Bacteria - Sensing the Environment

Getting about in the world

Bacteria have a wide variety of means of locomotion, depending on the type. Some bacteria are non-motile, relying entirely upon passive flotation and Brownian motion for dispersal. However, most are motile; at least during some stage of their lifecycle. Motile bacteria move with "intent", gathering in regions which are hot or cold, light or dark, or of favourable chemical/nutrient content. This is obviously a useful attribute. Many species glide across the substratum. This may involve specific organelles, such as the filament bearing goblet-shaped structures in the walls of Flexibacter (Moat, 1979). Often, however, no specific organelles appear to be involved and gliding may be attributable to the slime covering of some bacteria or the streaming of outer membrane lipids of others (e.g. Cytophaga (Moat, 1979)) or to other mechanisms currently being elucidated. The spiral-shaped Spiroplasma corkscrews its way through the medium by means of membrane-associated fibrils resembling eukaryotic actin (Boyd, 1988). Gonogoocci exhibit twitching motility (intermittent, jerky movements) due to the presence of pili (fine filaments, 7 nm diameter, less than one micrometer long) which branch and rejoin to form an irregular surface lattice. However, more than half of motile bacteria use one or more helical, whip-like appendages, about 24 nm diameter and up to 10 mm long, called flagella (sing. flagellum).

Flagella would be of limited use if bacteria could not sense their environment - they might help disperse cells to new environments but energy would be better used if bacteria could sense where they were going and respond appropriately. The success of flagella is evident from their wide occurrence and high expense (each flagellum comprises 1% of a bacterium's protein (Neidhardt, 1987) and about 2% of its genome (some 50 genes) for their synthesis and control. Perhaps the main advantage of flagella propulsion is their speed. Escherichia coli is 2 mm long and has a single flagellum that propels it at about 20 mm/s. In contrast speeds for gliding motility range from 1-10 mm/s. Flagella, therefore, allow bacteria to respond faster to changes in stimuli and enhance dispersal over a large area. The pattern of flagellation varies with each species. Some classification schemes divide flagellated bacteria into two groups: the Pseudomonadales have one or more flagella at one or both poles of the cell (they are polarly flagellated). The Eubacteriales have a random distribution of flagella, sometimes covering the whole surface of the cell (they are peritrichously flagellated).

Chemosensing mode in Escherichia coli

Bacteria can move towards or away from stimuli. In the vast majority of bacteria (all except the very largest) this occurs by chemokinesis, when the stimulus is a chemical such as a food substance or a noxious irritant. Chemokinesis is an indirect process in which the bacterium alternately swims for a stretch and then changes direction and changes its frequency of turning in response to environmental stimuli, which it can be demonstrated by computation leads to the bacteria gathering near food sources whilst avoiding noxious stimuli. Note that the definition used here of chemokinesis is in keeping with classical zoological and botanical definitions, but bacteriologists usually define the terms differently and refer to the phenomenon just described as chemotaxis. Call it what you like, the mechanism will now be described.

When unstimulated, free-swimming
Escherichia coli swarmer cells make long straight runs with infrequent tumbles (see figure below). During the straight runs, the flagella rotate CCW (counter-clockwise) and the flagella come together and work as a single bundle, possibly driven together by hydrodynamic forces (note that the cell body rotates in the opposite sense, CW, to the flagella). During the tumbles, the flagella reverse rotation to the CW (clockwise) sense, the wavelength of their filaments also changes and the pitch increases and the flagella mechanically uncouple and fly apart and the cell tumbles (rotates randomly). Chances are that after a tumble when the flagella resume CCW rotation and reform a bundle that the bacterium will start off in a new direction: it has turned by a random number of degrees. Recent evidence shows that only one or two flagella in a bundle are required to uncouple to effect a tumble.

