Dr Lluís Montoliu Centro Nacional de Biotecnología (CNB-CSIC), Department of Molecular and Cellular Biology, Campus de Cantoblanco, 28049 Madrid, Spain. E-mail: firstname.lastname@example.org
Pigmentary genes were among the first mammalian genes to be studied, mostly because of the obvious phenotypes associated with their mutations. In 1990, tyrosinase, encoding the limiting enzyme in the melanin synthesis pathway, was eventually assigned to the c (albino) locus by classical rescue experiments driven by functional constructs in transgenic mice. These pioneer reports triggered the study of the regulation of endogenous tyrosinase gene expression by combining different amounts of upstream regulatory and promoter regions and testing their function in vivo in transgenic animals. However, faithful and reproducible transgenic expression was not achieved until the entire tyrosinase expression domain was transferred to the germ-line of mice using artificial-chromosome-type transgenes. The use of these large tyrosinase transgenic constructs and the ease with which they could be manipulated in vitro enabled the discovery of previously unknown but fundamental regulatory regions, such as the tyrosinase locus control region (LCR), whose presence was required in order to guarantee position-independent and copy-number-dependent expression of tyrosinase transgenes, with an expression level, per copy, comparable to that of an endogenous wild-type allele. Subsequently, functional dissection of elements present within this LCR through the generation of new artificial-chromosome type tyrosinase transgenes has revealed the existence of different regulatory activities. The existence of some of these units had been suggested previously by standard-type transgenic analyses. In this review, we will discuss both independent approaches and conclude that optimal tyrosinase transgene expression requires the use of its complete expression domain.
The available first drafts of mammalian genomes from human and mouse have shown that, approximately, 5% of DNA could correspond to genes, classically identified as coding information, whereas the rest of sequences, frequently referred to as `junk' or (better) intergenic DNA, remain largely unknown and uncharacterized. However, most regulatory elements, responsible for the proper expression of a gene in time and space, are located in this intergenic DNA. Hence, eukaryotic genes are organized on chromosomes as contiguous but independent units called expression domains (1–5). These expression domains are thought to remain insulated from neighbouring sequences (and surrounding genes) and are believed to contain all regulatory elements necessary for correct gene expression. The existence of boundary elements, insulating adjacent but unrelated expression domains, has been recently demonstrated in a limited number of loci and proposed as one of the most important regulatory elements of a gene (6, 7).
Gene transfer experiments represent classical approaches used by investigators to functionally identify regulatory elements, especially by testing DNA constructs with different amounts of upstream and/or downstream regulatory sequences in transgenic animals. However, it is not unusual to find pivotal regulatory sequences located at long distances from the promoter of the gene [i.e. human beta-globin locus (8), mouse GATA-3 gene (9), mouse Myf-5 gene (10) and mouse tyrosinase gene (11, 12)]. These distant regulatory elements usually will not be properly included in standard-type transgenic constructs (plasmids), with a limitation in cloning capacity for heterologous sequences, resulting in suboptimal in vivo performance. These problems are generally manifested by low or undetectable expression levels, variegation, ectopic expression or a combination of some or all of these effects, usually known as position effects (13–15). The progressive addition of previously unknown regulatory regions from a given expression domain might overcome, in some cases, these position effects (16). However, since 1993 it has been known that artificial chromosome-type constructs [first yeast artificial chromosomes (YACs) but thereafter BACs and PACs, vectors characterized by much larger cloning capacities than standard plasmids, ranging from 100 to 1000 kb] are the best way to ensure that most of the regulatory elements characteristic of a given expression domain are included in a transgenic construct (5, 16).
Pigmentary genes, such as tyrosinase, constitute highly valuable experimental models to functionally analyse and dissect expression domains in detail (16–18). Direct evaluation of functional tyrosinase constructs is possible in albino genetic backgrounds by simple visual inspection as tyrosinase is the key enzyme of melanin biosynthesis and, along with related enzymes [including tyrosinase-related protein 1 (TYRP1) and dopachrome tautomerase (DCT)] form a multienzymatic protein complex within melanosomes, which are responsible for the production of pigment (19).
