Steven A. Benner, Foundation for Applied Molecular Evolution

Science Addressing Questions Not Easily Addressed by “the” Scientific Method. The Origins of Life

1. A multidisciplinary workshop built a model for the geochemical formation of RNA

In October 2018, a group of planetary, biological, and molecular scientists gathered in Atlanta for an unusual three-day workshop. On its last day, the scientists closed their PowerPoints, set aside individual projects, and opened an interactive discussion with just one goal: To resolve paradoxes that prevent the “origins of life problem” from being studied as “normal science” (Kuhn, 1966).[1]

The interactive discussion moved beyond talks that had been presented by individual participants from their individual laboratories in their individual fields. Freed from constraints, for three hours, the group built a model for origins that combined all three fields. The “Atlanta Model” that emerged included a chemical path to make RNA from starting materials that were likely available on a Hadean Earth. It identified geological events that likely made those materials available. It estimated dates when those events most likely made oligomeric RNA.

The Atlanta Model is described in detail in a joint publication of the participants (Benner et al., 2019). The discussion that generated the model was summarized by Robert Service, a science journalist who witnessed the interactive discussion (Service, 2019).

In brief, the Atlanta Model hypothesizes that life on Earth most likely emerged 4.36 ± 0.1 billion years ago (Ga) via an abiotic synthesis of oligomeric RNA. That RNA arose from:

(a)   formaldehyde (H2C=O) and glycolaldehyde (HOCH2HC=O) generated by ultraviolet light and/or electrical discharge in a Hadean atmosphere rich in carbon dioxide (CO2). These

(b)   reacted with volcanic sulfur dioxide (SO2) to form carbohydrate-bisulfite adducts that

(c)   rained into constrained aquifers that intermittently evaporated, where

(d)   the carbohydrate-bisulfite adducts, constrained by minerals containing borate, silicate, and phosphate matured to give 5-carbon sugars. These then

(e)   generated carbohydrate cyclic phosphates that

(f)    gave nucleosides via direct reaction with nucleobases that had

(g)   been formed from precursors having reduced nitrogen atoms that had been generated in

(h)   an atmosphere that had been transiently reduced by a large impact, which was

(i)    date-estimated from various geological clocks.

The nucleosides were then converted to phosphorylated materials in

(j)    aquifers evaporating and re-hydrating to support

(k)   reaction between nucleosides and phosphate, phosphite, nickel, and borate minerals to give nucleoside 5’-polyphosphates, which then

(l)    reacted on silicate minerals to form mineral-stabilized oligomeric RNA.

Our task today is to ask, with much incredulity: How is it possible to say such things about events that occurred billions of years ago, none directly observable?

We also consider the more general question: Can such models ever enjoy the respect that we give models emerging from “the” scientific method? Over the past century, many have introduced methods and demarcation criteria, including logical positivists (Smith, 1986), Popper (2002), Quine (1953), Hempel, and others. As we were taught for our 8th grade science fairs, a “scientific” activity requires careful observations, precise measurements, and tests of well-structured hypotheses. None of these seem directly possible when seeking to understand the origins of life.

However, science has long recognized that such “big” questions can be indirectly studied by asking other, perhaps related, “small” questions (Benner, 2009). Indeed, almost no one found on a list of people said to be investigating the origins of life is directly investigating the origin of life. Rather, they are investigating microbial metabolism or synthetic organic chemistry that, at funding time, they say is relevant to the origin of life.

Here, “problem selection” becomes very important. Scientists generally continue to do research in the future that is similar to research that they have done in the past. Funding agencies encourage this. Thus, it is easy for scientists to go down rabbit holes in their work. Even if a project began with the intention to indirectly study how life originated on Earth, it often evolves, drilling deeper and deeper into narrower and narrower problems that are less and less relevant to the initial intention.

Most importantly, we must remember the Feynman dictum: “People are easy to fool, and the easiest person to fool is ourselves” (Feynman, 1974). Scientists are like all humans. We are good at rationalizing. We easily rediscover old ideas and brand them as our own (Saladino et al., 2018). We advocate for our ideas by cherry picking data that agree with them while excluding data that disagree. Indeed, origins of life research holds many examples of the “anything goes” description of science provided by Paul Feyerabend, my former colleague at the ETH. Paul noted that scientists use “any trick, rational, rhetorical or ribald” to get a community to accept their views, with the publications, grants, and awards to follow (Feyerabend, 1975). This has certainly been the case in the field of origins.

2. Background. RNA as the first genetic molecule to support Darwinism

The Atlanta workshop started with a “Standard View” for the origin of life on Earth (Neveu et al. 2013). That view holds that Darwinian evolution began when abiotically created RNA molecules became the informational part of a system that catalyzed the template-directed synthesis of RNA, with replicable errors (Rich, 1962). This is a minimalist, molecular, and reductionist view of evolution. An “existence proof” for such RNA molecules comes from the laboratory of Philipp Holliger in Cambridge, which found examples of RNA molecules that catalyze the template-directed synthesis of RNA (Attwater et al. 2013).

This Standard View has long enjoyed support because it manages a key “origins” problem emerging from the details of modern molecule biology. Today, the information in DNA encodes the information of RNA, which encodes the information of catalytic proteins. If RNA emerged first and was able to perform both genetic and encoded catalytic roles in the first forms of life on Earth, this avoids the chicken-or-egg problem (which came first, DNA for genetics, or proteins for catalysis?). This origin scenario then gave an in an “RNA World” (Gilbert, 1986).

Other chemistry from the modern biosphere also implicates RNA in early episodes of life on Earth. The ribosome, the molecular machine that makes proteins, is itself an RNA catalyst. Ada Yonath, in her contribution to the Atlanta workshop, provided information about the performance of ancient RNA from ancient ribosomes, consistent with the Standard View.

An RNA World model also accounts for the chemistry of RNA cofactors in modern metabolism (White, 1976; Visser & Kellogg, 1978a,b). In life on Earth today, various chemical reactivities required for metabolism are not enabled by the encoded amino acids in our proteins. Rather, they are provided by cofactor moieties that are attached to small pieces of RNA (Fig. 1). The RNA appendage is seen as a vestige of a handle that held the reactive moiety to RNA catalysts in the RNA World. An analysis of these cofactors and modern genomic sequences adumbrated models for the genomic and metabolic activities in the RNA world at the time when encoded synthesis of proteins first emerged (Benner et al. 1989).

3. The “scientific method” and the origin of life

3.1 The “RNA First” model for the origins of life is ripe for philosophizing

The “RNA First” model might be viewed as a hypothesis. However, it would not serve well an 8th grader participating in a science fair, as it contains few of the instructed elements required for a prize. Where is the experiment that would confirm a historical model, something that would fit a “positivist” model? What experiment might “falsify” the model, to satisfy Karl Popper and Gunther Wächtershäuser (1995)? Thomas Kuhn (1966) might remark on the absence of opportunities for “normal science” in the field. However, pointing out that a field needs a revolution is hardly a prescription for getting one.

The difficulty of placing origins research into a standard “scientific method” has been noted by its more philosophical participants. For example, one of many philosophical (and quotable; Krishnamurthy, 2018) aphorisms from Albert Eschenmoser, also of the ETH, noted that the origin of life cannot be discovered; it must be re-invented (Eschenmoser, 2007, 2009). “Re-invention” is outside of the 8th grade science fair canon.

John Sutherland has also contributed richly to the philosophy of the discipline. For example, in 2011, Powner and Sutherland (2011) suggested that they had found “a new modus operandi” for prebiotic chemistry, one that allowed geology to inform chemistry, and vice versa. Sutherland (2016) criticized a view, which he attributed to Gerald Joyce (2002), that “Darwinian evolution needs informational molecules, so RNA must have come first”. He then criticized another view, attributed to Wächtershäuser (1992), that “[y]ou can’t get by without building blocks and energy, so metabolism must have come first”. He then criticized another view, attributed to Plankensteiner et al. (2005), that “[g]enetics and metabolism without catalysis is hard to imagine, so proteins must have come first”. He then criticized another view, attributed to Segré et al. (2001), that “[t]he development of Darwinian selection is hard to imagine without compartments, so membranes must have been there at the outset”.

Sutherland then advised the community to pursue a philosophically “more holistic approach” to the origins problem. His advice was to let “chemical results … constrain geochemical scenarios”, and allow the geochemical scenarios “back-inform the chemistry [leading] to refinements in both the [geological] scenario and the chemistry”. Sutherland (2017) noted that by doing so, the field might not be at its end, or even the beginning of its end, but at the end of its beginning.

The need for geological and chemical models to interact under such a modus operandi has been recognized in this community for over a half a century (see, for example, Bernal (1961), Cairns-Smith (1977; 1982; 2005), Scorei et al. (1999), Cimpoiasu et al. (1999), Prieur (2001), Ricardo et al. (2004). Indeed, the Atlanta workshop brought together many scientists who followed this modus. Let us see how the Atlanta discussion developed.

3.2 Paradoxes as a tool to focus research effort on the most important problems

The goal of the interactive discussion in Atlanta was to resolve paradoxes that prevent the origins of life problem from being studied as a “normal science”. Several prominent researchers in the field have disputed this approach, arguing that paradoxes are no different from statements of “difficult problems”. Let us develop the concept so that we (at least) understand what such a dispute might be about.

Paradoxes are constructed in a way that, using logical formalism, moves from universally accepted facts via Aristotelian deductive logic to a conclusion that its constructors find to be unacceptable. To be a paradox, as opposed to simply a statement that a problem is “difficult”, formal logic must be invoked.

It is best if the target of the paradox is a model that the constructors themselves like. For example, for those liking the RNA-first model for the origin of the Darwinism, a useful paradox moves via deductive logic from “obviously correct” premises to the conclusion that RNA cannot possibly have emerged abiologically. This is a way that a paradox can mitigate human confirmation biases, a way to manage the issues raised by Feynman. Indeed, science students may be taught to construct paradoxes as a “game” to train them to these ends.

Constructing paradoxes is also a way to productively focus research effort. This use is well understood in physics. As Niels Bohr once remarked (Moore, 1966), “How wonderful that we have met with a paradox. Now we have some hope of making progress”. That remark reflects the ability of paradoxes, even if they turn out to be misperceived, to focus attention on important things, and direct our attention away from activities that will have little impact.

