Gene Expression Profiling in Multiple Sclerosis Brain

Abstract

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system and the leading cause of non-traumatic neurological disability in young adults in the United States and Europe. The clinical disease course is variable and starts with reversible episodes of neurological disability in the third or fourth decade of life. Microarray-based comparative gene profiling provides a snapshot of genes underlying a particular condition. Several large scale microarray studies have been conducted using brain tissue from MS patients. In this review, we summarize existing data from different gene expression profiling studies and how they relate to understanding the pathogenesis of MS.

1. Introduction

Multiple sclerosis (MS) is an inflammatory demyelinating disease of the central nervous system (CNS). MS affects more than two million people worldwide, making it the leading cause of non-traumatic neurological disability in young adults in North America and Europe (Hauser and Oksenberg, 2006; Noseworthy, 1999; Noseworthy et al., 2000; Trapp and Nave, 2008; Weinshenker, 1998). The majority (~85%) of MS patients have a biphasic disease course, beginning with the primary phase termed relapsing-remitting MS (RR-MS). During this phase, patients experience alternating episodes of neurological disability and recovery that can last for many years (Hauser and Oksenberg, 2006; Noseworthy, 1999; Noseworthy et al., 2000; Trapp and Nave, 2008). Within 25 years, ~90% of RR-MS patients develop a secondary-progressive disease course (SP-MS), which is characterized by steady neurological decline (Noseworthy et al., 2000; Trapp and Nave, 2008; Weinshenker et al., 1989). About 10% of MS patients also exhibit a disease course with steady decline in neurological function without recovery and are classified as primary progressive MS (PP-MS). A small minority of MS patients (~5%) suffer from a disease course with progressive neurological decline accompanied by well demarcated acute attacks with or without recovery. This disease course is classified as progressive-relapsing MS (PR-MS).

Since the late 1990’s, MS research has refocused on the role of axonal and neuronal pathology, and neurodegeneration is generally accepted as the major cause of irreversible neurological disability in MS patients. Axonal transection and degeneration occur in the setting of acute inflammatory demyelination (Trapp et al., 1998) and as a consequence of chronic demyelination (Bjartmar et al., 2000; Dutta et al., 2006; Ganter et al., 1999; Lovas et al., 2000). As axon pathology and frequency of transected axons in acute MS lesions correlate with the degree of inflammation (number of immune cells) (Ferguson et al., 1997; Trapp et al., 1998), early axonal transection is thought to occur due to vulnerability of demyelinated axons to inflammation. Activated immune and glial cells release a plethora of substances including proteolytic enzymes, matrix metalloproteases, cytokines, oxidative products, and free radicals that can damage axons (Hohlfeld, 1997; Nave and Trapp, 2008). Inducible nitric oxide synthase (iNOS), a key enzyme required for synthesis of nitric oxide (NO), is significantly increased in acute MS lesions (Bo et al., 1994; Liu et al., 2001). Additionally, glutamate-mediated excitotoxicity is observed in many acute and chronic neurodegenerative conditions (Lipton and Rosenberg, 1994). Another possible mechanism of axonal degeneration in MS is a specific immunologic attack on the axon, suggested by the strong correlation between inflammation and axonal transection (Trapp and Nave, 2008; Trapp and Stys, 2009; Weiner, 2009). The terminal axonal ovoids are often surrounded by macrophages and activated microglia in acute MS lesions (Trapp et al., 1998). Whether these cells are directly attacking axons, protecting axons or removing debris remains to be determined. Antibodies to axonal components in the CNS have not, however, been localized to MS lesions (Hafer-Macko et al., 1996; Ho et al., 1998). T-cells in individual MS patients change depending on disease activity (Khoury et al., 2000), CD4+ T cells are the most prominent cells in active lesions but are absent in chronic MS lesions (Bennett and Stuve, 2009). In some cases CD8⁺ cells outnumber the CD4⁺ T cells, thereby suggesting the former as driving cytotoxicity (Crawford et al., 2004). Both CD4⁺ and CD8⁺ T-cells have been identified as possible mediators of axonal transection in MS lesions (Babbe et al., 2000; Skulina et al., 2004), in EAE mice (Huseby et al., 2001), and in vitro (Giuliani et al., 2003; Medana et al., 2001). Further, some reports indicate that axonal subpopulations may be targeted by immune-mediated mechanisms (Evangelou et al., 2001; Ganter et al., 1999; Lovas et al., 2000). Additionally, since most axons survive the acute demyelinating process, it seems unlikely there is a specific immunological attack against axons. Despite the current paucity of direct evidence supporting a specific immunological attack on axons in MS, the possibility of cell-mediated mechanisms of axon loss is worthy of investigation.

