Mice lacking global Stap1 expression do not manifest hypercholesterolemia

Background Autosomal dominant familial hypercholesterolemia (ADH; MIM#143890) is one of the most common monogenic disorders characterized by elevated circulatory LDL cholesterol. Initial studies in humans with ADH identified a potential relationship with variants of the gene encoding signal transducing adaptor family member protein 1 (STAP1; MIM#604298). However, subsequent studies have been contradictory. In this study, mice lacking global Stap1 expression (Stap1−/−) were characterized under standard chow and a 42% kcal western diet (WD). Methods Mice were studied for changes in different metabolic parameters before and after a 16-week WD regime. Growth curves, body fats, circulatory lipids, parameters of glucose homeostasis, and liver architecture were studied for comparisons. Results Surprisingly, Stap1−/− mice fed the 16-week WD demonstrated no marked differences in any of the metabolic parameters compared to Stap1+/+ mice. Furthermore, hepatic architecture and cholesterol content in FPLC-isolated lipoprotein fractions also remained comparable to wild-type mice. Conclusion These results strongly suggest that STAP1 does not alter lipid levels, that a western diet did not exacerbate a lipid disorder in Stap1 deficient mice and support the contention that it is not causative for hyperlipidemia in ADH patients. These results support other published studies also questioning the role of this locus in human hypercholesterolemia. Supplementary Information The online version contains supplementary material available at 10.1186/s12881-020-01176-x.

STAP1 (MIM#604298), also called B-cell antigen receptor downstream signaling 1 protein (BRDG1) or stem cell adaptor protein 1, was first discovered in immune cells with the highest expression documented in appendix, lymph nodes, and spleen [7]. The gene encoding STAP1 protein is located on chromosomal region 4q13.2 with 10 exons in the protein coding region [7]. Its relationship to lipid homeostasis was first suggested in 2007 in a clinical study finding a correlation between increased levels of circulatory triglycerides and STAP1 gene expression in leukocytes [9]. Later, in 2014, Fouchier and colleagues discovered that variants of the STAP1 gene were associated with the FH phenotype in a linkage analysis of a Dutch family with FH4 and suggested robust support for genetic causality [6]. Subsequent studies reported STAP1 gene variants observed in such FH patients also manifested significant cardiovascular events [10][11][12]. However, other studies found no association between STAP1 variants and hypercholesterolemia [13][14][15][16]. An analysis of seven families with FH phenotype failed to observe the co-segregation of four rare predicted pathogenic variants of STAP1 [17]. Loaiza et al more recently demonstrated no marked changes plasma lipid profiles of carriers of STAP1 variants compare to controls as well as in a mouse model [18]. Therefore, we sought out to understand the STAP1 role by using a mouse knockout model, to ask if loss of STAP1 would lead to hyperlipidemia.
In this study, mice lacking global expression of Stap1 gene (Stap1 −/− ) were characterized for various metabolic parameters on standard diet and after 16 weeks of western diet (WD) regime. The present work presents an independent characterization of Stap1 −/− mice to that studied by Loaiza et al with some differences between models and confirms the argument against STAP1 having a role in causing FH.

Knockout mice and diets
Heterozygous Stap1 KO mice (Stap1 +/− ) on C57BL/6 N background generated through reporter-tagged insertion were imported from Wellcome Trust Sanger Center, via EUCOMM [19][20][21][22]. The 'knockout first' tm1a (Stap1 t-m1a(KOMP)Wtsi ) mutation bears the IRES:lacZ trapping cassette and a floxed promoter-driven neo cassette inserted into the intron of the gene in order to disrupt the gene function [22]. Brother-sister matings of the heterozygotes generated homozygous knockout mice (Stap1 −/− ) and corresponding WT (Stap1 +/+ ) littermate controls. Tail snip genotyping was performed to assess the genotype of each mouse before recruiting into experiments; amplimers of 266 bp and 236 bp correspond to Stap1 +/+ and Stap1 −/− , respectively (Table). Mice were fed with irradiated standard rodent-chow diet (CD; Envigo 7912) [23] and water ad libitum, with equal light and dark cycles. Baseline measurements were performed in 10 to 11 week-old male and female mice on rodent chow and then placed on a high-cholesterol western diet (WD; Envigo TD.88137) for another 16 weeks [24]. Chow and western diets provided 17 and 42% kcal energy from fats [23,24] and were procured from Harlan Teklad, Madison, WI, USA. All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Cincinnati (Protocol No. 16-04-19-01). Terminal blood samples were obtained under full inhaled isoflurane anesthesia. For all studies, except where indicated, each group consisted of 6 mice, based upon prior data that show the power to detect 20% differences for the most variable procedure (OGTT) where the Power needed is n = 5, but we included and extra mouse to ensure Power was not due to loss of a mouse for any inadvertent loss. The investigators were not blinded to the samples or mouse genotypes, as there are no datasets collected which required subjective analyses. However, all mice were group housed and each cage contained mice of both genotypes to minimize bias and variation. Mice were sacrified by CO2 inhalation, followed by cervical dislocation and bilateral thoracotomy.

