Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Evidence of inflammatory immune signaling in chronic fatigue syndrome: A pilot study of gene expression in peripheral blood

Evidence of inflammatory immune signaling in chronic fatigue syndrome: A pilot study of gene... Background: Genomic profiling of peripheral blood reveals altered immunity in chronic fatigue syndrome (CFS) however interpretation remains challenging without immune demographic context. The object of this work is to identify modulation of specific immune functional components and restructuring of co-expression networks characteristic of CFS using the quantitative genomics of peripheral blood. Methods: Gene sets were constructed a priori for CD4+ T cells, CD8+ T cells, CD19+ B cells, CD14+ monocytes and CD16+ neutrophils from published data. A group of 111 women were classified using empiric case definition (U.S. Centers for Disease Control and Prevention) and unsupervised latent cluster analysis (LCA). Microarray profiles of peripheral blood were analyzed for expression of leukocyte-specific gene sets and characteristic changes in co-expression identified from topological evaluation of linear correlation networks. Results: Median expression for a set of 6 genes preferentially up-regulated in CD19+ B cells was significantly lower in CFS (p = 0.01) due mainly to PTPRK and TSPAN3 expression. Although no other gene set was differentially expressed at p < 0.05, patterns of co-expression in each group differed markedly. Significant co-expression of CD14+ monocyte with CD16+ neutrophil (p = 0.01) and CD19+ B cell sets (p = 0.00) characterized CFS and fatigue phenotype groups. Also in CFS was a significant negative correlation between CD8+ and both CD19+ up-regulated (p = 0.02) and NK gene sets (p = 0.08). These patterns were absent in controls. Conclusion: Dissection of blood microarray profiles points to B cell dysfunction with coordinated immune activation supporting persistent inflammation and antibody-mediated NK cell modulation of T cell activity. This has clinical implications as the CD19+ genes identified could provide robust and biologically meaningful basis for the early detection and unambiguous phenotyping of CFS. cell dysfunction and activation has been demonstrated in Background Chronic fatigue syndrome (CFS) is estimated to cost the CFS by several groups [2-4]. Though similar in terms of American economy over $9 billion each year in lost pro- broad lymphocyte classes CFS and non-fatigued subjects ductivity [1]. Among other components chronic immune can be readily distinguished when specific immune cell Page 1 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 subsets are examined. For example Klimas et al. [2] report a tion in CFS from gene expression profiles of mixed lym- significant expansion CD26+ (DPP-IV) activated T cells in phocyte populations. In particular we construct gene sets CFS subjects. This multifunctional molecule plays a major that capture elements of abundance and activity assigna- role in the regulation, development, maturation and migra- ble to specific immune cell subsets thereby facilitating tion of T helper (Th) and natural killer (NK) cells as well as direct integration with flow cytometry results. Data from in B cell immunoglobulin switching [5]. Moreover abnor- a large population-based study of CFS [16] is then exam- mal expression of CD26+ is found in autoimmune diseases ined for changes of immune set expression across two sep- [6]. More recently CFS patients were also reported to have arate CFS classification approaches. In addition, patterns significantly fewer CD3+/CD25- T cells and significantly of coordinated expression linking these immune sets were more CD20+/CD5+ B cells [7], a subset associated with investigated using simple correlation networks. These net- auto-antibodies. Significantly fewer CD56+ NK cells were works were examined for shifts in topology and point to also observed in recent work by Racciatti et al. [8]. Though patterns of immune signaling in CFS that are consistent important, flow cytometry results such as these leave many with chronic inflammation. These observations could questions regarding cellular state unanswered. Microarray constitute a signature of CFS or a component thereof. profiling of gene expression on the other hand offers a glimpse of pathway activation in disease pathogenesis at Methods molecular resolution. Microarray analysis of cDNA profiles Subjects and diagnostic classes in peripheral blood mononuclear cells (PBMC) have Recently a dataset for a 2-day in-hospital study of CFS in revealed altered expression in CFS of several immune genes the general population of Wichita Kansas was made avail- [9,10] involved in response to oxidative stress, NK cell able [16]. Referred to as the Wichita Clinical study, this activity and elements of antigen processing. Instability in investigation included a highly comprehensive spectrum immune response and restructuring of immune cell signal- of detailed clinical and laboratory measures and PBMC ing under exercise challenge has also been observed [11]. expression profiles for 20,000 genes. From this dataset a Unfortunately microarray profiling is commonly per- final analysis group of 111 female subjects was obtained formed on mixed cell populations producing an average by excluding the few male subjects and subjects with con- profile from which it is very difficult to dissect the contribu- founding medical or psychiatric conditions. Subjects in tions of relative cell abundance, cell activation state and this dataset were classified as CFS using the CDC Symp- cell-cell signaling. More importantly, this averaging can tom Inventory, Multidimensional Fatigue Inventory obscure significant changes in the state of minority cell sub- (MFI) and Short Form 36 (SF-36) instruments [17,18]. populations. This classification will be referred to as "empiric" and resulted in 39 CFS, 37 non- fatigued (NF), and 35 subjects These challenges notwithstanding, a review of this evidence with insufficient symptoms or fatigue (ISF). A second clas- strongly suggests that CFS pathogenesis is likely to include sification proposed by Vollmer-Conna and colleagues a characteristic immunologic component in at least one [19] used latent class analysis (LCA) of 440 clinical and subset of the patient population [12]. However the exact biological measurements to delimit 5 fatigue classes, a nature of this immunologic component remains the object non-fatigued class and 2 unassigned individuals. Obese of considerable debate at least in part because of an inabil- subjects with prominent post-exercise fatigue, hypnoea ity to cast gene expression profiles in the useful context of and disturbed sleep formed Class 1. Reasonably healthy immune cell demographics. In an attempt to address this subjects with few symptoms, low depression scores and issue methods have been proposed to dissect global gene good sleep composed Class 2. Subjects in Class 3 resem- expression profiles into discrete elements assignable to bio- bled those in class 1 but also displayed low heart rate var- logic processes [13-15]. The assignment of genes to discrete iability during sleep and low 24-hour cortisol levels. Class modules or sets has been successful in several respects. A 4 was populated with healthier, less depressed individuals first contribution involves simply reducing the dimension- having restful sleep but suffering muscle pain. Finally ality of >55,000 gene expression measures to that of say 10 Classes 5 and 6 both captured less obese but highly symp- or so gene sets. The interpretability of results is further tomatic and depressed individuals with prominent post- enhanced by associating sets with basic cellular functions. exercise fatigue. Individuals in Class 6 also displayed dis- Finally the numerical robustness is greatly improved turbed sleep with low heart rate variability and low corti- through the averaging of changes in expression over many sol. The patient demographics for each of these genes. In addition gene sets are transportable across micro- classification systems are summarized in Table 1 and the array platforms making it possible to compare studies alignment between these systems is described in Table 2. based on different technologies. The collection and processing of PBMCs including hybrid- ization to MWG microarrays (MWG Biotech, Ebersberg, In this work we explore the use of discrete gene sets in Germany) are described in Vernon and Reeves [16]. extracting useful information regarding immune dysfunc- Details of the microarray data preprocessing including Page 2 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 1: Demographic data for 111 subjects from the Wichita clinical study Empiric Classification LCA Classification CFS IFS Controls LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 LCA-0/2 (n = 39) (n = 35) (n = 37) (n = 23) (n = 17) (n = 11) (n = 14) (n = 11) (n = 35) Mean Age (SD) 51.4 (8.2) 50.3 (8.2) 51.6 (9.0) 50.9 (7.6) 54.7 (5) 44.4 (8.7) 48.2 (10.2) 55.8 (3.4) 51.3 (9) Mean Years Ill (SD) 16.7 (11.0) 14.4 (10.0) 2.8 (5.0) 15.5 (10.7) 11.3 (4) 16.0 (12.3) 16.8 (10.5) 16.4 11.2 14.3 (12.9) Race [n (%) ] White 35 (90.0) 32 (91.4) 36 (97.3) 21 (91) 17 (100) 10 (90.9) 11 (78.6) 11 (100) 33 (94.3) Black 1 (2.6) 3 (8.6) 1 (2.7) 1 (4.4) 0 (0) 1 (9.1) 1 (7.1) 0 (0) 2 (5.7) Multiple Race 2 (5.1) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 2 (14.3) 0 (0) 0 (0) Other 1 (2.6) 0 (0) 0 (0) 1 (4.4) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Onset Type Gradual 32 (82.0) 28 (80.0) 10 (27.0) 18 (78.3) 14 (82.4) 9 (81.8) 11 (78.6) 10 (90.9) 8 (22.9) Sudden 6 (15.4) 3 (8.6) 0 (0.0) 3 (13) 1 (5.9) 1 (9.1) 3 (21.4) 1 (9.1) 0 (0.0 Undetermined 1 (2.6) 4 (11.4) 27 (73.0) 2 (8.7) 2 (11.8) 1 (9.1) 0 (0) 0 (0) 27 (77.1) BMI [n (%) ] <25 5 (13.8) 10 (28.6) 7 (18.9) 0 (0) 4 (23.5) 8 (72.7) 3 (21.4) 3 (27.3) 4 (11.4) 25–30 20 (51.3) 14 (40) 18 (48.7) 9 (39.1) 6 (35.3) 3 (27.3) 9 (64.3) 7 (63.6) 18 (51.4) >30 14 (35.9) 11 (31.4) 12 (32.4) 14 (60.9) 7 (41.2) 0 (0) 2 (14.3) 1 (9.1) 13 (37.1) SD is standard deviation normalization, outlier detection and false discovery cor- MWG microarrays used in the Wichita Clinical study. We rection are available in Broderick et al. [9]. further dissected these subset-specific profiles into dis- crete non-overlapping sets composed of genes at least 2- Gene set development fold up-regulated or 2-fold down-regulated preferentially Extracting elements that represent the abundance and in each cell lineage. An additional gene set was defined for activity of a specific leukocyte subset was approached by NK cell activity and regulatory T cell activity was estimated identifying discrete sets of genes that are uniquely or pre- from the expression of the FoxP3 gene (AF277993). Indi- dominantly expressed in a given cell type [20-22]. Cur- vidual MWG gene probes belonging to each immune gene rently discrete gene sets offer the simplest and most set as well as NCBI gene annotation and PANTHER func- immediately accessible method for analysis across micro- tional annotation [24,25] are listed in the supplementary array technological platforms. We constructed a number data file [Additional file 1]. of gene sets a priori for CD4+ T cells, CD8+ T cells, CD19+ Statistical analysis B cells, CD14+ monocytes and CD16+ neutrophils using data collected on Affymetrix microarrays (Affymetrix, The aggregate expression G of each gene set a was com- Santa Clara, CA, USA) by Lyons et al. [23]. Of the 12,022 puted as the average of the Ln-transformed expression genes surveyed, 2,641 were differentially expressed Ln(g ) of each gene i across the k member genes in the set i, a (Equation 1). In a first level of analysis a classical Wil- between individual lymphocyte subsets. Of these original 2,641 distinguishing genes, 268 were present on the coxon non-parametric test was used to evaluate the differ- Table 2: A cross-reference of systems for diagnostic assignment Empiric Classification (CFS research case definition) CFS ISF Controls LCA Category LCA Class Description (n = 39) (n = 35) (n = 37) n(%) n(%) n(%) Controls (0–2) Well (n = 33) or Unassigned (n = 2) 1 (2) 0 (0) 34 (91) 1 Obese hypnoea (n = 23) 15 (38) 8 (23) 0 (0) 3 Obese hypnoea and stressed (n = 17) 5(13) 11 (31)1 (3) 4 Interoception – muscle pain (n = 11) 1(3) 9 (26) 1(3) 5 Interoception depression (n = 14) 10 (26) 4 (11) 0 (0) 6 Multisymptomatic, depressed, stressed (n = 11) 7 (18) 3 (9) 1 (3) Page 3 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 ential expression of immune gene sets for both Results classification systems. As suggested by Efron and Tib- Alignment of empiric and LCA classifications A cross tabulation of the empiric classification and LCA shirani [15] the performance of these gene sets was also compared to that obtained with randomly populated sets classification is presented in Table 2. There was good of the same size. A null distribution was computed from alignment of non-fatigued subjects with 90% of empiric the analysis of 1000 instances of random gene selections NF controls residing in LCA classes 0 (Well) and 2 (Unas- and 1000 random permutations of the diagnostic labels. signed). Together LCA