Functional evaluation of 16 SCHAD missense variants: Only amino acid substitutions causing congenital hyperinsulinism of infancy lead to loss‐of‐function phenotypes in vitro

Short‐chain 3‐hydroxyacyl‐CoA dehydrogenase (SCHAD), encoded by the HADH gene, is a ubiquitously expressed mitochondrial enzyme involved in fatty acid oxidation. This protein also plays a role in insulin secretion as recessive HADH mutations cause congenital hyperinsulinism of infancy (CHI) via loss of an inhibitory interaction with glutamate dehydrogenase (GDH). Here, we present a functional evaluation of 16 SCHAD missense variants identified either in CHI patients or by high‐throughput sequencing projects in various populations. To avoid interactions with endogenously produced SCHAD protein, we assessed protein stability, subcellular localization, and GDH interaction in a SCHAD knockout HEK293 cell line constructed by CRISPR‐Cas9 methodology. We also established methods for efficient SCHAD expression and purification in E. coli, and tested enzymatic activity of the variants. Our analyses showed that rare variants of unknown significance identified in populations generally had similar properties as normal SCHAD. However, the CHI‐associated variants p.Gly34Arg, p.Ile184Phe, p.Pro258Leu, and p.Gly303Ser were unstable with low protein levels detectable when expressed in HEK293 cells. Moreover, CHI variants p.Lys136Glu, p.His170Arg, and p.Met188Val presented normal protein levels but displayed clearly impaired enzymatic activity in vitro, and their interaction with GDH appeared reduced. Our results suggest that pathogenic missense variants of SCHAD either make the protein target of a post‐translational quality control system or can impair the function of SCHAD without influencing its steady‐state protein level. We did not find any evidence that rare SCHAD missense variants observed only in the general population and not in CHI patients are functionally affected.

expressed in HEK293 cells. Moreover, CHI variants p.Lys136Glu, p.His170Arg, and p.Met188Val presented normal protein levels but displayed clearly impaired enzymatic activity in vitro, and their interaction with GDH appeared reduced. Our results suggest that pathogenic missense variants of SCHAD either make the protein target of a post-translational quality control system or can impair the function of SCHAD without influencing its steadystate protein level. We did not find any evidence that rare SCHAD missense variants observed only in the general population and not in CHI patients are functionally affected.

K E Y W O R D S
congenital hyperinsulinism of infancy, HADH, loss-of-function mutations, SCHAD, short-chain 3-hydroxyacyl-CoA dehydrogenase, variants of unknown significance

| INTRODUCTION
An increasing challenge of clinical medicine is how to handle the wealth of information provided by high-throughput genetic analyses. For a given patient, techniques such as whole-exome and whole-genome sequencing may reveal multiple, rare genetic variants that are of unknown significance with regard to health implications. 1 Predicting the functional effect based on bioinformatics analyses alone is still unreliable, particularly for missense mutations; that is, when the genetic variant results in amino acid substitutions at the protein level. This problem has, for example, been illustrated by studies of HNF1A variants implicated in monogenic diabetes. 2 A concerted effort to functionally evaluate missense variants of the HADH gene has so far not been performed. This gene encodes short-chain 3-hydroxyacyl-CoA dehydrogenase (SCHAD; EC 1.1.1.35), a mitochondrial protein expressed in all cells. It exerts a general metabolic function by catalyzing the third step of β-oxidation of short-and medium-chain fatty acids. 3 In addition, SCHAD has a specific role in glucose homeostasis by inhibiting insulin secretion in the pancreatic β-cells, 4-7 most likely through an inhibitory effect on the activity of another metabolic enzyme: glutamate dehydrogenase (GDH). 8,9 Thus, a number of recessive mutations in the HADH gene have been found to cause congenital hyperinsulinism of infancy (CHI; OMIM # 609975). [10][11][12] CHI is a disease characterized by inappropriately elevated plasma concentrations of insulin, resulting in episodes of hypoglycemia that may become life-threatening if not treated correctly. 13,14 The gnomAD data set contains 155 HADH missense variants, of which 154 are rare with an allele frequency < 0.01. 15 Only few of the SCHAD-CHI case reports have examined protein expression and enzymatic activity of the mutation in question. 10,16,17 In addition, a study of the C. elegans SCHAD protein experimentally evaluated the structural and functional impact of two amino acid substitutions in the conserved dimerization interface of the protein. 18 Here we have initiated a systematic assessment of SCHAD amino acid substitutions by establishing a toolkit of prokaryotic and eukaryotic expression vectors for producing the protein variants as well as a SCHAD knockout cell line for functional testing. We have tested 16 naturally occurring missense variants for stability, subcellular localization, enzymatic activity and GDH interaction. Overall, our data showed that SCHAD variants reported in CHI patients display various loss-of-function phenotypes, whereas functional defects were not seen in any rare variant of the general population.

