Arsenic cannot replace phosphorus at the origin of life

Arsenic and old lace

When I read that Science and NASA were holding a press conference on the role of arsenic as a central element of life elsewhere in the cosmos, and that it might have played a central role in the origin of life, I could not believe my eyes. But it was.  I immediately wrote to Science  with a short version of the text below, but received no reply (except that my comment had been received). So I decided to publish it in The Journal of Cosmology, a fairly obscure journal. My text was quickly accepted, but curiously amended to blame only the peer reviewers and editors and to defend the scientists involved in this nightmare.

It is true that scientific publications are now like advertisements (and scientists do indeed have to pay to be published, in many journals). It becomes all the more essential that scientists react and try to conform to the scientific method. Science should return to its original construction process, the critical generative method.

We should all have remembered Frank Westheimer's 1987 article, aptly titled: "Why nature chose phosphates", in the very same journal, Science: 235: 1173–1178, where he explained the peculiar situation encountered by the metastability of phosphate bonds, and why phosphate is so important to life as we know it. The fact that we have forgotten this important article shows how science has drifted from knowledge creation to a kind of signalling and 'display' between people, in a world of mass media communication. Arsenic is a toxic contaminant of water worldwide, so it is essential to avoid spreading misconceptions about this metalloid.

Here is my first reaction in a letter to Science (unpublished, december 3, 2010):

Wolfe-Simons and co-workers claim that they have identified a bacterial species where arsenic plays most of the role played by phosphorous in all known extant living species (1). I welcome the idea of discovering novel examples of the extraordinary propensity of life to capture all kinds of processes that may help it to transmit information over generations. Yet, arsenic is very unlikely as a major component of the backbone of nucleic acids, despite its similarity with phosphorus, essentially because it would form extremely unstable macromolecules.

Life provides us already with a caveat in this domain. Quite a few modifications of the (deoxy)ribose-phosphodiester backbone exist in nature, but they seldom affect the phosphate group. In a case in point one atom of sulfur replaces an oxygen of phosphate at places of the backbone of DNA in Streptomyces lividans (2). This modification results in the degradation of DNA in vitro by oxidative, double-stranded, site-specific cleavage during normal and pulsed-field gel electrophoresis, making completion of genome sequencing a difficult task. Several genomes from Cyanobacteria possess the same modification, with similar instability of their DNA at the modified positions (accounting for lack of completion of several genome projects). This modification, which was long unidentified, is comparable to the phosphorous - arsenic change, as sulfur belongs to the same family as oxygen in Mendeleiev's table, but it is located at a much less crucial position in the backbone.

Performing experiments with "pure" compounds is notoriously difficult. Many investigators have observed that Escherichia coli in media containing "pure" sodium salts will grow, albeit slowly. This is not due to sodium replacing potassium but to a remarkable enrichment of potassium present in elusive traces. Another - unfortunately common - practice is to grow bacteria in synthetic media where common salts (potassium, magnesium and iron) are included, without any addition of essential metals such as zinc or manganese (3). Yet bacteria grow, and grow well. I do not suspect them to be able to transmute matter, or to universally substitute zinc or manganese by iron of magnesium. In the same way the well known "fluoride effect" results from extraction of aluminum from glassware, making AlF4, which, in fact, mimicks phosphate (4).

But there is a stronger argument that should be taken into consideration. I have belonged to a team that sequenced the genome of an arsenic-loving organism, Herminiimonas arsenicoxydans (5). This organisms uses, indeed, arsenic to perform electron transfers that permit it to manage energy. It has also a way to detoxify this element. A remarkable conjecture can be proposed to explain how it does so. While this organism does not use selenium (as other bacteria such as E. coli do, to make selenocysteine, the 21^st amino acid), it has the counterpart of the selD gene. I note that Halomonas elongata possesses the counterpart of this gene (HELO_4105 ). SelD in "standard" bacteria is the enzyme that makes selenophosphate, used to modify serine on a tRNA that translates UGA codons into selenocysteine. Selenium is the counterpart of sulfur in Medeleiev's table, it occupies a volume that is slightly larger than the latter. A similar situation is met with arsenic and phosphorous. Hence, a small distortion of the standard SelD catalytic site would easily accomodate sulfur instead of selenium and arsenate instead of phosphate. The role of SelD would then be to make thioarsenate. This compound is insoluble (it is found in stones), and this would be an excellent way to detoxify arsenic, while removing it from the medium, leaving room for minuscule traces of phosphate to be accumulated and play their normal role.

