TI INSTRUMENTO CARICOM (CSME) SIECA CAN MERCOSUR

As clusters of KrasG12D cells are eliminated and as EphA2 is required for the decrease in RFP area in vivo the requirement of EphA2 for the elimination of mutant clusters was investigated.

First it was necessary to establish whether loss of functional EphA2 affects RFP cluster density over time (Figure 3.8A). As expected, in the RFP-Eph model RFP clusters occur at a similar density to controls at 7 dpi (RFP-Eph = 38.3 mm-

2 _{± 12.76 Vs. Control = 37.5 mm}-2 _{± 5.6). The loss of functional EphA2 also did}

not result in changes to the number of clusters over time in RFP-Eph mice (7 days = 38.3 mm-2 Vs. 35 days = 36.2 mm-2 ± 11.0). Furthermore, the distribution of clusters of different sizes in RFP-Eph mice is similar at 7 and 35 dpi (Figure 3.9A). In summary, this suggests that loss of EphA2 alone does not lead to changes to total number of clusters or clusters of a specific size.

Next the effect of EphA2-deficiency on KrasG12D clusters was investigated. At 7 dpi, the density of KrasG12D clusters in KC-Eph mice was similar to controls and KC (Figure 3.8A). However, at 35 dpi significantly more RFP clusters were observed in KC-Eph mice compared to KC (KC = 18.1 mm-2 _{± 5.8 Vs. KC-Eph =}

44.3 mm-2 _{± 11.8, Welch’s unpaired t-test, p = 0.0048). This suggests that loss}

of functional EphA2 is required for the observed decrease in clusters. As the density of clusters in KC-Eph animals not only increased but returned to similar levels as controls this suggests that EphA2-deficiency inhibits KrasG12D cell loss and is sufficient to completely abrogate the loss of KrasG12D clusters.

Following this the density of clusters was broken down by cluster size to further elucidate the effect of EphA2-deficiency on KrasG12D clusters. This demonstrated that in EphA2-deficient animals the density of KrasG12D clusters of all sizes was similar over time (Figure 3.9B). Analysis of individual cluster sizes up to 5000 µm2 _{showed no significant change in density between 7 and 35 dpi}

(Welch’s t-test and Mann-Whitney test, all p >0.05). Moreover, a significant increase in KrasG12D clusters up to 4200 µm2 _{in size was observed KC-Eph mice}

when compared with KC mice (Table 3.2). This supports the hypothesis that EphA2-deficiency leads to the retention of clusters of KrasG12D _{cells up to 4200}

observed but the density in an EphA2-deficient model returned to a similar level as controls (Figure 3.9C), with no significant difference between the density of clusters up to 5000 µm2 _{in KC-Eph mice compared with Kras wild-type control.}

In summary, EphA2-deficiency alone does not alter the distribution of clusters of different sizes over time. However, EphA2-deficiency leads to a significant increase in the number of KrasG12D clusters present at 35 dpi up to 4200 µm2 _{in size. Furthermore, the number of clusters in KC-Eph mice at 35 dpi}

is similarto Kras wild-type controls. This suggests a size threshold of around 4400 µm2 _{below which Kras}G12D _{clusters surrounded by normal neighbours are}

Figure 3.8: EphA2 is required for the elimination of mutant clusters within the context of a normal epithelium. (A) At 7 dpi there is no significant difference in cluster density between control (Pdx1-CreERT; Rosa26LSL-RFP, n = 17892 clusters), KrasG12D _(Pdx1-CreERT_{; LSL-KrasG12D; Rosa26}LSL-RFP_{, n = 15263 clusters)}

KrasG12D;EA2 (Pdx1-CreERT; LSL-KrasG12D; Rosa26LSL-RFP; EphA2tm1Jrui,n = 13422 clusters) and EA2 (Pdx1-CreERT; Rosa26LSL-RFP; EphA2tm1Jrui,n = 16942 clusters). This allows direct comparison of later time points between genotypes. (B) After 35 dpi the number of Kras WT clusters on a background of EphA2-defiency (EA2) is not significantly different to control or KrasG12D; EA2. There are significantly more KrasG12D clusters in EphA2-deficient mice at 35 dpi. than KC mice. With the number of clusters in Kras EA2 similar to control this suggests loss of EphA2 completely inhibits normal neighbour driven elimination. Data represents mean ± s.d. with each point representing a biological replicate. Cluster density was compared using Welch’s unpaired t-test. P<0.001 = **

Figure 3.9: Elimination of small clusters of KrasG12D cells requires EphA2 in vivo. (A) The distribution of Kras WT clusters in a model of EphA2 loss of function, EphA2 (Pdx1-CreERT; Rosa26LSL-RFP; EphA2tm1Jrui), remains relatively changed between 7 (n=16942 clusters) and 35 dpi (n=23305). (B) The density of KrasG12D clusters on a EphA2 LOF background, KrasG12D _{EphA2 (Pdx1-Cre}ERT_{; LSL-KrasG12D; Rosa26}LSL-RFP_; EphA2tm1Jrui) was also similar at 7 (n=13422 clusters) and 35 dpi. (n=18402 clusters). (C) KrasG12D clusters in EphA2-deficient mice appear at similar levels to Kras WT controls (n=12967 clusters) at the same time point, suggesting EphA2 is required for the elimination of small clusters from the pancreas. Data represents mean ± s.d. (shaded

p-value Cluster size (mm2) Kras EA2 35 days Vs. KrasG12D

35 days Fold change

KrasG12D clusters analysed (n=4) KrasG12D EA2 clusters analysed (n=5) 200 0.0094 3.1 1035 3686 400 0.0125 2.9 478 1951 600 0.0254 2.7 378 1462 800 0.0057 2.9 269 1053 1000 0.016 2.3 278 897 1200 0.0211 2.3 202 687 1400 0.0313 2.1 214 604 1600 0.0047 2.2 183 503 1800 0.0133 2.4 164 494 2000 0.0823 1.7 170 408 2200 0.0603 1.8 135 334 2400 0.0303 2.1 120 331 2600 0.0168 2.1 122 317 2800 0.0318 1.8 106 274 3000 0.069 1.7 103 226 3200 0.139 1.7 93 227 3400 0.0043 3.1 79 218 3600 0.0766 1.7 79 183 3800 0.0041 1.9 68 157 4000 0.0138 1.8 79 161 4200 0.0091 1.8 63 143 4400 0.0584 1.8 62 134 4600 0.7836 1.1 77 121 4800 0.2017 2.0 67 123 5000 0.0643 1.9 61 131 5200 0.2931 1.3 60 105

Table 3.2: Table of p-values for EphA2-deficient and KC mice at 35 dpi. Table listing p-values calculated to test for statistical difference in relative loss RFP clusters per cluster size bin utilising either a Welch’s unpaired t-test or a Mann- Whitney t-test. Red = p-value <0.05. The analysis suggests KrasG12D expressing clusters of less than 4600 µm2 _{are significantly increased in an EphA2-deficient}

3.2.9 Tissue-wide homeostasis is unaltered by mosaic KrasG12D expression