This research was based on the premise that, as broilers and layers have been genetically selected for different traits, the fact that adult broilers exhibit contrasting ovarian phenotypes under non-restricted conditions from layers is a result of genetic selection and there is ample evidence that this is the case
. Consequently, the effect of this genetic selection on expression of genes suspected of being important for reproductive development was undertaken, as regulation of their expression is likely to be the cause, directly or indirectly, of the phenotypic differences observed in broilers and layers. These are obvious targets for research to explain these differences but are also of fundamental interest in reproductive biology in a wider context. The study therefore used results from a range of experimental and analytical approaches to identify candidate genes for regulation of follicle selection and recruitment in the broiler breeder ovary. Two different approaches were used in complementary fashion to analyse the large dataset produced by the microarray. BioLayout Express, the more novel component of our analytical approach, has many built-in features that complement its use as a basic analytical tool. The most fundamental is the ability to visualise the data as a 3D network of datapoints that can be subsequently analysed by the internal clustering algorithm. The results can also be filtered by any available annotation or expression profile across samples to facilitate identification of possible associations between them. This makes it much easier to ask different questions of a single dataset and was an essential asset in switching focus between developmental stages and effect of nutrition. Using these different approaches to analyse the microarray data, as well as using information from different patterns of follicular development a number of genes that show changes in expression that would be consistent with a role in follicle selection have been identified. Three of these candidates have also been shown to exhibit altered expression levels as a result of ad libitum feeding.
From the 37 prioritised candidates considered for multi-level QPCR profiling, 10 candidates (including FSHR and TGFBR1 for validation of the approach) had sufficient experimental and/or literature-sourced evidence for basic hypothesis generation as to their role in follicle selection and recruitment. All of these candidates are associated with regulation of apoptosis
[43–48], cell growth
[20, 49, 50] and survival
 or steroidogenesis
 where there is any documented function. While these processes are all prerequisite to follicle survival, there is insufficient information at present to create a single model system incorporating all of our candidates, although FSHR, TGFBR1 and SMAD3, with the added possibility of MOSPD1, do interact with common signalling pathways. However, PDGFRL is by far the most intriguing novel candidate.
PDGFRL produces a product which is homologous with the functional domains of Platelet-derived Growth Factor Receptors that are involved in intrafollicular cell signalling associated with steroidogenesis in mice
. QPCR profiling in layers clearly shows significant (P <0.001) and substantial peaks in expression at the 5–6 mm and F2 stages, i.e. immediately prior to selection and ovulation respectively. This evidence would support a function in regulatory feedback mechanisms. The BioLayout Express profile from the broiler breeder microarray data suggests that the gene is downregulated in 5–7 mm follicles relative to the stroma and F1. This is consistent with the broiler breeder QPCR expression pattern for feed restricted birds. In contrast, the expression levels across tissues in ad libitum fed birds remain relatively constant. In ad libitum fed birds, where hierarchical follicle number is increased, the upregulation of PDGFRL expression, relative to feed restricted birds, suggests that it is likely to be in activation or upregulation of positive feedback signalling to the HPG axis. The observed downregulation in 5–7 mm follicles from broiler breeders is in marked contrast to the layers, which would support the hypothesis of potential dysregulation of part of the steroid-based feedback mechanisms, given what is already known of the PDGFR family in other species.
The QPCR results from experiment 2 show that FSHR and TGFBR1, genes known to be involved in follicle growth, peak in their expression during early prehierarchical development. This agrees with previous results
[20, 49]. Interestingly, both FSHR and TGFBR1 show a prominent peak at 8–10 mm, indicating that follicles immediately post-recruitment may have a heightened sensitivity to the ligands of these receptors at this stage. As both receptors activate pathways leading to cell growth, proliferation and differentiation, higher expression at those stages is not surprising. SMAD3, a known downstream signal mediator of TGFB family receptors demonstrates a very similar pattern of expression to TGFBR1. However, TGFBR1 and SMAD 3 did not show a significant effect in response to ad libitum feeding, whereas FSHR was significantly downregulated in the F1 follicle under ad libitum feeding. It is most likely that, as the lower FSHR expression shown in ad libitum fed broilers is more comparable with previous studies
, the increased expression in feed restricted birds leads to negative feedback resulting from steroidogenic factors. Little functional significance has been placed on FSH control in the F1 and further investigation is warranted to explore the potential roles for FSHR in this follicle in light of the results reported here. MOSPD1 has been implicated in mesenchymal cell differentiation
 and is upregulated in ovarian cancer
. BioLayout Express analysis showed it clustered with FSHR in broiler breeders and QPCR profiling in layers corroborated this. MOSPD1 is a membrane-associated protein
 and may be involved in supporting or mediating signal transduction from the FSH receptor but further work will be required to determine this.
