Isolation and Quantification of Metabolite Levels in Murine Tumor Interstitial Fluid by LC/MS.

Cancer is a disease characterized by altered metabolism, and there has been renewed interest in understanding the metabolism of tumors. Even though nutrient availability is a critical determinant of tumor metabolism, there has been little systematic study of the nutrients directly available to cancer cells in the tumor microenvironment. Previous work characterizing the metabolites present in the tumor interstitial fluid has been restricted to the measurement of a small number of nutrients such as glucose and lactate in a limited number of samples. Here we adapt a centrifugation-based method of tumor interstitial fluid isolation readily applicable to a number of sample types and a mass spectrometry-based method for the absolute quantitation of many metabolites in interstitial fluid samples. In this method, tumor interstitial fluid (TIF) is analyzed by liquid chromatography-mass spectrometry (LC/MS) using both isotope dilution and external standard calibration to derive absolute concentrations of targeted metabolites present in interstitial fluid. The use of isotope dilution allows for accurate absolute quantitation of metabolites, as other methods of quantitation are inadequate for determining nutrient concentrations in biological fluids due to matrix effects that alter the apparent concentration of metabolites depending on the composition of the fluid in which they are contained. This method therefore can be applied to measure the absolute concentrations of many metabolites in interstitial fluid from diverse tumor types, as well as most other biological fluids, allowing for characterization of nutrient levels in the microenvironment of solid tumors.

