Searching for protein partners of short-chain 3-hydroxyacyl-CoA dehydrogenase (SCHAD) reveals keratin 8 as a novel candidate for interaction in pancreatic β-cells

Abstract

Background: Short-chain 3-hydroxyacyl-CoA dehydrogenase (SCHAD) is a ubiquitously expressed mitochondrial enzyme with a role in the degradation of fatty acids. Because the protein also is a negative regulator of insulin secretion in pancreatic β-cells, inactivating mutations in the SCHAD gene (HADH) cause congenital hyperinsulinism of infancy (CHI) and severe hypoglycemia. Here we sought to identify novel interaction partners of SCHAD that might be particularly relevant for the endocrine pancreas.

Results: Employing the SCHAD protein as bait, we performed yeast 2-hybrid screening of a cDNA library made from human islets of Langerhans. Surprisingly, the screening revealed the intermediate filament protein keratin 8 (K8) as a putative interaction partner of SCHAD with very high confidence. Previous reports have linked K8 to glucose homeostasis, and we confirmed the SCHAD interaction by co-immunoprecipitation in HEK293 cells. SCHAD and K8 expression were then characterized in the human β-cell model EndoC-βH1. By using proximity ligation assay, we demonstrated that stimulating the cells with a high level of glucose triggered a transient increase in the interaction. However, when studying knockout mice, we found that the loss of either K8 or SCHAD did not change the expression level of the other interaction partner. Still, when K8 knockout mice were challenged with a ketogenic diet, upregulation of SCHAD expression was blunted compared to the upregulation observed in wildtype littermates.

Conclusions: We propose that the SCHAD protein interacts with K8 in a way that might be relevant for proper functioning of the pancreatic β-cell. Whether the SCHAD-K8 interaction influences the phenotype of CHI remains to be demonstrated.

Keywords: HADH; KRT8; Congenital hyperinsulinism of infancy; Intermediate filaments; Keratin 8; Proximity ligation assay; SCHAD; Short-chain 3-hydroxyacyl-CoA dehydrogenase; Yeast 2-hybrid screening.

Fig. 1

A putative SCHAD-K8 interaction identified by yeast 2-hybrid (Y2H) screening. (A) Schematic representation of the SCHAD protein and its domains, and the SCHAD protein variants used as bait in the two Y2H screens. Bait 1 lacked the mitochondrial import signal (MIS), while Bait 2 lacked the MIS and the dimerization domain. (B) Bait 1 resulted in the SCHAD protein itself being identified as the only prey with a high predicted biological score (PBSc). (C) Bait 2 resulted in two preys with high PBSc: K8 and cytospin-A. Legend: Brown boxes = coiled-coil domains; yellow boxes = the specific interaction domains (SID) identified by the experiment; numbers = amino acid residue number. Numbers for K8 are according to its long isoform (P05787-2)

Fig. 2

Subcellular localization and co-immunoprecipitation (co-IP) of SCHAD and K8 from HEK293 cells. (A) HEK293 cells transfected with V5-tagged SCHAD were fixed and stained with anti-V5 (green) and anti-K8 (red) antibodies. Counterstaining of nuclei was with DAPI (blue). Imaging was done by regular fluorescence microscopy. The arrow points to an area of yellow signal, indicating co-localization. Scale bar = 10 μm. (B) HEK293 cells were transfected with V5-tagged SCHAD and crosslinked with 1% paraformaldehyde. The cell lysate was immunoprecipitated with the anti-K8 M20 antibody, followed by western blotting of the precipitate. The upper part of the membrane was immunostained for K8 and the lower part for SCHAD. The arrowheads indicate endogenous K8 and SCHAD, whereas the arrow points to tagged SCHAD protein resulting from the transfection. Two different amounts of cell lysate were loaded on the gel (0.5 and 5 µg protein) to visualize the weak K8 band in the input of the experiment. The strong SCHAD signal in the 5-µg lane had to be covered (white box) to avoid overexposure. Negative control (N) was control agarose resin. The IP lane shows the signals in the immunoprecipitate. The experiment was repeated 3 times with similar results. (C) HEK293 cells without endogenous SCHAD protein (SCHADKO) were transfected with normal SCHAD-V5 (WT) or a mutant SCHAD-V5 variant without a mitochondrial import signal (Δ1–12). After crosslinking, the cell lysates underwent co-IP by either an anti-V5 antibody or a general anti-IgG antibody. Finally, the immunoprecipitates were subjected to mass spectrometry. Two technical replicates were analyzed. The figure shows the Z-score for keratins identified in HEK293 cells (K1, K8, K10, K18, K19)

