Advanced Institute of Mathematics , Statistics and Computer Science ( AIMSCS ) Author ( s ) :

Systemic lupus erythematosus (SLE) commonly accredited as “the great imitator” is a highly complex disease involving multiple gene susceptibility with non-specific symptoms. Many experimental and computational approaches have been used to investigate the disease related candidate genes. But the limited knowledge of gene function and disease correlation and also lack of complete functional details about the majority of genes in susceptible locus, encumbrances the identification of SLE related candidate genes. In this paper, we have 15 studied the human immunome network using various graph theoretical centrality measures integrated with the gene ontology terms to predict the new candidate genes. As a result, we have identified 8 candidate genes which may act as potential targets for SLE disease. Introduction Systemic lupus erythematosus (SLE) is a polygenic and multi-factorial disease, which often manifests different clinical phenotypes \citep{Castro08}. The conventional therapies for SLE include antimalarial drugs, immunosuppressive medications, nonsteroidal anti20 inflammatory drugs (NSAIDs), which are non-specific and immunosuppressive. So the research is concentrated on developing the targeted therapies. Investigation of genetic predisposition through gene expression profiling and linkage analysis in multiple populations generates large sets of potential candidate genes. This approach successfully predicts genes for the diseases that cause a single risk, but fails to identify the genes causing complex disease \citep{Glacier02}. This necessitated the development of in silico approaches such as ontology based, computation-based, and text based for the analysis of complex diseases\citep{Zhu07}. In silico methods utilize the 25 information of protein interactions, GO terms, gene expression data, sequence features, protein domains, protein function, orthologous connections, chromosomal regions, pathways, mutations (SNPs), chemical components, disease probabilities etc for predicting the candidate gene. Recently, several online tools have been developed for prioritizing candidate genes, which usually combine the different in silico 30 approaches\citep{Tran10}. For example, SUSPECTS \citep{Adie06} ranks genes by matching sequence features, GO terms, interpro domains, and gene expression data. ToppGene \citep{Chen09} uses functional annotations, protein interaction networks to prioritize disease specific genes. Different tools like Polysearch \citep{Cheng08}, MimMiner \citep{Driel06}, and BITOLA \citep{Hris05} relies on biological data mining. Posmed, a computational based approach prioritizes candidate genes using an inferential process similar to artificial neural network comprising documentrons \citep{Yoshi09}. Some tools like Phenopred use disease phenotype information 35 which employs/associates data from gene-disease relations \citep{Radi08}, protein-protein interaction data, protein functional annotation at a molecular level and protein sequence data to detect novel gene-disease associations in humans. All these online tools have been successfully used for the prediction of candidate gene in diseases like epilepsy \citep{Sha10}, and osteoporosis \citep{Huang08}, type II diabetes \citep{Tiffi06} and gene prioritization, depending on information of chromosomal location or genes differentially expressed in a tissue. But the above approaches have failed in case of SLE as it involves genes of differential expression patterns in tissues, influenced 40 by various environmental factors. The limited information about the markers of SLE also contributed to their failure\citep{Ref crispin et al 2010}. In such a situation, the network centrality measures coupled with the ontological terms favoured the identification of candidate genes for SLE. Analysis of Protein -protein interaction networks aids for inferring the function of uncharacterized proteins. In the recent past many 5 network based analyses have been developed for protein function prediction, identification of functional modules and disease candidate gene prediction etc. With the advent of sophisticated technologies for the functional annotation of genes, the candidate genes prioritization have become increasingly facile. GO terms are used for the systematic annotation of genes. In the present work, we study the human immunome network in combination with the graph theoretic centrality measures and GO terms 10 in order to identify candidate genes for SLE disease. For this purpose we have adopted the procedure developed by CsabaOrtutay {\it et~al}. \citep{Ortu09}. In our work, we have applied more number of centrality measures such as eccentricity, centroid values, SPbetweenness, CF-closeness, CF-betweenness, Katz Status, Eigen vector and Page Rank apart from closeness and degree centrality used by Csaba Ortutay {\it et~al}. This study effectively analyzes candidate genes because of the consideration of more graph centrality measures. This method can be applied to investigate other immune diseases for which sufficient information is present in databases. 15 Results We have constructed the human immunome network from the data collected from the HPRD \citep{Pra09} and the immunome database \citep{Ortu06}. The core immunome network contains 548 proteins with 1372 interactions (Figure: 1). In this work, we have adopted the method developed by CsabaOrtutay to predict the SLE related candidate genes \citep{Ortu09}. CsabaOrtutay and co-works in their work 20 used three centrality measures such as degree, closeness and efficiency for the prediction of primary immunodeficiency (PID) related genes. In our study, we have applied ten centrality measures \citep{Jun06} to find the potential of individual proteins from the immunome network. From the Fig 1 which shows the Spearman correlation coefficients for all pairs centrality measures, we conclude that the centrality measures of the immunome network are not coincidental. The other topological properties like the degree distribution that follows power law behavior ($\gamma$ = 1.86), the average degree is 4.995 and the diameter is 12 were also calculated 25


