The mammalian gastro‐intestinal tract is a NOT a major extra‐pineal source of melatonin

In 1992, a paper reported that the melatonin content of the rat duodenum was 24 000 ± 2000 pg/g tissue (range: 4000–100 000 pg/g) while the pineal melatonin content was 580 000 ± 36 000 pg/g. The data has been used for the last 30 years to infer that the gut produces 400 hundred times more melatonin than the pineal gland and that it is the source of plasma melatonin during the daytime. No‐one has ever challenged the statement. In this review, evidence is summarised from the literature that pinealectomy eliminates melatonin from the circulation and that studies to the contrary have relied upon poorly validated immunoassays that overstate the levels. Similarly studies that have reported increases in plasma melatonin following tryptophan administration failed to account for cross reactivity of tryptophan and its metabolites in immunoassays. The most extraordinary observation from the literature is that in those studies that have measured melatonin in the gut since 1992, the tissue content is vastly lower than the original report, even when the methodology used could be overestimating the melatonin content due to cross reactivity. Using the more contemporary results we can calculate that rather than a 400:1 ratio of duodenum: pineal melatonin, a ratio of 0.05–0.19: 1 is likely. The gut is not a major extra‐pineal source of melatonin; indeed it may well not produce any.


| INTRODUCTION
For more than 40 years many researchers across the world have been interested in the presence and potential actions of melatonin in the gastrointestinal tract.Reviews on the topic have included statements such as; "It has been generally accepted that the pineal gland is not the only source of melatonin in vertebrates 1 ; "Following pinealectomy, the light/dark cycle of plasma melatonin levels disappears, while its day-time blood concentrations are attenuated but sustained mainly due to its release from the GIT." 2 ; "It is now generally accepted, that whereas night time levels of melatonin in blood are mostly of pineal origin, daytime melatonin concentrations in blood are produced mostly in the GIT 3 ; "It was speculated that (the) majority of circulatory melatonin during the day was derived from the GI tract since pinealectomy does not alter the daytime serum melatonin level." 4 Another apparently influential concept is epitomised by the following quote; "Considering the large surface area of the GI tract and the relatively high concentrations of melatonin per gram of GI tissue it was calculated that the amounts of melatonin generated in GI tract of mammals would exceed the amounts of melatonin generated in pineal gland by roughly 400 times". 4To date, no-one has critically evaluated the quality of the research that underpins the hypothesis that there is endogenous melatonin in the GIT and that it has a major physiological role. 41 | Is the pineal gland the sole source of plasma melatonin?
A key pillar of the hypothesis of the gut being a major extra-pineal source of melatonin is the statement that pinealectomy does not decrease the levels of daytime plasma melatonin.A search of the literature identified more than 20 papers in which plasma melatonin had been assayed following removal of the pineal gland in rats (Table 1), 7 in sheep (Table 2) and a further 9 human studies involving subjects with ablated pineal tumours or denervated pineal glands (Table 3).Before assessing any individual studies, it is important to discuss the underlying methodologies behind melatonin measurement.
In the mid/late 1970s several research groups developed radioimmunoassays that were capable of detecting melatonin in plasma. 63While most laboratories attempted to partially purify and concentrate melatonin by extraction with solvents or chromatography before immunoassay, one popular method involved direct assay of plasma. 12When the assays were used by groups other than the original team, it was often not clear whether the original protocols were followed.
The most common species used in studies on the effects of pineal gland removal is the laboratory rat. Figure 1 shows the mean plasma melatonin levels in laboratory rats from collections at mid light and mid dark taken from published studies using various assay methodologies.The most sensitive and specific assays are those that use mass spectrometry.When plasma melatonin was assayed in rats by mass spectrometry 31 the melatonin concentration 1 h before lights off was 4 pg/mL, increasing to 52 pg/mL 4 h after lights off.There was no measurable melatonin (<1 pg/mL) in the plasma of pinealectomized animals at either time point.More recently Lee et al. 33 reported that blood melatonin showed significant circadian oscillations from 0.46 pg/mL in day to 14.6 pg/mL at night and that after pinealectomy, melatonin levels fell below previous basal levels even at night, to 0.14 pg/mL.
