Research note 037 — 2026-09-04
Fermentation Creates Bitterness, Then Breaks It Down — Miso, Soy Sauce, Sake
Miso, soy sauce, and sake are all fermented foods in which koji mold breaks down the protein and glycosides of soybean or rice. That breakdown does not move bitterness in a single direction. From the same starting point — 「麹菌が原料を分解する」 (in translation: "koji mold breaks down the raw material") — soy sauce releases the soybean's glycosides into the lees, miso keeps its bitterness while holding onto the byproducts of protein breakdown, and in sake, koji enzymes cut up rice protein to generate new bitter peptides, after which milling, activated carbon, enzyme treatment, and storage move that bitterness again. Bitterness is not the product of a single fermentation step, but a product that follows a different path for each system. This note looks at miso, soy sauce, and sake in turn.
On this page (9 sections)
- What this page establishes
- Soybean starts with bitterness and astringency from glycosides
- Soy sauce releases glycosides into the lees, and koji enzymes break them down
- Miso's bitterness comes from the breakdown of soybean protein
- Sake's peptides are born from koji, and moved by milling, activated carbon, enzymes, and storage
- The word zatsumi includes bitterness, but is not only bitterness
- Honest notes
- Frequently asked
- What is not known
What this page establishes
- In soy sauce, one analysis has confirmed that nearly all of the soybean's glycosides move into the soy sauce lees, with almost none detected in the soy sauce itself (the degradation rate of saponin Bb was 68.7%).
- In sake, 5 bitter peptides born when koji enzymes break down rice glutelin have been isolated and identified.
- The substance responsible for miso's bitterness has not yet been chemically identified (for Hatcho miso, the explanation that it derives from protein breakdown remains inferential).
- Multiple experiments have shown that how much rice polishing and activated-carbon treatment reduce sake's bitter peptides varies in effect depending on the rice variety and the amount of treatment.
Soybean starts with bitterness and astringency from glycosides
Even before fermentation begins, soybean already carries the seeds of bitterness and astringency. Isoflavone glycosides (genistin, daidzin, etc.) isolated from raw soybean are reported to show bitterness when subjected to a sensory test.
Scope: Isoflavone glycosides (daidzin, genistin, etc.) isolated from raw, unprocessed soybean (sensory test)
Note: Okubo (1988) conducted a sensory test on genistin and daidzin, the main isoflavonoids of soybean, and their aglycones (genistein and daidzein), and showed that the isoflavonoid components exhibit a strong astringent taste and bitterness at a taste threshold of 10⁻²–10⁻³ mM.
Show data
compound-class-soy-isoflavone-glycosides —contributes_to→ sensation-bitterThe same isolated components are also reported to show a weak astringent taste — astringency. Bitterness and astringency arise together from the same glycoside, as separate sensations.
Scope: Isoflavone glycosides isolated from raw, unprocessed soybean (sensory test)
Show data
compound-class-soy-isoflavone-glycosides —contributes_to→ sensation-astringentSoybean also has another group of glycosides: saponins. A study examining 239 miso samples entered in the National Miso Tasting Competition reports a tendency for the content of the main saponin component to differ by type of miso, being highest in the order soybean miso, salty rice miso, sweet rice miso, and sweet miso.
Scope: 239 miso samples entered in the National Miso Tasting Competition (soybean miso, rice miso, barley miso, and specialty miso; per dry weight)
Note: Kudo et al. (1992) analyzed the glycoside components of 239 miso samples entered in the National Miso Tasting Competition by TLC and HPLC. The content of the main saponin component Bb was highest in the order soybean miso > salty rice miso > sweet rice miso > sweet miso, and tended to be higher the higher the blending ratio of raw soybean (though not in direct proportion). Group A saponins, meanwhile, were not detected in any of the 239 miso samples analyzed, and partial deacetylation or partial degradation of the sugar chain is inferred.
