Cathepsin B is a novel gender-dependent determinant of cholesterol absorption from the intestine.

We used a mouse C57BL/6J×CASA/Rk intercross to map a locus on chromosome 14 that displayed a gender-dependent effect on cholesterol absorption from the intestine. Studies in congenic animals revealed a complex locus with multiple operating genetic determinants resulting in alternating gender-dependent phenotypic effects. Fine-mapping narrowed the locus to a critical 6.3 Mb interval. Female subcongenics, but not males, of the critical interval displayed a decrease of 33% in cholesterol absorption. RNA-Seq analysis of female subcongenic jejunum revealed that cysteine protease cathepsin B (Ctsb) is a candidate to explain the interval effect. Consistent with the phenotype in critical interval subcongenics, female Ctsb knockout mice, but not males, displayed a decrease of 31% in cholesterol absorption. Although studies in Ctsb knockouts revealed a gender-dependent effect on cholesterol absorption, further fine-mapping dismissed a role for Ctsb in determining the effect of the critical 6.3 Mb interval on cholesterol absorption.

enterocytes in the form of chylomicrons. Although clarifi cation of these metabolic steps largely increased our understanding of the absorption process, the molecular mech anisms in control of cholesterol absorption are not well understood.
To identify genes in control of the absorption process, we used an intercross between two inbred mouse strains, C57BL/6J and CASA/Rk, that differ markedly in cholesterol absorption rates and plasma plant sterol levels ( 4 ), a surrogate measure of cholesterol absorption. This effort allowed mapping of a locus on chromosome 14 that we designated Plast14 with a linkage map that displayed two broad peaks at markers D14Mit18 (48.4 Mb) and D14Mit123 (66.8 Mb) ( 4 ). To validate the effect of the Plast14 locus, we fi rst generated a congenic strain that was designated 14KK and consisted of a chromosome 14 CASA/Rk 100.1 Mb interval introgressed onto the C57BL/6J genetic background. Studies on cholesterol absorption in 14KK congenics revealed a gender-dependent effect, with decreased cholesterol absorption observed in males, but not in females. Interestingly, generation of two subcongenic strains, 14PKK and 14DKK, which consisted of the centromeric 20.8 Mb and telomeric 70.4 Mb of the 14KK interval, respectively, displayed decreased cholesterol absorption in both genders ( 5 ). Here we describe the fi ne-mapping of the 14DKK subcongenic interval. These studies revealed a complex alternated gender-dependent effect on cholesterol absorption. Furthermore, jejunum RNA-Seq analysis in subcongenics of the critical 6.3 Mb interval, along with studies in knockout mice, identifi ed cathepsin B (Ctsb), a ubiquitously expressed lysosomal cysteine protease, as a novel gender-dependent determinant of cholesterol absorption from the intestine.
Abstract We used a mouse C57BL/6J×CASA/Rk intercross to map a locus on chromosome 14 that displayed a genderdependent effect on cholesterol absorption from the intestine. Studies in congenic animals revealed a complex locus with multiple operating genetic determinants resulting in alternating gender-dependent phenotypic effects. Finemapping narrowed the locus to a critical 6.3 Mb interval. Female subcongenics, but not males, of the critical interval displayed a decrease of 33% in cholesterol absorption. RNA-Seq analysis of female subcongenic jejunum revealed that cysteine protease cathepsin B (Ctsb) is a candidate to explain the interval effect. Consistent with the phenotype in critical interval subcongenics, female Ctsb knockout mice, but not males, displayed a decrease of 31% in cholesterol absorption. Although studies in Ctsb knockouts revealed a gender-dependent effect on cholesterol absorption, further fi ne-mapping dismissed a role for Ctsb in determining the effect of the critical 6. The absorption of dietary cholesterol from the intestine is an important determinant of plasma cholesterol level, a major risk factor for atherosclerotic cardiovascular diseases. The uptake of dietary cholesterol is critically dependent on Niemann-Pick C1 like 1 (NPC1L1), a transmembrane permease that localizes to the brush border membrane and endocytic recycling compartment of intestinal enterocytes (1)(2)(3). The uptake of cholesterol across the apical membrane is followed by esterifi cation at the endoplasmic reticulum by sterol O -acyltransferase 2 (ACAT2), packaging, and secretion from the basolateral pole of corrected for multiple testing using false discovery rate (FDR). In addition to analysis of expression level, RNA-Seq allows querying for SNPs. SNPs were identifi ed by aligning jejunum sequences from subcongenic G mice to the mm 9 C57BL/6J reference mouse genome (UCSC). Functional signifi cance of nonsynonymous SNPs was examined for potential deleterious effects by using the dbNSFP database ( 8 ).

