A novel panel of yeast assays for the assessment of thiamin and its biosynthetic intermediates in plant tissues

Summary Thiamin (or thiamine), known as vitamin B1, represents an indispensable component of human diets, being pivotal in energy metabolism. Thiamin research depends on adequate vitamin quantification in plant tissues. A recently developed quantitative liquid chromatography–tandem mass spectrometry (LC–MS/MS) method is able to assess the level of thiamin, its phosphorylated entities and its biosynthetic intermediates in the model plant Arabidopsis thaliana, as well as in rice. However, their implementation requires expensive equipment and substantial technical expertise. Microbiological assays can be useful in deter‐mining metabolite levels in plant material and provide an affordable alternative to MS‐based analysis. Here, we evaluate, by comparison to the LC–MS/MS reference method, the potential of a carefully chosen panel of yeast assays to estimate levels of total vitamin B1, as well as its biosynthetic intermediates pyrimidine and thiazole in Arabidopsis samples. The examined panel of Saccharomyces cerevisiae mutants was, when implemented in microbiological assays, capable of correctly assigning a series of wild‐type and thiamin biofortified Arabidopsis plant samples. The assays provide a readily applicable method allowing rapid screening of vitamin B1 (and its biosynthetic intermediates) content in plant material, which is particularly useful in metabolic engineering approaches and in germplasm screening across or within species.

The following Supporting Information is available for this article:

S1
Supporting Fig. S1. Schematic overview of the sample preparation procedure for the LC-MS/MS method. LC-MS/MS determination of thiamin, the precursors HMP and HET, and the phosphate derivatives TMP and TPP in Arabidopsis thaliana, as described in detail by (Verstraete et al., 2020).

S2
Supporting Fig. S2. Schematic representation of yeast microbiological assays for thiamin determination. The microbiological assay protocol was adapted from the assay utilizing the Saccharomyces cerevisiae thiazole biosynthesis mutant, thi4 (Kall, 2003;Raschke et al., 2007;Chatterjee et al., 2011;Mangel et al., 2017). Abbreviation: OD, optical density.  This was performed to assess whether some discrepancies in measured MTE (molar thiamin equivalent) from the different metabolites can in part be explained by a specific metabolite stock solution. The measured MTE values upon spiking with the different thiamin-related metabolites are depicted for yeast assay RWY16 (a; can be rescued by HMP(-P(P)), TMP, thiamin and TPP)), thi20/21 (b; can be rescued by HMP-PP, TMP, thiamin and TPP), thi4 (c; can be rescued by HET(-P), TMP, thiamin and TPP) and thi6 (d; can be rescued by TMP, thiamin and TPP). Bars indicate the mean ± SE of 5 samples, whether or not spiked with a particular metabolite. Significant differences were determined via parametric tests, as using the Shapiro-Wilk test, the experimental data were found to follow a normal distribution. ANOVA test revealed presence of significant differences within the groups, identified using the post-hoc Tukey's test. Different lower case letters indicate significant differences between groups (p < 0.05).

S5
Supporting Fig. S5. Analysis of 60 different metabolite-spiked plant extracts with different yeast assays. Plant samples, originating from 15 days old WT Arabidopsis complete seedlings, grown on thiamin-free half strength MS medium were used as starting material for the analyses. Sample extracts were spiked with HMP, HET, TMP, thiamin or TPP to test the ability of the yeast assays to identify the increase in a specific thiamin-related metabolite (see schematic representation of the experimental setup in Fig. S3). The molar concentration of the metabolites spiked was set at a specific level, aimed at surmounting the molar thiamin equivalent (MTE) of non-spiked plant samples threefold. Data represent the mean ± SE of three technical replicates. Samples were defined as harboring an either low  (THI1), both AtTHIC and AtTHI1 (TT) or AtTHIC, AtTHI1 and AtTH1 (TTT), described by (Strobbe et al., 2021b), were utilized. The datasets (in A and B) were tested for normality using the Shapiro-Wilk test. In case of normality, statistically significant differences were detected via a two-sided T-test (scedasticity depending on the outcome of a preceding F-test). In case of non-normality, Mann-Whitney U test was used to identify significant differences. Significant differences are indicated by a single asterisk (p < 0.05) or double asterisks (p < 0.01). For the engineered lines (B), the significant difference depict comparison with the wild type (WT).