This hypothesis and others should be further explored before we propose that life as we know it can develop using arsenic in the place of phosphorus.

1. Wolfe-Simon F, Blum JS, Kulp TR, Gordon GW, Hoeft SE, Pett-Ridge J, Stolz JF, Webb SM, Weber PK, Davies PCW, Anbar AD, Oremland RS. A Bacterium That Can Grow by Using Arsenic Instead of Phosphorus. Science 2010 dec 3
2. Zhou X, He X, Liang J, Li A, Xu T, Kieser T, Helmann JD, Deng Z. A novel DNA modification by sulphur. Mol Microbiol. 2005 Sep;57(5):1428-38.
3. Miller, JH A short course in bacterial genetics. A laboratory manual and handbook for Escherichia coli and related bacteria. 1992. Cold Spring Harbor Laboratory Press
4. Bigay J, Deterre P, Pfister C, Chabre M. Fluoroaluminates activate transducin-GDP by mimicking the gamma-phosphate of GTP in its binding site. FEBS Lett. 1985 Oct 28;191(2):181-5.
5. Muller D, Médigue C, Koechler S, Barbe V, Barakat M, Talla E, Bonnefoy V, Krin E, Arsène-Ploetze F, Carapito C, Chandler M, Cournoyer B, Cruveiller S, Dossat C, Duval S, Heymann M, Leize E, Lieutaud A, Lièvremont D, Makita Y, Mangenot S, Nitschke W, Ortet P, Perdrial N, Schoepp B, Siguier P, Simeonova DD, Rouy Z, Segurens B, Turlin E, Vallenet D, Van Dorsselaer A, Weiss S, Weissenbach J, Lett MC, Danchin A, Bertin PN. A tale of two oxidation states: bacterial colonization of arsenic-rich environments. PLoS Genet. 2007 Apr 13;3(4):e53.

Naturally, we do not know well how cells cope with a considerable level of arsenic, but we begin to have some ideas of the many biochemical solutions to the riddle. For example we have found, in the sequencing of the genome of Heminiimonas arsenicoxydans, that oxido-reduction played a major role. Yet, the best way to cope with a toxic compound is to fix it under an insoluble form. Many biological polymers may allow this. Similar situations are met with many metals, magnesium, calcium, iron and manganese in particular. As a matter of fact metallurgic industry copes with arsenic by forming insoluble compounds. Research on arsenophilic bacteria should focus on biochemical processes resulting in the formation of insoluble or low toxicity arsenic compounds. The hypothesis proposed here (formation of monothioarsenate) is just a conjecture. Many others should have been proposed.

April 2012: After discussion with the geochemist Raoul-Marie Couture we wrote an article proposing a detailed scenario whereby some bacteria could synthesise monothioarsenate, a fairly innocuous derivative of arsenic. The scenario is hypothetical and quite wild, of course, but it shows that we should explore many biochemical hypotheses before trying to challenge our standard knowledge of the constraints of the laws of physics. This paper is published in Environmental Microbiology.

RM Couture , A Sekowska, G Fang, A Danchin
Linking selenium biogeochemistry to the sulfur-dependent biological detoxification of arsenic
Environ Microbiol (2012) Apr 20. doi: 10.1111/j.1462-2920.2012.02758.x.

