A REVIEW OF THE CHEMISTRY AND BIOCHEMISTRY OF SEED SHOOT FLOUR AND FRUIT PULP OF THE PALMYRAH PALM (BORASSUS FLABELLIFER L) E.R.JANSZ*,

This review with 74 references, focuses on the chemistry and biochemistry of palmyrah seed shoot and palmyrah fruit pulp. While palmyrah fruit pulp is highly underutilized due to its bitterness, there is so far no evidence of its toxicity. On the other hand, palmyrah flour, which has been consumed for centuries, has many reported toxic effects, viz., neurotoxicity, hepatotoxicity, immunosuppression, clastogenic and mutagenic effects. This review discusses the abovs in relation to studies directed a t utilization of palmyrah fruit pulp based on its sugar, pectin and carotenoid content and the effect a group of steroidal saponins termed flabelliferins will have on utilization. The structural studies and bioactivities of these flabelliferins are discussed. Also discussed is the reported toxicity of odiyai flour.


INTRODUCTION
The palmyrah palm (Borassus flabellifer L) is widespread in the arid tropics of South America, East Africa, India, Sri Lanka and South-East Asia.I It is a feature of the landscape of North-East Sri Lanka where it is called the tree of life. It is estimated that at present there are about 11-12 million trees in Sri-Lanka,2.3 with about 3.5 million each in the Jaffna peninsula and the Wanni districts of Killinochchi and M~llaitivu.~ In Jaffna, palmyrah represents a major socio-economic factor even at the present time as it is estimated that about 100,000 families are directly or indirectly employed in palmyrah-based industries.: From the point of industry, the sap products are most important, as the influorescence sap (sweet toddy) gives many products "4+5,6.7~"ncluding fresh toddy, fermented sap (toddy), arrack (distilled spirit), treacle, jaggery and palm sugar. In addition, many microbiological studies have been conducted on fermentation of lo However, these aspects have not been considered to be within the scope of this review, which is mainly chemical and biochemical. Further, any new finding of sap product research and development has not been documented in refereed journals. In addition, techniques have largely been adapted from the coconut industry where methods are well known.

Palmyrah fruit pulp
Overview: The fruit of palmyrah varies in size, colour and has been classified into more than 10 morphological types.15More recently, 4 distinct morphological fruit types have been described.16J7 A thick leathery pericarp encloses 1 to 3 (frequently 3) seeds embedded in fibre enmeshed with a yellow or more rarely orange fruit pulp.16J7 A thick viscous liquid called palmyrah fruit pulp (PFP)5 can be extracted with water (1:1 or 1:2 v/v) either manually or by a fruit pulp extractor.lS  reportedlo the free sterols (0.3%) stigmast-5en-3 P 01 (24a Et) and lanosterol, In a subsequent study sitosterol, sito-5en-3gol (24a Et) was identified as the only free sterol,lg while other workers claimed that there were no significant quantities of free steroL20 The latter conclusion was arrived at by a study of more than 10 PFPs from different locations.20