Mutations within the tyrosinase gene inactivating its function result in oculocutaneous albinism type I, the most common type of albinism (reviewed in Ref. 20). A complete deficiency of tyrosinase protein is one of the few examples of a mutant phenotype the expression of which is constant regardless of genetic background (21). In mice, the classical albino mutation corresponds to a single-point mutation in the first exon of the tyrosinase gene, which results in an amino acid mutation Cys103Ser, leading to the accumulation of a non-functional protein (22, 23); reviewed in Ref. (18). Recently, this mutant albino tyrosinase protein has been found abnormally retained at the endoplasmic reticulum (24).
These pigmentary proteins (tyrosinase, TYRP1 and DCT) show 40% homology and form the tyrosinase-related protein family (25). Their corresponding pigmentary genes share a transcriptional regulatory pathway operated by the transcription factor Microphthalmia, a protein that integrates several independent signal transduction pathways (26, 27). In the mouse tyrosinase gene, Microphthalmia targets two specific sequences, an M-box at –100 bp and an E-box at the vicinity of the start of transcription (28, 29). As little as 270 bp from the mouse tyrosinase promoter, including the M and E boxes, are sufficient to match the expression pattern of the endogenous gene in cells (28, 30, 31) and in transgenic experiments (17, 32, 33). Analogous sequences are found within the 5′ upstream regulatory region of Tyrp1 and Dct genes (26, 29).
The mouse tyrosinase gene is tightly regulated during development. It is expressed from developmental day +16.5 in skin and hair bulb melanocytes, derived from neural crest, and from day +10.5 in the retinal pigment epithelium (RPE) cells, derived from the optic cup (33, 34).
THE MOUSE TYROSINASE GENE IN TRANSGENIC EXPERIMENTS
The albino phenotype has been corrected in mice (17, 23, 31–33, 35–40), rabbits (41), fish (42, 43) and other vertebrates expressing tyrosinase functional transgenes. Moreover, the direct and visual detection of pigmentation in tyrosinase transgenic animals generated in albino genetic backgrounds was repeatedly proposed by independent teams as a marker in coinjection strategies for rapid detection of successful transgenesis (44–48).
However, all these standard tyrosinase constructs driven by limited amount of tyrosinase regulatory sequences (ranging from 270 to 5500 bp 5′ upstream promoter sequences) displayed high degree of variability in pigmentation between independent lines (39, 40, 45–48), resulting in variegation (23, 38, 47, 48), and did not normally reach levels found in the wild-type phenotype (35, 36, 44, 46–49). These abnormal expression patterns have been currently explained by position effects, directly correlated with the random site of integration of these standard tyrosinase transgenes in the host genome. Taken together, these results suggested that other regulatory regions within the tyrosinase expression domain were required to sustain faithful tyrosinase transgenic expression, independent of integration site.
In contrast, the generation of transgenic mice with a 250-kb YAC (YRT2, see Figs 1 and 2) covering the whole mouse tyrosinase locus totally rescued the albino phenotype of recipient animals (11), resulting in mice that were indistinguishable from agouti wild-type mice. The expression of YRT2 transgenes was independent of site of integration, dependent of transgene copy-number and with an expression value, per copy, comparable to that of the endogenous gene (11, 49, 50). Furthermore, visual abnormalities commonly associated with albinism, which include central retina underdevelopment, rod photoreceptor cell deficit and misrouting of axons from retinal ganglion cells at the optic chiasm (51), appeared also faithfully corrected in YRT2 YAC tyrosinase transgenic mice, as compared with non-transgenic albino mutant individuals (52, 53). These results clearly indicated the existence of important regulatory elements, absent in previous standard constructs but responsible in YAC tyrosinase transgenes to overcome observed position effects.
THE DISCOVERY OF A LOCUS CONTROL REGION (LCR) AT –12 KB IN THE MOUSE TYROSINASE GENE
Yeast artificial chromosome YRT2 was 250 kb in size, including about 100 kb of promoter and immediately 5′ upstream regulatory sequences, five exons (and their corresponding four introns) and limited 3′ downstream regulatory sequences plus additional 150 kb of further distant 5′ upstream sequences, potentially encompassing unknown regulatory elements that were responsible for the correct expression of YAC tyrosinase transgenes (see Fig. 2) (11, 54). One of these elements, previously identified during the characterization of the molecular basis of the chinchilla-mottled (Tyrc–m) allele of the tyrosinase gene (55), was found to contain a hypersensitive DNase I site which operated as a cell-specific enhancer in vitro and in vivo (31, 39).
