The history of research on microbial rhodopsins offers a novel perspective on the history of the molecular life sciences. Events in this history play important roles in the development of fields such as general microbiology, membrane research, bioenergetics, metagenomics and, very recently, neurobiology. New concepts, techniques, methods and fields have arisen as a result of microbial rhodopsin investigations. In addition, the history of microbial rhodopsins sheds light on the dynamic connections between basic and applied science, and hypothesis-driven and data-driven approaches. The story begins with the late nineteenth century discovery of microorganisms on salted fish and leads into ecological and taxonomical studies of halobacteria in hypersaline environments. These programmes were built on by the discovery of bacteriorhodopsin in organisms that are part of what is now known as the archaeal genus Halobacterium. The transfer of techniques from bacteriorhodopsin studies to the metagenomic discovery of proteorhodopsin in 2000 further extended the field. Microbial rhodopsins have also been used as model systems to understand membrane protein structure and function, and they have become the target of technological applications such as optogenetics and nanotechnology. Analysing the connections between these historical episodes provides a rich example of how science works over longer time periods, especially with regard to the transfer of materials, methods and concepts between different research fields.
Microbial rhodopsins: exemplary membrane proteins
For more than a century, research on microbial rhodopsins and the organisms that contain them has shed light on numerous facets of biology. Researchers have made important discoveries of diverse unanticipated organisms and physiological capacities in saline environments. In 1970, the first microbial rhodopsin, bacteriorhodopsin, was isolated from the cell membrane of Halobacterium. This finding had a major impact on protein structure and membrane research. Recently, an even broader distribution of previously unknown microbial rhodopsin genes, proteorhodopsins, has added another dimension to the ecological, genetic and physiological understanding of these exemplars of membrane proteins.
Although microbial or type I rhodopsins were first found in the membranes of organisms that are now known as Archaea, these proteins were also subsequently detected in eukaryotic microorganisms and bacteria (Spudich et al., 2000; Lanyi, 2004). They function as light-dependent transporters or photoreceptors, either through ion transport or through signal transduction. The protein is formed by a single amino-acid chain with a characteristic fold that encompasses seven transmembrane α-helices. A light-sensitive retinal cofactor is bound via a Schiff base to a conserved lysine residue of the protein. Upon absorption of a single photon, the retinal isomerizes from the all-trans to the 13-cis form, causing a cyclic sequence of spectroscopically detectable intermediates. The production of these intermediates is accompanied by conformational changes in the protein. At the end of this photocycle, the protein returns to its initial state. The mechanism of microbial rhodopsin function has many similarities to type II rhodopsins, which include the photosensitive receptor proteins of animal retinas (Vinothkumar & Henderson, 2010). However, these visual pigments function as G-protein-coupled receptors (GPCRs) and the cofactor is released from the opsin protein during the photocycle, whereas the retinal in microbial rhodopsins remains permanently bound (Terakita, 2005) [for historical accounts of the history of research on visual rhodopsins and other seven transmembrane receptors, see Costanzi et al. (2009) and Lefkowitz (2004)].
Although type I and II rhodopsins share the retinal cofactor and an architecture consisting of seven transmembrane helices, it is unclear whether the two protein families have resulted from a single progenitor or are an example of convergent evolution (Spudich et al., 2000). John Spudich and colleagues prefer the latter explanation on the basis of current genome sequence data as well as protein structures. Yet, they also point out that all known examples of type I and II rhodopsins are from evolutionarily distant organisms, and that further genome projects might reveal a missing link between the two families. Because a broad, but patchy distribution of type 1 rhodopsins has been found in Archaea, Bacteria and Fungi, Sharma et al. (2006) suggest that lateral gene transfer has been the principal force for the distribution of these proteins. In addition to questions about the evolutionary relationships between microbial and visual rhodopsins, research on these two fields has intersected at other points in the last decades, most notably in the field of structural biology, but also in studies of protein dynamics (Spudich et al., 2000; Vinothkumar & Henderson, 2010).
In the history of microbial rhodopsin research, bacteriorhodopsin plays a central role (Hampp, 2000a; Lanyi, 2004). The protein has a molecular mass of 26.8 kDa and forms two-dimensional crystalline patches in the cellular membrane of Halobacterium salinarum. Because of the conspicuous colour of the protein, these patches are called ‘purple membrane’ whether in cells or as biochemical preparations. Bacteriorhodopsin is known as a ‘light-driven proton pump’ because a proton is transported out of the cell during the protein's photocycle. The purple membrane thereby provides the organism with the means for a phototrophic lifestyle. Halorhodopsin, also discovered in H. salinarum, is a light-dependent anion importer involved in osmotic homeostasis (Kolbe et al., 2000). Microbial rhodopsins can also have sensory functions. Through protein–protein interactions, these sensory rhodopsins mediate phototactic responses in their host organisms (Spudich, 2006). Channelrhodopsins, such as those found in the green alga Chlamydomonas, are involved in the perception of light, but function by triggering passive ion fluxes across the membrane (Stehfest & Hegemann, 2010). All the evidence on the recently discovered proteorhodopsins, which were found through metagenomic analyses of bacterial DNA from microbial communities in marine environments, indicates a bioenergetic rather than a signalling function (Béjà., 2000; Martinez et al., 2007; Gómez-Consarnau et al., 2010).
Today, microbial rhodopsin research is a flourishing research field in which new understandings of rhodopsin diversity, function and evolution are contributing to broader microbiological and molecular knowledge. In addition, a variety of inventions have been devised on the basis of microbial rhodopsins – from photosensitive security cards to devices that can control specific neuronal activities. As well as being a dynamic field now, microbial ‘rhodopsinology’ has been a source of major insight in earlier membrane research, structural biology, photobiology, microbial ecology and physiology. Here, we link these many dimensions of microbial rhodopsin research together and show that the long history of the field enriches understanding of not only the development of a particular research programme but also the general history of the life sciences. Aspects of this story, especially in regard to the early research on halophilic microorganisms, are presented in Aharon Oren's detailed and multidimensional discussion of the biology of these organisms (Oren, 2002). An extensive collection of sources on early halophile research was also compiled by Helge Larsen (1922–2005), a professor of biochemistry at the university of Trondheim (Larsen, 1962). Additional insight into early research on the purple membrane and bacteriorhodopsin can be gleaned from Walther Stoeckenius's reflections on that period (Stoeckenius, 1994).
