SEARCH

SEARCH BY CITATION

Keywords:

  • Gene expression;
  • lung transplantation;
  • primary graft dysfunction

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

We hypothesized alterations in gene expression could identify important pathways involved in transplant lung injury. Broncho alveolar lavage fluid (BALF) was sampled from donors prior to procurement and in recipients within an hour of reperfusion as part of the NIAID Clinical Trials in Organ Transplantation Study. Twenty-three patients with Grade 3 primary graft dysfunction (PGD) were frequency matched with controls based on donor age and recipient diagnosis. RNA was analyzed using the Human Gene 1.0 ST array. Normalized mRNA expression was transformed and differences between donor and postreperfusion values were ranked then tested using Gene Set Enrichment Analysis. Three-hundred sixty-two gene sets were upregulated, with eight meeting significance (familywise-error rate, FWER p-value <0.05), including the NOD-like receptor inflammasome (NLR; p < 0.001), toll-like receptors (TLR; p < 0.001), IL-1 receptor (p = 0.001), myeloid differentiation primary response gene 88 (p = 0.001), NFkB activation by nontypeable Haemophilus influenzae (p = 0.001), TLR4 (p = 0.008) and TLR 9 (p = 0.018). The top five ranked individual transcripts from these pathways based on rank metric score are predominantly present in the NLR and TLR pathways, including IL1β (1.162), NLRP3 (1.135), IL1α (0.952), IL6 (0.931) and CCL4 (0.842). Gene set enrichment analyses implicate inflammasome–mediated and innate immune signaling pathways as key mediators of the development of PGD in lung transplant patients.


Abbreviations
ALI

acute lung injury

DAMPs

danger-associated molecular patterns

MYD88

myeloid differentiation primary response gene 88

NLR

nucleotide binding oligomerization domain (NOD) like receptor

NTHI

NFkB activation by nontypeable Haemophilus influenzae

PAMPs

pathogen-associated molecular patterns

PGD

primary graft dysfunction

ROS

reactive oxygen species

TLR

toll-like receptor

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Primary graft dysfunction (PGD) is the most common cause of early death after lung transplantation [1]. PGD is characterized by hypoxemia and radiographic infiltrates occurring in the allograft within 72 h of transplantation [2]. Ten to thirty percent of all subjects receiving lung transplantation develop PGD [3], which is associated with both short- and long-term morbidity and mortality [2, 4].

Though the mechanisms of PGD remain incompletely understood, donor [5] and other factors have been implicated in PGD risk [3]. mRNA expression profiling is an ideal tool for identifying key pathways that are involved in the pathogenesis of complex syndromes, such as acute lung injury [6]. We sought to define pathways important in PGD development by comparing changes in donor lung gene expression before transplant with those after reperfusion, using a multicenter cohort study design. We used a gene set enrichment approach employing known biological pathways to capture changes seen in many genes within a cellular pathway in PGD subjects [7]. We hypothesized that specific pathways could be identified that are differentially expressed during the transplant procedure in patients who develop clinically significant PGD.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Study population and data collected

Subjects were selected from the Clinical Trials in Organ Transplantation-03 study (CTOT-03) which is a multicenter, prospective cohort study of solid organ transplant recipients (NCT00531921). CTOT-03 is a research consortium of five centers conducting clinical and mechanistic studies designed to investigate mRNA profiles with early outcomes. At study completion 294 participants had been enrolled. Institutional review board approval and informed written consent from both recipients and organ donor proxies were obtained prior to the recruitment of subjects. A subset of consecutively consenting lung transplant subjects was enrolled between January 22, 2008 and August 19, 2010 from three centers (Penn, Columbia and Wisconsin). Clinical data were collected prospectively. PGD grade was determined using the consensus definition of the International Society of Heart and Lung Transplantation using two blinded readers with adjudication as previously described [8, 9]. We used any Grade 3 PGD occurring within the first 72 h following lung transplantation as our primary case definition [8]. To minimize confounding secondary to donor and recipient factors independent of suspected PGD mechanism, we utilized a nested case control strategy (Figure 1). From the total enrolled lung transplant cohort of 106 subjects, we selected all 23 patients with PGD and matched controls on donor age (categorical variable: <45 vs. ≥45) [10] and pretransplant recipient diagnosis (cystic fibrosis, emphysema or pulmonary fibrosis) to ensure similarity between cases and controls. The minimum age difference was then used to select the “best” match when there was more than one potential control subject identified per case.

image

Figure 1. Enrollment design.

