Effects of Hot Air Drying on the Nutritional and Phytochemical Composition of Radish (Raphanus sativus L.) Microgreens

ABSTRACT

Cruciferous microgreens are recognized for their high content of phytochemicals, particularly glucosinolates. Their high moisture content and water activity make them highly perishable after harvest and highlight the need for effective preservation methods that maintain their nutritional quality. This study evaluated the impact of hot air drying at 45°C, 65°C, or 95°C compared to freeze‐drying on the content and bioaccessibility of nutrients and phytochemicals in radish microgreens. Total phenolic content (TPC) and glucoraphenin (GE) exhibited opposite responses to heat: TPC decreased at 45°C (by 9%) and 65°C (21%), but did not change at 95°C, whereas GE decreased only at 95°C (by 21%). Vitamins B₁ and B₉ were stable across drying treatments, whereas vitamins B₂, B₃, and C were reduced by up to 35%, 36%, and 63%, respectively, with the exposure to the three temperatures. The in vitro bioaccessibility of TPC and vitamins B₁, B₃, B₉, and C ranged from 13% to 68% and did not differ between drying treatments. The bioaccessibility of vitamin B₂ was higher in samples dried at 65°C compared to the other treatments. GE and monomeric anthocyanin content (MAC) could not be quantified following in vitro digestion. The overall metabolomes of the samples were compared by liquid chromatography–mass spectrometry. Partial least squares‐discriminant analysis (PLS‐DA) showed separation by drying treatment, with glucosinolates and flavonoids driving the separation. Overall, this research indicates that hot air drying represents an effective approach for postharvest nutrient preservation in radish microgreens and may facilitate the use of radish microgreen powder in formulated foods.

Nomenclature

GAE

gallic acid equivalents

GE

glucoraphenin

LC–MS

liquid chromatography–mass spectrometry

MAC

monomeric anthocyanin content

MC

moisture content

PLS‐DA

partial least squares‐discriminant analysis

PPO

polyphenol oxidase

SD

standard deviation

TPC

total phenolic content

VIP

variable importance in projection

1. Introduction

Cruciferous vegetables from the Brassicaceae family, including radish, broccoli, and kale, are recognized for their health benefits, particularly in reducing the risk of chronic diseases, such as cancer and cardiovascular disease (Ren et al. 2024). These vegetables are rich in essential micronutrients, polyphenols, and glucosinolates (Podsędek et al. 2023). In addition to their health‐promotion effects, these polyphenols and glucosinolates play a role in the flavor and color of cruciferous vegetables (Mezzetti et al. 2022; Wieczorek et al. 2018)

Microgreens are small seedlings harvested at an early stage of growth and are gaining attention for their enhanced bioactivity (Demir et al. 2023). Microgreens, particularly those from the Brassicaceae family, have been found to contain higher levels of functional compounds, such as sulforaphane, compared to their mature counterparts (Di Gioia et al. 2023; Di Bella et al. 2020). For instance, Baenas et al. (2023) reported that oral administration of aqueous extracts of broccoli sprouts, but not mature broccoli, reduced the volume of C6 glioblastoma allografts in rats. This antitumor effect was attributed to the threefold higher concentration of sulforaphane in sprouts compared to mature broccoli.

Although cruciferous microgreens offer potential health benefits and potential utility in diversifying the diet, they are highly perishable with shelf lives of only 1–2 days at ambient temperature or 7–14 days under refrigeration (Xiao et al. 2014; Kou et al. 2013). Significant losses in nutritional quality occur even under refrigerated storage conditions (Vale et al. 2015; Waje et al. 2009). This rapid degradation presents a challenge when trying to increase the use of microgreens in the food system, especially in regions with limited access to refrigeration, and underscores the need for effective preservation techniques that do not compromise nutritional quality and bioactivity (Akter et al. 2022).

Hot air drying is a cost‐effective and widely used food preservation method that uses convective heat transfer to reduce moisture, leading to inhibition of microbial spoilage and postharvest enzymatic reactions (EL‐Mesery et al. 2023; Akter et al. 2022). However, it can also lead to undesirable changes in the chemical composition of vegetable products due to the elevated temperatures employed. Factors such as drying temperature and time can impact the retention of bioactive compounds and influence the nutritional and sensory quality of the final product (Boateng 2024; Li et al. 2020). For instance, hot air drying of cruciferous vegetables, including broccoli and radishes, has been shown to cause reductions in glucosinolates (Luo et al. 2022) and phenolic compounds (Dziki et al. 2020). Interestingly, Vargas et al. (2022) reported that hot air drying resulted in significantly greater vitamin C retention in broccoli and spinach compared to freeze‐drying. The authors attributed this to heat inactivation of enzymes that degrade vitamin C. Nevertheless, the authors pointed out that their results may extrapolate to other products.

While freeze‐dried microgreens are already marketed as nutrient‐dense powders for smoothies, teas, and healthy drinks, these products rely on expensive and energy‐intensive drying technologies. In contrast, hot air drying offers a low‐energy method that could make shelf‐stable microgreens more accessible for their utilization as functional powders, particularly in low‐resource settings. However, the effects of different hot‐air drying temperatures on the metabolite composition and bioaccessibility of bioactive compounds in microgreens have not been sufficiently explored. To our knowledge, there is only one study analyzing the effects of microwave and air frying on the metabolite profiles in broccoli microgreens (Tallei et al. 2024). A few others have examined the effect of drying on antioxidant capacity or broader classes of phytochemicals (e.g., total phenolic content [TPC]) of broccoli and radish microgreens (Dziki et al. 2020; Gunjal et al. 2025).

This study evaluated the impact of hot air drying at different temperatures on the nutritional properties of radish microgreens. The concentrations of glucoraphenin (GE), a panel of water‐soluble vitamins, total monomeric anthocyanin, and phenolic compounds, as well as the in vitro bioaccessibility of these components, were determined. Changes in the overall metabolome were also evaluated. Given the growing interest in the use of microgreens to enhance diet quality, a better understanding of the impact of hot air drying on the content of nutrients and nonnutrient phytochemicals in the dried product is needed. Such information can support rational integration of the dried powder into formulated foods to achieve desired nutrient and phytochemical profiles.

