Bromelain and papain: proteolysis, absorption, and clinical limits

Bromelain and papain do cleave proteins. The harder question is how much enzyme retains activity in the intestinal lumen after ingestion, how much reaches blood unchanged, and whether that fraction is sufficient for an anti-inflammatory or fibrinolytic effect. A laboratory casein assay and a plasma protein-concentration curve measure different links in that chain. Conflating them has fueled exaggerated claims, from “neutralizing” gluten to dissolving clots with a tablet.

Method and limitations

This is a narrative synthesis of 28 relevant records in the supplied bibliography, not a systematic review. On 9 October 2026, the primary studies by Castell, Bradbrook, Metzig, Akhtar, Majid, Hale, Engwerda, Walker, Braun, Baur, Brehler, and Masson were checked directly to resolve material quantitative and bibliographic discrepancies. Other studies are considered within their reported designs. No exhaustive search, formal risk-of-bias assessment, GRADE rating, or external peer review was performed. Purified enzyme, commercial mixtures, model systems, animals, and people are considered separately: sharing the name “bromelain” does not make these materials interchangeable.

First identify the enzyme and the activity being measured

Bromelain is a collective name for proteolytic fractions from Ananas comosus, not one chemically homogeneous molecule. Stem bromelain (EC 3.4.22.32) and fruit bromelain (EC 3.4.22.33) differ in composition, charge, glycosylation, and substrate specificity. The stem also contains ananain and other proteases; extraction leaves nonproteolytic proteins and low-molecular-weight components as well. A commercial “500 mg bromelain” product therefore cannot be equated with 500 mg of purified, active stem endopeptidase. Papain (EC 3.4.22.2) is a defined member of the C1 family, CA clan, obtained from latex of unripe Carica papaya. Raw latex also contains chymopapain, caricain, and glycine endopeptidase; their presence depends on the starting material and purification. Historical injectable use of chymopapain for chemonucleolysis does not characterize oral papain safety.

For an enzyme, source, active fraction, and assay conditions matter. It is useful to distinguish mass dose $m$ (mg), total activity $A$ (assay units), specific activity $A/m$, and activity retained in a particular medium after a specified time. GDU measures gelatin hydrolysis; MCU measures milk clotting. The assays differ in substrate and endpoint. A GDU value cannot be converted universally to MCU with one factor or to clinical “potency” without the manufacturer’s method and batch standard. FCC, FIP, and USP units also refer to defined analytical procedures. Comparing trials requires the product composition, activity per dose through shelf life, coating, and dosing schedule.

Crystallographic studies by Drenth and Kamphuis revealed two papain domains with a catalytic cleft and a disulfide-bond network that stabilizes the fold. C1 peptidases are characterized by conserved active-site Cys and His; residue numbering Cys25–His159, as well as Gln19 and Asn175, primarily refers to papain. Residue numbers and specific disulfide bridges cannot be transferred to each bromelain fraction without sequence alignment. A molecular mass of roughly 20–30 kDa explains why these proteins can enter pores in certain laboratory membranes, but does not by itself prove passage through healthy intestinal epithelium at a pharmacological dose.

How Cys–His cleaves a peptide bond

In the active form, Cys and His make a nucleophilic thiolate–imidazolium pair. The substrate fits so that the carbonyl of the bond to be cleaved faces the Cys sulfur; nearby groups help stabilize the oxygen’s negative charge in a tetrahedral intermediate. The cycle has acylation and deacylation stages, rather than one-step “protein dissolution”:

$$E\!–\!S^-+R_1CO\!–\!NHR_2\;\longrightarrow\;[E\!–\!S\!–\!C(O^-)R_1]\;\longrightarrow\;E\!–\!S\!–\!COR_1+H_2NR_2,$$ $$E\!–\!S\!–\!COR_1+H_2O\;\longrightarrow\;E\!–\!SH+R_1COOH.$$

First, His participates in proton transfer and departure of the amine product; it then helps activate water to hydrolyze the thioester acyl-enzyme. The net reaction is $R_1CO\!–\!NHR_2+H_2O\ightarrow R_1COOH+H_2NR_2$; the enzyme is regenerated. The ionization state of the active site depends on pH and its microenvironment. The free thiol is vulnerable to oxidants and heavy-metal binding; a reducing agent or chelator may increase measured in vitro activity, but that does not establish therapeutic synergy between orally consumed N-acetylcysteine and bromelain.

