Source: Journal of Chronic Fatigue Syndrome
        Vol. 14, #3, pp 7-25
Date:   Autumn 2007
URL:    http://jcfs.haworthpress.com  
        

Transcriptome Analysis of Peripheral Blood Mononuclear Cells from Patients
with Chronic Fatigue Syndrome
--------------------------------------------------------------------------
Hanna Grans, PhD; Birgitta Evengard, MD; Peter Nilsson, PhD
- Hanna Grans is affiliated with the Division of Clinical Bacteriology, F82,
  Karolinska Institutet, Stockholm, Sweden and the Department of Proteomics
  and Department of Gene Technology, KTH­Royal Institute of Technology, 
  Stockholm, Sweden.
- Birgitta Evengard is Professor, Division of Clinical Bacteriology, F82, 
  Karolinska Institutet, Stockholm, Sweden.
- Peter Nilsson is affiliated with the Department of Proteomics and Department
  of Gene Technology, KTH­Royal Institute of Technology, Stockholm, Sweden.
- Address correspondence to: Hanna Gräns, PhD, Division of Clinical 
  Bacteriology, F82, Karolinska Institutet, Karolinska University Hospital,
  Huddinge, Stockholm SE-141 86, Sweden (E-mail: Hanna.Grans@ki.se) or 
  Peter Nilsson (E-mail: nipe@kth.se).

Received: 06/16/06
Accepted: 05/24/07


ABSTRACT

Objective
Chronic fatigue syndrome (CFS) is an illness defined by unexplained disabling
fatigue lasting longer than six months, together with at least four out of
eight specified symptoms. The etiology and pathophysiology of CFS are to a
large degree unknown. Since much remains unclear about CFS we wanted to
investigate transcript expression levels in peripheral blood mononuclear
cells to identify genes that are involved in CFS.

Method
Transcript expression profiles for 20 CFS patients were compared with 14
healthy controls using microarray technology. Results were verified with
real-time PCR.

Results
We have identified significantly differentially expressed genes comparing a
female CFS patient subgroup with gradual illness onset and no previously
documented infection with female healthy controls. We have also created a
list of genes with indicated, but not verified, expression differences from
comparisons between other sub-groups and healthy controls. These genes are
candidates for further study of potential involvement in CFS.

Conclusion
Our results stress the necessity of subgrouping the heterogeneous CFS patient
cohort. The mRNA expression differences identified here may be causal factors
for the illness or symptoms observed in these patients, or a result of
altered functions of other cellular components involved in the illness. The
role of these genes in the CFS pathology needs further investigation.

KEYWORDS. Chronic fatigue syndrome, microarray technology, gene expression, 
transcriptome


INTRODUCTION

Fatigue is a common symptom in disease and illness. Unexplained disabling
fatigue lasting longer than six months, together with at least four out of
eight specified symptoms, is termed chronic fatigue syndrome (CFS) (1). The
prevalence of CFS is estimated to be 0.20-0.45% in the Western countries with
a predominance of women (2-4 times higher compared with males) (2, 3).
Population studies excluding medical examination also show a female
preponderance, but higher prevalence rates (1.4-2.4%) (4, 5).

Both the etiology and pathophysiology of CFS are unknown. Immune system
dysfunction, deficient neuroendocrine-immune communication, stress related
disturbance in the hypothalamic-pituitary-adrenal axis, and viral/bacterial
infections as an illness trigger have been postulated as possible causative
factors (6). Most of the transcript expression studies so far have reported
immune system dysfunction, although none of the genes identified as
differentially expressed is in common for any two studies. The poor overlap
seen between different studies could, however, be due to the heterogeneous
CFS patient cohort (7).

So far peripheral blood mononuclear cells (PBMCs) have been used in CFS gene
expression studies (8-14). The cells show small individual variability, and
can serve as indicators for abnormal processes going on throughout the body
(15).