Figure 1 shows that in some studies the plasma melatonin concentration in pinealectomised rats, while generally not showing a night time rise, was still 1-2 orders of magnitude higher than levels detected using mass spectrometry assays.Interestingly the results published by Yu et al. 9 (Table 1) that used the early Pang RIA 10 were subsequently declared to be invalid by the Brown group, 23 raising questions about use of the antibody reagent in all prior reports.It is from studies using immunoassays that led some researchers to propose an extra-pineal source of melatonin, while ignoring the early mass spectrometer results.A similar 95% reduction in the excretion of the melatonin metabolite, 6-hydroxymelatonin assayed by mass spectrometry following pinealectomy 64 was apparently ignored.
There have been at least 7 studies addressing the question of melatonin levels in sheep following pinealectomy (Table 2 and Figure 2).The authors of these papers reported that levels following pinealectomy were below the sensitivity of their assays, which ranged from <4 to <30 pg/mL.The exceptions were papers by Kennaway et al. 34 who reported levels of 30-100 pg/mL in pinealectomised sheep and McMillen and Nowak 41 who reported levels of 8.4 pg/mL.In the former case, samples were collected continuously and used very high concentrations of heparin and this may have caused interference with the assay despite the chromatography step used in this first generation RIA.The conclusion from these studies is that pinealectomy prevents the nocturnal rise in melatonin and that daytime levels in both intact and pinealectomised sheep are below the sensitivity of the assays used.When plasma melatonin was measured in pineal intact sheep using the very sensitive and specific Buhlmann RIA, the daytime levels were below 2 pg/mL. 65he question arises about the source of melatonin in humans and in subjects who have had their pineal gland removed due to the presence of a tumour or who have had a cervical spinal injury inhibiting the neural pathway required to stimulate pineal melatonin production.High quality mass spectrometry assays have shown that humans have daytime plasma melatonin levels well below 4 pg/mL and some third generation immunoassays 63 are available that also report levels this low.As shown in Table 3 the destruction of the pineal gland or its denervation results in apparent total loss of circulating plasma melatonin levels.
T A B L E 1 Plasma melatonin levels (pg/mL) at mid-light and mid-dark in pineal intact and pinealectomised rats.Abbreviation: PX, pinealectomy.

Authors
a Subsequently Brown et al. 23 concluded that those results are invalid and that disappearance of melatonin from the circulation of undisturbed pinealectomized rats should be one criterion for a valid melatonin radioimmunoassay.
T A B L E 2 Plasma melatonin levels (pg/mL) at mid-light and mid-dark in pineal intact and pinealectomised sheep.Neuwelt & Lewy, 1983 [51]   Pineal tumour removal (n = 1) The first paper to suggest the presence of melatonin in the GIT investigated the effects of extracts of human appendix on the skin lightening response of a frog (species unknown). 66In the study, an alkaline extract applied to frog skin resulted in changes in melanophores that were characteristic of the response to melatonin.The authors further suggested that rabbit intestinal extract contained a substance that had similar mobility to melatonin on a thin layer chromatography plate.There was also evidence of serotonin in

Authors
Mean plasma melatonin levels in control (Con) and pinealectomised (Px) rats during the light and dark phase retrieved from 20 publications documented in Table 1 (A).Melatonin levels in human subjects who had their pineal glands removed (Px) or had denervated pineals (Dp) due to cervical spinal injury retrieved from 6 publications documented in Table 3 (B).Samples assayed by mass spectrometry are indicated by the triangle symbol ▽.
F I G U R E 2 Mean plasma melatonin levels in control (Con) and pinealectomised (Px) sheep during the light and dark phase retrieved from 6 publications documented in Table 2.The shaded region on the graphs indicates the reported sensitivities of the assays used.The true values are likely to be even lower than the plotted data points.
the extract.The data from this paper were duplicated in a subsequent paper. 67he first report of the cellular presence of melatonin in rat esophagus, stomach, intestine liver, spleen and pancreas was published by Bubenik et al. 68 The study used an anti-melatonin antibody 69 which had a 1.5% cross reaction with N-acetyl serotonin as determined by RIA (determined at a final dilution of 1:1625 10 ).Before use, the antiserum was diluted either 1:4 or 1:6 and saturated with an "appropriate" amount of BSA.As one test of the specificity of the staining the anti-melatonin serum was saturated with melatonin, but no results of the test were reported.Indeed, the only result shown was one figure said to demonstrate "specific" fluorescence in Lieberkuhn's crypts.The distribution of melatonin-like immunoreactivity was stated as higher in the stomach and duodenum, low in the jejunum and ileum and rising again toward the rectum corresponding to the distribution of serotonin-producing argentaffin cells.
