Research note 007 — 2026-08-30 (expanded 2026-09-08)
Did Salt-Rubbing Ever Remove Bitter Gourd's Bitterness?
The short answer
Under a condition of soaking in brine for 60 minutes and making juice, bitterness did not differ. Among things you can try in home cooking, adding katsuobushi (dried bonito flakes) lowered bitterness.
To try at home
Add katsuobushi to goya chanpuru or raw bitter gourd. In tests where people tasted and compared, bitterness dropped significantly (p<0.01 in chanpuru, p<0.001 in raw bitter gourd).
In an experiment adding it to bitter gourd extract, katsuobushi stock (the simmered liquid) also lowered bitterness, but not as much as adding the flakes themselves.
As for salt-rubbing, this experiment does not show that it removes bitterness. What was measured was a processing condition of 60 minutes in 3.5% brine, and the bitterness score did not move.
Not yet known
No study was found that measured home salt-rubbing itself (rubbing with salt and squeezing out the liquid).
Why katsuobushi works. In an experiment adding the spent flakes left after making stock to bitter gourd extract, bitterness also dropped strongly, and the authors think bitter compounds are adsorbed onto the solid.
Which compound leads bitter gourd's bitterness is still not settled.
Salt-rubbing bitter gourd before cooking is standard practice, based on the idea that it "draws out the bitterness." The experiment read in this note, however, is not household salt-rubbing itself. Under a processing condition of 60 minutes in 3.5% brine before juicing, a sensory panel found no significant difference in the bitterness score before and after the brine treatment. What moved instead was sourness. A non-significant result is not proof that the effect is zero, but at the least, the change we perceive from salt-rubbing cannot be explained by a reduction in bitterness alone.
bitter melon — perceived bitterness is unchanged by: Fruit pieces soaked in 3.5% (w/v) NaCl solution at 25 C for 60 minutes before juicing (bitterness 10.6 +/- 1.1 on a 15 cm line scale versus 10.7 +/- 1.7 untreated, p > 0.05)supported
Scope: Bitter gourd juice evaluated by a trained panel of 10 in 1 Malaysian study
Note: A null result contrary to the folk expectation that salt-rubbing reduces bitterness. The same soak significantly reduced sourness rather than bitterness. Single study, single fruit lot.
For juice from fruit soaked in 3.5% (w/v) NaCl solution at 25 C for 60 minutes, bitterness was 10.6 +/- 1.1 on an unstructured 15 cm line scale, versus 10.7 +/- 1.7 for untreated fruit; the paper states verbatim that no significant (p > 0.05) difference in juice bitterness was observed between untreated bitter gourd and fruit treated with NaCl solution, while the same soak significantly (p < 0.05) reduced sourness.
Limitations: A single study with a small, young panel. Whether the untreated control also underwent a blanching step is not included in our extraction, so the effects of NaCl alone and NaCl plus blanching cannot be fully separated here.
Context: Verbatim: 'Only extracted ion chromatographs for blanched bitter melon and bitter melon with MY 68 agent showed the absence of momordicosides K and L.' Heat blanching and the commercial masking agent MY 68 rendered the bitterness-related glycosides (momordicoside K and L) undetectable by LC/ESI/MS, whereas oil cooking (frying) and drying (sun, oven, freeze) did not. This paper contains no description of NaCl soaking.
Limitations: The endpoint is compound detectability, not sensory bitterness. NaCl soaking itself is not among the conditions tested. Abstract-level. The link between K/L and bitterness assumed here has itself been contested since 2024.
Show dataingredient-bitter-gourd —has_perceived_bitterness_unchanged_by→ "Fruit pieces soaked in 3.5% (w/v) NaCl solution at 25 C for 60 minutes before juicing (bitterness 10.6 +/- 1.1 on a 15 cm line scale versus 10.7 +/- 1.7 untreated, p > 0.05)"
One study shows that soaking bitter gourd in 3.5% brine for 60 minutes barely moved its sensory bitterness score.
What moved instead was sourness — the same treatment is recorded as significantly lowering sourness.
In tests where people tasted and compared, what lowered bitterness in both raw bitter gourd and chanpuru was adding katsuobushi.
Multiple sources confirm that molecule-enclosing techniques such as beta-cyclodextrin lower bitterness, but how well they work varies with the molecule paired and the ratio used.
Which compound is responsible for bitter gourd's bitterness has still not been agreed upon among researchers for more than 40 years.
What actually moved bitterness was a cooking step other than salt
The bitterness that did not move with a brine soak dropped when katsuobushi was added. In tests where people tasted and compared, bitterness was rated significantly lower with katsuobushi in both goya chanpuru and raw bitter gourd (p<0.01 in chanpuru, p<0.001 in raw bitter gourd). In an experiment adding it to bitter gourd extract, katsuobushi stock (liquid simmered in hot water) also lowered bitterness (p<0.05), but did not reach the large drop seen with the flakes themselves. Adding the solid left after making stock (the spent flakes) to the extract lowered bitterness strongly. The authors think the effect comes from bitter compounds being adsorbed onto the solid flakes rather than staying in the liquid.
Scope: Layer: food. Katsuobushi added to raw bitter gourd and to cooked bitter gourd.
