Degradation of the Vaccine Additive Thimerosal by L-Glutathione and L-Cysteine at Physiological pH

Abstract

Humans are being exposed to a variety of potentially toxic metal compounds through the diet and/or the intravenous administration of metal-containing medicinal drugs. The organomercurial thimerosal (THI) is a bactericidal that is present in vaccines, but its potential degradation by biomolecules in vivo is incompletely understood. To probe its interaction with low-molecular-weight thiols that are highly abundant within cells, we have employed an LC-based analytical approach in conjunction with a mercury-specific detector. The injection of THI into a C₁₈-HPLC column equilibrated with mobile phases that contained increasing concentrations of up to 15 mM of glutathione (GSH) and 30% acetonitrile revealed the elution of a GS-EtHg adduct in conjunction with THI, as evidenced by electrospray ionization mass spectrometry. These results were confirmed by ¹⁹⁹Hg-NMR spectroscopy. While these results imply a rapid degradation of THI by GSH at physiological pH, it is important to point out that our results were obtained in aqueous solutions containing 30% (v:v) acetonitrile. Further studies need to confirm if the GS-EtHg adduct is also formed in biological fluids. Our results nevertheless demonstrate that GSH and L-cysteine (Cys) are potential targets of THI at physiological pH, which is relevant to better understand its side effects, including previously reported effects on Ca²⁺ channels.

1. Introduction

The exposure of mammals to metal (loid) species from the environment [1,2,3] as well as medicinal drugs that contain metals [4], such as the organomercury compound thimerosal (THI) (Figure 1A) is being increasingly recognized as a major public health concern due to the associated potential short and long-term toxic effects [4]. Since potentially toxic metal species will invade the bloodstream, bioinorganic reactions therein play a fundamental role in the context of better understanding their toxicological chemistry. The interaction of toxic metal species with biomolecules in the bloodstream and/or organs may involve their binding to proteins as well as their interaction with cytosolic biomolecules [4]. As these latter interactions can contribute to their long-term adverse health effects at the organ level, they need to be better understood in order to evaluate their involvement in the etiology of adverse human health effects. THI has been shown to remain in the bloodstream for days after its administration to mice [5]. Since the red blood cell (RBC) membrane can be easily penetrated by a variety of toxic metal species [6] and since RBC cytosol contains ~2.5 mM of glutathione (GSH, Figure 1C), the interaction of THI with GSH therein is among the first interactions that are of potential human health relevance [7].

THI has been widely used as a preservative in many biological and pharmaceutical products, including vaccines, since the 1930s [8,9]. THI is a potent bactericidal, has antifungal activity [10], and effectively inhibits the SARS-CoV2 main protease [11]. While THI remains one of the most widely used preservatives in multidose vaccines, particularly in low-resource countries, this organomercury compound has been the subject of controversy because of concerns about a possible link between human exposure to thimerosal in vaccines and autism [12]. From a bioinorganic chemistry point of view, rather few studies have investigated the interaction of THI with thiols. While some studies have demonstrated the binging of a THI-degradation product to hemoglobin using mass spectrometry [13] and metallomics approaches [7], no studies have investigated its reaction with GSH at physiological pH. The related organomercury compound phenylmercuric acetate has been recently shown to be rapidly degraded by L-cysteine at physiological pH (Cys, Figure 1B) using a LC-based approach. We therefore decided to investigate the fate of THI at pH 7.4 in the presence of physiologically relevant GSH-concentrations as it is the most abundant small molecular weight thiol in mammalian cells at concentrations up to 19 mM [14]. The employed LC-based approach allows the specific detection of Hg-species and is therefore ideally suited to observe the degradation of THI provided that the degradation products can be chromatographically separated from the parent compound [15] and structurally characterized using analytical techniques, such as electrospray ionization mass spectrometry (ESI-MS) and ¹⁹⁹Hg NMR spectroscopy [16]. Our investigations are solely aimed at understanding the biochemical reaction between THI and GSH at physiological pH to gain insight into the corresponding bioinorganic chemistry. As such the anticipated results are insufficient to causally link human exposure to THI with the etiology of autism in children [17], but the results may nevertheless provide an important first step to uncover bioinorganic processes that may also unfold in mammalian cells and explain adverse effects therein. To the best of our knowledge our investigations appear to be the first of their kind to address the bioinorganic chemical reaction between THI and GSH at physiological pH in the presence of 30% of acetonitrile (ACN).