Measurement of whole-body composition
Whole-body fat and lean mass were measured in live mice using ECHOMRI™-100H whole body composition analyzer (Echo Medical Systems, Texas, USA) according to the manufacturer's specifications. Absolute weights (g) of fat and lean mass of each mouse were used to calculate their % changes [(absolute fat or lean mass weight (g)/whole body weight(g))*100], a widely employed method to determine the diet induced obesity phenotype.

Oral glucose tolerance test (OGTT)
For the OGTT experiment, the experimental mice were initially acclimatized in the procedure room overnight followed by 4 h fast period in the next day morning. After baseline blood glucose measurements, 50% Dextrose (Cat# D16-1; ThermoFisher) in 0.9% saline was administered via oral gavage at a dose of 2 g/kg of body weight [24]. Time dependent changes in blood glucose were measured via tail vein bleeds at 0, 15, 30 and 60 min using Accu-Chek Nano electronic glucometer (Roche Applied Science, Indianapolis, IN, USA).

Fast performance liquid chromatography (FPLC)
Lipoprotein separation for total cholesterol estimation in plasma samples was performed using FLPC as described previously [24]. Briefly, blood was collected from 4 h fasted mice through the submandibular plexus and then centrifuged at 2500×g for 10 min at 4°C. The isolated plasma was then measured for basal total cholesterol to exclude any outliers (to ensure pooling was performed between comparable samples) and 200 μL of pooled plasma from each group was loaded onto the Akta pure FPLC to separate different lipoproteins by column chromatography (10/300GL, superpose 6). A total of 52 fractions with volume of 500 μL each were collected and analyzed for cholesterol content using an enzymatic kit as detailed below.

Plasma biochemistry
Plasma triglycerides and cholesterol were estimated using Infinity Triglyceride (TR22421) and Cholesterol (TR13421) calorimetric kits (Thermo Fisher Scientific, Middletown VA, USA) per manufacturer instructions. For estimation of tissue lipids after the western diet phase, livers from Stap1 +/+ and Stap1 −/− mice were homogenized, normalized for equal protein content, and then measured for total cholesterol and triglycerides using the same calorimetric kits used for plasma samples.

Tissue harvesting and histology of liver
Mice were euthanized for terminal tissue collection at the end of 16 weeks WD regime. Four-hour pre-fasted mice were euthanized by CO 2 exposure and bilateral thoracotomy. Liver, spleen, and kidneys were blotted dry, weighed, and then flash frozen in liquid N 2 . A small portion of liver was rinsed in chilled PBS (pH -7.4) and fixed in 10% neutral buffer formalin (5725; Fisher Scientific, Pittsburg, PA, USA) for paraffin embedding and hematoxylin and eosin (H&E) staining [25]. About 5 μm of paraffin embedded sections were cut, processed in graded volumes of ethanol, and then stained with H&E. Images of the stained liver sections were captured at 20x and 40x objective lens magnification using Olympus BX61 microscope.

QPCR analysis of gene expression
To determine the tissue expression profile of Stap1, total RNA were isolated from different tissues of Stap1 +/+ mice using Qiagen RNeasy Mini Kit (74,104; Qiagen Inc., Germantown, MD, USA). Isolated RNAs were reverse transcribed to cDNA using SuperScript First-Strand Synthesis kit (11904-018; Invitrogen, Carlsbad, CA, USA) and then assessed for the Stap1 gene expression on Applied Biosystems 7300 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). Each sample was run in triplicates using power SYBR Green PCR master mix (4,367,659; Applied Biosystems, Foster City, CA, USA). Comparative C T method was used for quantitation post data normalization to Atp5po expression as the house keeping gene to compensate the variations between input RNA amounts [25]. The list of different primers used for PCR experiments was tabulated in the Table S1 (Additional file 1).

