Genomic Expression Analyses Reveal Lysosomal, Innate Immunity Proteins, as Disease Correlates in Murine Models of a Lysosomal Storage Disorder

Niemann-Pick Type C (NPC) disease is a rare, genetic, lysosomal disorder with progressive neurodegeneration. Poor understanding of the pathophysiology and a lack of blood-based diagnostic markers are major hurdles in the treatment and management of NPC and several additional, neurological lysosomal disorders. To identify disease severity correlates, we undertook whole genome expression profiling of sentinel organs, brain, liver, and spleen of Balb/c Npc1−/− mice relative to Npc1+/− at an asymptomatic stage, as well as early- and late-symptomatic stages. Unexpectedly, we found prominent up regulation of innate immunity genes with age-dependent change in their expression, in all three organs. We shortlisted a set of 12 secretory genes whose expression steadily increased with age in both brain and liver, as potential plasma correlates of neurological and/or liver disease. Ten were innate immune genes with eight ascribed to lysosomes. Several are known to be elevated in diseased organs of murine models of other lysosomal diseases including Gaucher’s disease, Sandhoff disease and MPSIIIB. We validated the top candidate lysozyme, in the plasma of Npc1−/− as well as Balb/c Npc1nmf164 mice (bearing a point mutation closer to human disease mutants) and show its reduction in response to an emerging therapeutic. We further established elevation of innate immunity in Npc1−/− mice through multiple functional assays including inhibition of bacterial infection as well as cellular analysis and immunohistochemistry. These data revealed neutrophil elevation in the Npc1 −/− spleen and liver (where large foci were detected proximal to damaged tissue). Together our results yield a set of lysosomal, secretory innate immunity genes that have potential to be developed as pan or specific plasma markers for neurological diseases associated with lysosomal storage and where diagnosis is a major problem. Further, the accumulation of neutrophils in diseased organs (hitherto not associated with NPC) suggests their role in pathophysiology and disease exacerbation.


Introduction
Niemann-Pick Type C (NPC) is a neurodegenerative, lysosomal disorder caused by defects in function of either genes Npc1 or Npc2, although in 95% of patients disease is caused by defect in Npc1 [1]. There is resulting defect in cellular transport of lipids, characterized by accumulation of both unesterified cholesterol and sphingolipids in late endosomal/lysosomal compartments. Inflammatory changes have been reported in the liver, spleen and brain of NPC animals [2,3,4,5] and anti-inflammatory treatments have been shown to reduce disease burden in mice [4,6]. Prior work suggests that antisense mediated knock down of Npc1 in C57BL/6 mice results in tumor necrosis factor a (TNF-a)-dependent accumulation of inflammatory cells in liver [2,7]. Foamy macrophage accumulation in liver [2,3,8], activation of microglia in brain [9] and impaired development and reduced natural killer T (NKT) cells in spleen and thymus have been reported [10,11] in NPC null mice. Changes in inflammatory cells and protein markers [4,7,12] appear consistent with organ specific (largely the brain) analysis of transcripts [5,13,14]. Expression arrays have also been utilized to investigate transcriptional changes in cell culture [15,16]. However comprehensive, unbiased, genome wide analyses of changes in gene expression in a leading organ of interest, the brain, across the life span, especially as animals transition from a phenotypically asymptomatic state to manifesting major disease symptoms, is not yet available. Further whether age-dependent gene expression in the brain is linked if at all, to that in the liver and/or spleen two organs that manifest early disease symptoms, is also not known. Genes expressed in an age-dependent manner in both brain and liver (the source of plasma proteins) would facilitate identification of blood-based biomarkers that reflect cerebral disease.
Consistent with increase in their inflammatory mechanisms, NPC disease cells and/or animals have been shown to be refractory to infection by HIV-1 and Brucella abortus [17,18]. However resistance of NPC cells and animals to infection may also occur because cholesterol and endosomal trafficking are known to play critical roles in vacuolar infection of virus, bacteria and parasites in a variety of different hosts [19,20,21,22]. More recently, NPC1 has been shown to act as an invasion receptor for Ebola and Marburg viruses [23,24], suggesting a direct role for NPC1, possibly independent of cholesterol trafficking in the infection of filoviridae. However, whether cellular mechanisms controlling microbial proliferation in organ systems are altered, is not known.
Salmonella enterica serovar Typhimurium (S. typhimurium), a Gram-negative, rod shaped, facultative intracellular bacterial pathogen, is a major cause of food-borne enterocolitis in humans as well as a typhoid-like disease in mice [25,26]. Due to the ease with which it can be genetically manipulated, quantitatively analyzed both in vitro and in mouse models of infection, Salmonella is often used as a model system to investigate cellular and organismal processes of mammalian hosts. Replication in the liver and spleen is essential for dissemination of Salmonella [26,27]. These organs also manifest the earliest pathologies of NPC. However, whether NPC1 defects influence Salmonella virulence, and/or proliferation in vivo, is not known. In both liver and spleen, if loss of the Npc1 gene influences expression of genes important for host response to Salmonella infection, the underlying basis can be rapidly validated with well-developed cellular assays and other functional read outs.
We have performed non-biased, genome wide expression profiling analyses to discover increase in a restricted subset of innate immunity transcripts as a major transcriptional change in the brain, across the life span of the Npc1 2/2 mouse. Expression profiling of liver and spleen also established up-regulation of innate immunity transcripts. By comparative analyses of up regulated brain and liver genes, we identify 12 secretory proteins that have potential to be developed as plasma correlates measuring transition to NPC disease in the brain. As a proof of concept, we validated the top hit lysozyme in plasma. Further we confirmed functional elevation of innate immunity mechanisms in both liver and spleen by following resistance to infection by S. typhimurium as a model organism. We also report for the first time, neutrophil elevation in liver and spleen of Npc1 2/2 mice that may play a role in NPC pathophysiology and disease exacerbation.

