A spotlight on dosage and subject selection for effective neuroprotection: exploring the central role of mitochondria

Neurons are notoriously vulnerable cell types. Even the slightest change in their internal and/or external environments will cause much distress and dysfunction, leading often to their death. A range of pathological conditions, including stroke, head trauma, and neurodegenerative disease, can generate stress in neurons, affecting their survival and proper function. In most neural pathologies, mitochondria become dysfunctional and this plays a pivotal role in the process of cell death. The challenge over the last few decades has been to develop effective interventions that improve neuronal homeostasis under pathological conditions. Such interventions, often referred to as disease-modifying or neuroprotective, have, however, proved frustratingly elusive, at both preclinical and, in particular, clinical levels. In this perspective, we highlight two factors that we feel are key to the development of effective neuroprotective treatments. These are: firstly, the choice of dose of intervention and method of application, and secondly, the selection of subjects, whether they be patients or the animal model. We use the method of red to near-infrared light (λ = 600–1300 nm) treatment as our prime example of why these factors are so important. We then suggest that mitochondria within the distressed neurons form central players in the process and that these organelles, already known to be able to induce cell death, can be the targets for successful neuroprotective intervention.

Helping neurons in distress: When the brain suffers an ailment, for example after stroke, head trauma or neurodegenerative disease, its constituent neurons become distressed. In many instances, however, they do not die immediately. Rather, they go through a series of pre-death mechanistic stages. Initially, they undergo intrinsic damage but may not show any obvious sign of dysfunction, presumably because their self-protective mechanisms maintain a level of normal function. Over time, however, and with an increase in the insult severity, these survival mechanisms become less effective and neurons start to develop progressive pathology, leading finally to their death. One of the greatest challenges to neuroscience research over the last fifty years or so has been the development of effective interventions that can prevent or slow this progression to cell death. Unfortunately, these so-called disease-modifying or neuroprotective interventions have proved exceptionally difficult to develop. This is particularly the case in clinical studies and, although generally more successful, the same effect can occur in preclinical animal studies (Olanow et al., 2008). The factors that have contributed to this lack of success are many and vary across studies of different neurological conditions, but two are consistently evident. These are (1) issues with the choice of dose of intervention and its method of application and (2) problems with the selection of patients or animal models. We would like to focus on these two factors, using examples mainly from our area of expertise, the application of red to near-infrared light (λ = 600–1300 nm) on body tissues. These wavelengths of light (henceforth referred to as “light”) have been shown to improve the function and survival of all types of neurons across the central and peripheral nervous systems suffering distress, by stimulating mitochondrial function to produce more adenosine triphosphate energy, increasing mitochondrial membrane potential and reducing toxic levels of reactive oxygen species (ROS); in addition to these short-term gains, light triggers more long-term benefits by activating the downstream expression of stimulatory and protective genes (Hamblin, 2018, 2019).

Just the right dose: The choice of an appropriate dose of intervention and the method of its application has been a consistent problem in neuroprotection. For both clinical and to a lesser extent preclinical studies, failures could be attributed to the chosen doses being either too weak or too strong to generate an effect. For many treatments, such as light (as well as a range of pharmaceutical agents), the dose-response relationship appears to be biphasic (Figure 1A). That is, medium doses (in the middle of the bell curve) are more effective than either lower or high doses (at either end of the curve; Figure 1A–A’; Hamblin, 2018, 2019). Hence, it is likely that most neuroprotective interventions would be beneficial only within a rather narrow dose range and this range could be easily missed (Olanow et al., 2008; Hamblin, 2019). We can offer two examples in this regard. First, in preclinical studies using animal models of disease, most have used light at doses within the middle of the bell curve (0.1–15 J/cm²) and have reported neuroprotection and improvements in behavior (Hamblin, 2018, 2019). By contrast, studies that have used doses outside of the bell curve have reported little or no light-induced neuroprotection or behavioral improvement (e.g., Sipion et al., 2023). In addition, light treatment, depending on the dose, can have either a stimulatory (i.e., higher doses) or inhibitory (i.e., lower doses) effect on neurons, complicating the issue even further (see Yang et al., 2021). Second, in a phase 3 clinical trial examining the effect of light on stroke patients, the study was terminated prematurely because of a lack of statistical significance. This failure was notwithstanding some encouraging early findings in phases 1 and 2 of the clinical trial series, together with several positive preclinical studies in animal models. Many reasons have been offered for this failure at such a late stage of clinical trial, but one of the key issues was that the overall dose was too low to generate a positive outcome (Hamblin, 2018). In short, these examples of preclinical and clinical failures may have been avoided if the selected doses and their application were more appropriate for the study undertaken.

Figure 1.

The importance of dose and subject selection in neuroprotection.

(A) A dose-response curve, using red and near-infrared light treatment as an example. This treatment is most effective, generating maximum benefits, at medium doses (mid-range, in the middle of the bell curve), and less so at either low or high doses. (A’) Schematic diagrams outlining the dose-response effect of red and near-infrared light. Using the example of mildly-stressed neurons (with minor damage to mitochondria, conditions where benefits from treatment are most favorable), improvements in cell function and mitochondrial health are more likely with medium doses (red shading from the light device), rather than low (light red shading) or high (dark red shading) doses; with the latter two doses, no changes are the more likely. (B) Schematic diagrams of the importance of subject selection, whether it be patients or animal models. When considering mildly-stressed neurons, as in patients early in the disease process or less severe animal models, medium doses of light (the most effective in generating beneficial outcomes) are more likely to offer improvements in cell function and mitochondrial health. When considering severely-stressed neurons, as in patients late in the disease process or more severe animal models, the medium doses of light are less likely to offer improvements in cell function and mitochondrial health (i.e., no changes). Created with Apple Keynote.