Specificity is described with the Schechter–Berger nomenclature: substrate residues $P_3,P_2,P_1\mid P'_1,P'_2$ fit corresponding pockets $S_3,S_2,S_1\mid S'_1,S'_2$. Papain’s $S_2$ pocket is particularly important and prefers many hydrophobic residues at $P_2$. This preference is probabilistic: the surrounding bond sequence, folding of a large protein, and site accessibility alter the rate. Papain and bromelain do not become trypsin because some cleavages occur after Lys/Arg: trypsin is a serine protease with a different pocket architecture and cleaves mainly after basic residues. Pepsin is an aspartic protease of acidic environments. Broad in vitro substrate range does not mean these enzymes can cleave every structurally protected bond in native collagen or every immunogenic food epitope.

Initial rate on a defined soluble substrate is often described by $v_0=k_{cat}[E]_{active}[S]/(K_m+[S])$, while $k_{cat}/K_m$ is useful at low $[S]$. These values depend on substrate, pH, ionic strength, temperature, and how much of the preparation is actually active. Values for BAEE, gelatin, or a small chromogenic peptide cannot be inserted into a calculation of gluten or fibrin hydrolysis in vivo. A processing-temperature optimum above body temperature does not mean the molecule has maximal activity at 37 °C or will survive the stomach unchanged. A pH profile must be reported with substrate and exposure time: “active from pH 4–9” does not mean equal rate across that range.

Food and mucosa: what happens in the intestinal lumen

An exogenous endopeptidase can cleave accessible bonds in proteins in a food bolus into shorter polypeptides. Gastric, pancreatic, and brush-border enzymes then hydrolyze those products. Tenderizing meat with papain or bromelain is established food processing: prolonged contact, controlled temperature and pH, and a disrupted matrix. After ingestion, dose, contact time, and access to the native collagen triple helix are different. Thus, the ability to hydrolyze gelatin or denatured collagen does not prove clinical correction of “poor collagen digestion.” Casein, soy proteins, and actomyosin are acceptable laboratory substrates, but fragment-size data must be distinguished from gastric emptying, symptoms, and nutritional status.

GI proteases have physiological limits. Acidic gastric juice can inactivate or proteolyze some plant protease; food temporarily buffers pH, but the degree of protection varies with meal composition, acid secretion, and gastric-emptying time. An enteric coating changes where a product is released; it does not guarantee transport of enzyme into blood. Gastric acidity and pepsin activation explain why changing pH affects both endogenous enzyme activity and survival of an added enzyme. In confirmed exocrine pancreatic insufficiency, pancreatic enzyme replacement therapy has its own clinical indications and nutritional monitoring; plant proteases do not justify replacing that regimen without supervision.

Gluten: disappearance of a peak does not equal safety

Proline- and glutamine-rich regions of gliadin are relatively resistant to ordinary digestion. Shan and colleagues described the immunogenic α-gliadin 33-mer and the role of overlapping epitopes in celiac disease. Testing a candidate enzyme requires more than seeing protein cleavage in a tube: all reaction products should be identified by mass spectrometry, HLA-DQ2/DQ8-dependent T-cell epitopes assessed for persistence or loss, activity tested in a real food matrix, and rate shown to be adequate over gastrointestinal transit time. Separate studies of caricain and enzyme combinations explored this idea, but a caricain-containing mixture, papain, and bromelain are not the same intervention. In particular, one cannot claim that either of the latter two fully converts the 33-mer into non-immunogenic di- and tripeptides at an ordinary dose.