Paradox construction thus offers a tool to prevent research from going down rabbit holes. If one sees that the outcome of the evolved research program, even if it is fully successful, will have no impact on the logic of the paradox, it may be time to select a different program.

4. Illustrating the role of paradoxes in guiding origins research

4.1. A Miller-inspired paradox. Ribose, and therefore RNA, cannot emerge abiologically

Let us illustrate the value of paradoxes by considering ribose, the “R” in RNA. Ribose is rather unstable to self-reaction to give complex mixtures of products. This is a property of all carbohydrates, which are also often known as sugars. Indeed, this property is known to anyone who has ever heated sugar in the kitchen and watched it caramelize.

A quarter century ago, Stanley Miller, a founder of modern prebiotic chemistry, quantitatively measured the instability of ribose, an excellent example of quantitative and observational science, good for a science fair project (Larralde et al. 1995). He and his coauthors found half-lives for the decomposition of ribose in water to be 44 years on ice at neutral pH 7, but only 73 minutes at 100°C at pH 7. This is, of course, the caramelization that is known to every chef (amateur and professional).

Miller concluded that the instability of ribose “preclude[s] the use of ribose and other sugars as prebiotic reagents except under very special conditions. It follows that ribose and other sugars were not components of the first genetic material” [emphasis added]. With more Aristotelian formality, Miller inspired a paradox whose conclusion is incompatible with an RNA First model for life’s origin, unacceptable to any scientist who wants that model to be true:

(a)   Ribose decomposes at a quantitatively measured rate,

(b)   Any molecule that decomposes at this rate cannot be part of an abiotically-formed genetic material,

(c)   Therefore, ribose could not have been part of an abiotically-formed genetic material.

Those of us who accept these premises and accept Aristotelian logic are compelled, it would seem, to abandon the Standard View for the origin of life. The only hope we might find in Larralde et al. (1995) is the qualification: “except under very special conditions”.

4.2. The instability of sugars is rooted in the C=O carbonyl group

Of course, scientists rarely accept logic that contradicts a preferred theory, as many have noted (Feyerabend, 1985, 1993), Boyd (1991). Instead, scientists are inclined to try to rescue the theory. This inclination is not necessarily bad, Feynman aside. The premises of a paradox may be flawed in a non-obvious way. If an attempt to rescue the theory makes obvious those flaws to give new insights, it is worthwhile.

Here, the rigor of physical organic chemistry puts the instability of ribose into a broader context. Instability is not a problem just with ribose specifically. Rather, ribose has a feature of its molecular structure that causes it to decompose in aqueous solution. This feature is a carbon atom doubly bonded to an oxygen atom. This “carbonyl group” (C=O) confers special reactivity on a molecule if its carbon atom is bonded to another carbon atom that is bonded to a hydrogen atom. This allows any molecule having this structural feature (not just ribose) to “enolize” (Fig. 2).

Enolization is the first step in a manifold of processes that destroy carbohydrates (Fig. 2). First, the “enolate” can react with another carbonyl molecule to give an “aldol” reaction. Alternatively, the enolate can suffer “elimination”, “tautomerization”, and reaction with other “electrophiles” (Fig. 2). The rate of enolization increases as the temperature increases and as the pH moves away from neutrality. For example, at pH 10.5, rates of enolization of carbonyl compounds at room temperature are not measured in years, but in minutes (Illangkoon, 2010).

The “aldol product” arising from the reaction of a carbohydrate molecule with an enolate often (but not always) ends up having a carbonyl group of its own (Fig. 2). Further, often but not always, the carbon of the C=O group in the product is attached carbon bonded to a hydrogen atom. In this case, the product can itself again enolize, tautomerize, eliminate, and undergo aldol reactions, leading to more products, and eventually to caramel and tar. These self-reactions prevent the accumulation of most carbonyl compounds, except perhaps under “very special conditions”.

Enolization occurs in water, but many compounds can encourage (catalyze) this process and participate in it, including acids, bases, and those that contain reduced nitrogen (e.g. ammonia, NH3). If the products become large enough to absorb ultraviolet light, the manifold of complexifying reactions becomes larger and larger. The result is “tar”, the natural fate on Earth of biological organics removed from Darwinism, including, oil, coal, and graphite.

Miller was not alone in recognizing that ribose and other C=O carbonyl compounds might be too reactive to have participated in the origins of life, at least not in amounts that might self-react. For example, a quarter century ago, Sutherland and Weaver (1994) noted that the reactivity of C=O group in glycolaldehyde phosphate required the C=O moiety be introduced last in a laboratory synthesis of this compound. Eschenmoser had proposed glycolaldehyde phosphate to be an intermediate in the abiological synthesis of ribose (Wagner et al., 1990), in part because he expected the phosphate group to control the base-catalyzed enolization of glycolaldehyde.

5. Descending rabbit holes unconstrained by paradoxes. Examples in origins research

With this understanding of physical organic chemistry, even non-chemists can evaluate a sequence of reactions that is proposed to have occurred abiologically. They must inspect a path-hypothesis to find any C=O compounds that it might contain.

5.1. An Orgel hard problem: The formation of cytidine

For example, Fig. 3 contains a path-hypothesis for a prebiotic synthesis of cytidine, one of the four building blocks of RNA (Powner et al., 2009). You can play along. Identify in Fig. 3 all of the compounds containing a C=O carbonyl groups, and circle those if the carbon of the C=O group is attached to a second carbon having a carbon-hydrogen bond.

Players of this game find three carbonyl compounds in this particular path-hypothesis. One of these is glycolaldehyde (labeled 10 in Fig. 3). Another is glyceraldehyde, labeled 9 in Fig. 3. The third, labeled 5, is not on the specific path that concerns us at the moment, and we will set this aside.

It turns out glycolaldehyde 10 and glyceraldehyde 9 actually enolize more rapidly than ribose. This is because they cannot form the 5-membered cyclic ring that ribose can (see Fig. 5 for reactions that C=O compounds can undergo that transiently control their reactivity). Therefore, glycolaldehyde 10 and glyceraldehyde 9 decompose faster than ribose. Thus, the reasoning that generates the Miller-inspired paradox, which precludes ribose as a prebiotic reagent, would seem to also apply even more strongly to glycolaldehyde 10 and glyceraldehyde 9. Thus, Fig. 3 does not solve the Miller-inspired paradox; instead, it would seem to aggravate it by requiring large amount of two less stable carbonyl compounds.

Work suitable for high school science fair projects confirms this. This work has studied in detail the reactions of glycolaldehyde 10 and glyceraldehyde 9, both free in solution (Harsch et al. 1984) and in the presence of borate (Ricardo et al., 2004; Kim et al., 2011), a species that has long been considered in prebiotic chemistry (Cimpoiasu et al. 1999)(Scorei et al. 1999) (Prieur, 2001) (Scorei & Cimpoiasu, 2006).

We start by discussing the “self-reaction” of each. What happens if many molecules of glycolaldehyde 10 or glyceraldehyde 9 are present in solution without any other molecules?

The answers are well known. Specifically, if the enolate of 10 encounters an unenolized molecule of 10, the two react in the aforementioned aldol reaction (Fig. 4, top). Adding carbon atoms, the two carbons in 10 would sum with the two carbons in the enolate to give one of two 4-carbon carbohydrates, which are called threose and erythrose (Fig. 4, top). Likewise, if the enolate of compound 9 encounters an unenolized molecule of 9, the two react also in an aldol reaction (Fig. 4, top). Adding the three carbons in 9 to the three carbons in the enolate give one of many 6-carbon carbohydrates; fructose is one of these (Fig. 4, top).

The path-hypothesis in Fig. 3 does not mention these self-reaction possibilities. We can ourselves, however, add these to Fig. 3 to give Fig. 4. This expanded scheme makes clear that the path-hypothesis in Fig. 3 requires glycolaldehyde 10 to wait without enolization until a molecule of 8 comes along. Compound 9, glyceraldehyde, must wait without enolization until a molecule of 11 arrives.

This creates a new problem. Enolization is, in the jargon of the physical organic chemist, a “pseudo-first order reaction”. Its rate depends only on the pH of the aqueous solution. The reaction of 10 with 8 and with 11 are “second order reactions”. The reactions desired in Fig. 3 occur in competition with enolization only if the pH is near neutral and the concentrations of 8 and 11 are high.

In the published work (Powner et al., 2009), the concentrations of glycolaldehyde and glyceraldehyde were high, the materials were pure, and the mixing of reagents was timed by a graduate student. How was this to be managed on the Hadean Earth without a graduate student’s attention? In a paper reporting part of the Fig. 3 path hypothesis, Sutherland speculated that the aminooxazole 11 might be synthesized in separate location, with the synthesis “followed by [its] evaporation, sublimation, and subsequent rain-in” at a “separate location” to get the desired outcome (Anastasi et al., 2007). The geological plausibility of this has been questioned.

This notwithstanding, even if the Hadean had a source of and 10, the path-hypothesis does not resolve the Miller-inspired paradox. On the contrary, it doubles it by replacing one unstable carbonyl compound (ribose) with two even less stable carbonyl compounds (glycolaldehyde and glyceraldehyde). Having lost our focus on the paradox, we have gone into a rabbit hole.

One might propose to resolve these new problems by having scarce glycolaldehyde dropped into a large amount of cyanamide 8. This way, perhaps the scarce glycolaldehyde might encounter before it has a chance to enolize. This raises a new problem, however: 8 is itself a reactive molecule. In particular, cyanamide 8 hydrolyzes in water (half life of 1 day at pH 12 and 50°C) to give 6 (Buchanan & Barsky, 1930); it also reacts with itself. So cyanamide also cannot accumulate to await addition of scarce glycolaldehyde 9.

Experimentally in laboratory experiments controlled by graduate students, Powner et al. managed the competing enolization of glycolaldehyde 10 and glyceraldehyde 9 by providing a phosphate buffer to maintain the pH at about 7, where enolization is the slowest. Importantly, their work found several unexpected ways that phosphate influences the fates of various species in this path-hypothesis.

However, the requirement for phosphate generated another problem, as its influence was seen at only high concentrations (1 molar, 1 M). Most phosphate salts have low solubility, far below 1 M; sodium and potassium phosphates the only ones that routinely deliver 1 M phosphate to a solution. In contrast, the salt of phosphate with the ferrous ion (which features in other path-hypotheses, see below) is the largely insoluble mineral vivianite.

