Characterization of the mechanisms by which missense mutations in the lysosomal acid lipase gene disrupt enzymatic activity

Abstract

Hydrolysis of cholesteryl esters and triglycerides in the lysosome is performed by lysosomal acid lipase (LAL). In this study we have investigated how 23 previously identified missense mutations in the LAL gene (LIPA) (OMIM# 613497) affect the structure of the protein and thereby disrupt LAL activity. Moreover, we have performed transfection studies to study intracellular transport of the 23 mutants. Our main finding was that most pathogenic mutations result in defective enzyme activity by affecting the normal folding of LAL. Whereas, most of the mutations leading to reduced stability of the cap domain did not alter intracellular transport, nearly all mutations that affect the stability of the core domain gave rise to a protein that was not efficiently transported from the endoplasmic reticulum (ER) to the Golgi apparatus. As a consequence, ER stress was generated that is assumed to result in ER-associated degradation of the mutant proteins. The two LAL mutants Q85K and S289C were selected to study whether secretion-defective mutants could be rescued from ER-associated degradation by the use of chemical chaperones. Of the five chemical chaperones tested, only the proteasomal inhibitor MG132 markedly increased the amount of mutant LAL secreted. However, essentially no increased enzymatic activity was observed in the media. These data indicate that the use of chemical chaperones to promote the exit of folding-defective LAL mutants from the ER, may not have a great therapeutic potential as long as these mutants appear to remain enzymatically inactive.

Introduction

Low-density lipoprotein (LDL) is the major cholesterol-transporting lipoprotein in plasma and is cleared from plasma by binding to the cell-surface LDL receptors (1). After LDL has been internalized by LDL receptor-mediated endocytosis, LDL is released from the LDL receptor in the sorting endosome and transported down the endosomal/lysosomal tract for degradation in the lysosome (1). In the lysosome, apolipoprotein B of LDL is hydrolyzed by cathepsins (2) and cholesteryl esters and triglycerides are hydrolyzed by lysosomal acid lipase (LAL) (3,4). After hydrolysis of cholesteryl esters, free cholesterol interacts with Niemann–Pick type C2 and is then transported out of the lysosome by the action of Niemann–Pick type C1 (5). Cholesterol that is released from the lysosome increases the cholesterol concentration in the endoplasmic reticulum (ER) that elicits regulatory mechanisms to reduce the endogenous cholesterol synthesis, reduce the synthesis of LDL receptors and increase cholesterol esterification for storage in lipid droplets (1).

The importance of LAL for lipid metabolism is illustrated by the clinical features of subjects lacking LAL activity due to homozygosity or compound heterozygosity for mutations in the LAL gene (LIPA) (OMIM# 613497). These subjects, who have Wolman disease, present with liver disease, malnutrition and adrenal cortical insufficiency which appear during the first months of life, usually die within 6-12 months after birth (3). Those who are homozygous or compound heterozygous for milder mutations in LIPA and who have retained some enzymatic activity, have cholesteryl ester storage disease (CESD). These patients present with hepatomegaly, increased levels of transaminases and hypercholesterolemia during childhood or later in life (3,6–8). Typically, CESD patients have a LAL activity that is <10% of normal (9,10) that illustrates that therapeutic measures, which may cause only a minor increase in LAL activity, could make these patients symptom-free.

LAL is synthesized as a 399 residue protein and its catalytic triad consists of residues Ser₁₇₄, Asp₃₄₅ and His₃₇₄ (11,12). Structurally, LAL has a core domain (LAL-CD) and a cap domain (12). The cap domain of 125 residues consists of a non-flexible part and a 40 residue flexible lid (12,13). Displacement of this lid is required for substrates to access Ser₁₇₄ of the catalytic triad (13). After synthesis, LAL undergoes cleavage of the signal peptide and N-linked glycosylation in the ER (11,14). After trimming of the N-linked sugars and addition of mannose-6-phosphate in the Golgi apparatus, LAL binds to the mannose-6-phosphate receptor for subsequent transport to the lysosome (15,16). However, some of the synthesized LAL is secreted and can be found in plasma or in the medium of cultured cells (4,17).