Discoveries of halobacteria in applied microbiology and microbial ecology
A crucial factor in the discovery story of halophilic microorganisms was the striking red colour that many of these organisms display in mass growth. Red waters were already reported in the Bible and some Chinese manuscripts, and the Historia Naturalis by the Roman naturalist Pliny discusses in 77–79 AD the regionally differing colours of sea salt (Pliny XXXIII: xxxi, Baas Becking, 1931). [Oren (2002) cites biblical episodes where water was turned into blood: e.g. the Nile during the first Plague of Egypt (Exodus 7: 17–25; see also 2 Kings 3: 22).]. Saltern workers in San Francisco Bay saw the red tinge as an indication of increasing salt concentration in the brine, which was then transferred to other ponds for crystallization (Tressler, 1940). The discovery of microorganisms as the causal agents of redness in brines was catalysed by the use of sea salt in the preservation of food such as fish. In the late nineteenth century, a time of industrialization and increasing global trade, as well as the development of bacteriology and infection biology, microbial analyses were established as part of food hygiene practices (Hardy, 1999). Because salted and dried codfish sometimes turned red in damp summers, the US Fish Commission commissioned an investigation from William G. Farlow (1844–1919), a professor of cryptogamic botany at Harvard. Farlow attributed the red discoloration of the fish to the growth of a ‘very minute plant’ named Clathrocystis roseo-persicina (Farlow, 1880). He also stated that the reddish cells originated from the Mediterranean salt used for preservation. To prevent contamination, which not only lowered the price of the product, but was also suspected to cause food poisoning, Farlow recommended the use of salt from less contaminated sources. Farlow's analysis and subsequent publications had legal and economic impacts. In France, a sale ban on reddened codfish was announced in 1885 (Mauriac, 1889). In the same year, a Spanish newspaper reported Farlow's findings and warned that Spanish salterns could be affected by dwindling sales to the fish-producing regions of the north Atlantic (Anonymous, 1886).
A scientific problem that continued to plague microbiological studies was the taxonomy of the halophile isolates, which were continually reclassified for many decades after their initial discovery [for a detailed account of early isolates of halophilic microorganisms and their various reclassifications, see Tindall (1992)]. Researchers working with fish or salt samples often found several types of cells, and it remained difficult to establish the identity of different isolates purely on the basis of traits such as cell shape or colour. An important step was to match organisms from fish samples with those found in salterns. A study of San Francisco Bay saltwater ponds identified bacteria that were believed to be degrading organic material at extremely high salt concentrations (Pierce, 1914). One of these bacteria, the ‘red chromogenic bacillus’, was found to have a colour similar to that of saltern brines and to smell like those brines when cultured. These cultures were also shown to thrive on salt cod.
In the early 1920s, two extensive studies reviewed the current knowledge of halophiles on salt-fish and attempted to improve their classification through more specific culturing and staining methods. Using fish from Norway that had been shipped to North German ports, the German mycologist and phytopathologist, Heinrich Klebahn (1859–1942), managed to isolate from fish a gram-negative, rod-shaped microorganism that formed reddish colonies. He named it Bacillus halobius ruber (Klebahn, 1919; for an English translation, see DasSarma et al., 2010). He noted two phenomena that would continue to be of importance throughout the history of halobacterial research. Klebahn observed that the cells swell and lyse upon transfer into water. Reflecting on the relationship between the organism and its environment, he stated that ‘the process reminds one of the behavior of certain deep sea creatures that burst due to their internal pressure, when they are taken out of the deep’ (translation by DasSarma et al., 2010, p.12). Klebahn also managed to extract the pigments responsible for the red colour and to characterize them spectroscopically.
Just two years later, another comprehensive paper on halophiles was published in Canada, which was a second hub of salt-fish production and trade. Bacteriologists Francis C. Harrison and Margaret E. Kennedy examined various fish and salt samples and isolated an organism that they named Pseudomonas salinaria (Fig. 1; Harrison & Kennedy, 1922). They described the lifecycle of this organism as ‘pleomorphic’, meaning that it possessed various cell shapes. The nomenclature was probably bestowed without knowledge of Klebahn's work, which was published in German during the period of economic and political instability after World War I, when the dissemination of science carried out in central Europe was limited.
During the interwar period, halophile research was increasingly conducted by environmentally inclined microbiologists, some of whom had strong connections to the Delft school of microbiology (Oren, 2002). L.G.M. Baas Becking, who moved from the Netherlands to the United States and became director of the Jacques Loeb Laboratory at Pacific Grove in California, studied organisms in Californian saltwater lakes (Quispel, 1998). For him, the presence of life in seemingly hostile environments had broader implications than it did for the microbiologists examining fish and salt samples. He called these niches a ‘borderland of physiological possibilities’ and thought they suggested ‘that physiology heretofore has been founded on the properties of liquids such as freshwater, seawater and blood-serum’ (Baas Becking, 1928, p. 9).
In the Netherlands, halophiles were studied by microbiologist and biochemist Albert Jan Kluyver, who was involved in the isolation of Pseudomonas beijerincki and its identification as the causative agent of spoilage in salted beans (Larsen, 1973). With help from both Kluyver and Baas Becking, Helena Petter, a PhD student in the Botanical Laboratory at the University of Utrecht, isolated bacteria from salted cod and herring as well as from sea salt samples. She named these organisms Bacterium halobium rather than Bacillus halobius because she did not observe any spore formation (as Klebahn had not earlier). In her paper, she also introduced the term ‘bacterioruberin’ for the coloured pigments, which she purified in crystalline form. These substances appeared to be similar to carotenoids, but did not match any known forms (Petter, 1931). Trijntje Hof, working in Kluyver's laboratory at the Technical University of Delft, noticed the lack of attention that had been paid to bacteria in earlier studies of saline lake ‘biocoenoses’ (ecological communities) and observed: ‘Nevertheless, there can be scarcely any doubt that bacteria will form an essential link in the cycle of life occurring in these lakes’ (Hof, 1935, p. 93). Although the Netherlands had no saltwater lakes suitable for her research, she examined materials such as salt samples, brine mud and salted fermentation products in order to compare her findings with those from lakes (Hof, 1935). These studies successfully combined microbiological methods with ecological approaches and biochemical analyses, but the taxonomic problems persisted.