Below: Patterns of movement in Escherichia coli

Escherichia coli chemotactic behavior

What do we mean by frequency of tumbling? The bacterial flagella have a switch mechanism, alternating between two states: CW rotation and CCW rotation, with the rotation rate remaining more or less constant. The Gibbs free energy for the CCW (swimming) to CW (tumbling) transition is about +40 kJ mol-1, and this energy is provided for by random thermal fluctuations. When the cell is triggered to tumble more, it simply increases the probability of tumbling by reducing the free energy required, making it more likely that thermal noise will provide the necessary energy.

Above left - a computer-simulated 'random walk' of a bacterial cell starting at the center of the arena and tumbling at random intervals. Right - two such simulated cells tracked over 15000 seconds each. Despite starting at the center they soon reach the edge of the arena (by random chance); they have been programmed to bounce of the walls at the arena's edge.

When Escherichia coli nears the source of an attractant chemical, moving up a concentration gradient, it's tumbling frequency decreases and it makes longer runs. If tumbling frequency was simply inversely proportional to attractant concentration, then computer simulations reveal that this will actually lead to a repellent effect: the bacterium will tend to move away from the attractant source. However, when the computerized virtual cell is given a memory, allowing it to compare the concentration at different points in time, it is able to measure the gradient rather than simply the concentration. By reducing tumbling as it moves up the gradient toward the attractant source, then it will indeed become 'attracted' to the source (meaning that it spends longer near the source). This is indeed found to be the case: bacteria have a memory of a few minutes that allows them to make temporal comparisons of chemical concentrations in their environment.

Computer-simulated bacterial cells programmed to respond to the (exponential) concentration gradient of a chemo-attractant placed in the center of the arena. Each cell has been given a one second memory with which to measure changes in attractant concentration over time. The cells tumble less if they move up the concentration gradient and tumble more if they travel down the gradient (within preset minimum and maximum tumbling frequencies). Each picture shows 4 cells traced for 5000 seconds each.
Left - each cell started at the edge of the arena at the bottom right corner at about 5 o'clock. Right - the bacteria start nearer to the arena center. Note how the bacteria remain near the attractant even though they are in constant motion.

Repellents work in a converse manner - the bacteria tumble more frequently as they approach the repellent (as the gradient increases) and so tend to avoid it:

Computer-simulated bacteria programmed to respond to a chemo- repellent at the center of the arena, by tumbling more frequently if they travel up the gradient and tumbling less frequently if they travel down the gradient (within preset minimum and maximum tumbling frequencies). Each cell had a one second memory to enable temporal comparisons in repellent concentration. Left - one cell starting at the center and was tracked for 15000 seconds. Right - 3 cells began in the center of the arena and were tracked for 15000 seconds each. They can be seen to clearly avoid the repellent and stay near the edge of the arena.

Chemosensing modes in other flagellated bacteria

Some alpha-subgroup bacteria also have a CW/CCW flagella switch, for example in Caulobacter crescentus CW rotation leads to swimming, CCW rotation to tumbling (the opposite sense as in Escherichia coli). Others, however, have flagella that rotate in one direction only. In other members of this group very different strategies are employed. Rhodobacter sphaeroides has a single right-handed flagellum in a lateral position which can push the cell at up to ~35 mm/s. This flagellum rotates only CW (or CCW in some variants). If an anaerobically grown cell detects a decreasing gradient  of chemoattractant (if it is moving from the source of the attractant) then its probability of turning increases (note that this response is opposite in sense to that of E. coli, but equivalent). During turning, the flagellum stops and assumes the form of a planar spiral. This whole flagellum spiral slowly rotates, turning the cell by a random amount. Sinorhizobium meliloti has 2-6 short, peritrichously or lophotrichously arranged right-handed flagella with fixed CW rotation, during which they form a propulsive bundle. The flagella rotate continuously, however, asynchronous rates of rotation cause the bundle to fly apart and the cell to tumble. more information on these other sensory response modes can be found here.

Mechanism of chemosensing

This section of the article is somewhat more technical.

