Marles & Farnsworth 1995 Antidiabetic Plants

)

Phytomedicine Vol. 2 (2), pp. 137-189, 1995 @ 1995 by Gustav Fischer Verlag, Stuttgart Jena . New York Review alilcie

Antidiabetic plants and their active constituents 1

R. J. MARLEsa and N. R. FARNSWORTHb

Department of Botany, Brandon University, Brandon, MB R7 A 6A9, CANADA b Program for Collaborative Research in the Pharmaceutical Sciences, College of Pharmacy, University of

Illinois at Chicago, Chicago, Illinois 60612, U.S.A.

Summary

Diabetes mellitus is a debilitating and often life-threatening disease with increasing incidence in rural populations throughout the world. A scientific investigation of traditional herbal remedies for diabetes may provide valuable leads for the development of alternative drugs and therapeutic strategies. Alternatives are clearly needed because of the inability of current therapies to control all of the pathological aspects of diabetes, and the high cost and poor availability of current ther-apies for many rural populations, particularly in developing countries. This review provides in-formation on more than 1200 species of plants reported to have been used to treat diabetes and/or investigated for antidiabetic activity, with a detailed review of representative plants and some of great diversity of plant constituents with hypoglycemic activity, their mechanisms of action, methods for the bioassay of hypoglycemic agents, potential toxicity problems, and promising directions for future research on antidiabetic plants. The objective of this work is to provide a starting point for programs leading to the development of indigenous botanical re-sources as inexpensive sources for standardized crude or purified antidiabetic drugs, and for the discovery of lead compounds for novel hypoglycemic drug development.

Key words: antidiabetic plants, botany, chemistry, mechanism of action, bioassays for hypogly-cemic agents.

Introduction

At least 30 million people throughout the world suffer from diabetes mellitus. Life expectancy may be halved by this disease, especially in developing countries where its prevalence is increasing and adequate treatment is often un-available. Even in developed countries such as the USA, where sophisticated therapy is widely available, more deaths are attributed to diabetes than to lung cancer, breast cancer, or motor vehicle accidents (World Health Organiza-tion 1985).

Diabetes not only kills, but is a major cause of adult blindness, kidney failure, gangrene, neuropathy. heart at-tacks, and strokes. In the USA, where there are an estimat-ed 14 million diabetics (Bransome 1992), the economic im-

1 Reprinted. with revisions and additions, from Economic and Medicinal Plant Research Volume 6. Academic Press ltd., 1994, with permission.

pact of the disease is enotmous. In 1987 an estimated 5.7 million hospital days were attributed to the treatment of di-abetic complications, with an additional 2 million labor days lost to out-patient physician visits and work loss. Di-rect medical costs due to diabetes are estimated to have been $9.6 billion, and indirect costs for short-term morbid-ity, long-term disability, and mortality (more than 80,000 deaths) are estimated to have been $10.6 billion (Center for Economic Studies in Medicine 1988).

There has been a striking emergence of non-insulin-de-pendent diabetes mellitus as a major health problem in populations undergoing modernization of life-style, both in developing nations and in rural areas of developed coun-tries (Bennett 1983, Bransome 1992, World Health Organ-ization 1985, Gohdes 1986, Schraer et al. 1988). The enor-mous costs of modem treatment indicate that alternate strategies for the prevention and treatment of diabetes must be developed. Since almost 90 % of the people in rural are-

138 R. J. Maries and N. R. Farnsworth

as of developing countries still rely on traditional medicines for their primary health care, and scientific investigations of traditional medicines have led to the discovery of at least 88 drugs now in professional use worldwide (Soejarto and Farnsworth 1989), a synthesis of local traditional and mod-ern knowledge and techniques for the management of dia-betes should be feasible. A rationally designed interdiscipli-nary research program could lead to the development of in-digenous, renewable, medicinal plant resources as practical and cost-efficient alternatives. The purpose of this review is to provide the information needed for the design of such a project.

Background: Diabetes Classification and Modern Therapy

Diabetes mellitus comprises a group of etiologically and clinically heterogeneous disorders with a common set of symptoms: excessive thirst and hunger, muscular weakness and weight loss, excessive urination, and elevation of the blood glucose level which, when it exceeds the renal thresh-old, results in the excretion of glucose in the urine. These symptoms were described by the ancient Egyptians in the Ebers Papyrus about 3500 years ago (Hengesh and Hol-comb 1981), and by the Greek physicians Aretaeus the Cappadocian (A.D. 30-90) and Galen (A.D. 130-200) (Farnsworth and Segelman 1971).

There are three main types of diabetes mellitus recog-nized by the World Health Organization (1985). Insulin-de-pendent diabetes mellitus (IDDM) requires daily injections of insulin to prevent a catabolic cascade culminating in di-abetic ketoacidosis, coma, and death. It is characterized by the virtual absence of p-cells from the islets of Langerhans in the pancreas, and a level of insulin secretion insufficient to restrain excessive secretion of glucagon or to counter its enhancement of hepatic glucose and ketone production. The loss of J3-cells may be due to exogenous chemicals from the environment or diet, viral infection, or immunological factors such as an autoimmune disorder in genetically vul-nerable individuals (Unger and Foster 1985).

Non-insulin-dependent diabetes mellitus (NIDDM, also known as Type II or maturity-onset) occurs predominantly in older people, e.g. 16.8 % of persons over 65 years of age in the United States have NIDDM, and it is often associat-ed with obesity (I1arde and Tuck, 1994). NIDDM repre-sents a variety of diabetic states in which the J3-cells are usu-ally low in number relative to a-cells and insulin secretion is usually sufficient to oppose the ketogenic actions of glu-cagon but not to prevent hyperglycemia. The basal rate of hepatic glucose production is elevated in subjects with NIDDM and this is positively correlated with the degree of fasting hyperglycemia. This increased rate of glucose re-lease by the liver results from impaired hepatic sensitivity to insulin, reduced insulin secretion through impaired p-cell

responsiveness to glucose, and increased glucagon secretion through a reduced ability of glucose to suppress glucagon. The efficiency of glucose uptake by the peripheral tissues is also impaired due to a combination of decreased insulin se-cretion and defective cellular insulin action (insulin resis-tance) (Porte and Kahn 1991). Receptor mediated insulin resistance may be a consequence of various factors includ-ing increased serine/threonine phosphorylation of the re-ceptor with decreased tyrosine phosphorylation, receptor desensitization, auto-antibodies to the receptor and inherit-ed structural defects in the insulin receptor. Defects in insu-lin action could also arise at post-receptor events particu-larly glucose transport. Other circulating hormones such as islet amyloid polypeptide (amylin) may also cause insulin resistance (Pillay and Makgoba 1991).

Malnutrition-related diabetes mellitus (MRDM) refers to the condition of young diabetics in tropical developing countries with a history of nutritional deficiency and a set of symptoms which fail to meet the criteria used to classify IDDM and NIDDM. The subclass, "fibrocalculous pan-creatic diabetes" (FCPD), is believed to be associated with the consumption of foods containing cyanogenic glyco-sides, such as cassava (Manihot esculenta Crantz, Euphor-biaceae). The other main subclass, "protein-deficient pan-creatic diabetes" (PDPD), is believed to be associated with early childhood malnutrition conditions such as kwashior-kor in which p-cell damage occurs (World Health Organ-ization 1985, McMillan and Geevarghese 1979). Both FCPD and PDPD may be forms of NIDDM complicated by dietary factors, and thus not necessarily associated with liv-ing in a tropical developing country (Alberti 1988).

Although this classification of diabetes mellitus is actual-ly too simplistic to properly explain the etiology of the dis-ease in most individuals, since a wide range of factors may determine the expression of diabetic symptoms (Rossini et al. 1988), it is still a clinically useful scheme for determin-ing the appropriate therapeutic method.

Modern therapy of IDDM began with the discovery of the involvement of the pancreas in diabetes by von Mering and Minkowski in 1889, and the demonstration by Banting and Best in 1921 that an extract of beef pancreata could successfully lower blood glucose levels in pancreatecto-mized dogs. Their use of a pancreatic extract in a human di-abetic in 1922 marked the first use of the pancreatic antidi-abetic principle, insulin, in the treatment of diabetes mellit-us. Several different preparations of bovine, porcine, and human insulin (1, in Fig. 1) are now available, including lente or long-acting forms, and a regimen of daily injections represents the current standard of therapy for IDDM (Hen-gesh and Holcomb 1981).

Insulin acts by binding to a cell membrane tetrameric protein receptor which consists of two extracellular a- and two transmembrane p-subunits. Binding of insulin to the a-subunit causes autophosphorylation of J3-subunit intracel-lular tyrosine residues. The activated insulin receptor then


gl y~ i le~v.lgluglncys-cys-thr-ser- j le-cys- ser-leu- tyr"gln-Leu-glu-asn-tyr-cys-asn 8 9 10 s~

I s

I I

ph.-val- n-gln-hi .-Iou-cys-gl y-ser-hi s-Ieu-val-glu-ala-leu-tyr-Ieu- val-cys-gl y-g lu-.rg-gl y-phe-ph.- tyr-thr

Boyine insul in: Porcine insuL in:

1

2

8 9 10 30 .ll-ser-val thr-ser- i Ie

ala ala

.---J pro-Iy.-thr

30

3

EN~S02NHC(=O)NH-o ~ OCH3 '.0 CI 4

HN-

140 R. J. Marles and N. R. Farnsworth adult pancreas transplantation (Hellerstrom et al. 1988). Treatment with an insulin pump or multiple daily injections still differs from the natural situation because exogenous insulin is entering the peripheral circulation first rather than the liver and these methods do not perfectly imitate the pulsatile character of insulin secretion nor the rapid post-prandial rise of insulin levels and subsequent rapid drop. Microencapsulation of pancreatic islets has been at-tempted in spontaneously diabetic models, but variable re-sults were obtained due to fibrosis of the capsules. This can be minimized by purification of the capsule material (sodi-um alginate-poly-L-lysine) and decreasing the size of the microcapsules (Clayton et al. 1993). Hybrid artificial pan-creas consist of insulin-secreting pancreatic tissue sur-rounded by a membrane that protects the tissue from rejec-tion by the immune system following implantation. Unre-solved problems with this mode of therapy include biocom-patibility, oxygen supply limitations, and prevention of im-mune rejection (Colton and Avgoustiniatos 1991).

Due to the immunological nature of some cases of 100M, treatment with immune suppressants has been at-tempted. Cyclosporin is only temporarily effective in 100M and its use is not recommended (Faulds et al. 1993).