Thus, the analysis of the path-hypothesis in Fig. 3 finds that it does not resolve the Miller inspired paradox in any general way. The path-hypothesis avoids ribose, a positive. But it requires large amounts of two less stable carbonyl compounds, glycolaldehyde 10 and glyceraldehyde 9. And it requires phosphate at concentrations that are incompatible with most common counterions, giving us a new problem to solve.

5.2. Path-hypotheses to solve this “Orgel hard” problems need not be necessary

The path-hypothesis in Fig. 3 was criticized specifically by Robert Shapiro (Wade, 2009). before his untimely death. Shapiro served as the needed gadfly for self-uncritical prebiotic chemistry. Shapiro (2007) likened such multistep pathways to “a golfer who had successfully played a golf ball around an 18 hole course, could assume that the ball could then play the same course by itself, through a combination of the wind and other natural forces”.

What motivated Powner et al. (2009) to develop this path-hypothesis to create cytidine in the first place? It was not to get prebiotic ribose. In 2004, Ricardo et al. showed that in a Hadean environment with borate, glycolaldehyde 10 and glyceraldehyde 9 in the amounts required by the Fig. 3 path-hypothesis would have reacted with each other via enolization to give large amounts of borate-stabilized ribose and other pentoses. Indeed, absent the formation of the glycolaldehyde and glyceraldehyde in different locations, it would have been difficult for ribose and other pentoses not to be formed in substantial amounts if glycolaldehyde and glyceraldehyde were present in the amounts specified by Powner et al. (2009). This alternative path is captured in Fig. 4.

Indeed, borate alone partly resolves the Miller-inspired paradox with respect to ribose. Borate forms a stable complex with the cyclic form of ribose that is formed in the presence of borate (Ricardo et al. 2004). As noted in Fig. 5, the cyclic form of ribose lacks a C=O carbonyl group. Stabilization of the cyclic forms of ribose by borate complexation offers additional stability by reversibly locking ribose into this cyclic form. Indeed, borate offers a way out of any Miller-inspired paradox for molecules that contain C=O moieties where such cyclic forms are possible (Fig. 5). Such cyclic forms are not available for glycolaldehyde 10 or glyceraldehyde 9.

Nor was the Fig. 3 path-hypothesis motivated by a need for a route to create prebiotic cytosine, the nucleobase portion of cytidine. Cytosine is itself formed in respectable yields from cyanoacetylene or cyanoacetaldehyde 5 and urea 6, the compounds involved in the path-hypothesis in Fig. 3 (Ferris et al., 1968).[2] Thus, if the nitrogen-containing compounds in Fig. 3 had been available in a Hadean environment, it would have been difficult not to form cytosine.

If cytosine and ribose are both available, the question arises: Would it not be simpler to just make cytidine (the ribonucleoside) by combining some derivative of cytosine with some derivative of ribose? This would directly form the glycosidic bond joining the base to the sugar.

Powner et al. (2009) considered this possibility, but felt that it presented an “insuperable” chemical problem. Indeed, they placed in Fig. 3 an “x” over a conjectural step that might combine cytosine and ribose directly. To explain this “x”, Powner et al. (2009) cited a paper by Leslie Orgel (2004) to suggest that “the condensation of ribose 4 and cytosine 3 does not work”.

This citation of Leslie Orgel indicates his well-deserved place in the culture of prebiotic chemists. Indeed, problems in that culture can be distinguished as “Orgel easy” or “Orgel hard”. This distinction is made by examining literature from the Orgel laboratory starting a half-century ago. If Orgel’s lab showed that a compound could be formed under what it felt were “plausibly prebiotic” conditions, then the culture sees this compound as available in the Hadean. If Orgel’s lab stated that a compound could not be formed under what it felt were “plausibly prebiotic” conditions, then the culture sees this compound as unavailable in the Hadean.

This is the source of the “x” in the path hypothesis in Fig. 3. The x-ed out step is “Orgel hard”. This was the premise that motivated the development of the path hypothesis in Fig. 3.

Historians of science will ask: What experiments did Orgel actually do to support the conclusion that problems in directly forming a glycosyl bond between cytosine and ribose are “insuperable”? The specific reference that Powner et al. (2009) cite states only that “[n]o direct synthesis of pyrimidine nucleosides from ribose and uracil or cytosine has been reported”. This may indicate (or not) that someone in Orgel’s laboratory had attempted this reaction without success. This statement is, however, removed by a few logical steps from the conclusion that this process is problematically “insuperable”.

Sanchez and Orgel in 1970 did publish a work-around that differs from the path proposed by Powner et al. (2009) in using ribose, rather than glycolaldehyde and glyceraldehyde, as a starting material (Sanchez & Orgel, 1970). The downstream steps of Powner et al. (2009) follow closely those of Sanchez and Orgel (1970).

However, seeking to form a bond between cytosine and ribose without activating either reactant is likely to be difficult on energetic grounds. The reaction formally involves the removal of water in a process that is likely to have occurred in water. As Powner et al. (2009) point out, this reaction is thermodynamically unexpected.

However, a glycosidic bond between a cytosine and a ribose can be formed from derivatives of one, or the other, precursor. For example, in 2000, Krishnamurthy et al. pointed out that amidophosphates could convert sugars such as ribose into their 1,2-cyclic phosphates (Fig. 4, right) (Krishnamurthy et al., 2000). Any Hadean aquifer having 1 M phosphate (as required for the path-hypothesis in Fig. 3), upon evaporation, certainly had polyphosphates. Any Hadean environment having cyanamide (as required for the path-hypothesis in Fig. 3) certainly had the ammonia needed to react with polyphosphates to give amidophosphates. These cyclic phosphates have no C=O carbonyl group, and therefore mitigate (at least in part) the Miller-inspired paradox; they could have accumulated without self-reaction (Fig. 4).

Further, they could have reacted directly with cytosine to give the Orgel hard glycosidic bond. Kim and Kim (2019) showed that such cyclic phosphates react with many nucleobases, including cytosine, to form the corresponding nucleoside upon evaporative heating (Kim & Kim, 2019). Thus, if the premises of Fig. 3 are accepted, cytidine can be formed even without the novel parts of the path-hypothesis in Fig. 3.

Thus, there was no need to go down the rabbit hole in the first place. If we presume that the Hadean had large amounts of phosphate, cyanamide, glycolaldehyde, and glyceraldehyde, it had not only ribose and cytosine, but also the precursors to make cytidine in processes that are as workable in the laboratory as those in Fig. 4.

5.3. Selecting between closely alternative path hypotheses using Occam’s razor

This does not mean, of course, that the path-hypothesis in Fig. 3 did not operate in the Hadean. Prebiotic chemists like Gerald Joyce, who describe RNA as a “prebiotic chemist’s nightmare” (Joyce & Orgel, 1999), will have better dreams with either path-hypothesis. They might even prefer having both, with one being a backup for the other.

However, advocates for the path-hypothesis in Fig. 3 have argued that their model is preferred over one involving borate based on an “Occam’s razor” argument (Sutherland 2016). Those familiar with the Sagan-Druyan film Contact are aware of scientists seeking to explain their preference for one model over an alternative using a concept of “simplicity”. Ellie Arroway, played by Jodie Foster, argued for a principle known as “Occam’s razor”.

Occam’s razor is actually not a good criterion for comparatively evaluating models in science, since different people having different world-views have different opinions about what is “simple”. Occam’s razor is especially difficult to apply when comparatively evaluating two very different models. From Contact, Ellie Arroway may have been transported 130 light years in 10 seconds to converse with (fictional) human-shaped aliens who had invaded her brain. Or she may have been a dupe of a grant hoax of the (equally fictional) billionaire S.R. Hadden. Which modeler is “simpler” depends on one’s view of the world.

Occam’s razor is more easily applied when the competing models are similar, and even more if they share premises. For example, the path-hypothesis in Fig. 3 requires large amounts of glycolaldehyde and glyceraldehyde. Never mind where they come from. A world with large amounts of pre-formed glycolaldehyde 10 and glyceraldehyde 9 would also have had access to large amounts ribose (if borate were present), by simple mixing (Ricardo et al. 2004). Further, a world having access to one molar phosphate and reduced nitrogen (such as NH3) would have had access to amidophosphates to make the 1,2-cyclic phosphate from that ribose. And a world with abundant cyanamide and cyanoacetylene most likely offered abundant cytosine, which forms cytidine by condensing with ribose 1,2-cyclic phosphate.

The two models are identical in their premised required environments. Which is simpler? This is simply a matter of perspective. And as Feynman, Feyerabend, and many others have pointed out, the perspectives of scientists often find simpler the models that they themselves are advocating.

6. Going further down the rabbit hole adds paradoxes

In part because of criticisms like those of Robert Shapiro, Sutherland’s group then set out to find path-hypotheses to make glycolaldehyde 9 and glyceraldehyde 10. One recent path-hypothesis is summarized in Fig. 6, adapted from Xu et al. (2018). It features:

(a)  the reaction of atmospherically formed hydrogen cyanide (HCN) with dissolved ferrous iron (Fe2+), thought to be the dominant oxidation state of iron in the Hadean oceans (Bray et al., 2018), to form ferrocyanide complexes,

(b) the reaction of cyanide ion, containing one carbon atom, with formaldehyde to form a “cyanohydrin” (which lack a C=O moiety),

(c)  photochemical reduction with UV light of the -CN group in the cyanohydrin with ferrocyanide, to give glycolaldehyde with one more carbon,

(d) repeating the cycle using sulfur dioxide (SO2) as the reductant.

Again, non-chemists can play an evaluation game by circling the C=O carbonyl compounds in Fig. 6 that might suffer enolization. There are several. We then ask about the expected rate of enolization of these relative to other processes, in particular, their reaction with cyanide. Again, it is important to remember that the rate of enolization is “pseudo first order”, depending on the pH of the solution. The reaction of a C=O compound with cyanide is second order; its rate depends on the concentration of total cyanide and pH. Cyanide itself is unstable in water, hydrolyzing to give formamide and (then) ammonium formate (Fig. 7).

