An important way of assessing and improving the strength of an assay is to introduce perturbations into the system and analyze the outcome. The three major problems that the previous operators of this assay had were experimental failure due to detachment of the cell-laden agarose droplet from the well during normal operations, a low signal-to-noise ratio (SNR) due to migration on the BSA negative control, and inaccuracy in data analysis due to a soft boundary between the droplet area and the migration area. Each issue was addressed by altering key variables in the method until the assay was optimized.

The first issue that was addressed was the incessant detachment of the majority of the droplets, causing the experiment to fail. In an attempt to resolve this, the agarose droplet migration assay was performed on three different brands of polystyrene 96-well plate. The plates were the Corning tissue culture plate (Corning Inc., Corning, NY) that was used by the previous operator of this type of assay (Nigro, 2001), the FalconÔ tissue culture plate (Becton Dickinson Labware, Franklin Lakes, NJ) that was used extensively in our lab for migration assays (Clark et.al., 2002), and the Immulon 4 immunoassay plate. When the 1 μl of agarose was dropped onto the well in the Corning plate, it spread to an average diameter of 2.7 mm. Once the droplet had hardened, more than 50 % of the droplets would detach from the surface of the well during one of the rinsing steps, even when extreme caution was exercised. The since the gel spread out so that the contact angle was fairly small, the soft edge between the gel area and the migration area persisted. When the 1 μl of agarose was dropped onto the well in the FalconÔ plate, it spread to an average diameter 3.6 mm. The droplets were not well retained in these wells either. Due to low surface tension, the contact angle of the agarose droplet on the FalconÔ surface was the smallest, therefore making the distinction between the gel periphery and the inner boundary of the migration area difficult to determine. When the 1 μl of agarose was dropped onto the well in the Immulon-4 plate, however, it only spread to an average diameter of 1.5 mm. 99 % of the droplets were retained throughout the experiments and the extreme care needed when performing procedures of the agarose-droplet migration assay could be relaxed without negative consequences. The ease of experimentation was therefore increased using the Immulon-4 plates. In addition, the contact angle between the gel and the plate was large, giving a sharper edge distinction than the contact angles between the agarose and other aforementioned plates. This helped improve the accuracy of the data analysis as well.

The second issue that was addressed was to increase the SNR of the assay. As there was some cellular migration across the BSA in the negative control wells, albeit only around 1/10 of the migration in the FN wells, the concentration of this effort was on reducing this migration in the negative control. The first attempt at eliminating migration in the negative control was an exploration of a HD-BSA blocking step after the protein- coating step of the agarose-migration assay. In this study, the wells were incubated at 37 ºC with HD-BSA concentrations ranging from 0 mg/ml to 50 mg/ml for times spanning 0 to 3 hours. This HD-BSA step was followed by a 24- hour incubation period during which the wells were immersed in BSA containing either 20 mg/ml or 50 mg/ml BSA. It was determined that the HD-BSA step did not significantly reduce migration of the cells, and so that step was eliminated from the agarose-gel migration assay protocol. Cellular migration was not further inhibited by the increased concentration of BSA in the media so 20 mg/ml BSA was adopted in the assay protocol.

The reason why the cells have been able to outmigrate in the negative control wells is probably due to the cells producing and then migrating on their own fibronectin, cellular fibronectin. Therefore, attempts were made to limit the cellular production of fibronectin. The first such attempt was to reduce the concentration of cells within the agarose droplet. The SNR of the agarose- droplet migration assay was compared between cellular concentrations within the agarose of 3.3x10⁷ cells/ml, 1.1x10⁷ cells/ml, and 3.3x10⁶ cells/ml. Not surprisingly, it was found that the lowest concentration of cells yielded the least migration on the negative control. However, the lowest concentration of cells yielded less migration on the FN positive control as well, producing a lower overall SNR than did the medium concentration of cells. The inverse is true about the highest concentration of cells, which also yielded at lower SNR than did the medium concentration. 1.1x10⁷ cells/ml was therefore chosen as the optimal cellular concentration in the agarose droplet.

Agarose-droplet migration assays were performed with and without the presence of PDGF. PDGF is a powerful stimulus for fibroblast migration and fibronectin production. As such, its presence or absence from the media in the assay undoubtedly influences the fibroblast migration in both the FN positive control and the BSA negative control. Although experiments carried out in a PDGF-free environment had much lower migration within the negative control wells, the SNR was greater when PDGF was present. This was due to the robust stimulation of fibroblast migration in the presence of PDGF.

A further attempt to limit the cellular production of fibronectin, thereby reduce the cellular migration on the negative control, was attempted by starving the cells by replacing the DMEM with SF-DMEM as their media for 2 hours prior to their addition to agarose. It has been shown in our lab that the deprivation of serum to a fibroblast in culture diminishes the responsiveness of the cell, reducing fibronectin synthesis. Additionally, other cells were exposed to 25 µg/ml cycloheximide for 2 hours prior to their addition to agarose. Cycloheximide is an antibiotic that inhibits protein synthesis in cells by blocking translation of messenger RNA. The exposure of the fibroblasts to this compound should inhibit them from producing fibronectin. The results of this attempt at increasing SNR are summarized in figure 2.

Figure 2. SNR of agarose-droplet migration assay using cells that are serum starved, cells that are inhibited by cycloheximide, and cells that are not starved or inhibited (control). The control group has the highest SNR.

