582 PETER MITCHELL 



molecules or other particles to pass against the natural direction of the 

 diffusion or escaping tendency, or act, as Cohen and Monod [i] have 

 suggested, as Maxwell demons. This, we can say at least with the certainty 

 of the physicist, is not possible. 



Membrane structure and transport function in bacteria 



The relative simplicity of the structure of bacteria makes them 

 especially suitable for the study of transport processes at the molecular 

 level of dimensions [2, 3, 4]. In this paper I shall concentrate attention 

 upon bacterial membranes, but will attempt to develop a simple conception 

 of the relationship between transport function and physicochemical 

 structure that may be of general validity in biology. 



Broadly speaking there are four main experimental approaches to the 

 analysis of membrane transport which can be summarized under the 

 following headings: 



1. Osmotic barrier function of the plasma membrane: General 

 impermeability function ; studied by net permeation measurements. 



2. Osmotic link function of the plasma membrane : Specific transport 

 function; studied by observations on the specificity and kinetics of 

 the transport process, interpreted in terms of the catalysis of mole- 

 cular complex-, molecule-, ion-, electron-, and group-translocation. 



3. Structure of the plasma membrane: Chemical and catalytic com- 

 position ; studied by orthodox chemical and biochemical methods. 



4. Correlation of structure and function in " synthetic " or reconstituted 

 membrane systems. 



The first three of these approaches have been pursued in parallel in my 

 laboratory. Our studies of the osmotic barrier function, beginning with 

 the introduction of the term osmotic barrier 1 1 years ago [5], can be 

 roughly summarized by saying that in general bacterial plasma membranes 

 are permeable to small molecules carrying three water molecules or less 

 (e.g. glycerol), but they are impermeable to molecules carrying more than 

 four water molecules (e.g. glutamate, phosphate, succinate, and glucose). 

 There are, of course, factors other than the degree of hydration that 

 influence the rate of permeation of different solutes into bacteria. For 

 example, D-ribose permeates much more rapidly than L-arabinose and 

 other pentoses, probably because in the ribose molecule all the hydroxyl 

 groups are on the same side of the ring so that one side of the molecule is 

 hydrophilic while the other is hydrophobic. There are also differences 

 between the permeability of the plasma membrane of different organisms 

 to a given solute. For example. Staphylococcus aureus and Micrococcus 

 lysodeikticus are quite permeable to alkali thiocyanates while Escherichia 