The relevance of this element was evaluated in vivo by generating specific deletions within YAC tyrosinase transgenes resulting in the description of a novel LCR at –12 kb in the mouse tyrosinase gene (12). LCRs are regulatory elements defined functionally, according to the pioneering work carried out with the human β-globin genes (8, 56, 57), which harbour the best example of known LCRs, a set of complex regulatory sequences that enable a linked transgene to be expressed in a position-independent and copy-number-dependent manner, with a value per copy comparable to that of the endogenous alleles. At present, a number of LCR and LCR-like regulatory sequences have been identified in various genes (recently reviewed in Ref. 58).
In YAC tyrosinase transgenics, the albino phenotype was totally rescued in animals whose transgenes kept the LCR (YRT2 and YRT3, see Figs 1 and 2). In contrast, pigmentation (and hence, transgene expression) was much weaker in transgenic mice in which the LCR was deleted or substituted (YRT4 and YRT5, see Figs 1 and 2) (12). Furthermore, detailed phenotypic analyses of YRT4 transgenic mice showed variegated expression in skin, iris, choroid and retina, all tissues that contain tyrosinase-expressing cells (59). In retinas of YRT4 transgenic mice, a gradual mosaic of RPE cells was observed, with central areas largely devoid of pigment, whereas in more peripheral regions pigmentation levels increased to normal levels. This pattern was reminiscent of initial stages of embryogenesis, where pigmentation first accumulates at the periphery of the RPE and later progresses towards the centre of the retina (60). Further, YRT4 transgenic mice, lacking the pivotal LCR element of the mouse tyrosinase expression domain, abnormally preserved this early retinal pigmentation pattern throughout embryo development, to the adult stage. Finally, there was a temporal delay observed in the start of pigment accumulation in retinas of YRT4 transgenic mice. All these data supported the idea that LCR was required to properly trigger and maintain the tyrosinase expression domain in a transcriptional permissive status throughout retina development (59).
The converse experiment, namely the addition of the LCR fragment to a standard tyrosinase minigene, was carried out independently by two laboratories (31, 39). Ganss and colleagues extended the initial ptrTyr4 transgenic construct (35), containing up to 5.5 kb 5′ upstream and tyrosinase promoter sequences linked to a tyrosinase minigene, with the addition of the LCR entire fragment (3.7-kb long). The pigmentation level achieved in this new transgenic experiment (phsTyr4 transgene) improved the expression achieved with previous standard minigene tyrosinase constructs. Moreover, there was a direct correlation between transgene copy-number and tyrosinase expression (and thus, pigmentation) (31, 61).
Similar experiments were addressed by Porter and Meyer in 1994. In this case the authors literally rescued an almost non-functional standard tyrosinase construct (carrying about 2.5 kb 5′ upstream and promoter sequences from the mouse tyrosinase gene fused to a human tyrosinase cDNA; PT transgene) by the addition of the tyrosinase LCR. These important regulatory sequences strongly promoted expression of tyrosinase transgenes in mice (UPT transgene) and resulted in most lines displaying high pigmentation levels (39). However, these authors noticed that transgenic animals carrying the LCR fragment, with moderate number of transgene copies (<8) had no detectable pigment in RPE cells, whereas surrounding tissues such choroid, colonized by melanocytes, strongly displayed high level of tyrosinase expression. This observation prompted them to suggest that LCR had little or no effect in optic-cup derived RPE cells, but behaved as a transcriptional enhancer in neural-crest derived melanocytes. However, these data were not confirmed by Ganss and colleagues. In their report, the presence of LCR determined increasing expression of tyrosinase both in RPE and surrounding melanocytes from the choroid (31). Furthermore, recent studies addressing the phenotype of YAC tyrosinase transgenes, lacking the LCR fragment, unequivocally demonstrated a fundamental role of LCR sequences in vivo both for initiating and supporting proper tyrosinase expression throughout retina development, specifically within the RPE cells (59).