Although both the starving as well as the cycloheximide inhibition of fibroblasts both clearly reduced migration within the BSA negative control group, they also greatly reduced migration within the FN positive control group. It has been shown in our lab (preliminary data) that cycloheximide interferes with the PDGF- induced receptor activation on fibroblast surfaces. Thus it follows that the treatment of cells with cycloheximide inhibited migration on FN as well. Additionally, the cycloheximide- inhibited, as well as the starved cells, were more susceptible than the control cells to cellular damage due to environmental conditions, as their metabolic activity was dampened. Thus, the conditions of the assay preparation may have damaged a larger percentage of the cells in the starved and the cylcoheximide- inhibited conditions than in the normal condition. This cellular damage might have further inhibiting migration on FN.

The third issue that was addressed was an improvement in the resolution of the edge between the outer limit of the gel and the inner rim of the cellular migration during data analysis. To accomplish this a couple alternate stains were evaluated. With crystal violet in boric acid (pH 9.0) as the standard, crystal violet in (2-[4-Morpholino]-Ethane Sulfonic Acid) with 150mM NaCl (MES) (pH 6.0) and Hanks balanced salt solution (Ponseau stain) were assessed. Neither of these stains was able to increase the resolution of the gel- migration boundary. In the end, I was able to increase the resolution of that boundary greatly by adding additional washing steps and by removing the final rinse of PBS from the well prior to image capture. These simple steps reduced background noise and distortion of the image.

The method for measuring the migration area contains intrinsic variability due to human error. To determine this error, measurements of the same image were made repeatedly (n=5), independently drawing the boundaries each time, and compared. Analysis of the results gave a standard error of only 2 %. This allowed us to disregard the human intrinsic error in the data analysis of the agarose-droplet migration assay.

Once the type of plate to be used was chosen, the protein-coat concentration necessary for optimal fibroblast migration in the two-dimensional agarose- droplet migration assay was found. The determining experiment run was an intact fibronectin concentration study. As intact plasma fibronectin was to be used as the positive control to assess the migration potential of the recombinant fibronectin fragments, it was important to achieve a concentration that allowed relatively robust migration However, it had to be below the threshold of migration “saturation,” where any increase in protein concentration will not yield any additional migration. This was to provide the largest possible range of fibronectin fragment migration values that can be accurately compared to migration on intact fibronectin. The BSA negative control was coated at 100 µg/ml.

The commercially available human plasma fibronectin that was used in these series of experiments was acquired from Chemicon Inernational and verified as intact through SDS-PAGE, as shown in Figure 3.

Lanes 1 2 3 4 5 6 7

FN

BSA

66.3 kDa

220 kDa

Figure 3. Coomasie blue- stained 4-20% Tris-HCl graduated protein gel (Bio-Rad) from SDS-PAGE of FN. Lane 1 is the Gibco BRL protein ladder (Bands 220, 130, 90, 70, 60, 40, 30, 20, 15, and 10 kda) lane 2 is 10 mg intact fibronectin and lanes 3-7 are BSA standards (20 mg, 10 mg, 5 mg, 2.5 mg, 1.25 mg). This gel shows that FN is intact with little or no degradation

Data from multiple concentration experiments were pooled together by normalizing the migration areas to 100 mg/ml FN and summarized in Figure 4. .

Figure 4. Agarose-droplet migration assay dose response curve on intact fibronectin. The study shows a positive correlation of fibroblast migration to FN concentration. However, this correlation disappears at around 150 mg/ml. ANOVA tests show that the variability between FN>=1 and BSA was not random (p<0.02). However, ANOVA tests show that the variability between FN=150, FN=200, and FN=300 was random and insignificant (p> 0.5).

From the dose response curve it was determined that 100 mg/ml FN coating would be optimal with the particular experimental setup used in these studies. This point was chosen as it is near the maximal migration limit of fibroblasts, but not yet at the saturation point. These characteristics were important to most accurately measure the relative migrations between FN fragments and intact FN. As a negative control, BSA was coated around the droplet at a concentration equivalent to that of the positive control (100 mg/ml). This concentration is applicable as a negative control even for fragments coated at lower concentrations. This is due to a negative control study, showing there is no significant difference between coating the wells with 100 mg/ml BSA to coating the wells with no BSA at all (PBS), as described in Figure 5. In addition, it can be inferred from this data and from additional experiments that any coating concentration of BSA between 0 mg/ml and 100 mg/ml is equivalent in its lack of support of fibroblast migration as well.

Figure 5. Agarose-droplet migration assay dose response on BSA. This study shows the equivalency of coating at 100 mg/ml BSA and pure PBS. ANOVA test confirmed that the variability between 100 mg/ml BSA and 0 mg/ml BSA was not random (p> 0.75)

There were minor variations in gel size between wells in many of the experiments. To ensure that these differences in gel size did not influence the statistical outcome of the experiments, additional manipulations were carried out to normalize migration areas to the variable sizes of the agarose droplets. In this independent calculation, the area was determined according to the formula (Total Area - Area of Droplet)/ Area of Droplet. This formula gives the migration as a percentage of the droplet size. The statistics were carried out the same way as described in above. When the data is normalized to take the size of the gel into account, the outcome of the experiments remains essentially unchanged.

The commercially available 110K human plasma fibronectin digest that was used was verified as intact through SDS-PAGE, as shown in Figure 6.