Download figure to PowerPoint

Collection and processing of biological samples

BALF was collected in the donor operating room before procurement and again 1 h after reperfusion to focus on early mechanisms contributing to PGD. Twenty milliliters of normal saline was instilled in one subsegmental location in order to recovery a minimum of 5 mL of BALF. The BALF was placed in a sterile 120 mL specimen cup and immediately placed on ice. The sample was processed as soon as possible but no greater than 8 h after collection. The sample was transferred into 50 mL sterile centrifuge tubes and centrifuged at 2100 rpm for 10 min at room temperature. The supernatant was aliquoted and the cell pellet was resuspended in Trizol and stored at −80°C.

Gene expression array and quality control

We used the Affymetrix Human Gene 1.0 ST Array, which has whole-transcript coverage for 36 079 total RefSeq transcripts and 21 014 genes [11]. One hundred nanograms of total RNA was converted to first-strand cDNA using reverse transcriptase primed by a poly(T) oligomer that incorporated the T7 RNA polymerase promoter. Second-strand cDNA synthesis was followed by in vitro transcription (Affymetrix One-Cycle Target Labeling Kit, Affymetrix, Cleveland, OH) for linear amplification and biotinylation of each transcript, and the resulting cDNA was fragmented and assessed by Bioanalyzer. Affymetrix Command Console and Expression Console were used to quantitate expression levels for targeted genes; default values provided by Affymetrix were applied to all analysis parameters. Border pixels were removed, and the average intensity of pixels within the 75th percentile was computed for each probe. Probe sets for positive and negative controls were examined in Expression Console, and Facility quality control parameters were confirmed to fall within normal ranges. Probe intensities were exported as Affymetrix cel files. Cel files were imported into Partek Genomics Suite (v6.6, Partek, Inc. St. Louis, MO) where RMA normalization was applied, resulting in log2-transformed expression intensities for each transcript in each sample.

Differential expression and gene set enrichment analysis (GSEA)

Three independent biological replicates of each sample (two time points for each of 23 PGD and 23 control participants) for each condition (control preprocurement: Time 0, PGD; control postreperfusion: Time 1 PGD; case preprocurement: Time 0 PGD+; case postreperfusion: Time 1 PGD+) were assayed on microarrays. Principal components analysis by sample was performed to confirm that replicates within each condition grouped with most similarity, and to identify any outlier samples. Genes were then ranked to reflect greatest changes from donor to postimplantation by the following transformation ([mean Log2,Time 1,PGD+ − mean Log2,Time 1,PGD−] − [mean Log2,Time 0,PGD+ − mean Log2,Time 0,PGD−]), and the resulting ranked gene list was tested for networks of gene interactions using GSEA (v2.07, Broad Institute, Cambridge, MA) using curated gene sets (Molecular Signatures Database, v3.0, C2, Cannonical Pathways including KEGG, BIOCARTA, REACTOME and GO, Cambridge, MA) [7]. The logic of the transformation was based on the premise that there are significant changes in gene expression after transplantation not all of which are related to the development of PGD and therefore this strategy would highlight only those pathways important in the mechanism of PGD.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

One hundred eight subjects were enrolled and 106 subjects were transplanted. Twenty-three subjects, 21.7% (95% CI: 14.3, 30.8), developed Grade 3 PGD within 72 h. Baseline demographics between PGD cases and matched controls are summarized in Table 1. There were no statistically significant differences in donor age, gender, race, cause of death, smoking history, bronchoscopic findings, ischemic time and preservation fluid type or recipient-related factors such as age, gender, race, pulmonary artery pressure and transplant type. Additionally, we were unable to demonstrate statistically significant differences in intraoperative interventions, including cardiopulmonary bypass (CPB) use or time. As would be expected, postoperative need for mechanical ventilation differed significantly between PGD and non-PGD controls (Table 1).

