Abstract

Chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) is a multisystem disease, the pathogenesis of which remains undetermined. We set out to determine the precise abnormalities of gene expression in the blood of patients with CFS/ME. We analyzed gene expression in peripheral blood from 25 patients with CFS/ME diagnosed according to the Centers for Disease Control and Prevention diagnostic criteria and 50 healthy blood donors, using a microarray with a cutoff fold difference of expression of ⩾2.5. Genes showing differential expression were further analyzed in 55 patients with CFS/ME and 75 healthy blood donors, using quantitative polymerase chain reaction. Differential expression was confirmed for 88 genes; 85 were upregulated, and 3 were downregulated. Highly represented functions were hematological disease and function, immunological disease and function, cancer, cell death, immune response, and infection. Clustering of quantitative polymerase chain reaction data from patients with CFS/ME revealed 7 subtypes with distinct differences in Medical Outcomes Survey Short Form-36 scores, clinical phenotypes, and severity

Chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) is characterized by severe and debilitating fatigue, abnormal sleep behavior, impaired memory and concentration, and musculoskeletal pain [1]. In developed nations, the population prevalence is estimated to be 0.5% [2, 3]. Studies have identified various features relevant to the pathogenesis of CFS/ME, such as viral infection; abnormal immune function; exposure to toxins, chemicals, and pesticides; stress; hypotension; abnormal lymphocyte levels; and neuroendocrine dysfunction. However, the precise underlying mechanisms of disease and the means by which they interrelate in patients with CFS/ME remain to be clarified [4, 5]

Various studies have analyzed gene expression in the peripheral blood of patients with CFS/ME, and in each study genes associated with immunity and defense were prominent [6–17]. Unfortunately, in several of these studies quantitative polymerase chain reaction (PCR) confirmation was not performed, so the results may be unreliable [6, 8–10, 14–17]. The genes identified in the studies that used PCR confirmation suggest that CFS/ME has a complex pathogenesis [7, 11–13]. However, because none of these studies were comprehensive in terms of the number of human genes examined, our knowledge of the precise metabolic pathways involved in CFS/ME remains incomplete

The goal of this study was to determine the precise abnormalities of gene expression in the blood of patients with CFS/ME. We enrolled patients with a rigorously defined CFS/ME phenotype and compared them to healthy blood donors by means of a microarray that represented the entire human genome and by quantitative PCR confirmation. Using this approach, we identified differential expression of 88 human genes in patients with CFS/ME. Among these genes, highly represented functions were hematological disease and function, immunological disease and function, cancer, cell death, immune response, and virus infection. Clustering of gene expression data revealed 7 CFS/ME subtypes with distinct clinical phenotypes and associated disease severity

Subjects, Materials, and Methods

Subject enrollment, clinical characterization, and blood samplingTwenty-five patients with CFS/ME from the Dorset CFS Service in southeast England were enrolled for the microarray study. Patients with CFS/ME whose blood was used for subsequent PCR studies comprised those in the microarray study along with an additional 30 patients from clinics in 3 United Kingdom cities (Dorset, Bristol, and London; 1 patient from Leicester was under the care of a clinic in London) and New York, New York. CFS/ME was diagnosed on the basis of Centers for Disease Control and Prevention (CDC) criteria [1]. Patients with psychiatric disease were excluded on the basis of findings of the Minnesota International Neuropsychiatric Interview, thus ensuring that no patients had major psychiatric disease or were abusing alcohol or drugs. In addition, patients who had smoked tobacco during the previous 12-month period and/or had taken antibiotics, steroids, or antidepressants during the previous 3-month period were excluded from the study

Healthy blood donors were used as a comparison group for both the microarray and real-time PCR studies. For the microarray study, 50 healthy blood donors were enrolled from the Dorset National Blood Service (NBS) and matched to the CFS/ME group at a ratio of 2:1 on the basis of age, sex, and geographical location. For subsequent PCR studies, the comparison group comprised the donors involved in the microarray study plus an additional 25 donors, all enrolled from the NBS. Restrictions imposed by the NBS on persons who are allowed to donate blood are outlined elsewhere [13]. Blood donors were excluded from the study if they had smoked tobacco during the previous 12-month period and/or had taken antibiotics, steroids, or antidepressants during the previous 3-month period

In accordance with the recommendations of the International Chronic Fatigue Syndrome Study Group [18], the following questionnaires were completed for all enrolled subjects (patients and control donors): the Chalder Fatigue Scale [19], to assess the severity of physical and mental fatigue; the Medical Outcomes Survey Short Form-36 (SF-36), to determine the level of disability; the Somatic and Psychological Health Report, to characterize accompanying symptoms; the Pittsburgh Sleep Questionnaire, to assess abnormal sleep behaviors; and the McGill Pain Questionnaire, to assess the type and severity of pain. For patients with CFS/ME, neurocognitive testing was performed using the Spatial Span (SSP) and Verbal Recognition Memory (VRM) modules of Cantab software (Cambridge Cognition). In another study, the SSP test yielded abnormal findings for patients with CFS/ME [20], and we found similar abnormalities in patients with CFS/ME enrolled in this study