Results
Genome-wide Gene-expression Analyses in Brain, Liver and Spleen of Npc1 2/2 Mice from Weaning through Advanced Neurodegeneration Progressive neurological dysfunction is a prominent feature of NPC disease, and hence understanding correlates in the brain is of critical importance to understanding disease progression. To comprehensively cover the life span, we examined transcripts in brain from animals from ,20 days to 80 days ( Figure S1A) reflecting the period from weaning, when the animals are completely asymptomatic to advanced neurodegeneration and significant weight loss (,30% reduction is observed by 60-80 days) characteristic of this model [28,29,30]. Across this range, six time points (days 20-25, 37-40, 54-55, 59-62, 67-71, and 81-84) were utilized to closely map the life span of Npc1 2/2 mice. For each point, transcripts of brains from two Npc1 2/2 mice were compared to age matched, Npc1 +/2 mice. Npc1 +/+ animals were also included for days 20-25 and 59-62 and 67-71 (as outlined in Figure S1A) to enable comparative analysis across all three genotypes. Due to technical difficulties, RNA yield from one animal (Npc1 2/2 , 71 days) was low and thus at this time point transcript data from only one Npc1 2/2 animal was included (whose exclusion had no adverse effect on inferences drawn from the global data set: data not shown). Our data analysis selected for genes whose expression was significantly altered between Npc1 2/2 mice relative to Npc1 +/2 and/or Npc1 +/+ mice at all time points (see Materials and Methods, Table S1). Because one gene may be represented multiple times on the array chip, we removed the replicates. As shown in Figure 1A, and Table S1, 115 genes were up regulated (red), and 71 were down regulated (blue), suggesting that less than one percent of the total number of genes were consistently changed in the brain throughout the animal's life span.
Since enlargement of the liver and spleen are early indicators of NPC, we were also interested in examining corresponding changes in these organs. For both liver and spleen, we examined three of the six time points utilized for brain analysis. Thus pairs of Npc1 2/ 2 mice relative to age-matched Npc1 +/2 at 20-25 days, 54-55 days and 67-71 days were analyzed for each organ ( Figure S1B). This age range was sufficient to cover animals immediately post weaning and the transition from asymptomatic (20-25 days) into symptomatic animals (that arises between 45 to 60 days) in this model [28,30,31]. As shown in Figure 1B and Table S2, in the liver, 1421 genes showed consistent change through these age groups in Npc1 2/2 mice. 817 were reliably up regulated and 604 were down regulated. In contrast in the spleen, 218 and 17 transcripts were respectively up-and down-regulated in Npc1 2/2 mice compared to Npc1 +/2 ( Figure 1C and Table S3). Hence, compared to the brain and spleen, the liver showed the greatest number of changes manifest throughout the life span, which is consistent with significant liver dysfunction associated with this disease.
Our data are consistent with prior studies in the literature examining transcriptional changes in the brain at individual time points or multiple time points over a short age range [5,13,14,28,36]. Thus, genes like Lyz1/2, Cd84, Cd68, C1qa, C1qb, Ifit3, Ptprc, H2-d1, H2-k1 etc have been previously shown to be increased early in mouse brain [13]. Additional innate immunity genes previously described in the brain of NPC mice are Mpeg1, Gpnmb, Ctss, Ctsd, Ctsz, Grn, Clec7a, Itgax, Gp49a, Hexb, Lgls3bp, Tyrobp etc [5,13,14]. It should be noted that at a given time point, a relatively large number of genes are altered as described earlier [5,13]. However our data show that smaller subsets of these genes are consistently up regulated across the animal life span.
In the liver, both the number of genes and fold change in gene expression were greater compared to the brain. Changes in gene expression seen in the top 20 up regulated genes were relatively large and ranged from ,80 to 15 fold (Table 2). InnateDB identified 123 genes to be innate immunity genes of which 101 were up-and 22 were down-regulated (Table S2, shown in bold). In the top 20 most up regulated genes, eleven are reported to have roles in innate immunity and/or antimicrobial activity against viruses, bacteria and/or fungi ( Gene expression analysis in the spleen also suggested up regulation of innate immunity genes. InnateDB identified 35 genes of which 32 were up-and 3 were down-regulated (Table S3, shown in bold). Of the top 20 up regulated genes, 6 were innate immunity genes, five (Clec7a, Atf3, Mmp12, Msr1 and Elane) of which were found in InnateDB (Table 3, marked in bold). The sixth Gpnmb [41], was also up regulated in the brain and liver. Additional, prominent up regulated innate immunity genes were annexins (Anxa1, Anxa4), Ctsb, Ctsd, Lgals1, Lgals3, that were over expressed in brain and liver and Mmp9 and Camp, also over expressed in liver (Table S2, marked in bold).

Prioritization of Plasma Correlates Predictive of Cerebral Disease
There is as yet, no blood-based biomarker for NPC and this greatly delays diagnosis of the disease, which can take on average of five years [42,43] Recent studies suggest that elevation of oxysterols in plasma could well be developed into the first bloodbased diagnostic for NPC [44]. However, despite their maximal elevation in Npc1 2/2 , oxysterols also show slight increase in Npc1 +/ 2 animals. Further, oxysterols may not respond to substrate reduction therapies such as miglustat (Zavesca) that reduces levels of sphingolipids rather than cholesterol [45], suggesting need for multiple biomarkers.
