Efficient Extraction from Mice Feces for NMR Metabolomics Measurements with Special Emphasis on SCFAs

Nuclear magnetic resonance (NMR) spectroscopy is one of the most promising methods for use in metabolomics studies as it is able to perform non targeted measurement of metabolites in a quantitative and non-destructive way. Sample preparation of liquid samples like urine or blood serum is comparatively easy in NMR metabolomics, because mainly buffer and chemical shift reference substance are added. For solid samples like feces suitable extraction protocols need to be defined as initial step, where the exact protocol depends on sample type and features. Focusing on short chain fatty acids (SCFAs) in mice feces, we describe here a set of extraction protocols developed with the aim to suppress changes in metabolite composition within 24 h after extraction. Feces are obtained from mice fed on either standard rodent diet or high fat diet. The protocols presented in this manuscript are straightforward for application, and successfully minimize residual bacterial and enzymatic activities. Additionally, they are able to minimize the lipid background originating from the high fat diet.

. Comparison spectra of the extracted metabolites using ultra-sonication and/or shock freeze for extraction of mice feces on a standard diet. For clarity only the spectral region of interest, harbouring resonances of the short chain fatty acid (SCFA) is shown.

Combination of double extraction cycles
The experiments are performed in accordance to the protocol suggested by Wu et al. [1], where the extraction is repeated to the same feces material twice and the two extraction supernatants are combined and further lyophilised. In Figure S2 is shown the outcome from the first and second extraction and as well the NMR spectrum of the combined phase.   Figure 5 in main text.

Methanol extraction
In an attempt to simplify the multi solvent extraction a simple methanol extraction is tested, where cold methanol-d4 is added the mice feces, followed by vortexing, an ice bath extraction time and centrifugation. 700 µl of cold methanol-d4 are added to 20 mg mice feces. After a 30 s mixing with the vortex, the sample is placed in an ice bath for 10 minutes and then centrifuged at 4°C and 15000 x g for 10 minutes. The supernatant is taken and transferred to a 5 mm NMR tube. 2 mM TSP is added as a reference signal. This simple protocol has allowed us to stop bacteria and the enzymatic activity, however, the extraction is quite inefficient in the case of mice feces on high fat diet as shown in Figure S4. In the case of HFD mice samples the amounts of extracted SCFA are at the level of noise and the comparison with the spectrum of mice on standard diet requires at least eight times downscaling ( Figure S4). Yet, a high level of lipid background is extracted in both types of samples. In addition, the NMR spectrometer has problems with correctly locking on the desired solvent peak. Figure S4. Comparison of the methanol extraction applied to feces from mice on standard diet, where the signal intensities in the spectrum are scaled down eight times, and from mice on high fat diet.

Lipid extraction protocols
A mice feces lipid extraction is performed according by Kraus et al. using 600 µl 0.15 M sodium chloride solution is mixed with the mixture of chloroform-d and methanol-d4 (2:1) at RT [3]. This protocol was applied as described by Kraus et al. and since the mixed solvents prevented the sample changes over time, it serves as a basis and was further optimized. Here, we describe below in details all the optimisation steps. We have tested the lipid extraction by using either phosphate buffer or sodium chloride solution, performing the extraction steps at various temperatures and the centrifugation conditions. The exact procedure is the following: to the mice feces are added 600 µl 0.15 M sodium chloride (or phosphate buffer) and 600 µl of a 2:1 volume mixture of CDCl3 and methanol-d4. Optimal extraction is tested on identical sample sets kept at either -20 °C, 4 °C or +20 °C. The samples are vortexed for 30 s. As we are targeting at the SCFA, we have tested the temperature for centrifugation to -20 °C, 4 °C or, +20 °C for one sample per set. Thus we have had tested all variations for cold or RT extraction followed by either cold or RT centrifugation. In order to achieve better and clearer phase separation an increased centrifugation speed is evaluated. It is not found beneficial to increase the centrifugation to 11000 x g. The comparison of the optimal temperature for the extraction showed that 4 °C is most optimal as maximal amount of extracted SCFA versus minimal background signals could be achieved (see Figure 7, main text). The optimal result of the above optimisation and what we called Protocol 1 for lipid extraction was achieved with extraction at 4 °C and centrifugation at the same temperature.
As the amounts of background signals are still high, in addition we have performed further two washing steps with chloroform to remove the lipid content from the aqueous phase. After each washing step with the addition of 200 µl CDCl3, the samples are centrifuged at 4°C and 1100 x g for 2 minutes. At the end additional 200 µl CDCl3 are added and the samples are centrifuged at 4°C and 1100 x g for 10 minutes. Together with the addition of a crystal NaN3 500 µl of the aqueous phase are transferred to a NMR tube for measurements. This procedure we refer to as Protocol 1 with washing steps. As a comparison we have performed equivalent phosphate buffer extraction instead of 0.15 M sodium chloride in a second iteration for the protocol optimization and we have identified the sodium chloride version as better suited for extraction of SCFA from mice feces as shown in Figure 6 in main text.
In final optimisation iteration we weight the 20 mg of feces into Precellys tubes with ceramic beads. The 600 μl sodium chloride (0.15 M) and 600 μl mixture with CDCl3 and MeOD-d4 (2:1, CDCl3:MeOD-d4) are added into the tube. The mixture is vortexed for 30 s. As a next step the sample is homogenised four times for 20 s at 10°C with 6000 rpm with a waiting time of 120 s in the homogenizer. The mixed solution is transferred in a micro centrifuge tube and centrifuged at 0°C and 1485 x g for 10 minutes. 500 μl of the aqueous upper phase are taken and filled in 5 mm NMR tube for measurement with the addition of a crystal NaN3. This procedure we refer to as Protocol 2. Figure S5. Comparison of the metabolites in the aqueous and chloroform phases originating from optimised protocol for lipid extraction, referred to as Protocol 1. Figure S6. Comparison of Protocol 1 and 2 with the basic protocol with shock freeze: (A) feces of mice fed on a standard diet and (B) on high fat diet (HFD) (the intensity is upscaled four times). For clarity only a selected spectral range (0.6 ppm till 2.0 ppm) of the proton NMR spectra is shown and the most prominent metabolites are assigned. Protocol 1 and 2 clearly extract more efficiently the SCFA, based on their increased intensity for both diets. Figure S7. 1H NMR spectrum of the extracted metabolites from mice on a standard diet with phosphate buffer protocol with shock freeze Figure S8. 1H NMR spectrum of the extracted metabolites from mice on a HFD with phosphate buffer protocol with shock freeze Figure S9. 1H NMR spectrum of the extracted metabolites from mice on a HFD with ultrafiltration (3kDa cut off filter) Figure S10. 1H NMR spectrum of the extracted metabolites from mice on a standard diet with Protocol 1 without washing steps.