Conclusiones Finales

HSO;

R-N=G=S

Figure 1.2: Formation of ITCs from their glucosinolate precursors. Adapted from Hecht

Chapter 1: introduction

1 .2 .2 C h e m o p re v e n tiv e a c tiv ity o f is o th io c y a n a te s

I t is generally accepted th a t cancers induced by genotoxic chemicals begin with initiation, a process characterised by a DNA dam age which rem ains unrepaired a n d /o r misrepaired. N orm ally, m ost of such initiated cells are elim inated by apoptosis or by the cells of the im m une system . Avoidance of apoptosis gives such initiated cells a growth advantage over the norm al cells, which, as a resuit of cionai expansion, form foci of altered cells, a process called promotion. As the DNA of tu m o u r cells is unstable and prone to fu rth er m utations, total loss of control over th e ir survival, growth and proliferation will eventually result in malignancy.

The mechanism of anticarcinogenic activity of ITCs is m ultifactorial. I t is believed th a t ITCs block em ergence of carcinogen-induced aberran t cells as a result of th e ir ability to modulate m etabolism of carcinogenic compounds in favour of detoxification and excretion. Moreover, ITCs suppress tum our promotion and progression, operating mainly through a concerted modulation of m ultiple cellular signalling pathw ays involved in the regulation of apoptosis, cell grow th, proliferation, differentiation and inflam m ation (Hecht et a l., 2 0 0 0 ; Myzak and Dashwood, 2 0 0 6 ; Fimognari and H relia, 2 0 0 7 and Juge et al., 2 0 0 7 ).

1 .2 .3 IT C : In h ib itio n o f tu m o u r in itia tio n

It is hypothesised th a t inhibition of cytochrome P450 and induction of phase I I enzym es is one of the m ajo r mechanisms of tum o u r prevention by ITCs a t the initiation stage (Thornalley, 2 0 0 2 ; Hecht 1 9 9 9 ). The notion th a t ITCs a lte r the balance between Phase I/Phase I I m etabolism to favour carcinogen deactivation has been strongly supported by experim ental evidence. PEITC, consumed as a serving of w atercress, enhanced urinary excretion of 4 -m e th y ln itro s a m in o -l-(3 -p y rid y l)-l-b u ta n o n e (N N K ) in smokers (H echt et ai., 1 9 9 9 a ). Cruciferous vegetables adm inistered sim ultaneously with 2 a m in o 3

-Chapter 1: introduction

m ethyiim idazol [4 ,5 -f] quinoline (IQ ) inhibited form ation of colonic and hepatic preneopiastic lesions in the F344 rats (Knasm uller, 2 0 0 3 ). ITCs such as BITC and PEITC administered to rodents prior to, or around the tim e of, tre a tm e n t with high doses of carcinogens, such as azoxym ethane, NNK, d ib en z[a,h ]an th racen e (DBA), benzo(a)pyrene (B (a )P ), 5-m ethylchrysene (5 -M e C ) and N -nitrosom ethylbenzylam ine (NMBA), have been shown to effectively inhibit tum origenesis a t different organ sites (Morse et a l., 1991; Jiao e t al., 1994; Chung et a l., 19 9 6 ; Hecht, 199 9 a and Hecht et al., 2 0 0 2 ).

The chem opreventive activity of ITC appears to be tissue-specific, possibly reflecting the tissue-specific distribution and relative abundance of m etabolizing enzym es. BITC adm inistered to A/J mice concurrently with 5-M eC , DBA and B (a)P , potent anim al lung carcinogens present in tobacco sm oke, significantly reduced th e incidence and multiplicity of the lung neoplastic lesions, while exerting no effect on stomach tum ours in all PAH-treated groups (W atten b erg , 1 9 8 7 ).

1 .2 .3 .1 M o d u la tio n o f Phase I e n zy m e a c tiv itie s bv IT C

The m ajority of dietary, environm ental and occupational carcinogens such as PAH, polycyclic arom atic am ines (PAA), heterocyclic am ines (H A ) and nitrosam ines are indirect acting carcinogens th a t require m etabolic transform ation by phase I enzym es to ultim ate carcinogens, a process referred to as bioactivation. Such reactive interm ediates are electrophilic in nature and covaiently bind to critical cellular m acrom olecules such as DNA and proteins. Although th e m ajority of DNA adducts are effectively rem oved by the cell defence system s, some can be misrepaired and, if this occurs a t critical sites of critical genes such as pro-oncogenes a n d /o r tum our-suppressing genes, can give rise to initiated cells.

