*4.1. NMR Sample Preparation*

Reference 1 and Reference 2 were prepared as previously described [15]. Human plasma extraction: metabolites were extracted from 1 mL of human plasma via a methanol and chloroform liquid–liquid extraction. The aqueous phase was transferred to a 15 mL Falcon tube and freeze-dried. The powder was reconstituted in 180 μL of 50 mM phosphate buffer at pH 7.4 in D2O, and then immediately transferred to a 3 mm NMR tube for NMR data collection. The NMR standard, DSS-d6, was added to each sample for chemical shift referencing.

## *4.2. NMR Experiments and Data Processing*

All NMR spectra were acquired on a Bruker AVANCE III solution-state NMR spectrometer equipped with a liquid helium-cooled TCI (H/F, C, N), deuterium lock, and a cryoprobe operating at a frequency of 599.773010 MHz for proton and 150.822998 MHz for carbon. NUS schedules were generated using a Poisson gap distribution with a sinusoidal weight of two and random seed generator [51]. The same 25% NUS schedule and seed were used for all experiments. All NMR data were collected at 298 K.

The spectral widths along the direct and the indirect dimensions were set at 9578.544 and 24,132.982 Hz, respectively. The number of complex points in the direct dimension was set at 512, and in the indirect dimension set at 32 with a 25% NUS sampling schedule. The number of scans for the 1D 1H experiments was set to 64 (d1 = 1.5 s), 72 (d1 = 1.2 s), 84 (d1 = 0.8 s), and 92 (d1 = 0.6 s). The number of scans for the 2D 1H-13C HSQC experiments was set to 36 (d1 = 1.5 s), 48 (d1 = 1.2 s), 72 (d1 = 0.8 s) and 84 (d1 = 0.6 s), respectively. The scan numbers were selected such that the total acquisition time for each 1D- and 2Dexperiment was on average 246 s, and 69 min., respectively. The transmitter frequency offset was set to 75 ppm in the 13C dimension and 4.7 ppm in the 1H dimension.

The spectral data were processed using the NMRPipe software package, as previously described [52]. The NUS data were reconstructed using iterative soft thresholding according to the hmsIST algorithm [51] to generate the same number of direct dimension data points and twice the number of indirect dimension data points, 512 (N2) × 256 (N1). Both the NUS and US NMR data were zero-filled, Fourier-transformed and manually phasecorrected to yield a final digital resolution of 2048 (N2) × 2048 (N1) points. Chemical shift queries, metabolite identifications and quantifications were performed using the COLMARm NMR webserver (http://spin.ccic.ohio-state.edu/index.php/colmar (accessed on 4 February 2021) [35]. The metabolite list is presented in the Supplementary Materials section for Reference 1 (Table S1) and Reference 2 (Table S2). The resonance assignments were used as previously reported [15].

#### **5. Conclusions**

In this study, we demonstrated the ability of Gd to improve the sensitivity of 2D NUS NMR spectra for the analysis of metabolomics samples. The addition of Gd led to an overall improvement in S/N and mean signal intensity for metabolites in both a model mixture and plasma samples. In the model mixture, the addition of Gd led to a 1.25-fold improvement in NMR signal intensities, which resulted in 1.7- and 1.6-fold improvements in LOD and LOQ, respectively. Interestingly, a significant improvement (>2-fold increase) was observed for metabolites with the lowest peak intensities, which suggests that the combination of Gd with NUS may improve the coverage of the plasma metabolome. The addition of Gd also maintained the highly linear intensity measurements that were correlated with a wide range of metabolite concentrations (50 μM to 2 mM). The reproducibility of intensity measurements, as noted by a decrease in %CV for both the model mixture (8% to 7%) and the plasma samples (15% to 10%), was similarly improved with the addition of Gd. Collectively, our results suggest that supplementing metabolomics samples with 0.25 mM Gd can improve the sensitivity of 2D NUS 1H-13C HSQC spectra and enhance the overall quality of the resulting data analysis. The routine adoption of PRE by the metabolomics community may expand the utility of multidimensional NMR to empower future biomarker discoveries.

**Supplementary Materials:** The following are available online. Figure S1. Chemical structure of Gadobutrol (Gd), Figure S2. Effect of addition of relaxation agents on 1H-13C HSQC 1D 1H spectral intensity for a model mixture (Reference 1) of metabolites in time equivalent experiments, Figure S3. Plots showing (a) mean peak intensity and (b) mean S/N for Reference 1 with and without addition of Gadobutrol, Figure S4. Linear regression curve of Reference 2 metabolites, Table S1. The list of

metabolites in Reference 1 model mixture, Table S2. The list of metabolites in Reference 2 model mixture, Table S3. Comparison of LOD and LOQ at 0.8 D1 and at 1.5 D1 with and without Gd.

**Author Contributions:** Conceptualization, R.P. and E.M.O.; investigation, C.H., N.T. and B.Z.; data curation, C.H.; writing—original draft preparation, C.H. and E.M.O.; writing—review and editing, R.P. and E.M.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Olaris, Inc., and, in part, by funding from the National Science Foundation under Grant Number (1660921) to R.P. and the National Institutes of Health Grant Number (P20 GM113126, NIGMS, Nebraska Center for Integrated Biomolecular Communication) to R.P. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

**Institutional Review Board Statement:** Not Applicable.

**Informed Consent Statement:** Not Applicable.

**Data Availability Statement:** Data available on request due to proprietary restrictions. The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank Leonardo O. Rodrigues, Chen Dong, Srihari Raghavendra Rao, Tan Chui-Mae, Josephine J. Wolf, Kanchan Sonkar, Monty Mahantesh Kothiwale, Charles Sang, Gerhard Wagner, Amber Balazs and David Emmons, for their many valuable discussions that motivated the writing of this communication.

**Conflicts of Interest:** The authors report that this work (design of the study, collection, analysis, and interpretation of data) was supported by Olaris, Inc. Chandrashekhar Honrao, Nathalie Teissier and Elizabeth O'Day are employees at Olaris, Inc. Bo Zhang is a former employee. Robert Powers is a member of the Scientific Advisory Board at Olaris, Inc.

**Sample Availability:** Samples are or not available from the authors.

#### **References**