Composition of PFP
There is no controversy on saponin content, which has been reported to be in the range of 0.15 to 0.4 mg.100g1.10,20~21 Carbohydrates: The main digestible carbohydrates are simple sugars1° of which sucrose, glucose and fructose (6.6,3.5 and 3.4 g-100gi respectively) dominate. There are traces of rharnnose8 in PFP and 1.5g. 100gl oligosaccharides8 (unidentified). Pectin is present and reported as 4.4g.100g1 lo and 6.7g.100g1. 22 A branched glucan has also been reportedz3 but its anomeric configuration was not determined.
Carotenoids: Carotenoids were first reported as 3.2mg.100g-' but clearly varies considerably as does the colour of PFP. A range of 1-10mg.lOOgl zz and most recently 2-253mg.100g-' l6jZ0 was reported. The carotenoids were earlier assumed to be p carotene,1° however, spectra using a scanning spectrophotometer on a hexane extract showed that PFP most commonly had a h max of 422-428nm16 but occasionally a h max of 434nm and 437nm16p20 showing that carotenoids are mixtures and probably vary in composition. Examination of a pooled PFP of h max 427nm (the common type) separated by medium pressure liquid chromatography showed the presence of a mixture of 4 main c a r o t e n~i d s .~~,~~ They were a-carotene and p-zeacarotene (structurally pro-vitamin A)24,z5 and lycopene and zeta-carotene (non pro vitamin A).24,25 These carotenoids are'labile and easily subject to oxygenati~n.~" Food Preparations: The fruit pulp provides a number of traditional preparation^.^^ These include a type of chewing gum (pinnatu), jams, cordials, sauces, palumellows (cf marshmallow), toffees, and can be a component in ice cream, biscuits, fruit bars and other confecti~nary.~.~~ The major production data are shown in Table 2.
With a potential for 10-15 k tonnes per annum, it is clear that only a fraction of the PFP is utilized. This is mainly on account of a marketing problem due to the bitterness of PFP.27.28 While those with an acquired taste prefer bitterness, this attribute is a deterrent to wide utilization to those not familiar with PFP (PDB 1992, Personal Communication) The nature of the bitter compound was not known until less than a decade ago. Until this time suggestions had been that the causative molecule was an alkaloid or fatty acid. (Report GTZ, 1989, unpublished) The flabelliferins: This family of compounds first attracted attention when one of them was identified as the cause for b i t t e r n e s~.~~ Further studies showed that there was a range of at least 14 flabelliferins16, one of which was a n t i -m i~r o b i a l .~~.~~ These compounds were steroidal s a p o n i n~~~~~~ and the term flabelliferin was coined from the specific name fZal~eLLifer.~~ Several lines of study directed a t isolating and identifying flabelliferins as well as studies on bioactivity have since been reported. In addition, the flabelliferins appear to play a key role in the determination of the future modes of utilization of PFP.16,17s20p21 Separation techniques: Steroidal saponins of PFP were first reported by an extraction of a methanolic extract of dried PFP from Jaffna which had been successively extracted using petroleum ether (40-60°C) and chloroform in a Soxhlet extractor. Only a monoglucoside and monorhamnoside were reported.1° This was probably due to use of inadequate water in the extracting medium, which is needed to isolate the flabelliferins of longer carbohydrate moiety from PFP.27 Curiously in that study i0 no reference was made to the bitterness of PFP.1° The two monoglycosides were separated by pre-prepared TLC of an acetone extract.1° Eventually, the bitter flabelliferin (subsequently called F-11) was isolated using a methanolic extract of fresh fruit pulp from Kalpitiya. Removal of fat solubles was by petroleum ether and sugars by dry cellulose chromatography and acetone extraction. The flabelliferins crystallized as needles from a n acetone extract.27 Two flabelliferins F-I (a tetraglucoside M.W. 1062) and F-I1 a (tetraglycoside,M.R7. 1030) were isolated.27 F-I1 was an intensely bitter flabelliferin.27 This technique was not good enough to separate more complex flabelliferin mixtures and as would be seen later, flabelliferins from different fruits gave vastly different pr~files.l~.l'.~~ In another s t~d y~l ,~~ using PFP from Hambantota, 4 flabelliferins were isolated by incorporating a flash chromatography step at the end of the isolation procedure. The flabelliferins isolated were F-I1 (bitter, tetraglycoside) F, & F, (triglycosides) and F, (diglycoside) all with rhamnose (rha) terminii.21,30 As flabelliferin profiles increased to show up to 14 16,17 in certain PFPs, separating them became more difficult. Various techniques were used to achieve separation. These included, selective solvent extraction (which is a good technique for separating small carbohydrate moiety flabelliferins from large oneslS), solvent gradient ~hromatography,'~2~~~~ the c h r o m a t~t r o n ,~~~~ (which is a good separation technique for F-11) and MPLC which is the most economic, time saving and efficient method which is applicable to nearly all different types of flabelliferin profiles.19~20~21,22~33,34 By this technique using a gradient of Hexane CH,Cl, + ethyl acetate methanol it has been possible to separate 5 flabelliferins including F,(triglycoside) F,(diglycoside) F, (diglucoside) F, (monoglucoside) and F, (MW 884,l rha, 2 glc).
With 3 earlier isolates, this brought the total number of flabelliferins isolated to 8, at least another 6 remain to be isolated.16J7 A new technique involving direct separation of carotenoids, sugars and flabelliferins without the use of working up techniques has been reported.35 Structural studies oh flabelliferins (a) The aglycone: Isolated flabelliferins have been hydrolysed by enzymes21 or trifluoroacetic a~i d~~J~~~~~~~ to produce a n aglycone (sapogenin). Direct MS analysis and GCIMS a n a l y s i~~~p~~ of a silylated derivitive has shown its molecular weight to be 414. After this point there has been controversy, the aglycone was first identified as spirost-5en-3P 01, then as stigmast -5en -3 P 01 (24 a Et)20 then as sitoster01~~ and finally unambiguously as P-  (c) Flabelliferin 11 (F-II): This is the bitter flabelliferin, MW 1030 with a rhamnose (rha) terminus.27 It has been shown to have 2 glc and two rha.27Hydrolysis of F-I1 with naringinase yields rha31 a flabelliferin F, (MW 868) shown to be identical with another flabelliferin F,, 21 and an a g l y~o n e .~~"~ F-I1 was hydrolyzed by a heat stable a amylase (specific for a-1-4 and to a lesser extent a-1-6) 21727, from which a tentative structure of the carbohydrate moiety can be deduced to be as follows. A further controversy has been reported. Namely that this may not be the only bitter compound in PFP as only 40% of the bitter PFP is hydrolyzed by aa m y l a~e .~~J~ This seems to show that there are two bitter compounds: (1) very bitter hydrolyzed by both naringinase 2133738 and a-amylase and a less bitter flabelliferin not susceptible to hydrolysis by a-amylase (no Glc a Glc bond). Although both have similar R, clearly the sugar sequence (in a linear chain) is different. A similar bitter compound has been detected in palmyrah (d) Flabelliferin B (FB): This is the antimicrobial flabellifri11,2~ MW 86g3O with 2 rha and 1 glc (rha t e r m i n u~) .~~ Methylation analysis giving alditol acetates has shown that the structure is b r a n~h e d~~s~~ and the linkages of the rha to the P glucose connected to p sitosterol are a-1,2 and a-l,4. The coupling constants and chemical shifts in lH nmr gave the anomeric configuration^.^^^^^ The structure of the carbohydrate moiety is therefore as follows ( Figure 3) Rha a-1,2

