The importance of chromium (Cr) for glucose metabolic regulation has been seen in clinical states of relatively severe Cr deficiency, characterized by impaired glucose tolerance, fasting hyperglycaemia and eventually lipid disorders 10 11 . Severe deficiencies are mostly limited to parenteral nutrition but analysis of diet plans currently in use in United States reveals a lack of Cr in many diets, potentially linked to various chronic diseases 12 . The safe and adequate daily intake of Cr was considered to be in the range 50-200 mg. However, now 30 mg/d seems a newly admitted value. Cr is found in relatively high amounts in barley 13 . Cr is poorly absorbed and only toxic in high doses, depending on the milieu, the tissues and the valence of the metal: hexavalent Cr is highly toxic whereas the in vivo toxicity of trivalent Cr is debated 10 14 . After a period of intense utilization, Cr seems to be less popular nowadays, as recently reviewed 15 . Studies performed over the last 20 years in both animals and humans have been largely criticized and whether Cr is indeed an essential element has been questioned 15 16 .

Cr has no effect in healthy individuals and, because these are not Cr-deficient, there is no need for supplementation. Cr has been administered and tested in various forms such as Cr-chloride, Cr-nicotinate, Cr-proprionate, Cr-histidinate or Cr-picolinate, the latter being by far the best investigated form because of its solubility in water at neutral pH. However in acidic milieu it hydrolyzes and it has been shown that only about 1% of absorbed Cr is found in the bloodstream 15 . Most of the studies published about biological effects of Cr have indeed used Cr-picolinate. Recently new forms aimed at improving Cr bioavailability have been investigated: a complex of Cr and D-phenylalanine 17 , Cr-dinicocysteinate 18 and a combination of Cr and biotin 19 .

Several mechanistic studies showed improvements in insulin receptor/postreceptor signalling, leading to increased glucose transport by enhanced activity of the hormone-sensitive GLUT-4 transporters. These effects can be due to the receptor itself, its control mechanisms or its immediate environment in the cell membrane. Cr-picolinate was shown to induce various effects: increase in AMPKinase and p38 MAPkinase activities 20 , increase in receptor phosphorylation 21 and increase in insulin receptor mRNA (in addition to mRNA for GLUT-4, glycogen synthase and UCP3) 22 . Increased synthesis of IGF receptors, able to functionally replace failing insulin receptors, was also described 23 . Changes in the environment of insulin receptors were shown as reductions in plasma membrane cholesterol 24 25 , leading to improved cytoskeletal functioning and enhanced insulin sensitivity. Decreases in TNFα 26 27 , resistin 28 , interleukin-6, CRP 27 as well as increases in vitamin C or adiponectin 29 were all positive mechanisms to reduce insulin resistance by Cr.

Finally, one should mention that Cr can bind directly to insulin, in particular to insulin dimers 30 , thereby possibly stabilizing the hormone structure and/or modifying its receptor binding. Interestingly the Cr binding to insulin receptors occurs on a sequence of 3 aminoacids which are identical to those described for GTF decades ago (see below).

The appearance of glucose intolerance or even fasting hyperglycaemia in demonstrated cases of Cr deficiency have rapidly led to investigations on the potential of Cr supplementation in insulin resistance and diabetes. Furthermore, diabetic patients have about 1/3 less Cr in various biological samples 31 32 33 . The reasons for this have recently been discussed 15 . While it was shown that inorganic Cr was poorly active, Mertz and coworkers identified as early as 1955 a complex found in brewer's yeast, liver and kidney which was able to stimulate glucose transport in the presence-but not in absence- of insulin and which contained Cr. They termed it GTF [glucose tolerance factor]. Its activity required insulin and was dependent on its Cr content, ideally 4 ions/molecule, while 8 ions/molecule revealed toxic 34 35 . GTF (MW about 1500) appeared to be a mixture of Cr, nicotinic acid and the 3 aminoacids constitutive of glutathione 36 . Replacement of Cr by other metals did not reproduce GTF biological activity. It increased insulin efficacy in adipocytes 35 37 at relatively high concentrations but no effects was found in diabetic patients despite demonstrable increases in Cr content of various tissues 38 . Presently the GTF story is highly criticized because of missing informations and doubtful technical purification procedures 15 . The GTF story continued with description of a similar structure termed LMWCr.