Note: Based on the abstract's description (the Japanese portion of the body text was garbled due to a font-embedding problem in the PDF, so specific figures such as sensory scores could not be confirmed). The authors conclude that katsuobushi has a debittering effect by adsorbing (binding) the bitter substances.
English abstract: 'Addition of Katsuobushi reduced the bitterness of both raw and cooked bitter gourd.' States explicitly that adding katsuobushi reduced the bitterness of both raw and cooked bitter gourd.
Limitations: The Japanese portion of the PDF body text was garbled due to a font-embedding problem, so the sensory-score figures and panel details could not be confirmed.
Scope: Layer: food. Katsuobushi 'extract' added to bitter gourd (a condition without solid material).
Note: The authors interpret that debittering is due mainly to adsorption (binding) by the solid material rather than to components dissolved into the extract. 2026-09-26: Reread in full (Figure 2, p.187). The English abstract's phrase "significant debittering effect was not achieved by using the extract" means 「かつお節と同程度の著しい効果は得られなかった」 (in translation: "an effect as pronounced as with katsuobushi itself was not obtained"), and in Figure 2 there is a significant (p<0.05) decrease with the extract addition.
Fig. 2: adding an equal volume of katsuobushi extract (hot-water extract) to bitter gourd extract lowered the bitterness score below the control (distilled water) (p<0.05). However, the effect was weaker than that of katsuobushi itself in Fig. 1, and with concentrated katsuobushi extract the suppression did not strengthen; bitterness rose slightly because of the extract's own off-taste.
Limitations: The panel was 5 people (mean age 25.0±6.7) selected from students and staff. A single cultivar and a single lot of bitter gourd.
Cooking itself can also change how bitterness is perceived. In an experiment evaluating bitter gourd mixed into 4 kinds of dishes — chili, tomato sauce, curry, and minestrone soup — dishes using tomato, garlic, and basil scored higher on the bitterness item (degree of liking) than the strongly spiced curry. The authors interpret this as showing that stronger spicing does not necessarily hide bitterness, and that the tomato-and-garlic combination worked better. This is a way of moving bitterness through cooking itself, separate from salt-rubbing.
bitter melon is documented as having The liking score for the bitterness attribute (9-point hedonic scale, 1 = dislike extremely to 9 = like extremely) differed by recipe: chili 5.0±2.2, tomato sauce 5.9±2.0, curry 3.3±2.2, minestrone soup 5.1±2.3 (rated by 49 non-diabetic adults, mean±SD). The authors state that the 3 dishes other than curry masked bitterness with ingredients such as tomato, garlic, and basil, and among them tomato sauce had the highest score.preliminary
Scope: Layer: food. 50 g of bitter gourd mixed into a 1-cup (250 mL) serving of each dish (chili, tomato sauce, curry, minestrone soup). Frozen within 7 days of cooking, reheated to 70-73°C and held at 60°C on the day of evaluation, and served with white rice or pasta. Texture, bitterness, aroma, and overall liking were evaluated.
Note: The bitterness score is a hedonic (liking) evaluation, not a rating of bitterness intensity itself. Curry had the lowest bitterness liking of the 4 dishes, and the authors interpreted this as showing that spice does not necessarily mask bitterness.
Hedonic scores for the bitterness attribute of the 4 dishes (mean±SD, 9-point scale, 1 = dislike extremely to 9 = like extremely): chili 5.0±2.2, tomato sauce 5.9±2.0, curry 3.3±2.2, minestrone soup 5.1±2.3. The authors state "tomato, garlic and basil that masked their bitterness," and identify tomato sauce as masking bitterness the most.
Limitations: Because the bitterness score is a liking evaluation rather than an intensity measurement, whether bitterness intensity itself changed cannot be determined from this evaluation alone. The specific spice composition of the curry recipe has not been confirmed within the scope of the text summary.
Show dataingredient-bitter-gourd —is_documented_as_having→ "The liking score for the bitterness attribute (9-point hedonic scale, 1 = dislike extremely to 9 = like extremely) differed by recipe: chili 5.0±2.2, tomato sauce 5.9±2.0, curry 3.3±2.2, minestrone soup 5.1±2.3 (rated by 49 non-diabetic adults, mean±SD). The authors state that the 3 dishes other than curry masked bitterness with ingredients such as tomato, garlic, and basil, and among them tomato sauce had the highest score."
What was significant, in the same line of research, was adding gum arabic. Beta-cyclodextrin (a cyclic sugar that encloses bitter molecules) also substantially raised juice acceptability — a technique, also used in pharmaceutical bitterness masking, of "enclosing" the molecule so it never touches the tongue.
bitter melon — perceived bitterness is suppressed by: Addition of 5% gum arabic to juice (8.6 +/- 0.6 versus control 10.7 +/- 1.7 on a 15 cm line scale, p < 0.05; 7.3 +/- 1.4 at 15%)supported
Scope: Bitter gourd juice evaluated by a trained panel of 10 in 1 Malaysian study
Note: In the same study, bitterness showed a negative correlation with viscosity, so part of the effect may be masking by matrix/mouthfeel rather than removal of the bitter compounds.
