Review: Are moles senescent?

Melanocytic nevi (skin moles) have been regarded as a valuable example of cell senescence occurring in vivo. However, a study of induced nevi in a mouse model reported that the nevi were arrested by cell interactions rather than a cell‐autonomous process like senescence, and that size distributions of cell nests within nevi could not be accounted for by a stochastic model of oncogene‐induced senescence. Moreover, others reported that some molecular markers used to identify cell senescence in human nevi are also found in melanoma cells—not senescent. It has thus been questioned whether nevi really are senescent, with potential implications for melanoma diagnosis and therapy. Here I review these areas, along with the genetic, biological, and molecular evidence supporting senescence in nevi. In conclusion, there is strong evidence that cells of acquired human benign (banal) nevi are very largely senescent, though some must contain a minor non‐senescent cell subpopulation. There is also persuasive evidence that this senescence is primarily induced by dysfunctional telomeres rather than directly oncogene‐induced.

What OIS might look like after a natural oncogenic mutation in vivo is not well defined however, as Damsky and Bosenberg (2017) noted.Overexpression of BRAF V600E in cultured human diploid melanocytes does lead to rapid arrest, within days (Gray-Schopfer et al., 2006;Michaloglou et al., 2005).However, a natural BRAF V600E   point mutation in an epidermal melanocyte would not give overexpression, only normal abundance of the oncoprotein, or at least 50% normal, being initially heterozygous.Induction of such a point mutation in BRAF leads to proliferation of cultured normal human melanocytes (Zeng et al., 2018), rather than replication stress and OIS.McNeal et al. (2021) also report that if BRAF V600E was overexpressed at a level sufficient to arrest melanocytes grown with the mitogenic phorbol ester TPA, but now in melanocytes cultured without TPA, it could stimulate proliferation rather than arrest.Moreover, resumed proliferation was observed in BRAF-arrested cultures on removal of either TPA or the BRAF V600E expression, so at least some cells were reversibly arrested rather than senescent.
A recent paper proposed further that nevus cells are not arrested through OIS, nor any kind of senescence, but are quiescent: arrested reversibly through cell interactions (Ruiz-Vega et al., 2020).This conclusion was based on data from a mouse model of Braf V600E -induced nevi, compared to a stochastic mathematical model for OIS timing.
They also proposed that this concept could extend to human benign nevi.This conclusion, apparently in conflict with previous ideas and data, has attracted attention and perhaps generated some confusion.Accordingly, this review aims to re-examine the available evidence on nevi and senescence, including the data of Ruiz-Vega et al. in comparison to human nevi.

