P, Blitzer MG, Wulfsberg EA & Cowan TM (1998) Molecular characterization of a unique patient with epimerase-defi- ciency galactosaemia. J Inherit Metab Dis 21, 341–350. 18 Wohlers TM, Christacos NC, Harreman MT & Frido- vich-Keil JL (1999) Identification and characterization of a mutation, in the human UDP-galactose-4-epimer- ase gene, associated with generalized epimerase-defi- ciency galactosemia. Am J Hum Genet 64, 462–470. 19 Henderson JM, Huguenin SM, Cowan TM & Frido- vich-Keil JL (2001) A PCR-based method for detecting known mutations in the human UDP galactose-4¢-epim- erase gene associated with epimerase-deficiency galacto- semia. Clin Genet 60, 350–355. 20 Park HD, Park KU, Kim JQ, Shin CH, Yang SW, Lee DH, Song YH & Song J (2005) The molecular basis of UDP-galactose-4-epimerase (GALE) deficiency galacto- semia in Korean patients. Genet Med 7, 646–649. 21 Chhay JS, Vargas CA, McCorvie TJ, Fridovich-Keil JL & Timson DJ (2008) Analysis of UDP-galactose 4¢-epimerase mutations associated with the intermediate form of type III galactosaemia. J Inherit Metab Dis 31 , 108–116. 22 Timson DJ (2005) Functional analysis of disease-caus- ing mutations in human UDP-galactose 4-epimerase. FEBS J 272, 6170–6177. 23 Quimby BB, Alano A, Almashanu S, DeSandro AM, Cowan TM & Fridovich-Keil JL (1997) Characteriza- tion of two mutations associated with epimerase-defi- ciency galactosemia, by use of a yeast expression system for human UDP-galactose-4-epimerase. Am J Hum Genet 61, 590–598. 24 Wasilenko J, Lucas ME, Thoden JB, Holden HM & Fridovich-Keil JL (2005) Functional characterization of the K257R and G319E–hGALE alleles found in patients with ostensibly peripheral epimerase deficiency galactosemia. Mol Genet Metab 84, 32–38. 25 Kingsley DM, Kozarsky KF, Hobbie L & Krieger M (1986) Reversible defects in O-linked glycosylation and Functional analysis of GALE mutation in ldlD cells Y L. Bang et al. 1960 FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS LDL receptor expression in a UDP-Gal ⁄ UDP-GalNAc 4-epimerase deficient mutant. Cell 44, 749–759. 26 Garcia-Mata R, Gao YS & Sztul E (2002) Hassles with taking out the garbage: aggravating aggresomes. Traffic 3, 388–396. 27 Johnston JA, Ward CL & Kopito RR (1998) Aggre- somes: a cellular response to misfolded proteins. J Cell Biol 143, 1883–1898. 28 Schulz JM, Ross KL, Malmstrom K, Krieger M & Fridovich-Keil JL (2005) Mediators of galactose sensi- tivity in UDP-galactose 4¢-epimerase-impaired mamma- lian cells. J Biol Chem 280, 13493–13502. 29 Wasilenko J & Fridovich-Keil JL (2006) Relationship between UDP-galactose 4¢-epimerase activity and galac- tose sensitivity in yeast. J Biol Chem 281, 8443–8449. 30 Shyu AB, Wilkinson MF & van Hoof A (2008) Messen- ger RNA regulation: to translate or to degrade. EMBO J 27, 471–481. 31 Kopito RR (2000) Aggresomes, inclusion bodies and protein aggregation. Trends Cell Biol 10, 524–530. 32 Ramm M, Wolfender JL, Queiroz EF, Hostettmann K & Hamburger M (2004) Rapid analysis of nucleotide- activated sugars by high-performance liquid chromatog- raphy coupled with diode-array detection, electrospray ionization mass spectrometry and nuclear magnetic resonance. J Chromatogr A 1034, 139–148. Supporting information The following supplementary material is available: Fig. S1. Summary of the functional consequences of human GALE mutations. Fig. S2. Allelic interaction between the mutant alleles identified in Korean galactosemia patients. This supplementary material can be found in the online version of this article. Please note: Wiley-Blackwell is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. Y L. Bang et al. Functional analysis of GALE mutation in ldlD cells FEBS Journal 276 (2009) 1952–1961 ª 2009 The Authors Journal compilation ª 2009 FEBS 1961 . Functional analysis of mutations in UDP-galactose-4- epimerase (GALE) associated with galactosemia in Korean patients using mammalian GALE-null cells You-Lim. reported nine mutations in Korean patients with epimerase-deficiency galactosemia. In order to determine the functional consequences of these mutations,