The next report 70 involved the analysis of frozen cryostat sections of human appendices (removed during appendectomy and so presumably abnormal).The sections were incubated in several drops of an antimelatonin antiserum (diluted 1:4) generated according to a previous method. 69The specificity of the antiserum was determined by the addition of melatonin or N-acetyl serotonin.No data was shown confirming that the specificity of the antibody used was similar to that generated for the original publication 69 or that the "neutralising" of the antibody eliminated fluorescence.The authors stated that enterochromaffine cells, localised at the basal sections of the mucosal crypt of the human appendix in the majority of cases possess intense yellow luminescence after treatment with antiserum for melatonin.A single figure showing just 7 cells was published.
Holloway et al. 71 next investigated the presence of melatonin in rat colon.Sections prepared similarly to the above papers were covered with 1-2 drops melatonin antibody (diluted 1:10).The antibody used was generated according to a previous method 72 and had significant cross reactivity with 6-hydroxymelatonin (3.2%), N-acetyl serotonin (1.5%) and N-acetyl tryptamine (0.8%).Specificity of the immunohistochemistry procedure was tested using anti-melatonin serum saturated with melatonin (100 mg/mL).Intense fluorescence was apparent in the epithelial layer of the colon but was absent from the submucosa and muscularis.Treatment of tissue sections with normal sheep serum, phosphate buffered saline, or antiserum passed through an affinity column, produced low levels of fluorescence.Saturation of antiserum with melatonin did not eliminate fluorescence.The failure of this control greatly weakens the conclusion that "Melatonin-specific fluorescence was present in the epithelial layer of the rat colon." Bubenik 73 extended his previous work 68 by studying the effects of post natal development (d16 to d30), time of day, melatonin treatment and pinealectomy on melatonin immunoreactivity in the rat intestine using a melatonin antibody developed by Pang et al. 10 No specificity controls were reported.The author stated that "unfortunately we were not able to perform a precise localisation of melatonin in the tissue of GIT.It seems that to a certain extent, melatonin was marginally displaced during the histological procedure.However, because of the rather quantitative purpose of this study, we did not spend the extra time to modify our technique to pinpoint the cellular localisation of N-acetyl indoles." Menendez-Pelaez et al. 74 investigated the nuclear localisation of melatonin in the Harderian gland, liver, gut, kidney, spleen, cerebellum, cortex and pineal gland of rats by immunohistochemistry using an antiserum from Guildhay (G/S/704-8483)."As controls, antigenabsorbed antisera and sheep serum were used.All control sections were consistently devoid of any staining as a result of endogenous biotin."All tissues studied revealed melatonin-like immunoreactivity which appeared preferentially located in the cell nucleus.
Anisimov et al. 75 investigated the presence of melatonin in tumour-free tissues from glandular stomach, duodenum, ileum and descending colon from rats.In contrast to previous studies, the authors used deparaffined and dehydrated tissues.Slides were incubated with "ultraspecific" rabbit polyclonal antimelatonin antiserum (CIDtech Research Inc., Mississauga, Canada, 1:100) which may have been the antibody reported by Brown et al. 23 No specificity controls were reported.Melatonin-like immunoreactivity was detected in stomach, duodenum, colon and ileum in the control rats.
In a recent study by Soderquist et al., 76 formalin fixed human tissue specimens were incubated with rabbit antimelatonin (0100-0203 AbD Serotec, diluted 1/500).No specificity data is available for this discontinued antibody.A neutralisation test was performed using melatonin (0.1 mg/mL) and serotonin (0.1 mg/mL).Staining with melatonin antibody was completely blocked by preincubation with the antigen.Preincubation with serotonin, partially blocked the melatonin immunostaining.Immunoreactive cells for melatonin were found throughout the GI tract and pancreas.Intense melatonin-like immunoreactivity was seen in enterochromaffin cells.Throughout the GI-tract distinct melatonin-like immunoreactivity was found in numerous enterochromaffin cells and in a majority of the endocrine cells in the pancreas.
In summary, the immunohistochemical evidence for melatonin in the GIT does not appear to be strong.In some studies, controls involving "saturation" of antibody binding sites with melatonin or serotonin failed to eliminate the signals, in others no controls were reported.Despite some of the antibodies having significant cross reaction with related indoles like N-acetyl serotonin, N-acetyl tryptophan or N-acetyl tryptamine by RIA these were not tested as controls in the tissue studies.