The first edition of Bergey's Manual of Determinative Bacteriology (1923) and the fourth (1934) name the organism that was first described by Harrison and Kennedy as Serratia salinaria (Bergey et al., 1923; Bergey, 1934). It was assigned, presumably, for the red hue of the cultures. When chromogenic, rod-shaped halophiles were also isolated from leather, these were consequently designated as Serratia cutirubrum, meaning ‘red hide’ (Lochhead, 1934). However, using a trait such as colour as the key characteristic of a genus was subsequently questioned. It was found that not only the pigments of S. marcescens and those of other halophilic microorganisms were different but also that the physiology and the flagellar organization of these organisms were distinct (Lochhead, 1943). Because of their similarity to pseudomonads, the two halophile species reappeared in Bergey's 1948 edition as P. salinaria or cutirubrum, which was the nomenclature proposed originally by Harrison and Kennedy (Burkholder, 1948; Larsen, 1973).
A solution to this taxonomic muddle was provided by the Israeli microbiologist, Benjamin Elazari-Volcani (1915–1999), who had also been a visitor to Kluyver's laboratory at Delft (Larsen, 1973). Elazari-Volcani's first paper on halophiles dates back to 1936, when he published a short note, ‘Life in the Dead Sea’, in Nature, authored by his birth name Benjamin Wilkansky. He isolated several types of aquatic microorganisms, some of which were orange-red when cultured (Wilkansky, 1936). In a 1944 article, he describes several obligately halophile bacteria and suggests Halobacterium as a subgenus of Flavobacterium on the basis of the similarity of his isolates to those of Peirce, Klebahn, Harrison and Kennedy, Petter and others (Volcani, 1944). Elazari-Volcani (1957) contributed Halobacterium as the genus name to Bergey's Manual [the name Halobacterium had already been introduced in the 1930s for halophilic rods, but this original classification did not catch on (Tindall, 1992)].
The reformed genus encompassed microorganisms from natural brines as well as those from salt-fish and other salted goods. It worked as a unifying category for these apparently related isolates (Elazari-Volcani, 1957; Oren & Ventosa, 1999). The type species H. salinarium (later salinarum) was characterized as a gram-negative, rod-shaped bacterium. Today, former species such as Halobacterium cutirubrum or Halobacterium trapanicum, originally isolated from the salterns of Trapani in Sicily, are all considered to be different strains of H. salinarum (Oren, 2002). For Elazari-Volcani, as for Klebahn and Petter, and indeed, for the existing history of halophile research, the unusual pigmentation of the microorganisms was crucial to the observation, identification and biochemical analyses of the organisms. However, using the colour as a characteristic caused taxonomic problems when this highly visible trait had to be evaluated against microscopical or biochemical distinctions. The physiological function of the pigments remained unclear.
The first microbial rhodopsin: bacteriorhodopsin, membranes and bioenergetics
In the late 1960s, biological membrane research was an experimentally and conceptually heterogeneous territory. Electron micrographs showed cell membranes as dark shaded lines that could be distinctively stained by heavy metal salts. It could not, however, be determined unambiguously how the lipids and proteins found in membrane material from cell fractionations were spatially organized (Stoeckenius & Engelman, 1969; Bechtel, 2006). Moreover, the phenomenon of the active transport of, for example, ions or sugars across lipid barriers, was difficult to address with techniques such as radioactive transport assays for intact cells. Little was known about the material structures responsible for the transport process (Pardee, 1968). In addition, the membranes of chloroplasts and mitochondria formed the battleground for the controversy over oxidative phosphorylation and Peter Mitchell's chemiosmotic theory. According to the latter, membranes maintained a proton gradient that was required for ATP synthesis, whereas competing ‘chemical theories’ argued for the existence of a phosphorylated high-energy intermediate (Mitchell, 1961; Weber, 2002; Morange, 2007). Differing models of membrane organization in biochemistry textbooks of this era clearly illustrate the lack of consensus among scientists on these topics (Fig. 2).
Biochemical studies of membranes were complicated by the difficulties in obtaining membrane material in sufficient amounts and by the problems of separating protein and lipid components. These factors may explain why Walther Stoeckenius, an electron microscopist at Rockefeller University who was studying myelin organization as well as erythrocytes and mitochondrial membranes, decided to work on halobacterial membranes in the mid-1960s. The publication that initially stimulated his interest reported a disintegration of the halobacterial cell envelope fractions into distinct molecular components when they were transferred into media of low osmotic strength (Stoeckenius, 1994). These data suggested a subunit model of the membrane as a planar array of lipoprotein particles. This was in contrast to the then prevalent Davson–Danielli model, which implied a lipid bilayer with protein adjacent to both sides (Fig. 2). However, the ‘auto-fractionation’ of halobacterial membranes in distilled water also provided a technical advantage as it considerably facilitated analyses of the different fractionation products (Henderson, 1977; Oren, 2002). In addition, the fractionation products obtained from halobacterial membranes could easily be distinguished by their characteristic colours (Stoeckenius & Rowen, 1967).
Stoeckenius started to work on halobacterial membranes while still at Rockefeller. His project continued at the University of California Medical School at San Francisco, where he moved in 1969 with two new co-workers, Allen Blaurock and Dieter Oesterhelt. Oesterhelt, a trained chemist, was on sabbatical leave from Munich, where he had earned a doctorate in 1967 under the supervision of Nobel laureate biochemist, Feodor Lynen. Blaurock, a physicist, had worked with X-ray techniques in Maurice Wilkin's lab at King's College London. When Blaurock and Oesterhelt examined a curiously purple-coloured membrane fraction by X-ray diffraction, they found a sharp diffraction pattern that indicated a high degree of molecular order (Stoeckenius, 1994). In addition, Oesterhelt observed that the purple material turned yellow upon the addition of organic solvents such as ether (Stoeckenius, 1994; Fig. 3). When Blaurock told him about a similar colour change in the frog retina he was working with at King's, Oesterhelt was inspired to look for the chemical substance that formed the visual pigment in the eyes of higher animals: vitamin A aldehyde or retinal (D. Oesterhelt, pers. commun.).