We have here investigated mRNA transcript expression levels in PBMCs for 20
CFS patients and 14 healthy controls using microarray technology. We use a
different microarray platform and a different analysis approach compared
with previous studies, and verify our results with real-time PCR. The
identification of differentially expressed genes could give an important
increased insight into the pathophysiology of the illness.


METHODS

Study Population

The study cohort consisted of 20 CFS patients (Table 1) from a clinic for
infectious diseases (Stockholm, Sweden) fulfilling criteria for CFS and 14
healthy age-matched and sex-matched controls (1). The two groups have similar
age and sex distribution with a mean age of 38 years.


Sample Preparation

PBMCs were isolated from study participants immediately following blood draw
(Venglect(r) Evacuated blood collection tubes, heparin, Terumo(r), Leuven,
Belgium) after written informed consent (ethical approval, 130/02, Karolinska
University Hospital, Huddinge, Sweden). Total RNA was extracted from 10
million cells (TRIzol(r)Reagent, Invitrogen, Carlsbad, CA, USA), and the
quality was checked with the 2100 Bioanalyzer instrument (Agilent
Technologies, Palo Alto, CA, USA). Two micrograms of total RNA were amplified
(RiboAmpTM RNA Amplification Kit, Arcturus, Mountain View, CA, USA).

Starting with 500 ng of amplified RNA, an indirect amino allyl-dUTP labeling
protocol was used to prepare labeled cDNA (KTH Microarray Center). The
patient and control samples were labeled with Cy5, and the common reference
sample pool with Cy3. The nucleotide concentrations in all preparation steps
were quantified with spectrophotometry (NanoDrop Technologies, Wilmington,
DE, USA).


Microarray Experiment

Thirty-one successful hybridizations (17 patients and 14 controls) were
performed with indirect experimental design to allow all kind of group
comparisons (16). Three patients were omitted from microarray experiments due
to degraded RNA (two) and uneven hybridization (one).

In-house manufactured cDNA microarrays with 29,760 cDNA fragments
representing 18,953 unique human UniGene IDs were used (ArrayExpress,
A-MEXP-114) (17).

Scanned images were analyzed (GenePix® Pro 5.1, Axon instruments) using
median features with median local background subtraction. The definition of a
positive intensity signal was median background plus two standard deviations
(SDs). The SD of the background signal intensity was calculated using the
lower 55% of the background signal intensity values.

Data were pre-processed using the R environment for statistical computing and
the KTH-package, and was loess print-tip normalized (18, 19). An empirical
bayes method (B-test) was used to rank genes most likely to be differentially
expressed (20). Genes qualified for statistical testing had ratio values for
at least half of the samples and an absolute log ratio (M-value) larger than
0.3 (1.25 fold-change, between compared groups, e.g., patient/control). Genes
with B scores >1 in the B-test and p-values <0.0001 in the t-test (two-sided
with unequal variance) were selected for further verification experiments. As
several B-tests generated B values <1 for the highest ranked genes, the top
genes from the ranking lists were included for further investigation together
with genes with B scores >1, since so little is known about the
pathophysiology.

Proteins coded by genes with high rank scores were used for protein pathway
analysis using PathwayAssist (Ariadne Genomics, Rockville, MD, USA) and
evaluation of gene ontology categories. All data with detailed information
will be publicly available at ArrayExpress upon publication (17).


Gene Expression Validation

Results were validated using real-time PCR (LightCycler 2.0, Roche, Basel,
Switzerland). Complementary DNA was synthesized from 1 g of DNase treated
total RNA using SuperScriptTM III system (Invitrogen).

Primer pairs were designed using BeaconDesigner 4 (Premier Biosoft
International, Palo Alto, CA, USA). 18S rRNA and GAPDH were used for
normalization. Double stranded DNA-binding dye SYBR Green was used with hot
start and 40-50 cycles of 95 C for 5 seconds, 62/64 C for 5 seconds and 72 C
for 10 seconds (LightCycler FastStart DNA Master SYBR Green I, Roche).
Samples were run in duplicates and re- sults were analyzed using the standard
curve method. Microarray cDNA clones were also sequence verified.