Fig. 3

SCHAD and K8 interact during insulin secretion in the human β-cell line EndoC-βH1. (A) Immunofluorescent staining of EndoC-βH1 cells using antibodies against SCHAD and K8. The separate images of K8 and SCHAD are maximum projections of several planes through the cells, while the merged image and inset are single planes. Signals from K8 (green) and SCHAD (red) are in close proximity at some spots (arrows). Scale bar = 5 μm. (B) Proximity ligation assay (PLA) of EndoC-βH1cells undergoing glucose-stimulated insulin secretion. The cells were fixed 10, 30 or 40 min after addition of 20 mM glucose. Cells kept at 0 mM glucose were fixed at the same time points for reference. PLA was performed using antibodies against SCHAD and K8. Each orange dot represents a co-localization signal. (C) Quantification of the PLA signals in B. Data are represented as mean ± SEM. ** indicates p < 0.01, *** indicates p < 0.001, n.s. = not significant. A total of 300 cells were analyzed

Fig. 4

K8 and SCHAD expression in SCHAD and K8 knockout mice. (A) Western blots for the detection of K8 in lysates of total pancreas and isolated islets for WT and SCHADKO mice (n = 3 or 4). SCHAD protein was analyzed as confirmation of the knockout in SCHADKO samples and β-actin served as loading control. (B) Western blots for the detection of SCHAD in whole pancreas and isolated islets of WT and K8KO mice (n = 3). Knockout of K8 was confirmed by detection of K8, and GAPDH or β-tubulin served as loading control

Fig. 5

Pancreatic expression of SCHAD in K8 knockout mice fed a ketogenic diet. (A, B) Western blots for the detection of SCHAD in lysates of pancreas samples from male and female K8 wildtype (WT) (n = 11) and K8KO (n = 10) mice fed a control or ketogenic diet. K8 was detected for the confirmation of the wildtype or knockout genotype. Detection of β-actin served as loading control. The stippled line in B indicates a technically unsuccessful lane that was removed. (C) Quantification of SCHAD protein levels of (A) and (B) combined, normalized to respective β-actin loading controls. Data are represented as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01

Fig. 6

Structural modelling of a K8-SCHAD-NAD complex using AlphaFold 3. (A) A possible protein complex consisting of one K8 protein (UniProt ID P05787-2), one SCHAD protein (UniProt ID Q16836) and one NAD molecule is modelled and shown in colors representing the pLDDT score (an estimate of the confidence in the structure prediction). (B) The predicted alignment error (PAE) matrix shows a confident prediction of SCHAD while K8 has regions of both good and poor prediction, indicated by the intensity of the green color in the squares representing each protein (darker is more confident). The black lines indicate the chain boundaries. A region in the head of K8 (amino acids, AA 2–26), indicated by the blue boxes, shows a possible interaction with both SCHAD and NAD. The corresponding region of the protein complex is magnified in the blue circle. The SCHAD protein is shown with surfaces (Chimera-X) where pink represents the mitochondrial import signal, yellow the NAD-binding domain and green the dimerization domain. K8 is shown in purple (AA 2–26), blue (AA 268–419, corresponding to the Y2H specific interaction domain) and grey (other parts). The bound NAD is not visible in this surface model