Table 1. Baseline demographics
 PGD (n = 23)No PGD (n = 23)p-value
  1. Data are presented as mean (standard deviation) for continuous variables and as number (percent) for categorical variables. CF, cystic fibrosis; LAM, lymphangioleimatosis; PASP, pulmonary artery systolic pressure; CPB, cardiopulmonary bypass.

Donor
Age36.0 (16.32)39.7 (12.85)0.409
Female5 (21.7)10 (43.5)0.116
Race  0.641
Asian01 (4.3) 
African American4 (17.4)4 (17.4) 
Caucasian15 (65.2)34 (73.9) 
Unknown/not reported4 (17.4)1 (4.3) 
Cause of death  0.916
Anoxia3 (13.0)3 (13.0) 
Cerebrovascular10 (43.5)11 (47.8) 
Head trauma7 (30.4)5 (21.7) 
Other3 (13.0)4 (17.4) 
Nonsmoker13 (56.5)14 (60.9)0.765
Normal bronchoscopy19 (82.6)18 (78.3)>0.999
Total ischemic time447 (222)378 (201)0.273
Recipient
Age56.4 (8.79)55.4 (10.27)0.724
Female5 (21.7)8 (34.8)0.326
Race  0.234
African American3 (13.0)0 
Caucasian20 (87.0)21 (91.3) 
Unknown/not reported02 (8.7) 
Diagnosis  >0.999
Bronchiectasis/CF4 (17.4)4 (17.4) 
COPD/LAM5 (21.7)5 (21.7) 
Pulmonary fibrosis12 (52.2)12 (52.2) 
Other2 (8.7)2 (8.7) 
Bilateral transplant16 (69.6)10 (43.5)0.078
Intra-operative factors
PASP56.9 (25.36)60.8 (33.03)0.671
CPB needed14 (60.9)8 (34.8)0.077
CPB duration260.8 (89.01)201.1 (91.3)0.156
Mechanical ventilation
24 h post-op20 (87.0)11 (47.8)0.005
48 h post-op14 (60.9)3 (13.0)<0.001
72 h post-op10 (43.5)3 (13.0)0.022

Pathway differences in mRNA expression preprocurement to postreperfusion between PGD cases and controls are presented in Table 2. A total of 362 gene sets were upregulated in recipients who developed PGD of which eight gene sets met significance with a familywise-error rate (FWER) of less than 0.05 (Figure 2). A complete list of gene sets that did not meet our cutoff can be found in the supplement (Table S1). The pathway with the highest normalized enrichment score (NES), a standardized metric that accounts for differences in gene set size and in correlations between gene sets and the expression dataset, was the KEGG nucleotide binding oligomerization domain like receptor (NLR) inflammasome pathway (Figure 2A; NES 2.44; FWER adjusted p < 0.001). The second highest pathway was the KEGG toll-like receptor (TLR) signaling pathway (Figure 2B; NES 2.22; FWER adjusted p < 0.001). Six other pathways met our FWER cutoff and included: BIOCARTA IL-1 receptor (IL1R) pathway (NES 2.19; FWER adjusted p = 0.001), REACTOME myeloid differentiation primary response gene 88 (MYD88) (NES 2.16; FWER adjusted p = 0.001), BIOCARTA NFkB activation by nontypeable Haemophilus influenzae (NTHI) signaling pathway (NES 2.16; FWER adjusted p = 0.001), REACTOME-activated TLR4 signaling pathway (NES 2.10; FWER adjusted p = 0.008), REACTOME TLR 9 cascade (NES 2.07; FWER adjusted p = 0.018) and BIOCARTA toll pathway (NES 2.06; FWER adjusted p = 0.018). Individual enrichment plots are presented in the supplement (Figure S1). Each of these pathways has been described as either activating or affecting innate immune function. Further analysis of the individual transcripts from these pathways based on the rank metric score, a measure of an individual transcript's correlation with the PGD phenotype, demonstrated significant overlap within the NLR and TLR pathways. The highest ranked individual transcripts include IL1β (1.162), NLRP3 (1.135), IL1A (0.952), IL6 (0.931), CCL4 (0.842), TLR6 (0.805), TLR1 (0.796), TNFAIP3 (0.792) and TLR4 (0.677). The enriched individual transcripts from the eight pathways meeting the FWER of 0.05 are presented in Table 3 and the complete ranked gene list for positively correlated transcripts is presented in the supplement (Table S2).