The path-hypothesis in Fig. 3 managed problematic enolizations by keeping the pH near neutrality, about 7. This was done with large amounts of inorganic phosphate. With the path-hypothesis in Fig. 6, neither a low pH nor large amounts of phosphate rescues those enolizable species. First, phosphate precipitates with Fe2+ to give the aforementioned mineral vivianite. Second, in a series of studies, Toner and Catling (2019) show that ferrocyanide is formed at 25°C only if the pH is above 10. At this pH, rates of enolization of all of the C=O compounds in Fig. 6 are measured in minutes (Illangkoon, 2010), and the rate of hydrolytic destruction of cyanide has reached its fastest plateau (Miyakawa et al., 2002). The propensity of HCN to itself undergoes hydrolysis prevents the problem from being solved by adding scarce C=O compounds to an excess of cyanide.

Adding further to the challenge is the requirement in the path-hypothesis in Fig. 6 for high-energy photons at each reductive step. In their 2009 contribution, Powner et al. (2009) remarked on the unexpected stability of beta-ribocytidine-2’,3’-cyclic phosphate upon UV irradiation. They noted that “irradiation ... destroy[s] most other pyrimidine nucleosides and nucleotides”. Here, they require this destructive irradiation throughout the process.

Further, trained chemists expect aminooxazole to be degraded by ultraviolet light as well. In an interesting set of studies, Todd et al. recently confirmed this photodecomposition (Todd et al. 2019).[3] If UV irradiation is necessary for a process to made glycolaldehyde 9 and glyceraldehyde 10 by sequential photochemical reductions, and if UV irradiation destroys another key precursor 11 and most other RNA building block products, a paradox that now needs resolution might be:

  • UV photons at 254 nm must be present for the path-hypothesis in Fig. 6 to operate, as they must be present for certain steps.
  • UV photons at 254 nm must not be present for the path-hypothesis in Fig. 6 to operate, as they destroy molecules that must be present for certain steps.

But we are still not free from paradoxes. The photoactive ferrocyanides that participate in this path-hypothesis must be on the surface of the Hadean Earth; indeed, they must be unshielded on the surface. That surface must be below an atmosphere that must be delivering substantial amounts of HCN, if only to balance the loss of HCN by hydrolysis. However, HCN is synthesized in large amounts only in atmospheres where the amounts of methane (CH4) or other reduced compounds are substantial relative to the amount of atmospheric CO2. Such atmospheres are mostly hazy.

Unfortunately, as we have gone further down the rabbit hole to get larger amounts of glycolaldehyde and glyceraldehyde, the Miller-inspired paradox related to the intrinsic instability of carbonyl compounds has survived. At least in Fig. 3, we had only two unstable C=O compounds, glyceraldehyde in glycolaldehyde, to worry about. Their instability was controlled by a phosphate buffer at neutral pH. And if the path-hypothesis in Fig. 3 struck us as not being “simple” enough, we could get cytidine directly from the ribose that unavoidably would form, the high concentrations of phosphate to make aminophosphates, the nitrogenous compounds to make cytosine, and the processes proposed by Krishnamurthy et al. (2000) and Kim & Kim (2019).

In Fig. 6, the glyceraldehyde and glycolaldehyde are transient intermediates, not pre-formed reservoirs. This is good. However, we have moved in Fig. 6 to a pH where they both enolize in minutes. We now require an atmosphere to make HCN that is likely opaque to the photons required in the ferrocyanide-mediated cyanohydrin reduction invoked at many points in the path-hypothesis of Fig. 6. The paradoxes write themselves.

This observation has been the inspiration of work now being done by Kevin Zahnle and David Catling to attempt to manage the new haze paradox, modeling that we will discuss later (Zahnle et al. 2020).

7. The path-hypothesis emerging from the Atlanta workshop

This example is illustrative about how a field can go down a rabbit hole that began under a premise that later work found to be wrong, here, the “insuperable” problems of any path that might combine directly a derivative of ribose with a derivative of cytosine. This is a problem with research groups, with entire fields, and with entire cultures, and is certainly not limited to origins of life. Avoiding it requires a discipline that we seek to train in students: Every now and then, we ask students return to first principles to understand why we are doing what we are doing. Formulating paradoxes that (apparently) stand between the state-of-the-art and the “big picture” goal is a useful way to train scientists to do this. This is why the discussion in the unusual Atlanta workshop was set to resolve paradoxes.

First, it is well accepted that photochemistry high up in a moist atmosphere with CO2 generates in substantial amounts of formaldehyde. The amount of H2C=O formed is largely independent of the amount of methane (CH4) or other reducing agents in the atmosphere (Harman et al. 2013). Further, various quantitative estimates for the amounts of H2C=O formed in the Hadean atmosphere are available (Pinto et al. 1980).

Lacking a carbon carrying a hydrogen atom adjacent to the C=O group, formaldehyde cannot enolize. Further, unlike HCN, H2C=O has no hydrolytic path to destruction. Aqueous pools of H2C=O are stable against self-reaction, if the pH is not so high as to support the Cannizzaro reaction (pH >> 10). Thus, the Atlanta discussants regarded H2C=O as being reliably available in the Hadean atmosphere at amounts estimated by Pinto et al. (1980).

Much smaller amounts of the 2-carbon glycolaldehyde (HOCH2-HC=O) were also indisputably formed by analogous processes in the Hadean atmosphere. However, the amount of glycolaldehyde formed is quite dependent on the amount of methane (CH4) or other reduced species in the atmosphere (Harman et al. 2013). In the current consensus model for the Hadean Earth, Earth’s mantle had already approached its modern oxidation state (Wade & Wood, 2005), with measurements on surviving zircons placing it near the fayalite-magnetite-quartz redox buffer (FQM -0.5 ± 2.3) (Trail et al. 2011). This means that relatively little glycolaldehyde was formed in the standard Hadean atmosphere. Even if substantial reducing power was present, HOCH2-HC=O was likely to be only ~ 1 ppm relative to H2C=O. This means that glycolaldehyde can participate in prebiotic chemistry only as a catalyst, not as a stoichiometric reagent (as in the path-hypothesis in Fig. 3).

Critical to the fate of both formaldehyde and glycolaldehyde, sulfur emerging from the FMQ Hadean mantle was predominantly SO2 rather than H2S. Indeed, the rate of SO2 likely to emerge from Hadean volcanoes is on the same order of magnitude as the rate of formation of formaldehyde.

An atmosphere that had both SO2 and H2C=O could not avoid the generating product of the reaction between the two. In aerosols, H2C=O, HOCH2-HC=O, and other C=O compounds, react reversibly with SO2 to form “bisulfite addition products” (Fig. 7) (Graedel & Weschler, 1981). These products are sulfonates having the general formula R-CH(OH)SO3 (Fig. 7). This reaction requires no intelligent guidance. Indeed, these sulfonates are formed in atmospheric aerosols above Earth today, where H2C=O and SO2 are products of human activities (e.g. burning coal).

What forms first, a sulfonate or a cyanohydrin? Here, atmospheres having little CH4, NH3, or other reduced compounds form very little HCN. Thus, in such atmospheres in redox equilibrium with the bulk mantle, the bisulfite addition product of H2C=O (HMS) is the dominant or exclusive form. The same for glycolaldehyde, in the small amounts that it is formed. Another aphorism of Albert Eschenmoser, that “the answer [to the origins of life] has to come from revisiting the chemistry of HCN” (quoted in Sutherland, 2016), is hardly relevant to a Hadean atmosphere that lacks HCN.

Neither sulfonate (H2C(OH)SO3- or HOCH2-CH(OH)SO3-) is volatile; both are soluble in water. Therefore, both must rain out into any aquifer that the Hadean offered (as indeed they do today). If the aquifer and the water evaporates, the two form evaporite sulfonate minerals, similar to borates and halites (Kawai et al. 2019). Neither has a C=O group. Therefore, even in water, these compound do not directly react. In dry form, the sulfonates are quite stable. Therefore, they must accumulate, absent their being washed into a global ocean.

Depending on the precise environment, the expected estimated accumulation is ~ 0.1 milligram of H2C(OH)SO3per century precipitating per cm2, or about one gram/m2. This is based on the estimate (Pinto et al. 1980) that ~ 3 x 108 molecules of formaldehyde were formed per cm2 every second in the Hadean atmosphere; with 3 x 109 sec/century, this corresponds to 1018 molecules, or 10-6 moles.

The addition reaction between C=O compounds and bisulfite arising when SO2 dissolves in water is reversible (Fig. 7). The rates of the forward and reverse reaction are well-studied functions of temperature and pH (Sorensen & Andersen, 1970)(Kok et al. 1986)(Dong & Dasgupta, 1986). Thus, if a bed of evaporite sulfonate minerals is rehydrated, H2C(OH)SO3- slowly bleeds free H2C=O into the aquifer. The HOCH2-CH(OH)SO3- slowly bleeds glycolaldehyde into the same aquifer, in much smaller (ppm) amounts. This chemistry would again have been hard to avoid in any constrained aquifers on the Hadean Earth.

We can play the “carbonyl game” again (Fig. 8). Glycolaldehyde that leaks from its sulfonate derivative can enolize with a rate depending on the pH, temperature, concentration of dissolved borate, and other features of the environment. However, if the enol of glycolaldehyde does form, it is a million fold more likely (or more) to encounter a H2C=O molecule (which cannot enolize) than a second glycolaldehyde molecule.

This means that the enolate of glycolaldehyde will react with formaldehyde in a 2 + 1 = 3 carbon reaction to yield glyceraldehyde; the 2 + 2 formation of erythrose or threose is a million fold disfavored because of the low concentration of glycolaldehyde.

The rates associated with this process are well measured, including in the presence of the borate (Kim et al. 2011) (Fig. 9). Further, the high reactivity and relative abundance of H2C=O means that H2C=O traps enolates arising from all enolizable C=O compounds. Even at low concentrations (10 mM), H2C=O prevents competing aldol reactions, beta eliminations, hydride shifts (Appayee & Breslow, 2014), and other processes that give tars (Fig. 2). It even prevents protonation of the enolate at pH values greater than ~ 9. Thus, in a quantitative study, even at pH >11 with Ca(OH)2 and temperatures as high as 80°C, H2C=O prevents tar formation of carbohydrates; only after H2C=O is consumed does “yellowing” characteristic of tar formation begin (Ricardo et al. 2006).

The fate of the glyceraldehyde is likewise constrained. It can itself react with the bisulfite necessarily in the environment to form its own bisulfite addition product (Fig. 7). This product again lacks a C=O group; it can accumulate. However, should it enolize, the glyceraldehyde enolate will first encounter H2C=O because of its relatively high concentration, not glycolaldehyde or another molecule of glyceraldehyde. In the presence of borate, the predominant product arising from this encounter is erythrulose (with 4 carbons, from a 3 + 1 = 4 reaction) (Fig. 8).

