Final Report for GNE14-082
Hydrogen sulfide (H2S) is a leading cause of sulfurous off-aromas in wines. H2S can increase during post-bottling storage under low oxygen conditions, but the precursors responsible for in-bottle H2S formation are not well understood. We had previously observed that elemental sulfur (S-0) pesticide residues on grapes can generate H2S under reductive storage conditions. Because S-0 is widely used among grapegrowers as a relatively safe and inexpensive powdery mildew control, determining appropriate S-0 levels to avoid excess H2S precursors in wine is of importance to the wine and grape industries.
To determine appropriate limits on S-0 residues, grape juice was spiked with S-0 (0-100 mg/L) and fermented to dryness. The resulting wines were bottle aged for 3 months. Free H2S increased during storage proportional to the original S-0 spike, corresponding to the conversion of approximately 1% S-0 residue to soluble H2S precursor, with some variation across yeast strains. Based on this work, we established an S-0 limit of 7 mg/kg to avoid an increase of H2S during storage. These results were shared with the regional wine and grape industries through conference talks, extension publications and online webinars.
To determine the chemical structure of S-0 derived H2S precursors, S-0 spiked wines were separated into different fractions by column chromatography. Using synthesized standards and LC-MS, we tentatively identified glutathione trisulfide as a precursor of H2S in S-0 fermented wines. These results were presented at scientific conferences, and should help in future goals of developing corrective methods for wines produced from S-0 contaminated grapes.
Powdery mildew (PM) is a common disease affecting grapevines, particularly in the northeastern region of the United States. The most common fungicide for control of this disease, elemental sulfur (S-0), is preferred due to its low cost, efficacy, and environmental sustainability. However, S-0 has been shown to result in hydrogen sulfide (H2S) “rotten egg” aroma formation during winemaking. Furthermore, my own work has demonstrated that this formation can continue in the wine bottle, particularly in cases of low oxygen ingress (e.g. screw cap closures). These consequences discourage the use of S-0 against PM and grape growers instead rely on less environmentally and economically sustainable PM fungicides close to harvest. Prior to this work, recommendations for sulfur spray limits at harvest were based upon the amount of S-0 necessary for increasing H2S production during fermentation. The goal of this project was to define a more useful limit, based upon the amount of residue necessary to form H2S during storage.
The objectives as described in the original plan of work are:
- Define new limits for use of S-0 fungicides on the basis of H2S formed during storage rather H2S formed during fermentation.
- Identify the wine soluble precursors responsible for H2S formed during storage.
The first objective was achieved. From our experimental results, we were able to correlate TCEP-releasable H2S post-fermentation to free H2S production during bottling. Based upon this conversion, we concluded that sulfur residues at harvest should not exceed 7 mg/L. This recommendation is in agreement with the 10 mg/L S-0 spray recommendation made in a previous study (Kwasniewski, 2014).
Towards the second objective, we have tentatively identified a supplementary source of H2S that is specific to wines fermented on elemental sulfur residues. A glutathione trisulfide derivative was identified in a treatment wine, whereas it was non-detectable in the control wine. Glutathione trisulfide was demonstrated to have TCEP-releasable H2S. Future work for a more complete conclusion will involve spiking glutathione trisulfide into wine and demonstrating release of H2S under reductive conditions. An additional future direction will be the untargeted investigation of other polysulfide derivatives.
Barriers to the completion of the second objection include the extremely low concentration of the analyte, which makes detection difficult without advanced tools (e.g. high resolution mass spectrometry, relevant software). Furthermore, the extremely polar nature of the analytes posed a road block for isolation by solid phase extraction or flash chromatography.
Materials. Gas detection tubes for H2S (Gastec 4LT) were purchased from Nexteq (Tampa, FL). Methanol (MeOH), acetonitrile (ACN), dichloromethane (DCM), formic acid, tris(2-carboxyethyl)phosphine (TCEP), glutathione disulfide (GSSG), elemental sulfur (S-0), potassium metabisulfite (PMBS), diammonium phosphate (DAP), and ammonium carbonate, were purchased at ≥99% purity from Sigma-Aldrich (St. Louis, M). Alka-seltzer tablets (Bayer Healthcare, Morristown, NJ) and organic grape juice (Cascadian Farms, Skagit Valley, WA) were purchased locally. Distilled de-ionized water was used for all experiments.