Table 2. Genes set enrichment analysis
NameSourceNESNOM p-valFDR q-valFWER p-val
  1. NES, normalized enrichment score (accounts for differences in gene set size and in correlations between gene sets and the expression dataset to allow for comparisons across gene sets); NOM p-val, nominal p value (estimates the statistical significance of the enrichment score for a single gene set); FDR, false discovery rate (estimated probability that a gene set with a given NES represents a false positive finding); FWER, familywise-error rate (conservative correction that seeks to ensure that the list of reported results does not include even a single false-positive gene set) [7].

NOD-like receptor signaling pathwayKEGG2.443<0.001<0.001<0.001
TOLL-like receptor signaling pathwayKEGG2.222<0.001<0.001<0.001
IL1R pathwayBIOCARTA2.19<0.0010.00030.001
MYD88 cascadeREACTOME2.161<0.0010.00020.001
NTHI pathwayBIOCARTA2.157<0.0010.00020.001
Activated TLR4 signalingREACTOME2.103<0.0010.00140.008
TLR9 cascadeREACTOME2.066<0.0010.00260.018
TOLL pathwayBIOCARTA2.064<0.0010.00230.018
image

Figure 2. Most significantly enriched gene sets. The top portion of the plot shows the running enrichment score (ES) for the gene set from decreasing values of the rank list. The score at the peak of the plot (the score furthest from 0.0) is the ES for the overall gene set. The middle portion of the plot shows where the members of the gene set appear in the ranked list of genes. The bottom portion of the plot shows the value of the ranking metric as the list of ranked genes decreases in value. The ranking metric measures an individual transcript's correlation with the PGD phenotype.

Download figure to PowerPoint

Table 3. Significantly enriched transcripts in GSEA pathways
NLR signaling pathwayTLR signaling pathwayIL1R pathwayMYD88 cascadeNTHI pathwayActivated TLR4 signalingTLR9cascadeToll pathway
  1. Numbers represent the rank metric score of each gene which is used to position the gene in the ranked list. The ranking metric measures an individual transcript's correlation with the PGD phenotype.

IL1B1.162IL1B1.162ILIB1.162TLR60.805IL1B1.162TLR 60.805TLR60.805TLR60.805
NLRP31.135IL60.931ILIA0.952TLR10.796NFKB10.629TLR10.796TLR10.796TLR40.677
IL60.931CCL40.842IL60.931TLR40.677TLR20.608TLR40.677TLR40.677NFKB10.629
TNFAIP30.792TLR60.805IL1RN0.676IRAK30.636MAPK140.602IRAK30.636IRAK30.636TLR20.608
MEFV0.632TLR10.796IL1RAP0.673TLR20.608IL30.558TLR 20.608TLR20.608MAPK140.602
NFKB10.629TLR40.677IRAK30.636TLR80.453MAP2K60.542TLR 80.453TLR80.453MAP2K60.542
NAIP0.613NFKB10.629NFKB10.629TLR100.406TNF0.360TLR 100.406TLR100.406TLR100.406
MAPK140.602TLR20.608MAPK140.602TLR50.345NFKBIA0.309TLR50.345TLR50.345PGLYRP10.358
NLRC40.565MAPK140.602MAP2K60.542IRAK40.341CREBBP0.258IRAK40.341IRAK40.341FOS0.336
ILS0.558CCL30.583IL1R10.411LY960.259IKBKB0.238TBK10.320LY960.259NFKBIA0.309
CXCL10.539IL80.558IRAK20.410  DUSP10.229LY960.259PIK3C30.248EIF2AK20.282
CXCL20.524MAP2K60.542TNF0.360  SMAD30.214    LY960.259
CASP50.518PIK3CG0.464NFKBIA0.309        IKBKB0.238
CARDS0.413TLR80.453          CD140.200
CCL70.413TNF0.360            
CASP10.394TLR50.345            
BIRC20.394IRAK40.341            
TNF0.360FOS0.336            
BIRC30.358IFNAR10.321            
NFKBIA0.309TBK10.320            
CASP80.299NFKBIA0.309            
IKBKB0.238CASP80.299            
MAPK100.218LY960.259            
CCL20.199RAC10.258            
ERBB2IP0.196STAT10.254            
XIAP0.189PIK3CA0.240            
NOD20.187IKBKB0.238            
MAPK110.170PIK3CD0.232            
CCL110.164MAPK100.218            
  PIK3R50.200            
  CD 140.200            

