Mice with Alopecia, Osteoporosis, and Systemic Amyloidosis Due to Mutation in Zdhhc13, a Gene Coding for Palmitoyl Acyltransferase

Protein palmitoylation has emerged as an important mechanism for regulating protein trafficking, stability, and protein–protein interactions; however, its relevance to disease processes is not clear. Using a genome-wide, phenotype driven N-ethyl-N-nitrosourea–mediated mutagenesis screen, we identified mice with failure to thrive, shortened life span, skin and hair abnormalities including alopecia, severe osteoporosis, and systemic amyloidosis (both AA and AL amyloids depositions). Whole-genome homozygosity mapping with 295 SNP markers and fine mapping with an additional 50 SNPs localized the disease gene to chromosome 7 between 53.9 and 56.3 Mb. A nonsense mutation (c.1273A>T) was located in exon 12 of the Zdhhc13 gene (Zinc finger, DHHC domain containing 13), a gene coding for palmitoyl transferase. The mutation predicted a truncated protein (R425X), and real-time PCR showed markedly reduced Zdhhc13 mRNA. A second gene trap allele of Zdhhc13 has the same phenotypes, suggesting that this is a loss of function allele. This is the first report that palmitoyl transferase deficiency causes a severe phenotype, and it establishes a direct link between protein palmitoylation and regulation of diverse physiologic functions where its absence can result in profound disease pathology. This mouse model can be used to investigate mechanisms where improper palmitoylation leads to disease processes and to understand molecular mechanisms underlying human alopecia, osteoporosis, and amyloidosis and many other neurodegenerative diseases caused by protein misfolding and amyloidosis.

Palmitoyl post-translational modification has recently emerged as an important mechanism for modulating protein targeting, trafficking, stability and protein-protein interactions, and plays roles in numerous cellular processes, including signaling, apoptosis and neuronal transmission [1,3].
Although palmitoylation was first described over 30 years ago, the genes coding for enzymes involved in protein palmitoylation, the palmitoyl acyltransferase (PATs), have only recently been discovered [7,9]. To date, at least 23 members of PATs have been identified in the mammalian genome [9,10]. This family of proteins contains a cysteine-rich domain (CRD) with a core Asp-His-His-Cys (DHHC) motif that is essential for PAT activity [7,9,11]. The presence of so many PATs in a single organism could be due to differences in substrate specificities, intracellular localizations or tissue distributions. For example, DHHC2 and DHHC15 are more specific to PSD-95 and GAP-43, DHHC9 and DHHC18 are specific to H-Ras and N-Ras, while DHHC3 and the closely related DHHC7 have broad substrate specificities [2,8]. In neuronal tissue, DHHC13 and 17 modulate huntingtin palmitoylation and DHHC8 modulates paralemmin-1 [11]. The substrate specificity appears to be determined by the regulatory domains outside the DHHC domains of the enzymes [8,11].
Despite the functional importance of protein palmitoylation at the cellular and biochemical levels, its physiological role and its relevance to disease processes is not clear. Oncogenic Ras proteins and huntingtin are direct targets for palmitoylation thus, they may be involved in the disease process. Disregulation of DHHC2 may be involved in cancer metastasis [12]. The sole mouse model of DHHC deficiency is the Zdhhc8 knockout; these mice have a mild behavior phenotype with a decrease in exploratory activity and a deficiency in prepulse inhibition. These behavioral changes are only observed in female mice. The phenotypes together with genetic evidence may support the hypothesis that DHHC8 is a risk factor for schizophrenia [13]. Systematic knockdown of the Zdhhc genes has not been done, which could provide some unexpected physiological roles for DHHC proteins. Here we report on mice with a mutation in the Zdhhc13, a gene coding for palmitoyl acyltransferase, which catalyzes the reaction of protein palmitoylation [11]. Mutant mice exhibit a severe phenotype and profound pathology involving multi-organ/systems. These mice, developed cachexia, alopecia, osteoporosis, systemic amyloidosis, failed to thrive and succumbed to early death.

Results
The mutant mice were analyzed either on a C57BL/66129S6/ SvEv or C57BL/66129S6/SvEv 6 C3H mixed genetic background. The phenotypes described herein are all penetrant in both genetic backgrounds.

