Understanding the mechanisms of HPV-related carcinogenesis: Implications for cell cycle dynamics

Abstract

The role of human papillomavirus (HPV) as a causative agent for epithelial cancers is well-known, but many open questions remain regarding the downstream gene regulatory effects of viral proteins E6 and E7 on the cell cycle. Here, we extend a cell cycle model originally presented by Gérard and Goldbeter (2009) in order to capture the effects of E6 and E7 on key actors in the cell cycle. Results suggest that E6 is sufficient to reverse p53-induced quiescence, while E7 is sufficient to reverse p16^INK4a-induced quiescence; both E6 and E7 are necessary when p53 and p16^INK4a are both active. Moreover, E7 appears to play a role as a "growth factor substitute", inducing cell division in the absence of growth factor. Low levels of E7 may permit regular cell division, but the results suggest that higher levels of E7 dysregulate the cell cycle in ways that may destabilize the cellular genome. The mechanisms explored here provide opportunities for developing new treatment targets that take advantage of the cell cycle regulatory system to prevent HPV-related cancer effects.

Keywords: Cancer; Cell cycle; HPV; Mathematical modeling.

Figure 1:

Diagram of the network structure for the model, showing the original Goldbeter modules, as well as the additional p53, INK4, E6, and E7 components. Modules based on those in [1] are shown as grey shaded regions as follows: 1) AP1/E2F module, 2) cyclin D module, 3) cyclin E module, 4) cyclin A module, 5) cyclin B module. Newly added p53, INK4, E6, and E7 modules are shown in blue, with dashed ovals indicating viral proteins and dotted circles indicating growth factor. Note that E7 is shown in the diagram twice for ease of visualization.

Figure 2:

Timecourse data from the baseline model with no additional modules active. This result is consistent with that shown in Goldbeter [1] figure 2B. The color-bar above the figure indicates cell cycle phases: yellow for G1, blue for S-G2, and red for G2-M. The transition between G1 and S-G2 was defined to be peak p27, the S-G2-to-G2-M transition was defined at peak cyclin A complex, and the G2-M-to-G1 transition occurred at peak Cdc20a.

Figure 3:

Timecourse results for cyclin B/CDK1 complex. In each panel, the Control simulation is the baseline model without any additional modules added, and each “+” indicates which modules are added to the baseline (“Control”) model for each timecourse. In panel (A), activating INK4 halts the cell cycle, which can then be restarted (with a faster period) by activating E7; in (B), p53 activation halts cell cycle oscillations, which cannot be restored with E7 alone; in (C), E6 activation restores the cell cycle in the presence of p53, and addition of E7 results in a faster cell cycle; in (D), cell cycle restoration by E6 can have varying effects on the period depending on E6 level; and in (E), when p53 and INK4 are both activated, both E6 and E7 are needed to restore the cell cycle. These panels taken together illustrate which combinations of models result in periodic behavior (dividing cell) vs. a constant steady state (quiescence), and also visualize changes to the cell cycle period In each panel, time is measured in hours (with the initial 400 hours of simulation omitted to focus on the steady state behavior), whereas concentration levels are all relative.

Figure 4:

Timecourse results for cyclin E (black), cyclin A (red), and cyclin B (green) complexes, with each of the twelve possible combinations of modules (here, “+X” indicates that module X is active) (expanding on Figure 3 to compare multiple Cyclins and explicitly track the cell cycle phase). Above each timecourse window is a bar indicating progression through the cell cycle; yellow indicates G1 phase, blue indicates S/G2 phase, and red indicates G2/M phase. As before, time is measured in hours (with the initial 400 hours omitted), whereas concentration levels are relative.

Figure 5:

The ‘+p53’ panel from Figure 4, expanded to view the full time course. Addition of p53 results in low-amplitude oscillations that are significantly slower than the baseline model behavior, likely corresponding to either non-cycling or an extended period of cell cycle arrest. As before, time is measured in hours, whereas concentration levels are relative.

Figure 6:

Heatmap depicting the peak level of cyclin E/CDK2 complex in the model with the p53 and INK4 modules active, with varying levels of E6 and E7 added to the system. Various regions with qualitatively distinct behavior are labeled, with representative examples of their behavior given in Figure 7. Black dots indicate points with abnormally high peak Cyclin E/CDK2 level (orders of magnitude higher than elsewhere) likely indicative of integration error. We observe that varying levels of E6 and E7 produce markedly different levels of Cyclin E complex, with intermediate levels of E6 and E7 producing the highest levels of Cyclin E complex.

Figure 7:

Sample timecourses from the labeled regions in Figure 6. The color-bar above each figure indicates cell cycle phases: yellow for G1, blue for S-G2, and red for G2-M. As before, time is measured in hours (with the initial 400 hours omitted), whereas concentration levels are relative.

Figure 8:

Exploring E7 as a growth factor substitute. Here are presented timecourse results for cyclin E, cyclin A, and cyclin B complexes, under the following conditions (left to right, top to bottom): (A) In the baseline model with no growth factor present. (B) With no growth factor present and 0.2 units of E7. (C) With 100 units of growth factor present. (D) With no growth factor present and 5 units of E7. The color-bar above each figure indicates cell cycle phases: yellow for G1, blue for S-G2, and red for G2-M (no bar is shown for the first panel as the concentrations are all constant). As before, time is measured in hours (with the initial 400 hours of omitted), whereas concentration levels are relative.

Figure 9:

Examining growth factor and E7 impact on Rb. Here are presented timecourse results for the five retinoblastoma protein states, in: (A) the baseline model with 1 unit of growth factor, (B) the baseline model with 100 units of growth factor, (C) the baseline model with no growth factor and 0.2 units of E7, (D) the baseline model with no growth factor and 5 units of E7. The color-bar above each figure indicates cell cycle phases: yellow for G1, blue for S-G2, and red for G2-M. As before, time is measured in hours (with the initial 400 hours omitted), whereas concentration levels are relative. The dynamics shown here illustrate how E7 replicates growth factor effects, but in a non-bottlenecked way.