Erythrulose can also enolize. However, borate controls the direction of enolization (Fig. 9) and the products that arise when its enolate reacts with H2C=O. Again, if H2C=O is in excess, the enolate of erythrulose almost reacts to form one of two diastereomeric branched 5-carbon sugars, from a 4 + 1 = 5 reaction. Borate and formaldehyde suppress other reactions (Fig. 2) that give tar.

This is a “maturation” process. It will happen if carbonyl compounds that almost certainly were raining from the Hadean atmosphere are left to themselves in the presence of borate, bisulfite, and excess formaldehyde in a constrained aquifer. It does not require the intervention of a graduate student; it is a hard-to-avoid outcome of a carbohydrate pool.

Indeed, it is the process that if, prolonged, leads to caramelization. Thus, the key question, if maturation is to yield any species (such as the ribose of (Larralde et al. 1995)) useful in an abiological synthesis of RNA is: How will the maturation be constrained before tar forms?

The answer to this question lies in part in the structures of the molecules that emerge via this maturation. In particular, the branched pentoses are privileged in several of their structural features. First, they are the first compounds in the maturation that can form a cyclic structure that lacks a C=O group, all by themselves (Fig. 8, top right). Glycolaldehyde, glyceraldehyde, and erythrulose all cannot.

Therefore, these branched pentoses are also the first molecules in the maturation of a carbohydrate pool derived from Hadean sulfonate minerals that are able to form a tight complex with borate (Fig. 8, top right). This makes the branched pentoses the first higher carbohydrates to be formed in this maturation that can accumulate.

Finally, for those playing the carbonyl game, these branched pentoses, even in their ring-open form that has a C=O moiety, do not have an adjacent carbon carrying a hydrogen atom (Fig. 8, top right). Thus, the branched pentoses are the first molecules to emerge in the process that cannot enolize, even in the absence of bisulfite or other molecules that might react to remove their C=O bond. Thus, they may accumulate as stable organic minerals.

Further, if the carbonyl compounds mature from their bisulfite adducts, this maturation is self-quenching. Every time a new carbon-carbon bond is formed via an aldol addition of species arising from the dissociation of a sulfonate precursor, a dissolved bisulfite molecule is released without a C=O partner. Thus, as the aldol reactions proceed in a constrained aquifer, the matured products encounter higher and higher concentrations of bisulfite. These increasingly higher concentrations of bisulfite in a constrained aquifer increase the conversion of carbonyl compounds to sulfonates, slow further reactions, and slow the formation of tar.

Thus, the Atlanta discussants recognized that maturation controlled by bisulfite, borate, excess formaldehyde, and perhaps even amidophosphates, mitigates the Miller-inspired paradox arising from the measured instability of ribose (Larralde et al. 1995) discussed a quarter century earlier. This mitigation follows directly from current models for the Hadean atmosphere, the Hadean Sun, and the Hadean FMQ mantle. Even the early view that borate minerals were not present in adequate abundance in the Hadean crust due to insufficient plate tectonics has been set aside (Hazen et al., 2008)(Grew et al., 2011). Sizeable amounts of borate are present on the accessible surface of Mars, which has never had plate tectonics (Gasda et al. 2017). Further, the very recent observation of ribose in meteorites (Furukawa et al. 2019) is possible non-laboratory evidence that such maturation actually can be observed in a natural environment. Finally, something suitable for a science fair project.

8. What happens next?

8.1. The fate of carbohydrates that are stabilized

Sutherland has criticized this model by noting that the metastability of the borate complex of the branched pentose is a flaw, not an asset, of the maturation (Sutherland, 2016). He wondered how that complex would be mobilized to form nucleosides.[4]

Of course, the fate of the borate complexes of the branched pentoses depends on the balance between incoming H2C=O, borate, and bisulfite. If HOH2CSO3- continues to precipitate into a constrained aquifer, the amount of branched pentose will come to exceed the amount of stabilizing borate. This will generate a free branched pentose.

Again, the branched pentose molecules that are freed from a borate ligand cannot enolize; they lack an adjacent carbon with a C-H. They can, however, undergo a retroaldol fragmentation reaction (Fig. 8). That 5 à 2 + 3 fragmentation gives the now-familiar glycolaldehyde and glyceraldehyde, the first as its enol (Fig. 8). Indeed, the fact that these branched pentoses cannot enolize is the reason why the normally slower retroaldol reaction manifests itself in these branched pentoses.

In the absence of H2C=O, glycolaldehyde enolate and glyceraldehyde formed from this fragmentation can combine directly to give ribose and other linear (not branched) pentoses (Ricardo et al. 2004). Thus, if formaldehyde is absent, the branched pentose-borate complex will gradually mature to give the more stable borate complex of ribose and other linear pentoses. These too can accumulate as reservoirs for future prebiotic synthesis, as they lack a C=O group in their cyclic forms complexed by borate.

Alternatively, if H2C=O is present, the glycolaldehyde enolate and glyceraldehyde enolate can fix more H2C=O. This is preferred as long as the amount of H2C=O is significant. The result is a catalytic cycle, where multiple molecules of H2C=O are fixed for each molecule of glycolaldehyde that was originally presented.

This overall cycle was shown experimentally by Neveu et al. (2013) and Kawai et al. (2019), who studied the reaction manifold using 13C-labeled formaldehyde. To make the reaction rates large enough to be conveniently measured, the reactions were followed at pH 10.5, a bit higher than that expected in aquifers exposed to serpentinizing basalts.

8.2. The Atlanta discussion combined the carbohydrate maturation with other ideas

Thus, the October 2018 Atlanta workshop participants had at their disposal some largely inevitable chemistry that would have delivered linear pentoses, like ribose, in mineral stable forms via “hands free” maturation of bisulfite-trapped C=O compounds emerging from the Hadean atmosphere. This control comes from the special reactivity of H2C=O, mineral species such as borate, and bisulfite that emerged from an FMQ mantle. Fig. 10 summarizes this chemistry.

Further, the workshop participants had available a process that would transform the matured carbohydrates to their cyclic phosphates, provided that the conditions required by Krishnamurthy et al. (2000) were available. They also had chemistry able to form the “Orgel hard” glycosidic bonds of the nucleosides. They also had available multiple ways to phosphorylate nucleosides to give mono-, di-, and triphosphates, stereoselectively in the presence of borate. And they had multiple possible ways to convert those phosphorylated nucleosides into oligomeric RNA that might be stabilized on silica phases.

The overall model is captured in Fig. 10. However, astute readers (as well as the Atlanta workshop participants) recognize a paradox. We already noted that the Hadean Earth likely had an FMQ mantle delivering correspondingly oxidized minerals to the surface of the Earth beneath a redox-neutral atmosphere with little methane, carbon monoxide, ammonia, H2S, or dihydrogen. This explicitly excluded substantial amounts of HCN (as noted above), HCCCN, H2NCN, and other compounds that make the formation of the nucleobases “Orgel easy”. Our focus on the Miller-inspired paradox related to the instability of enolizable C=O compounds had not solved another paradox; we needed reduced species that are not present, according to the premises of the model. Another paradox.

The multidisciplinary Atlanta group understood, however, that the Hadean environment was neither static nor placid. Rather, it was interrupted by impacts, some rather large. Much literature views the impactors as sources of HCN and other reduced compounds directly. This itself generated a paradox; an impactor large enough to deliver these (apparently) needed precursors in the quantity needed would create the heat that destroyed them. Here, attempts to resolve this paradox have focused on local environments.

The Atlanta group therefore turned to the view that the impactors were not the sources of the needed reduced organic materials themselves, but rather a source of reducing power that would create a reduced atmosphere that would allow the reduced organic materials to be made there. Substantial impacting bodies (>1021 kg) would have had their own iron cores, which would have likely shattered on impact with the Hadean Earth. This shattering would deliver molten iron (Fe0) to the Hadean atmosphere.

The Fe0 delivered by the impactor must have reduced water, N2, CO2, and other species in the atmosphere. It must also have delivered the Ni2+ (about 20% of a typical iron meteorite) to allow nucleoside triphosphate formation (Benner et al. 2019a). Any NH3 generated from the reduced iron would have been available for the Krishnamurthy-Eschenomser process to activate cyclic trimetaphosphate (or a reactivity equivalent) to generate ribose-1,2-cyclic phosphate.

And, of course, the reduction due to the impactor would have allowed the atmospheric formation of HCN, HCCCCN, NCCN, H2NCN, and other well recognized precursors of the nucleobases, as well as their hydrolysis products, including formamide, urea, and cyanoacetaldehyde. All of these are necessary to make the formation of RNA nucleobases Orgel-easy. Admittedly, a productive atmosphere would have been hazy, although Zahnle et al. (2020) have noted that production of HCN (if not the other species) may have continued for a short time after the haze had cleared.

The model is agnostic with respect to the size and the date of the relevant impactor. Important only is that the impact that enables RNA formation not be followed by a subsequent impactor that sterilized the planet. Participants at the Atlanta workshop pointed out that the last sterilizing impactor may have been a Moon-sized body, called Moneta. Weighing in at 1023 kg, Moneta would have been big enough to deliver the late veneer of siderophiles (Mojzsis et al. 2019), the heavy metals (like platinum and gold) that arrived on Earth after Earth’s core closed. Moneta was certainly sterilizing, and would likely have re-set most of the geological clocks that we have on Earth.

A smaller 1021 kg Ceres-sized impactor is likely to have also been sterilizing. However, would not have reset all of the clocks that we find on Earth as it delivered a reducing atmosphere. A still smaller 1020 kg Vesta-sized impactor would not have sterilized the Earth or reset the clocks, but still could transiently generate a productively reducing atmosphere.

Inspired by the output of the Atlanta workshop, Zahnle and his coauthors have now modeled a spectrum of outcomes of impactors of different sizes (Zahnle et al., 2020) and the productivity of the reducing atmosphere that they generated. This work includes estimates for the amount of time following an impact the atmosphere remained productively reduced, how hazy the atmosphere was during its productive period, and how much material would be generated.