Clinical Phenotypes
General appearance. Affected mice appeared normal at birth, but by postnatal day 7 were small in size and developed hypotrichosis; these features differentiated affected from normal siblings. The affected male mice had poor weight gain and weighed 50% less than the unaffected siblings ( Figure 1A). Affected females mice also showed similar poor weight gain (data not shown). In addition, these mice, regardless of sex, had a shortened life span; about 50% died before 7.5 months of age and only 20% survived beyond one year of age ( Figure 1B). Furthermore, the affected mice showed generalized hypotrichosis and hair loss particularly certain body parts, some hairs remained over the head and back, although, they were thin, short and had decreased luster ( Figure 2). Skin was loose with wrinkling and folding ( Figure 2C). Kyphosis was evident beginning at day 28 ( Figure 2, also shown in Figure 3A). When the gene responsible for these phenotypes was identified as Zhddc13 (see below under Identification of the mutated gene), it was clear that only homozygous Zdhhc13 -/-

Author Summary
Palmitoylation, the addition of palmitate (a fatty acid) to protein, is one of the most common post-translational lipid modifications and has recently emerged as an important mechanism for modulating protein targeting, trafficking, stability, and protein-protein interactions. However, its physiological role and its relevance to the disease processes are not at all clear. Here we reported that mice with mutation in Zdhhc13, a gene coding for palmitoyl acyltransferase that catalyzes the reaction of protein palmitoylation, exhibited a severe phenotype and profound pathology involving multi-organ/systems. These mice showed wasting, weight loss, hair loss (alopecia), reduced bone mineral density (osteoporosis), and generalized amyloid deposition, which resulted in early death. Our results established a direct link between protein palmitoylation and regulation of the important diverse physiological functions and indicated that its absence can result in profound disease pathology. This mouse model will be useful for further understanding the molecular mechanisms underlying human alopecia, osteoporosis, and many other neurodegenerative diseases caused by protein misfolding and amyloid deposition. exihibited abnormal phenotypes, while heterozygous Zhddc13 +/2 displayed normalcy like the wild-type (+/+).
Hematology and blood chemistry. Complete blood counts obtained from affected mice at 4 weeks and 30 weeks of age were comparable to the wild-type, except that adult mutant mice showed neutrophilia (mutant = 58.862. 24    elevations in BUN (up to 180 mg/dl), CPK and total bilirubin. Serum calcium, magnesium and C-reactive protein were all within normal limits in the affected mice at both 4 weeks and 30 weeks of age.
Bone studies. Radiographic examinations showed severe kyphosis with marked increased spinal angle in the affected mice ( Figure 3A). Osteoporosis was also profound, as evidenced by a decrease in the trabecular number of femur ( Figure 3B and Table 1) and by other trabecular bone parameters, including a decrease in bone volume density (BV/TV) and bone mineral density (BMD), along with an increase in the structure model index (SMI), which indicated an abundance of rod-like trabeculae ( Table 1). These features of osteoporosis could be seen as early as 4 weeks of age (data not shown).
Histopathological analyses. Post mortem examinations revealed hepatosplenomegaly (2-3 times normal), severe muscle wasting and reduced white and brown adipose tissue. Skin histopathology of the affected mice showed hyperkeratosis and epidermis hyperplasia with thin dermis and scanty adipose tissue ( Figure 4A). Hair follicles in different stages, such as anagen, catagen, and telogen, could be observed in wild-type mice ( Figure 4D), while mutant mice had significantly fewer active hair follicles with most remaining in the late telogen phase. There were no hair shafts in the affected hair follicles and the upper portion was dilated ( Figure 4C). Amyloid deposition was observed in the entire dermis in the affected mice as homogenous eosinophilic substances by H&E stain ( Figure 5A). This was confirmed by Congo red stain as pink-red deposits ( Figure 5B) and under a polarizing microscope as yellowgreen birefringence ( Figure 5C).
Amyloid depositions were also found in most of the other major organs examined, except for muscle. Amyloids were seen in liver, spleen, kidney, adrenal gland, pancreas, salivary glands, heart, lung, intestine and brain. The amyloids deposited in organs and their severity in young (4 weeks) and older animals (20-32 weeks) are summarized in Table 2. In general, there was a progressive increase of amyloid with age.
In the liver, amyloid deposits were observed mainly in sinusoids around perivascular areas (Portal and central veins) ( Figure 6A). Enlarged Kupffer cells containing amyloid substance were also seen and in severe cases, massive amyloid deposits disrupted the hepatic architecture (data not shown). Mild amyloid deposition in the space of Disse or sinusoid was also found in young mice as early as 4 weeks of age (Table 2).
In the spleen, accumulations of amyloid appeared in peri-white pulp, connective tissue frameworks of red pulp and central arterial walls of white pulp ( Figure 6B). Mild amyloid deposition was also observed in young animals; however, the arterial wall findings were not detectable in young mutants ( Table 2).
In the kidney, amyloid accumulated in the glomerulus ( Figure 6C), renal tubules ( Figure 6D) and perivascular areas. Thickenings of the basement membranes of glomerular capillaries and mesangial matrix were common lesions in mutants. In the advanced stage, the glomerular capillaries were obliterated, and the glomerular structures were completely destroyed. The renal tubules became dilated with large amounts of filtrated substances (as an eosinophilic substance, presumably albumin that is observed as a result of a glomerular filtration defect) and some tubular epithelial cells showed atrophy. In young animals, renal amyloidosis was milder, but could be seen in both glomerulus and tubules ( Table 2).
Massive amyloid deposits were also observed in the adrenal glands in adult mice. Amyloid accumulated in the zona fasciculate, zona reticularis, and in part of the zona glomerulosa and medulla ( Figure 6E).
In the pancreas, amyloid accumulated in both exocrine pancreatic tissues and islet of Langerhans ( Figure 6F) and, in the advanced stage, acinar cells were degenerated and completely replaced by amyloid. Acinar cells in salivary glands were also full of amyloid ( Figure 6G).
In the heart, amyloid deposition could be seen in the blood vessel walls, and only in severe cases, amyloid was observed in the myocardium ( Figure 6H). Small amounts of amyloid deposits could also been found in other organs such as lung, intestine and brain. No amyloid deposition was found in skeletal muscle.