Mutations in LIPA that cause Wolman disease are typically non-sense mutations, frame-shift mutations or splice-site mutations, while the majority of mutations that cause CESD are missense mutations (18). However, it may sometimes be difficult to determine whether missense mutations in LIPA are pathogenic or neutral genetic variants by the use of commonly used tools such as literature search, bioinformatics prediction programs and exploration of mutation databases. Thus, we have recently performed in vitro testing of a large series of missense mutations in LIPA in order to determine how these mutations affect LAL activity (19). As a result, novel pathogenic mutations have been identified, and previously reported pathogenic mutations have been reclassified as being neutral genetic variants (19).

The mechanism by which missense mutations outside the catalytic triad cause defective enzyme activity is often poorly understood. By affecting the structure of LAL, these mutations could indirectly disrupt the catalytic site or affect the intracellular transport from the ER to the lysosome. Because of the potential of relieving the symptoms in CESD patients by obtaining only a minor increase in LAL activity, there is a need to identify the mechanism by which the individual mutations disrupt the function of LAL. These data could then form the basis for developing novel mutation-specific therapies. In this study we have selected 23 pathogenic missense mutations in LIPA in order to characterize the mechanism by which they cause defective LAL activity. Moreover, studies aimed at increasing the amount of functional LAL by the use of chemical chaperones, have been performed.

Results and Discussion

Intracellular transport and secretion of mutant LAL caused by missense mutations in LIPA

We have previously identified 23 pathogenic missense mutations in LIPA that affect 20 of the residues in the mature LAL (19). All the respective LAL mutants had <10% of normal enzyme activity (19). Because none of the 23 mutants affects residues of the catalytic triad, the reduced enzymatic activity could be due to lost or altered protein structure that indirectly disrupts the catalytic site, inhibits the normal processing of LAL or inhibits the intracellular transport.

To study whether the mutants were transported normally out of the ER and were found in media, HepG2 cells were transiently transfected with the respective mutant LIPA plasmids, and the amounts of mutant LAL in the lysates and media were determined by western blot analysis (Fig. 1). Some variation in the amount of mutant LALs in lysates were observed that could be due to differences in intracellular degradation or intracellular transport, or be due to differences in transfection efficiencies. However, marked differences in the amounts of mutant LAL were observed in the media. For the 12 mutants Q85K, G87V, W95R, N119S, H129P, L200P, P202L, S289C, H295Y, G342R, L357P and I391S, no or only minor amounts were found in the media. Thus, these mutants may be defective in their transport out of the ER or through the Golgi apparatus. The three mutants N98K, D145E and T288I were the mutants that were most efficiently secreted. However, some caution should be exerted when interpreting data obtained by overexpression of the mutants.

Figure 1

Secretion of mutant LAL. HepG2 cells were transiently transfected with each of 23 missense mutant LIPA plasmids, and the amounts of mutant LAL in lysates and media were determined by western blot analysis using an antibody against a C-terminal V5 tag. GAPDH was used as a loading control. The molecular weight markers representing 50 and 75 kDa are indicated. One representative western blot from three separate western blot analyses is shown.

Structural and conservation analysis of human LAL and its mutants

To provide more information about the mechanism by which the 23 mutants have <10% of normal enzymatic activity, we have investigated how the 20 residues affected by mutations contribute to the structure of LAL and to what extent they are evolutionary conserved.

Full-length sequences of human LAL and 89 orthologs from 43 placental mammals, 13 bird species, 10 reptiles, amphibians, the coelacanth, 17 fishes and a cartilaginous fish together with human paralogs encoded by the genes lipase family member F (LIPF) (gastric triacylglycerol lipase) (OMIM# 601980), lipase family member K (LIPK) (OMIM# 613922), lipase family member N (LIPN) (OMIM# 613924) and lipase family member M (LIPM) (OMIM# 613923), were aligned in order to investigate how the residues of these proteins are conserved in the LAL family. The different acid lipase paralogs have originated by a number of gene duplications that predates the origin of vertebrates (20). Sequence identifiers of these 94 homologs are shown in Supplementary Material, Table S1, and a subset of the aligned LAL sequences is shown in Figure 2.