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

In this study, we have defined key pathways involved in the development of PGD, using GSEA. Like most complex traits, the PGD phenotype likely manifests with multiple alterations in gene expression highlighted by fewer dominant pathways. The strength of GSEA lies in its ability to utilize biological information (e.g. published information about well-characterized biological pathways) to guide analysis so that multiple changes in individual genes acting as part of a network within a background of profound physiologic perturbation, as seen in transplantation, can be analyzed with an interpretable result [7]. Using this methodology, our results suggest inflammasome and innate immune-mediated processes are actively involved in the pathophysiology of PGD, when also taking into account changes seen during transplantation in control subjects.

Inflammasomes are proinflammatory macromolecular complexes that activate caspase-1, which results in IL-1β activation [12]. These complexes are part of an increasingly recognized stereotyped innate immunologic response to tissue damage. This threat is recognized in the host by pattern recognition receptors that are either localized in the cell membrane (e.g. TLRs) or in the cytoplasm (e.g. NLRs). Though the specific pathways of inflammasome activation are incompletely understood, danger- and pathogen-associated molecular pattern (DAMPs and PAMPs) recognition plays an important role [13]. Stimuli for activation of these pathways include bacterial and viral pathogens [14], lysosomal disruption [15], neutrophil or mitochondria derived reactive oxygen species (ROS) [16, 17] and cell apoptosis [17, 18]. Therefore, our findings may indicate that response to either pathogens (likely donor derived) or response to cell and tissue damage signals are key in PGD pathogenesis and warrant further investigation.

We have identified IL-1β, the major inflammasome-regulated effector cytokine, as being significantly correlated to several pathways potentially involved in PGD pathogenesis. IL-1β has consistently been demonstrated as a key mediator in other types of acute lung injury, and recent investigations by our group have implicated genetic variation in this pathway to associate with differential acute respiratory distress syndrome (ARDS) risk [12, 19-24]. It is possible that a subset of the donor population had subclinical lung injury present at the time of procurement that was augmented by ischemia and reperfusion leading to clinical PGD. Both direct (pneumonia, aspiration, hyperoxia, pulmonary contusion and reperfusion) and indirect (trauma, sepsis and transfusion) causes of ALI are recognized in critically ill patients [25, 26]. Several of these exposures are present in organ donors and are significant contributors to organ unsuitability for transplant [27]. Among organs used for transplant with these exposures, PGD may reflect organ injury which may have progressed to ALI had the donors not been transplanted [27]. Furthermore, brain death directly causes neurogenic pulmonary edema and ALI. Diffuse organ inflammatory responses associated with brain death impair lung function after implant and contribute to PGD [31-33]. Alternately, IL-1 signaling in response to inflammasome activation may represent a pathophysiological amplification of the immune response after reperfusion, perhaps as a result of different levels of tissue damage occurring during organ preservation.

