MITF-M, a “melanocyte-specific” isoform, is expressed in the adult retinal pigment epithelium

Dear Editor

The gene residing at the microphthalmia locus that causes white and small eye phenotype in mice when mutated was cloned and named microphthalmia-associated transcription factor (Mitf) nearly two decades ago (Arnheiter, 2010; Hodgkinson et al., 1993). In the ensuing years, MITF has become one of the most extensively studied transcription factors in the field of pigment cell research. It turns out to be a critical and indispensable factor for several cell lineages, including melanocytes, cells of the retinal pigment epithelium (RPE), mast cells, and osteoclasts (Goding, 2000). In addition, it plays critical, although not yet fully defined, roles in melanoma development and metastasis (Hoek and Goding, 2010). Within the eye, MITF function is required for normal RPE development, and it has been directly implicated in regulating the expression of pigment-related genes (Bharti et al., 2006; Goding, 2000; Martinez-Morales et al., 2004) as well as non-pigment genes, such as BEST1 (Esumi et al., 2007; Masuda and Esumi, 2010).

An interesting feature of MITF is its gene structure and intricate pattern of alternative splicing that involves the use of multiple promoters. The mouse and human MITF genes both contain at least ten distinct first exons, seven of which are coding and three are non-coding (Bharti et al., 2008; Li et al., 2010; Shibahara et al., 2001). “Alternative promoter usage” creates ten different MITF isoforms at the mRNA level, leading to MITF proteins with eight distinct N-termini connected to a common C-terminal major portion of the protein (Bharti et al., 2008; Shibahara et al., 2001). The first cloned mouse Mitf is designated as the M isoform (Mitf-M). MITF-M has been universally regarded as being melanocyte-specific based on the finding that it is highly expressed in melanocytes but it has not been detected in other cell types, including RPE cell lines (Shibahara et al., 2001). In mouse RPE specification and differentiation, Mitf-H and Mitf-D have been shown to play important roles (Bharti et al., 2008). The functional difference among the MITF isoforms is an important area that has not yet been fully explored.

We have been studying the regulatory mechanisms of gene expression in the RPE using BEST1 as a model system. During the course of our studies, we repeatedly detected MITF-M expression in both adult RPE obtained directly from human donor eyes and cultured human RPE primary cells with a few passages (Esumi et al., 2007; Masuda and Esumi, 2010). Because the results were unexpected and contradicted the conventional dogma that MITF-M is melanocyte-specific, we decided to further investigate the expression of MITF-M in adult RPE cells. Additionally, since MITF isoforms in the RPE have been studied in either mouse embryonic RPE or human RPE cell lines, but not in fully differentiated and functionally mature RPE cells, we also quantitatively compared the mRNA level of MITF-M with that of three other MITF isoforms (A, H, and D) which are known to be expressed in or important for RPE development (Bharti et al., 2008). Human donor eyes were obtained through the National Disease Research Interchange (NDRI), and bovine eyes were obtained from a local slaughterhouse (Supplementary Method S1). RPE primary cultures obtained from human donor eyes were prepared and described previously (Masuda and Esumi, 2010).

First, we analyzed the expression of MITF-M using conventional reverse transcription PCR (RT-PCR), along with three RPE markers (OTX2, RPE65, and KRT8), two melanocyte markers (PAX3 and MLANA), and β-actin (ACTB) as control. MITF-M mRNA was clearly detected in both human adult RPE and RPE primary cells with a few passages (M1), although the expression levels of MITF-M were significantly higher in two melanoma cell lines, SK-MEL-5 and SK-MEL-23, as compared to the RPE cells (Fig. 1A). Not surprisingly, MITF-M expression was undetectable in the two most commonly-used human RPE cell lines, ARPE19 and D407, or cell lines that are unrelated to RPE or melanocytes, SK-N-MC and HEK293. All of the three RPE markers were nicely detected in adult RPE, but the M1 primary RPE cells did not show detectable RPE65 expression (Fig. 1A). Since RPE65 is a critical enzyme in the visual cycle, one of the most unique functions of mature RPE, the results suggest that M1 cells likely dedifferentiated to some extent during culture. Consistent with this possibility, although M1 cells showed a nice cobblestone-like appearance, they lacked visible pigmentation. The most critical issue was to rule out the possible contamination of the RPE sample with choroidal melanocytes, which could have explained the presence of MITF-M transcripts in our RPE sample. To assess this possibility, we analyzed expression of two melanocyte markers, PAX3 and MLANA. While both markers were nicely expressed in the melanoma cells, neither of them was significantly detected in the RPE or M1 (a barely detectable MLANA band appeared only after long exposure). We further tested the possibility of contamination using quantitative RT-PCR (RT-qPCR). The PAX3 mRNA level detected in the RPE sample was 0.66% and 0.45% of that in SK-MEL-5 and SK-MEL-23 cells, respectively; for MLANA, the levels were 0.31% and 1.7%, respectively (Supplementary Fig. S1). These relative levels were significantly lower than those of MITF-M, of which the mRNA level in the RPE sample was 1.2% and 13.4% of that in SK-MEL-5 and SK-MEL-23, respectively (Fig. 1B).

