Hopanoids are pentacyclic triterpenoids - polymers of isoprene with a backbone containing 30 carbon molecules and five rings. The figure at the right shows bacteriohopanetetrol (BHT) - an example of a hopanoid commonly found in the environment.

Hopanoids are derived from bacteria - but not all bacteria. By the most recent estimates, less than 10% of bacterial species, and perhaps less than 2% of the bacterial cells in the ocean contain hopanoids. Yet hopanoids are commonly detected in sedimentary rocks that are hundreds of millions of years old. In one case, hopanoids have been reported in rocks that are 2.7 billion years old.

Can these ancient hopanoids tell us something about the microbes that produced them? What kinds of environments did these organisms live in? What kinds of physiologies did they have? We know very little about the physiological function of hopanoids in bacteria living today - so examining these bacteria is a way to start.

In the Marx lab, we are investigating hopanoids in the model organism Methylobacterium extorquens. We are constructing mutant strains of this bacterium that are deficient in various aspects of hopanoid synthesis. Some of our mutants are unable to modify hopanoids with methyl groups or side chains, or can not make hopanoids at all.

We have found that Methylobacterium mutants deficient in hopanoid production exhibit a strong growth phenotype in liquid, i.e. they grow very poorly compared ot the wild type. This growth phenotype is particularly pronounced during growth on single-carbon compounds. One of the ideas we are exploring is whether the ability of Methylobacterium to grown on single-carbon compounds is related in part to its abundant production of hopanoids in its membrane.

The figure shows two flasks containing Methylobacterium grown in liquid media. The flask on the left contains wild-type (normal) Methylobacterium, while the flask on the right contains our mutant strain of Methylobacterium that produces no hopanoids. This figure demonstrates the severe flocculation and poor growth characteristics of hopanoid-free Methylobacterium. We are examining this trait, along with others, to try to determine what role hopanoids play in Methylobacterium, what this means for hopanoids in evolution, and what this can tell us about the hopanoids we detect in the environment and in the rock record

The Lost City Hydrothermal Field is an alkaline, low temperature hydrothermal field located near the Mid-Atlantic Ridge. Its fluids are rich in calcium, hydrogen, and methane and the microbial community in the vent chimneys is dominated by methane-cycling archaea from the family Methanosarcinales.

Are these archaea producing methane? Or consuming it? The abundant methane in the vent fluids might constitute a rich source for methane-consuming archaea, yet the abundant hydrogen in the vent fluids provide a thermodynamic drive for methane production. We analyzed biomarkers from the carbonate chimneys to determine the answer. The biomarkers are dominated by isoprenoidal and non-isoprenoidal diether lipids. Isoprenoidal diethers mainly derive from methanogenic archaea. Non-isoprenoidal dithers are commonly associated with sulfate-reducing bacteria, although their precise microbial source at Lost City is unknown.

It turns out that in environments where archaea are hypothesized to consume methane, lipids are always depleted in ¹³C relative to methane. In methane producers, lipids are enriched relative to methane. At Lost City, the latter pattern holds: archaea are producing methane.

The most unusual feature of the lipids at Lost City is that they lack the telltale depletion in ¹³C that is usually found in biologically-derived products. This is attributed to carbon-limitation in this ecosystem. Methane is also enriched in ¹³C, with values between -9‰ and -14‰ relative to VPDB.

The most likely explanation? Lost City Methanosarcinales are consuming H₂ and CO₂, and producing methane. If we examine the cell density in the towers, its obvious that they can't be producing all the methane that we observe in the vent fluids. Some of this methane must be coming from other sources, such as abiotic Fischer-Tropsch type (FTT) reactions, or Sabatier reactions. There is too much methane relative to ethane and propane for the reaction to be a pure FTT process, and the maximum temperatures are too low to catalyze this efficiently... yet the peridotite massif doesn't seem to be checmically reduced enough to effectively catalyze Sabatier reactions. So it's not yet completely clear what the precise source of methane at Lost City really is.

Contents of ¹³C in archaeal diethers and methane

¹³C contents of methane and archaeal lipids i) at sites where anaerobic methanotrophy is occuring (HMMV = Haakon Mosby Mud Volcano, HR = Hydrate Ridge, ERB = Eel River Basin, NW BS = Northwest Black Sea); ii) at Lost City; iii) in Methanogens grown on each of three substrates: H₂/CO₂, acetate, and trimethylamine (TMA). Figure from Bradley et al., 2009 GCA 73:102-118

Yellowstone National Park is a fascinating ecosystem on many scales. It is probably the most famous concentration of hot springs and geysers in the world. My interest is in examining the microbial life inhabiting these hot springs, understanding their physiology and the potential for their preservation in the rock record, and trying to determine to what extent these ecosystems are good models for environments on early Earth.