In celiac disease, clinical outcomes include mucosal status, symptoms, and immune markers under controlled gluten exposure, not merely disappearance of the original chromatographic peak. The reviewed evidence does not show that bromelain or papain protects sufficiently from accidental contamination to change a strict gluten-free diet. Non-celiac wheat/gluten sensitivity has different candidate triggers and diagnostic criteria; transferring celiac findings to it is also unjustified. Cow’s-milk protein can be fragmented while some epitopes retain allergenicity, so hydrolysis of β-casein is not equivalent to preventing an immune reaction.

Proteins in the mucin gel and microbial biofilms can also be substrates in vitro. But mucins form a protective layer, not simply an obstacle to digestion; excessive thinning could impair the barrier. The link from mucin proteolysis in a model, through biofilm disruption, to human recovery requires separate tests. Mynott and Chandler’s work in piglets and rabbit intestinal loops showed that bromelain can alter enterotoxigenic-bacteria interactions or toxin responses in those systems. This does not prove prevention or treatment of bacterial diarrhea in people. Soy trypsin-inhibitor experiments should likewise be read as substrate tests: the stability and post-cooking behavior of Kunitz or Bowman–Birk inhibitors differ, and proposed “synergy” with pancreatin has not been confirmed against a clinical endpoint.

From gut to blood: evidence of presence and its quantitative limit

Most dietary protein is digested, but it is also incorrect to claim that a mature intestine is absolutely impermeable to every intact macromolecule. Epithelial and M-cell endocytosis can transport antigens, and permeability depends on mucosal condition. These general routes do not establish a bromelain-specific apical receptor, lysosomal escape through its own proteolysis, or safe “opening” of tight junctions. Each proposed transport route requires a labeled intact molecule, compartment tracking, exclusion of free label, and an independent check that activity is retained. None justifies assigning a systemic bioavailability percentage by analogy with a small-molecule drug.

Castell and colleagues’ key human study included 19 healthy men. They detected immunoreactive bromelain in plasma after oral administration, confirmed an undegraded fraction by immunoprecipitation, and found at least partial retention of proteolytic activity. With repeated 3 g/day dosing, the concentration reached up to 5000 pg/mL, or 5 ng/mL, by 48 hours; from the curve, the authors estimated a mean 10.8 μg of bromelain present in plasma over the 3–51-hour interval. The estimated half-life of the plasma signal was 6–9 hours. Protein was detected in association with α₂-macroglobulin and α₁-antichymotrypsin. This is important evidence for small-scale transfer of intact protein in people, not proof that 15–40% of the dose entered blood.

It is essential to distinguish three denominators: the fraction of a dose that crossed the intestinal barrier; the mass present in measured plasma at a given time; and the ability of a biochemical assay to detect enzyme activity in a sample. The ratio of 10.8 μg to 3 g is about $3.6\times10^{-6}$, but even this is not measured absolute bioavailability: it uses an estimate of the amount in plasma during repeated dosing, not the integral of all incoming and outgoing fluxes. The published data do not support a reliable calculation of $F$, volume of distribution, free active enzyme concentration in tissues, a universal $T_{max}$ across products, or a systemic papain dose. Attributing a plasma $C_{max}$ in micrograms per milliliter to this study is an error of three orders of magnitude relative to its upper value in nanograms per milliliter.

α₂-Macroglobulin acts as a trap for many protease classes: cleavage of its “bait region” triggers a conformational change that restricts access of large protein substrates to the trapped enzyme. A small test peptide may still reach the active site under some conditions; this explains how plasma can contain immunoreactive enzyme and partially measurable activity at the same time. But fibrin fibers, large leukocyte surface receptors, and extracellular-matrix proteins do not automatically become accessible substrates for this complex. The complexes are also captured by receptor-mediated pathways and cleared from circulation. The model in which a “protective capsule delivers bromelain specifically to inflamed tissue and releases it there” is not demonstrated by human pharmacokinetic data; old precise percentages of organ accumulation cannot be transferred to clinical use.