Show data
compound-class-soyasaponin-group-b —has_measured_content_in→ ingredient-misoSoy sauce releases glycosides into the lees, and koji enzymes break them down
In the process of becoming soy sauce, this glycoside does not disappear — it simply changes destination. A study examining raw soybean, finished koji, soy sauce, and soy sauce lees found no glycoside components detected in the soy sauce itself; all of it had moved into the soy sauce lees. Bands corresponding to the saponin aglycone were detected in the finished koji and the soy sauce lees, and it is discussed that an enzyme breaking down this glycoside is at work during the koji-making process.
Scope: The glycoside fraction of commercial soy sauce (TLC analysis)
Note: Kudo et al. (1992) prepared glycoside fractions from raw soybean, finished koji, soy sauce, and soy sauce lees, and compared them by TLC, showing that no glycoside components (saponins, isoflavones) are detected in the soy sauce itself, with most of them moving into the soy sauce lees. Because a band corresponding to soyasapogenol B was detected only in the finished koji and the soy sauce lees, it is inferred that an enzyme that breaks down the soybean saponin B group into its aglycone is produced during the koji-making process. The total group B saponin concentration in the soy sauce lees was about 2.5 times that of the raw soybean, and the degradation rate over the course of brewing was calculated at 68.7% for Bb and 33.0% for Ab.
Show data
compound-class-soyasaponin-group-b —is_excluded_from→ ingredient-shoyuAfter the glycoside has left for the lees, the bitterness remaining in soy sauce is attributed to a different component. In the background description of a study comparing commercial raw soy sauce and heat-treated soy sauce, soy sauce's bitterness is attributed to bitter-tasting amino acids such as isoleucine and to peptides, and this study used the concentration of phenylalanine, a bitter amino acid, as its indicator of bitterness.
Scope: Commercial soy sauce (background description in Imamura & Katayama 2017's examination of the validity of sensory-evaluation terms)
Note: In the process of selecting a quantitative indicator for sensory evaluation, Imamura & Katayama (2017) state 「しょうゆの苦味は，isoleucineなどの苦味を呈するアミノ酸やペプチド類に起因すると考えられている」 (in translation: "the source states that soy sauce's bitterness is attributed to bitter-tasting amino acids such as isoleucine, and to peptides"), and use phenylalanine, a bitter amino acid, as the quantitative analytical indicator of bitterness. This description is a background explanation based on a prior source cited by the same paper (reference 23, a secondary citation whose original was not checked), and the same paper itself does not newly isolate or identify a bitter peptide.
Show data
ingredient-shoyu —has_bitterness_attributed_to→ compound-class-bitter-peptidesAnd soy sauce's bitterness also moves in a single step that comes after fermentation is finished. When raw soy sauce and heat-treated soy sauce from the same manufacturers were compared by a trained panel, the heat-treated soy sauce showed stronger bitterness, bitter aftertaste, and astringency/egumi.
Scope: 6 brands each of commercial raw soy sauce and heat-treated (hiire) soy sauce (6 matched pairs from the same manufacturers), compared by QDA (quantitative descriptive analysis) with a trained panel of 13, women aged 32–62
Note: In Imamura & Katayama's (2017) QDA, heat-treated soy sauce showed significantly higher, or a differing tendency in, evaluation scores for bitterness, astringency/egumi, and bitter aftertaste compared with raw soy sauce (Wilcoxon signed-rank test). In the principal component analysis as well, bitterness, astringency/egumi, and bitter aftertaste showed strong positive loadings on PC1 (0.95, 0.94, and 0.93 respectively; 69.3% contribution rate), and heat-treated soy sauce was plotted in the positive direction of PC1. Note that this is the heat-sterilization process itself, not a change in the fermentation process.
Show data
process-shoyu-hi-ire —increases→ sensation-bitterMiso's bitterness comes from the breakdown of soybean protein
In miso, it is not glycosides but the traces of protein breakdown that are the main account given for bitterness. In an experiment extracting Hatcho miso with ether and butanol, a difference in the strength of bitterness was confirmed between the samples before and after extraction, but the bitter substance itself has not been chemically identified. There has long been an explanation that nitrogen compounds such as peptides and amines are the cause, but the description stops at there not yet being definitive confirmation.