Figure 2

Conservation of functionally important and mutated residues in LAL. Multiple sequence alignment of human LAL, orthologous proteins from 10 vertebrates and the 3 human paralogs encoded by the genes LIPF (gastric triacylglycerol lipase), LIPK and LIPN. The figure shows the location of the poorly conserved, mainly hydrophobic, signal peptide (yellow bar), cap domain (pink/blue bar, residues 205–329 of human LAL), with the flexible lid region (pink bar, residues 233–272). The residues of the catalytic triad (red) and the Cys residues (orange) forming the conserved disulfide bridge (black solid line) in the lid, Cys₂₄₈ and Cys₂₅₇ in human LAL, are highlighted. The six Asn residues (green) of human LAL found in an Asn-X-Ser/Thr N-glycosylation consensus sequence have previously been shown to be glycosylated (22). The 20 residues affected by pathogenic mutations studied in the current work are highlighted in black. The sequences are from human (NCBI RefSeq identifiers NP_000226.2, NP_004181.1, NP_001073987.1 and NP_001095939.1), Mus musculus (NP_067435.3), Felis catus (XP_003993923.1), Gallus gallus (XP_421661.3), Alligator mississippiensis (XP_006272309.1), Chelonia mydas (XP_007071640.1), Xenopus laevis (NP_001089697.1), Latimeria chalumnae (XP_006011159.1), Danio rerio (NP_998569.1), Lepisosteus oculatus (XP_015202324.1) and Callorhinchus milii (XP_007895060.1).

If one considers residues in human LAL that are identical in at least 93 of the 94 homologs in our vertebrate dataset as being highly conserved, then 98 of the 399 residues (25%) in human LAL are highly conserved. However, among the residues affected by pathogenic mutations that have been investigated in the current study, 14 of 20 residues (70%) are highly conserved. Of the remaining six residues, hydrophobic residues Leu₂₆₄, Leu₂₉₄ and Ile₃₉₁ are not conserved as such, but their properties are conserved. During evolution they have only been replaced by other hydrophobic residues such as Val, Met, Leu or Ile. Finally, Asp₁₄₅ is conserved in 92 of the 94 homologs and Arg₁₂₇ and Leu₃₅₇ are found to be identical in all human paralogs. Thus, the 20 functionally important residues studied in this work are among the most conserved in human LAL, indicating that there is strong evolutionary pressure to keep their identity to uphold protein function.

A structural model of human LAL with an accessible active site in the open configuration, generated by comparative modeling, is shown in Figure 3A. All six Asn residues of the N-glycosylation consensus sequence Asn-X-Ser/Thr (21) of human LAL have previously been shown to be glycosylated (22). These six Asn residues are highlighted in Figure 3A together with the catalytic triad. None of the 20 residues affected by pathogenic mutations are found in the N-glycosylation consensus sequence at the six glycosylation sites (Figs 2 and 3B). Thus, the 23 mutations are not likely to directly influence protein glycosylation. The biological significance of the six N-glycosylation consensus sequence sites, analyzed with respect to their evolutionary conservation (Fig. 2), is discussed in Supplementary Material.

Figure 3

Non-secreted pathogenic LAL mutants are mainly caused by mutations of residues stabilizing LAL-CD. (A) Structural model of human LAL showing the location of LAL-CD (gray) with the fully conserved catalytic triad residues (rendered as sticks with red carbons), and the cap domain (pink and blue, residues 205–329) with the flexible lid (pink, residues 233–272). The lid has a fully conserved disulfide bridge between residues Cys₂₄₈ and Cys₂₅₇ (sticks with yellow carbons). Six Asn residues, all located on the protein surface, that previously have been shown to be glycosylated (22), are highlighted with yellow spheres at their α-carbons. (B) Structural model of human LAL (left) and rotated 180° (right) showing the location of the 20 residues affected by pathogenic mutations studied in the current work. The eight residues that upon mutation result in mutant LALs that are secreted (see Fig. 1) are shown with green balls at their α-carbons. All these residues are either building the structure of the cap domain, or they are involved in maintaining the interaction between the cap domain and LAL-CD. Residues that upon mutation result in at least one non-secreted mutant are highlighted with red balls at their α-carbons. Ten out of 12 of these residues are stabilizing LAL-CD, while Ser₂₈₉ and His₂₉₅ are located in the cap domain. The figure is rendered as a Cα trace with other coloring and rendering of catalytic residues and glycosylation sites as in (A).