From a genetic standpoint, MS is considered to be a complex polygenic disease characterized by a modest inherited risk for disease susceptibility. The risk for MS in many parts of North America is 1:1,000, with a 10 fold greater inheritance risk for females. There is greater susceptibility towards development of MS in families, relatively high risk in Northern Europeans and relatively low risk in Africans, Asians and American Indians (Rosati, 2001). The only consistent MS-associated gene is the HLA-DRB1 gene on chromosome 6p21 (Oksenberg et al., 2008), which accounts for 16–60% of the genetic susceptibility to MS (Haines et al., 1998). The interleukin-7 receptor alpha chain and interleukin-2 receptor alpha chain have been recently identified as additional (<0.4%) inheritable risk factors for developing MS (Gregory et al., 2007; Hafler et al., 2007). A number of studies support a role for environment in MS disease susceptibility. Data suggest that environmental exposure is time-sensitive and occurs before the age of 14 (Ascherio and Munger, 2007a; Ascherio and Munger, 2007b; Munger and Ascherio, 2007). While viruses, in particular Epstein barr virus (EPV) is the most common infectious factor, environmental agents like vitamin D has also been associated with the disease. Risk assessment studies have indicated genetic factors as major contributors of familiar clustering of MS patients (Ebers et al., 2000). Based on several years of genetic research, MS has been clustered with complex genetic diseases, a group of common disorders characterized by modest disease risk heritability and multifaceted gene-environment interactions (Oksenberg and Barcellos, 2005).

Gene expression profiling, or microarray analysis, has enabled the measurement of thousands of genes in a single RNA sample. Recent advances in bioinformatics have led to both an increased number of genes analyzed as well as increased reproducibility between arrays. Advances in data analysis have also been successful in identifying complex interactions between transcriptomes (Baranzini, 2009; Geschwind and Konopka, 2009).