The best-studied chemosensing mechanism of
Escherichia coli is the Che phosphorelay system summarised in the diagram below. In Escherichia coli the chemosensors are methyl-accepting chemotaxis proteins (MCPs) that span the inner membrane as dimers (pairs). These are connected to a phosphorelay system. In such phosphorelay systems, the signal is relayed by protein phosphorylation, which involves enzymes called phosphatases (which remove phosphate) phosphorylases (which attach phosphate) and kinases (which transfer phosphate from one molecule to another). These receptors tend to be concentrated at the cell poles. Together with several of the chemotaxis (Che) proteins they form a MCP:CheW:CheA:CheY receptor complex.

When stimulated by a repellent

CheW functions as a relay linker, connecting the MCP to CheA. When an MCP is stimulated by a chemorepellent binding to it, the receptors cluster in groups of about 7 MCP dimmers together with about 7 CheA dimers and are in their activated state. This state activates CheA which autophosphorylates (phosphorylates itself) to CheA-P.

CheA-P is an active kinase and phosphorylates CheY to the active CheY-P form.

CheY-P then dissociates from the receptor complex and diffuses to the flagella motor where it binds FliM in the C ring (see flagella structure) and it also binds the regulator phosphatase CheZ which undergoes delayed activation as it oligomerizes to its active form. This binding to CheZ increases the probability of tumbling (CW rotation) by reducing the activation energy of the switch. After a delay CheZ becomes active and this acts as a phosphatase, removing phosphate from CheY-P to yield CheY which reassociates with the receptor complex. Thus, CheZ acts as a time-delay which resets the system, a form of adaptation which helps confer the memory needed for the system to respond to concentration gradients.

A second adaptation system also exists, which is methyl-dependent. CheA-P not only phosphorylates and activates CheY, but it also phosphorylates and activates
CheB, at a slower rate, producing CheB-P after a time-delay. CheB-P is a specific methyl-esterase which cleaves a methyl group from MCP (MCP-Me → MCP). The methylated form is the active form which stimulates CW flagella rotation and tumbling. The demethylated form becomes inactive and CCW rotation and straight runs become favoured again - returning the cell to its pre-stimulus state. Chemorepellents thus stimulate MCP demethylation. This periodic resetting of the receptor occurs every few minutes, giving the cell a memory of an equivalent duration. CheR also helps to reset the system by antagonising CheB-P as CheR is a methyl transferase which methylates MCPs.

Above: Che phosphorelay signalling system. Binding of a chemical repellent (ligand) to the receptor (MCP) increases tumbling frequency when repellent concentration increases. The protein CheW (W) connects the MCP to the protein CheA (A). When the repellent binds MCP, CheA becomes activated by autophosphorylation and then phosphorylates CheY, activating it too. The active phosphorylated cheY diffuses to the flagellar motor and binds to it, along with CheZ to which the active CheY also binds. This increases the probability of tumbling by tending to switch the flagella into CW rotation mode.

When stimulated by an attractant

Attractants work in a converse manner. When an attractant binds to the MCP, the receptors deactivate and de-cluster into separate dimmers (with associated CheA dimers). This inhibits the autophosphorylation of CheA, deactivating it and reducing the concentration of CheY-P, resulting in an increased probability of CW flagella rotation and a reduced probability of tumbling.

Sugars are chemoattractants that do not bind MCPs directly. Instead they either bind their sugarspecific
periplasmic binding protein (PBP) which is involved in sugar transport into the cell as well as chemotaxis. [Recall that the periplasm is the region between the inner and outer membranes in the envelopes of gram negative bacteria and contains various PBPs which bind to nutrients.] Galactose, maltose and ribose work this way. At least the PBP for maltose is known to be localised at the cell poles along with the MCPs. Alternatively, some sugars, e.g. glucose, mannose and mannitol, bind to a specific Enzyme II of the PEP-dependent sugar phosphotransferase system (PTS) (PEP is phosphoenolpyruvate). The PTS Enzyme I modulates the kinase activity of the MCP-CheW-CheA receptor complex by phosphorylating CheA and simultaneously phosphorylates the sugar as it is imported through the EII/EIII proteins. [More explanation to be added here.]

