A new method for isolation of interstitial fluid from human solid tumors applied to proteomic analysis of ovarian carcinoma tissue

Abstract

Major efforts have been invested in the identification of cancer biomarkers in plasma, but the extraordinary dynamic range in protein composition, and the dilution of disease specific proteins make discovery in plasma challenging. Focus is shifting towards using proximal fluids for biomarker discovery, but methods to verify the isolated sample's origin are missing. We therefore aimed to develop a technique to search for potential candidate proteins in the proximal proteome, i.e. in the tumor interstitial fluid, since the biomarkers are likely to be excreted or derive from the tumor microenvironment. Since tumor interstitial fluid is not readily accessible, we applied a centrifugation method developed in experimental animals and asked whether interstitial fluid from human tissue could be isolated, using ovarian carcinoma as a model. Exposure of extirpated tissue to 106 g enabled tumor fluid isolation. The fluid was verified as interstitial by an isolated fluid:plasma ratio not significantly different from 1.0 for both creatinine and Na(+), two substances predominantly present in interstitial fluid. The isolated fluid had a colloid osmotic pressure 79% of that in plasma, suggesting that there was some sieving of proteins at the capillary wall. Using a proteomic approach we detected 769 proteins in the isolated interstitial fluid, sixfold higher than in patient plasma. We conclude that the isolated fluid represents undiluted interstitial fluid and thus a subproteome with high concentration of locally secreted proteins that may be detected in plasma for diagnostic, therapeutic and prognostic monitoring by targeted methods.

Figure 1. Overview of the tumor fluid validation assay.

A: Illustration of the principle for validation of tumor fluid as interstitial fluid. The concentrations of the extracellular markers, Na⁺ and creatinine were compared in isolated tumor fluid and plasma. When the isolated fluid derives solely from the interstitial space, the concentrations of the markers are equal to that in plasma, while addition of intracellular fluid will result in a lower tumor fluid:plasma ratio. To assess the presence of intracellular proteins, mass spectrometric analysis was used, the resulting protein lists were filtered based on known house keeping and extracellular proteins and the change in the number of spectra for intracellular proteins relative to albumin was used as an indicator of intracellular protein contamination. B: Measured tumor fluid:plasma ratio for creatinine and Na⁺ for tumor fluid (black) and frozen tumor fluid (grey). Values are mean ± SEM. *: Significantly different from tumor fluid (p<0.05) (Mann-Whitney test).

Figure 2. Measurement of creatinine by HPLC.

A: The complete chromatogram from HPLC creatinine analysis. Deproteinized samples were injected on two strong cation exchange columns in series at pH 4.68, where creatinine will be positively charged and retained. TCA and other contaminants are not retained at this pH and were washed out in the first five minutes, while creatinine was passed into a third column for further separation and fixation. By changing the buffer pH to 7.01 in the third column the creatinine molecule was rendered neutral, reducing retention and thus producing a sharp and well-defined peak with absorbance at UV 234 nm. After a focusing step the buffer pH changes back and the columns were regenerated. B: Calibration curve for creatinine with standard concentrations 0, 10, 40, 80, 100 and 200 mM. C: Chromatograms for creatinine standards 0, 10, 40 and 80 µmol/L used for calibration. D: By analyzing tumor fluid and plasma samples before and after enzyme hydrolysis by creatininase and creatinase the method's specificity could be assessed. The chromatograms from analysis of plasma, tumor fluid before and after enzyme hydrolysis, and a blank run are shown.

Figure 3. Colloid osmotic pressure (COP) in plasma and tumor fluid samples.

COP was measured in paired plasma and tumor fluid samples using a colloid osmometer with a 30 kDa cut off-membrane. Corresponding pressure measurements in plasma and tumor fluid have been connected.

Figure 4. Chromatographic evaluation of tumor fluid, plasma and ascites by size-exclusion chromatography.

A: Size exclusion-chromatography of native tumor fluid, plasma and ascites, results are normalized in respect to albumin. As expected, albumin was a major constituent, and all three samples were dominated by major plasma proteins. The chromatogram for tumor fluid indicates that there were some low molecular weight proteins present in tumor fluid that were not found in ascites or plasma. B: After immunoaffinity depletion of the 14 most abundant plasma proteins, the ascites and plasma samples were still similar, except from an albumin peak for ascites that can indicate incomplete depletion of albumin. Tumor fluid reveals large differences in the protein composition compared to plasma and ascites, indicating the presence of tumor specific proteins in tumor fluid. Results were normalized in respect to the peak at a retention time of 29 minutes.

Figure 5. Overview of proteomic results for tumor fluid compared to plasma.

Pooled plasma and tumor fluid samples of three patients with ovarian cancer and a control pool taken from five women operated for suspected ovarian carcinomas later shown to be benign were immunoaffinity depleted, fractionated by reversed phase and strong cation exchange chromatography at protein level and nano-reversed phase liquid chromatography at peptide level before analysis by tandem mass spectrometry. Venn-diagram of the three proteomes identified from tumor fluid (left), patient plasma (lower right) and control plasma (upper right). Analyzing patient plasma compared to control plasma yielded few new proteins, while the tumor fluid contains a large amount of proteins not detected in plasma.

Figure 6. Comparison of tumor fluid proteome with published protein data.

The proteins found in tumor fluid in the present study (right) compared to proteomes presented for ascites by Gortzak-Uzan et al. (lower left) and ovarian cancer cell line cultures of Faca et al. (upper left). In ascites there were 268 proteins in addition to the proteins found by proteomic analysis of ovarian cancer cell lines cultures, while in the tumor fluid proteome, although with a much lower total number of proteins identified, there were 454 proteins that were neither detected in ascites nor in cell line cultures.