Even though none of the 23 mutations affect the 3 residues of the catalytic triad (Figs 2 and 3B), 4 of the 20 residues affected by pathogenic mutations are found close to the catalytic triad residues either in the sequence (Gly₃₄₂) or in the structure (within 5 Å of the catalytic triad; Gly₈₇, Leu₂₀₀ and Pro₂₀₂). Gly₃₄₂ is located just upstream of catalytic residue Asp₃₄₅ in the same loop structure, while Gly₈₇ is located in a second loop at the bottom of the active site pocket. Mutations G342R (c.1024G>T), G342W (c.1024G>T) and G87V (c.260G>T) are very likely to impair the flexibility of the loops and destroy the structure of the active site, and most likely also affect the folding of the entire LAL-CD (Fig. 3). Leu₂₀₀ and Pro₂₀₂ are contributing to a third loop near the catalytic triad. They are both highly conserved and are involved in stabilizing the structure of LAL-CD.

Because only four of the affected residues are spatially close to the active site and can potentially contribute to loss of catalytic activity directly, the major fraction of the mutations most likely destroys the folding and three-dimensional structure of the protein. Residues Trp₉₅ and Leu₂₀₀ (both highly conserved) and Ile₃₉₁ (conserved as being hydrophobic) (Fig. 2) are contributing to the hydrophobic core of LAL-CD and are interacting with other hydrophobic and highly conserved residues such as Leu₁₀₄, Phe₁₉₈ and Phe₃₇₇, or residues that are conserved as being hydrophobic such as Val₁₇₁, Trp₃₄₀ and Leu₃₈₇ (Figs 2 and 4A and B). Mutations W95R (c.283T>A) and I391S (c.1172T>G) insert hydrophilic side chains into the hydrophobic core of LAL-CD and are very likely to destroy the three-dimensional structure. The same is probably true for mutation L200P (c.599T>C) due to the lack of flexibility of the Pro residue backbone and the different sizes of the Pro and Leu side chains. Although Leu₃₅₇ to a certain extent is solvent-exposed, its side chain points into the core of the protein. Mutation L357P (c.1070T>C) may therefore also lead to destabilization of LAL-CD.

Figure 4

Residues involved in stabilizing LAL-CD. (A) Several of the residues affected by pathogenic mutations are buried deep in LAL-CD, in three main regions. Coloring and rendering of the model is identical to that in Figure 3, with the protein backbone shown as a Cα trace and selected side chains in stick representation. (B) Residues Trp₉₅, Leu₂₀₀ and Ile₃₉₁ (cyan) contribute to the hydrophobic core of LAL-CD and interact with other conserved residues with lipophilic side chains (green). (C) Residues Gln₈₅ and Asn₁₁₉ (cyan) are involved in a highly conserved H-bonding network (yellow dotted lines) that keeps together three different segments of the protein backbone deep in LAL-CD. (D) His₁₂₉ (cyan) stabilizes a protrusion on LAL-CD through H-bonds (yellow dotted lines) to backbone atoms of Thr₁₃₁ and Asp₅₇. Residues are here shown in stick representation.

The highly conserved residues Gln₈₅, Asn₁₁₉ and His₁₂₉ (Figs 2 and 4C and D) also help stabilizing LAL-CD. Gln₈₅ is buried deep in LAL-CD and its side chain H-bonds to the side chain of the highly conserved Ser₁₂₀ (Fig. 4C). Similarly, Asn₁₁₉ is located inside LAL-CD where the side chain H-bonds to the side chains of Ser₉₂ and Asn₁₂₃, of which the latter is highly conserved (Fig. 4C). Finally, His₁₂₉ H-bonds to the backbone atoms of Asp₅₇ and Thr₁₃₁ and stabilizes a protrusion on the surface of LAL-CD (Fig. 4D). In all three cases, the residues are involved in highly specific and conserved H-bonding networks in LAL-CD, and the mutations Q85K (c.253C>A), Q85R (c.254A>G), N119S (c.356A>G), H129P (386A>C) and H129R (c.386A>G) are very likely to destroy the structure of this domain.