Production of wines with H2S precursor. Grape juice samples were prepared from reconstituted frozen organic grape juice concentrate. Juice was spiked with wettable sulfur in concentrations of 0, 20, and 100 mg/L, prepared in duplicate 1-L lots. Samples were fermented to dryness using a commercial yeast with prophylactic nutrient addition. Upon completion, samples were racked once, and sparged until free H2S was not detectable by 4LT gas detection tubes. At this point, wines were also assessed for levels of TCEP-releaseable H2S. Wines were bottled under nitrogen in 187-mL wine bottles, and sealed with oxygen scavenging crown caps. Oxygen level was monitored during bottling to ensure that it did not rise above 5%.
For the assessment of yeast strain effect on H2S production from elemental sulfur residues, fermentations were prepared in triplicate 100-mL lots, and elemental sulfur was added to grape juice at 2 levels: 0 mg/L and 100 mg/L (control and treatment). The yeast strains used included: 58W3, BRL97, EC1118, Alchemy I, and CY3079. (3 fermentation replicates * 2 treatments * 5 yeast strains = total of 30 samples.)
Quantification of free H2S. To quantify free H2S, an aliquot of approximately 30 g of wine was weighed into a plastic squeeze bottle and two alka-seltzer tablets were added to deaerate the sample. The cap, fitted with a 4LT H2S detection stick, was immediately replaced tightly on the bottle so that no gas was allowed to escape. The sample was left until all fizzing had stopped. The length of color change on the H2S detection tube was used to quantify the H2S that was sparged from the sample, according to a calibration curve.
Tentative Quantification of Latent Precursor. To assess TCEP-releaseable H2S, samples were sparged as described above, and a few granules of tris(2-carboxyethyl)phosphine (TCEP) were added to the wine sample and allowed to react for 5-10 minutes. The sparging process was then repeated through addition of another alka seltzer tablet and immediate replacement of the cap. The new length of color change on the H2S detection tube was used to tentatively quantify the “latent” H2S, according to a calibration curve.
Flash chromatography was performed on a Combiflash RF75 system. Two methods were tested. Reversed-phase: The column was a 5.5g C18 “gold” column (Teledyne Isco). Solvent A was 0.1% formic acid in water and solvent B was 0.1% formic acid in MeOH, and the gradient was as follows: 5% B, held for 2.5 minutes, linear gradient to 95% B for 10 minutes, 95% B held for 10 minutes. The flow rate was 18 mL/min and the equilibration volume was 28.7 mL. 30 mL of wine which had been fermented on 0 mg/L S0 (control) or 100 mg/L S0 (treatment) was distilled down to approximately 2 mL on a rotary evaporator and injected onto the equilibrated column. Absorbance was monitored at 214 nm and 280 nm. 18 fractions were collected and assayed for free and TCEP-releaseable H2S.
Normal phase (HILIC method): 50mL of wine (treatment and control) was dry-loaded on acid-washed Celite. The column used was a 24g silica “gold” column. Solvent A was DCM and solvent B was MeOH. The flow rate was 35 mL/min and the equilibration volumne was 252 mL. The gradient was as follows: 2.5% B, held for 2 minutes, linear gradient to 80% B for 10.5 minutes, and held at 80% B for 1.5 minutes. Absorbance was monitored at 214 nm and 280 nm. 30 fractions were collected and assayed for free and TCEP-releaseable H2S.
Synthesis of Glutathione Trisulfide (GS3G). GS3G was prepared according to the method of Moutiez et al. (Moutiez et al., 1994). 500 mg (0.8 mmol) of glutathione disulfide (GS2G) was added to a solution of 261 mg (8 mmol) elemental sulfur (S0) in EtOH/CHCl3/CS2/NH4OH (45/5/2/2). The reaction was stirred at 30 °C for 2.5 hours, and acidified to pH2 with concentrated HCl. The solvent mixture was removed by rotary evaporation and the solute was reconstituted in distilled water. The resulting mixture was filtered in a Buchner funnel to remove S0, followed by filtering in a 0.2 µm PTFE membrane filter.
