Expression of oncogenic BRAF^V600E in melanocytes induces Schwannian differentiation in vivo

Dear Editor

Although melanocytes represent one of several terminally differentiated neural crest-derived lineages, they exhibit a surprisingly high degree of phenotypic plasticity (Adameyko et al., 2009, Real et al., 2006). Interestingly, the cell fate switching in melanocytes can be observed in vivo under pathologic conditions including nevi, as nevus cells phenotypically resemble Schwann cells in multiple ways (Aso et al., 1988). However, it is unclear whether the phenotypic switch occurs coincidental with nevus formation or whether it functionally contributes to the tumor suppressive effect of nevigenesis.

BRAF^V600E is the most prevalent mutation in melanoma (Girotti et al., 2014). Our laboratory generated a transgenic mouse in which BRAF^V600E is expressed in melanocytes (Goel et al., 2009). The BRAF^V600E mice exhibited skin hyperpigmentation, melanocytic hyperplasia and developed melanomas at a low frequency; moreover, metastatic melanomas can be induced in combination with Cdkn2a^+/-, Tp53^+/-, Arf^-/- or UV irradiation (Luo et al., 2013).

Two founder lines of the transgenic mice with high and low BRAF^V600E expression, denoted 470 and 476, respectively, were backcrossed to C57Bl/6 mice. The resultant progeny retained the phenotypes observed in the mixed background, with profound hair graying being prominent (Figure 1A). Histology confirmed accumulation of pigmented cells in cutaneous regions and deep dermis, consistent with elevated pigment contents (Figure S1A-B). The premature hair graying phenotype, which was observed as early as postnatal day 7 (P7, Figure S1C), was a result of white and black hair mixture (Figure S1D). Hair pigmentation is a direct contribution of hair follicle melanocytes. Using tyrosinase-related protein-2 as a lineage tracker, we found BRAF^V600E melanocytes displayed a developmental defect in proper localization, leaving the majority of hair follicles devoid of melanocytes, and thereby melanin, in aged skins (Figure S2A-B).

Figure 1. BRAF^V600E transgenic mice develop benign nerve sheath tumors in the skin.

(A) Morphology of 6-month-old BRAF^V600E mice and wild-type littermate. (B) Immunohistochemical (IHC) staining of Ki-67 on skins from age-matched BRAF^V600E and WT mice. Arrowheads indicate Ki-67 positive cells in the dermis. (C) Senescence-associated β-gal staining in a BRAF^V600E skin. Arrow indicates senescent cells; arrowheads suggest pigmented lesions without senescence. (D) H&E staining of skin sections from BRAF^V600E mouse at different ages. Arrowheads indicate Schwannian masses in the subcutis or adjacent to the hair follicle bulb; circles highlight Schwannian tumors in the superficial dermis. Boxed areas are enlarged as indicated. (E-F) Expression of neurofibromin 1 and Schwannian markers, including myelinating Schwannian markers myelin basic protein (Mbp) and myelin protein zero (Mpz), as well as non-myelinating Schwannian marker glial fibrillary acidic protein (Gfap) in primary melanocytes from WT and BRAF^V600E mice. Data represent mean+/-SD. n=3, *p<0.05, **p<0.01, student t-test. (G) IHC staining of human BRAF^V600E (VE) in skins from age-matched WT and BRAF^V600E mice. (H) H&E staining of skins from BRAF^V600E mice (476 line) with or without deletion of Arf or Cdkn2a. Bars represent 100μm if not otherwise indicated.

Although BRAF^V600E melanocytes undergo hyperproliferation (Figure 1B), spontaneous melanomas are rare. Oncogene-induced senescence undoubtedly serves as the primary barrier to malignant transformation (Dhomen et al., 2009); however, not all BRAF^V600E melanocytes displayed a senescent phenotype (Figure 1C, arrowhead). Instead, ectopic melanocytes in BRAF^V600E skins emerged as cutaneous masses, which were initially detected in P13 subcutis and became evident near hair follicle bulk at P22 (Figure 1D, arrowhead). These masses progressed over time to form intermediately to highly pigmented tumors in both superficial dermis and subcutis (Figure 1D, circle). Morphologically, the masses contained spindle and epithelioid cells forming oval or polygonal nests and fascicles with smooth and defined boundaries, reminiscent of Schwannian differentiation. Histopathologically, the masses resembled hyperpigmentation overlying (plexiform) neurofibroma, a benign nerve sheath tumor derived from Schwann cells occurring frequently with neurofibromin 1 (Nf1) inactivation.