To develop a prioritized set of plasma proteins that are linked to correlates of disease in the brain, we identified genes of soluble secretory proteins that are up regulated in the NPC brain as well as the liver (the major source of plasma proteins) at all time points. This led to the identification of 18 genes namely Lyz1 (Lysozyme1), Lyz2  (Table S5). Of these, 12 showed progressive, age dependent change in both brain and liver, that is desired in a disease marker (Figure 3 and Figure S2). In the Figure 1. Genome-wide gene-expression profiling on brain, liver and spleen of Npc1 2/2 mice. (A) Brain. Pie chart displays 0.46% up regulated and 0.28% down regulated transcripts relative to the total number of transcripts (24615) expressed in the brain of Npc1 2/2 mice across the life span. Total RNA from brain of 27 mice (11 Npc1 2/2 and 16 age matched controls) age ranging from 20-84 days (6 age groups, see Figure S1A for details) were isolated and gene expression was analyzed using affymetrix microarray chips (see Materials and Methods). The expression level of 186 genes (115 up and 71 down) in the Npc1 2/2 mice changed by 1.5 fold or higher (p,0.05). (B) Liver. Pie chart displays 4.44% up regulated and 3.28% down regulated transcripts relative to the total number of transcripts (18377) expressed in the liver of Npc1 2/2 mice across the life span. Total RNA from the liver of 12 mice (6 Npc1 2/2 and 6 age matched Npc1 +/2 ) age ranging 20-71 days (3 age groups, see Figure S1B for details) were isolated and transcript expression was analyzed as described for the brain. The expression of 1421 genes (817 up and 604 down) in the liver of Npc1 2/2 mice changed by 1.5 fold or higher (p,0.05). (C) Spleen. Pie chart displays 1.3% up regulated and 0.11% down regulated transcripts relative to the total number of transcripts (15348) expressed in the spleen of Npc1 2/2 mice across the life span. Experimental set up and analysis criteria were identical to that described for the liver. The expression of 218 genes (201 up and 17 down) in the spleen of Npc1 2/2 mice changed by 1.5 fold or higher (p,0.05). doi:10.1371/journal.pone.0048273.g001  also Table S4). (C) Bar diagram shows the top 10 biofunctions enriched in the spleen of Npc1 2/2 compared to Npc1 +/2 . A total of 58 genes (49 up and 9 down regulated) associated with the immune response were enriched in the spleen of Npc1 2/2 across all time points (see Table S4 Usp18: similar to ubiquitin specific protease UBP43 3.91 Trem2: triggering receptor expressed on myeloid cells 2 3.88 Genes marked in bold are related to innate immunity and the genes marked in bold and also underlined are innate immunity genes catalogued by InnateDB. doi:10.1371/journal.pone.0048273.t001 order of their elevation in the brain, these are Lyz1, Lyz2, C1qb, Lgals3, C1qa, Grn, Ctss, Ctsd, Timp2, Man2b1, Hexb and Ctsb (Table 4 and Figure 3). Remarkably, other than Man2b1 and Hexb, the remaining ten are innate immune genes of which eight are lysosomal. All 12 may be putative, plasma predictors of the transition to cerebral disease.
Elevated Lysozyme Activity in the Plasma of Npc1 2/2 Mice As validation, we selected our top hit lysozyme, whose transcripts showed highest elevation in the brain, and also linear increase in the liver. Our interest was to determine a measure of lysozyme levels in the plasma. To facilitate rapid quantification, we pursued lysozyme's well defined muramidase activity assay in plasma. As shown in Figure 4, levels of active lysozyme were indeed elevated in Npc1 2/2 mice 3-4 weeks old (representing 21-28 days at weaning and soon after) relative to age matched, Npc1 +/ + and Npc1 +/2 . Further, plasma from Npc1 2/2 mice showed progressively increased lysozyme activity reaching a peak at 7-8 week of age. At 9-10 weeks (most animals die by 11 weeks in Npc1 nih model), lysozyme activity levels plateau. This is in contrast to Lyz1 and Lyz2 transcript levels which increase steadily from 7-8 weeks to 9-10 weeks (See Figure 3). Lysozyme is known to be inactivated at high concentrations and thus it is possible that activity levels do not accurately measure total protein at advanced stages. Nonetheless, it increased up to 8 weeks and in particular during transition from phenotypically asymptomatic (5-6 weeks) to symptomatic state (,7-8 weeks) [28,31]. The data shown in Figure 4 are derived from both male and female animals, suggesting age dependent elevation of lysozyme was independent of gender. The assay could be carried out using 2 to 20 ul of plasma, suggesting it is sensitive and has a large dynamic range.
Elevation of Lysozyme in BALB/c Npc1 nmf164 Mice and its Reduction in Response to Treatment with Cyclodextrin, an Emerging Therapeutic Although the Npc1 null (Npc1 nih ) mouse captures the progression of human disease, most patients show point mutations rather than a truncation in the gene. We therefore examined the BALB/c Npc1 nmf164 (Npc1 nmf ) mouse with milder disease progression due to partial loss of NPC1 function as a result of a single point mutation (D1005G) in the cysteine rich domain of the protein, which is one of the most common regions for human mutations. Previous studies suggest that Npc1 nmf in the C57BL/6J background have a life span of ,112 days and develop progressive disease [30]. They show delayed weight loss starting from 9-10 weeks and the rate was slower than the Npc1 nih mice. Histological analyses of brain, liver and spleen showed abnormal cholesterol accumulation, and purkinje cell loss at a slower rate than the Npc1 nih [30]. We found that BALB/c Npc1 nmf have a similar life span (,120-125 days) and disease progression to that of C57BL/6J Npc1 nmf164 mice. Typically they exhibited weight loss from 12 weeks and by the end of 16 weeks ,15-20% weight loss was observed ( Figure 5A).