Therapy of NIDDM involves modifications of lifestyle and diet, an exercise regimen, and use of oral hypoglycemic agents (Fig. 1). Dietary modifications include limiting total caloric intake, increasing the percentage of calories from complex carbohydrates and reducing the intake of fats and cholesterol (Ilarde and Tuck, 1994). To be effective, thera-peutic interventions for NIDDM must reduce heptic glu-cose production either by improving islet dysfunction and raising plasma insulin levels, or improving the effectiveness of insulin on the liver (Porte and Kahn 1991). The use of oral hypoglycemic drugs may be effective in controlling blood glucose levels, but may not prevent all the complica-tions of diabetes (Unger and Foster 1985).

Observations in the 1940's that certain sulfonamide anti-biotics, used to treat typhoid fever and pneumonia, caused the side-effect of hypoglycemia led to the development of the sulfonylurea hypoglycemic agents. Tolbutamide (2) was the first, approved for use in the United States in 1957. Al-though the "first generation" drug chlorpropamide (3) causes side effects more often, the incidence of severe hypo-glycemia, the major lethal side effect, is at least as high with the "second generation" glyburide (4). To lower the inci-dence of this problem gliclazide (5) was developed. Its mechanisms of action involve stimulation of insulin secre-tion through the j3-cell sulfonylurea receptor, involving clo-sure of K+ ATP channels, and possibly through a direct ef-fect on intracellular calcium transport. and also reduction of hepaeic glucose production and improvement in glucose clearance. This is accomplished without changes in insulin receptors, suggesting a pose-receptor effect on insulin ac-tion perhaps by stimulation of hepaeic fructose-2,6-bis-phosphatase and muscle glycogen synthetase. Additional

effects of gliclazide beneficial for the prevention of diabetic microangiopathic complications include reduction of plate-let adhesion, aggregation and hyperactivity, and increased fibrinolysis (Campbell et al. 1991, Alberti et al. 1994). Pos-sible detrimental side effects of sulfonylureas include hypo-natremia, dermatological reactions, hepatitis, and hemato-logic effects (Ferner 1988).

None of the currently available sulfonylureas completely normalize insulin secretion and action (Beck-Nielsen et al. 1988). Failure to respond to sulfonylurea drugs may be pri-mary (25 % to 30 % of initially treated patients) or secon-dary, occurring in 5 % to 10 % of patients per year (Ilar-de and Tuck 1994), e.g. when NIDDM is a transitional state preceding 100M. Combination therapy with insulin and a sulfonylurea agent only slightly improved glycemic control in NIDDM patients - less exogenous insulin was needed but fasting serum insulin levels showed no difference berween treatment groups. This therapy did not produce near-normal blood glucose levels and so it is not recom-mended for poorly controlled NIDDM patients receiving insulin (Peters and Davidson 1991).

Another class of drugs, the biguanides, was also shown to be effective, and phenformin (6) was approved in 1959, but due to its association with fatal lactic acidosis it was re-called in 1977. Metformin (7) is a less toxic biguanide which can be a useful adjunct in NIDDM therapy, since it improves peripheral sensitivity to insulin through a stimu-lated tissue glucose uptake by a transporter-linked system (Sirtori and Pasik 1994).

Other therapeutic options for NIDDM include the use of insulin-sparing antihyperglycemic agents such as a-glucosi-dase inhibitors, thiazolidinediones, chloroquine or hy-droxychloroquine, or fibric acid derivatives such as clofi-brate. Other experimental agents include fatty acid oxida-tion inhibitors and dichloroacetate. To prevent the compli-cations arising from the spectrum of clinical and metabolic abnormalities which arise from insulin resistance other spe-cific agents may be used including antihypertensives, lipid lowering agents and sorbitol inhibitors (Ilarde and Tuck 1994). Insulin-like growth factor-1 (IGF-1) produced by re-combinant DNA technology is being used in NIDDM pa-tients to stimulate glucose uptake, improve glucose toler-ance. decrease hyperinsulinemia and decrease hypertrigly-ceridemia. Since it improves metabolic control in patients with extreme insulin resistance it is a useful therapeutic ad-junct (Kolaczynski and Caro 1994).

From this brief overview of diabetes classification and modern therapy, it can be seen that current methods of treatment for all types of diabetes mellitus fail to achieve the ideals of normoglycemia and the prevention of diabetic complications. Promising fields of research such as pan-creatic transplants offer little hope to the majority of the world's diabetics, for whom such procedures will be too ex-pensive and difficult to obtain. Most developing countries cannot even afford adequate conventional therapy at the

1987 average U.S. price Qf $14.28 per oral hYPQglycemic prescriptiQn, Qr $14.23 per insulin prescriptiQn (Center fQr EcQnQmic Studies in Medicine 1988).

Further prQblems with cQnventiQnal therapy in develQP-ing cQuntries include insulin supply, storage, and injectiQn, dietary cQntrQl and cQmplicatiQns frQm malnutritiQn, a lack Qf trained health care wQrkers, and a lack Qf educatiQn fQr the patients (Gill 1988). In such situatiQns the incidence Qf diabetes-related mQrtality is far greater than in well-served urban areas. There is, therefQre, a clear need fQr al-ternate SQurces Qf bQth Qral and parenteral antidiabetic drugs and alternate strategies fQr diabetes therapy.

Plants and the Treatment of Diabetes Mellitus

TraditiQnal medicines fQr the treatment Qf diabetes mel-litus are probably based mainly Qn treatment Qf its QbviQUS symptQms Qf prQnQunced thirst and PQlyuria. Even glycQ-suria was recQgnized as a symptQm Qf diabetes in ancient Ayurvedic medical texts such as the Sushruta Samhita and Charaka Samhita (Nagaraj an et al. 1982). The Greek phy-sician Aretaeus recQmmended treatment Qf diabetes by treatment Qf the profQund thirst. FQr this he recQmmended starting with a purgative to' strengthen the stQmach, fQl-lQwed by cQnsuming water bQiled with autumn fruit (a gQQd SQurce Qf sQluble fibre and cQmplex carbQhydrates like pectin), milk, gruels Qf a variety Qf whQle grains (an ex-cellent SQurce Qf sQluble and insQluble fibre and glycans), and astringent wines (alcQhQl is hYPQglycemic accQrding to' Hengesh and HQlcQmb 1981, LQmeQ et al. 1988). He alsO' recQmmended a crude drug Qf animal Qrigin: venQm Qf the "dipsas" viper, which in bite victims causes a severe thirst. Aretaeus suggested it CQuid be used as a mithridate, i.e., a PQisQn which is deliberately administered in small, gradual-ly increasing dQses in Qrder to' develQP an immunity to' the effect Qf the PQisQn (Adams 1856). In fact, the venQm Qf the Middle Eastern viper PiscivQrus PiscivQrus (Crotalidae) was fQund to' be hYPQglycemic when administered i.v. at a dQse Qf lOJ.1g!kg in nQrmal rats and rabbits, but was inac-tive against allQxan-induced hyperglycemia in rats (Taha 1982).

MQre than 1200 species Qf Qrganisms have been used eth-nQpharmacQIQgically Qr experimentally to treat symptQms Qf diabetes mellitus (see the Appendix). They represent mQre than 725 genera in 183 families, extending phylQge-netically all the way frQm marine algae and fungi to ad-vanced plants such as the cQmpQsites. The mQst frequently cited families are shQwn in Table 1. These are very large and widely distributed families, SO' the large number Qf spe-cies repQrted to' have been used traditiQnally Qr experimen-tally fQr the treatment Qf diabetes may be cQincidental. The phylQgenetic distance between even this select group Qf families is a strong indicatiQn Qf the varied nature Qf the ac-tive cQnstituents. Thus, while chemQtaxQnQmic studies are

Antidiabetic plants and their active cQnstituents 141

Table 1. Plant Families Most Often Cited for Antidiabetic Activi-ty.

Family Species cited Total species'

Fabaceae 127 18,000 Asteraceae 98 21,300 Lamiaceae 36 3,500 Liliaceae 35 6,460 Poaceae 30 10,000 Euphorbiaceae 30 7,000

"According to Thorne (1981).

Qften useful in the discQvery O'f new plants with biO'IO'gical-ly active cQnstituents, it will be necessary to learn mO're abQut particular grO'UPS Qf hYPQglycemic natural prO'ducts and their mechanisms Qf actiO'n befO're this methQd Qf drug discO'very can be successfully emplQyed.

Half O'f the species fQund in Qur literature review have been used in traditiO'nal medicine to' treat symptQms O'f dia-betes. Half O'f these traditiO'nal remedies have had sO'me ex-perimental testing fO'r hypoglycemic activity, e.g., in nor-mal, glucO'se-loaded, allO'xan- or streptQzQtO'cin-induced di-abetic, or naturally diabetic subjects. DistinctiQns Qf the ex-perimental mQdel used are clearly impO'rtant fO'r gaining an understanding Qf the mechanism Qf actiQn of these bQtani-cal drugs. Further details Qf the biQassay methQds cO'mmO'n-ly used and their significance fO'r the discQvery Qf new anti-diabetic agents will be prQvided belQw.

A summary of the results of screens fQr blQQd glucQse lO'wering activity, presented in Table 2, shQWS that 81 % Qf thQse traditiO'nal antidiabetic plants tested gave PQsitive re-sults. Even fO'r thQse plants fO'r which nO' traditiQnal use was mentiQned, 47 % Qf those species screened were active. This rate Qf PO'sitive results is higher than Qne WQuid expect by random chance - perhaps 10 % WO'uid be reasQnable, based Qn the number of active species Qbtained by the U.S. NatiO'nal Cancer Institute's random screening O'f mO're than 35,000 species fQr antitumor activity (Spjut and Perdue 1976). The high percentage Qf active plants prO'bably re-flects, at least in part, the great variety Qf PQssible active cO'nstituents and mechanisms O'f actiQn, the PQssibility that nO't all negative results were repQrted, and fO'r the nO'n-tra-ditional plants, Qther cO'nsiderations made in selecting them fO'r study (e.g. chemQtaxQnO'mic). Nevertheless, it is clear from the above results that the study of traditiQnal reme-dies fO'r diabetes mellitus yields an excellent return in po-tential new sources of antidiabetic drugs.

If the same O'r a closely related plant is used traditiO'nally for the same purpQse in mQre than Qne cQuntry, it suggests

Table 2. Activity of Traditional Antidiabetics vs. Other Plants.