The nutrient environment that a cancer cell has access to is predominantly composed of interstitial fluid (Wiig and Swartz, 2012). Understanding the nutrient content of tumor interstitial fluid would provide insight into the metabolic constraints imposed upon tumor cells by their environment. There exist multiple methodologies for isolating interstitial fluid from normal organs and from tumors (Wiig et al., 2010). However, early attempts to measure the nutrient content of interstitial fluid were limited by their inability to measure multiple metabolites, and consequently our knowledge of nutrient availability in tumors is restricted to a few metabolites in a limited number of animal tumor models (Burgess and Sylven, 1962;Gullino et al., 1964). The advent of mass spectrometry has allowed for detection of many metabolites simultaneously. However, despite technological advances, metabolomics studies are complicated by the fact that components present in biological fluids can suppress or enhance the detection of specific metabolites. These discrepancies in detection of metabolites between different biological fluids are termed "matrix effects," and are a major confounding factor in comparing metabolite concentrations between different biological fluids and in accurately quantitating metabolites in those fluids (Panuwet et al., 2016;Sullivan et al., 2019).
Here we demonstrate a method for centrifugation-based isolation of tumor interstitial fluid and the subsequent absolute quantitation of numerous metabolites within that fluid using stable isotope dilution, a technique in which stable isotope-labeled internal standards for metabolites of interest are added to experimental samples. These stable isotope internal standards are subject to the same matrix effects as the corresponding metabolite in the sample and can be distinguished by their increased mass compared to the metabolites in the sample. To measure many metabolites simultaneously, we first quantitate the concentrations of 13 C metabolites from an extract of polar metabolites from yeast that are cultured with 13 C isotopically labeled glucose as the sole carbon source. This quantitated yeast extract is then used as an internal standard that allows for reliable quantification of targeted metabolites in biological samples while minimizing systematic error from matrix effects. This approach provides a robust method to quantitate polar metabolites in biological fluids and complements similar existing isotope dilution based methods, such as the commercially available Biocrates AbsoluteIDQ kits (Gieger et al., 2008) that primarily quantify non-polar lipids in biological samples.
The absolute quantitation of metabolite levels enabled by this protocol can allow for direct comparison of interstitial fluid composition in diverse tumor types, providing the opportunity to systematically interrogate nutrient availability in animal models of diverse cancers and human tumor samples. Further, the absolute quantification of metabolite levels in interstitial fluid allows for the generation of tissue culture media that mimics physiological conditions found in a tumor, thus expanding the range of in vitro/ ex vivo experiments that can be carried out under physiological nutrient conditions. Most broadly, this protocol provides a method to absolutely quantify many metabolites simultaneously in complex biological fluids, which can be used to study the metabolic composition of any biological material. 3 www.bio-protocol.org/e3427  a. The number of biological replicates needed for studies will depend on the variability between samples. For animal studies, where a large number of variables can be controlled (i.e., tumor genetics, tumor size, animal genetics, animal diet, time of interstitial fluid isolation), variability will likely be smaller than for human samples. Additionally, the number of replicates required will depend on the intended purpose of the experiment to be performed.
For example, if the intended purpose is to determine if there is a nutritional difference in the interstitial fluid between two tumor types, it is important to determine an effect size between the groups in addition to variability between samples in order to estimate sample sizes needed. We recommend generating pilot data or utilizing previously published data on TIF 3. We recommend two people work together to isolate TIF and cardiac blood to increase the speed of TIF harvest to prevent alterations in TIF composition due to prolonged periods of ischemia occurring between euthanasia and TIF harvest. In our own experiments, dissection was completed in ~2 min. and we found limited evidence of ischemia altering tumor metabolite levels However, depending on experimental goals, the full analysis utilizing all 7 standard pools may not be required. Subsets of the chemical standard pools can be used that cover analytes of interest if the full analysis is not needed. Note though that individual metabolites in the standard pools provided in this protocol have been carefully selected so as to avoid metabolites with the same m/z (isomeric and isobaric species) being in the same pool. In addition, metabolites that could be generated by in-source fragmentation from larger metabolites have been separated. 6. The description of the liquid chromatography-mass spectrometry analysis in this protocol (Procedure D) is a rough guideline for experienced operators of such instruments to perform the analysis described. Successful mass spectrometry analysis of the samples will require a trained UHPLC and Thermo Scientific hybrid quadrupole-Oribtrap mass spectrometer operator.
B. Isolation of TIF and plasma from tumor bearing animals 1. Prepare a TIF isolation tube (Figure 1 A).
a. Take a nylon filter and place it over the top of a 50 ml conical tube.
b. Tape the filter down using lab tape. Make sure the filter is affixed somewhat loosely to the top of the conical tube, such that the tumor can push the filter down slightly into the tube. 2. Prepare materials in advance to allow for rapid mouse dissection.
a. Put pre-chilled (4 °C) saline (~25 ml) into a Petri dish for washing the tumor.
b. Make a ~4 cm square piece of Whatman paper for drying the tumors after the saline rinse.
c. Have a 25G TB syringe ready for cardiac blood collection.
d. Label and pre-chill one EDTA-coated plasma collection tubes on ice for 10 min prior to mouse dissection.
3. Euthanize the mouse by cervical dislocation. 4. Spray around the incision site with 70% ethanol to prevent contamination of samples with hair. We have analyzed TIF samples both before and after 2 months of storage at -80 °C (avoiding freeze-thaw cycles) and detected similar metabolite concentrations after storage 12 www.bio-protocol.org/e3427   5. If using an UHPLC system that has a separate needle wash, fill this with acetonitrile. 6. Connect a SeQuant ® ZIC ® -pHILIC 5 μm 150 x 2.1 mm analytical column to the Guard column using the connector supplied in the Guard kit.
7. Connect the column and guard to your UHPLC system using standard techniques.
8. Set the column oven temperature to 25 °C. 9. Set the autosampler temperature to 4 °C. 10. Set initial conditions: set the flow rate to 0.150 ml/min with 80% B. Record initial pressure value.

Note: ZIC-pHILIC columns cannot tolerate such high back pressures and injection volumes as typical reverse phase columns. Keep an eye on the back pressure and do not let it exceed the maximum pressure recommended by the manufacturer. It is good practice to set a maximum pressure in your method that is below that set by the manufacturer to avoid damage to the column.
11. Equilibrate the column with starting conditions (80% B) for 30 min prior to running anything on the system. 12. Check the mass calibration on the mass spectrometer. If the mass has not been calibrated within the last week, or if it fails the mass check, recalibrate using the standard calibration mixes recommended by the manufacturer. In addition, perform a custom low-mass calibration by spiking glycine and aspartate into the calibration mix, or as recommended by the manufacturer. 13. Ensure your entire LCMS system performance is acceptable by running system suitability tests, such as injecting a mixture of amino acids onto your column and into the MS. Check for signal intensity as well as peak shape and separation. Table 1 below for the UHPLC gradient:  15. Operate the mass spectrometer in full-scan, polarity-switching mode, with a scan range of 70-1000 m/z. Include an additional narrow-range scan from 220 to 700 m/z in negative mode to improve detection of nucleotides. Use the parameters shown below in Tables 2 and 3    c. Follow with a system suitability test (SST) injection. We use 80% methanol containing 13