| THE MOUS E MODEL US ED BY RU IZ-VEG A E T AL .
The murine melanoma model used by Ruiz-Vega et al. (2020) was that of Dankort et al. (2009).Mice were genetically engineered such that cells of the melanocytic lineage inducibly acquire the Braf V600E mutation, and express the oncogene from its natural (Braf) promoter.Thus, the oncogene should not be overexpressed.The mutation was activated postnatally through a Tyr::CreER transgene, by 4-hydroxytamoxifen (4-OHT).
Painting of 4-OHT on the back skin of 2-to 4-day-old C57BL6/J mice resulted in growth of numerous, microscopic melanocytic colonies in the dermis, at least 5-10 colonies/mm 2 from images shown, and with a median diameter of 154 μm.These stopped growing by 21 days, and were described as nevi (Ruiz-Vega et al., 2020).They were reported to be composed of smaller cell clusters or nests, as often seen in human nevi (though no clear histological images were provided).These nests had a reported median diameter of around 40 μm.
This biology immediately seems very different from that of human acquired nevi, which are usually defined as at least 2 mm in diameter (Bataille et al., 2007;Shain & Bastian, 2016), range up to 5 mm diameter or more, and are typically spaced centimeters apart.
Congenital nevi can be much larger.Acquired nevi grow most commonly along the dermoepidermal border (simple lentigo, junctional nevus), or in both epidermis and dermis (compound nevus) (Mooi & Krausz, 2007).A recent preprint reports that a subclass of dysplastic nevi also shows a lentiginous pattern, associated with driver mutations other than BRAF and with older patient age (Lorbeer et al., 2023).
Ruiz-Vega et al. presented convincing evidence that their melanocytic colonies were not arrested via OIS.The evidence included single-cell gene-expression profiles in two cell sets (from whole skin of these mice), both taken to be nevus cells because they included melanocytic markers, and were absent from skin of normal mice.
Neither of these sets showed any match to any of several gene expression signatures taken from different types of cell senescence.
Secondly, the authors compared the size distribution of their nevi (log-normal), to a mathematical model of OIS in which they assumed stochastic (random) arrest of individual cells with a certain probability per cell division or per time, following oncogene activation.This model predicted highly skewed size distributions on a log axis, rather than log-normal distributions as observed.This was very good evidence that their nevi were not arrested by a stochastic process.Additional considerations, for example sizes versus locations of cell nests within nevi, and the crowding of the nevi (median spacing 79 μm), supported the conclusion that the arrest involved cell interactions, rather than being an autonomous and permanent type of arrest, like senescence (or at least like the initiation of senescence).
In comparison, the similar Braf V600E /Tyr::CreERT2 mouse strain of Dhomen et al. (2009) gave somewhat different biological results when the oncogene was induced 2-3 months after birth.A smaller number of nevus-like lesions was obtained, somewhat larger than those of Ruiz-Vega et al. (images suggest diameters around 0.5-2 mm), and spaced further apart.This difference likely reflects the changing numbers of epidermal melanocytes in mice, which peak around 2-4 days of age, at ~1000/mm 2 in the closely related C57BL/10 J mouse strain, falling to zero by around 30 days when | 393 BENNETT all melanocytes are normally in hair-follicles (Hirobe, 1982).These lesions were positive for 2/2 tested cell senescence markers, βgalactosidase, and p16 CDKN2A (markers detailed later) (Dhomen et al., 2009).The apparently fewer but larger lesions of Dhomen et al. contrast with those seen by Ruiz-Vega et al., consistently with interactions between Ruiz-Vega et al.'s crowded micro-nevi mutually halting lesional growth, but with clones being able to grow larger to the point of senescence if they have more space.Likewise, Braf V600Edriven nevus-like clones in mice, initiated at 9 weeks, ceased growth while quite well-spaced, and resembled senescent melanocytes in expressing p16 and BCL-W (BCL2L2) (Kohli et al., 2022).These lesions interestingly showed specific sensitivity to combined "senolytic" inhibitors of the BCL-family and MCL1 anti-apoptotic proteins (Kohli et al., 2022).
In short, Ruiz-Vega et al. ( 2020) appear correct concerning their small, crowded melanocytic colonies induced in a Braf V600E mouse model just after birth, that the arrest was due to cell interactions rather than senescence.It is questionable however whether these should be called nevi, since they are so different from human nevi.
Other groups who induced Braf V600E in similar but older mice found that the sparser clones grew larger and then did show evidence of senescence, more like typical human nevi (Dhomen et al., 2009;Kohli et al., 2022;Wang et al., 2023).Ruiz-Vega et al. (2020) proposed that their conclusion on nonsenescence could also be applied to human nevi, on two specific grounds.One was the clinical evidence that some human nevi can regenerate after excision.The rate of this recurrence varies between reports, from 0.3% to 27% (Vilain et al., 2016).Irrespective of the reasons for such variation, this clearly does imply that some human nevi contain some nevus melanocytes capable of dividing.Indeed, rare mitoses can be observed within excised nevi (Mooi & Krausz, 2007).However, this does not imply that all or most nevus cells can divide.

| NE S T-S IZE MODELING AND HUMAN NE VI
The other ground was the authors' measured size distribution for nests within a sample of five human nevi.They observed a lognormal distribution of radii of nests, implying that these nests did not fit a stochastic model for OIS, assuming nests represent subclones of the nevus.As a side-issue, not all types of human nevi even have distinct nests.The common simple lentigo consists of a patch of more-closely-spaced than normal melanocytes in the basal epidermis (Mooi & Krausz, 2007).These are rarely excised and studied, however.Excised nevi tend to be larger and thicker, those suspected of malignancy.
Even restricting modeling to nevi containing nests, there are problems with testing predictions of a simple model against a collection of nest radii pooled from several people, since humans are not inbred and have nevus size distributions that differ widely between individuals, so their nests could also differ in size, depending on many possible factors.The core assumption of the model, that only the stochastic process would be generating the variation in size, is not applicable here.Genetic factors are known to affect human nevus size (next section).The person's age at the time of nevus initiation is another factor.Congenital nevi (initiated before birth), for instance, can be far larger than any acquired nevus.This led to the speculation that telomere length is a determinant of human nevus size (Bastian, 2003), a speculation later confirmed (next section).If a stochastic process were the only determinant, then age and telomeres should not matter.Conversely, if nevus cells were capable of dividing but reversibly arrestable by paracrine interactions such as an inhibitory factor(s) made by all colony cells, we would expect cell division to stop in the lesion's center, but to continue around the edges where cells have fewer neighbors, as seen with cell colonies in culture (Stoker & Rubin, 1967).To complicate matters, it has been reported that senescent melanocytes in epidermis can induce senescence in neighboring cells in paracrine fashion (Victorelli et al., 2019).
However, by definition this would not be reversible.
Stochastic molecular events are believed to regulate other important cellular processes including reversible telomere uncapping (Rodriguez-Brenes & Peskin, 2010), and initiation of DNA synthesis (Lee et al., 2010).However, a stochastic model for OIS lacks an obvious molecular basis.Our understanding is that "standard" OIS works through the largely deterministic processes of replication stress and DNA damage (Bartkova et al., 2006;Di Micco et al., 2006), rather than through some unknown event(s) that occurs at random and until then allows cells to carry on proliferating.
For all these reasons, the nevus nest size distribution does not weigh strongly against cell senescence in human nevi, especially given the variety of evidence in support-genetic, molecular and biological.These findings, apparently not considered by Ruiz-Vega et al., will now be reviewed.