Patients and control subjects gave informed written consent, in accordance with guidance of the Wandsworth Research Ethics Committee (approval number 05/Q0803/137). For patients in New York, approval of the local institutional review board was obtained. Human experimentation guidelines of the US Department of Health and Human Services were followed in this study

Fifteen-milliliter blood samples from patients with CFS/ME and healthy blood donors (as part of routine blood donation) were collected into PAXgene tubes (PreAnalytix), and total RNA was extracted using the PAXgene blood RNA kit (PreAnalytix), according to the instructions of the manufacturer. RNA quality and amount were confirmed by microspectrophotometry (Nanodrop). Total RNA samples used in this study had an absorbance ratio (A260/280) of 1.9–2.0

Microarray analysisRNA specimens were shipped as ethanol precipitates to the Penn Microarray Facility (Philadelphia, PA), where mass and qualitative assays were repeated, using the Nanodrop spectrophotometer and Agilent Bioanalyzer, respectively. Bioanalyzer traces indicated that intact ribosomal bands as well as various background and partial degradation bands typical of blood RNA were present in all samples. All microarray analyses were conducted with the GeneChip human genome U133+2 microarray (Affymetrix) and the One-Cycle target labeling and control reagents kit (Affymetrix), in accordance with the manufacturer’s recommendations. The average cRNA yield was 48 μg for samples from the control group and 51 μg for samples from patients with CFS/ME. Samples from 10 patients with CFS/ME were randomly selected as technical replicates for hybridization to a second GeneChip, to assess concordance. Microarray Suite 5.0 (Affymetrix) was used to quantitate expression levels for targeted genes; default values provided by the manufacturer were applied to all analysis parameters. A weighted mean value of probe fluorescence (corrected for nonspecific signals by subtracting the mismatch probe value) was calculated using the Tukey 1-step biweight estimate. This signal value, a relative measure of the expression level, was computed for each assayed gene. Global scaling was performed to allow comparison of gene signals across multiple microarrays

Microarray data were normalized in Excel 2003 (Microsoft) and imported into GeneSpring 7.3 (Agilent Technologies) within a genome consisting of the entirety of the probes on the microarray minus the probes that were specific for >1 gene (i.e., those with the suffix “_s_at” or “_x_at”); this effectively excluded probes whose specificity could not be assigned with certainty to any one gene and reduced the genome to 39,174 probes. Data were analyzed by use of a class comparison experiment to obtain a list of gene probes that showed differences in expression between test and control groups with a fold difference cutoff of 2.5 and a P value of ⩽.05 (an arbitrary level designed to select for 5% of the probes whose values showed the greatest differences between the CFS/ME and control groups). Probe values for each of these genes were reviewed, and a gene was included only if the majority of probe values were in agreement with the mean fold difference of the probe that originally flagged the gene

Quantitative PCRQuantitative PCR (Applied Biosystems) with TaqMan primers and probes was used to confirm the significance of genes identified in array experiments by the comparative method, using custom 384-well low-density arrays (LDAs) and the ABI Prism 7900HT instrument (Applied Biosystems), with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the endogenous control gene. Experiments were performed in triplicate, using a protocol described elsewhere [13]. Data were displayed using SDS software (version 2.2 [ABI]), and discordant data between replicates were omitted. Results for each LDA were calculated and loaded into SDS software (version 2.2, Enterprise Edition), to facilitate analysis of data involving up to 80–100 LDAs per experiment

The threshold cycle (Ct) for the test gene in each sample was compared to that of a calibrator sample to calculate a ΔCt value. ΔCt values were then normalized to the Ct value for GAPDH in respective samples to give ΔΔCt values. Relative quantities of each mRNA of interest (RQ; defined as 2−ΔΔCt) were then calculated. Samples showing a difference of ⩾100 between minimum and maximum RQs (which is indicative of poor replicate concordance) were excluded. The t test was used to compare RQs for the patients with CFS/ME with RQs for the control subjects. Genes with mean RQs that differed significantly (defined as a P value of ⩽.05) between the groups in the same direction as in the microarray were included in our CFS/ME-associated gene signature

Transcription factor binding sitesPromoter sequences for each of the 83 genes identified after real-time PCR analysis were extracted from the Mammalian Promoter Database at Cold Spring Harbor Laboratory (Cold Spring Harbor, NY; available at: http://rulai.cshl.edu/cgi-bin/CSHLmpd2/hspd.pl) and analyzed for overrepresentation of transcription factor binding sites, using tools from the Institute of Bioinformatics (Beijing, China; available at: http://www.bioinfo.tsinghua.edu.cn/∼zhengjsh/OTFBS); this analysis is described elsewhere [21]. Human transcription factors identified by this approach were tested in the same samples, using available real-time PCR assays (ABI 7500 Fast PCR System [Applied Biosystems]) performed in accordance with the method described above but in a 96-well format