Cathepsin B proteolytic activity assay
Cathepsin B activity was determined using the fl ourimetric innozyme kit (EMD Biosciences; Billerica, MA, Cat# CBA001). In brief, freshfrozen jejunum samples were homogenized in extraction buffer, centrifuged (14,560 g , 20 min, 4°C), and total protein was determined using Bicinchoninic Acid Assay reagents. Samples were assayed in duplicate of 50 µg protein. After 30 min incubation at 37°C, activity was determined by excitation at 380 nm and emission at 460 nm wavelengths. Ctsb activity was expressed as cleaved fl uorescent 7-amino-4-methylcoumarin product per mg protein per min.

Jejunum quantitative PCR
Tissues were kept at Ϫ 80°C in RNA-later solution (Invitrogen), until processed. Frozen tissues were homogenized in Trizol reagent using TissueLyser II (Qiagen) according to the manufacturer's instructions. Total RNA samples were treated with DNase I and reverse-transcribed using Superscript II (Invitrogen). Gene expression levels were determined using the SYBR Advantage qPCR Premix (Clontech; Mountain View, CA) on a BioRad CFX384 real-time PCR detection system. Specifi c gene expression was inferred from gene-specifi c standard curves after normalization to either polyadenylate binding protein 1 (Pabpc1) or ␤ -actin.

Statistical analysis
Group differences in cholesterol absorption, gene expression, and Ctsb activity were analyzed using Student's unpaired t -test. Unless otherwise indicated, values throughout this manuscript are expressed as mean ± SD.

The chromosome 14 locus displays a complex genderdependent effect on cholesterol absorption
We have previously reported that the 14DKK subcongenic strain (herein designated as subcongenic strain A), consisting of a CASA/Rk chromosome 14 interval of 70.4 Mb introgressed onto the C57BL/6J background, displayed

Animal breeding and genotyping
Generation of chromosome 14DKK congenics (here referred to as subcongenic A) on the C57BL/6J genetic background has been previously described ( 5 ). Construction of further subcongenic strains of the 14DKK interval was generated through backcrossing onto C57BL/6J animals, followed by microsatellite markers and single-nucleotide polymorphism (SNP) genotyping for identifi cation of recombinant events. Recombinant animals were then bred to homozygosity and studied for cholesterol absorption as described below. Ctsb knockout (Ctsb-KO), Gulonolactone oxidase knockout (Gulo-KO), and dedicator of cytokinesis-5 knockout (Dock5-KO) ( 6 ) mice, all on the C57BL/6J genetic background, were generously provided by Dr. Terence Dermody at Vanderbilt University School of Medicine and Dr. Nobuyo Maeda at the University of North Carolina, respectively. Ascorbic acid-defi cient Gulo-KO mice were given drinking water supplemented with L -ascorbic acid (330 mg/l; Sigma Aldrich, Cat# A7506) ad libitum. Animals were housed in a 10/14 h light/ dark cycle with free access to chow diet (18% protein, 44% carbohydrate, 6% fat; Harlan Laboratories, Teklad Cat# 2918) and water. All experiments were approved by the Institutional Animal Care and Use Committee.