classes 1 (40%) and 5 (26%) con- tained two thirds of the subjects assigned to the empiric CFS class. However ISF subjects were distributed almost Ln(g ) equally across LCA classes 1 (23%), 3 (31%), and 4 ia , (1) i =1 (26%). Conversely most LCA class 3 and 4 subjects were G = identified as ISF and most subjects in LCA classes 1, 5 and 6 were assigned an empiric CFS classification. To examine the patterns of association linking immune gene sets simple linear association networks were con- Differential expression of a priori defined immune cell structed using the Pearson correlation coefficient r as a, b gene sets the metric describing similarity in the expression of gene In a first level of analysis the differential expression of set a with that of gene set b. Statistical significance of cor- immune gene sets across disease phenotypes and control relation was assessed using the t statistic in Equation a, b groups for both classification systems was evaluated. (2). This statistic has a Student's t-distribution with Results in Table 3 show that the median expression of the degrees of freedom n-2 under the null hypothesis of no CD19+ B cell up-regulated gene set was significantly lower correlation [26], where n is the number of microarray in CFS (p = 0.01) and ISF (p = 0.05) subjects when com- measurements. C is the covariance in the expression of a, b pared to the NF group. Expression of this gene set was also gene set a with gene set b and E() is the expected value significantly repressed in LCA class 3 (p = 0.04) and mar- operator or the mean. ginally so in LCA class 5 (p = 0.09) when compared to control subjects in LCA classes 0 and 2. Recall that 11 of () n −2 tr = 17 cases in LCA class 3 were also designated ISF. Similarly ab,, ab (2) (1−r ) 10 of the 14 LCA class 5 cases were designated CFS. NK ab , gene set expression was marginally increased in the CFS group (p = 0.07). Though not significant the null proba- Where, bility for NK cell expression was lowest among the LCA classes for LCA-3 (p = 0.11). Finally expression of the T ab , regulatory set (FoxP3) was marginally repressed in LCA r== ;CE[(G−E(G ))⋅ (G−E(G ))] ab,, ab a a b b C C aa,, bb class 1 (p = 0.09) which contained 40% of the CFS sub- jects though no significant difference was found for the A cutoff for the resulting probability p (t > t ), above a, b a, b larger CFS group (p = 0.31). which we accept the null hypothesis, can be established in a variety of ways [27]. It should be noted however that The performance of these gene sets was compared to that these require specific assumptions regarding network obtained with randomly populated sets of the same size as topology such as network edge sparseness or the appear- well as by random classification assignment of each sub- ance of highly cliquish disconnected sub-networks. As ject. Null distribution results indicated that both the such they are generally more relevant to the study of large CD19+ B cell (up-regulated) and NK cell gene sets per- networks. Instead we examined the dependency of the formed significantly better than random sets of equivalent network size S, or the sum of the edge weights w on the a, b size in discriminating CFS from NF (p < 0.05) (Figure 1). choice of threshold p-value (Equation 3). We compared Performance of the T regulatory gene set (FoxP3) was mar- curves obtained for NF and CFS networks, identifying ginal at best (p~0.15) in terms of uniqueness in differen- threshold p-values where networks differed primarily in tial expression. In addition a detailed analysis of structure from those where they differed in both structure individual genes in the CD19+ up-regulated set indicated and size. that no single gene was differentially expressed even though the parent set was expressed at the p = 0.01 level. M−1 M This reaffirms that high levels of measurement noise can Sw = ab , ∑ ∑ be effectively managed by aggregating genes into biologi- a =1 ba > (3) cally relevant sets. Details of this analysis are listed in where wr=≤ if p p ab,, ab ab, threshold Table 4 and illustrated graphically in Figure 2 and w = 0 if pp > ab , ab , threshold Figure 3. Page 4 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 3: Changes in median expression and corresponding null probability values () for pair-wise comparison of disease classes and the non-fatigued control group under both classification systems 3-Class (NF Controls) 7-Class (Controls = LCA-0 U LCA-2) Cell Type Expression Number of ISF CFS LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 Level +/- genes CD8 T cells Up-regulated 5 0.06 (0.43) 0.01 (0.75) 0.06 (0.52) 0.00 (0.64) 0.04 (0.24) 0.01 (0.80) 0.23 (0.27) Down- regulated CD14 Up-regulated 78 0.04 (0.29) 0.05 (0.83) 0.02 (0.62) 0.11 (0.15) 0.09 (0.42) 0.02 (0.94) 0.05 (0.33) Monocytes Down- regulated CD16 Up-regulated Neutrophils Down- 185 0.02 (0.20) 0.02 (0.25) 0.02 (0.29) 0.00 (0.65) 0.03 (0.25) 0.02 (0.50) 0.04 (0.30) regulated CD19 B cells Up-regulated 6 -0.17 (0.05) -0.28 (0.01) -0.27 (0.29) -0.31 (0.04) -0.12 (0.13) -0.25 (0.09) -0.17 (0.14) Down- 2 -0.22 (0.33) -0.22 (0.29) -0.18 (0.19) 0.07 (0.71) 0.10 (0.76) -0.08 (0.78) -0.08 (0.16) regulated CD4/8/25 T NA 1 -0.23 (0.19) -0.23 (0.31) -0.35 (0.09) 0.19 (0.37) -0.25 (0.11) -0.17 (0.32) 0.28 (0.40) reg cells NK cells NA 4 0.09 (0.92) 0.30 (0.07) 0.19 (0.20) 0.32 (0.11) 0.12 (0.68) 0.12 (0.45) -0.01 (0.57) Emergence of characteristic patterns of association the patterns of association linking immune gene sets sim- between immune gene sets in CFS ple linear correlation networks were constructed. Results Conventional analysis of microarray data remains focused in Figure 4 show network size for the empiric NF and CFS on the detection of differentially expressed genes or gene classes as a function of cutoff p-value for edge weight sig- sets. However it is important to realize that genes nificance. Both NF and CFS networks were identical in expressed at similar levels across patient groups may still overall size at a cutoff p-value of 0.05. Changes at this play an important role in the disease process. To examine level of edge weight significance consisted therefore of a Gene Set CD19+ Up Gene Set NK aggregated 1 1 Random Gene Set 0.9 0.9 Random Diagnostic Label delta median = -0.3367 0.8 0.8 0.7 0.7 0.6 0.6 P(x>0.2790) ~ 0.025 0.5 0.5 P(x<-0.3367) ~ 0.006 0.4 0.4 0.3 0.3 0.2 0.2 Random Gene Set delta median = 0.2790 0.1 Random Diagnostic Label 0.1 0 0 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Delta Median CFS - NF Random Permutations Delta Median CFS - NF Random Permutations (A) (B) Cum cell Figure 1 up-regulated gene set ulative probability plot anof d thΔ d e NK cell gene set ifferential expression of CFS versus NF for random gene sets similar in size to the CD19+ B Cumulative probability plot of Δ differential expression of CFS versus NF for random gene sets similar in size to the CD19+ B cell up-regulated gene set and the NK cell gene set. Page 5 of 13 (page number not for citation purposes) Cumulative Probability (1000 permutations) Cumulative Probability (1000 permutations) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 4: Changes in median expression and corresponding null probability values () for pair-wise comparison of disease classes and the non-fatigued control group for each individual gene in the CD19+ B cell Up-regulated gene set 3-Class (NF Controls) 7-Class (Controls = LCA-0 U LCA-2) Gene ISF CFS LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 SP140 -0.03 (0.60) 0.02 (0.49) -0.07 (0.43) 0.10 (0.60) 0.01 (0.72) -0.08 (0.29) 0.07 (0.92) CD22 -0.16 (0.10) -0.09 (0.31) -0.14 (0.17) -0.58 (0.02) -0.30 (0.04) 0.02 (0.35) 0.05 (0.70) QRSL1 -0.03 (0.95) -0.21 (0.91) 0.29 (0.28) -0.10 (0.70) 0.23 (0.50) -0.16 (0.89) -0.49 (0.42) PTPRK -0.12 (0.55) -0.21 (0.18) -0.12 (0.76) -0.07 (0.82) -0.22 (0.68) -0.20 (0.17) -0.07 (0.64) P2RY10 0.01 (0.21) -0.02 (0.51) 0.08 (0.99) -0.08 (0.06) 0.28 (0.82) 0.11 (0.94) 0.20 (0.33) TSPAN3 -0.10 (0.54) -0.20 (0.19) -0.24 (0.39) -0.17 (0.83) 0.13 (0.38) -0.08 (0.83) -0.02 (0.82) re-organization of edges only. The curves in Figure 4 the CFS network. Topologies emerging at both p < 0.05 diverge at p~0.10 and maintain a similar offset form one and p < 0.10 thresholds were examined as they contain another as p-value increases. As a result comparisons of complementary information. Detailed results of pair-wise network topology conducted at the p < 0.10 level included correlation between gene sets may be found in Table 5 for edge re-assignment as well as the addition of new edges to Gene SP140 Gen Gene CD22 e CD22 Gene QRSL1 12 12 p=0.4863 p=0.3060 11 11 10 10 9 9 8 8 7 7 p=0.9131 6 6 CFS ISF NF CFS ISF NF CFS ISF NF Gene PTPRK Gene P2Y10 Gene TSPAN3 12 12 p=0.1939 p=0.5127 11 11 p=0.1800 10 10 9 9 8 8 8 7 7 6 6 CFS ISF NF CFS ISF NF CFS ISF NF Box classes Figure 2 and whisker plot for the expression of each gene in the CD19+ up-regulated gene set in each of the 3 empiric illness Box and whisker plot for the expression of each gene in the CD19+ up-regulated gene set in each of the 3 empiric illness classes. Boxes indicate the lower quartile, median and upper quartile values. Whiskers are located at extreme values within 1.5 times the inter-quartile range from the ends of each box. Outliers are displayed with a red '+'. Each plot is annotated with the null probability for the difference in median expression between the NF and CFS subject groups. Page 6 of 13 (page number not for citation purposes) Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 1.4 Gene Set CD19+ Up-regulated NF controls 9.5 CFS 1.2 p=0.0100 9.0 0.8 8.5 0.6 8.0 0.4 0.2 7.5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 7.0 Significance cutoff p-value Threshold p-value Ne weights (Equ value for the empiric Figure 4 twork size ation 3) and plotte S defineNF d as and CFS classes the sum o d as a function f all networ of cutoff p- k edge 6.5 Network size S defined as the sum of all network CFS ISF NF edge weights (Equation 3) and plotted as a function of cutoff p-value for the empiric NF and CFS classes. B regulated gene set Figure 3 ox and whisker plin each ot for the of the 3 empiri expression of the CD19 c illness classes + up- Box and whisker plot for the expression of the CD19+ up-regulated gene set in each of the 3 empiric trophil gene set were completely absent in NF even at the illness classes. Boxes indicate the lower quartile, median p < 0.10 level. Also apparent in CFS was the emergence of and upper quartile values. Whiskers are located at extreme values within 1.5 times the inter-quartile range from the ends a significant negative correlation between the expression of each box. Outliers are displayed with a red '+'. The plot is of CD8+ and CD19+ up-regulated gene sets (p = 0.02). annotated with the null probability for the difference in Moreover CD19+ B cells appeared altered with up and median expression between the NF and CFS subject groups. down-regulated sets no longer maintaining a strong direct correlation in ISF or CFS. Interaction with NK cell gene set expression was also a distinguishing feature in particular the empiric classification system and Tables 6 and 7 for for ISF. Instead of appearing as a transitional state the LCA classification system. between NF and CFS, ISF exhibited a distinct co-expres- sion pattern characterized by a significant interaction of Heat maps depicting the edge weights ra, b linking gene NK cell and monocyte gene sets (p < 0.05). Contrary to sets are presented in Figure 5 for both empiric and LCA NF, the NK and CD19+ down-regulated gene sets corre- classes. The LCA control classes 0 and 2 exhibited a pat- lated negatively (p < 0.10) in ISF. tern of gene set co-expression at the p < 0.10 identical to that of the empiric NF group though in the latter this pat- In much the same way as the ISF group, several of the LCA tern was also retained at the p < 0.05 level. In both the NF groups were characterized by a lack of coordinated activity and LCA-0/LCA-2 networks CD 19+ B cell up-regulated between immune gene sets. Indeed no significant correla- and down-regulated gene sets correlated tightly behaving tions existed for LCA-4 even at the p < 0.10 level. This was as one set (r = 0.43, p = 0.008). T regulatory and NK cell also true of LCA-3 and LCA-6 classes at the p < 0.05 level. gene sets both supported significant positive interaction Though also quite sparse, heat maps for LCA-1 and LCA- with one or both of the CD19+ B cell sets. In addition 5 each recovered specific features of the CFS and ISF heat CD8+ T cell activity and CD14+ monocyte activity were maps. LCA-1 demonstrated a significant positive correla- significantly antagonistic. In contrast the network tion (p < 0.05) between CD14+ monocyte and CD16+ obtained for the CFS subjects displayed a shift in interac- neutrophil sets, a CFS feature. At the p < 0.10 level the tions towards the upper left hand corner of the heat map. same heat map showed a negative correlation between the Indeed significant interactions appeared linking the NK and CD19+ B cell down-regulated set, an ISF feature. expression of the CD14+ monocyte gene set with that of Unique to LCA-1 was a positive correlation between T reg the B cell (CD19+ up-regulated) and the CD16+ neu- (FoxP3) and CD14+ monocyte gene set expression (p < trophil gene sets. The neutrophil set also shared signifi- 0.10), a trait not identified in CFS and actually reversed in cant co-expression with the CD8+ T cell gene set (p < LCA-3. Similarly the heat map for LCA-5 contained 2 fea- 0.05) in CFS. Interestingly interactions with the neu- tures specific to the CFS group, namely a strong positive Page 7 of 13 (page number not for citation purposes) Ln Transformed Gene Expression Network Size S Graph Size Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 5: Detailed results of gene set correlation r (null probability p ) for empiric classes. a, b a, b CFS CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.43 (0.01) 0.13 (0.42) 0.14 (0.39) 0.33 (0.04) 0.27 (0.10) -0.25 (0.12) CD14+ Up 0.54 (0.00) 0.29 (0.07) -0.12 (0.46) 0.18 (0.26) -0.08 (0.61) CD19+ Up 0.16 (0.33) -0.38 (0.02) 0.12 (0.46) 0.15 (0.36) CD19+ Down 0.15 (0.37) -0.23 (0.16) -0.04 (0.79) CD8+ Up 0.00 (1.00) -0.29 (0.08) T reg 0.02 (0.88) NK cell ISF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.13 (0.46) 0.11 (0.51) -0.13 (0.47) -0.16 (0.37) 0.19 (0.27) 0.09 (0.62) CD14+ Up -0.24 (0.17) 0.09 (0.62) -0.26 (0.13) -0.11 (0.52) 0.49 (0.00) CD19+ Up -0.08 (0.64) 0.02 (0.91) 0.26 (0.13) -0.03 (0.88) CD19+ Down 0.22 (0.21) 0.00 (0.99) -0.30 (0.08) CD8+ Up 0.17 (0.33) -0.26 (0.13) T reg -0.04 (0.80) NK cell NF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.10 (0.55) 0.11 (0.53) 0.08 (0.64) -0.27 (0.11) 0.15 (0.38) -0.01 (0.96) CD14+ Up 0.12 (0.49) -0.06 (0.74) -0.38 (0.02) -0.10 (0.54) 0.02 (0.91) CD19+ Up 0.43 (0.01) -0.02 (0.90) 0.37 (0.02) 0.25 (0.13) CD19+ Down 0.05 (0.79) 0.37 (0.02) 0.35 (0.04) CD8+ Up -0.10 (0.58) 0.20 (0.24) T reg 0.20 (0.23) NK cell correlation linking CD19+ B cell with CD14+ monocyte tion providing additional insight into potential subclasses up-regulated sets (p < 0.05) along with a strong negative of CFS. The commonalities between these classification correlation linking the former with CD8+ gene set expres- systems are readily observed in the patterns of gene set co- sion. These 2 features were not shared with the LCA-1 expression. Indeed the empiric CFS group seems to group. These results reaffirmed the strong links between present an aggregation of the gene set co-expression pat- the CFS subject group and LCA classes 1 and 5 in addition terns observed in LCA classes 1 and 5. However, the differ- to suggesting that immune set co-expression might offer ential expression of gene sets only achieves statistical insight into the distinct nature of these apparent sub- significance in the case of the coarser empiric classes with classes of CFS. the larger group sizes providing better noise reduction. Specifically in the empiric CFS class we found a significant decrease in the median expression for a set of 6 genes pref- Discussion In this work we dissected PBMC gene expression profiles erentially up-regulated in isolated CD19+ B cells com- into components that were preferentially expressed in sev- pared to non-fatigued controls. Expression of this CD19+ eral isolated lymphocyte subpopulations. We also used 2 B cell up-regulated gene set also discriminated ISF from systems to stratify subjects into illness groups. The LCA controls at 0.05 confidence level. In a recent study of CFS class structure was inferred directly from a comprehensive occurrence both in the presence and absence of viral infec- set of clinical and biological indicators. All indicators tion Racciati et al. [8] found no significant differences in were equally weighted and contrary to common practice CD19+ cell abundance. Robertson et al. [7] recently no subset was assigned greater relevance a priori. In con- reported significantly higher abundance of CD20+/CD5+ trast the empiric classification which was based on a con- B cells, a subset associated with the production of auto- sensus of opinions from expert clinicians. Results confirm antibodies, in patients with depression. These findings strong links between both systems with the LCA classifica- together with our observations of depressed CD19+ gene Page 8 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 6: Detailed results of gene set correlation r (null probability p ) for LCA-0/2, 1, 3 classes. a, b a, b LCA-0/2 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.17 (0.33) 0.15 (0.39) 0.09 (0.62) -0.19 (0.27) 0.09 (0.62) -0.10 (0.56) CD14+ Up 0.14 (0.41) -0.03 (0.88) -0.30 (0.08) -0.11 (0.52) 0.00 (0.98) CD19+ Up 0.43 (0.01) -0.04 (0.82) 0.40 (0.02) 0.19 (0.28) CD19+ Down -0.01 (0.93) 0.39 (0.02) 0.30 (0.08) CD8+ Up -0.05 (0.77) 0.06 (0.73) T reg 0.04 (0.82) NK cell LCA-1 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.55 (0.01) -0.03 (0.91) 0.16 (0.47) 0.28 (0.20) 0.16 (0.46) -0.25 (0.25) CD14+ Up 0.25 (0.26) 0.20 (0.37) -0.03 (0.90) 0.36 (0.10) -0.04 (0.84) CD19+ Up -0.14 (0.53) -0.21 (0.34) 0.13 (0.55) 0.06 (0.77) CD19+ Down 0.33 (0.13) -0.19 (0.38) -0.41 (0.05) CD8+ Up 0.09 (0.69) -0.32 (0.14) T reg -0.02 (0.91) NK cell LCA-3 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.13 (0.61) 0.07 (0.80) -0.18 (0.49) -0.20 (0.45) 0.35 (0.16) -0.31 (0.22) CD14+ Up -0.31 (0.23) 0.15 (0.57) -0.36 (0.16) -0.45 (0.07) 0.15 (0.56) CD19+ Up 0.37 (0.14) -0.06 (0.82) 0.29 (0.26) -0.07 (0.78) CD19+ Down 0.20 (0.44) 0.15 (0.56) 0.18 (0.48) CD8+ Up 0.21 (0.42) -0.05 (0.84) T reg 0.19 (0.47) NK cell expression and altered association between up and down- intracellular perforin [34]. In this work we observe an regulated B cell functions would suggest that the function increased expression of the NK cell gene set. Of the 4 of these cells might be compromised in CFS subjects. Cole genes used to capture NK cell function the expression of et al. [28] reported a selective reduction of mature B lym- NKG2A/C (NM 002260) was most increased. The binding phocyte function in subjects who experienced chronic of NKG2A to its natural ligand, human non-classic class I high levels of social isolation including suppression of leukocyte antigen (HLA) E, is known to induce its immu- several transcription factors involved B cell differentiation noreceptor tyrosine-based inhibition motif (ITIM) and such as Ikaros/ZNF1A1. Genes encoding for members of suppress cytotoxic cell effector activity [35]. Moreover the zinc finger protein family were also identified in pre- NKG2A is also known to be co-expressed on activated Th2 vious work by this group as prominent contributors to the but not Th1 lymphocytes [36]. A bias towards Th2-type CFS symptom space [9]. A closer look at the 6 genes that immune response in CFS patients has also been suggested constitute the CD19+ up-regulated set showed that the on the basis of intracellular T cell cytokine profiles by PTPRK and TSPAN3 genes, both associated with immune Skowera et al. [37]. Interestingly this also aligns with cell adhesion and development, were the most sup- altered expression of the PTPRK gene mentioned above as pressed. Down-regulation of PTPRK, a TGF-β target gene, Asano et al. [38] report impaired Th1 function with is known to be down-regulated by the Epstein-Barr virus PTPRK deletion in rats. Therefore our observations sup- (EBV) [29], an infectious agent known to trigger CFS ported findings of increased suppression of cytotoxic [30,31]. Down-regulation of TGF-β has been reported in activity in CFS and hinted at increased Th2 activity though CFS by Tomoda et al. [32]. the latter were not specifically addressed in this analysis. NK cell activity is suppressed in CFS [33] and this Neutrophils for their part are only found at trace and con- decreased cytotoxity has been associated with reduced taminating amounts in most PBMC preparations [39] so Page 9 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 7: Detailed results of gene set correlation r (null probability p ) for LCA-4, 5, 6 classes. a, b a, b LCA-4 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.41 (0.21) -0.02 (0.96) -0.09 (0.80) -0.14 (0.68) 0.00 (0.99) 0.14 (0.67) CD14+ Up -0.05 (0.88) -0.13 (0.71) -0.12 (0.73) -0.01 (0.98) -0.28 (0.41) CD19+ Up -0.21 (0.53) 0.09 (0.78) 0.28 (0.40) 0.13 (0.71) CD19+ Down 0.44 (0.18) 0.05 (0.89) -0.48 (0.13) CD8+ Up 0.19 (0.57) -0.04 (0.91) T reg -0.11 (0.75) NK cell LCA-5 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.44 (0.12) 0.35 (0.22) 0.05 (0.87) 0.32 (0.27) 0.17 (0.56) 0.04 (0.90) CD14+ Up 0.86 (0.00) 0.23 (0.44) -0.40 (0.16) 0.35 (0.21) 0.40 (0.16) CD19+ Up 0.15 (0.60) -0.48 (0.08) 0.35 (0.22) 0.36 (0.20) CD19+ Down -0.16 (0.58) 0.16 (0.60) -0.16 (0.59) CD8+ Up -0.22 (0.45) -0.12 (0.69) T reg 0.07 (0.82) NK cell LCA-6 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.34 (0.30) -0.02 (0.95) 0.20 (0.55) -0.15 (0.65) 0.45 (0.16) 0.23 (0.50) CD14+ Up 0.10 (0.76) 0.57 (0.07) -0.14 (0.69) -0.40 (0.22) 0.20 (0.56) CD19+ Up 0.35 (0.30) -0.33 (0.32) -0.43 (0.18) 0.31 (0.35) CD19+ Down 0.15 (0.67) -0.31 (0.36) 0.32 (0.33) CD8+ Up 0.11 (0.74) -0.46 (0.16) T reg 0.07 (0.84) NK cell SD is standard deviation it is interesting to note that the neutrophil gene set arose (IL-1, IL-6, IL-8, GM-CSF). IL-8 attracts more neutrophils as a core element in the emergence of coordinated and together with GM-CSF causes these to once again immune activity. In particular the CD16+ neutrophil gene degranulate. With the corresponding release of additional set and the CD14+ monocyte gene set shared significant MPO, the cycle starts once again. The TNF-α initiated cas- co-expression. Not only do these arise from the same cade induces IL-6 which is used by B cells for maximum hematopoietic CD34+ progenitor cell [40] but since the antibody secretion usually IgM. In addition to the present immune community is highly integrated the presence or analysis, a preliminary examination of cytokine data col- absence of neutrophils will also be mirrored in the state of lected in the Wichita study pointed to an increase in TNF- the remaining cell population. The CD14+ monocyte set α in CFS subjects (data not shown) as documented previ- also shared significant co-expression with the CD19+ B ously by Moss et al. [42]. cell gene sets. Together this neutrophil-monocyte-B cell immune interaction triad is highly consistent with a In addition to this core network, we also observed that model of chronic inflammation proposed by Lefkowitz CD8+ T cell set expression correlated negatively with that and Lefkowitz [41]. According to this model once an of the NK and CD19+ up-regulated B cell sets. In one pos- event initiates inflammation, neutrophils are among the sible mechanism linking these three cell types, IgG anti- first cells to arrive at the site. They degranulate releasing bodies binding to GD3 on the surface of CD4+ and CD8+ MPO into the microenvironment which together with T cells could elicit signals for proliferation of these subsets iMPO binds to macrophage MMR receptor and induces and expression of the IL-2 receptor CD25. NK cells have release of TNF-α. The latter functions in an autocrine been shown to selectively inhibit this antibody-mediated manner and along with iMPO initiates a cytokine cascade proliferation of CD8+ T cells by Claus et al. [43] perhaps Page 10 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 NF Empiric CFS ISF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell LCA-3 LCA-0 & 2 LCA-1 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell 0.8 LCA-6 LCA- 4 LCA- 5 0.6 CD16+ Down 0.4 CD14+ Up 0.2 CD19+ Up CD19+ Down -0.2 -0.4 CD8+ Up -0.6 T reg -0.8 NK cell -1 a Figure 5 H cu se well as for LCA contro toff sign at maps of gene set c ificance p <0.05 ( o-exp l clasl ression ex ses ( ) for empir LCA-pr 0i, c classes for non-fati esse 2) an d as li d for all LCA disease classes near correl gued ation co (NF) eff co icintro ent r ls, in at cut suffici oefnt fati f signifgue icansymptoms ce p <0.10 ( (ISF) and CFS ) and at a,b a, b a,b Heat maps of gene set co-expression expressed as linear correlation coefficient r at cutoff significance a, b p <0.10 () and at cutoff significance p <0.05 (l) for empiric classes for non-fatigued (NF) controls, insuffi- a,b a,b cient fatigue symptoms (ISF) and CFS as well as for LCA control classes (LCA-0, 2) and for all LCA disease classes. through down-regulation of autologous mixed lym- structed for LCA classes 1 and 5 suggested that B cell phocyte reaction (MLR). This basic analysis of immune involvement in these processes may serve as factor for dis- gene set co-expression points therefore to the existence of criminating between distinct subsets of CFS subjects. immune signaling processes in CFS that adhere to at least one known mechanism of chronic inflammation and sup- Although several very plausible immune response mecha- port possible antibody-mediated NK cell modulation of T nisms were recovered by this analysis it must be empha- cell activity. Furthermore association networks con- sized that the use of discrete gene sets has several Page 11 of 13 (page number not for citation purposes) CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 limitations. In particular it becomes increasingly difficult tion, supervised the analysis and drafted the manuscript. to identify genes that are exclusively or even predomi- All authors