Analysis of gene function and interactionMean RQs from TaqMan real-time PCR testing of the 88 CFS/ME-associated genes were analyzed with Ingenuity Pathways Analysis (IPA) (Ingenuity) to link CFS/ME-associated genes into networks based on recognized interactions. IPA was also performed to discern the top associated diseases and disorders, molecular and cellular functions, associated physiological system development and function, and canonical pathways on the basis of over-representation analysis, in which, for each pathway, the fraction of CFS/ME-associated genes involved in that pathway was compared to the fraction of human genes involved in that pathway. For each pathway, the probability of involvement of the respective number of CFS/ME-associated genes was expressed as a P value; values of ⩽.05 were taken to be significant

Clustering of quantitative PCR–generated gene values for patients with CFS/MERQs for all 88 CFS/ME-associated genes were normalized and clustered by use of Genesis software [22]. For each of the 7 CFS/ME subtypes identified by means of this approach, mean RQs were calculated for each gene and used to generate fold differences (defined as the value for patients with CFS/ME divided by the value for healthy control subjects) for each gene in each CFS/ME subtype. Mean fold differences for each gene in each CFS/ME subtype were then clustered with and without normalization/median centering by use of Cluster software, version 2.11, and visualized by use of Treeview software, version 1.60 [23]

Results

Subjects and clinical characterizationFifty-five patients with CFS/ME who fulfilled CDC diagnostic criteria and 82 healthy blood donors were enrolled in the study; 25 patients and 50 donors were evaluated in the microarray study, and 55 patients and 75 healthy donors were evaluated in the quantitative PCR study. Patient and clinical details are summarized in table 1. This study included several patients with CFS/ME whose disease was severe and necessitated bed rest for much of the day and patients who were able to attend an outpatient clinic (table 1)

Table 1

Demographic and clinical characteristics and results of diagnostic tests for patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) and healthy blood donors involved in microarray and real-time polymerase chain reaction (PCR) studies

Table 1

Demographic and clinical characteristics and results of diagnostic tests for patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) and healthy blood donors involved in microarray and real-time polymerase chain reaction (PCR) studies

Microarray analysisAlthough overall RNA quality was similar between the 2 groups, as indicated by Bioanalyzer traces, cRNA amplification yields, and GAPDH integrity, the microarrays from patients with CFS/ME showed half as many genes that met the Microarray Suite 5.0 “Present” detection threshold, but the average level of expression of these genes was 1.9 times the average level in control samples. Technical microarray replicates showed a concordance correlation coefficient of 0.978. GeneSpring analysis of microarray data identified 1789 probes with a differential expression in patients with CFS/ME that was statistically significantly different from that in healthy controls at a fold difference cutoff of ⩾2.5. However, when all probe values for each of these genes were reviewed, for only 182 genes were the majority of probe values in agreement with the mean fold difference of the probe that originally flagged that gene

Quantitative PCRStatistically significant differential expression with the same profile as that in gene arrays was confirmed for 82 of 182 genes tested, with upregulation of 79 genes and downregulation of 3 genes (PIK3R1, IL7R and HIF1A) (table 2 and figure 1)

Table 2

Characteristics of differentially expressed genes in chronic fatigue syndrome/myalgic encephalomyelitis

Table 2

Characteristics of differentially expressed genes in chronic fatigue syndrome/myalgic encephalomyelitis

Figure 1

Mean relative quantity of mRNA transcripts (RQ) in patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME; hatched bars) and healthy control subjects (black bars) for 88 genes that were differentially expressed in patients with CFS/ME. Error bars indicate variance from the mean in each case. The chart is truncated at an RQ of 34, to enhance the clarity of data for the greatest number of genes

Figure 1

Mean relative quantity of mRNA transcripts (RQ) in patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME; hatched bars) and healthy control subjects (black bars) for 88 genes that were differentially expressed in patients with CFS/ME. Error bars indicate variance from the mean in each case. The chart is truncated at an RQ of 34, to enhance the clarity of data for the greatest number of genes