Western Blot

PBMC lysates (106 cells) were separated onto a 12% SDS polyacryl amide gel
(Bio-Rad Laboratories, Inc., Hercules, CA, USA), and transferred to a PVDF
filter (Bio-Rad). The blocked PVDF membrane was incubated with primary hCD83
antibody (1:2000) (R & D Systems, Minneapolis, MN, USA), and secondary
HPR-coupled goat IgG antibody (1:3000) (R & D Systems).


RESULTS

Microarray Experiments

In order to verify that the applied method is suitable to distinguish
differential gene expression in these data set a comparison was made between
the sexes (Figure 1a). This comparison yielded 23 significantly
differentially expressed known genes, of which the majority were Y-linked or
X-linked (data not shown).

No indication of gene expression differences was found between the entire
groups of CFS patients versus healthy controls. A B-value of zero indicates
that the theoretical chance of differentially expression is 50%. None of the
genes had a positive B-value (Figure 1b).

The larger transcriptional differences between the sexes compared with CFS
patients versus healthy controls led us to focus on the sexes separately.
Comparing only female patients with female controls indicated genes with
higher probability of being differentially expressed (higher B values).
Further analysis was performed using only female patients and controls.


Significantly Differentially Expressed Genes in Female Patient Subgroups

In order to elucidate all possible new insights into the relatively unknown
pathophysiology and etiology of CFS, an unbiased and complete scheme of
pairwise group comparisons based on epidemiological data (Table 1) was
performed to identify differentially expressed genes.

Statistical tests comparing female CFS patients with no previous documented
infection (n=10) with female controls (n=12) and female patients with a
gradual illness onset (n=9) with female controls generated eight
overlapping genes with high ranking scores. Comparing the female patients
with both absence of previously documented infection and gradual illness
onset (n=8) with healthy controls yielded seven of the eight overlapping
genes on top of the ranking list indicating possible significant gene
expression differences (Figure 1c). Out of the five known genes significant
differences was verified for three genes (CD83, BOLA1 and NRK1) using
real-time PCR (Table 2). Significance was achieved using both 18S rRNA and
GAPDH for normalization. A trend of differential expression was seen for
SYNC1 (p=0.06). The fifth gene, WDR47, was by sequencing of the microarray
cDNA clone found to be another cDNA fragment with no similarity to any known
human gene.

In order to extend the value of the finding that CD83 is significantly
differentially transcribed, a limited initial experiment on the protein level
was performed. The human glycoprotein CD83 is a 45 kDa protein (21) and we
showed that the CD83 protein is detectable in PBMC lysates with western blot
analysis, which is a good starting point for subsequent deeper analysis on
the protein level. A weak band of correct size was identified in the majority
of the PBMC lysates analyzed, but it was not possible to determine any
significant differences in protein expression in this initial experiment.


Indication of Differential Expression in Other CFS Patient Subgroups

Indication of transcriptional differences was seen between several of the
other patient subgroups, although no statistical significance was achieved.
The interesting genes from all group comparisons are listed in Table 3. The
trend of down-regulation of CPT1A and FOSL2 in female patients with a
non-infectious illness onset compared with an infectious illness onset was
verified (Table 2). The trend for up-regulation of NEDD4L in female patients
with gradual illness onset compared with sudden onset was also verified
(Table 2). No transcriptional differences were observed over time of CFS
illness or between varying numbers of fulfilled symptoms.

Hierarchical clustering of all female samples using the genes in Table 3
gathered most of the patients in the subgroup consisting of patients with no
previously documented infection and gradual illness onset together, and the
other CFS patients were spread out mainly in patient clusters (Figure 2).


Linking Genes to Biology

Forty-three genes in Table 3 had a known protein product with a denoted locus
link ID. The proteins were searched for common cell processes and regulators
in the ResNet database (Ariadne Genomics). Five of the proteins were involved
in proliferation. Differentiation, proteolysis and contraction involved at
least two proteins (Figure 3a). Six out of the nine proteins involved in any
of the cellular processes had at least one regulator in common with another
protein (Figure 3b).