read and approved the final manuscript. nantly expressed in specific cell lineages when these share many commonalities of function and goal. This issue was Additional material reflected in by the small size of the gene sets identified in this work from lymphocyte subset expression profiles. An Additional File 1 approach that promises to be more robust and more Supplementary_data_Panther_Gene_List_Annotation. Additional data revealing still involves the direct use of the genome-wide file 1 is a set of tables listing the genes used in each gene set along with expression for these cell populations. This remains an the functional annotation available in the PANTHER classification sys- active area of research [44]. However, even this simple tem. This file also contains the median values and standard deviations within each illness class of the Ln-transformed gene expression for each analysis points to dramatic differences in immune net- gene. work topology and cell signaling in CFS and we expect Click here for file these differences to be largely conserved in more elaborate [http://www.biomedcentral.com/content/supplementary/1744- analyses. Furthermore the methodology outlined and 9081-4-44-S1.xls] issues raised in this work demonstrate the importance of developing approaches that effectively integrate flow cytometry with cytokine and gene expression profiling. In particular it underscores the importance of looking Acknowledgements beyond differential expression of individual components The authors would like to thank Dr. WC Reeves and the members of his towards changes in their patterns of coordinated activity CFS Research Group at the Centers for Disease Control and Prevention in Atlanta for their many helpful discussions. This work was supported by and formally recognizing the network properties of the research grant # N031000099 from the University of Alberta's Faculty of immune system. Medicine and Dentistry. List of abbreviations References CD: cluster of differentiation; CDC: Centers for disease 1. Reynolds KJ, Vernon SD, Bouchery E, Reeves WC: The economic Control and Prevention; cDNA: complementary DNA; impact of chronic fatigue syndrome. Cost Eff Resour Alloc 2(1):4. 2004 Jun 21 CFS: Chronic Fatigue Syndrome; DPP-IV: dipeptidyl 2. Klimas N, Salvato F, Morgan R, Fletcher MA: Immunologic abnor- peptidase-4; EBV: Epstein-Barr virus; FoxP3: forkhead box malities in chronic fatigue syndrome. J Clin Microbiol 1990, P3 gene; GD3: ganglioside D3; GM-CSF: granulocyte mac- 28(6):1403-1410. 3. Straus SE, Fritz S, Dale JK, Gould B, Strober W: Lymphocyte phe- rophage colony-stimulating factor; HCMV: human notyping and function in chronic fatigue syndrome. J of Clin cytomegalovirus; Ig: immunoglobulin; iMPO: inactive Immunol 1993, 13(1):30-40. myeloperoxidase; IL: interleukin; ISF: insufficient symp- 4. Tirelli U, Bernardi D, Improta S, Pinto A: Immunologic abnormal- ities in chronic fatigue syndrome. J Chronic Fatigue Syndrome toms of fatigue; MFI: Multidimensional Fatigue Inven- 1996, 2(1):85-96. tory; MLR: mixed lymphocyte reaction; MMR: 5. Yan S, Marguet D, Dobers J, Reutter W, Fan H: Deficiency of CD26 results in a change of cytokine and immunoglobulin secre- macrophage; mannose receptor; NF: non-fatigued; NK: tion after stimulation by pokeweed mitogen. Eur J Immunol natural killer cell; NKG2A/C: killer cell lectin-like receptor 2003, 33(6):1519-27. subfamily C, member 1/2 or KLRC1/C2; LCA: latent class 6. Boonacker E, Van Noorden CJ: The multifunctional or moon- lighting protein CD26/DPPIV. Eur J Cell Biol 2003, 82(2):53-73. analysis; MPO: myeloperoxidase; PANTHER: Protein 7. Robertson MJ, Schacterle RS, Mackin GA, Wilson SN, Bloomingdale ANalysis THrough Evolutionary Relationships; PBMC: KL, Ritz J, Komaroff AL: Lymphocyte subset differences in peripheral blood mononuclear cell; PTPRK: protein tyro- patients with chronic fatigue syndrome, multiple sclerosis and major depression. Clin Exp Immunol 2005, 141(2):326-32. sine phosphatase, receptor type, K; SF-36: Short Form 36; 8. Racciatti D, Dalessandro M, Delle Donne L, Falasca K, Zingariello P, TGF: transforming growth factor; TNF: tumor necrosis fac- Paganelli R, Pizzigallo E, Vecchiet J: Study of immune alteration in patients with chronic fatigue syndrome with different etiolo- tor; Th: T helper cell; TSPAN3: tetraspanin 3. gies. J Immunopath Pharm 2004, 17(2):57-62. 9. Broderick G, Craddock RC, Whistler T, Taylor R, Klimas N, Unger ER: Identifying illness parameters in fatiguing syndromes Competing interests using classical projection methods. Pharmacogenomics 2006, The authors declare that they have no competing interests. 7(3):407-19. 10. Carmel L, Efroni S, White PD, Aslakson E, Vollmer-Conna U, Rajeevan MS: Gene expression profile of empirically deline- Authors' contributions ated classes of unexplained chronic fatigue. Pharmacogenomics ALA carried out statistical analysis of gene expression pro- 2006, 7(3):375-86. files, gene set design and significance analysis and helped 11. Whistler T, Jones JF, Unger ER, Vernon SD: Exercise responsive genes measured in peripheral blood of women with chronic draft the manuscript. CB participated in the design of the fatigue syndrome and matched control subjects. BMC Physiol study and assisted with the statistical analysis and helped 5(1):5. 2005 Mar 24; draft the manuscript. SDV helped conceive of the study, its 12. Siegel S, Antoni M, Fletcher M, Maher K, Segota M, Klimas N: Impaired natural immunity, cognitive dysfunction, and phys- design and coordination and helped to draft the manu- ical symptoms in patients with chronic fatigue syndrome: script. GB conceived of the study, its design and coordina- Page 12 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 preliminary evidence for a subgroup? J Psychosom Res 2006, patients with chronic fatigue syndrome and normal controls. 60:559-566. Psychiatry Res 2005, 134(1):101-104. 13. Segal E, Friedman N, Koller D, Regev A: A module map showing 33. Fletcher MA, Maher K, Klimas NG: Natural killer cell function in conditional activity of expression modules in cancer. Nat chronic fatigue syndrome. Clin Applied Immunol Rev 2002, Genet 2004, 36(10):1090-8. 2:129-139. 14. Dinu I, Potter JD, Mueller T, Liu Q, Adewale AJ, Jhangri GS, Einecke 34. Maher KJ, Klimas NG, Fletcher MA: Chronic fatigue syndrome is G, Famulski KS, Halloran P, Yasui Y: Improving gene set analysis associated with diminished intracellular perforin. Clin Exp of microarray data by SAM-GS. BMC Bioinformatics 8(1):242. Immunol 2005, 142(3):505-511. 2007 Jul 5; 35. Kabat J, Borrego F, Brooks A, Coligan JE: Role that each NKG2A 15. Efron B, Tibshirani R: On testing the significance of sets of immunoreceptor tyrosine-based inhibitory motif plays in genes. Ann Appl Statist 2007, 1(1):107-129. mediating the human CD94/NKG2A inhibitory signal. J 16. Vernon SD, Reeves WC: The challenge of integrating disparate Immunol 2002, 169(4):1948-1958. high-content data: epidemiologic, clinical, and laboratory 36. Freishtat RJ, Mitchell LW, Ghimbovschi SD, Meyers SB, Hoffman EP: data collected during an in-hospital study of chronic fatigue NKG2A and CD56 are coexpressed on activated TH2 but syndrome. Pharmacogenomics 2006, 7(3):345-354. not TH1 lymphocytes. Hum Immunol 2005, 66(12):1223-1234. 17. Reeves WC, Wagner D, Nisenbaum R, Jones JF, Gurbaxani B, Solo- 37. Skowera A, Cleare A, Blair D, Bevis L, Wessely SC, Peakman M: High mon L, Papanicolaou DA, Unger ER, Vernon SD, Heim C: Chronic levels of type 2 cytokine-producing cells in chronic fatigue fatigue syndrome – A clinically empirical approach to its def- syndrome. Clin Exp Immunol 2004, 135(2):294-302. inition and study. BMC Medicine 2005, 3:8. 38. Asano A, Tsubomatsu K, Jung CG, Sasaki N, Agui T: A deletion 18. Wagner D, Nisenbaum R, Heim C, Jones JF, Unger ER, Reeves WC: mutation of the protein tyrosine phosphatase kappa (Ptprk) Psychometric properties of the CDC Symptom Inventory gene is responsible for T-helper immunodeficiency (thid) in for assessment of chronic fatigue syndrome. Popul Health Metr the LEC rat. Mamm Genome 2007, 18(11):779-786. Epub 2007 Oct 2005, 3:8. 2 19. Vollmer-Conna U, Aslakson E, White PD: An empirical delinea- 39. Jison ML, Munson PJ, Barb JJ, Suffredini AF, Talwar S, Logun C, Ragha- tion of the heterogeneity of chronic unexplained fatigue in vachari N, Beigel JH, Shelhamer JH, Danner RL, Gladwin MT: Blood women. Pharmacogenomics 2006, 7(3):355-364. mononuclear cell gene expression profiles characterize the 20. Einecke G, Broderick G, Sis B, Halloran PF: Early loss of renal tran- oxidant, hemolytic, and inflammatory stress of sickle cell dis- scripts in kidney allografts: relationship to morphologic ease. Blood 104(1):270-80. 2004 Jul 1; changes and alloimmune effector mechanisms. Am J Trans- 40. Ferrari F, Bortoluzzi S, Coppe A, Basso D, Bicciato S, Zini R, Gemelli plant 2007, 7(5):1121-1130. C, Danieli GA, Ferrari S: Genomic expression during human 21. Famulski KS, Broderick G, Einecke G, Hay K, Cruz J, Sis B, Mengel M, myelopoiesis. BMC Genomics 8:264. 2007 Aug 3; Halloran PF: Transcriptome analysis reveals heterogeneity in 41. Lefkowitz DL, Lefkowitz SS: Macrophage-neutrophil interac- the injury response of kidney transplants. Am J Transplant 2007, tion: A paradigm for chronic inflammation revisited. Immunol 7:2483-95. Cell Biol 2001, 79:502-506. 22. Mueller TF, Einecke G, Reeve J, Sis B, Mengel M, Jhangri GS, Bunnag 42. Moss RB, Mercandetti A, Vojdani A: TNF-alpha and chronic S, Cruz J, Wishart D, Meng C, Broderick G, Kaplan B, Halloran PF: fatigue syndrome. J Clin Immunol 1999, 19(5):314-316. Microarray Analysis of Rejection in Human Kidney Trans- 43. Claus C, Schlaak J, Dithnayer M, Meyer zum K-H, Dippold B, Dippold plants Using Pathogenesis-based Transcripts Sets. Am J Trans- W: Inhibition of anti-GD3-ganglioside antibody-induced pro- plant 2007, 7(12):2712-2722. liferation of human CD8+ Tcells by CD16+ natural killer 23. Lyons PA, Koukoulaki M, Hatton A, Doggett K, Woffendin HB, cells. Eur J Immunol 1994, 24:1208-1212. Chaudhry AN, Smith KG: Microarray analysis of human leuco- 44. Gosink MM, Petrie HT, Tsinoremas NF: Electronically subtract- cyte subsets: the advantages of positive selection and rapid ing expression patterns from a mixed cell population. Bioin- purification. BMC Genomics 2007, 8:64. formatics 2007, 23(24):3328-3334. Epub 2007 Oct 22 24. Thomas PD, Kejariwal A, Campbell MJ, Mi H, Diemer K, Guo N, Ladunga I, Ulitsky-Lazareva B, Muruganujan A, Rabkin S, Vandergriff JA, Doremieux O: PANTHER: a browsable database of gene products organized by biological function, using curated pro- tein family and subfamily classification. Nucleic Acids Res 2003, 31(1):334-341. 25. PANTHER Classification System [http://www.pantherdb.org] 26. Bernstein S, Bernstein R: Schaum's Outline of Theory and Problems of Elements of Statistics II New York (NY): McGraw-Hill; 1999. 27. Elo LL, Järvenpää H, Orešiè M, Lahesmaa R, Aittokallio T: System- atic construction of gene coexpression networks with appli- cations to human T helper cell differentiation process. Bioinformatics 2007, 23(16):2096-2103. 28. Cole SW, Hawkley LC, Arevalo JM, Sung CY, Rose RM, Cacioppo JT: Social regulation of gene expression in human leukocytes. Genome Biology 2007, 8(9):R189. 29. Flavell JR, Baumforth KR, Wood VH, Davies GL, Wei W, Reynolds GM, Morgan S, Boyce A, Kelly GL, Young LS, Murray PG: Downreg- Publish with Bio Med Central and every ulation of the TGF-Beta target gene, PTPRK, by the epstein- scientist can read your work free of charge barr virus encoded EBNA1 contributes to the growth and survival of Hodgkin's lymphoma cells. Blood 2008, "BioMed Central will be the most significant development for 111(1):292-301. 2007 Aug 24; disseminating the results of biomedical researc h in our lifetime." 30. Lerner AM, Beqaj SH, Deeter RG, Fitzgerald JT: IgM serum anti- Sir Paul Nurse, Cancer Research UK bodies to Epstein-Barr virus are uniquely present in a subset of patients with the chronic fatigue syndrome. In Vivo 2004, Your research papers will be: 18(2):101-106. available free of charge to the entire biomedical community 31. Hickie I, Davenport T, Wakefield D, Vollmer-Conna U, Cameron B, Vernon SD, Reeves WC, Lloyd A, Dubbo Infection Outcomes Study peer reviewed and published immediately upon acceptance Group: Post-infective and chronic fatigue syndromes precipi- cited in PubMed and archived on PubMed Central tated by viral and non-viral pathogens: prospective cohort study. BMJ 333(7568):575. 2006 Sep 16; yours — you keep the copyright 32. Tomoda A, Joudoi T, Rabab el-M, Matsumoto T, Park TH, Miike T: BioMedcentral Cytokine production and modulation: comparison of Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 13 of 13 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral and Brain Functions Springer Journals

Evidence of inflammatory immune signaling in chronic fatigue syndrome: A pilot study of gene expression in peripheral blood

Download PDF
13 pages

Loading next page...
 
/lp/springer-journals/evidence-of-inflammatory-immune-signaling-in-chronic-fatigue-syndrome-IeN4WcNiLC

References (44)

Publisher
Springer Journals
Copyright
Copyright © 2008 by Aspler et al; licensee BioMed Central Ltd.