Scope: Hatcho miso (soybean miso made using soybean alone)
Note: Ishii (1966) confirmed, through solvent fractionation and sensory testing, that the ether and butanol extracts of Hatcho miso show a strong bitterness (accompanied by an astringent taste), and states 「八丁味噌の苦味は従来からペプチド，アミンのような窒素化合物によるといわれているが，まだ確証はない」 (in translation: "the source states that Hatcho miso's bitterness has traditionally been attributed to nitrogen compounds such as peptides and amines, but there is not yet definitive confirmation"). The chemical isolation and identification of the bitter substance itself had not been completed as of this paper.
Show data
ingredient-miso —has_bitterness_attributed_to→ compound-class-bitter-peptidesIn the same study's conclusion, an inferential conclusion is offered: that Hatcho miso's bitterness is generated by the breakdown of soybean components, and derives from soybean protein. The author himself also notes, alongside this, the possibility that trace lipid components are having an effect, so it is not narrowed down to a single cause.
Scope: The ripening process of Hatcho miso (soybean miso made using soybean alone)
Note: From the conclusion of Ishii (1966): 「八丁味噌の苦味はその成分上の特長から考えれば，大豆成分の分解により生成されるものであり，大豆蛋白に由来すると考えられるであろう」 (in translation: "judging from the compositional characteristics of Hatcho miso, its bitterness would be considered to be generated by the breakdown of soybean components, and to derive from soybean protein"). This is an inferential summary drawn by the author himself from general-component and amino-acid composition analysis, and it does not involve isolation and identification of the bitter substance. The author also notes, alongside this, the possibility that trace lipid components are involved, since Hatcho miso contains more lipid than other misos.
Show data
ingredient-miso —derives_bitterness_in_part_from→ organism-glycine-maxFor miso in general, a secondhand empirical rule is recorded: that bitterness develops once the ratio of soluble nitrogen to total nitrogen reaches 65% or above. Yet in the same study, the measured value for Hatcho miso is about 55%, and even though it does not meet this standard, the author states that it is already bitter. This mismatch shows that how far protein breakdown has progressed cannot be described by a single ratio alone.
Scope: Miso in general (the original source is a secondary citation of Yoshikawa 1957). The point of comparison is the measured value for Hatcho miso
Note: Ishii (1966) introduces the statement 「一般に味噌は可溶性窒素/全窒素の比が65%以上になると苦味を生ずるといわれている(吉川，醗工55,501(1957))」 (in translation: "the source states that, in general, miso develops bitterness when the soluble-nitrogen/total-nitrogen ratio reaches 65% or above (Yoshikawa, Hakko Kogaku 55,501 (1957))"), while also pointing out, within the same source, a mismatch between the general rule and the measured value, stating 「八丁味噌ではこの比は大体55%であるがすでに苦味がある」 (in translation: "in Hatcho miso this ratio is about 55%, yet it is already bitter"). Yoshikawa (1957) is a secondary citation; the original was not checked.
Show data
ingredient-miso —has_perceived_bitterness_associated_with→ "The soluble-nitrogen/total-nitrogen ratio reaching 65% or above (a report described as a general tendency for miso)"That enzymatic breakdown of protein readily produces bitterness is, in itself, widely reported well beyond miso. A review sums up the bitterness that frequently occurs with enzymatic hydrolysis as being due mainly to the formation of low-molecular-weight peptides made of hydrophobic amino acids.
Scope: Enzymatic food protein hydrolysates.
Note: Not "the lower the molecular weight, the more bitter": the fraction below 1000 Da has the weakest bitterness (Cho 2004, see limits).
Show data
compound-class-bitter-peptides —cause_bitterness_of_protein_hydrolysates→ "The bitterness that frequently occurs with enzymatic hydrolysis of protein is due to the formation of low-molecular-weight peptides mainly composed of hydrophobic amino acids (a review's summary)."Sake's peptides are born from koji, and moved by milling, activated carbon, enzymes, and storage
In sake too, bitterness is born not from the raw material itself but as a product of breakdown. Steamed-rice protein is digested by koji-mold enzymes in the sake mash (moromi), generating, from the acidic subunit of rice glutelin, an oligopeptide with a characteristic structure at its N-terminus; this has been shown to possibly be involved in sake's taste.