Figure 1. Analysis of MITF-M expression in adult RPE cells.

(A) MITF-M expression in human adult RPE. Total RNA was prepared from human adult RPE (labeled as RPE), M1 human RPE primary cells (M1), and six human cell lines that are derived from RPE (ARPE19 and D407), melanoma (SK-MEL-5 and SK-MEL-23), neuroblastoma (SK-N-MC), and transformed embryonic kidney (HEK293). First-strand cDNA was synthesized from 2 μg of total RNA with random primers, and the mRNA levels of the indicated genes were analyzed by PCR using gene-specific primers (Supplementary Table S1). (B) Quantitative analysis of MITF-M expression. The mRNA level of MITF-M was analyzed by RT-qPCR with the same RNA samples as analyzed in A. Based on threshold cycle values, the mRNA level of MITF-M was normalized by the geometric mean of three reference genes, CREBBP, FBXL12 and SRP72 (Supplementary Table S1), and results were presented as relative mRNA level. The values represent the means and SEM (bar). (C) MITF-M expression in adult bovine RPE. The mRNA levels of MITF-M, MLANA, and RPE65 were compared between bovine RPE and choroid samples by RT-qPCR using FBXL12 as control (Supplementary Method S1 and Supplementary Table S2). The ratio of mRNA level in the RPE to that in the choroid (RPE/choroid ratio) was 0.195, 0.008, and 46.7 for MITF-M, MLANA, and RPE65, respectively. The difference in the RPE/choroid ratio between MITF-M and MLANA was statistically analyzed by t-test (*p = 0.0000925). The values represent the means and SD (bar). (D) Expression of MITF-A, -H, and -D. Three additional MITF isoforms were analyzed and presented in the same manner as in B. (E) Absolute quantification of MITF isoforms. A cDNA fragment of each MITF-M, A, H, and D isoform was made by RT-PCR and used to generate a standard curve for quantification in RT-qPCR analyses (Supplementary Method S2). The absolute amount of cDNA in human adult RPE and M1 RPE primary cells was calculated from a standard curve for each isoform.

To test whether MITF-M is indeed expressed in mature RPE in different species as well as to further rule out the possible contamination of not only choroidal melanocytes but also other choroidal cells in the RPE sample, we used bovine eyes because of the availability of ample fresh materials. We extracted RNAs from 6 RPE and 8 choroid samples independently and analyzed the mRNA levels of MITF-M, MLANA, and RPE65 by RT-qPCR using FBXL12 as control for normalization (Supplementary Method S1). The ratio of mRNA level in the RPE samples to that in the choroid samples (RPE/choroid ratio) was 0.195 for MITF-M and 0.008 for MLANA, and the difference was statistically significant (p = 0.0000925) (Fig. 1C), indicating that MITF-M is expressed in adult RPE cells at a level above that which can be explained simply by contamination with choroidal cells. The RPE/choroid ratio for RPE65 was 46.7, suggesting that the possible reverse contamination of RPE in the choroid samples was minimal. These results also indicate that MITF-M expression is a common feature of adult mammalian RPE cells.

Next, we compared the expression levels of MITF-M, MITF-A, MITF-H, and MITF-D as well as pan-MITF including all isoforms in human RNA samples by RT-qPCR. MITF-M mRNA was present in both RPE and M1 at the levels that were 10- and 100-fold higher than those in ARPE19 and D407, respectively, but 100- and 10-fold lower than those in SK-MEL-5 and SK-MEL-23, respectively, as described earlier (Fig. 1B). The overall expression pattern of MITF-M was similar to that of pan-MITF (Supplementary Fig. S2). In contrast, MITF-A was expressed in all cell-types at relatively comparable levels (Fig. 1D). The expression of MITF-H and MITF-D showed a similar pattern in that they were expressed at the highest level in M1, with the levels that were approximately 4.5 and 3.0 times higher than those detected in adult RPE, respectively (Fig. 1D). The expression of H and D isoforms was consistently higher in ARPE19 than in D407, but low or undetectable in the melanoma cells. These results suggest that the expression of H and D isoforms may decrease as RPE cells proceed toward terminal differentiation.

