Of the millions of proteins on the planet it is unremarkable most exist elsewhere, just as it's likely most parts in a car can be found in other machines. Further, these aren't identical proteins, merely 'homologous' ones, where there's a stretch of, say, a 70% match over 40% of the protein, so that makes this finding not surprising (lug nut in engine A like fastener in engine B). A Type 3 Secretory System has about 1/3 of the proteins in a flagellum (depends which T3SS, which flagellum), but to get from one to the other needs probably ten thousand new nucleotides in a specific constellation, and nothing close to that kind of change has been observed in the lab w/ fruit flies or E. coli. So, it's possible, but still improbable, like turning one of my programs into another via random change and selection, there are many similarities in all my programs, but it just would take too long. Possible is not probable, and unlike cosmological improbabilities, there's no anthropic principle to save this. Pointing out homologs still leaves the problem of tranversing a highly spiked fitness landscape, but if this is ever demonstrated on say, a new protein complex in E. coli, I'd say, you win (but more complex than moving a gene closer to a promoter region as in the citT).