As shown in Figure 5B, levels of active lysozyme were indeed elevated in Npc1 nmf mice 3-4 weeks old (representing 21-28 days, the time of weaning and soon after) relative to age matched, controls. Further, plasma from Npc1 nmf mice also showed progressively increased lysozyme activity reaching a peak at 10-11 weeks of age. At 14-15 weeks lysozyme levels plateau (and it is possible that here again, lysozyme is inactivated at high concentration). Most animals die by 17-18 weeks in this model. The data shown in Figure 5B is derived from both male and female animals, suggesting that elevation in lysozyme may be a useful correlate for disease, especially at the early phases, when diagnosis is difficult but needed. Genes marked in bold are related to innate immunity and the genes marked in bold and also underlined are innate immunity genes catalogued by InnateDB. doi:10.1371/journal.pone.0048273.t003 With the emergence of new therapeutics for NPC, there is urgent need for correlates whose levels mirror improvement of disease course as a consequence of treatment. Cyclodextrin has emerged as the most effective compound at retarding NPC disease in mice [46]. Previous studies suggest that weekly injections of HPbCD (2-hydroxypropyl-beta-cyclodextrin) to Npc1 nih (a BALB/ c strain) ameliorates the disease and extend the survival [47,48]. Similarly, weekly injections of HPbCD to Npc1 pf/pf mice (a knockin BALB/c strain carrying point mutations resulting in failure to cholesterol binding and manifestation of NPC disease) also show improvement in disease status [49]. We therefore treated Npc1 2/2 mice with HPbCD or vehicle control (0.2% DMSO in 0.9% saline) with once a week drug injections starting at age 21-27 days. At 50-55 days, untreated Npc1 2/2 mice had ,1.4-1.8 fold higher plasma lysozyme activity compared to Npc1 +/+ or Npc1 +/2 (age 42-49 days). The plasma lysozyme activity of the vehicle treated Npc1 2/2 mice remained elevated (comparable to untreated Npc1 2/2 ). However, it was significantly reduced in Npc1 2/2 mice treated with HPbCD ( Figure 5C). Thus, lysozyme may be an early disease correlate that measures responsiveness to a drug during the asymptomatic stage.
Functional Validation of Elevated Innate Immunity Genes in Liver and Spleen of Npc1 2/2 Mice Microbial systems provide rapid mechanisms of functional validation of innate immunity and there is prior evidence that defect in NPC1 results in attenuated intracellular infection by HIV-1 and Brucella abortus [17,18]. We therefore infected mice with the Gram-negative bacterium S. typhimurium which can be used as model organism to understand the cellular response underlying innate immunity. We selected mice of age at 6-8 weeks, because this was approximately in the middle of the age range of animals examined in our microarray studies. Since we wanted to directly assess bacterial proliferation in the spleen and liver (and bypass the gut) the animals were infected through intraperitoneal (i.p) route. The bacterial load in spleen and liver was determined at 48 hours post infection (hpi) by measuring colony forming units. As shown in Figure 6 A-B, for both liver and spleen, we found comparable Figure 3. Age dependent over expression of 12 potential secretory genes in brain and liver of Npc1 2/2 mice. The raw signal intensity of all 12 genes obtained after Dchip analysis of brain and liver at three distinct time points (1 corresponds to 20-25 days, 2 corresponds to 54-55 days and 3 corresponds to 67-71 days for liver and 81-84 days for brain) were taken and the mean was calculated. Mean signal intensity obtained from 2 Npc1 +/2 mice was subtracted from the mean values of 2 Npc1 2/2 mice corresponding to same age group. The process was undertaken for each gene at all three time points for both brain and liver. The difference obtained for each time point is plotted. doi:10.1371/journal.pone.0048273.g003 bacterial loads in Npc1 +/+ and Npc1 +/2 mice. However, there was ,8-10 fold reduction in bacterial load in the organs of Npc1 2/2 mice.
Since the spleen is readily amenable to comprehensive cellular analysis of innate immunity, we examined the numbers of CD335 + natural killer (NK) cell, CD11c + dendritic cells (DC), CD11b + F4/ 80 + monocytes and macrophages (Mo/MO), and CD11b + Gr-1 hi neutrophils in splenic single cell suspensions of Npc1 2/2 and Npc1 +/2 animals (Figures 7A and Figure S3). Again, we selected mice of age at 6-8 weeks, because the reason described above. Flow cytometric analysis showed no effect on counts of NK cell or dendritic cells. Further, while the total number of CD11b + F4/80 +    Figure S3). To test whether increased levels of neutrophils seen in Figures 7A were functional in Npc1 2/2 spleens, we undertook cellular analyses of splenic cells after infection with S. typhimurium infection. As shown in Figure 7B, the levels of NK cells were unchanged, CD11b lo F4/80 hi Mo/MO decreased and CD11c + dendritic cells increased to the some extent in Npc1 2/2 versus Npc1 +/2 mice (Figures 7B and Figure S3). Importantly, CD11b + Gr-1 hi neutrophils were greatly increased in Npc1 2/2 compared to Npc1 +/2 , ,167610 5 versus ,74610 5 , respectively p,0.000001. The reduced bacterial proliferation seen in Npc1 2/2 spleen is well explained by the fact that as much as ,14% cells were neutrophils compared to only ,7% in Npc1 +/2 Together these data suggest an increased innate immunity function associated with neutrophils in the Npc1 2/2 spleen.
The increase in neutrophils is consistent with the innate immune cell footprint observed in microarray. As an additional follow up, we functionally validated neutrophils accumulation by immunohistochemistry (IHC), using spleen from Npc1 2/2 and Npc1 +/2 littermates aged 48-52 days, which is an intermediate time point in the life span. Neutrophils (Gr-1 + cells stained in brown) were primarily observed in the marginal zone and in the red pulp of the spleen in both Npc1 +/2 and Npc1 2/2 mice ( Figure 8A). However, a massive accumulation of neutrophils was seen in the red pulp of Npc1 2/2 mouse ( Figure 8A, panel A3-4) compared to Npc1 +/2 mouse (Fig. 8A, panel A1-2). Since S. typhimurium showed attenuated growth in liver, we examined whether neutrophils were also elevated in the liver. Prior data in the literature have suggested accumulation of foamy macrophages in liver but neutrophils have not been investigated [2,3,8]. As shown in Figure 8B, (panel B3-4), giant foci of neutrophils (Gr-1 + cells stained in brown) were seen in the liver of Npc1 2/2 mouse, compared to the liver of Npc1 +/2 mouse ( Figure 8B, panel B1-2). Notably, damage to liver tissue in the region of neutrophil accumulation was seen in Npc1 2/2 mice. In summary the data support that neutrophils infiltrate spleen and liver of Npc1 2/2 mice.