Subject
Biomedicine; Neurosciences; Neurology; Behavioral Therapy; Psychiatry
eISSN
1744-9081
DOI
10.1186/1744-9081-4-44
pmid
18822143
Publisher site
See Article on Publisher Site

Abstract

Background: Genomic profiling of peripheral blood reveals altered immunity in chronic fatigue syndrome (CFS) however interpretation remains challenging without immune demographic context. The object of this work is to identify modulation of specific immune functional components and restructuring of co-expression networks characteristic of CFS using the quantitative genomics of peripheral blood. Methods: Gene sets were constructed a priori for CD4+ T cells, CD8+ T cells, CD19+ B cells, CD14+ monocytes and CD16+ neutrophils from published data. A group of 111 women were classified using empiric case definition (U.S. Centers for Disease Control and Prevention) and unsupervised latent cluster analysis (LCA). Microarray profiles of peripheral blood were analyzed for expression of leukocyte-specific gene sets and characteristic changes in co-expression identified from topological evaluation of linear correlation networks. Results: Median expression for a set of 6 genes preferentially up-regulated in CD19+ B cells was significantly lower in CFS (p = 0.01) due mainly to PTPRK and TSPAN3 expression. Although no other gene set was differentially expressed at p < 0.05, patterns of co-expression in each group differed markedly. Significant co-expression of CD14+ monocyte with CD16+ neutrophil (p = 0.01) and CD19+ B cell sets (p = 0.00) characterized CFS and fatigue phenotype groups. Also in CFS was a significant negative correlation between CD8+ and both CD19+ up-regulated (p = 0.02) and NK gene sets (p = 0.08). These patterns were absent in controls. Conclusion: Dissection of blood microarray profiles points to B cell dysfunction with coordinated immune activation supporting persistent inflammation and antibody-mediated NK cell modulation of T cell activity. This has clinical implications as the CD19+ genes identified could provide robust and biologically meaningful basis for the early detection and unambiguous phenotyping of CFS. cell dysfunction and activation has been demonstrated in Background Chronic fatigue syndrome (CFS) is estimated to cost the CFS by several groups [2-4]. Though similar in terms of American economy over $9 billion each year in lost pro- broad lymphocyte classes CFS and non-fatigued subjects ductivity [1]. Among other components chronic immune can be readily distinguished when specific immune cell Page 1 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 subsets are examined. For example Klimas et al. [2] report a tion in CFS from gene expression profiles of mixed lym- significant expansion CD26+ (DPP-IV) activated T cells in phocyte populations. In particular we construct gene sets CFS subjects. This multifunctional molecule plays a major that capture elements of abundance and activity assigna- role in the regulation, development, maturation and migra- ble to specific immune cell subsets thereby facilitating tion of T helper (Th) and natural killer (NK) cells as well as direct integration with flow cytometry results. Data from in B cell immunoglobulin switching [5]. Moreover abnor- a large population-based study of CFS [16] is then exam- mal expression of CD26+ is found in autoimmune diseases ined for changes of immune set expression across two sep- [6]. More recently CFS patients were also reported to have arate CFS classification approaches. In addition, patterns significantly fewer CD3+/CD25- T cells and significantly of coordinated expression linking these immune sets were more CD20+/CD5+ B cells [7], a subset associated with investigated using simple correlation networks. These net- auto-antibodies. Significantly fewer CD56+ NK cells were works were examined for shifts in topology and point to also observed in recent work by Racciatti et al. [8]. Though patterns of immune signaling in CFS that are consistent important, flow cytometry results such as these leave many with chronic inflammation. These observations could questions regarding cellular state unanswered. Microarray constitute a signature of CFS or a component thereof. profiling of gene expression on the other hand offers a glimpse of pathway activation in disease pathogenesis at Methods molecular resolution. Microarray analysis of cDNA profiles Subjects and diagnostic classes in peripheral blood mononuclear cells (PBMC) have Recently a dataset for a 2-day in-hospital study of CFS in revealed altered expression in CFS of several immune genes the general population of Wichita Kansas was made avail- [9,10] involved in response to oxidative stress, NK cell able [16]. Referred to as the Wichita Clinical study, this activity and elements of antigen processing. Instability in investigation included a highly comprehensive spectrum immune response and restructuring of immune cell signal- of detailed clinical and laboratory measures and PBMC ing under exercise challenge has also been observed [11]. expression profiles for 20,000 genes. From this dataset a Unfortunately microarray profiling is commonly per- final analysis group of 111 female subjects was obtained formed on mixed cell populations producing an average by excluding the few male subjects and subjects with con- profile from which it is very difficult to dissect the contribu- founding medical or psychiatric conditions. Subjects in tions of relative cell abundance, cell activation state and this dataset were classified as CFS using the CDC Symp- cell-cell signaling. More importantly, this averaging can tom Inventory, Multidimensional Fatigue Inventory obscure significant changes in the state of minority cell sub- (MFI) and Short Form 36 (SF-36) instruments [17,18]. populations. This classification will be referred to as "empiric" and resulted in 39 CFS, 37 non- fatigued (NF), and 35 subjects These challenges notwithstanding, a review of this evidence with insufficient symptoms or fatigue (ISF). A second clas- strongly suggests that CFS pathogenesis is likely to include sification proposed by Vollmer-Conna and colleagues a characteristic immunologic component in at least one [19] used latent class analysis (LCA) of 440 clinical and subset of the patient population [12]. However the exact biological measurements to delimit 5 fatigue classes, a nature of this immunologic component remains the object non-fatigued class and 2 unassigned individuals. Obese of considerable debate at least in part because of an inabil- subjects with prominent post-exercise fatigue, hypnoea ity to cast gene expression profiles in the useful context of and disturbed sleep formed Class 1. Reasonably healthy immune cell demographics. In an attempt to address this subjects with few symptoms, low depression scores and issue methods have been proposed to dissect global gene good sleep composed Class 2. Subjects in Class 3 resem- expression profiles into discrete elements assignable to bio- bled those in class 1 but also displayed low heart rate var- logic processes [13-15]. The assignment of genes to discrete iability during sleep and low 24-hour cortisol levels. Class modules or sets has been successful in several respects. A 4 was populated with healthier, less depressed individuals first contribution involves simply reducing the dimension- having restful sleep but suffering muscle pain. Finally ality of >55,000 gene expression measures to that of say 10 Classes 5 and 6 both captured less obese but highly symp- or so gene sets. The interpretability of results is further tomatic and depressed individuals with prominent post- enhanced by associating sets with basic cellular functions. exercise fatigue. Individuals in Class 6 also displayed dis- Finally the numerical robustness is greatly improved turbed sleep with low heart rate variability and low corti- through the averaging of changes in expression over many sol. The patient demographics for each of these genes. In addition gene sets are transportable across micro- classification systems are summarized in Table 1 and the array platforms making it possible to compare studies alignment between these systems is described in Table 2. based on different technologies. The collection and processing of PBMCs including hybrid- ization to MWG microarrays (MWG Biotech, Ebersberg, In this work we explore the use of discrete gene sets in Germany) are described in Vernon and Reeves [16]. extracting useful information regarding immune dysfunc- Details of the microarray data preprocessing including Page 2 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 1: Demographic data for 111 subjects from the Wichita clinical study Empiric Classification LCA Classification CFS IFS Controls LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 LCA-0/2 (n = 39) (n = 35) (n = 37) (n = 23) (n = 17) (n = 11) (n = 14) (n = 11) (n = 35) Mean Age (SD) 51.4 (8.2) 50.3 (8.2) 51.6 (9.0) 50.9 (7.6) 54.7 (5) 44.4 (8.7) 48.2 (10.2) 55.8 (3.4) 51.3 (9) Mean Years Ill (SD) 16.7 (11.0) 14.4 (10.0) 2.8 (5.0) 15.5 (10.7) 11.3 (4) 16.0 (12.3) 16.8 (10.5) 16.4 11.2 14.3 (12.9) Race [n (%) ] White 35 (90.0) 32 (91.4) 36 (97.3) 21 (91) 17 (100) 10 (90.9) 11 (78.6) 11 (100) 33 (94.3) Black 1 (2.6) 3 (8.6) 1 (2.7) 1 (4.4) 0 (0) 1 (9.1) 1 (7.1) 0 (0) 2 (5.7) Multiple Race 2 (5.1) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 2 (14.3) 0 (0) 0 (0) Other 1 (2.6) 0 (0) 0 (0) 1 (4.4) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Onset Type Gradual 32 (82.0) 28 (80.0) 10 (27.0) 18 (78.3) 14 (82.4) 9 (81.8) 11 (78.6) 10 (90.9) 8 (22.9) Sudden 6 (15.4) 3 (8.6) 0 (0.0) 3 (13) 1 (5.9) 1 (9.1) 3 (21.4) 1 (9.1) 0 (0.0 Undetermined 1 (2.6) 4 (11.4) 27 (73.0) 2 (8.7) 2 (11.8) 1 (9.1) 0 (0) 0 (0) 27 (77.1) BMI [n (%) ] <25 5 (13.8) 10 (28.6) 7 (18.9) 0 (0) 4 (23.5) 8 (72.7) 3 (21.4) 3 (27.3) 4 (11.4) 25–30 20 (51.3) 14 (40) 18 (48.7) 9 (39.1) 6 (35.3) 3 (27.3) 9 (64.3) 7 (63.6) 18 (51.4) >30 14 (35.9) 11 (31.4) 12 (32.4) 14 (60.9) 7 (41.2) 0 (0) 2 (14.3) 1 (9.1) 13 (37.1) SD is standard deviation normalization, outlier detection and false discovery cor- MWG microarrays used in the Wichita Clinical study. We rection are available in Broderick et al. [9]. further dissected these subset-specific profiles into dis- crete non-overlapping sets composed of genes at least 2- Gene set development fold up-regulated or 2-fold down-regulated preferentially Extracting elements that represent the abundance and in each cell lineage. An additional gene set was defined for activity of a specific leukocyte subset was approached by NK cell activity and regulatory T cell activity was estimated identifying discrete sets of genes that are uniquely or pre- from the expression of the FoxP3 gene (AF277993). Indi- dominantly expressed in a given cell type [20-22]. Cur- vidual MWG gene probes belonging to each immune gene rently discrete gene sets offer the simplest and most set as well as NCBI gene annotation and PANTHER func- immediately accessible method for analysis across micro- tional annotation [24,25] are listed in the supplementary array technological platforms. We constructed a number data file [Additional file 1]. of gene sets a priori for CD4+ T cells, CD8+ T cells, CD19+ Statistical analysis B cells, CD14+ monocytes and CD16+ neutrophils using data collected on Affymetrix microarrays (Affymetrix, The aggregate expression G of each gene set a was com- Santa Clara, CA, USA) by Lyons et al. [23]. Of the 12,022 puted as the average of the Ln-transformed expression genes surveyed, 2,641 were differentially expressed Ln(g ) of each gene i across the k member genes in the set i, a (Equation 1). In a first level of analysis a classical Wil- between individual lymphocyte subsets. Of these original 2,641 distinguishing genes, 268 were present on the coxon non-parametric test was used to evaluate the differ- Table 2: A cross-reference of systems for diagnostic assignment Empiric Classification (CFS research case definition) CFS ISF Controls LCA Category LCA Class Description (n = 39) (n = 35) (n = 37) n(%) n(%) n(%) Controls (0–2) Well (n = 33) or Unassigned (n = 2) 1 (2) 0 (0) 34 (91) 1 Obese hypnoea (n = 23) 15 (38) 8 (23) 0 (0) 3 Obese hypnoea and stressed (n = 17) 5(13) 11 (31)1 (3) 4 Interoception – muscle pain (n = 11) 1(3) 9 (26) 1(3) 5 Interoception depression (n = 14) 10 (26) 4 (11) 0 (0) 6 Multisymptomatic, depressed, stressed (n = 11) 7 (18) 3 (9) 1 (3) Page 3 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 ential expression of immune gene sets for both Results classification systems. As suggested by Efron and Tib- Alignment of empiric and LCA classifications A cross tabulation of the empiric classification and LCA shirani [15] the performance of these gene sets was also compared to that obtained with randomly populated sets classification is presented in Table 2. There was good of the same size. A null distribution was computed from alignment of non-fatigued subjects with 90% of empiric the analysis of 1000 instances of random gene selections NF controls residing in LCA classes 0 (Well) and 2 (Unas- and 1000 random permutations of the diagnostic labels. signed). Together LCA classes 1 (40%) and 5 (26%) con- tained two thirds of