| G ENE TI C E VIDEN CE ON S ENE SCEN CE IN H UMAN NE VI
Genetics provides some of the strongest and broadest evidence connecting cell senescence with human nevi, as well as melanoma suppression, from studies of both population genetics and somatic genetic changes in sporadic melanoma.These will be considered in turn.

| Human population genetics of nevi
The population genetics of melanoma overlaps with that of nevi.
Around 10% of human melanomas worldwide occur in people with a family history of melanoma (with some geographical variation), and over 50 genetic loci associated with melanoma risk have now been identified in genome-wide association studies (GWAS) (Bishop et al., 2002;Landi et al., 2020;Law et al., 2015).Some of these families (not all) also transmit a tendency to develop large numbers of nevi including unusually large nevi.This is known as dysplastic nevus syndrome (DNS) or familial atypical multiple mole-melanoma (FAMMM) syndrome (Rashid et al., 2022).Of note, nevus number is not entirely separable from nevus size in clinical practice, since counts normally include lesions over 2 mm in diameter, whereas lentigos of diameter <2 mm are also common on adult human skin (as many of us can see on our own skin).Thus, a general tendency for larger lentigo sizes would lead to higher nevus counts.The phenotype of large nevi is found from twin studies to have high heritability of ~65% (Newton- Bishop et al., 2020).A meta-analysis of GWAS studies generated 12 significant loci associated with nevus density (counts), all of which (with one possible exception, KITLG) were also associated with melanoma risk (Duffy et al., 2018).The target genes reflected multiple biological pathways, not surprisingly including pigmentation (protects against mutagenic UV light), but also many were linked to telomere maintenance and cell senescence.