INTRODUCTION
Complex tissues such as the lung are composed of structural hierarchies such as alveoli, alveolar ducts, and lobules (1). Recent evidence indicates that these structures reflect not only mechanical relationships, but also functional units that selectively participate in integrated processes such as tissue regeneration (2) and the development of bronchioalveolar carcinoma (3). Because bulk analyses of cells and gene expression -ignoring these regenerative units -have failed to illuminate the interactions between cells participating in these processes (4,5), there is growing interest in the spatial sampling of the cells within regenerative units.
The initial attempts to refine the bulk analysis of cells within regenerating tissues used flow cytometry (FC) (6). FC is a multidimensional high-throughput technology that can analyze and sort individual cells based on their phenotypic characteristics -in most cases, sorting is based on cell surface molecule expression (7,8). This approach has provided useful insights into the molecular expression of cell populations participating in lung regeneration (9)(10)(11)(12); nonetheless, the relatively large number of cells sorted by FC has potentially obscured the cell-cell interactions within individual regenerative units.
To address these limitations, there is growing interest in isolating cells within individual regenerative units; in lung regeneration, the anatomic unit appears to be the alveolar duct. This goal has been aided by two developments. First, computer-controlled laser microdissection of tissue has facilitated the rapid isolation of small anatomic units within complex tissues. The ability to harvest anatomic units, such as the alveolar duct, under physiologic conditions has enabled detailed single-cell analysis. Second, microfluidic devices capable of capturing and handling single cells allowed for the analysis of the small number of cells comprising anatomic units such as the alveolar duct. Furthermore, these isolated cells can now be analyzed for gene expression at the level of individual cells. The result is an opportunity to define not only the signaling interactions between cells, but also the molecular interactions and signaling pathways within cells (9,11,13).
In this report, we demonstrate the use of laser microdissection to isolate cells in the murine lung alveolar duct and the microfluidic isolation and gene expression analysis of individual cells within this structural unit.

ANIMALS
Male mice, 8-10-week-old wild type C57BL/6 (Jackson Laboratory, Bar Harbor, ME, USA) were anesthetized as previously described (14). The care of the animals was consistent with guidelines of the American Association for Accreditation of www.frontiersin.org Laboratory Animal Care (Bethesda, MD, USA) and approved by our Institutional Animal Care and Use Committee.

ANESTHESIA AND INTUBATION
The animals were anesthetized with an intraperitoneal injection of ketamine 100 mg/kg (Fort Dodge Animal Health, Fort Dodge, IA, USA) and xylazine 6 mg/kg (Phoenix Scientific, Inc., St. Joseph, MO, USA). The animals were intubated under direct visualization with a standard 20 g angiocatheter (BD Insyte, Sandy, UT, USA) (14) for subsequent agarose instillation.

CORROSION CASTING AND SEM
The right ventricular outflow tract was cannulated via right ventriculotomy with a 2 mm olive-tipped cannula (Acufirm 1428LL, Dreieich, Germany). The pulmonary vessels were perfused with 15-20 ml of 37°C saline followed by a buffered 2.5% glutaraldehyde solution (Sigma, St Louis, MO, USA) at pH 7.40. After casting of the pulmonary microcirculation with PU4ii (VasQtec, Zurich, Switzerland), diluted with 20% methylmethacrylate monomers (Sigma), and caustic digestion, the microvascular corrosion casts were imaged after coating with gold in an argon atmosphere with a Philips ESEM XL30 scanning electron microscope as previously described (13).

EMBEDDING MEDIUM
Agarose at 3% (w/v) was thoroughly mixed with distilled H 2 O and warmed to 37°C. The trachea, cannulated with a 20 g Angiocath (BD Insyte), was infused with the 37°C agarose using the lowest pressure necessary to inflate the peripheral lung (typically 20 cm H 2 O pressure). At total lung capacity, the trachea was clamped and the lung block was placed in 4°C saline and allowed to harden.