2. Results and Discussion

Since THI is widely used in a variety of pharmaceutical products numerous previous studies have investigated its stability in aqueous solutions, such as vaccines to determine its shelf life [18,19]. While the results from these studies provided insight into the integrity of THI under storage conditions, much less is known about the potential degradation of THI by biomolecules after its intramuscular injection into people. As soon as THI enters cells it will encounter small molecular weight thiols, such as Cys and GSH (Figure 1B,C) which are present at mM concentrations. Indeed, recent studies have allowed to observe the degradation of the organomercurial compound phenylmercuric acetate (PMA) by Cys at physiological pH [15]. To understand the degradation of another organomercury compound that is of pharmaceutical interest, we have employed a LC-approach which encompassed a Hg-specific detector, namely a flame atomic absorption spectrometer (FAAS) to study the degradation of THI. THI was injected into a C₁₈-HPLC column and eluted using mobile phases that contained different GSH concentrations at physiological pH.

The results obtained for the injection of THI using a GSH-free mobile phase revealed the elution of a single Hg peak (t_r = 137 s; Figure 2, red line), which exhibited a Hg recovery of 82.2%.

The addition of up to 15 mM GSH concentrations to the mobile phase resulted in a concentration dependent decrease in the intensity of the Hg-peak that eluted at 137 s and the elution of a new Hg-peak that was progressively more intense with a 35 s reduced retention time compared to THI (Figure 2, blue, green, brown and black lines). With regard to the Hg peak areas of the Hg-specific chromatograms, Hg recoveries between 87.5 and 114.6% were obtained for the GSH-containing mobile phases (

Table 1. Peak areas of the Hg peaks obtained after the analysis of THI with different GSH concentrations in the mobile phase.

Concentration of GSH (mM) in Mobile Phase	Retention Time (s)		Peak Area (Area Units) (Area Percentage [%])		Sum Area (Area Units) (Sum Area Percentage [%])
	Peak 1	THI	Peak 1	THI
0	-	137 ± 1 *	-	8.43 ± 1.07 (82.2 ± 10.7%)	8.43 ± 1.07 (82.2 ± 10.7%)
2.5	107 ± 2	137 ± 1	1.71 ± 0.07 (16.7 ± 0.8%)	8.11 ± 0.37 (79.0 ± 4.3%)	9.82 ± 0.38 (95.7 ± 4.4%)
5	111 ± 1	138 ± 1	2.79 ± 0.06 (27.2 ± 1.0%)	6.19 ± 0.20 (60.3 ± 2.6%)	8.98 ± 0.21 (87.5 ± 2.8%)
10	103 ± 1	135 ± 1	4.75 ± 0.06 (46.3 ± 1.5%)	5.73 ± 0.37 (55.8 ± 4.0%)	10.48 ± 0.37 (102.1 ± 4.3%)
15	105 ± 1	136 ± 1	6.23 ± 0.04 (60.7 ± 1.8%)	5.53 ± 0.02 (53.9 ± 1.6%)	11.98 ± 0.29 (114.6 ± 2.4%)

* n = 3.

). Using the 15 mM GSH mobile phase the two Hg-containing fractions were collected and ESI-MS analysis revealed that the first Hg-peak (t_r = 105 s) corresponds to a GS-ethylmercury (EtHg) adduct with a m/z ratio of 536.0787 (Figure 3), while the second Hg-peak (t_r = 137 s) produced a peak in the ESI-MS mass spectrum of 383.0041 m/z ratio which is congruent with the elution of ‘intact’ THI (Figure 4). These results imply that increasing physiologically relevant GSH concentrations in the mobile phase progressively degraded THI on the column. With the 15 mM GSH mobile phase, 54% of THI was eluted intact (Table 1), while the remaining Hg corresponded to the GS-EtHg adduct.

