In an attempt to visualize the difference between heat-killed intact cells and released inclusion bodies, the following stains were tested: PI, BacLight Green, and SYTOX Green. The flow cytometer was set to capture FS, SS, and fluorescence data from a variety of heat-killed E. coli samples. Heat-killed intact samples stained with PI exhibited the expected large shift in fluorescence intensity ((Figure 2)A) relative to unstained control (not shown). Staining of the same heat-killed cells, further subjected to one or two rounds of homogenization, or staining of granules isolated from the heat-killed cells yielded successively lower fluorescence signal ((Figure 2)A), although all samples yielded a brighter signal than unstained controls. Because intact samples gave the brightest fluorescence signal while isolated granules gave the weakest, it appeared that PI bound both cells and granules and that lower fluorescence intensity indicated increased homogenization and granule release. To test this hypothesis, PI-stained homogenized material was examined by epifluorescence microscopy ((Figure 3)A). These images verify that granules as well as cells take up nuclear stain. It is possible that the observed fluorescence of the inclusion bodies could, in theory, be caused by contamination with nucleic acids either as a result of entrapment during expression or as a result of sticking to the inclusion bodies following cell disruption. To test this, we have treated the granules with a DNase preparation and did not observe a significant decrease in the fluorescence (data not shown). Therefore, it is more likely that the dye molecules diffuse into the submicron pores in the inclusion bodies and become trapped without binding to nucleic acid. As per the fluorescence histograms ((Figure 2)C and (4)) and as illustrated in (Figure 3)A, the relative amount of fluorescence is substantially lower than that caused by the binding of the dye to nucleic acids within intact cells. As noted above, the relative fluorescence of SYTOX is ∼500-fold greater when bound to nucleic acids. As per the histograms, this is approximately the difference observed between the fluorescence of the intact cells versus inclusion bodies, suggesting that dye entrapment is the mechanism.

Figure 2.

Figure 3.

Based upon these experiments, it appeared that flow cytometric analysis in conjuction with nucleic acid staining could distinguish isolated granules from intact heat-killed E. coli. Additionally, based on histogram results, this staining seemed sensitive enough to detect sample differences in the intervening homogenization rounds leading to granule isolation.

While PI staining proved promising, there was a need to further resolve the signal overlap seen among intact E. coli, homogenized E. coli, and isolated granule samples. Therefore, several other stains were tested (BacLight Green and SYTOX Green) to determine whether greater resolution could be achieved. In comparison with PI, sample resolution did not improve when stained with BacLight Green ((Figure 2)B). However, sample resolution did improve dramatically when stained with SYTOX Green ((Figure 2)C and (3)B). Of the several concentrations of the stain tested (2, 4, 20, and 30 µL/mL), 4 µL/mL yielded the best results.

Additional experiments (data not shown) demonstrated that the flow cytometric method using SYTOX Green is capable of distinguishing free inclusion bodies from unruptured (intact) cells and cell debris. Purified inclusion bodies recovered from the manufacturing process stream following centrifugation and washing were used as a standard in the experiments that were performed. We have verified that these inclusion bodies are aligned with the histogram peaks for fluorescence and also populate the identified region in a dot density plot of FSS versus fluoresence. The inclusion body region is clearly distinguishable from both intact cells and debris allowing the events to be enumerated. Furthermore, we have demonstrated the identity of the inclusion body peak in the fluorescence histogram by spiking the samples with the purified inclusion bodies and observing the number of events associated with this peak increase accordingly.