A 2006 publication entitled ‘Catalytic Antioxidants to Treat Amyotrophic Lateral Sclerosis’ summarized the therapeutic effects of metalloporphyrins (MPs) and metallotexaphyrins (MTs) in the G93A superoxide dismutase (SOD1) mouse model of amyotrophic lateral sclerosis (ALS) [1]. It also contained discussions related to: 1) important caveats of the ALS SOD1 mouse model, 2) characterizations of catalytic antioxidants and the rationale for therapeutic benefits relative to classical dietary/endogenous antioxidants, 3) evidence for oxidant-related therapeutic effects, and 4) preliminary evidence for novel therapeutic mechanisms based on ‘adaptive resistance’ rather than antioxidant activity per se. The key points and broad concepts presented in that publication will be examined herein to determine whether they have held up over time, and/or need to be modified or even abandoned outright.

The ALS mouse study published in 2005 [2] was among the first to utilize a symptom onset-administration paradigm. Most studies up till then had treated mice presymptomatically – something that cannot be done in humans as there are no reliable predictors of human disease and no one knows they have ALS until a differential diagnosis is made. (SOD1 mutations are the only known cause of ALS and humans can be tested for SOD1 mutations, but that accounts for only 2% of all cases). Onset-type studies were promoted to better mimic the human condition, i.e., to determine whether disease progression, once begun, could be slowed or stopped, and to examine whether earlier failures of animal studies to predict the effects in humans was related to differences in study design (e.g., presymptomatic in mice versus onset in humans).

Despite the strong rationale for onset-type studies in animals, this design was not universally adopted. Indeed, human trials continue to be based on presymptomatic animal studies, if at all. That is, often times cell culture studies or even anecdotal ‘evidence’ provides the justification for human trials. In other cases, rationale for human trials is straightforward and obvious, such as small molecules, antibodies, or antisense messages directed against a suspected or known gene or protein, such as SOD1. Reliance on results of animal studies (to predict human outcomes) has proved problematic not only in ALS, but in a number of other diseases as well. However, the situation in ALS is particularly acute, and urgent. Not only is the survival after onset relatively short (2-4 years) in most cases of ALS, but every spinal motor neuron lost results in a new deficit, and the loss is irreversible. Thus, the battle to preserve motor function and stave off major disability mean treatments must begin promptly and aggressively. This often precludes lengthy and rigorous preclinical studies to help establish a strong rationale.

Given the sheer number of disease conditions where oxidative injury has been implicated (including ALS, [3]), why have classical dietary/endogenous antioxidants like ascorbate, tocopherol, glutathione, etc., even in large doses, proved ineffective? Sacrificial reaction with relevant biological oxidants and slow re-reduction rates may offer clues. Because such antioxidants are sacrificial, one molecule is consumed for each molecule of reactive species scavenged. This lowers the effective concentration of antioxidant available for the next round of scavenging. While mechanisms exist in vivo to re-reduce these endogenous antioxidants, they are typically too slow to preserve the antioxidant at adequate (protective) concentrations, particularly during periods of continuous oxidant production. Catalytic antioxidants, on the other hand, can react with multiple oxidants many orders of magnitude faster, and can be re-reduced under physiological conditions almost as fast as the initial oxidant scavenging reaction. Thus, effective concentrations are more readily maintained. This is critical, as the effectiveness of an antioxidant to scavenge is a function of its second order rate constant times its effective molar concentration.

Manganese-porphyrins (and some iron-containing porphyrins) constitute the most efficient catalytic antioxidants to date in terms of reaction rates with various oxidants, albeit none appear to be oxidant-specific. Broad reactivity toward oxidants is likely to be advantageous, but it does complicate the interpretation of mechanism of action, at least as an antioxidant. For example, MPs that are good SOD mimics also tend to be good peroxynitrite decomposition catalysts. Other chemical classes of macrocyclic manganese and iron compounds have been synthesized and tested, and have the potential to be therapeutic via non-redox-active mechanisms [4]. Most recently, transition metal complexes of flavonoids like quercetin have been prepared and tested [5]. Flavonoids and other compounds with phenolic hydroxyl groups are reasonably good antioxidants; adding metals to these hydroxyls could attract negatively charged oxidants and thereby facilitate quenching reactions. However, these compounds would still be sacrificial, and the subsequent release of the free transition metal could be cytotoxic.

All in all, the electronics of the porphyrin nucleus – crafted by Nature as a stable, metal-coordinating moiety – is unlikely to be equaled in terms of facilitating rapid redox cycling of the central transition metal, while keeping the metal atom tightly bound. The rates of redox cycling for a given oxidant can be altered dramatically by side chain composition, as the elegant work of Batinic-Haberle and colleagues have shown [6,7]. Side chain modifications also alter overall formal charge, solubility, cell membrane permeability, and blood-brain-barrier penetration – critical to effective drug design.