Total no. tested Total active

Traditionals

295" 238 (81 %)

Others

541 254 (47%)

d Out of a total of 582 known traditionally used plants

142 R.]. Marles and N. R. Farnsworth

Table 3. Most Widely Used Traditional Antidiabetic Plants.

Scientific name Countries where used traditionally

CUCURBITACEAE Momordica charantia Saudi Arabia, West Africa, Pakistan, India, Sri Lanka, Thailand, Fiji, Bimini, Panama, Puerto Rico,

Belize, Jamaica, Trinidad, Virgin Islands, England APOCYNACEAE Catharanthus roseus Australia, England, Thailand, Natal, Mozambique, India, Philippines, Vietnam, Dominican Republic,

Jamaica ANACARDIACEAE Anacardium occidentale Ecuador, Colombia, Mexico, Venezuela, Jamaica, Madagascar, India, Thailand, England MYRTACEAE Syzygium cumini Eucalyptus globulus

India, Pakistan, Thailand, West Indies, USA, Portugal West Indies, Mexico, Guatemala, China

FABACEAE Lupinus albus Trigonelfa foenum-graecum

Canary Islands, India, Israel, Portugal, Morocco Israel, Egypt, France, India

LILIACEAE Aloe vera Allium cepa Allium sativum

Haiti, India, Tunisia, Kuwait, Saudi Arabia India, Saudi Arabia, North Africa, Peru India, Saudi Arabia, Mexico, Venezuela

BIGNONIACEAE Tecoma stans India, Mexico, Guatemala, Virgin Islands, Cuba URTICACEAE Urtica dioica England, USA, Guatemala, Nepal, India ASTERACEAE Taraxacum officjnale Europe, Costa Rica, Mexico, USA CYPERACEAE Kyllinga monocephala India, Ethiopia, Indonesia, South America (country not specified) EUPHORBIACEAE Phyllanthus emblica Phylfanthus niruri MELIACEAE Azadirachta indica MORACEAE Morusalba ROSACEAE Poterium ancistroides APIACEAE Daucus carota

India, Nepal, Tibet, Pakistan Indonesia, India, West Indies, Brazil

India, Fiji, Saudi Arabia, Trinidad

India, USSR, China, Peru

Spain, Greece, Syria, Israel

India, China, England, USA

either cultural contact between the countries or indepen-dent discovery. In either case, the conservation of that tra-ditional use indicates a higher probability that the tradi-tional practitioners found the remedy to be effective. Table 3 lists the twenty most widely used traditional antidiabetic plants. With the notable exception of Kyllinga, all of these species have already been studied and shown to be active or have active constituents, and for most of them the identity of the probable active constituents is known. Several of these plants will be discussed in detail below.

Seventeen of the twenty most widely used traditional antidiabetic plants, and many others too, are used in India. The Indian subcontinent has an extensive indigenous phar-macopoeia, including the Ayurvedic, Unani, and folkloric medical systems, which has already supplied the world with

such useful drugs as reserpine, from Rauvolfia serpentina, which is used as an antihypertensive and tranquilizer (Tyler et al. 1981). Reserpine is also reported to be hypoglycemic in normal animals and animals made hyperglycemic by pre-treatment with epinephrine (Ricci and Ricordati 1955). In-dian traditional medicines may very well supply the world with some new antidiabetic drugs.

Several reviews of plants with known antidiabetic activ-ity or traditional use as antidiabetic remedies, prepared on a more limited scale than the current work, have been pub-lished (Farnsworth and Segelman 1971, Ajgaonkar 1979, Oliver-Bever and Zahnd 1979, Oliver-Bever 1980, Nagara-jan et al. 1982, Mossa 1985, Oliver-Bever 1986, Day and Bailey 1988, Bailey and Day 1989, Handa et al. 1989, Rah-

o:fOOH ccrCOOH 1 "-'::: ~ NH2 1 "-'::: ~ ~ NH ~ NH 8 9

U C02H a COOH l,fi :::,...1 N NH2 11 12

Fig. 2. Plant growth regulators with hypoglycemic activity.

man and Zaman 1989, Ivorra et al. 1989, Winkelman 1989).

The object of this review is to present a comprehensive literature review of plants associated with the treatment of diabetes mellitus, and to discuss with detailed examples their potential as sources for new antidiabetic drugs. The primary source of the information on antidiabetic plants presented here was the NAPRALERT (Natural Products Alert) computer database of the Program for Collaborative Research in the Pharmaceutical Sciences, College of Phar-macy, University of Illinois at Chicago.

Hypoglycemic Constituents and Mechanisms of Action

To understand how plant constituents can be hypoglyce-mic in animals, it is worthwhile to consider the reasons why compounds with hypoglycemic activity occur in plants. In general, discussions of medicinal agents from plants center on plant secondary metabolites, i.e., non-ubiquitous con-stituents with no known essential role in the plant's metab-olism.

It has been postulated that bioactive plant secondary me-tabolites may play a role in chemical defense mechanisms (Ehrlich and Raven 1964, Berenbaum 1983). While the precise mechanisms that may be involved in chemically me-diated coevolution between plants and herbivores or path-ogenic organisms are controversial (Strong et al. 1984, Spencer 1988), it has been suggested that natural selection would ensure the survival for reproduction of those indi-viduals of a species having the gene coding for production of a toxin, while individuals without the toxin would be consumed (Williams et al. 1989).

Most hypoglycemic plant constituents, such as the Cath-aranthtts alkaloids, might fit in this category, but there are other rather common plant constituents for which this ex-planation is not entirely satisfactory. At the cellular and molecular levels, plants and animals are not very different


~O'/COOH I~ CI

10

a C02-Ih a C02H N+ :::,...1 I CH3 OH 13 14

in their metabolic processes. Glucose is the metabolic ener-gy source and most important biosynthetic precursor in plants, so glucose undergoes storage and mobilization under hormonal control in plants as it does in animals. Plant growth regulators such as indole-3-acetic acid (Fig. 2, 8) and natural and synthetic analogs such as indole-3-bu-tyric acid, indole-3-propionic acid, L-tryptophan (9), and p-chlorophenoxyacetic acid (10) , inhibit insulinase in vitro and are hypoglycemic in vivo in normal rats (Mirsky et al. 1956). Nicotinic acid (11) and anthranilic acid (12) also in-hibit insulinase and potentiate simultaneously administered insulin. An inhibitor of indole-3-acetic acid oxidase from Phaseo/us vulgaris fruit exocarps also has hypoglycemic ac-tivity. The hypoglycemic alkaloid trigonelline (13), from Trigone//a (oenum-graecum, is a plant growth inhibitor and produces dormancy.

Salicylic acid (14) is also a plant growth inhibitor and hy-poglycemic agent (Oliver- Bever and Zahnd 1979). Thus, plant metabolism-regulating constituents can also be ani-mal metabolism-regulating agents. The variety of ways in which this may be possible will become more clear with the discussion of hypoglycemic mechanisms of action to follow.

Possible active hypoglycemic constituents have been re-ported for 88 (16 %) of the plants used traditionally as antidiabetics and 62 (11 "!o) of the other plants screened. There are more than 200 pure compounds from plant sources reported to show blood glucose lowering activity. Table 4 provides a summary of the chemical classes of these compounds. The wide variety of chemical classes indicates that a variety of mechanisms must be involved in the lower-ing of the blood glucose level. Some of these compounds may have therapeutic potential, while others may produce hypoglycemia as a side-effect of their toxicity, especially he-patotoxicity.

Some of the compounds reported to be active in vitro or at high doses in vivo, e.g., ~-sitosterol-D-glucoside (daucos-terol, Fig. 3, 16), occur so widely in nature that therapeutic activity seems unlikely. This could be due to their low con-centration in the plant or co-occurrence with complexing


Table 4. Hypoglycemic Natural Products

Chemical class

Alkaloids Carbohydrates Coumarins Cyanogenic gycosides F1avonoids Glycopeptides Inorganic salts Iridoids Lipids

Number active

38 66

4 1 7

20 3 4 6

or counteracting constituents. Some examples of plants with known active constituents and known mechanisms of action will be described below to show the range of active constituents and mechanisms of hypoglycemic action.

Pep tides and Terpenoids from Momordica The most widely used traditional remedy for diabetes

mellitus is Momordica charantia L. (Cucurbitaceae), com-mon names for which are "bitter gourd," "balsam pear," "cundeamor," and "cerasee." The fruit, leaf, and stem have been used to make an antidiabetic decoction (Rivera 1941, Rivera 1942, Pons and Stevenson 1943, Ram 1956, Oakes and Morris 1958, Khan and Burney 1962, Lotlikar and Rajarama Rao 1966, Jain and Sharma 1967, Morton 1967, Olaniyi 1975, Ayensu 1978, Halberstein and Saunders 1978, Aslam and Stockley 1979, Gupta et al. 1979, Arna-son et al. 1980, Oliver-Bever 1980, Nagarajan et al. 1982, Morrison and West 1982, Mossa 1985, Bailey et al. 1986, Singh 1986). In antihyperglycemic bioassays using oral, subcutaneous, and intravenous dosing of mice, rats, and rabbits pretreated with anterior pituitary extract, alloxan, or streptozotocin, and in diabetic humans, it gave different and often apparently conflicting results (Chatterjee 1964, Khanna et al. 1981, Mossa 1985, Bailey et al. 1985, Weli-hinda et al. 1986). In a clinical trial with NIDDM patients, 73 % of the patients showed improved glucose tolerance with oral administration of M. charantia fruit juice (Weli-

ij~H H . HO OH

15

Chemical class

Pep tides and amines Phenolics (simple) Phenolpropanoids Steroids Stilbenes Sulfur compounds Terpenoids Vitamins Xanthenes

Number active

15 4 1 7 1 2

17 2 1

hinda et al. 1986). Some confusion also prevails with hypo-glycemic testing in normal animals (Rivera 1942, Pons and Stevenson 1943, Morrison and West 1982, Karunanayake et al. 1984, Meir and Yaniv 1985, Welihinda and Karuna-nayake 1986).

Several active compounds have been isolated from M. charantia (Fig. 3), and some mechanistic studies have been done. Khanna et at. (1981) have reported the isolation from the fruits, seeds, and tissue culture of seedlings, of "poly-peptide-p," a 17-amino acid, 166-residue polypeptide which did not cross-react in an immunoassay for bovine in-sulin. This peptide was shown to be "insulinomimetic" when administered subcutaneously in rodent and primate experimental assays and in a limited clinical trial with both juvenile- and maturity- onset diabetic patients. A number of other polypeptides from M. charantia seeds have been studied in vitro for the insulin-like activities of stimulation of lipogenesis and inhibition of corticotropin-induced lipol-ysis. The mechanism was suggested to involve interaction of the peptides with a-adrenergic or cotticotropin receptors (Ng et al. 1986).