Current methods of organ preservation rely on hypothermia and specially formulated perfusates to decrease metabolic injury. It is recognized that all non-enzymatic and enzymatic processes are reduced by 1.5–3-fold per 10°C temperature decreases from baseline and cold preservation has been the foundation for organ preservation [28]. Recent evidence suggests that hypothermia in the presence of oxygen can result in ROS-mediated hypothermic injury [28]. Among solid organs, the lung is unique because it is stored inflated and does not experience similar degrees of hypoxia [28-30]. Further, evidence in a rodent model demonstrates oxygen dependent dose related lung injury with increasing percentage of oxygen in the gas mixture used for inflation [28, 31]. Thus, the injury seen in PGD may be consequent to oxidant stress with resultant DAMP-regulated inflammasome activation, which may be a target for future ex vivo lung perfusion strategies. Alternatively, hidden donor microbial pathogens alone or in combination with subclinical organ injury may amplify ischemia/reperfusion injury to the extent that recipient antioxidant potential is overwhelmed.

There are several limitations of this study. First, our study design relied on the evaluation of cells from BALF at two time points (donor in situ and postreperfusion). While BALF has the advantage of sampling multiple cell types from a larger section of lung than lung biopsies, this methodology may introduce a sampling bias as cells from this source would be expected to be comprised mostly of alveolar macrophages, neutrophils, B cells, T cells and epithelial cells [32]. Additionally, this mixed cell population would be exclusively donor derived preprocurement and, likewise, donor alveolar cells would be most likely to contribute to gene expression measured after 1 h postreperfusion [33]. This sampling method was specifically chosen to highlight early pathways important in PGD pathogenesis. However, we acknowledge that there may be parenchymal and vascular contributions that may have been present in lung biopsy samples and not BAL, for example. Despite this limitation, our results are consistent with previous reports that TLR signaling pathways are important in lung injury [34]. Second, it is not possible to establish functional causation with our study design. However, our results serve to prioritize inflammasomes, and innate immune activation and modulation, for future study in PGD. Third, given the limited number of samples used for analysis it is possible that important changes in gene expression in single genes may have been overlooked and we may have been unable to demonstrate statistical differences in baseline demographics known to affect PGD. Fourth, our decision to use the FWER, a more conservative correction for multiple hypothesis testing than the false discovery rate (FDR), limited the number of pathways we considered significant. This conservative approach potentially excluded pathways important in PGD pathogenesis; however, we list the full results in the supplemental material (Table S1). Finally, our study lacks an external validation step. Unfortunately, there are no existent external datasets that can be used for replication; for this reason we have adopted a conservative multiple test correction.

In summary, we have demonstrated an association between inflammasome and innate immune activation and PGD. The effects of these pathways suggest importance of the inflammasome and innate immune signaling in PGD pathogenesis. This suggests a complex mechanism of injury initiation and modulation and therefore should stimulate further inquiry into the role of tissue damage from storage, as well as novel methods of identifying donor pathogens, such as donor microbiome analyses.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

This work was funded by NIH AI063589, HL087115, HL081619, HL096845, HL116656, HL090021 and RWJ AMFDP11642.

Disclosure

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's web-site.

FilenameFormatSizeDescription
ajt12283-sm-0001-SuppData.docx555K

Table S1: Expanded gene set enrichment analysis. NES, normalized enrichment score (accounts for differences in gene set size and in correlations between gene sets and the expression dataset to allow for comparisons across gene sets); NOM p-val, nominal p value (estimates the statistical significance of the enrichment score for a single gene set); FDR, false discovery rate (estimated probability that a gene set with a given NES represents a false positive finding); FWER, familywise-error rate (conservative correction that seeks to ensure that the list of reported results does not include even a single false-positive gene set).

Table S2: Ranked gene list of PGD positively correlated transcripts.

Figure S1: All significantly enriched gene sets meeting FWER corrected cutoff. The top portion of the plot shows the running enrichment score (ES) for the gene set from decreasing values of the rank list. The score at the peak of the plot (the score furthest from 0.0) is the ES for the overall gene set. The middle portion of the plot shows where the members of the gene set appear in the ranked list of genes. The bottom portion of the plot shows the value of the ranking metric as the list of ranked genes decreases in value. The ranking metric measures an individual transcript's correlation with the PGD phenotype.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.