The majority of melanoma/FAMMM-associated mutations in
CDKN2A affect the p16-specific segment (exon 1α) or the shared portion with ARF, and cause defects in the binding and inhibition of CDK4 by p16 (Bishop et al., 2002;Rashid et al., 2022).Melanoma/ FAMMM-associated mutations in CDK4 likewise lead to defective p16-binding.Thus, genetic defects in the p16/CDK4/RB1 cellsenescence pathway can lead to large and numerous moles as well as melanoma.ARF functions to upregulate the other chief mediator of cell senescence, p53 (Figure 1).ARF inhibits the degradation of p53, otherwise mediated by the E3 ubiquitin-protein ligase MDM2 (mouse double minute 2 homolog) (Ozenne et al., 2010).Germline mutations specifically affecting ARF have also been reported in a few melanoma families, though indirect transcriptional effects of these mutations on p16 have not been ruled out.These families, and those with combined p16/ARF defects, also appear to have an increased risk of nerve-sheath and other tumors (Sargen et al., 2016).
Another potential connection between nevi and cell senescence concerns telomeres, the protective structures at the ends of linear chromosomes, consisting of highly repetitive DNA coated in the protein complex shelterin (Gorgoulis et al., 2019).
The ribonucleoprotein enzyme telomerase is required for maintenance or extension of telomere length, and includes subunits TERT (telomerase reverse transcriptase) and TERC (telomerase F I G U R E 1 Senescence signaling in humans: core pathways.Simplified to show the roles of key intermediates relevant to nevi.See Bennett (2016) for more detail and interacting pathways.Blue: pro-senescence signaling; red: anti-senescence signaling.T-bars: inhibition.Darker shades indicate components subject to significant mutation in familial and/or sporadic melanoma (Cancer Genome Atlas Network, 2015; Landi et al., 2020;Law et al., 2015;Tate et al., 2019).RB represents the RB protein family (RB1, RBL1, RBL2).p21 is also called CDKN1A.Dashed arrow: upregulation of p16 by senescence/DNA-damage signaling is functionally established but the mechanism is unknown.S and M phases refer to the cell cycle.The p16 pathway appears more important than the p53 pathway in human nevi and generally in primary senescence of human melanocytes (see text for details).
| 395 BENNETT RNA component).TERT protein is absent or nearly so in most normal human somatic cells, resulting in telomere shortening in dividing cells with age, and replicative senescence through DNAdamage signaling from short telomeres following extensive division (Figure 1) (Gorgoulis et al., 2019).Most immortal cancers (cells capable of indefinite proliferation) reactivate TERT expression, usually through promoter mutations (Gorgoulis et al., 2019).
Telomere length distribution at birth varies between individuals.
Concerning nevi, Bataille et al. (2007) observed that an individual's age-normalized telomere length was correlated with their numbers of melanocytic nevi, also with numbers of large nevi.The TERC and TERT loci are linked to high nevus counts (Duffy et al., 2018).
These and other telomere-related genes including several for shelterin components are also high-penetrance loci for melanoma risk (Newton- Bishop et al., 2020).A genetic score predicting telomere length was specifically found to be a strong predictor of melanoma risk (Newton- Bishop et al., 2020).Rare germline mutations in the TERT promoter are also associated with familial melanoma, and large, atypical and multiple nevi have also been reported in members of these families (Horn et al., 2013;Zaremba et al., 2022).An interesting biological finding was that when normal epidermal melanocytes were explanted from multiple-melanoma patients carrying no known melanoma susceptibility gene, they consistently showed significantly longer culture lifespans than those from single-melanoma patients, implying a general, systematic connection between individual melanoma susceptibility and delayed melanocyte senescence (Kohli, Tolomio, et al., 2017).
Taken together, these genetic findings imply that telomeres must have a role in nevus size, just as Bastian (2003) surmised, and so must the p16/CDK4 pathway.This suggests contributions of short telomeres as well as (or upstream of) the p16 pathway in control of human nevus size.Yet Michaloglou et al. (2005) assessed telomere length within human nevi and found no significant difference from that of neighboring normal cells.This apparent conflict is revisited below in the light of more-recent data.

| Evidence from genetics of nevus-to-melanoma progression
If nevi are arrested by cell senescence, then to resume proliferation and progress to dysplasia or melanoma, genetic (or epigenetic) alterations disrupting cell senescence will be expected.To become immortal, human cancers need both to inactivate the signaling pathways that effect senescence arrest, and to gain telomerase function, allowing telomere maintenance (Hayashi et al., 2015).We know the requirements for immortalization of human melanocytes in culture: telomere maintenance and full immortality can be conferred in cultured human melanocytes by the combination of p16 pathway loss/ inactivation and exogenous expression of TERT (Gray-Schopfer et al., 2006;Sviderskaya et al., 2003).Both these changes (p16 loss and TERT upregulation) are seen in the great majority of advanced human melanomas and indeed are now the two commonest known alterations in melanomas.Activating TERT promoter mutations are found in around 70% of melanomas, or 85% of metastatic melanomas (Horn et al., 2013;Huang et al., 2013;Newton-Bishop et al., 2020).For p16, if we assume that loss by homozygous deletion, deleterious mutation or silencing by DNA methylation will usually be mutually exclusive (only one needed), then we can add their incidences together, in which case p16 is defective in ~88% of melanomas (Bennett, 2016).
The incisive studies of Shain, Bastian et al. (Shain et al., 2015;Shain & Bastian, 2016) have related such changes to specific stages of progression.They sequenced 293 cancer-related genes from sporadic human melanomas and their spatially associated precursor lesions (benign nevi and/or "intermediate" lesions between nevus and melanoma) (Shain et al., 2015).They observed that nevi typically showed only a single mitogenic driver mutation (BRAF, NRAS), while associated intermediate lesions (and corresponding melanomas) displayed the same driver plus most commonly (77%) an activating TERT promoter mutation; or a heterozygous (usually) deletion or mutation in CDKN2A, or both.Homozygous CDKN2A deletions were observed only in invasive melanomas, where they were frequent.This work showed that the commonest first step in escape from the growth arrest in benign nevi is upregulation of TERT via a promoter mutation, or alternatively a defect in CDKN2A.
A recent preprint from the same center reported that TERT promoter mutations in dysplastic nevi were seen only in those nevi without BRAF mutations, but with other MAPK (proliferative signaling) pathway mutations such as in NRAS, instead.Dysplastic nevi with TERT promoter mutations had shorter average telomere lengths than others, and were found in older patients (Lorbeer et al., 2023).Chiba et al. (2017) further reported that a single TERT promoter mutation within a nevus is insufficient to prevent overall telomere shortening, but repairs the shortest telomeres, allowing resumed proliferation.Telomere shortening continues into early melanomas, leading to telomeric crisis and chromosomal instability at this stage, which can be overcome by further TERT upregulation (Chiba et al., 2017).This accords with the common presence of crisis markers in primary melanomas and their common lack of immortality when explanted (Soo, MacKenzie Ross, Kallenberg, et al., 2011).

