Cultures of M. xanthus, DK1622, and several of its mutants were grown as described in Kaiser and Warrick (2011). Bacteria were propagated as swarms by inoculating an agar plate with bacteria harvested from the edge of a mature swarm using the tip of a sterilized round toothpick, and incubating at 20°C. Most likely, clusters of aligned cells carried on the toothpick helped nucleate the swarm. Time-lapse photo-microscopy was carried out as described in Kaiser and Warrick (2011).

In 1978, we set out to find which extracellular signals M. xanthus employs, using a simple bioassay. Two signals were identified from a set of conditionally defective mutants unable to form spore-filled fruiting bodies in single-species pure culture (Hagen et al., 1978; LaRossa et al., 1983). When different mutants were mixed with wild-type cells or with other mutants, some of those mixtures were able to form fruiting bodies with spores. Pairwise mixing of 57 mutants identified two extracellular signals: A and C.

Medium conditioned by Myxococcus development was found to include both a heat-stable and a heat-labile form of A-signal activity. In 1992, Plamann et al. found that heat-labile A-signal was a mixture of proteases and proteins that were degraded by the proteases (Plamann et al., 1992). That same year Kuspa et al. showed that heat-stable A-signal was a set of six amino acids and small peptides containing those amino acids (Kuspa et al., 1992). It appears that amino acids are the primary A-signal molecules, while the extracellular release of proteases and proteins generates first peptides, then A-signal amino acids.

Singer and Kaiser (1995) found that fruiting body development is induced by starvation and that A-signal helps M. xanthus assess the nutrient available for developmental protein synthesis. When Myxococcus is challenged by starvation, it must choose between initiating fruiting body development with differentiation of spores or slow growth at a rate compatible with the level of nutrient available. In nature, myxobacteria feed on particulate organic matter in the soil. Because death follows for the great majority of cells (more than 99%), the option of choice depends on each cell’s projection of future nutrient availability. If nutrient is on its way to exhaustion, then slowing growth to match the level of residual nutrient leads to slower and slower growth, until death from starvation follows. Manoil found that limitation for any amino acid or starvation for carbon, energy, or phosphorous induces fruiting body development (Manoil and Kaiser, 1980). Neither O2 deprivation nor purine or pyrimidine starvation induces development (Kimsey and Kaiser, 1991). Since a complete set of amino-acylated tRNAs is needed for protein synthesis, the absence of one or more amino-acylated tRNA is readily perceived. In M. xanthus, as in many other bacteria, the absence or shortage of any one of the charged tRNAs being called for by messenger RNA causes a ribosome to synthesize guanosine tetra- (and penta-) phosphate, (p)ppGpp. Pyrophosphate, or P~P, is transferred from ATP to GTP, and a stringent response is triggered. Singer showed that (p)ppGpp was both necessary and sufficient to initiate fruiting body development (Singer and Kaiser, 1995). Giglio et al. found that development is initiated and propagated by a sequence of enhancer binding proteins, responding to the level of A-signal (Giglio et al., 2011). Recently, Sarwar confirmed that M. xanthus uses a stringent response and the cell-density-dependent A-signal to predict the future nutrient availability (Sarwar and Garza, 2012).

A-signal is water soluble and diffusible, and the model emerging from Giglio et al. is that some threshold number of starving cells is needed to begin development (Giglio et al., 2011). A census of the cell population is then taken to determine whether the consensus is adequate. By sensing the level of nutrient available, each cell naively casts a vote as to whether it is time to begin fruiting body development. When it produces A- signal, it casts its vote for development. But the vote is advisory only, for if nutrient is added, all the cells begin to grow again; they haven’t yet committed themselves to sporulation. By definition, quorum signals are soluble and diffusible and they are used by many bacteria to determine the number of bacteria in their neighborhood belonging to their species as well as the number that belong to different species. Bacillus subtilis, for example, uses a particular pentapeptide, PhrC, to regulate competence for DNA transformation and to regulate the initiation of sporulation (Auchtung and Grossman, 2008), with each signal dependent upon a particular biochemical model. Vibrio harveyi and Vibrio cholerae use a variety of homoserine lactones as quorum sensors (Hammer and Bassler, 2008). Two questions arise for every quorum sensor: (1) What is the chemical identity of the signal molecule and (2) what is the biochemical model for the receiver?

