Formation of Sodium Cluster Ions Complicates LC-MS Metabolomics Analyses

Metabolomics represents the large-scale study of small molecules in complex biological samples and offers direct insights into metabolism.^1,2 The comprehensive and unbiased measurement of metabolites by untargeted strategies can provide a systematic global view of changes in phenotype under different physiological conditions, from genotype influence to metabolic, disease-related, hormonal and/or environmental-induced dysfunctional states.³ Untargeted metabolomics is a hypothesis-generating approach that enables the discovery of unknown metabolites with key biologically roles, e.g., in metabolism, signaling, regulation, etc. Untargeted metabolomics has found widespread application in many areas, including the study of disease pathogenesis, biomarker identification, diagnostics, and drug discovery.^4,5

Liquid chromatography coupled with mass spectrometry (LC-MS) is a widely used platform for high-throughput identification and quantification of metabolites in biological samples.^6,7 Thousands of peaks can be accurately detected in biological extracts using high-resolution LC-MS instruments.^6,8 Nevertheless, global studies of metabolites for diagnostics and pathological insights have been limited by the time and expertise required for data interpretation. The large datasets generated by untargeted mass spectrometry metabolomics frequently contain experimental artefacts and chemical interferents (i.e. solvent contaminants, plasticizers, etc.), not necessarily consistent across samples, and often difficult to deconvolute, potentially leading to misidentification.^9,10

During the development of untargeted metabolomics methods for the study of polar components from human biofluids, using hydrophilic interaction chromatography (HILIC) in two LC-QTOF MS systems, we have observed an intense LC peak across samples from various biological origin, which included human urine, plasma, serum and cell extract, as shown in Figures 1 and 2. This intense peak eluted at ~7.3 min and exhibited an intensity of 10⁵–10⁷ counts per second (cps). The same peak was not detected during injections of instrument blanks (consisting of pure MilliQ water, acetonitrile, methanol), nor sample preparation blanks (pure MilliQ water, acetonitrile or methanol processed through the sample extraction protocol). We obtained similar results, i.e. no LC peak detected, with the injection of mobile phase (MP) A and B (70:30, v/v).

Figure 1.

Base peak chromatograms obtained by HILIC-ESI-QTOF MS in positive ion mode, showing formation of sodium acetate cluster peak at 7.3 min. (A) human urine (B) human serum (C) human blood plasma (D) MCF7 cell extract (E) MilliQ water with the addition of NaCl, (F) Milli-Q water (18 MΩ).

Figure 2.

Base peak chromatograms (BPC) obtained in negative ion mode, showing the detection of a sodium acetate cluster peak. (A-F) Injections of the same samples described in Figure 1.

The HILIC-ESI-MS/MS analyses were performed with an Agilent 1200 SL LC system coupled to an Agilent 6520 Q-TOF mass spectrometer (Agilent Technologies, Santa Clara, CA) using an Agilent Jet Stream Technology electrospray ionization source, and a Sciex ExionLC AC-X500 Q-TOF system (AB Sciex, Framingham, MA) equipped with a TurbolonSpray ion source. Chromatographic separation was initially carried out with a Waters Xbridge BEH Amide column (serial n° 01313021116974) and later with a Waters Xbridge BEH HILIC (serial n° 01053429315404) stationary phase, both with 2.5 μm particle size and 2.1 mm × 150 mm dimensions. MP A was composed of 10 mM ammonium acetate in 95% water, 3% acetonitrile, 2% methanol, and 0.2% acetic acid, while MP B was prepared with 10 mM ammonium acetate in 93% acetonitrile, 5% water, 2% methanol, and 0.2% acetic acid. The gradient elution mode was as follows: 0–6.5 min, 94–78% B; 6.5–12.0 min, 78–39% B; 12.0–18.5 min, 39% B; 18.5–19.0 min, 39–94% B; 19.0–35.0 min, 94% B. The flow rate was 0.3 mL min⁻¹, the column temperature was 35 °C, and the injection volume was 5 μL. The acquisition parameters were: capillary voltage ±3.8 kV; cone voltage ±40 V; source temperature 120 °C; desolvation temperature 325 °C; cone gas flow 20 L h^–1; nebulizer pressure 45 psi; desolvation gas flow 600 L h^–1. Nitrogen was used as the drying gas.

The full-scan MS1 survey of peak at ~7.3 min showed the main ions m/z 925, 843, 761, 679, 597, 515, 433, 351, 269, 187 in (+)-ESI, and m/z 961, 879, 797, 715, 633, 551, 469, 387, 305, 223, 141 in (-)-ESI, as illustrated in Figure 3. According to the literature, this fragmentation pattern was identified as sodium adducts of a fluorinated compound, polytetrafluoroethylene, caused by grease used on 1100 Agilent LC pump modules.¹¹ Indeed, the neutral losses of 82 Da between consecutive in-source fragments suggest the sequential neutral elimination of trifluoroethylene repeat units (C₂HF₃; 82.0030 Da), with a mass error < 5 ppm, as depicted on Scheme 1A. However, the contaminant was not observed in blanks, indicating that its formation relies on components present in samples of biological origin.

Figure 3.

ESI-MS of sodium ion cluster in (A) positive and (B) negative ESI modes.