As for many therapeutic trials, concern must be raised as to the concentrations used. Especially in vitro and in animals studies are usually performed with irrelevant concentrations of the active compound, leading to so-called "pharmacological" effects. Accordingly these data should be viewed with appropriate caution. In vitro many beneficial effects of Cr salts have been reported for insulin resistance or diabetes; Cr chloride increased glucose transport in fat and muscle cells as well as in cardiomyocytes 17 25 35 . This was partly explained by higher translocation of Glut-4 transporters 20 . It also decreased oxidative stress, glycosylation and lipid peroxidation in erythrocytes and monocytes under hyperglycaemic conditions 26 39 40 41 42 .

In animals Cr-picolinate reduced AUC for glucose in type1 diabetic rats 37 . Insulin resistance improved in high fat/low Cr diet-fed rats 43 , in sucrose-fed rats 44 , in Zucker diabetic fatty rats 29 , in GK rats 45 46 , in JCR-LA corpulent rats 47 and in db/db mice 48 . A study in obese Zucker rats failed to report improvements 49 .

In addition to positive effects on glucose and insulin metabolic profiles, some studies also reported reduction in triglycerides 37 and stimulation of β-oxidation 50 .

In contrast to animal studies, human studies look much more controversial. This may be due to large differences in Cr administration protocols. Positive reports with Cr exist in type2 diabetics treated with sulfonylureas 51 or in combination with either biotin 52 or vitamins C and E 53 . Many human studies failed to see improvements in diabetes, as evidenced by reviews and meta-analyses 54 55 56 . It is likely that patient selection is cardinal 57 and that Cr supplementation is primarily of interest in patients suffering from clearcut Cr deficiency if the aim is to reduce hyperglycaemia 58 or as an adjuvant to already-existing treatment with oral antidiabetics. Changes in blood lipids in humans were not conclusive.

Based on mechanistic hypotheses (see below) the best application might be insulin resistance (prediabetes, metabolic syndrome), although this is also controversial: reduction in insulin resistance was observed in diabetics 53 59 but also in obese women suffering polycystic ovary syndrome 60 . Another study in metabolic syndrome found no effect, however 61 .

Zinc (Zn) is an essential micronutrient which has an important role in the functioning of hundreds of enzymes 62 , in insulin metabolism and acts as an efficient antioxidant 63 64 . Consequently many observations relate to Zn deficiency, although clinical signs of deficiency are quite modest. Zn is found mainly in cereals, meat, seafood and dairy products 65 . Although usual intakes of Zn are harmless, the range between safe and unsafe is relatively narrow 66 . Because Zn has no storage form, there is need for a constant supply and Zn availability to cells is particularly well regulated, albeit poorly understood 67 68 . Zn absorption can be reduced by iron or inflammatory bowel diseases. It is transported across cell membranes via two families of transporters: ZnT which stimulate Zn efflux and Zip which stimulate its influx 69 . Over 20 transporters are identified and most are intracellular 70 . The implication and regulation of Zn transporters in chronic diseases has been reviewed 71 72 . Concerning metabolic diseases (insulin resistance, metabolic syndrome, diabetes), Zn is considered important mainly because 1) it plays a major role in the stabilization of insulin hexamers and the pancreatic storage of the hormone 73 and 2) it is an efficient antioxidant 74 , while oxidative stress is considered to be a main component in initiation and progression of insulin resistance and diabetes 7 75 .

Severe Zn deficiency is not frequent but concerns have nevertheless been raised about Zn levels in diabetic patients because of increased excretion due to polyuria. Whereas some studies have reported Zn deficiency in type 2 diabetes 31 76 77 78 , others failed to find significant differences with healthy subjects 79 80 . In type 1 diabetes, Zn deficiency is expected to increase the pancreatic damage 81 . A recent report described reduced levels of Zn in obese, insulin resistant subjects 82 . Lower Zn plasma concentrations were found in type 2 diabetics but they were not related to glycemic status or diabetes duration or components of the metabolic syndrome 83 . Interestingly it was reported that diabetic have elevated levels of copper 78 80 and it could be that copper is in fact linked to metabolic syndrome and diabetes 80 84 . Nevertheless, reduced Zn levels in diabetics appear to be related to increased risk for coronary artery disease 77 and mortality 85 .

Several modes of action have been described to explain the improved action of insulin by Zn. It appears that Zn can have direct insulin-like effects, which may be due to inhibition of the important glycogen-regulating enzyme GSK3 86 . Other mechanisms include stimulation of the postreceptor proteins Akt and PI3-kinase. Zn can also reduce cytokines such as IL-1β as well as NFκB 87 88 . Zn induces metallothionein synthesis, whereby Zn may have indirect efficacy 87 . Finally it has been suggested that the important Zn concentrations for its action are the intracellular -rather than plasma- levels 89 .