Immunohistochemistry Staining
Immunohistochemistry staining was performed using antiamyloid A, anti-kappa light chain and anti-lambda light chain to confirm the amyloidosis and to differentiate AA and AL type of amyloidosis. In liver, amyloid AA and AL k were the major amyloid detected, primarily in sinusoids and around the portal vein, while the amount of AL l amyloid was less ( Figure 7, upper panel). A similar pattern was also observed in kidney glomerulus and tubular cells in which AA and AL k were the predominant amyloids ( Figure 7, lower panel). Both AA and AL k could also be detected in the kidneys of the young animals ( Table 2).

Identification of the Mutant Gene
To map the gene responsible for these abnormal phenotypes, affected mice in the B66129 mix background were out-crossed to C3H/HeJ mice to generate N1 (B6 and C3H hybrid) affected offspring, then were intercrossed. The offspring of this intercross (N1F1) were used in the genomic analysis. Whole genome SNP homozygosity mapping revealed one region located between 46.4 and 64.7 Mb (18.3 Mb) of chromosome seven with 90% B6 homozygosity in consecutive SNPs ( Figure 8A). Fine mapping narrowed down the candidate region to within 2.4 Mb (between 53.9 and 56.3 Mb) on chromosome 7 ( Figure 8B). This region contained 64 genes. Because our affected mice showed generalized amyloidosis, we concentrated on the amyloid related genes located in this region which included Saa1l (serum amyloid A-like 1), Saa3 (serum amyloid A 3), Saa4 (serum amyloid A 4), Saa1 (serum amyloid A 1), Saa2 (serum amyloid A 2) and Zdhhc13 (zinc finger, DHHC domain containing 13). Direct DNA sequencing of genes in affected mice revealed a homozygous A to T substitution in exon 12 of Zdhhc13 (c.1273A.T) ( Figure 8C). The parents were heterozygous for this mutation. Further study showed that the homozygous c.1273A.T mutation (2/2) completely segregated with the abnormal phenotypes. Siblings, as well as parents, that were heterozygous for this mutation (+/2) were phenotypically normal. This A to T substitution resulted in a stop codon (AGA.TGA) (arg-425-stop codon) and predicted a truncated protein. No other mutations were found in the remaining exons of gene Zdhhc13 or in its promoter region. All other amyloid-related genes in the candidate region were also normal without detectable mutations. Real time RT-PCR showed that the tissue expression of Zdhhc13 mRNA was significantly reduced to 26.23% in the liver and 15.59% in the kidney of the affected mice compared to the wild type, ( Table 3).
The decreased mRNA indicated nonsense RNA decay due to the premature stop codon in our mutant Zdhhc13 mice.
To confirm that we had identified the correct gene, we obtained a gene trap allele in Zdhhc13 from the SIGTR. This allele produced the same phenotype as the ENU-induced allele, with the

