Recently, several studies on Staphylococcus aureus (S. aureus) have been hindered due to poor performance from assays using the µSiM-Canalicular Array (CA). As discussed here, we attempted to improve assay performance by coating our membranes with fibronectin. This did not work well and instead left us in a state of confusion as we did not have much information to go off of in order to improve the performance of the assay. As a result, we decided it would be best to try to visualize the process of bacterial division in µSiM-CA devices loaded with GFP tagged S. aureus using the confocal microscope. Our approach and results are shown below.

Study Goals

Our goal was to accomplish two things. First we wanted to try to see if we could use confocal z-stacks to determine where S. aureus was with relation to the membrane (on top of, inside of, or on the bottom of). Second, we wanted to determine at which time point S. aureus divided from one side of the membrane (the apical side) to the other side (the basal side).

Approach

To accomplish these goals, we set up an experiment using two µSiM-CA’s containing 0.5 µm microporous membranes and two µSiM-CA’s containing nonporous membranes. By using both types of membranes, we hoped to be able to more accurately determine where S. aureus was with relation to the porous membrane by comparing it’s position to that of the S. aureus on the nonporous membrane. Since we were unsure of when S. aureus divided through the membrane (or if there was even a consistent time with which this occurred), we decided to plan on taking confocal images every hour for 12 hours straight. By doing so, we hoped to capture close estimates as to when S. aureus divided to the basal side of the device.

Preparing Devices

µSiM-CA devices were prepared by wetting the bottom channel with PBS, filling the well with 100 µL of PBS, and blocking a port with a sticker seal. Membrane permeance was tested by injecting 40 µL of PBS in the open port and ensuring the injection all went through the membrane while facing little resistance. After passing our permeance check, we aspirated the well and filled it with 100 µL of GFP tagged S. aureus. After loading devices, one of the non-porous membranes broke so the experiment was only conducted on three separate µSiM-CA’s containing 2 porous membranes and 1 nonporous membrane. Devices were sealed with PDMS blocks (~5 mm thick) in order to allow them to be imaged in the microscopy core while maintaining gas permeance. To create an adequate seal, PDMS blocks were laid on top of the µSiM-CA after wetting and the interface between the PDMS bottom and the acrylic top of the device was sealed with nail polish. Upon sealing, devices were stored in petri dishes for transport over to river campus.

Imaging

Images were taken with the 60X water immersion objective and the 1.5X magnifier on the base of the confocal microscope. The 488 nm laser was used with the 525 nm filter. 

Results and Conclusions

Determining where S. aureus is with relation to the membrane

Images were taken across various time points at a somewhat regular interval for each of the three devices based upon the length of imaging and availability of the scope. As time progressed, it was quickly realized that the dry environment in the microscopy core was drying out the devices at different rates (presumably due to how well they were sealed). Once the bottom channel of a device dried out, it became un-imageable. Both porous devices dried out at different rates, but the nonporous device never dried out. Because of this, only the nonporous device has a full set of images up to 12 hours. Side by side, images of both a top-down view (each main panel) and side profiles of membranes (to the right and below each main panel) are shown in Figure 1. For the porous membranes, the same fluorescence images are shown overlaid with DIC images in Figure 2 to give a better idea of where the bacteria is relative to the membrane.

From Figures 1 and 2 we can see it is easy to get an idea of bacteria density on the membrane, but it is very challenging to determine if bacteria is on top of, inside of, or on the bottom of the membrane. When looking at all three µSiM-CA devices from the side, there are few immediately noticeable differences in the position of the bacteria on the membrane. Our hope was that the nonporous membrane would give us a defined line with which bacteria are held at and the porous membranes would not have such a defined line since bacteria are able to enter the membrane. In Figure 1, comparing all three µSiM-CA devices shows that there is a slightly more defined line on the nonporous membrane, but as there is in the porous membranes, there are a few things that stick out below this line. There are certainly more of these things sticking below the lines in the porous membranes, but the presence of a few in our nonporous control make it difficult to say whether or not these things are present because they are in fact bacteria within the pores or if they are simply artifacts created by the confocal when stitching our image together. In Figure 2, I try to develop a better sense of where these protrusions are by overlaying DIC images with our fluorescence images. The hope was DIC would give us more a defined line indicating where our membrane was in relation to the bacteria on the membrane. Unfortunately, the DIC images of the nonporous membrane were of bad quality, so I was unable to overlay the nonporous images. Nonetheless, while somewhat more interesting, all the overlays really did were reaffirm that where we thought the membrane was located was in fact where the membrane was located. On both porous membranes, the lines visible in fluorescence imaging (indicating where the bacteria are accumulating) are exactly where the DIC membrane lines are as well. The overlay did not help distinguish bacteria position as it too suffered from stitching artifacts. Ultimately what we came to realize is the current setup on the confocal scope does not have the resolution to definitively tell us whether bacteria is on top of, inside of, or on the bottom side of our membranes.

