The blind spot: a 28-fold miss on the dominant compound
Duan et al. (2026), in Food Chemistry: X (doi:10.1016/j.fochx.2026.104410), set out to make selenium speciation cheap enough for routine use — replacing ICP-MS with atomic fluorescence spectrometry behind orthogonal strong-anion/strong-cation-exchange chromatography — and ended up demonstrating that the standard sample prep for Cardamine hupingshanensis was silently discarding the plant’s main selenium compound.
In leaves of hydroponically grown C. hupingshanensis (80 mg Se L⁻¹ as selenite, 14 days), 72.06 ± 0.32% of total Se is water-soluble, and the water-soluble pool is dominated by free selenoamino acids rather than protein. Yet the conventional one-step protease digestion recovered almost none of the characteristic storage compound selenolanthionine (SeLan): measured SeLan rose from 5.06 ± 0.28 to 143.74 ± 2.70 µg Se g⁻¹ (a 28.41-fold increase) when a water-first step was inserted ahead of enzymatic hydrolysis, with SeLan found almost exclusively in the aqueous extract. Even the buffer choice matters at this level — extracting in Tris-HCl (pH 4.5) instead of water collapsed SeLan from 157.30 ± 8.14 to 42.72 ± 3.74 µg Se g⁻¹. Total speciated recovery in the hyperaccumulator climbed from 30.20 ± 2.55% (one-step) to 70.97 ± 0.22% (sequential) and 80.65 ± 8.51% once the whole procedure was moved into a nitrogen glove box; for wheat grain the aerobic-to-anaerobic gain was 80.79 ± 2.80% → 90.02 ± 5.93%, with SeMet rising 98.73 ± 4.50 → 112.17 ± 7.87 µg Se g⁻¹ (13.61%).
Why the pool hides
The mechanism is metabolic, and it is the defining one for Se hyperaccumulators. Non-hyperaccumulators like wheat assimilate Se into SeMet and SeCys that are non-specifically stitched into protein, so protease digestion is the right key. Hyperaccumulators avoid that toxic dead-end: they store Se as free, non-protein selenoamino acids — and a protocol built to liberate protein-bound Se simply passes them by, while the enzyme incubation’s shifting pH and ionic strength degrade the labile fraction. Detection adds a second filter: on the SAX columns used in most published Se speciation, SeLan, SeCys2 and selenomethionine oxide co-elute at one retention time (the paper’s Fig. 2A), so an analyst running SAX alone cannot even see SeLan as a distinct peak — and the authors note that SeCys2 values reported by SAX-only methods may have been overestimated through exactly this co-elution.
The corrected picture of C. hupingshanensis is strikingly different from the SAX-era literature. The optimized method’s final profile (Table 1): SeLan 47%, MeSeCys 23%, γ-GluMeSeCys 3%, SeO3²⁻ 7%, SeCys2 not detected, SeMet not detected (81% recovery). Compare Yuan et al. 2013, the species’ canonical speciation reference: SAX, protease-only — SeCys2 reported at 78% of total Se, with no SeLan data at all. Both et al. 2018, using cation-exchange separation, had already found SeLan at ~40% and no SeCys2. Same plant, two chromatographies, two different chemistries — and the new work shows the discrepancy is not academic: it is a preparation artifact you can dial up or down 28-fold with a buffer swap.
Why it matters for Se agromining and Se products
This is not a niche analytical quibble, because C. hupingshanensis is not a niche plant. It is the flagship Chinese Se hyperaccumulator — our database carries a verified maximum of 1,965 µg Se g⁻¹ from its type locality near Enshi, Hubei — and it is already commercialized: sold into Se-enriched foods and investigated as a feed additive for Se-enriched eggs and meat. For every such product, the chemical form of Se is what governs bioavailability and safety assessment, not the total on the label. A QA lab in a producing region that runs the textbook protease-plus-SAX workflow on this species would certify a profile in which the actual dominant compound does not exist and a minor oxidation product is inflated into the lead species — and would do so with 96% apparent recovery, looking rigorous the whole time. The paper’s real contribution to the Se economy is the instrument, not just the prep: AFS detection makes the full orthogonal method affordable for exactly the mid-scale food and feed labs that sit downstream of Se biofortification and Se phytomining operations.
There is also a genuine physiological result buried in the methods work: γ-glutamyl-MeSeCys, never before reported in this species, now shows up as a real metabolite — and the same compound pairing (MeSeCys 53%, γ-GluMeSeCys 47%) fills the leaf trichomes of Stanleya pinnata (Freeman et al. 2006), suggesting a conserved γ-glutamylation step on the MeSeCys branch of hyperaccumulator Se metabolism that bears directly on how stored Se is transported, sequestered, and volatilized.
The honest limits
Bound the claims. The plants were 14-day hydroponic seedlings at a single 80 mg Se L⁻¹ selenite dose, from one commercial Enshi nursery source; all experiments n = 3. Recovery of a species during extraction is evidence about extraction, not proof of the in-vivo pool — the glove-box choreography that gets recovery to 80–90% is far from routine, and the authors themselves concede the anaerobic mechanism for SeLan loss is unresolved. The paper’s own housekeeping needs care: its abstract claims recoveries improved “from 20.82%” for C. hupingshanensis, but 20.82% appears nowhere in the body (whose worst figure is 23.26 ± 3.63%); one sentence in §3.3 says the authors “did not detect significant levels of SeLan” in their early extracts, which flatly contradicts their headline result and can only be a typo for SeCys2 (read it that way); and the “8,000 mg Se kg⁻¹” maximum quoted in the introduction is a literature figure the authors repeat, not a measurement — our species record stays at the verified 1,965 µg g⁻¹. None of this touches the core finding, which is measured, replicated, and internally consistent across four independent extraction experiments.
Bottom line
For Se — an element where commerce, nutrition and remediation all ride on chemical form — speciation is only as good as the extraction. This paper shows the field’s default protocol for the most commercialized Se hyperaccumulator was blind to its dominant compound and nearsighted about its minor ones, provides the cheap orthogonal instrumentation to fix it, and adds γ-GluMeSeCys to the metabolic map of C. hupingshanensis, tying it to the same storage chemistry documented in S. pinnata. Anyone quoting a Se species profile for any hyperaccumulator should now ask one question first: what did the prep throw away? See the selenium hub for context.
Provenance: based on the open-access full text (CC BY-NC-ND, Food Chemistry: X 39:104410, published 3 September 2026), read in full via Europe PMC (PMC13572759, PMID 42741486); nearly every load-bearing number is body-only and absent from the abstract. Every load-bearing number was verified verbatim against the primary source by three independent reviewers (Claude, GPT, Kimi — each self-retrieving the full text), with zero cross-vendor discrepancies; the source’s own internal inconsistencies (the abstract-only 20.82% baseline, the §3.3 “SeLan” typo, and the literature-cited 8,000 mg kg⁻¹ maximum) are disclosed above. See /methodology/.