Transcription factorsAnalysis of the promoter sequences of the 82 human genes specified above revealed over-representation of binding sites for the following transcription factor genes: REPIN1, SP1, ETS1, GABPA, GTF3A, EGR1, EGR2, EGR3, NFKB1, NHLH1, EGR4, REST and the Epstein-Barr virus (EBV) R transactivator gene BRLF1. Of these 13 transcription factor genes, results of real-time PCR assays were available for all except BRLF1; NFKB1 had already been tested because it was one of the original 82 genes. Real-time TaqMan PCR was used to test the remaining 11 genes in 25 patients with CFS/ME and 25 healthy controls, and significant upregulation in patients with CFS/ME was confirmed for 7 genes (REPIN1, ETS1, GABPA, NFKB1, EGR1, EGR3 and NHLH1) in patients with CFS/ME (table 2 and figure 1), suggesting that upregulation of these genes may be key to the overall gene signature observed in the CFS/ME group. This is supported by the fact that the fold differences for transcription factors were higher than those for most of the other genes. The transcription factor genes EGR2, GTF3A and SP1 were also upregulated in patients with CFS/ME; however, P values ranged from .18 to .09. Assays for REST and EGR4 did not provide usable data for the cDNA concentrations used

Analysis of gene function and interactionEighty-eight CFS/ME-associated genes were analyzed using IPA. The following 5 networks were revealed (table 3), all of which could be linked together in a single network (data not shown): hematological disease, cell development, and hematological system development and function (figure 2A); cell morphology, gene expression, and cellular assembly and organization (figure 2B); protein synthesis, cell cycle, and gene expression (figure 2C); gene expression, cellular development, and hematological system development and function (figure 2D); and cell cycle, protein degradation, and protein synthesis (figure 2E)

Table 3

Top gene networks associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Table 3

Top gene networks associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Figure 2

Five large and distinct gene networks (A–E) identified using Ingenuity Pathways Analysis (Ingenuity); functional themes of each are shown in tables 3–5. Genes are colored according to levels of expression determined in the present study: red denotes upregulation, and blue denotes downregulation. Color intensity reflects the magnitude of the fold difference between patients with CFS/ME and healthy subjects

Figure 2

Five large and distinct gene networks (A–E) identified using Ingenuity Pathways Analysis (Ingenuity); functional themes of each are shown in tables 3–5. Genes are colored according to levels of expression determined in the present study: red denotes upregulation, and blue denotes downregulation. Color intensity reflects the magnitude of the fold difference between patients with CFS/ME and healthy subjects

Diseases and disorders, molecular and cellular functions, and physiological systems for these genes are shown in table 4. Prominent functional features involved immunity, inflammation, apoptosis, neurological disease and function, infectious disease, and cancer. Many canonical pathways were implicated by these genes (table 5); the most important themes were immunity, inflammation, apoptosis, and neurological disease and function

Table 4

Top gene functions associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Table 4

Top gene functions associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Table 5

Top canonical pathways associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Table 5

Top canonical pathways associated with the chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) signature, according to Ingenuity Pathways Analysis

Genomic CFS/ME subtypesClustering of RQs from patients with CFS/ME identified 7 subtypes consisting of 2, 5, 2, 19, 7, 14, and 3 patients. Clustering of mean fold differences (calculated as the mean subtype RQ of patients divided by the mean RQ for control subjects) for the 88 CFS/ME-associated genes revealed a distinct profile of gene expression in each subtype and clustering of genes with similar profiles of expression in the different subtypes (figure 3). Clustering of normalized/median centered fold differences for each gene in each subtype is shown in figure 3A which highlights the distinct nature of the gene signature for each CFS/ME subtype. Clustering of logarithm (base 2) fold differences are shown in figure 3B which emphasizes the predominance of upregulation for most genes in all CFS/ME subtypes

Figure 3

A Clustering of normalized/median centered fold differences, calculated as the mean relative quantity of mRNA transcripts (RQ) in patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) for each subtype divided by the mean RQ in healthy control subjects, for 88 genes associated with CFS/ME. B Clustering of logarithm (base 2) fold differences, calculated as the mean RQ in patients with CFS/ME for each subtype divided by the mean RQ in healthy controls, for 88 genes associated with CFS/ME

Figure 3

A Clustering of normalized/median centered fold differences, calculated as the mean relative quantity of mRNA transcripts (RQ) in patients with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) for each subtype divided by the mean RQ in healthy control subjects, for 88 genes associated with CFS/ME. B Clustering of logarithm (base 2) fold differences, calculated as the mean RQ in patients with CFS/ME for each subtype divided by the mean RQ in healthy controls, for 88 genes associated with CFS/ME

Analysis of mean age and sex ratios for each subtype revealed that subtypes 3, 5, and 7 occurred only in females, subtype 2 was predominantly male, and the remaining subtypes occurred in both males and females; age differences were less clearly demarcated. The clinical phenotype was distinct between subtypes; subtypes 1 and 7 were the most severe, followed sequentially by subtypes 2, 4, 5, 6, and 3. Analysis of variance testing revealed significant differences in the SF-36 total score (P=.016), social functioning (P=.03), and emotional role (P=.003) between the groups, whereas the difference between groups approached statistical significance for general health (P=.08) and mental health (P=.08)