The paper states verbatim that adding 5% gum arabic significantly (p < 0.05) reduced the bitterness of bitter gourd juice. The table values, on a 15 cm line scale, are 10.7 +/- 1.7 for the control versus 8.6 +/- 0.6 with 5% gum arabic and 7.3 +/- 1.4 with 15% gum arabic.
Limitations: In the same study, viscosity was negatively correlated with bitterness, and masking from increased viscosity is not separated from binding of the bitter compounds.
Context: Verbatim from the abstract: "Juice with 1.5% beta-cyclodextrin demonstrated highest score (7.7 +/- 0.3) for sensory acceptability compared to the control (3.8 +/- 0.7)", and the NMR/ROESY/FTIR results "indicated formation of an inclusion complex by interaction of hydrophobic triterpenoidal region of momordicosides with beta-cyclodextrin". That is, an independent group reproduced the sensory improvement of bitter gourd juice by a carbohydrate-type additive.
Limitations: The additive is beta-cyclodextrin (inclusion-complex formation), not gum arabic; the endpoint is overall acceptability, not bitterness intensity; an abstract-level reading.
Context: Verbatim: 'adding 0.6% CMC resulted in lowering the bitterness significantly by 19%' (a trained-panel rating on a 10 cm unstructured bitterness scale). The proposed mechanism is that CMC acts as 'a physical barrier that restricts the interaction between bitterants and taste buds by increasing viscosity'. This independently reproduces, with a different viscous polysaccharide (carboxymethylcellulose), the same class of effect that Rashima 2017 reported with 5% gum arabic.
Limitations: CMC is not gum arabic. There was no significant reduction at 0.2% or 0.4% CMC, and even at 0.6% the reduction was only 19%, far smaller than beta-CD (>90% at 0.75%). The matrix is an extract, not juice. The paper notes that CMC 'cannot be applied to low-viscosity products'.
Show dataingredient-bitter-gourd —has_perceived_bitterness_suppressed_by→ "Addition of 5% gum arabic to juice (8.6 +/- 0.6 versus control 10.7 +/- 1.7 on a 15 cm line scale, p < 0.05; 7.3 +/- 1.4 at 15%)"
Scope: Bitter gourd juice with 0.25-2% beta-cyclodextrin added, in 1 Indian study
Note: The reported value is hedonic acceptability, not a bitterness-intensity rating. The abstract has no bitterness-intensity value.
Juice with 1.5% beta-cyclodextrin showed the highest sensory acceptability score (7.7 +/- 0.3) compared with the control (3.8 +/- 0.7). NMR, ROESY, and FTIR showed formation of an inclusion complex through interaction between the hydrophobic triterpenoid region of momordicosides and beta-cyclodextrin.
Limitations: An abstract-level extraction. The score is acceptability, not bitterness intensity, and panel size and scale are not given in the abstract. A slight (10%) decrease in alpha-glucosidase inhibition is also reported.
An independent group found, in bitter gourd extract, verbatim that 'With the addition of only 0.25% beta-CD, the bitterness was reduced by about 33%', and that with 0.75% or 1% beta-CD 'the bitterness was reduced by more than 90%', a contrast with the 19% reduction seen with 0.6% carboxymethylcellulose. A trained panel (n = 10, ISO 8586-2014, 10 cm unstructured scale) rated bitterness intensity, and FTIR/1H-NMR/TGA supported inclusion-complex formation by beta-CD. Total saponin content was not significantly affected.
Limitations: This measures the bitterness intensity of an extract rather than the hedonic acceptability of juice as in Deshaware 2018, and the effective concentration also differs (0.75% versus 1.5%). Because of the different matrix, direct numerical comparison is not possible, but the bitterness-removal mechanism via beta-CD inclusion complexation is independently reproduced.
For bitter gourd extract itself, separately from acceptability (liking), there is an experiment that directly measured the strength of bitterness. Adding just 0.25% beta-cyclodextrin is reported to lower bitterness by about 33%, and increasing it to 0.75% or 1% is reported to lower it by more than 90%. Set apart from the earlier result that acceptability rose when it was added to juice, this is data showing that the bitterness intensity itself moved.
bitter melon — perceived bitterness is suppressed by: Addition of beta-cyclodextrin (bitter gourd extract; a 0.25% addition reduced bitterness by about 33%)supported
Scope: Layer: food. Beta-cyclodextrin added to bitter gourd extract. Separate from the acceptability claim (same source group), this line records the reduction in bitterness intensity itself.
Note: Split out from an existing claim that recorded the rise in acceptability, isolating only the reduction in bitterness intensity.
An independent group found, verbatim, in bitter gourd extract, that 'With the addition of only 0.25% beta-CD, the bitterness was reduced by about 33%', and that with 0.75% or 1% beta-CD 'the bitterness was reduced by more than 90%', contrasting with a 19% reduction using 0.6% carboxymethyl cellulose. A trained panel (n = 10, ISO 8586-2014, 10 cm unstructured scale) rated bitterness intensity, and FTIR/1H-NMR/TGA supported formation of an inclusion complex with beta-CD. Total saponin content was not significantly affected.