Scope: The acidic subunit of glutelin, a protein in steamed rice, is digested in sake mash by koji-mold-derived enzymes, generating a characteristic oligopeptide (a bitter peptide) bearing a pyroglutamyl group at its N-terminus; this has been shown to possibly be involved in sake's taste.
Note: An account by Hashizume Katsumi (2008). Based on a digestion test of steamed rice using 4 rice-protein mutant lines (esp1-4) and their parent strains (Nipponbare and Yamada Nishiki), and on molecular structural analysis and sensory evaluation of a bitter-peptide preparation isolated from commercial junmai sake.
Show data
ingredient-sake —has_bitterness_generated_by→ compound-class-sake-glutelin-peptidesOne chemical reaction common to koji fermentation in general is involved in how this peptide comes to have its structure. A glutamine residue freed by enzymatic breakdown by koji mold can spontaneously cyclize into pyroglutamic acid partway through heating or ripening, and this same reaction also occurs at a peptide's N-terminus. Short-chain pyroglutamyl peptides have been identified in sake.
Scope: In koji-fermented seasoned foods (miso, soy sauce, sake, etc.), glutamine (Gln) freed by enzymatic breakdown by koji mold, or a Gln residue at a peptide's N-terminus, spontaneously cyclizes into pyroglutamic acid (pyroGlu) during heating or ripening. Short-chain pyroglutamyl peptides such as pyroGlu-Leu and pyroGlu-Gln have been identified in sake (Table 1).
Note: Zheng et al. (2022) review, pp. 6–8. Since the sake bitter peptides isolated by Hashizume (2008) likewise carry a pyroglutamyl group at their N-terminus, this is cited as general background for the mechanism. However, no statement calling the short-chain peptides listed by Zheng et al. (2022), such as pyroGlu-Leu, themselves 『苦味』 (in translation: "bitterness") could be found in that review, so this is not used as direct evidence for bitterness (the evidence's relation is set to contextualizes).
Show data
compound-class-sake-glutelin-peptides —has_biosynthesis_controlled_by→ "Protein breakdown by koji-mold enzymes, and the non-enzymatic cyclization of a peptide's N-terminal glutamine residue into pyroglutamic acid (pyroGlu)"What was actually isolated and identified were 5 oligopeptides carrying a pyroglutamyl group at their N-terminus, all of them derived from the N-terminal sequence of rice glutelin's acidic subunit.
Note: Hashizume (2008), Table 3 and Figure 3. Purified standards fractionated by reverse-phase HPLC from junmai sake (Yamada Nishiki, 70% rice polishing ratio).
Show data
compound-class-sake-glutelin-peptides —has_composition→ "The 5 pyroglutamyl (<Q) oligopeptides isolated and identified: <QLFNPS (calculated mass 687.0 / observed 688.5, LC-MS), <QLFNPSTNP (999.1/1000.4), <QLFNPSTNPWH (1322.4/1323.6, MALDI-TOF-MS), <QLFNPSTNPWHSP (1506.6/1507.7), <QLFGPNVNPWHNP (1501.6/1502.7). All derive from the N-terminal sequence of the acidic subunit of rice glutelin (31–32 kDa)."When these peptides were purified and synthesized for further study, detection and recognition thresholds are reported to differ from peptide to peptide, and the taste characteristics are also reported to differ by peak, described as 「苦味，不快な後味」, 「苦味，エグ味」, and 「エグ味，苦味，雑味」 (in translation: "bitterness, unpleasant aftertaste," "bitterness, egumi," and "egumi, bitterness, zatsumi").
Note: Hashizume (2008), Table 5. Evaluation of a purified preparation dissolved in a 15% ethanol solution.