Gene ontology analysis of the genes in Table 3 revealed several proteins with
a role in lipid metabolism, fatty acid beta oxidation and oxidoreductase
activity.


DISCUSSION

The underlying mechanisms causing CFS are to a large degree unknown. There is
a continuous need to identify genes and proteins being differentially
expressed in CFS patients, both between patients and controls and between
different subgroups of patients, to increase the understanding of the
pathophysiology.

The CFS patients have a heterogeneous symptom profile. This makes the
diagnosis and treatment more difficult. Studies have attempted to subgroup
CFS patients according to clinical symptoms, for example, but no consistently
superior grouping strategy has so far been identified (7). The heterogeneity
can obscure differences between CFS patient subgroups, which emphasize the
need for subgrouping patients (7).

Here differential gene expression was only found when the patients were
subgrouped.

The choice of stringency level during microarray data analysis is being
reflected in the final outcome in terms of the number of differentially
expressed genes. We have here defined three genes as being thoroughly
verified as significantly differentially expressed between a subgroup of
female CFS patients and controls. The three genes, CD83, BOLA1 and NRK1, have
recently been briefly announced (22).

The CD83 protein is a maturation marker for dendritic cells, and is also
present in activated lymphocytes. The immunological role of the protein is
not yet fully elucidated. Down-regulation of CD83 due to infection by
different viruses has been observed, and suggested to be a viral mechanism to
escape host-specific immune responses (23). Inhibition of the stimulatory
function of dendritic cells lead to impaired antiviral T-cell responses (23).
Down-regulation of CD83 mRNA levels in the CFS patient subgroup may lead to
lower expression levels of the protein, which could in turn lead to disturbed
T-cell activity. Differential gene expression levels for other genes involved
in T-cell activation have been reported in several CFS studies (9, 10,12).
This finding indicates that more focus on infections, as one important part
of the CFS pathophysiology, is motivated.

A number of psychiatric and medical treatments have been tested for CFS
patients. Nicotinamide adenine dinucleotide (NAD+) is one pharmacological
treatment tested, but without any great success (24). We have here, although,
found that NRK1 (nicotinamide riboside kinase 1) is up-regulated in the CFS
patient subgroup. This enzyme is involved in the synthesis of NAD+ through
nicotinamide mononucleotide using nicotinamide riboside as the precursor.

The third gene, BOLA1 (BolA-like protein 1 [CGI-143]) has not been associated
with any known function. It has been mapped to the gene ontology entry of
transcription regulator activity. The BolA protein family in Escherichia coli
is believed to play an important role in general stress responses (heat
shock, acidic stress, oxidative stress, car- bon-starvation stress and
osmotic shock).

CPT1A is a key regulator in the mitochondrial oxidation of longchain fatty
acids and deficiencies could lead to a lower rate of fatty acid oxidation.
Kaushik et al. have reported up-regulation of genes involved in initiation of
translation in the mitochondria comparing CFS patients with controls (9).
FOSL2 has been implicated as a regulator of cell proliferation,
differentiation, and transformation.

NEDD4L is an ubiquitin ligase that controls cell surface expression of kidney
epithelial Na+ channels by ubiquitin-mediated endocytosis and lysosome
targeting.

The genes being indicated to be differentially expressed between pa- tient
subgroups and subgroups versus controls were analyzed according to their
functional categories with the aim to establish a biological role and pattern
of those genes. The 43 genes with known protein in Table 3 are involved in
widely spread biological processes. Gene ontology searches showed that a
number of the genes are involved in metabolism. Whistler et al. have also
identified genes involved in metabolism (14).

The mRNA expression levels do not seem to change over time of illness, but a
thorough time series experiment would be required to verify this indication.
The case definition requirement of fulfilling at least four out of eight
specified symptoms has lately been questioned (25, 26). Neither this study
nor another microarray study have identified differences between patients
with varying number of symptoms (14).