S7
Supporting Fig. S7. Assessing the ability of the thi6 yeast assay to identify high thiamin lines. Here, MTE values acquired via thi6 yeast assay of the lines, harboring above 1.5-fold increase in total vitB1 level, as measured by LC-MS/MS and compared to WT, are presented. The engineered lines overexpressing AtTHIC (THIC), AtTHI1 (THI1), both AtTHIC and AtTHI1 (TT) or AtTHIC, AtTHI1 and AtTH1 (TTT), described by (Strobbe et al., 2021b), were utilized. Eight engineered lines depicted above 1.5-fold enhancement of total B1 levels, when measured by LC-MS/MS analysis (Fig. 6a). Mean values ± SE of 3 (transgenic) or 6 (WT) biological replicates are shown. This graph zooms in on the result of the thi6 yeast assay of these lines, to examined whether this above 1.5-fold enrichment in in planta vitB1 level is recognized by the thi6 assay. These lines are indeed found to exhibit MTE values above WT, albeit very limited for line TT-25.5 (WT level is set to 1, red dashed line). The black dashed line corresponds to the 1.5-fold increase of MTE over WT, which is the threshold level needed to be reached by LC-MS/MS metabolite level to be selected here. Given the presented correlation between the LC-MS/MS and thi6 assay data (relative values; see  (TT) or AtTHIC, AtTHI1 and AtTH1 (TTT), described by (Strobbe et al., 2021b), were utilized. (a) As the only difference in the set of metabolites detected by the THI6 and THI20/21 assays is that the latter is theoretically also detecting the pyrimidine intermediate HMP-PP (Fig. 1b), their combined usage could be considered a crude method of estimating HMP-PP presence in samples. Unfortunately, this pyrophosporylated entity cannot be detected in the LC-MS/MS method (as it depends on measurement of phosphatase treated samples, thereby not enabling the distinction between HMP-P and HMP-PP) (Verstraete et al., 2020), making it impossible to verify the relative presence of this metabolite. As HMP-PP is a labile compound, it could have deteriorated in the cooking step (sterilization of plant samples). Therefore, thi6 and thi20/21 assays were conducted on samples of transgenic plants (selected to have a high probability of HMP-PP presence, due to high pyrimidine accumulation), which were filter sterilized (no cooking step was included in sample preparation).The higher MTE measured in the thi20/21 assay as compared to the thi6 assay observed in line   serving as an estimate for pyrimidine (HMP-(P(P))) accumulation. The datasets were tested for normality using the Shapiro-Wilk test. In case of normality, statistically significant differences were detected via a two-sided T-test (scedasticity depending on the outcome of a preceding F-test). In case of non-normality, Mann-Whitney U test was used to identify significant differences. Significant differences are indicated by a single asterisk (p < 0.05) or double asterisks (p < 0.01). For the engineered lines (B), the significant difference depict comparison with the wild type (WT). estimate for thiazole (HET-(P)) accumulation. The datasets (in A and B) were tested for normality using the Shapiro-Wilk test. In case of normality, statistically significant differences were detected via a two-sided T-test (scedasticity depending on the outcome of a preceding F-test). In case of non-normality, Mann-Whitney U test was used to identify significant differences. Significant differences are indicated by a single asterisk (p < 0.05) or double asterisks (p < 0.01). For the engineered lines (B), the significant difference depict comparison with the wild type (WT).

S11
Supporting Fig. S11. Preliminary results of utilizing thi4 yeast assay to estimate sum of total vitB1and thiazole content in brown and polished rice seeds. Genetically engineered rice lines, originating from a thiamin biofortification study (Strobbe et al., 2021a), were assessed via yeast assay and compared to LC-MS/MS methodology. (a) LC-MS/MS analysis of total B1 + thiazole content of both brown (unpolished) as polished rice seeds. (b) Total B1 + thiazole as estimated by thi4 yeast assay. The yeast assay is able to detect differences between WT and engineered lines, both in polished and unpolished rice seed samples. Bars indicate mean ± SE of two repeats. Significant differences as compared to wild type (WT), as identified using a two-sided heteroscedastic T-test, are indicated by a single asterisk (p < 0.05) or double asterisks (p < 0.01). used in this study, are indicated. A list of growth restoring metabolites is provided, which is based on the genetic and biosynthetic knowledge of the specific mutant line. The source of which the specific strain was obtained is shown.
Studies characterizing the mutated yeast genes are cited. The genetic background in which the mutant lines were created, are presented. It is important to note that the different lines contain additional auxotrophies (amino acids and uracil), which do not need to be taken into consideration, as all lines are grown on media supplemented with amino acids and uracil, provided by the Complete Supplement Mixture (Formedium, DCS0011). different plant sample were subjected to the four yeast assays (RWY16, thi4, thi6 and thi20/21) and were scored for each separate assay, the results of which are shown in Fig. S5. This allowed allocation of each sample to the baseline group or high metabolite group, which is indicated in the table as low and high, respectively. Attributing one sample to the 'low' or 'high' group, was performed for each of the four yeast assays. This allowed, based on the known array of metabolites rescuing the specific mutant strain (Fig. 1c), to score the samples based on the expected outcome (grid score) of a certain metabolite (Fig. 3b). For instance, a sample spiked with HMP is expected to fall into the 'high' group using the RWY16 assay, while being part of the 'low' or baseline group in the thi4, thi6 and thi20/21 assays (e.g. sample 1). This allowed, correct allocation of all spiked samples into the following groups (HMP spiked, HET spiked, B1 spiked and mock (non-spiked)).