| GIT MELATONIN CONTENT IN RATS
In what appears to be the first attempt to quantitate melatonin in the gut, Bubenik 73 wrote that "The presence of melatonin in the gut was detected by radioimmunoassay (Tsui, personal commun.).However, this technique is not yet sensitive enough to determine the absolute levels of this hormone."This is despite the same RIA being used successfully to measure melatonin in rat serum 77 (Table 4).
In the first apparently successful study quantifying melatonin in gut tissue, Huether et al. 1 homogenised duodenal pieces in perchloric acid/EDTA/sodium bisulphite and directly injected the supernatant into their HPLC apparatus for electrochemical detection.In rats the melatonin concentration was determined to be 24 000 ± 2000 pg/g tissue (range: 4000-100 000 pg/g).Subsequently Brammer also used HPLC with fluorescence detection to measure melatonin in rat duodenum, but found that the melatonin content was less than their detection limit of 2500 pg/g. 78Moreover, there was no evidence of melatonin synthesis in the duodenum from N-acetyl serotonin either in vivo or in vitro and there was no melatonin synthesis from 5-hydroxy tryptophan or 5-methoxy tryptophan in vitro.
Bubenik and Brown 22 using an updated RIA 23 reported that the melatonin content of rat stomach, jejunum, ileum, and duodenum ranged from approximately 340-560 pg/g which was not reduced by pinealectomy.Zeman et al. 79 analysed melatonin in rat duodenum and colon by RIA 12 after chloroform extraction and concentration of melatonin in the duodenum of 250 pg/g.Stebelova et al. 80,81 in the same group published 2 papers on plasma melatonin assayed by direct RIA 12 and gut melatonin after chloroform extraction.In the 2006 paper, the duodenum melatonin concentration was 300-1000 pg/g, whereas in the 2010 paper the concentration was 90 pg/g.Pal et al. 82 incubated gastric tissue in medium for 2 h and then assayed melatonin using a direct melatonin ELISA kit (ER1169; Wuhan Fine Biotech Co.) and reported that the melatonin content of the duodenum, colon and stomach was approximately 500 pg/g for all 3 tissues.Finally, Zagajewski et al. 83 investigated the effects of admminstration of vehicle by gavage at 1000 h on gastric mucosa levels of melatonin assayed by an HPLC procedure.Melatonin was not detectable in either plasma or tissue (<3.9 pmol/mL or/ 100 mg; <904 pg/mL or/100 mg, respectivley).
Melatonin has been reported in gut tissue of other species (Table 5).In sheep, Pan et al. 84 homogenised gut tissue and assayed the supernatant by direct ELISA (Beijing Sinouk Institute of Biological Technology).No details of any validation studies of this assay were provided.Tissue melatonin contents of 3200-4200 pg/g were reported and plasma melatonin levels were consistently 4-50 pg/mL throughout the day.In another ruminant study Bubenik et al. 85 determined melatonin in the luminal fluid, mucosa, and muscularis of gut tissues of cattle by RIA after extraction. 23They concluded that T A B L E 4 Tissue and blood levels of melatonin in rats.

Authors
References the melatonin levels detected in the mucosa and muscularis of the bovine GIT substantially exceeded the levels observed in the peripheral blood.Melatonin was also detected in the gut of bovine fetuses using the same methodology 86 Bubenik and colleagues conducted several studies on gut melatonin in pigs. 85,87,88The melatonin content was measured by their RIA and varied from 25 to 875 pg/g across studies and tissues.While in 2 studies, the daytime plasma melatonin levels were <5 pg/mL, in another the daytime melatonin was an extraordinary 98 pg/mL. 88here have been 2 studies conducted on mice. 89,90In the first study gut tissue from Swiss-Webster mice were assayed by RIA. 23Melatonin content varied from 116 to 580 pg/g.The study on C57Bl/6 mice which are known to be melatonin deficient 48 used liquid chromatography coupled with tandem mass spectrometry with positive electrospray ionisation to analyse melatonin in gut tissue. 90Colon melatonin was 0.45 pg/g while plasma melatonin was 0.16 pg/mL.