Expression of Zdhhc13
We carried out Northern analysis of a variety of tissues to determine where Zdhhc13 is expressed. We found that Zdhhc13 is expressed in most adult tissues, but at low levels in the liver, skin, and lung ( Figure 9A). We hypothesized that Zdhhc13 may be needed most during development of these tissues, so we analyzed expression of Zdhhc13 in liver, skin, lung and brain at three different time points: postnatal day (P) 2, P8 and P30. Zdhhc13 was expressed most highly in the liver, lung, and brain at P2, showing that transcripts are developmentally regulated ( Figure 9B). In contrast, Zdhhc13 is expressed most highly in skin at P8, when hair follicles are maturing. We examined the gene trap allele, which contains a lacZ reporter gene, for expression of Zdhhc13 in the skin, and found that it is expressed in the epithelium surrounding the hair follicles, consistent with a role in hair growth ( Figure 9).

Mutation in Zdhhc13 Affects Protein Palmitoylation
Huntingtin is a known substrate of Zdhhc13 [11]. To demonstrate the palmitoylation defect caused by the mutation, HEK 293 T cells were co-transfected with huntingtin-myc and Zdhhc13-flag (WT and mutant). We examined the palmitoylation levels using acyl-biotin exchange assay after immunoprecipitation of huntingtin with anti-myc antibody and found that the huntingtin was palmitoylated by the wild-type Zdhhc13. The ability of mutant Zdhhc13 to palmitoylate huntingtin was greatly reduced by the mutant Zdhhc13, to a level indistinguishable from the endogenous palmitoyl activity present in the control ( Figure 10A).
Moreover, when we examined the IgG light chain purified from serum of wild and mutant mice, we observed much less  Adrenal gland Cortex  palmitoylated signals in the mutant mice as compared to the wild type mice ( Figure 10C).