Show data
compound-class-sake-glutelin-peptides —has_bitter_recognition_threshold→ "5 purified isolated peptides (peaks 9,13,17,18,20) had detection thresholds of 0.02-0.18 mM and recognition thresholds of 0.08-0.66 mM (geometric mean of 6 assessors). Taste characteristics differed by peak: "bitterness, unpleasant aftertaste," "bitterness, egumi," "egumi, bitterness, zatsumi (off-taste)," and so on. The index of concentration in sake divided by recognition threshold (a gauge of contribution to taste) ranged from 6 to 60."This peptide bitterness is not constant once it is born, either. When a digestion test was run on steamed rice with the rice polishing ratio changed in steps, one variety showed a tendency for the amount of high-molecular-weight peptide generated to decrease as the rice polishing ratio fell, but another variety showed no such tendency. Whether milling reduces the source of bitterness depends, in its answer, on the variety.
Scope: In a steamed-rice digestion test on 3 varieties prepared with rice polishing ratios stepped down from 90% to 40%, Nipponbare showed a tendency for the amount of high-molecular-weight peptide to decrease as the rice polishing ratio fell (i.e., as milling became more thorough), but Yamada Nishiki and Gohyakumangoku showed no such tendency.
Note: Hashizume (2008), Figure 4. Measured data showing that the relationship between rice polishing ratio and bitter-peptide generation depends on the raw-rice variety and is not uniform. Based on a digestion test with koji mold (strain RIB128) (pH 4.3, 37°C, left standing for 1–2 days).
Show data
compound-class-sake-glutelin-peptides —are_not_uniformly→ "Changes in the amount of high-molecular-weight peptide generated in steamed-rice protein digests, as a function of rice polishing ratio (90–40%)"This bitter peptide also moves through post-brewing processing. The more the amount of activated-carbon treatment was increased, the more the peptide concentration in sake fell, and commercial futsū-shu (regular sake), which uses more activated carbon, had a lower concentration than untreated daiginjo and junmai.
Scope: The more the amount of activated-carbon treatment (200/500/1000 mg/l) was increased, the more the peptide concentration in sake (peaks 13,17,18,20) fell, and with 1000 mg/l treatment most fell below the detection limit. In untreated daiginjo (2001–2004 vintage, 40% rice polishing ratio), peak 13 was 10.5–12.1 mM, compared with a lower 0.8–3.5 mM in commercial futsū-shu (regular sake) A–C (consistent with the tendency for futsū-shu using more activated carbon to be lower).
Note: Hashizume (2008), Table 4. Mean ± standard deviation (3 measurements).
Show data
compound-class-sake-glutelin-peptides —is_reduced_by→ "Activated-carbon treatment"In treatment using enzymes, it is reported that the effect of reducing bitterness comes packaged together with other changes. Adding a peptidase preparation with high acid carboxypeptidase activity lowered bitter peptides, but with a different enzyme preparation, yet another peptide increased instead — the effect on sake's taste changes depending on which enzyme is chosen.
Scope: According to an abstract by Hashizume et al. (publication year not confirmed, original not read in full) introduced in "Research Achievements on Alcoholic Beverages in Reiwa 4" (Journal of the Brewing Society of Japan, vol. 118, no. 4, p.220), adding a peptidase preparation with high ACP activity to sake reduced bitter peptides and increased (pGlu)L-ethyl. The source reports that, on the other hand, adding an enzyme preparation with high (pGlu)-peptide ethyl ester-forming activity and low ACP activity increased (pGlu)LF-ethyl and (pGlu)LFNP-ethyl in sake (without increasing (pGlu)L-ethyl), and it was suggested that this increase in oligopeptide ethyl esters gives a good effect on sake's taste. A separate abstract by Hashizume et al. in the same journal, p.233, states explicitly that an enzyme showing carboxypeptidase-type activity "contributes to the reduction of bitter peptides while also being involved in the generation of dipeptides with poor taste," suggesting that reducing bitterness can carry a quality trade-off as well.
Note: Original not read in full. A secondary citation from an abstracting journal (a summary by the editorial staff). Even for enzyme treatment of the same peptide lineage, the treatment "reduction of bitter peptides = desirable" coexists with the treatment "increase in the esterified derivative = a good effect on taste"; this is included as an example showing this two-sidedness in how bitterness is treated. The addition of the enzyme preparation with high ACP activity itself also carries a two-sided effect, reducing bitter peptides while increasing (pGlu)L-ethyl (see also the separate Hashizume et al. abstract on p.233 of the same journal, vol.118 no.4).