Statistical analyses
An unpaired t-test (without or without Welch's corrections) or Mann-Whitney U test was performed to assess significant between Stap1 −/− and Stap1 +/+ mice within each sex.

Validation of Stap1 gene deletion
The targeting strategy for generation of global Stap1 knockout mice is depicted in Fig. 1a. Measurement of transcript levels of Stap1 in WT mice by qPCR revealed gene expression in a variety of tissues with the highest levels demonstrated in spleen (Fig. 1b). The complete absence of Stap1 expression was confirmed with qPCR in Stap1 −/− mice liver, kidney, and spleen (Fig. 1c). RT-PCR analysis of exons 2-5, 2-6, and 2-7 using total RNA isolated from livers and spleens demonstrated sequential exon splicing of exons 5, 6, and 7 in Stap1 +/+ mice and no products in Stap1 −/− mice (Fig. 1d). The single slightly larger amplicon for exon 2-7 in Stap1 −/− spleen (Fig. 1d, open arrowhead) was sequenced and demonstrated an off-target product on chromosome 2 (Additional file 2).

Baseline characteristics
Baseline measurements on standard chow diet were assessed at 10-11 weeks of age and body weights (Fig. 3a), whole-body fat (Fig. 2a) and lean mass (Fig. 2a) were similar between Stap1 +/+ and Stap1 −/− mice. Further biochemical assessment found no significant differences in glucose tolerance (Fig. 2b), plasma total cholesterol (Fig. 2c), or plasma triglycerides (Fig. 2d) between knockout mice and wildtype mice. Lipoprotein cholesterol profiles in male knockout and wild-type mice were analyzed by FPLC (Fig. 2e) and were also found to be unaffected by the loss of Stap1.

Effects of western diet (WD)
To determine the role of Stap1 under metabolic challenge, mice were fed western diet (42% kcal from fat). Prior to the commencement of the western diets, baseline body weights showed no statistical differences between wildtype and knockout mice (Stap1 +/+ males 25.2 g +/− 3.2 vs Stap1 −/− males 23.5 +/− 3.1, Stap1 +/+ females 22.5 +/− 2.6 vs Stap1 −/− females 19.7 +/− 1.3, n = 6 for each group). Body weights were measured weekly (Fig. 3a), while body composition (Fig. 3b) and tissue weights (Fig. 3c and d) were assessed the end of the WD. No significant differences were noted between Stap1 −/− mice over Stap1 +/+ littermate controls in any of these parameters. Similar trends were also observed in glucose tolerance testing, monitored to ensure there were no differential changes in intermediary metabolism induced by the western diet and loss of Stap1 (Fig. 3e).
Plasma cholesterol and triglyceride levels ( Fig. 4a and  b), and plasma FPLC cholesterol profiles ( Fig. 4c and d) showed the diets led to a significant change in the lipoprotein distribution from baseline (compare to Fig. 2e) as expected for a high fat diet, but there were no differences between wild-type and Stap1 knockout mice.
Investigation of liver histology with hematoxylin and eosin staining (H&E) showed changes compatible with fat accumulation as would be expected after 16 weeks of WD, but found no differences in morphology between groups ( Fig. 5a and b). Moreover, the elevated levels of liver total cholesterol and triglycerides ( Fig. 5c and d) were comparable between WD fed Stap1 +/+ and Stap1 −/− mice.