| Human-mouse molecular differences in cell senescence and nevus arrest
Cell-senescence mediators appear to have undergone relatively recent evolution, consistent with the widely varying lifespans and cancer rates of different vertebrates.For example, ARF in mice (p19 ARF ) is different from and larger than p14 ARF in humans (Ozenne et al., 2010).p16 is not even present in the chicken, only an even-smaller ARF and p16's paralog p15 (INK4B; CDKN2B) (Kim et al., 2003).Mice also have much longer telomeres than humans and yet mouse replicative cell senescence occurs quite rapidly, within weeks, in normal cell-culture conditions, while human cells often take many months to senesce.Mouse cell senescence in culture is not triggered by extensive division and telomere shortening (as in human cells), but typically by oxidative damage through the near-atmospheric oxygen level usually used for culture, well above tissue levels (Coppé et al., 2010).Likewise, mouse cells including melanocytes (Sviderskaya et al., 2002) do not require TERT activation to become immortal, only Cdkn2a loss.Another mammal, the elephant, has 20 copies of the TP53 (p53) gene, and low cancer incidence (Sulak et al., 2016).Caution is thus required in extrapolating senescence-related findings between different mammals and vertebrates; core pathways (Figure 1) are largely shared, but can differ markedly in detail and triggering.
Of the two CDKN2A protein products, p16 broadly appears the more important in humans (see sections 4.1 and 4.2), while ARF appears the more important in mice.Mice with ARF deletion and normal p16 developed a variety of spontaneous tumors, not including melanoma (Kamijo et al., 1997), while p16-null, ARF-competent mice also developed tumors, but at later ages and of only a few types, including melanoma (Sharpless et al., 2001).The difference is also seen in melanocyte senescence and its role in nevus arrest and tumor suppression.Dhomen et al. (2009) reported that p16 was not required for the arrest of nevi initiated in older Braf V600E /Tyr::CreERT2 mice.They did not study ARF, but Ferguson et al. (2010) found that intact ARF was required for arrest of nevi induced by neonatal UV irradiation in other model mice, including Tyr::Nras Q61K , where specific abrogation of ARF led to "skipping" of a nevus stage and early onset of melanoma.
Regarding requirements for senescence, mouse melanocyte strains explanted from skin of ARF-null, p16-competent mice completely failed to senesce, and grew immediately as immortal cells (Ha et al., 2007).In comparison, p16-null, ARF-competent mouse melanocytes were also immortal from the outset, but were dependent for survival on keratinocytes or keratinocyte-derived growth factors (Ha et al., 2007), recalling the similar dependence seen in human p16-null, ARF-competent melanocytes (Sviderskaya et al., 2003).
In cultured human melanocytes however, p16 (not ARF) is required for the normal timing of replicative senescence.Either germline or engineered loss of CDKN2A or of active p16 from human melanocytes, with normal ARF, permits many more cell doublings than in normal adult melanocytes with wild-type p16 (Fung et al., 2013;Gray-Schopfer et al., 2006;Sviderskaya et al., 2003;Zeng et al., 2018).Delayed senescence then occurs in culture, with activation of the p53-p21 pathway (Gray-Schopfer et al., 2006).p53 signaling also appears in melanoma progression in vivo, but appears incomplete and ineffectual in arresting the tumor (MacKenzie Ross et al., 2013), consistent with the finding that p53 remains wild-type in around 80% of human melanomas They showed that diploid human melanocytes were thereby converted to invasive melanoma-like lesions upon xenografting to mice.Without TERT, noninvasive nevus-like lesions were formed (Chudnovsky et al., 2005).

