“Monoallelic germline methylation and sequence variant in the promoter of the RB1 gene: a possible constitutive epimutation in hereditary retinoblastoma”

The studied family shows an autosomal dominant pattern characteristic of hereditary
Rb with several family members affected by the condition in each generation 3]. However, the number of affected members (18 and 33 %) in generations II and III
is 50 % as expected in this type of inheritance. The higher number of affected males
with respect to females also deviates from this ratio. On the other hand, none of
the 10 obligate carriers developed Rb and all of those affected have unilateral Rb.
These familial data indicate that this corresponds to the small but significant number
of families in which Rb is inherited with incomplete penetrance (IP) 3].

In contrast to what was observed in generations II and III, in generation IV, corresponding
to the family of this study, 4/7 children (57 %) were affected, showing a change to
complete penetrance. However, the moderate expression of unilateral tumors is preserved
as well as the bias of affected males.

It has been suggested that these biases are due to a higher mutation rate in spermatogenesis
than in oogenesis, meiotic drift, and to the existence of imprinted genes 4]–7], 23]. Klutz et al. 24] studied two non-related families with Rb and IP who carried the same mutation. This
showed variation in the phenotypic expression of Rb and a higher number of affected
members when the father was identified as the transmitter of the mutant allele.

To identify the paternal germline mutation that has conditioned IP in the reported
family, complete sequencing of RB1 would be required because the type of mutations in the families with IP are not part
of the spectrum of germline mutations found in most families with Rb and almost every
family has its own mutation 25], 26]. The mutations conferring IP in general cause a quantitative decrease in the expression
or a partial loss of the RB1 suppressor function 3], 10], 17], 24]. It has been suggested that in families with IP mutations, LOH is oncogenically insufficient
because the homozygosity of the predisposing allele still retains suppressive activity
and the carriers would be asymptomatic 17], 20]. For Rb development, a mutation with complete loss of function in the normal allele
is required 17]. Because these mutations are less common (30 vs. 70 % LOH), the proportion of those
affected in these families may be lower 3], 17]. This would explain the lower number of those affected in generations II and III.
Another explanation for the lower number of those affected may be related to the differential
expression of RB1, consecutive to its normal imprinting state 7]. In this case, the preferential expression of the maternal allele could substitute
the low expression of the putative paternal germline IP mutation, which would avoid
the development of Rb in asymptomatic carriers of this family.

Regarding the change of penetrance specifically in the RB-F60 nuclear family, the
possibility that could help explain this change is the finding of the c. ?[187T??G;
188T??G] sequence variant between the ATF and E2F sequences. In the first, Sakai
et al.17] found the G??T transversion (position 189 upstream from the start codon) in a family
with hereditary Rb and IP, which allowed identifying the binding site of this nuclear
factor in the core of the RB1 promoter, necessary for transcriptional activation of RB1 and the oncogenic suppression 17]. Whereas E2F is involved in gene repression 18], studies with transgenic reporters have shown that mutations at a single E2F site
are critical for gene repression, further suggesting that this factor may contribute
to the regulation of the transcription of RB127]. The change observed in this study is found in positions 187 and 188, and the transversion
is also different. Mutations in the RB1 promoter are rare. In the ATF sequence, three cases of mutations were found 4], 25], 26], whereas in E2F, no similar reports were found 25], 26].

Three aspects stand out in reference to the sequence variant observed in this family:
(a) it involves two adjacent bases, one in an activator site and the other in a repressor
site; this variant was not found reported 25], 26]; (b) the variant was unexpectedly found in PBL DNA from the index case’s mother,
who transmitted it to two generations including all their children (generation IV)
and the index case’s daughter (generation V), suggesting germline occurrence that
has been segregated with a dominant inheritance pattern; (c) and it coincides or is
associated with methylation mainly of recognition sequences for transcription factors
in the RB1 promoter. Methylation is apparently allele-specific because, in general, clones without
this change do not show consistent methylation.

