Micro-dissection and integration of long and short reads to create a robust catalog of kidney compartment-specific isoforms

Hongyang Li, Ridvan Eksi, Daiyao Yi, Bradley Godfrey, Lisa R. Mathew, Christopher L. O'Connor, Markus Bitzer, Matthias Kretzler, Rajasree Menon, Yuanfang GuanPLOS Computational Biology 2022 · DOI

TL;DR

This paper presents a novel pipeline for precisely identifying gene isoforms within specific microscopic compartments of the kidney (glomeruli and tubulo-interstitial regions). By integrating long-read and short-read sequencing data, they've created a more accurate and comprehensive catalog of kidney-specific transcripts, revealing distinct functional pathways in each compartment.

What Problem Does This Paper Address?

Understanding how genes function within specific tissues is crucial, but this becomes complex when considering alternative splicing, which allows a single gene to produce multiple protein isoforms. This complexity is particularly challenging at the microscopic level within a tissue like the kidney, which has distinct functional compartments (glomeruli for filtration, tubulo-interstitial for reabsorption and secretion). Traditional RNA sequencing methods often struggle to accurately identify and quantify these isoforms, especially when dealing with short reads that can lead to ambiguous assignments and errors in expression analysis. The paper aims to overcome these limitations by developing a method to dissect these microscopic compartments and then use advanced sequencing and computational techniques to build a robust catalog of their unique isoforms.

Methods

The core of this work lies in a multi-pronged experimental and computational approach:

  • Micro-dissection: Healthy human kidney tissue was physically separated into glomerular and tubulo-interstitial compartments. This is a critical step for compartment-specific analysis.
  • Long-Read Sequencing (PacBio Iso-Seq): This technology generates full-length transcript sequences, which are essential for accurately identifying the complete structure of isoforms and resolving ambiguities that short reads can't.
  • Short-Read Sequencing (Illumina RNA-Seq): This provides high-throughput, lower-error data that is used to validate the findings from long-read sequencing, particularly for splice junctions and transcription start/end sites.
  • Integrative Computational Pipeline:
    • Alignment: Both long and short reads were aligned to the human genome.
  • Isoform Identification: PacBio reads were processed to identify consensus full-length transcripts (CFLs).
  • Feature Identification: Transcription start sites (TSS), splice junctions, and transcription end sites (TES) were identified from both long and short reads.
  • Validation: A rigorous validation process was employed, requiring splice junctions to be supported by either short-read data or existing annotations.
  • Collapsing and Annotation: Validated isoforms were collapsed to remove redundancy, and then compared against the GENCODE annotation set to classify them as known, potentially novel, or intergenic transcripts.

Datasets

  • Human Kidney Cortex Samples: Obtained from five patients undergoing tumor nephrectomies and 20 healthy samples from the European Renal cDNA Bank (ERCB) study.
  • GENCODE Annotation Set (version 24): Used as a reference for validating identified transcripts and features.
  • NCBI RefSeq: Mentioned for comparison of gene models.

The paper introduces a new dataset of compartment-specific isoforms derived from their micro-dissected kidney samples, which serves as a valuable resource for the research community.

Key Insights and Ideas

  • Long reads are crucial for isoform resolution: Relying solely on short reads leads to significant errors in isoform identification. Long reads provide the necessary completeness to accurately define transcript structures.
  • Micro-dissection is key to functional understanding: The kidney's distinct compartments have different physiological roles, and this study demonstrates that they also possess distinct isoform expression profiles, which are likely to underlie these functional differences.
  • Integration is more than the sum of its parts: Combining long and short reads provides a powerful synergy. Long reads identify potential isoforms, while short reads validate their critical features, leading to a more robust and reliable catalog.
  • Novel isoforms abound: The study identified a substantial number of potentially novel transcripts in both glomerular and tubulo-interstitial compartments, highlighting the incompleteness of current annotation databases for specialized tissues.

Key Results

  • Identification of Novel Transcripts:
    • Glomerular compartment: 14,739 annotated transcripts and 17,268 potentially novel transcripts.
  • Tubulo-interstitial compartment: 14,259 annotated transcripts and 13,118 potentially novel transcripts.
  • Validation Success: Approximately 65% of PacBio junctions were validated in glomerular compartments, and a similar percentage in tubulo-interstitial compartments.
  • Distinct Functional Pathways: Preliminary analysis revealed significantly different enriched KEGG pathways between the glomerular and tubulo-interstitial compartments, underscoring the functional specialization at the isoform level. For example, glomerular-only transcripts were enriched for NAFLD and Rap1 signaling, while tubulo-interstitial transcripts were enriched for metabolic and ribosomal pathways.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that exact TSS and TES coordinates from PacBio data don't always agree perfectly with annotated coordinates, and that Quiver polishing of PacBio reads can sometimes obscure introns.
  • The study focused on healthy tissues; understanding isoform changes in disease states is a future direction.
  • While short reads help validate junctions, they can still be ambiguous for TSS and TES identification.
  • Future Work:
    • Applying this micro-dissection and isoform identification pipeline to other tissues.
  • Connecting the identified novel transcripts to isoform function prediction methods to elucidate their roles in activated pathways.
  • Investigating isoform changes between healthy and disease states within the same microscopic structures.
  • Integrating these findings with single-cell level sequencing to understand expression variability.

Summary updated

BibTeX
@article{li2022micro,
  author = {Hongyang Li and Ridvan Eksi and Daiyao Yi and Bradley Godfrey and Lisa R. Mathew and Christopher L. O'Connor and Markus Bitzer and Matthias Kretzler and Rajasree Menon and Yuanfang Guan},
  title = {{Micro-dissection and integration of long and short reads to create a robust catalog of kidney compartment-specific isoforms}},
  journal = {PLOS Computational Biology},
  year = {2022},
  doi = {10.1371/journal.pcbi.1010040},
}
Figure