Immune and inflammatory pathways: the experiment’s location matters

A potential mechanism begins at the surface of an accessible cell: a protease may cleave an extracellular receptor or ligand domain, changing leukocyte migration or activation. Hale and colleagues studied changes in surface molecules after treating cells with bromelain; candidate targets included CD44 and L-selectin (CD62L). CD62L supports initial leukocyte “rolling” on endothelium, while CD44 participates in hyaluronan interactions. A change in flow-cytometry fluorescence should be confirmed with recovery after protease inhibition, measurement of shed ectodomain, and cell-viability checks: a lower signal may reflect cleavage of the antibody epitope, internal regulation, or surface damage. Even confirmed cleavage on a treated cell does not show that an oral dose produces the corresponding concentration of free protease in blood. In Hale’s experiment, plasma sharply changed marker sensitivity to bromelain: for some targets, estimated EC₅₀ exceeded 1000 μg/mL in 100% plasma, whereas it was about 1 μg/mL without plasma. These values belong to that particular in vitro setup, but illustrate the strength of inhibitor context; they cannot be compared directly with 5 ng/mL after oral dosing and called the same action. Fitzhugh’s experimental neutrophil-migration models showed a functional response, but translation to treating inflammation in people is a separate step.

The NF-κB pathway includes IKK-dependent phosphorylation of IκB, degradation of the inhibitor, and movement of p65/p50 into the nucleus. Stimulated cell systems have reported changes in NF-κB, COX-2, iNOS, and cytokines after bromelain treatment. This does not prove direct, selective binding of bromelain to IKKα/β, unchanged COX-1, or absence of mucosal injury. Proteolytic modification of a surface receptor, a change in stimulus dose, cell toxicity, or extract impurities can produce similar endpoint signals. The arachidonic-acid pathway likewise cannot be simplified to “PGE₂ and TXA₂ fall while PGI₂ is always preserved”: different cells and enzymes produce different mediators, and measurements in culture are not their balance in a patient.

Kallikrein converts kininogen to bradykinin; through the B₂ receptor, bradykinin participates in pain, vasodilation, and vascular permeability. A cysteine protease may hydrolyze suitable peptides in a tube, but its rate in the presence of physiological inhibitors, access to interstitium, and contribution to edema resolution in people are not established. PAR-2 can also be activated or “disarmed” by cleavage at different positions; without site mapping, any protease contact cannot be presumed analgesic. Findings on substance P, Th2 cytokines, Th17, or FoxP3/Treg from mouse or cell experiments should be described as hypotheses about modulation, not established immunotherapy for asthma, colitis, or autoimmune disease. Secor’s allergic-airway-inflammation experiments describe a mouse model, not clinical prevention of asthma.

Direction of effect matters: proteolysis may weaken one signaling chain while strengthening another if it creates an active ligand or disrupts a barrier. In cell and mouse experiments, Engwerda and colleagues observed both increased T-cell proliferation through accessory cells, decreased IL-2 production, and stronger antigen-specific B-cell responses. This combination is incompatible with the simple label “suppresses all cellular immunity.” A clinical conclusion would require prespecified disease outcomes, a control comparison, and free-exposure data, not a cumulative list of possible targets.

Fibrin, plasmin, and platelets: three different experiments

Fibrin forms when thrombin cleaves fibrinogen, and factor XIIIa cross-links fibrin chains. Physiological lysis depends on conversion of plasminogen to plasmin by t-PA and other activators, regulated by PAI-1 and α₂-antiplasmin. If purified bromelain breaks down native or denatured fibrin in a tube, that demonstrates fibrinolytic activity in the specified assay. Dissolving an intravascular clot after ingestion additionally requires sufficient free enzyme at the clot, access to fibers under blood flow, retained action in the presence of inhibitors, and better clinical outcomes at acceptable bleeding risk. Detection of bromelain bound to α₂-macroglobulin is precisely why caution is needed: a large fibrin network is unlike a small chromogenic substrate that can reach the center of a trap.