Chapter 1: introduction

Inhibition of Phase I enzym es and consequent interruption of bioactivation p athw ay(s) of parent carcinogens by phytochemicais such as ITCs is associated with decreased DNA adduct form ation, m utation frequency, and, ultim ately, inhibition or retardation of tumorigenesis. I t is feasible to assess the effects of ITC on the e x te n t of DNA/protein adduct form ation both in vitro and in vivo and this can be used as a biom arker of the carcinogen bioactivation/deactivation and, u ltim ately, of IT C effects on carcinogen m etabolism . P re -tre a tm e n t with PEITC decreased the unscheduled DNA synthesis in mucosal fragm ents of ham sters following exposure to NMBA (S oit e t al., 2 0 0 3 ) and formation of B(a)P-induced DNA and protein adducts in A/J mice (Sticha e t al., 2 0 0 2 ).

Sim ilarly, PEITC reduced the levels of PhlP-DNA and PhlP-protein adducts (Dingley et al., 2 0 0 3 ). Exposure to high levels of carcinogen(s) m ay overw helm the cellular defence systems (K nasm uller et a l., 1 9 9 6 ). A study utilising dietary relevant doses of SFN and PhIP provided strong evidence th a t both ITC concentrations and levels of carcinogen affect the chem opreventive efficacy of ITCs. The highest degree of protection against PhlP-m ediated DNA dam age afforded by SFN was observed a t the lowest concentrations of carcinogen (Bacon e t al., 2 0 0 3 ), highlighting the im portance of assessment of the chem opreventive potential of ITC at dietary relevant levels of carcinogens and ITC.

A lternatively, a direct assessm ent of enzym e level/ac tivity can be used as an indicator of chem opreventive potential of ITC. Individual ITCs vary considerably in th eir potency and selectivity as CYP inhibitors (Jiao et al., 1996; Sm ith et a l., 19 9 3 ; Sm ith et al., 1 9 9 6 ). It was suggested th a t the length of aliphatic carbon chain m ight influence th e potency of cytochrome P450 inhibition (Ham ilton et al., 1 9 9 4 ). The data supporting this theory indicated th a t the potency of arylalkyl isothiocyanates to inhibit rat C Y P lA l/2 and CYP2B1 increases if the length of the aliphatic carbon chain was Increased up to C6 (Conaway et a i., 1 9 9 6 ). In contrast, BITC, which has a very short a iky I chain, showed a potent m echanism -based inhibition of rodent C Y P lA l/2 , CYP2B1, CYP2E1 and CYP3A2 as

Chapter 1: introduction

weli as human CYP2B6 and CYP2D6 (Goosen e t a i., 2 0 0 1 ; Canistro et a i., 2 0 0 4 ) and rabbit CYP2E1 (M oreno et a l., 1 9 9 9 ). While m any ITCs w ere effective inhibitors of most cytochrome P450 enzym es in vitro (C onaw ay e t a l., 199 6 ; Goosen e t a l., 2 0 0 1 ; Moreno et al., 1999 and Canistro e t al., 2 0 0 4 ), the in vivo effects are inconsistent. ITCs have been shown to eith er induce or inhibit P450 content and activity, depending on the specific ITC studied, experim ental conditions, tre a tm e n t protocol and tissue exam ined.

For exam ple, a sh o rt-term adm inistration of PEITC to rodents resulted in inhibition of CYP2E1 and m arked induction of CYP2B1 (Ishizaki et a l., 1 9 9 0 ). In contrast, hepatic and pulmonary CYP2E1 activity was elevated a fte r chronic dietary adm inistration of PEITC to mice (Sm ith e t a l., 1 9 9 3 ). D ietary Brussels sprouts reduced IQ -in itia te d coion and iiver aberrant foci in rats despite elevated hepatic CYP1A2 activity (Kassie et a l., 2 0 0 3 a ).

CYP1A2 activity was also elevated in humans a fte r consumption of Brassica vegetables (Lam pe et a l., 2 0 0 0 ). Both P450 inhibition and induction w ere observed in rats fed broccoli samples from different cultivars and growth conditions indicating complex interactions of the individual ITC (Vang et al., 2 0 0 1 b ).