 Here, the important word is “transiently”. An atmosphere that gets its reducing power from the dispersion of molten iron from an impactor does not keep that reducing power for long. Gasses continue to come from the mantle, which is not reduced by any impactor smaller than the hypothetical Moneta, and not significantly reduced even with Moneta itself.

Further, the reducing property of the atmosphere is lost as dihydrogen escapes to space. With Moneta, which is modeled to have generated as much as 90 bars of H2 (Genda et al., 2017a, 2017b), the half-life for the restoration of an unproductive redox neutral atmosphere is measured in the tens of millions of years. For a Vesta-sized impactor, the time constant for the loss of a productive atmosphere is measured in the tens of thousands of years.

For those hoping to accumulate more RNA precursors, bigger impactors are better. However, most recent models for the impact history of the Earth have impactor size monotonically decreasing over time. Thus, a Ceres-sized impactor may have occurred 1 ± 1 times after a Moneta-sized impactor. If it occurred, it was sterilizing, and therefore is the impactor most likely to be relevant to the abiotic formation of RNA. In contrast, if this impactor did not arrive, and Moneta was the last sterilizing impactor, Moneta is likely to be the most relevant, as it created the longest-lived productively reducing atmosphere. The later Vesta-sized impactors (of which there were likely several) are conceivably important, but RNA formation would need to have occurred rapidly.

8.3. Dating when RNA most likely emerged

The highlight of the Atlanta discussion then came as Ramon Brasser pointed out to the chemistry and molecular biology participants that by combining the chemical and geological information, we might estimate when this chemistry most likely occurred. The impact energy of a Moneta would likely have reset the radiogenic clocks in the crust (Abramov et al., 2013). A survey of ~ 200,000 detrital zircons from the Jack Hills in Western Australia shows the oldest of these at ~ 4.38 Ga (Harrison et al., 2017) (Valley et al., 2014). This suggests that RNA formation must have occurred more recently.

Mojzsis et al. (2019) placed these dates for terran material within the context of Solar System-wide events. Of special importance is the date for the onset of the migration of the giant planets. They offer a timeline that relates radiometric ages from asteroidal meteorites to dynamical models that account for late accretion and consider the lack of reset ages after ca. 4.45 Ga. This, they argue, confined the onset of the migration to no later than about 4.48 Ga, with planetesimals and asteroids continuing to strike the inner planets in agreement with crater chronology.

This suggests a starting constraint on when the window of opportunity for RNA formation opened if Moneta was the relevant impactor. It could be no earlier than 4.48 Ga, and no later than 4.45 Ga. An impactor having a size able to deliver most of the veneer in one body would have created a magma ocean effectively everywhere on the surface of the Earth. This would have (as noted above) reset the geological clocks and destroyed any organic molecules made earlier. Thus, if 10 million years after a 4.48 Ga impact is needed for a surface to cool enough to allow it to accumulate organics, the window opened at 4.47 Ga.

With Hloss following a simple exponential with a half-life for of 40 million years for a Moneta-sized impact, atmospheric production of primary nucleobase precursors might be half at 4.44 Ga, a quarter at 4.40 Ga, and an eighth at 4.36 Ga, if the rate of formation of primary precursors is approximated as a linear function of H2 partial pressure. The model may be biased to favor later dates, reflecting (presumably) increasing amounts of sub-aerial land. This estimates a date for the optimal accumulation of RNA precursors at ~ 4.36 Ga ± 0.1 Ga. This, according to the model, is the time when the formation of RNA was most probable, as well as the formation of RNA-based Darwinism under an RNA-First model.

Smaller impactors, as noted above, could also have transiently made the Hadean atmosphere productive for the formation of RNA nucleobase precursors. These would have supported Orgel-easy processes to make the nucleobases themselves, nucleobases that would then react with cyclic phosphates via the Kim & Kim (2019) process. However, the time would have been shorter, and the amounts of material would have been smaller. This implies that these are less probable as geological events to initiate an RNA world. This consideration is especially true given a model where the first RNA organisms were heterotrophic, relying on the environment to provide them their first foodstuffs.

9. The next generation of paradoxes

This is how the Atlanta workshop in October 2018, freed from PowerPoints, was able in an interactive discussion to identify a chemical path to make RNA from starting materials likely available on a Hadean Earth, identify geological events to make those materials available, and estimated dates when those events most likely made RNA. With the delivery by an impactor of reducing power to the Hadean atmosphere, the natural maturation of carbonyl compounds to give pentoses, and various sources of phosphate (and phosphite, with reduced phosphorus from an impactor as outlined by Gull et al. (2015)), a generation of paradoxes was laid to rest. Indeed, nearly all of the paradoxes that had been mentioned in the call for proposals issued by the Foundation for Applied Molecular Evolution in a grant program supported by the John Templeton Foundation have one or more resolutions available to them. RNA is much less of a “prebiotic chemistry nightmare”.

Other problems remain, of course. One of these relates to the amount of semi-arid dry land at the relevant time on Earth. Estimates for the amount of continental crust in the Hadean cover a wide range (Korenaga, 2018), indicating the difficulties of modeling this feature of early Earth. Further, the amount of continental crust need not translate directly into the amount of dry land. Of course, a very large impactor, by reducing water, would have increased the amount of available semi-arid dry land.

Further, considerably more work must be devoted to study carbohydrate maturation. Analytical methods remain difficult. In particular, for convenience, experiments exploring this maturation process have largely been run at pH 10-10.5; this allows the rates of these processes to be measured on a convenient time scale (pace the objection by Ritson et al. (2018) that carbon dioxide in the atmosphere would have buffered this pH, making the enolization reactions slower and therefore more difficult to study). More work is needed to understand how the product manifold changes at lower pHs, across temperature, and under other conditions that may have been more abundant in constrained Hadean aquifers. Further, amidophosphates can phosphorylate glycolaldehyde and glyceraldehyde as well as ribose. Work must be done to see if this generates a paradox demanding resolution.

Our understanding of the status of phosphorus on the Hadean surface is also limited. Many processes for formation of, for example, cyclic trimetaphosphate require elevated temperatures, for example 600°C (Gard, 1990). While these temperatures were almost certainly available on a surface with volcanism, little of this parameter space has been explored.

Also missing from this overview is a mechanism to obtain homochirality. As discussed elsewhere (Benner, 2017), homochirality is required for an evolvable informational molecule. The Atlanta group found no mechanism to get nucleosides that all have the same enantiomeric form. Because the Simons Collaboration had team members addressing this issue, the Templeton Foundation consortium had not focused in this area. We are awaiting insights from the Simons Collaboration into this very important problem.


Figure legends

Figure 1. Two representative cofactors that have a piece that delivers reactivity not available in the 20 standard amino acids (in black) attached to a piece of RNA (in orange). While this does not demand that Darwinism began on Earth through the abiological formation of RNA, it does provide evidence that an early episode of life on Earth used RNA in a larger capacity than is used at present.

Figure 2. Compounds that contain a carbonyl group (C=O) whose carbon atoms are attached to another carbon atom that carries a hydrogen atom can undergo enolization. This initiates a series of reactions that include reaction with the C=O group of second carbonyl compound in an “aldol reaction”. Important to the propensity of carbohydrates (often called sugars) to form tar is that the products of these reactions are often (but not always) themselves new carbonyl compounds whose carbon atoms are attached to another carbon atom that carries a hydrogen that can undergo enolization. Unless these reactions are stopped promptly by an attentive graduate student, the mixture can become quite complex. The rate of enolization is a function of pH. For simple carbohydrates like glycolaldehyde and glyceraldehyde, at pH 10.5 in room temperature, the rates of enolization are measured in minutes.

Figure 3. The path-hypothesis to make cytidine, from Powner et al. (2009). Carbonyl groups are boxed in red. The “Orgel-hard” formation of a glycosidic bond is marked with a red x.

Figure 4. The path-hypothesis to make cytidine, from Powner et al. (2009), to which are added reactions that likely occurred under conditions specified by Powner et al. (2009). Reactions in red are destructive. Reactions in blue are constructive. If we accept the premises of Powner et al. (2019) on their own terms, cytidine is available prebiotically without the novel process that they propose. However, by introducing two new C=O compounds, both less stable to self-reaction than ribose, the path does nothing to resolve the Miller-inspired paradox. Indeed, it makes it worse.

Figure 5. Ways to stabilize C=O compounds via a reversible reaction that generates a product lacking that a C=O group. Some (e.g. sulfonates) products cannot further react. Others, like the cyanohydrin, can hydrolyze irreversibly to act as a sink for the C=O compound.

Figure 6. A path-hypothesis that exploits the ability of Fe2+ to sequester cyanide, sulfite from volcanic SO2 to serve as a reductant, and high-energy photons in a process that sequentially lengthens short linear carbohydrates to longer carbohydrates. Reduced HCN, also proposed to be of atmospheric origin (here, from the late heavy bombardment), is used with metals (Na+, K+, Ca2+) to create ferrocyanide (Keefe & Miller 1996), which is proposed to serve as a source for HCN, a catalysts for the reduction of organic nitriles, and as a source of the cyanamide needed to generate an aminooxazole. Purine ribonucleosides would presumably be made by other processes. Adapted from Xu et al. (2018) and Ritson et al. (2018).

Figure 7. Volcanic sulfur dioxide becomes sulfurous acid in water aerosol particles, giving them a pH of ~4. When rained on to surface having alkaline pH, the bisulfite anion is formed. This reacts with carbonyl C=O groups to give their bisulfite addition products, which are sulfonates. This reaction is quite general, as discussed in the text. The sulfonates are quite stable to self-reaction and other degradation paths. Their only reaction at lower temperatures is their reversible dissociation to give back the carbonyl compound. Therefore, sulfonates slowly bleed reactive species into aqueous mixtures. When formaldehyde is in excess, it captures enolates of carbohydrates before they can unproductively react to give tars.

Figure 8. Maturation of carbohydrates when formaldehyde is in excess.

Figure 9. Carbohydrate maturation in the presence of borate, which guides enolizaton of four carbon carbohydrates and directs the regiochemistry of the attack of HCHO on their enol.