Discussion
Using ENU-mutagenesis, we have identified mice with severe phenotypes manifested by failure to thrive, alopecia, osteoporosis, systemic amyloidosis and early death. We found that a nonsense mutation (R425X) in Zdhhc13 was the cause of these abnormal phenotypes. To ensure that other mutations induced by ENU would not confound the observed phenotypes, mice examined after 6 generations of outcross breeding continued to show 100% phenotype and genotype correlations [14]. Further, a gene trap allele of Zdhhc13, which contains a vector inserted into the first intron, exhibits the similar phenotypes (data not shown).
Zdhhc13, also named huntingtin-interacting protein-like (HIP14L), shares 51% identity and 69% similarity with huntingtin-interacting protein-14 (HIP14) or Zdhhc17 between #45 and #611 of the 622 residues [15]. Both Zdhhc13 and 17 belong to a family of enzymes that are involved in attaching lipids to proteins, the palmitoyl acyltransferases (PATs). Zdhhc13, in addition to being a PAT, is also a mediator of Mg 2+ transport. Inhibition of palmitoylation by 2-bromopalmitate (2BP) diminished Mg 2+ transport by about 50% [16]. The causal mutation (R425X) in Zdhhc13 mice would predict the synthesis of a truncated protein lacking the zinc-finger DHHC-CRD domain (#426-476) and the active site (C456) for the formation of an S-palmitoyl cysteine intermediate ( Figure S1). Thus, it is unlikely that this truncated protein can perform any palmitoylation function, consistent with the similarity of phenotypes with the gene trap allele. Because our mutant mice had normal serum magnesium and calcium levels and demonstrated no clinical evidence of magnesium deficiency, we propose that the observed phenotypes originated from the loss of the enzymatic function of Zdhhc13 as a PAT. Indeed, we have shown that mutation in the Zdhhc13 affects the protein palmitoylation.
The exact mechanism by which the mutation of Zdhhc13 resulted in such diverse pathologies is not clear. Amyloidosis is a devastating group of disorders in which normally soluble proteins are misfolded and aggregate to form insoluble amyloid fibrils with  a b-sheet structure and presumably trigger an unfolded protein response (UPR) and its downstream pathways, including autophagy and cell death by apoptosis [17][18][19]. There are two major types of systemic amyloidosis. AA-amyloidosis, also called secondary or reactive amyloidosis, is a consequence of prolonged high level expression, mainly in the liver, of the acute-phase protein Serum Amyloid A precursor protein (SAA). AA-amyloidosis is usually associated with chronic inflammatory conditions but it can also be caused by mutations in a constitutively expressed protein, resulting in its greater tendency to aggregate [20,21]. The second type of amyloidosis is called AL-amyloidosis. It is either a primary or a multiple myeloma-associated amyloidosis; its fibrils are derived from fragments of monoclonal immunoglobulin light chain (l and k) condensed into b-pleated sheet structures as a result of incomplete breakdown in the autophagolysosomes [22]. It is rare to find both AA and AL types in the same patient, and such cases only account for ,2-3% of all amyloidosis patients [23].
Our Zdhhc13 mutant mice manifested both AA and AL amyloids ( Figure 5, Figure 6, Figure 7). Liver, spleen, kidneys, skin, adrenal and salivary glands are the most affected organs resulting in hepatosplenomegaly, nephromegaly, sialadenosis (non-inflammatory swelling of the salivary glands) and skin involvement. We cannot attribute this observed amyloidosis to inflammation, as CRP levels were normal (Table S2). Nor can we attribute it to multiple myeloma, as blood and bone marrow contained no excess plasma cells (data not shown). Furthermore, sequences of all amyloid-related genes in the candidate regions were normal except for the single mutation (R425X) in Zdhhc13. The coexistence of AL and AA systemic amyloidosis could be attributed to the fact that the presence of AL type amyloid fibrils acting as an amyloid-enhancing factor (AEF) and enhance the AA amyloid deposition [23][24][25][26]. Alternatively, a deficiency of Zdhhc13 may play a role in the amyloidogenesis. Palmitoylation is known to affect protein stability by influencing a protein's access to an ubiquitinating enzyme [27]. Palmitoylation is also known to protect huntingtin from aggregation [28], and prevent oligomerization of certain proteins [29]. Moreover, defective palmitoylation results in aggregation of amyloid b-sheet, which leads to the formation of fibril [30].
Therefore, we propose that a deficiency of Zdhhc13 PAT activity may have caused amyloidosis in our mice. The lack of Zdhhc13 PAT activity affected palmitoylation of a set of unspecified protein targets and compromised their conformational stability and subcellular localization, eventually causing systemic amyloidosis of both the AA and AL types [26]. Coincidentally, a computer algorithm predicted the presence of palmitoylation site(s) on both SAA and the light chains of IgG [31]. Consistent with this notion was our demonstration of reduced level of palmitoylation of IgG light chain in the mutant mice.
Deficiency of protein palmitoylation in the Zdhhc13 mutant mice could also explain the apparent osteoporosis because palmitoylation regulates osteoblast differentiation through bone morphogenesis protein (BMP)-induced Osterix expression [6]. Deletion of Osterix leads to a loss of mature osteoblasts and a lack of calcified bones (Osteoporosis), other signaling pathways such as NF-kB may be also involved in the development of the severe osteoporosis phenotype observed as early as weaning [32].
The Zdhhc13 mutant mice also showed significant skin pathology with hypotrichosis, alopecia and loose skin with wrinkling and folding. Histopathology revealed epidermal hyperplasia with a thin dermis, inactive hair follicles and amyloid deposition. Expression analysis shows that expression of Zdhhc13 is upregulated at the time of follicle maturation (P8), consistent with a direct role of Zdhhc13 in hair formation. Although the exact molecular mechanisms are not clear, these phenotypes are consistent with defects in the NF-kB signaling pathways [33]. Another possible mechanism responsible for the skin pathology is the BMP-induced MAPK pathway as protein palmitoylation plays an important role in BMP-induced MAPK pathway activation [6]. BMP is involved not only in osteoblast differentiation but also in epidermal proliferation and differentiation, hair follicle cycling and innervations [34]. Since Zdhhc13 normally upregulates both NF-kB and MAPK signaling pathways [35] the Zdhhc13 mutation may significantly affect these pathways, leading to the disease phenotypes.
In summary, we report that deficiency of a single palmitoyl acyltransferase (Zdhhc13) can cause severe systemic phenotypes, including failure to thrive, cachexia, osteoporosis, alopecia, multiorgans/systems dysfunction secondary to systemic amyloidosis and early death. Our results established a direct link between protein palmitoylation and regulation of important diverse physiological functions and indicated that its absence can result in profound disease pathology. This mouse model will be useful for further investigation of the mechanisms by which improper palmitoylation leads to disease processes. The identification of target proteins of ZDHHC13 would be an important first step for understanding the molecular mechanisms underlying human alopecia, osteoporosis and many neurodegenerative diseases caused by protein misfolding and amyloidosis.