The C-signal, rather than sensing a quorum, is a morphogenetic paracrine signal. C-signal is required for cellular aggregation, spore differentiation, and gene expression induced by starvation (Kim and Kaiser, 1990b,d). Surprisingly, cell motility (apparently A-motility) is required for proper intercellular transmission of the signal (Kim and Kaiser, 1990c). Due to the complexity of its action and the many poorly understood proteins necessary for M. xanthus development, only fragments of the C-signal transduction pathway can be written down; a number of fragments are compiled in one review (Søgaard-Andersen, 2008). C-signal, itself, is cell-bound, and signal exchange requires direct contact between two cells, unlike the A-signal that can diffuse between cells. Active C-signal was purified from the membrane fraction of whole cells, using a detergent to dissolve the signal protein. The model bioassay depended upon restoring aggregation and sporulation to a mutant lacking a csgA gene (Kim and Kaiser, 1990c). The mutants arrested fruiting body development before aggregation (Kaiser, 2003) and they formed very few, if any, spores. A 17 kDa protein was purified from starved wild type cells that could restore activity to csgA mutants (Kim and Kaiser, 1990c). C-signal activity was not recovered from extracts of growing cells (not starved) or from csgA mutants (Kim and Kaiser, 1990a). Subsequent experiments showed that p17 is the molecular signal and that it is produced by the PopC cell-surface protease acting on p25 (Kruse et al., 2001). C-signal carries information as to cell density and to cell position with respect to other cells. It was a surprise to find that non-motile mutants of M. xanthus arrested fruiting body development at exactly the same morphological stage as the csgA mutants (Kroos et al., 1988). This observation suggested that C-signaling might occur only between aligned cells in end-to-end contact. Seung Kim tested this hypothesis by mechanically placing non-motile cells into end-to- end alignment (Kim and Kaiser, 1990b). The asymmetry of the long rod-shaped M. xanthus cells was used to orient them lengthwise as they tumbled from suspension into the narrow grooves produced by scoring agar with a fine grained aluminum oxide abrasive paper. Phase contrast microscopy revealed that cells, which had settled into the grooves, were indeed oriented with their long axes parallel to the axis of the groove (Kim and Kaiser, 1990b). Because the grooves were 5–10 μm wide, cells were also found lying side-by-side in the grooves. For that reason, the experiment did not exclude C-signaling between the sides of two cells. Further experiments are required to test whether side-by-side contacts are involved in C-signaling. A related difficulty is that a C-signal receptor in the receiving cell has yet to be identified. Gronewold discovered positive feedback in the C- signal reception circuit, controlled by the act operon of 5 co-transcribed genes (Gronewold and Kaiser, 2001, 2002). Positive feedback appears to raise progressively the number of C-signal molecules per cell from a few at 3 h post starvation to several hundred by 18 h (Gronewold and Kaiser, 2001, 2002). This rise coordinates C-signal-dependent gene expression, and eventually it triggers spore differentiation with a concomitant loss of cell motility (Kroos and Kaiser, 1987; Julien et al., 2000). There may be a functional similarity between C- signal, the PatS- and the HetN-dependent formation of heterocysts in the filamentous, nitrogen-fixing cyanobacteria Anabaena PCC 7120 (Aldea et al., 2008)