The cytochrome P450 enzym es have rem arkable substrate specificity. Suppression of bioactivation can be highly specific and depends on the structures of both the ITC and the carcinogen, reflecting ITC-induced disruption of the bioactivation pathw ay specific to the carcinogen in question. For exam ple, BITC was shown to be a highly effective inhibitor of B (a)P -initiated lung tum origenesis, while PEITC effectively suppressed form ation of NNK-induced neoplasm lesions (Conaw ay et a l., 2 0 0 0 ; Hecht, 2 0 0 0 ; Hecht et a l., 2 0 0 2 ) and inhibited human C Y P lA 2-m ediated oxidative m etabolism of NNK (Sm ith et a l., 1 9 9 6 ). Although intake of a m ixture of ITCs theoretically m ay have a synergistic effect o r/an d protect against a w ider range of carcinogens, a m ixture of dietary BITC and PEITC (1 and 3 pm ol/g d ie t), failed to reduce the level of B (a)P

-Chapter 1: introduction

induced DNA and protein adducts, although It Inhibited the form ation of NNK-lnduced adducts (Boysen e t a l., 2 0 0 3 ).

1 .2 .3 .1 .1 M o le c u la r m e c h a n is m s o f cyto ch ro m e P 4 5 0 in h ib itio n b v IT C

Both m echanism -based and com petitive Inhibition have been described as mechanisms of Cytochrom e P450 enzym es Inactivation. PEITC com petitively Inhibited NNK bloactlvatlon by mouse liver microsomes, but in vivo PEITC seem s to act as a mechanism -based Inhibitor of NNK m etabolism (Sm ith et a l., 1 9 9 3 ). Mechanism-based Inactivation Is defined as loss of catalytic activity of the P450 enzym e as a result of Irreversible adducts form ation following bloactlvatlon of the enzym e's substrate to a reactive Interm ediate. The m echanism -based Inhibitors are selective for specific P450 enzymes and, unlike com petitive Inhibitors, do not readily dissociate. Benzyl Isothlocyanate (B IT C ), for exam ple, was a m echanism -based Inactivator of rabbit CYP2E1 (M oreno e t a l., 1 9 9 9 ), rat C Y P lA l, CYP1A2 and CYP2B1 as well as human CYP2B6 and CYP2D6 (Goosen et al., 2 0 0 1 ). BITC Is metabolised to a reactive benzyl cyanate and benzylam lne th a t covalently bind to and Inactivate the enzym e. The covalent binding prevents fu rth er substrate binding and Is thought to occur at the active site of the apoprotein (Goosen et al., 2 0 0 1 ; Moreno et a l., 1 9 9 9 ). The Inactivation of microsomal P450s caused by covalent modification or destruction of the haem m oiety has not been observed experim entally (Goosen et a l., 2 0 0 1 ). O th er enzym es, such as human CYP1A2 and CYP2C9, appear to be Inhibited by PEITC com petitively, whereas human CYP3A4 was Inhibited both com petitively and non-com petetlvely (N akajim a et a!., 2 0 0 1 ).

1 .2 .3 .1 .2 E ffe c t o f th io l c o n ju g a te s o f IT C on c y to c h ro m e P 4 5 0

Thiol conjugates of ITCs have also been shown to Inhibit CYP enzym es, although they are less potent Inhibitors than th e ir parent compounds. Thiol conjugates of ITCs Inhibited

Chapter 1: introduction

CYP2E1, C Y P lA l/2 and CYP2B1 (Jiao et a!., 199 6 ; Conaway et a!., 1 9 9 6 ). The Inhibitory potency of Individual thiol conjugates In aqueous solutions Is determ ined by the rate of their deconjugation to release the parent ITC. Inhibition of C YPIA and CYP2B by aqueous dithlocarbam ates (D TC ) was greater with Increased pre-lncubatlon tim e and positively correlated with the exten t of decomposition of the IT C conjugates, Indicating that both the structure of parent ITC and the rate of the conjugate dissociation Influence enzym e Inhibition. H ow ever, In the presence of physiological concentrations of GSH, deconjugation of DTC Is expected only to be m arginal. I f the rates o f dissociation of thIol-ITCs in plasma and tissues are slow, as anticipated, then this would suggest th a t ITC-lnduced Inhibition of P450 enzym es Is unlikely to be a principal mechanism of th eir chem opreventlve activity (C onaw ay, 2 0 0 1 ).

