Does high-temperature-treated soy protein alter amino acid availability and social behavior in mice?

Modeling moist-heat sterilization of SPI at 121°C

The study targets a common food-processing step: moist-heat sterilization. The authors suspended soy protein isolate (SPI) in water at 10% concentration and heated it at 121°C for 20 minutes, mimicking the retort conditions used for canned, bottled, and retort-pouch foods. They note that dry heating is already known to damage protein nutritional value, whereas the effects of heating protein in the presence of water are less understood, which is the gap this work addresses.

The comparison baseline is untreated SPI. The paper first compares protein content, amino acid composition, reducing sugars, and ammonia between the two preparations, then simulates gastrointestinal digestion with proteases in vitro, and finally feeds each SPI as the sole protein source to mice to track serum amino acids and behavior. The key to reading the paper is keeping three measurement levels apart: chemical changes in the SPI itself, the products left after in vitro digestion, and the physiological and behavioral responses of living mice.

Nitrogen barely falls while amino acids drop nearly a tenth

Kjeldahl analysis measured crude protein (nitrogen content × 6.14) falling from 737.1 mg/g to 724.6 mg/g, a drop of only about 1.7%. Amino acid analysis, however, showed protein content falling from 701.90 mg/g to 634.43 mg/g, about 9.6%. The two numbers are not contradictory: the Kjeldahl method counts nitrogen atoms, while amino acid analysis counts recoverable amino acid residues. The gap indicates that most residues were chemically modified or destroyed rather than lost as volatile nitrogen.

Except for Met and Trp, amino acid levels in both HCl and NaOH hydrolysates of HT-treated SPI were significantly lower than in untreated SPI. The authors checked several possible routes. Glucose and galactose derived from contaminating oligosaccharides such as raffinose and stachyose fell significantly, but the total monosaccharide reduction was about 10 µmol/g, far below the roughly 1 mmol/g loss of amino acids. Free ammonia rose by about 20 µmol/g, less than one-fiftieth of the amino acid loss. Maillard reactions and deamidation therefore explain only a small share, and the authors propose that most residues were converted into non-volatile nitrogen compounds such as imine-type Schiff bases.

Digestion leaves insoluble protein and larger peptides behind

After in vitro digestion with pepsin and pancreatin, the untreated SPI digest was completely soluble, whereas the HT-treated SPI digest contained 53.0 ± 9.9 mg of insoluble material per gram of SPI, and amino acid analysis confirmed that this material contained protein. Heating therefore generated insoluble proteins resistant to endoproteinases.

The soluble fraction was then treated with carboxypeptidase A and leucine aminopeptidase and analyzed by size exclusion chromatography. HT treatment increased the proportion of peptides in the 200–500 Da range and decreased peptides below 200 Da. In other words, even after exopeptidase treatment, a population of larger peptides remained incompletely degraded. The authors attribute this to modified amino acid residues forming digestion-resistant peptides, which is the material basis for the reduced amino acid availability described later.

Serum amino acids diverge only at 120 minutes

In the single-dose gavage experiment, total serum free amino acids did not differ significantly between groups at 30 minutes; by 120 minutes, the HT-treated group was significantly lower than the untreated group. For individual amino acids, Thr, Ala, and Met were significantly lower at 60 minutes, and Trp was significantly lower at 120 minutes. This timing suggests the difference is not a simple shift in absorption speed but reflects the total amount and form of amino acids released by digestion.

Notably, neither untreated nor HT-treated SPI raised total serum amino acids above the fasted baseline. The authors also measured 3-methylhistidine as an indicator of muscle protein breakdown; the two groups did not differ significantly, with only a trend toward a decrease after gavage (P = 0.09). The core conclusion is therefore that HT treatment lowers amino acid bioavailability, not that SPI overall raises serum amino acids. Each group had n = 4, and a single fasted group served as the baseline, a limitation to keep in mind when reading the time course.

Social novelty recognition is the only impaired behavior

In the six-week feeding experiment, diets supplied 20% protein with either untreated or HT-treated SPI replacing casein in the standard AIN-93G formula as the sole protein source. Body weight did not differ significantly between groups at any point. The open-field and elevated plus maze tests showed no group differences in anxiety-like behavior, entries, or total locomotor activity, and the object recognition test showed no differences in close-range exploration of the novel object or in recognition index.

The three-chamber social test had two phases. In session 1 (sociability), both groups showed stay and close-range exploration indices for Stranger 1 significantly above the 50% chance level, with no group difference. In session 2 (social novelty), only the untreated SPI group showed an index for Stranger 2 significantly above 50%, and the HT-treated group's index was significantly lower than the untreated group's. HT-treated SPI thus specifically impaired the ability to distinguish a new from a familiar individual without broadly changing social motivation or anxiety. Behavioral tests used n = 12 per group, the main source of strength for the behavioral conclusions.

Peripheral serotonin falls while tryptophan and brain serotonin hold

After the same six weeks of feeding, serum serotonin was significantly lower in the HT-treated group than in the untreated group, but serum Trp and whole-brain serotonin did not differ significantly. This combination matters: if the behavioral deficit were simply a shortage of tryptophan supply driving down central serotonin synthesis, serum Trp and brain serotonin should fall together, and they did not.

The authors therefore propose that high-temperature treatment may generate Maillard-type compounds or modified peptides that interfere with Trp metabolism or the peripheral serotonin pathway, rather than acting through reduced Trp supply alone. This must be read as the authors' mechanistic hypothesis based on correlational results; the paper does not directly show that a specific compound caused the serotonin drop or the behavioral change. Serotonin and Trp were measured by ELISA, and behavioral testing followed the feeding period, so the evidence is associative.

The conclusion is bounded by 20% protein and specific heating

A key boundary comes from the authors' preliminary work: on a 10% protein diet, HT-treated SPI already significantly reduced feed efficiency in rats, whereas on a 20% protein diet feed efficiency did not differ. Choosing the 20% level for long-term feeding means the behavioral and serotonin changes appeared even under relatively adequate protein, which makes the behavioral finding more notable but also means the results cannot be extrapolated directly to low-protein diets or to humans.

The heating conditions are equally specific: a 10% aqueous suspension, 121°C, 20 minutes. In the discussion, nitrogen loss plus ammonia production together account for only about 15% of the amino acid residue loss, leaving the rest speculative. The judgment most worth remembering from this paper is therefore that moist high-temperature treatment can cut SPI amino acid availability by roughly a tenth even when nitrogen content barely changes, and can selectively affect social novelty recognition and peripheral serotonin under a protein-adequate long-term diet, while anxiety, sociability, and object recognition remain unaffected, so the damage is selective rather than general behavioral toxicity.

Sources used in this answer