Background/Purpose

Lupus nephritis (LN) progresses from mild focal inflammation, to diffuse proliferative nephritis, to fibrosis and end-stage renal disease. Though the understanding of LN has progressed, there is a need to use global, data-driven research methodologies to elucidate its molecular pathogenesis. As a foundation for this goal, we aimed to develop a quantitative proteomics workflow that can directly study formalin-fixed, paraffin-embedded (FFPE) archived clinical tissues. This applicable workflow could therefore provide a powerful tool to study the progression of LN, albeit if we can trust that data obtained are without substantial sample processing bias.

Methods

To obviate the need for large pieces of human LN tissue, we used kidney tissues from lupus-susceptible NZM.2328 mice that develop glomerulonephritis that mimics LN in humans. Identical transverse kidney tissue cuts from 10-month-old female NZM-2328 mice with high-grade proteinuria were processed as FFPE and fresh frozen tissue (FFT). FFPE and FFT sections were digested with trypsin and stable isotope labeled for protein identification and quantification. Our workflow includes a combination of methodologies including filter aided sample preparation (FASP), in-solution dimethyl isotope labeling, strong cation exchange StageTip fractionation, along with nano-LC MS/MS through an Orbitrap XL mass spectrometer. Two separate experiments were run where three conditions were studied in each: two exact technical replicate FFT conditions and one FFPE condition. Within our workflow, combining FASP and in-solution dimethyl isotope labeling, relative quantitative values were obtained via direct comparison of each pair of conditions within each experiment.

Results

We developed and validated a workflow that allows for a direct comparison of FFPE tissue to FFT. Through our workflow validation experiments, we observed an almost 100% protein identification overlap between FFPE and FFT from a LN kidney. A consistent identification of over 1400 proteins in both FFPE and FFT indicate no selection bias with tissue processing. Although, quantification differences did exist when comparing FFPE-to-FFT, the quantitative changes (quantification ratios) of proteins in FFPE tissues were consistent across replicate experiments. This reliability is seen with global hierarchical clustering as well as with specific protein categories such as TGFβ signaling, the KEGG SLE annotated pathway, the GSEA annotated lupus CD4 T cell vs. myeloid function upregulation, and B cell function related proteins, which have been implicated in LN pathogenesis.

Conclusion

Our methodology is the first to directly compare FFT and FFPE tissue in a manner that can be readily applied to archived clinical samples. Our results demonstrate the utility of this workflow by its ability to equally identify proteins between FFPE and FFT, minimizing sample processing bias, and by providing consistent protein quantification values of FFPE tissue between technical replicates and across separate experiments. We conclude that this clinically oriented proteomics workflow, when applied to archived, FFPE tissue can be reliably utilized to study LN pathogenesis.