| Plasmids
The complete coding sequence of the human HADH gene (transcript variant 2, NCBI reference sequence NM_005327.4) was synthesized (DNA 2.0) flanked by sites for the restriction enzymes EcoRI and XhoI. For eukaryotic SCHAD expression, the synthetic gene was transferred to expression vector pcDNA3.1/V5-His B (Invitrogen). The insert was cloned in-frame with the Cterminal V5/His tag using EcoRI/XhoI and the Quick Ligation Kit (NEB).
For prokaryotic expression, the synthetic gene served as template to amplify HADH without the mitochondrial import signal. Overhangs with BsmI and NcoI sites were added in the 5 0 -end and an Acc65I site in the 3 0 -end. PCR primers were 5 0 -CATCATCGTCTCCCATGGGATCC-TCCTCGTCCAC-3 0 and 5 0 -CATCATGGTACCATCA-CTTGTATTTGTAAAATCCTTCTC-3 0 . Phusion High-Fidelity PCR polymerase (Thermo Fisher Scientific) was used with annealing at 53.2 C for 20 seconds. The PCR product was blunt-end-cloned into the vector pJet1.2 (CloneJET PCR Cloning Kit, Thermo Fisher Scientific). Subsequently, the HADH insert of the pJet1.2 vector was ligated to the pETM41-His/MBP expression vector inframe with an N-terminal 6His/Maltose Binding Protein (MBP) tag (Quick Ligation Kit). Restriction enzymes Acc65I/BsmBI were used for the insert and Acc65I/NcoI for the vector.
Sixteen missense variants chosen from the gnomAD data set 15 (transcript ENST00000309522) and from CHI case reports were introduced in the plasmids by using the QuickChange II XL Site Directed Mutagenesis Kit (Agilent). This kit was also used to delete the SCHAD mitochondrial import signal (Δ2-12 construct). Primers used to produce each variant are listed in Table S1.
The plasmid Plu-CMV-hGDH for over-expressing human GDH was kindly provided by Dr Charles Stanley, Philadelphia.
For propagation, all plasmids were heat shocktransformed into One Shot TOP10 competent E. coli. Plasmids were prepared by the QIAfilter Plasmid Midi Kit (QIAGEN) and controlled by linearization/agarose gel electrophoresis and Sanger sequencing.
Empty vector (EV) pcDNA3.1/V5-His B was used as negative control in the expression studies. Theoretical molecular masses were estimated using the ProtParam tool on the ExPASy server. 19 All sequence analyses for site-directed mutagenesis and for CRISPR-Cas9 genome editing (below) were performed by the SnapGene software (GSL Biotech).
Treatment with the proteasome inhibitor MG132 (Sigma-Aldrich) was done 24 hours after transfection. The culture medium was removed, and cells were incubated with 5 μM MG132 or an equivalent volume of DMSO in fresh medium for the specified amount of time.
To produce HEK293 whole-cell lysates, cells were incubated in RIPA buffer (Thermo Fisher Scientific) and centrifuged at 14 000g for 15 minutes at 4 C to obtain a clear supernatant. Protein concentration was measured using the Pierce BCA protein assay kit (Thermo Fisher Scientific).
For western blotting, the samples were mixed with loading buffer and reducing agent, heated at 70 C for 10 minutes and subjected to SDS-PAGE before transfer to a PVDF membrane. The proteins of interest were detected using specific antibodies and visualized by enhanced chemiluminescence using Amersham ECL Prime Western Blotting detection reagent (GE Healthcare) and a GBOX I Chemi XR5 imager (Syngene).