Another active constituent, charantin, has been isolated from both M. charantia and M. foetida, and identified as a mixture of two steroid glycosides: ~-sitosterol-D-glucoside (15) and 5,25-stigmastadien-3-~-01-D-glucoside (16). Anti-hyperglycemic activity in alloxan-treated rabbits and de-pancreatized cats dosed p.o. or i.v. was equivocal, but hy-poglycemic activity was observed in normal rabbits, rats,

16 Fig. 3. Steroid glycosides of Momordica charantia reported to be hypoglycemic.

and cats dosed p.o., i.p., or i.v. (Lotlikar and Rajarama Rao 1966). Studies performed in vitro with M. charantia fruit extracts indicated a significant enhancement of glucose up-take in muscle tissue and of glycogen accumulation in mus-cle and hepatic tissue, but no effect on glucose uptake or triglyceride synthesis in adipose tissue (Meir and Yaniv 1985, Welihinda and Karunanayake 1986). Inhibition of glucose uptake by intestinal fragments was also observed and attributed to a glycosidic constituent of the fruit ex-tract (Meir and Yaniv 1985). Thus, there appear to be con-stituents of M. charantia with both pancreatic and extra-pancreatic effects with therapeutic potential for diabetic patients. Caution is advised, however, because a mildly tox-ic lectin has been reported from the seeds and outer rind of the fruits, which is capable of interfering with protein syn-thesis in the intestinal wall (Lampe and McCann 1985).

Alkaloids from Catharanthus The Madagascar periwinkle (Catharanthus roseus [L.] G.

Don, Apocynaceae), is another widely used traditional rem-edy for diabetes, and a proprietary preparation, Vinculin, was marketed in England as a "treatment" for diabetes. Pharmacological studies have been conducted on periwin-kles since the 1920's, and while two studies of leaf aqueous extracts administered orally to rabbits (Asthana and Misra 1979) and dogs (Morrison and West 1982) reported a hy-poglycemic response, many other experiments with a varie-ty of laboratory animals and limited clinical studies have given negative or at best equivocal results (Noble et al. 1958, Farnsworth 1961, Svoboda et al. 1959 and 1964, Farnsworth and Segelman 1971, Swanston-Flatt et al. 1989).

Despite these disappointing results, Svoboda et al. (1964) tested for hypoglycemic activity a number of alkaloids (Fig. 4) isolated from C. roseus during an investigation of the plant's oncolytic activity, which was discovered by Noble et al. (1958) while investigating the plant's reputed antidiabet-ic activity. Hypoglycemic activity was observed for catha-ran thine (17), leurosine (18), lochnerine (19), tetrahydroal-stonine (20), vindoline (21), and vindolinine (22). Adminis-tered orally in a dose of 100 mg/kg, leurosine sulfate and vindolinine hydrochloride were more hypoglycemic than tolbutamide, the commercial antidiabetic sulfonylurea used as a positive control (Svoboda et al. 1964). Svoboda et al. (1964) suggested that toxicity of crude extracts and frac-tions (e.g., several of the alkaloids are potent cytotoxic agents) may have made their experimental antidiabetic ver-ification difficult, but that further study of C. roseus as a natural antidiabetic agent would be worthwhile.

Some progress has been made in this direction. The Cath-aranthus and Vinca alkaloids, vincamine (23) and (-)-ebur-namonine (24), have been shown to induce an extensive de-crease in rat brain tissue glucose, a concomitant increase in lactate and pyruvate concentrations and the lactateJpyru-


vate ratio, and an increase in ATP contents and energy charge potential (Benzi et al. 1984). Tetrahydroalstonine (20), administered orally in rats with alloxan-induced hy-perglycemia, produced a triphasic response of a rapid-onset hypoglycemia, a recovery period from 2-12 hours post-treatment, and then a prolonged hypoglycemic effect last-ing more than 48 hours post-treatment (Kocialski et al. 1972).

In an in vitro study of the mechanism of action of the quinoline derivatives, quinolate and 3-mercaptopicolinate, Snell (1979) reported that hepatic gluconeogenesis from lactate or alanine, and the release from muscle of alanine, is inhibited through inhibition of cytosolic and mitochondrial phosphoenolpyruvate carboxykinase. The mechanism in-volves a direct effect which is facilitated by complex forma-tion between the agent and Fe2+ or Mn2+, an inhibitory ac-tion on the ferroactivator-mediated Fe2+ activation of cyto-solic phosphoenolpyruvate carboxykinase, and indirect ef-fects by lowering of cytosolic oxaloacetate concentrations through blocking the translocation of anions such as 2- ox-oglutarate from mitochondria, and inhibiting cytosolic aspartate aminotransferase.

Certainly the active alkaloids of Catharanthus could serve as models for the development of new antidiabetic drugs. Eleven indolizine alkaloids, synthesized as analogs of vincarnine, vindoline, and vindolinine, were tested for oral hypoglycemic activity in fasted rats, but the best was only one third as active as tolbutamide (De and Saha 1975).

Sulfur Compounds from Allium The hypoglycemic principles of onion (Allium cepa L.,

Liliaceae) and garlic (A. sativum L.) are the sulfur-contain-ing compounds, allyl propyl disulfide (25 in Fig. 5) and di-allyl disulfide oxide (allicin, 26). Active in normal and al-loxan-diabetic animals and patients with NIDDM, but not pancreatectomized animals, they are believed to act by competing with insulin, which has a disulfide linkage, for endogenous sulfhydryl-rich insulin-inactivating com-pounds (Augusti et al. 1974, Oliver-Bever and Zahnd 1979). However, an oral feeding study of garlic bulbs given to normal or streptozotocin-diabetic mice showed reduced hyperphagia and polydipsia but no effect on hyperglycemia or hypoinsulinemia (Swanston-Flatt et al. 1990).

Inorganic Ions from Atriplex The saltbush (Atriplex halimus L., Chenopodiaceae) was

investigated for antidiabetic activity after sand rats (Psam-momys obesus), that in nature feed extensively on the leaves of this plant, developed diabetic symptoms after be-ing captured and fed laboratory rat chow or fresh vegeta-bles. The sand rats have a genetic predisposition to diabetes that seems to be prevented by the presence of chromium, manganese, and magnesium salts in the saltbush leaves.


17

H

19

21

23

Fig. 4. Hypoglycemic alkaloids of Catharanthus roseus.

Studies of the leaf ash and chromium in vitro showed a po-tentiation of insulin-stimulated glucose utilization by epi-didymal fat cells of chromium deficient rats. The mecha-nism may involve Cr2+ inactivation of an insulin-inactivat-ing enzyme (Aharonson et al. 1969, Oliver-Bever and Zahnd 1979). The reputed hypoglycemic activity of the "glucose- tolerance factor" of brewer's yeast, Saccharomy-ces cerevisiae, which has been attributed to trivalent chro-mium (CrJ+), was contradicted by long-term feeding studies in genetically diabetic mice, in which no beneficial t;ffect was seen (Flatt et al. 1989). However, chromium does pot-entiate the action of insulin in vitro and in vivo. Maximal in vitro activity requires mineral complexation, e.g. a chromi-um-nicotinic acid complex. Clinical trial results were vari-

18

20

22

CXXJp I ~N N -o CH2CH3

24

able but the majority of patients showed an improved effi-ciency of insulin. (Mertz 1993).

Chronic administration of magnesium salts has also been shown to be beneficial in the treatment of NIDDM. Hypo-magnesemia is a common finding in diabetic subjects. Mag-nesium is a necessary cofactor for many enzymes and is in-volved in protein synthesis. Treatment with magnesium salts resulted in a net increase in acute insulin response and the rate of glucose disappearance after glucose loading (Paolisso et al. 1989, White and Campbell 1993).

Other minerals may also playa role in diabetes pathogen-esis and therapy. The protein tyrosine kinase associated with the insulin receptor has been shown to be Mn2+ depen-dent (Reddy and Kahn 1988). Vanadium is another trace

2S

CH2=CH-CH2-S( .... O)-S-CH2-CH=CH2 26

Fig. 5. Hypoglycemic sulfur compounds from Allium spp.

mineral whose salts have insulin-like properties in animal models of insulinopenia or insulin resistance in vitro and in vivo, due to stimulation of glucose metabolism. Like most dietary trace minerals vanadium is toxic in excess so its therapeutic potential is being investigated carefully (Brich-ard et al. 1991).

Amino Acids from Blighia Ingestion of unripe akee fruit (Blighia sapida Koenig, Sa-

pindaceae) causes the often fatal disorder " \-omiting sick-ness" in Jamaica. The emetic constituents were discovered to be the cyclopropanoid amino acid, hypoglycin A (27 in Fig. 6), and its y-L-glutamyl dipeptide, hypoglycin B (28). which are also potent hypoglycemics. They appear to act by inhibiting ~-oxidase enzymes, thus blocking oxidation of long-chain fatty acids. Since the fatty acids are no longer available as an energy source, hepatic glycolysis is stimulat-ed to provide an alternate source, and the increased utiliza-tion of glucose brings about a fall in blood glucose levels. Hypoglycin A is twice as potent a hypoglycemic as hypo-glycin B; the latter is also teratogenic, so these compounds are too toxic to be used therapeutically, though they may provide models for the development of new hypoglycemic agents (Feng and Patrick 1958, von Holt et al. 1966, Tana-ka et al. 1972, Oliver-Bever and Zahnd 1979).

In order to find a more specific inhibitor of free fatty acid oxidation, Kanamaru et al. (1985) screened microbial me-tabolites for substances that would inhibit the oxidation of long-chain fatty acids in rat liver mitochondria. This re-search led to the discovery of the ~-aminobetaines, emerice-din (29) and its more potent synthetic derivative emeria-mine (30), from the fungus Emericella quadrilineata IFO 5859 (Trichocomaceae). Emeriamine was shown to be a po-tent and specific inhibitor of carnitine palmitoyltransferase I, and both compounds had dose-dependent oral hypoglyce-mic and anti ketogenic activities in fasted normal, strepto-zotocin-diabetic, and genetically obese (Zucker) rats.

