Doenjang Fermentation: What Bacillus Subtilis Does to Soy Isoflavones

Doenjang’s fermentation starter, meju, is dominated by Bacillus subtilis, a spore-forming bacterium, not by Aspergillus oryzae, the mold that Japan’s miso industry treats as its indispensable koji organism. Nam, Lee, and Lim sequenced 16S rRNA genes from nine local and two commercial doenjang samples in a 2012 study in the International Journal of Food Microbiology and identified 208 bacterial species across the samples, with each batch reflecting a distinct, region-specific community rather than a single controlled culture. That is the core proposition of this piece: doenjang’s fermentation biology, its isoflavone chemistry, and its human health evidence are each more specific, and less settled, than the “fermented soy paste is good for you” shorthand common in wellness writing suggests, and none of that evidence resolves the tension created by doenjang’s very high sodium content.

The practical starting point of doenjang is meju, a brick of boiled, mashed soybeans left to dry and ferment in open air for one to two months, typically hung from rafters where ambient bacteria and airborne mold spores colonize it. Jeon and colleagues, screening bacterial isolates from low-salt doenjang fermentations in a 2016 study in the Journal of Microbiology and Biotechnology, found that Bacillus species dominate this stage because of their tolerance for the fluctuating temperature, humidity, and eventual high salt concentration of the process, and they specifically evaluated strains for the proteolytic and antimicrobial activity needed to break down soy protein and outcompete spoilage organisms.

• Strongest evidence Bacillus-dominated fermentation microbiology and glycoside-to-aglycone isoflavone conversion, confirmed by direct sequencing and chromatography across multiple labs.
• Moderate human evidence A large sodium-adjusted Korean cohort links daily jang intake to lower metabolic syndrome and hypertension risk in men.
• Mixed, contradictory evidence Korean case-control and meta-analysis data link fermented soy paste to elevated, not reduced, gastric and colorectal cancer risk.
• Unresolved tradeoff Doenjang runs roughly 4 to 5 grams of sodium per 100 grams, a genuine cardiovascular counterweight to any fermentation benefit.
• Animal-only so far Isoflavone bioavailability gains and blood-pressure benefits from doenjang have been shown in rats, not yet in a human trial.

Doenjang’s bacterial fermentation is a genuinely different biology from Japanese miso’s Aspergillus koji process

Miso production begins with koji: rice, barley, or soybeans inoculated with a controlled Aspergillus oryzae culture, which is then combined with a salted soybean mash and left to mature, sometimes for years, under a fermentation regime dominated by halotolerant organisms such as Tetragenococcus halophilus and the osmotolerant yeast Zygosaccharomyces rouxii. Doenjang’s meju stage skips the controlled mold inoculation step entirely. Fungi including Aspergillus, Mucor, and Rhizopus do colonize the drying meju brick from the ambient environment, but Oh and colleagues, characterizing microbial succession across meju fermentation in a 2024 study in Food Science and Biotechnology, found that Bacillus, Lactococcus, and Enterococcus populations shift over the fermentation timeline and that Bacillus subtilis functions as the organism most responsible for the proteolytic breakdown of soy protein into peptides and free amino acids, a role koji mold enzymes play in miso.

This distinction is not a minor taxonomic footnote. Bacillus subtilis’s protease and amylase enzymes operate under different pH, temperature, and salt tolerances than Aspergillus oryzae’s, and Bacillus fermentation additionally produces enzymes and byproducts, including menaquinone and bioactive peptides, that koji-based miso fermentation does not generate in comparable quantities. Nam, Lee, and Lim’s sequencing work also found that the specific bacterial composition of doenjang varies meaningfully by region and household, meaning “doenjang” describes a family of related but chemically distinct fermentation outcomes rather than a single standardized product, a variability that complicates any blanket health claim made about “doenjang” as though every jar were chemically identical.