Scheme 1.

Mechanisms of dissociation suggested for the neutral losses of 82 Da. (A) sequential neutral elimination of trifluoroethylene (TrFE, C₂HF₃; 82.0030 Da) from polytetrafluoroethylene (PTFE), and (B) consecutive elimination of CH₃COONa (82.0031 Da) units from the sodium acetate cluster ion.

Biofluids and other complex samples matrices generally have high levels of salt, which affect the ESI response through either ionization suppression or enhancement.¹² We spiked blanks with NaCl solution and injected them into the Q-TOF MS system. The addition of NaCl promoted the detection of a peak with similar intensities to those observed in our samples (10⁵–10⁷ cps), as illustrated in Figure 1E in (+)-ESI and Figure 2E in (-)-ESI mode. We also observed the contaminant peak in samples containing a mixture of mobile phases A and B (7:3, v/v) after spiking with NaCl solution. The direct infusion of each component of the mobile phase in the Q-TOF added to NaCl revealed that Na⁺ induced the appearance of cluster ions from sodium acetate buffer, with the general structure Na⁺(CH₃COO⁻Na⁺)_n.

Adduct formation is a well-known phenomenon of ion formation in ESI/MS.¹³ Small ions such as alkali metal cations (e.g Na⁺, K⁺), anions of strong acids (e.g., phosphate, trifluoroacetate) or salt ions (ammonium acetate, ammonium formate) coordinate with the analyte during the evaporation/fission of the charged droplets in the ionization source, in a process that is still poorly understood.¹⁴ Coordination reactions have often been used to enhance the detection of compounds with low ionization behavior.¹⁵ The strategy has also been applied to investigate fragmentation pathways and to support structural elucidation.¹⁶ The chemical interactions generated in the ESI source prompt changes in the geometry of ion-ligands and can produce more intense and/or unique fragment ions.¹⁶ On the other hand, adduct formation can result in multiple ionization processes and unpredictable fragmentation, as well as cluster ion formation.¹⁷

Cluster ions, formed by at least two chemical species bonded by noncovalent interactions and in coordination with ionic species such as Na⁺ or Cl⁻,¹⁸ can be generated in the ESI source by incomplete electrophoretic separation of an ionic salt, possibly due to excess ions at the droplet surface. During the process of droplet fission, unseparated chemical species would be retained in the droplet together with the excess ions, forming unseparated ion pairs that constitute the cluster.¹⁴ In this case, the sequential neutral losses of 82 Da between consecutive in-source ions represent CH₃COONa (82.0031 Da) units from the ion pair, as illustrated in Scheme 1B.

Interestingly, in HILIC experiments, small inorganic ions are retained on the column and eluted in high amounts during the chromatographic run, favoring the ion cluster formation as a LC-peak. The salt ion cluster peak can reduce the detection of co-eluted analytes, promote the formation of adducts and lead to convoluted MS spectra.¹⁷

We observed the intense LC peak related to sodium cluster ions when we used both BEH Amide and HILIC columns under similar experimental conditions. The peak only diverged in the retention time (6.0 min for the HILIC columns). We also investigated the impact of substituting ammonium acetate/acetic acid for the ammonium formate/formic acid buffer in the mobile phases. We observed the formation of an intense LC peak at ~7.0 min (peak intensity of 10⁵–10⁷ cps), after addition of NaCl solution, when using BEH Amide column, and the characteristic fragmentation pattern of repeating similar masses, in this case, the sequential loss of 68 Da, as illustrated in Figure 4A–B. The result suggested that ammonium formate/formic acid addition can also lead to sodiated cluster ions, with the formula Na⁺(HCOO⁻Na⁺)_n, and a similar dissociative mechanism, with loss of HCOONa units (67.9874 Da), as shown in Figure 4C.

Figure 4.

(A) Base peak chromatogram (BPC) showing the sodium formate cluster peak at 7.0 min. BPC obtained by HILIC-ESI-QTOF MS in positive ion mode. (B) ESI-MS of sodium formate ion cluster in positive ESI mode. (C) Proposed mechanism of neutral losses of 68 Da, i.e., sequential elimination of HCOONa (67.9874 Da) units.

These findings are consistent with recent literature,^17,19 and illustrate how difficult it remains to investigate the full chemical composition, i.e. the set of small organic molecules, of complex biological matrices. The variation derived from alkali metal complexation as well as cluster ion formation dramatically impacts metabolite fragmentation, charge states, and adduct distributions and ratios. These complications can lead to significant challenges for metabolite identification and for hypothesis-generating approaches in untargeted metabolomics using ESI-MS. For this reason, mass features detected by MS-based untargeted metabolomics do not necessarily provide a comprehensive description of the chemical constitution of samples, but rather a combination of metabolite specific responses along with an instrument-specific snapshot of the response of molecular compounds to the mass spectrometer.²⁰ One step towards improving untargeted MS data quality and robust scientific outcomes is to remove contaminants and interferents, such as salt cluster combinations.²¹ An accurate characterization of cluster ions can reduce the complexity of inflated mass feature lists, and help the identification of a higher percentage of the myriad adducts, fragments, and charge states retrieved by metabolites and other species in co-elution.