Making stable informational polymers in water at 300K limits chemical variations within extremely narrow borders. This is why the basic atoms of life — those that are found in meteoritic molecules — are hydrogen, carbon, nitrogen and oxygen. In these same conditions, management of energy to support life used the unique property of phosphorus to make energy-rich metastable phosphoester or polyphosphate bonds. Analysis of the genome of arsenic-loving bacteria suggests that arsenic can nevertheless be accumulated in bacteria via formation of innocuous derivatives that may decorate inert (mostly non informational) biopolymers. However, arsenic cannot replace phosphorus in this core function of life. This has been previously firmly established by numerous biochemical experiments. Likewise, recent claims by Wolfe-Simon et al. (2010) that this replacement could happen, have not been experimentally verified, and are based on experiments lacking proper controls; the purpose of which was to substantiate these beliefs. However, the authors are not sole to be blamed, but the journals that try to maintain their high impact factors at all cost, publishing articles that should never have reached the public.

As a present for the new year, back in 2008, a prophecy appeared as a peer-reviewed pre-publication. It anticipated that arsenic would be found in the backbone of nucleic acids of living organisms, replacing the ubiquitous phosphorus. The prophecy, as is often the case with this type of beliefs, also suggested a place on Earth where this would happen: Lake Mono in California (Wolfe-Simon et al., 2008). On April 6^th, 2008, this prophecy was communicated to the world by a popularization journal (Reilly, 2008). Now, at the end of 2010, as a Christmas present (in Continental Europe, december 2^nd, the chosen date might have recalled the sun of the Napoleonian Victory of Austerlitz), the NASA sent a sensational press release calling on journalists to tell them that, yes, the prophecy had come true, and not on an exotic planet, but on our old mother Earth and exactly at the place where this was predicted to happen (Wolfe-Simon et al., 2010).

In pre-scientific times, the sayings of prophets were the norm, and nobody would be bothered. Some would follow, some would be miscreants. One may dispute the demarcation between science and non-science, but, in any event, a core criterium to accept facts as belonging to science is to understand that we cannot get direct access to truth (if it ever exists) but only make models of Nature. In this process we must avoid trying to prove that the model is right, but, rather, try to find where models are an inadequate representation of Nature (Popper, 1972). If this fairly standard way or proceeding (explicitly stated in the NIH guidelines, for example) had been followed, rather than running after fame and recognition by mass media (a kind of democratic voting to demarcate what should be true and what should not via people’s acclamations), the Wolfe-Simon et al., (2010) paper should never have reached the pages of scientific journals, but would have remained among the many opinion pamphlets that thrive everywhere, especially on the Internet.

To substantiate this contention I review here basic arguments that should have come to mind before publication: chemical consequences of placement of atoms in the periodic table of element; biochemical experiments that demonstrate fragility of modifications of the phosphodiester bond, and biochemical data documenting instability of arsenate derivatives of nucleotides; ability of living organisms to concentrate elements present in traces in the environment, and analysis of genomic data that suggest a biochemical process permitting cells to alleviate arsenic toxicity.

Many ideas of what life could be have been discussed for centuries and even millenia. The most imaginative authors made it far from the life we know (see for example Fred Hoyle's Black Cloud (Hoyle, 1957)), but the present contentious article assumes that life is based on principles that are identical to those governing our life, but with one big difference: arsenic would replace phosphorus in its construction (Wolfe-Simon et al., 2010). This places us right in front of a textbook approach of life: why are the elements we find in living organisms in limited number, and why are they those we know?

The short answer is straightforward. Life develops around 300K, with water as its bathing medium. A core property of its components is that beside a limited number of small molecules (a few tens of atoms), it is made of macromolecules: giant polymers obtained by elimination of a water molecule between a small alphabet of basic building blocks, amino acids and nucleotides. This seemingly simple arrangement has a remarkable property in terms of information: while the backbone of these polymers is invariable, the side chains can be arranged in an infinite variety of combinations.

This informational view must be associated with physico-chemical constraints at the required temperature. Making molecules, and a fortiori polymers, implies forming stable covalent bonds. A covalent bond appears when electrons share their presence between two or more atoms’ nuclei. Now, the electrons associated to a nucleus are arranged along specific constraints ruled by quantum physics laws. And the consequence of these laws is that the various atoms of the universe can be arranged along rows and columns, according to the way they match their electrons with the charge of their nuclei. This constitutes the periodic table of elements (Figure 1).