MCPs also respond to changes in pH and temperature. Oxygen acts as an attractant for
Escherichia coli and its receptor is Aer (an MCP homologue) which is cytoplasmic and lacks methylation sites for adaptation. Aer also binds FAD (flavine adenine dinucleotide, an electron/hydrogen carrier used in respiration) which acts as an internal signal for the redox state of the electron transport chain (ETC).

A summary of this complex process of chemosensing in enteric bacteria, such as
Escherichia coli, is shown in the figures and table below:

MCP receptors

Attractants bind either directly to MCPs or via periplasmic binding proteins, the latter serving as adapters to increase the range of chemical ligands to which the MCPs can respond. The chemosensory ligand must first diffuse across porins (protein pores or channels) in the outer membrane (OM). IM: inner membrane; OM: outer membrane; PP: periplasm.

MCP binding repellent

Above: when a repellent (or PBP + attractant) binds the Escherichia coli MCP, such as phenol, or equivalently when an attractant dissociates from the MCP, then CheA autophosphorylates itself and then passes the phosphate on to a second messenger called CheY (CheA may potentially phosphorylate several CheY molecules in this way). Phosphorylated CheY, CheY-P, is active. CheB also becomes activated by phosphorylation. This is the characteristic two-component system (TCS) using a phosphorelay mechanism, which is characteristic of signalling in bacteria (compare this to the eukaryotic kinase cascade). CheW is an adaptor protein, attaching CheA to the MCP so CheA can respond to ligand binding the MCP. CheB-P demethylates the MCP. The more methylated the MCP, the more it activates CheA. Thus, demethylation by CheB (an esterase) reduces activity in the system by reducing CheA phosphorylation. This resets the MCP sensor should the concentration of repellent remain unchanged, this is adaptation to the stimulus and is a form of memory. It would be interesting to see to what extent adaptation occurs to repellents compared to attractants. (There are few repellents in the Escherichia coli case and establishing their existence has been problematic, but other bacteria may possibly sense a wider range of repellents). This adaptation is brought about by MCP methylation / demethylation which depends on the enzymes CheB and CheR.

MCP binding repellent

Above: when an attractant (or PBP + attractant), such as a specific sugar like galactose, binds the Escherichia coli MCP, CheA phosphorylation and activation are reduced and so phosphorylation and activation of CheY is also reduced. Similarly phosphorylation and activation of CheB is reduced. CheR (a methyl transferase) remains active at a more-or-less constant rate regardless and is now able to demethylate the MCP without competing with CheB. Methylation of the MCP is thus reduced, which tends to increase the activity of the MCP and CheA, resetting the system should the concentration of attractant not change. This is adaptation, which is crucial if the cell is to respond to changes in stimulus concentration.

Note: much of the details of this system are still being elucidated by experiment. There is some uncertainty as to the mode of CheW and CheA binding - CheW is described as linking the MCP to CheA by some sources, whilst others claim that CheA and CheW bind to MCP competitively. Here we present only models (the so-called 2:2:2 model) which will be altered as and when my own researches turn up positive developments. It has also been suggested that CheA binds between the two MCP dimers, facilitating their dimerisation.

Effect of CheY

When CheY is phosphorylated to CheY-P by CheA, it becomes active. In this active form it disassociates
from the MCP receptor and binds to FliM, a protein component of the C ring in the cytoplasm (see
motility for a description of the flagella structure). This switches the rotation bias of the flagellum to CW
rotation, which favours tumbles. CheY-P is phosphorylated when the concentration of a repellent increases, or equivalently, when the concentration of an attractant reduces.