LASER MICRODISSECTION
The Arcturus XT LCM System (Life Technologies) was used for all ultraviolet (UV) laser dissection. The UV laser was a specially adapted beta-test laser for wet tissue applications. The enhanced UV cutting laser was a solid-state, diode-pumped (345 nm) laser with adjustable current (0-100%) and pulse frequency (10-5000 Hz). After mounting on the Arcturus stage, the surface of the lung slice was blotted dry. The alveolar ducts were optically targeted using a Nikon Eclipse Ti-E microscope with 10× and 20× Nikon CF objectives with high-intensity LED illumination. The stage was adjusted using a motorized trackball actuated in Xand Y -axes with 1 µm precision. The UV laser was empirically adjusted between 30 and 50% of maximal current to ensure complete penetration of the tissue slice. The Arcturus XT software was used to target tissue for UV dissection.

ENZYMATIC DIGESTION
Enzymatic digestion of the lung reflected a previously published protocol (15). Briefly, collagenase Type IV and DNase I (Sigma) was used to dissociate the tissue. The digestion was performed at 37°C under constant agitation until dissolved. The digest was filtered through 35 µm nylon mesh in preparation for microfluidic analysis.

C1 MICROFLUIDICS CHIP
The C1 single-cell auto prep array integrated fluidic circuit (chip) is the microfluidic component of the Fluidigm C1™ Single-Cell Auto Prep System. The medium chip (10-17 µm cell size) was used to capture single cells (96 on one chip) obtained from the laser microdissection. The chip was used to perform single-cell PCR using a custom 96 gene panel designed for the alveolar duct.

FLOW CYTOMETRY
After standard enzymatic digestion of the lung (10), FC cell sorting using anti-CD31 mAb (APC or PE-Cy7; rat IgG2a, clone 390, eBioScience) (18) was used to isolate the CD31 + endothelial cells. Endothelial cell gene expression was assessed by PCR array as previously described (6).

PCR GENE EXPRESSION
Comparison of single-cell and FC values were based on Ct values (6). For population analyses, the gene expression based on the ∆Ct method (19) used an average of five housekeeping genes. The ∆Ct values were normalized to an arbitrary 100-point linear scale for comparison with single-cell values.

PRECISION-CUT LUNG SLICES
Structural analysis of the murine lung using SEM demonstrated a mean diameter of the intact alveolar duct of 177 ± 31 µm (Figure 1). To obtain thickness-calibrated lung slices encompassing the alveolar ducts, 200 µm thick precision-cut lung slices were sectioned after intra-airway instillation of a 3% (w/v) agarose embedding medium. At 37°C, low melting point agarose had sufficiently low viscosity to permit instillation into the peripheral airspaces -important for the preservation of alveolar architecture. When cooled to 4°C, the agarose-lung block demonstrated stability for precision sectioning. The agarose-lung block, mounted on Leica VT1000 S specimen holder and immersed in ice-cold PBS, was serial sectioned (Figure 2). For single-cell genomics studies, the lung slices were mounted on a PEN membrane frame slide to Frontiers in Oncology | Molecular and Cellular Oncology  provide structural support during laser dissection. The frame slide facilitated the retrieval and subsequent processing of the dissected specimen (Figure 3).

OPTICAL TARGETING
The hierarchical anatomic structure of the lung facilitated the visual identification of the alveolar duct. The dehydrated lung used in laser capture loses structural detail when examined by light microscopy (not shown). In contrast, the well-hydrated lung used in laser microdissection -in the absence of counterstains or cell-specific labels -retains structural detail throughout the 200-300 µm thick precision-cut lung slices (Figure 4). Preliminary identification of the alveolar duct was confirmed by varying the optical plane and identifying the feeding bronchiole ( Figures 4A,B). The computer-controlled laser software allowed for precise selection of regions encompassing the alveolar duct; the demarcated regions were harvested by the UV cutting laser (Figures 4C,D). The laser power, speed, and focal point were varied to optimize single-pass dissection of the tissue and minimize thermal damage. The required laser power was consistently reduced by 50% with absorbent blotting of the section surface followed by re-application of media immediately following UV dissection.