Guanidines from Galega Seeds of the traditional antidiabetic plant. "goat's rue:'

(Galega officinalis L., Fabaceae) contain the guanidine de-rivative, galegine (31 in Fig. 7). Like synthetic biguanide hypoglycemics (6, 7), galegine blocks succinic dehydroge-


27

28

(CH3hN+CH2CH(NHCOCH3)CH2COO-

29

(CH3hN'CH2CH(NH2)CH2COO-

30 Fig. 6. Inhibitors fo fatty acid oxidation.

nase and cytochrome oxidase, thus increasing anaerobic glycolysis and decreasing gluconeogenesis, resulting in en-hanced glucose uptake and hypoglycemia. Biguanides are also known to inhibit glucose absorption from the intestine (Oliver-Bever and Zahnd 1979).

Vitamins, Coumarins, and Steroids from Trigonella Fenugreek (TrigonelLa foenum-graecum L.), seeds con-

tain a number of hypoglycemic principles, although an oral feeding study performed with normal and streptozotocin-diabetic mice showed no significant effect of seed consump-tion on basal glucose and insulin, insulin-induced hypogly-cemia, glycosylated hemoglobin, or pancreatic insulin con-centration (Swanston-Flatt et al. 1989). Trigonelline (Fig. 2: 13), which is the N-methyl derivative and main human metabolite of the vitamin nicotinic acid (niacin, 11), has a weak and transient hypoglycemic effect when administered orally to diabetic patients. It acts by slowing the metab-olism of nicotinic acid, also present in Trigonella, which is known to increase glucose uptake from the blood and its subsequent oxidation, if administered orally. Nicotinic acid is hyperglycemic if administered parenterally, by means of impairment of carbohydrate utilization (Mishkinsky et al. 1967, Shani et al. 1974). Taken orally, nicotinic acid is con-verted in the body into nicotinamide, which is an inhibitor of the enzyme poly(ADP-ribose) synthetase, responsible for the depletion of NAD from pancreatic ~-cells, and is also a potent hydroxyl-radical scavenger, by which mechanisms

148 R.]. Marles and N. R. Farnsworth NH

HN=< 2 NH

~=< 31 7

Fig. 7. Comparison of the structures of galegine and metformin.

nicotinamide can prevent the ~-cell toxicity of streptozoto-cin and alloxan (Ledoux et al. 1988). Free-oxygen radicals are important mediators of ~-cell destruction in IDDM, and nicotinamide's antioxidant activity has been shown to have some effect on preventing IDDM in high-risk individ-uals and has a slight effect on residual insulin secretion in newly diagnosed patients. Other antioxidants have been tested in animal models with results suggesting prevention of diabetes (Ludvigsson 1993).

Vitamin E (a-tocopherol, 32 in Fig. 8), which occurs in seed oils and green leafy vegetables, has been shown at dos-es of 600-1200 mg daily to reduce the levels of glycosylat-ed hemoglobin in diabetic subjects independently of chang-es in plasma glucose, which may help reduce the incidence of diabetic complications (Ozden et al. 1989, Ceriello 1991).

Coumarin (33), another constituent of Trigonella, is pro-foundly hypoglycemic in normal and alloxan-diabetic rats (Shani et al. 1974). The mechanism for this observation probably involves hepatotoxicity. Coumarin is hepatotoxic in rats and dogs, where it is metabolized through 3-hy-droxycoumarin to reactive quinone metabolites that bind covalently to microsomal proteins. In humans and other primates, however, coumarin is metabolized through 7-hy-droxycoumarin to a glucuronide conjugate that is rapidly excreted, and no hepatotoxicity occurs (Cohen 1979). S_co-poletin (34), another coumarin constituent of Trigonella, exerted borderline hypoglycemic effects in normal and al-loxan-diabetic rats at high doses (Shani et al. 1974). Fenu-greekine (35), a steroidal sapogenin-peptide ester, is an-other hypoglycemic constituent (Ghosal et al. 1974).

Complex Carbohydrates and Postprandial Blood Glucose

Seeds of a number of other members of the Fabaceae are used traditionally to treat diabetes. In addition to direct hy-poglycemic effects of their constituents, dietary effects are also important. Clinical studies of high legume diets showed improvement in many of the indices of blood glu-cose control, especially postprandial levels. Beans are high in complex carbohydrates which are more slowly digested than other types of starch. Non-cellulosic types of dietary fiber such as carob gum and guar gum, high-molecular-weight galactomannans from Ceratonia siliqua L. and Cya-mopsis tetragonoloba (L.) Taub . respectively, slow intesti-nal absorption of glucose by slowing gastric emptying and

HO

32

eel o 0 33 34

3S Fig. 8. Some antidiabetic constituents of Trigonella foenum-grae-cum.

by thickening the unstirred water layer adjacent to the in-testinal villi (Leeds 1981, Karlstrom et al. 1987). Modifica-tion of the physical and chemical characteristics of the in-testinal contents by leguminous gums might also modify the release of gastrointestinal hormones which influence in-sulin secretion and gastrointestinal motility (Forestieri et al. 1989). Provision of purified guar fiber as tablets taken with meals significantly reduced low-density lipoprotein choles-terollevels but did not improve excessive postprandial gly-cemia in ~DDM patients in whom near-normal fasting plasma glucose levels had been obtained with diet, sulfo-nylurea, or human ultralente insulin therapy (Holman et al. 1987). Patient compliance may be a problem with pure guar gum due to its unpalatability and tendency to cause abdominal distension and diarrhea, but incorporation into high-carbohydrate foods has been shown to provide even more effective blunting of the postprandial glycemic profile without gastric distress (Briani et al. 1987).

Some legumes also contain low levels of lectins, which if incompletely destroyed by inadequate cooking, might ac-celerate intestinal motility and increase mucus secretion, thus modifying absorption of glucose (Leeds 1981). The antidiabetic activities of a number of other plant gums were attributed to inhibition of gluconeogenesis and stimulation of peripheral glucose utilization, not to interference with intestinal absorption of glucose (AI-Awadi and Gumaa 1987). Some structure-activity relationships of hypoglyce-mic plant mucilages have been studied (Tomoda et al. 1987). Intestinal bacterial fermentation of leguminous olig-

HO ~ ~2OH HO HOHN~3 HO

~20H HO o HO

~2OH HO 36

OH

37

39

o HO

HO OH

~r~~H HO~~ HO OH

38

Fig. 9. Hypoglycemic intestinal enzyme inhibitors.

osaccharides and fiber, in addition to producing a feeling of satiety that might aid in compliance with a fixed diet, pro-duces short-chain fatty acids which are then absorbed and affect metabolic processes relevant to diabetic control, such as hepatic gluconeogenesis (Leeds 1981).

A microbial product, acarbose (36 in Fig. 9), isolated from strains of Actinoplanes sp. (in the order Actinomyce-tales) (Hillebrand 1987), is known to inhibit the intestinal a-glucosidases, y-amylase, sucrase, and maltase. This ac-tion reduces the release of glucose from carbohydrates, re-sulting in a dose-related delay in, or reduction of, the post-prandial increase in blood glucose and triglycerides, dimin-ished prevalence of diabetic nephropathy, as well as in-creased insulin binding in muscle (Hillebrand 1987, Yoshi-kuni 1988, Le Marchand-Brustel et al. 1990, Hanefield et al. 1991).

Castanospermine (37), an indolizidine alkaloid isolated from Castanospermum australe A. Cunn. (Fabaceae), is an-other example of an intestinal enzyme inhibitor with hypo-glycemic activity. Structurally, castanospermine shares sim-ilarities with the pyranose form of glucose in the orienta-tion of its hydroxyl groups. It blocks the hyperglycemic re-sponse to oral doses of sucrose through inhibition of disac-charase, but does not reduce glucose-induced hyperglyce-mia (Rhinehart et al. 1987). Moranoline (38), isolated from


mulberry (Morus alba L., Moraceae) root bark and also leaves of Jacobinia (Acanthaceae) and cultures of Bacillus and Streptomyces, inhibits intestinal a-glucosidase potently but only weakly inhibits l3-glucosidase, glucoamylase, and a-amylase (Yoshikuni 1988). Miglitol (39), prepared semi-synthetically from moranoline, is an a-glucosidase inhibi-tor which, unlike acarbose, is absorbable from the gastroin-testinal tract. It may exert inhibitory effects on nonintesti-nal a-glucosidase present in various cell types, and has been clinically evaluated as a hypoglycemic agent in both IDDM and NIDDM (Reuser and Wisselaar 1994).

Hypoglycemic Glycans

Hikino's research group (Hikino et al. 1985a-c, 1986a-c, 1988, Konno et al. 1985a-e, Takahashi et al. 1985a,b, 1986, Tomoda et al. 1987, 1990) has isolated a variety of glucans, peptidoglucans, and heteroglycans from plants used in oriental traditional medicine. These complex carbo-hydrates, with molecular weights ranging from approxi-mately 1000 - >10,000,000 amu, were shown to have re-markable hypoglycemic activity when administered intra-peritoneally (i.p.) to normal, alloxan-hyperglycemic, and spontaneously diabetic mice.

The mechanism of action of the glucan aconitan A, from Aconitum carmichaeli Debeaux (Ranunculaceae), involves significant potentiation of the activity of hepatic phospho-fructokinase. Acceleration of glycolysis in the liver was ac-companied by some increase in hepatic total glycogen syn-thetase, but liver glycogen content and plasma and liver cholesterol and triglyceride contents were unchanged, indi-cating that the conversion of glucose into glycogen or lipids does not contribute to the hypoglycemic activity of aconi-tan A. Plasma insulin levels and insulin binding to isolated adipocytes also were unaffected. Stimulation of glucose up-take and metabolism in small intestine tissues was ob-served. Thus, stimulation of glucose utilization in the liver and peripheral tissues is the main mechanism for the hypo-glycemic activity of aconitan A (Hikino et al. 1989a).

Ganoderan B, a glycan from Ganoderma lucidum Kar-sten (Polyporaceae), increases the plasma insulin level in normal and glucose-loaded mice, increases the activities of hepatic glucokinase, phosphofructokinase, and glucose-6-phosphate dehydrogenase, decreases the activities of he-patic glucose-6-phosphatase and glycogen synthetase, and reduces hepatic glycogen content. The observed stimu-lation of glucose metabolism in a homogenate of the small intestine suggests that acceleration of glucose util-ization may also occur in peripheral tissues (Hikino et al. 1989b).