Finally, we wanted to know the absolute level of MITF-M in comparison with that of other three RPE-expressing MITF isoforms (A, H, and D) in human adult RPE cells. A cDNA fragment of each isoform was amplified from M1 RNA by RT-PCR, purified by agarose gel fractionation, quantified by spectrophotometric analysis, and used to generate a standard curve for quantification in RT-qPCR analyses (Supplementary Method S2). The absolute level of MITF-M mRNA in adult RPE is comparable to that of MITF-H and MITF-A, two major MITF isoforms in the RPE (Fig. 1E). These results indirectly imply that MITF-M may be functional in the adult RPE, although its exact function in RPE cells remains to be determined. It is interesting that the proportional profile of MITF isoforms in our adult RPE sample is different from that reported for mouse embryonic eyes (Bharti et al., 2008). In addition, the MITF isoform profile of M1 RPE primary cells, in which MITF-H composed a large proportion, was different from that of adult RPE cells which was composed of H, M, and A isoforms at a comparable ratio, suggesting that the isoform expression can change during the course of primary culture that involves dedifferentiation and proliferation from the post-mitotic mature state (Fig. 1E).

There are a number of possible reasons why the expression of MITF-M in the RPE has been overlooked to date. Earlier studies used RPE cell lines such as ARPE19 and D407 as surrogate RPE cells to analyze the expression of different MITF isoforms; however, RPE cell lines, like most cell lines, do not accurately represent their in vivo counterparts (Masuda and Esumi, 2010). Additionally, MITF function and its isoforms have been studied mostly in embryonic RPE to date (Bharti et al., 2008; Bharti et al., 2006; Martinez-Morales et al., 2004), rather than in adult, mature tissue. Another possible factor is that since the expression of MITF-M is much lower in the RPE compared to that in melanocytes, if the assays employed were not sufficiently sensitive, then the RPE-derived MITF-M transcripts might not have been detected. For mouse studies, impure samples could have been an issue. Due to the small size of mouse eyes, expression analysis generally utilized RNA extracted from RPE/choroid mixed tissues that contain choroidal melanocytes; as a result, when Mitf-M expression was detected in such preparations, investigators interpreted that the observed Mitf-M mRNA was derived from choroidal melanocytes.

It should also be noted that Mitf-M expression has previously been reported in non-melanocytes. Mitf-M mRNA was detected by RT-PCR in cultured mast cells from both mouse spleen (CMCs) and peritoneal mast cells (PMCs), and Mitf-M expression was higher in PMCs than CMCs (Oboki et al., 2002). In addition, while Mitf-E was preferentially expressed in CMCs, only Mitf-M mRNA was detectable in PMCs. Since CMCs and PMCs are regarded as immature and fully matured mast cells, respectively, the authors suggested that mast cells express different MITF isoforms specific to their differentiation stages (Oboki et al., 2002). Our results suggest a similar scenario that MITF-M might be expressed in the RPE in its more differentiated stages, with MITF-H and -D being expressed also in a differentiation stage-dependent manner but in an opposite direction to MITF-M. It is interesting to note that Mitf^mi-bw (Mitf black-eyed white) mutant mice that lack Mitf-M expression due to an insertion of an L1 element into intron 3 of the Mitf gene show normal eye development with pigmentation, but lack choroidal melanocytes (Yajima et al., 1999). Therefore, unlike the situation with melanocytes, MITF-M is not essential for RPE development. However, a hint for the possible functional role that MITF-M may play in adult RPE cells is our finding that the amount of MITF-M mRNA is comparable to that of MITF-H and MITF-A, two major MITF isoforms in the RPE. Further studies will be needed to define the role of MITF-M in mature RPE, and to determine if it has functions that are distinct from those of the other MITF isoforms. To the best of our knowledge, this is the first report that described the proportional composition of MITF isoforms in adult RPE, which is different from that in mouse embryonic RPE (Bharti et al., 2008) or human RPE primary cultures (this study). Therefore, the characteristic MITF isoform profile may serve as one of the mature RPE markers in the process of producing fully differentiated RPE cells from embryonic stem (ES) and/or induced pluripotent stem (iPS) cells.

In summary, our results demonstrate that terminally differentiated bovine and human adult RPE cells express MITF-M. Our findings raise a number of questions: 1) when and at what stage of differentiation do RPE cells begin to express MITF-M, 2) is MITF-M expressed in a differentiation-dependent manner, 3) what functional role does MITF-M play in adult RPE, 4) is the function of MITF-M in adult RPE redundant with that of the other expressed isoforms, or does it have some unique function, and 5) how is the expression of MITF-M regulated in adult RPE cells? All of these questions remain to be answered, but it is clear that the conventional wisdom that the MITF-M isoform is melanocyte-specific needs to be revised.