Yes, SacB can indeed cause toxicity in E. coli even in the absence of sucrose, especially when placed under strong promoters, which appears to be the case in your experiment. While SacB is classically known to be toxic in the presence of sucrose due to the production of levan polymers by levansucrase activity, its overexpression alone, driven by high-expression promoters, can impose a significant metabolic burden on the cell. This can result in cellular stress due to misfolded proteins, membrane disruption, or saturation of the protein secretion machinery, particularly since SacB contains a signal peptide for secretion in its native Bacillus host. E. coli is not well-equipped to handle this kind of overexpression, especially if the protein accumulates in the periplasm or forms aggregates in the cytoplasm.

The consistent recovery of clones with frameshift mutations in the SacB gene or the promoter region strongly suggests that the bacteria are selectively mutating these sequences to eliminate SacB expression, thereby alleviating the associated toxicity. This type of negative selection is common when cloning toxic genes under strong regulatory elements. The low transformation efficiency you observed also supports the idea that cells expressing SacB at high levels are either dying or growing poorly, reducing their representation among the colonies you recover.

Even though you confirmed that sucrose is not present in your media or buffers, trace amounts can sometimes be found in complex media components like yeast extract or tryptone. However, your observation that SacB under its endogenous promoter does not cause toxicity further supports the idea that the issue lies with overexpression rather than sucrose-dependent toxicity.

In summary, high-level SacB expression, even without sucrose, can be harmful to E. coli and lead to selection for inactivating mutations. To reduce this effect, you might consider using a weaker promoter, an inducible system, or tightly repressing expression during the cloning steps (e.g., by using glucose-containing media to repress certain promoters, or using strains with tighter repression systems).

I would agree with the above answer. I haven't heard of this happening specifically with SacB, but it would make some amount of sense to me as SacB has a signal peptide for the Sec pathway and is exported into the periplasm when expressed in Gram-negative bacteria. I am not 100% sure where I remember this from so don't quote me, but I believe over-expressed proteins with signal peptides can sometimes "jam" the Sec pathway.

Unfortunately, the selective toxicity of SacB in Gram-negatives requires it to be exported into periplasm as far as I know, so the suggestions above for using weaker/inducible promoters are probably your best bet.

Side note:

If these constructs are destined for non-E. coli bacteria and you can't swap the strong promoters for that reason, you could try adding lacO sites around the promoter. That way the endogenous E. coli LacI protein can repress the promoter to some extent when passaged in E. coli, but when transferred to a species without a LacI homolog the promoter should be constitutive again. If you need to go this route, I would recommend using the lacOid version of the lac operators since it results in stronger binding:

(see: Article Quality and position of the three lac operators of E.coli de...

Ideally you have one lacOid site immediately upstream of the -35 site and one overlapping the -10 site, but even flanking the rough promoter region with lacOid sequences can repress expression to some extent via steric hinderance. Due to the copy number differences between the single chromosomal lacI and the multicopy plasmid, this wouldn't result in total repression but hopefully to prevent toxicity during routine passage. There's also lacIq strains of E. coli that overexpress LacI for further repression if that's needed.