A separate type of evidence is platelet treatment in vitro. At 10 μg/mL bromelain, Metzig and colleagues inhibited thrombin-induced aggregation of isolated human platelets; a lower concentration affected adhesion in an endothelial experiment. In the same study, an oral dose of 60 mg/kg in rats changed thrombosis in a mesenteric-vessel laser model. These results inform a mechanistic hypothesis, but 10 μg/mL in a tube is 2,000 times the 5 ng/mL upper plasma concentration reported by Castell for a different product and regimen in people. This comparison does not prove indirect effects impossible, but rules out treating the two experiments as equivalent. Metzig’s work is not a trial of stroke or venous-thrombosis prevention or of acute clot treatment in patients.

Reports of changes in aggregation, fibrinogen, PAI-1, or t-PA across separate systems cannot be combined into one confirmed cascade. D-dimer is a specific product of endogenous breakdown of cross-linked fibrin; “D-dimer degradation” is not the same as beneficial clot dissolution. There is no basis to claim that bromelain selectively blocks platelet ADP receptors, leaves normal hemostasis unchanged, or necessarily raises human t-PA. The label “systemic enzyme therapy” does not replace pharmacokinetic and clinical demonstration. Oral bromelain and papain should not be treated as thrombolytic drugs or substitutes for anticoagulants.

Human trials: outcomes, formulation, and control

Postoperative edema, pain, trismus, a joint-function scale, and laboratory aggregation are different endpoints. A significant change in one does not establish a change in another. Small randomized studies of different bromelain preparations in dental surgery measured pain, facial volume, mouth opening, and quality of life. For example, Majid and Al-Mashhadani assigned 45 people after removal of one impacted lower third molar to oral bromelain, diclofenac, or placebo; bromelain was given at 250 mg four times daily and diclofenac at 25 mg four times daily. A positive pain/edema signal in this brief model does not establish a universal 30–45% reduction in swelling, 40% lower analgesic use, or equivalence for major surgery. Interpretation depends on assessment timing, concomitant analgesics, facial measurement method, and enzyme activity, not milligrams alone.

In knee osteoarthritis, a mixture should not be called monotherapy. In Akhtar’s randomized double-blind study, 103 patients with a painful episode received a combination of bromelain, trypsin, and rutoside or diclofenac for six weeks. The primary measures were the Lequesne index and symptom index. In both groups, the Lequesne score fell to about 9.4 from baselines of 13.0 and 12.5; the authors met the prespecified non-inferiority criterion for the mixture versus the active comparator. No placebo group, short duration, and multiple ingredients prevent attributing the effect to bromelain alone or calling the treatments interchangeable for long-term care. The study did not establish gastric protection, a fivefold reduction in erosions, or bleeding prevention.

Other human observations show limits of inference. In an open study, 77 adults without an arthritis diagnosis but with recent mild knee pain completed one month of 200 or 400 mg bromelain per day in Walker’s study. WOMAC symptoms improved from baseline in both groups; doses were randomly assigned, but there was no placebo group or blinding. This was not a trial treating confirmed osteoarthritis. Masson described reduced pain and swelling in 59 people with blunt injuries who received bromelain alongside usual care; without a comparable control group, a causal effect cannot be separated from natural recovery and co-interventions.

Braun’s observational pediatric cohort in acute sinusitis included 116 patients from 19 centers: 62 received bromelain, 34 bromelain plus standard therapy, and 20 standard therapy. Mean symptom duration was 6.66, 9.06, and 7.95 days, respectively. Groups were not randomized; baseline severity and choice of concomitant treatment may have differed. The observation therefore does not prove that the enzyme shortened a bacterial infection. One child with known pineapple allergy had a mild allergic reaction, a relevant safety detail.