Limitations: This measures the bitterness intensity of an extract rather than the hedonic acceptability of a juice, as in Deshaware 2018, and the effective concentrations also differ (0.75% vs. 1.5%). Because of the difference in matrix, a direct numerical comparison is not possible, but the debittering mechanism via beta-CD inclusion complexation is independently reproduced.
Show dataingredient-bitter-gourd —has_perceived_bitterness_suppressed_by→ "Addition of beta-cyclodextrin (bitter gourd extract; a 0.25% addition reduced bitterness by about 33%)"
The technique of inclusion complexation is used for bitterness masking in oral pharmaceuticals generally, not only in bitter gourd. But it does not work uniformly. A systematic review of taste masking in oral formulations introduces the technique of covering a bitter-tasting active ingredient in a host-guest inclusion complex, and notes that among the prior studies it cites, there are combinations, such as fluoxetine or captopril, where the bitterness nearly disappears, and others, such as spironolactone, where a complex forms but the taste does not improve. Even with the same beta-cyclodextrin, how it works changes with the drug it is paired with and the molar ratio.
cyclodextrins can reduce the perceived bitterness of Bitter-tasting active pharmaceutical ingredients. Through host-guest inclusion complexation that lowers the concentration of free drug reaching the taste buds. The effect depends on the type of cyclodextrin and the drug:CD molar ratio, and is not obtained for every drugcontested
Scope: Oral pharmaceutical formulations. A 2023 systematic review of literature indexed in PubMed, Web of Science, Scopus, and the Cochrane Library on cyclodextrins and taste masking
Note: The same review reports both clear successes (fluoxetine/beta-CD, captopril/beta-CD) and clear failures (spironolactone/HP-beta-CD, cetirizine/gamma-CD). The individual primary studies have not been read here; the evidence level is the review's own reporting.
The reported successes include fluoxetine complexed with beta-cyclodextrin (6 volunteers scored the uncomplexed fluoxetine as 3 [bitter] or 4 [very bitter] at 10 seconds, while the complex was scored 1 [tasteless]), and captopril at a 1:2 molar ratio with beta-cyclodextrin (of 6 volunteers, 3 rated the orally disintegrating tablet as pleasant, 2 as slightly sweet, and 1 as slightly bitter, while the uncomplexed captopril tablet had a burning, metallic flavor).
Limitations: The panel was 6 people, unblinded, and the scale differs between studies; the primary reports have not been read here. The proposed mechanism (the complex has no affinity for the taste receptor, or the free drug falls below threshold) is the review's explanation, not a measurement.
Contradicts: For spironolactone and HP-beta-cyclodextrin, a complex formed but the taste did not improve, confirmed in 24 healthy volunteers. Brief-access taste-aversion results agreed: rat preference rose for spironolactone solutions with 1%, 2%, and 6% HP-beta-CD, but lick counts fell at 18% HP-beta-CD, and the review concludes that the taste-masking effect for spironolactone is not very strong.
Limitations: Review-level reporting. The primary studies have not been read here. A single drug-cyclodextrin pair.
Limits: For cetirizine dihydrochloride, of 13 tasters, 5 reported no bitterness and 7 reported only mild bitterness for a cetirizine/beta-CD 1:5 solution, while 11 of the 13 described a cetirizine/gamma-CD 1:5 solution as extremely bitter. A drug:CD ratio of 1:5 tasted better than 1:2. Likewise, for propantheline bromide and oxyphenonium bromide, natural and modified beta-CD reduced bitterness significantly better than alpha-CD or gamma-CD.
Limitations: Efficacy is specific to the type of cyclodextrin and the molar ratio, and does not support the general statement that cyclodextrins mask bitterness. Review-level reporting; the primary studies have not been read.
Show datacompound-class-cyclodextrins —can_reduce_perceived_bitterness_of→ "Bitter-tasting active pharmaceutical ingredients. Through host-guest inclusion complexation that lowers the concentration of free drug reaching the taste buds. The effect depends on the type of cyclodextrin and the drug:CD molar ratio, and is not obtained for every drug"
There is also an instrumental example that looks more closely at the pairing of drug and inclusion agent. A formulation mixing propranolol and hydroxypropyl-beta-cyclodextrin at a 1:1 molar ratio had a bitterness score, measured by electronic tongue, 15% lower than propranolol alone, and increasing the ratio to 1:2 lowered it by about 30%. This is not a human-panel evaluation, but it shows that even with the same drug and the same inclusion agent, the degree of reduction changes with the mixing ratio alone.
Scope: 0.2% w/v aqueous propranolol solution. Insent SA 402B taste-sensing system, 4 sensors, each sample measured 2 times. No human panel
Note: Instrumental measurement only. Because of how the anchor was set (drug alone = 10), the percentages are relative to the unformulated drug and are not an absolute bitterness value.
The bitterness score was calculated with HP-beta-CD alone set at 0 and propranolol alone at 10. The propranolol:HP-beta-CD = 1:1 mol:mol formulation had a total bitterness score (bitterness 1 + 2) of 8.5, a 15% reduction versus propranolol; the 1:2 mol:mol formulation was 7.3, about a 30% reduction. PCA of the sensor data separated the samples along PC1 (99.2% of variance explained), and only 2 bitterness sensors were discriminant; aftertaste bitterness and aftertaste astringency were not discriminant.