While our side profiles were challenging to interpret, our top down views were not. These views showed that at first, no matter if the membrane was porous or nonporous, there was not a noticeable difference in density of bacteria on the membrane. As time went on, the density of the µSiM-CA device containing a porous membrane that had bacterial transmission seemed to remain the same while the other two devices’ bacterial density increased substantially. Above both the nonporous membrane and the porous membrane that did not have bacterial transmission, thick mounds of bacteria can be seen. The bacteria above the porous membrane that had bacterial transmission did not seem to do this, maintaining more of a monolayer as time goes on. This potentially could provide clues as to why some porous membranes containing µSiM-CA devices have bacterial transmission while other do not.

Determining when S. aureus divides from one side of the membrane (the apical side) to the other side (the basal side)

After initially taking confocal images of just the membrane, I decided at the ~4 hour mark to begin to take large z-stacks imaging both the membrane and the bottom channel at the same time. Using this approach, I could show the bacteria above the membrane and any bacteria that may have gotten to the bottom channel (as well as any in between). The resulting images are shown below in Figures 3 and 4.

Figure 3 shows that predictably, no bacteria made it to the bottom channel of the µSiM-CA device containing a nonporous membrane. Since this result was expected to save time while imaging, I only took stacks for the nonporous membrane containing device at three different time points. Interestingly, Figure 3 also shows that bacteria only made it to the bottom of one of the µSiM-CA devices containing a porous membrane. Stack sizes vary as I increased the distance I searched for bacteria from the membrane in case thickness differences or drying artifacts affected where bacteria lay. Despite changing my probe depth, I still did not see any bacteria in the first porous membrane containing device. This affirmed the conclusion that bacteria only made it through the second device containing a porous membrane. Unfortunately, because the devices dried out, I was unable to run the experiment indefinitely until bacteria did go through the first porous membrane containing device. All we learned from this attempt with regard to when bacteria gets through the membrane is what we were already seeing in our CFU analyses, despite being seemingly similar, some devices have bacteria go through the membrane within 6 hours and other devices do not.

Focusing on the bacteria that did get through the membrane in the second µSiM-CA device which contained a porous membrane, some interesting patterns arise. Figure 3 shows that the bacteria always seems to be aggregated and in some cases it can actually be seen forming what look to be colonies. It was also good to see bacterial density increase over time in the bottom channel as bacteria continue to divide.

Proof of drying artifacts

As a sanity check, I made Figure 5 to show clear evidence that the devices were in fact drying out. All three images within the Figure show how after drying, all that is left on the membrane and in the bottom channel of the µSiM-CA devices are bubbles of water. It was interesting to see that even despite a majority of the fluid evaporating in the bottom channel, bacteria could still be seen within the left over droplets of liquid in the bottom channel of the device that it was able to divide into the bottom channel of.

Summary and Future Directions

Altogether, this experiment did not completely answer both of our initial questions, but it did allow us to gauge our new technique and provide us with data that we could use to craft a plan moving forward. A summary of these findings are below.

Technique

1. This technique does not currently provide enough detail to definitively distinguish between bacteria that is on top of, inside of, or on the bottom of the membrane
2. If not attentive to local humidity, the devices will dry out within 12 hours
3. The technique does provide a great view of bacteria on the membrane, allowing us to observe bacterial density and the presence or lack of a monolayer
4. It also easily shows if bacteria is in the bottom channel

Data

1. There seemed to be a difference in bacterial density on the membrane between the two porous devices that did and did not have transmission of bacteria into their bottom channels
2. The membrane that did have transmission across it maintained a lower bacterial density over time
3. This membrane also seemed to have something closer to a bacterial monolayer than the porous membrane that did not have bacterial transmission

Moving forward

Once again seeing variability in an assay using the µSiM-CA, we decided it would be best to take a step back and try to nail our assay protocol down (as we suspected this could be a potential source of assay variability). Using a combination of our new technique and the CFU analysis which is done in the Gill and Dunman labs, the next thing we aimed to do was standardize our assay protocol between each lab group and establish a baseline performance with which we could improve upon. From there we sought to make incremental changes on the protocol, observing their affect on assay performance until we got to a point where we were satisfied (>90% transmission with positive controls).