| Cell-free expression of SCHAD-V5 plasmids
Cell-free expression of SCHAD variants was achieved by incubating the pcDNA3.1-SCHAD-V5-His plasmids in TNT T7 Quick Coupled Transcription/Translation master mix (Promega). Each reaction was incubated for 90 minutes at 30 C, and 3 μL were analyzed by western blotting.

| Immunostaining and microscopy
HEK293 SCHAD KO cells were grown on poly-Llysine coated glass coverslips. To label the mitochondria, cells were incubated for 30 minutes in 200 nM MitoTracker Red CMXRos (Invitrogen) diluted in prewarmed cell medium. Cells were fixed for 15 minutes with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100 in PBS and blocked with 1% BSA, 22.5 mg/mL glycine, 0.1% Tween in PBS. To stain for SCHAD-V5, the cells were incubated consecutively with the anti-V5 and anti-mouse Alexa Fluor 488 conjugate antibodies. Each incubation was 1 hour at room temperature, with three PBS washes of 5 minutes in between. Coverslips were mounted in ProLong Gold Antifade Reagent with DAPI (Cell Signaling Technology). The confocal images were collected using a TCS SP5 confocal microscope with a 63×/1.4NA HCX Plan-Apochromat oil immersion objective,~1.0 airy unit pinhole aperture, and appropriate filter combinations (Leica Microsystems). Images were acquired with 405 diode, argon and DPSS 561 lasers and processed using the LAS AF software.

| Multiangle light scattering measurements
The molecular mass of MBP-SCHAD was calculated by SEC-MALS using a Superdex 200 HR 10/30 column coupled to a light-scattering device. Bovine serum albumin (BSA) was used as standard.

| SCHAD enzymatic assay
The enzymatic reaction was started by adding 0.07 μg of purified MBP-SCHAD protein to 1.0 mL 100 mM potassium phosphate buffer (pH 7.0) containing 0.1 mM DTT, 0.3 mg/mL BSA (fatty acid-free) and saturating concentrations of acetoacetyl-CoA (50 μM) and NADH (0.15 mM). Absorbance at 340 nm was measured every minute for 5 minutes at 37 C.

| Statistical analysis
The software R 20 with reshape2 21 and multcomp 22 was used. Plots were produced using the package ggplot2. 23 To evaluate differences in the enzymatic activity of the SCHAD variants, a one-way analysis of variance (ANOVA) and a post-hoc Dunnett's test were used. To evaluate differences in the amount of bound GDH, a onesample T-test was used.

| Selection of SCHAD variants for functional testing
To perform a systematic functional evaluation of SCHAD protein variants, we chose 16 missense variants present in human populations and CHI patients ( Table 1). The population variants were selected to be distributed throughout the domains of SCHAD, and their positions are illustrated in Figures 1A and S1A. Specifically, nine variants were selected from the gnomAD data set, 15 most of them having an allele frequency in the range 0.0002 to 0.005 (Table 1). We also included p.Pro86Leu, which is the most frequent SCHAD coding variant and the only one with an allele frequency > 0.01. 27 Additionally, seven pathogenic variants found in CHI cases were included. Of these, six have been described in case reports 10,16,17,[24][25][26] whereas one, p.His170Arg, has been identified in a CHI patient via our own clinical laboratory service (unpublished).
To characterize functional aspects of these variants in vitro, we introduced them by site-directed mutagenesis into plasmids pcDNA3.1 (for expression in mammalian cells as SCHAD-V5/His fusion protein) and pETM41-His/MBP (for expression in E. coli as MBP-SCHAD fusion protein).