1 .2 .3 .2 IT C ; e ffe c t o f phase I I m e ta b o lis m

Reactive m etabolites and ultim ate carcinogens, generated as a result of bloactlvatlon by P450s, m ay be detoxified by conjugation with glucuronic acid, GSH and sulphate, catalysed by phase I I enzym es. The biological purpose of conjugation Is to Increase polarity and excretion of highly lipophilic compounds such as m ost carcinogens. The relative abundance of Phase I and I I enzym es Is critical and carefully regulated. ITCs such as PEITC have been shown to shift this balance in favo u r of conjugating enzym es.

1 .2 .3 .2 .1 M o d u la tio n o f p h a se I I e n z y m e s in a n im a ls b v IT C s

Feeding Brussels sprouts to rodents resulted In upregulatlon of hepatic UGT and decreased IQ -Induced dam age to hepatocytes (H um blot et a l., 2 0 0 4 ). PEITC Induced activity/levels of GST, QR and UGT In the rodent liver, lung and gastric mucosa (Van Lleshout et a l., 1 9 9 8 a , 1998b; Guo et al., 1 9 9 2 and Kassle e t a l., 2 0 0 3 a ). AITC significantly Induced QR and GST In rat urinary bladder (M unday and M unday, 2 0 0 2 ).

The cysteine conjugates of BITC and PPITC sim ilarly Induced GST In m any tissues of

Chapter 1: introduction

fem ale A/J mice (Zheng et a!., 1 9 9 2 ), Indicating th a t certain ITC m etabolites retain their bloactlvlty.

The Induction of phase I I enzym es by ITC appears to be tissue-specific. For exam ple, AITC significantly Induced QR and GST In rat urinary bladder. M oreover, while Induction of QR reached a plateau a fte r 2 weeks of tre a tm e n t with A ITC , the activity of bladder GST continued rising even a fte r 21 days, suggesting th a t duration of tre a tm e n t also Influences the exten t of Phase I I enzym e Induction (M unday and M unday, 2 0 0 2 ). A single oral adm inistration of PEITC to rats (1 m m ol/kg bw ) resulted In significant elevation of the liver QR (5 -fo ld ) and GST activities (1 .5 -fo ld ) w hereas activities of these enzymes w ere not significantly affected In the lung (Guo e t a l., 1 9 9 2 ). G S T T l was significantly elevated In gastric mucosa but not In the esophagus, colon, or liver of male W /A rats a fte r dietary adm inistration of PEITC (van Lleshout e t a l., 1 9 9 8 a ; van Lleshout et al., 19 9 8 b ).

1 .2 .3 .2 .2 P h ase I I e n z y m e m o d u la tio n in h u m a n s b v IT C

D ietary consumption of vegetables rich In glucoslnolates by hum ans has also been associated with phase I I enzym e Induction, m odulation of m etabolism and, ultim ately, excretion of carcinogens and hence decreased risk of cancer. Consumption of tw o ounces of watercress, which contains gluconasturtlln, the glucoslnolate precursor of PEITC, Increased the urinary excretion of NNAL and Its glucuronlde, the m etabolites of the tobacco-specific nitrosam lne NNK In smokers (H echt et a l., 1 9 9 5 ). A recent epidemiological study In Chinese men has found a link between concentration of ITC metabolites In urine, G S T M l/T l polymorphism and Incidence of lung cancer. The G S T M l/T l-n u ll sm okers th a t had detectable levels of ITC m etabolites In th e ir urine w ere at the lowest risk of lung cancer, while G S T M l/T l-n u ll smokers who had no detectable urinary ITC m etabolites w ere at the greatest risk. This Is the first epidem iological study

Chapter 1: introduction

th a t established a direct link between ITCs and reduced risk of lung and colon cancer.