We have examined bulk organic material and lipids from a number of hot springs in Yellowstone, with a particular focus on the alkaline hot springs in the Sentinel Meadows area of the Lower Geyser Basin. Interestingly, many of the biomarkers here are enriched in ¹³C as they are at Lost City - indicating that ¹³C enrichment may be a common feature in alkaline hydrothermal environments. The hydrothermal ecosystem here is much more complex however, with numerous inputs from photosynthetic organisms and also from the surrounding Sentinel Meadows ecosystem

The hydrothermal system is dominated by Crenarchaeota in the higher-temperature regions, and by photosynthetic organisms including cyanobacteria in the lower temperature regions. The Crenarchaeota produce abundant tetraether lipids, including some that appear to have central core moities of more than 40 carbons, which is highly unusual. Through examination of both the structures and the carbon isotope content of these lipids, we are starting to get a better picture of the carbon cycling in this system. By participation in an interdisciplinary team studying microbiology, community genomics, and geochemistry of this system, we hope to get a better understading of how this system functions.

Polar lipids are those lipids that are strongly amphiphilic, meaning that they contain both a non-polar part (such as fatty acids) that is fat-soluble, and also a strongly polar part that is water-soluble. Examples are phospholipids and glycolipids, both of which have very polar headgroups connected through gylcerol to a non-polar core moiety. Lipids such as sterols are not strongly amphiphilic and are termed "neutral lipids".

The polar headgroups of glycolipids and phospholipids are easily lost after cell death. Although the rate at which this happens is not precisely known, intact polar lipids (IPLs) are thought to be a good marker for living cells. The structure of the headgroup is also additional information that we can use to distinguish taxonomic groups and discern environmental processes.

Polar lipids are separated by High-Pressure Liquid Chromatography (HPLC) and analyzed by Mass Spectrometry (MS). By connecting these instruments together (HPLC-MS), we can rapidly assess the lipids in an environmental or laboratory sample.

For example, at Lost City we determined that the polar headgroups associated with non-isoprenoidal diethers derived from bacteria were glycolipids. This was highly unusual because bacterial glycolipid diethers have never been detected before - all known bacterial glycolipids had fatty acid cores, and all known bacterial diethers were phospholipids. The unusual combination in Lost City hydrothermal vents could be an evolutionary relic. Alternatively, it may reflect the unique chemistry of this environment. Lost City has an alkaline pH and vent chimneys contain abundant brucite. This combination of fluid and mineral is expected to render phosphate insoluble. Without any phosphate in vent waters, bacteria are likely to conserve this element (required for DNA & RNA, optional for lipids) as much as possible. One way to do that? Make your lipids with sugars instead of phosphate!

Future work will include examining the relative degradation rate of various polar lipids, testing the hypothesis that polar lipids constitute a biomarker for living cells, and examining the degree to which tetraether lipids with different polar headgroups contain differences in core moieties - with an eye towards testing proxies such as TEX₈₆.

Methanogens have likely been an important part of Earth's biosphere for billions of years. In order to reconstruct their imprint on the Earth, I am interested in understanding how isotopes undergo fractionation as they pass through the biosynthetic networks of methanognes.

As elements are processed through biosynthetic networks, they undergo isotope fractionation. Carbon is the central element of biomolecules, and the distribution of carbon isotopes in natural products is critical to our interpretations of past environments.

In order to interpret the isotopic content of natural products in the environment or in the rock record, we must understand the source of isotope fractionation. In biosynthetic networks, carbon isotopes are particularly useful. The magnitude of carbon isotope fractionation depends on the ¹³C content of the substrates, along with the substrate concentrations, the kinetic or equilibrium isotope fractionations imposed by various enzymes involved in processing the substrate, and the amount of carbon that flows to different parts of the biosynthetic products. The ultimate destination of carbon is the pool of metabolites, which includes DNA, RNA, protein, and lipids, along with products such as carbohydrates, intracellular metabolites (e.g. NADPH), and biological byproducts like CO₂ or CH₄. Needless to say, it can quickly get complicated.

Luckily, we can learn a lot of useful information by looking simply at bulk end-products. For example, in methanogens we can examine the ¹³C content of bulk biomass, individiual lipids, and produced methane. If we know the starting isotope composition and concentration of the substrates, we can measure the difference in ¹³C content between substrate and products.

We have done work with the methanogen Methanosarcina barkeri, examining growth on four substrates: H₂/CO₂, acetate, methanol, and trimethylamine. By varying the concentrations of these substrates we have observed the how this variable affects the observed isotope fractionations. By examining a variety of biological products (biomass, methane, and lipids) we can look for patterns that let us work backwards in the environment, by examining products and inferring information about the availability of substrates.

Future plans include a more refined look at the isotopic composition of methanogen products and development of a new tracer method that will (perhaps) be able to distinguish the source of methane in geological environments.