2. Materials and Methods

2.1. Chemical Reagents

Analytical standards of vitamins thiamine (B₁), nicotinic acid (B₃), folate (B₉), and l‐ascorbic acid (C) were purchased from Sigma–Aldrich (St. Louis, MO, USA). Methanol (analytical grade), acetonitrile (liquid chromatography–mass spectrometry [LC–MS] grade), and formic acid (purity grade) were purchased from Thermo Fisher (Waltham, MA, USA). Ultrapure water (18.2 MΩ·cm) was prepared in‐house (Q‐POD Millipore system, Danvers, MA, USA) and was used throughout the study. All the other reagents were of the highest grade commercially available.

2.2. Microgreens Samples and Growing Conditions

Untreated ‘Red Rambo’ radish (Raphanus sativus L.) seeds (99% germination rate, 11.2 g/1000 seeds) were purchased from Johnny's Selected Seeds (Winslow, Maine, USA) and seeded at 2 seeds/cm² into PRO‐MIX BX (Premier Tech Ltd., Rivière‐du‐Loup, Canada), a peat‐perlite growing mix, in black growing trays (12 cm × 16 cm × 5 cm) with drainage holes at the bottom. Microgreens were grown in a Conviron PGR‐15 reach‐in style plant growth chamber (Controlled Environments Inc., Pembina, ND, USA) at the Penn State Greenhouse Facilities (University Park, PA, USA). The growth chamber was set to 20°C and 70% relative humidity. During germination, growing trays were covered with a black polyethylene film and were kept in the dark. Once germination was complete, the black polyethylene film was removed, and radish seedlings were exposed to 200 µmol/m²/s radiation provided by broad‐spectrum fluorescent lights with a photoperiod of 16 h/day, realizing a daily light integral (DLI) of 11.5 mol/m²/day. During germination, growing trays were misted daily from the top with deionized water using a spray bottle. After complete germination, growing trays were irrigated via subirrigation. Upon reaching the stage with fully expanded cotyledons (10 days after seeding), radish microgreens were harvested from each growing tray by cutting shoots above the substrate using clean scissors.

2.3. Drying Procedure

2.3.1. Drying Curves

The moisture content (MC) in 2 g of fresh microgreens was determined using a moisture analyzer MA37 (Sartorius, Göttigen, Germany). Weight was recorded at 2‐min intervals, and the drying process continued until no measurable weight loss was observed and the samples reached less than 10% MC on a wet basis (Li et al. 2020; Tian et al. 2016). To investigate the drying kinetics of microgreens, drying curves were constructed based on the measured MC, the moisture ratio (MR, the unitless form of moisture), and the drying rate (DR) (Figure S1B–D).

MC (g water/g dry matter) at each time point was calculated using Equation (1):

$$\mathrm{MC}\left(\frac{\mathrm{g}\mathrm{water}}{\mathrm{g}\mathrm{dry}\mathrm{matter}}\right)=\frac{\left(M_0-M_{\mathrm{water}}\right)-M_{\mathrm{dry}\mathrm{matter}}}{M_{\mathrm{dry}\mathrm{matter}}},$$ (1)

where M ₀ is the initial mass (g), M _water is the weight of evaporated water at the time point (g), and M _dry matter is the weight of dry matter (g).

MR over time was calculated using Equation (2):

$$\mathrm{MR}=\frac{\mathrm{M}{\mathrm{C}}_t-\mathrm{M}{\mathrm{C}}_{\mathrm{e}}}{\mathrm{M}{\mathrm{C}}_0-\mathrm{M}{\mathrm{C}}_{\mathrm{e}}},$$ (2)

where MC _t is the MC at time t and MC₀ and MC_e are the initial and equilibrium MCs on a dry basis, respectively.

Drying rate was calculated using Equation (3):

$$\mathrm{DR}=\frac{\mathrm{M}{\mathrm{C}}_{t_2}-\mathrm{M}{\mathrm{C}}_{t_1}}{t_2-t_1},$$ (3)

where $\mathrm{M}{\mathrm{C}}_{t_1}$ and $\mathrm{M}{\mathrm{C}}_{t_2}$ represent MCs at two sequential time points, while t ₁ and t ₂ represent the corresponding times.

Five drying models were evaluated for their validity in describing radish microgreens dried under hot air (Table S1). The performance of each model was assessed based on standard error (SSE), reduced chi‐squared (χ ²), and root‐mean‐square deviation (RMSE), comparing predicted and experimental points using Equations (4) to (6). The model with the lowest SSE, χ ², and RMSE was considered the best fit. Our results showed that the logarithmic model provided the best fit for our data at 45°C and 65°C, while the Page model was more suitable at 95°C (Table S1).

$$\mathrm{SSE}=\sum _{i=1}^N{\left(\mathrm{M}{\mathrm{R}}_{\exp ,i}-\mathrm{M}{\mathrm{R}}_{\mathrm{pre},i}\right)}^2,$$ (4)

$${\chi }^2=\frac{{\sum }_{i=1}^N{\left(\mathrm{M}{\mathrm{R}}_{\exp ,i}-\mathrm{M}{\mathrm{R}}_{\mathrm{pre},i}\right)}^2}{N-z},$$ (5)

$$\mathrm{RMSE}={\left[\frac{1}{N}\sum _{i=1}^N{\left(\mathrm{M}{\mathrm{R}}_{\exp ,i}-\mathrm{M}{\mathrm{R}}_{\mathrm{pre},i}\right)}^2\right]}^{1/2},$$ (6)

where MR_exp is the experimental moisture ratio, MR_pre is the predicted moisture ratio, N is the number of observations, and z is the number of coefficients in the equations that have been used.