GDF9 was not identified through the original microarray analysis but was included due to its location near the putative QTL for follicle number in the chicken and because studies in sheep report an association between mutations in GDF9 and increased ovarian follicle number and ovulation rate
[56, 57]. Despite showing little change in expression between tissues in the initial screen in layers, GDF9 does show significant downregulation in response to ad libitum feeding in broiler breeders in the stroma and in 6–8 mm follicles (P = 0.005). In conjunction with results from mamalian studies this result would imply an inhibitory effect on follicle number. It is interesting to note that the expression profile for GDF9 in broiler breeders, regardless of diet, is comparable with other species
[58–61], indicating a high level of inter-species conservation for this gene. Reported inter-species sequence conservation from the UCSC Chicken Genome Browser supports this, with sheep being most comparable in terms of exon coverage. Further investigation is underway to determine if there are mutations in the chicken, as there are in sheep, that might be associated with multiple ovulation.
SLIT3 and VDAC1 have both been shown to be involved in pro-apoptotic signalling
[43–46] and are located in the putative QTL for follicle number on chromosome 13. SLIT3 is also involved in ovary and follicle development in sheep
 and its expression profile in layers is consistent with phases of increased apoptosis. Expression profiling of SLIT3 and VDAC1 in BioLayout Express for broiler breeders is consistent with the layer QPCR profiling. However, there was no significant differential expression either between tissues or dietary regimes in the broiler breeder QPCR validation for either candidate.
YAP1 is believed to be involved in cell survival signalling through regulation of the p53 signalling pathway
. The BioLayout Express profile of the broiler breeders suggests that it is upregulated in late prehierarchical follicles. However, profiling in layers shows higher expression in early prehierarchical follicles and the F5 follicle, where pro-survival signals would be expected to occur more prominently. Upregulation in ad libitum fed birds could not be validated by QPCR. This does not negate a role for YAP1 in follicular development, however it is unlikely to be responsible for multiple ovulation in broiler breeders.
GULP1 is expressed in macrophages and is involved in engulfment of apoptotic cells
[47, 48]. Profiling in layers is consistent with this activity. While this is not likely to be a candidate for follicle selection, it does highlight the transitional stages of the follicle as it progresses through development.
RIGG1740, KRT75 and SPTY2D1 were also investigated for a potential role in follicle development, however their expression profiles in layers, in conjunction with the level of available evidence (summarised in Table
5) did not indicate a central role in follicle recruitment.
Conclusions and further work
Our prime candidates, PDGFRL, GDF9 and FSHR, although only indirectly linked, all have strong cases for further investigation. GDF9 and FSHR are not novel candidates, indeed, FSHR was intially included in this study as a form of positive control and the identification of a previously unreported dietary effect on its expression was unexpected. PDGFRL however, is a novel candidate, and its implicated role in regulation of steriodogenesis, along with its response to ad libitum feeding makes it of primary importance.
Further investigation of PDGFRL, as well as FSHR and GDF9, is clearly warranted, and would yield many valuable insights. For example, work to determine localisation of expression of PDGFRL within the different cell types in the follicle wall is clearly needed to further substantiate its proposed role in steroid-based feedback. Successful examination of these candidates should bring us one step closer to solving the problem of multiple ovulation and allow for the prospect of relaxing feed restriction to improve the welfare of broiler breeder chickens.