Solid Phase Extraction and Sample Preparation for LC-MS. A solid phase extraction method was optimized using a 30mg/ 1cc OASIS MAX cartridge (Waters Corporation, Milford, Maxxachusetts) for the extraction of glutathione trisulfide. The SPE column was primed with 1mL MeOH followed by 1mL of 5mM (NH4)2CO3. 1mL of sample was adjusted to pH8-9 and loaded, following which the cartridge was aspirated to dryness. The cartridge was washed with 1mL 50mM (NH4)2CO3, followed by 1mL ACN. Finally, the analyte was eluted with 3mL 10mM trifluoroacetic acid. The eluate was evaporated to dryness and reconstituted in the starting mobile phase. Alternatively, in the interest of recovery, a 30mL sample of wine was pre-concentrated to 1mL and filtered in a 0.2 µm PTFE membrane filter prior to LC-MS injection.
Liquid Chromatography – Mass Spectrometry (LC-MS). For initial analyses, the instrument used was a Thermo Finnigan Surveyor MS Pump Plus equipped with a Surveyor Autosampler Plus, and interfaced to a TSQ Discovery MAX mass spectrometer (Gentech Scientific, Arcade, NY). For subsequent high resolution analyses, the instrument used was an Orbitrap Elite mass spectrometer and liquid chromatography system. Gradient elution of GS2G, GS3G, and subsequent polysulfides was adapted from the method of Zhu et al. (Zhu et al., 2007), and was performed employing a Phenomenex Jupiter 5u C18 300A column (150mm x 2 mm i.d., 5 µm; Phenomenex, Inc., Torrance, CA, USA). Mobile phase A was 0.1% formic acid in water and mobile phase B was 0.1% formic acid in MeCN. The gradient was performed at a flow rate of 0.21 mL/min at room temperature, as follows: 1.6% B held for 2.5 minutes, 6.4 % B at 2.51 minutes, 6.4% B at 4.5 minutes, 56% B at 4.51 minutes, 80% B at 11 minutes, 80% B at 15 minutes, and 1.6% B at 15.5 minutes, followed by a 8.5 minutes of equilibration. Operating conditions were ESI positive ion mode with a voltage of 4.0 kV, sheath gas 50, auxiliary gas 24, sweep gas 7. The source temperature was 350 °C. The scan was a full MS scan in FTMS at 120,000 resolving power (at m/z 400) across m/z 350-750 in profile mode followed by two MS/MS data-dependent scans acquired from 0 to 24 minutes. Injection volume was 20 µL. Data analysis was performed using Xcalibur 2.2.
Prior studies have shown that H2S can accumulate during storage from odorless precursors. Several potential precursors have been suggested, but the only verified source is metal-sulfide complexes (Franco-Luesma et al., 2014). Such complexes cannot account for all of the H2S released during storage. In particular, our own work has shown that TCEP, though relatively ineffective at releasing metal-bound sulfide, can release H2S in wines fermented on elemental sulfur residues Table 1. This led us to hypothesize that elemental sulfur has the ability to form a supplementary latent source of H2S.
In an additional experiment, we assessed the ability of different strains to produce the unknown precursor when fermented in the presence of elemental sulfur residues. The results are shown in Figure 1. Through a Multiple ANOVA we could ascertain that variation across fermentation and analytical replicates was non-significant at α=0.01, while yeast strain did have a significant effect on both free and latent H2S production (p<0.001). The data from table 1 and Figure 1 were used to calculate new recommendations for elemental sulfur spraying limitations, as described in the following section (“Impact of Results/Outcomes”).
Based upon the above results, we further hypothesized that the precursor may exist as polysulfide derivatives, e.g. glutathione trisulfide, cysteine trisulfide, and/or higher polysulfide derivatives and assymetrical polysulfides. To test this hypothesis, an LC-MS/MS method was developed for the detection of gluatathione polysulfide derivatives. Glutathione polysulfide derivatives were synthesized and used to optimize the method. The Jupiter C18 300A column allowed good retention and separation of the glutathione polysulfide derivatives. Retention times were as follows: 3.5 minutes (GS2G), 8.5 minutes (GS3G), and 9.6 minutes (GS4G). It was ascertained that all three derivatives were stable at wine pH over 6 months of storage.
A solid phase extraction method was optimized for the isolation of glutathione polysulfides. The very polar analytes were able to be retained on an OASIS MAX SPE cartridge. However, recovery was severely compromised when the anlaytes were spiked into a wine matrix. Additional experiments verified that this was due to a combination of protein and organic acids (e.g. tartaric acid, malic acid), but a great deal of difficulty was met in removing these interferences. Therefore, the sample used for LC-MS injection was a pre-concentrated wine sample with no sample clean-up.