Finally, there has been one paper addressing human gut melatonin. 91Plasma and tissue were obtained from patients undergoing diagnostic or therapeutic endoscopic retrograde cholangiopancreatography between 8 and 12 a.m.Samples were extracted with chloroform and assayed by RIA using Stockgrand reagents.The concentration of melatonin measured in plasma was 55 ± 38 pg/mL, while tissue melatonin levels were more than 1600 pg/g with even the liver reported to have 800 pg/g melatonin.
To summarise the results of the rat studies, there is no consensus on the actual amounts of melatonin in gut tissue, with values varying by 266-fold between studies.Of obvious concern is that in the papers that reported melatonin levels in blood collected during the light period as well as tissue, the melatonin levels were vastly in excess of levels determined by the gold standard mass spectrometry assay.We might conclude then that the tissue values are also not reliable.In other species there is considerable variability within and between species and even within the same laboratories.

| STIMULATION OF MELATONIN PRODUCTION BY EITHER 5-HYDROXY TRYPTOPHAN OR TRYPTOPHAN
As previously mentioned, it was presumed that the plasma melatonin levels measured during the daytime and which often appeared to persist after pinealectomy, originated from extra pineal gland sources.The intestines were first proposed to be the major source.If this were correct, then it was argued that administering the precursors of Note: Plasma melatonin is pg/mL; tissue melatonin is pg/g.melatonin, 5-hydroxy tryptophan and tryptophan might be expected to drive melatonin production in the GIT and for plasma melatonin to also increase.Huether et al. 1 were the first to investigate this possibility.When rats were injected with 150 mg/kg tryptophan, melatonin in the duodenum increased from 24 ± 2 ng/g (n = 3) to 350 ± 87 ng/g (n = 3).Interestingly there was also a 6-fold increase in pineal gland melatonin, despite presumed low levels of AANAT.Daytime serum melatonin levels increased from 35 ± 7 pg/mL to 93 ± 25 pg/mL after 1 h.Oral tryptophan administration resulted in 158 pg/mL melatonin in the serum, while 200 mg/kg intra peritoneal tryptophan administered to pinealectomised rats increased melatonin from 49 to 212 pg/mL.The assay used in this study was a modified direct scintillation-proximity radioimmunoassay 16 that used Stockgrand melatonin antisera (GS 704-6483).Yaga et al. 17 administered tryptophan (300 mg/kg) orally to rats and collected blood 1 h later for assay by direct RIA 12 using a different Stockgrand antiserum, G/S/704-8483.Following the tryptophan treatment, plasma melatonin increased from 26.3 ± 4.6 pg/mL to 89.7 ± 6.2 pg/mL in ad libitum fed animals.The melatonin levels in pinealectomised rats changed from approximately 18 to 45 pg/mL.In this study, there was no significant change in pineal gland melatonin content.Sanchez et al. 49 treated rats for 7 consecutive days with a daily oral dose of saline or 125 mg/kg of tryptophan 1 h before lights off and collected blood at regular intervals across the day and night at the end of the experiment.The plasma was assayed using a commercial melatonin kit (ITISA Biomédica), but no details of this assayed can be found.The authors concluded that the tryptophan treatment increased melatonin levels over the 24 h period.Daytime plasma melatonin levels increased from 50 to 100 pg/mL across the light period in both groups, while an effect of tryptophan was only really evident at one time point, 4 h after lights off.Zagajewski et al. 83 investigated the effects of admminstration of tryptophan (50 mg/kg) or vehicle by gavage in rats at 1000 h on serum and gastric mucosa levels of tryptophan, serotonin, and melatonin assayed by HPLC.The administration of tryptophan resulted in a 3-fold increase in the content of tryptophan in both gastric mucosal tissue and serum, compared to the vehicle within 15 min.The concentration of serotonin increased by approximately 50% in tissue and 3-fold in the serum.Administration of tryptophan increased the mean concentration of melatonin from <3.9 pmol/ 100 mg to 2500 ± 20 pmol/100 mg in tissue and from <3.9 to 100 ± 9 pmol/mL in serum.
There have been 2 reports of the effects of melatonin precursor administration on plasma melatonin in sheep. 84,92In the first study, sheep were injected intra peritoneally with tryptophan (500 mg/kg) and plasma melatonin was assayed by RIA following solvent extraction. 20There was no consistent change in melatonin when tryptophan was administered either in the day or night.By contrast, administration of 5-hydroxy tryptophan (20 or 200 mg/kg) increased melatonin 7-and 20fold during the daylight but not at night.Infusion of tryptophan (up to 50 mg/sheep per hour) into the duodenum of sheep for 6 h for 7 days increased plasma 5-hydroxy tryptophan from 200 ng/mL to approximately 240 ng/mL 84 andlasma melatonin from approximately 45 to 55 pg/mL.Melatonin was assayed by a direct ELISA (Beijing Sinouk Institute of Biological Technology); no validation or other information is available for this kit.
