Transcriptomic analysis identifies novel candidates in cardiorenal pathology mediated by chronic peritoneal dialysis

Peritoneal dialysis (PD) is associated with increased cardiovascular (CV) risk. Studies of PD-related CV pathology in animal models are lacking despite the clinical importance. Here we introduce the phenotypic evaluation of a rat model of cardiorenal syndrome in response to chronic PD, complemented by a rich transcriptomic dataset detailing chronic PD-induced changes in left ventricle (LV) and kidney tissues. This study aims to determine how PD alters CV parameters and risk factors while identifying pathways for potential therapeutic targets. Sprague Dawley rats underwent Sham or 5/6 nephrectomy (5/6Nx) at 10 weeks of age. Six weeks later an abdominal dialysis catheter was placed in all rats before random assignment to Control or PD (3 daily 1-h exchanges) groups for 8 days. Renal and LV pathology and transcriptomic analysis was performed. The PD regimen reduced circulating levels of BUN in 5/6Nx, indicating dialysis efficacy. PD did not alter blood pressure or cardiovascular function in Sham or 5/6Nx rats, though it attenuated cardiac hypertrophy. Importantly PD increased serum triglycerides in 5/6Nx rats. Furthermore, transcriptomic analysis revealed that PD induced numerous changed transcripts involved with inflammatory pathways, including neutrophil activation and atherosclerosis signaling. We have adapted a uremic rat model of chronic PD. Chronic PD induced transcriptomic changes related to inflammatory signaling that occur independent of 5/6Nx and augmented circulating triglycerides and predicted atherosclerosis signaling in 5/6Nx LV tissues. The changes are indicative of increased CV risk due to PD and highlight several pathways for potential therapeutic targets.


Supplemental Methods:
Additional Details of RNA-sequencing Frozen tissues were pulverized under liquid nitrogen, with ~100mg placed in Trizol® for RNA extraction following manufacturer's instructions (Ambion™). Samples were submitted to Genomic Sciences and Precision Medicine Center (GSPMC) for RNA quality assessment using an RNA Fragment Analyzer and quantification by Qubit (ThermoFisher Scientific). A total of 500 ng of high-quality RNA from each sample was used for automated mRNA library construction utilizing the Illumina TruSeq stranded mRNA library prep kit with unique dual 8 base pair indexes for samples on each run.
The library quality was evaluated by Kapa qPCR Quantification, MiSeq Nano 50 cycle QC run, prior to proceeding with sequencing. Samples were multiplexed 64 samples per flow cell and pair-end sequenced with an Illumina NovaSeq6000 sequencer.
In analysis, adapter sequences were removed from the output reads; sequences with low quality (base quality < 13) at both ends of reads were further trimmed and trimmed reads with less than 25 bp were removed using the Trim Glore tool [1]. The trimmed short sequencing reads were aligned to the rat reference (rn6) using HISAT2 (v2.2.1) with mammalian default parameters [2]. Transcript construction, quantification and normalization of transcript abundance was performed using StringTie2 (v2.1.5) [3].
Statistical analysis of sequencing results was performed comparing LV RNA from 5/6Nx rats that received PD and those that did not. This comparison was also made in remnant kidney RNA from these groups. Differential gene expression was determined using the R package DESeq2 [4]. The Benjamini-Hochberg method was used to control false discovery rate in differential expression analysis to provide an adjusted P-value. A list of DEGs with corresponding log2 fold-change for each comparison of interest   was uploaded to IPA for core analysis utilizing the "Ingenuity Knowledge Base" (IPA; QIAGEN, Hilden, Germany) as the "Reference Set". Data resulting from IPA Graphical Summary in the Subcellular view are presented here, which allows visualization of enrichment established pathways with DEGs from a given data set. The enrichment is ranked within these pathways to provide a "-log (P-value)" for enrichment based on the # of differentially expressed genes in the dataset relative to the overall number of genes represented in the pathway. Specifically, in this analysis, we utilized the "Graphical Summary" graphics with default settings, which include a color-coding of pathways that are predicted to be activated or inhibited based on "z-score". This z-score is also referred to as the activation score because it is used as an index of biological function based on established directionality of effects resulting from gene-gene interactions.

Metascape Analysis
The list of DEGs from each comparison group were uploaded and species R. norvegicus was selected for both input and analysis. Express analysis was then performed. The Metascape [5] tool then utilized this input information to generate "Enriched Ontology Clusters" for graphical presentation of datasets.

Sequencing Results
The mapping rate for this dataset was >96% for all samples (Table S1). The number of genes with an adjusted P-value <0.05 in each comparison (Differentially Expressed Genes; DEGs) are indicated in Table S2, along with the number of mapped genes from this list that represented in Metascape, and IPA tools used in pathway analysis.

Changes in LV Tissue resulting from PD in Sham and 5/6Nx Rats.
The LV DEGs with the top 15 largest log2 fold-change increase or decrease in response to PD in either the Sham group or the 5/6Nx group are listed in Tables S3 and   S4, respectively. Also listed are the P-value and adjusted P-value for that comparison.
The IPA Graphical Summary Representation highlights numerous processes and pathways are predicted to be activated (orange) or inhibited (blue) with PD based DEGs in Sham ( Figure S1A) and 5/6Nx rats ( Figure S2A). Metascape analysis identified numerous enrichment terms based upon the same DEG input from our LV tissues in Sham ( Figure S1B) and 5/6Nx rats ( Figure S2B).

Changes in Kidney Tissue Resulting from PD in Sham and 5/6Nx Rats.
We performed the same analysis with the DEGs from kidney tissues as described above for LV tissue. This Metascape analysis yielded fewer enriched pathways and there were not clear pathological processes identified in both Sham ( Figure S3A) and 5/6Nx rats ( Figure S4A). The enrichment through IPA analysis of Sham kidneys almost exclusively represented upregulated genes ( Figure S3B). When MAP was applied to the 5/6Nx kidney dataset there were no pathways that were predicted to either be activated or inhibited based on our limited input ( Figure S4B).