All the LAL mutants discussed thus far are predicted to affect the structure and stability of LAL-CD. In general, they are not, or only very weakly, detected in the media of cells transfected with the respective mutant plasmids (Fig. 1). These residues are highlighted in red in Figure 3B. The remaining mutants fall in two classes. The first class of mutants are modifying residues that are located in and contributing to the structure of the cap domain, that is, L264P, R276K, T288I, S289C, L294S, H295Y and Q298H (Fig. 3B). The second class are mutants affecting residues in LAL-CD that mainly contribute to the interaction between the core and cap domains, that is, N98K, R127W and D145E (Fig. 3B). Examples of residues from both classes are shown and briefly discussed in Supplementary Material, Figure S2. All of these 10 mutants, except S289C and H295Y, are secreted into the medium (Fig. 1) and are highlighted in green in Figure 3B. The mutants S289C and H295Y are both affecting residues that are buried deep in a central α-helix of the cap domain.

In conclusion, our structural analysis has shown that with few exceptions, mutants that do not end up in media are caused by mutations that destroy the folding of LAL-CD, whereas mutants that are found in the media, are caused by mutations that affect the cap domain. The likely explanation is that when critically important residues that build the structure of LAL-CD are mutated, the domain structure is destroyed and the mutant LALs undergo degradation in the ER. When residues that build the cap domain or contribute to the interaction between the cap and core domains are mutated, the mutant LALs are likely to still have an intact LAL-CD, but a misfolded and non-functional cap domain. Apparently, the misfolding is not severe enough to result in protein degradation already in the ER, and the mutant LALs are shuttled to the Golgi apparatus and are to some extent secreted. A summary of the pathogenic mechanisms for the 23 missense mutations appears in Supplemenatary Material, Table S2.

Induction of ER stress by mutants that are not being secreted

Ten of the 12 non-secreted LAL mutants discussed above and marked red in Figure 3B affect the structure and stability of LAL-CD. A likely explanation why these mutants are very weakly, or not at all, detected in the media of transfected cells (Fig. 1) is that when LAL-CD does not fold properly in the ER, mutant LAL may expose hydrophobic regions, aggregate and cause ER stress (23). The remaining two mutants S289C and H295Y that affect the cap domain, are expected to behave similarly. As a consequence of ER stress, the unfolded protein response consisting of a series of pathways aimed to restore the normal cellular state, becomes activated. One of these pathways is ER-associated degradation whereby the misfolded mutant proteins are ubiquitinated and subjected to proteasomal degradation (24). Thus, ER-associated degradation is a likely mechanism for the failure of a mutant LAL to be secreted. Another pathway that is activated by ER stress is phosphorylation of inositol-requiring enzyme 1 that activates its endoribonuclease activity and causes skipping of 26 nucleotides in mRNA of XBP1 (OMIM# 194355) (25). Thus, analysis of XBP1 mRNA can be used to identify ER stress. As representatives for LAL mutants that are not being secreted, mutants Q85K and S289C were selected for analysis of whether they induce ER stress.

mRNA was isolated from CHO T-REx cells stably transfected with the wild-type (WT)-LIPA plasmid, the Q85K-LIPA plasmid or the S289C-LIPA plasmid and assayed with respect to mRNA of XBP1. Whereas, no abnormal XBP1 mRNA was detected in cells transfected with the WT-LIPA plasmid, a 26 nucleotide shorter transcript was observed in cells transfected with each of the two mutant LIPA plasmids (Fig. 5). Assuming that these two mutants are representative for mutants that are not being secreted, the underlying mechanism for the failure to be secreted appears to involve abnormal folding of the mutants in the ER. This induces ER stress and ER-associated degradation. ER-associated degradation has previously been shown to be the mechanism for intracellular degradation for several other mutant lysosomal enzymes including mutant β-hexosaminidase that causes Tay–Sachs disease and mutant glucocerebrosidase that causes Gaucher disease (26). One feature of lysosomal enzymes that may make these proteins especially prone to abnormal folding in the ER is their acidic nature which lowers their thermal stability in neutral environments such as the ER (27).