Interestingly, individuals with the wild type of G S T M l/T l genes had no appreciable protection from th e consumption of Brassica vegetables. I t has been hypothesised th a t the potent chem opreventlve activity of ITCs observed In G S T M l/T l-d e flc le n t Individuals was due to decreased excretion of ITCs so th a t th eir tissue concentrations would be considerably Increased, a n d /o r elevated global GST activity th a t m ight be sufficient to compensate for the function of missing enzym es encoded by silenced genes (Seow et al, 2 0 0 2 ). Moreover, ITC-lnduced Increase In plasma GSTA and peripheral lymphocyte GSTM was m ore pronounced In fem ale than In m ale subjects (Thornalley, 2 0 0 2 ). The m ajor lim itation of the above studies Is lack of differentiation between Individual glucoslnolates and ITC s, because the cyclocondensation m ethod often used to estim ate ITC content does not allow differentiating between Individual ITCs. The biochemical effects observed In hum an volunteers after consumption of Brassica vegetables are most likely to be a net result of bloactlvlty of the complex m ixture of all plant glucoslnolates.

In other words, no individual glucoslnolate can account for the cancer chem opreventlve activity observed a fte r consumption of Brassica vegetables (Vang et a l., 2 0 0 1 a ). For exam ple, w atercress contains a t least three phase I I enzym e Inducers, two of which, 7- m ethylsulflnylheptyl and 8-m ethylsulflnyloctyl Isothlocyanates, are 1 0 -2 5 tim es more potent than PEITC (Rose et a l., 2 0 0 0 ).

1 .2 .3 .2 .3 M e c h an is m o f p h a se I I in d u c tio n b v IT C

In vitro studies have shown significant differences In the phase I I enzym e Inducing

potential of Individual ITCs which appears to correlate strongly with the Intracellular accumulation of the ITCs. In tu rn , the extent of the Intracellular accum ulation of the ITC appears to be dependant on the ITC structure. The length of the side chain Is of little Importance while the oxidation state has a pronounced effect on the Inducer activity as can be Illustrated by comparing the Inducing potency of erucin and SFN, the la tte r being

Chapter 1: introduction

more than twice as potent as erucin at Inducing OR and GST activities In the Murine Hepatoma Cells (Zhang et a l., 1 9 9 2 ).

ITCs are highly reactive towards GSH. Among ITCs, erucin and SFN have the highest rate of spontaneous, non-enzym atic reaction with GSH (Kolm e t a l., 1 9 9 5 ). The Intracellular accum ulation of ITCs is accompanied by a rapid decline In the GSH levels and brings about oxidative stress. Experim entally Induced depletion of GST In Hepa cells prior to tre a tm e n t with ITCs enhanced th eir QR and GST Inducing potency (Zhang and Talalay, 1 9 9 8 ). M oreover, low llpophlllclty and high reactivity tow ards GSH have been shown to Increase Inhibitory activity of ITC toward N N K-lnltlated lung tum orlgenesis (Jiao et al., 1 9 9 4 ), supporting the central role of GSH depletion and oxidative stress In Induction of phase I I enzym es by ITCs. Being highly reactive tow ards cellular GSH, the ITCs rapidly accum ulate mostly as GSH conjugates, also known as dithlocarbam ates (DTCs), and Induce depletion of cellular GSH following tran sp o rter-m ed iated expulsion of DTCs out of the cell (Z h an g , 2 0 0 0 ). ITCs w ere characterised as monofunctional Inducers th at selectively upregulate Phase I I enzym es w ithout concurrent Induction of Phase I enzymes (M lao et a l., 2 0 0 4 ). The 5'-flanklng regulatory region of genes encoding phase I I enzymes such as OR and GST contain Antloxldant Response Elem ent (ARE), a binding site for the nuclear transcriptional factor Nrf2 (erythrold p 4 5 -re la te d fa c to r-2 ) activated by ITCs (Thornalley, 2 0 0 2 ).

IT C

K eapl

^ K eap l ^

Nuclear Alteratered ^ —---^ C n rf2 ^ translocation redox

homeostasis

ARE DNA

Figure 1.3: Activation by isothiocyanates of N rf2 p a th w a y reg u latin g expression genes encoding phase I I enzym es

Chapter 1: introduction

The Nrf2 norm ally resides In the cytosol bound to Kelch-llke ECH-assoclated protein 1 (K e a p l). In response to oxidative stress, Nrf2 dissociates from K e a p l and translocates to the nucleus w here It binds to ARE sequence In the upstream prom oter of phase I I genes (Fig. 1 .3 ). The m olecular mechanism of N rf2 -K e a p l com plex dissociation In response to ITC -lnduced oxidative stress Is not fully understood (Thornalley, 2 0 0 2 ).