Cholesterol absorption rates
Cholesterol absorption studies were conducted in subcongenic strains and in knockout homozygous or heterozygous mice at 12-15 weeks of age. Animals were individually housed in metabolic cages and received a gastric bolus of 50 µl olive oil supplemented with [ 14 C]cholesterol and [ 3 H] ␤ -sitostanol. After 24 h, feces were collected, dried overnight, ground with a mortar and pestle, extracted in Folch solution (2:1 choloroform-methanol), and counted for radiolabeled lipids, and cholesterol absorption rates were calculated as previously described ( 7 ).

RNA-Seq analysis of jejunal tissue
Five female C57BL/6J and fi ve female homozygous subcongenic G animals were fasted for 6 h, anesthetized by intra-muscular injection of ketamine/xylazine, and euthanized toward the end of the light phase of the light-dark cycle. Intestinal lumens were fl ushed with ice-cold PBS, and the proximal 5 cm of the jejunum (distal to the Treitz ligament) was harvested and stored in RNAlater solution (Invitrogen; Carlsbad, CA) at 4°C overnight. Jejunal total RNA was extracted using Trizol reagent (Invitrogen), and polyadenylated RNA was isolated using the Oligotex mRNA kit (Qiagen; Valencia, CA). Polyadenylated mRNA was then processed according to Illumina protocol. Briefl y, the mRNA was fragmented using divalent cations, and double-stranded cDNA was synthesized using random primers. After conversion of overhangs into blunt ends, an adenosine base was added at the 3 ′ end and sequencing adapters ligated to the cDNA. Fragments of ‫ف‬ 200 bp were selected through gel excision and purifi ed before amplifi cation. The resulting libraries were then sequenced using an Illumina Genome Analyzer IIx to generate 51 bp paired-end reads. Sequencing reads were then aligned against a reference mouse genome (UCSC mm 9) using TopHat (v1.0.14; doi:10.1093/bioinformatics/btp120) and reads that were not properly paired were discarded from further analysis. For each RNA sample, we obtained an average total of 29,056,676 ± 3,084,265 properly paired reads. For this analysis, we focused on well-annotated genes and calculated, for each sample, the number of reads mapped to each RefSeq gene. We then used negative binomial regression analysis to test whether gene expression was statistically associated with the animals' genotypes (after exclusion of outlier samples: >3 SD of the mean). All P values were genes (FDR < 0.05) none of which are involved in either cholesterol absorption or lipoprotein assembly. The subcongenic G 6.3 Mb interval harbors a total of 77 RefSeq genes, out of which 11 genes were differentially expressed using the genome-wide signifi cance cutoff. As shown in Table 1 , qPCR studies confi rmed differential expression in two genes, cathepsin B (Ctsb) and gulonolactone oxidase (Gulo), but did not confi rm the remaining nine genes (six genes displayed expression below the qPCR detectable limits, whereas in the other three genes, differential a decrease of 49% and 40% in cholesterol absorption in males and females, respectively ( 5 ). To identify cholesterol absorption-modifying genes in the subcongenic A interval, we fi rst constructed six subcongenic strains that were designated subcongenics B through G ( Fig. 1A and supplementary Table I ). As shown in Fig. 1B and C , male and female subcongenic B animals, which cover the telomeric 46.7 Mb of strain A, had no effect on cholesterol absorption. In contrast, subcongenic strains C through G, which span the centromeric end of the subcongenic A interval, displayed a complex gender-dependent effect. In strain C, which covers the centromeric 31.6 Mb of the subcongenic A interval, subcongenic females, but not males, displayed a decrease of 32% in cholesterol absorption. Interestingly, strains D and E, which cover the telomeric 20.1 Mb and centromeric 12.1 Mb of the subcongenic C interval, respectively, displayed the opposite gender effect, in which subcongenic males, but not females, displayed a decrease of 40% and 45% in cholesterol absorption, respectively. Moreover, recombinant events within the strain E interval, which generated strains F and G, resulted in yet another alternated gender effect. Therefore, strains F and G, which cover the centromeric 5.5 Mb and telomeric 6.3 Mb of the subcongenic E interval, displayed a decrease of 25% and 33% in cholesterol absorption in subcongenic females, but not in subcongenic males. These fi ndings strongly suggest that the centromeric 32 Mb of the subcongenic A CASA/Rk interval harbors several gender-dependent determinants of cholesterol absorption from the intestine.