the subjects assigned to the empiric CFS class. However ISF subjects were distributed almost Ln(g ) equally across LCA classes 1 (23%), 3 (31%), and 4 ia , (1) i =1 (26%). Conversely most LCA class 3 and 4 subjects were G = identified as ISF and most subjects in LCA classes 1, 5 and 6 were assigned an empiric CFS classification. To examine the patterns of association linking immune gene sets simple linear association networks were con- Differential expression of a priori defined immune cell structed using the Pearson correlation coefficient r as a, b gene sets the metric describing similarity in the expression of gene In a first level of analysis the differential expression of set a with that of gene set b. Statistical significance of cor- immune gene sets across disease phenotypes and control relation was assessed using the t statistic in Equation a, b groups for both classification systems was evaluated. (2). This statistic has a Student's t-distribution with Results in Table 3 show that the median expression of the degrees of freedom n-2 under the null hypothesis of no CD19+ B cell up-regulated gene set was significantly lower correlation [26], where n is the number of microarray in CFS (p = 0.01) and ISF (p = 0.05) subjects when com- measurements. C is the covariance in the expression of a, b pared to the NF group. Expression of this gene set was also gene set a with gene set b and E() is the expected value significantly repressed in LCA class 3 (p = 0.04) and mar- operator or the mean. ginally so in LCA class 5 (p = 0.09) when compared to control subjects in LCA classes 0 and 2. Recall that 11 of () n −2 tr = 17 cases in LCA class 3 were also designated ISF. Similarly ab,, ab (2) (1−r ) 10 of the 14 LCA class 5 cases were designated CFS. NK ab , gene set expression was marginally increased in the CFS group (p = 0.07). Though not significant the null proba- Where, bility for NK cell expression was lowest among the LCA classes for LCA-3 (p = 0.11). Finally expression of the T ab , regulatory set (FoxP3) was marginally repressed in LCA r== ;CE[(G−E(G ))⋅ (G−E(G ))] ab,, ab a a b b C C aa,, bb class 1 (p = 0.09) which contained 40% of the CFS sub- jects though no significant difference was found for the A cutoff for the resulting probability p (t > t ), above a, b a, b larger CFS group (p = 0.31). which we accept the null hypothesis, can be established in a variety of ways [27]. It should be noted however that The performance of these gene sets was compared to that these require specific assumptions regarding network obtained with randomly populated sets of the same size as topology such as network edge sparseness or the appear- well as by random classification assignment of each sub- ance of highly cliquish disconnected sub-networks. As ject. Null distribution results indicated that both the such they are generally more relevant to the study of large CD19+ B cell (up-regulated) and NK cell gene sets per- networks. Instead we examined the dependency of the formed significantly better than random sets of equivalent network size S, or the sum of the edge weights w on the a, b size in discriminating CFS from NF (p < 0.05) (Figure 1). choice of threshold p-value (Equation 3). We compared Performance of the T regulatory gene set (FoxP3) was mar- curves obtained for NF and CFS networks, identifying ginal at best (p~0.15) in terms of uniqueness in differen- threshold p-values where networks differed primarily in tial expression. In addition a detailed analysis of structure from those where they differed in both structure individual genes in the CD19+ up-regulated set indicated and size. that no single gene was differentially expressed even though the parent set was expressed at the p = 0.01 level. M−1 M This reaffirms that high levels of measurement noise can Sw = ab , ∑ ∑ be effectively managed by aggregating genes into biologi- a =1 ba > (3) cally relevant sets. Details of this analysis are listed in where wr=≤ if p p ab,, ab ab, threshold Table 4 and illustrated graphically in Figure 2 and w = 0 if pp > ab , ab , threshold Figure 3. Page 4 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 3: Changes in median expression and corresponding null probability values () for pair-wise comparison of disease classes and the non-fatigued control group under both classification systems 3-Class (NF Controls) 7-Class (Controls = LCA-0 U LCA-2) Cell Type Expression Number of ISF CFS LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 Level +/- genes CD8 T cells Up-regulated 5 0.06 (0.43) 0.01 (0.75) 0.06 (0.52) 0.00 (0.64) 0.04 (0.24) 0.01 (0.80) 0.23 (0.27) Down- regulated CD14 Up-regulated 78 0.04 (0.29) 0.05 (0.83) 0.02 (0.62) 0.11 (0.15) 0.09 (0.42) 0.02 (0.94) 0.05 (0.33) Monocytes Down- regulated CD16 Up-regulated Neutrophils Down- 185 0.02 (0.20) 0.02 (0.25) 0.02 (0.29) 0.00 (0.65) 0.03 (0.25) 0.02 (0.50) 0.04 (0.30) regulated CD19 B cells Up-regulated 6 -0.17 (0.05) -0.28 (0.01) -0.27 (0.29) -0.31 (0.04) -0.12 (0.13) -0.25 (0.09) -0.17 (0.14) Down- 2 -0.22 (0.33) -0.22 (0.29) -0.18 (0.19) 0.07 (0.71) 0.10 (0.76) -0.08 (0.78) -0.08 (0.16) regulated CD4/8/25 T NA 1 -0.23 (0.19) -0.23 (0.31) -0.35 (0.09) 0.19 (0.37) -0.25 (0.11) -0.17 (0.32) 0.28 (0.40) reg cells NK cells NA 4 0.09 (0.92) 0.30 (0.07) 0.19 (0.20) 0.32 (0.11) 0.12 (0.68) 0.12 (0.45) -0.01 (0.57) Emergence of characteristic patterns of association the patterns of association linking immune gene sets sim- between immune gene sets in CFS ple linear correlation networks were constructed. Results Conventional analysis of microarray data remains focused in Figure 4 show network size for the empiric NF and CFS on the detection of differentially expressed genes or gene classes as a function of cutoff p-value for edge weight sig- sets. However it is important to realize that genes nificance. Both NF and CFS networks were identical in expressed at similar levels across patient groups may still overall size at a cutoff p-value of 0.05. Changes at this play an important role in the disease process. To examine level of edge weight significance consisted therefore of a Gene Set CD19+ Up Gene Set NK aggregated 1 1 Random Gene Set 0.9 0.9 Random Diagnostic Label delta median = -0.3367 0.8 0.8 0.7 0.7 0.6 0.6 P(x>0.2790) ~ 0.025 0.5 0.5 P(x<-0.3367) ~ 0.006 0.4 0.4 0.3 0.3 0.2 0.2 Random Gene Set delta median = 0.2790 0.1 Random Diagnostic Label 0.1 0 0 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Delta Median CFS - NF Random Permutations Delta Median CFS - NF Random Permutations (A) (B) Cum cell Figure 1 up-regulated gene set ulative probability plot anof d thΔ d e NK cell gene set ifferential expression of CFS versus NF for random gene sets similar in size to the CD19+ B Cumulative probability plot of Δ differential expression of CFS versus NF for random gene sets similar in size to the CD19+ B cell up-regulated gene set and the NK cell gene set. Page 5 of 13 (page number not for citation purposes) Cumulative Probability (1000 permutations) Cumulative Probability (1000 permutations) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 4: Changes in median expression and corresponding null probability values () for pair-wise comparison of disease classes and the non-fatigued control group for each individual gene in the CD19+ B cell Up-regulated gene set 3-Class (NF Controls) 7-Class (Controls = LCA-0 U LCA-2) Gene ISF CFS LCA-1 LCA-3 LCA-4 LCA-5 LCA-6 SP140 -0.03 (0.60) 0.02 (0.49) -0.07 (0.43) 0.10 (0.60) 0.01 (0.72) -0.08 (0.29) 0.07 (0.92) CD22 -0.16 (0.10) -0.09 (0.31) -0.14 (0.17) -0.58 (0.02) -0.30 (0.04) 0.02 (0.35) 0.05 (0.70) QRSL1 -0.03 (0.95) -0.21 (0.91) 0.29 (0.28) -0.10 (0.70) 0.23 (0.50) -0.16 (0.89) -0.49 (0.42) PTPRK -0.12 (0.55) -0.21 (0.18) -0.12 (0.76) -0.07 (0.82) -0.22 (0.68) -0.20 (0.17) -0.07 (0.64) P2RY10 0.01 (0.21) -0.02 (0.51) 0.08 (0.99) -0.08 (0.06) 0.28 (0.82) 0.11 (0.94) 0.20 (0.33) TSPAN3 -0.10 (0.54) -0.20 (0.19) -0.24 (0.39) -0.17 (0.83) 0.13 (0.38) -0.08 (0.83) -0.02 (0.82) re-organization of edges only. The curves in Figure 4 the CFS network. Topologies emerging at both p < 0.05 diverge at p~0.10 and maintain a similar offset form one and p < 0.10 thresholds were examined as they contain another as p-value increases. As a result comparisons of complementary information. Detailed results of pair-wise network topology conducted at the p < 0.10 level included correlation between gene sets may be found in Table 5 for edge re-assignment as well as the addition of new edges to Gene SP140 Gen Gene CD22 e CD22 Gene QRSL1 12 12 p=0.4863 p=0.3060 11 11 10 10 9 9 8 8 7 7 p=0.9131 6 6 CFS ISF NF CFS ISF NF CFS ISF NF Gene PTPRK Gene P2Y10 Gene TSPAN3 12 12 p=0.1939 p=0.5127 11 11 p=0.1800 10 10 9 9 8 8 8 7 7 6 6 CFS ISF NF CFS ISF NF CFS ISF NF Box classes Figure 2 and whisker plot for the expression of each gene in the CD19+ up-regulated gene set in each of the 3 empiric illness Box and whisker plot for the expression of each gene in the CD19+ up-regulated gene set in each of the 3 empiric illness classes. Boxes indicate the lower quartile, median and upper quartile values. Whiskers are located at extreme values within 1.5 times the inter-quartile range from the ends of each box. Outliers are displayed with a red '+'. Each plot is annotated with the null probability for the difference in median expression between the NF and CFS subject groups. Page 6 of 13 (page number not for citation purposes) Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Ln Transformed Gene Expression Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 1.4 Gene Set CD19+ Up-regulated NF controls 9.5 CFS 1.2 p=0.0100 9.0 0.8 8.5 0.6 8.0 0.4 0.2 7.5 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 7.0 Significance cutoff p-value Threshold p-value Ne weights (Equ value for the empiric Figure 4 twork size ation 3) and plotte S defineNF d as and CFS classes the sum o d as a function f all networ of cutoff p- k edge 6.5 Network size S defined as the sum of all network CFS ISF NF edge weights (Equation 3) and plotted as a function of cutoff p-value for the empiric NF and CFS classes. B regulated gene set Figure 3 ox and whisker plin each ot for the of the 3 empiri expression of the CD19 c illness classes + up- Box and whisker plot for the expression of the CD19+ up-regulated gene set in each of the 3 empiric trophil gene set were completely absent in NF even at the illness classes. Boxes indicate the lower quartile, median p < 0.10 level. Also apparent in CFS was the emergence of and upper quartile values. Whiskers are located at extreme values within 1.5 times the inter-quartile range from the ends a significant negative correlation between the expression of each box. Outliers are displayed with a red '+'. The plot is of CD8+ and CD19+ up-regulated gene sets (p = 0.02). annotated with the null probability for the difference in Moreover CD19+ B cells appeared altered with up and median expression between the NF and CFS subject groups. down-regulated sets no longer maintaining a strong direct correlation in ISF or CFS. Interaction with NK cell gene set expression was also a distinguishing feature in particular the empiric classification system and Tables 6 and 7 for for ISF. Instead of appearing as a transitional state the LCA classification system. between NF and CFS, ISF exhibited a distinct co-expres- sion pattern characterized by a significant interaction of Heat maps depicting the edge weights ra, b linking gene NK cell and monocyte gene sets (p < 0.05). Contrary to sets are presented in Figure 5 for both empiric and LCA NF, the NK and CD19+ down-regulated gene sets corre- classes. The LCA control classes 0 and 2 exhibited a pat- lated negatively (p < 0.10) in ISF. tern of gene set co-expression at the p < 0.10 identical to that of the empiric NF group though in the latter this pat- In much the same way as the ISF group, several of the LCA tern was also retained at the p < 0.05 level. In both the NF groups were characterized by a lack of coordinated activity and LCA-0/LCA-2 networks CD 19+ B cell up-regulated between immune gene sets. Indeed no significant correla- and down-regulated gene sets correlated tightly behaving tions existed for LCA-4 even at the p < 0.10 level. This was as one set (r = 0.43, p = 0.008). T regulatory and NK cell also true of LCA-3 and LCA-6 classes at the p < 0.05 level. gene sets both supported significant positive interaction Though also quite sparse, heat maps for LCA-1 and LCA- with one or both of the CD19+ B cell sets. In addition 5 each recovered specific features of the CFS and ISF heat CD8+ T cell activity and CD14+ monocyte activity were maps. LCA-1 demonstrated a significant positive correla- significantly antagonistic. In contrast the network tion (p < 0.05) between CD14+ monocyte and CD16+ obtained for the CFS subjects displayed a shift in interac- neutrophil sets, a CFS feature. At the p < 0.10 level the tions towards the upper left hand corner of the heat map. same heat map showed a negative correlation between the Indeed significant interactions appeared linking the NK and CD19+ B cell down-regulated set, an ISF feature. expression of the CD14+ monocyte gene set with that of Unique to LCA-1 was a positive correlation