The charge of the agmatine analogues AO-Agm [N-(3-aminooxypropyl)guanidine], GAPA [N-(3-aminopropoxy)guanidine] and NGPG [N-(3-guanidinopropoxy)guanidine] is deficient as compared with that of agmatine and they are thus able to inhibit agmatine transport in liver mitochondria. The presence of the guanidine group is essential for an optimal effect, since AO-Agm and NGPG display competitive inhibition, whereas that of GAPA is non-competitive. NGPG is the most effective inhibitor (Ki=0.86 mM). The sequence in the inhibitory efficacy is not directly dependent on the degree of protonation of the molecules; in fact NGPG has almost the same charge as GAPA. When the importance of the guanidine group for agmatine uptake is taken into account, this observation suggests that the agmatine transporter is a single-binding, centre-gated pore rather than a channel.
Agmatine [1-(4-aminobutyl)guanidine] is a bivalent amine at physiological pH and an arginine metabolite as it is formed by decarboxylation of arginine catalysed by ADC (arginine decarboxylase) . It is widely distributed in bacteria, plants and invertebrates, and was also recognized in eukaryotes in the mid-1990s . It is known to bind to α2-adrenergic and imidazoline receptors and to have properties as a neurotransmitter or neuromodulator (for a review, see ). Agmatine also displays clinical properties such as neuroprotection and tumour suppression , while its biochemical properties include induction of ornithine decarboxylase antizyme  and spermidine/spermine N1-acetyltransferase , and inhibition of nitric oxide synthase . In mammals, agmatine is taken up from exogenous sources, such as food, and is produced by the intestinal flora . After its absorption it is accumulated by several organs, but particularly by the liver . In hepatocyte primary cultures, agmatine uses the same transporter as putrescine , but can also be transported by the organic cation and the extraneural monoamine transporter . Agmatine is found in mitochondria  together with its metabolic enzymes ADC and agmatinase [11–13]. Indeed, the imidazoline receptor, I2, which binds agmatine, is also located on a domain of MAO (monoamine oxidase) in RLM (rat liver mitochondria) . The close relationship between agmatine and mitochondria is further emphasized by the recent finding of agmatine transport in RLM . This process is mediated by a specific uniporter displaying electrophoretic behaviour and is thought to be a channel or a single-binding, centre-gated pore . Agmatine transport is only non-competitively inhibited by the propargylamines , well-known inhibitors of MAO. In general, naturally occurring polyamines do not display any inhibition; that induced by spermine is due to a membrane potential drop . The basic amino acids are also ineffective as agmatine transport inhibitors . The new, recently synthesized, charge-deficient agmatine analogues AO-Agm [N-(3-aminooxypropyl)-guanidine], GAPA [N-(3-aminopropoxy)guanidine] and NGPG [N-(3-guanidino-propoxy)guanidine]  can be used to clarify the role and function of agmatine in the cell. They have been employed in the present study to find out whether they inhibit agmatine transport in RLM and to gain further information about the origin of the agmatine transporter.
RLM were isolated by conventional differential centrifugation in a buffer containing 250 mM sucrose, 5 mM Hepes (pH 7.4) and 1 mM EGTA, which was omitted from the final washing solution . Protein content was measured by the Biuret method .
Incubations were carried out at 20°C with 1 mg of mitochondrial protein/ml in the following standard medium: 250 mM sucrose, 10 mM Hepes/HCl (pH 7.4), 5 mM succinate and 1.25 μM rotenone. Variations and other additions are indicated in the description of each experiment. Uptake of [14C]agmatine, [14C]arginine, [14C]ornithine and [14C]lysine was determined by a centrifugal filtration method . Electrical membrane potential (ΔΨ) was measured in a closed thermostatically controlled stirred vessel by monitoring the distribution of the lipophilic cation TPP+ (tetraphenylphosphonium) across the mitochondrial membrane with a selective electrode prepared in our laboratory .
As reported previously , [14C]agmatine, at 1 mM concentration, is transported in liver mitochondria at approx. 70 nmol/mg of protein in 20 min (Figure 1A). Agmatine uptake is energy-dependent, since it is completely blocked in de-energizing conditions [absence of substrate or presence of KCN or CCCP (carbonyl cyanide 3-chlorophenylhydrazone)]. This transport displays the non-ohmic force–flux relationship typical of an electrophoretic behaviour . Extrapolation at zero time of the time course of agmatine transport reveals instantaneous binding of approx. 15 nmol/mg of protein. The presence of idazoxan, an inhibitor of the imidazoline receptor, I2, strongly inhibits this binding, but does not affect the net transport of agmatine. Figure 1(B) shows that arginine, the biosynthetic precursor of agmatine, does not inhibit either the binding or the transport of agmatine, as well as the other basic amino acids lysine and ornithine, as also observed previously . Under our experimental conditions, arginine is not significantly transported in RLM, whereas the other amino acids are normally accumulated (Figure 1B, inset), confirming previously published results . The surprising observation that agmatine transport is not inhibited by arginine, which has the same guanidine group as agmatine, encouraged us to find out whether other guanidine compounds, namely AO-Agm, GAPA and NGPG (structures shown in Figure 2A), are equally ineffective. As is shown in Figure 2(A), they significantly inhibit net agmatine uptake with differing efficacy in 30 min: AO-Agm, 42%; GAPA, 21%; and NGPG, 55%. All three compounds slightly reduce the instantaneous binding of agmatine to RLM (Figure 2A). Since their interaction with the mitochondrial membrane has no effect on ΔΨ (Figure 2A, inset), their inhibition of agmatine transport is hardly due to a drop of the driving force. The saturation kinetics of agmatine transport in the presence of the inhibitors is depicted in Figure 2B1. The corresponding Lineweaver–Burk plot (Figure 2B2) demonstrates that inhibition by NGPG and AO-Agm is competitive (Ki=0.86 mM and 1.3 mM respectively), whereas that of GAPA is non-competitive. In aqueous solutions at pH 7.4, agmatine is a diamine with two positive charges. This could be the form electrophoretically transported in RLM as mentioned above, although agmatine may always be transported electrophoretically as a univalent cation, since one univalent form is very stable and has a very high dipole moment . Our three analogues most probably are also transported electrophoretically via the agmatine transporter.