In addition to the oxidative injury aspect, the ALS SOD1 mouse model also presented as a unique opportunity to dissect the relevant activity of the catalytic antioxidants. Prototype manganese porphyrin (MnP) compounds were conceived and synthesized before peroxynitrite was known to exist in vivo. These MnP’s were developed as scavengers of superoxide and thus termed “SOD mimics” or “mimetics”. Only later was it determined that many of the MnP’s (and iron-Ps) could scavenge peroxynitrite at similar rates. Because the ALS SOD1 mouse model was based on a 6-8-fold overexpression of SOD1 mutant enzyme, mice had several-fold more SOD1 activity in all tissues, including the CNS. This fact had already established that SOD1 mutants caused disease via a gain-of-function, not a loss of SOD1 activity. If SOD1 mutant-mediated injury still occurred in the presence of enormous enzyme activity, then adding more SOD activity in the form of MP’s would have no effect. However, an iron-porphyrin (FeTCPP) and a manganese-porphyrin (MnTDE-2-ImP, AEOL10150), both of which are redox-active, were found to extend survival of ALS SOD1 mice [2,8].

In both human ALS and the G93A SOD1 mouse model of ALS, evidence for oxidative neuronal injury abounds. It was never clear whether such oxidative injury was the main cause of motor neuron loss, or whether it was secondary reaction to some other, more fundamental injury. Treatment with novel catalytic antioxidants seemed to offer a chance to help distinguish between these possibilities. It turned out that both redox-active porphyrins (iron and manganese) tested in the mouse model dramatically decreased levels of oxidative markers associated with peroxynitrite production [2,8]. Moreover, these effects correlated with enhanced survival of spinal motor neurons and retention of motor function. Thus, these results were consistent with the MPs acting as peroxynitrite decomposition catalysts.

The question of redox-based therapeutic benefit versus other non-redox-based benefits is perhaps best addressed via treatment with a redox-inactive MP analog. However, no redox-inactive MP has been tested in the ALS mouse model thus far. Such attempts were made using a cobalt porphyrin (CoTCPP), but that compound proved to be acutely toxic (Crow et al., unpublished). One possible clue came from studies in isolated cardiac myocytes where doxorubicin-mediated killing of myocytes was inhibited by ZnTCPP, a redox-inactive MP [9]. Protection correlated with up-regulated HO-1 and decreased activity of caspase-3. While correlation did not provide proof of mechanism, the overall results indicated that a redox-inactive MP could still be protective against some types of oxidative injury.

Mechanistic clues came from another source – a five-coordinate ‘texaphyrin’ containing gadolinium as the central metal. The manganese analog (manganese texaphyrin or MnT) was first tested in the ALS mouse model based on the fact that it was found to be a modest peroxynitrite decomposition catalyst. MnT did extend survival of the mice, but not as much as the manganese-porphyrins [10]. A proof-of-concept study employed a gadolinium analog (Motexafin gadolinium or MGd) which had virtually no antioxidant properties and was, in fact, mildly pro-oxidant under physiological conditions. Surprisingly, it was also found to be therapeutic in the ALS SOD1 mice [11]. MGd had already been given to over 600 human patients in a trial to evaluate radiation-enhanced killing of metastatic tumors. Results from that study suggested that MGd might actually protect normal neurons. Based on these unexpected findings with both redox-active and redox-inactive metal complexes, 2-D proteomics/mass spectrometry techniques were employed to help determine potential mechanisms of action.

Although quantitative 2-D proteomics was then in its infancy, changes in protein expression of 1.5-fold and greater could readily be seen. As originally stated, up-regulation of four general classes of proteins were seen with all therapeutic porphyrins and texaphyrins: 1) heat-shock proteins, 2) metal-binding, 3) neurofilaments, and 4) proteins associated with mitochondrial energy production [1]. (See Sheng et al. [12] for a comprehensive review of MP effects and potential mechanisms in various disease conditions). Heat-shock proteins help misfolded proteins refold properly. To the extent that SOD1 mutants misfold and aggregate – which is widely held to be the case – increases in HSPs could certainly be therapeutic.

Metal-binding proteins like (metallo)thioneins are rich in thiols, which are potent antioxidants, and their ability to bind free metals provide a secondary mechanism for limiting cytotoxicity. Although the reasons remain unclear, it had been shown early on that increases in the heavy chain of neurofilament (NF-H) were neurotoxic in transgenic mice, but that increases in the neurofilament light chain (NF-L) could rescue that phenotype [13]. Thus, up-regulation of NF-L production by MP’s and MT’s seemed a plausible non-redox therapeutic mechanism as well.