Figure 5

Mutant LAL that is not secreted causes ER stress. CHO T-REx cells were stably transfected with the WT-LIPA plasmid or the mutant LIPA plasmids Q85K-LIPA or S289C-LIPA. Tetracycline (1 μg/ml) was used to induce expression of the transgenes. mRNA was isolated, reverse transcribed and a relevant part of the XBP1 gene was amplified by PCR and subjected to agarose gel electrophoresis. CHO T-REx cells stably transfected with the WT-LIPA plasmid and incubated with DTT (5 mM) overnight was used as a positive control for ER stress. The band representing XBP1 mRNA without skipping of the 26 nucleotides is indicated as XBP1, and the band representing skipping of the 26 nucleotides is indicated as ∆26-XBP1. One representative gel from three separate experiments is shown.

Use of chaperones to rescue LAL mutants from ER-associated degradation

An established therapeutic option for CESD patients is enzyme replacement therapy whereby intravenously administered recombinant LAL is taken up by mannose-6-phosphate receptors at the cell membrane (28). At the acidic milieu of the sorting endosome, recombinant LAL is released from the mannose-6-phosphate receptor for subsequent transport down the endosomal/lysosomal tract to the lysosome. However, the use of enzyme replacement therapy is severely restricted by its high cost. LAL mutants that have retained some enzymatic activity but are defective in exiting the ER could possibly have sufficient enzymatic activity to prevent clinical manifestations if measures are taken to promote their exit from the ER. This could be accomplished by the use of various chemical agents referred to as chemical chaperones. Chemical chaperones have proved to be effective in rescuing folding-defective mutants in cell culture studies, and they have also been used in human clinical trials as therapies for metabolic diseases such as cystic fibrosis and Fabry disease (29–32).

The 18 missense LAL mutants that were not detected in the media of transfected cells or that were found in reduced amounts in the media, were selected for study. The chemical chaperones used were trimethylamine N-oxide (TMAO), 4-phenylbutyrate (4-PBA), glycerol, dimethyl sulfoxide (DMSO) and carbobenzoxy-Leu-Leu-leucinal (MG132). TMAO is a cellular osmolyte that has been found to induce conversion of an unfolded protein to its native state (33). 4-PBA is a hydrophobic chaperone that interacts with exposed hydrophobic segments of the unfolded protein to assist folding in the ER (34). Glycerol and DMSO are chaperones that alter the solvent conditions by sequestering water molecules (35). MG132 is a proteasomal inhibitor that has been found to increase the activity of mutant lysosomal enzymes by preventing proteasomal degradation (36,37).

HepG2 cells transiently transfected with each of the 18 mutant LIPA plasmids were cultured in the presence or absence of each of the five chemical chaperones, and the amounts of mutant LAL in the lysate and media were determined by western blot analysis. Figure 6 shows the effects of these agents on the mutants Q85K and S289C that are not being secreted. The enzymatic activity of the two mutants in lysates has been shown to be 3 and 7% of normal, respectively (19). Whereas, MG132 markedly increased the amounts of the Q85K and S289C mutants both in lysates and in media, essentially no or minimal effects were observed for any of the four other chemical chaperones (Fig. 6). However, the marked increase in the amounts of Q85K and S289C mutants was not reflected by a significant increase in enzymatic activity in the media (Fig. 6). For the other 16 missense mutants, the effects of these five chemical chaperones were similar to those observed for the mutants Q85K and S289C with respect to the amount of enzyme and enzymatic activities in the media, with only MG132 having a marked effect on the amounts of mutant LAL in the media (data not shown).