Good selection: The second factor for consideration is subject (patient or animal model) selection (Figure 1B). Regardless of the ailment, there has been a long-standing approach of “the earlier the treatment, the better”. The thinking here is to try and save the neurons during the initial stages of their distress before they suffer too much damage. Once the damage progresses beyond a critical point, there is little or no chance of saving the neurons (Figure 1B; Porciatti and Chou, 2022). We can offer several examples in this regard. First, in a toxin-induced, animal model of Parkinson’s disease, light has been reported to improve cell survival effectively at lower doses, but not at higher doses of the toxin, after the cellular damage becomes too extensive (El Massri et al., 2016). This experimental finding would be akin to the selection of patients early in the disease as against those at later stages after they have suffered considerably more cell death. Second, and further to the latter example, in a pilot study on patients with age-related macular degeneration, light was shown to have no effect on patients that had progressed to intermediate forms of the disease (Grewal et al., 2020). It is possible that more success would have been obtained if the patients selected were at an earlier stage of the disease. Third, in animal models of Alzheimer’s disease, previous studies have shown that light can induce positive outcomes (see above); most of these studies used transgenic models that generated pathology at a relatively slow rate over a longer period (e.g., APP/PS1 model; Hamblin, 2019). In a recent study, however, a model that generated more pathology over a much shorter period (5×FAD) was used and a light-induced effect was not observed (Sipion et al., 2023). These examples highlight the need to identify the pathology early and to provide treatment before there is too much cellular damage.

In summary, effective neuroprotection is a delicate balance; there needs to be just the right level of intervention delivered at just the right time. These fine margins have no doubt contributed to the many failures at clinical, but also at preclinical levels as well.

Mitochondria as a central target? At this point, we would like to consider whether there is a central target that regulates this delicate balance between cell life and death. Although the mechanisms are not entirely clear, previous studies have suggested that there is indeed a central target responsible for this regulation, namely mitochondria. These organelles have been reported to form a convergence point for both the process of cell death and, quite paradoxically, for the signaling leading to protection. It is remarkable that the same organelle is not only critical for the life of the cell, but is also responsible for its death (Correia et al., 2010).

Mitochondria as killers: When mitochondria become dysfunctional – for example, due to aging, genetic mutation, hypoxia, trauma, or toxic insults – cell homeostasis is disrupted. First, adenosine triphosphate production decreases and there is a loss of readily available energy to drive intrinsic cell physiology. Second, calcium buffering is impaired, leading to abnormal cell activity and internal damage. Finally, there is an increase in the production of ROS resulting in higher oxidative stress, generating a self-propagating chain reaction of cell membrane lipid peroxidation, destruction of membrane lipids, and a high mutation rate of mitochondrial DNA. Taken together, this oxidative damage can lead to the activation of pro-apoptotic pathways and cell death. Mitochondria induce cell death by releasing mitochondrial intermembrane space proteins that then activate the killer caspases in the apoptotic cascade. This process has been viewed as an all-or-nothing event that once triggered, cell death is inevitable. Recent studies have, however, challenged this view. The process can be incomplete and not all mitochondria appear to be involved, as some may remain functionally viable. A threshold concentration of mitochondrial intermembrane space proteins may need to be achieved before the overall death process is triggered. Although the first signs of this death process can be evident after ten minutes, the entire process, from the initial trigger to death, can take anywhere from hours to days. Hence, early intervention is critical (at least within days of insult) to save the majority of mitochondria, before a threshold of intermembrane space proteins is reached (Correia et al., 2010; Tait and Green, 2013; Yang et al., 2020).

Mitochondria as protectors: Although high levels of ROS expression after mitochondrial dysfunction lead to cell death, a brief burst of low levels of ROS produced by mitochondria could, by contrast, promote cell survival by triggering self-protective mechanisms. This process could be considered a form of brain tolerance or acquired resilience, whereby neurons can protect themselves against future damage by preconditioning or adapting to lower levels of the toxic insult, in this case, ROS (Correia et al., 2010; Tait and Green, 2013; Yang et al., 2020). For example, it has long been known that preconditioning either the heart or the brain with short periods of ischemia can alleviate future damage incurred by a subsequent prolonged ischemic insult (Stone et al., 2019). In addition, the low levels of ROS can induce a protective mitogenic response, allowing for a cascade of cell-to-cell communication over long distances, even across different tissues. This could be a basis for the so-called “abscopal” effect, where the treatment of cells of one region of the body (e.g. thigh) can lead to protection and repair of cells in another, more distant region of the body (e.g. brain: Mitrofanis, 2019). Thus, effective neuroprotection may depend on controlling the ROS levels and the acquisition of resilience could be one way to improve the mechanisms of neuroprotection.

Conclusions: Mitochondria appear to hold the key to effective neuroprotection. If they have not suffered too much damage caused by a particular insult, they can still be saved by an intervention at just the right dose; not too little, not too much. If allowed to go beyond a critical point where the mitochondrial damage becomes excessive, then no amount of intervention can save them and the process of cell death ensues. This delicate situation has no doubt contributed to the many failures and frustration of exploratory neuroprotection studies at both preclinical and, in particular, clinical levels. A better understanding of the basic science behind how mitochondria can control the balance between cell death and protection will hopefully lead to more effective neuroprotective treatments in the future. It is also clear that preconditioning the mitochondria to a future insult so that they gain an acquired resilience, may be central to improving the efficiency of the neuroprotection process.

This work was supported by Fonds Clinatec and COVEA France (to JM).