APOBEC-related mutagenesis and neo-peptide hydrophobicity: implications for response to immunotherapy

Figures & data

Table 1. Consequences of a single iteration of APOBEC-related mutagenesis on the overall hydrophobicity of the human coding genome (per in silico computation)^a

	A. CONSIDERING ALL STRETCHES			B. CONSIDERING ONLY MUTATED STRETCHES^b
	Hydrophobicity (AU)^c			Hydrophobicity (AU)
	Before mutagenesis	After mutagenesis	Difference	Before mutagenesis	After mutagenesis	Difference
Number of stretches	4096			3744
Median	−0.00003438	−0.00003418	+ 2.9 x 10^−8	0	−0.000003384	+ 1.0 x 10^−7
25% percentile	−0.0006361	−0.0006178	−1.0 x 10^−7	−0.000486	−0.0004733	−1.3 x 10^−7
75% percentile	0.0003892	0.000396	1.0 x 10^−6	0.0004366	0.000448	1.3 x 10^−6
Mean	−0.0001756	−0.0001698	+ 5.7 x 10^−6	−0.000009969	−0.000003683	+ 6.3 x 10^−6
Standard deviation	0.001392	0.00139	2.1 x 10^−5	0.001202	0.001199	2.2 x 10^−5
Standard error	0.00002175	0.00002172	3.2 x 10^−7	0.00001964	0.00001959	3.5 x 10^−7
Lower 95% CI	−0.0002182	−0.0002124	5.1 x 10^−6	−0.00004847	−0.00004209	5.6 x 10^−6
Upper 95% CI	−0.0001329	−0.0001272	6.4 x 10^−6	0.00002853	0.00003472	7.0 x 10^−6
Sum	−0.7191	−0.6955	+ 0.0235	−0.03732	−0.01379	+ 0.0235
Wilcoxon signed rank test
Sum of signed ranks (W)	2.79 x 10^6			2.79 x 10^6
P-value	< 0.0001			< 0.0001

Abbreviations: AU = arbitrary unit; CI = confidence interval.

^aAPOBEC-related mutagenesis patterns 2 and 13 (as described by ref.¹⁹) were used. Table 1A shows the result of the analysis using all existing 6-nucleotides stretches (mutated or not), whereas Table 1B shows the result of the analysis using only 6-nucleotides stretches presenting a mutation. Alterations on the reciprocal strands were not included because of an existing bias against mutations in the reciprocal strand¹⁹. See Supplemental Table 1 for calculations with the use of the reciprocal strand.

^bMutated stretches are nucleotide stretches presenting a different nucleotide sequence after application of the mutagenesis pattern.

^cThe overall hydrophobicity of the coding genome was obtained by summing the hydropathic indices of all existing codons (using the hydrophobicity scale described by ref.³⁴), weighted by their frequency of observation (using the codon usage described by ref.³⁷).

Figure 1. Amino-acid distribution and relative hydrophobicity of the human coding genome after 1 and 20 APOBEC-related iterations (by in silico computation), as compared as other causes of mutagenesis.

Panel A shows changes in human coding genome overall hydrophobicity after multiple iterations of APOBEC-related mutagenesis, as computed in silico (n = 1 to 100 iterations). All nucleotides stretches (mutated or not) were included, reciprocals were not considered. Results obtained by applying the same method to alternative mutation signatures are shown for comparison purposes.Panel B and C show changes in hydrophobicity distribution after a single (B) or multiple (C) APOBEC-related mutagenesis iteration of all 6-nucleotide stretches whether or not they would be expected to be mutated; reciprocals wre not included. The figure breaks down the changes by hydrophobicity categories. There is a loss of hydrophilic amino acids (left side of the mountain graphic) and a gain of hydrophobic amino acids (right side of the mountain graphic).Panel D and E show the percentage of change of each amino acid (D) and codon stop (E) after 1 and 20 iterations of APOBEC-related mutagenesis.Amino-acid code – alanine (Ala, A), arginine (Arg, R), asparagine (Asn, N), aspartic acid (Asp, D), cysteine (Cys, C), glutamic acid (Glu, E), glutamine (Gln, Q), glycine (Gly, G), histidine (His, H), isoleucine (Ile, I), leucine (Leu, L), lysine (Lys, K), methionine (Met, M), phenylalanine (Phe, F), proline (Pro, P), serine (Ser, S), threonine (Thr, T), tryptophan (Trp, W), tyrosine (Tyr, Y), valine (Val, V), stop codon (X, *).