Flowchart of bacterial flagella operation - new version

Notes: CheR is active at the same basic rate in both cases (that is it is not affected by binding or detachment of attractant or repellent from the MCP). This will methylate the MCP at a basal rate. However, when the concentration of repellent increases (or concentration of attractant decreases) then CheB is phosphorylated to CheB-P and activated. CheB-P demethylates the MCP, opposing the action of CheR. Thus, when active, the methylation of MCP decreases. Binding of repellent initially activates the MCP, then the delayed action of CheB-P deactivates it, leading to adaptation of the response.

Color and label your own diagram of Escherichia coli and its nanobrain! (Note these diagrams are not to scale: the size of the signaling components have been exaggerated for clarity).

MCP binding repellent

MCP binding repellent

MCP binding repellent

For more on chemokinesis in bacteria:

Modeling Chemokinesis

Integral feedback control and asymptotic tracking

One key feature of this signaling pathway is the aforementioned property of adaptation. The E. coli chemotactic system is dynamic: it responds primarily to changes in ligand (attractant or repellent) concentration rather than the absolute concentration of ligand. This is important since the bacterium wants to move closer to an attractant, such as a food source, and further away from a source of repellent: it is the concentration gradient that must be sensed as already discussed. The key thing about adaptation is that after a stimulus, such as an increase in repellent concentration activates the MCP receptors then after a certain short time interval the receptors return to their preset (resting) level of activity even if the ligand concentration remains elevated. Similar arguments apply to a decrease in attractant concentration activating the system or an increase in attractant / decrease in repellent inhibiting the system: the system detects the change then returns to its set level. This means that if repellent concentration increases from say 2 (in arbitrary units) to 4, and remains at 4, the system responds initially to the change from 2 to 4 and then returns to its preset level of activity.

A similar adaptation (though by different mechanisms) occurs with the human olfactory system. If you enter a smelly room, perhaps one in which the rubbish needs taking out, then you will be greeted by the unpleasant odor. If you remain in the room then after a while you are no longer aware of the odor: your olfactory system has adapted and returned to its preset level. If you go outside for a bit, into fresh air, and then return, however, then you will notice the smell again, for a while. Your olfactory system is specialized to detect changes in odor profiles rather than background odors: the former indicate change and changes are generally much more important. For a more sensitive animal, a new odor or an increase in the odor might indicate an approaching predator!

Adaptation returns the system to its preset level. This preset level is that at which the system is most sensitive to further changes: adaptation not only filters out unimportant constant background signals, but also ensures that the sensitivity of the system to further changes is maintained. In other words: the output from the sensory system is independent of the input level (stimulus of ligand or odor) in steady-state (when environmental signals are constant). Biologists call this perfect adaptation, engineers call it asymptotic tracking.

The most intuitive way to compensate for an error is to generate an output signal proportional to the size of the error, what we call a proportional controller. As an example, consider a heater attempting to keep the temperature of a room at the set temperature. If the temperature drops too low the heater comes on and switches off once the error = set temperature - actual temperature. However, the elements will remain hot for a while and so overcompensate. At best the actual temperature will oscillate around the set temperature. The residual or steady-state error that causes such overshoots or oscillations can be prevented by using integral control, which is employed in most actual heaters. Integral control is necessary to achieve asymptotic tracking, in which the system returns precisely to the set-level. An integral is essentially a sum of many small changes (not an exact definition) and integral controllers generate an output based on how the error changes over a period of time: they respond to the gradient or rate of change of the input signal. This is precisely what E. coli does: it samples ligand concentration over an interval of time (and space) To determine the gradient of the change and it is this gradient the bacterium responds to. Precise adaptation or asymptotic tracking is brought about by an integral controller. A third control system commonly employed in engineering is the differential controller, which uses collected input data to predict future trends: is the temperature going up or down, for example. Some control circuits may combine all three methods as a PID controller.

To achieve asymptotic tracking, it has been shown that the CheB / CheR system must have the following properties: the rate of receptor methylation must depend on receptor activity and not methylation level and not directly on ligand concentration; only inactive receptors should become methylated and only activated receptors demethylated. Indeed, this appears to be the case in E. coli, at least to a first approximation (Yi et al., 2000). Once again, Mother Nature has already beaten engineers to it! Nevertheless, the fact that both devise similar systems is reassuring.