Since Cys is also present in various cell cytosols at comparatively lower concentrations than GSH (e.g., hepatocytes contain ~0.5 mM Cys [15]), we replicated the LC-experiments with Cys-containing mobile phases. While the employed Cys concentrations are supra physiological, the obtained results nevertheless resembled those obtained with equimolar concentrations of GSH in the mobile phase (Figure 5). With regard to the obtained Hg peak areas of the obtained Hg-specific chromatograms, Hg recoveries between 78.9 and 95.7% were obtained for the Cys-containing mobile phases and 85.1% for the Cys-free mobile phase (

Table 2. Peak areas of the Hg peaks obtained after the analysis of THI with different Cys concentrations in the mobile phase.

Concentration of Cys (mM) in Mobile Phase	Retention Time (s)		Peak Area (Area Units) (Area Percentage [%])		Sum Area (Area Units) (Sum Area Percentage [%])
	Peak 1	Peak 2	Peak 1	Peak 1
0	-	140 ± 1 *	-	8.73 ± 0.30 (85.1 ± 3.8%)	8.73 ± 0.30 (85.1 ± 3.8%)
2.5	115 ± 2	140 ± 1	0.51 ± 0.01 (4.9 ± 0.2%)	8.55 ± 0.36 (83.3 ± 4.3%)	9.06 ± 0.37 (88.3 ± 4.4%)
5	111 ± 1	139 ± 1	0.58 ± 0.06 (5.7 ± 0.6%)	8.44 ± 0.14 (82.2 ± 2.8%)	9.02 ± 0.20 (87.9 ± 3.2%)
10	121 ± 1	132 ± 1	3.84 ± 0.31 (37.4 ± 3.2%)	4.83 ± 0.49 (47.1 ± 4.9%)	8.67 ± 0.80 (84.5 ± 8.2%)
15	116 ± 1	133 ± 0	5.46 ± 0.39 (53.2 ± 4.1%)	4.36 ± 0.25 (42.5 ± 2.7%)	9.82 ± 0.64 (95.7 ± 6.8%)
20	117 ± 1	134 ± 1	5.95 ± 0.17 (57.9 ± 2.4%)	2.15 ± 0.19 (20.9 ± 1.9%)	8.10 ± 0.36 (78.9 ± 4.2%)

* n = 4.

). Similarly to the GSH LC-results we observed the parent THI Hg-peak at the same retention time (t_r = 137 s) as well as the elution of a progressively more intense Hg-peak (t_r = 115 s), which consistently eluted at a 10 s increased retention time compared to the corresponding GS-EtHg adduct (Figure 2). While a fraction containing the putative Cys-EtHg adduct was submitted to ESI-MS analysis, the results were not as clear as those obtained for the GS-EtHg adduct. Nevertheless, these Cys HPLC results suggest that the putative Cys-EtHg adduct that was formed on the column is slightly more hydrophobic than the GS-EtHg adduct. Compared to the 15 mM GSH mobile phase, 42.5% of Hg eluted in the form of intact THI, while 53.2% eluted as the putative Cys-EtHg adduct. These results imply a fast degradation of THI by Cys as compared to GSH at comparable mobile phase concentrations.

To substantiate these HPLC/ESI-MS results for THI we conducted ¹⁹⁹Hg-NMR experiments, the results of which are depicted in Figure 6. The ¹⁹⁹Hg-NMR spectrum of a THI solution in a phosphate buffer that contained 30% of ACN revealed a well-defined complex multiplet at −743 ppm (Figure 6A). When GSH was added to a THI solution to achieve the same mM concentration ratio as in the LC-experiments, the observed multiplet (Figure 6A) shifted slightly to −740 ppm and it collapsed to a broad singlet (Figure 6B). This observation is indicative of a more complex coupling environment and likely conformational isomerism while conclusively confirming a chemical reaction between the Hg-center of THI with GSH by ¹⁹⁹Hg-NMR spectroscopy.

