Mycolactones are polyketide-derived lipid virulence factors made by the slow-growing human pathogen, Mycobacterium ulcerans. Three unusually large and homologous plasmid-borne genes (mlsA1: 51 kb, mlsB: 42 kb and mlsA2: 7 kb) encode the mycolactone type I polyketide synthases (PKS). The extreme size and low sequence diversity of these genes has posed significant barriers for exploration of the genetic and biochemical basis of mycolactone synthesis. Here, we have developed a truncated, more tractable 3-module version of the 18-module mycolactone PKS and we show that this engineered PKS functions as expected in the natural host M. ulcerans to produce an additional polyketide; a triketide lactone (TKL). Cell fractionation experiments indicated that this 3-module PKS and the putative accessory enzymes encoded by mup045 and mup038 associated with the mycobacterial cell wall, a finding supported by confocal microscopy. We then assessed the capacity of the faster growing, Mycobacterium marinum to harbor and express the 3-module Mls PKS and accessory enzymes encoded by mup045 and mup038. RT-PCR, immunoblotting, and cell fractionation experiments confirmed that the truncated Mls PKS multienzymes were expressed and also partitioned with the cell wall material in M. marinum. However, this heterologous host failed to produce TKL. The systematic deconstruction of the mycolactone PKS presented here suggests that the Mls multienzymes are necessary but not sufficient for mycolactone synthesis and that synthesis is likely to occur (at least in part) within the mycobacterial cell wall. This research is also the first proof-of-principle demonstration of the potential of this enzyme complex to produce tailored small molecules through genetically engineered rearrangements of the Mls modules.