We have shown that a complex mutation was associated with the lavender phenotype in Japanese quail, and genotyping this mutation in an informative cross confirmed that the mutation co-segregated with the phenotype. The region affected by the complex mutation is highly conserved between birds and mammals, and MLPH, PRLH, RAB17 and LRRFIP1 are found in the same order and relative orientation in Zebra Finch (chromosome 7), Chicken (chromosome 7), Mouse (chromosome 1) and Human (chromosome 2). The same region can be found in the genome of the turkey, but it is distributed on several contigs because an assembly is not yet available. This conserved synteny indicates that the genes involved in the present mutation and those already studied in mammals or chicken are orthologous, and that this region is not generally affected by genomic instability.
The characterization of the detailed genomic structure of the mutation leading to the lavender phenotype in quail showed that an homologous phenotype between quail and chicken, demonstrated by the production of lavender intergeneric hybrids
, was apparently the consequence of very different mutational events: a single base-pair substitution in chicken
 versus a complex and large mutation with three successive chromosomal rearrangements (two inversions and one deletion), affecting four consecutive genes in quail. The reconstruction of the most likely sequence of chromosome rearrangements leading to the lavender allele raises the question of the molecular mechanism involved. Intermediate steps (2 and 3, Figure
3) between wild-type and lavender alleles were not observed in the population (PCR screening of the breakpoints, data not shown). Then, it was not possible to determine whether the rearrangements occurred as three unrelated events, with intermediate genotypes now lost or not represented in the population, or in only one step. If the three mutations occurred independently, something may have increased the probability of recurrent chromosome rearrangements in the same region. Otherwise, a complex and dramatic change in genome organization took place during a single meiosis, leading to the lavender allele. Screening the genomic region with chicken or quail sequence data (acquired during the study) did not reveal any particular structure, such as repeats or transposable elements that might have facilitated chromosomal rearrangements.
Very few mutations similar to the one observed here, i.e. overlapping chromosomal rearrangements, have been reported. In Human, complex chromosome rearrangements have been observed in cancer cells, and some are even characteristic of specific cancers
, and other genetic diseases. For example, the Wiskott-Aldrich syndrome results from a complex mutation due to two deletions and one inversion in a 4.5-kb region
. In that example, the three chromosomal events were contiguous, but it has not been confirmed that they were overlapping, and the family study revealed that the complex mutation appeared in a single step. Similarly, another complex human mutation
 resulted from two deletions and one inversion in a 2-Mb region, and all were contiguous. The pattern of the mutation suggested two overlapping chromosomal changes: an inversion followed by a deletion, and the family study indicated that this complex mutation occurred as a single event. These observations of similar de novo complex chromosomal changes suggest that the chromosomal rearrangements leading to the lavender quail may have also taken place in a single step.
Mechanisms proposed to explain complex mutational events combining inversion and deletion are based on strand misalignment of sequences with complementary features during DNA replication or meiosis. We have carefully checked the breakpoint regions for a common DNA pattern. These regions were not located on repeated or transposable elements, but some sequences flanking breakpoints shared a common G(C)3-5A pattern (Additional file
1: Table S2). Interestingly, a 2.2-kb DNA sequence (between BP4 and BP5, Figure
3) appears to be involved in all the chromosomal changes. The first inversion changed its sequence orientation, the second one restored it, and the deletion removed this DNA portion together with the region containing RAB17, PRLH and MLPH, leaving only a short 246-bp region (between BP6 and BP5, Figure
3) translocated 30 kb downstream by the two inversions. The sequence of this DNA segment did not show any particular structure, however.