Heritability mode of cholesterol absorption in subcongenic G females
We previously published the linkage map on chromosome 14 ( 4,5 ). This map is characterized by two peaks, one at D14Mit18 and the other at D14Mit123, that map at 48.4 Mb and 66.8 Mb, respectively. The second peak at 66.8 Mb maps within the subcongenic G interval. Therefore, we focused our gene discovery efforts on the CASA/ Rk G interval. First, we examined the heritability mode of the cholesterol absorption trait in subcongenic G females. As shown in Fig. 2A , whereas subcongenic G females that are homozygous for the CASA/Rk G interval displayed signifi cantly decreased cholesterol absorption rates, heterozygous females displayed cholesterol absorption rates that were indistinguishable from their control C57BL/6J females. These fi ndings indicate that in subcongenic G females, suppression in cholesterol absorption is inherited in an autosomal recessive mode.

RNA-Seq analysis of jejunum tissues identifi ed Ctsb, Gulo, and Dock5 as three candidate genes to explain the G interval effect on cholesterol absorption
To identify cholesterol absorption-modifying genes at the subcongenic G interval, we subjected the jejuna of homozygous subcongenic G and C57BL/6J control females to RNA-Seq analysis. As detailed in supplementary Fig. I , this effort recovered a total of 28 million properly paired reads that were aligned to the mouse genome, and identifi ed 674 genome-wide differentially expressed RefSeq

Studies in knockout animals confi rmed that Ctsb is a gender-dependent determinant of cholesterol absorption from the intestine
Subcongenic G females displayed an increase of 12.5fold in Gulo expression levels ( Table 1 ). To determine whether decreased jejunal synthesis of vitamin C may increase the absorption of cholesterol in C57BL/6J females, we examined the effect of ascorbic acid supplementation on cholesterol absorption. As shown in Fig. 3A , C57BL/6J females treated with ascorbic acid displayed cholesterol absorption rates that were indistinguishable from their untreated controls. These fi ndings did not support a role for vitamin C in modifying the absorption of cholesterol from the intestine. Next, we examined the absorption of cholesterol in Gulo, Dock5, and Ctsb knockouts, all on the C57BL/6J background. As shown in Fig. 3B , Gulo knockout females and males displayed cholesterol absorption rates that were indistinguishable from those of their C57BL/6J controls. These fi ndings did not support a role for Gulo in decreasing the absorption of cholesterol in subcongenic G females. Next, we examined the role of Dock5. As shown in Fig. 3C , Dock5 knockout females and males displayed cholesterol absorption rates that were similar to those of their age-matched C57BL/6J controls. Again, these fi ndings did not support a role for Dock5 in modifying the absorption of cholesterol. Finally, to examine the role of the Ctsb gene, we studied the Ctsb knockout animals (Ctsb-KO). When compared with their C57BL/6J controls, Ctsb-KO animals were defi cient in Ctsb RNA ( Fig. 4A ), Ctsb proteins ( Fig. 4B ) and their jejunum homogenates lacked Ctsb proteolytic activity ( Fig. 4C ). Interestingly, as shown in Fig. 4D , when compared with control C57BL/6J females, Ctsb-KO females displayed a signifi cant 31% decrease in cholesterol absorption (75.2 ± 8.0% vs. 51.7 ± 11.4% in control and Ctsb-KO mice, respectively; P < 0.001). Moreover, targeting of Ctsb displayed a gender-dependent effect on cholesterol absorption, inasmuch as Ctsb-KO males displayed absorption that was indistinguishable from that of control males. It is of note, however, that heterozygous Ctsb females displayed a 50% decrease in Ctsb expression ( Fig. 4A ), which was associated expression was not confi rmed). Ctsb is a ubiquitously expressed lysosomal cysteine protease. When compared with the jejunum of C57BL/6J control females, the jejunum of subcongenic G females displayed a 42% decrease in Ctsb expression ( Table 1 ) and a 30% decrease in the proteolytic activity of this enzyme ( Fig. 2B ). Gulo catalyzes the conversion of gulonolactone to L -ascorbic acid, the fi nal step in the vitamin C biosynthesis pathway ( 9 ). When compared with C57BL/6J controls, subcongenic G females displayed a 12.5-fold increase in jejunum Gulo expression ( Table 1 ). RNA-Seq allows querying expressed genes for SNPs. Comparison of RNA sequences in C57BL/6J control and subcongenics of the CASA/Rk G interval identifi ed 502 SNPs, out of which 26 were nonsynonymous ( Table 2 ). We examined the functional signifi cance of these sequence variations by querying the dbNSFP database, which allows functional predictions of amino acid substitutions ( 8 ). As shown in Table 2 , out of 26 nonsynonymous SNPs, seven code for highly conserved amino acids, with three predicted to have deleterious effects. Interestingly, all three deleterious amino acid substitutions mapped to a single gene, dedicator of cytokinesis 5 (Dock5), a guanine nucleotide exchange factor that activates the small-G protein Rac.  Results are expressed as mean properly paired reads aligned to each RefSeq gene after correction for outliers (>3 SD); N = 5 per genotype; UD, undetectable . on cholesterol absorption. This set of experiments strongly suggests that suppressed Ctsb expression cannot, by itself, explain the suppression of cholesterol absorption by the subcongenic G interval.
with cholesterol absorption indistinguishable from that of their controls ( Fig. 4D ). These fi ndings were in disagreement with our Ctsb expression and phenotype in subcongenic G females, where a 58% decrease in Ctsb expression ( Table 1 ) was associated with a 33% decrease in cholesterol absorption ( Fig. 1B ). Taken together, the fi ndings in heterozygous Ctsb females did not support a role for Ctsb in modifying the absorption of cholesterol in subcongenic G females.