Figure 10. Schematic for a path-hypothesis that yields RNA nucleotides by direct joining of preformed canonical nucleobases to preformed ribose derivatives via a glycosidic bond (magenta) (Kim & Benner 2017)(Kim & Kim 2019). The stereochemistry of various chiral molecules is arbitrary. This path-hypothesis invokes reservoirs of carbohydrates (red) arising from formaldehyde and traces of glycolaldehyde stabilized by SO2 (yellow) (Kawai et al. 2019) emerging from a Hadean mantle having an oxygen fugacity (redox state, fO2) near the fayalite-magnetite-quartz buffer (fO2 = FMQ -0.5 ± 2.3) (Trail et al., 2011). Black lines (left) indicate aldol reactions that carbohydrates undergo if released from their bisulfite adducts in the presence of borate and trimetaphosphate transformed with ammonia (Krishnamurthy et al. 2000). This path-hypothesis requires sub-aerial surfaces intermittently submerged by water. Last, it requires HCN, HCCCN, H2NCN, and other reduced atmosphere-generated primary precursors (in blue, not all shown). Formation of these depends strongly on the redox state of the atmosphere. For one critique of this path-hypothesis, see the Supplementary Information from Ritson et al. (2018).


Acknowledgements: The material is based in part upon work supported by NASA under award NNX14AK37G. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Aeronautics and Space Administration. This publication was also made possible through the support of a grant from the John Templeton Foundation 54466. The opinions expressed in this publication are those of the author(s) and do not necessarily reflect the views of the John Templeton Foundation.


Abramov, O., Kring, D.A., & Mojzsis, S.J. (2013). The impact environment of the Hadean Earth. Chemie der Erde-Geochemistry 73, 227-248.

Anastasi, C., Crowe, M.A., Sutherland, J.D. (2007) Two-step potentially prebiotic synthesis of alpha-d-cytidine-5’-phosphate from D-glyceraldehyde-3-phosphate. J. Am. Chem. Soc. 129, 24-25.

Appayee, C., Breslow, R. (2014) Deuterium studies reveal a new mechanism for the formose reaction involving hydride shifts. J. Am. Chem. Soc. 136, 3720-3723.

Attwater, J., Wochner, A., Holliger, P. (2013) In-ice evolution of RNA polymerase ribozyme activity. Nature Chem. 5, 1011.

Benner, S.A. (2009) The Life, the Universe and the Scientific Method. Gainesville FfAME Press, 312 pp.

Benner, S.A. (2017) Detecting Darwinism from molecules in the Enceladus plumes, Jupiter’s moons, and other planetary water lagoons. Astrobiol. 17, 840-851.

Benner, S.A., Bell, E.A., Biondi, E., Brasser, R., Carell, T., Kim, H.-J., Mojzsis, S.J., Omran, A., Pasek, M.A., Trail, D. (2019) When did life likely emerge on Earth in an RNA-first process? ChemSystChem 1:e190003 doi:10.1002/syst.201900035

Benner, S.A., Ellington, A.D., Tauer, A. (1989) Modern metabolism as a palimpsest of the RNA world. Proc. Natl. Acad. Sci. USA 86, 7054-7058.

Benner, S.A., Kim, H.J., Biondi, E. (2019a). Prebiotic chemistry that could not not have happened. Life, 9(4), 84.

Bernal, J.D. (1951) The Physical Basis of Life. Routledge and Kegan Paul, London, UK.

Boyd, R. (1991). The Philosophy of Science. MIT Press.

Bray, M.S., Lenz, T.K., Bowman, J.C., Petrov, A.S., Reddi, A.R., Hud, N.V., Williams, L.D., Glass, J. B. (2018). Ferrous iron folds rRNA and mediates translation. bioRxiv, 256958.

Buchanan, G.H., Barsky, G. (1930) The hydrolysis and polymerization of cyanamide in alkaline solutions. J. Am. Chem. Soc. 52, 195-206.

Cairns-Smith, A.G. (1977) Takeover mechanisms and early biochemical evolution. BioSystems 9, 105-109

Cairns-Smith, A.G. (1982) Genetic Takeover and the Mineral Origins of Life. Cambridge University Press, UK.

Cairns-Smith, A.G. (2005) Sketches for a mineral genetic material. Elements 1, 157-161

Cimpoiasu, V.M., Steinbrecher, Gy., Scorei, R., Petrisor, I., Brad, I., Olteanu, I., Sbirna, L.B. (1999) TD-NMR titration studies on complexation of borate with polyhydroxylated organic compounds. Implications for chemical evolution at high temperatures. Proceedings of ISSOL, July 11-16 1999 San Diego California, Abstracts, pp. 68.

Dong, S., Dasgupta, P.K. (1986) On the formaldehyde-bisulfite hydroxymethane sulfonate equilibrium. Atmospheric Environment 20, 1635-1637.

Eschenmoser, A. (2007) The search for the chemistry of life’s origin. Tetrahedron 63, 12821-12844.

Eschenmoser, A. (2009) The search for the chemistry of life’s origin. Pontifical Academy of Sciences, Acta 20, 181-199.

Ferris, J.P., Sanchez, R.A., Orgel, L.E. (1968) Studies in prebiotic synthesis: III. Synthesis of pyrimidines from cyanoacetylene and cyanate. Mol. Biol. 33, 693-704.

Feyerabend, P.K. (1975) Against Method. New Left Books, New York.

Feyerabend, P.K. (1985) Realism, Rationalism and Scientific Method: Volume 1: Philosophical Papers (Vol. 1). Cambridge University Press.

Feynman, R. (1974) Cargo Cult Science. Caltech Commencement Address. Reproduced in: Surely You’re Joking, Mr. Feynman. Norton, New York.

Furukawa, Y., Chikaraishi, Y., Ohkouchi, N., Ogawa, N.O., Glavin, D.P., Dworkin, J.P., Nakamura, T. (2019) Extraterrestrial ribose and other sugars in primitive meteorites. Proc. Natl. Acad. Sci. USA 116, 24440-24445.

Gard, D.R. (1990) Method of Forming Amidophosphates in Aqueous Solutions. US Patent 4956163.

Gasda, P.J., Haldeman, E.B., Wiens, R.C., Rapin, W., Bristow, T.F., Bridges, J.C., Schwenzer, S.P., Clark, B., Herkenhoff, K., Frydenvang, Lanza, N.L., Maurice, S., Clegg, S., Delapp, D.M., Sanford, V.L., Bodine, M.R., McInroy, R. (2017) In situ detection of boron by ChemCam on Mars. Geophysical Research Lett. 44, 8739-8748.

Genda, G., Iizuka, T, Sasaki, T., Ueno, Y., Ikoma, K. (2017a) Ejection of iron-bearing giant-impact fragments and the dynamical and geochemical influence of the fragment re-accretion. Earth Planetary Sci. Lett. 470, 87-95.

Genda, H., Brasser, T., Mojzsis, S.J. (2017b) The terrestrial late veneer from core disruption of a lunar-sized impactor. Earth Planetary Sci. Lett. 480, 25-32.

Gilbert, W. (1986). Origin of life: The RNA world. Nature 319, 618.

Graedel, T.E., Weschler, C.J. (1981) Chemistry within aqueous atmospheric aerosols and raindrops. Rev. Geophys. 19, 505-539.

Grew, E.S., Bada, J.L., Hazen, R.M. (2011) Borate minerals and origin of the RNA world. Origins of Life and Evolution of Biospheres 41, 307-316

Gull, M., Mojica, M.A., Fernández, F.M., Gaul, D.A., Orlando, T.M., Liotta, C.L., Pasek, M.A. (2015). Nucleoside phosphorylation by the mineral schreibersite. Scientific Reports 5, 17198.

Harman, C.E., Kasting, J.F., Wolf, E.T. (2013) Atmospheric production of glycolaldehyde under hazy prebiotic conditions. Origins Life Evolution Biosphere 43, 77-98.

Harrison, T.M., Bell, E.A., Boehnke, P. (2017). Hadean zircon petrochronology. Reviews Mineralogy Geochemistry 83, 329-363.

Harsch, G., Bauer, H., Voelter, W. (1984). Kinetik, Katalyse und Mechanismus der Sekundärreaktion in der Schlußphase der Formose‐Reaktion. Liebigs Annalen der Chemie 1984, 623-635.

Hazen, R.M., Papineau, D., Bleeker, W., Downs, R.T., Ferry, J.M., McCoy, T.J., Sverjensky, D.A., Yang, H. (2008). Mineral evolution. American Mineralogist 93, 1693-1720.

Illangkoon, H.I. On the origins of life: The prebiotic synthesis of carbohydrates, primitive catalytic cycles & engineering the genetic lexicon. Dissertation. University of Florida, Gainesville, FL, 2010.

Joyce, G.F. (2002) The antiquity of RNA-based evolution. Nature 418, 214-221.

Joyce, G.F., Orgel, L.E. (1999) Prospects for understanding the origin of the RNA world. In: The RNA World, 2nd ed, edited by R.F. Gestland, T.R. Cech, and J.F. Atkins, Cold Spring Harbor Press, Cold Spring Harbor, NY, pp 49-78

Kawai, J., McLendon, D.C., Kim, H.J., Benner, S.A. (2019) Hydroxymethanesulfonate from volcanic sulfur dioxide. A mineral reservoir for formaldehyde in prebiotic chemistry. Astrobiol. 19, 506-516.

Keefe, A.D., Miller, S.L. (1996) Was ferrocyanide a prebiotic reagent? Origins life Evolution Biosphere 26, 111-129.

Kim, H.J., Benner, S.A. (2017) Prebiotic stereoselective synthesis of purine and noncanonical pyrimidine nucleotide from nucleobases and phosphorylated carbohydrates. Proc. Natl. Acad. Sci. USA 114, 11315-11320.

Kim, H.J., Kim, J. (2019). A prebiotic synthesis of canonical pyrimidine and purine ribonucleotides. Astrobiology 19, 669-674.

Kim, H.J., Ricardo, A., Illangkoon, H.I., Kim, M.J., Carrigan, M.A., Frye, F., Benner, S.A. (2011) Synthesis of carbohydrates in mineral-guided prebiotic cycles. J. Am. Chem. Soc. 133, 9457-9468.

Kok, G.L., Gitlin, S.N., Lazrus, A.L. (1986). Kinetics of the formation and decomposition of hydroxymethanesulfonate. J. Geophys. Res.: Atmospheres 91(D2), 2801-2804.

Korenaga, J. (2018) Crustal evolution and mantle dynamics through Earth history. Phil. Trans. Royal Society A 376, 20170408.

Krishnamurthy (2018) Experimentally investigating the origin of DNA/RNA on early Earth. Nature Comm. 9, 5175.