Because retinal was only known to occur in visual rhodopsins, Stoeckenius considered this a ‘wild idea’ (Stoeckenius, 1994). To generate evidence in its support, Oesterhelt used methods established in rhodopsin research. First, Blaurock told him about a detergent used in the solubilization of rhodopsin (Stoeckenius, 1994). In addition, biochemical studies of vitamins and vision by scientists such as the Swiss chemist, Paul Karrer, and the American physiologist, George Wald, had resulted in an entire body of techniques to detect retinal and its mode of binding to proteins (Wald, 1968). Oesterhelt imported these methods of organic chemistry into his Halobacterium project and thus provided several lines of evidence for retinal being bound to the protein of the purple membrane (Oesterhelt & Stoeckenius, 1971). Because of this cofactor, Oesterhelt and Stoeckenius named the protein ‘bacteriorhodopsin’ in their 1971 paper (Oesterhelt & Stoeckenius, 1971). In the same issue of Nature New Biology, the Stoeckenius group presented X-ray diffraction and electron microscopy data that suggested a hexagonal arrangement of the purple membrane protein and indicated that bacteriorhodopsin would form two-dimensional crystals in the purple membrane of halobacteria (Blaurock & Stoeckenius, 1971).
On the basis of the similarities to visual pigments, the authors of the two 1971 papers speculated about a function of bacteriorhodopsin as a photoreceptors but did not present any physiological data. In his 1994 recollection of the period, Stoeckenius reported phototactic experiments on Halobacterium that were stimulated by a talk Max Delbrück gave in 1969 on the photoresponses of the fungus Phycomyces (Stoeckenius, 1994). Yet, he also stated that a connection between these experiments and the membrane project was not made immediately. Oesterhelt also conducted similar experiments after his return to Munich in the spring of 1971, and his laboratory notebooks document that phototaxis was observed (pers. commun.). Neither of these results were published at the time. Within a few years, most notably after data on the physiological function of the purple membrane had been added, the number of publications on the subject increased rapidly (Fig. 4a and b). An entire research programme arose at the intersection of structural biology and biochemical studies in less than a decade. In retrospect, the multidisciplinary configuration of a physicist (Blaurock), a (bio)chemist (Oesterhelt) and a membrane researcher (Stoeckenius) could well have generated an enriched and supportive context for the discovery of bacteriorhodopsin.
The purple membrane was soon revealed to function as a light-dependent proton exporter (Oesterhelt & Stoeckenius, 1973). A structural model of the protein was established by Richard Henderson and Nigel Unwin from the Medical Research Council Laboratory of Molecular Biology (LMB) in Cambridge. Henderson's and Unwin's paper depicted the α-helices of bacteriorhodopsin as parallel graded columns perpendicular to the membrane plane (Fig. 5). This publication is the most extensively cited of all papers on bacteriorhodopsin or the purple membrane from 1970 to 1980. This high rate of citation might indicate the more general relevance of this particular publication to membrane and structural research [The Science Citation Index shows 1591 citations for this paper as of February 2011. The next closest paper is Oesterhelt & Stoeckenius's (1971) paper with 1277 citations.]. Henderson and Unwin qualify the purple membrane as ‘a simple example of an ‘intrinsic’ membrane protein, a class of structures to which many molecular pumps and channels must belong’ (Henderson & Unwin, 1975, p.31). Transport was thought to result from quasi-mechanical conformational changes of the polypeptide chain. Bacteriorhodopsin also played a major role in the resolution of the controversy over oxidative phosphorylation, when it was shown that proteoliposomes containing it (in combination with the mitochondrial FOF1-ATPase) synthesized ATP upon illumination. (Racker & Stoeckenius, 1974). Because a transmembrane proton gradient alone sufficed for ATP synthesis, this experiment was interpreted in favour of Peter Mitchell's chemiosmotic mechanism of energy generation (Weber, 2002).
Bacteriorhodopsin thereby became an extraordinarily successful model system for membrane transport proteins. An important factor in this development can be found in the properties of halobacterial cells and the purple membrane fraction. Before the advent of heterologous expression or protein affinity chromatography techniques, the purple membrane provided a highly tractable material for membrane studies. It could be prepared easily and in fairly large amounts, and it contained only one protein in a crystalline arrangement. In addition, the material was robust and able to resist electron radiation, detergents and removal from its native environment better than many other membrane proteins.
The rest of the 1970s and 1980s produced more discoveries of microbial rhodopsins. One of them was halorhodopsin, which was initially characterized by physiological techniques using intact cells in order to measure ion fluxes and membrane potentials. This method of characterization led to the assumption that it was a sodium importer (see e.g. Matsuno-Yagi & Mukohata, 1977; Mukohata & Kaji, 1981). Only later was it shown through photobleaching and reconstitution with retinal in a bacteriorhodopsin-deficient strain that a rhodopsin was responsible for the electrical potential, and the corresponding protein was finally isolated (Lanyi & Weber, 1980; Lanyi & Oesterhelt, 1982). Biophysicist Janos Lanyi and his group at the University of California at Irvine then showed that the protein actually exported anions, such as chloride, which result in similar electrical effects (Schobert & Lanyi, 1982). The phototactic responses of halobacteria were also associated with retinal proteins. After first being called ‘slow cycling rhodopsin’, the phototactic receptor was relabelled as ‘sensory rhodopsin I’ (Hildebrand & Dencher, 1975; Bogomolni & Spudich, 1982; Spudich & Bogomolni, 1984). Similarly, the previously named ‘phoborhodopsin’, linked to a photorepellent response, was relabelled ‘sensory rhodopsin II’ (Tomioka et al., 1986; Marwan & Oesterhelt, 1987; Spudich & Bogomolni, 1988).
Halobacterium had become a well-established laboratory organism for membrane research by the end of the 1970s. Research was carried out in numerous laboratories engaged in microbial physiology, biochemistry and structural biology. The Federal Republic of Germany, the United Kingdom and the United States were major research hubs, as well as Israel and the Soviet Union. There, bioenergetic and protein chemistry studies of bacteriorhodopsin were carried out at the Moscow State University and the Shemyakin Institute of Bioorganic Chemistry (Drachev et al., 1974; Ovchinnikov et al., 1979). The expansion of microbial rhodopsin research influenced general halophile research concerned with taxonomy, physiology and genetics, while the structural studies of bacteriorhodopsin influenced research on GPCRs and eukaryotic rhodopsins (Caplan & Ginzburg, 1978; Ovchinnikov, 1982).