Pathway of H₂/CO₂ methanogenesis in Methanosarcina barkeri

CO₂ is reduced to methane through a pathway involving seven steps, and the intermediate H₄MPT (tetrahydromethanopterin). Interestingly, this series of reductive steps is very similar to the steps that happen during the geochemical reduction of CO₂ to methane. Methyl-H₄MPT is the split point, where some of the carbon branches off of the methane pathway to become biomass, most of which is synthesized through acetate (leading to lipids) or pyruvate (leading to most other products).

Sulfate Reducing Bacteria are among the most important biogeochemical players in anaerobic environments. These bacteria respire by transferring electrons from a donor molecule to sulfate, thereby reducing the sulfate to sulfide. The pathway by which this happens, and how this bears on the fractionation of sulfur isotopes, has been a subject of debate for decades. This is important, because sulfur isotopes are an important tool for tracking the oxidation state of the crust through Earth's history.

Recently I have been involved in a project to try to work out the isotopic consequences of this pathway in greater detail. This work (with David Johnston and Wil Leavitt) draws upon recent crystal structures published in the biochemical literature. These suggest that the classic pathways for sulfur isotope fractionation, while generally correct, are not correct in detail. In fact, sulfate reduction is a highly branched pathway in which sulfite plays a critical intermediate role.

A branching network of reactions for dissimilatory sulfate reduction. Dotted line represents cellular boundary. Dark black arrows indicate the 'pseudo-Rees' pathway for reudction of sulfate to sulfide. Colors indicate specific reactions within the thionate loop: dark blue = trithionate formation, light blue = thiosulfate formation, green = trithionate reduction, crimson = thiosulfate reduction, grey = cellular leaks. Each of the reductions in the thionate loop regenerates sulfite.

Well I don't want to give away all my secrets! However, there are some obvious directions to take my current work...

My interests are in applying organic geochemistry to understand modern and ancient environments, and in using geochemical tools to understand biogeochemistry, evolution, ecology, and climate.

I will continue to work on genes and biomarkers. Specifically, I am interested in hopanoids, and also very interested in tetraether lipids - lipids that span the bilayer in some Archaea as well as in some Bacteria. Identifying the genes involved in synthesizing these lipids, and the physiological controls on their regulation will be critical to understanding the environmental distribution of these lipids. These lipids are used as important paleo-indicators of temperature. This may be valid in many instances, but perhaps not in others. Genes should provide the insignt into understanding this.

Many classes of lipids used as biomarkers have little-known genetic bases. In this vein, I am collaborating with Gordon Love to identify the distribution of genes involved in the synthesis of certain sterols that may mark the first appearance of animals in the geochemical record.

I also plan to continue understanding how biosynthetic pathways control the isotopic content of biological products. I am particularly interested in major isotopes (C, H, N, O, S). In lipids, this is mainly limited to the distrbution of carbon and hydrogen isotopes. The D/H ratio of lipids is very interesting, since it is probably controlled both by the source water available to an organism, and by the energy metabolism of the organism (since lipid D/H derives from the NADPH pool). I have some ideas to test controls on this in living organims... check back in a few months

Crenarchaeol. A characteristic bilayer-spanning membrane lipid of marine Thaumarchaeota, an isomer of which forms part of the basis of The TEX₈₆ temperature proxy

Lipids can be extracted from biological, geological, or environmental in a number of different ways, but all boil down to separating compounds soluble in organic solvents (like hexane, dichloromethane, or chloroform) from those soluble in water or not soluble at all. Lipids are found in the organic phase, while molecules like DNA and protein will be found in the aqueous phase. Insoluble lipids - those bound together in macromolecules like kerogen - can be further analyzed by techniques like pyrolysis.

The lipid extract contains many different compounds. To analyze these, the compounds must be separated from each other. The separation is achieved by chromatography. A crude separation can be done in the laboratory by passing the lipid extract over a column of silica gel, and separating it into fractions such as saturated hydrocarbons, aromatic hydrocarbons, alcohols, fatty acids, etc. After this crude separation, the resulting fractions can be further separated by a specialized instrument coupled to a mass-spectrometer used to detect them. There are different ways to do this, depending on the nature of the compounds being analyzed.

GCMS (gas chromatography-mass spectrometry) is a method by which compounds are separated in the gas phase on a capillary column. The capillary column is placed in a temperature-controlled oven and slowly heated, while a flow of helium passes through the column. As each compound elutes from the capillary column, it passes into the mass spectrometer, where it is ionized and fragmented into its constituent pieces. The mass of each of these pieces is detected, forming a mass spectrum that is used as a molecular fingerprint to determine the identity of the compound. GCMS is ideal for compounds that are volatile - i.e. that can be put into the gas phase at less than 300 degrees Celsius.