2.3.2. Hot Air Oven‐Drying

Hot‐air drying experiments were conducted at 45°C, 65°C, and 95°C (Dziki et al. 2020; Li et al. 2020) in a forced air oven (VWR, Radnor, PA, USA), which was preheated to the target temperature for 30 min prior to drying. Immediately after harvest, 45 g of fresh radish microgreens were evenly spread on a thin stainless‐steel wire mesh placed on top of the oven rack to allow uniform airflow. The oven was equipped with an internal fan located at the bottom of the chamber, ensuring even distribution of the heated air. The exhaust air escaped through a ventilation hole at the back of the oven. The temperature inside the drying chamber was digitally monitored and controlled using an integrated thermocouple. Samples were dried for the time needed to achieve less than 10% MC, as determined by monitoring weight loss over time and reference drying curves built during the drying process using a moisture analyzer. Once the drying time was completed, dried microgreens were cooled in a desiccator, and the final MC was confirmed at 105°C using a moisture analyzer. For freeze‐dried samples, fresh microgreens were placed in Ziploc bags immediately after harvest and frozen at −80°C for 24 h. Frozen microgreens were then freeze‐dried (Virtis Genesis 35XL, SP Scientific, Warminster, PA, USA) at −50°C under vacuum (280 mTorr) for 72 h. After drying, all samples were flash‐frozen in liquid nitrogen, ground into a fine powder using a coffee grinder, passed through a No. 60 sieve, and stored in moisture barrier pouches at −20°C for further analysis. Freeze‐dried samples were used as a control for hot air drying.

2.4. Color

The color of freeze‐dried and hot air‐dried radish microgreen powders was determined for each drying experiment using a colorimeter (Chroma Meter CR‐400; Konica Minolta, Japan) and the CIELAB system with a reference to illuminate D65 and a visual angle of 10°. Prior to the color measurements, the equipment was calibrated with a standard black and white ceramic tile. The color coordinates L* (lightness), a* (red to green), and b* (yellow to blue) were measured. The color difference (ΔE) between the hot air‐dried samples was calculated using the following equation (Al‐Dairi et al. 2021; Prieto‐Santiago et al. 2020):

$$\mathrm{\Delta }E=\sqrt{{\left({L^{\ast }}_f-{L^{\ast }}_i\right)}^2+{\left({a^{\ast }}_f-{a^{\ast }}_i\right)}^2+{\left({b^{\ast }}_f-{b^{\ast }}_i\right)}^2,}$$ (7)

where the subscripts refer to the readings of the thermally treated samples “f” and the sample used as a control “i.”

2.5. Phytochemical Composition

2.5.1. TPC

Dried microgreens (0.5 g) were suspended in 7 mL of acetone:water:acetic acid (80:20:0.1, v:v:v), stirred for 15 min at room temperature, and filtered through Whatman Paper No. 1. This was repeated twice, and then the filtrates were combined and dried to near dryness under vacuum (CentriVap Vacuum Concentrator; Labconco, Kansas City, MO, USA). The resulting residue was reconstituted in 2 mL of ultrapure water and filtered through 0.45‐µm nylon filters. An aliquot of extract (40 µL) was diluted with ultrapure water to a total volume of 1600 µL. The TPC of the dried microgreens was determined using the Folin–Ciocalteu method, as reported by Musci and Yao (2017), with slight modifications. Samples’ absorbance was measured at 765 nm using a microplate spectrophotometer (Multiskan GO microplate Spectrophotometer; Thermo Scientific, Waltham, MA, USA). Absorbance values were compared to a standard curve of gallic acid, and TPC was expressed as milligrams of gallic acid equivalents (GAE) per 100 grams of samples on a dry weight basis (Singleton et al. 1999).

2.5.2. Monomeric Anthocyanin Content (MAC)

Anthocyanin‐rich extracts were prepared by adding 0.1 g of dried samples to 4 mL of methanol:water:acetic acid (85:14:1, v:v:v), sonicating at 25°C for 15 min (VEVOR Ultrasonic Cleaner, China), and incubating at 4°C for 5 h (Mezzetti et al. 2022). Samples were then centrifuged at 1365 × g (Sorvall Legend XTR; Thermo Scientific) for 15 min, and the supernatants were filtered through 0.45‐µm nylon filters. The MAC was determined in triplicate using a pH differential method (Giusti and Worlstad 2001). Extract (80 µL) was combined with potassium chloride (0.025 M, pH 1) or sodium acetate (0.4 M, pH 4.5) to a final volume of 1.5 mL.

Absorbance in each buffer was measured at 510 and 700 nm using a microplate spectrophotometer (Multiskan GO microplate Spectrophotometer; Thermo Scientific). MAC was calculated as cyanidin‐3‐O‐glucoside equivalents, using Equation (8), and normalized to dry weight.

$$\begin{array}{ccc}\hfill \mathrm{MAC}& =& [{\left(A_{510}-A_{700}\right)}_{\mathrm{pH}1}-{\left(A_{510}-A_{700}\right)}_{\mathrm{pH}4.5}]\hfill \\ & & \times \left(\mathrm{MW}\times \mathrm{DF}\times 1000{\epsilon }^{-1}\right),\hfill \end{array}$$ (8)

where MW (molecular weight of cyanidin‐3‐O‐glucoside) = 449.2 g/mL; DF (dilution factor) = 18.75 (1500/80), and $\epsilon$ (molar extinction coefficient of cyanidin‐3‐O‐glucoside) = 26,900 L/mol·cm.

2.5.3. GE Content

GE was extracted from dried radish microgreens by combining 0.1 g of powder with 1 mL of 70% aqueous methanol and sonicating the mixture at 25°C for 10 min (Jo et al. 2022; Baenas et al. 2014). Samples were then placed into a heating bath (VEVOR) at 75°C for 1 h and vortexed every 5 min. Subsequently, samples were centrifuged at 4824 × g for 15 min at 4°C, and the supernatants were collected and brought to near dryness under vacuum (Labconco). The residue was reconstituted with ultrapure water and filtered through a 0.45‐µm nylon filter. GE was quantified by LC–MS analysis using an Agilent 1290 Infinity UPLC (Agilent Technologies, Palo Alto, CA, USA) interfaced with an Agilent 6460 Triple Quad mass spectrometer (MS; Agilent Technologies), which was equipped with an electrospray ionization interface (ESI) in the negative mode. The scan range of the MS was set to m/z 100–2000 with a scan time of 300 ms. The capillary voltage and nozzle voltage were 3000 and 1000 V, respectively. The desolvation gas (N₂) temperature was 350°C, and the flow rate was 4 L/min. The sheath gas (N₂) temperature was 250°C, and the flow rate was 5 L/min. Chromatographic separation was achieved using an Acquity UPLC HSS C18 column (2.1 mm × 100 mm, 18 µm; Waters Corp., Milford, MA, USA), kept at a temperature of 35°C. The mobile phase was composed of aqueous formic acid (0.1% v/v) (A) and acetonitrile containing formic acid (0.1% v/v) (B). The initial mobile phase consisted of a 2‐min isocratic period at 0.1% B, followed by a linear increase to 1.5% B at 10.5 min and to 95% B at 11 min. This was followed by a 4‐min isocratic period at 95% B. The mobile phase conditions were returned to 0.1% B at 15 min and allowed to re‐equilibrate for 4 min. The flow rate of the eluent was 0.4 mL/min, and the injection volume was 5 µL. Data were acquired using the LCMS Workstation Data Acquisition Console and processed using Qualitative Analysis 10.0 (Agilent Mass Hunter).