Figure 3: F,
This branched structure is the only one of this type so far elucidated of all the flabelliferins. Branching was confirmed by the detection of 2 diglycosides and one monoglycoside on exposing the F, to a mixed culture of y e a~t .~~,~~ The scheme given in Figure 4, shows that only a branched carbohydrate moiety can produce the products formed.

Figure 4: Proof of Branched Structure of F,
A problem that arose is that the treatment of F, with naringinase (a b glucosidase and a fi rharnnosidase) yielded rha and one diglycoside,2* which indicated a p rha. This can be explained by the presence of two flabelliferins occurring in different samples: one having 2 a rha links and the other 1 rha a and the other P.
Thus showing that there is yet much to be revealed in flabelliferin structures.

Figure 5: Alternative structures of F,
(dl Flabelliferin C: Not much is known about this structure, as it has not been subject to sugar sequencing. By FAB/MS, its molecular weight is identical to F, (868) with a rha terminus 30but it is not antimi~robial.~~ It is not susceptible to naringinase and a-amylase hydr~lysis.~~ It, therefore, has no P bonds and no a-1,4 or a-1,6 glucose bonds. There is no evidence of a M-162 fragmentation peak on FAB/MS analysis and therefore does not have a glucose terminus branch. Its most likely sequences are as given in Figure 5.
(dl Flabelliferin D: This was shown to be a diglycoside with 1 rha (terminal) and lglc with a MW of 722 by FAB/MS analy~is,3~ This was supported by GC/EI/MS analysis which also gave a MW of 722.34 Chemical shifts and coupling constants with lHnmr showed that glc was in (3 configuration and rha was a and methylation analysis showed a 1,4 bond.34 Its structure is therefore as given in Figure 6 (3 sitosterol-glc -rha Traditionally, this is done by heating the palmyrah fruit (proximal end upward) on hot coals. Froth emerges which wheil removed takes with it part of the bitterness (PDB, personal communicatiori) The only scientific method of debittering has so far been enzymatic, using naringinase (p glycosidase and P rhamnosidase activity) or heat stable a- amylase. 20,21927,39940925 AS indicated in section 2.4.2 there is one very bitter compound hydrolysable by both enzymes3s and a less bitter compound hydrolyzed only by n a r i n g i n a~e .~~ Debittering is a key step for more widely utilizing PFP in the form of jams and cordials.
It should be noted that debittering also leads to hydrolysis of some other non-bitter flab el lifer in^.^^^^^ Antimicrobial effect: An inhibitor was found to inhibit mixed batches of yeast culturem41 Inhibitors were also detected in PFP (whole) that inhibit a number of bacteria and fungi. LD,, value lmgml-' was as follows; Pseudomonas aeruginosa (0.02) Klebsiella pneumoniae (0.13), Escherichia coli (0.03) Bacillus licheniformis (0.13) Saccharomyces cervisiae (0.06) Aspergillus oryzae (0.26). Earlier the antibacterial and anti-yeast growth effect was shown to be due to F,. Both gram-positive and gram-negative bacteria inhibited at about 35pg.disk-l by the Bauer-Kirby method.
This sensitivity is comparable to Ampicillin. The bacterial strains inhibited were: -Staphylococcus aureus, Staphylococcus epidermidis, Escherichia coli, Pseudomonas aeruginosa, Proteus rettigeri, Aciinetobacter calcoaceticus var Lowffi~.21.29 This activity was lost when PFP was exposed to naringinase hydrolysis due to the destruction of FB.21939 It has been proposed that this antibacterial effect could be of use in an ointment for topical applications to heal sores and ulcers.43 Knowledge of the structure of F,33 wou1.d be of advantage as would the fact that some of the bacteria mentioned above, have strains that are pathogenic and resistant to many common antibiotics. Further, growth of Saccharomyces cervisiae strain S, , F3 was inhibited by F, a t 60pg ml-l and F-11 a t 250 60yg ml-1.21929A mixed culture of baker's yeast was also inhibited.4'This provides a basis for the observation that PFP does not spoil easily.