Panaxans A-E, glycans of ginseng (Panax ginseng CA. Meyer, Araliaceae), show different mechanisms of action despite their similar structures. Panaxans A and B stimulate hepatic glucose utilization by increasing the activity of glu-cose-6-phosphate dehydrogenase, phosphorylase-a, and


40

Fig. 10. Sapogenin of ginsenosides and panaxosides: protopana-xadiol Rj = H, ptotopanaxatriol Rj = OH; sugars in glycosides are attached to oxygens at R,-R j

phosphofructokinase. Panaxan A decreases the activity of glucose-6-phosphatase but does not affect hepatic glyco-gen content. Panaxan B has no effect on glucose-6-phos-phatase but decreases glycogen synthetase activity and he-patic glycogen content. Panaxan A does not affect plasma insulin levels and insulin sensitivity, but panaxan B elevates the plasma insulin level by potentiating insulin secretion from pancreatic islets and enhances insulin sensitivity by increasing insulin binding to receptors (Suzuki et al. 1989a,b).

Ginseng contains a number of other hypoglycemic con-stituents, with different mechanisms of action. Adenosine was isolated from a water extract of the rhizomes by bioas-say-guided fractionation, and was shown to enhance lipo-genesis and cyclic adenosine monophosphate (cAMP) accu-mulation in adipocytes, which possess specific adenosine receptors. Some of the sterol glycosi-des known as ginseno-sides (40 in Fig. 10) inhibited adrenocorticotropin-induced lipolysis and at the same doses suppressed insulin-stimulat-ed lipogenesis, while others stimulated the release of insulin from cultured islets (Waki et al. 1982. Ng and Yeung 1985).

Plant Constituents that Modulate Intracellular Second-Messengers

Pancreatic ~-cell membranes possess adenosine triphos-phate (ATP)-sensitive K+ channels which, in the absence of glucose, allow an efflux of K+ to contribute a hyperpolariz-ing membrane current that maintains the hyperpolarized resting membrane potential of the cell. Metabolites of glu-cose and amino acids inhibit this channel. causing a reduc-tion in the hyperpolarizing current, which leads to ~-cell_ depolarization and voltage-dependent Ca~+ uptake. Bind-ing of Ca2+ to calmodulin results in microfilament contrac-tion, resulting in exocytosis of insulin from storage gran-ules. Intracellular ATP is believed to have a second-messen-ger role in inhibiting the K+ channel by almost 99 %, thus making the ~-cell very sensitive to changes in channel activ-ity (Cook et al. 1988, Misler et al. 19891. Tolbutamide (2) specifically mimics the effects of glucose stimulation, depo-

larizing the ~-cells by inhibiting the ATP-sensitive K+ chan-nel, which has been suggested to be the ~-cell receptor for sulfonylureas. The alkaloid quinine (41 in Fig. 11) is also a potent blocker of this channel, although, unlike the sulfon-ylureas, it also blocks Ca2+-activated K+ channels (Cook and Ikeuchi 1989).

Intracellular cAMP also acts as a second-messenger in the ~-cells. Increasing the intracellular cAMP concentration potentiates cholecystokinin- and glucose-stimulated insulin release. The mechanism involves synergistic action with the influx of Ca2+ that occurs as a consequence of the glucose metabolite-induced increase in intracellular K+ (Hill et al. 1987). The physiological actions of glucagon result from stimulation of cAMP synthesis, which in pancreatic ~-cells forms part of the pancreatic hormone regulatory mecha-nism (Lamer 1980). The role of second-messengers in insu-lin action has been reviewed by Saltiel (1990).

The most famous plant product for the stimulation of intracellular cAMP is forskolin (42), a diterpene from Cole-us forskohlii (Poir.) Briquet (Lamiaceae). It is an adenylate cyclase activator which increases intracellular cAMP by stimulating its biosynthesis. Theophylline (43) and other methylxanthenes from Camellia sinensis (L.) Kuntze (Thea-ceae) and flex guayusa Loesner (Aquifoliaceae), and papav-erine (44) from Papaver somniferum L. (Papaveraceae), are phosphodiesterase inhibitors which increase intracellular cAMP by preventing its breakdown (Gearien and Mede 1981, Hill et al. 1987, Zawalich et al. 1988). Theophylline is orally hypoglycemic when administered chronically to normal rats, but this in vivo effect was not attributed to its phosphodiesterase inhibition, but rather due to its induc-tion of intracellular Cal. efflux. Increased extracellular Ca l + might enhance calcium-stimulated ATPases, which would result in decreased cellular ATP levels, enhanced li-polysis, and reduced glycogenolysis. This effect is also seen with administration of caffeine (45) (Tobin et al. 1976).

Sodium salicylate (salt of 14) inhibits cyclooxygenase, thus preventing the metabolic cascade from arachidonic ac-id to the prostaglandins. Inhibition of ~-cell PGE2 synthesis increases glucose-induced insulin secretion because this prostaglandin binds to specific ~-cell receptors that are cou-pled to regulatory components that inhibit adenyl ate cy-clase. Inhibition of this enzyme would lead to a decrease in intracellular cAMP (Robertson 1988). Additionally, arach-idonic acid (46) itself is an insulin secretagogue, acting to mobilize Ca2+, increasing its free cytosolic concentration, and to activate protein kinase C (Metz 1988).

Carbohydrate components of the diet stimulate the re-lease of the hormone "gastric inhibitory polypeptide," which is thought to influence insulin secretion by elevating islet ~-cell cAMP levels. The activity of cAMP is also syner-gized by phosphoinositide-derived second-messenger mole-cules generated during the phospholipase C-mediated cleavage of membrane phospholipids in the ~-cell. This hy-drolysis is thought to be activated by the interaction of ex-


~ HO" "',,= " ~ 0 OH "", OCOCH3 ' H OH 41

44

& ~ OH "l?)::::' I '" HO

OH OH OH

47

Fig. 11. Modulators of intracellular second-messenger systems.

tracellular hormones and agonists with a specific mem-brane receptor (Zawalich 1988).

The flavonoid, (-)-epicatechin (47), isolated as the active principle of the traditional antidiabetic plant Pterocarptls marsupium Roxb. (Fabaceae), has been shown to cause an ATP-dependent enhancement of glucose-stimulated insulin secretion from isolated islets, and to cause a rise in islet in-sulin content in vivo in rats. Inhibition of cAMP phospho-diesterase and stimulation of insulin biosynthesis were sug-gested to be the mechanisms for the observed effects (Hii and Howell 1984). The flavonoids quercetin (48) and my-ricetin (49) have also been reported to be hypoglycemic (Rahman and Zaman 1989), but they are known to be po-tent inhibitors of protein tyrosine kinase (Geahlen et al. 1989), the activity of which is essential in the post-receptor-binding activity of insulin.

42

45

0

48

50

43

46

OH OH

HO OH

OH OH 0

49

When insulin binds to the extracellular a-subunit of its heterodimeric cell surface receptor, the insulin-receptor complexes aggregate along the plasma membrane and are then internalized rapidly. Activation of a Mn2+-dependent protein tyrosine kinase in the transmembrane ~-subunit en-sues, resulting in phosphorylation of the receptor and other proteins with phosphate groups from ATP (Reddy and Kahn 1988). Activation of a phosphatidylinositol-specific phospholipase C leads to hydrolysis of a membrane glycan phosphoinositide. This produces a cyclic inositol phos-phate-glucosamine second-messenger that activates phos-phodiesterase, decreasing intracellular cAMP, and also pro-duces diacylglycerol, which activates protein kinase C (Sal-tiel et al. 1986). Protein kinase C regulates a number of en-zymes and the insulin receptor through phosphorylation (van de Werve 1985a).

152 R.]. Marles and N. R. Farnsworth

Some tumor-promoting phorbol esters, such as 12-0-tet-radecanoylphorbol-13-acetate (TPA, 50), share structural similarities with diacylglycerol, and are potent activators of protein kinase C (van de Werve et al. 1985). Phorbol esters are diterpenes isolated from species of Euphorbia and a few other genera of the Euphorbiaceae (Kinghorn 1983), 30 species of which have been associated with the treatment of diabetes. Phorbol esters have been reported to have a num-ber of insulinomimetic effects, including stimulation of glu-cose transport, lipogenesis, and amino acid uptake. How-ever, they may reduce insulin receptor affinity for insulin, insulin stimulation of glucose transport and lipogenesis, and basal glycogen synthesis (Sowell et al. 1988). It has been suggested that phorbol ester-stimulated serine phos-phorylation of insulin receptors may be associated with a decrease in the affinity of the receptor for insulin and de-creased receptor tyrosine kinase activity, although conflict-ing results have been reported (van de Werve et al. 1985a, Obermaier et al. 1987, Sowell et al. 1988). Ishizuka et al. (1991) found that phorbol esters, glucose, and insulin translocatively activate protein kinase C, resulting in a sub-sequent down-regulation of protein kinase C and insulin-stimulated glucose uptake in adipocytes. This contributes to impaired responsiveness of the glucose transport system after prolonged insulin and/or glucose exposure. Phorbol esters can inhibit ai-adrenergic stimulation of glucose pro-duction by inhibiting phosphorylase activity, also through their effect on protein kinase C (van de Werve et al. 1985b). They can also inhibit glucagon-stimulated adenylate cy-clase, but the metabolic significance of this is much less than that of their inhibition of ai-adrenergic effects (-Garda- Sainz et al. 1985). Tumor promotion may also be explained by phorbol ester activation of protein kinase C (van de Werve et al. 1985a).

Plant Hypoglycemics Acting by Adrenergic Effects In addition to the ai-adrenergic inhibition described

above for tumor-promoting phorbol esters, a number alka-loids are known to affect blood glucose levels by a similar mechanism. In normal patients, there is no effect of a-, ~-, or a+~-blockade on the slope of glucose-potentiated insulin secretion. In patients with NIDDM, only selective a-adre-nergic blockade increases glucose-potentiated insulin secre-tion, through both a decrease in an endogenous overactive a-adrenergic stimulation and an increase in synaptic cleft norepinephrine levels, which results in an increase in islet ~adrenergic stimulation. Thus, a chronic decrease in islet a-adrenergic stimulation may be a useful adjunct to NIDDM management (Broadstone et al. 1987).

Ergot alkaloids, occurring in fungi such as Claviceps pttr-purea (Fries) Tulasne (Hypocreaceae) and at least one group of higher plants, Rivea corymbosa (L.) Hallier f. and closely related Ipomoea and Argyreia species (Convolvula-ceae), are a-adrenergic blockers which inhibit epinephrine-

induced hepatic glycogenolysis and hyperglycemia, but not glycogenolysis in skeletal muscle. These effects are not cor-related with their well-known smooth muscle effects, and may not be due to a specific a-receptor effect (Weiner 1980). Dihydroergotamine (51 in Fig. 12) and yohimbine (52), another a-adrenergic blocking alkaloid from Pausi-nystalia yohimbe (K. Schumann) Pierre (Rubiaceae), pre-vented epinephrine-induced inhibition of insulin release, but not diazoxide-induced inhibition (Henquin et al. 1982). However, yohimbine is also a monoamine oxidase inhibitor and is contraindicated for patients with diabetes (Tyler et al. 1993).

