We describe a new sensory response in the enteric bacterium Serratia marcescens. When grown in liquid media, the bacteria were short rods with one to two flagella and displayed classical swimming behavior. Upon transfer to a solid surface (0.7 to 0.8% agar medium), the bacteria underwent a dramatic change of form. They ceased septation, elongated, and expressed numerous (10 to 100) flagella that covered the lateral sides of the cells. The bacteria now displayed a different form of locomotion-swarming-which allowed them to rapidly move over the top of the solid surface. The differentiation to either swimmer or swarmer cells could be reversed by growth on solid or liquid medium, respectively. To identify conditions that influence this differentiation, the growth environment of S. marcescens was manipulated extensively. The swarming response was monitored by visual and microscopic observation of cell movement on solid surfaces, by immunofluorescent labeling followed by microscopic observation for the presence of elongated, profusely flagellated cells, as well as by estimation of induction of flagellin protein, using Western immunoblot analysis. Conditions that imposed a physical constraint on bacterial movement, such as solid or viscous media, were the most efficient at inducing the swarming response. No chemical constituent of the medium that might contribute to the response could be identified, although the existence of such a component cannot be ruled out. Both swimmer and swarmer cells had flagellin proteins of identical molecular weight, which produced similar proteolysis patterns upon digestion with trypsin. Combined with an earlier demonstration of a single flagellin-encoding gene in S. marcescens (R. M. Harshey, G. Estepa, and H. Yanagi, Gene 79:1-8, 1989), our results suggest that the same protein is a component of both swimmer and swarmer cell flagella. Thus, in response to a physical signal, a single flagellin gene must be differentially regulated. We discuss possible mechanisms for sensing such a signal.

Serratia marcescens exists in two cell forms and displays two kinds of motility depending on the type of growth surface encountered (L. Alberti and R. M. Harshey, J. Bacteriol. 172:4322-4328, 1990). In liquid medium, the bacteria are short rods with few flagella and show classical swimming behavior. Upon growth on a solid surface (0.7 to 0.85% agar), they differentiate into elongated, multinucleate, copiously flagellated forms that swarm over the agar surface. The flagella of swimmer and swarmer cells are composed of the same flagellin protein. We show in this study that disruption of hag, the gene encoding flagellin, abolishes both swimming and swarming motility. We have used transposon mini-Mu lac kan to isolate mutants of S. marcescens defective in both kinds of motility. Of the 155 mutants obtained, all Fla- mutants (lacking flagella) and Mot- mutants (paralyzed flagella) were defective for both swimming and swarming, as expected. All Che- mutants (chemotaxis defective) were also defective for swarming, suggesting that an intact chemotaxis system is essential for swarming. About one-third of the mutants were specifically affected only in swarming. Of this class, a large majority showed active "swarming motility" when viewed through the microscope (analogous to the active "swimming motility" of Che- mutants) but failed to show significant movement away from the site of initial inoculation on a macroscopic scale. These results suggest that bacteria swarming on a solid surface require many genes in addition to those required for chemotaxis and flagellar function, which extend the swarming movement outward. We also show in this study that nonflagellate S. marcescens is capable of spreading rapidly on low-agar media.

The NILE glycoprotein is a rat neuronal cell adhesion molecule which has been reported to be very similar in structure, function, and distribution to the mouse L1 glycoprotein. Here we report the complete nucleotide sequence of the NILE message (5,208 nucleotides) and the deduced amino acid sequence of the NILE polypeptide (1,257 amino acids). The predicted NILE protein is 96% identical to L1 at the amino acid level, confirming that the two molecules are homologues. The sequence information shows that NILE is a transmembrane molecule with an extensive ectodomain and a much smaller cytoplasmic domain. The extracellular portion of the molecule contains six immunoglobulin C-2 type domains followed by five fibronectin type III repeats. These two structural motifs are characteristic of several other cell adhesion molecules. The cytoplasmic tails of NILE and L1 are identical to each other and distinct from the cytoplasmic regions of all other cell adhesion molecules except Ng-CAM and neuroglian. Several possible sites for phosphorylation are present in the cytoplasmic tail of NILE. Antisera were produced against two NILE-beta-galactosidase fusion proteins containing distinct segments of the NILE polypeptide: the cytoplasmic domain and the segment containing fibronectin type III repeats. Immunoblots with these antisera and Northern blots with a NILE cDNA probe indicate that NILE continues to be expressed in most areas of the mature rat brain. This contradicts previous immunofluorescence data, which suggested that NILE was substantially down-regulated in maturing nerve fiber tracts. This raises the possibility that NILE could be masked in situ by interactions with other cell surface molecules.

No abstract