The results that were obtained for the interaction of THI with GSH at physiological pH in the presence of 30% ACN are all congruent with the formation of a GS-EtHg adduct at these conditions (Figure 7). Although our results were not obtained in 100% aqueous solution, previous studies have conclusively demonstrated that THI—after its incubation with RBC cytosol at 37° Celsius—degraded over a 6 h period and progressively co-eluted with hemoglobin (Hb) [7]. The binding of the EtHg-breakdown product of THI to Hb was confirmed by analyzing a fraction containing the putative EtHg-Hb adduct by X-ray absorption spectroscopy, which revealed a Hg-S bond length that was congruent with literature results [7]. Based on these previous studies obtained with RBC cytosol it appears that THI degrades within cells and the formed degradation product EtHg will then competitively bind to biological thiols, such as GSH and/or Hb based on steric factors. Future studies should investigate whether the GS-EtHg adduct can bind to cytosolic proteins, such as Hb using established metallomics techniques [7]. Based on previous results by others another promising future investigation is to evaluate if the GS-EtHg adduct will bind to calcium channels as there appears to be a link between THI and the perturbation of the Ca-metabolism in certain cell types [20,21].

3. Experimental

3.1. Chemicals and Solutions

Thimerosal (97–101%), L-glutathione reduced (>98%), L-cysteine (>98%), sodium phosphate dibasic (Na₂HPO₄, >98.5%), sodium phosphate monobasic (NaH₂PO₄, >98%), sodium chloride (NaCl, >99.5%), acetonitrile (ACN, Chromasolv, HPLC grade, >99.9%) and deuterium oxide (D₂O, 99.9 atom % D) were purchased from Sigma Aldrich (St. Louis, MO, USA).

A 0.1 M solution of THI was made every day by dissolving 10.1 mg in 250 μL 0.8% NaCl, prepared by dissolving 0.8 g of NaCl in 100 mL deionized (dI) water from a Simplicity UV water purification system (Millipore, Billerica, MA, USA). Mobile phases were prepared by mixing 100 mM of Na₂HPO₄ and 100 mM of NaH₂PO₄ solutions to obtain pH 7.4 using a VWR Symphony SB20 pH meter (Thermo Electron Corporation, Beverly, MA, USA). Then ACN was added to achieve a 30% final concentration (v:v). Mobile phases that contained GSH (0–20 mM) were prepared by adding the corresponding amount of GSH to the ACN containing phosphate buffer to give 150 mL of GSH-containing buffer (0.6915 g of GSH to obtain the 15 mM GSH buffer). Mobile phases containing up to 20 mM Cys were prepared in an analogous manner. Then, the pH of the final solution was adjusted to 7.4 with 0.4 M HCl and all mobile phases were filtered through 0.45 μm MCE nitrocellulose membrane filters (Millipore, Mandel Scientific, Guelph, ON, Canada).

3.2. Instrumentation

The HPLC system comprised an Azura P2.1S pump equipped with a ceramic pump head, a Rheodyne 9725 injector (Rheodyne, LLC, 600 Park Court. Rohnert Park, CA, USA), a 50 μL sample loop and a Gemini 5 μm C₁₈ HPLC column (Phenomenex, 411 Madrid Avenue Torrance, CA, USA) 150 × 4.6 mm, 5 μm particle size). The flow rate was 1.0 mL min⁻¹ and 0.1 M solutions of THI were injected. All separations were conducted at room temperature (22 °C) and Hg was detected using a Buck Model 200A flame atomic absorption spectrometer (FAAS, Buck Scientific, East Norwalk, CT, USA) at 253.7 nm. The HPLC column exit was connected to the FAAS inlet with polyethylene tubing (I.D. 0.13 mm, length 8 cm) and its void volume (t₀ = 92 s) was determined by injecting a NaCl solution and observing the yellow color of the flame using dI water as the mobile phase. To determine the recovery of Hg in the HPLC experiments, we injected 1000 μg of Hg into a hollow plastic tube (0.15 mm × 21 cm) and defined the area that was obtained from the FAAS (10.26 ± 0.3 area units) as 100%. The total Hg peak areas determined by integrating all Hg peaks were then expressed as a percentage of this theoretical peak area of 100%.