Transport of agmatine in RLM: effects of idazoxan, de-energizing agents and basic amino acids
Time-dependent inhibition of agmatine transport by agmatine analogues (
A), saturation kinetics (B1) and double-reciprocal plot (B2)
The results clearly show that AO-Agm, GAPA and NGPG are dissimilarly effective as inhibitors of the electrophoretic transport of agmatine in RLM (Figures 2A and 2B2). The observation that inhibition by AO-Agm and NGPG is competitive and that of GAPA is non-competitive (Figures 2B1 and 2B2) suggests that the guanidine group is of prime importance. In other words, this group competes with that of agmatine in binding to the site responsible for the transport (see below). NGPG is a more effective inhibitor than AO-Agm, with the guanidinopropoxy group having a pKa of approx. 7.5 providing more effective binding as compared with that of the amino-oxy group with a pKa of approx. 5.0. Non-competitive protective inhibition by GAPA indicates that the guanidinopropoxy guanidine group as, most probably, the aminopropoxy group, does not bind to the same site as the guanidine group. This suggests that the inhibition by GAPA is due to a conformational change of the agmatine-binding site, the so-called guanidine site.
Arginine's failure to inhibit agmatine transport is presumably attributable to its carboxy group, which strongly hampers its binding to the agmatine transporter despite the presence of the guanidine group.
It would seem that the presence of both a guanidine and a carboxy group is incompatible with transport across the RLM, as demonstrated by the lack of arginine uptake (Figure 1B, inset). Ineffective inhibition by the basic amino acids (Figure 1B) is most probably ascribable to their carboxy group. As is shown in Figure 2(A), the guanidine compounds slightly reduce the amount of initial binding of agmatine to RLM. Agmatine binds to RLM at two types of sites, S1 and S2, both with mono-co-ordination and a total binding capacity of approx. 83.25 nmol/mg of protein. Bound agmatine is distributed between S1 and S2 with 3.2 and 80.05 nmol/mg of protein respectively . The dissociation constants of both sites demonstrate that the binding affinity of S1 is approx. 200-fold higher than that of S2 . Since previous investigations of polyamine transporters have shown that the site with the higher affinity is tightly linked to the transport process , S1 (the guanidine site) is evidently responsible for the transport of agmatine. This explains why only a few nmol/mg of protein are prevented from binding to the mitochondrial membrane when the inhibitors are present (Figure 2A). For NGPG and AO-Agm, this amount is a part of the 3.2 nmol/mg of protein which can bind at maximum to S1 and whose detachment results in inhibition of the transport. In the case of GAPA, the reduction is slightly higher and probably due to a local conformational change at S1 when GAPA binds to S2. Besides, GAPA, according to the charge distribution, is closer to putrescine rather than agmatine, and this would explain the above-mentioned effect of the guanidinopropoxy group of this compound.
In conclusion, the general slight inhibition of total agmatine binding by these compounds reflects a direct or indirect release of agmatine from S1 only, as site S2 (most probably not involved in the transport) has a broad binding capacity by which the inhibitor can interact with it without affecting the agmatine binding. Agmatine, at 1 mM concentration, totally binds to RLM at approx. 15 nmol/mg of protein (Figure 2A). Idazoxan strongly inhibits agmatine binding (Figure 1A), due, most probably, to its interaction with S2, but does not inhibit its transport as it does not bind to site S1. This demonstrates that agmatine is not transported by the receptor I2. The sequence in the inhibition efficacy of the analogues is NGPG>AO-Agm>GAPA (Figure 2A). Their charges in the same sequence are approx. 1.5, 1.0 and 1.5 respectively, which means that their efficacy is not directly related to their charge.
In the light of the mechanisms proposed for agmatine transport , this illustration of the greater importance of the non-modified guanidine group for the inhibition as opposed to a close charge-dependence indicates that the agmatine transporter is a single-binding centre-gated pore rather than a channel.
Health Implications of Dietary Amines: A joint COST Action 922 and Biochemical Society Focused Meeting held at Medico-Chirurgical Hall, University of Aberdeen, U.K., 19–21 October 2006. Organized and Edited by H.M. Wallace (Aberdeen, U.K.).