For more details on adaptation see the pdf on network analysis of the Escherichia coli chemotactic system.

Geometry of the sensory array and cooperativity

Early models used to think of signaling pathways, such as the MCP-CheY pathway in E. coli, as consisting of molecules simply diffusing around at random, either in the volume of the cytosol of the cell or in a cellular membrane, and reacting when they encountered one-another. Although diffusion still plays a role, of course, more and more it is being realised that biomolecules form multimolecular complexes or nanomachines. Large molecules in particular, such as proteins often associate with one-another in such a way that one protein is not too far away from other proteins it must react with. First of all, this changes the kinetics: it may force proteins to react in certain stoichiometries (ratios) rather than reacting with a large number of diffusing individual proteins. For example, in eukaryotic cells proteins that make up a signaling module called a kinase cascade would ordinarily form an amplifier when the components are free to diffuse, but if the components are held into a multiprotein complex by adaptor or scaffold proteins the module acts as a noise filter instead. (A scaffold protein can be thought of as a large multi-adaptor protein that can bind to several other proteins and hold them in place). In bacteria, with their small cell size and simpler architecture, we might expect such complexes to occur less often, but they do occur and in E. coli and some other bacteria that have been studied, the chemotactic receptors form multiprotein complexes.

MCP chemosensory array

Above:a model of the chemosensory array of E. coli as seen when looking down onto the membrane (into the cell). The receptors are shown in green as trimers of dimers (pointing out of the page); the CheW adaptor proteins are shown as small blue circles: each dimer is strongly associated with one CheW; orange circles are CheA dimers. The diameter of the central 'pore' is 10 nm (1 nm = 10-9 m). The stoichiometry is 6MCP:3CheW:3CheA. Many of these hexagonal arrays are fitted together in a 2D sheet containing something like 1500 to 4500 MCP receptors (up to 125 of these hexagonal units) - the actual number is probably dynamic. This would form a patch about 0.2 to 0.6 μm in diameter (1 μm = 10-6 m). This field of sensors occurs at one or both ends in rod-shaped bacteria like E. coli. The large MCP proteins cause the membrane to curve and so to minimize energy this arrays assemble at the cell poles where the curvature is greatest. (Based on: Shimizu et al., 2000).

Below: an alternative representation:

MCP chemosensory array

Such complexes of receptors raises another possibility: cooperativity. Cooperativity between the receptors occurs when a ligand binding to one receptor increases the affinity of the other receptors and so enhances the binding of additional ligands. (Cooperativity may also involve the CheA, in which activation of one CheA increases the likelihood that other CheA units become activated when further ligands bind the receptors). The chief effect of this kind of cooperativity is to filter out noise from the signal. Noise can be external, e.g. the binding of a stray ligand, or internal, e.g. a CheA molecule momentarily flipping into its active state with low probability in the absence of ligand (thinking in terms of repellents to make things easier). This arrangement may also possibly amplify the signal, but note there is a trade-off between 'sensitivity' and noise reduction. The chemosensory apparatus of E. coli certainly filters out noise efficiently and also amplifies the signal considerably and these protein complexes are almost certainly integral to how this is done, although the exact mechanisms are not yet fully understood.

For more details of these phenomena see our pdf on modeling this system computationally using network analysis.

Responses to other stimuli

Bacteria can also respond to a variety of other stimuli, apart from specific chemicals, and responses to temperature, light, osmolarity, pH and touch have been observed. Photosynthetic bacteria may move towards the light to photosynthesise, or away from it if the light is too bright and potentially damaging. One particular response we shall look at here is the response to magnetism, in particular to the Earth's magnetic field. Bacteria that move magnetic North or South in a magnetic field are called magnetotactic.