Limitations: Instrumental only, no human panel; each sample was measured 2 times. The percentages are relative to unformulated propranolol, and the reference points were assigned rather than measured against a bitterness standard.
Context: The review cites an EMA report on the propranolol hydrochloride formulation Hemangiol, stating that the e-tongue result 'demonstrated ~80% masking effect compared to the reference in water', and that subsequent clinical trials observed neither a signal of poor acceptability nor early treatment discontinuation for reasons of taste.
Limitations: The ~80% figure and the cyclodextrin figures of 15-30% cannot be compared directly: the formulation, the e-tongue scoring convention, and the reference state differ. The underlying EMA report has not been read here.
A similar shape appears in a separate experiment on herbal tinctures (within the scope of the abstract). Adding beta-cyclodextrin to 5 tinctures including white horehound lowered the bitterness direction, as measured by electronic tongue, in every case, but the degree of reduction differed by tincture, and it fell the most for the white horehound tincture, which contains the bitter molecule marrubiin. Even within a single technique — inclusion complexation — whether it works, and how much, is determined by the target molecule and its pairing. The story of suppressing bitter gourd's bitterness through inclusion complexation can also be read as one instance of this general tendency.
Scope: Layer: solution. PCA analysis using an Alpha Astree II electronic tongue (7 potentiometric sensors). Caffeine was used as the standard for the bitterness direction. The authors state that the electronic-tongue data cannot provide a quantitative assessment of bitterness.
"The addition of CDs reduces the perceived bitterness of tinctures. The efficacy depended on the types of CD and mother tincture; beta-CD was the most effective in reducing bitterness of marrubium mother tincture. Complementary GC-MS analysis confirmed the formation of an inclusion complex of beta-CD and the highly bitter molecule, marrubiin, present in the marrubium mother tincture."
Limitations: Checked only at the abstract level. This is electronic-tongue data, and the authors themselves state that it cannot provide a quantitative assessment of bitterness. Not a human sensory panel.
Bitter gourd's bitter compounds also move a receptor-like pathway outside the tongue. Adding a bitter-compound-rich fraction to a mouse enteroendocrine cell line induced GLP-1 secretion in a dose-dependent manner, with stimulation rising up to 3.35-fold at 100 microg/mL. This pathway was partially blocked by the drugs probenecid and U-73122, and is reported as a result consistent with a pathway involving a bitter receptor and PLC-beta-2. However, the authors themselves state that whether a bitter receptor is truly involved is a speculation requiring further investigation, and what corresponds to the human bitter receptor in mouse cells is not known. Separate evidence exists that giving a bitter stimulus to the human stomach raises GLP-1, but that was obtained with quinine, not bitter gourd's cucurbitanes, and there is no receptor-level confirmation either. The fact that something moved inside a cell, and feeling it as bitterness on a human tongue, are matters at different levels.
bitter melon stimulates_glp1_secretion_via In mouse STC-1 enteroendocrine cells, a probenecid- and U-73122-partially-sensitive pathway consistent with involvement of a bitter receptor and PLC-beta-2 (bitter-compound-rich fraction: maximal 3.35-fold stimulation at 100 microg/mL)preliminary
Scope: Aqueous extract and fractions of wild bitter gourd applied to a mouse enteroendocrine cell line
Verbatim: BGP-bi (a bitter-compound-rich fraction enriched in 2 momordicosides, prepared by Okabe's method) induced GLP-1 secretion in a dose-dependent manner, reaching a maximal stimulation of 3.35-fold over vehicle-treated cells at 100 microg/mL. Both probenecid and U-73122 partially inhibited GLP-1 production induced by WE and by BGP-bi (P < 0.05). Of 6 isolated cucurbitane triterpenoids, the maximal stimulation exceeded 6-fold for compound 5 (19-nor-cucurbita-5(10),6,8,22-(E),24-pentaen-3beta-ol, 246 microM) and exceeded 3-fold for compound 6 (karavilagenine E, 440 microM).
Limitations: The authors state, verbatim, that the speculation that a bitter receptor may be involved requires further investigation, because the mouse and STC-1-cell orthologs of these human bitter receptors, and their specific agonists/antagonists, remain unknown. This is a mouse cell line, and there is no human data on this mechanism. The inhibition was partial.
Context: Human in vivo evidence exists that a bitter stimulus in the stomach can raise plasma GLP-1 and lower postprandial blood glucose, but it was obtained with quinine, not with bitter gourd's cucurbitanes, and there is no receptor-level confirmation, so it does not transfer the mouse STC-1 result for bitter gourd to humans.
Limitations: Both the stimulus and the preparation differ from the bitter gourd study. There is no TAS2R antagonist arm.
Limits: Verbatim: "no immunostaining evidence exists depicting a colocalization between EEC markers and T2Rs, although alpha-gustducin and TRPM5 have both been detected in serotonin and GLP-1-producing ECCs." This weakens the inference that enteroendocrine GLP-1 release in response to bitter compounds is mediated by T2Rs on those cells.
Limitations: This is a 2017 assessment, and this review does not review the bitter gourd study itself.