Effect of weight gain by mice: Feeding trials using ICR mice showed that feed containing 10% PFP caused a statistically significant weight loss,44 compared to control WHO breeding feed (p=0.023) despite feeds being iso-caloric and feed intake being similar. The effect was reversed by n a r i n g i n a~e .~~ However, since naringinase hydrolyses both F11 and F,, the results were ambiguous. Selectingvarieties of PFP containing F-11 and no F, (bitter) and F, and no F-11 (non-bitter), showed45 that a t 10% PFP it was only the bitter PFP that reduced weight gain (p=0.014) compared to control. The non-bitter sample gave increased weight gain compared to control (p=0.8~10-~), showing that if F-11 was not present, PFP was probably a nutritious feed for mice. The probable mechanism of action of F-11 was inhibition of the Na+/K+ ATPase. Only F-11 among flabelliferins inhibited the ATPase of ghost red blood ~e l l s .~~T h u s , it is postulated that F-11 reduces glucose uptake, while food intake is unaffected. This effect can have implications in lowering the glycaemic index of food; a situation, which can be advantageous to diabetics.
Some indicators on biosynthesis offlabelliferins: No biosynthetic enzymes have been isolated. Speculations can be made by considering the structure of the flabelliferins where structures have been e l u~i d a t e d .~~~~~,~~ The only sugars in the carbohydrate moiety are rhamnose and glucose.27 Each can be a or p.20,21 There appears to be no specificity for the acceptor molecules in carbohydrate chain lengthening.20 Even though the maximum number of residues in the carbohydrate chain has so far been found to be 4, the potential for varied structures is mind-boggling. It is possible to conclude that since rha is a minor component in PFP,8 that the corresponding nucleoside diphosphotransferase must be very active and 1 or with a low Km value.
Three other such transferases are required to explain the structures so far elucidated.

Utilization of PFP
Diversity offlabelliferin profiles: The key to the wider utilization of PFP is the nature of its flabelliferin profile. PFP can be bitter, containing F-11 and detract from food use.27 It can be an anti-yeast -growth f a c t~r~~,~% n d more importantly retard f e r m e n t a t i~n .~~~~~~~~,~~ The matter is complicated by the existence of a number of flabelliferins (at least 14). On some of these will depend the strategy of utilization. Utilization could have been benefited if an easily distinguishable morpholagical character was correlated to flabelliferin profile. However, none of the characteristics studied viz., size, colour of fruit, colour of pulp were of use for such correlations.16 Although F-11 could be detected by t a~t e '~.~~ and F, by fermentati~nl'.~~ neither property can find use in a commercial or field scenario.
There appeared to be some correlation between location and flabelliferin profile but the extent of sampling in the study was ins~fficient.'~.~~ Fermentation ofPFP: Fermentation of PFP has been r e~0 r t e d . l~~~~ There was a strong correlation in lower rates of fermentation and F, content. Both F, and F-11 produce lag in f e r m e n t a t i~n~l *~~ at concentrations of 1mg.ml-I. However, fermentation efficiency in all cases, bar one, was good, 85-95%. The sample which did not ferment was collected from Polonnaruwa and had a t least 10 flabelliferin~.l~*~~ This observation was not possible to explain from F, content alone.
It had been previously shown that exposure to naringinase increased the rate of ferrnentati~n.~~ This was not surprising as naringinase hydrolyzed F, .21,25,31 PFP has also been used for citric acid fermentati~n.~~ Utilization of palmyrah constituents: The useful constituents of PFP had been described as far back as 1967.4s Pectin can be isolated from PFP by the calcium salt on alcohol precipitation of acid e x t r a~t .~~~~~ Pectin was reported to have an acetyl value of 3.5 and methyl value of 5.4 and had good gel strength. 22,49 This is of value as commercial food pectin. Alternately, PFP can be depectinized by p e c t i n a~e~~,~~ to yield a clear juice on filtrati~n.~' This has been fermented to give a bitter wine (Personal communication, PDB). Fermentation and distillation with or without pectinase action will yield a waste-containing, colour and flabelliferins. Without pectinase action the texture, foam characteristics, colour and anti-microbial properties of PFP have been made use of as toothpaste after fermentation of sugar.43 The colour could have value only as a food colouring since carotenoids are oxidised during distillation of ethanoLZ4 The most feasible commercial use is aerobic fermentation, where the presence of 14-16% sugarlo makes the yield of alcohol sufficiently high to be viable. Furthermore, in the scenario of palmyrah sap and PFP yielding sugar substrates during different times of the year that are complementary, it will be possible to run a distillery all year r o~n d . l~,~~