Due to the extremely low expected concentrations of the targeted analytes in the real samples (S-fermented wines) improved instrumentation was needed in order to detect the compounds. Therefore, analyses were run on an Elite Orbitrap Mass Spectrometer to exploit the high mass defect of the sulfur atom in high resolution MS. With this technique, we were able to tentatively identify the targeted compound in the treatment wine, though it was undetectable in the control wine. Future work will be needed for a more robust affirmation of this hypothesis (see “Areas Needing Additional Study” below).
The LC-MS/MS method for detection of glutathione polysulfide derivatives was not able to be successfully translated to a flash chromatography method. We originally observed that GSH and its polysulfide derivatives were not able to be retained on a typical C18 HPLC column. Zhu et al., hypothesized that the larger pore size (300A) of the Phenomenex Jupiter C18 column was what allowed retention of these hydrophilic compounds.
Several flash chromatography methods were attempted on the S-0 fermented wines in an attempt to isolate the H2S-releasing fraction. The TCEP assay requires >15 mL of wine to detect released H2S, so HPLC fractionation was not an option. Using the reversed-phase flash chromatography method, the H2S-releasing fraction was not able to be retained on the C18 column.
The HILIC method was then tested to see if it was possible to retain the H2S-releasing fraction. This method resulted in artifacts – the total TCEP-releasable H2S of the combined fraction exceeded that of the unfractionated wine sample. By using DCM and MeOH as solvents A and B, and correcting for artifacts against a control sample, we were able to determine that the H2S-releasing fraction eluted in the later fractions (%MeOH > 50), but did not elute cleanly (Figure 2). This suggests that the H2S reservoir exists as one or more water-soluble compounds. For instance, the reservoir could exist as multiple polysulfide derivates (e.g. glutathione trisulfide would elute before higher order polysulfide derivatives, but all would be capable of releasing H2S).
Translating bench results into practical recommendations
During late Summer and Fall 2015, I combined data sets on conversion rates of sulfur residues to H2S precursors and ultimate post-bottling H2S formation in order to calculate a practical limitation on sulfur spray residues for winemakers and grape growers. Specifically, I used my experiments to calculate the percent of “latent” H2S (as determined by our own developed reductive assay) which would be converted into free H2S following 3 months of storage under reductive bottling conditions. Then, using the data on the formation of latent H2S across a range of yeast strains, I was able to calculate a maximum sulfur residue concentration which could be recommended to producers. This value was comparable to that previously determined in our lab based upon larger-scale trials. This recommendation has been communicated at presentations (see “oral presentations”, below) including a Northern Grapes Project Webinar and several conferences, and has also been included in written communications geared towards industry members (see “written communication”, below).
Education & Outreach Activities and Participation Summary
The scope of the extension work for this project includes i) communicating my results to industry and extension agents both orally (webinars, conferences) and ii) in written forms (online fact sheets and technical articles). An additional unplanned outcome was iii) mentoring an undergraduate Summer Scholar student in applied research related to defining appropriate spray limits.
- Oral Presentations:
Objective: To provide practical knowledge to industry members with application to winemaking and grape growing, including new developments in our understanding of hydrogen sulfide formation in wine and best practices for avoiding off-aromas without compromising sustainability. To present work in settings that are interactive between industry, extension, and academic members of the winemaking community; to communicate research findings and gain feedback and practical insight.
Oral presentations, event format, date, and number/background of attendees are described in Table 2.
Webinar: I developed and presented a webinar for the USDA supported Northern Grapes Project as part of their webinar series. Due to the challenging growing conditions of New England, Northern New York, and the Upper Midwest, powdery mildew, and consequently sulfur residues, can be a cause for concern in these regions. The webinar was developed with feedback from Chris Gerling, an extension associate at Cornell, as well as my advisor Gavin Sacks. I presented a webinar on known causes of hydrogen sulfide off-aromas in wines, presented a previously developed method for the rapid and inexpensive quantification of sulfur residues on grapes, presented previously published data investigating the persistence of sulfur residues on grapes during the vinification process, and provided recommendations for the best practices to avoid off-aromas based upon our findings. Following the Webinar, I responded to questions from viewers regarding the content. This Webinar is still available for viewing on the Northern Grapes Project Website.