Mouse Lines
The first recessive mutant allele was generated by a conventional ENU mutagenesis regimen. [36]. Briefly, multiple doses of N-ethyl-N-nitrosourea (ENU) (100 mg per kg body weight) were injected to mutagenize spermatogonia of C57BL/6J males (G0 generation). Recessive mutations were isolated in the third generation of breeding to females that carried either the balancer chromosome 129S6.Inv(11)8Brd or 129.Rex mutations, both of which had been made congenic on a 129S6/SvEvTac genetic background (N = 10). The mutant mice reported here were identified by their small size and hypotrichosis as early as postnatal day 7 and were designated as skin and coat mutation 4 (skcm 04Jus ). The mutant line was inherited as a recessive trait that segregated independently of the chromosome 11 balancer and the phenotype was completely penetrant in the 129S6/SvEv genetic background, and, later on, in the C3H background. The experimental protocols in this study were reviewed and approved by the Institutional Animal Care and Utilization Committee of Academia Sinica.
The gene trap allele was produced from embryonic stem (ES) cells AC0492 obtained from the Sanger Institute Gene Trap Resource (SIGTR). ES cells were injected into C57BL/6J blastocysts by the Darwin Genetics Core at Baylor College of Medicine. Chimeras were obtained, mated to C57BL/6J mice, and the allele was transmitted through the germline to generate Zdhhc13 SIGTR mice. Subsequent experiments were performed on mice from this mixed 129/B6 genetic background. After initial genotyping of ES cells for the gene trap allele per protocols available from the resource, the phenotype was used to follow transmission of the allele. Zdhhc13 is located at 56 Mb on Chromosome 7, which is only 7 Mb from Oculocutaneous albinism 2 (Oca2; pink-eyed dilution), which is located at 63 Mb. Therefore, mice carrying the gene trap allele were also pink-eyed and had dilute coat colors because of the Oca2 mutation carried in the 129/Ola ES cells used to generate the gene trap.

Blood Chemistry
Blood samples were obtained through an incision of the tail artery or by cardiac puncture at the time of sacrifice and collected in a heparinized tube (MICROTAINER, BD Diagnostics, Franklin Lakes, NJ). Complete hemogram was carried out using Abbott Cell-DYN 3700 Veterinary Haematology Analyzer (Abbott Laboratory, Illinois, USA). Thin blood smears were taken directly from the tail artery, fixed with absolute methanol for 5 minutes and stained by modified Wright's Giemsa stain for Plasma cells (Plasma B cells) identification. Plasma was analyzed using the FUJI DRI-CHEM SYSTEM 3500s (Fuji Photo Film Co. Ltd.) for measurement of aspartate aminotransferase (AST; U/l), alanine aminotransferase (ALT; U/l), creatinine phosphokinase (CPK; U/l), total cholesterol (TCHO; mg/dl), total protein (TP; g/dl), albumin (ALB; g/dl), globulin (GLO; g/dl), total bilirubin (TBIL; mg/d), blood urea nitrogen (BUN; mg/dl), C-reactive protein (CRP; mg/dl), calcium (Ca; g/dl) and magnesium (Mg; g/dl).

Micro-Computed Tomography (Micro-CT) Analysis of Bone
For trabecular bone analysis and 3D images, a micro-CT scanner (Skyscan-1076, Skyscan, Belgium) was operated at 50 kV, 200 uA, 0.4u of rotation step, 0.5 mm Al filter and 9 um/pixel of scan resolution. For bone mineral density (BMD) analysis, it was operated at 50KV, 200 uA, 1u of rotation step, 0.5 mm Al filter and 35 um/pixel of scan resolution. Cross-sections were reconstructed using a cone-beam algorithm (software Cone_rec; Skyscan, Belgium). Files were then imported into CTAn software (Skyscan) for three-dimensional analysis and three-dimensional image generation. BMD for each femur was measured by CTAn, which was calibrated using of phantoms with known BMD (0.25,0.75 g/cm 3 ).

Histopathology
Mice were sacrificed with overdoses of sodium pentobarbital for the histop-athological examinations. After flushing with normal saline, mice were perfused through the heart with 4% paraformaldehyde in 0.1M PBS, pH7.4, the perfusion flow rate, (4 ml/ min) was controlled by an infusion pump (Bio-Rad, Econo Pump). A total of 37 organs and tissues, including heart, lung, liver, kidney, spleen, pancreas, adrenal gland, salivary gland, brain, skin, adipose tissues, skeletal muscles and bone, were removed, embedded in paraffin, cut into 5 mm sections and stained with hematoxylin-eosin (H&E) for general pathological examinations. Other serial sections were also processed for Congo Red staining to detect amyloid; and immunohistochemistry staining for amyloid classification. Bone marrow was aspirated from both femora for bone marrow smears immediately after sacrificing by cervical vertebral dislocation. Smears were fixed with absolute ethanol for 5 minutes and stained by modified Wright's Giemsa stain for Plasma cells identification.