It is possible that the diverse origin of all these transgenes could have slightly altered the expression pattern of tyrosinase. For instance, the basal standard tyrosinase minigene used by Ganss and colleagues resulted in moderate to high pigmentation levels in transgenic mice (as reported in Ref. 35). Further, no mosaicism was observed with (phsTyr4 transgene; 31) or without (ptrTyr4 transgenes; 35) the addition of LCR sequences. In contrast, the basal standard tyrosinase construct used by Porter and Meyer (PT transgene) contained somehow less 5′ regulatory sequences (2.5 vs. 5.5 kb) and an heterologous intron-less tyrosinase cDNA from another species (human), including its own 3′ untranslated region. Transgenic mice generated with this basal cDNA construct failed to express tyrosinase and rescue the albino phenotype of recipient animals (only two out of six lines showed ruby eyes and faint traces of pigment in their bodies), thus suggesting an already abnormal expression pattern in the initial construct that could be partially compensated in most lines (two out of eight lines showed variegation) by the addition of the entire LCR fragment. In support of a suboptimal tyrosinase expression (i.e. mosaicism) associated with the use of cDNA-only tyrosinase transgenic constructs, it must be noted that all other independent laboratories that used different mouse cDNA-only based tyrosinase constructs obtained mosaic expression patterns in a variable proportion of reported transgenic animals (23, 36, 38, 45, 47, 48). But, no variegation has been described in transgenic mice generated with standard tyrosinase minigenes that include the very long first endogenous intron (31, 35, 46). It is also plausible that the addition of intronic sequences, in this case, could have served to simply stabilize nascent transgene mRNA transcripts, without directly contributing to the regulation of tyrosinase gene expression, as it had been proposed before as a general observation, valid for most transgenic experiments for homologous (62, 63) and heterologous intronic sequences (64, 65). The use of artificial chromosome type transgenes ensures the inclusion of all genomic sequences that could contribute to efficiently transcribe and process a given expression domain (5, 16).
The in vivo role of the mouse tyrosinase LCR has also been addressed in transgenic animals expressing different amounts of mouse tyrosinase 5′ upstream and promoter sequences driving the expression of a prokaryotic lacZ gene (66). Previous analyses made with standard tyrosinase-lacZ reporter transgenes, carrying 270, 2200 or 6100 bp mouse tyrosinase 5′ upstream and promoter sequences had served to report, in a limited number of transgenic lines, histochemical detection of β-galactosidase expression in developing and adult mouse brain (67–70). These unexpected observations have been used to propose brain-specific expression of tyrosinase gene (reviewed in Ref. 70), an ectopic location for endogenous tyrosinase gene (33). However, the addition of mouse tyrosinase LCR sequences to standard tyrosinase-lacZ reporter transgenes completely abolished β-galactosidase expression in developing mouse brain (66). The authors discussed their findings by suggesting a repressive activity within the LCR, which would be responsible for the inactivation of lacZ expression in the brain.
Alternatively, it might be argued that standard tyrosinase-lacZ reporter transgenes could be prone to position effects, because of the presence of bacterial lacZ coding sequences, increasingly being recognized as a potential source of abnormal expression patterns in transgenic experiments, which include silencing, variegation and/or ectopic expression (71–76). Therefore, the addition of mouse tyrosinase LCR sequences (characterized by their properties to overcome or compensate for position effects) (12, 31) could contribute to re-establish the normal expression pattern, suppressing ectopic expression sites, such as, possibly, brain. Detailed analysis of the expression pattern of endogenous tyrosinase gene using direct approaches is still required to help clarifying this controversy.
Contrary to the variability observed in all standard tyrosinase transgene approaches (17, 31, 35, 39, 46–48, 66), all YAC-derived tyrosinase transgenes displayed reproducible and consistent phenotypes (11, 12). All lines carrying entire YAC-tyrosinase expression domains, including the LCR (YRT2 and YRT3, see Fig. 2) had a wild-type pigmented phenotype whereas all YAC-tyrosinase constructs lacking the LCR (YRT4 and YRT5, see Fig. 2) displayed suboptimal tyrosinase expression resulting in variegated and weak, but otherwise comparable, pigmentation levels (12). These observations strongly suggest that gene expression studies should be addressed, whenever possible, with transgenic approaches most similarly reflecting their situation in the endogenous alleles. Such conditions are normally found with artificial chromosome-type transgenes (16).