2.0. Large scale gene expression studies in MS

A majority of the large scale gene expression studies have been conducted using either peripheral blood monocytes (PBMCs) or brain tissue isolated from MS patients. Gene expression signatures of circulating leukocytes have been particularly popular due to ready availability from a large number of patients. PBMCs can be easily derived from a Ficoll gradient extraction of fresh blood samples. PBMC microarray studies are particularly useful to predict disease course and examine patient response to treatment, as the circulating leukocytes are one of the first indicators of external changes in the internal milieu. Several review articles have comprehensively addressed gene expression studies using PBMCs from MS patients (Comabella and Martin, 2007; Goertsches and Zettl, 2007; Habek et al., 2010). Comparative expression profiling of PBMCs isolated from MS patients and controls has been performed to detect markers specific to the disease. In one of these series of studies, Ramanathan et al. compared PBMCs from 15 control and 15 RR-MS patients and found several genes involved in activation of T- and B-cells (Ramanathan et al., 2001). Differences in expression of immune and cell cycle genes were reported in PBMCs of MS patients compared to control (Bomprezzi et al., 2003). Comparison of PBMCs isolated from 72 MS patients and 22 control patients identified significant decreases in DNA repair, replication, and chromatin remodeling (Satoh et al., 2005). PBMC profiling has also been used for comparing mRNA levels between patients in different stages of the disease to identify markers specific to the disease course. Achiron et al. identified 721 genes involved in activation of T-cells, epitope spreading and evasion of immune regulation in patients with acute MS relapse, when compared to MS patients in remission (Achiron et al., 2004). A separate study found dysregulated genes during the relapse and remission phases belonging to the biological categories of apoptosis and inflammation (Arthur et al., 2008). Using data from a separate PBMC profiling of 22 MS patients, it is was suggested that differential expression of apoptotic genes may influence the susceptibility of activated lymphocytes to expand and induce acute relapse and persistent inflammation in patients with relapsing-remitting multiple sclerosis (RRMS) (Achiron et al., 2007). PBMCs have also been used to identify differences between the responder and the non-responder MS patients towards interferon therapy. Trying to identify genes which could predict a favorable response to interferon therapy, IL-8 levels was identified as a useful predictor of the positive responder group (Sturzebecher et al., 2003). Action of interferon beta therapy is not purely anti-inflammatory (Wandinger et al., 2001) and within the first 4 weeks of IFN-β administration, a NF-κB-centered sub-network of genes was highly expressed (Hecker et al., 2010). Peripheral mononuclear blood cells from 25 RRMS patients were studied over a period of 2 yrs to detect the effect of IFN-beta administration. MS4A1 (CD20), a known target of B-cell depletion therapy, was significantly down-regulated after one month (Goertsches et al., 2010). Taken together these studies indicate that there is still a lot to be determined about the mechanism of IFN-beta action and the identification of IFN-beta regulated genes.

There are however several limitation to using PBMCs for large scale gene expression studies, the primary being that the gene expression profile does not represent neural components. Comabella and Martin comprehensively reviewed the “biological classification” of transcripts altered in MS brain and PBMCs (Comabella and Martin, 2007). Their results showed that there was a clear over-representation of mRNAs related to categories of “immune” in PBMC and “CNS related” in brain microarray studies. The gene expression profile generated from blood also may in part reflect influence of other factors not pertinent to the disease process itself (Goertsches and Zettl, 2007). There are also reports of significant influence of inter-individual and time variation upon blood derived gene expression profiles (Eady et al., 2005; Whitney et al., 2003). In a landmark study, Baranzini et al. studied mRNA and sequence variation in CD4(+) lymphocytes from three MS-discordant, monozygotic twin pairs (nature 2010). No reproducible differences were detected between co-twins for the expression of approximately 19,000 genes in CD4(+) T cells (Baranzini et al., 2010). Microarray studies using MS brains are often plagued by recovery of good quality RNA from a low number of samples and targeting the secondary phase of the disease. Nevertheless, studies using MS brains have provided several clues to the pathogenesis of the disease and we will focus on these data for the major portion of this review.

2.1. MS White matter microarray studies

Traditionally MS has been considered a white matter disease, hence it is not surprising that the majority of MS brain profiling studies have been concentrated on white matter. Based on histological analysis of demyelination and infiltration of immune cells, MS white matter can be subdivided into either normal appearing white matter (NAWM) or white matter lesions (WML). White matter areas without any signs of demyelination and abundant microglial activation could be referred as normal appearing white matter (NAWM). Based on immunohistological analysis of immune and myelin markers, white matter lesions can however be classified into active (characterized by abundant perivascular and parenchymal leukocyte infiltration throughout the lesion), chronic active (hypocellular center and hypercellular lesion rim) or chronic inactive (hypocellular) (Trapp et al., 1998; van, V and De Groot, 2000).