The genetics of microbial rhodopsins and halobacteria
Surprisingly, the entire field of bacteriorhodopsin research managed without any genetic techniques or concepts until the 1980s. Analyses of halobacterial DNA had been carried out in the 1960s (e.g. Moore & McCarthy, 1969), and halobacterial ribosomes had been studied by Donn J. Kushner and colleagues at the Division of Biosciences of the National Research Council (NRC) in Ottawa (e.g. Bayley & Kushner, 1964; Oren, 2002) [due to their stability, the ribosomes of halophiles such as Halobacterium marismortui (now Haloarcula marismortui, see Oren et al., 1990) have played an important role in determining ribosomal structure (e.g. Yonath, 2009)]. These publications did not, however, considerably influence bacteriorhodopsin research. The field developed separately from classical molecular biology and the early years of genetic engineering. Microbial genetics had already been ‘molecularized’ due to the discovery of the double helix structure of DNA in 1953 and by insights into transcription and translation, as well as the emergence of recombinant DNA in the early 1970s (Morange, 1998). Protein science had also been influenced by structural methods and concepts such as protein crystallography, Pauling's α-helix and the structures of myoglobin and haemoglobin from Cambridge's LMB (de Chadarevian, 2002). Membrane research (as well as many other fields of the life sciences) adopted a different path of molecularization than bacterial genetics and structural biology. In the case of early bacteriorhodopsin science, the entire research programme – including culturing Halobacterium, preparing the purple membrane, and analysing the structural, biochemical and physiological properties of the protein – was carried out as if Halobacterium were an organism without genes and genetic apparatus.
Several factors could be suspected to lie behind the slow import of genetics into membrane research. First of all, conditions such as the high osmotic strength of the media or the lack of resistance to common antibiotics must have impaired the adaptation of molecular genetic techniques. Yet, in order to understand why molecular research on halophiles could flourish without these technologies, another characteristic of these organisms needs to be taken into account, namely the fact that Halobacterium is extremely phenotypically variable. Colonies change their appearance from different shades of pink, red or orange to white, or from opaque to translucent, depending on conditions such as light or salt concentration, as well as ‘spontaneously’ (Larsen, 1973; Oren, 2002). In contrast to the problems of taxonomy, the variability of Halobacterium had some advantages for molecular studies. Although it was difficult to isolate, standardize, stock and distribute strains, variants of morphological or physiological traits could be easily obtained by plating or subjecting cultures to freeze–thaw cycles. The strain Halobacterium halobium R1, which was the basis of the isolation of bacteriorhodopsin, was a spontaneous mutant. Stoeckenius's team had initially received a strain of H. halobium from the NRC in Ottawa, where the physiology of the organism had been studied for a long time by Norman Gibbons and colleagues from the Division of Applied Biology (Murray, 1978). From this stock, the Stoeckenius group simply picked colonies of a deeper red colour and translucent appearance (caused by the lack of gas vacuoles) and used them for their membrane project (Stoeckenius & Kunau, 1968). Other strains that differed in their amount of bacteriorhodopsin or their stability were then isolated from H. halobium R1 by chemical mutagenesis and screenings (Stoeckenius et al., 1979). All the bacteriorhodopsin research on these strains could be carried out on the basis of phenotypic traits, without any specification of the genotype or the genetic maps.
The success of the field and the wealth of valuable data generated show that genetics was not needed to study the biochemistry and biophysics of bacteriorhodopsin. Carefully selected, maintained and bred strains produced sufficient amounts of the purple membrane. The lack of gas vacuoles facilitated biochemical purification. Generally speaking, this episode of membrane research could be called ‘molecular biology without genetics’. By this, we mean that knowledge about the actual genetics of the organism was not strictly necessary for its analysis in terms of molecules. However, although bacteriorhodopsin research initially developed more or less independently of and in parallel to molecular genetics, the introduction of genetic techniques in the 1980s changed the field's path of research significantly.
In 1979, the isolation of a large plasmid from H. halobium was reported (Weidinger et al., 1979). Well-known phenotypic variations, such as the presence of gas vacuoles or the purple membrane, were mapped onto insertions, rearrangements or deletions of plasmid DNA (Pfeifer et al., 1981). For the first time, a molecular genetic explanation was given for the variability of Halobacterium. Genetic investigation was expanded by H. Gobind Khorana's group at MIT, which had begun to focus on bacteriorhodopsin. The phenotypically detected inactivation of bacteriorhodopsin synthesis was correlated with a transposable element inserted at a specific site in the genome (Simsek et al., 1982). This discovery was based on the finding by Khorana's group of a specific gene coding for bacteriorhodopsin (Dunn et al., 1981). Fundamental to this study, which involved cDNA techniques, was the determination of the amino-acid sequence of bacteriorhodopsin. This project was carried out both by Khorana's group and by a team led by the Soviet biochemist, Yuri Ovchinnikov (Khorana et al., 1979; Ovchinnikov et al., 1979). These early genetic studies were conducted within another period of major molecularly driven change in Halobacterium research. In 1980, Carl Woese and colleagues assigned the genus to the kingdom of archaebacteria, which they had distinguished from eubacteria through sequence homology analyses of rRNA (Fox et al., 1980). Subsequent phylogenetic studies of Halobacterium produced more detailed evolutionary understanding of the biology of these organisms (Sapienza & Doolittle, 1982).
By the mid-1980s, research on microbial rhodopsins had undergone many important changes and these proteins were increasingly examined using the tools of molecular genetics. A functional characterization of bacteriorhodopsin by mutagenesis, for example, was among the most important achievements expected from the fusion of microbial rhodopsin research and molecular genetics (Lo et al., 1984). However, Halobacterium genetics was still lacking some of the tools used in organisms such as Escherichia coli. Because vectors such as phages or conjugative plasmids were not available for Halobacterium, it was impossible to introduce DNA fragments in the form of mutated alleles of rhodopsin genes into the organism's genome. Any mutation studies had to be conducted heterologously in E. coli. The development of shuttle vectors for species of Halobacterium in the late 1980s provided a means for this kind of manipulation (Lam & Doolittle, 1989).