Figure 6

Effect of chemical chaperones on the amounts of mutant LAL secreted. HepG2 cells transiently transfected with the WT-LIPA plasmid or the mutant LIPA plasmids Q85K-LIPA or S289C-LIPA, were incubated with chemical chaperones TMAO (100 mM), glycerol (2.5%), DMSO (2%), 4-PBA (5 mM) or MG132 (2.5 μM) overnight. Cell lysates were prepared and media were collected for western blot analyses using an antibody against the C-terminal V5 tag. GAPDH was used as a loading control. The molecular weight markers representing 50 and 75 kDa are indicated. One representative western blot from three separate western blots is shown. The enzymatic activity of WT-LAL and mutant LAL was determined in the media of transfected cells. The enzymatic activity in the presence of each chaperone was expressed relative to that obtained in the absence of that chaperone, which was set as 1.0. The enzymatic activities shown are the mean of three separate experiments.

The LAL-increasing effect of MG132 on the amounts of the mutants Q85K and S289C is assumed to be due to reduced proteasomal degradation that results in higher amounts of LAL intracellularly and thereby increased amounts of secreted LAL. Because MG132 also increased the amounts of WT-LAL in lysates and media (Fig. 6), a certain amount of WT-LAL appears to undergo proteasomal degradation. The finding that the enzymatic activity in the media of cells expressing the WT-LAL only increased by 33% in the presence of MG132 reflects that the assay is not linear at high enzymatic activities (Supplementary Material, Fig. S1). The somewhat lower molecular weights of WT-LAL and of the 18 mutants in the media of cells were cultured in the presence of MG132, is assumed to reflect an effect of MG132 on glycosylation of LAL.

Proteasomal inhibitors have previously been shown to increase the amounts of other lysosomal enzymes such as glucocerebrosidase (26,38,39). Moreover, proteasomal inhibitors such as bortezomib, ixazomib and carfilzomib have been approved for clinical use as treatment of lymphomas (40). Whether MG132 or other proteasomal inhibitors could relieve the symptoms of CESD patients who have mutant LAL that undergoes ER-associated degradation remains to be determined. However, our data may suggest that increasing the amount of mutant LAL with markedly reduced enzymatic activity such as Q85K and S289C seen in patients with CESD, may have a limited therapeutic potential.

Material and Methods

Reagents

TMAO, glycerol, DMSO, 4PBA, MG132, 1,4-dithiothreitol (DTT), 4-methylumbelliferyl-palmitate (4-MUP) were from Sigma-Aldrich Corp. (St. Louis, MO). All other reagents were of analytical grade.

Codon numbering of LIPA

Codon numbering of LIPA (RefSeq NM_000235.3) was according to the nomenclature recommended by the Human Genome Variation Society (41) with the ATG initiation codon being codon #1. The nucleotide changes of the 23 LIPA mutations are shown in Supplementary Material, Table S2.

Cell culturing

HepG2 cells (European Collection of Cell Cultures, Wiltshire, UK) were cultured in Modified Eagle’s medium (Gibco Life Technologies, Paisley, UK) containing 10% fetal bovine serum, 50 μg/ml streptomycin, 50 U/ml penicillin, 2 mM L-glutamine and non-essential amino acids. The cells were grown on collagen-coated plates (BD Biosciences, San Jose, CA). CHO T-REx cells (Invitrogen, Carlsbad, CA) are stably transfected with a tetracycline repressor that enables tetracycline-induced (1 μg/ml) expression of genes cloned into plasmids containing the tetracycline operator 2 element. CHO T-REx cells were maintained in Ham’s F-12 medium (Biowest, Nuaillé, France) supplemented with 10% fetal bovine serum, 50 μg/ml streptomycin, 50 U/ml penicillin, 2 mM L-glutamine, 100 μg/ml blasticidine and 100 μg/ml zeocin.

Plasmids and transfections

Construction of the WT and missense mutant LIPA plasmids with a C-terminal V5-His tag has previously been described (19). Transient transfection of HepG2 cells was performed using FuGENE HD (Roche Diagnostics GmbH, Mannheim, Germany) according to the manufacturer’s instruction. CHO T-REx cells (Invitrogen, Carlsbad, CA) were stably transfected with the indicated plasmids as previously described (42).