The agreement between the individual CFS studies is not good, although, some
categories of biological processes are recurrent, such as immune responses
and T-cell activation (9, 10, 12). All of the studies have used different
microarray platforms and analysis approaches.

The relatively small numbers of patients in the different CFS subgroups
analysed in this study resulted in a very small number of obtained
significantly differentially transcribed genes between subgroups. A larger
patient cohort focused on the subgroups is needed to obtain significance for
more genes in the candidate list. Still, the candidate list generated by the
transcript profiling screening provides a manageable number of suitable
targets for initial analysis on the protein level. This will be an important
subsequent step in order to generate increased knowledge regarding which gene
products that might be involved in the disease, either as a result or cause,
and for potential biomarker discovery.

The discovery of biological markers for CFS to increase the understanding of
illness pathophysiology and for facilitating diagnostics is only starting to
be uncovered. In this study we have identified three differentially expressed
genes, which have been thoroughly verified, and an extensive candidate list
of genes for further study of involvement in CFS. The difference in mRNA
expression may contribute to some of the symptoms observed in CFS, may be a
marker for the illness or a marker for altered functions of other cellular
components. The altered levels may provide an entry point for identifying
interesting potentially disease-causing molecules for further study.


The authors thank Annelie Walden for the microarray production.
This work was supported by Wallenberg Consortium North, the Swedish Research
Council, and the Swedish Council for Working Life and Social Research. The
authors declare that they have no competing interests.


TABLE CAPTIONS

There are no electronic versions of the tables available.

Table 1. Patient cohort data
Table 2. Top ranked genes from a comparison of a subgroup of female CFS
         patients with healthy female controls, and clinical patient subgroups,
         which have been verified by real time PCR
Table 3. Genes identified with microarray as potentially differentially
         expressed comparing patient subgroups and subgroups to controls


FIGURE CAPTIONS

Figure 1.
Volcano plots of the results from B-test group comparisons. The B-test
results from comparisons between: (a) the female study participants and the
male study participants, (b) the CFS patients and the controls (both females
and males), and (c) the female patients with gradual and non-infectious
illness onset versus the female controls.

Figure 2.
Clustering of female patients and controls using the top ranked genes from
all comparisons. The dendrogram shows the relationship between the female
study participants, and in the heat map the gene expression levels are
visualized in the heat map.

Figure 3.
Cell processes and regulators in common for the genes in Table 3. Biological
networks of the genes that: (a) are involved in cell processes that are
associated with at least one additional protein in the table, and (b) have a
regulator in common with at least one more protein in the table.