Conference Presentations (July 2015, November 2015, March 2016)
The American Society of Enology and Viticulture Eastern Section (ASEV-ES) Annual conferences have a strong industry presence and as such allow for communication between academic, industry, and extension partners. ASEV-ES 2015 had about 80 attendees, ¼ of which were industry members. I received ideas and feedback from other academic members, but also communicated findings to and receive feedback from industry members to whom this work is applicable. Similarly, the CRAVE (Cornell Recent Advances in Viticulture and Enology) annual conference presented a similar, short format for summarizing my recent work for extension agents. While the annual conference of the American Chemical Society (ACS) is more academic, there are many industry members from who attend the Ag & Food Division symposia. Elle Friedberg, the summer scholar who worked with me during 2015, presented our findings on the post-bottling re-emergence of hydrogen sulfide across a range of different yeast strains for small-scale fermentations.
2. Written communication –
Creation and distribution of a Fact Sheet for reference material on sulfur pesticide residues and sulfurous off-aromas
Objectives: To provide a quick resource for winemakers and growers on sulfur pesticide residues, their role in sulfurous off-aromas, and best practices to avoid or remediate these problems. The communication of these findings in a simple and straight-forward format will forward the understanding of off-aroma reappearance in wines, and propagate the implementation of good winemaking practices. Specifically, the reference sheet includes:
- Overview of potential hydrogen sulfide precursors and their causes, including sulfur pesticide residues;
- Fate of residues before, during and after post-fermentation
- Potential for sulfur residues to contribute to formation of sulfurous off-aromas in bottle
- Recommended maximum residue limits based on wine style, approximate pre-harvest intervals to achieve these limits, and strategies for measurement
- Recommended practices to minimize sulfur residues prior to fermentation
Article for publication in trade or extension publication (August 2016)
Objective: To provide a 4-5 page article describing research findings in a non-technical format; to give winemakers and grape growers practical insight into how our research findings can be applied in local winemaking practices; and to update recommendations of sulfur spray limitations based upon an improved parameter.
Although related work has recently been accepted for publication in the American Journal of Enology and Viticulture (AJEV), and further academic publications are in the works, these articles are not readily accessible to industry members, nor do they emphasize the practical applications of this work. A more relevant extension activity is the publication of an article emphasizing the outcome of our work instead of the chemical mechanisms and analytical methods used to obtain results. To this end, we have published such an article in Appellation Cornell as a Research Focus in the Fall 2016 issue. The article can be found at:
3. Mentoring Summer Scholar (6/2015-8/2015)
Objective: To redefine sulfur spray limitations on the basis hydrogen sulfide produced under post-bottling conditions; to provide an undergraduate summer scholar the opportunity to conduct graduate-level research in the Cornell Food Science Department.
The primary practical objective of this work from a winemaker’s or grape grower’s standpoint was to define new recommendations for sulfur spraying and subsequent wine making practices. Current recommendations are based upon the hydrogen sulfide produced from sulfur residues during fermentation, but our preliminary findings indicated that a more appropriate recommendation would be based upon the hydrogen sulfide formed post bottling. With this end in mind, my experiments for this project have focused on elucidating the chemistry behind post-bottling hydrogen sulfide formation; specifically, in isolating and identifying the elemental sulfur-derived precursor. While this understanding is instrumental in identifying the best practices for avoiding off-aromas in wine, a more direct approach to defining sulfur spray limitations was carried out as the project of an undergraduate student who worked with me as part of the Cornell Food Science Summer Scholar Program in 2015. The summer scholar in the Sacks’ lab, Elle Friedberg, selected 5 wine yeasts and conducted small-scale fermentations in triplicate for grape juice that had been spiked with elemental sulfur formulations such as one would encounter in harvested grapes. Using our previously developed techniques, Elle was able to quantify the “latent” hydrogen sulfide formed during fermentation, and calculate, using my previous data, how much hydrogen sulfide would be formed during storage of these wines. From this, we could extrapolate a maximum recommended elemental sulfur residue at the time of inoculation.
Areas needing additional study
Future work will be to verify the presence of glutathione trisulfide in treatment vs. control S-0 fermented wine, and its ability to release H2S under reductive conditions. Additionally, a non-targeted approach to identifying other potential polysulfide precursors will be attempted in winter 2016-17.