| Expression of SCHAD missense variants in HEK293 cells
SCHAD is a ubiquitous protein that occurs as homodimer in its functional form. 18,28 To avoid the interference that dimerization with endogenous wildtype protein could cause in cellular studies with exogenously expressed variants, we produced a HEK293 SCHAD KO cell line using CRISPR-Cas9 targeted genome editing. Disruption of exon 1 of the HADH gene was verified through DNA sequencing of selected colonies ( Figure S2). Knockout at the protein level was confirmed through western blotting using different anti-SCHAD antibodies ( Figure 1B). A band of the expected molecular mass for SCHAD (~34 kDa) was observed only for the lysates of HEK293 WT cells, and not for SCHAD KO cells.
We then tested expression of the SCHAD variants by transiently transfecting the plasmids into HEK293 SCHAD KO cells. Figure 1C shows selected variants in HEK293 SCHAD KO cells to summarize the most relevant findings, whereas Figure S3 presents all 16 variants expressed in both HEK293 WT and SCHAD KO cells. Analysis of the cellular lysates by western blotting, using an anti-V5 antibody ( Figure 1C) or an anti-SCHAD antibody ( Figure S3), revealed a band of the expected molecular mass (~37 kDa) for the WT-V5/His fusion protein.
Most missense variants exhibited the same size and expression levels as the WT protein. The five exceptions were p.Gly34Arg, p.Phe92Cys, p.Ile184Phe, p.Pro258Leu, and p.Gly303Ser. The p.Phe92Cys variant, found in the gnomAD dataset, displayed slower electrophoretic mobility than the WT protein, but had the same expression level. The four other variants, all CHI-associated, were consistently expressed at reduced levels. This reduction was more pronounced for p.Gly34Arg, p.Ile184Phe, and p.Pro258Leu than for the p.Gly303Ser variant. However, CHI variants p.Lys136Glu, p.His170Arg, and p.Met188Val appeared similar to the WT protein and the remaining population variants. All 16 variants yielded the same expression pattern in normal HEK293 (WT) as in the SCHAD KO cell line ( Figure S3).
Next, we evaluated the expression and subcellular localization of the variants by immunofluorescence of transiently transfected KO cells. Figure 2 exemplifies how variants exhibiting reduced protein levels in Figure 1C, such as p.Pro258Leu and p.Gly303Ser, also resulted in lower amounts of SCHAD-V5 when directly visualized in the cells. Images of the other variants are presented in Figure S4. For variants with lower protein levels, we did not observe a uniform overall reduction in SCHAD levels within the expressing cells. A few cells displayed a fluorescence pattern comparable to that of WTtransfected cells, while most cells showed from very low to undetectable SCHAD levels.
By confocal microscopy, we found that all variants were correctly targeted to the mitochondria as shown by co-localization of SCHAD-V5 with a mitochondrialspecific stain ( Figure S5).

| Translation and protein degradation of SCHAD variants with reduced protein levels
To further test expression of variants p.Gly34Arg, p.Ile184Phe, p.Pro258Leu, and p.Gly303Ser, that is, those displaying low protein levels in HEK293 cells, we used a cell-free expression system. Equal amounts of the plasmids were incubated with the cell-free expression reaction mix followed by western blotting. In stark contrast to the results obtained with transfected cells, these four variants now displayed similar protein levels to WT SCHAD ( Figure 3A). This confirmed that the plasmids were fully functional with protein synthesis being as efficient as from the WT plasmid. Moreover, this observation hinted at the activity of a cellular quality control F I G U R E 2 Expression of selected SCHAD variants in HEK293 SCHAD KO cells as determined by immunofluorescence. The cells were transiently transfected with the indicated V5-tagged SCHAD variants, fixed 48 hours-post transfection, stained using anti-V5 primary antibody and fluorescent anti-mouse IgG secondary antibody (green), and counterstained with DAPI (blue). Images are representative fields from two experiments, each with two technical replicates (×200). WT = wildtype, EV = empty vector mechanism, not present in the cell-free expression system, as explanation for the low variant levels in HEK293 cells.
To investigate whether the four variants were degraded by the ubiquitin-proteasome system, we blocked this pathway in transfected cells by using the proteasome inhibitor MG132. Figure 3B demonstrates that the variants p.Gly34Arg, p.Ile184Phe, and p.Pro258Leu exhibited clearly increased protein levels after 24 hours of MG132 treatment, suggesting that proteasomal degradation is of importance for their instability. There was also a certain effect on the level of p.Gly303Ser, an increase that was confirmed in two additional, independent experiments.

