Escherichia coli contains a major phosphofructokinase isoenzyme, phosphofructokinase 1, which is allosteric, and a minor isoenzyme, phosphofructokinase 2. The pfkB1 mutation is known to increase the amount of phosphofructokinase 2 and allow growth on sugars of mutants lacking phosphofructokinase 1; it does not affect growth on substances such as glycerol or lactate (i.e., 'gluconeogenic growth'). However, gluconeogenic growth is markedly impaired in strains with a different allele, pfkB1*. We show here that strains with pfkB1* contain an altered form of phosphofructokinase 2, called phosphofructokinase 2*, which has been purified. Phosphofructokinase 2* is cold labile and has slightly different kinetic characteristics from phosphofructokinase 2, which include being less sensitive to inhibition by fructose 1,6-bisphosphate. The Km for fructose 6-phosphate is low (about 5 X 10(-5) M) in both phosphofructokinase 2 and phosphofructokinase 2*. However, in strains lacking phosphofructokinase 1, a high level of phosphofructokinase 2 is associated with unusually high concentrations of hexose monophosphates during growth on glucose, while a strain with phosphofructokinase 2* instead of phosphofructokinase 2 grows more rapidly on glucose and contains lower levels of hexose monophosphates. In gluconeogenic conditions, by contrast, hexose monophosphate levels are normal in phosphofructokinase 2 strains, while the impaired growth of phosphofructokinase 2* strains is associated with high levels of fructose 2,6-bisphosphate and very low levels of hexose monophosphates. These results show that phosphofructokinase 2, as studied in vitro, should no longer be regarded as a 'non-allosteric' protein, a conclusion also reached by Kotlarz and Buc on the basis of different types of experiments [Eur. J. Biochem. 117, 569-574 (1981)]. The fact that mutational alteration of phosphofructokinase 2 allows more rapid growth on glucose but severely impairs gluconeogenic growth is an indication of the significance of the regulation in vivo. The more rapid growth of the mutant on glucose might be explained on the basis of decreased sensitivity to an inhibitor (possibly, but not necessarily, fructose 1,6-bisphosphate), although other models are possible. A variety of speculations are offered as to the mechanism of gluconeogenic impairment.