2.5.4. Untargeted Metabolomic Analysis

Extracts were prepared as described in Section 2.5.2 but with chlorpropamide (1 µM) added as an internal standard, and samples were filtered through 0.2‐µm nylon syringe filters before analysis. Untargeted metabolomic analysis was carried out using a Vanquish UHPLC chromatographic system coupled to an Exploris 120 mass spectrometer (Thermo Fisher Scientific, Dreieich, Germany) equipped with an H‐ESI ion source. Chromatographic separation was achieved using an Acquity CSH C18 column (2.1 × 100 mm, 1.7 µm, Waters Corp) at a temperature of 40°C. The mobile phase consisted of (A) 0.1% aqueous formic acid and (B) acetonitrile containing 0.1% formic acid. The mobile phase was an initial isocratic period at 3% B for 5 min. The flow rate was 0.25 mL/min. The sample injection was 2 µL. MS data were acquired in both the negative and positive ionization modes at a resolution of 120,000 over a scan range of 80–1000 Da. During the analysis, ion spray voltage of 4000 V for positive mode and −2600 V for negative mode, sheath gas of 241.32 kPa, aux gas of 68.94 kPa, sweep gas of 6.89 kPa, vaporizer temperature of 150°C, and ion transfer tube temperature of 3258°C were used. The raw MS data were imported into the ProteoWizard‐ms convert (http://proteowizard.sourceforge.net/) for data file format conversion (Adusumilli and Mallick 2017). Subsequently, MS DIAL (.NET Framework 4.0 or later; RAM: 4.0 GB or more) was used for preprocessing the data, which included peak alignment, peak picking, and compound identification. Normalization was conducted using Normalization by a Pooled Sample from the Group (PNQ) and the Pareto method in Metaboanalyst 5.0 (https://www.metaboanalyst.ca) (Chong et al. 2019). To assess the effects of the different treatments and visualize group clusters, partial least squares‐discriminant analysis (PLS‐DA) was conducted. Variable importance in projection (VIP) scores and loading plots were used to identify the variables contributing to the separation among the groups.

2.6. Nutrient Composition

2.6.1. Water‐Soluble Vitamins

Water‐soluble vitamins were extracted according to the method described in Mather et al. (2024). Vitamins B₁, B₂, B₃, B₉, and C were quantified as thiamine, riboflavin, nicotinic acid, folate, and l‐ascorbic acid, respectively, by UPLC–MS using the method of El‐Hawiet et al. (2022), with some modifications. The analysis was performed using a Agilent 1290 Infinity UPLC system (Agilent Technologies) interfaced with an Agilent 6460 Triple Quad MS (Agilent Technologies), which was equipped with an ESI ion source. Chromatographic separation was achieved using a Waters Acquity UPLC HSS C18 column (2.1 mm × 100 mm, 1.8 µm). The mobile phase consisted of (A) 0.1% aqueous formic acid and (B) acetonitrile containing 0.1% formic acid. The elution sequence began with a 1‐min isocratic period at 10% B, followed by linear increases to 50% B at 6 min. The mobile phase was returned to 10% B over 2 min and then re‐equilibrated for 2 min. The flow rate of the eluent was 0.4 mL/min. The injection volume was 5 µL. Vitamins B₁, B₂, and B₃ were monitored in the positive ion mode, whereas vitamins B₉ and C were monitored in the negative ion mode. The MS conditions were optimized as follows: capillary voltage, 3500 V; nozzle voltage, 1000 V; desolvation gas (nitrogen) temperature, 350°C at 10 L/min; and sheath gas temperature (nitrogen), 300°C at 9 L/h. Analytes were detected via multiple reaction monitoring mode to achieve the best specificity and signal‐to‐noise ratio for each of the vitamins (Table S2). Data were acquired using the LCMS Workstation Data Acquisition Console and processed using Qualitative Analysis 10.0 (Agilent Mass Hunter).

2.6.2. Crude and Digestible Protein

The crude protein content of microgreen powders was measured based on the Dumas method, using a LECO FP 828p Nitrogen analyzer (St. Joseph, MI, USA) and calculated by multiplying the total nitrogen content by a conversion factor of 6.25 (Fujihara et al. 2001). The digestible protein content was determined using the pH drop method (Hsu et al. 1977). In brief, microgreen samples were suspended in water (6.25 mg/mL), and the pH was adjusted to 8.0. Samples were then combined with an equal volume of protease cocktail (trypsin [1.6 mg/mL], chymotrypsin [3.1 mg/mL], and pepsin [1.3 mg/mL], pH 8.0), incubated at 37°C for 10 min, and measured for final pH. In vitro protein digestibility (%) was calculated using Equation (9):

$$\mathrm{In}\mathrm{vitro}\mathrm{digestibility}\left(\%\right)=210.46-18.10\times \left(\mathrm{pH}\mathrm{at}10\min \right),$$ (9)

where the y‐intercept (210.46) and slope (18.10) were derived by Hsu et al. (1977).