Show dataingredient-bitter-gourd —stimulates_glp1_secretion_via→ "In mouse STC-1 enteroendocrine cells, a probenecid- and U-73122-partially-sensitive pathway consistent with involvement of a bitter receptor and PLC-beta-2 (bitter-compound-rich fraction: maximal 3.35-fold stimulation at 100 microg/mL)"
The lead role in bitter gourd's bitterness is still undecided
A 1982 structural study named momordicosides K and L as the bitter principle, a 2024 dose-threshold analysis named different compounds (momordicine IV and others), and a 2025 review shows yet another line-up — even after 40 years, the lead role remains contested. And among the reported compounds, a large number have been reclassified as artifacts arising during the extraction process.
bitter melon — the identity of the principal bitter compounds is contested between studies: Momordicoside K and L (1982 structural study, restated in later debittering studies) versus momordicine IV and charantoside B (2024 dose-over-threshold analysis) versus 5 principal cucurbitane-type triterpenoids (2025 review)contested
Scope: Immature bitter gourd fruit studied across independent groups from 1982 to 2025
Note: At the compound level the attributions are not mutually exclusive, but no single set of compounds is consistently named across studies, and isolation artifacts complicate every attribution.
Verbatim from the abstract: "The structures of momordicosides K and L, bitter principles in the fruits of Momordica charantia L. (Cucurbitaceae), were elucidated as 7-O-beta-D-glucopyranosides of 3beta,7beta-dihydroxy-25-methoxy-cucurbita-5,23-dien-19-al and 3beta,7beta,25-trihydroxy-cucurbita-5,23-dien-19-al, respectively, from spectral and chemical evidence." The introduction restates that K and L were "isolated as bitter principles from the immature fruits of Momordica charantia L."; the experimental section defers the isolation itself to an earlier Part III paper (Chem. Pharm. Bull. 30:3977 (1982)), and this paper itself is purely a structural study (1H/13C-NMR, CD, acetylation, hydrogenation, methanolysis).
Limitations: Nowhere in the full text is there sensory-panel, threshold, or bitterness-intensity data; the term 'bitter principle' is carried over from the Part III isolation paper. Momordicoside K has a 25-methoxy group and was crystallized from MeOH, and the paper itself demonstrates an easy acid-methanol conversion of this skeleton (0.01 N HCl-MeOH, room temperature) -- the very type of chemistry invoked by the 2024-2025 artifact reassessment.
A 2018 abstract restates this attribution verbatim: 'triterpene glycosides namely momordicoside K and momordicoside L causes bitterness in bitter-gourd, thereby reducing consumer acceptance'.
Limitations: This restates the 1982 attribution; it is not an independent sensory identification.
A 2024 dose-over-threshold analysis concluded that momordicine IV (5) and charantoside B (6) should be regarded as the principal bitter compounds of bitter gourd. Momordicoside L (1) was one of five known compounds isolated, and for compounds 3-8, a strong bitterness with recognition thresholds of 3.6-13.6 ppm in 3% ethanol was reported.
Limitations: The abstract also states that many of the isolated compounds are isomerization products of natural precursors, and that the primary bitterness contributors in fresh fruit should be those precursors, so this attribution applies to the analyzed extract rather than to fresh fruit.
A 2025 review states verbatim that bitter gourd's bitterness is attributed to 5 key cucurbitane-type triterpenoids, but the abstract does not name the compounds.
Limitations: The 5 compounds cannot be identified from the abstract; the full text could not be obtained, and overlap with the 1982 and 2024 attributions could not be confirmed.
A 2007 processing study puts the 1982 attribution into practice: its stated purpose was 'to determine the effect of the processing methods on momordicosides K and L contents' of a vegetable that 'has an unpleasant bitter taste', treating K and L as the analytical targets tied to bitterness. This records that the attribution of momordicoside K/L as the principal bitter compounds served as a working hypothesis in applied food chemistry through the period from Okabe 1982 to the 2024 re-attribution.
Limitations: It assumes the link between K/L and bitterness rather than testing it (no sensory verification of these compounds); the methanol extraction is precisely the condition later argued to generate cucurbitane-type artifacts.
Verbatim: "Bitter gourd's bitter taste is linked to a presence of high concentration of triterpene glycosides known as saponins (Deng et al. 2022), especially momordicosides K and L" -- showing that the 1982 K/L attribution was still being restated as the prevailing view as late as 2023, immediately before Cai 2024's dose-over-threshold re-attribution to momordicine IV and charantoside B.
Limitations: A secondary citation rather than primary sensory data. Included to record the transmission lineage of the contested attribution.
Show dataingredient-bitter-gourd —has_contested_identity_of_principal_bitter_compounds→ "Momordicoside K and L (1982 structural study, restated in later debittering studies) versus momordicine IV and charantoside B (2024 dose-over-threshold analysis) versus 5 principal cucurbitane-type triterpenoids (2025 review)"
cucurbitane-type triterpenoidsOf 244 cucurbitane-type triterpenoids reported from bitter gourd, 141 (58%) were judged to be isolation artifacts. Only 45 (18%) have a plausible biosynthetic pathway supporting a natural origin includes reported members judged to be isolation artifacts: supported
Scope: The body of literature on bitter gourd cucurbitane-type triterpenoids reported over roughly 40 years. An assessment by 1 research group
Note: The artifact judgment derives from a review and 1 mechanistic study by the same group. No independent replication of this classification has been found.