Table 2. Comparison of change in hydrophobicity of the neo-peptide library (8- to 10-mer peptides) of TCGA tumors (whole exome sequencing) with and without APOBEC-related mutagenesis^a.

	TOP 30% OF TUMORS BY MUTATION BURDEN
	APOBEC-mutagenesis NEGATIVE (n = 239)	APOBEC-mutagenesis POSITIVE (n = 230)	p-value
Change in hydrophobicity, by tumor^b
Mean [CI, 95%] (AU)	+ 3,374 [2,987–3,761]	+ 8,702 [7,506–9,898]	< 0.0001
Median [range] (AU)	+ 2,763 [−1,692–22,428]	+ 5,587 [765–70,444]
Expression-weighted change in hydrophobicity, by tumor^f
Mean [CI, 95%] (AU)	+ 2.6 [−8.9–14.2] x10^8	+ 22.2 [17.7–26.6] x10^8	< 0.0001
Median [range] (AU)	+ 5.1 [−1,344–215] x10^8	+ 11.5 [−63–291] x10^8
	LOW 70% OF TUMORS BY MUTATION BURDEN
	APOBEC-mutagenesis NEGATIVE (n = 2,859)	APOBEC-mutagenesis POSITIVE (n = 156)	p-value
Change in hydrophobicity, by tumor
Mean [CI, 95%] (AU)	+ 563 [538–588]	+ 1,172 [1,042–1,301]	< 0.0001
Median [range] (AU)	+ 417 [−2,550–4,032]	+ 1,075 [−617–3,260]
Expression-weighted change in hydrophobicity, by tumor
Mean [CI, 95%] (AU)	+ 2.5 [0.9–4.1] x 10^8	+ 6.6 [1.9–11] x 10^8	< 0.0001
Median [range] (AU)	+ 0.4 [−410–2,000] x 10^8	+ 1.6 [−44–296]] x 10^8
	ALL TUMORS
	APOBEC-mutagenesis NEGATIVE (n = 3,098)	APOBEC-mutagenesis POSITIVE (n = 386)	p-value
Change in hydrophobicity, by tumor
Mean [CI, 95%] (AU)	+ 780 [734–826]	+ 5,659 [4,856–6,462]	< 0.0001
Median [range] (AU)	+ 473 [−2,550–22,428]	+ 2,802 [−617–70,444]
Expression-weighted change in hydrophobicity, by tumor
Mean [CI, 95%] (AU)	+ 2.5 [0.8–4.2] x 10^8	+ 16 [13–19] x 10^8	< 0.0001
Median [range] (AU)	+ 0.5 [−1300–200] x 10^8	+ 6.1 [−63–296] x 10^8

Abbreviations: AU = arbitrary unit; CI = confidence interval.

^a An analysis of hydrophobicity was performed on a set of highly mutated pan-cancer tumors (top 30% of mutational load in TCGA), not presenting mismatch repair or polymerases delta and epsilon alterations, classified as APOBEC-related mutagenesis positive or negative using the P-MACD estimates provided by the Broad GDAC Firehose website (https://gdac.broadinstitute.org). The lower 70% of tumors by mutational burden was analyzed as well. See Supplemental Table 2 for data obtained for full-length transcripts.

^b The change in hydrophobicity induced by the mutagenesis on all possible 8- to 10-mer neo-peptides of a tumor is shown. The peptide hydrophobicity was evaluated before and after mutagenesis, by summing residue hydrophobicity indices. The hydrophobicity change induced by the mutagenesis was calculated for each peptide, and these changes were then summed by tumor.

^c The expression-weighted change in hydrophobicity was calculated in a manner identical as described above, except that each hydrophobicity score was multiplied by the corresponding level of transcript expression.

Figure 2. Overall change in hydrophobicity in human tumors positive for APOBEC-related mutagenesis.

Panel A. Correlation between the overall neo-peptide hydrophobicity change and the total number of APOBEC-related mutations in highly-mutated human tumors positive for APOBEC-related mutagenesis.Panels B-E. Comparison of neo-peptide hydrophobicity change in highly mutated tumors with or without APOBEC-related mutagenesis, in different subgroups of tumors (non-melanoma (B, D) and melanoma tumors (C, E)), considering the hydrophobicity scores before (B, C) or after (D, E) weighting by the mRNA-expression levels. Mean and 95% confidence intervals are represented in bold bars.

Figure 3. Response and progression-free survival comparison between patients treated with checkpoint blockade presenting differential APOBEC-related mutagenesis phenotypes.