Conclusions on utilization of PFP
* PFP can be utilized as a traditional p r o d~c t .~~.~~ * Most PFP (90%) can be used as a fermentation b a~e , '~.~~ hydrolysis of FB will increase rate of f e r m e n t a t i~n .~~ * Sweet PFP16 can be used directly for jams and cordials.
* About 40% of bitter PFPs can be debittered with a cheap heat stable a-amylase for food use.38 * Less bitter PFP can be utilized traditionally.
* PFP components can be separated and utilized.

Palmyrah seed shoot
Overview: The palmyrah fruit contains 1, 2, or 3 seeds (frequently 3). The seed kernel contains a g a l a c t~m a n n a n .~~~~ On germination the seed produces a shoot (fleshy food storage scales) which gives rise to the product kottaikilangu, on boiling and drying or on drying alone.8 This shoot grows to 12-15 cm in height before it is . harvested.52 If the seed shoot is sun dried it is referred to as odiyal. The shoot may be boiled and dried in which case it is called pluk~diyal.~ Odiyal and plukodiyal may be ground and sieved to give palmyrah flour (odiyal flour or plukodiyal flour) The flour can be made into a number of foods which are used traditionally. The odiyal is usually consumed as a porridge called Khool and a steamed product called pittu. (PDB, personal communication) Plukodiyal is generally converted into oil cakes, and a product called palmyrah cunchy which is a type of biscuit made by mixing with wheat flour, sugar and margarine rolled to a pastry, cut and baked in an oven.8 Also made out of plukodiyal is a product called palmposha which again has about 70% palmyrah flour, roasted green gram, rice, soya, ~ugar,.etc.~

Composition ofpalmyrah flour
There have been only two major reports on the ~u b j e c t . '~,~~ These findings are summarized in Table 3 along with some other analytical data. With regard to major nutrients and energy value, odiyal flour is comparable to common flours like rice and wheat flour. 52 The main carbohydrate is starches The starch has a low viscosity and gelatinization temperature but a good setting property exhibiting unique properties as a food ~t a r c h .~ The starch is devoid of bitterness inherent in palmyrah shoot and has a large grain size of about 40p.m, similar to potato starch.54 Sugar content is low (< 1%) comprising: sucrose, glucose, and fructose.1° Odiyal has considerable fibre and should lead to a low glycaemic index. The metal ion content of odiyal flour is given in Table 5. Sodium and calcium content is higher than normal cereal flours, otherwise metallic ion content is normal.