Tracing the history of microbial rhodopsin research clearly sheds light on the contributions of very different fields to membrane research. Whereas bacteriorhodopsin research had begun with studies of the protein in its crystalline state and was followed only much later by studies of the genetic organization of the organism, other forms of membrane transport research developed differently. The maltose transporter and the lactose and histidine permeases, for example, were discovered in model genetic organisms such as E. coli or Salmonella typhimurium in the 1950s. Consequently, these membrane transporters were known first by their genetic localization and organization on the chromosome. They were then characterized by screening for anomalous phenotypes such as transport deficiency, deregulation or the uptake of other substrates (Schwartz, 1987; Guan & Kaback, 2006). The actual proteins responsible for physiological functions were isolated and characterized only much later, some not until the 1990s. Crystal structures are only just being published (e.g. Oldham et al., 2007). These parallel tracks of research in membrane transport generally and bacteriorhodopsin specifically began to merge in the 1990s with the advent of more broadly usable recombinant DNA tools, as well as affinity chromatographies and biophysical imaging techniques [other membrane biology techniques, such as SDS gel electrophoresis, were also crucial (Vinothkumar & Henderson, 2010)].
Proteorhodopsin, metagenomics and high-throughput discovery
In 2000, another microbial rhodopsin was unexpectedly discovered. This time, it was not found through protein research or genetic characterization, but through the emerging tools of metagenomics. Metagenomics was initially defined as the culture-independent study of community genomes in natural environments in the 1990s (Handelsman et al., 1998; Handelsman, 2004; Committee on Metagenomics, 2007). These days, metagenomics is often ‘targeted’ to specific genes rather than whole genomes, and DNA can sometimes be from cultured communities (Cuvelier et al., 2010; Svraka et al., 2010). However, importantly, the field is not focused on single organisms, but on environmental genes, and the aim of such studies is to gain a better understanding of microorganismal biodiversity, both genetically and functionally.
Proteorhodopsin genes were first discovered in 2000 in an uncultured marine gammaproteobacterium group, called SAR86, that was part of a community sample drawn from the waters of Monterey Bay, California (Béjà., 2000). These organisms were considered to be exclusively chemoorganotrophs, because phototrophy without chlorophyll was not suspected to exist in oceanic microorganisms. In this community genome, however, was a gene that had sequence homology to bacteriorhodopsin. The ORF was found in a 130 kb stretch of environmental DNA of bacterial origin from which the flanking regions of an identified rRNA gene were sequenced. The inferred amino acid sequence of the flanking nucleotides was recognized as similar in structure to those of bacteriorhodopsin (Béjà., 2000, 2001).
To see whether the function was also similar to known microbial rhodopsins, copies of the proteorhodopsin sequence were inserted into retinal-enhanced laboratory E. coli in the hope that proteorhodopsin would be expressed and light-driven proton translocation could be observed. The experiment succeeded, to the extent that it launched a whole new field. The appropriate pigmentation was observed in the host cells, and proton translocation was measured by pH changes in light-exposed cells (Béjà., 2000, 2001). These experiments mimic and reproduce the bacteriorhodopsin studies carried out in the early 1970s that established the protein's function as a proton exporter (e.g. Oesterhelt & Stoeckenius, 1973). The kinetics of proteorhodopsin's conformational shifts were measured by absorption changes in membranes after light exposure, indicating that proteorhodopsin was a proton pump and not a sensory rhodopsin. Again, the reasoning behind this extrapolation was based on data from earlier rhodopsin research, in which sensory rhodopsins displayed much slower photocycles than those involved in energy conversion. The proteorhodopsin experiments were first carried out in vitro in transgenic E. coli and later in membranes prepared from bacterioplankton collected from Monterey Bay surface waters (Béjà., 2000, 2001). The latter approach is of particular importance because it transforms metagenomic strategies of studying entire communities of organisms into a biochemical experimental system.
Scientists who participated in the proteorhodopsin discovery describe proteorhodopsin as a fortunate discovery of metagenomic analysis, made doubly lucky because unlike the bacteriorhodopsin gene, which did not express perfectly in E. coli, proteorhodopsin did. This was just as well because the native organisms were unculturable (DeLong, 2005; DeLong & Béjà 2010). One of the key marine microbiologists involved, Oded Béjà, describes the initiation of PR inquiry as ‘flying blind’ and a matter of luck (Sreenivasan, 2001). According to Béjà, marine DNA was sequenced simply because it was there in the environmental sample, and the putative rhodopsin-encoding sequence was analysed in the hope that it would be similar to bacteriorhodopsin. This casual description of the research does not take into account, however, the fact that the marine genome survey was undertaken very methodically, with a highly systematic approach aimed at linking functional genes to specific groups of organisms. And it is also clear that the systematic transfer of techniques from bacteriorhodopsin drove the field forward the minute the discovery of the new microbial rhodopsin was made. The pattern of inquiry was tightly structured by following the system of investigation carried out on bacteriorhodopsin in the 1970s. Using these studies as protocols was not only a practical way of proceeding, but a strategic one. Exclusively genetic claims about proteorhodopsin would not have had the same effect of creating a research programme encompassing protein and membrane biochemistry.
This highly organized import of another research system led to a massive knowledge base being built up rapidly about the protein, its genes and their contributions to communities of organisms. Proteorhodopsin genes were found to be diverse and extensively distributed, not only in multitudes of marine bacteria, but even in freshwater microorganisms (Venter et al., 2004; Rusch et al., 2007; Sharma et al., 2008). The discovery led to novel understandings of survival strategies and the role of phototrophy in biogeochemical cycles (de la Torre et al., 2003; Fuhrman et al., 2008). Extensive insight was gained into the evolutionary relationships between different photosystems (Frigaard et al., 2006; Sharma et al., 2006; Spudich, 2006; McCarren & DeLong, 2007). Biogeographical analyses have shown important differences in proteorhodopsins in different geographical and ecological contexts (Papke et al., 2003; Sabehi et al., 2003; Pašić., 2005). Proteorhodopsin studies have, in fact, radically revised previous knowledge of the nature and prevalence of light utilization in ocean waters (DeLong, 2005; Martinez et al., 2007), even though a great deal remains to be understood about the physiological functions and fitness benefits of these proteins (Giovannoni et al., 2005; Gómez-Consarnau et al., 2007, 2010; Stingl et al., 2007; Walter et al., 2007; Fuhrman et al., 2008). In fact, the struggle to establish the physiological role of proteorhodopsins in their native organisms was because the only culturable organisms bearing proteorhodopsin are still not genetically manipulable (DeLong & Béjà 2010) – a situation similar to the one that occurred in bacteriorhodopsin studies.
