Fig 1.
ADC target identification methodology workflow.
Workflow depicting approach used for identification of 82 prioritized ADC targets starting from a list of 20K protein coding genes.
Fig 2.
Expression of 82 prioritized ADC targets across normal and tumor tissues along with evidence based filtering annotations using five criterias*.
A) A heatmap depicting expression levels of potential ADC targets across 44 normal tissue types. B) A heatmap depicting expression levels of potential ADC targets across 20 tumor types based on their quasi H score. *1) Literature: targets for which there is existing literature evidence elucidating their potential role in tumor biology. 2) Antibody: targets against which antibodies have been generated 3) Protein family: targets belong to a protein family where other proteins isoforms of which have been employed for the advancement of ADC in either clinical or preclinical setting 4) Preclinical: targets tested in preclinical setting 5) Clinical: targets tested in clinical setting.
Fig 3.
Expression levels of 16 targets with >150 quasi scores across more than 7 tumor types.
A heatmap depicting quasi H-score across 20 tumor types for 16 potential ADC targets with >150 score in more than 7 tumor types.
Fig 4.
Scoring of 82 prioritized ADC targets based on five evidence based filtering criterias.
A) Radar plot generated using five criterias mentioned in the method section to give scores between 1 to 5 in order to rank potential ADC targets. It shows 82 prioritized targets in a circular fashion and each point on the plot represents a corresponding score for the aligned target. B) A wordcloud representing potential ADC targets based on the five criteria annotations. Wordcloud is a representation of a score for each of the 82 prioritized targets by color and size of the word. The targets with the same score are represented by the same color and font size with 5 being the highest score and 1 being the lowest score.
Fig 5.
Impact of mutations on expression levels of ADC targets identified in our analysis across tumor subtypes.
A) Impact of KRAS mutation on expression levels of multiple targets in Pancreatic Adenocarcinoma (PAAD) B) Impact of EGFR mutation on expression levels of multiple targets in Low Grade Glioma (LGG) C) Impact of STK11 and KEAP1 mutation on expression level of MSLN in Lung Adenocarcinoma (LUAD) D) Impact of BRAF mutation on multiple targets in Thyroid Carcinoma (THCA). The annotations are given as “Mut” for mutated gene and “Wild” for wild type gene.
Fig 6.
Mutations impacting MSLN and NECTIN4 target expression level across tumor subtypes.
A) Radar plot shows log2 fold change of MSLN target expression level across multiple tumor subtypes and mutations. B) Radar plot shows log2 fold change of NECTIN4 target expression level across multiple tumor subtypes and mutations.
Fig 7.
Workflow used to identify potent payload candidates in a tumor selective manner.
Workflow depicting strategy employed to identify tumor selective potential payload narrowing down from ~50K to 729 compounds.
Fig 8.
Heatmap depicting compounds with sensitivity ranging between pM to 1nM.
Heatmap depicting clustering of 33 compounds based on sensitivity patterns across 9 NCI60 cancer indications. This figure represents a narrowed-down list of compounds that demonstrate specific or heightened sensitivity towards the desired cancer type. The trend of sensitivity of cancer indications towards compounds is ascending as we move in the direction of the arrowhead. It is feasible to identify compounds that exhibit distinct activity either in solid tumors or hematological malignancies, as shown in green box compounds, such as Nogamycin and Vengicide exhibit activity against prostate and breast cancer, while no activity was seen against heme malignancies at this sensitivity range. Red box highlights compounds, such as Vedelianin and Trichloroplatinum which exhibit differential activity, with blood cancers showing greater sensitivity.
Fig 9.
Heatmap depicting compounds with sensitivity ranging between picomolar to ≦ 10nM.
Heatmap depicting clustering of 65 compounds (overlap with both subclasses) based on sensitivity patterns across 9 NCI60 cancer indications. Arrowhead represents eribulin mesylate, an ADC payload identified in this subclass which is currently under active clinical investigation.
Fig 10.
Heatmap representing clinically tested ADC payload.
Heatmap depicting clustering of 6 compounds identified in our screening based on sensitivity patterns across 9 NCI60 cancer indications. As highlighted by green boxes utilization of MMAE, expressed moderate activity against renal cell carcinoma, a discernible pattern reported by other studies, which aligns with our analysis [68].