2.6.3. Minerals

The relative abundance of macrominerals (P, S, N, K, and Ca) and microminerals (Mn, Fe, Cu, and Zn) in dried microgreens was measured using an Epsilon 1 X‐Ray Fluorescence (XRF) spectrometer (Malvern Panalytical, Malvern, UK) according to the manufacturer's instructions. Results are expressed as percentages.

2.7. In Vitro Bioaccessibility

The bioaccessibility of TPC, MAC, GE, and water‐soluble vitamins in dried radish microgreens was determined using an established in vitro protocol (Cilla et al. 2022; De la Fuente et al. 2019; Minekus et al. 2014). Due to the low starch content of the samples, the oral phase was omitted, and only the gastric and small intestine phases were used. For the gastric phase, 0.15 g of powdered radish microgreens was mixed with 5 mL of ultrapure water and combined with 4.55 mL of simulated gastric fluid containing 7.8 mg/mL pepsin and 0.16 mM CaCl₂. The gastric phase pH was then adjusted to 3 using 1 M HCl, diluted to 10 mL with water, and incubated in a shaking bath at 37°C for 2 h. Subsequently, to simulate the intestinal fluid, 8 mg/mL pancreatic solution, 65 mg/mL bile salt solution, and 0.65 mM CaCl₂ were added to the gastric phase. The intestinal phase was adjusted to pH 7 using 1 M NaOH, diluted to 20 mL with water, and incubated in a shaking bath at 37°C for 2 h. After incubation, the digested samples were cooled on ice and centrifuged at 3100 × g for 80 min at 4°C. The supernatant was considered the bioaccessible fraction and stored at −80°C. Prior to quantification of the analytes of interest as described above, the supernatant was filtered through 0.2‐µm nylon filters.

2.8. Statistical Analysis

Drying experiments were repeated three times. For the fitting of drying curves, the constants of each model equation were evaluated through nonlinear regression analysis using Minitab Statistical Software Version 22.1. All nutritional and phytochemical analyses were performed for each of the three drying replicates, and the results are presented as mean ± standard deviation (SD). Statistical analysis was conducted with R Studio (version 2023.06.0+421). For TPC, MAC, GE, vitamins, protein, and mineral content analyses, a one‐way analysis of variance (ANOVA) with a Dunnett's post hoc test was used to determine whether the hot air‐dried samples were different from the freeze‐dried samples. Differences were considered significant at a p‐value of <0.05. Metabolomics data were analyzed by PLS‐DA using Metaboanalyst 5.0 (Alberta, Canada) to identify metabolites that discriminated samples based on drying protocol.

3. Results and Discussion

3.1. Drying Process

The drying time to reach moisture levels below 10% MC, which is considered microbiologically safe for vegetables (Afolabi 2014), was 438, 212, and 74 min at 45°C, 65°C, and 95°C, respectively (Figure S1A). These results are similar to what has been reported previously for acerola, radish, and broccoli sprouts (Galieni et al. 2020; Li et al. 2020; Nóbrega et al. 2015). The shorter drying times required to reach the target MC at higher drying temperatures can be explained by the principles of heat transfer. As drying temperature increases, more thermal energy reaches the sample, which accelerates the diffusion of water molecules from the interior to the surface and the subsequent escape to the surrounding air (Kosasih et al. 2020). The MC of radish microgreens dried for chemical analysis ranged from 3.4% to 8.3% with significant differences between the freeze‐dried and the hot air‐dried samples (Table S3). The difference in MC observed between the freeze‐dried and hot air‐dried samples could be due to the protocol used for freeze‐drying, which may have left some residual moisture in the sample (Afolabi 2014; Ratti 2001). All drying conditions resulted in water activity values below 0.6 (Table S3), which is reported to inhibit microbial growth (Karuppuchamy et al. 2024; Chitrakar et al. 2019).

3.2. Color

The fresh radish microgreens used in this study were bright green with intense pink‐purple coloration (Figure S2A). Since color is an important quality factor for consumer acceptability, color analysis was performed on dried microgreens. Hot air drying led to a darker color compared to the control, with samples showing reddish‐brown hues (Figure S2B). The L* values of all hot air‐dried samples were reduced compared to the freeze‐dried sample but were not significantly different from one another (Table 1). These results are similar to a previous study showing that hot air drying led to a 30% decrease in L* in broccoli sprouts compared to freeze‐drying (Dziki et al. 2020). In the present study, hot air drying at 95°C, but not 45°C or 65°C, resulted in a significant increase in a* compared to freeze‐drying, whereas drying at both 65°C and 95°C led to significant decreases in b* compared to freeze‐drying (Table 1). The decrease in the brightness and increase in the red/brown color in hot air‐dried samples may be attributed to both enzymatic and nonenzymatic browning. Particularly, the changes observed at 45°C and 65°C are likely due to polyphenol oxidase (PPO)‐ and peroxidase (POD)‐catalyzed browning, which is enhanced at mild and moderate temperatures (Loh et al. 2018). In contrast, the significant change in red coloration at 95°C is more likely due to nonenzymatic Maillard reactions, as enzymatic activity declines at high temperatures due to protein denaturation (Soto‐Hernández et al. 2017).

TABLE 1.

Color parameters of dried radish microgreens.

Drying	L*	a*	b*	∆E
−50°C	46.1 ± 1.2	−1.5 ± 0.9	11.8 ± 1.1	—
45°C	32.1 ± 0.6*	0.4 ± 0.1	10.7 ± 0.6	14.71
65°C	32.7 ± 1.1*	0.3 ± 0.1	9.4 ± 0.2*	13.71
95°C	32.3 ± 1.1*	3.0 ± 1.4*	9.8 ± 0.3*	14.02

Note: Data represent mean ± standard deviation (n = 3).

***p < 0.001, **p < 0.01, and *p < 0.05 by one‐way ANOVA with Dunnett's post hoc test compared to control (−50°C).

3.3. Phytochemical Composition and in Vitro Bioaccessibility

3.3.1. TPC

TPC in radish microgreens dried at 45°C and 65°C were 10% and 21% lower, respectively, than the freeze‐dried samples. This reduction likely reflects oxidative degradation of phenolic compounds by PPO and POD, which have been reported to be active at temperatures below 70°C (Loh et al. 2018). Samples dried at 95°C were not significantly different from the freeze‐dried control in terms of TPC (Table 2). This may be due to rapid inactivation of PPO and POD at higher temperatures (>80°C). Additionally, the shorter drying time at 95°C may have contributed to lower oxidation of phenolic compounds (Soto‐Hernández et al. 2017).