There have been several human studies in this area.When 5-hydroxy tryptophan (10-12 mg/kg) was administered orally to adults and plasma melatonin assayed by RIA, 93 there was no significant plasma melatonin response up to 6 h after administration. 94When 5-hydroxy tryptophan (60 mg) was infused into healthy control subjects, 95 plasma melatonin increased from 15 to 18 pg/mL after 90 min.Plasma melatonin was assayed by RIA with 125 I-melatonin as tracer (Euro-Diagnostics).No details about this kit can be found.
Following an intravenous infusion of 5 g tryptophan during daytime., 96 plasma melatonin measured by direct radioimmunoassay with the GS 704-6483 antiserum from Guildhay Antisera rose from 15 to 234 pg/ mL within 60 min.When tryptophan (50 mg/kg) was infused into healthy subjects 97 starting at 0800 h, 1000 h, or 1200 h and melatonin assayed by direct RIA, 12 plasma melatonin was 124 ± 59 pg/mL, 98 ± 76 pg/mL, and 95 ± 25 pg/mL at 0900 h, 1100 h, and 1300 h, respectively.Inexplicably the melatonin levels following the placebo infusion were not presented.The infusion of tryptophan (5 g) over a 40-min period resulted in peak levels of plasma tryptophan (265 200 000 pg/mL) after 60 min. 98lasma N-acetyl tryptophan levels peaked at 676 500 pg/ mL.Plasma melatonin measured by a direct Scintillation-Proximity radioimmunoassay 16 peaked at 290 pg/mL.

| ASSESSMENT OF THE METHODOLOGY
The vast majority of studies investigating the presence of melatonin in the GIT have used immunochemical methods, either immunohistochemistry, RIA or ELISA.Natural melatonin will not provoke an immune response to produce antibodies.To generate an antibody that will bind to melatonin, melatonin needs to be conjugated to a larger molecule that is immunogenic and rabbits, sheep or goats immunised with the conjugate.Usually a small molecule like melatonin is chemically reacted with a protein such as bovine serum albumin or thyroglobulin.When the immunogen is injected, the protein with its attached melatonin in this case is broken down into individual short epitopes and presented to B cells.These cells produce antibodies that are specific to those epitopes.Thus we can expect that a huge number of antibodies might be produced, some that recognise small sections of the protein and some that recognise sections of protein that have the melatonin attached.How the melatonin is chemically attached to the carrier protein is critical for determining specificity; and for such a small molecule as melatonin this has presented a problem since there are few sites on the molecule that are chemically "accessible."The various immunogens that have been used to create antibodies against melatonin for use in assays has been reviewed previously. 63Depending on how the melatonin was conjugated to its protein will determine what other compounds the subsequent antibody may bind.
In the case of immunohistochemistry of a small compound like melatonin, there are major potential difficulties.First of course is the expectation that the levels of the hormone are likely to be low and so high affinity antibodies will be required to detect the melatonin.The early studies reported above 68,70,71 used antiserum diluted between 1:4 to 1:10 and assumed that the cross reaction profile would be the same as that used in an RIA at much higher dilution.Even at these higher dilutions the antibodies were not specific as determined by use of melatonin saturation experiments.No paper ever determined whether incubation of the antibody with the original immunogen or antibodies to the carrier proteins blocked the signal.
All of the antibodies used for the immunohistochemistry studies were used in RIAs to measure melatonin in plasma and tissues.It is an attractive prospect to be able to directly assay melatonin by simply adding an antibody and tracer directly to a sample of plasma, partition the free and bound hormone and measure either the bound or free tracer.Such a so-called direct assay relies entirely on the specificity of the antibody for accuracy.A direct plasma RIA was developed by Fraser et al. 12 for use in human circadian rhythm studies, measuring endogenous melatonin.The original assay used 3 H-melatonin as a tracer and as such the detection of the radioactivity required the addition of scintillation fluid to the sample.A key component of the assay was the preparation of the standards in charcoal stripped human plasma to ensure that at least the protein components of the samples and standards were similar.A modification of this approach (scintillation proximity assay) was subsequently developed that did not require centrifugation or any separation of free and bound tracer with the entire assay conducted in one tube 16 again for humans.