| Enzymatic activity of purified SCHAD variants
Although enzymatic activity of SCHAD can be measured in cell and tissue lysates, 3 we opted for using purified protein due to the greater flexibility of the assay and full control over the amount of SCHAD tested. Even though we were able to express all variants in E. coli cells, the purification efficiency differed considerably from variant to variant ( Figure S6 and data not shown). In fact, the levels of p.Gly34Arg, p.Phe92Cys, and p.Ile184Phe were undetectable at the end of the SEC purification step and therefore no further work was performed on them. Notably, for all variants we employed a purification protocol optimized for WT SCHAD. Thus, none of the conditions were adjusted to the individual characteristics of any specific variant as an attempt to improve the yield.
We assessed the identity and purity of the samples by SDS-PAGE, western blotting and Coomassie staining. For all purified variants, a band around 75 kDa was detected by anti-SCHAD ( Figure 4A) and anti-His (data not shown) antibodies. This agrees with the predicted molecular mass of~77 kDa for MBP-SCHAD. Coomassie staining revealed a few additional, weak bands of higher molecular mass and unknown identity in all cases F I G U R E 3 Expression of unstable SCHAD variants in a cell-free system and in HEK293 cells treated with a proteasome inhibitor. A, Equal amounts of the vectors expressing the indicated variants were incubated with the cell-free expression master mix for 90 minutes. Three microliters of each reaction were analyzed by western blot using an anti-V5 antibody. WT = wildtype, EV = empty vector. B, Twentyfour hours after transfection with the indicated variants, HEK293 SCHAD KO cells were treated with 5 μM MG132 or only DMSO for the specified times. To be able to visualize the unstable variants, we loaded five times the amount of whole-cell lysate (5 μg) per well than loaded in Figure 1C. The blots were analyzed by anti-V5 antibody, whereas anti-β-tubulin and anti-β-catenin antibodies were used for monitoring the loading and controlling the efficiency of MG132 treatment, respectively ( Figure S7A). Some variants, most notably p.Ile33Met, exhibited additional bands of lower molecular weight.
To test if recombinant MBP-SCHAD was purified in the functional dimeric form, we estimated the molecular mass of purified MBP-SCHAD WT by SEC-MALS ( Figure S7B). While the predicted monomeric molecular mass was~77 kDa, the calculated value by SEC-MALS analysis was~168 kDa. This indicated that MBP-SCHAD WT was purified in the dimeric form. As the missense variants all eluted at the same rate as WT SCHAD ( Figure S6), we expected them to be purified as dimers, too.

| Interaction of pathogenic SCHAD variants with GDH
We were therefore left with principally two classes of SCHAD pathogenic variants, one with severely reduced cellular amounts of protein and another with impaired enzymatic activity in the presence of normal protein levels. The variants of the latter class might still have the capacity to inhibit GDH, thereby carrying out normal F I G U R E 4 Purified recombinant MBP-SCHAD variants from E. coli and their enzymatic activity. A, Western blot of all successfully purified variants. Purified protein samples (0.3 μg) were analyzed using the Abmart anti-SCHAD antibody. B, The enzymatic activity of the purified variants was tested on three different days, each time with three measurements. Each dot represents the mean of a triplicate measurement and the horizontal lines show the overall mean. *P < .05, ***P < .001 SCHAD function in insulin secretion. We therefore evaluated the ability of the variants of the second class (p.Lys136Glu, p.His170Arg, and p.Met188Val) to bind GDH by using co-IP (Figures 5 and S8). We co-transfected HEK293 SCHAD KO cells with GDH and the SCHAD variant to be tested, and then treated the cells with a crosslinking reagent before cell lysis and co-IP. This ensured that we captured protein interactions within the mitochondria. The set-up was validated by including a SCHAD variant (Δ2-12) lacking the mitochondrial import signal and therefore unable to locate with GDH inside the mitochondria. The positive control (WT SCHAD) and all negative controls (EV, Δ2-12, IgG) behaved as intended. Somewhat unexpectedly, GDH was still able to co-IP with the three pathogenic SCHAD variants ( Figure 5A, B). Nevertheless, the GDH levels were lower for the missense variants than for WT SCHAD in eight of nine independent Co-IPs tests performed ( Figure 5C), although for each variant the reduction did not quite reach statistical significance when assessed separately.