As the rank of the row increases, the (negative) electrons of the outer shell of the atom become more and more loosely attached to the (positive) nucleus. In a nutshell, this makes that the best candidates for making stable covalent links are, beside hydrogen, some of the atoms present in the second row of the periodic table. Further down, the atoms involved are mainly involved in electrostatic bonds (much weaker than covalent bonds) and in electron exchanges (oxido-reduction reactions). Another constraint, more anecdotal but often ignored, concerns three atoms: lithium, beryllium and boron. All three are rare in the universe for cosmological reasons (nucleosynthesis during the first stages of formation of our universe (Bernas et al. 1967)). As a consequence, this limits the basic atoms of life to hydrogen, carbon, nitrogen and oxygen (typically the atoms combined in extraterrestrial molecules, when they are found).

As shown in the figure, other atoms are also involved in life. Most, in fact, are playing important roles in specific features of catalytic reactions needed to construct, modify and destroy covalent bonds, electrostatic interactions (metals) and more or less complicated electron exchanges (transition metals and complex anions such as molybdate or tungstate).

Two exceptions remain, that have to be accounted for. Sulfur (the higher homolog of oxygen) and phosphorus (the higher analog of nitrogen). The former, being in a deeper row than oxygen, is mainly involved in exchanges of electrons (its oxido-reduction potential varies from -2 to +6), in formation of energy-rich thioesters and via a remarkable affinity for iron, in making iron-sulfur clusters that have probably had a seminal role on the origin of life on Earth, where it is quite abundant (Wickramasinghe, 1973). The latter, phosphorus, is the candidate that Wolfe-Simon and colleagues proposed to see replaced by arsenic. This is not chemically possible, as we shall discuss now.

Phosphorus, in living cells, is essentially used as phosphate, involved in three major processes: carrying and transporting energy, forming the backbone of nucleic acids, and acting as an energy-dependent tag for regulation. In fact the association of phosphorus with oxygen atoms, making the phosphate anion, has the property to make long chains (polyphosphates) when desiccated. These chains are, of course, prone to hydrolysis, but remarkably, in a highly metastable way. This means that while the stable forms are those which results from hydrolysis (the phosphate anions), to reach this stage, the compound needs to go through a very high activation energy barrier. This is the reason why the phosphate bond has been named energy-rich, since the discovery of the role of ATP by Fritz Lipmann (Lipmann, 1975). And this is why phosphate is the basic currency of energy in life. This metastability extends to the formation of phosphoesters, and in particular phosphodiester bonds. And this permitted the formation of nucleic acids, which use a negatively charged backbone to carry and protect from the environment fragile base pairs in a double helical structure. We note here that the two ubiquitous elements of the third row of Mendeleiev's table that are present in all living cells, sulfur and phosphorus, are involved in the storage of energy (thioesters and phosphates/phosphoesters).

Life warns us already of the intrinsic fragility of these metastable bonds. Modifications of the (deoxy)ribose-phosphodiester backbone exist in nature, but they seldom affect the phosphate group. There is a modification, however, that is relevant in the present context. In Streptomyces lividans a DNA modification system replaces a side oxygen of phosphate by an atom of sulfur at some specific places of the DNA backbone (phosphorothioation (Zhou et al. 2005)). This modification makes DNA unstable in vitro by oxidative, double-stranded, site-specific cleavage during normal and pulsed-field gel electrophoresis, making completion of genome sequencing a difficult task. A database of genomes possessing the same modification, which makes the completion of the corresponding genome sequencing difficult, has been published (Ou et al. 2009). This modification, which was long unidentified, has some conceptual similarity to the phosphorous - arsenic swap, as sulfur belongs to the same family as oxygen in Mendeleiev’s table, but it is located at a much less crucial position in the backbone. It also tells us that beside hydrolysis, oxidation may be central in the instability of arsenic, if it were to replace phosphorus.