The proteorhodopsin programme has thus become what some commentators call the ‘ecological poster child of metagenomics success’ (Kowalchuk et al., 2007, p. 479), due to its demonstration that so-called data-driven approaches (i.e. metagenomics) could be integrated with more traditional experimental strategies. Not only does the proteorhodopsin story illustrate this process of integration, but it shows how such fusions produce new knowledge and an innovative research programme.
Microbial rhodopsins as models and technologies
Since the 1970s, further methods and research technologies have been pioneered in microbial rhodopsin research. The photocycle and function of bacteriorhodopsin have been studied (e.g. Oesterhelt & Stoeckenius, 1973) and the protein was among the first membrane proteins to be reconstituted and made to function in a chimeric arrangement with another protein (Racker & Stoeckenius, 1974). The dynamics of the protein were studied using spectroscopical methods, and different intermediates of the photocycle were identified after cryo-trapping or in real time (e.g. Stoeckenius et al., 1979, Lanyi, 2004). These techniques had a higher time resolution than methods such as the stopped-flow assays used for kinetic studies of other enzymes and transporters.
An early view of bacteriorhodopsin structure was gained by a novel method of low-dose electron microscopy, which used unstained 2D-crystalline specimens (Henderson & Unwin, 1975). In 1990, a model of the protein from near-atomic resolution diffraction data at 3.5 Å was obtained by high-resolution electron-cryomicroscopy (Henderson et al., 1990). However, bacteriorhodopsin proved to be somewhat more recalcitrant with regard to X-ray analysis. The X-ray structure of the photosynthetic reaction centre provided the first atomic resolution structure of an integral membrane protein in 1985, earning a Nobel prize in chemistry for Johann Deisenhofer, Robert Huber and Hartmut Michel (Deisenhofer et al., 1985). As Deisenhofer and Michel report in their Nobel lecture, earlier crystallization studies of bacteriorhodopsin provided a background for this achievement (Deisenhofer & Michel, 1988; e.g. Michel et al., 1980; Michel & Oesterhelt, 1980). X-ray structures of bacteriorhodopsin followed only much later, increasing the resolution down to 1.55 Å towards the end of the 1990s (Belrhali et al., 1999; Luecke et al., 1999). Around that time, another crystallographic study of the purple membrane provided insights into the stabilization of the bacteriorhodopsin–lipid complex (Essen et al., 1998).
By 2009, the NCBI Protein Data Bank had more than 60 structural datasets on bacteriorhodopsin from electron microscopy, X-ray crystallography and NMR (Vinothkumar & Henderson, 2010). The robustness of microbial rhodopsins and the availability of structural data have enabled microbial rhodopsins to be used as model systems for dynamic approaches to protein structure that include molecular modelling or spin-labelling spectroscopy (Altenbach et al., 1990; Kühlbrandt, 2000).
Efforts to use microbial rhodopsins in applied technologies have also been developed during the last 30 years of research. Between 1980 and 2000, at least 60 Japanese, 20 US and 10 European patents with a main claim related to bacteriorhodopsin were filed (Hampp, 2000a). Technological applications of bacteriorhodopsin or the purple membrane can be classified according to three functional characteristics of the structure. These are the photoelectric function of the purple membrane, its photochemical ability to translocate protons, and its photochromic properties (Oesterhelt & Hampp, 2004). The capacity for oriented layers of purple membrane to create light-dependent voltage has been utilized in photoelectrical detection devices such as artificial retinas (Chen & Birge, 1993). The photochemical capacity of the purple membrane to translocate protons has been included in designs for the desalinization of sea water via engineered membranes or the direct generation of an electric current (Oesterhelt, 1976; Eisenbach et al., 1977). Such devices would require very large, oriented unimolecular sheets of the purple membrane, and these have not yet been constructed (Hampp, 2000a). Recently, microbial rhodopsins have also been suggested as bases for the in vivo and in vitro synthesis of biofuels by engineered microorganisms or membranes (Walter et al., 2010). Many applications of the purple membrane are based on its photochromic properties, such as the films developed by Soviet researchers in the 1970s and 1980s (Vsevolodov, 1998). In the 1990s, devices were made that used purple membrane for optical technologies, such as holographic recording, and for information-storage media (Hampp, 2000b). More recently, patches of mutated bacteriorhodopsin that change their colour upon illumination and display a photocycle on a timescale of minutes have been tested as security pigments in ID cards (Fig. 6; Hampp, 2006).
An important characteristic for all of these applications is the robustness of purple membrane, which is unusual for biological substances (Oesterhelt & Hampp, 2004). Most applications require large amounts (several grams) of purple membrane, which makes techniques to scale up culturing and biochemical preparation prerequisites of profitable production (Oesterhelt & Hampp, 2004). In Germany, a biotechnological company has been created to establish the commercially viable production of purple membrane variants (Hampp, 2006). Many of the technologies mentioned above require genetically engineered variants of bacteriorhodopsin with modified photocycles, and this explains why many publications and patents date from the early 1990s, when genetic tools such as shuttle vectors made it possible to synthesize recombinant bacteriorhodopsin in large amounts. Hampp (2000a) speaks of a ‘second evolution’ of purple membrane through genetic engineering, which enhances desirable properties without losing the naturally evolved rapid light response or the robustness of the protein. However, despite ongoing interest, there is no publicly accessible information showing that any of these technologies has been successfully commercialized yet.
Another strategy, using recombinant DNA technology to make microbial rhodopsins function in other organisms, has recently become the focus of a dynamic field of application (Scanziani & Hausser, 2009; Szobota & Isacoff, 2010). In technologies now called ‘optogenetics’, neurons of various animals are transfected with microbial rhodopsin genes by viral vectors. The proteins expressed and transferred into the membranes of these cells can then be used as light-activated ‘switches’ to trigger or suppress action potentials. Optogenetic technologies mostly use the chloride-importer halorhodopsin of Natronomonas pharaonis as an inhibitor of action potentials (Han & Boyden, 2007; Zhang et al., 2007), or channelrhodopsin-2 of the alga, Chlamydomonas reinhardtii, to trigger neuronal responses (Nagel et al., 2003). Channelrhodopsins, as mentioned earlier, are a member of a family of rhodopsins found in eukaryotic microorganisms (Spudich et al., 2000; Stehfest & Hegemann, 2010). They have been studied in Chlamydomonas since the 1980s and shown to be inward-directed proton and cation channels rather than transporters (Harz & Hegemann, 1991; Spudich et al., 2000).