Fermentation converts most soy isoflavone glycosides into aglycones, without increasing the soybean’s total isoflavone content

Raw soybeans carry their isoflavones, primarily daidzin and genistin, attached to a sugar molecule in glycoside form. Fermentation’s enzymatic activity, driven largely by microbial beta-glucosidases from the Bacillus- and mold-associated community, cleaves that sugar group, converting daidzin to daidzein and genistin to genistein, the aglycone forms. Lee and colleagues, comparing household-made and commercial doenjang directly in a 2023 study in Molecules, found that household doenjang samples, which typically undergo longer, more variable fermentation with a more diverse bacterial community, carried substantially higher aglycone-to-glycoside ratios than commercial doenjang, which is often produced with shorter fermentation windows and more standardized inputs. Household samples also showed greater bacterial diversity overall, tying the degree of isoflavone conversion directly to fermentation depth and microbial community complexity rather than treating it as a fixed property of the food category.

This conversion is real and measurable, but it is not the same claim as “fermentation increases doenjang’s isoflavone content.” Li and colleagues, working with a whole-cell biocatalysis system paired with protease treatment in a 2017 study in the Journal of Microbiology and Biotechnology, were explicit that the goal of their accelerated process was to reach the aglycone and gamma-aminobutyric acid profile of doenjang aged more than a year, using a fraction of the time, precisely because natural long fermentation depletes the total isoflavone pool even as it shifts the remaining fraction toward the aglycone form. Isoflavones are consumed as substrate by the fermenting microbial community and are also subject to further microbial degradation into other metabolites over long fermentation timescales. The honest summary is that doenjang typically contains less total isoflavone than the raw soybeans it was made from, but a larger share of what remains is in the aglycone form.

The bioavailability advantage of aglycone isoflavones is documented in rodents, not yet confirmed in a doenjang-specific human trial

Aglycones are absorbed faster and more completely than glycosides because gut and brush-border enzymes must first hydrolyze the glycoside’s sugar group before absorption can proceed, a rate-limiting step that fermentation effectively completes before the food is even eaten. Lee and colleagues tested this directly in a 2018 study in the Journal of Food Science, comparing isoflavone pharmacokinetics after boiled soybean versus doenjang administration in sham-operated and ovariectomized, estrogen-deficient rats. In the estrogen-deficient animals, doenjang produced isoflavone metabolite exposure, measured by area under the curve, comparable to boiled soybean despite doenjang’s lower total isoflavone content, indicating that fermentation’s bioavailability gain can offset its content loss, at least under the estrogen-deficient physiological state the study modeled.

That finding is worth taking seriously as mechanism, and it is equally worth stating plainly that it comes from an ovariectomized rat model, not a human pharmacokinetic trial conducted on doenjang itself. Related human pharmacokinetic work exists for cheonggukjang, a different, unsalted fermented soybean product made without the long brine-maturation step doenjang requires, which is not the same food and cannot be assumed to behave identically. No published human trial has yet measured doenjang-specific isoflavone absorption kinetics in people. The gap between a well-designed rodent bioavailability study and a demonstrated human absorption advantage is exactly the kind of gap that deserves to stay visible rather than being smoothed over into a general “fermentation boosts isoflavone absorption in people” claim.

A large, sodium-adjusted Korean cohort links daily fermented soybean paste intake to lower metabolic syndrome and hypertension risk

The most substantial human evidence on doenjang’s health outcomes comes from Jeong and colleagues’ 2023 analysis in Frontiers in Nutrition, a hospital-based cohort of 58,701 Korean adults that measured combined “jang” intake, meaning chungkookjang, doenjang, doenjang soup, and ssamjang together, via food frequency questionnaire. After adjusting for total sodium intake and standard confounders, higher daily jang consumption was inversely associated with metabolic syndrome risk in men, with the association strongest for body fat and hypertension components specifically. The sodium adjustment is the methodologically important detail: the researchers explicitly built the analysis around the fact that jang is high in sodium, described in the paper at roughly 4.43 grams per 100 grams for doenjang specifically, and still found an inverse association with metabolic syndrome risk after accounting for it statistically.