But there is even more direct biochemical evidence that makes impossible for arsenic to play the role of a stable component of a nucleic acid backbone in an aqueous environment. Arsenic has been known as a poison for millenia. In recent times biochemists looked into the many reasons accounting for that role. Beside involvement in oxido-reduction reactions (which can be used both to evolve arsenite into its more innocuous form arsenate, and to recover energy), arsenic can indeed replace phosphorus in many phosphorolytic reactions and even form carbohydrate arsenates that are similar to their phosphate analogs (Lagunas & Sols, 1968). However this is limited to exchange reactions that cannot lead to stable and useful compounds. It is also possible to begin to construct compounds along the line predicted to exist if phosphorus could replace arsenic, for example by constructing an arsenical analog of adenosine diphosphate (an essential pre-requisite if the prophecy were true). While difficult, a synthesis in which the phosphono-oxy group of ADP was replaced by the arsenomethyl group (arsenate itself was far too unstable at this position, requiring replacement of the bridging oxygen by a methylene group) was achieved by Dixon and colleagues. The product was unable to compete for ADP in any of its standard substrates (Webster et al. 1978). A further demonstration puts the final nail in the coffin: the arsonomethyl analogue of AMP can be used as a substrate for adenylate kinase. It permits transfer of a phospho group from ATP, but like all anhydrides of arsonic acids breaks down immediately, transforming the enzyme into an ATPase (Adams et al. 1984). The situation would be even worse if the hypothetical arsenical analog of ATP could have existed.

Performing experiments with « pure » compounds is notoriously difficult. Many investigators have observed that Escherichia coli in media containing « pure » sodium salts will grow, albeit slowly and to a limited extent. This is not due to sodium replacing potassium but to a remarkable enrichment of potassium present in elusive traces. Another  - unfortunately common - practice is to grow bacteria in synthetic media where common salts (potassium, magnesium and iron) are included, without any addition of essential metals such as zinc or manganese (Miller, 1992). Yet bacteria grow, and grow well. I do not suspect them to be able to transmute matter, or to universally substitute zinc or manganese by iron or magnesium. In the same way the well known « fluoride effect » results from extraction of aluminum from glassware, making AlF4, which, in fact, mimicks phosphate (Bigay et al. 1985). Specific transporters are meant to import and export essential metals, as well as to regulate their homeostasy in a variety of environments where they usually exist as trace elements (Haas et al. 2009).

In Wolfe-Simon and colleagues experiments the traces of phosphate present as contaminant at most steps in the building up of their growth medium is certainly sufficient to permit growth of the cells, with phosphorus playing its normal role. The main question to answer, then, is not whether phosphate remains present with its standard role, but how the cells cope with an environment which is arsenic-replete.

The genome of several arsenic-loving bacteria has been sequenced, in particular that of Herminiimonas arsenicoxydans (Muller et al. 2007). This organisms uses arsenic to perform electron transfers that permit it to manage energy while alleviating some of its toxicity. But this cannot be enough, most probably, to detoxify this element. Analysis of the genome sequence, as well as that of other arsenophilic bacteria (including immediate kins of the Halomonas species used in Wolfe-Simon and colleagues’ study) suggests a remarkable conjecture to explain how it does so. While H. arsenicoxydans does not use selenium (as other bacteria such as E. coli do, to make selenocysteine, the 21^st amino acid), it has the counterpart of the selD gene. I note that Halomonas elongata possesses the counterpart of this gene (HELO_4105). SelD in « standard » bacteria is the enzyme that makes selenophosphate, used to modify serine on a tRNA that translates UGA codons into selenocysteine.

.

Selenium is the counterpart of sulfur in Medeleiev’s table, it occupies a volume that is slightly larger than the latter. A similar situation is met with arsenic and phosphorous. Hence, a small distortion of the standard SelD catalytic site would easily accomodate sulfur instead of selenium and arsenate instead of phosphate. The role of SelD would then be to make monothioarsenate (Figure 2). This compound can react with a variety of compounds, biopolymers or minerals that it would decorate with arsenic in an innocuous form. This would explain why arsenophilic bacteria contain arsenic.

This hypothesis and others should be further explored before we propose that life as we know it can develop using arsenic in the place of phosphorus.

The idea that arsenic could have replaced phosphorus as a central component of nucleic acids should never have been published in a scientific journal.

This work benefited from discussions with Philippe Marlière. It was supported by MICROME, a Collaborative Project funded by the European Commission within its FP7 Programme, contract number 222886-2.