Zn is a remarkable antioxidant: it acts at specific sites where it can compete for iron and copper; it further binds to SH groups in proteins, protecting them from oxidation 90 . Because Zn controls metallothionein expression it is involved in cellular redox regulation 91 92 . Oxidative stress can be corrected by dietary Zn as demonstrated by an elevation of hepatic antioxidant enzymes 93 . In contrast, a recent study showed that in vitro even micromolar concentrations were cytotoxic under prevailing H2O2-induced oxidative stress 94 . The latter finding is in line with many observations on prooxidant potential of antioxidants, according to doses and present situation 95 96 .

Despite the potential interest in Zn in diabetes only some investigations have been published (reviewed in 62). However the conclusions are far from clear. In animals Zn prevented the development of type 1 diabetes induced by alloxan or streptozotocin 87 . In a study on Zn restriction during pregnancy in rats, offsprings presented reduced body weight, increased fat mass and low lean mass and showed reduced insulin response to glucose challenge. Zn supplementation corrected the situation only in females 97 . In db/db mice Zn reduced hyperglycaemia and hyperinsulinaemia 98 . Administration of Zn for 3 months reduced proteinuria 99 . Zn restriction increased oxidative stress and the levels of isoprostanes 100 , in line with antioxidant effects of Zn. Excessive Zn, on the other hand, increased metabolic syndrome in rats fed a high fat/sucrose diet 101 ; this could be reversed by copper and magnesium-enriched poultry egg 102 .

High Zn slightly reduced the risk for type 2 diabetes in a prospective study performed in 8300 women and the effect was limited to the Zn-deficient subgroup 103 . Another study, however, reported aggravation of glucose intolerance in Zn-deficient diabetic patients 104 . In microalbuminuric type 2 diabetic patients Zn lowered homocysteine levels while increasing vitamin B12 and folate levels 105 . In obese, non-diabetic subjects insulin sensitivity was improved without changes in leptin levels 106 . However no preventive effect against diabetes development could be found in obese women 107 . Oxidative stress was reduced by Zn in healthy elderly subjects 108 .

Some particular forms of Zn have been tested recently. Zn-α2-glycoprotein [ZAG] is an adipokine which stimulates energy expenditure in skeletal muscle and brown adipose tissue, resulting in reductions in body weight, glycaemia, triglycerides and NEFAs as found in ob/ob mice. It increases the pancreatic insulin content and improves the glucose tolerance test 109 . Its levels are lower in obese human subcutaneous and visceral adipose tissue and liver, but interestingly does not appear to be related to insulin resistance 110 .

Zn-dithiocarbamate and a complex of Zn with garlic-originated allixin improved diabetes in KK mice 111 112 .

Selenium (Se) is relatively well absorbed from diet, better so if it is in an organic form 113 . It acts as an antioxidant in the form of selenoproteins which contain selenocysteins 114 . About 10 different selenoproteins are described, the best known being glutathione peroxidases, thioredoxin reductases and iodothyronine deiodinases 115 . Selenoproteins are also responsible for the transport of Se to tissues 116 117 . Severe Se deficiency is rare, while reduced Se levels are seen in diabetics together with increased oxidative stress. In offsprings of diabetic patients Se correlates inversely with CRP and with insulin resistance if Se levels are below 80 μg/l 118 . Excess of Se results in excessive oxidative stress and inflammation as well as diseases like idiopathic scoliosis in adolescents 119 or amyotrophic lateral sclerosis 120 . Presently a lively debate exists as to the benefit of supranutritional doses of Se in cancer 121 .

In view of the potent antioxidant 122 123 and anti-inflammatory effects of Se and the prominent role played by these disorders in insulin resistance and diabetes, supplementation with Se in such diseases appears worthwhile. Interestingly there are only few studies available.

In vitro Se inhibited hyperglycaemia or hyperinsulinaemia-induced expression of adhesion molecules via reduction in p38MAPkinase 124 . It also reduced NFκB, thereby reducing inflammation, CRP and soluble L-selectin 125 .

In type 1 diabetic rats Na-selenite protected mitochondria from oxidative stress 126 . Another study showed a better effect with Se-methionine than with Na-selenate 127 , confirming other data 128 . In db/db mice selenate reduced gluconeogenesis and inhibited phosphotyrosine phosphatases by 50% 129 . This effect was attributed to formation of selenite from selenate.