between T reg the B cell (CD19+ up-regulated) and the CD16+ neu- (FoxP3) and CD14+ monocyte gene set expression (p < trophil gene sets. The neutrophil set also shared signifi- 0.10), a trait not identified in CFS and actually reversed in cant co-expression with the CD8+ T cell gene set (p < LCA-3. Similarly the heat map for LCA-5 contained 2 fea- 0.05) in CFS. Interestingly interactions with the neu- tures specific to the CFS group, namely a strong positive Page 7 of 13 (page number not for citation purposes) Ln Transformed Gene Expression Network Size S Graph Size Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 5: Detailed results of gene set correlation r (null probability p ) for empiric classes. a, b a, b CFS CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.43 (0.01) 0.13 (0.42) 0.14 (0.39) 0.33 (0.04) 0.27 (0.10) -0.25 (0.12) CD14+ Up 0.54 (0.00) 0.29 (0.07) -0.12 (0.46) 0.18 (0.26) -0.08 (0.61) CD19+ Up 0.16 (0.33) -0.38 (0.02) 0.12 (0.46) 0.15 (0.36) CD19+ Down 0.15 (0.37) -0.23 (0.16) -0.04 (0.79) CD8+ Up 0.00 (1.00) -0.29 (0.08) T reg 0.02 (0.88) NK cell ISF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.13 (0.46) 0.11 (0.51) -0.13 (0.47) -0.16 (0.37) 0.19 (0.27) 0.09 (0.62) CD14+ Up -0.24 (0.17) 0.09 (0.62) -0.26 (0.13) -0.11 (0.52) 0.49 (0.00) CD19+ Up -0.08 (0.64) 0.02 (0.91) 0.26 (0.13) -0.03 (0.88) CD19+ Down 0.22 (0.21) 0.00 (0.99) -0.30 (0.08) CD8+ Up 0.17 (0.33) -0.26 (0.13) T reg -0.04 (0.80) NK cell NF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.10 (0.55) 0.11 (0.53) 0.08 (0.64) -0.27 (0.11) 0.15 (0.38) -0.01 (0.96) CD14+ Up 0.12 (0.49) -0.06 (0.74) -0.38 (0.02) -0.10 (0.54) 0.02 (0.91) CD19+ Up 0.43 (0.01) -0.02 (0.90) 0.37 (0.02) 0.25 (0.13) CD19+ Down 0.05 (0.79) 0.37 (0.02) 0.35 (0.04) CD8+ Up -0.10 (0.58) 0.20 (0.24) T reg 0.20 (0.23) NK cell correlation linking CD19+ B cell with CD14+ monocyte tion providing additional insight into potential subclasses up-regulated sets (p < 0.05) along with a strong negative of CFS. The commonalities between these classification correlation linking the former with CD8+ gene set expres- systems are readily observed in the patterns of gene set co- sion. These 2 features were not shared with the LCA-1 expression. Indeed the empiric CFS group seems to group. These results reaffirmed the strong links between present an aggregation of the gene set co-expression pat- the CFS subject group and LCA classes 1 and 5 in addition terns observed in LCA classes 1 and 5. However, the differ- to suggesting that immune set co-expression might offer ential expression of gene sets only achieves statistical insight into the distinct nature of these apparent sub- significance in the case of the coarser empiric classes with classes of CFS. the larger group sizes providing better noise reduction. Specifically in the empiric CFS class we found a significant decrease in the median expression for a set of 6 genes pref- Discussion In this work we dissected PBMC gene expression profiles erentially up-regulated in isolated CD19+ B cells com- into components that were preferentially expressed in sev- pared to non-fatigued controls. Expression of this CD19+ eral isolated lymphocyte subpopulations. We also used 2 B cell up-regulated gene set also discriminated ISF from systems to stratify subjects into illness groups. The LCA controls at 0.05 confidence level. In a recent study of CFS class structure was inferred directly from a comprehensive occurrence both in the presence and absence of viral infec- set of clinical and biological indicators. All indicators tion Racciati et al. [8] found no significant differences in were equally weighted and contrary to common practice CD19+ cell abundance. Robertson et al. [7] recently no subset was assigned greater relevance a priori. In con- reported significantly higher abundance of CD20+/CD5+ trast the empiric classification which was based on a con- B cells, a subset associated with the production of auto- sensus of opinions from expert clinicians. Results confirm antibodies, in patients with depression. These findings strong links between both systems with the LCA classifica- together with our observations of depressed CD19+ gene Page 8 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 6: Detailed results of gene set correlation r (null probability p ) for LCA-0/2, 1, 3 classes. a, b a, b LCA-0/2 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.17 (0.33) 0.15 (0.39) 0.09 (0.62) -0.19 (0.27) 0.09 (0.62) -0.10 (0.56) CD14+ Up 0.14 (0.41) -0.03 (0.88) -0.30 (0.08) -0.11 (0.52) 0.00 (0.98) CD19+ Up 0.43 (0.01) -0.04 (0.82) 0.40 (0.02) 0.19 (0.28) CD19+ Down -0.01 (0.93) 0.39 (0.02) 0.30 (0.08) CD8+ Up -0.05 (0.77) 0.06 (0.73) T reg 0.04 (0.82) NK cell LCA-1 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.55 (0.01) -0.03 (0.91) 0.16 (0.47) 0.28 (0.20) 0.16 (0.46) -0.25 (0.25) CD14+ Up 0.25 (0.26) 0.20 (0.37) -0.03 (0.90) 0.36 (0.10) -0.04 (0.84) CD19+ Up -0.14 (0.53) -0.21 (0.34) 0.13 (0.55) 0.06 (0.77) CD19+ Down 0.33 (0.13) -0.19 (0.38) -0.41 (0.05) CD8+ Up 0.09 (0.69) -0.32 (0.14) T reg -0.02 (0.91) NK cell LCA-3 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down -0.13 (0.61) 0.07 (0.80) -0.18 (0.49) -0.20 (0.45) 0.35 (0.16) -0.31 (0.22) CD14+ Up -0.31 (0.23) 0.15 (0.57) -0.36 (0.16) -0.45 (0.07) 0.15 (0.56) CD19+ Up 0.37 (0.14) -0.06 (0.82) 0.29 (0.26) -0.07 (0.78) CD19+ Down 0.20 (0.44) 0.15 (0.56) 0.18 (0.48) CD8+ Up 0.21 (0.42) -0.05 (0.84) T reg 0.19 (0.47) NK cell expression and altered association between up and down- intracellular perforin [34]. In this work we observe an regulated B cell functions would suggest that the function increased expression of the NK cell gene set. Of the 4 of these cells might be compromised in CFS subjects. Cole genes used to capture NK cell function the expression of et al. [28] reported a selective reduction of mature B lym- NKG2A/C (NM 002260) was most increased. The binding phocyte function in subjects who experienced chronic of NKG2A to its natural ligand, human non-classic class I high levels of social isolation including suppression of leukocyte antigen (HLA) E, is known to induce its immu- several transcription factors involved B cell differentiation noreceptor tyrosine-based inhibition motif (ITIM) and such as Ikaros/ZNF1A1. Genes encoding for members of suppress cytotoxic cell effector activity [35]. Moreover the zinc finger protein family were also identified in pre- NKG2A is also known to be co-expressed on activated Th2 vious work by this group as prominent contributors to the but not Th1 lymphocytes [36]. A bias towards Th2-type CFS symptom space [9]. A closer look at the 6 genes that immune response in CFS patients has also been suggested constitute the CD19+ up-regulated set showed that the on the basis of intracellular T cell cytokine profiles by PTPRK and TSPAN3 genes, both associated with immune Skowera et al. [37]. Interestingly this also aligns with cell adhesion and development, were the most sup- altered expression of the PTPRK gene mentioned above as pressed. Down-regulation of PTPRK, a TGF-β target gene, Asano et al. [38] report impaired Th1 function with is known to be down-regulated by the Epstein-Barr virus PTPRK deletion in rats. Therefore our observations sup- (EBV) [29], an infectious agent known to trigger CFS ported findings of increased suppression of cytotoxic [30,31]. Down-regulation of TGF-β has been reported in activity in CFS and hinted at increased Th2 activity though CFS by Tomoda et al. [32]. the latter were not specifically addressed in this analysis. NK cell activity is suppressed in CFS [33] and this Neutrophils for their part are only found at trace and con- decreased cytotoxity has been associated with reduced taminating amounts in most PBMC preparations [39] so Page 9 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 Table 7: Detailed results of gene set correlation r (null probability p ) for LCA-4, 5, 6 classes. a, b a, b LCA-4 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.41 (0.21) -0.02 (0.96) -0.09 (0.80) -0.14 (0.68) 0.00 (0.99) 0.14 (0.67) CD14+ Up -0.05 (0.88) -0.13 (0.71) -0.12 (0.73) -0.01 (0.98) -0.28 (0.41) CD19+ Up -0.21 (0.53) 0.09 (0.78) 0.28 (0.40) 0.13 (0.71) CD19+ Down 0.44 (0.18) 0.05 (0.89) -0.48 (0.13) CD8+ Up 0.19 (0.57) -0.04 (0.91) T reg -0.11 (0.75) NK cell LCA-5 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.44 (0.12) 0.35 (0.22) 0.05 (0.87) 0.32 (0.27) 0.17 (0.56) 0.04 (0.90) CD14+ Up 0.86 (0.00) 0.23 (0.44) -0.40 (0.16) 0.35 (0.21) 0.40 (0.16) CD19+ Up 0.15 (0.60) -0.48 (0.08) 0.35 (0.22) 0.36 (0.20) CD19+ Down -0.16 (0.58) 0.16 (0.60) -0.16 (0.59) CD8+ Up -0.22 (0.45) -0.12 (0.69) T reg 0.07 (0.82) NK cell LCA-6 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down 0.34 (0.30) -0.02 (0.95) 0.20 (0.55) -0.15 (0.65) 0.45 (0.16) 0.23 (0.50) CD14+ Up 0.10 (0.76) 0.57 (0.07) -0.14 (0.69) -0.40 (0.22) 0.20 (0.56) CD19+ Up 0.35 (0.30) -0.33 (0.32) -0.43 (0.18) 0.31 (0.35) CD19+ Down 0.15 (0.67) -0.31 (0.36) 0.32 (0.33) CD8+ Up 0.11 (0.74) -0.46 (0.16) T reg 0.07 (0.84) NK cell SD is standard deviation it is interesting to note that the neutrophil gene set arose (IL-1, IL-6, IL-8, GM-CSF). IL-8 attracts more neutrophils as a core element in the emergence of coordinated and together with GM-CSF causes these to once again immune activity. In particular the CD16+ neutrophil gene degranulate. With the corresponding release of additional set and the CD14+ monocyte gene set shared significant MPO, the cycle starts once again. The TNF-α initiated cas- co-expression. Not only do these arise from the same cade induces IL-6 which is used by B cells for maximum hematopoietic CD34+ progenitor cell [40] but since the antibody secretion usually IgM. In addition to the present immune community is highly integrated the presence or analysis, a preliminary examination of cytokine data col- absence of neutrophils will also be mirrored in the state of lected in the Wichita study pointed to an increase in TNF- the remaining cell population. The CD14+ monocyte set α in CFS subjects (data not shown) as documented previ- also shared significant co-expression with the CD19+ B ously by Moss et al. [42]. cell gene sets. Together this neutrophil-monocyte-B cell immune interaction triad is highly consistent with a In addition to this core network, we also observed that model of chronic inflammation proposed by Lefkowitz CD8+ T cell set expression correlated negatively with that and Lefkowitz [41]. According to this model once an of the NK and CD19+ up-regulated B cell sets. In one pos- event initiates inflammation, neutrophils are among the sible mechanism linking these three cell types, IgG anti- first cells to arrive at the site. They degranulate releasing bodies binding to GD3 on the surface of CD4+ and CD8+ MPO into the microenvironment which together with T cells could elicit signals for proliferation of these subsets iMPO binds to macrophage MMR receptor and induces and expression of the IL-2 receptor CD25. NK cells have release of TNF-α. The latter functions in an autocrine been shown to selectively inhibit this antibody-mediated manner and along with iMPO initiates a cytokine cascade proliferation of CD8+ T cells by Claus et al. [43] perhaps Page 10 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 NF Empiric CFS ISF CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell LCA-3 LCA-0 & 2 LCA-1 CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell 0.8 LCA-6 LCA- 4 LCA- 5 0.6 CD16+ Down 0.4 CD14+ Up 0.2 CD19+ Up CD19+ Down -0.2 -0.4 CD8+ Up -0.6 T reg -0.8 NK cell -1 a Figure 5 H cu se well as for LCA contro toff sign at maps of gene set c ificance p <0.05 ( o-exp l clasl ression ex ses ( ) for empir LCA-pr 0i, c classes for non-fati esse 2) an d as li d for all LCA disease classes near correl gued ation co (NF) eff co icintro ent r ls, in at cut suffici oefnt fati f signifgue icansymptoms ce p <0.10 ( (ISF) and CFS ) and at a,b a, b a,b Heat maps of gene set co-expression expressed as linear correlation coefficient r at cutoff significance a, b p <0.10 () and at cutoff significance p <0.05 (l) for empiric classes for non-fatigued (NF) controls, insuffi- a,b a,b cient fatigue symptoms (ISF) and CFS as well as for LCA control classes (LCA-0, 2) and for all LCA disease classes. through down-regulation of autologous mixed lym- structed for LCA classes 1 and 5 suggested that B cell phocyte reaction (MLR). This basic analysis of immune involvement in these processes may serve as factor for dis- gene set co-expression points therefore to the existence of criminating between distinct subsets of CFS subjects. immune signaling processes in CFS that adhere to at least one known mechanism of chronic inflammation and sup- Although several very plausible immune response mecha- port possible antibody-mediated NK cell modulation of T nisms were recovered by this analysis it must be empha- cell activity. Furthermore association networks con- sized that the use of discrete gene sets has several Page 11 of 13 (page number not for citation purposes) CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell CD16+ Down CD14+ Up CD19+ Up CD19+ Down CD8+ Up T reg NK cell Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 