Methods S1 Yeast strains
A relevant overview of Saccharomyces cerevisiae thiamin biosynthesis, as well as the utilized strains, is shown in Fig. 1. The applied microbiological assays were adapted from the originally described yeast assay using the Saccharomyces cerevisiae thiazole biosynthesis mutant, thi4 (Kall, 2003;Raschke et al., 2007;Chatterjee et al., 2011;Mangel et al., 2017). Additionally, assays utilizing thiamin auxotrophic yeasts, mutated in the pyrimidine branch of thiamin biosynthesis, were included. These employ the strains RWY16 (Wightman and Meacock, 2003) and thi20/21 (Llorente et al., 1999;Kawasaki et al., 2005), which are unable to synthesize thiamin due to defective HMP synthesis and HMP-P phosphorylation, respectively. For this work, the mutant thi20/21 was obtained following the indications in the original publications (Llorente et al., 1999;Kawasaki et al., 2005), while the RWY16 mutant strain was kindly provided by Dr. R. Wightman. Moreover, Saccharomyces cerevisiae strains thi6 (Nosaka et al., 1994) and thi80 (Nosaka et al., 1993) were included. The former is unable to condense the pyrimidine and thiazole moieties to form TMP, while the latter is impaired in the synthesis of TPP. As these involve mutations further downstream in thiamin biosynthesis, these strains are hypothesized to allow specificity towards B1 vitamers when used in assays, limiting interference from intermediates.  (Strobbe et al., 2021a). Upon collection, plant tissue samples were individually weighed, flash frozen in liquid nitrogen and subsequently stored at -80°C until further processing.

Methods S2 Sample preparation
The protocol for plant extract preparation was adapted from (Kall, 2003;Raschke et al., 2007;Mangel et al., 2017). Samples were pulverized and homogenized using a Retsch milling machine. A subsequent extraction and sterilization step was performed by addition of 1.5 ml of 22 mM sulfuric acid and incubating the samples at 95°C for 60 min (shaking at 500 rpm). The following steps (including the preparation of the assay) were performed in sterile conditions.
Neutralization of the samples was achieved by addition of 240 µL sterile 3 M sodium acetate, ensuring a pH of 5.7. Excess of plant material (cell debris) was removed by centrifugation (5 min at 14000 g and 4°C). Subsequently, the clear supernatant was transferred to a new (sterile) tube.
The final plant extracts were stored at -80°C until usage in the yeast assays (see further).

Methods S4 Yeast cultures
The different yeast strains were propagated on yeast extract-peptonedextrose (YPD) medium, which allows growth of the auxotrophic mutant lines (Raschke et al., 2007;Mangel et al., 2017). A single colony growing on solid YPD medium (15 g/L agar, Sigma-Aldrich) was selected per each strain and sub-cultured on liquid thiamin free yeast medium (TFYM) supplemented with 100 ng/mL thiamin. TFYM comprises yeast nitrogen base without amino acids and without thiamin (CYN4701, ForMedium), enriched with complete supplement medium (DCS0011, ForMedium) and 1% sucrose (following the manufacturer's recommended concentrations). In practice, 6.9 g CYN4701, 790 mg DCS0011 and 10 g sucrose are dissolved in 500 ml distilled water and autoclaved to yield double concentrated TFYM (2x TFYM). The yeast strains were grown for 1 day at 28°C shaking at 200 rpm, until an optical density (OD) above 1 was reached. Subsequently, the cultures were pelleted (1700 g, 5 min) and resuspended in 0.85% NaCl, which was repeated 3 times to eliminate any residual presence of thiamin. The yeast cells were again pelleted (1 700 g, 5 min) and resuspended in TFYM (no vitB1) to reach an OD of 0.5.
The yeast cultures were stored at 4°C until usage in the microbiological assays.

Methods S5 Assay protocol, conditions and data acquisition
The yeast assays were performed using 24-well plates (CELLSTAR, Greiner Bio-One), in sterile conditions. The following constituents were added to each well: 250 μL of 2xTFYM (double concentrated TFYM), 50 μL yeast (OD 0.5), 233 μL water, 67 μL sample (plant extract or standard). Each standard concentration was analyzed in quadruplicate. The 24-well plates, harboring 600 μL assay volume in each well, were incubated at 28°C for 17 hours (200 rpm). The optical density of the wells was measured using a TECAN Infinite 200 Pro plate reader, which allows rapid acquisition of the OD data. The OD data of the thiamin standards (4 of each concentration) were utilized to obtain a doseresponse curve. For all analyzed yeast strains, a dose-response curve was constructed, which was used to determine the thiamin level (or molar thiamin equivalent, MTE) present in the plant extracts. The amount of metabolites measured was calculated from the MTE present in the plant extracts, taking the initial weight of the plant sample into consideration (± 120 mg). Another set of samples was supplemented with only sterile extraction buffer, referred to as 'mock'.

Methods S6 Spiking of plant extracts
The sample sets were randomized and assessed using the described yeast assays.