In one of the earlier reports of gene profiling using NAWM tissue from MS brains, Graumann et al compared 10 MS cases with 7 matched controls (Graumann et al., 2003). The study revealed up-regulation of several genes involved in endogenous neuroprotection. In a separate study conducted by the same group, increased mRNA levels of both pro- and anti-inflammatory genes were reported in NAWM tissue (Zeis et al., 2008). Increased levels of mRNAs related to anti-oxidative stress were reported in a single biopsy sample derived from the normal area adjacent to an MS lesion (Zeis et al., 2009). These studies suggest a possible balance between the neuroprotective and immune response associated with NAWM in MS brains (Kinter et al., 2008). Increased expression of interferon and TNF-alpha receptors was detected in an acute lesion from a single MS patient compared with NAWM (Whitney et al., 1999). In a subsequent study, several lesions from two MS cases and white matter from control cases with tissue from EAE mice were investigated (Whitney et al., 2001). The results showed 5-lipoxygenase as a candidate gene present in the MS disease state but not in control brains. In a comparison between chronic active and active plaques with control white matter, about 46% of genes were found to have similar directions of mRNA expression in both types of lesions (Tajouri et al., 2003). Comparison of gene profiles derived from active lesions and NAWM showed no difference in genes related to cellular immune response while there was a distinct difference in the expression profile of genes related to the humoral immune response (Lindberg et al., 2004). In a separate RT-PCR analysis, comparison between 8 MS lesions with 8 control brains also showed a complex pattern of cytokine expression (Baranzini et al., 2000). Comparative analysis of chronic active and inactive lesions confirmed the existence of significant differences in the transcriptional profiles of genes associated with inflammatory markers and apoptosis between these two lesion types (Mycko et al., 2003; Mycko et al., 2004). In a study using 2 controls and 4 MS patients, Lock et al. selected Fc receptors and GM-CSF (Granulocyte colony-stimulating factor) among several candidates for unique expression in chronic and active lesions and also showed that these genes could ameliorate EAE in mouse models (Lock et al., 2002). Gene expression studies conducted in MS brain tissue is summarized in Table 1.

Table 1.

Summary of microarray studies carried out in MS brains

Sample	Tissue	Main Findings	Ref
1 MS (PP)	WM	62 genes were differentially expressed in acute lesions compared with normal white matter. Major biological category of genes includes chemokines, cell adhesion and cell cycle.	(Whitney et al., 1999)
2 MS (PP, RR) 3 Ctl (Pooled)	WM	Genes involved in biosynthesis of the proinflammatory leukotrienes were increased. Major finding include validation of 5-lipoxygenase in MS and EAE tissues.	(Whitney et al., 2001)
4 MS (SP) 2 Ctl	WM	88 genes were consistently increased or decreased across all four MS samples. Granulocyte colony stimulating factor mRNA increased in acute, while immunoglobulin Fcγ-receptor was increased in chronic plaques.	(Lock et al., 2002)
5 MS (SP) Ctl (No. unknown)	WM	69 Genes were common between two plaque types. These were classified in biological pathways associated with cytoskeletal reorganization, cell motility, immune response.	(Tajouri et al., 2003)
4 MS (SP)	WM	Comparison between margin and center of acute and chronic WM lesions were carried out. Genes identified in the center and the margin of active lesions mainly belonged to inflammation while those in inactive lesions belonged to cell death.	(Mycko et al., 2003; Mycko et al., 2004)
6 MS (SP) 12 Ctl	WM	Comparison of active lesions and NAWM with control tissues show 123 genes in lesions, and 47 genes in NAWMMS to be differentially expressed. Genes related to immunoglobulin synthesis and neuroglial differentiation was increased in lesions.	(Lindberg et al., 2004)
10 MS (SP,PP) 7 Ctl	WM	The study revealed upregulation of a significant number of genes in NAWM involved in endogenous neuroprotection including HIF1αCREB, PI3K/Akt signaling pathway. Genes involved in inflammation were specifically regulated in NAWM of MS patients.	(Graumann et al., 2003; Zeis et al., 2008)
6 MS (SP, PP) 6 Ctl	GM	This study compared non-lesion motor cortex to control motor cortex. Genes related to mitochondrial function and inhibitory neurotransmission were decreased. Conversely genes belonging to the CNTF-mediated neuroprotection were increased	(Dutta et al., 2006; Dutta et al., 2007)