Above: Magnetospirillum (Aquaspirillum)

Magnetotactic bacteria are Gram-negative, flagellated and motile bacteria. They swim at about 100 mms-1. They are microaerophilic, aquatic and contain magnetosomes. These bacteria come from many taxonomic groups. The magnetosomes are intracellular magnetic grains that give each cell a permanent magnetic dipole moment. The local geomagnetic field (about 0.5 G) places a magnetic torque upon the bacteria, aligning them with the magnetic field.

Magnetospirillum magnetotacticum (Aquaspirillum magnetotacticum) has Fe3O4 magnetite (lodestone) magnetosomes. Each cell contains about 20 cuboidal-octahedral crystals. Each crystal is about 420 angstroms (42.0 nm) in diameter and the crystals are arranged in a chain. Single magnetic domains in magnetite range from 400-1000 angstroms. Full alignment requires the ratio of the interactive magnetic energy with the applied field to the thermal energy to be greater than about 10. A ratio of 16 is obtained for 22 500 angstrom particles. As we shall see, there are other mechanisms that stabilize the structure and keep the crystals aligned.

Thus, the magnetosomes function as a ferromagnetic
biocompass. At 0.5G and 30oC, 80-90% of the cells are fully aligned. The cells clump together if they contain too many magnetosomes. Recorded levels of magnetotactic bacteria yield 103-104 cells / ml of slurry in aquatic environments (in New England). These bacteria are absent from heavily polluted waters and absent in limestone caverns, thermal springs, thawed Antarctic sediments and iron-rich seeps. They are abundant in water purification plant settling basins, ponds with organic sediments, and sewage-treatment oxidation ponds.

In the Northern Hemisphere, magnetotactic bacteria swim predominantly northward, whilst in the Southern Hemisphere they swim predominantly southward. The vertical component of the geomagnetic field, due to Earth's curvature, is directed upwards in the Southern Hemisphere, is zero at the geomagnetic equator, and is directed downwards in the Northern Hemisphere. Therefore, the bacteria are directed downwards in both hemispheres and so are abundant in sediments, but absent in surface water. At the geomagnetic equator there are approximately equal proportions of each polarity and the bacteria swim horizontally.

Magnetospirillum magnetotacticum is a microaerophilic chemoheterotroph that metabolises organic acids, like fumaric, tartaric and succinic acids as its sole carbon (and energy?) source. Nitrate is the principle electron-acceptor for microaerobic growth on tartrate. Traces of oxygen are required. (As a substrate for oxygenases? For haem production?). Nitrate is reduced to nitrous oxide and ammonia in the process. Thus, this bacterium is a denitrifier. In low oxygen tensions (< 0.2 kPa) these bacteria also reduce ethyne. In high oxygen tension (above about 6 kPa) they do not produce magnetosomes and are not magnetotactic. This bacterium lacks catalase activity and so is sensitive to high oxygen concentrations. They have polar or bipolar flagellation. Pili on coccoid forms, extracellular polysaccharide on spirilla, allow them to adhere to sediment particles.

Magnetosomes are typically (though not always) arranged in one or more linear chains. Each magnetosome particle is a single crystal of magnetite or greigite enclosed in a phospholipid bilayer membrane. This membrane is derived by invagination of the inner membrane but contains specialized proteins. This causes some confusion in nomenclature - does 'magnetosome' refer to the whole chain or just to one of the particles? Each magnetosome particle/vesicle is a tiny magnet and the particles within each chain are aligned -(N-S)-(N-S)-(N-S)- which helps to stabilize the chain as a magnetic north pole is attracted to a magnetic south pole. However, as anyone who has ever played with magnets will know, this arrangement is metastable, meaning that a slight displacement of one of the magnets can cause the whole chain to collapse into an aggregated cluster of magnets. Additional structures are needed to stabilize this arrangement (see the diagram below). In
Magnetospirillum the actin homologue protein MamK forms a protein filament track along which the magnetosome particles are aligned (see below). There is evidence to suggest that the protein MamJ forms linkers that connect each vesicle to the track. It is also possible that additional structures link neighbouring magnetosome vesicles together. (See review by Thanbichler and Shapiro, 2008).