It can be seen from Table 1 that assays conducted on rat plasma using direct RIA in all but one case produced daytime melatonin levels in pineal intact and pinealectomised rats that were 10-100 times those obtained by mass spectrometry.Whether charcoal stripped rat plasma was used was not stated and full validation studies were not reported.Aside from possible cross reaction issues, non specific interactions and colour quenching of the photons generated by the scintillation fluid could all have resulted in false high values being recorded.
Even in the case of assays that involved extraction and reconstitution of samples in buffer before assay, some of the assays used for Table 1 produced results that are difficult to accept.For example 2 Bubenik papers reported on the changes in melatonin after food intake in pigs 87,99 and used the same radioimmunoassay reagents and technique. 23In Figure 3 the mean melatonin levels are shown from the papers.Several things are very obvious; F I G U R E 3 Mean plasma melatonin levels in pigs redrawn from 87,99 that were assayed by the same group, using the same methodology.It is not clear when samples were collected during light in panel (A) but was apparently between 6 and 20 h for panel (B).
in one study the fasting levels start low (4-5 pg/mL) and increased to 15 pg/mL after 3 h following re-feeding.In the other study mean melatonin levels ranged from 20 to 40 pg/ mL across 24 h.In a different pig study, the same authors reported morning plasma melatonin levels of 98.9 pg/mL. 88e really must question the validity of results produced for these and other studies from this group that used these radioimmunoassays since there is no internal consistency.As a reference, it should be noted that pigs do have a circadian rhythm of melatonin with daytime levels less than 1.5 pg/mL and peak values ranging from 12.2 to 35.9 pg/mL depending upon the season. 100t is important to evaluate the methodology behind the experiments that were designed to determine whether administration of tryptophan or 5-hydroxy tryptophan increased plasma melatonin through intestinal melatonin production.Little attention appears to have been given to the potential cross reaction of the administered amino acids with the antibodies used in the direct immunoassays.If we consider that the normal plasma tryptophan concentration in rats is approximately 100 µM or 100 000 000 pM, then a 0.0001% cross reaction would give a signal of 100 pM (23 pg/mL).According to the original assay description of the most widely used direct radioimmunoassay, 12 the cross reaction was 0.4% for 6-hydroxymelatonin, 0.35% for N-acetyl tryptamine and 0.03% for N-acetyl serotonin, with cross reactivity of the other compounds tested only stated to be <0.01%.Note that N-acetyl tryptamine has been measured in rat and human plasma in close to a 1:1 molar ratio to melatonin, 101 suggesting that this compound could contribute to plasma melatonin immunoreactivity especially after tryptophan administration.We do not know what other compounds were tested for the Fraser assay, but tryptophan administration results in a rapid increase in the levels of tryptophan, N-acetyl tryptophan, 5-hydroxy tryptophan and 5-methoxy tryptophan, 98 which also may contribute to the apparent melatonin signal.The above comments should not be taken as a criticism of the Guildhay/Stockgrand direct assay, which when used in humans according to the original methodology 12 has proved to be a very valuable tool.But when used in other species to assay plasma and tissue homogenates without proper validation and in circumstances where increased levels of tryptophan metabolites may appear, then the results must be questioned; there is a clear expectation that the melatonin levels will be over-estimated.The RIA used in early studies by the Bubenik group, was not a direct assay, but involved chloroform extraction of the plasma or tissue. 10n 1985 Brown et al. 23 published a modified RIA and wrote that the previous assay system detected measurable melatonin following pinealectomy and concluded that "those earlier results are invalid and that disappearance of melatonin from the circulation of undisturbed pinealectomized rats should be one criterion for a valid melatonin radioimmunoassay."They published an extensive table of cross reacting compounds, many of which had significant potential cross reaction but only if used without extraction.In the current context, however, one compound stands out as a problem; N-acetyl tryptamine, a potential metabolite of tryptophan following its administration showed a 0.11% cross reaction and is expected to be extracted into chloroform.As stated above, this compound is circulating naturally at levels similar to melatonin.