2.2. Grey matter microarray

Recent evidence supports major cortical involvement in the pathogenesis of MS (Trapp et al., 1999; Trapp and Nave, 2008). Cortical demyelination is a prominent feature of postmortem MS brains (Bo et al., 2003a; Bo et al., 2003b; Kidd et al., 1999; Kutzelnigg and Lassmann, 2005; Peterson et al., 2001). Demyelinated cortices are not evident macroscopically in postmortem brain slices because they do not change color like white matter lesions. Estimates of cortical lesion load have been limited to immunocytochemical analysis of postmortem brains and may equal or exceed white matter lesion load (Bo et al., 2003a; Kutzelnigg and Lassmann, 2005). Neuronal damage in motor and sensory cortex would negatively impact ambulation in MS patients and thus cortical pathology must contribute to neurological decline. It is imperative therefore to study cortical regions to understand the basis of neurological disability associated with MS. There is very little data on microarray studies performed in MS grey matter. In one of the first set of studies, tissue isolated from MS (patients were SP-MS and PP-MS) non-lesioned motor cortex was compared with control motor cortex. Using unbiased comparisons of 33,000 mRNA transcripts and employing statistical stringency including false discovery rate, 555 transcripts were significantly altered (Fig. 1A). Among them, 488 were decreased and 67 were increased in MS cortex (Dutta et al., 2006; Dutta et al., 2007). When grouped into ontology-based biological processes, altered genes showed decreases in two major gene families belonging to oxidative phosphorylation and synaptic transmission. Oxidative phosphorylation is catalyzed by five large multi-protein complexes that are encoded by both nuclear and mitochondrial DNA. Among the 119 nuclear-encoded mitochondrial electron transport chain genes on the arrays, 103 were called present in all samples and 26 were significantly (p<0.05) decreased in multiple sclerosis samples (Fig. 1B). The changes were validated using RT-PCR. To investigate whether mitochondrial transcripts were reduced in upper motor neurons, quantitative in situ hybridization was performed using mitochondrial complex I protein NDUFA6 with proteolipid protein mRNA served as control. When viewed by darkfield microscopy, mitochondrial mRNAs were highly enriched in upper motor neurons in both control (Fig. 1CI) and multiple sclerosis (Fig. 1CII) sections, while proteolipid protein mRNAs were enriched in glial cells (Fig. 1CIII, IV). In addition, mitochondria isolated from MS cortex had reduced function of respiratory chain complexes I and III (Fig. 1D). Due to the redistribution of Na⁺ channels and the resulting increased influx of sodium, ATP consumption is greatly increased in demyelinated axons (Peterson et al., 2005; Waxman, 2006). The mitochondria that reach chronically demyelinated axoplasm are likely to be compromised and have a reduced capacity for ATP production caused by decreased neuronal transcription of nuclear encoded mitochondrial genes as mentioned earlier (Dutta et al., 2006). Since reduced ATP production can protect cells from noxious stress and reduce apoptosis, reduction of mitochondrial gene expression may therefore be a part of a neuroprotective response resulting in demyelinated axonal segments being at risk for degeneration. Microarray analysis therefore supports previous studies (Smith, 2007; Waxman, 2005; Waxman, 2006) that implicate reduced ATP production as a mechanism of degeneration of chronically demylinated axons in multiple sclerosis.

Fig 1. Microarray data and downstream validation techniques performed in MS motor cortex.