Trials of enzyme combinations in injuries and dental procedures also cannot be retold as results for purified papain. For DOMS after eccentric exercise, creatine kinase is an indirect biomarker of muscle damage; changes in CK, subjective soreness, and strength recovery should be considered separately. The supplied evidence does not support promising athletes a fixed 35–50% CK reduction or recovery two to three days sooner after a particular bromelain dose. A health-economic calculation requires actual costs, comparable treatment, and adverse-event accounting; without those data, claims of savings on proton-pump inhibitors or hospital days remain assumptions.

Antibiotics, cancer, and other clinical extrapolations

Historical studies examined bromelain given with specific antibiotics. The drug and measurement must be distinguished: Bradbrook, Morrison, and Rogers studied oral tetracycline in 1978, not ampicillin; this was a small absorption study, not proof of a threefold rise in every antibiotic concentration or improved infection cure. Other older amoxicillin/tetracycline studies have their own formulations, regimens, and analytic limitations. Current antibiotic prescribing must account for established dose, microbiological diagnosis, and drug interactions; these data do not justify adding an enzyme for presumed “delivery through fibrin.”

In cell-based cancer models, a protease may alter surface proteins, growth factors, and a cell line’s response to stimulation. This does not show that oral bromelain reaches a tumor, exposes all neoantigens to CD8 cells, neutralizes TGF-β, or prevents metastasis. α₂-Macroglobulin itself binds various signaling proteins; attributing a change in their activity specifically to captured bromelain requires a controlled experiment. The reviewed evidence provides no clinical oncology indication for bromelain or papain. Likewise, effects on T cells in animal models cannot be converted into a treatment regimen for ulcerative colitis, asthma, or autoimmune disease.

Using enzymology to assess a product and its risks

The phrase “with food for digestion, fasting for systemic effects” describes a plausible difference in contact with food substrate, not a validated chronopharmacologic distinction. Fasting changes pH and enzyme exposure to acid and pepsin; coating changes the release site. Neither guarantees a higher blood concentration of free active enzyme. A professional description of an intervention should specify stem or fruit source, presence of papain/caricain or trypsin, activity units and assay, coating, dose, timing, food conditions, concomitant medicines, and a prespecified outcome.

The most direct risk is hypersensitivity to pineapple or papaya proteins; sensitization to papain, bromelain, and chymopapain has been described, especially with occupational exposure. People with latex allergy may have cross-reactivity with some fruits, but percentages and specific molecular epitopes depend on population and IgE assay. A positive laboratory test does not guarantee a clinical reaction, while prior anaphylaxis to a specific enzyme warrants avoiding re-exposure pending allergy-specialist assessment. The claim that α₂-macroglobulin completely hides the enzyme from antibodies and prevents anaphylaxis is wrong: exposure may occur at the mucosa, and complexes do not make a foreign protein non-allergenic.

For hemostasis, distinguish a theoretical or preclinical interaction from an established clinical risk. If a patient uses warfarin, a direct oral anticoagulant, or an antiplatelet drug, is planning surgery, or has bleeding, informing the care team about a protease supplement and agreeing on its use is more important than following an invented universal “stop 48–72 hours before.” INR tracks warfarin effect; it is not a universal test for apixaban or bromelain. Reports of good tolerance in brief, small trials do not prove that rare bleeding, drug interactions, or pregnancy safety are absent; data for concentrated products in these settings are insufficient.

Plant cysteine proteases have a real place in enzymology and food technology. Oral bromelain has convincing evidence of a small active fraction being detected in plasma and limited clinical signals for symptoms after dental surgery and in combination products. Between that and a promise of “systemic thrombolysis” lie quantitative exposure, access to a large substrate, causal mechanism, and trials of clinical outcomes. Systemic human data for papain are even more limited. Expert decisions should remain tied to the specific preparation and measured outcome, not to the word “protease” on a label.