SINGLE-CELL ISOLATION
The laser-dissected samples were enzymatically digested, microfiltered, and prepared for microfluidic analysis (Figure 5). The PEN membrane-associated samples were incubated in collagenase www.frontiersin.org FIGURE 3 | The tissue slice was mounted on a frame slide. The slide is composed of a standard optical microscope slide with a metal frame that supports a PEN membrane to facilitate laser microdissection (A). In cross-section (B), the frame slide creates a reservoir for retrieval of the dissected tissue. Type IV buffer for 15-30 min at 37°C with constant agitation. The acellular matrix and PEN membrane-associated debris were microfiltered with a 35 µm mesh. Routine examination of the filtered debris demonstrated virtually no residual cells. Cell yields varied from 1 to 3 × 10 3 cells per alveolar duct. Cell viabilities, assessed by vital dye exclusion as well as live/dead assay microfluorimetry, ranged from 70 to 95% depending on treatment conditions. Live/dead confocal microscopy demonstrated dead cells at the laser-dissected sample margins (not shown), but cell viability from laser-dissected tissue (87 ± 5) was not statistically different from enzymatic digestion alone (95 ± 4) (p > 0.05). The cells were analyzed using the Fluidigm C1™ Single-Cell Auto Prep System (Fluidigm, South San Francisco, CA, USA). In the 96-well C1 system, a typical result was the capture of 75 single cells, 13 wells with 2 or more cells, and 2 wells with no cells.

SINGLE-CELL PCR
The C1 chip used a microfluidic circuit to capture the isolated single cells (96 on one chip). The single-cell poly-adenylated RNA was converted to full-length cDNA for universal amplification of the cDNA and custom PCR analysis. Our custom PCR arrays were developed based on gene expression studies using FC-based phenotypic isolation of lung cells (6,(10)(11)(12)20). Results of the C1 PCR studies were analyzed by unsupervised hierarchical clustering (Figure 6). Upon filtering based on array QC readouts, data on 72 single-cells were available. In addition, we have eliminated the genes that showed low variation (SD less than 1) followed by feature selection. Thus, we obtained a smaller matrix of 72 singlecells × 42 marker genes. Clustering based on selected features (using the Sparse Hierarchical Clustering module in GenePattern software from Broad Institute of MIT) demonstrated single-cell clusters (Figure 7) that corresponded to previously identified cell types (e.g., endothelial cells, myofibroblasts, alveolar macrophages, and alveolar epithelial cells). The information content of singlecell analysis was apparent when gene expression of endothelial cells (Pecam1 gene expression) isolated by the C1 chip was compared to gene expression of endothelial cells (CD31 + surface expression) isolated by FC. Although the expression of many genes was similar in both groups (Figure 8A), some genes demonstrated variable expression patterns that were not reflected in the population-based FC analysis (Figure 8B).