Toxic effects
Neurotoxicity: This is by far the most widely studied toxic effect. The/symptoms of neurotoxicity were first reported in 1971.55 The authors had interpreted the symptoms to be a secondary effect of hepatotoxicity. Wistar albino rats fed on a diet of kottaikilangu showed within 4-5 days symptoms of ruffled coat and apathetic behavior. They stopped taking food a t that stage and on the 6-7 day showed nodding movement of the head, uncoordinated spasms of forelimbs and falling over backward. This progressed to ataxia, immobility of hind limbs, followed by total immobility, laboured respiration and finally death within10 days. All 6 batches of kottaikilangu tested gave the symptoms.55 Nearly a decade later, another group of worked6 highlighted neurotoxicity being the cause for symptoms described previously. In addition they observed malaise, upright posture and salivation as early symptoms. In the study, unlike in the previous study, diet was clearly specified, being a 1:l mixture of palmyrah flour and MRC 41 B diet. Feeding aqueous extracts by oesophagal tube to Wistar rats and the brine shrimp assay monitored the toxicity of the extract and its fractions. The studP6 reported a 400-fold enrichment of the neurotoxin by methanol: water (19) extraction and removal of impurities by lead acetate precipitation. The toxin was purified using absorption on a strong cation exchange resin and elution with aqueous ammonia. Using Sephadex G-25, the molecular weight of the toxin was estimated as 1400. The toxin was reported56 to be stable in neutral alkaline medium and was hydrophilic in nature. The authors56 concluded that the toxin had no COOH group but had an ionized quaternary N group. They56 postulated that the delay of onset of symptoms was possibly due to a necessity for the toxin to undergo metabolic transformation for absorption by the intestine or by the nervous system (the latter seems more likely).
This introduces the concept of a pro-toxin (cf pro-drugj rather than a neurotoxin in palmyrah flour. This may give a clue to why neurotoxicity is not reported in major palmyrah flour consumers among humans.
Recent studies57 have shown that depending on the source of odiyal flour, the time taken for appearance of neurotoxic symptoms can change, viz., Jaffna>Mannar>Kalpitiya.
That is, the Kalpitiya flour showed the most severe neurotoxicity. Further, wet heat (boiling and plukodiyal preparation) did not destroy Dry heating of odiyal (8O0C, 45 Min) destroyed Although the toxin is very soluble in water, washing odiyal did not remove toxicity, but rather reduced it.57 Steaming as in the case of pittu preparation did not affect The previous56 conclusions that the toxic compound was a quaternary ammonium salt containing a glycoside was not borne out as the to.xic fraction was very water soluble and gave a ninhydrin positive reaction, pointing to a primary amine. Its chrornatogaphic behaviour also pointed out to the possibility of it being an amine. 5s However, the conclusion that the toxin is a neurotoxin and not a hepatotoxin was substantiated by the fact that the neurotoxic symptoms paralleled the elevation of serum aspartate aminotransferase (AST) and not serum alanine aminotransferase CALT).53 This is because ALT is diagnostic of liver disease, and a n elevation ofAST without increase of ALT is consistent with neurological damage.
It is possible, therefore, that the hepatotoxic symptoms observed in 197155 and 197659 have no relationship with the neurotoxic effect. The fact that subsequent s t~d i e s~~,~~ had not shown hepatotoxicity indicates that the two causative factors are different.
Regarding the nature of the neurotoxin, it has become clear that the MW 1400 isolate56 is an adduct or mixture of an amine and flabelliferi~~~~"~ (contains a glycoside). What is not clear is whether some sort of synergism exists as the mixture gives rise to more toxic effects than the purified amine.58 It is also possible that the amine in the absence of the flabelliferin is lost to some extent during purification. This is possible as amines not containing a strong ionic group can be volatile.
The hepatotoxic effect: The symptoms of neurotoxicity described above were first ascribed to he pa tot ox in^.^^ Lines of evidence for a hepatotoxin were two fold, Firstly, in-vitro studies on liver mitochondria1 enzymes and secondly histopathological stUdies.65,59t60 The in-vitro studies involved succinic dehydrogenase and "succinic oxidase" .
The latter name is now not used and is usually synonymous with succinic dehydrogenase. The source of enzymes was not mentioned, the experimental design did not introduce cofactors and the interpretation of results of 0, uptake by Warburg respirometer was open to question. Therefore, the importance of these in-vitro studies is difficult to assess.55 Histopathalogical evidence was ~t r o n g ;~~"~ this included the observation of swollen liver mitochondria, veno-occlusive l e s i o n~, 6~~~~ hydrotropic degeneration of the centrilobular cells and fatty degeneration of the centrilobular and periportal cell~.~~However, the study reported that there was no parallel between the intensity of mitochondrial disfunction as evidenced by lysosomal acid phosphatase, malic dehydrogenase and a-glycerophosphate dehydrogenase activities with the extent of lesions observed in the liver.55 Histopathological studies also showed minimal oedema of the cerebral cortex and congestion of the kidney.55 A subsequent study by the same group confirmed veno-occlusive lesions of the liver of rats after prolonged feeding with palmyrah flour.59 Chronic hepatic lesions included intra-luminal fibrosis of the centrolobular veins, bile duct proliferation and increase in reticular fibrosis.59 No thrombosis and hepatic megalocytosis was seen.59 The study concluded that the toxic factors responsible are different from pyrrolizine alkaloids and dimethylnitrosoamine, which produce similar lesions. 55759 In another study by the same pregnant Wistar rats immediately after parturition were fed on loo%, 50% and 33% palmyrah flour. Studies on 44 suckling rats showed that the infants did not show neurotoxic symptoms but significant neonatal deaths occurred. When the maternal rats were fed on loo%, 50% and 33% palmyrah flour, the sucklings died in 1-7 days and 2 weeks. Survival was therefore dose dependant.
HistopathalogyG0 of neonate livers showed that the livers had dark patches. Internal examinations showed free pleural fluid and peritoneal fluid and enlarged and pale kidneys. External abnormalities included subcutaneous hemorrhages a t extremities of tail and limbs and in thoracic abdominal walls.60 The liver showed gross dilation of the sinusoids, foecal hemorrhages and mild to intense centrilobular sinusoidal c o n g e~t i o n .~~ The authors concluded that the hepatotoxin of palmyrah flour could be transferred by milk to produce toxicity in suckling rats.60 However, another group of researchers found no evidence of a hepatotoxic effect of palmyrah flour, although neurotoxicity was observed. They concluded that hepatotoxin and neurotoxin were different entities.