Krishnamurthy, R., Guntha, S., Eschenmoser, A. (2000) Regioselective α-phosphorylation of aldoses in aqueous solution. Angew. Chem. Int. Ed. 39, 2281-2285.

Kuhn, T.S. (1966) Structure of Scientific Revolutions. 3rd ed. Chicago: Univ. Chicago Press.

Larralde, R., Robertson, M.P., Miller, S.L. (1995) Rates of decomposition of ribose and other sugars. Implications for chemical evolution. Proc. Natl. Acad. Sci. USA 92, 8158-8160.

Miyakawa, S.; Cleaves, H.J.; Miller, S.L. (2002) The cold origin of life: A. Implications based on the hydrolytic stabilities of hydrogen cyanide and formamide. Orig. Life Evol. Biosph. 32, 195-208.

Mojzsis, S.J., Brasser, R., Kelly, N.M., Abramov, O., Werner, S.C. (2019) Onset of giant planet migration before 4480 million years ago. Astrophys. J. 881, 44-56.

Moore, R. (1966) Niels Bohr: The Man, His Science, & the World They Changed, p. 196.

Nelson, K.E., Robertson, M.P., Levy, M., Miller, S.L. (2001) Concentration by evaporation and the prebiotic synthesis of cytosine. Origins Life Evolution Biosphere 31, 221-229.

Neveu, M., Kim, H.-J., Benner, S.A. (2013) The “Strong” RNA World hypothesis. Fifty years old. Astrobiology 13, 391-403.

Orgel, L.E. (2004) Prebiotic chemistry and the origin of the RNA world. Crit. Rev. Biochem. Mol. Biol. 39, 99-123.

Pinto, J., Gladstone, G., Yung, Y. (1980) Photochemical production of formaldehyde in Earth’s primitive atmosphere. i>Science 210, 183-185.

Plankensteiner, K., Reiner, H., Rode, B.M. (2005) Prebiotic chemistry: The amino acid and peptide world. Current Organic Chemistry, 9(12), 1107-1114.

Popper, K.R. (2002) The Logic of Scientific Discovery. Routledge.

Powner, M.W., Gerland, B., Sutherland, J.D. (2009) Synthesis of activated pyrimidine ribonucleotides in prebiotically plausible conditions. Nature 459, 239-242.

Powner, M.W., Sutherland, J.D. (2011) Prebiotic chemistry: A new modus operandiPhilosophical Trans. Royal Soc. B: Biological Sciences 366, 2870-2877.

Prieur, B.E. (2001). Étude de l’activité prébiotique potentielle de l’acide borique. Comptes Rendus de l’Académie des Sciences-Series IIC-Chemistry, 4(8-9), 667-670.

Quine, W.V.O. (1953) From a Logical Point of View. Cambridge, Harvard University Press.

Ricardo, A., Carrigan, M.A., Olcott, A.N., Benner, S.A. (2004) Borate minerals stabilize ribose. Science 303, 196.

Ricardo, A., Frye, F., Carrigan, M.A., Tipton, J.D., Powell, D.H., Benner, S.A. (2006) 2- Hydroxymethylboronate as a reagent to detect carbohydrates. Application to the analysis of the formose reaction. J. Org. Chem. 71, 9503-9505.

Rich, A. (1962). On the problems of evolution & biochemical information transfer. In Horizons In Biochemistry, 103-126.

Ritson, D.J., Battilocchio, C., Ley, S.V., Sutherland, J.D. (2018) Mimicking the surface and prebiotic chemistry of early Earth using flow chemistry. Nature Comm. 9, 1821.

Saladino, R., Šponer, J.E., Šponer, J., Di Mauro, E. (2018) Rewarming the primordial soup. Revisitations and rediscoveries in prebiotic chemistry. ChemBioChem 19, 22-25.

Sanchez, R.A., Orgel, L.E. (1970). Studies in prebiotic synthesis: V. Synthesis and photoanomerization of pyrimidine nucleosides. J. Molecular Biol. 47, 531-543.

Scorei, R., Steinbrecher, Gy., Cimpoiasu, V.M., Petrisor, I., Scorei, V., Mitrut, M. (1999) Boron compounds in the primitive Earth. Implications for prebiotic evolution, Proceedings of ISSOL, July 11-16 1999 San Diego, California, Abstracts, pp. 33.

Scorei, R., Cimpoiaşu, V.M. (2006). Boron enhances the thermostability of carbohydrates. Origins of Life Evolution Biospheres 36, 1-11.

Segré, D., Ben-Eli, D., Deamer, D.W., Lancet, D. (2001). The lipid world. Origins of Life and Evolution of the Biosphere, 31(1-2), 119-145.

Service R. (2019) Seeing the dawn: Evidence lines up to offer a new view of how life on our planet may have emerged. Science, 363, 116-118.

Shapiro, R. (1999) Prebiotic cytosine synthesis: A critical analysis and implications for the origin of life. Proc. Natl. Acad. Sci. USA 96, 4396-4401.

Shapiro, R. (2002) Comments on ‘Concentration by evaporation and the prebiotic synthesis of cytosine’. Origins Life Evolution Biosphere 32, 275-278.

Shapiro, R. (2007) A simpler origin for life. Scientific American 296, 46-53.

Smith, L.D. (1986) Behaviorism and Logical Positivism: A Reassessment of the Alliance. Stanford University Press. p. 314.

Sorensen, P.E., Andersen, V.S. (1970). The formaldehyde-hydrogen sulphite system in alkaline aqueous solution. Kinetics, mechanisms, and equilibria. Acta Chem. Scand, 24, 1301-1306.

Sutherland, J.D. (2016) The origin of life. Out of the blue. Angew. Chemie Internat. Ed. 55, 104-121.

Sutherland, J.D. (2017) Opinion: Studies on the origin of life. The end of the beginning. Nature Reviews Chem. 1, 0012

Sutherland, J.D., Weaver, G.W. (1994) Synthesis of bis (glycoaldehyde) phosphodiester and mixed glycoaldehyde-triose phosphodiesters. Tetrahedron Lett. 35, 9109-9112.

Todd, Z.R., Szabla, R., Szostak, J.W., Sasselov, D.D. (2019) UV photostability of three 2-aminoazoles with key roles in prebiotic chemistry on the early earth. Chem. Comm. 55, 10388-10391.

Toner, J.D., Catling, D.C. (2019) Alkaline lake settings for concentrated prebiotic cyanide and the origin of life.Geochimica Cosmochimica Acta, 260, 124-132.

Trail, D., Watson, E.B., Tailby, N.D. (2011) The oxidation state of Hadean magmas and implications for early Earth’s atmosphere.Nature 480, 79-82.

Valley, J.W., Cavosie, A.J., Ushikubo, T., Reinhard, D.A., Lawrence, D.F., Larson, D.J., Clifton, H., Kelly, T.F., Wilde, S.A., Moser, D.E. Spicuzza, M.J. (2014) Hadean age for a post-magma-ocean zircon confirmed by atom-probe tomography. Nature Geoscience 7, 219.

Visser, C.M., Kellogg, R.M. (1978a) Bioorganic chemistry and the origin of life. J Mol Evol 11, 163-168.

Visser, C.M., Kellogg, R.M. (1978b) Biotin. Its place in evolution. J Mol Evol 11, 171-187.

Wächtershäuser, G. (1992). Groundworks for an evolutionary biochemistry: The iron-sulphur world. Progress Biophysics Molecular Biol. 58, 85-201.

Wächtershäuser, G. (1995) The uses of Karl Popper. Royal Institute of Philosophy Supplements, 39, 177-189.

Wade, J., Wood, B.J. (2005) Core formation and the oxidation state of the Earth. Earth Planetary Science Lett. 236, 78-95.

Wade, N. (2009) Chemist shows how RNA can be the starting point for life. New York Times, 05/14/science/14rna.html? pagewanted=all

Wagner, E., Xiang, Y.B., Baumann, K., Gück, J., Eschenmoser, A. (1990) Chemie von α-Aminonitrilen. Aziridin-2-carbonitril, ein Vorläufer von rca-O3-Phosphoserinnitril und Glycolaldehydphosphat. Helvetica Chimica Acta, 73, 1391-1409.

White, H.B. III (1976) Coenzymes as fossils of an earlier metabolic state. J. Mol. Evol. 7, 101-104.

Xu, J., Ritson, D.J., Ranjan, S., Todd, Z.R., Sasselov, D.D., & Sutherland, J.D. (2018). Photochemical reductive homologation of hydrogen cyanide using sulfite and ferrocyanide. Chemical Comm. 54, 5566-5569.

Zahnle K.J., Lupu, R., Catling, D.C. (2020). Creation and evolution of impact-generated reducing atmospheres of early Earth. Planetary Science J., submitted.


* Paper adapted from a presentation to the 2018 Plenary Session of the Pontifical Academy of Sciences at the Vatican in November 2018.


[1] The Origins Of Life 2018 Conference

[2] But see Shapiro’s criticism (Shapiro, 1999), Miller’s reply (Nelson et al., 2001), and Shapiro’s comments to the reply (Shapiro, 2002).

[3] Todd et al. do not abandon the path-hypothesis, hoping that an unidentified photostable molecule might act as a sunscreen to protect the aminooxazole that is used on the path-hypotheses in Fig. 3 and Fig. 6. A challenge then emerges to find a screening molecule that can selectively screen out the photons that cause the undesired decomposition of the aminooxazole without screening out the photons needed to perform the photoreduction step involving ferrocyanide complexes. If the photons have the same wavelength, the paradox writes itself.

[4] Sutherland’s (2016) criticisms (as well as those in Ritson et al. (2018)) overlook the maturation process in the path-hypothesis of Fig. 7, including the privileged structures and special reactivities of the branched pentoses, mistakenly argue that the dissociation of borate complexes require “acidification and repeated addition and evaporation of methanol” to be released, misunderstand the physical organic chemical rates and concentration-dependence of various reactions, and mistakenly confuse the path-hypothesis in Fig. 7 with the “formose reaction”. The “formose reaction” (Alexander Butlerov, 1828-1886) is the formation of sweet tasting material (formose) by the self-reaction of H2C=O upon heating in Ca(OH)2 which has mechanistic challenges missing from the path-hypothesis in Fig. 7. Nevertheless, those criticisms are worth considering.