| PROLIFER ATI ON -REL ATED MARKER S AND PERMANENT ARRE S T
As mentioned, some human nevi can regenerate after incomplete excision, proving that these contained some non-senescent nevus cells, but this does not tell us how many cells.Senescence markers label up to 90% of acquired nevus cells where quantitated (section 6), leaving the possibility of a minority of non-senescent cells.Such cells are not easy to detect, however.Nevi show no or very few mitotic figures, varying with stage of evolution (e.g., Mooi & Krausz, 2007).Proliferation markers are likewise reported absent or almost absent from excised human nevi, including Ki67 (Johmura et al., 2014;Michaloglou et al., 2005); cyclins A, D1 and D3 and survivin/BIRC5 (Alonso et al., 2004), and cyclin B (Johmura et al., 2014).Lack of division markers may reflect either senescence, quiescence (reversible arrest), or a mixture.Melanocytes in human nevi are by definition more crowded than normal epidermal melanocytes, whether forming nests or not (Mooi & Krausz, 2007), so density-dependent arrest of some residual non-senescent cells is possible.
A strong test for permanent versus reversible arrest is explantation into culture, whereupon any local inhibitory tissue factors are diluted out and quiescent cells such as normal melanocytes can resume division.Soo, MacKenzie Ross, Kallenberg et al. (2011) prepared such explants from samples of 22 nevi and reported that the great majority of human benign nevus cells failed to proliferate, under culture conditions where normal human melanocytes grew well.The newly-explanted nevus cells often showed a flat and/or multidendritic morphology, typical of senescent human melanocytes (Soo, MacKenzie Ross & Bennett, 2011).Some nevus cultures (15/22) did yield a small number of colonies of proliferating melanocytes among the arrested cells; these dividing cells could be subcultured but senesced after one to several passages, showing several markers of senescence (Soo, MacKenzie Ross, Kallenberg, et al., 2011).They were not able to exclude that the rare proliferating cells were normal melanocytes potentially trapped in the nevus.McNeal et al. (2015) also explanted human nevi (n = 27).They did not observe flat and multidendritic cells, possibly because of a different explantation protocol.Still, they observed no proliferation, again under conditions where normal melanocytes grew well, and they used sequencing to demonstrate the heterozygous presence of the BRAF V600E oncogene in all the arrested cultures tested.These explantation studies provide persuasive evidence that few to no cells within most human benign nevi can proliferate.

| MOLECUL AR AND MORPHOLOG IC AL MARKER S OF S ENE SCEN CE IN H UMAN NE VI
It has been noted that some molecular senescence markers found in nevi, such as DNA damage markers, can also be found in malignant melanomas (MacKenzie Ross et al., 2013;Tran et al., 2012).This has led to questioning of whether these can really be considered as senescence markers or as evidence that nevus cells are senescent (Tran et al., 2012).However, cancer cells are a special case.Senescence gene-expression programs are complex (Gorgoulis et al., 2019), and can be upstream of or parallel to the actual effectors of arrest (Figure 1; Bennett, 2016).Accordingly, when cancer cells escape from senescence by mutations and deletions in the arrest pathways, they do not necessarily lose other elements of the program such as the SASP (senescence-associated secretory program-secretion of cytokines, proteases etc; Gorgoulis et al., 2019).DNA-damage signaling from telomeres can likewise continue after abrogation of its downstream arrest pathways (as seen in early melanoma: MacKenzie Ross et al., 2013), unless or until TERT is sufficiently reactivated to suppress telomere dysfunction (Suram et al., 2012).Accordingly, we should separate three different aims of studying senescence markers in nevi: 1.To ascertain whether nevi are generally senescent, to understand their arrest.Suitable for this would be previously established markers, functionally associated with cell senescence, and absent from normal, non-senescent melanocytes.