Imunohistochemical staining failed to show accumulation of neutrophils in the brain (Figure 8 C). Thus, anti microbial lysosomal secretory protein transcripts (such as lysozymes, cathepsins etc) elevated in the brain come from other cellular sources. Although we did not determine the exact source of over expression of lysozyme and other markers, a likely source in brain may be microglia and/or astrocytes that have been shown to be activated in brain [9]. Since lysozyme and other secretory lysosomal proteins are expressed in variety of cells, additional sources cannot be ruled out. Lysozyme and other lysosomal proteins particularly in plasma could be derived from neutrophils which are known to store high proportion of these proteins in the secretory granules which are specialized lysosomes [50,51]. Elevation of neutrophils in the liver and spleen of NPC mice strengthens this hypothesis. Autopsies are rarely undertaken in Elevation of Innate Immunity in NPC Disease PLOS ONE | www.plosone.org humans to analyze diseased organs (since the cause of death is known to be due to NPC) and thus the involvement of neutrophils in spleen and liver in human disease have not been addressed. However, standard blood work up is carried out in patients and does not reveal notable cellular hematological abnormalities. Our analysis of mouse blood likewise revealed that cellular parameters in both Npc1 2/2 and Npc1 +/2 mice remained in the normal range (Table 5). Notably, there was no significant change in either circulating neutrophils or macrophages in the blood (Table 5).

Discussion
The examination of transcriptional changes seen from freshly weaned animals to late in neurodegeneration, in three major organ systems, enabled prediction of innate immunity trends that could not be obtained from single or a few time points in each organ. Our data reveal that in the brain, a restricted set of innate immune processes are activated early in this organ, exacerbated with age and are the dominant conserved response through the animal's life span. Prior analysis of individual time points reveal increase in innate immune transcripts in the brain, but a relatively large number of genes are changed at any given time point, which obscured discernment of conserved patterns detectable at all stages. Specifically, we see age-dependent elevation of lysosomal proteins in the brain, suggesting elevation of these proteins, possibly in a systemic way in many different cell types. The most likely reason is that NPC1 is a lysosomal protein and thus its systemic loss induces a compensatory response in other lysosomal components in all cells. Consistently, over expression of Cathepsin D (CTSD) has been reported in the brain of murine models of several other lysosomal diseases such as Gaucher's disease, Sandhoff disease, GM1 gangliosidoses, Neimann-Pick A [52]. Elevated Ctsb transcripts have also been observed in the brain of Sandhoff and Tay-Sachs patients [53]. In addition to innate immune markers, we also see elevation of transcripts of alpha-Nacetylglucosaminidase (Naglu) and HexosaminidaseB (Hexb), genes linked to lysosomal diseases MPS IIIB and Sandhoff disease respectively.
Our data also reveal that over expression of lysosomal, innate immune proteins in the brain is conserved in liver and spleen of NPC. Conservation in secretory, soluble, lysosomal proteins shared between brain and liver enabled prioritization of candidate proteins that correlate to cerebral disease and are likely to be detected in plasma. Our work here validated the top hit lysozyme. Recent studies [54] suggest that LGALS3 and CTSD may be suitable disease markers in patient plasma. These markers were selected on the basis of transcriptional expression in the Npc1 2/2 liver alone in absence of data from brain. This study originally prioritized Lgals3 (highly up regulated) Plau (moderately up regulated) and Ctsd (mildly up regulated). However, only LGALS3 and CTSD were validated in patient plasma. Plau is absent in our list, however both Lgals3 and Ctsd are included (Table 4). Based on our data of transcript elevation in the brain, Ctsd is likely a better index of neurological disease, because it is moderately up regulated in both the brain and the liver. In contrast Lgals3 may be a preferred marker for liver disease since we find that it is not substantially up regulated in the brain. Cluzeau et al., 2012 [54] demonstrated that Lgals3 and Ctsd transcripts reduced in response to HbCD in Npc1 2/2 mice. We show that lysozyme levels decrease in plasma in response to HbCD in mice carrying a point mutation in NPC1. Together, these data strongly validate our predictions of lysosomal, secretory innate immune proteins alone or in combination, may provide useful surrogate disease markers for NPC in plasma. As indicated earlier, many are also up regulated in other lysosomal disorders, suggesting they may also developed as pan or specific plasma markers for neurological diseases associated with lysosomal storage and where diagnosis is a major problem.
To further validate our gene expression data we compared them to prior gene expression studies undertaken in NPC whole animals and/or cultured cells [5,13,14,15,28,36]. Since NPC is a lipid storage disease, we examined whether there were changes in genes related to metabolism of lipids and fatty acids. Indeed, 180 genes and 117 genes were respectively linked to lipid and fatty acid metabolism (Table S4). This is consistent with prior analysis of single time points analyzed from Npc1 2/2 mice [5,13,14]. In contrast, we found no major changes in Liver X receptor (LXR) pathways which regulate levels of cellular cholesterol [36] but consistent with prior reports that there is no significant activation of LXR genes [28,36,46] in NPC organs. As exception, Abcg1, Lpl and Pltp were slightly elevated (Table S2), but this was also noted by Cluzeau et al., 2012 [54]. Prior gene expression analyses in the brain by qPCR, revealed over expression of genes involved in extracellular cholesterol trafficking (Apod, Apoe), intracellular cholesterol trafficking (Lipa, Npc2), sterol synthesis and metabolism (Cyp7b1, Cyp11a1) and cell abundance (Gfap, Pcp4) [28]. Our data suggest that other than Npc2 and Gfap, none were consistently up regulated across the animal life span in the brain. However inflammatory genes such as Cd68, Itgax, Itgb2, C3ar1, Cd44, Cyba, Fcgr2b, Grn, Ptprc etc were consistently up regulated in NPC brain (Table S4). Genes related to calcium regulation (Camk1, Camta2) and oxidative stress (Cyba, Cybb, Jund, Ncf2, Ncf4) reported in cell culture studies [15] were also consistently up regulated in the liver of NPC mice (Table S2). Cluzeau et al., 2012 [54] reported age dependent expression of 18 genes in the liver. Our study confirms 14 out of 18 genes showing good correspondence between the two data sets in the liver. These genes correspond to pathways of lipid homeostasis (Abcg1, Hexa, Lpl), cell adhesion and extracellular remodeling (Itgax, Itgb2, Mmp12), immune response and inflammation (Ctss, Gpnmb, Lyz2), developmental signaling (Rragd), oxidative stress (Cyba, Cybb), synaptic plasticity (Syngr1) ( Table S2).
