Band-selective NMR experiments for suppression of unwanted signals in complex mixtures

Band-selective NMR experiments are presented that allow selective suppression of unwanted signals (SUN) from the spectra of complex metabolite mixtures. As a result, spectral overlap and dynamic range problems are substantially reduced and low-intensity signals normally covered by dominant signals can be observed. The usefulness of the experiments is exemplified with selective suppression of sugar signals from the NMR spectra of fruit juice and a plant sample. Other possible applications include blood, milk, and wine samples.

Electronic Supplementary Material (ESI) for RSC Advances. This journal is © The Royal Society of Chemistry 2020 S2

Pulse sequences
The pulse sequences for the 1D SUN experiments are shown in Fig. S1a and S1b. Narrow black rectangles denote 90° hard pulses while wide black rectangles denote 180° hard pulses. Shaped white bars represent band-selective inversion or excitation 180° pulses. IBURP-2 shapes were used for both suppression and excitation experiments. Typically, a 3.5 ms IBURP pulse centred in the middle of the glucose region yields a bandwidth of 1500 Hz, which is enough for suppression of glucose. The Bruker WaveMaker tool was used for construction of shaped pulses targeting more than one region of the spectrum. TOCSY transfer is achieved by using the DIPSI-2 mixing scheme with a mixing time of 10-150 ms depending on the spin system (here, 100 ms was used for the rice root sample whereas 50 ms was used for the orange juice and the artificial mixtures). White trapezoids with arrows denote lowpower 180° chirp pulses of 20 kHz bandwidth aimed for suppression of zero-quantum coherences. Their durations were set to 20 and 15 ms before and after the DIPSI-2 mixing, respectively. All gradient pulses except for G0 have a duration of 1 ms. G1 and G2 (amplitude of 14.9 and 5.3 G cm -1 , respectively) are used to eliminate magnetization of signals within the bandwidth of the selective pulse (bandselective suppression, Fig S1a) or outside of the bandwidth of the selective pulse (band-selective excitation, Fig. S1b). G3 is a spoil gradient pulse with an amplitude of 3.4 G cm -1 . G0 (amplitude of ca. 2.4 and 3.2 G cm -1 before and after the DIPSI-2 mixing, respectively) is a weak pulsed field gradient applied simultaneously with the chirp pulses to suppress zero quantum coherence. All gradient pulses are followed by a recovery delay of 200 μs. The phase cycle is given in Table S1.
To achieve optimal suppression of unwanted signals, it might be necessary to adjust the 90° pulse length and the precise phases of the band-selective pulses (by phase corrections). The TOCSY mixing time might also need some optimization in order to obtain the best recovery of remaining signals.
The 2D TOCSY and HSQC versions of the SUN experiments are shown in Fig. S1c and S1d. The 2D SUN-TOCSY experiment has two TOCSY transfers, where the first is applied to restore signals in the sugar region of the spectrum and the second is applied to yield TOCSY correlations in the f1 dimension. The two TOCSY steps could be used with different mixing times, but for the examples herein, the same mixing time was used. The 2D TOCSY was run in the States-TPPI manner, with phases presented in Table S1.
The HSQC step of the 2D SUN-HSQC experiment is almost identical to the Bruker pulse program hsqcedetgpsisp.2, which is a multiplicity-edited HSQC with PEP sensitivity enhancement and adiabatic inversion and refocusing pulses. Chirp pulses of 20 kHz bandwidth and 500 us duration were utilized for 180° 13 C inversion and composite chirp pulses of 2 ms duration for 180° 13 C refocusing. Delays were set to τ = 1.7 ms, Δ = 3.45 ms, Δ′ = 862 μs, and δ = 1.2 ms. G4 and G5 amplitudes were set to 38.5 and 9.7 G cm -1 , respectively, with a duration of 1 ms, followed by a recovery delay of 200 μs. 13 C decoupling was obtained with the GARP-4 decoupling scheme. The experiment was run in the echoantiecho mode, with gradient selection obtained by the reversal of the G4 gradient pulse. The phase cycle is given in Table S1.  Table S1. Unless indicated otherwise, pulses are applied along the x-axis. Band-selective excitation versions of the SUN-TOCSY and SUN-HSQC versions are obtained by eliminating the φ2 and φ4 180° 1 H pulses. Table S1. Phase cycling of pulse sequences in Fig S1 (A-D).

Pulse Sequence (Bruker format)
Quick guide for starting-up 1. Copy the pulse program from this document to a text file, save it as "SUN1d" and put it in the folder named Bruker\Topspin(X.X)\exp\stan\nmr\lists\pp\user. 2. Run a 1 H NMR experiment and decide whether band-selective suppression or excitation will be used. 3. Define peak(s) in the middle of the desired region(s) for band-selective suppression or excitation. 4. Make a copy of the 1 H NMR experiment and change the pulse program to "SUN1d". 5. Set probe/solvent dependent parameters with the command "getprosol" and fill in acquisition parameters that are not automatically adjusted, such as gradient files and gradient strengths. The correct settings are provided in the pulse program. 6. Copy the peak list from the 1 H NMR experiment to the SUN experiment. This can be done with AU programs such as copyPL or getPL (can be found at the Bruker web library). 7. The default setting is band-selective suppression without water presaturation. To use bandselective excitation instead, add "-DEXCIT" in the ZGOPTNS field. For water presaturation, add "-DPRESAT" in the ZGOPTNS field. If both excitation and water presaturation are desired, add both ("-DEXCIT -DPRESAT") in the ZGOPTNS field. 8. Define the shape function as userA1, for instance iburp2. (For Topspin versions older than 3.6, the shape function must be defined in the pulse program. In this case, change "userA1" to the desired shape function on the line starting with "sp2:wvm".) 9. Define the bandwidth (in ppm) of the band-selective suppression or excitation region(s) as cnst18. The bandwidth is equal to the width of the selected region(s) and must be the same if more than one region is selected with the peak list. 10. Define the TOCSY mixing time (d9). Typically 50-100 ms gives efficient TOCSY transfer. 11. Run "wvm -a" to calculate the selective pulse and to add the result into the current experiment set-up. 12. Check the receiver gain with rga and start the experiment with zg.
13. If signals that should be suppressed still appear, the bandwidth for band-selective suppression (cnst18) may need to be increased to allow defocusing of the entire spin system, or the bandwidth for band-selective excitation (cnst18) may need to be decreased to avoid excitation of the unwanted spin system(s). In band-selective excitation, the region for excitation can also be moved away from the unwanted signals by adjusting the peak list. Adjusting p1 can also improve the suppression of unwanted signals. 14. To optimize the phase of signals in the suppressed region, if necessary, phase corrections of ph2 and ph4 may be added with the "phcor" command. Try out small changes from zero to obtain optimized phase and suppression. Also p1 can be changed to optimize the phase of the spectra. In addition, gpz0 can be adjusted to optimize the suppression of zero-quantum coherences.