LCMS (liquid chromatography-mass spectrometry) separates compounds dissolved in a liquid solvent by passing them over a solid-phase column that binds to organic compounds. The organic compounds initially 'stick' to the column. As the chromatography proceeds, the polarity of the solvent is slowly changed, until compounds unstick from the column and are washed by the solvent into the mass-spectrometer. The mass spectrometer then ionizes and detects the compounds in a manner similar to GCMS. LCMS is ideal for nonvolatile compounds, including highly polar compounds such as hopanoids and Intact Polar Lipids (IPLs).

Most of my lipid analyses at Harvard are done in the Pearson Laboratory

Analyses of stable isotopes of carbon and nitrogen in bulk organic material is achieved by first combusting the biological material to CO₂ and NO_x compounds, followed by reduction of NO_x to N₂. The CO₂ and N₂ gases can then be introduced into an isotope-ratio-monitoring mass spectrometer (irMS), where the relative abundance of ¹⁵N/¹⁴N or ¹³C/¹²C can be measured. Other treatments can be used to prepare materials for analysis of their sulfur, oxygen, or hydrogen isotope ratios.

The irms functions by first ionizing the analyte gas (e.g. CO₂) and accelarating the positively-charged ions along a flight tube. The flight tube is surrounded by a magnet, which deflects the ions from their trajectory. The degree to which they are deflected depends on their mass - lighter ions (e.g. those containg ¹²C) are deflected more than heavier (e.g. those containing ¹³C). Faraday cups then collect the ions at the end of their flight path, inducing a voltage that can be measured. The relative ratio of the voltages can be compared to a standard, yielding a measurement of the ¹³C/¹²C of the sample.

We can measure the ¹³C/¹²C ratio of individual lipids by connecting a gas-chromatograph to an in-line combustion interface coupled to an irMS. This means that we can simply inject our lipid extract onto the GC. The data we get out are a series of peaks with isotope ratio information. The irMS tells you only the carbon isotope ratio of each peak - not its identity. The identity can be associated with a particular structure by comparing the GC-irMS chromatogram to our GCMS data.

Genetics is a powerful tool that can be used to query the function of lipids. By disrupting individual genes that we think are associated with lipid biosynthesis, we can examine the suite of products that occur when that gene is absent. Subsequently, we can reintroduce the disrupted gene to the cell to be sure that the gene disruption is indeed associated with the phenotype we observed.

Most of the genetics I do in the Marx lab involves the targeted deletion of individual genes. We use a system that relies on single-crossover homologous recombination, selection on antibiotics (generally ampicillin, chloramphenicol, and tetracycline), and then counterselection against sucrose using a sacB locus.

Using these techniques, we have been able to disrupt a number of the genes involved in hopanoid biosynthesis in Methylobacterium. These genes are arranged in three clusters on the genome, in the structure shown below.

Hopanoid Gene Clusters in Methylobacterium. There are three clusters. Cluster I contains the hpnF gene: squalene-hopene cyclase, which produces the precursor hopene (aka diploptene) from squalene. Cluster II contains the hpnF and hpnG genes, which are critical in side chain formation. Cluster III contains a single gene, hpnP, which is involved in placing a methyl group on the A-ring, forming a methylhopanoid.

Experimental Evolution is a powerful tool that can be used in the laboratory to gain a better understanding of the function of genes or molecules. The idea is simple: by disrupting a gene, some cellular function no longer operates. However, during many generations of evolution in the laboratory it is possible that compensatory mutations may arise in some cells. If these mutations confer a selective advantage, those cells will come to dominate the population. Essentially we "break" a bacterium... and then ask it to fix itself!

Experimental evolution works by simply growing cells in batch cultures, and performing a serial passage of cells from one flask to the next. Once a flask is fully grown, a small proportion of it is removed and serves as the inoculum for the next flask. This is repeated... and repeated... and repeated.

The prediction is that compensatory mutations will reflect the function of the genes that have been disrupted. So, for example, if hopanoids are involved in membrane permeaility, than an evolved hopanoid-free mutant might be expected to acquire mutations in functions that deal with transporting things across the membrane. Mutations can be identified by using next-generation sequencing techniques to resequence the entire genome of individual evolved isolates. The resequenced genome is compared to the ancestor, and mutations are identified. These can then be reintroduced into the ancestral strain, one-by-one, to identify those mutations which confer the greatest selective advantage.

An experimental evolution protocol. Cells are plated on an agar plate. A colony is picked and inoculated into the flask. These cells may not grow very well (e.g. they are mutants defective in some function). After each flask is grown, a small proportion is used to inoculate the subsequent flask. After many generations, cells which have acquired beneficial mutations come to dominate the population. The cells at the end of the experiment have a higher fitness than the ancestral cells. This fitness is conferred by mutations, which can be identified by sequencing the genomes of the population at the end of the experiment and comparing to the genome of the ancestral cell

Three in review, more on the way. Check back soon!