Lysozyme is a small, stable protein present in blood as well as additional secretions like saliva and thus particularly suited to being developed as a simple disease test. Elevation of lysozyme in both Npc1 nih and Npc1 nmf mice strongly suggest that secretory lysosomal protein markers may be associated with both severe and milder disease progression as observed in patients. Additionally, plasma lysozyme levels provide a simple test to follow the effectiveness of a drug in mouse models of NPC. Curiously, although transcript levels of lysozyme continue to increase with age in both brain and liver, the enzymatic activity of lysozyme plateaus at later stages in both the Npc1 nih and Npc1 nmf models. One possibility is that as the disease becomes severe, lysozyme protein denatures and loses its activity due to prolonged oxidative stress [5], but this will require additional study. Nonetheless, at a minimum, lysozyme activity provides a useful marker in preclinical development of new therapeutics.
Our study also provide insights into activation of innate immune functions as well as comprehensive analysis of innate immune cells in the spleen of Npc1 2/2 mice, and thus established for the first time, that defect in NPC1 leads to increased infiltration of neutrophils in the spleen and liver. Indeed, up regulation of genes coding for neutrophils-specific proteins such as NCF4 (neutrophil cytoplasmic factor 4, increased ,7 fold) in the liver along with its interacting proteins NCF2 (neutrophil cytoplasmic factor 2) and CYBA (cytochrome b-245, alpha polypeptide, also known as p22phox) predicted infiltration of neutrophils to liver. Neutrophils or polymorphonuclear lymphocytes (PMNLs) are essential innate immune cells, and the host's first line of defense against various bacterial and fungal infections. They are laden with various cytotoxic granules enriched with different powerful antimicrobial molecules such as cationic peptides, proteases, lactoferrin, myeloperoxidase etc [55,56]. They undergo respiratory burst and produce reactive oxygen intermediates to target microbial pathogens [55,57]. In addition to microbial killing, granule components also mediate cell to cell interaction, adhesion and extravasation. Elevated neutrophils in spleen, liver of Npc1 2/2 mice could be attributed due to increased chemoattraction and extravasation, without steady state elevation in blood. The over expression of matrix metalloproteases, galectins, integrins, phygocyte oxidases, adhesins etc in spleen and/or liver of Npc1 2/2 mice supports enhanced neutrophils migration to these organs.
Although, we did not carry out cellular analysis of liver, in addition to neutrophils, large 'foamy' macrophages were readily detected in sections through NPC liver (data not shown) and undoubtedly contribute to an inflammatory response, as has been previously reported [2,8]. Indeed recent studies suggest that removal of macrophages by Ccl3 deletion aggravates the NPC disease [14] suggesting macrophages may be protective in NPC. It should be noted that although neutrophils are required to resolve inflammation, their sustained activation, degranulation and release of cytotoxic molecules leads to tissue injury [58]. Indeed, Flow cytometric analysis of innate immune cells in spleen of Npc1 +/2 and Npc1 2/2 mice. Splenocytes from un-infected Npc1 +/2 and Npc1 2/2 female littermates (age 6-8 weeks) mice were isolated and stained with anti-CD335 for NK cells, anti-CD11c for dendritic cells (DC), anti-F4/80 and CD11b for monocytes and macrophages (Mo/MO), anti-Gr-1 and CD11b for neutrophils. The data represent the mean from two independent experiments with a total of 6 mice (3 each experiment). Error bars show the mean6SD. Gating parameters are indicated in Figure S3. (B) Flow cytometric analysis of innate immune cells in the spleen of S. typhimurium infected Npc1 +/2 and Npc1 2/2 mice. Mice at 6-8 weeks were infected with S. typhimurium intraperitoneally (see Materials and Methods) and splenocytes were prepared at 48 hpi. Innate immune cells were analyzed as described in A. The data represent the mean from three independent experiments with a total of 6 mice (2 each experiment). Error bars indicate the mean6SD. Statistical significance was determined using Student's t test. Gating parameters are indicated in Figure S3. doi:10.1371/journal.pone.0048273.g007 neutrophil apoptosis followed by their phagocytosis by macrophages is an essential mechanism for regulating neutrophil functions and is an important control point in the development and resolution of inflammation [59,60]. Neutrophil numbers are not compromised in Ccl3 mutant mouse [61]. In the absence of macrophage function, the tissue would be exposed to cytoxic molecules released from apoptotic neutrophils and may thereby aggravate the injury. Future studies directed towards understanding the neutrophils function in the Ccl3/Npc1 double knockout mouse may provide a better understanding of neutrophil and macrophage involvement in NPC disease. In addition whether neutrophils are elevated in human NPC spleen and liver needs to be investigated.