After adjustment for multiple comparisons, significant associations were found between specific subtypes and clinical phenotypes. Subtype 7 had the most pain, the lowest SF-36 scores (along with subtype 1), and the most-severe individual symptoms, including swollen glands, sore throat, and headaches; subtype 1 had the worst cognition and mental health and poor sleep, despite being associated with the least pain; subtype 4 had moderate neurocognitive function and cognitive defects, combined with moderate levels of bodily pain and sleep problems; subtype 5 had the best mental health but poor neurocognitive function, gastrointestinal complaints, and the most marked muscle weakness and postexertional malaise; and subtype 2 had marked postexertional malaise, muscle pain, and joint pain but poor mental health (figures 4A and 4B)

Figure 4

A Medical Outcomes Survey Short Form-36 (SF-36) domain and total scores for each chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) subtype: physical function (PF), physical role (RP), bodily pain (BP), general health (GH), vitality (VIT), social functioning (SF), emotional role (RE), mental health (MH), and total score (Total). B Scores indicating the occurrence and severity of the following 11 clinical symptoms and 2 neurocognitive functions for each CFS/ME subtype: headache (HA), sore throat (ST), swollen glands (GLA), cognitive defect (COG), muscle pain (MP), joint pain (JP), muscle weakness (MW), postexertional malaise (PEM), sleep problems (SLE), fainting/dizziness (F/D), gastrointestinal complaints (GI), numbness/tingling (N/T), spatial span (SSP), and verbal recognition memory (VRM). C Histogram showing the number of patients with CFS/ME of each subtype, by geographical location

Figure 4

A Medical Outcomes Survey Short Form-36 (SF-36) domain and total scores for each chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) subtype: physical function (PF), physical role (RP), bodily pain (BP), general health (GH), vitality (VIT), social functioning (SF), emotional role (RE), mental health (MH), and total score (Total). B Scores indicating the occurrence and severity of the following 11 clinical symptoms and 2 neurocognitive functions for each CFS/ME subtype: headache (HA), sore throat (ST), swollen glands (GLA), cognitive defect (COG), muscle pain (MP), joint pain (JP), muscle weakness (MW), postexertional malaise (PEM), sleep problems (SLE), fainting/dizziness (F/D), gastrointestinal complaints (GI), numbness/tingling (N/T), spatial span (SSP), and verbal recognition memory (VRM). C Histogram showing the number of patients with CFS/ME of each subtype, by geographical location

Subtypes 4 and 6 were predominant in Dorset, subtype 4 was predominant in London and New York, and subtype 5 was predominant in Bristol (figure 4C)

Discussion

In this article, we document the differential expression of 88 human genes in CFS/ME. We have confirmed involvement of all 16 genes reported in our pilot study [13] in the same direction as previously reported, except IL10RA, which was downregulated in the previous study [13] but upregulated in the present study. This discrepancy remains unexplained, but because IL10RA is a critical gene required for T cell activation and since IL10RA levels correlate with measures of general health, we believe it to be particularly important

The functions of these genes present a complex picture with links to several diseases and pathways (tables 3–5). Prominent themes that are well recognized in patients with CFS/ME are immunity, inflammation, and infection [5, 13]; cell death [13, 24]; cancer [25, 26]; and neurological disease [13, 27–29]

One of the most important viral triggers of CFS/ME is EBV, and this virus very likely plays an important role in perpetuation of disease, because it is reactivated by stress [30]. Within the gene signature identified in this study, the following 12 human genes that we found to be upregulated in patients with CFS/ME have been shown elsewhere to be upregulated, either directly or indirectly, by EBV infection: NFKB1, EGR1, ETS1, GABPA, CREBBP, CXCR4 and EBI2 [31]; HIF1A; JAK1; IL6R; IL7R; and PIK3R1. A particularly interesting gene is EBI2 which was upregulated in 55% of patients with CFS/ME, one of whom was a 26-year-old woman with CFS/ME triggered by laboratory-documented EBV infection 10 years earlier. The EBV genes BRLF1 and BZLF1 mediate the switch from latent to lytic phases of EBV infection, and during this process they transactivate many human genes. It is interesting that BRLF1 was identified as being overrepresented in the transcription factor analysis and that IgG specific to the Zebra protein (the BZLF1 gene product) has been reported previously in patients with CFS/ME [32]

Enteroviruses are another very important viral trigger of CFS/ME [33]. Upregulation of EIF4G1 transcript variant 5, which was found in this study and elsewhere [8, 13], is targeted during infection by various viruses, including enteroviruses, to subvert cellular machinery for the production of viral proteins