In 1977, Jonathan Hodgkin, then a postdoctoral fellow in my laboratory, discovered that CglB, a protein that is essential for A-motility, and later found to be an outer membrane lipoprotein, could be transferred from one cell to another cell, provided the cells had come into end-to-end contact with each other (Hodgkin and Kaiser, 1977). However, only the CglB⁺ phenotype was transferred, which was detected by a gain of A-motility; the cglB gene was not transferred! Hodgkin found that the CglC, CglD, CglE, and CglF phenotypes were capable of being similarly transferred (Hodgkin and Kaiser, 1977). Hodgkin also found that the Tgl protein and not the tgl gene is transferred (Hodgkin and Kaiser, 1979). Tgl protein was shown to be an assembly factor that non-covalently linked 12–14 PilQ monomers together into an active PilQ multimer secretin (Rodriguez-Soto and Kaiser, 1997a,b). Subsequently, it was shown that substantial amounts of Tgl and CglB proteins were being transferred as a result of the transient contact with a mutant recipient – enough of each to restore full A⁺S⁺ motility (Nudleman et al., 2005). Recently, two host proteins, TraA and TraB, were shown to be essential in both donor and recipient for contact-mediated, outer-membrane lipoprotein transfer of this type (Pathak et al., 2012). In addition, TraA and TraB were found to mediate the transfer of lipids that modify swarming (Pathak et al., 2012). Although CglB, CglC, and CglD proteins have type II signal sequences, the transfer of CglE and CglF proteins, which have type I signal sequences, suggests that lipid molecules in addition to proteins without any lipid modification, like Tgl, can be transferred from one cell to another, provided they have been targeted to the outer-membrane (Pathak and Wall, 2012). Bhat et al.have identified many beta-barrel and lipoproteins in the outer membrane of M. xanthus by LC- MS/MS (Bhat et al., 2011). Because only a small fraction of M. xanthus outer membrane proteins can be transferred by cell contact, one can imagine that the few that can be transferred are not only exposed to other cells but organized for signaling to them.

In the fore-mentioned group of exposed and organized proteins, CglB, CglC, CglD, CglE, and CglF are proposed to link a pair of adjacent cells together in a particular way – to link them through a pair of multi-protein structures known as focal adhesions (Mignot et al., 2005, 2007) and to link them, thus, for less than a minute before the two cells separate and move on to link similarly to other cells, and signal to them. The focal adhesions are found on the sides of M. xanthus cells; each adhesion can be seen as a discrete series of fluorescent foci. Despite the multitude of different proteins that cluster in the focal adhesions, no amino acid sequences related to microtubule-based kinesin motors or actin-based myosin motors (Vale and Milligan, 2000) have been identified (see Luciano et al., 2011, for example). Indeed, the 15 or more A-motility proteins in the clusters associated with focal adhesions seem well-suited for a signaling pathway for two reasons. One, many of the proteins can bind one another in specific pairs as measured by GST affinity chromatography, identified in Table 1 of Nan et al. (2010). Two, different binding proteins are found to favor localization in different subcellular compartments of an M. xanthus cell. Proceeding inward, they are found on the outer surface of the outer membrane, in the periplasmic space, on the outer surface of the peptidoglycan sacculus, associated with the inner membrane, and finally in the cytoplasm, where the pacemaker proteins are located (Kaiser and Warrick, 2011). Due to their broad spatial distribution, the hypothetical signal would be capable of linking methylated-FrzCD in the pacemaker of one cell through pairs of protein 1 transiently bound to protein 2 links to CglB – one of the 15 A-motility proteins found in a focal adhesion – on the surface of that cell. CglB etc. on the surface of the first cell could assemble together with CglB etc. on the surface of the second cell, transiently forming a junction with a specific structure between the 2 cells. From that junction, the signal would link through the same series of protein 1" protein 2 pairs until it reached FrzCD in the pacemaker of the second cell. When completed, as shown in Figure 2, the series of signal links could plausibly bring the pacemakers of both cells to the same phase of their oscillatory cycles. The circuit of Figure 2 offers a concrete and thus testable example of the signaling alternative.