Recent
showed that Wistar rats fed on 50% palmyrah flour showed no gross evidence of liver abnormalities, and more significantly, serum alanine aminotransferase activity which is a sensitive indicator for liver cell damage, even sub-clinically, showed no elevation in comparison to that of rats fed on WHO test feed.
That studP3 also suggested that nutrition may play a part in the expression of toxic effect. It was concluded that if results of different studies are to be compared, publications must include details such as feed composition, feed intake and weight ' gains or losses. The obvious hepatic lesions seen by one g r o~p~~,~~ and not by others 53,56 is puzzling, especially as the palmyrah kottaikilangu originated from the same location, viz., Kalpitiya in the North -West of Sri Lanka. Further, the group reporting hepatotoxicity had been careful to eliminate chance contamination of feed by aflatoxins and pesticide residues.55 There is no clue as to the identity of the hepatotoxin.

The immunosuppressive effect
Wistar rats on prolonged feeding with palmyrah flour produced malignant lymphoma^.^^,^^ This has been attributed to immunosuppressive factors in the flour. It is clear that this factor is not associated with the.neurotoxic effect of palmyrah The basic h y p o t h e~i s~~ was that lymphoid neoplasias arose from a toxic constituent of palmyrah flour. This is manifest by depressed humoral and cell mediated immune response in rats after short term, high dose or long term medium dose feeding of palmyrah flour. In a study where inbred mice were fed with 40% palmyrah flour, a delayed type hypersensitivity response (DTH) to sheep red blood cells was observed.65 DTH was noted aRer 6 days and maximum suppression occurred in the sensitization period. The immunosuppressive effect was transferable to normal mice fed on normal feed by way of viable spleen cells. The cells transferring the effect were T cells with a Ly-l surface antigen (negative to Ly-2 antigen).65 No lesions were observed in the liver. The authors concluded that oral feeding with palmyrah induced T suppressor cell generation, which was able to suppress the DTH response to sheep red blood cells.
Using haemaggluttinating antibody titres and haemolytic plate forming count in the spleen following immunization with sheep red blood cells induced another showing the humoral and cell-mediated response of rats fed with 25% palmyrah flour. The immune competence of animals was depressed significantly in both cell-mediated and hurnoral response. Lymphocytes from peripheral blood failed to respond to phytohaemagglutinin stimulation. The studV4 concluded that it is possible that these immunological alterations were etiologically related to malignant lymphomas, which develop in rats after prolonged feeding. The authors state " that palmyrah flour is the only staple food demonstrating significant -;iterations in immune competence of experimental rats.
It is widely q~o t e d~~,~~,~~ that prevalence of malignant lymphoma^ in the North-East of Sri Lanka (in which palmyrah flour is commonly consumed) is 3-4 times that of the rest of the country. However, there is no access to data from the original epidemiological study.
Unlike the other toxins of palmyrah flour, the immunosuppressive factor has been isolated and its structure e l u~i d a t e d .~~ It was found to be 1[(17a, 23(t)-dammara-20,23-diene-3P 25 dioll (Fig 8)

Figure 8: The immunosuppressive agent
It is a triterpenoid and was isolated by selective solvent extraction and purified by chromatographic techniques." The isolation procedure was activity-directed by using the murine mixed lymphocyte reaction (MCR). The yield was 0.5mg.kg1 palmyrah flour. Spectroscopic analysis by lHnmr, 13Cnmr and FABMRMS on the pure product, revealed the compound to be a dammarane.68 Dammaranes are widespread in plants but mainly have a 17P substitutuent (not 17a as found in the studpa). This is the first time a 17a configuration has been reported for a natural product.
The compound has been made commercially viable by synthesis from a major component of natural "Dammar resin" in only 7 steps giving an overall yield of 4.3%. It is an extensively potent immunosuppressant with an Ic,, of 10ng.ml-' on the MLR assay.68 Of the toxins of palmyrah flour this is the only one characterized.

