Resolving Sphingolipid Isomers Using Cryogenic Infrared Spectroscopy

Abstract 1‐Deoxysphingolipids are a recently described class of sphingolipids that have been shown to be associated with several disease states including diabetic and hereditary neuropathy. The identification and characterization of 1‐deoxysphingolipids and their metabolites is therefore highly important. However, exact structure determination requires a combination of sophisticated analytical techniques due to the presence of various isomers, such as ketone/alkenol isomers, carbon–carbon double‐bond (C=C) isomers and hydroxylation regioisomers. Here we demonstrate that cryogenic gas‐phase infrared (IR) spectroscopy of ionized 1‐deoxysphingolipids enables the identification and differentiation of isomers by their unique spectroscopic fingerprints. In particular, C=C bond positions and stereochemical configurations can be distinguished by specific interactions between the charged amine and the double bond. The results demonstrate the power of gas‐phase IR spectroscopy to overcome the challenge of isomer resolution in conventional mass spectrometry and pave the way for deeper analysis of the lipidome.


IMS Data: CCS
DT-IMS-MS measurements were performed on a modified Synapt G2-S HDMS instrument (Waters Corporation, Manchester, UK) containing a drift tube instead of the commercial travelling wave cell. [1] Ions were generated by nano electrospray ionization from a 10 µM solution of the lipid in methanol. Drift times were converted into collision cross sections (CCS) using the Mason-Schamp equation. [2]

IR Spectra and Optimized Structures
IR spectra were measured using an experimental setup described previously. [3] Ions were generated by nano electrospray ionization from a 100 µM solution of the lipid in methanol (Sigma-Aldrich).
After m/z-selection in a quadrupole mass filter, the ions are deflected into a hexapole ion trap filled with helium buffer gas. The walls of the ion trap were cooled with liquid nitrogen (~80 K) to allow for buffer gas cooling. After pumping out the buffer gas, the trapped ions are picked up from the trap by a pulsed beam of superfluid helium droplets, which are generated by expanding pressurized helium into the vacuum via a cold nozzle. The doped droplets leave the trap and coincide with the pulsed IR beam generated by the Fritz Haber Institute free-electron laser (FHI FEL). [4] In the case of resonant absorption of multiple photons, the ion is released from the droplet and detected on a time-of-flight detector. Spectra were recorded by scanning the spectral range of interest in steps of 2 cm -1 . The laser was more strongly focused in the regions from 900 to 1150 cm -1 and 1550-1800 cm -1 to increase the fluence in these wavenumber ranges.
The conformational space of the sphingolipids was sampled using the genetic algorithm FAFOOM. [5] Each generated structure was optimized at PBE+vdW TS /light level of theory in FHI-aims. [6] The conformational search included rotation of all rotatable bonds. For each sample, a subset of the energetically most stable structures were re-optimized at PBE0-D3/6-311+G(d,p) level of theory followed by a frequency calculation in Gaussian 16. [7] The following figures visualize the match between experimental and computed IR spectra. The displayed theoretical spectra were obtained from specific low-energy conformers (conf) numbered according to the labels in Tables S3 to S18. They were selected based on the match of the absorption frequencies in the region between 1400-1650 cm -1 with the experimental spectrum (example shown in Figure S8). ΔF is the harmonic free energy relative to the lowest-energy conformer (see Tables S3-S18). It is a useful value in this experiment, which is conducted at constant temperature and volume. The selected conformer is not always the lowest-energy conformer according to ΔF with the employed level of theory.    Figure S5: IR spectra of 1-deoxysphingolipids and their 1-deoxymethylsphingolipid analogues. a) The absorption bands of 1-deoxySA and 1-deoxymethylSA overlap largely. b) The main absorption band of 14Z-1-deoxySO is slightly shifted compared to 13Z-1-deoxymethylSO. Figure S6: IR spectra of the OH-regioisomers 4E-1-deoxySO and 4E-3-deoxySO. The frequencies of N-H bending vibrations coincide, whereas differences occur in the low wavenumber range.

Calculated Spectra in the 3 Micron Range
The following calculated spectra illustrate the expected N-H stretching vibrations of the deoxysphingolipids in the 3 micron range. Experimental data are not available for this wavenumber range. For each structure, only the spectrum of the selected conformer according to the best match in the experimentally accessible region is shown (numbering according to Tables S3-S18). Figure S7: Calculated IR spectra of the selected conformers of all investigated deoxysphingolipids in the region between 2500-3800 cm -1 . The expected N-H stretching vibrations of the NH3 + group are indicated by red markers in each spectrum. According to the calculated spectra, the band patterns derived from the NH3 + vibrations in the 3 micron range are very diagnostic and allow for isomer distinction. O-H Stretching vibrations are located at higher wavenumbers (3700-3800 cm -1 ). Table S3: List of distinct structures of 1-deoxySA re-optimized at PBE0-D3/6-311+G(d,p) level of theory in Gaussian 16. Each conformer has a distinct label (conf_XX or ref_XX) and the final energy ΔE (including zero-point energy) and harmonic free energy ΔF at 78 K assigned to it. The energetics are relative to the lowest-energy conformer. All energies are displayed in kJ mol -1 .   Table S5: List of distinct structures of 4E-1-deoxySO re-optimized at PBE0-D3/6-311+G(d,p) level of theory in Gaussian 16. The reference conformers are structures provided by the authors of ref. [8] optimized by the same procedure.  [8] 4.57099 3.65677 ref_02 [8] -0.28093 0.32832  [8] 0 0  [8] 6.67402 5.90611 (NH3 + ) (3.87) 1 ref_02 [8] 11.07173 11.33412 NH3 + 2.27  [8] 20.75458 20.77038   [8] 0 0