Hierarchical clustering of significantly altered transcripts grouped control (C1–C6) and multiple sclerosis (MS1-MS6) motor cortex samples separately supporting disease-related gene expression patterns. Among the decreased transcripts in MS motor cortex, twenty-six belonged to the electron transport chain (Panel B). Mitochondrial complex I (NDUFA6) mRNA was decreased in neurons in multiple sclerosis motor cortex (Panel CII) compared to control (Panel CI) whereas proteolipid protein mRNA densities were similar control (Panel CIII) and multiple sclerosis (Panel CIV) cerebral cortex. Activity of electron transport complexes I and III are decreased in mitochondrial enriched fractions from motor cortex of multiple sclerosis patients (Panel D). (Reprinted from Dutta et al., 2006). Majority of the increased transcripts in MS motor cortex belonged to CNTF signalling pathway and are schematically represented in Panel E. Confocal image of MS cortex immunostained with antibodies specific to CNTF (green) and GFAP (red) shows CNTF positive cortical neurons. (Reprinted from Dutta et al., 2007). Scale bar CI-IV-25um F=20 um.

In the same microarray dataset, there were decreased levels of pre- and post-synaptic components of inhibitory neurotransmitters. Reduced inhibitory innervation leads to up-regulation of neuroprotective pathways (Hardingham et al., 2002) (Jalabi et al, 2010, under review) and therefore increased transcripts were searched for possible neuroprotective pathways. Using gene classification methods, we found 9 out of the 67 increased transcripts to be members of the ciliary neurotrophic factor (CNTF) family (Fig. 1E). CNTF is an established neurotrophic factor, which enhances neuronal survival during development and in disease. Translational and transcriptional products of CNTF-related genes were quantified and localized in control and MS cortices (Dutta et al., 2007). CNTF, the tripartite CNTF receptor complex and downstream CNTF signaling molecules, including the anti-apoptotic molecule Bcl2, were increased in neurons in MS cortex. Interestingly, CNTF was detected in cortical neurons in MS brains but not in glial cells (Fig. 1F). An active and functionally significant role for CNTF in MS patients is supported by the report that MS patients with CNTF-null mutations have an earlier disease onset and more aggressive disease course (Giess et al., 2002). Increased expressions of neurotrophic genes represent part of the endogenous defense mechanisms mounted by the MS brain to maintain neuronal integrity and combat progressive neurological decline. The microarray results described here therefore provides a comprehensive overview of the different validation and downstream applications that have been applied to microarray data related to multiple sclerosis brain tissue.

3. Conclusion and Future Challenges

In conclusion, microarray-based technology has successfully been applied to MS brain tissue and provided several important clues in identifying several candidates/pathways that were not revealed by conventional techniques in MS research. The genetic complexity of MS provides an ideal platform for the application of gene expression profiling studies. Microarray based gene expression profiling studies have often been criticized as mere “fishing” experiments rather than as hypothesis-driven research. This has led to widespread skepticism in the scientific community as a whole towards this technology. This is in part supported by the large number of studies, including MS research, that are comprised of large gene lists, poor sample collection, lack of statistical testing, lack of downstream validation and lack of reproducibility. Instead of applying the data to think ‘out of the box’, researchers have tried to fit their data into an existing ‘box’. The main question is how are we going to change this? One should keep in mind that this is a technology that has limitations and researchers should spend time validating the results by independent methods. These include RT-PCR, in situ hybridization, immunoblots and immunohistochemistry on independent samples. Due to the heterogeneity of MS, it is critical that the researcher follow extensive quality control guidelines for sample collection for microarray studies using MS tissue. Authors should provide information about how the tissue was collected, how specific regions were identified, etc. This will improve the reproducibility of the data and ultimately the reliability of the technology.

As we continue to develop genetic technologies, the gene expression profiling field will also continue to grow with refined probes, increased genome coverage and greater reproducibility. There is a lot of potential for application of large scale gene expression in MS research, especially using brain tissue. Gene expression studies will help understand cortical pathology in MS. This is one of the areas of MS research that has been underdeveloped to date. Additional technologies to incorporate pathology, neuroimaging and large scale sequencing studies to complement gene expression analysis should also be utilized. Future research into regulatory factors like transcription factors and microRNAs that control gene expression in MS should also be undertaken.