Show data
compound-class-sake-glutelin-peptides —has_bitterness_quality_reduced_by→ "Addition to sake of a peptidase preparation with high acid carboxypeptidase (ACP) activity."And storage — time itself — also moves bitterness. In a table gathering the evaluation terms for the flavor changes that accompany sake storage, bitterness is listed as a taste that increases with storage for pasteurized (heat-sterilized) sake, whereas the same table's entry for unpasteurized sake (before pasteurization) records no increase or decrease in bitterness.
Scope: According to Table 4 of Utsunomiya et al. (2006) (evaluation terms describing flavor changes accompanying sake storage), for pasteurized sake, the tastes "bitterness, stimulant taste (smooth)" increase with storage while "astringency, stimulant taste (rough)" decrease. The same table's entry for unpasteurized sake (before pasteurization) does not record any increase or decrease in bitterness.
Note: Based on the storage flavor-change table (Table 4), excluding hineka and namahineka (stale off-flavors). Note that the behavior of bitterness during storage is written up differently for unpasteurized sake and pasteurized sake.
Show data
ingredient-sake —has_perceived_bitterness_affected_by→ "Increase: storage of pasteurized (heat-sterilized) sake."The word zatsumi includes bitterness, but is not only bitterness
Among the words used to describe sake's flavor, there are also composite terms that include bitterness without being bitterness itself. 「雑味」 (in translation: "zatsumi") is defined as a composite sensation and subjective sensation described as bitterness, umami, astringency, and other unpleasant mouthfeel tastes.
Show data
concept-zatsumi —is_defined_as→ "A composite sensation and subjective sensation described as bitterness, umami, astringency, and other unpleasant mouthfeel tastes (the operational definition in Utsunomiya et al. 2006's terminology system)"Within this definition, bitterness is explicitly stated as one of the constituent elements.
Note: An explicit connection between zatsumi and bitterness, based on Utsunomiya et al.'s (2006) operational definition (bitterness, umami, astringency, and other unpleasant mouthfeel tastes).
Show data
concept-zatsumi —includes→ sensation-bitterYet in the study that compiled the terminology system for evaluating sake's flavor quality, zatsumi is excluded from the list of terms that carry standard reference samples, on the grounds that it is too strongly a composite, subjective sensation.
Scope: "Zatsumi" is a term currently in common use, but because it carries a strong sense of a compound, subjective sensation — "bitterness, umami, astringency, and other unpleasant mouthfeel" — it was excluded from this vocabulary.
Show data
concept-zatsumi —is_excluded_from→ "The standard terms (86 terms in Table 1) of the sake flavour-quality assessment vocabulary of Utsunomiya et al. (2006)"Meanwhile, the discussion section of the study that discovered sake's bitter peptides sums up that part of the reason for the previously vaguely understood empirical rule — 「タンパク質レベルが高い原料米で造った清酒には雑味が多い」 (in translation: "sake made from rice with a high protein level has more zatsumi") — could be captured in terms of protein level, namely the generation of glutelin-derived bitter peptides. Here, zatsumi, excluded from the terminology system, and the chemically identified bitter peptide are spoken of as connected.
Scope: The author discusses that part of the reason for the previously vaguely understood empirical rule 『タンパク質レベルが高い原料米で造った清酒には雑味が多い』 (in translation: "sake made from rice with a high protein level has more zatsumi") could be captured in terms of protein level, namely the generation of glutelin-derived bitter peptides.
Note: A statement in the closing section (p.579) of Hashizume (2008). This is the author's own interpretive summary, treated as an account that explains an empirical rule long discussed in the industry through the mechanism of bitter-peptide generation. Note that this is a statement in the discussion section, not an experiment that established causation statistically.
Show data
concept-zatsumi —is_associated_with→ compound-class-sake-glutelin-peptidesFor bitterness itself, a separate term with a standard reference sample is provided. The standard reference sample for the term 「苦味」 (in translation: "bitterness") is tyrosol, and its discrimination threshold, standard addition amount, and content in sake are given as numbers.
