The inventors have discovered that lipid A can be genetically eliminated from Caulobacter crescentus, dependent upon inactivation of the transcriptional regulator Fur and the presence of anionic sphingolipids called ceramide phosphoglycerate. The inventors identified and characterized genes responsible for ceramide phosphoglycerate synthesis. The inventors propose that other Gram-negative bacteria, including E. coli, can be engineered to eliminate lipid A by inactivating their Fur homologs, introducing genes for the synthesis of ceramide phosphoglycerate, or both. Bacteria thus engineered could be used for the endotoxin-free production of small molecule or protein-based pharmaceuticals, therapeutic bacteriophage, RNAs, or endotoxin-free therapeutic bacteria.BACKGROUND The bacterium Escherichia coli is used as a platform for the manufacture of 20-30% of the biopharmaceuticals currently marketed. E. coli, like other Gram-negative bacteria, possesses an outer membrane containing the glycolipid lipopolysaccharide (LPS). The innermost portion of LPS, lipid A, anchors LPS in the outer leaflet of the outer membrane. Lipid A, historically known as endotoxin, is a potent stimulator of the innate immune system in mammals. Even small amounts of endotoxin in the bloodstream can induce an unregulated, systemic inflammatory response known as sepsis. A major hurdle and cost in E. coli-based pharmaceutical production is the removal of endotoxin from each final product. Endotoxin removal strategies are developed on a case-by-case basis to find conditions in which the stable lipid A contaminant can be chemically separated from the desired product while not adversely affecting product recovery or activity. Industrial biotechnology could benefit from additional bacterial production platforms that eliminate the need for extensive processing to remove endotoxins. The challenge is that lipid A is almost always an essential structural component of the OM, meaning that it cannot be eliminated without causing the death of the bacterium. To date, only four species that normally contain lipid A have yielded mutant strains that completely lack lipid A and its biosynthetic precursors. However, these species are not well-developed platforms for industrial biotechnology. An E. coli strain (KPM22) has been developed that survives with only lipid IVA, an intermediate in the lipid A biosynthesis pathway. Lipid IVA contains fewer acyl chains than mature lipid A, causing a ~1000-fold reduction in its endotoxin activity. A modified version of this “endotoxin-free” strain is currently marketed by Lucigen under the trade name ClearColi (https://www.lucigen.com/faq-clearcoli.html).