TABLE 2.

Phytochemical content and bioaccessible fraction of dried radish microgreens.

Drying temperature	Total	Bioaccessible	Bioaccessibility (%)
	mg/100 g DW
Total phenolic content (TPC)
−50°C	2303.2 ± 159.3	1343.8 ± 229	58.3
45°C	2101.3 ± 71.3	1296.9 ± 384	61.7
65°C	1810.0 ± 100.5**	1186.4 ± 286	65.5
95°C	2068.6 ±77.7*	1461.7 ± 178	68.6
Monomeric anthocyanin content (MAC)
−50°C	219.6 ± 6.4	2.17 ± 0.5	< 1%
45°C	111.1 ± 7.6***	1.83 ± 0.6	<1%
65°C	111.3± 12.3***	1.93 ± 0.5	<1 %
95°C	114.8 ± 8.8***	1.87 ± 0.7	<1 %
Glucoraphenin (GE)
−50°C	707.1± 111.9	ND	ND
45°C	748.6 ± 61.9	ND	ND
65°C	703.2 ± 84.2	ND	ND
95°C	562.4 ± 153.2**	ND	ND
Thiamine (B1)
−50°C	1.39 ± 0.3	0.81 ± 0.1	58.3
45°C	0.68 ± 0.6	0.46 ± 0.06	67.6
65°C	1.29 ± 0.7	0.62 ± 0.1	48.1
95°C	1.35 ± 0.4	0.81 ± 0.2	60.0
Riboflavin (B2)
−50°C	0.83 ± 0.11	0.05 ± 0.02	6.02
45°C	0.67± 0.12	0.04 ± 0.01	5.97
65°C	0.52 ± 0.07*	0.08 ± 0.03	15.4*
95°C	0.54 ± 0.08*	0.06 ± 0.04	11.1
Niacin (B3)
−50°C	1.25 ± 0.08	0.19 ± 0.01	15.2
45°C	0.75 ± 0.19*	0.17 ± 0.03	22.6
65°C	0.81 ± 0.2*	0.19 ± 0.04	24.0
95°C	0.72 ±0.16*	0.19 ± 0.03	26.3
Folate (B9)
−50°C	1.36 ± 0.34	0.18 ± 0.19	13.2
45°C	1.21 ± 0.46	0.20 ± 0.08	18.1
65°C	1.46 ± 0.41	0.42 ± 0.05	23.5
95°C	0.89 ± 0.25	0.34 ± 0.03	38.2
Ascorbic acid/ascorbate (C)
−50°C	18.9 ± 8.06	2.38 ± 0.26	13.8
45°C	8.02 ± 2.44*	2.13 ± 1.28	30.7
65°C	7.29 ± 1.79*	2.45 ± 0.83	28.9
95°C	6.22 ± 0.54*	2.44 ± 0.54	39.8

Note: Data represent mean ± standard deviation (n = 3).

Abbreviation: ND, not detected.

***p < 0.001, **p < 0.01, and *p < 0.05 by one‐way ANOVA with Dunnett's post hoc test compared to control (−50°C).

Prior work on the effect of drying temperature and time on TPC has yielded mixed results, which appear to depend on the material being studied. For instance, one study on mature‐stage radishes reported no significant differences in TPC between hot air drying (50–80°C) and freeze‐drying (Li et al. 2020), whereas another study found that an increase in the drying temperatures up to 80°C caused a significant decrease in TPC, while little or nonsignificant effect was observed when comparing drying at 40°C and 60°C to fresh and freeze‐dried broccoli sprouts (Dziki et al. 2020). Zang (2015) observed that broccoli dried at 70°C retained higher TPC than broccoli dried at 50°C. The authors suggested that the longer drying time required at 50°C allowed greater enzyme‐mediated oxidation of phenolic compounds to occur. Conversely, Yap et al. (2022) reported that drying chilies at temperatures above 60°C for shorter times increases TPC compared to longer drying times at lower temperatures. The authors explained that this difference was due to thermal breakdown of cell walls in the chilies, leading to the release of bound phenolics, which could then be extracted and measured, while at lower drying temperatures, PPO and POD caused degradation of phenolics over time. Gan et al. (2017) reported similar findings with mung bean sprouts, but attributed increases in TPC at higher temperatures to the reaction of Maillard reaction products with the Folin–Ciocalteu reagent rather than to an increase in phenolic content. In the present study, the in vitro bioaccessibility of TPC of dried radish microgreens ranged from 58% to 68% (Table 2), which is consistent with previous studies on cruciferous microgreens (De la Fuente et al. 2019).

3.3.2. MAC

The MAC of freeze‐dried radish microgreens observed in this study was consistent with the previously reported values for Red Rambo radish sprouts (Baenas et al. 2015). Hot air‐dried samples exhibited a 47%–49% reduction in MAC compared to the freeze‐dried control samples. This decrease may be due to thermal degradation. Heat can promote the degradation of anthocyanins through several mechanisms. One is the hydrolysis of glycosidic bonds, which results in the formation of less stable aglycones (Garba et al. 2015). Additionally, thermal treatment can lead to the opening of the pyrylium ring and chalcone formation, precipitating the degradation of anthocyanins (Méndez‐Lagunas et al. 2017). Finally, anthocyanins can polymerize with PPO‐produced quinones, leading to pigmentation loss (Nunes et al. 2005). The bioaccessible fraction of MAC after in vitro digestion was less than 1% for all samples (Table 2). Based on previous work by De la Fuente et al. (2019), who were unable to detect MAC after in vitro gastrointestinal digestion of cruciferous microgreens, it is speculated that these compounds are oxidized or otherwise degraded during digestion protocol and that these results do not reflect the actual bioaccessibility of MAC from these samples but are rather an artifact of the method.