Suppressed Ctsb expression does not explain the effect of the subcongenic G interval on cholesterol absorption
Decreased cholesterol absorption in Ctsb-KO females unequivocally attested to its role in modifying the absorption in female mice ( Fig. 4D ). In contrast, the absence of a phenotypic effect in Ctsb heterozygous females ( Fig. 4D ) challenged the role of Ctsb in suppressing the absorption of cholesterol in subcongenic G females. To resolve the role of Ctsb, we examined the absorption in recombinant strains of the subcongenic G interval. As shown in Fig. 1A , we generated three additional subcongenic strains, designated as strains H through J, that covered the telomeric 4.5 Mb (strain H), and centromeric 2.8 and 1.8 Mb (strains I and J, respectively) of the CASA/Rk G interval. It is of note that the telomeric H strain harbors the Ctsb C57BL/6J allele, whereas both centromeric strains (I and J) harbor the CASA/Rk allele of Ctsb. Interestingly, all three strains, regardless of the Ctsb allele genotype, displayed significantly decreased Ctsb expression ( Fig. 5 ). However, as shown in Fig. 1B, C , none of these strains had an effect  These alternating effects strongly suggest that i ) the subcongenic A interval harbors a number of genetic determinants in control of the gender effect, and ii ) these genetic determinants are in complex interactions in determining the phenotype. Further studies are needed to identify these genetic determinants and to clarify the nature of the interactions. Fine-mapping of the chromosome 14 locus, together with studies in Ctsb-KO mice, identifi ed Ctsb as a genderdependent determinant of cholesterol absorption from the intestine. Ctsb is a ubiquitously expressed cysteine protease that has been localized to lysosomal, endosomal, plasma membrane, and the extracellular compartments. Furthermore, Ctsb has been shown to operate at acidic and neutral pH optima of 4.5-5.5 and 7.4, respectively ( 13,14 ). Therefore, Ctsb has the potential to affect different DISCUSSION In this study, we fi ne-mapped a chromosome 14 locus that displays a gender-dependent effect on cholesterol absorption from the intestine down to a critical interval of 6.3 Mb. Jejunum RNA-Seq analysis in subcongenic females of the critical interval identifi ed Ctsb as a candidate gene to explain the interval effect. Studies in Ctsb knockout animals confi rmed a role for Ctsb as a determinant of cholesterol absorption in female but not in male mice. However, further fi ne-mapping of the critical interval dismissed the role of Ctsb in determining the critical 6.3 Mb interval effect on cholesterol absorption.
Previous mouse studies indicated a gender-dependent effect on cholesterol absorption ( 10,11 ). In humans, a population survey of plasma plant sterol levels on the Micronesian island of Kosrae ( 12 ) found signifi cantly lower plasma levels in females. Nevertheless, the molecular basis of this gender dimorphic effect is unknown. We used a mouse cross between inbred strains C57BL/6J and CASA/Rk to identify a locus on chromosome 14 in control of plasma plant sterol levels. A congenic strain of this locus (14KK) clearly displayed a gender-dependent effect ( 5 ). In agreement with their parental 14KK congenics, a gender-dependent effect clearly characterized the phenotype in subcongenics that covered the telomeric 70.4 Mb of the 14KK interval (herein designated as subcongenics A-G). Moreover, subcongenic strains C-G, which covered the centromeric 31.6 Mb of the A interval, displayed an alternating gender effect on cholesterol absorption ( Fig. 1A, B ).  