Discussion
This study tested the hypothesis that loss of Stap1 can lead to hypercholesterolemia using Stap1 −/− mice. The Stap1 mice, targeted by a knockout-first, conditional targeting vector were obtained from the Welcome Trust Sanger Center and characterized. Although there is no reliable antibody for specific detection of endogenous STAP1 protein by western blot (we tested commercial antibodies from ThermoFisher and Abcam but found these unsuitable, see Supplementary files), a knockout was confirmed at the DNA and mRNA level instead. Normal Stap1 RNA expression observed in different tissues of Stap1 +/+ mice and compared with the protein atlas database (https://www.proteinatlas.org/ENSG0000 0035720-STAP1/tissue); relatively high levels of expression in spleen were noted but very minimal expression was seen in liver. The higher levels of RNA also coincide with the proportional amount of STAP1 protein in spleen of Stap1 +/+ mice as measured through targeted proteomics [18]. QPCR analysis of livers, spleens, and kidneys of Stap1 −/− mice did not detect any Stap1 mRNA, confirming the knockout (Fig. 1c). RT-PCR analysis of the mRNA produced expected amplicons when using primers targeted to exon 2 through 4 in Stap1 −/− mice (data not shown), but not with primers specific for exons 5, 6 or 7 (Fig. 1d) suggestive that the transcriptional message was preserved until exon 4 and was interrupted due to reporter insertion in the intronic region This construct is a 'knockout-first', which can be converted to a conditional, upon deletion of the LacZ-Neo cassette. The mice used in this study remain as knockout and retain this cassette. A tissue survey of Stap1 mRNA was performed using qPCR in tissues collected from Stap1 +/+ mice (panel b, n = 2). Note that compared to a house-keeping gene expression (see Methods), only spleen and lung showed some expression with very low levels detected in other tissues. Stap1 −/− mice (KO) showed no detectable qRT-PCR products (panel c, n = 2-3) compared to wild-type mice (WT) in spleen, liver or kidney. To further confirm a lack of legitimate transcript expression, RT-PCR analysis of exon splicing for exons 2-5, 2-6, and 2-7 in liver and spleen was performed (panel d). The labels '5, 6, and 7' for each tissue indicates the location of the reverse primers targeting exons 5, 6, or 7 respectively, used with a forward primer located in exon 2. Spleen mRNA from WT tissues amplified the correct expected size of product and size increments (exon 2-5; 251 bp, exon 2-6; 482 bp, and exon 2-7; 536 bp indicated by filled arrowheads). No products were noted when exons 2-5, or 2-6 were used for KO spleen mRNA. A slightly larger product was noted for exons 2-7 (indicated by open arrowhead), though on sequencing this was found to be a non-specific product from mis-priming (see Text). Error bars denote ±1 SD between exon 4 and 5. RT-PCR using primers targeted to exon 2 through 7 unexpectedly produced a product in Stap1 −/− mice (Fig. 1d), but sequencing showed that it was an off target product from chromosome 2 (supplemental data), whereas Stap1 is located on chromosome 4.
Obesity and insulin resistance are commonly associated with hyperlipidemia and high caloric fat-enriched diets are widely accepted to induce hyperlipidemia [26]. A western-type diet (TD.88137) was chosen for these studies for its ability to induce hypercholesterolemia, obesity and atherosclerosis depending on the genetic background of the mice used for study [24,27,28]. Previous studies have shown that 16 weeks of WD regime is sufficient to induce severe hypercholesterolemia in mice [29,30]. Augmentation of whole-body fat mass and markedly elevated circulatory cholesterol (especially males) was observed in Stap1 +/+ and Stap1 −/− mice at the end of 16 weeks WD (Figs. 3 and 4). However, these changes were not exacerbated, as hypothesized, in mice lacking global Stap1 expression and suggest the loss of Stap1 does not exacerbate hyperlipidemia.
The link between STAP1 and hypercholesterolemia was proposed when some FH4 patients were reported to have variants of the gene, in the absence of changes in known hyperlipidemia loci [6]. A review of genetic variants associated with FH noted 2580 in LDLR, 896 in Together with the fact that the phenotype of STAP1 variant carriers was milder than other forms of ADH [6], one possibility is whether these are 'false positive' hits. Nevertheless the association between STAP1 and ADH was supported further by other studies showing elevated circulatory cholesterol in patients with STAP1 variants [10][11][12]31]. However, this link has been highly debated with studies reporting no association between STAP1 and ADH [13][14][15][16][17], or no difference in plasma lipid profiles in carriers of STAP1 variants and controls [18]. Moreover, Loaiza et al recently also found the absence of any significant difference in lipid phenotype in Stap1 −/− mice compared to controls, even after metabolic challenges [18]. The current study is a verification of these findings. Nevertheless, a KO mouse model is a loss of function model and does not address the possibility of a gain of function variants in Stap1. The fact that the previously identified