What is in a mechanism?

Redoxcope by Francisco R. M. Laurindo

If you care about science, you care about mechanisms. Or at least you should, if you care about doing good science. More than ever, there is a wide consensus that the quality of science is as good as the depth of mechanistic insights it carries. Powerful mechanisms appear everywhere: in articles from top journals, in discussions with good scientists, in decisions about grant priorities, academic career, etc. This is also uncomfortably felt in the rejection letters one gets nowadays, in which the lack of sufficient mechanistic insights is a chief reason for not achieving a high-impact publication or the approval of a grant application. The size of such mechanistic trend has grown so fast and so conspicuously that I believe we can legitimately talk about a “mechanistic revolution of the modern scientific process”. This is particularly evident in the biochemical/biomedical literature, the main focus of our Redoxoma. Overall, such revolutions raise the need for a change in culture and a challenging need to develop and perfect intellectual and material resources. If facing these challenges is crucial even for basically following the crowd, they become essential for those who, like us, aspire to achieve international-level leadership in our community. Here we briefly discuss some aspects of this mechanistic revolution and challenges to be met in the pursuit of modern high-impact science.

What in fact is “sufficient mechanistic insight” ?

Despite the importance of the mechanistic framework to define excellence in science, to the point that many journals explicitly require from reviewers a clear opinion about this issue, the definition of a “good” and “meaningful” mechanism is in itself fuzzy and contaminated with subjectivity. It is easier in a way to say what a good mechanism is NOT. It is not a collection of data, as frequently seen, that simply are added to decorate the paper, but actually contribute little to close or to add depth to the main article story. One could bring several examples of these “decorative pseudo-mechanisms”. A frequent one is to investigate the effect of some central signaling proteins in the target event being studied, such as some MAP kinases in growth processes, apoptotic cascade proteins in cell stress, developmental proteins in repair responses and so on. Such converging signaling hubs are not surprisingly involved in essentially any complex cellular biological event, without being in fact a decision node or a true signal regulator. Another example is to pull out a sophisticated or exotic protein that is very poorly known and for which little criticisms can be constructed. These experiments can superficially provide meaningful information because they compose letter soups that add a flavor of complexity. But in fact, such signals may “affect”, but not “mediate” the response and thus add little to our understanding of the target process. Another apparent flavor of complexity is to add some complicated or novel technique that sophisticates the data - sometimes adding complicated mathematical formulas - but actually does not result in substantial further insights. An additional pitfall is what I would call “recurrent empty cycling”, in which apparently mechanistic diagrams are constructed not along the top-down axis of the target event being investigated, but rather along sterile lateral ramifications of modular or autodeterministic paths. Examples include papers in which, for example, apoptosis or autophagy signaling is exhaustively detailed, but the mechanisms that promote apoptosis or autophagy – the real advance – are only superficially addressed. Furthermore, it is possible to recall many studies claiming “novel” phytochemical antioxidants on the basis of their reactivity in vitro with oxidants, such as hydroxyl radical and ABTS⁺. The found mechanism - usually radical scavenger - is then transferred to experiments in vivo without consideration of other more likely mechanisms, such as the oxidative activation of the Nrf2 signaling pathway [1].

Although it is not difficult to grasp the paucity of mechanistic depth in these examples, it is less than obvious to understand what is really lacking here. I will risk saying that these examples lack the fundament of a good mechanistic insight, which is to provide a novel upgraded level of operational capacity in the system being studied or, in other words, to allow one to make accurate cause-effect predictions at an enhanced level. Thus, we might envision that a good scientific investigation is one that advances into such operational predictions. Consequently, a question we should add to all our projects is “what contribution to enhance operational predictions of this system will be brought by this investigation?”.

Is the mechanistic revolution healthy for science?

The so-called mechanistic revolution is here to stay, so we had better adapt to it. That does not prevent us from critically looking at whether this is a beneficial move at the end. As the overall mechanistic tendency has to do with quality, the trend to this state of mind is certainly positive with respect to improved outcomes in conceptual knowledge and translational applications. The latter is particularly relevant and not as frequently considered as it should in this regard by those trying to push innovation and development. While innovation is essential, it should be pointed that there is increasingly less possibility of innovation without mechanistic advances. No one is willing to invest heavily in any diagnostic test or therapeutic strategy that is not well grounded on good evidences for their mechanisms of action. Conversely, it is licit to propose that the relative lack of science applications (patents, products, etc) despite the numerical growth in science publications over the last years in Brazil may have its roots in the insufficient mechanistic quality of our science. This is indeed supported by the lack of parallel increase in impact indexes of such publications (http://docslide.com.br/documents/qualidade-e-impacto-das-revistas-brasileiras-do-portal-seer.html).