Optogenetic tools are used to stimulate cell populations or regions of organs such as the brain, in the same way that was achieved earlier by electrophysiology (Scanziani & Hausser, 2009). Microbial rhodopsin tools are non-invasive, however, and do not require any pharmacological agents or chemicals. Moreover, their photostimulation of neurons has a high spatial and temporal resolution (Szobota & Isacoff, 2010). By implanting optical devices into the brain of laboratory animals, rhodopsin responses can be used to influence behavioural traits such as locomotion, circadian rhythms or learning (Szobota & Isacoff, 2010). Rhodopsin-mediated photostimulation of pacemaker cardiomyocytes in embryonal zebrafish hearts has been used to modulate the rhythmicity of cardiac contractions (Arrenberg et al., 2010). The creation of chimeric proteins such as light-gated GPCRs is also feasible, and this further expands the spectrum of biological functions that can be manipulated (Szobota & Isacoff, 2010). Optogenetic technology is also envisaged for biomedical therapies of neurological diseases (e.g. Gradinaru et al., 2009), but the problems of viral transfection remain unsolved. The extensive ethical issues connected to such applications are obvious (e.g. the use of viral vectors, the manipulation of human subjects) and will need close attention.
Overall, these applied technologies are generating biotechnological strategies that open up novel realms of application even as they generate further basic insights into biological processes. The capacity for fields to shift between applied and basic research and back again calls into question standard understandings of the relation between science and technology and suggests that a better conceptualization of their dynamic interaction is needed. Technological applications have also been the focus of public perceptions of microbial rhodopsin research over the last three decades. The New York Times, for example, speculated in 1977 that the purple membrane ‘may yield a scientific goldmine’ due to its possible uses in desalinization and solar energy generation (Brody, 1977). Many other articles from popular science magazines, journals or newspapers adopt a similar angle, highlighting in particular the potential of bacteriorhodopsin as an optical data storage medium (e.g. Anonymous, 1992; Sperlich, 1993). Recently, optogenetics has attracted a great deal of similar media attention (e.g. Chen, 2007).
The history of research on halophiles and the photosensitive proteins of their membranes spans more than a century of the life sciences, from the first microbiological examinations of salt-fish to recent developments in nanotechnology and the neurosciences. Our study highlights the contributions of research fields such as general microbiology, the physiology of vision, membrane biology and metagenomics, and emphasizes the contribution of microbial rhodopsin research to the molecular life sciences. The twentieth century should thus be seen as not just the ‘century of the gene’, which began with the rediscovery of Mendel's laws and led to the determination of DNA structure and the rise of genetic engineering (Fox Keller, 2002), but as a century in which several major strands of molecular research in biology were woven together. Over the last hundred years, numerous fields, including general microbiology and membrane research, have successfully contributed to the molecular life sciences. The example of optogenetics shows how different approaches can merge into an integrated and increasingly technological mode of treating life, but still increase knowledge about its basic processes.
The history of rhodopsin research also reveals that productive interactions between different fields of science occur not only through the adaptation of theories, concepts or models, but very much at the level of materials or experimental systems. In the microbial rhodopsin case, entire regimes of experimentation were repeatedly transferred between research fields. In the early days of purple membrane research, for example, an established body of chemical techniques from vitamin and visual rhodopsin research was imported to detect the retinal cofactor. When genetics was imported into microbial rhodopsin research in the 1980s, techniques and expectations were not imposed piecemeal or occasionally, but as a system of inquiry focused on the genetic material of the phenomenon. And in the birth of proteorhodopsin research, the techniques so successful in bacteriorhodopsin research were deliberately and systematically imported into the investigation of proteorhodopsin structure and function.
The history of microbial rhodopsin also calls into question distinctions commonly made between basic and applied science, or hypothesis-driven and data-driven science. Microbial rhodopsin research shows how scientific fields can originate in highly applied contexts from which questions and findings are ‘translated’ into more traditional laboratory-based and natural history activities. As bioenergeticist Efraim Racker (1913–1991) said:
‘If there is no money available for fundamental research start working on a project of applied research. If you proceed logically, you will soon be doing basic research’ (Racker, 1976, p. 7).
Our history also illuminates how techniques are transferred from one research context to very different settings, and that in this dynamic and often iterative process the originating field can benefit. The proteorhodopsin research story can be interpreted as a challenge to the common notion that science advances exclusively by testing hypotheses (Allen, 2001). The discovery of microbial rhodopsins in proteobacteria was not made on the basis of a separation between hypothesis testing and data gathering, but through the integrative combination of approaches that is similar to the process said to be essential for successful systems biology (Kell & Oliver, 2004).
A final consideration is what the history of microbial rhodopsin research might have to say for the relationship between the history of science and science itself. First, it needs to be borne in mind that the history of science can be written in many ways. Although we have largely left out the broader cultural and social framework of science (e.g. the creation or modification of institutions, and the interplay between overarching technological developments and expectations), this is not to say that these factors did not play significant roles in rhodopsin research. Discussing them would require a different paper, however. Our account, which focuses on materials and experimental systems, has depicted an array of highly creative and inventive scientific activities. New fields of research were opened up when researchers pursued unusual phenomena, such as the colour of brines and salt-fish, or the colour changes of the purple membrane. Curiosity, intuition and the ability to connect very different pieces of knowledge have played indisputably major roles in this process of discovery. Individual scientific projects did not follow prescribed routes of inquiry with specific goals. From a longer historical view, halophile and microbial rhodopsin research illustrate the adventitious and even meandering road to findings of great impact. In an era of focused funding and narrow top-down research agendas, the microbial rhodopsin story will be worth recalling as an example of how science proceeds from one influential finding to another, and in the process, transforms all it illuminates.
We thank Norbert Hampp (Marburg), Peter Hegemann (Berlin), Richard Henderson (Cambridge), Janos Lanyi (Irvine) and Dieter Oesterhelt (Martinsried) for discussions on the subject. Norbert Hampp, Richard Henderson, Dieter Oesterhelt and W.H. Freeman & Co/Worth Publishers (New York) gave permission to reproduce the figures. We are grateful to our referees and the Editor, who helped us clarify several important points in our discussion. Our research was funded by the ESRC Centre for Genomics in Society (Egenis) at the University of Exeter, UK, and the Max Planck Institute for the History of Science, Berlin. M.A.O.'s contribution was completed with Australian Research Council Future Fellowship funding at the University of Sydney.