Above: The magnetosome and associated stabilizing structures in Magnetospirillum.

Some bacteria have two chains of magnetosomes at opposite sides of the cell and it is possible that these are positioned by their mutual magnetic repulsion. Large cells require a larger torque to turn them and the large rod- shaped cells of Magnetotacticum bavaricum, which are 8-10 mm long, have 200-1000 crystals arranged in 5 chains. However, it is possible that this produces far more torque than needed and magnetosomes may have additional functions.

The individual crystals of magnetite are 35-120 nm in diameter. A magnetite crystal contains one or more domains - regions in which the molecules all have their magnetic moments aligned parallel, reinforcing one another. If the crystals were much larger than domain boundaries would occur, where the alignment of the molecules switches between crystal domains with magnetic poles pointing in different directions (typically opposite one another). A crystal with several domains whose magnetic moments tend to oppose and cancel one another will have a reduced magnetic moment. If the crystals were much smaller, then they would lose their permanent magnetism, due to the thermal motion of the atoms. Thus, the size chosen is optimal for maximizing the intensity of the crystal's magnetic field.

Some bacteria use greigite (Fe3S4) instead of magnetite. In particular these bacteria occur in environments where oxygen availability is low and sulfur-availability high. Greigite is less effective, however, as it makes a weaker magnet (one third as strong). At least one known species of bacterium has crystals of both greigite and magnetite (a given magnetosome particle must be one or the other and mixed crystals do not occur). Such bacteria encounter environments of variable oxygen content - if oxygen is scarce then they incorporate sulfur instead.


Photosynthetic bacteria require light and are thus sensitive to it and, if motile, able to locate the best region of illumination to maximize their photosynthesis. For example the cyanobacterium Synechocystis will crawl towards light by extending type 4 pili (T4P or Tfp) which adhere to the substrate and then retract, dragging the cell forwards. Evidence suggests that the whole spherical cell acts as a lens to focus the light in order to determine the direction towards the light source, as illustrated below:

This begs the question: how small can a lens be and still focus light? Microlenses have been manufactured down to about 10 μm diameter. As a first 'rule of thumb' one might expect a lens to have to be at least ten times the wavelength of light in diameter to work as a conventional lens. For blue light at 500 nm wavelength, this would require a lens of 5 μm diameter or greater. A typical cell of Synechocystis, however, is only about 2 μm in diameter. This does not mean that it can not refract and focus light, only that diffraction effects will become very significant and perhaps a sharp image could not be formed. However, it only needs to focus the light partially to create a measurable difference in light intensity around the circumference of the cell (assuming the photoreceptor to be in or adjacent to the cell envelope). This prokaryote is probably about as small as it can be and still perform this function effectively. The direction opposite to the spot where the focused light is brightest is the direction the cell needs to move in (towards the right in the above diagram).

Related Articles:

Bacterial Flagella

Gliding in Bacteria

Models of Flagella Rotation

Motility in Spirochaetes

Network analysis of the E. coli chemotaxis signaling pathway (pdf)

Evolution of Bacterial Flagella

See also quorum sensing - how bacteria sense the presence of one-another!

Bacterial Biofilms

Article updated 16/9/14, 12/7/2020, 16/7/2023


Shimuzu, T. S., Le Novėre, N., Levin, M.D., Beavil, A.J., Sutton, B.J. and Bray, D. 2000. Molecular model of a lattice of signalling proteins involved in bacterial chemotaxis. Nat. Cell Bio. 2: 792-796.

Yi, T.-M., Huang, Y., Simon, M.I. and Doyle, J. 2000. Robust perfect adaptation in bacterial chemotaxis through integral feedback control. PNAS 25;97(9):4649-53. PMID:10781070 PMCID:PMC18287 DOI:10.1073/pnas.97.9.4649