In humans lower levels of sialic acid and triglycerides were reported in young adults having the highest dietary Se intake 130 . Reduction in complement factor 3 [C3] where also reported, which may be beneficial considering the link between C3 and body mass index, skin thickness and triglycerides 131 . However no positive effect was found on diabetes prevention and some data even pointed towards an increased risk 132 . A cross-sectional study in almost 9000 american adults as well as another analysis reported a positive link between high Se levels and diabetes 133 134 . Therefore, as is the case for use of high Se in cancer 135 , caution applies to utilization of Se supplementation in diabetes, especially since quite high doses are required to observe beneficial effects 113 129 .

Vanadium (Va) has been known for about one century to possess antidiabetic properties 136 . However only over the past 2 decades have concrete therapeutic applications been performed. In humans it is uncertain whether Va is an essential element and it is poorly absorbed in its inorganic form. In vivo Va is transformed into vanadyl cations via glutathione and complexes with ferritin and transferrin 137 . Indeed Va is known to accumulate in some tissues and concerns about its safety have been raised. Bones, liver, kidney lesions as well as loss of weight and gastrointestinal discomfort are well described side-effects of Va, in particular with inorganic Va, whereas organic Va complexes appear more safe 137 138 . Recent reports on Va toxicity show mitochondrial lesions, manifested as swelling and opening of the membrane transition pore due to oxidative stress 139 . Lung cancer in mice is another recently described complication with Va pentoxide 140 .

Va is usually described as "insulin mimetic", in view of several in vitro experiments looking at insulin receptor signaling. The most evident mechanisms is an inhibition of phosphotyrosine phosphatases, leading to increased phosphorylation of IRS-1, PKB, GSK3 and FOXO1 141 142 . Inhibition of phosphatases is supposed to result from production of peroxyvanadate 137 . Restoration of glucokinase activity was reported in the liver of Va-treated diabetic rats 143 . An ex vivo study in Va-treated gerbils suggested that Va was potentiating, rather than mimicking, insulin and that the main action was on GLUT-4 trafficking 144 . This could explain its poor effects in insulin resistant, normoglycaemic states 145 146 .

In animals; many positive reports exist showing beneficial effects of various Va salts in mild or severe diabetes. For example, Va sulphate or more elaborated organic complexes such as bis(maltolato)oxovanadium [BMOV] have proven efficacy in both type 1 and type 2 models 144 147 . More recently arylalkylamine derivatives reduced glycaemia in various models, alone 148 or in combination (as low doses) with substrates of semicarbazide-sensitive aminooxidase SSAO 149 .

In humans many trials were performed. However, as exposed in a review 150 , almost all these trials were short-term and included few patients. Glycaemia-reducing activity was seen at doses of 100 mg/d or more 145 151 152 153 . No efficacy was found, however, in obese non diabetic subjects 145 or in patients with impaired glucose tolerance 146 . Improvements in insulin-sensitivity were claimed to be due to a hepatic effect rather than to an improvement in peripheral insulin resistance 154 . Recently the newer derivatives BEOV and BMOV have entered clinical phase II investigations 155 .

Lithium (Li) is a complex trace element mainly used in neurological disorders and more particularly for treating mood and bipolar disorders. Its clinical management is difficult and Li is likely toxic 156 157 .

The interest in Li is based on its ability to inhibit GSK3 158 159 , a key enzyme regulating tissue glycogen levels 160 161 . Glycogen metabolism, on the other hand, is a particularly important component of glucose homeostasis 162 . GSK3 is now also claimed to be involved in bipolar disorders 163 . Very interestingly the prevalence of metabolic dysregulation and diabetes is highly augmented in bipolar disorders 164 165 . Transient diabetes has been associated with Li withdrawal, pointing to the potential effect of this trace element 166 .

In vitro Li increased both basal and insulin-stimulated glucose transport, as well as glycogen synthesis in soleus muscle of insulin resistant, obese Zucker rats 167 . The effect operated via increasing p38MAPKinase, confirming previous similar findings 168 .

In vivo Li paradoxically increased glycaemia when administered intravenously to diabetic rats, as shown by increases in glucagon 169 . Another study showed paradoxical effects of oral Li in rat liver: while glycogen synthase was activated and glycogen phosphorylase was decreased, liver glycogen nevertheless decreased 170 .

Altogether these data do not support the use of lithium in diabetes despite a cellular target (GSK3) which is potentially of particular interest.