A separate mechanistic study by Mun, Park, and Cha, published in Nutrients in 2019, gives a plausible physiological pathway for the hypertension finding, though again in an animal model. Sprague-Dawley rats fed a high-salt diet alongside doenjang showed significantly lower blood pressure and lower plasma renin than rats fed the same high-salt diet without doenjang, alongside increased urinary sodium and potassium excretion in the doenjang group. The authors’ interpretation, that eating a salty fermented food is not automatically equivalent to a direct sodium-driven blood pressure increase, is a reasonable reading of their own rat data, and it offers one candidate mechanism, altered renin-angiotensin signaling and mineral excretion, for the cohort’s human association. It is not, on its own, proof that the same mechanism operates the same way in humans eating doenjang as part of an ordinary diet, and no human blood pressure trial isolating doenjang from a broader dietary pattern has been published to confirm it.

The same population’s cancer epidemiology complicates any simple protective narrative for fermented soy paste

Any honest account of doenjang’s human evidence has to include the cancer literature, because it does not point uniformly in a favorable direction. Kim and colleagues’ 2011 meta-analysis in Cancer Science, pooling 20 studies of fermented soy food intake in Japanese and Korean populations, found that high fermented soy food consumption was associated with significantly increased gastric cancer risk, an odds ratio of 1.22, while non-fermented soy food intake was associated with decreased risk in the same pooled analysis. Shin and colleagues’ 2015 case-control study in PLOS ONE, comparing 901 Korean colorectal cancer cases with 2,669 controls, found that while total soy product intake was broadly associated with reduced colorectal cancer risk, high intake of fermented soy paste specifically was associated with elevated colorectal cancer risk in men.

Neither finding necessarily contradicts the metabolic syndrome cohort data; a food can plausibly carry a favorable effect on one disease pathway and an unfavorable association with another, and observational nutritional epidemiology routinely produces exactly this kind of pattern rather than a single unified verdict. But it does mean that a writeup of doenjang’s health profile that cites only the isoflavone chemistry and the metabolic syndrome cohort, while omitting the fermented-soy gastric and colorectal cancer signal from the same broad literature, is presenting a curated rather than a complete picture. The mechanistic candidates proposed for the cancer association center on sodium’s disruption of the gastric mucosal barrier and on nitrosamine or biogenic amine exposure from prolonged salt fermentation, both plausible pathways given doenjang’s salt content and extended maturation period, though the case-control and cohort designs available cannot establish causation on their own.

Sorting the evidence by rigor shows a bounded, sodium-complicated story rather than a clean superfood narrative

Ranked by strength, the fermentation microbiology and the isoflavone glycoside-to-aglycone conversion sit at the top: both are documented through direct genomic sequencing and chromatographic measurement across multiple independent Korean research groups, and there is little scientific dispute about the basic chemistry. Below that sits the metabolic syndrome and hypertension cohort association, a genuine, sodium-adjusted, large-sample human finding, but one drawn from a single hospital-based cohort analysis rather than a randomized trial, and one that describes association, not proven causation. Below that again sits the isoflavone bioavailability advantage, mechanistically plausible and demonstrated in an estrogen-deficient rat model, but not yet confirmed in any human pharmacokinetic study specific to doenjang. Running directly against all of it is a separately replicated epidemiological signal, from a meta-analysis and an independent case-control study, that fermented soy paste intake tracks with higher, not lower, gastric and colorectal cancer risk in the same Korean and Japanese populations, plausibly tied to the same sodium content that the metabolic syndrome study had to statistically adjust away.

The evidence-consistent position is that doenjang is a fermented food with a well-characterized, Bacillus-driven microbiology distinct from miso’s koji tradition, and one that measurably converts its isoflavones into a more readily absorbed chemical form even as fermentation reduces the total isoflavone pool available to convert. Its association with lower metabolic syndrome and hypertension risk in a large Korean cohort is real and worth taking seriously, but it has not been tested in a controlled human trial, and it exists in the same literature as a documented, biologically plausible association between fermented soy paste intake and elevated gastric and colorectal cancer risk. Doenjang’s sodium content is not a footnote to either finding; it is a shared candidate mechanism running through both the metabolic syndrome cohort’s need for statistical adjustment and the cancer studies’ proposed pathway, which makes it the single variable most likely to determine which effect actually dominates for any individual eating pattern, and current epidemiology has not settled that question either way.

Dr. Michael Torres is the Food Chemistry Columnist at Daily Bite Lab. He holds a PhD in Food Science from Cornell University, with research focused on plant chemistry, nutrient bioavailability, and the bioactive properties of phytochemicals.