Western blot analysis

Western blot analyses of culture media or cell lysates were performed as previously described (43,44). Briefly, 24 h after transfection, HepG2 cells were washed and the medium was replaced with Opti-MEM without phenol red (Gibco Life Technologies, Paisley, UK) for a 16 h incubation. Medium was collected and cell debris was removed by centrifugation. Cells were harvested by scraping in lysis buffer (1.0% Triton X-100, 150 mM NaCl, 10 mM Tris-HCl, pH 7.4) and sonicated. Cell debris was removed by centrifugation. Protein concentrations were determined by BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL) using bovine serum albumin as a standard. Media and lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 4–20% Tris-HCl Criterion Precast Gels (Bio-Rad, Hercules, CA) and transferred to immuno-blot polyvinylidene difluoride membranes (Bio-Rad, Hercules, CA). Non-specific binding sites were blocked using 5% Blotting Grade Blocker non-fat dry milk (Bio-Rad, Hercules, CA). To detect transgenic LAL, the membranes were immunostained with an anti-V5 antibody (Invitrogen, Carlsbad, CA). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control and was detected by the use of an anti-GAPDH antibody (Sigma-Aldrich, St. Louis, MO).

RT-PCR analysis of XBP1 splicing to identify ER stress

Total RNA was isolated from CHO T-REx cells stably transfected with LIPA plasmids using QIAamp RNA Blood Mini Kit (Qiagen, Hilden, Germany). One microgram of RNA was reverse-transcribed using Qiagen Onestep RT-PCR kit (Qiagen, Hilden, Germany). Hamster XBP1 was amplified using the forward primer 5′-CTCGCTTGGGAATGGATGTG-3′ and the reverse primer 5′-GGTAGACCTCTGGGAGTTC-3′ (25). PCR cycling conditions included a 15 min polymerase activation step at 95°C, followed by 40 cycles of 1 min denaturation at 94°C, 1 min annealing at 60°C and 1 min extension at 72°C. PCR products were subjected to electrophoresis on a 2% agarose gel at 50 V for 4 h and stained with GelRed Nucleic Acid Gel Stain (Biotium, Inc., Fremont, CA). RNA isolated from CHO T-REx cells that had been incubated with 5 mM DTT for 24 h was used as a positive control for ER stress.

Measurement of LAL activity

The assay of Dairaku et al. (45) as described by Vinje et al. (19) for measuring LAL activity in lysates was optimized with respect to sample volume for measuring LAL activity in the media of transfected cells (Supplementary Material, Fig. S1). Briefly, 4 μl of conditioned medium in a final volume of 40 μl was added to a 96-well plate together with 50 μl of a 0.345 mM 4-MUP substrate solution. After incubation for 3 h at 37°C, the reaction was terminated by the addition of 200 μl of 150 mM EDTA (pH 11.5). The plate was read on a Synergy H1 Plate Reader (BioTek, Winooski, VT) with an excitation wavelength of 365 nm and an emission wavelength of 450 nm. The enzymatic activity in medium from empty-transfected cells was subtracted to correct for the activity of endogenous lipases.

Bioinformatics analysis of LAL

Protein sequences that are homologous to human LAL were obtained from the NCBI RefSeq database resources (46) employing standard BLAST sequence searching (47). Only sequences with sequence identity above 50% of that of human LAL were included in the dataset. The NCBI Entrez Gene (48), Ensembl (49) and UniProt (50) databases were used to evaluate and ensure good quality of the selected sequences. Multiple sequence alignments were generated with MUSCLE (51), visualized with Jalview (52) and shown with the Clustal X coloring scheme. The structural model of residues 24–395 of human LAL was generated with standard homology modeling from the template structure of Canis familiaris gastric triacylglycerol lipase in the open conformation (13). SWISS-MODEL with default settings were used for structure modeling (53). Sequence identity between human LAL and the template structure (Protein data bank identifier 1K8Q) was 60%. With no insertions or deletions in the alignment, the quality of the model is considered to be very good. All protein structure illustrations were generated with the PyMOL (Molecular Graphics System, version 2.2, Schrödinger, LLC).