steps along the absorption process. For example, Npc1l1 has been shown to play a critical role in the absorption process. Localization of Npc1l1 to fl otilin-associated cholesterol-enriched microdomains is followed by clathrinmediated vesicular endocytosis to the endocytic recycling compartment ( 2,3 ). Localization of Ctsb to the enterocyte brush border membrane or the endocytic compartment may interfere with Npc1l1 recycling, thereby modifying the absorption process. Moreover, Ctsb may modify the transfer of cholesterol from the endocytic recycling compartment to the endoplasmic reticulum and thereby decrease the esterifi cation and packaging of cholesterol into chylomicrons. Understanding the cholesterol absorption inhibitory mechanisms that operate in females is expected to clarify the gender-dependent effect of Ctsb on the absorption process.
It is of note that whereas subcongenic G females displayed signifi cantly decreased cholesterol absorption rates, subcongenic strains that cover the centromeric and telomeric ends of this interval (subcongenics H through J) had no effect on cholesterol absorption. In subcongenics H, I, and J, decreased Ctsb expression with no effect on cholesterol absorption strongly challenges the role of this gene in determining the critical 6.3 Mb interval phenotypic effect. It is therefore likely that the subcongenic G interval harbors two or more genetic determinants that act in a mutually inclusive fashion to control the absorption of cholesterol. Our RNA-Seq experiment, in conjunction with our studies in Ctsb, Gulo, and Dock5 knockouts, strongly challenges the role of RefSeq genes at the critical interval. Further experimentation is needed to identify the genetic determinants at this interval. It is also of note that subcongenics G, I, and J are carriers of the CASA/Rk allele of Ctsb, whereas subcongenic H animals are carriers of the corresponding C57BL/6J allele ( Fig. 1A ). Interestingly, regardless of the Ctsb allele, females of all four subcongenic strains uniformly displayed a decrease of ‫ف‬ 50% in Ctsb expression ( Table 1 and Fig. 5 ). These fi ndings strongly suggest that the critical CASA/Rk 6.3 Mb interval harbors two or more CASA/Rk genetic elements that act in a mutually exclusive fashion to suppress the expression of Ctsb. It is likely that one or more elements map to the telomeric 4.5 Mb of the CASA/Rk critical interval (which corresponds to the subcongenic H interval), and the others map to the centromeric 1.9 Mb of this interval (and correspond to the subcongenic I and J intervals).
In summary, fi ne-mapping studies along with gene expression efforts and studies in knockout animals identifi ed Ctsb as a gender-dependent modifi er of cholesterol absorption from the intestine. To our knowledge, this is the fi rst report on the involvement of a cysteine protease in the