ROLE OF THE LCR-CORE (A AND B BOXES) IN ESTABLISHING LCR FUNCTION
Two binding boxes for nuclear factors within the LCR core, originally known as HS-1 and HS-2 and now referred as boxes A and B, were identified by in vitro analysis corresponding to the previously described hypersensitive DNase I site (31). These short sequences, showing some degree of homology to CREB and AP-1 binding sites, retained most of the LCR enhancer function and were subjected to direct mutational analysis in vivo. Using homologous recombination techniques that operate in yeast cells we have prepared new YAC-tyrosinase transgenes carrying specific mutations in A, B or both A and B boxes (77). These mutations (clean deletions of 20–60 base pairs) have been generated using the pop-in/pop-out method, which does not leave any sign of selectable marker or unwanted heterologous sequences at the end of the process. These new YACs have been termed YRT2ΔA, YRT2ΔB and YRT2ΔAB (see Fig. 2) (77). Transgenic mice have been generated with all of them (Giraldo and Montoliu, unpublished). Detailed experimental data will be published elsewhere. However, for the discussion of this review it can be anticipated that all transgenic mice carrying YRT2ΔA (five lines), YRT2ΔB (two lines) or YRT2ΔAB (five lines) mutant constructs analysed so far displayed variable pigmentation levels, below wild-type agouti animals but clearly above pigmentation levels achieved with YRT4 and YRT5 transgenic animals, in which not only A and B boxes but the entire LCR fragment was eliminated (YRT4) or substituted and interrupted by a yeast selectable marker (YRT5) (12). These results suggest that A and B boxes, although proven to be indispensable for in vitro enhancer function on tyrosinase promoter (31), co-operate in vivo with surrounding sequences, within the LCR. Therefore, specific deletion of A and B boxes in artificial-chromosome type tyrosinase transgenes results in mild position effects (which include some mosaics and, mostly, pigmentation weaker than wild-type animals), less pronounced than those observed after deletion of the entire LCR fragment (12).
The limited and nonexclusive role of A and B boxes in establishing faithful LCR function was already suggested by previous experiments with standard transgenic animals. Porter and Meyer extended their exhaustive transgenic analysis to new constructs containing different DNA subfragments located within the LCR (39, 40). In good agreement with our findings with YRT2ΔA, YRT2ΔB and YRT2ΔAB transgenic animals, the addition of a smaller 1 kb DNA subfragment covering A and B boxes (3PT transgene) could only rescue the suboptimal cDNA-based tyrosinase construct in two out of four lines generated, though only one line displayed high pigmentation level (39). Shorter DNA subfragments located 3′ from the A and B boxes (2PT transgene) could only trigger tyrosinase expression, albeit at very low levels, in one out of four lines (39).
More interestingly, the same team analysed an additional DNA subfragment (5PT transgene, extending upstream from the A and B box, up to the 5′ end of the entire LCR fragment) that permitted the generation of pigmented transgenic mice (40). Nine independent lines were established out of which six were pigmented at the variable level, most of them being mosaic (40). Again, in these analyses the use of cDNA-based tyrosinase transgenes strongly compromises the correct interpretation of results. Three lines carrying single copy 5PT transgenes were unable to show pigmentation. In addition, the presence of multiple transgene copies (up to 37 copies) could not overcome strong position effects (most likely enhanced by the use of a cDNA-only approach) manifested by somatic mosaicism (40). However, besides suboptimal transgene performance, these results are also in agreement with our analysis of YAC-tyrosinase transgenic mice carrying mutations in the A and B boxes and altogether firmly support the existence of additional co-operative elements (other than A and B sequences, most likely located 5′ upstream of them) within the LCR region that would contribute to the correct expression of tyrosinase locus.
These putative co-operative elements have been proposed to be of S/MAR type, as previous in vitro analysis indicated specific S/MAR-binding activity in the 5′ part of the LCR region (54). However, other activities and other types of regulatory elements, such as boundaries, should be considered and could be potentially present within the increasingly complex mouse tyrosinase LCR, as it has been progressively demonstrated in other LCRs over the past years (4, 6–8, 78, 79). These additional but complementary regulatory activities, presumably present within the mouse tyrosinase LCR, will require further characterization both at the structural and functional level before cellular specificity of these regulatory elements could be established. In addition, these observations obtained from the analysis of different artificial constructs and fusions, generally based on suboptimal standard type of transgenes, should be eventually tested at the endogenous allele location (using homologous recombination strategies in ES cells and, for instance, CRE/loxP approaches) or, most comparable and affordable, by using artificial-chromosome type transgenes (and their mutant and homologous recombinant derivatives obtained in yeast or bacterial cells) that accurately recapitulate and reproduce the behaviour of endogenous alleles in ectopic genomic locations.
The authors wish to thank the financial support from Spanish Ministry of Science and Technology through Grants BIO97-0628 and BIO2000-1653. G. Jeffery, M.A. Chinchetru and A. Lavado are greatly acknowledged for critical revision of this manuscript.