To investigate to what extent nevi can also contain non-senescent
cells.This could involve multi-senescence-marker studies on tissue sections (seeking cells that lack all such markers), or positive markers/functional studies of ability to proliferate, as discussed above (Section 5).
3. To seek senescence-related markers that may help with differential diagnosis: present in nevi and reliably lost in early melanoma.
These may be only a small subset of senescence markers in nevi.
Among markers considered below, to the author's knowledge, only p16 is currently in clinical use as a diagnostic aid (Joselow et al., 2017).
No one molecular marker has been found that is present in all kinds of senescent cells and only in senescent cells.Thus, the presence of several established markers is required to demonstrate cell senescence clearly (Gorgoulis et al., 2019).An impressive range of molecular markers of senescence has indeed been detected by immunohistochemistry and immunofluorescence in human nevi (Table 1).
Nuclear location is important because mutant, inactive p16 can be highly expressed but cytoplasmic only (Gray-Schopfer et al., 2006).
All investigators detect a mosaic expression pattern for p16, that is, not all nevus cells are labeled, and sometimes fewer than 50%.At least eight other established molecular senescence markers have been reported in human nevi (Table 1).Most of them are detected in most but not all nevus cells (Table 1), the highest quantitated rate being for DNA-damage marker 53BP1 (around 90% of cells).
An unlabeled subpopulation does not in itself rule out all cells of a nevus being senescent, since completely senescent melanocytic cultures, showing no increase in cell number, typically show only about 60%-80% cells positive for any one senescence marker (Soo, MacKenzie Ross, Kallenberg, et al., 2011), much as with senescent human fibroblasts and β-galactosidase (Dimri et al., 1995).The report of McNeal et al. (2015) on p15(CDKN2B) (paralog of p16) was interesting, as they reported higher detection levels than for p16, in nevi and BRAF V600E -transduced melanocytes.They also reported common loss of p15 in melanoma, although genetic CDKN2B defects were not found to arise commonly at the stage between melanoma and adjacent nevus (Shain et al., 2015).
An important marker common in nevi and other benign tumors is telomere-associated foci of DNA damage (TAFs, also known as TIFs: telomere-induced foci): colocated labeling for a DNA-damage marker and a telomere probe, indicating dysfunctional telomeres (Lai et al., 2018;Suram et al., 2012;Victorelli et al., 2019).The majority of cells in human nevi were found to show TAFs (Suram et al., 2012).
TAFs can represent either critically short telomeres, with shelterin loss, or stable DNA damage within longer telomeres.Damage within telomeres is poorly repaired, exactly because the shelterin complex inhibits DNA repair.It can thus persist stably and trigger senescence.
Such persistent telomeric DNA damage may be the cause of TAFs and senescence in nevi, since their TAF telomeres are not unusually short (Suram et al., 2012).Such DNA damage may arise through ultraviolet irradiation; through reactive oxygen species that are generated by melanin exposed to ultraviolet, especially pheomelanin (Brenner & Hearing, 2008); other unknown pathways, or a combination of these.Interestingly though, it has been noted that there is a minimum telomere length able to bind the PNA probe used in fluorescence (Q-FISH) measurements of telomere length, so the shortest telomeres will give no signal by this method (Lai et al., 2018).
Accordingly, it is possible after all that nevi contain critically short telomeres, not visible by Q-FISH in the TAF method, as a cause of senescence.This would help to explain the finding that a single TERT promoter mutation, found to extend only the shortest telomeres (Chiba et al., 2017), is the commonest genetic change linked to progression of benign nevi (Shain et al., 2015).It also aligns with data in the preprint of Lorbeer et al. (2023), that dysplastic nevi with TERT promoter mutations contain shorter telomeres than other dysplastic nevi (so clonal TERT expression would give a selective advantage).
Certain morphological senescence features are also seen in some types of human nevi.These include large cells with large, pale nuclei and prominent nucleoli; and a multinucleate cell subpopulation (Gray-Schopfer et al., 2006;Mooi & Krausz, 2007), also partial tetraploid DNA content (Johmura et al., 2014).There is evidence that diploid human cells can arrest for senescence in both G1 and G2 phases of the cell cycle, or in G1 with tetraploidy after skipping mitosis (Johmura et al., 2014;Mao et al., 2012).This may sometimes be because the M-phase CDK, CDK1, can be inhibited by p21, the universal CDK inhibitor (Figure 1; Sherr & Roberts, 1995), although unlikely in benign nevi, where little or no p21 is detectable (Gray-Schopfer et al., 2006;MacKenzie Ross et al., 2013).It may also happen through inhibition of CDK4 by p16, since CDK4 activates the G2/M master regulator FOXM1 (Anders et al., 2011), required for transcription of various mitotic effectors.