Beta-adrenergic blocking agents reduce the hyperglyce-mic response to epinephrine by blocking its stimulation of cAMP production. Epinephrine-induced glycogenolysis in heart and skeletal muscle and lipolysis in isolated rat adi-pocytes is inhibited. By these mechanisms, the non-selective

~-adrenergic blocking agent, propranolol, slows the post-insulin recovery of glucose concentration and prevents the usual rebound of plasma glycerol, while not affecting plas-ma glucose or insulin concentrations in normal individuals, or the rate or magnitude of the fall of plasma glucose after insulin (Weiner 1980). Beta-adrenergic blocking agents can also reduce insulin resistance caused by ~-adrenergic stimu-lation (Attvall et al. 1987). Kimura et al. (1988) suggested a possible ~-adrenergic blockade mechanism for the hypo-glycemic activity of an orally-administered aqueous extract of Ganoderma lucidum (Leyss. ex Fr.) P. Karst (Ganoder-mataceae).

Reserpine (53), from Rauvolfia serpentina (L.) Benth. ex Kurz (Apocynaceae), is an adrenergic blocking agent that causes intracellular depletion of catecholamines and serot-onin. Uptake of catecholamines is also antagonized by inhi-bition of the ATP-Mg2+-dependent uptake mechanism of the chromaffin granule membrane. A transient sympa-thomimetic effect is seen only after parenteral administra-tion of relatively large doses; pharmacological effects of the released catecholamines are minimal unless monoamine oxidase has been inhibited (Weiner 1980). Reserpine en-hanced the hypoglycemic effect of insulin and the hypergly-cemic effect of epinephrine in normal subjects. In glucose tolerance tests it inhibited the hyperglycemic response, even in diabetic patients (Ricci and Ricordati 1955). However, hypoglycemia is not reported as a significant side-effect of reserpine, nor are interactions with other hypoglycemic drugs listed (American Pharmaceutical Association 1976).

Photosensitizers and IDDM

Insulin-dependent diabetes may arise through T-Iympho-cyte mediated ~-cell destruction. One possible novel ap-proach to interrupting this pathogenic process is photo-pheresis, whereby lymphocytes would be treated with a photosensitizer such as 8-methoxypsoralen (54 in Fig. 13) and UVA radiation to cause a change in the antigenicity of

OH

SI S2

S3 Fig. 12. Adrenergic-blocking hypoglycemic alkaloids.

the treated lymphocytes. This is postulated to cause a vac-cination-like effect in the patient when they are retrans-fused at repeated intervals into the patient. This has proved effective in other autoimmune diseases and is now in clini-cal trials for IDDM (Ludvigsson 1993). Photosensitizers have been isolated from more than 30 flowering plant fam-ilies (both monocots and dicots) and represent a wide range of chemical classes including: polyacetylenes, thiophenes, lignans, porphyrins, quinones, chromenes, benzofurans, fu-roflavonoids, furocoumarins (e.g. 54), furochromones, fu-roquinoline alkaloids, and (3-carboline alkaloids (Downum 1986, Hudson 1990). A thiophene such as a-terthienyl (55) may have an advantage over 8-methoxypsoralen in these applications because of its lack of genotoxicity (MacRae et al. 1980, Tuveson et al. 1986). Structure-activity relation-ship studies of thiophenes have shown the possibility of achieving some cell or organism specificity despite the gen-eral mechanism of action involving singlet oxygen genera-tion (Maries et aI. 1992).

Plant Hypoglycemic Drug Screening Methodology

Scientific investigation of traditional medicines, as in the examples provided above, has resulted in the discovery of a number of promising leads for new antidiabetic agents. Modern drug discovery requires a systematic approach to optimize time and resource use for testing the maximum number of samples in bioassays which hopefully are predic-tive for therapeutic efficacy. These approaches to bioassay-guided antidiabetic drug discovery can be divided into two main classes: in vivo and in vitro techniques.


ff)I 'o~oAo OCH3 S4 5S

Fig. 13. Plant-derived photosensitizers.

In Vivo Techniques

Techniques for the study of hypoglycemic activity in vivo employ animals with normoglycemia or induced hypergly-cemia, as well as diabetic humans. Methods used to experi-mentally induce hyperglycemia include loading with glu-cose or cholesterol, treatment with drugs such as alloxan, streptozotocin, 2,4-dinitrophenol, and diazoxide, hor-mones such as epinephrine, glucagon, corticotropin, soma-totropin, and anterior pituitary extract, and surgical proce-dures such as partial or full pancreatectomy. Genetically obese and hyperglycemic animals such as Zucker falfa rats (e.g. Rosen et aI. 1986, Young et a!. 1991), yellow KK mice (e.g. Kanamaru et al. 1985), spontaneously diabetic mice of strain C57BUKsj-db/db (Suzuki and Hikino 1989), and sand rats (Psammomys obesus) (e.g. Aharonson et al. 1969) have also been used.

The most popular in vivo models for diabetes are rodents treated with alloxan (56 in Fig. 14) or streptozotocin (57). The history and mechanism of alloxan, a pyrimidine deriv-ative, has been reviewed by Lenzen and Panten (1988), who point out that, while it is a very selective toxin of pancreat-ic (3-cells through its inhibition of glucokinase, thus making it a good model for diabetes mellitus, there are a number of problems with its use. Alloxan's chemical instability, rapid metabolism, thiol reactivity, and effects of factors such as diet, age, and species, have made it almost impossible to es-tablish a clear relationship between the dose of alloxan and its effective concentration in the pancreas. Thus it is diffi-cult to be certain of the extent of (3-cell inhibition and ne-crosis from a particular dose of alloxan.

Streptozotocin, also known as streptozocin, is a natural nitrosourea glycoside isolated from Streptomyces achromo-genes, which also causes degeneration of pancreatic (3-cells. A single dose in the neonatal rat can produce an experimen-tal model of NIDDM, characterized by deficient insulin biosynthesis and release in response to glucose and dimin-ished pancreatic insulin content. There is a selective lack of sensitivity of the l3-cells to glucose and glyceraldehyde, but continued response to other secretagogues. The insensitiv-ity of the islets to glucose is associated with deficient islet glucose metabolism, probably due to a streptozotocin-in-duced alteration in islet mitochondrial function (Portha et al. 1988). Looking at adipocyte insulin binding and glucose transport, however, Fantus et a!. (1987) concluded that the neonatal streptozotocin-injected rat model did not provide a complete representation of human NIDDM.


;;~; H~OH NHCON(NO)CH3

S6 S7

Fig. 14. Commonly used drugs for creating models of diabetes mellitus.

For a model of IDDM where there is a total absence of p-cell function, pancreatectomy is sometimes used. However, at least in rabbits, due to the extended distribution of the pancreas and its close association with the duodenum, a to-tal pancreatectomy may not be feasible or totally successful if attempted. An in vivo bioassay employing surgical re-moval of the pancreas and a complementary injection of al-loxan was shown to give blood glucose values significantly different from those of animals with only surgical interven-tion (Ibanez-Camacho et al. 1983).

A further complication of in vivo hypoglycemic screening was described during an investigation of aqueous extracts of Tecoma stans Juss. (Bignoniaceae). Although initial in vi-vo hypoglycemic screening of the extract gave inconclusive results, chemical investigations resulted in the isolation of two monoterpene alkaloids, tecomanine (58 in Fig. 15) and tecostanine (59), which were shown to be hypoglycemic when administered i.v. or p.o. in rabbits (Hammouda et al. 1964, Hammouda and Amer 1966). The crude aqueous ex-tract of the plant, when administered i.v. to fasted dogs or i.p. to glucose-loaded rats, produces a sharp but transient (10 min) fall of arterial blood pressure and a transient (180 min) but significant hyperglycemia due to induction of hepatic glycogenolysis and subsequent elicitation of insulin release. This was followed by a slight hypoglycemia with a maximum decrease of the blood glucose level occurring from five to six hours after injection (Lozoya-Meckes and Mellado-Campos 1985, Meckes-Lozoya and Ibanez-Cama-cho 1985). Further investigations determined that the in-itial hypotension and hyperglycemia could be abolished by administration of antihistamines or by filtration of the ex-tract with a 0.5 pm pore-size membrane capable of retain-ing high molecular weight compounds such as proteins and kinins, which might cause the release of histamine. The late hypoglycemic effect remained, and thus is not secondary to the initial hyperglycemia (Meckes-Lozoya and Lozoya 1989).

A number of other plants, including Allium cepa, A. sati-va, Brassica o/eracea, Hordeum vulgare, Oplopanax hor-ridum, Phaseo/us vulgaris, Saccharomyces cerevisiae, Urti-ca dioica, and Vaccinium myrtillus have been reported to contain hyperglycemic as well as hypoglycemic constituents (Oliver-Bever and Zahnd 1979). Caution should therefore be employed in interpreting the results of in vivo adminis-tration of crude extracts.

S8 S9

Fig. 15. Hypoglycemic alkaloids from Tecoma.

There is extensive evidence for involvement of both cellu-lar and humoral immune phenomena in the destruction of pancreatic p-cells characteristic of IDDM (Spencer and Cudworth 1983, Bottazzo 1986, Montana et al. 1989). Im-munosuppressive drug therapy has been recommended in some cases of IDDM (e.g. Vardi et al. 1986). An enzyme-linked immunosorbent assay (ELISA) has been developed as a means of quantifying humoral immune responses in rats exposed to immunomodulating chemicals (Koller et al. 1983). This assay could be used for screening plant extracts and isolates for immunomodulating activity.

Unquestionably, in vivo bioassays are essential to prove the value of new hypoglycemic agents. However, whole an-imal tests reveal relatively little about the mechanism of ac-tion of the compound, and it can be seen from the previous section that there are a great many mechanisms by which blood glucose levels may be reduced. Some of these mecha-nisms, such as those involving hepatotoxicity, are obvious-ly not useful in diabetes therapy. The lack of perfect models for NIDDM and IDDM, coupled with financial restrictions on obtaining and maintaining animals, and social restric-tions on extensive use of animals in experimentation, indi-cate that a more practical approach would involve a series of in vitro prescreens before testing a potential new hypo-glycemic agent in animals. This is in agreement with the po-sition statement of the American Diabetes Association (1990) that antidiabetic research should use alternative methods to live animals when appropriate.