The theme of neurological disease and function was prominent among the 88 CFS/ME-associated genes, almost all of which are expressed in multiple areas of the human brain (data not shown). Involvement of specific genes highlights the importance of neuregulin signaling, neurotrophin/TRK signaling, axonal guidance signaling, dopamine receptor signaling, and Huntington disease signaling (table 5). NTE was upregulated in blood of patients with CFS/ME, both in this study and our pilot study [13]. NTE is the primary site of action of organophosphate (OP) compounds, such as sarin, which cause axonal degeneration and paralysis due to inactivation of its serine esterase activity [34]. In the nervous system of adult chickens, OP-modified NTE initiates neurodegeneration. NTE probably regulates neuron-glial interactions during development and possibly also during adult life [35]. Exposure to OPs may trigger CFS/ME [36] and Gulf War illness [37]. Furthermore, the heterogeneity of NTE levels seen in patients with CFS/ME is consistent with a role for OPs in only a subset of patients with CFS/ME. In this study, 3 of 28 patients with CFS/ME for whom NTE data passed quality control analysis had high NTE levels, compared with uniformly low levels in healthy blood donors. This is in contrast to data from our pilot study, in which 10 of 17 patients with CFS/ME showed high NTE levels [13]. The discrepancy between the studies is interesting, given that in our pilot study all patients with individual PCR results for NTE were from Dorset, whereas in the present study 11 patients were from Dorset; 1 was from Leicester, United Kingdom; 5 were from Bristol; 5 were from London; and 6 were from New York

Another interesting gene that affects neurological function is EIF2B4 mutations within which are associated with vanishing white matter disease [38]. This is interesting in view of the white matter hyperintensity lesions that have been documented in patients with CFS/ME [27–29]

We have identified involvement of several genes of the interleukin-6 (IL-6) signaling pathway, consistent with the findings of several previous studies [39, 40]. Although we found upregulation of the genes encoding IL-6R and IL-6ST, the 2 IL-6 receptors, we did not find evidence for upregulation of the gene encoding IL-6 itself, which may explain the inconsistent findings by different groups in this area

It is intriguing that it is possible to identify CFS/ME subtypes on the basis of expression values for these 88 genes and even more so that these subtypes have distinct clinical phenotypes, with such marked differences in the occurrence of particular symptoms and their severity. It has been recognized for some time that subtypes of CFS/ME exist, and it has been thought that these subtypes may, at least in part, reflect particular etiological factors [41]. A symptom-based approach has had some success in identifying musculoskeletal, inflammatory, and neurological subtypes [42]; however, these groups had only minor differences between them in overall functional severity, in contrast to the subtypes in the present study. A more detailed analysis of clinical and molecular features of these subtypes will be presented in a separate article. Further work is urgently required to validate and develop these findings

Acknowledgments

We thank Anne Faulkner and the CFS Research Foundation (Hertfordshire, United Kingdom), for generous funding of the project; Sir Joseph Hotung, for funding the salaries of J.R.K. and B.B; Dr. Frank Boulton, Ms. Julie Williams, Mr. Peter Rogers, Ms. Diana Carr, and the NBS teams of East Dorset, for their help in enrolling and collecting samples from healthy blood donors; Dr. Jon Sherlock (Applied Biosystems), for advice on TaqMan PCR; and all the patients with CFS/ME and blood donors for their participation. We also acknowledge the kind leadership of the late David A. J. Tyrrell [43], which was key in the early stages of this project

References

1
Fukuda
K
Straus
SE
Hickie
I
Sharpe
MC
Dobbins
JG
Komaroff A. The chronic fatigue syndrome: a comprehensive approach to its definition and
study
International Chronic Fatigue Syndrome Study Grou
Ann Intern Med
 , 
1994
, vol. 
121
 (pg. 
953
-
9
)
2
Department of
Health
A report of the CFS/ME Working Group
January
 , 
2002
3
Papanicolaou
DA
Amsterdam
JD
Levine
S
, et al.  . 
Neuroendocrine aspects of chronic fatigue syndrom
Neuroimmunomodulation
 , 
2004
, vol. 
11
 (pg. 
65
-
74
)
4
Komaroff
AL
Buchwald
D
Chronic fatigue syndrome: an updat
Annu Rev Med
 , 
1998
, vol. 
49
 (pg. 
1
-
13
)
5
Chronic fatigue syndrom
J Clin Virol
 , 
2006
, vol. 
37
 (pg. 
139
-
50
)
6
Vernon
SD
Unger
ER
Dimulescu
IM
Rajeevan
M
Reeves
WC
Utility of the blood for gene expression profiling and biomarker discovery in chronic fatigue syndrom
Dis Markers
 , 
2002
, vol. 
18
 (pg. 
193
-
9
)
7
Powell
R
Ren
J
Lewith
G
Barclay
W
Holgate
ST
Almond
J
Identification of novel expressed sequences, up-regulated in the leucocytes of chronic fatigue syndrome patient
Clin Exp Allergy
 , 
2003
, vol. 
33
 (pg. 
1450
-
6
)
8
Whistler
T
Unger
ER
Nisenbaum
R
Vernon
SD
Integration of gene expression, clinical, and epidemiologic data to characterize chronic fatigue syndrom
J Transl Med
 , 
2003
, vol. 
1
 (pg. 
10
-
8
)
9
Whistler
T
Jones
JF
Unger
ER
Vernon
SD
Exercise responsive genes measured in peripheral blood of women with chronic fatigue syndrome and matched control subject
BMC Physiol
 , 
2005
, vol. 
5
 pg. 
5
 