An analysis of the 244 reported cucurbitane-type triterpenoids shows, verbatim, that the majority (141, 58%) are isolation artifacts, and that only 45 (18%) have a plausible biosynthetic pathway supporting a natural origin.
Limitations: The classification criteria are in the full text, which is not accessible; the percentages are quoted from the abstract.
Based on demonstrated transformations, the authors state, verbatim, the view that all 40 of the 5,19-acetals, 46 methyl ethers, 6 5,19-ethyl acetals, and 7 ethyl ethers are extraction/isolation artifacts and should not be regarded as natural constituents of bitter gourd.
Limitations: The generalization from momordicine I to all the listed derivative classes is presented as the authors' view, extrapolated from 1 model compound.
Context: The 1982 primary literature itself records the transformation chemistry invoked by the 2025 artifact reassessment: methanolysis of momordicoside K in 0.01 N HCl-MeOH at room temperature for 1.5 hours yielded a methyl ether (K-ag-4) and 2 5,19-methyl acetals (K-ag-2 and -3; acetal carbons at delta 112.1 d and 114.9 d), and the authors state, verbatim, "K-ag-2 and -3 are not the genuine aglycones of K," explaining that they form via a C7 carbonium cation intermediate through "ring closure between C5 and C19-oxygen." That is, mild acidic methanol can actually convert a genuine glycoside of this skeleton into a 5,19-acetal/methoxy derivative.
Limitations: The 1982 experiment is a deliberate post-isolation derivatization under specified acidic conditions; how often an equivalent transformation occurs during ordinary extraction, and the artifact count of 141 out of 244, are claims of the Liu/Cai 2025 paper, not of this source.
Context: The review positions the chemical diversity of bitter gourd's cucurbitane-type triterpenoids as arising from both an enzymatic biosynthetic pathway and subsequent non-enzymatic modification (including transformation during the isolation/extraction process). This is consistent, at the mechanistic level, with the finding of acid-condition transformation in Cai 2025.
Limitations: This is a general, abstract-level positioning statement, and which category each individual compound falls into cannot be confirmed without the full text.
Show datacompound-cucurbitane-triterpenoids —includes_reported_members_judged_isolation_artifacts→ "Of 244 cucurbitane-type triterpenoids reported from bitter gourd, 141 (58%) were judged to be isolation artifacts. Only 45 (18%) have a plausible biosynthetic pathway supporting a natural origin"
bitter melon — bitterness varies among cultivars: Of three Guangdong cultivars grown under equivalent conditions, D was sweeter, Q was intermediate, and B was more strongly bitter. Geranyl 3-O-xylopyranosyl-glucopyranoside, which the authors identify as the substance responsible for bitterness, was strongly upregulated in B and Q but not in Dpreliminary
Scope: 3 bitter gourd cultivars (D, B, Q) in 1 multi-omics study
Note: Bitterness levels are based on the authors' experimental research and customer feedback; there is no description of a formal sensory panel. The study is limited to 3 cultivars.
Verbatim: "where D was sweeter, Q was average, and B was high in bitterness comparative to others"; "glycosylated terpenoids (e.g. goyaglycoside, kuguaglycoside) were markedly higher in cv. D compared to other cultivars"; and "the substance geranyl 3-O-xylopyranosyl-glucopyranoside, responsible for bitterness, was found highly upregulated in B and Q but not in D".
Limitations: The bitterness ranking relies, verbatim, on the authors' "experimental research and customer feedback", with no description of a formal sensory-panel method; the authors state that this study is limited to 3 specific cultivars. Note the apparent tension in which bitter-type glycosides (goyaglycoside, kuguaglycoside) are higher in the sweeter cultivar D, while a different geranyl glycoside tracks with the bitter cultivars.
Show dataingredient-bitter-gourd —has_bitterness_varying_among_cultivars→ "Of three Guangdong cultivars grown under equivalent conditions, D was sweeter, Q was intermediate, and B was more strongly bitter. Geranyl 3-O-xylopyranosyl-glucopyranoside, which the authors identify as the substance responsible for bitterness, was strongly upregulated in B and Q but not in D"
Outside the academic dispute, the lead role is often described as if already settled on a single compound. A text introducing Okinawan native vegetables states that goya's distinctive bitterness is due to momordicin, contained in the peel. While researchers are divided over which compound plays the lead role, everyday explanations are often written as if the answer were already decided.
Scope: Layer: food (a description concerning Okinawan native vegetables).
"Momordica charantia (G) is called 'goya' on the island. The peculiar bitterness of goya is due to momordicin, which is contained in the skin."
Limitations: This sentence carries no citation number, and whether it is primary data or a summary of general knowledge cannot be determined from the text. There is no quantitative value or analytical method for momordicin at that point.