Data analysis
A. Identify metabolite peaks. This protocol will describe peak identification in Thermo Scientific Xcalibur, but could be adapted with any other peak identification method.
1. Generate a processing method file that will be used to identify peaks for each metabolite of interest: a. Create a new processing method.
b. Load a .raw file containing LC/MS data derived from an external standard sample that contains the metabolite of interest as well as a 13 C internal standard for that metabolite.
Note: Typically using an external standard sample that is in the middle of the standard curve 17 www.bio-protocol.org/e3427 e. In the processing method, select either positive ionization mode or negative ionization mode depending on whether you will be searching for a positively charged ion or a negatively charged ion. iii. Open a .raw file of a different external standard sample with the metabolite of interest at a lower concentration.
1) Note which peaks that match the exact mass of the ion of interest decrease in area.
2) Repeat with each of the external standard samples that contain the metabolite of interest, checking which peak areas track with the expected amount of the metabolite.
3) Refer to MS/MS data to confirm peak identification.
iv. Open a .raw file that does not contain the metabolite of interest. Ensure that any candidate peaks are not present in this .raw file.
v. Search for the exact mass of the 13 C labeled version of the ion of interest.

Note: This peak should be approximately the same area in all samples.
vi. Check that the retention time of the 13 C labeled standard peak exactly matches that of the candidate peak.
vii. Repeat for all metabolites of interest.
g. Assign 13 C labeled standards as internal standards for their corresponding 12 C metabolites.
For metabolites with no 13 C internal standard, assign a 13 C metabolite with a similar retention time as the internal standard. a. Once all peaks have been automatically picked, manually inspect every peak for each metabolite and for each sample to ensure that all peaks have been correctly identified.
Some common examples of errors that occur with automatic peak picking algorithms: i. Incorrect peak was picked: this can occur for isobaric compounds with similar retention times, such as leucine and isoleucine.
ii. Peak was not fully picked from baseline to baseline.
iii. Peak was picked but overlaps with a second peak: this occasionally happens where biological samples have an overlapping peak that was not present in the external standards. If this is the case, this metabolite should not be quantitated using this LC/MS method and an alternative method of chromatographic separation should be identified.
b. Export the ratio of peak areas for the sample versus the 13  Metabolites often respond non-linearly at high concentrations. If the standard curve has points that are much higher than the concentrations present in experimental samples, the highest points on the standard curve can be removed. Just ensure that the relative peak areas for all samples fall within the linear range of the standard curve.
c. Non-linear metabolites should be excluded from quantitative analysis, as this lack of linearity will prevent accurate quantitation by isotope dilution.
C. Determine the concentrations of the internal standards that were added to all samples.
Solve for the concentration of 13 C internal standard present in each external standard sample using the following relationship: actual concentration C 12 metabolite actual concentration C metabolite 13 = relative peak area C metabolite 12 relative peak area C 13 metabolite Note: This relationship can be used to calculate the concentration of the 13  b. Any data that shows a negative concentration should be removed. 20 www.bio-protocol.org/e3427  combined metabolites using a Mixer Mill MM301 with five 5 mm diameter stainless steel grinding balls. Perform 6 cycles of 1 min mixing at 25 Hz followed by 3 min resting. Store the now mixed chemical standard library powder stocks at -20 °C prior to use. For use, resuspend each mixed chemical library in HPLC grade water at 5 mM concentration as indicated for each library below.
Custom chemical standard libraries can be produced by acquiring desired pure chemical standards and mixing the pure chemical standards in equimolar amounts. When generating libraries, it is important to ensure that each library will not contain metabolites that have the same exact mass, as it is not then possible to determine the correct retention time for both metabolites when compounded into the same library. Consider putting these metabolites into separate pooled libraries (Tables 5-11).