3.3.3. GE Content

GE and glucoraphasatin are the main glucosinolates present in radishes (Demir et al. 2023). Due to the lack of a commercially available glucoraphasatin standard, only GE content was determined in this study. GE concentration in microgreens dried at 95°C was 20% lower than that in freeze‐dried microgreens, whereas the concentration in those dried at 45°C or 65°C was not significantly different from the freeze‐dried samples (Table 2). Our findings align with the results of previous drying studies in mature cruciferous vegetables. Vargas et al. (2022) observed a 17%–23% reduction in glucosinolate levels in broccoli and kale after drying at 70°C. Similarly, Luo et al. (2022) reported total glucosinolate losses of 27%–51% in broccoli, cauliflower, cabbage, and radish after oven‐drying at 60°C. The authors attributed the decrease in glucosinolate concentration to the thermal degradation of plant tissue, which facilitated the interaction between glucosinolates and myrosinase, leading to the formation of isothiocyanates, thiocyanates, and nitriles (Oloyede et al. 2021; Wang et al. 2019). As with MAC, GE concentrations in the bioaccessible fraction following in vitro digestion were below the level of quantification for all treatments. This reduction is attributed to the alkaline conditions and enzymatic activities in the small intestine phase, which promoted the conversion of glucosinolates into their breakdown products (Abellán et al. 2021).

3.3.4. Untargeted Metabolomic Analysis

Untargeted metabolomic analysis in conjunction with PLS‐DA was employed to identify overall changes in the chemistry of radish microgreens subjected to drying treatments. A total of 8155 features were identified in the negative ion mode, and 5259 features were identified in the positive ion mode. The PLS‐DA score plots showed that the first two dimensions accounted for 51.8% and 57% of the total variance in the negative and positive ion mode data, respectively (Figure 1A,D). Drying temperature was the primary driver of separation in component 1, whereas differences among sample replicates accounted for variation along component 2. VIP analysis was used to determine which features drove separation between the treatment groups, and the relative abundance of the 20 most discriminant metabolites (VIP score >1) was determined for negative (Figure 1B,C) and positive ion (Figure 1E,F) modes. Only eight of the top 20 metabolites in the negative ion mode and five in the positive ion mode were tentatively identified based on their m/z values and major fragment ions (Table S4). Although the identities of the top three metabolites driving separation in the negative ion mode could not be determined, the similarity in their m/z values and fragment ions suggests that they share similar chemical structures. The tentatively identified metabolites driving the separation between the control and the hot air‐dried samples were primarily anthocyanins, glucosinolates, organic acids, and amino acids. These metabolite‐based separation patterns are consistent with the differences observed in TPC, MAC, and GE, particularly between the control and 45°C and 65°C treatments. Heatmaps presenting the relative abundance of each feature in the VIP score plots across the treatment groups show that higher temperatures appear to decrease the abundance of some metabolites while increasing that of others.

FIGURE 1.

Multivariate analysis of untargeted LC–MS‐based metabolomics data for dried radish microgreens. Partial least squares‐discriminant analysis (PLS‐D) score plots of data collected in (A) negative or (D) positive ion modes. The first 20 important MS features driving the separation of treatments in components 1 and 2 using negative (B, C) or positive ion mode (E, F); n = 3 replicates for treatment groups. Ellipses in the score plots represent 95% confidence intervals.

Using m/z, fragmentation data, and existing literature on the phytochemistry of radishes, 72 compounds were tentatively identified (Figure 2). Among these, 33% were phenolic compounds, 17.6% glucosinolates, and 12.5% vitamins. Clear differences in the relative abundance of the tentatively identified compounds were observed across the drying temperatures (Figure 2). Specifically, freeze‐dried samples contained higher relative abundance of glucosinalbin, glucoerucin, GE, neoglucobrassicin, and glucobrassicin. In contrast, it was observed that samples dried at 95°C exhibited a higher relative abundance of flavonoids. The higher relative concentrations could be attributed to the rapid inactivation of PPO at this temperature. The optimal temperature range for PPO activity has been reported to be 25°C–55°C (Huang et al. 2023), where the rate of enzymatic reaction is accelerated. This may explain the lower flavonoid abundance observed in samples dried at 45°C and 65°C. Additional studies using authentic standards are necessary for a more accurate comparison of the overall metabolome across drying treatments.

FIGURE 2.

Relative abundance of 72 tentatively identified features in dried radish microgreens; n = 3 replicates for treatment groups. The drying temperatures are listed from left to right as follows: −50°C (red); 45°C (green); 65°C (blue), and 95°C (sky blue). The heatmap was built using Metaboanalyst 5.0 software.

3.4. Nutrient Composition

3.4.1. Water‐Soluble Vitamins

The impact of drying on the concentration of a panel of water‐soluble vitamins was determined, using freeze‐dried samples as a control. Measured thiamine content did not differ across drying treatments (Table 2) but was approximately two to four times higher than previously reported for radish sprouts and microgreens (Rani and Singh 2021; Zieliński et al. 2005). The concentrations of riboflavin and niacin in the freeze‐dried samples in the present study were similar to those previously reported by others for radish microgreens (Rani et al. 2024; Rani and Singh 2021; Zieliński et al. 2005). Riboflavin and niacin concentrations were 37% and 35%–40% lower in samples dried at 65°C and 95°C, respectively, compared to freeze‐dried samples. No significant differences in folate content were observed between freeze‐dried and hot air‐dried radish microgreens (Table 2). A previous study found that the application of high temperatures ranging between 70°C and 95°C caused reductions of 16%–41% in thiamin, riboflavin, and niacin in black mulberry fruits (Sernikli and Kadakal 2020). Data regarding the relationship between hot air‐drying temperatures and water‐soluble vitamins’ retention in radish microgreens, however, have not been reported. The vitamin C levels found in this study were lower than those previously reported for freeze‐dried radish (76% lower) or fresh radish microgreens (58% lower) (Rani et al. 2024; Rani and Singh 2021; De la Fuente et al. 2019). This may be the result of differences in growing conditions or the variety of radishes used in the different studies. Vargas et al. (2022) reported that hot air drying resulted in greater retention (>20%) of some B vitamins and vitamin C in mature broccoli and spinach compared to freeze‐drying. However, in the present study, hot air drying reduced vitamin C by 57%–67% compared to the freeze‐dried controls. This difference may reflect varying compositions between mature vegetables and microgreens, where microgreens may be more susceptible to heat and oxidative degradation. Consistent with our findings, Bhatt et al. (2023) reported vitamin C losses of 47%–69% in tray‐dried broccoli, safflower, amaranth, fenugreek, mustard, and spinach microgreens at temperatures oscillating between 50°C and 52°C. Heat, oxygen, and light can accelerate the oxidative degradation of ascorbic acid to dehydroascorbic acid and other nonnutritional products. Despite the significant reduction in the concentration of some of the water‐soluble vitamins due to hot air drying, no significant differences were observed between drying treatments regarding the bioaccessible fractions of thiamine, niacin, folate, and ascorbic acid (Table 2). The in vitro bioaccessibility of riboflavin increased in the microgreens dried at 65°C. Overall, bioaccessibility values were low, which may be attributed to the instability of vitamins under varying pH and temperature conditions during in vitro digestion.