Beyond differences between cultivars, ripeness is also a factor that moves bitterness. The description of the cultivar used for genome sequencing states that as the fruit ripens, the peel changes from green to yellow or orange, and the pulp changes from bitter white to sweet red. The bitterness that barely moved with salt-rubbing has considerable room to move with the choice of cultivar or the timing of harvest.
bitter melon — sensory attribute: During ripening, the peel changes from green to yellow/orange at maturity, and the pulp changes from bitter white to sweet red.supported
Scope: A description of fruit developmental stages in the cultivar (Momordica charantia var. abbreviata) used for telomere-to-telomere genome sequencing. 3 stages were set: IM (14 days after anthesis), BR (28 days after anthesis), and RS (35 days after anthesis).
Note: The original text reads 『the pulp turns from bitter white to sweet red』. Pulp refers to the fruit-flesh tissue, and the paper treats peel and pulp as the 2 tissues analyzed for metabolites; the word for the aril enclosing the seeds does not appear in the body text. Correspondence with the commonly stated tissue name in 『仮種皮が甘くなる』 (in translation: "the aril becomes sweet") cannot be confirmed from this paper, and is recorded here as a limitation.
States: "During the process of fruit ripening, the fruit peel turns from green to yellow/orange at maturity, the pulp turns from bitter white to sweet red."
Limitations: The term 『aril(仮種皮)』 (in translation: aril) does not appear in the body text; pulp is used as a tissue category alongside peel. Correspondence with the specific tissue called the aril has not been confirmed.
Show dataingredient-bitter-gourd —has_sensory_attribute→ "During ripening, the peel changes from green to yellow/orange at maturity, and the pulp changes from bitter white to sweet red."
The null result for salt-rubbing comes from 1 study (a soaking condition intended for juice processing), and the conditions differ from the home operation of "rubbing with salt and wringing out the liquid." The possibility that some bitter compounds are physically removed along with the liquid has not been tested in this experiment.
For beta-cyclodextrin, there are claims that measured acceptability (liking) and claims that measured bitterness intensity itself, and this note shows both separately. The inclusion-complex data for propranolol and marrubiin are instrumental measurements such as electronic tongue, not evaluations by a human sensory panel.
The source on herbal tinctures was checked only to the abstract level; the full text has not been read. The katsuobushi source was checked in full text (2026-09-26). Chanpuru is a dish with several ingredients, so it is not a comparison isolating katsuobushi alone. The panel for the stock experiment was 5 people. The adsorption view rests on an indirect experiment looking at how bitter compounds eluted.
The GLP-1 pathway was observed in mouse cells, and involvement of a bitter receptor is, by the authors' own account, still at the stage of speculation; there is no confirmation in humans.
The sentence stating that "momordicin is the source of the bitterness," in the material on Okinawan native vegetables, carries no citation number, and whether it is primary data or a summary of general knowledge cannot be determined from the text.
The bitterness score for the dishes is a hedonic (liking) evaluation, not a measurement of bitterness intensity itself.
Frequently asked
Does salt-rubbing remove bitterness from bitter gourd?
The bitterness of juice pressed from fruit soaked in 3.5% brine for 60 minutes was, in an evaluation by a trained panel of 10, barely different from untreated fruit (no statistically significant difference). Sourness, under the same treatment, is recorded as significantly lower. This experiment is limited to 1 study, and moreover it uses a soaking condition intended for making juice, which differs from the home operation of rubbing with salt and then wringing out the liquid. The possibility that some bitter compounds are physically reduced along with the liquid has not been confirmed in this experiment.
What can be used to lower bitter gourd's bitterness?
There are several operations reported in experiments to have lowered bitterness itself. Adding gum arabic to juice, and adding beta-cyclodextrin, which encloses bitter molecules, both have measurements showing a reduction in bitterness strength. For katsuobushi, adding the flakes themselves is reported to lower bitterness in both raw and cooked preparations, and, in an experiment adding it to bitter gourd extract, the stock (extract liquid) also lowered it, though not as much as the flakes. Cooking combinations also differ, with a record showing that dishes using tomato and garlic scored higher on the bitterness item (degree of liking). These are, however, the results of individual experiments, each differing in method, cultivar, and panel size.
So what compound is bitter gourd's bitterness actually caused by?
This has not yet settled on a single answer. A 1982 structural study named momordicosides K and L as the bitter compounds, a 2024 analysis named a different group of compounds, and a 2025 review lists yet another line-up as the main bitter agents. There is also a reassessment, of 244 bitter-gourd-derived compounds reported over roughly 40 years, judging that the majority arose as artifacts during isolation. Bitterness strength is also reported to differ by cultivar, so for now this can only be treated as a matter involving multiple compounds and conditions, rather than pointing to a single "bitter compound."
What is not known
No experiment directly measuring the home operation itself — rubbing with salt and then wringing out the liquid — was found in this inventory (only experiments dealing with soaking alone exist).
The mechanism by which adding gum arabic lowers bitterness (whether it is sensory masking through viscosity, or binding with the bitter substances) is not distinguished by this experiment alone, and has not yet been confirmed.
No experiment directly comparing, under the same conditions as a treatment like salt-rubbing, how much bitterness moves with ripeness was found in this inventory.
No source directly confirming, with a sensory panel, how much bitterness the candidate compounds actually have on a human tongue is found in this inventory.