Both the ATF sequence mutations and the methylation of the promoter are oncogenic
9], 10], 13], 17], suggesting that this sequence variant associated with methylation could also be
oncogenic.

On the other hand, the methylation pattern seen in PBL in this family has similarities
with the methylation pattern reported by Stirzaker et al. 13] in Rb tumors. This shows methylation in the 27 CpGs of the RB1 promoter, including binding CpGs to transcription factors but with varying methylation
density both among the CpGs and from tumor to tumor. Some individual CpGs were unmethylated,
highlighting the CpG from E2F. In our study, extensive methylation is also observed
but is more consistent in the CpGs where transcription factors bind, although the
CpG in ATF was found unmethylated as well as one of the two CpGs in E2F. However,
it is not possible to perform a quantitative comparison of methylation in specific
CpGs because no quantification was performed as in the study by Stirzaker et al.

It is interesting to compare the results with the findings of Dávalos et al. 12] in Fig. 4c regarding the methylation pattern of cultured cells in which the CTCF binding sites
were mutated in the RB1 gene promoter. It was demonstrated that CTCF protects the promoter from methylation.
There are striking similarities in the methylation pattern in the promoter from the
studied family with the methylation pattern of the cells lacking CTCF protection,
which also showed low expression levels of pRB. Similar analyses have not been performed
in this family.

The silencing of RB1 consecutive to the promoter methylation reported in Rb tumors 10], 13], 14] and the gain of methylation in the promoter of this gene consecutive to mutations
in key-binding sites to transcription factors 10]–12], 17], 18], 20] would allow us to suggest that in RBF60, this double finding apparently in the same
allele could correspond to an epimutation consecutive to the TT??GG transversion
positioned between an activating sequence and a repressor sequence as previously mentioned.

This supposition is sustainable on the basis that in some neoplastic diseases with
hereditary predisposition, similar alterations to those observed in this study have
been reported in which sequence variants coexist in adjacent or distant genes that
promote epigenetic modifications, specifically the methylation in the promoters of
specific genes 28]–30]. These changes show that in the etiology of these conditions, very complex genetic-epigenetic
interactions coexist and are involved in the transcriptional silencing which, among
others, is consecutive to antisense transcription 31]. These mechanisms are helping to understand this new field of epigenetic inheritance
and its hereditary transmission through epimutations 32], 33].

It should be emphasized that these particular types of epimutations are hereditary
because the epigenetic methylation modifications are secondary to changes in cis in
the gene sequences that occur at the germline level and are dominantly transmitted
to offspring 28], 30], 31].

Speculatively, we suggest that the sequence variant in RBF60 according to some currently
unknown mechanism induces methylation of the RB1 promoter. This originates a constitutive epimutation 32], 33] because methylation is found in the melanoma of the index case and in the PBL. However,
this finding in the mother was unexpected. In fact, she represented the control arm
of the study because no obvious pathological data were found in the maternal family.
As shown in the genealogy, it is clear that the transmission was only of paternal
origin.

Because the father died, polymorphic markers were analyzed in an attempt to demonstrate
the paternal origin of the mutant allele. However, the results suggest that both parents
were carriers of germline mutations. In the father, this is still unidentified as
previously mentioned but was demonstrated by the IP inheritance pattern and the mother
as a carrier and transmitter of a probable constitutive epimutation in the RB1 gene promoter.

The biparental contribution is further supported by the results obtained with the
microsatellite markers because specific segregation was observed not only of one allele
of the paternal D13 chromosome but also by the specificity of one of the alleles of
D13 chromosome of maternal origin, which is segregated with a unique haplotype in
those affected. This suggests a biparental germinal contribution of both D13 chromosomes
to the Rb phenotype. A biparental-specific share of the alleles of chromosome 16 was
also observed, which represents information of additional interest in this family
since deletion of the long arm of this chromosome (16q) is related to a particular
type of Rb 34].