| DISCUSSION
We here, for the first time, present a concerted functional assessment of naturally occurring SCHAD missense variants. We have evaluated protein expression, intracellular localization, enzymatic activity and GDH interaction. A strength of our study is that we used a cell line devoid of the SCHAD protein. Because this protein naturally forms dimers, 18,28 our approach allowed the functional testing of SCHAD variants without the interference of endogenous protein.
All tested SCHAD missense variants from verified CHI patients displayed loss-of-function phenotypes. This is in line with the other CHI mutations in the literature being nonsense mutations, exon deletions or mutations that affect the splicing machinery. 11 Thus, pathogenic HADH mutations are generally predicted to result in deficiency of SCHAD protein. They are recessively inherited in all reported pedigrees (eg, References 12 and 25). One functional HADH allele therefore provides a level of SCHAD protein function sufficient to avoid hypoglycemic episodes, and dominant-negative mutations are yet to be described.
We found altered properties for all seven CHIassociated SCHAD variants examined. A few case reports of missense mutations have investigated SCHAD protein levels and/or enzyme activity in patient fibroblasts. Consistent with our findings, Clayton et al 10  Forty-eight hours post transfection, cells were treated with a crosslinking reagent before lysis. Cell lysates (450 μL) with equal protein concentration (Input) were incubated with either V5 or IgG antibodies coupled to agarose beads. The proteins bound to the antibodies/beads (IP) were separated by SDS-PAGE, and SCHAD (GeneTex antibody) and GDH were visualized by western blotting. C, GDH and SCHAD western blot signals were quantified for the variants and normalized to the WT sample. The dots represent the calculated GDH/SCHAD ratio for each replicate with the horizontal lines showing the mean for each variant. The dashed line corresponds to the GDH/SCHAD ratio for the wildtype co-IP reaction observed a V max comparable to the value of healthy controls. Still, they noted that the enzymatic function was affected by an increased K m for binding to the cofactor NADH.
Our results suggest that SCHAD missense variants causing CHI exhibit at least two types of loss-of-function phenotypes. The first group, consisting of p.Gly34Arg, p.Ile184Phe, p.Pro258Leu, and p.Gly303Ser, displays the most common outcome for pathogenic missense mutations; that is, a decrease of the protein's half-life leading to lower amounts in the cell. 29 This can be explained by abnormal folding of the protein such that folding intermediates and/or improperly folded proteins are degraded through a protein-quality control system. Accordingly, we readily achieved protein expression in a cell-free eukaryotic system, and treatment with a proteasome inhibitor stabilized the protein levels in HEK293 cells. The small amounts of these variants still detectable by western blot and immunofluorescence may represent a fraction of the synthesized protein that managed to obtain a folded state and escape degradation before import to the mitochondria. 30 A second loss-of-function phenotype was seen for CHI variants p.Lys136Glu, p.His170Arg, and p.Met188Val. Here, enzymatic activity of SCHAD was clearly impacted, but without visible affection of steady-state protein levels. It should be pointed out, though, that biophysical stress (such as high temperature) might have demonstrated protein instability also in this group of missense variants.
Among all CHI variants studied, p.His170Arg was the most atypical since it is the only one located directly in the active site of the enzyme. The His170 residue is part of the catalytic His-Glu pair and functions as a catalytic base, involved in proton abstraction from the substrate. 28 In the three-dimensional conformation of SCHAD, p.Lys136 and p.Met188 are both located on the surface of the NAD-binding domain, opposite to the catalytic site ( Figure S1B). Due to the nature of these amino acids and their location, they could be involved in some interaction or binding, critical for proper functioning of the protein. Notably, Xu et al 18 have demonstrated that mutations distant from the catalytic site affect the enzymatic activity of C. elegans SCHAD by decreasing the stability of the catalytic intermediate formation. Considering that in the context of insulin regulation, SCHAD acts by inhibiting GDH, 8 we tested this particular protein-protein interaction, expecting that pathogenic variants with normal steady-state protein levels would exhibit abolished GDH binding. However, compared with WT SCHAD we found only a partial reduction of the GDH interaction. It is therefore not obvious how these three variants result in CHI.