Mutagenic and clastogenic effects
Palmyrah flour was tested for mutagenic effects69 on Salmonella typhimurium (strains TA 78 and TA 100) and Escherichia coli (strains Wp,, Wp,uvrA, CM 1881 and CM 891). Both boiled and raw odiyal showed dose related mutagenic response in the base pair sensitive strains. The investigation added mutagenicity to the wide range of toxicity of palmyrah flour discussed in the review. Another study on induction of sister-chromatid exchange70 in human blood lymphocytes by aqueous'extractives of palrnyrah flour showed that the phenomenon was dose related but there was no consistency between batches of subjects. Sister-chromatid exchange was found to be proportional to chromosome length. The finding supported incidence of human malignant lymphomas66 and toxicity in rats5= and bacteria@ reported to be due to constituents of palmyrah flour. The effect was mainly on group Achromosomes and resulted in chromatid and chromosomal breaks.'O Some large and small acentric fragments were also observed.70 These effects were dose dependent and consistently produced by different batches of palmyrah flour. However, the potency was less than in the mitomycin control. The study70 suggests that the non-random appearance of chromosome aberrations (also seen in human malignant lymphomas) may be due to a specific tumor-causing agent as observed in the case ofnon-Burkitts lymphomas (group A, C, D) bleomycin (group C) and adrinomycin (group A, D)

Other bioactiuities caused by palmyrah flour
Palmyrah flour mixed with that of other roots, on oral administration to albino rats induced morphological changes in the endometrial surface epithelium of the uterus when viewed through a scanning electron micro~cope.~~ The smooth pattern of the endometrical cells changed to haphazardly orientated groups of cells and loss of microvilli. It was postulated71 that this structural disparity affects the smooth functioning of the nidatory preparation of the endometrium. The experiment lends weight to the use of this preparation traditionally as an oral contraceptive in India.
The antkixobial activity of palmyrah flour has been reported42 both in boiled and unboiled' odiyal. (Table 5)

The flabelliferins ofpalrnyrah flour
Saponins were stated to be absent in palmyrah flour in an early Later Jeyaratnam1° reported 2 steroidal saponins in palmyrah flour. They were a monoglucoside and monorhamnoside. Another report27 indicated the presence of the flabelliferins F, and F-I in palmyrah flour. Recent s t~d i e s~~,~~ showed the presence of a large number of flabeliferrins in the flour from Kalpitiya. These include F-I, F, (anti microbial) F,, F,, F,, F, and a bitter flabelliferin not identical with F-I1 of PFP.
The separations were made by MPLC and Rf determined preparative TLC and structure identification on TLC and fragmentation pattern on enzyme h y d r~l y s i s .~~,~~ The bitter flabelliferin presented difficulties in isolation, as it was unusually soluble in water and not in methanol unlike the other tetraglycbsides and appeared to be strongly associated with a ninhydrin positive spot (probably an amine). This mixture of flabelliferins and amine produced some of the neurotoxic symptoms on Wistar rats.72 Removal of the amine from the bitter flabelliferin by cation exchange yielded a flabelliferin very soluble in methanol and a ninhydrin positive white solid, which produced some neurotoxic symptoms of the mixture but to a lesser degree.58 This flabelliferin was not the same as the very bitter flabelliferin of PFP as it could not be hydrolyzed by heat stable a-amylase and yields rhamnose but no aglycone on hydrolysis with naringinase.
The strong binding of the flabelliferin to other compounds has been observed previously in that they bind to a W fluorescent volatile compound in PFP.73 The role of flabelliferins in toxicity, especially neurotoxicity, is suggestive from results58 but details have still to emerge. Further, it is reported 42 that palmyrah flour has an antimicrobial factor. This may be F, as reported previously in PFPZ9 and /or some other flabelliferin. Clearly more work has to be done on the bioactivity of flabelliferins.

Other toxins ofpalmyrah
The root of palmyrah is known to contain a slow acting poison.1° The pollen from palmyrah is reported to have a 90k Dalton factor responsible for allergic reactions .74

CONCLUSIONS
The direction of research on these two basic raw materials from palmyrah, differs vastly. PFP is underutilized and shows little evidence of being toxic to mammals.
Odiyal is widely utilized in North -East Sri Lanka as a staple and has many reported toxic effects.
The strategy for PFP is therefore to increase utilization, while in odiyal much remains to be done including identification of the toxic factors, studying the effects of processing on toxicity, detoxification techniques and attempts to utilize toxins for the benefit of man as has been done for the immunosuppressive dammarane.
The full significance of the presence of numerous flabelliferins both in PFP and in flour is worthy of investigation including its significance from an evolutionary standpoint.