The ESI-MS experiments to identify the Hg-species in the column effluent were conducted using previously established parameters [15]. To identify the Hg-degradation product that was formed on the column when THI was injected with the 15 mM GSH containing mobile phase, the corresponding Hg-containing fractions were collected and analyzed by LC-ESI MS with an Agilent 1200 series HPLC coupled to an Agilent 6520 Q-TOF mass spectrometer (Agilent, Santa Clara, CA, USA). Separation of the Hg complexes from the buffer matrix was achieved with an Agilent Eclipse Plus C₁₈ column (2.1 mm × 100 mm, 3.5 μm particle size) and a 20 mM NH₄HCO₃/methanol gradient (0–1 min 20% MeOH, 1–9 min gradient to 90% MeOH, 9–10 min 90% MeOH). Flow rate, injection volume, and column temperature were 0.4 mL/min, 1 μL, and 45 C, respectively. Fraction 1 (~105 s) was analyzed in positive and negative ESI modes, and fraction 2 (~139 s) was analyzed in negative ESI mode only. Extracted ion chromatograms (EIC) of the Hg complexes were obtained from the TIC using the calculated theoretical mass ± 10 ppm.

The ¹⁹⁹Hg NMR experiments were recorded on a Avance III 600 spectrometer (Bruker, Karlsruhe, Germany) at a resonance frequency of 107.36 MHz using a broadband probe. The chemical shift was externally referenced to a saturated solution of HgCl₂ in D₂O which resulted in a signal at −1497 ppm. Spectra were acquired using a 90° pulse, a sweep width of 217.4 kHz, and 32,000 data points. A 0.3 s delay was used between the scans, and a total of 8000 scans were accumulated at 298 K. To conduct the actual measurement, we needed to make a 0.2 M solution of THI in a solution prepared by mixing 100 mM of Na₂HPO₄ and 100 mM of NaH₂PO₄ solutions each prepared with D₂O to obtain pH 7.4 followed by the addition of ACN to obtain 30% (v:v). We dissolved 80.8 mg of THI in 1.0 mL of the ACN containing 100 mM phosphate buffer. This solution was then filtered using a Pasteur pipet (VWR International, LLC, Radnor, PA, USA), which had been equipped with cotton on the inside into the 5 mm NMR tube and the ¹⁹⁹Hg-NMR spectrum was measured. Thereafter, we made a 1.0 mL solution of 0.2 M THI in 100 mM of phosphate-buffered solution containing 30% of ACN (v:v) of pH 7.4 prepared with D₂O and then added the amount of GSH to achieve a 30 mM solution (9.2 mg).

Raw LC-FAAS data were imported into Sigma Plot 15 and smoothed using the bisquare algorithm. Retention times (t_r) and Hg peak areas were determined using OriginPro software (Version 2025). All experiments were performed in quadruplets. Raw ESI-MS and ¹⁹⁹Hg NMR data files were imported into OriginPro software for spectral visualization.

4. Conclusions

Previously reported results pertaining to the Cys-mediated degradation of the organomercurial phenylmercuric acetate (PMA) at physiological pH to form a phenylmercury-cysteineate species prompted us to investigate the interaction of THI with the most abundant cytosolic low-molecular-weight thiol (GSH) at pH 7.4. Employing an LC-based approach allowed us to observe a progressive GSH-mediated degradation of THI, which was associated with the elution of a Hg-containing degradation product, which was identified as a GS-EtHg adduct by ESI-MS. The confirmation of this chemical reaction by ¹⁹⁹Hg-NMR experiments implies that under our chosen experimental conditions, THI is effectively degraded by GSH. To corroborate the LC-results that were obtained with GSH-containing mobile phases, we replicated these experiments with Cys-containing mobile phases and the same concentration range, obtaining results that were comparable, suggesting the formation of a Cys-EtHg adduct. Taken together, our results serve as a starting point to investigate if THI will similarly interact with GSH and Cys in biochemically complex cell lysates to form GS-EtHg and/or Cys-EtHg adducts. These adducts may interfere with other biochemical processes within cells (e.g., calcium channels) to provide a mechanistic link to adverse biochemical toxicological effects. The results of our investigations are also relevant to the development of better antivirals, as PMA and THI have been demonstrated to be effective inhibitors of the main protease of SARS-CoV 2 [11], which may involve processes similar to the established mechanism depicted in Figure 7.