10
Whistler
T
Taylor
R
Craddock
RC
Broderick
G
Klimas
N
Unger
ER
Gene expression correlates of unexplained fatigu
Pharmacogenomics
 , 
2006
, vol. 
7
 (pg. 
395
-
405
)
11
Grans
H
Nilsson
M
Dahlman-Wright
K
Evengard
B
Reduced levels of oestrogen receptor beta mRNA in Swedish patients with chronic fatigue syndrom
J Clin Pathol
 , 
2007
, vol. 
60
 (pg. 
195
-
8
)
12
Grans
H
Nilsson
P
Evengard
B
Gene expression profiling in the chronic fatigue syndrom
J Intern Med
 , 
2005
, vol. 
258
 (pg. 
388
-
90
)
13
Kaushik
N
Fear
D
Richards
SC
, et al.  . 
Gene expression in peripheral blood mononuclear cells from patients with chronic fatigue syndrom
J Clin Pathol
 , 
2005
, vol. 
58
 (pg. 
826
-
32
)
14
Carmel
L
Efroni
S
White
PD
Aslakson
E
Vollmer-Conna
U
Rajeevan
MS
Gene expression profile of empirically delineated classes of unexplained chronic fatigu
Pharmacogenomics
 , 
2006
, vol. 
7
 (pg. 
375
-
86
)
15
Broderick
G
Craddock
RC
Whistler
T
Taylor
R
Klimas
N
Unger
ER
Identifying illness parameters in fatiguing syndromes using classical projection method
Pharmacogenomics
 , 
2006
, vol. 
7
 (pg. 
407
-
19
)
16
Fang
H
Xie
Q
Boneva
R
Fostel
J
Perkins
R
Tong
W
Gene expression profile exploration of a large dataset on chronic fatigue syndrom
Pharmacogenomics
 , 
2006
, vol. 
7
 (pg. 
429
-
40
)
17
Fostel
J
Boneva
R
Lloyd
A
Exploration of the gene expression correlates of chronic unexplained fatigue using factor analysi
Pharmacogenomics
 , 
2006
, vol. 
7
 (pg. 
441
-
54
)
18
Reeves
WC
Lloyd
A
Vernon
SD
et al. Identification of ambiguities in the 1994 chronic fatigue syndrome research case definition and recommendations for
resolution
International Chronic Fatigue Syndrome Study Grou
BMC Health Serv Res
 , 
2003
, vol. 
3
 pg. 
25
 