DISCUSSION
In this report, we demonstrated laser microdissection of the lung alveolar duct for single-cell PCR analysis. Our approach involved three steps. (1) The initial preparation used mechanical sectioning of the lung tissue with sufficient thickness to encompass the structure of interest. In the case of the alveolar duct, the precision-cut lung slices were 200 µm thick and processed using near-physiologic conditions to preserve cell viability. (2) The lung slices were examined by transmission light microscopy to target the alveolar duct. The air-filled lung was sufficiently accessible Frontiers in Oncology | Molecular and Cellular Oncology by light microscopy that counterstains or fluorescent labels were unnecessary to identify the alveolar duct. (3) The enzymatic and microfluidic isolation of single cells for PCR allowed for the harvest of as few as several thousand cells for analysis. The data suggest the unique value of this approach to understanding the intraand intercellular interactions within the regenerating or neoplastic alveolar duct.
Harvesting viable cells from the alveolar duct requires that tissue slices of the appropriate thickness have to be obtained in nearphysiologic conditions. Manually sliced tissues have been used for nearly 90 years in various in vitro applications (21,22); however, the manually prepared slices have had limited reproducibility and viability (23). A mechanical tool to facilitate the cutting of tissue of precise thickness was introduced by the Krumdieck in 1980 (24,25). With progressive refinement, tissue microtomes now vibrate with varying amplitudes and frequencies to produce slices with highly reproducible thickness -thus, these sections are referred to as precision-cut tissue slices (26). With current vibrating microtomes, precision-cut tissue slices preserve the original organ architecture, the cellular heterogeneity, and extracellular matrix found in vivo (27). Precision-cut tissue slices, for example, have enabled the development of ex vivo systems for systematic pharmaco-kinetic profiling of tumors contained in their native 3-dimensional micro-environment (28).
A key feature of precision-cut lung slices in the lung is the use of agarose as structural support. Because live tissue is too soft for precise sectioning, and molten paraffin is too hot for live tissue, a biocompatible liquid with a low-temperature gelling point is used as an embedding medium. Agarose is typically used for lung embedding because it has sufficiently low viscosity at 37-40°C to be infused into the distal airspaces, yet is sufficiently rigid for microtome sectioning at 4°C.
Although there is a significant overlap between techniques, laser microdissection can be distinguished from laser capture microdissection (LCM) by the method of cell isolation. In the original LCM system developed at the National Institutes of Health, the target cells were identified by light microscopy and captured using an infrared laser (29). A transfer film positioned over the region of interest (typically tumor cells) was melted into the tissue using an IR laser. The dissected tissue microsample was then lifted from the histopathology slide for subsequent RNA, DNA, or protein analysis. In most contemporary LCM systems, a requirement for effective capture is the dehydration of the tissue -a requirement that precludes the study of live cells (30). In contrast, our laser microdissection approach uses a cutting UV laser to define the margins of the dissection under near-physiologic conditions. Although UV-induced thermal injury to the neighboring cells is demonstrable using live-dead assays; the dissection margins can be www.frontiersin.org  drawn to minimize the injury to the region of interest. In our application, we maintained near-physiologic conditions during the UV laser dissection with only blotting of the tissue surface to minimize excess liquid media prior to UV dissection. The UV laser-defined alveolar duct was subsequently transferred to the collection chamber without the use of an IR laser or transfer film; in most cases, the sample dropped into the chamber occasionally assisted by the application of liquid media.
A frequently under-appreciated characteristic of the lung is the large amount of extracellular matrix in the organ (31). In addition, some extracellular matrix components -most notably, elastin -are resistant to common enzymatic treatments (32). In our application, we have used a variety of enzymatic cocktails based on collagenase to facilitate dissolution of the laser-dissected tissue; however, the most important component has proven to be collagenase Type IV. In our protocols, we used 35-µm mesh to separate cells from the collagenase-resistant matrix debris. Routine re-examination of the filtered debris did not show residual cells suggesting that this approach does not result in the systematic loss of any particular cell subpopulation. Of note, cell filtration with mesh larger than 35 µm was prone to debris-induced occlusion of the microfluidics chip.
The cells isolated by microfluidics were studied using a custom 96 gene PCR panel. The results of prior FC population studies were used to select the genes used in this custom panel (6,(9)(10)(11)(12)20). FC-derived cell type-specific markers were incorporated into the panel to facilitate the spatial reconstruction of intercellular interactions within the alveolar duct. For example, the genes Acta2 and Pecam1 were added to the PCR panel to identify myofibroblasts and endothelial cells, respectively. Although the single-cell transcriptional data must be analyzed in the context of potential limitations, such as burst transcription (33,34), insights derived from these data should be useful for re-interpreting population-based data both in vivo and in vitro.
Single-cell analysis is particularly relevant to the process of carcinogenesis within the alveolar duct. Bronchioalveolar stem cells, in addition to contributing to the maintenance and repair of the normal alveolar duct, have been shown to give rise to lung adenocarcinomas (35). The experimental challenge is that multipotent stem cells, like those giving rise to bronchioalveolar carcinoma, exist in low frequency within the peripheral lung. Populationbased analyses are prone to overlook these cells and their transcriptional profile because of the "averaging" effect demonstrated in our study. Because of this limitation, we anticipate that single-cell analysis of the regenerating alveolar duct will provide important insights into the biology of bronchioalveolar stem cells.
Finally, these results demonstrate the ability to extract viable cells from a morphologically defined micro-region of a tissue sample. We successfully applied single-cell whole-transcriptome amplification and gene expression analyses to these samples -a process that allows for simultaneous region-specific and cell typespecific expression analysis. We anticipate even more interesting experiments are possible in the future. For example, similarly obtained viable single cells can facilitate a variety of experimental perturbations prior to endpoint analyses (e.g., expression profiling, nucleic acid sequencing, immunoassays, or proteomics). Furthermore, new statistical and bioinformatics techniques may be used to design efficient strategies for sampling cells at different locations in the tissue and at different time-points, and to study the genomics and transcriptomics of these cellular samples, thus leading to construction of spatio-temporal cell signaling networks. We are confident that such studies involving single-cell analysis will lead to a more nuanced understanding of the process of tissue regeneration.