Future studies will also focus on determining whether lysozyme and other lysosomal/secretory proteins are disease markers in human NPC patients as well as other lysosomal disorders. One early report suggests a modest increase in plasma lysozyme in four adult patients with Gaucher's disease [62]. Elevated lysozyme transcripts and protein have been found in neuronal cells in the brain of another lysosomal disorder mouse model Sanfilippo syndrome type B (also known as MPS IIIB) [63,64]. A linkage between lysozyme and hyperphosphoylated tau has been suggested in the MPS IIIB mouse brain [64]. At high concentration, lysozyme on its own is known to be amyloidogenic [65] and exposure of cultured rat neurons to oligomers of hen egg white lysozyme had been found to induce hyperphosphorylation of tau [66]. Thus, in addition to serving as secretory markers, lysozyme and other secreted lysosomal proteins expressed in glial and neuronal (and possibly endothelial) cells in the brain, may also exacerbate neurological disease.

Microarrays and Expression Analyses
Brain from 11 Npc1 2/2 and 16 control female mice (Npc1 +/+ and Npc1 +/2 ) age ranging from 20-84 days (see Figure S1A for details) and spleen and liver from 6 Npc1 2/2 and 6 Npc1 +/2 female mice age ranging from 20-71 days (see Figure S1B for details) were surgically harvested, kept in RNA later and stored at -20uC until used. RNA was isolated using Roche MagNa Pure Compact automated system and labeling was done using MessageAmp TM Premier RNA Amplification Kit (Invitrogen). Affymetrix mouse 430 2.0 array hybridizations were performed by 'UCLA Clinical Microarray Core', UCLA, Los Angeles, CA, USA, following standard Affymetrix GeneChip Expression Analysis protocol. RNA from each animal was profiled individually. The acquisition of array image was undertaken by using Affymetrix GeneChip Command Console 1.1 (AGCC). Subsequent raw data were analyzed using DNA-Chip Analyzer (D-Chip) with the.CEL files obtained from AGCC. This analysis was undertaken irrespective of consideration of littermates. A PM/MM difference model was used for estimating gene expression levels and combined with a quantile approach for data normalization. Thresholds for selecting significant genes were set at a relative difference $1.5-fold, absolute difference $100 signal intensity units and p,0.05. Genes that met all three criteria were considered as significantly changed. All data are available from NCBI, GEO accession number GSE39621.

Identification of Secretory Proteins that Show Agedependent, Over-expression in Brain and Liver
Genes up regulated in the brain of Npc1 2/2 mice across all time points, were further selected for secretory proteins identified by an N-terminus signal sequence, recognized by SignalP 4.0 (http://www.cbs.dtu.dk/services/SignalP/). The UniProt database (http://www.uniprot.org/) was also utilized to confirm the presence of a signal sequence and identify additional secretory proteins that lack conventional signal sequences. Proteins known to localize to membranes or predicted to have transmembrane domains as predicted by the UniProt database were filtered out. The resulting short list from the brain was cross referenced with genes over expressed in liver at all time points to yield 18 genes. For each of these genes, the mean signal intensities detected for age matched Npc1 +/2 (control) mice on the microarray chip was subtracted from that seen with Npc1 2/2 mice. This yielded 12 genes with progressive age-dependent increase at three distinct time points across the animal's life span in both brain and liver.

In vivo Infection of Mice
Salmonella enterica serovar Typhimurium SL1344 was grown in Luria-Bertani (LB) broth containing streptomycin sulfate (50mg/ ml). Female Npc1 +/+ , Npc1 +/2 and Npc1 2/2 mice (age 6-8 weeks) were used for the S. typhimurium infection. Bacteria from overnight cultures were pelleted by centrifugation for 5 min at 6000 rpm and were re-suspended in PBS. Mice were given 1610 4 bacteria in 100ml by i.p injection. Serial dilutions of inoculants were plated on selective media to determine the actual doses. At 48 hours post infection (hpi), mice were sacrificed. Spleen and liver were isolated, weighed, homogenized, serial dilutions were made and plated on selective media to determine the number of bacterial colony forming units (CFU).

Flow Cytometry
The number and types of different immune cells in spleen of female Npc1 +/2 and Npc1 2/2 littermates (6-8 weeks) were enumerated as follows. Spleens were harvested, splenocytes were prepared and cells were counted using a hemocytometer. S. typhimurium infection of mice was performed as described earlier and splenocytes were isolated 48 hpi. For flow cytometry, cells were stained with fluorophore conjugated antibodies to CD335 (FITC; for NK cells), CD11c (FITC; for dendritic cells), F4/80 (FITC; for macrophages) CD11b and Gr-1 (PE and APC respectively, for neutrophils). Cells positive for both F4/80 and CD11b were considered monocytes/macrophages whereas cells positive for CD11b and high Gr-1 expression were considered neutrophils. Depending on the requirements and fluorophore compatibility splenocytes were stained either separately or in combinations. Suitable isotype control for each antibody was included as controls and compensation was performed wherever required. 10 5 events were typically recorded in Beckman Coulter FC500 flow cytometer.