Scope: In the vocabulary's class "13. Bitterness" / the term "bitterness," defined as "bitter," the standard reference sample is set as tyrosol (2-(4-Hydroxyphenyl)ethyl alcohol, TCI 98%+). Detection threshold 1.4 g/L, standard addition level 4.9 g/L, content in sake 70-90 mg/L.
Show data
ingredient-sake —has_standardized_vocabulary_entry_in→ "The sake flavour-quality assessment vocabulary of Utsunomiya et al. (2006) (compiled by the National Research Institute of Brewing with the cooperation of the Japan Sake and Shochu Makers Association, the Nada Sake Research Association, and the Fushimi Brewers' Friendship Society)"Honest notes
- The bitter substance in Hatcho miso itself has not been chemically identified, and the conclusion that it derives from protein breakdown does not go beyond inference.
- The panel that evaluated sake's bitter peptides was a small one of 6 assessors, and the rice-polishing-ratio experiment compared only 3 varieties.
- The 65% soluble-nitrogen-ratio standard for miso in general is a secondary citation, and the original's measurement conditions could not be traced.
Frequently asked
Why does miso sometimes taste bitter?
In an experiment extracting Hatcho miso with ether and butanol, a statistical difference in the strength of bitterness before and after extraction has been confirmed, but the substance that produces the bitterness itself has not yet been identified. The view that nitrogen compounds (peptides and amines) are involved is an old one, but the author himself states explicitly, 「まだ確証はない」 (in translation: "there is not yet definitive confirmation"), and the influence of lipids is separately pointed out as well. The same study also introduces the rule of thumb spoken of for miso in general, 「可溶性窒素比65%を超えると苦味が出る」 (in translation: "bitterness appears once the soluble-nitrogen ratio exceeds 65%"), but notes that even though the measured ratio for Hatcho miso stays at about 55%, it is already bitter. Miso's bitterness is thought to be connected to protein breakdown, but identifying the causal substance itself has not yet been reached.
Where does sake's bitterness come from?
Steamed-rice protein, the starting material for sake, is broken down in the moromi by koji-mold enzymes, and it has been confirmed that bitter peptides carrying a pyroglutamyl group are born from part of rice's glutelin. The 5 isolated peptides each have different detection and recognition thresholds, and the quality of the bitterness is also reported to differ by sample, described as 「苦味,エグ味」 (in translation: "bitterness, egumi") and so on. In experiments that varied the rice polishing ratio, some varieties showed a decrease in the amount of peptide generated and others did not, so it cannot be said that milling alone lowers bitterness uniformly. It has also been confirmed that activated-carbon treatment and the addition of enzyme preparations can lower peptide concentration, but which taste remains behind depends on which treatment is chosen.
What is not known
- The substance responsible for Hatcho miso's bitterness itself has not been chemically identified.
- The rule of thumb cited for miso in general, 「可溶性窒素比65%」 (in translation: "a soluble-nitrogen ratio of 65%"), is a secondary citation, and no material has been found to confirm the measurement conditions it was originally based on.
- The sensory evaluation of sake's bitter peptides remained a small panel of 6, and the rice-polishing-ratio experiment was also a comparison of only 3 varieties; whether this can be generalized directly to other varieties has not yet been confirmed.
- The report concerning enzyme preparations with high ACP activity could not be confirmed against the original paper itself, and remains a secondary citation from an abstract.
Related:miso ／soy sauce (shoyu) ／sake ／bitter peptides ／sake bitter peptides from rice glutelin ／zatsumi ／soy isoflavone glycosides
Related topic hub: Bitterness, Astringency, and Egumi
Related research notes: Cooking Doesn't Erase Bitterness, It Edits It ／Bitterness Has a Clock
Source ledger: Source list
How to cite this page
Aoki, M. (NIGAMI LAB) (2026). Fermentation Creates, Then Breaks Bitterness. https://nigamilab.com/en/notes/fermentation-writes-and-erases-bitterness/
Each claim card has a stable anchor (#clm-…). The dataset is CC BY 4.0; see the data page. Dataset DOI: 10.5281/zenodo.22739101