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 Group. Annals of
    Internal Medicine 1994; 121(12):953-959.
 2. Jason LA, Richman JA, Rademaker AW, Jordan KM, Plioplys AV, Taylor RR
    et al. A community-based study of chronic fatigue syndrome. Archives of 
    Internal Medicine 1999; 159(18):2129-2137.
 3. Reyes M, Nisenbaum R, Hoaglin DC, Unger ER, Emmons C, Randall B et al.
    Prevalence and incidence of chronic fatigue syndrome in Wichita, Kansas. 
    Archives of Internal Medicine 2003; 163(13):1530-1536.
 4. Evengard B, Jacks A, Pedersen NL, Sullivan PF. The epidemiology of chronic
    fatigue in the Swedish Twin Registry. Psychological Medicine 2005; 35(9):
    1317-1326.
 5. Lindal E, Stefansson JG and Bergmann S. The prevalence of chronic fatigue
    syndrome in Iceland - a national comparison by gender drawing on four 
    different criteria. Nordic Journal of Psychiatry 2002; 56(4):273-277.
 6. Evengard B, Klimas N. Chronic fatigue syndrome: probable pathogenesis and
    possible treatments. Drugs 2002; 62(17):2433-2446.
 7. Jason LA, Corradi K, Torres-Harding S, Torres-Harding S, Taylor RR, King C.
    Chronic fatigue syndrome: the need for subtypes. Neuropsychology Review 
    2005; 15(1):29-58.
 8. Fang H, Xie Q, Boneva R, Fostel J, Perkins R, Tong W. Gene expression
    profile exploration of a large dataset on chronic fatigue syndrome.
    Pharmacogenomics 2006; 7(3):429-440.
 9. Kaushik N, Fear D, Richards SC, McDermott CR, Nuwaysir EF, Kellam P et al.
    Gene expression in peripheral blood mononuclear cells from patients with 
    chronic fatigue syndrome. J of Clinical Pathology 2005; 58(8):826-832.
10. Powell R, Ren J, Lewith G, Barclay W, Holgate S, Almond J. Identification
    of novel expressed sequences, up-regulated in the leucocytes of chronic 
    fatigue syndrome patients. Clinical and Experimental Allergy 2003; 33(10):
    1450-1456.
11. Steinau M, Unger ER, Vernon SD, Jones JF, Rajeevan MS. Differential-display
    PCR of peripheral blood for biomarker discovery in chronic fatigue syndrome.
    Journal of Molecular Medicine 2004; 82(11):750-755.
12. Vernon SD, Unger ER, Dimulescu IM, Rajeevan M, Reeves WC. Utility of the
    blood for gene expression profiling and biomarker discovery in chronic 
    fatigue syndrome. Disease Markers 2002; 18(4):193-199.
13. Whistler T, Jones JF, Unger ER, Vernon SD. Exercise responsive genes mea-
    sured in peripheral blood of women with chronic fatigue syndrome and 
    matched control subjects. BMC Physiology 2005; 5(1):5.
14. Whistler T, Unger ER, Nisenbaum R, Vernon SD. Integration of gene expres-
    sion, clinical, and epidemiologic data to characterize Chronic Fatigue 
    Syndrome. Journal of Translational Medicine 2003; 1(1):10.
15. Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC, Relman DA et al.
    Individuality and variation in gene expression patterns in human blood. 
    Proceeding of the National Academy of Science U S A 2003; 100(4):1896-1901.
16. KTH Microarray Center.(http://www.ktharray.se).
17. EBI Database ArrayExpress.(http://www.ebi.ac.uk/arrayexpress/).
18. R Development Core Team: R: A language and environment for statistical com-
    puting. R Foundation for Statistical Computing, Vienna, Austria 2004.
19. Smyth GK, Speed T. Normalization of cDNA microarray data. Methods 2003;
    31(4):265-273.
20. Lonnstedt I, Speed T. Replicated microarray data. Statistica Sinica 2002;
    12:31-46.
21. Zhou LJ, Schwarting R, Smith HM, Tedder TF. A novel cell-surface molecule
    expressed by human interdigitating reticulum cells, Langerhans cells, and 
    activated lymphocytes is a new member of the Ig superfamily. Journal of 
    Immunology 1992; 149(2):735-742.
22. Grans H, Nilsson P, Evengard B. Gene expression profiling in the chronic
    fatigue syndrome. Journal of Internal Medicine 2005; 258(4):388-390.
23. Lechmann M, Zinser E, Golka A, Steinkasserer A. Role of CD83 in the
    immunomodulation of dendritic cells. International Archives of Allergy 
    and Immunology 2002; 129(2):113-118.
24. Cairns R, Hotopf M. A systematic review describing the prognosis of chronic
    fatigue syndrome. Occupational Medicine (London) 2005; 55(1):20-31.
25. Reeves WC, Lloyd A, Vernon SD, Klimas N, Jason LA, Bleijenberg G. Identifi-
    cation of ambiguities in the 1994 chronic fatigue syndrome research case 
    definition and recommendations for resolution. BMC Health Services 
    Research 2003; 3(1):25.
26. Sullivan PF, Pedersen NL, Jacks A, Evengard B. Chronic fatigue in a
    population sample: definitions and heterogeneity. Psychological Medicine 
    2005; 35:1337-1348.

--------
(c) 2007 The Haworth Press