Stability of colon stem cell methylation after neo-adjuvant therapy in a patient with attenuated familial adenomatous polyposis

Recent studies reveal methylation of certain CpG rich sequences increases with age in normal human colon [1, 2]. These CpG rich sequences, like most CpG islands [3], are unmethylated at birth and appear to somatically accumulate cytosine methylation over many years. The etiology of such age-related methylation is uncertain, but one possible mechanism is random somatic errors during DNA replication [4]. By this mechanism, methylation would inherently increase with cell division because the CpG rich sequences are initially unmethylated. Consistent with random replication errors, methylation patterns (the 5' to 3' order of methylation) differ between cells within the same colon [4] and overall levels increase with age [1, 2].

Random epigenetic errors potentially records cell histories because methylation exhibits somatic inheritance [3]. Daughter cells generally inherit the same methylation pattern as their parents, but rarely replication errors occur which are subsequently passed to progeny (Figure 1). Somatic errors can only accumulate within stem cells because non-stem cells and their errors are lost within days from normal crypt cell differentiation and migration [4, 5]. Therefore, CpG rich sequences, like sequences used to reconstruct species evolution [6], potentially function as somatic epigenetic molecular clocks that can be used to reconstruct somatic cell evolution. Preliminary studies are consistent with this hypothesis because seemingly random human colon crypt methylation patterns appear to record life-long stem cell fates within crypt niches [4].

Although crypt methylation patterns from multiple different aged colons are consistent with a clock-like accumulation of epigenetic errors, environmental factors may also alter methylation. Age-related methylation levels differ between individuals of the same age [1, 2, 4], which may represent life-long biological differences between individuals, stochastic differences, or transient responses to environmental changes. It may be possible to distinguish between these alternatives through serial examinations of the same colon. A somatic molecular epigenetic clock predicts the amount of additional methylation is a function of the time between examinations. In contrast, if methylation is more labile, patterns may quickly change after a severe environmental insult. Here we sample methylation 62 days apart in the same colon before and after cytotoxic neo-adjuvant chemotherapy and radiation for a rectal adenocarcinoma. Methylation patterns were essentially unchanged, suggesting crypt stem cell epigenomes are relatively protected from transient environmental changes.

A 22 year-old male presented with anemia. Colonoscopy revealed eight hyperplastic to adenomatous polyps ranging in size from 0.6 to 3.0 cm scattered through the colon, and a four cm, locally invasive, ulcerative rectal adenocarcinoma five cm from the anus. The clinical impression was attenuated familial adenomatous polyposis (AFAP), with a germline heterozygous truncating APC mutation at codon 215 (CAG to TAG; Gln to stop). The first sample for methylation analysis (Day 1) was a biopsy of normal appearing sigmoid colon.

The patient received neo-adjuvant chemotherapy consisting of three, two week cycles of oxaliplatin and 5-fluorouracil, with the last cycle ending two weeks before colectomy. The patient exhibited Grade 3 toxicity with mucositis, stomatitis, diarrhea, and parasthesia. He also received radiation therapy to the pelvis (2,500 centigrays over five days), ending 19 days before colectomy. The specimens obtained for analysis were outside of the primary radiation field.

Colectomy revealed approximately 50 scattered small hyperplastic to adenomatous polyps (0.1 to 0.3 cm), and a 0.7 cm ulcerative lesion at the site of the rectal adenocarcinoma containing a few residual neoplastic glands with superficial invasion of the muscularis propria. Lymph nodes were negative for metastatic carcinoma. The second specimen for analysis (Day 62) was taken from a normal appearing region of the sigmoid colon.

The ability to extract historical information from normal appearing tissue is limited. However, recent advances in molecular phylogeny illustrate that genomes automatically record ancestry and time [6]. Although sequences are usually faithfully copied, errors rarely occur, which are also passed from generation to generation (Figure 1). Sequence comparisons can be used to reconstruct species evolution that occurred thousands to millions of years ago. The greater the differences between sequences, the greater the likely time since they shared a common ancestor.

Sequences comparisons depend on relatively constant error rates (a molecular clock hypothesis [6]) because time is proportional to the number of sequence differences. Although it is possible that mutation rates vary over time from environmental changes, species histories inferred from sequences are generally consistent with relatively constant mutation rates over millions of years [6]. In theory, molecular phylogeny can also be applied to somatic cell evolution because somatic cell genomes are also passed from cell to cell. Although somatic cell sequence comparisons are impractical because mutation rates are low relative to human lifetimes, epigenetic errors also potentially record ancestry because methylation exhibits somatic inheritance [3]. Epigenetic replication fidelity is less than sequence fidelity because age-related methylation is readily detected in the human colon [1, 2]. Colon crypt methylation patterns have been used to infer normal human crypt niche dynamics [4], and that most directly adjacent adult crypts are not closely related because their methylation patterns differ [9]. Stem cell survival can also be inferred from methylation pattern diversity (number of unique tags). For example, increased crypt diversity and enhanced crypt stem cell survival is observed in some FAP patients with truncating germline mutations that remove some but not B-catenin binding repeats [7]. In contrast and as with this patient, germline APC mutations in the N-terminus of APC have not been associated with enhanced stem cell survival [7], and are associated with fewer polyps and AFAP [10].

A general problem with the use of sequences to reconstruct ancestry is the stochastic nature of most errors. It is difficult to prove stochastic sequence errors actually accumulate in a clock-like manner, and error rates may potentially fluctuate in response to environmental changes. Highly variable error rates would obscure ancestral information because many errors may accumulate in relatively short periods.

Although it is difficult to test a molecular clock hypothesis for species evolution, it is possible to test a somatic epigenetic molecular clock through serial examinations of the same individual. This Case Report experimentally tests the hypothesis that age-related methylation predominantly reflect stem cell errors that start to slowly accumulate from birth. Consistent with this hypothesis, methylation patterns were not significantly different after 62 days despite intervening neo-adjuvant therapy. Neo-adjuvant chemotherapy targets rapidly dividing cells and normal colon is relatively unaffected, although diarrhea (Grade 3 toxicity) may and did occur in this patient. However, the therapy did not appear to alter stem cell methylation patterns, suggesting stem cells are protected from this type of therapy. Therefore, age-related methylation appeared to be relatively stable and primarily a function of the time or number of stem cell divisions since birth. Further serial examinations with more patients and longer intervals may provide additional evidence supporting or refuting this hypothesis.

In summary, age-related methylation appears to slowly accumulate in a clock-like manner in stem cells from birth. The stochastic nature of somatic epigenetic errors requires an algorithm for interpretation, which essentially translates the sophisticated methods used to extract ancestry from species genomes to reconstruct somatic cell histories. The unseen fates of human crypt stem cells may be potentially read from anyone because methylation automatically records their autobiographies.