Taken together, the therapeutic and proteomic results suggested that metalloporphyrins and metallotexaphyrins might share some properties in terms of up-regulating cellular defenses – a phenomenon which had been referred to as ‘preconditioning’ or perhaps more appropriately ‘adaptive resistance’. That is, a mild and prolonged pro-oxidant stimulus could cause cells to bolster naturally-occurring defense mechanisms. Much work remains to be done in this regard, but the concept is sound, and would explain why chronic low dose administration is more effective than acute dosing. It would also explain the bell-shaped dose-response curve seen with many of these compounds, where the optimal dose lies in a fairly narrow range, and the therapeutic effect is lost when the dose drops below or exceeds it [1].

In general, redox-active MPs possessing potent SOD and peroxynitrite decomposing activities have the more pronounced therapeutic effects. However, even these compounds can be mildly pro-oxidant via cycles of slow auto-oxidation and re-reduction by endogenous antioxidants [14]. Thus, their therapeutic effects may result from a synergy between oxidant scavenging and stimulation of adaptive resistance in cells. It follows that those compounds with poor reactivity toward superoxide and peroxynitrite would be less effective by only eliciting the adaptive resistance arm. This provides a testable hypothesis for future work, as the relative contributions of potent catalytic antioxidants compounds are contrasted with adaptive resistance mechanisms stemming from less redox-active, mildly pro-oxidant compounds. Moreover, the scale and scope of cellular changes associated with such adaptive resistance can be expanded using newer techniques and approaches.

Debate continues as to effects of MnTBAP (non-trivial name of manganese tetracarboxyl phenyl porphyrin or MnTCPP) in the ALS SOD1 model and in other model systems. Reference is often made to MnTBAP per se being essentially inactive, and that contaminants in various preparations of MnTBAP are responsible for its therapeutic effects [12]. Because FeTBAP (FeTCPP) – the iron analog – was both therapeutic in the ALS SOD1 mice [2,8] and a good peroxynitrite decomposition catalyst [14], it seemed reasonable that it might be the mystery contaminant in MnTBAP. (Synthesis of MPs involves addition of a metal ion, and iron is a common contaminant of manganese salts.) In any case, given the apparent potency of this ‘contaminant’, it seems that an effort to determine its identity, once and for all, would be justified. It may well be that the ‘contaminant’ could provide clues as to a more optimal MP structure, or possibly to an entirely new class of compounds.

Superoxide possesses an extra electron and is a poor oxidant; indeed, it’s ability to reduce cytochrome c is the basis for the most common SOD activity assay. Humans have three different forms of SOD – Cu, Zn SOD (SOD1), MnSOD (SOD2), and extracellular SOD (SOD3). Moreover, SOD1 constitutes as much as 1% of all soluble protein in some tissues. Why does Nature go to such lengths to scavenge a species that, in and of itself, is fairly innocuous? One product of the SOD reaction with superoxide is hydrogen peroxide – a more powerful oxidant. Yet if hydrogen peroxide was highly cytotoxic, it seems unlikely that Nature would have chosen this pathway for dismutation of superoxide. One can only suspect that SODs exist to prevent superoxide reacting with nitric oxide to give peroxynitrite – an oxidant of unrivaled potency, with a wide range of biomolecular targets. We may yet find that some type of peroxynitrite degrading system exists in biology. Until then, it seems prudent to continue to develop synthetic compounds that can effectively scavenge it.

Quite often the mechanisms of action of new, experimental compounds – even those that become FDA-approved – are not known. Many times claims are made that it’s an “antioxidant”, even when evidence is lacking, or when the compound itself lacks the structure and functional groups to give it true antioxidant properties. Because oxidative injury is such a common feature of so many disease conditions, “antioxidant” becomes a useful mechanistic catch-all. Thus, it’s prudent to be wary of antioxidant claims until the evidence confirms it.

It’s noteworthy that the manganese-porphyrin MnTDE-2-ImP (AEOL10150), first reported to extend survival of ALS SOD1 mice in 2005, remains as the only MnP to be administered to human patients [12]. That Phase I trial found doses up to 75 mg/kg being well-tolerated – a much higher dose than was needed to produce dramatic survival effects in mice. The clinical trial could not be sustained for financial reasons, but it paved the way for human testing in ALS and other neurodegenerative diseases. Work continues to find the optimal structural features that maximize activity, solubility, and bioavailability in the hope that a new opportunity will emerge to examine therapeutic effects of novel MP catalytic antioxidants in human disease.

The work referenced in this Commentary was partially supported by the National Institutes of Health grant R01 NS040819 to Dr. John P. Crow.