1 .2 .4 IT C -a s s o c ia te d in h ib itio n o f tu m o u r p ro g re s s io n

In addition to th e altered ratio of phase I/p h a s e I I enzym es th a t suppresses carcinogenesis a t the Initiation phase, ITCs inhibit exp erim entally induced tum orlgenesis at the post-lnltlatlon and promotion phases. D ietary BITC and PEITC Induced apoptosis In the lungs of rats following exposure to cigarette smoke for 28 consecutive days (D'Agostlnl et a l., 2 0 0 1 ). Furtherm ore, orally adm inistered NAC conjugates of BITC and PEITC also decreased a m ultiplicity of (B (a )P )-ln ltla te d lung adenom as In A/J mice when adm inistered using a post-lnltlatlon protocol. The Inhibition of tum orlgenesis was attributed to the Induction of apoptosis (Yang et a l., 2 0 0 2 ). ITC -lnduced apoptosis and Inhibition of cell proliferation have been reported In a num ber of studies (Huang et al., 1998; Chen et al., 1 9 9 8 ; Yu e t a l., 1998; G am et-P ayrastre et al., 2 0 0 0 ).

1 .2 .4 .1 P ro -a p o p to tic a c tiv ity o f IT C s

Pro-apoptotic activity of ITC Is another Im portant aspect of th e chem opreventlve potential of ITCs as the m ajority of human cancers are resistant to apoptosis. There Is some evidence th a t ITCs preferentially Induce apoptosis In cells with m alignant genotypes. For exam ple, PEITC and AITC-lnduced apoptosis and growth Inhibition was

> 10-fold g reater In hum an leukaem ia (H L60) cells com pared with norm al cells (Xu and Thornalley, 2 0 0 0 ; Xu and Thornalley, 2 0 0 1 a ).

Chapter 1: introduction

C om m itm ent of cells to apoptosis and Inhibition of progression through the cell cycle appear to Involve ITC -lnduced protein modification as a result of GSH depletion and oxidative stress (Xu and Thornalley, 2 0 0 0 ; Pa yen et a l., 2 0 0 1 ). I t has been suggested th at ITC-assoclated activation of caspases and m itogen activated protein kinase (MAPK) fam ily m em bers occurs as a result of thiol modification and subsequent activation of protein tyrosine kinase a n d /o r Inhibition of protein tyrosine phosphatase activity and, ultim ately. Increased overall protein tyrosine phosphorylation (Xu and Thornalley, 20 0 1 b ). Additionally, ITCs m ay alter expression of genes associated with apoptosis and regulation of cell progression through the cell cycle through change In the activity of stress-response transcription factors such as (activato r protein) AP-1 and N F -kB, the downstream targets of MAPK, as well as transcription a n d /o r activity of p53 protein (Yang et al., 2 0 0 2 and Juge e t al., 2 0 0 7 ).

ITC -m edIated apoptosis and Inhibition of cell cycle progression associated with ITCs appears to Involve Induction of oxidative stress caused by GSH depletion following expulsion of th e ITC -G S conjugates through the MRP protein channel (Xu and Thornalley, 2 0 0 0 ; Payen et al., 2 0 0 1 ). PEITC-lnduced apoptosis correlated with initial depletion of Intracellular GSH and accumulation of the GSSG and GSH conjugates of PEITC, and was prevented by high concentrations of GSH (Xu and Thornalley, 2 0 0 0 ; Xu and Thornalley, 2 0 0 1 a ). The data on Inhibitory activity of thiol-contalning antloxldant N-

ITC -m edIated apoptosis and Inhibition of cell cycle progression associated with ITCs appears to Involve Induction of oxidative stress caused by GSH depletion following expulsion of th e ITC -G S conjugates through the MRP protein channel (Xu and Thornalley, 2 0 0 0 ; Payen et al., 2 0 0 1 ). PEITC-lnduced apoptosis correlated with initial depletion of Intracellular GSH and accumulation of the GSSG and GSH conjugates of PEITC, and was prevented by high concentrations of GSH (Xu and Thornalley, 2 0 0 0 ; Xu and Thornalley, 2 0 0 1 a ). The data on Inhibitory activity of thiol-contalning antloxldant N-