| PERPETUATION OF THE MYTH
In the original Huether et al. 1 paper on the melatonin content of the intestine the authors stated that "Pineals and gut (upper duodenum) were rapidly removed and processed."Two years later the same data was reported in a review as being from the small intestine. 102umerous reviewers have nevertheless to this day used the original results to declare that the gut contains 400 times that of the pineal gland.The first explicit statement that the total amount of melatonin present at any time in the GIT is more than 400 times higher than the amount of melatonin present in their pineal glands, appears to have been made in 1997. 22The Huether et al. 1 paper has a number of serious problems that have never been addressed.The key data of that paper is reproduced here in Table 6.
It is clear that the basis of the statement "400 times that of the pineal gland" is derived from 80 ng divided by 0.2 ng and 2000 ng divided by 4.5 ng which were the ranges measured in an unstated number of animals.It would be more accurate to use mean data rather than the ranges and the only appropriate data we have from the paper are the mean pineal and duodenum contents of 3 animals from their table; 580 ± 36 ng/g and 24 ± 2 ng/g for pineal and duodenum respectively.If we use the maximum pineal weight of 3 mg and the duodenum weight of 20 g as stated, the calculated contents will be 1.74 and 480 ng to give a ratio of 275 times (Table 7).But Huether et al. 1 overestimated the weight of the rat duodenum 20-fold!The rat duodenum actually weighs no more than 1 g. 103Recalculating the ratio, we get the duodenum containing 13.8 times the pineal gland content.From Table 4, we can see that since the seminal paper in 1992 the levels of melatonin in the gut reported in the literature have tended to decrease considerably with Stebelova et al. 81 reporting the lowest content at just 0.090 ng/g or 90 pg in the whole rat duodenum.This means that the duodenum:pineal ratio (90/1740) is just 0.05-fold!(Table 7) Given the caveats on the use of immunoassays to measure melatonin in tissue due to cross reaction and poor validation, we might expect the ratio to be even lower.
We must ask the question why authors have persisted in uncritically perpetuating the myth of the gut being a major extra-pineal gland source of melatonin, and supposedly producing enough to influence circulating plasma levels during the daytime?It is extraordinary that if Bubenik and Brown 22 in 1997 had calculated the duodenum:pineal ratio using their own results, they would have come up with a ratio of 0.19:1 and not 400:1 and perhaps raised doubts about the importance of gut melatonin.Of course this is equally relevant to all the other scientist who have worked in the field who have simply restated comments by others before them.The great physiologist, Claude Bernard put it very clearly more than 150 years ago; "Men who have excessive faith in their theories or ideas are not only ill prepared for making discoveries; they also make very poor observations.Of necessity, they observe with a preconceived idea, and when they devise an experiment, they can see, in its results, only a confirmation of their theory.In this way they distort observation and often neglect very important facts because they do not further their aim". 104n conclusion, after evaluating the primary evidence of studies investigating the gut as a major extra-pineal source of melatonin, it is clear that daytime circulating levels of melatonin in mammals are negligible when the appropriate methodology is used to measure the hormone.The use of poorly validated immunoassays especially during the 1980s and 1990s tended to "Duodenum" 24 000 ± 2000 (n = 3) ~1 g a 24 000 13.8 Huether, 1994 [102]   "Small intestine" 24 000 ± 2000 (n = 3) ~6 g 600 000 82.7 Bubenik & Brown, 1997 perpetuate the idea that pinealectomy does not eliminate plasma melatonin.Similarly, poorly validated immunohistochemistry techniques and the failure of authors to recognise cross reactivity of other indoles in immunoassays of tissue and plasma following tryptophan administration led to continuing support for sites of extra-pineal melatonin production.But most importantly, a single (flawed) study 1 whose extraordinarily high gut melatonin levels have never been replicated has continued to drive the field for 30 years, despite many researchers producing results that would have debunked the hypothesis.The gut is not a major extra-pineal source of melatonin; indeed it may well not produce any.

T A B L E 5
Tissue and daytime blood levels of melatonin in other species.

T A B L E 7
Calculations of the ratio of rat duodenum melatonin content to pineal gland melatonin content from the literature.
The approximate weight of the rat duodenum is 1 g and the small intestine 6 g. a If the Huether et al. 1 paper actually used only duodenum, then the ratio would be approximately 14-fold.The melatonin concentrations are the maxima, minima and means reported by Huether et al. 1 and the mean values for the Bubenik & Brown, 22 and Stebelova et al. 81 studies.
1he original data from the seminal paper.1 Note: Values are ranges between minimum and maximum concentrations of melatonin measured in all control animals used throughout this study.