The complex mutation found in the Japanese quail is not lethal, although it has a major impact on the DNA sequence of MLPH, but also on PRLH, which is completely missing, RAB17, which is mostly deleted, and to a lesser extent on LRRFIP1. Consequently, only expression of a truncated gene is possible for MLPH, and RAB17 and PRLH are not expressed. This is confirmed by the RT-PCR results, with no expression of RAB17 in lavender, and no transcript for MLPH beyond exon 9. PRLH (Prolactin Releasing Hormone) is one of the hypothalamic peptide hormones that regulate the production of pituitary hormones, and it was first identified as a prolactin-releasing factor in mammals. PRLH also acts as a neuromodulator of pituitary products, and is involved in the control of metabolism, energy homoeostasis and food intake
. The absence of the whole DNA region for PRLH in lavender quail was not associated with any difference in plasma prolactin concentration. Indeed, secretion of prolactin in quail is probably under the control of the vasoactive intestinal peptide (VIP), as in turkey
 and zebra finch
, which would explain the unchanged prolactin levels of lavender quail. RAB17 is a member of the RAB family (Ras proteins involved in membrane trafficking, part of the RAS oncogene superfamily). It encodes the RAS-associated protein 17 found in epithelial cells
 and may be involved in membrane trafficking. It is almost completely deleted in lavender quail, but it is possible that its loss is compensated by another gene of the same family through functional redundancy, since the RAB gene family has many members (for example, 67 protein-coding genes in humans, and 53 protein-coding genes in chicken are annotated in Homologene NCBI database:
http://www.ncbi.nlm.nih.gov/homologene). RAB17 particularly, is phylogenetically very close to RAB5 genes (RAB5A, RAB5B and RAB5C)
. RAB17 expression
 is regulated by MITF in mammalian cell lines (melanomas) and melanocytes with RAB17 knockdown have reduced filopodia formation, leading to impaired melanosome transfer. Then, RAB17 might make some contribution to the quail plumage colour but it is probably not responsible for lavender in quail because hybrids between lavender chicken and quail would have the RAB17 function rescued by the normal chicken RAB17, and should have shown a wild-type, undiluted, plumage colour, which they did not
. LRRFIP1 (Leucine-Rich Repeat (in Flightless I) Interacting Protein-1) encodes a protein involved in the regulation of the TLR signalling pathway
 and the production of type-I interferon in response to pathogens
. This immune function has not been investigated in the present study, but it is likely that this gene was only affected marginally by the complex mutation because it lies mostly outside the deleted and inverted region of the lavender quail DNA sequence. Indeed, only its putative first exon was involved, whereas gene annotation in human and mouse indicates that only one alternatively spliced variant out of five described is starting from this first exon, with relatively few ESTs including this first exon.
Past results in chicken
, and dogs
 point to the large deletion in MLPH as the most likely cause of the “lavender” plumage colour in the Japanese quail. In the present study, the lavender genotype was also associated with slightly lower growth, as for albino, roux and silver mutations in quail. The percentage of the phenotypic variation of the body weight traits which was accounted for by the linear model (R2) did not increase by more than 3 points (Tables
3) when plumage colour was included (Full model), despite statistical significance. Yet, as similar associations between lower body weight and diluted plumage colour due to various mutations were also reported in the chicken (eg:
), these convergent observations in quail and chicken across several independent genes for plumage colour suggest that the association may be due to a pleiotropic effect related to melanin expression. Indeed, it was found recently that the feathers of roux, lavender and albino quail, which have lower body weight than wild-type birds, each had lower total melanin content than wild-type and yellow quail
 with similar adult body weight
. Further study will be needed to find the possible causes of this covariation because melanins have several suggested functions (eg:
[38, 39]). Most other associated phenotypic effects observed in lavender quail are probably not attributable to MLPH alone, since melanophilin defects in MLPH-mutated humans or mice have not been found to be associated with any noticeable phenotypic consequence other than the colour dilution
The very significant differences in higher residual feed intake and lower body temperature in lavender quail during the feed trial (Table
2) were associated with a fair portion of the variation as R2 increased by 27 and 13 points, respectively, when the effect of plumage colour was included in the linear model. This phenotypic pattern might be related to central homeothermy-related deficiency induced by the effect of the complex mutation on PRLH as was reported in PRLH-deficient rodents
. Of course, other genes linked to lavender might also be involved in some of the physiological changes observed in this study.