An estimate of the APOBEC-related mutagenesis was obtained for each patient using the AMSE tool. Patients were classified in low APOBEC-related mutagenesis phenotype for scores ≤ 0.727 and in high APOBEC-related mutagenesis phenotype for scores > 0.727. Panels A: Pie-charts of response in patients presenting a low versus high APOBEC-related mutagenesis score. Panel B: Kaplan-Meier curve for progression-free survival of patients with APOBEC-related mutagenesis score low versus high.Abbreviations: CI = confidence interval; HR = hazard ratio; OR = odds ratio.

Table 3. Factors affecting the outcome of patients treated with immunotherapy agents.

		Complete or partial response N (%)	Stable or progressive disease N (%)	P – value univariate^g	P-value multivariate^h	OR [95% CI]^i
Tumor type	NSCLC^j	7 (19%)	29 (81%)	> 0.9999
	Other tumors^k	12 (19%)	51 (81%)
TMB^l	Low	2 (4%)	44 (96%)	0.0006	0.006	0.09 (0.017–0.51)
	Intermediate	8 (24%)	25 (76%)	0.4211
	High	9 (45%)	11 (55%)	0.0027	0.651	1.34 (0.37–4.85)
APOBEC-related mutagenesis^m	Low	1 (3%)	28 (97%)	0.0106	0.031	10.99 (1.25–100)
	High	18 (26%)	52 (74%)
Type of immunotherapy	Anti-PD-1/PD-L1 monotherapy	16 (18%)	75 (82%)	0.1780
	Other immunotherapy^n	3 (38%)	5 (62%)

Abbreviations: CI = confidence interval; CTLA4 = cytotoxic T-lymphocyte associated protein 4; Mb = megabase; N = number; NSCLC = non-small cell lung cancer; OR = odds ratio; PD-1 = programmed death receptor-1; PD-L1 programmed death receptor-ligand 1; PFS = progression free survival; TMB = tumor mutational burden.

^a Calculated using Fisher’s exact test. Supplemental Table 4 shows the univariate analysis with all demographic factors.

^b All univariate p-values of ≤ 0.05 were included in the multivariate analysis, using a binomial logistic regression model.

^c OR > 1.0 implies higher chance of response.

^d Histologies of NSCLC included adenocarcinoma (N = 30) and squamous cell carcinoma (N = 6).

^e Other tumors include adrenal carcinoma (n = 1), appendix adenocarcinoma (n = 1), basal cell carcinoma (n = 2), bladder transitional cell carcinoma (n = 4), breast cancer (n = 3), cervical cancer (n = 2), colon adenocarcinoma (n = 5), cutaneous squamous cell carcinoma (n = 8), hepatocellular carcinoma (n = 3), head and neck (n = 13), Merkel cell carcinoma (n = 2), ovarian carcinoma (n = 2), pleural mesothelioma (n = 1), prostate cancer (n = 1), renal cell carcinoma (n = 6), sarcoma (n = 3), thyroid cancer (n = 3), unknown primary squamous cell carcinoma (n = 2), and urethral squamous cell carcinoma (n = 1).

^f TMB low = 1–5 mutations/Mb; TMB intermediate = 6–19 mutations/Mb; TMB high ≥ 20 mutations/Mb.

^g An estimate of the APOBEC-related mutagenesis was obtained for each patient using the AMSE tool, available at https://github.com/KwatME/mutational_signature. Patients were classified in APOBEC-related mutagenesis LOW phenotype for scores ≤ 0.727 and in APOBEC-related mutagenesis HIGH phenotype for scores > 0.727.

^h Other immunotherapy included OX40 (n = 2), anti-CD73 (n = 1), anti-CTLA4 (n = 2), OX40+ IDO (n = 1), anti-PD-1+ anti-CTLA4 (n = 1), and IDO+ anti-PD-1 (n = 1).

Alexandrov LB, Nik-Zainal S, Wedge DC, Aparicio SAJR, Behjati S, Biankin AV, Bignell GR, Bolli N, Borg A, Børresen-Dale A-L, et al. Signatures of mutational processes in human cancer. Nature. 2013;500(7463):415–421. doi:10.1038/nature12477.

 PubMed Web of Science ®Google Scholar

Kyte J, Doolittle RF. A simple method for displaying the hydropathic character of a protein. J Mol Biol. 1982;157(1):105–132. doi:10.1016/0022-2836(82)90515-0.

 PubMed Web of Science ®Google Scholar

Nakamura Y, Gojobori T, Ikemura T. Codon usage tabulated from international DNA sequence databases: status for the year 2000. Nucleic Acids Res. 2000;28(1):292. doi:10.1093/nar/28.1.292.

 PubMed Web of Science ®Google Scholar