limitations. In particular it becomes increasingly difficult tion, supervised the analysis and drafted the manuscript. to identify genes that are exclusively or even predomi- All authors read and approved the final manuscript. nantly expressed in specific cell lineages when these share many commonalities of function and goal. This issue was Additional material reflected in by the small size of the gene sets identified in this work from lymphocyte subset expression profiles. An Additional File 1 approach that promises to be more robust and more Supplementary_data_Panther_Gene_List_Annotation. Additional data revealing still involves the direct use of the genome-wide file 1 is a set of tables listing the genes used in each gene set along with expression for these cell populations. This remains an the functional annotation available in the PANTHER classification sys- active area of research [44]. However, even this simple tem. This file also contains the median values and standard deviations within each illness class of the Ln-transformed gene expression for each analysis points to dramatic differences in immune net- gene. work topology and cell signaling in CFS and we expect Click here for file these differences to be largely conserved in more elaborate [http://www.biomedcentral.com/content/supplementary/1744- analyses. Furthermore the methodology outlined and 9081-4-44-S1.xls] issues raised in this work demonstrate the importance of developing approaches that effectively integrate flow cytometry with cytokine and gene expression profiling. In particular it underscores the importance of looking Acknowledgements beyond differential expression of individual components The authors would like to thank Dr. WC Reeves and the members of his towards changes in their patterns of coordinated activity CFS Research Group at the Centers for Disease Control and Prevention in Atlanta for their many helpful discussions. This work was supported by and formally recognizing the network properties of the research grant # N031000099 from the University of Alberta's Faculty of immune system. Medicine and Dentistry. List of abbreviations References CD: cluster of differentiation; CDC: Centers for disease 1. Reynolds KJ, Vernon SD, Bouchery E, Reeves WC: The economic Control and Prevention; cDNA: complementary DNA; impact of chronic fatigue syndrome. Cost Eff Resour Alloc 2(1):4. 2004 Jun 21 CFS: Chronic Fatigue Syndrome; DPP-IV: dipeptidyl 2. Klimas N, Salvato F, Morgan R, Fletcher MA: Immunologic abnor- peptidase-4; EBV: Epstein-Barr virus; FoxP3: forkhead box malities in chronic fatigue syndrome. J Clin Microbiol 1990, P3 gene; GD3: ganglioside D3; GM-CSF: granulocyte mac- 28(6):1403-1410. 3. Straus SE, Fritz S, Dale JK, Gould B, Strober W: Lymphocyte phe- rophage colony-stimulating factor; HCMV: human notyping and function in chronic fatigue syndrome. J of Clin cytomegalovirus; Ig: immunoglobulin; iMPO: inactive Immunol 1993, 13(1):30-40. myeloperoxidase; IL: interleukin; ISF: insufficient symp- 4. Tirelli U, Bernardi D, Improta S, Pinto A: Immunologic abnormal- ities in chronic fatigue syndrome. J Chronic Fatigue Syndrome toms of fatigue; MFI: Multidimensional Fatigue Inven- 1996, 2(1):85-96. tory; MLR: mixed lymphocyte reaction; MMR: 5. Yan S, Marguet D, Dobers J, Reutter W, Fan H: Deficiency of CD26 results in a change of cytokine and immunoglobulin secre- macrophage; mannose receptor; NF: non-fatigued; NK: tion after stimulation by pokeweed mitogen. Eur J Immunol natural killer cell; NKG2A/C: killer cell lectin-like receptor 2003, 33(6):1519-27. subfamily C, member 1/2 or KLRC1/C2; LCA: latent class 6. Boonacker E, Van Noorden CJ: The multifunctional or moon- lighting protein CD26/DPPIV. Eur J Cell Biol 2003, 82(2):53-73. analysis; MPO: myeloperoxidase; PANTHER: Protein 7. Robertson MJ, Schacterle RS, Mackin GA, Wilson SN, Bloomingdale ANalysis THrough Evolutionary Relationships; PBMC: KL, Ritz J, Komaroff AL: Lymphocyte subset differences in peripheral blood mononuclear cell; PTPRK: protein tyro- patients with chronic fatigue syndrome, multiple sclerosis and major depression. Clin Exp Immunol 2005, 141(2):326-32. sine phosphatase, receptor type, K; SF-36: Short Form 36; 8. Racciatti D, Dalessandro M, Delle Donne L, Falasca K, Zingariello P, TGF: transforming growth factor; TNF: tumor necrosis fac- Paganelli R, Pizzigallo E, Vecchiet J: Study of immune alteration in patients with chronic fatigue syndrome with different etiolo- tor; Th: T helper cell; TSPAN3: tetraspanin 3. gies. J Immunopath Pharm 2004, 17(2):57-62. 9. Broderick G, Craddock RC, Whistler T, Taylor R, Klimas N, Unger ER: Identifying illness parameters in fatiguing syndromes Competing interests using classical projection methods. Pharmacogenomics 2006, The authors declare that they have no competing interests. 7(3):407-19. 10. Carmel L, Efroni S, White PD, Aslakson E, Vollmer-Conna U, Rajeevan MS: Gene expression profile of empirically deline- Authors' contributions ated classes of unexplained chronic fatigue. Pharmacogenomics ALA carried out statistical analysis of gene expression pro- 2006, 7(3):375-86. files, gene set design and significance analysis and helped 11. Whistler T, Jones JF, Unger ER, Vernon SD: Exercise responsive genes measured in peripheral blood of women with chronic draft the manuscript. CB participated in the design of the fatigue syndrome and matched control subjects. BMC Physiol study and assisted with the statistical analysis and helped 5(1):5. 2005 Mar 24; draft the manuscript. SDV helped conceive of the study, its 12. Siegel S, Antoni M, Fletcher M, Maher K, Segota M, Klimas N: Impaired natural immunity, cognitive dysfunction, and phys- design and coordination and helped to draft the manu- ical symptoms in patients with chronic fatigue syndrome: script. GB conceived of the study, its design and coordina- Page 12 of 13 (page number not for citation purposes) Behavioral and Brain Functions 2008, 4:44 http://www.behavioralandbrainfunctions.com/content/4/1/44 preliminary evidence for a subgroup? J Psychosom Res 2006, patients with chronic fatigue syndrome and normal controls. 60:559-566. Psychiatry Res 2005, 134(1):101-104. 13. Segal E, Friedman N, Koller D, Regev A: A module map showing 33. Fletcher MA, Maher K, Klimas NG: Natural killer cell function in conditional activity of expression modules in cancer. Nat chronic fatigue syndrome. Clin Applied Immunol Rev 2002, Genet 2004, 36(10):1090-8. 2:129-139. 14. Dinu I, Potter JD, Mueller T, Liu Q, Adewale AJ, Jhangri GS, Einecke 34. Maher KJ, Klimas NG, Fletcher MA: Chronic fatigue syndrome is G, Famulski KS, Halloran P, Yasui Y: Improving gene set analysis associated with diminished intracellular perforin. Clin Exp of microarray data by SAM-GS. BMC Bioinformatics 8(1):242. Immunol 2005, 142(3):505-511. 2007 Jul 5; 35. Kabat J, Borrego F, Brooks A, Coligan JE: Role that each NKG2A 15. Efron B, Tibshirani R: On testing the significance of sets of immunoreceptor tyrosine-based inhibitory motif plays in genes. Ann Appl Statist 2007, 1(1):107-129. mediating the human CD94/NKG2A inhibitory signal. J 16. Vernon SD, Reeves WC: The challenge of integrating disparate Immunol 2002, 169(4):1948-1958. high-content data: epidemiologic, clinical, and laboratory 36. Freishtat RJ, Mitchell LW, Ghimbovschi SD, Meyers SB, Hoffman EP: data collected during an in-hospital study of chronic fatigue NKG2A and CD56 are coexpressed on activated TH2 but syndrome. Pharmacogenomics 2006, 7(3):345-354. not TH1 lymphocytes. Hum Immunol 2005, 66(12):1223-1234. 17. Reeves WC, Wagner D, Nisenbaum R, Jones JF, Gurbaxani B, Solo- 37. Skowera A, Cleare A, Blair D, Bevis L, Wessely SC, Peakman M: High mon L, Papanicolaou DA, Unger ER, Vernon SD, Heim C: Chronic levels of type 2 cytokine-producing cells in chronic fatigue fatigue syndrome – A clinically empirical approach to its def- syndrome. Clin Exp Immunol 2004, 135(2):294-302. inition and study. BMC Medicine 2005, 3:8. 38. Asano A, Tsubomatsu K, Jung CG, Sasaki N, Agui T: A deletion 18. Wagner D, Nisenbaum R, Heim C, Jones JF, Unger ER, Reeves WC: mutation of the protein tyrosine phosphatase kappa (Ptprk) Psychometric properties of the CDC Symptom Inventory gene is responsible for T-helper immunodeficiency (thid) in for assessment of chronic fatigue syndrome. Popul Health Metr the LEC rat. Mamm Genome 2007, 18(11):779-786. Epub 2007 Oct 2005, 3:8. 2 19. Vollmer-Conna U, Aslakson E, White PD: An empirical delinea- 39. Jison ML, Munson PJ, Barb JJ, Suffredini AF, Talwar S, Logun C, Ragha- tion of the heterogeneity of chronic unexplained fatigue in vachari N, Beigel JH, Shelhamer JH, Danner RL, Gladwin MT: Blood women. Pharmacogenomics 2006, 7(3):355-364. mononuclear cell gene expression profiles characterize the 20. Einecke G, Broderick G, Sis B, Halloran PF: Early loss of renal tran- oxidant, hemolytic, and inflammatory stress of sickle cell dis- scripts in kidney allografts: relationship to morphologic ease. Blood 104(1):270-80. 2004 Jul 1; changes and alloimmune effector mechanisms. Am J Trans- 40. Ferrari F, Bortoluzzi S, Coppe A, Basso D, Bicciato S, Zini R, Gemelli plant 2007, 7(5):1121-1130. C, Danieli GA, Ferrari S: Genomic expression during human 21. Famulski KS, Broderick G, Einecke G, Hay K, Cruz J, Sis B, Mengel M, myelopoiesis. BMC Genomics 8:264. 2007 Aug 3; Halloran PF: Transcriptome analysis reveals heterogeneity in 41. Lefkowitz DL, Lefkowitz SS: Macrophage-neutrophil interac- the injury response of kidney transplants. Am J Transplant 2007, tion: A paradigm for chronic inflammation revisited. Immunol 7:2483-95. Cell Biol 2001, 79:502-506. 22. Mueller TF, Einecke G, Reeve J, Sis B, Mengel M, Jhangri GS, Bunnag 42. Moss RB, Mercandetti A, Vojdani A: TNF-alpha and chronic S, Cruz J, Wishart D, Meng C, Broderick G, Kaplan B, Halloran PF: fatigue syndrome. J Clin Immunol 1999, 19(5):314-316. Microarray Analysis of Rejection in Human Kidney Trans- 43. Claus C, Schlaak J, Dithnayer M, Meyer zum K-H, Dippold B, Dippold plants Using Pathogenesis-based Transcripts Sets. Am J Trans- W: Inhibition of anti-GD3-ganglioside antibody-induced pro- plant 2007, 7(12):2712-2722. liferation of human CD8+ Tcells by CD16+ natural killer 23. Lyons PA, Koukoulaki M, Hatton A, Doggett K, Woffendin HB, cells. Eur J Immunol 1994, 24:1208-1212. Chaudhry AN, Smith KG: Microarray analysis of human leuco- 44. Gosink MM, Petrie HT, Tsinoremas NF: Electronically subtract- cyte subsets: the advantages of positive selection and rapid ing expression patterns from a mixed cell population. Bioin- purification. BMC Genomics 2007, 8:64. formatics 2007, 23(24):3328-3334. Epub 2007 Oct 22 24. Thomas PD, Kejariwal A, Campbell MJ, Mi H, Diemer K, Guo N, Ladunga I, Ulitsky-Lazareva B, Muruganujan A, Rabkin S, Vandergriff JA, Doremieux O: PANTHER: a browsable database of gene products organized by biological function, using curated pro- tein family and subfamily classification. Nucleic Acids Res 2003, 31(1):334-341. 25. PANTHER Classification System [http://www.pantherdb.org] 26. Bernstein S, Bernstein R: Schaum's Outline of Theory and Problems of Elements of Statistics II New York (NY): McGraw-Hill; 1999. 27. Elo LL, Järvenpää H, Orešiè M, Lahesmaa R, Aittokallio T: System- atic construction of gene coexpression networks with appli- cations to human T helper cell differentiation process. Bioinformatics 2007, 23(16):2096-2103. 28. Cole SW, Hawkley LC, Arevalo JM, Sung CY, Rose RM, Cacioppo JT: Social regulation of gene expression in human leukocytes. Genome Biology 2007, 8(9):R189. 29. Flavell JR, Baumforth KR, Wood VH, Davies GL, Wei W, Reynolds GM, Morgan S, Boyce A, Kelly GL, Young LS, Murray PG: Downreg- Publish with Bio Med Central and every ulation of the TGF-Beta target gene, PTPRK, by the epstein- scientist can read your work free of charge barr virus encoded EBNA1 contributes to the growth and survival of Hodgkin's lymphoma cells. Blood 2008, "BioMed Central will be the most significant development for 111(1):292-301. 2007 Aug 24; disseminating the results of biomedical researc h in our lifetime." 30. Lerner AM, Beqaj SH, Deeter RG, Fitzgerald JT: IgM serum anti- Sir Paul Nurse, Cancer Research UK bodies to Epstein-Barr virus are uniquely present in a subset of patients with the chronic fatigue syndrome. In Vivo 2004, Your research papers will be: 18(2):101-106. available free of charge to the entire biomedical community 31. Hickie I, Davenport T, Wakefield D, Vollmer-Conna U, Cameron B, Vernon SD, Reeves WC, Lloyd A, Dubbo Infection Outcomes Study peer reviewed and published immediately upon acceptance Group: Post-infective and chronic fatigue syndromes precipi- cited in PubMed and archived on PubMed Central tated by viral and non-viral pathogens: prospective cohort study. BMJ 333(7568):575. 2006 Sep 16; yours — you keep the copyright 32. Tomoda A, Joudoi T, Rabab el-M, Matsumoto T, Park TH, Miike T: BioMedcentral Cytokine production and modulation: comparison of Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 13 of 13 (page number not for citation purposes)

Journal

Behavioral and Brain Functions – Springer Journals

Published: Sep 26, 2008

There are no references for this article.