Organ Harvest and Immunohistochemistry
Female, littermates, Npc1 +/2 and Npc1 2/2 mice (age 48-52 days) were sacrificed by asphyxiation using CO 2 The circulatory bed was washed with PBS (pH 7.4), and subsequently perfused with 10% neutral buffered formalin (,4% formaldehyde). The organs (brain, liver, lung and spleen) were surgically harvested and stored in 4% formaldehyde at room temperature (RT) until transfer to paraffin. Formalin paraffin-embedded tissue sections (3-4 mm) were dewaxed in xylene and alcohol. Antigen retrieval was done by pre-incubation of deparaffinized samples with 0.05% proteinase K (Dako, Germany) in 50mM Tris-HCl (pH 7.5) for 8 min at RT. After washing, the sections were immersed in 3% H 2 O 2 in distilled water for 20 min at RT to block endogenous peroxidase. After an additional wash with PBS, the sections were treated with 5% rabbit serum for 30 min, followed by successive incubation in avidin and biotin (Avidin/biotin blocking kit, Vector Laboratories) to block endogenous biotin. Anti-mouse Gr-1 (5mg/ ml in PBS with 2% rabbit serum) was applied to the sections for 60 min at RT. Secondary antibodies were biotinylated rabbit antirat IgG (mouse absorbed, Vector Laboratories). Reagents were prepared according to the manufacturer's instructions. The peroxidase complexes were revealed by incubation with 3,39diaminobenzidine-tetra-hydrochloride (DAB, Vector Laboratories) and the sections were lightly counterstained with Mayer's hemalum. The slides were then mounted in cytoseal XYL (Thermo Scientific, Kalamazoo, USA). Sections stained only with secondary antibodies served as controls. Pictures were acquired on a Nikon Olympus microscope, using a Nikon digital DS-Fi1-U2 camera controlled by NIS-Elements F3.0 Nikon software (all from Nikon Instruments INC, Tokyo, Japan). Images were visualized with A10 PL 106/0.25, or a DPIan Apo 406/1.00 oil-immersion or a DPIan Apo 1006/1.30 oil-immersion objective lens (Nikon).

Lysozyme Activity Assay in Mouse Plasma
Lysozyme activity in the plasma of Npc1 +/+ , Npc1 +/2 and Npc1 2/2 mice was measured using fluorescence based lysozyme assay kit (EnzCheck, Molecular Probes, Grand Island, NY, USA). The assay measures the lysozyme activity on Micrococcus lysodeikticus cell walls, which are labeled to such a degree that the fluorescence is quenched. Lysozyme action relieves this quenching; yielding an increase in fluorescence that is proportional to lysozyme activity.
Plasma from both female and male Npc1 nih mice corresponding to 50-500 mg protein (,2 to 10 ml in volume) was used in a 100 ml reaction volume. The reaction was carried out either at 37uC for 1 h (when 500 mg plasma protein was used) or at 37uC for 24 h (when 50 mg plasma protein was used). For Npc1 nmf164 mice, we used 50 mg plasma protein and the reaction mixture was incubated at 37uC for 24 h. Fluorescence was read using excitation/emission of 494/518 nm in a multiwall plate reader spectramax M2 (Molecular devices, CA, USA). The values obtained were normalized to 1 by dividing the numbers by the maximum value of lysozyme obtained among Npc1 +/2 mice. Purified chicken egg white lysozyme was used as a positive control.

Drug Injections and Blood Withdrawal
Starting at P21-27 and once a week thereafter, Npc1 nmf164 homozygous mutant female mice were injected i.p with 20% 2hydroxypropyl-beta-cyclodextrin (HPbCD, 4000 mg/Kg) prepared in 0.2% DMSO and 0.9% saline. Control mice received 0.2% DMSO in 0.9% saline. Blood via cheek bleed was collected from mice, age 50-55 days from both treatment groups in EDTA tubes (BD, CA). Plasma was separated by centrifugation at 2500 rpm for 15 min and stored at -70uC until used. For hematology analyses, 20 ml blood was collected in a microfuge tube coated and dried with 20 ml of 1.25 mg/ml EDTA. Blood cell parameters were analyzed by Hemavet 950 (Drew Scientific, Dallas).

Miscellaneous
All animal experiments were performed with the approval and authorization from the 'Institutional Review Board' and the 'Animal Care and Use Committee', University of Notre Dame. Student's t test was carried out to determine the statistical significance of the data. p#0.05 considered significant. Figure S1 Pictorial representation of the experimental design of whole-genome gene-expression analysis for brain, spleen and liver. (A) Chart displaying the experimental set up for the microarray experiment using brain from 27 mice (11 Npc1 2/2 and 16 controls) age ranging from 20-84 days. (B) Chart displaying the experimental set up for the microarray experiment using liver or spleen from 12 mice (6 Npc1 2/2 and 6 controls) age ranging from 20-71 days. +/+ denotes Npc1 +/+ , +/2 denotes Npc1 +/2 and 2/2 denotes Npc1 2/2 mice. (TIF) Figure S2 Age-dependent over expression of 18 secretory genes in brain and liver of Npc1 2/2 mice. The raw signal intensity of all 18 genes obtained after the Dchip analysis of brain and liver transcripts at three time points (1 corresponds to 20-25 days, 2 corresponds to 54-55 days and 3 corresponds to 67-71 days for liver and 81-84 days for brain) were taken and mean value was calculated. Mean signal intensity of 2 Npc1 +/2 mice was subtracted from the mean signal intensity values of 2 Npc1 2/2 mice between age-matched animals. The process was carried out for each gene at all three time points for both brain and liver. The difference obtained was plotted as a function of time.

Supporting Information
(TIF) Figure S3 Flow cytometric enumeration of different innate immune cells in spleen. Representative data showing the staining of splenocytes with different cell surface markers. Splenocytes from un-infected (left panel) and infected (right panel) with S. typhimurium at 48 hpi from Npc1 +/2 and Npc1 2/2 female litter mate mice (age 6-8 weeks) were isolated and stained with fluorophore conjugated antibodies, anti-CD335 for NK cells (panel A), anti-CD11c for dendritic cells (panel B), anti-F4/80 and CD11b for monocytes and macrophages (panel C) and anti-Gr-1 and CD11b for neutrophils (panel D). Monocytes and macropahges are represented into two sub groups, (i) F4/80 hi CD11b lo and (ii) F4/80 lo CD11b hi whereas cells positive for CD11b and have high expression of Gr-1 (Gr-1 hi CD11b + ) were considered neutrophils. Open hsitograms represent the staining with isotype control and gray histograms represent the staining by specific antibodies as mentioned. (TIF) Table S1 List of differentially expressed genes in the brain across the life span (20-84 days) of Npc1 2/2 mice. Up regulated innate immunity genes listed in InnateDB are shown in bold letters. (XLS)