Despite Nf1 transcript was unaffected, chronic activation of BRAF^V600E significantly triggered several Schwannian markers in melanocytes, suggesting oncogenic BRAF^V600E reprograms melanocytes toward the Schwannian lineage (Figure 1E-F). This Schwannian phenotype was indeed a consequence of the BRAF^V600E in vivo, as all the masses stained positive for BRAF^V600E protein (Figure 1G). This finding was not an artifact of insertional effects in the 470 BRAF^V600E line, as similar tumors were present in the 476 line in the context of Arf or Cdkn2a loss (Figure 1H), and Schwannian-like differentiation was also reported in other melanoma models in which the BRAF^V600E is expressed from the endogenous locus (Marsh Durban et al., 2013, Damsky et al., 2015). Intriguingly, the Schwannian masses seldom progressed to malignancy, remaining benign for the life-span of the animals, suggesting the trans-lineage differentiation between melanocytes and Schwann cells is tumor suppressive, supplementing senescence to inhibit melanoma formation in response to BRAF^V600E. Indeed, features characteristic of Schwann cells can be detected in human nevi, while invasive melanoma cells never adopt a Schwannian phenotype (Aso et al., 1988, Reed et al., 1999).

It is likely that melanocytes driven into the Schwannian differentiation “trap” by BRAF^V600E must be diverted back to the melanocytic pathway before they can undergo malignant transformation. AKT is likely to play a key role in this process, as AKT activation has been reported to retard Schwann cell differentiation (Sherman et al., 2012), whereas AKT activation positively correlates with melanoma progression. Interestingly, we found that active AKT was absent in the BRAF^V600E precancerous lesions containing Schwannian masses but abundant in melanomas (Figure 2A-B). We also observed that during progression of BRAF^V600E melanocytes to melanoma, expression of Schwannian markers was drastically reduced (Figure 2C), indicating suppression of the Schwannian phenotype occurs concomitant with transformation. Furthermore, when we treated BRAF^V600E mouse melanoma cells with AKT inhibitor, Schwannian markers were significantly elevated (Figure 2D), confirming AKT suppression enhanced Schwannian differentiation of melanoma cells.

Figure 2. AKT activation reverses the Schwannian phenotype.

(A-B) pAKT-S473 immunostaining (A) and immunoblot (B) on corresponding precancerous skin and malignant melanoma from the BRAF^V600E mice. (C) Expression of Schwannian markers in primary BRAF^V600E melanocytes and BRAF^V600E melanoma cells (4246 and 4228). (D) Expression of Schwannian markers is upregulated upon AKT inhibition in BRAF^V600E melanoma cells. Data represent mean+/-SD. n=3, *p<0.05, **p<0.01, student t-test. (E) A model of phenotypic switching between melanocytes and Schwann cells and its involvement in melanomagenesis. During development, the bipotent glial-melanocytic progenitors (GMP) can give rise to either differentiated melanocytes (Mc) or Schwann cells (Sch) in the skin under the guidance of lineage-specific transcription factors or microenvironmental cues. When an oncogenic mutation, such as BRAF^V600E, occurs in these progenitor cells, it biases the differentiation towards the Schwannian lineage leading to the formation of benign nerve sheath tumors, an apparent “dead end” event restricting melanoma development. Certain cells gain AKT activation resulting in suppression of the Schwannian program, leading to the development of tumor foci. Therefore, under the dual stimulation of oncogenic BRAF^V600E and Schwannian-suppressing AKT activation, melanoma is induced. However, it is unknown whether the melanoma-promoting effect of AKT activation is via a reversion into the melanocytic pathway or a direction transition into malignancy.

Taken together, oncogenic BRAF^V600E in melanocytes, in addition to instigate senescence, rewires the schwannian differentiation program, leading to benign nerve sheath tumors. As such schwannian tumors can remain benign for the life-span of the animals, they represent a dead end for oncogenic BRAF^V600E thereby functioning as a novel melanoma suppressive mechanism. An individual BRAF^V600E cell, in which AKT is hyperactivated, will downregulate the Schwannian program, and presumably reactivate the melanocytic pathway to arise as a tumor focus, ultimately leading to melanoma initiation (Figure 2E). This hypothesis not only provides another layer of functionality to AKT in melanomagenesis, namely, lineage switching from the tumor-suppressive Schwannian pathway to the tumor-prone melanocytic lineage, but also emphasizes the high degree of phenotypic plasticity of melanocytes. Thus, deciphering the mechanisms that regulate the plasticity and maintenance of neural crest-derived differentiated cells is likely to be an important step towards understanding mechanisms of melanoma progression.