In Vitro Techniques

Many in vitro techniques have been developed to eluci-date the varied mechanisms of action of hypoglycemic agents discovered by in vivo bioassays. For the purpose of screening large numbers of plant extracts and chromato-graphic fractions in order to isolate novel hypoglycemic agents, some of these in vitro bioassays should be employed as the first steps rather than the last steps of drug discovery.

Three aspects of the hypoglycemic response are common-ly studied in vitro: insulin release from the pancreatic islets, peripheral insulin binding and glucose utilization, and ef-fects on hepatic enzymes.

For studying the effect of potential hypoglycemic agents on the release of insulin, perfused pancreas, intact isolated islets, and dispersed islet cell techniques have been devel-

oped. Characteristics of insulin and glucagon release from these preparations have been studied comparatively by Weir et al. (1986). Most of the original work was done with tissues from rats, for which the experimental techniques of isolation and culture are well established (Larner and Pohl 1984a,b, 1985, Pipeleers 1986,1987, Pipeleers et al. 1991). More appropriate to large scale screening procedures are the techniques of Ricordi et al. (1986, 1988) for the mass isolation of porcine and human pancreatic islets. Much of the recent work on the mechanism of sulfonylureas at the cellular and subcellular level has been done with cultured ~cells (e.g., Boyd 1988, Gorus et al. 1988, Garvey 1992, Lienhard et al. 1992).

Non-insulin-dependent diabetes mellitus is not due just to a defect in the ~-cells, but rather to a collusion between

~-cells, the liver, and peripheral tissues (DeFronzo 1988, 1992, Mueckler 1990, Granner and O'Brien 1992). Hepat-ic involvement in diabetes and its therapy has been studied in primary cultures of rat hepatocytes (e.g., Salhanick et al. 1983, Rinninger et al. 1984, McCormick et al. 1986) using techniques developed by Fry et al. (1976) and Bellemann et al. (1977). More recently, a human hepatoma cell line has been used to study insulin receptors (McClain and Olefsky 1988). Hikino's group has done mechanistic studies on plant hypoglycemic agents with a variety of hepatic enzyme preparations (Hikino et al. 1989a,b, Suzuki and Hikino 1989).

For studying in vitro insulin resistance, insulin internal-ization, and glucose transport in peripheral tissues, the most common techniques involve cultures of skeletal mus-cle strips or cells (Beck-Nielsen et al. 1992) or adipocytes derived from rat epididymal pads (Jochen and Berhanu 1987) or from humans by surgical excision (Kashiwagi et al. 1983) or less invasive needle biopsy (Yki-]iirvinen et al. 1986). The effect of natural products on glucose uptake and metabolism in peripheral tissues has also been studied by use of fragments or a homogenate of the rat's small in-testine (Hikino et al. 1989a,b). These methods could also be adapted to use larger animal tissues available from slaughterhouses.

Screening techniques have been developed to detect in vi-tro natural products that show immunoreactivity with guinea pig insulin (de Pablo et al. 1986), inhibition of long chain farty acid oxidation (Kanamaru et al. 1985), eleva-tion of intracellular cAMP concentration (Swanson et al. 1988), and inhibition of protein-tyrosine kinase activity (Geahlen et al. 1989).

Finally, the measurement of insulin levels is a critical step in several of the bioassays. The original immunoassay tech-nique (Wright et al. 1968, 1971, Makulu et al. 1969) was replaced by a radioimmunoassay technique that earned a Nobel prize for its developer (Yalow 1978), and is still the most widely used technique. However, an enzyme-linked immunosorbent assay (ELISA) with increased sensitivity, high accuracy, and greater practicability (Kekow et al.


1988) may soon replace radioimmunoassay as the method of choice.

Toxicity of Hypoglycemic Plants

If most hypoglycemic plant constituents have arisen through coevolution as chemical defense compounds, then it should be recognized that for the source plant's survival, the best strategy is a non-selective toxin which will deter herbivory regardless of the species of herbivore attacking it. Often the development of new drugs from plants does not involve increasing the potency of the lead natural product because this has been optimized by millions of years of co-evolution. Rather, the task is to achieve optimum selectivity and minimize general toxicity. Quantitative structure-activ-ity relationship analysis is an essential tool for achieving this goal.

While a long history of traditional medicinal use may suggest that a plant is relatively non-toxic, this should be confirmed by in-depth literature review and properly-con-trolled experimental bioassays. Some of the reports of tox-icity for antidiabetic plants are derived from case studies or Poison Control Center reports of human poisoning or inju-ry. Species known to contain toxic constituents, such as pyrrolizidine alkaloids, were recorded in the database for this review as toxic, even though the actual concentration in the plant may not be known.

Information was also included from acute toxicity stud-ies. Usually performed by i.p. injection of extracts into ro-dents, they do not necessarily relate closely to human oral toxicity (Irwin 1962). Also, the technique is not employed with as much standardization as it should be, e.g., extracts are prepared differently, and mortality may be recorded af-ter 24 hours (der Marderosian et al. 1976) or 7 to 14 days (Klaasen 1980). For this study a plant was considered tox-ic if the median lethal dose (LDso) by i.p. administration in mammals was 500 mg/kg or less.

Approximately one-third (377 species) of the plants asso-ciated with the treatment of diabetes are to be considered toxic by the above criteria, while for another third their safety is uncertain. In some cases, such as the ingestion of unripe akee fruit (Blighia sapida Koenig, Sapindaceae), tox-icity is expressed in part as a profound hypoglycemia caused by the constituents, hypoglycin A and B (27 and 28 in Fig. 6). There are many other toxicological effects of plants which may result in hypoglycemia, such as hepato-toxicity or ~-adrenergic blockade. Many plants used to treat diabetes or shown experimentally to be hypoglycemic have toxic effects unrelated to their desired effect.

Toxicity is influenced by the plant part, method of prep-aration, route of administration, and test organism. For ex-ample, the leaves of Abrus precatorius L. (Fabaceae) are used in traditional medicine to treat diabetes, and both the leaves and roots have been used to sweeten foods. While

156 R. J. Maries and N. R. Farnsworth the leaves and roots are relatively non-toxic and non-muta-genic (Choi et al. 1989), the seeds contain the glycoprotein, abrin, one of the most potent of all known botanical toxins, with a minimum lethal dose of 0.7 pg/kg when adminis-tered i.v. to mice. Sublethal doses of abrin i.v. are hypogly-cemic (Fodstad et al. 1979), but since a single well-chewed seed can be fatal to a human (Lampe and McCann 1985), the seed or isolated abrin are not suitable alternative hypo-glycemic agents.

When calculated on the basis of dose per body weight, humans are generally vulnerable to a drug at a dose one-tenth that shown to have the same effect in experimental animals; when calculated per unit of body surface area, toxic effects in man are usually within the same dosage range as animals (Klaassen 1980). With i.p. administration, the peritoneal cavity offers a large absorbing surface from which drugs enter the circulation rapidly. The dose-re-sponse relationship might be quite different from oral ad-ministration, where absorption from the gastrointestinal tract is governed by a wide variety of factors, including pro-portion of the drug in non-ionized form, presence of food, gastric emptying time, decomposition of the drug by gastric acids and enzymes, diffusion rate across the gastrointestinal epithelium, and the "first-pass effect" of gastrointestinal epithelial and hepatic drug-metabolizing enzymes (Mayer et al. 1980).

Since a very small dose of some toxic drugs provides im-portant therapeutic effects, while a large dose of other drugs with low toxicity is required to achieve the desired ef-fect, more useful information would be provided by the Therapeutic Index, which is generally expressed as a ratio of the median lethal dose to the median effective dose (LDsJEDso), or the Certain Safety Factor (LD1IED99 ). However, such information is rarely available for plants other than those already well known to modern pharma-cology.

Allergenicity and photosensitization are other aspects of toxicity that would not be revealed by regular acute toxic-ity tests, yet are significant risks in the therapeutic use of plants, especially when employing members of the Anacar-diaceae (urushiol), Asteraceae (thiophenes, sesquiterpene lactones), Hypericaceae (hypericin), and Apiaceae (furano-coumarins). Lewis and Elvin-Lewis (1977) and Lampe and McCann (1985) have tabulated many of the plants believed to cause these problems. The mutagenicity of any com-pound with potential for therapeutic use should also be ex-amined (Ames et al. 1975, Skopek et al. 1978a,b).In gener-al, little is known about the chronic toxicity of plants. Since diabetes mellitus is a chronic condition with no known cure, antidiabetic drugs must be taken for the lifetime of the patient. It is therefore important that chronic toxicity stud-ies be performed before recommending a plant-derived drug for antidiabetic therapy.

Prospects for Future Antidiabetic Plant Research

For both the discovery of locally available alternative medicines to treat diabetics in developing countries, and for the commercial development of new botanical hypoglyce-mic agents and adjuncts to antidiabetic therapy, the best strategy will involve the study of traditional antidiabetic plants.

There will be a number of obstacles to overcome, not the least of which is financial. To bring a new drug to market, it will likely cost more than $300 million and 10 years to perform the pharmacological and toxicological testing re-quired by current strict regulations such as those of the u.S. Food and Drug Administration (Soejarto and Farnsworth 1989). Only pharmaceutical companies can afford this type of investment, and they will only undertake such projects if they can be assured of recovering their costs and making a profit, through patent protection. While it is possible to patent a natural product, particular applications, and de-rivatives made from it, it is difficult to obtain the degree of patent coverage for a plant isolate that would be desired by most companies (Tyler 1979).

The cost of bringing a new plant-derived drug to market could be reduced substantially by changes in government regulations regarding the methods for proving efficacy and safety of traditionally used drugs from natural sources. The current regulations of West Germany and those under de-velopment in Canada could serve as models (Morrison 1984, Tyler 1987, Blackburn et al. 1986, 1993, Liston 1986, 1987, 1990, Canada Department of National Health and Welfare 1992).

Despite the difficulties, the financial rewards of success in marketing plant-derived drugs are great. In the United States, 25 % of all prescriptions dispensed from community pharmacies in 1980 contained active principles prepared from higher plants. Consumers paid more than $8 billion for these prescription natural products, which include such essential medicines as vincristine, digitoxin, quinidine, and L-dopa (Farnsworth et al. 1985).

The supply of medicinal

Documents

Marles & Farnsworth 1995 Antidiabetic Plants