| CON CLUS I ON: WHAT K IND OF S ENE SCEN CE?
To summarize concerning the mouse lesions called nevi by Ruiz-Vega et al. (2020), they were unlike human nevi.There was good evidence that these crowded, microscopic colonies were arrested by cell interactions rather than cell senescence.However, extrapolation of this conclusion to human nevi seems unwarranted, in the face of wide-ranging evidence for senescence in banal, acquired human nevi, as reviewed above.Regeneration by some human nevi shows that some nevus cells can divide, especially in recently initiated nevi that are still growing, and also in some clinically suspicious nevi that are excised.However, marker and explantation studies indicate that any such non-senescent subpopulation is typically small.Should we call the senescence in nevi "oncogene-induced"?I agree with Suram et al. (2012), Damsky and Bosenberg (2017) and Ruiz-Vega et al. (2020) that it does not match the concept of OIS.As with fibroblasts, overexpression of BRAF V600E from a strong viral or housekeeping promoter in human melanocytes induced immediate OIS, after zero to a few cell divisions, not nearly sufficient to generate a visible nevus (Gray-Schopfer et al., 2006;Michaloglou et al., 2005;Zeng et al., 2018).This was not prevented by prior abrogation of p16, showing that OIS following oncogene overexpression does not require p16.However, when CRISPR-Cas9 was used to introduce the activating BRAF V600E mutation into genomic BRAF with its own promoter, in human melanocytes, the cells proliferated rather than undergoing immediate OIS (Zeng et al., 2018).This mimics nevi, which have obviously proliferated initially following the mutation.(Ressler et al., 2006) and 53BP1 (Suram et al., 2012).Presumed to be senescent.c Antigen retrieval at pH 9.0 recommended (Kohli, Mir, et al., 2017).Sviderskaya et al., 2003).Accordingly, the eventual senescence in nevi does not resemble OIS as seen in culture.Instead, nevi undergo oncogene-induced proliferation, followed after many divisions by cell senescence.
Reports of limited telomere shortening in nevi (Michaloglou et al., 2005) implied that nevus cell senescence did not resemble replicative senescence closely either.However, this now needs to be revisited following evidence that the shortest telomeres are invisible to a PNA probe (Lai et al., 2018), and preprint data that TERT promoter mutations in dysplastic nevi are associated with shorter telomeres (Lorbeer et al., 2023).At any rate there are abundant TAFs in cells of nevi, indicating either critically short telomeres or telomeric DNA damage, and TERT promoter mutations are associated with nevus progression (Chiba et al., 2018;Shain et al., 2015Shain et al., , 2016;;Suram et al., 2012).The broad implication is that senescence in human nevi may sometimes after all be replicative senescence, and at present can broadly be described as TDIS: telomere-dysfunction-induced senescence (Suram et al., 2012).These conclusions are summarized in Figure 2.
Some unanswered questions remain.One is that of how to integrate the genetic and biological evidence for a role of p16 in nevus arrest with the histopathological finding that p16 is detectable in only a proportion of nevus cells by immunostaining.Another is the issue that telomerase is thought to be active only in proliferating cells, not in arrested cells (Holt et al., 1997), so how does it begin to repair telomeres after arrest?Perhaps rare nevus cells acquire such TERT promoter mutations already in the growth phase of the nevus, especially given that senescence can be a gradual process rather than an abrupt switch: some human cells can continue some divisions with one or a few DNA-damage foci (Nassrally et al., 2019).Colebatch et al. (2019) did report subclonal TERT promoter mutations within clinically benign nevi.
To conclude, concerning whether moles are senescent: the literature provides strong evidence that cell senescence is the primary cause of arrest in human banal nevi and that this arrest is a key tumor-suppression mechanism for melanoma, although some such nevi must include a minority of reversibly arrested cells.The senescence in nevi has the properties of TDIS rather than canonical OIS.

ACK N OWLED G M ENTS
Dr Jaskaren Kohli (University College London, UK) is warmly thanked for valuable comments on the manuscript.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The author declares no conflict of interest.
(review:Bennett, 2016).Normal repression of TERT is also of course important in melanocyte senescence and normal nevus size.Reexpression of TERT plus abrogation of the p16/RB pathway were found to be necessary and sufficient to immortalize human melanocytes (Gray-Schopfer et al., 2006).Chudnovsky et al. also blocked human melanocyte senescence by viral transduction of three dominant mutations blocking the p16/RB and p53 pathways and expressing TERT, and they added activated NRAS as driver.

F
Updated model for human nevus senescence and escape.Activation of BRAF or NRAS by natural mutation (Δ) leads to transformation -proliferation through reduced requirement for external mitogens and endogenous upregulation of the MAPK mitogenic pathway.A combination of DNA damage within telomeres and telomere shortening then leads to TDIS in most cells, arresting growth as a banal nevus.A few cells are shown without TAFs, representing the possible presence of some non-senescent cells.Escape from this senescence is rare, and its most common first step is either a TERT promoter mutation or a CDKN2A defect.
(Fung et al., 2013;Gray-Schopfer et al., 2006rment of p16 delayed melanocyte replicative senescence (allowed additional population doublings), as previously reported(Fung et al., 2013;Gray-Schopfer et al., 2006; Senescence markers found in human banal/benign nevi. a Only congenital nevi tested, in both studies, by histochemistry (frozen sections).All other markers were assessed in formalin-fixed sections of acquired nevi.b Note, a few normal melanocytes in adult human skin, in numbers increasing with age, label with p16