Immunohistochemistry
Immunostaining used the following antibodies: rabbit antihuman l light chains polyclonal antibody, rabbit anti-human k light chains polyclonal antibody, mouse anti-human amyloid A monoclonal antibody (DakoCytomation). Tissue sections were pretreated with concentrated formic acid for 1 min, washed in trisbuffered saline (TBS) for 10 min, then incubated in Rodent Block M (BioCare) or tris-buffered saline Tween (TBST) containing 2% bovine serum albumin and 3% normal goat serum at 37uC for 30 min. Then, tissue sections were incubated with primary antibody, washed in TBST, incubated in 3% H 2 O 2 for 15 min at room temperature and then washed again in TBST followed by incubation with horseradish peroxidase-conjugated secondary antibody (anti-rabbit IgG and goat anti-mouse IgG (Jackson ImmunoResearch. West Grove, USA). Color was developed with 0.1% 3,39-diaminobenzidine.

Mapping of Gene Responsible for Abnormal Phenotypes
For the purpose of rough mapping, the affected skcm 04Jus mice, which were in a mixed 129S6/SvEv and C57BL/6 mixed genetic background at the N = 4 generation on 129S6/SvEvTac (obtained from Baylor College of Medicine) were outcrossed to the C3He/HeJ strain in Academia Sinica to generate N1 offspring, and N1 mice were then intercrossed with generate N1F1 offspring. DNA was collected from 32 affected N1F1 mice. A panel of 295 single nucleotide polymorphism (SNP) markers located on all 19 mouse autosomes and the X-chromosome for mouse strains C3H/HeJ, C57BL/6, DBA/2J or BALB/cByJ was selected from a SNP dataset containing 10,915 SNPs from 48 mouse strains (provided by Tim Wiltshire, Genomics Institute of the Novartis Research Foundation, San Diego, California). SNPs were chosen based on the criterion that the genotype of these strains at the 295 loci was different from that of C57BL/6J. Since point mutations were introduced into C57BL/6J genome by ENU, the recessive mutant phenotype will always associated with a homozygous B6 SNP genotype at the mutant locus. SNP genotyping using genomic DNA isolated from mouse tails (Puregene DNA purification kit, Gentra Systems, Minneapolis, MN, USA) was performed using high-throughput MALDI-TOF mass spectrometry [37,38]. Primers and probes flanking the SNPs were designed in multiplex format using SpectroDESIGNER software (Sequenom, San Diego, CA, USA). PCRs were performed in a volume of 5 ml containing 0.15 U of Taq polymerase (HotStarTaq, Qiagen, Valencia, CA, and USA), 5.0 ng of genomic DNA, 1.0 pmol of each PCR primer and 2.5 nmol of dNTP. Thermocycling conditions were one cycle at 94uC for 15 min, 45 cycles of 94uC for 20 s, 56uC for 30 s, 72uC for 30 s and one final cycle of extension at 72uC for 3 min. Unincorporated dNTPs were dephosphorylated using 0.3 U of Shrimp Alkaline Phosphatase (Hoffman-LaRoche, Basel, Switzerland) followed by primer extension using 9 pmol of each primer extension probe, 4.5 nmole of the appropriate dNTP/ ddNTP combination, and 1.28 U of Thermosequenase (Amersham Pharmacia, Piscataway, NJ, USA). Reactions were cycled at 94uC for 2 min, followed by 55 cycles of 94uC for 5 s, 52uC for 5 s and 72uC for 5 s. Following the addition of a cation exchange resin (SpectroCLEAN, Sequenom) to remove residual salt from the reactions, 15 nl of the purified primer extension reaction was spotted onto a 384-element silicon chip preloaded with 3hydroxypicoloinic acid matrix (SpectroCHIP, Sequenom), using the SpectroPOINT (Sequenom). SpectroCHIPs were analyzed using a Bruker Biflex III MALDI-TOF SpectroREADER mass spectrometer (Sequenom) and spectra processed with Spectro-TYPER (Sequenom).