Of the evaluated population variants, most had properties like that of normal SCHAD. Intriguingly, the p.Asp57Glu variant has been identified in a patient with fulminant hepatic failure consistent with a metabolic disorder involving defective fatty acid oxidation. The patient was heterozygous for p.Asp57Glu, which occurred together with the variant p.Ala40Thr. 31 In fact, a second variant in the Asp57 residue has been identified in a patient with Reye-like syndrome (p.Asp57Gly) who was compound heterozygous, carrying another unique variant, p.Tyr214His. 32 Possibly, these variants could affect the function of the SCHAD protein in fatty acid oxidation but not in insulin secretion.
Among the population variants, p.Phe92Cys stood out by consistently migrating with slower electrophoretic mobility. Still, its protein level was as for WT SCHAD, and we speculate that an extra post-translational modification involving the Cys92 residue might cause the apparent increase in molecular mass. As p.Phe92Cys was the only population variant that could not be purified after E. coli expression, we cannot rule out an effect on enzymatic function. Still, with an allele frequency of 0.006, this variant is the second most frequent SCHAD amino acid substitution observed in populations, and we consider it unlikely to have a pathogenic phenotype.
There is only one common variant (allele frequency > 0.01) of the HADH gene that results in change of protein sequence, namely p.Pro86Leu. When expression levels and enzyme activity of p.Pro86Leu were evaluated in fibroblasts of heterozygous carriers, properties comparable for non-carriers were found. 27 Considering that around 0.7% of the population are predicted to be homozygous for this SNP (based on allele frequency of 0.085 according to the gnomAD database), it was interesting to test the functional effects of this variant without any interference of the WT protein. We observed no obvious difference regarding expression or enzymatic activity when compared with normal SCHAD.
The phenotype of SCHAD deficiency is different from that of other genetic disorders in which genes encoding the enzymes of fatty acid oxidation are mutated. The protein has its highest expression level in the pancreatic β-cells, 5 and it is when the islet-specific function is disrupted that hypoglycemia ensues. 6 Notably, a recent analysis of the genetic architecture of diabetes indicated a small, protective effect of rare variants of the HADH gene ( Figure 2c in Reference 33), an effect that might be explained by heterozygous carriers of rare HADH variants having increased capacity for insulin secretion. A limitation of the present study is therefore that the amino acid substitutions were not tested in cell lines that reflect the β-cell environment, such as INS-1 or EndoC-βH1. Hence, it is conceivable that a variant that in the present study appears normal with regard to SCHAD expression and function, still could exhibit altered properties in β-cells.
Another limitation is that we included only one of the four SCHAD variants found in patients who had no signs of hyperinsulinism but presented with other metabolic diseases (fulminant hepatic failure and Reye-like syndrome). Only the enzyme kinetics of these non-CHI variants have so far been investigated. 31,32 It would have been interesting to investigate their properties both individually and in the combination that they appear in the patients, comparing with the CHI-causing variants. Finally, future studies should address whether the CHI variants with normal protein levels are unable to inhibit GDH despite being able to bind it, or whether these variants disturb other crucial interactions within the protein super complex in which SCHAD is postulated to be embedded. 34 Investigations of how GDH kinetics and protein complex formation are impacted by the normally expressed variants may be needed to fully clarify the mechanism of GDH activation in SCHAD-CHI.
Nevertheless, we have established a set of useful tools for functional testing of SCHAD protein variants. This toolkit includes a human cell line devoid of endogenous SCHAD expression as well as a series of eukaryotic/prokaryotic expression plasmids with missense variants distributed over the two major protein domains. We found no evidence that rare SCHAD amino acid substitutions not seen in CHI patients exhibited impaired function. This information will be of value when interpreting HADH genetic variants revealed by high-throughput sequencing projects and in clinical testing.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.