Background

The periplasm of Gram-negative bacteria is a quite remarkable cellular compartment. This space, incarcerated between the inner and outer bacterial membrane, contains proteins at an extraordinarily high concentration exceeding 300 mg ml⁻¹ (Oliver, 1996) and, in the absence of cellular sources such as ATP, functions almost energetically independent from its cytosolic counterpart. So far, structural knowledge about periplasmic proteins has been exclusively obtained using purified proteins isolated from their native environment. Thus, any structural and functional influence this special environment might impose on a protein are lost during purification. Efforts to study periplasmic proteins in-situ using biophysical methods such as in-cell NMR spectroscopy have been hampered by the low volume ratio of the periplasm, which contributes only 5-10% to the total bacterial volume (Brass et al., 1986).

We recently described a new method to characterize proteins within bacterial outer membrane vesicles (OMVs) using solution-state NMR spectroscopy (Thoma and Burmann, 2020). OMVs, which are naturally released by Gram-negative bacteria, contain periplasmic content in their lumen and can be non-destructively purified from bacterial growth cultures (Chutkan et al., 2013; Schwechheimer and Kuehn, 2015). Thus, OMVs represent an excellent platform to study periplasmic proteins, enriched in the luminal space, together with their natural environment. To be amendable for NMR studies, OMVs are uniformly labeled with stable isotopes (²H and ¹⁵N) and prepared in large quantities. Our approach proved particularly useful for the study of soluble periplasmic proteins within this environment, as the membrane components of OMVs remain largely invisible to solution-state NMR, due to slower molecular tumbling rates.

One key factor for obtaining large amounts of luminally enriched OMVs is the use of a suitable expression strain. Widely used expression hosts such as E. coli BL21(DE3) do not produce sufficient amounts of OMVs to be feasible for these types of experiments (Thoma et al., 2018). In contrast, strains lacking the major outer membrane protein OmpA have been shown to have a 26-fold increased OMV production (Schwechheimer et al., 2014). Moreover, deletion of OmpA omits possible background signals originating from its soluble peptidoglycan binding domain in NMR experiments (Thoma and Burmann, 2020). In our experimental set-up we relied on the strain BL21(DE3)ompA, which was kindly provided by Guido Grandi (Fantappiè et al., 2014) and which reproducibly yielded high amounts of luminally enriched OMVs. However, other alternatives such as BL21A are available (Meuskens et al., 2017) or could easily be tailor-made by deletion of the ompA gene in the expression strain of choice.

Another critical factor for obtaining large amounts of suitable OMVs is the time point, at which the bacteria are removed from the growth media. Bacterial growth should be maintained as long as possible, since the amount of OMVs present in the culture supernatant correlates directly with the cell density and consequentially with the length of the growth period. However, OMVs should be collected before bacterial growth ceases and cultures enter the stationary phase, because this will substantially increase contaminations with cytoplasmic proteins, inner membranes, and cell debris resulting from stress-induced explosive cell lysis (Turnbull et al., 2016; Thoma et al., 2018). Depending on whether cells are grown in H₂O or D₂O based media this stage is typically achieved after 3 to 5 h, respectively. Since bacterial growth is not only retarded by the use of D₂O but also underlies various influences such as the strain and plasmid used for expression, it is advisable to determine the growth behavior of the bacterial cultures prior to OMV preparation. Whereas deuteration is not strictly necessary for small to medium sized proteins (< 25 kDa), luminally enriched OMVs prepared from cells grown in D₂O based media typically yield NMR spectra of higher quality due to a decrease in background signals and a smaller NMR line-width (Thoma and Burmann, 2020).