Identification of the Mutant Gene
All exons, exon-intron junctions and 2.5 kb promoter regions of candidate genes, Saa11, Saa3, Saa4, Saa1, Saa2 and Zdhhc13, were amplified and sequenced. Primers were designed using the Primer3 program http://biotools.umassmed.edu/bioapps/ primer3_www.cgi. The primers used for the detection of an exon 12 mutation in the Zdhhc13 gene were F, 59-CTGGGTTGA-GAGTATTCCACA-39 and R, 59-GAGATTAGCCACA-GAGCTTCG-39. PCR reactions were performed in a final volume of 25 ml, containing 50 pmol of each primer (0.5 ul), 106 Taq Buffer (10 mM Tris-HCl (pH 8.3), 50 mM KCl) with 1.5 mM MgCl 2 (2.5 ul), 2.5 mM dNTPs (2.5 ul) and Taq DNA polymerase (5 U/ul) MDBio, Inc. (0.25 ul). Amplification conditions were an initial denaturation of 4 min. at 94uC, followed by 20 cycles of touchdown PCR in 30 s at 94uC, 30 s at 65uC (decrease 0.5uC per cycle), 40 s at 72uC; and a final 20 cycles in 30 s at 94uC, 30 s at 55uC, followed by 40 s at 72uC and then a final extension at 72uC for 5 min. All amplified PCR fragments were digested with shrimp alkaline phosphatase and ExoI to remove unincorporated primers and sequenced using the BigDye Terminator Cycle Sequencing Kit v1.1/3.1 (Applied Biosystems, Foster City, CA, USA) following the manufacturer's instructions. Sequencing products were separated on either ABI PRISM 3100 Genetic Analyzer or ABI PRISM 3700 DNA Analyzer (Applied Biosystems). Raw sequencing data were analyzed with the DNA Sequencing Analysis Software v3.7 (Applied Biosystems).

Expression Analysis
To examine the differences in the tissue expression of Zdhhc13, total RNA samples were extracted from liver and kidney of three 6-month old mutants as well as three aged-matched wild type C3He/HeJ mice using Trizol following the manufacturer's protocol. First-strand cDNA was synthesized using 1 ml oligo-dT 15 primer and 1 ml SuperScript III RT (200 U/1 ml) in 20 ml volume of 2 mg of total RNA, 1 ml of 10 mM dNTPs, 1 ml of reaction buffer (10 mM Tris-HCl pH 8.3, 2.5 mM KCl, 0.6 mM MgCl2), ml of RNase inhibitor40 U/ml, and 1 ml of 0.1 M DTT. Real-time quantitative RT-PCR analysis used the ABI PRISM 7700 Sequence Detection System (Applied Biosystems). RT-PCR amplification of Zdhhc13 was carried out using the following primer set: 59-GACTGGACGCTGCATAGGTT, forward strand in Zdhhc13 exon13, and 59-TGGCACAATGATTTGAC-CAG, reverse strand in Zdhhc13 exon 15. The primers were designed using Primer Express (Applied Biosystems).
The cDNA corresponding to 75 ng of reversed transcribed total RNA was amplified in a final volume of 20ml using Power SYBER green PCR Master mix in 20ml total reaction volume in duplicate assays for Zdhhc13 and endogenous B-actin as an internal control.
An analysis of the results was based on the Ct calculation, where Ct represents the cycle number at which fluorescence of the PCR samples crossed a given threshold. The expression level of b-actin was taken as the first ''calibrator'' to normalize the total Zdhhc13 mRNA in each tissue (DCt). Expression of, Zdhhc13 in each of the control mouse tissue was then taken as the second ''calibrator'' to normalize the expression of Zdhhc13 in the affected tissue accordingly (DDCt). Final results were given as the relative amounts of Zdhhc13 mRNA in the affected mouse tissues as compared to the control (2 DDCt ).
Northern analysis was carried out as previously described by Lorenzetti et al. [39], using 10 ug of total RNA isolated using RNA STAT-60 reagent (TEL-TEST, Inc., Friendswood, TX) according to the manufacturer's protocol, and transferred to a nylon membrane. The blot was hybridized using Ultrahyb (Ambion) with a probe for Zdhhc13 is the N-terminal 571 base pairs, which Supporting Information Figure S1 Predicted secondary structure of Zdhhc13 (Huntingtininteracting protein 14-related protein, HIP14-related protein a palmitoyltransferase ZDHHC13). ZDHHC is 660 AA with molecular weight of 70890 Da, it has 6 transmembrane domains and 6 ANK repeats (ANK 1-6), Zn_Fing (DHHC-TYPE) length is 51aa (from 426-476aa) and the Bompbias (Phe-rich) is 64aa (from 328-391aa). The active site (S-palmitoyl cysteine intermediate) is located at residue 456. Note that nonsense mutation arg-425-stop codon is located before both Zn_Fing and the active site of the protein.