Analysis of the EGFR and the ROS1 Projects

Application to the EGFR Pro ject

This oncology project focuses on the inhibition of the epidermal growth factor receptor (EGFR), which has been identified in many human epithelial cancers, colorectal, breast, pancreatic, non-small-cell lung, and brain cancer (Shaib et al., 2013).

For this project, of the 55 fingerprint features that demonstrated differential effects on the primary bioassay, FF-442307337 came out first based on a

TABLE 16.1

Subclasses of genes using a hypothetical example. If both a and в are significantly different from zero, the correlation between the gene expression and bioactivity is present but in contrast with scenario (a), the gene expression in scenario (b) is correlated with the bioactivity variable only due to the effect of the fingerprint feature, hence its adjusted association is zero, pj = 0. From the point of view of the structural optimization in the early drug development, the association observed in (b) is desirable while the one observed in (a) is an ideal genetic biomarker for bioactivity.

Pi ^ 0

Pi = 0

в = 0 &

aj = 0

(a) X and Y are correlated, the gene is differentially expressed.

(b) X and Y are correlated but are conditionally independent.

в = 0 & aj = 0

(c) X and Y are correlated, the gene is not differentially expressed.

(d) X and Y are uncorrelated.

feature-by-feature two-sample t-test of bioactivity data (Figure 16.4a). This substructure is prominent in less potent compounds, i.e., those with pIC50 values less than 6.5 (Figure 16.4b).

Several genes correlate with the inhibitory activity against the target. Figure 16.5 highlights the linear association between pIC50 from the anti-

FIGURE 16.4

FF -442307337 for the EGFR project.

proliferation assay and gene expression changes of two on-target cancer-related genes: FGFBP1 and KRAS.

The gene FGFBP1 encodes for the fibroblast growth factor carrier protein (FGF-BP1) whose over-expression is noted in cell lines, from lung (Brattstrom et al., 2002; Pardo et al., 2003), prostate (Tassi et al., 2006), pancreas (Kuwa- hara et al., 2003), and colon cancer (Hauptmann et al., 2003). By using the joint model, it has been shown that the expression is down-regulated via the MAPK/ERK pathways after EGF-stimulated inhibition of EGFR (Harris et al., 2000). Figure 16.5a shows that more potent compounds down-regulate FGFBP1 but up-regulate KRAS.

KRAS protein has a pivotal role in the transduction of EGFR signaling (Shaib et al., 2013), it encodes a small GTP binding protein that transmits the original signal from EGFR downstream to activate important cell functions, in particular, proliferation and survival (van Krieken et al., 2008). Upregula- tion of the KRAS gene in response to EGFR inhibition could be a negative feedback mechanism of the cell to trigger cell survival. Several authors have indicated KRAS as part of a potential mechanism of resistance to EGFR inhibition, which makes KRAS a key target oncogene (Zimmerman et al., 2013; Collins and di Magliano, 2014). This gene participates in a large number of signaling pathways including MAPK, ErbB, VEGF, and a number of biological processes.

On the structure-activity side, the chemical feature, FF-442307337, is also

TABLE 16.2

Results for FF -442307337 (EGFR) at 5% FDR.

p

7^0

0

= 0

396

61

“ 0

1099 2039

linked with differential expression of numerous genes. In addition, some of the correlations observed between the pIC50 and gene expression can be attributed to this substructure as the correlation changes after adjusting for this chemical feature (Figure 16.6).

Next, genes are classified into subgroups based on whether their expression changes are linked with the structure and/or the association remains linear after adjustments for the chemical structure. The number of genes for each subgroup are presented in Table 16.2.

KRAS and FGFBP1 seem to belong to different gene classes. Figure 16.7 shows the 5 most differentially expressed genes with the adjusted association remaining high after adjusting for the chemical structure including the gene FGFBP1, while the estimates for the top 10 genes are given in Table 16.3. The association observed between the gene FGFBP1 and pIC50 is still evident after adjusting for chemical structure (Figure 16.5b). Most of these genes are known to participate in biological processes involving cell proliferation (positive and negative), survival, and differentiation. Another set of differentially expressed genes following a similar pattern to gene KRAS is presented in Table 16.4 with the visualization of the top 5 genes in Figure 16.8. For this group, the joint model resulted in very low adjusted correlation (p-adj(p)>0.05) between the genes and the activity. Unlike FGFBP1, the adjustment has a considerable effect in the observed association (from unadjusted correlation, r=0.62 to adjusted correlation, p =0.34, see Figure 16.5b).

The substructure FF-442307337 is present in the majority of compounds that inhibit cell growth to a lesser extent than FGFBP1. Figure 16.9a shows the chemical structure of FF-442307337, an oxygen in ortho position of the aniline (highlighted substructure). The next compound is very similar to the less potent compound but without FF-442307337 and it is one of the highly potent compounds in this experiment (Figure 16.9b) along with the two reference compounds gefitinib and erlotinib (Figure 16.9c-d). However, other less potent compounds do not have this feature; this substructure is probably not the sole reason for compounds' lower activity.

TABLE 16.3

List of top 10 differentially expressed genes with high adjusted association (p- adj(p) < 0.05) after adjusting for FF -442307337 (EGFR).

Genes

Effect

p-adj( Effect)

r

P

P-adj {p)

FOSL1

1.19

0.01

-0.84

-0.76

0.00

FGFBP1

0.79

0.01

-0.84

-0.78

0.00

SEPP1

-0.64

0.01

0.81

0.73

0.00

SCGB2A1

-0.61

0.01

0.83

0.76

0.00

SH2B3

0.61

0.01

-0.79

-0.69

0.00

SLCO4A1

0.60

0.01

-0.79

-0.70

0.00

PHLDA1

0.58

0.01

-0.85

-0.77

0.00

RRM2

0.56

0.02

-0.77

-0.70

0.00

TXNIP

-0.53

0.00

0.75

0.58

0.00

CDC6

0.52

0.01

-0.80

-0.73

0.00

 
Source
< Prev   CONTENTS   Source   Next >