STAP1 variants were missense mutations suggests that a viable protein may potentially be expressed in those patients, though its functionality would be purely speculative. Fig. 3 Effect of western diet on growth parameters and body composition in mice lacking Stap1 gene. Panel a shows the growth curves in wildtype and knockout mice and no statistical differences were discernable in the patterns of weight gain on the high caloric diet (n = 6). Neither the percent fat or lean mass were altered in Stap1 −/− mice compared to age and sex-matched wildtype mice on a western diet (panel b, n = 6). No changes in weights of liver, spleen or kidneys were seen in male (panel c, n = 4-6) or female mice (panel d, n = 5-6) were noted. No changes in glucose tolerance tests were noted between knockout and wildtype controls either (panel e, n = 6). In case of male Stap1 −/− mice, kidney weights of only 2 mice were considered for analysis It was initially hypothesized that extra-hepatic mechanisms were responsible for STAP1 mediated hypercholesterolemia due to the fact that liver distribution was low [6,31]. Interestingly, recent experiments using single cell RNA-Seq indicated that its expression is highly specific to lymphoid cells (https://tabula-muris.ds.czbiohub. org/) and any expression detected by examining tissue RNA likely comes from the presence of these cells in these tissues. Stap1 expression measured by qPCR in the current study was low in liver and other tissues except for the spleen (Fig. 1b). This could be explained by the concentration of B-cells in the spleen and their presence in other organs, as the tissues were not perfused prior to collection. Some studies have associated STAP1 variations with cardiovascular disease [10][11][12], and immune cells have been linked with atherosclerosis and FH [32]. This study did not thoroughly investigate possible effects of STAP1 deletion on the immune system, but there were no obvious changes in B-cell subsets and IgG (data not shown). Further studies would be obviously necessary to fully characterize the effects, if any, on immune function. Nevertheless, transplant of Stap1 −/− bone marrow (BM) into Ldlr −/− female mice had no impact on plasma lipids as well on the development of atherosclerosis plaques [18]. Furthermore, there were no marked changes in B-lymphocyte populations in blood of carriers of STAP1 gene variants compared to age-and sex-matched family controls [18].
The study by Loaiza et al was well-designed and thorough, but differs from the current investigation in two important experimental variables: 1) the strategy used to generate Stap1 KO mice, using clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat-associated 9 (CRIS PR/Cas9) technology vs targeted homologous recombination of embryonic stem cells and 2) the much shorter time period for hyperlipidemia induction in global and tissue-specific KO mice (4 weeks vs 12 weeks). Since 16 weeks of WD regime is sufficient to induce severe hypercholesterolemia and the fact that FH patients show elevated levels of plasma cholesterol, the long-term regime that we used is ideally advantageous over the short duration regime used in the study by Loaiza et al. An important limitation of the study is loss-of-function models do not recapitulate the missense variants reported for STAP1 in humans. However, we are confident our current studies are valid as our characterization (in keeping with published data) finds no meaningful expression of STAP1 in cells other than B-cells. Thus, even for missense variants of STAP1 to have a pathological role in Fig. 4 Effect of 16 weeks of western diet (WD) on plasma lipid profiles. Panel a shows the plasma cholesterol and panel b the triglyceride levels in wildtype and knockout mice (n = 6 per group). Female mice (whether wild-type or knockout) seem to be resistant to a WD-induced hypercholesterolemia (panel a), but this may be a feature of the mouse strain background. FPLC cholesterol profiles of the plasma (panel c, males, panel d, females) showed that the diet in males led to a dyslipidemic profile, but loss of Stap1 did not affect these patterns. Error bars denote ±1 SD causing hyperlipidemia in humans, a mechanistic pathway to do so would need to be identified. Based upon our current understanding of lipid metabolism, no such meaningful pathway exists.

Conclusion
Global loss of Stap1 in mice does not result in an abnormal lipid phenotype, and this study validates the study by Loaiza et al, who reported similar findings. Additionally, expression STAP1 mRNA seems to be confined to B-cells, suggesting alteration of lipid metabolism in a direct manner to be not feasible. Genetic changes in STAP1 are therefore unlikely to be cause of familial hypercholesterolemia and variants reported in such rare families may represent a spurious association.
Additional file 2. Sequencing of off-target products on chromosome 2.
Additional file 3.   a and b). This was confirmed by analyses of liver cholesterol (panel c, n = 6 per group in males and 5 per group in females) and triglyceride (panel d, n = 6 per group in males and 5 per group in females) but loss of Stap1 does not affect these parameters. Error bars denote ±1 SD