19
Chalder
T
Berelowitz
G
Pawlikowska
T
, et al.  . 
Development of a fatigue scal
J Psychosom Res
 , 
1993
, vol. 
37
 (pg. 
147
-
53
)
20
Joyce
E
Blumenthal
S
Wessely
S
Memory, attention, and executive function in chronic fatigue syndrom
J Neurol Neurosurg Psychiatry
 , 
1996
, vol. 
60
 (pg. 
495
-
503
)
21
Zheng
Z
Wu
J
Sun
Z
An approach to identify over-represented cis-elements in related sequence
Nucleic Acids Res
 , 
2003
, vol. 
31
 (pg. 
1995
-
2005
)
22
Sturn
A
Quackenbush
J
Trajanoski
Z
Genesis: cluster analysis of microarray dat
Bioinformatics
 , 
2002
, vol. 
18
 (pg. 
207
-
8
)
23
Eisen
MB
Spellman
PT
Brown
PO
Botstein
D
Cluster analysis and display of genome-wide expression pattern
Proc Natl Acad Sci U S A
 , 
1998
, vol. 
95
 (pg. 
14863
-
8
)
24
Kennedy
G
Spence
V
Underwood
C
Belch
JJ
Increased neutrophil apoptosis in chronic fatigue syndrom
J Clin Pathol
 , 
2004
, vol. 
57
 (pg. 
891
-
3
)
25
Levine
PH
Fears
TR
Cummings
P
Hoover
RN
Cancer and a fatiguing illness in Northern Nevada—a causal hypothesi
Ann Epidemiol
 , 
1998
, vol. 
8
 (pg. 
245
-
9
)
26
Levine
PH
Atherton
M
Fears
T
Hoover
R
An approach to studies of cancer subsequent to clusters of chronic fatigue syndrome: use of data from the Nevada State Cancer Registry
Clin Infect Dis
 , 
1994
, vol. 
18(Suppl 1)
 (pg. 
S49
-
53
)
27
Lange
G
DeLuca
J
Maldjian
JA
Lee
H
Tiersky
LA
Natelson
BH
Brain MRI abnormalities exist in a subset of patients with chronic fatigue syndrom
J Neurol Sci
 , 
1999
, vol. 
171
 (pg. 
3
-
7
)
28
Greco
A
Tannock
C
Brostoff
J
Costa
DC
Brain MR in chronic fatigue syndrom
AJNR Am J Neuroradiol
 , 
1997
, vol. 
18
 (pg. 
1265
-
9
)
29
Schwartz
RB
Garada
BM
Komaroff
AL
, et al.  . 
Detection of intracranial abnormalities in patients with chronic fatigue syndrome: comparison of MR imaging and SPEC
AJR Am J Roentgenol
 , 
1994
, vol. 
162
 (pg. 
935
-
41
)
30
Glaser
R
Padgett
DA
Litsky
ML
, et al.  . 
Stress-associated changes in the steady-state expression of latent Epstein-Barr virus: implications for chronic fatigue syndrome and cance
Brain Behav Immun
 , 
2005
, vol. 
19
 (pg. 
91
-
103
)
31
Birkenbach
M
Josefsen
K
Yalamanchili
R
Lenoir
G
Kieff
E
Epstein-Barr virus–induced genes: first lymphocyte-specific G protein-coupled peptide receptor
J Virol
 , 
2003
, vol. 
67
 (pg. 
2209
-
20
)
32
Sairenji
T
Yamanishi
K
Tachibana
Y
Bertoni
G
Kurata
T
Antibody responses to Epstein-Barr virus, human herpesvirus 6 and human herpesvirus 7 in patients with chronic fatigue syndrom
Intervirology
 , 
1995
, vol. 
38
 (pg. 
269
-
73
)
33
Chia
JK
The role of enterovirus in chronic fatigue syndrom
J Clin Pathol
 , 
2005
, vol. 
58
 (pg. 
1126
-
32
)
34
Glynn
P
NTE: one target protein for different toxic syndromes with distinct mechanism
Bioessays
 , 
2003
, vol. 
25
 (pg. 
742
-
5
)
35
Glynn
P
Neural development and neurodegeneration: two faces of neuropathy target esteras
Prog Neurobiol
 , 
2000
, vol. 
61
 (pg. 
61
-
74
)
36
Tahmaz
N
Soutar
A
Cherrie
JW
Chronic fatigue and organophosphate pesticides in sheep farming: a retrospective study amongst people reporting to a UK pharmacovigilance schem
Ann Occup Hyg
 , 
2003
, vol. 
47
 (pg. 
261
-
7
)
37
Proctor
SP
Heaton
KJ
Heeren
T
White
RF
Effects of sarin and cyclosarin exposure during the 1991 Gulf War on neurobehavioral functioning in US army veteran
Neurotoxicology
 , 
2006
, vol. 
27
 (pg. 
931
-
9
)
38
Leegwater
PA
Pronk
JC
van der Knaap
MS
Leukoencephalopathy with vanishing white matter: from magnetic resonance imaging pattern to five gene
J Child Neurol
 , 
2003
, vol. 
18
 (pg. 
639
-
45
)
39
Chao
CC
Janoff
EN
Hu
SX
, et al.  . 
Altered cytokine release in peripheral blood mononuclear cell cultures from patients with the chronic fatigue syndrom
Cytokine
 , 
1991
, vol. 
3
 (pg. 
292
-
8
)
40
MacDonald
KL
Osterholm
MT
LeDell
KH
, et al.  . 
A case-control study to assess possible triggers and cofactors in chronic fatigue syndrom
Am J Med
 , 
1996
, vol. 
100
 (pg. 
548
-
54
)
41
Jason
LA
Corradi
K
Torres-Harding
S
Taylor
RR
King
C
Chronic fatigue syndrome: the need for subtype
Neuropsychol Rev
 , 
2005
, vol. 
15
 (pg. 
29
-
58
)
42
Janal
MN
Ciccone
DS
Natelson
BH
Sub-typing CFS patients on the basis of ’minor’ symptom
Biol Psychol
 , 
2006
, vol. 
73
 (pg. 
124
-
31
)
43
Taylor-Robinson
D
David Arthur John Tyrrell CBE: 19 June 1925–2 May 200
Biogr Mem Fellows R Soc
 , 
2007
, vol. 
53
 (pg. 
349
-
63
)
Potential conflicts of interest: none reported
Presented in part: 2007 Conference of the International Association for Chronic Fatigue Syndrome/Myalgic Encephalomyelitis, 10–14 January 2007, Fort Lauderdale, Florida
Financial support: Chronic Fatigue Syndrome Research Foundation, Hertfordshire, United Kingdom
R.P. and B.B. made equal contributions to this article