On the other hand, as with any good thing, there are risks and drawbacks to be considered. A basic problem is that, as mechanistic advances become the real treat, there is a natural parallel tendency to undermine of the importance of observation, particularly by the young investigators. This is clearly inadequate. Good descriptions steming from observation are the basis of risky well-based hypotheses and a necessary start-up point of important advances. Accurate observations remain the basis of every investigation and are a hallmark of good scientists. That is, sometimes it is very important to just look, think and get a good feeling of what is going on in a given investigation. And then observe more, think more, read more and wait for genuine good eurekas to pop up. And then, collect more data, observe more facts, read even more, think more and perfect your theory. At the end, this means that although you aim for a good theory, facts must always come first and uncontaminated. Otherwise, as Sherlock Holmes used to tell Watson, one distorts the data in order to fit the theory… which is a crucial mistake (see actual quotation below [2]). Thus, despite the mechanistic revolution, observations and descriptions remain essential and it is dangerous to insert a poorly conceived mechanism ahead of a good description, even if reviewers have criticized your investigation as being “too descriptive”. This creates interpretation bias that will sooner or later be deleterious and contribute to irreproducible results. Indeed, I personally believe that this bias is an important cause of the recently much debated irreproducibility of science [3]. Another related problem is that a state of mind too focused on constructing mechanisms may cause a fear to attempt some larger risk-taking investigative advances in favor of safer incremental steps.

The challenges

The mechanistic revolution poses considerable challenges to the modern scientist. These challenges are first intellectual, associated with the need to increasingly change the culture from developing superficial descriptive works to a profound mechanism-based approach. This requires strong multidisciplinarity and a clear willingness to get out of the comfort zone of one’s original formation and assured expertise. Such multidisciplinarity obviously involves working as group, but that is not enough. It is necessary to have a strong individual capacity of working at distinct levels of system complexity with a translational capacity that far exceeds the mere language translation. And of course, this requires communication skills and the willingness to exert them. But there are additional challenges, and they are of a logistical nature. Mechanistic work involves usually much –actually a whole lot much –more work than descriptive science, as can be easily seen by the extreme conceptual and technical complexity exhibited by modern high-impact papers. And this does not mean just methodically piling more and more information, but actually multiplying complexity at several levels. Clearly, this takes material and intellectual resources, being a substantial challenge to the community’s scientific system. And, importantly, it takes time. This puts a significant stress on top of our post-graduation system, which has cartorial rules about the duration and additional formalities involved in the PhD or post-doc formation. And, importantly, these stresses tend to uncover the weak spots in the scientific formation of our students and investigators.

How can we adapt to his picture?

A renewed and seemingly definitive trend established by the mechanistic revolution has been the need for collaboration and atypical combinations, which have been discussed elsewhere [4, 5]. Here I comment on the need to reformulate and adapt our science system to this revolution. Clearly, if we want to enhance the impact of our science, we must increasingly turn our approach from descriptive to mechanistic. And that is no small task, as it involves a community-wide , actually a nation-wide effort to change the culture of students, post-graduation committes, PhD thesis committees, financing agencies and scientists in general. Enforcing cooperative projects seems a good way to succeed in these aims, but is not enough if the groups do not actually sinergize. Overall, these efforts require time and especially continuity in order to change things at even the basic educational levels. But there are some things that can be done faster, I believe: 1) Rewarding quality, not quantity of publications across all system levels, at financing agencies, university and post-graduation system; 2) Establishing more clear deliverable landmarks in projects and evaluate them concerning grant priorities; 3) Monitor such deliverable landmarks in PhD or post-doc institutional periodic evaluation committees; 4) Insert evaluation items in grants concerning how the ideas and proposals will be mechanisticly treated, since some good ideas that cannot be addressed from the mechanism standpoint will not provide good contributions; 5) Clearly expose, in submitted projects or thesis proposals, the mechanistic values and pathways to be investigated; 6) Above all, mechanism-based investigation requires substantial versatility and capacity for rapid changes in direction. This is largely impossible in our country due to the inappropriate delays in importing equipment and consumables, which, in addition, often arrive in poor state of conservation. This would be a good doable point to jump-start our much-needed changes.

Take-to-the-lab message

Overall, joining the mechanistic revolution of the modern scientific process is crucial to tune our investigative efforts to the international standards of research excellence and, at the other end, to meet society’s expectations regarding applications. This requires facing great individual challenges that demand intellectual efforts, proactivity and entrepreneurship. In particular, adopting a multidisciplinary interactive attitude is probably the single most important change required for these advances. Indeed, it will be increasingly less feasible to perform a mechanistic revolution individually, further reinforcing the importance of growing as a system into this direction. And, let’s not forget, this advance ultimately sums up to improving institutional quality, while quality essentially emanates from meritocracy.

1. H. J. Forman, K. J. Davies, F. Ursini. How do nutritional antioxidants really work: nucleophilic tone and para-hormesis versus free radical scavenging in vivo Free Radical Biology & Medicine, 66: 24-35, 2014 | doi: 10.1016/j.freeradbiomed.2013.05.045
2. A. Conan-Doyle. Complete Sherlock Holmes quote: “It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” From “A Scandal in Bohemia”, available in http://sherlockholmesquotes.com/
3. R. Bolli. Reflections on the Irreproducibility of Scientific Papers Circulation Research, 117(8): 665-6, 2015 | doi: 10.1161/circresaha.115.307496
4. R. Uzzi, S. Mukherjee, M. Stringer, B. Jones. Atypical combinations and scientific impact Science, 342(6157): 468-72, 2013 | doi: 10.1126/science.1240474
5. S. Wuchty, B. F. Jones, B. Uzzi. The increasing dominance of teams in production of knowledge Science, 316(5827): 1036-9, 2007 | doi: 10.1126/science.1136099

Francisco R. M. Laurindo, Vascular Biology Laboratory, Incor, University of São Paulo Medical School

The author is grateful to Prof. Ohara Augusto for help with this text.