3.4.2. Crude and Digestible Protein Content

The total crude protein content was assessed based on nitrogen content, and no significant differences among the treatment groups were observed (Table 3). The protein content measured here was consistent with previous studies on radish microgreens (Poudel et al. 2024; Ghoora et al. 2020). Similarly, no statistically significant differences were observed in the in vitro digestible fraction or overall protein digestibility, which was greater than 79.6% across the treatment groups (Table 3). To our knowledge, this is the first report of the relative protein digestibility for dried radish microgreens. However, the pH‐drop method used in this study may not be sensitive enough to detect changes in protein conformation or amino acid availability. A complete amino acid profile analysis could provide deeper insights into the nutritional quality and biological value of the final product.

TABLE 3.

Crude and digestible protein in dried radish microgreens.

Drying temperature	Crude protein	Digestible fraction	Digestibility
	g/100 g DW		(%)
−50°C	36.4 ± 8.34	30.8 ± 6.55	84.7 ± 1.94
45°C	34.3 ± 7.35	27.6 ± 5.97	80.3 ±1.10
65°C	36.8 ± 3.21	29.2 ±1.62	79.6 ± 2.70
95°C	36.8 ± 2.51	29.4 ± 1.98	80.0 ± 1.13

Note: Data represent mean ± standard deviation (n = 3).

***p < 0.001, **p < 0.01, and *p < 0.05 by one‐way ANOVA with Dunnett's post hoc test compared to control (−50°C).

3.4.3. Minerals

Potassium and calcium were the most abundant macroelements (Table 4), which is consistent with previous findings (Di Gioia et al. 2023). Compared to freeze‐dried samples, the samples dried at 65°C and 95°C showed reduced potassium and sulfur abundance but increased calcium levels. Reema (2023) noted similar increases in calcium in tray‐dried safflower, amaranth, fenugreek, mustard, and spinach microgreens at 52°C. Among the microminerals analyzed, all except copper exhibited increased levels following hot air drying at some or all the tested temperatures. These changes were relatively small (<5%) and may therefore not be nutritionally meaningful.

TABLE 4.

Mineral relative abundance in dried radish microgreens.

	P	S	N	K	Ca	Mn	Fe	Cu	Zn
Drying temperature	%
−50°C	5.52 ± 0.13	13.10 ± 0.26	5.34 ± 1.16	43.2 ± 1.31	25.9 ± 0.89	0.09 ± 0.01	0.16 ± 0.03	0.01 ± 0.0	0.03 ± 0.0
45°C	4.01 ± 0.96*	9.52 ± 2.01**	5.28 ± 1.13	44.6 ± 3.49**	35.9 ± 4.01**	0.17 ± 0.04	0.35 ± 0.11*	0.02 ± 0.01	0.11 ± 0.02***
65°C	5.00 ± 0.08	9.96 ± 0.26*	5.58 ± 0.51	40.7 ± 1.23*	33.7 ± 2.73*	0.20 ± 0.01*	0.38 ± 0.05**	0.02 ± 0.0	0.12 ± 0.02***
95°C	4.51 ± 0.63	8.83 ± 0.37**	5.68 ± 0.37	41.6 ± 3.51**	35.6 ± 0.98	0.23 ± 0.08*	0.27 ± 0.01	0.03 ± 0.01	0.11 ± 0.01***

Note: Data represent mean ± standard deviation (n = 3).

***p < 0.001, **p < 0.01, and *p < 0.05 by one‐way ANOVA with Dunnett's post hoc test compared to control (−50°C).

4. Conclusion

The effects of hot air drying on the phytochemical profile of radish microgreens were compared to freeze‐drying. Increasing the drying temperature to 95°C induced significant changes in the phytochemical profile of the microgreens, but the resultant product retained 33%–65% of the water‐soluble vitamin content, 100% of the protein content, 70%–80% of the macromineral content, and 50%–80% of the phenolic and GE content found in the freeze‐dried samples. Future research is needed to evaluate the in vivo bioavailability of the nutrients and nonnutrient phytochemicals in dried radish microgreens, to explore the impact of dried radish microgreen powder on the sensory characteristics of formulated foods in order to establish achievable incorporation rates, and to determine how the results of these laboratory‐scale drying experiments translate to pilot‐ or production‐scale applications.

Author Contributions

Marjorie J. Jauregui: investigation, formal analysis, writing–original draft. Ezekiel R. Warren: methodology, writing–review and editing. Francesco Di Gioia: resources, writing–review and editing. Misha T. Kwasniewski: methodology, formal analysis, writing–review and editing, data curation. Joshua D. Lambert: conceptualization, formal analysis, supervision, funding acquisition, project administration, resources, writing–review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Fig.1: jfds70426‐sup‐0001‐figureS1.docx

Supporting Fig. 2: jfds70426‐sup‐0002‐figureS2.docx

Supporting Table 1: jfds70426‐sup‐0003‐tableS1.docx

Supporting Table 2: jfds70426‐sup‐0004‐tableS2.docx

Supporting Table 3: jfds70426‐sup‐0005‐tableS3.docx

Supporting Table 4: jfds70426‐sup‐0006‐tableS4.docx