Quantitative Proteomics Using Formalin-fixed, Paraffin-embedded Biopsy Tissues in Inflammatory Disease

Background: Investigations in human disease pathogenesis have been hampered due to paucity of access to fresh-frozen tissues (FFT) for use in global, data-driven methodologies. As an alternative, formalin-fixed, paraffin-embedded (FFPE) tissues are readily available in pathology banks. However, the use of formalin for fixation can lead to the loss of proteins that appear during inflammation, thus introducing an inherent sample bias. To address this, we compared FF and FFPE tissue proteomics to determine whether FFPE-tissue can be used effectively in inflammatory diseases. Methods: Adjacent kidney slices from lupus nephritic mice were processed as FFPE or FFTs. Their tissue lysates were run together using proteomics workflow involving filter-aided sample preparation, in-solution dimethyl isotope labeling, StageTip fractionation, and nano-LC MS/MS through an Orbitrap XL MS. Results: We report a >97% concordance in protein identification between adjacent FFPE and FFTs in murine lupus nephritic kidneys. Specifically, proteins representing pathways, namely, ‘systemic lupus erythematosus’, ‘interferon-α’, ‘TGF-β’, and ‘extracellular matrix’, were reproducibly quantified between FFPE and FFTs. However, 12%−29% proteins were quantified differently in FFPE compared to FFTs, but the differences were consistent across experiments. In particular, certain proteins represented in pathways, including ‘inflammatory response’ and ‘innate immune system’ were quantified less in FFPE than in FFTs. In a pilot study of human FFPE tissues, we identified proteins relevant to pathogenesis in lupus nephritic kidney biopsies compared to control kidneys. Conclusion: This is the first report of lupus nephritis kidney proteomics using FFPE tissue. We concluded that archived FFPE tissues can be reliably used for proteomic analyses in inflammatory diseases, with a caveat that certain proteins related to immunity and inflammation may be quantified less in FFPE than in FFTs.

. An example of protein identification and quantification process. Figure S4. Quantifying proteins in fresh frozen (FF) vs. formalin-fixed, paraffin-embedded (FFPE) lupus nephritic kidney relative to healthy kidney tissues.
Table S1. The 79 proteins that were most decreased in formalin-fixed, paraffin-embedded (FFPE) samples, compared to fresh frozen (FF), across both experiments: provided as a separate excel sheet (Supplemental Table S1.xlsx). Table S2. The proteins within the pathways illustrated in Figure 4B.

SUPPLEMENTARY FIGURES
Supplemental Figure S1. Experimental Design of the study. Four slices, each of 5µm, from a lupus nephritic kidney were processed as fresh frozen (FF) and formalin-fixed, paraffin-embedded (FFPE) tissues, and run in two independent experiments: FFT protein lysate was run as two exact technical replicates (FF1 and FF2) and a contiguous FFPE slice lysate (FFPE1); the three lysates were labeled and run together (experiment 1). In experiment 2, two mirror image, but not contiguous, kidney slices from the same kidney were run as two exact technical replicates (FF3 and FF4), and a FFPE (FFPE2). Further steps included trypsin digestion, triplex dimethyl isotope labeling, and the combination of each condition's peptides into a single solution. Solutions were fractionated via a stage-tip method and analyzed with nLC-MS/MS. Figure S2. Human kidney biopsies used in the study. H&E stained kidney biopsy sections from two each of SLE patients (lower panels, B and B1) and control subjects (upper panels, A and A1) and are shown. Control kidneys have normal appearing glomeruli and tubules. Kidney biopsies from SLE patients show mild mesangioproliferative LN with mesangial hypercellularity (B) and diffuse proliferative LN with marked glomerular proliferation and accentuated lobular architecture, karyorrhexis, crescent formation, and interstitial infiltration (B1). Four µm sections of these FFPE kidney biopsies were obtained from UCLA Pathology Core in accordance with the approved Institutional Review Board protocol. Tissues were deparafinnized with 100% xylene for 10 minutes, 100% ethylic alcohol, and rehydrated with descending ethylic alcohol 95% in water v/v, (1X), 80%(1X) and 50% (1X). Samples were prepared using the Liquid Tissue MS protein prep kit (Expression Pathology, Rockville, MD) following a previously described protocol [1]. Protein concentration in the sample was measured using the micro BSA protein assay method. The extracted tryptic peptide samples (3 µg) were fractionated on a HPLC column, and analyzed with a quadruple time-of-flight hybrid mass spectrometer (Applied Biosystems Q-STAR XL) with nanobore LC-MSMS capability and equipped with ESI, nanospray and APCI sources. We only identified proteins that were found in 95% confidence interval or higher in relation to the total ion score and strength as determined by the MASCOT software and SwissProt database. Automatic isotope correction was carried out by both software packages using the values supplied with the Applied Biosystems reagents.

Supplementary
Then, the UniProt database was used to elucidate the biological process, cellular location and molecular function of each individual protein. Pathway mapping was conducted using the publicly available DAVID Functional Annotation and Bioinformatics Analysis Software [2].

Supplementary Figure S3: An example of protein identification and quantification process.
A) In-solution dimethyl labeling and quantification for complement factor H. This shows the extracted ion chromatograms and the mass spectra from a triplex stable dimethyl isotope labeled peptide (H2N-LYYEESLRPNFPVSIGNK). This peptide was produced from trypsin cleavage and is specific to the complement factor H protein. Once differentially labeled, one FFPE condition as well as two technical replicate FF conditions were combined in a 1:1:1 ratio. This resulted in the simultaneous detection of each of the labeled peptide ion fragments. As noted under the MS1 peptide quantification window, an expected m/z shift increase was observed among the conditions. This allowed for the relative quantification of peptide between each pair of conditions via the area under the curve from the extracted ion chromatogram [3].

B)
Example of quantification of complement factor H in each pair of conditions. The relative quantification of complement factor H peptide between each pair of conditions in both experiments is illustrated. These values were derived from a geometric mean of the quantification from each of its detected peptides, thereby providing quantification for the complement factor H protein. The quantification of all proteins analyzed was conducted via the Maxquant analysis and Perseus visualization system in a manner identical to this. Figure S4: Quantifying proteins in fresh frozen (FF) vs. formalin-fixed, paraffin-embedded (FFPE) lupus nephritic kidney relative to healthy kidney tissues. Kidney slices from 10-month-old lupus-prone NZM.2328 female mice with high-grade proteinuria were processed as FFPE and FF tissues. Kidney slices from 10-month-old healthy BALB/c female mice were processed as FF tissues. Proteins extracted from BALB/c-FF, NZM-FF, and NZM-FFPE tissues were dimethyl isotope labeled, and processed together for quantitative proteomics evaluation, as described in Methods. A)