In this study we pursued the observation that high expression of an endogenous retrovirus of the ALVE type was associated with low growth in one of two chicken lines established by long term divergent selection for high or low body weight [5, 7]. We conclude that the high levels in the LWS line show Mendelian inheritance. LWS birds have more ALVE integrations than HWS birds, which in turn have a larger number of integrations compared with WL and RJF chickens. Using F9 birds from an advanced intercross between the two selected lines we tested if there was a correlation between body weight, ALVE integrations and expression levels. The results indicated that a minority of the integrations contributed to the higher levels and that high expression was significantly correlated to lower body weights of females but not males. The conserved correlation between high ALVE expression and low body weight in females after 9 generations of intercrosses indicates that ALVE loci conferring high expression are genetically linked to or constitute loci directly contributing to low body weight of LWS chickens in a sex-limited fashion.
The chicken genome contains four families of ERV elements classified as chicken repeat 1 (CR1) elements, ALVE s, avian retrotransposones from the chicken genome (ART-CHs) and endogenous avian retrovirus elements (EAV-0) . Although the microarray contained probes with different retroviral sequences, only ALVE-related sequences were identified as differentially expressed. The env gene in the ALVE proviral genome is a source for genetic diversity through recombination with exogenous viruses [33, 34]. The sequence diversity of this gene constitutes the basis for defining the six subgroups of ALV (A, B, C, D, E, J) and is related to variation in infection susceptibility, receptor interference as well as antibody neutralization . The env gene was used as target for the primer design for qPCR, qRT-PCR and for sequencing. The primers we used amplified the endogenous ev -loci of several ALVE subtypes, but did not match other types of retrovirus such as RSV or avian myeloblastosis virus. Primers against the ALVE pol gene confirmed the differential expression seen with the env primers (Fig. 3C).
Endogenous retrovirus elements are in most cases transmitted genetically . Transmission of ALV can occur via several natural routes . Exogenous ALVs are transmitted horizontally by infection between individuals or vertically from hen to progeny in ovo by congenital transmission [11, 36]. Horizontal transmission is relatively inefficient while congenital transmission is very efficient and leads to a high ratio of infected embryos . The ALVE elements exist in the chicken genome as partial or complete ALVE proviral genomes. Endogenous elements have in general a limited or restricted ability to transmit virus congenitally, in contrast to exogenous ALV that undergo highly efficient congenital transmission [37, 38]. However, it was demonstrated that some ev -loci that encode complete provirus genomes, particularly ev -12 and ev -21, can be transmitted at higher frequencies from subgroup E susceptible dams to susceptible progeny [25–27].
Susceptibility of chickens to ALV retroviral infection is regulated by subtype-specific cell membrane receptors that interact with the Env glycoprotein. Exogenous ALV subtypes B and D, and virus particles of endogenous ALVE infect through this interaction. Different types of receptors for ALV subtypes B, D and E are encoded by three alleles of the TVB locus. The TVB *S1 allele encodes tumour necrosis factor receptors that are required for the viral entry of all three subgroups while TVB *S3 permits viral entry of subgroup B and D but not E. The TVB *R allele produces truncated receptors that do not support entry of any ALV [28, 31, 32]. All of the successfully tested 19 individuals (G41) possessed the TVB *S1 allele that gives susceptibility for ALVE. This result is in agreement with that 83% of chickens from 36 broiler lines were homozygous for TVB*S1 . Hence, both HWS and LWS chickens are susceptible for ALVE infection and polymorphism in the TVB locus is neither a result of the long term selection nor is it likely to be involved in the high ALVE expressing phenotype.
The possibility that the LWS chickens propagated high ALVE expression via congenital infection from hen to egg was examined. We analyzed ALVE expression in an F1 generation after a reciprocal cross between the lines (G46). F1 siblings from the same LWS dam often had both high and low ALVE levels and LWS males transmitted high expression to their progeny (Fig. 3). Moreover, hens with high ALVE expression did not always transmit high expression to their progeny as would have been expected by congenital infection. Rather, their expression spanned the full range of expression levels seen in the parentals. Therefore, the high/low ALVE expression levels were likely to have been inherited and these data support a Mendelian mode of genetic transmission of ALVE expression. Furthermore, an exogenous ALV infection among parental LWS is less plausible because ALV-related disease symptoms have not been observed during the course of selection . It cannot be excluded that such infection has occurred and by recombination may have formed elements that triggered increased ALVE expression because there are examples of male-mediated congenital transmission of ALVE . The active transcription of ALVE in the tested tissues may also have introduced recombinant somatic ALVE pro-viral integrations .
The number of env gene integrations in RJF and WLs ranged from 2 to 7 per haploid genome. Both the HWS and LWS lines had more integrations than RJF and WLs. HWS individuals had significantly fewer integrations than LWS while the F9 birds had 8 to 22 env integrations per haploid genome, a number similar to that for the LWS line (Fig. 4B). The reported average for layer chickens is 1 to 3 elements, while that for meat-type chickens is 6 to 10 . Altogether 22 different ALVE loci have been identified in WLs and current estimates suggest that there may be over 50 different loci . Although the number of ALVE integrations in the genome pool of the White Plymouth Rock founder population for the selection experiment is not known, they probably had a similar number of ev -loci as the HWS and LWS lines (7 to 22 integrations). This number is little higher than the average meat bird, however, the qPCR in this study may be more sensitive than previously used methods.
HWS birds have low ALVE expression and fewer ALVE integrations than LWS birds suggesting that differential selection for growth has influenced both ALVE expression and integration number (Figs. 2 and 5). This hypothesis was supported by results from the F9 population where we observed a weak but significant correlation between integration number and expression (Fig. 5A). The results suggested that only a few of the integrations contributed to the high levels of expression. This assumption was further supported by the occurrence of sequence polymorphism for the env gene (Fig. 1C), and one sequence variant was exclusively found in LWS birds. Only this LWS-variant was found in cDNA from LWS birds and F9 individuals with high ALVE expression (Figs. 5C, D and 1C). In contrast, in genomic DNA from LWS chicken both the LWS- and HWS-variants were present and the HWS variant was more frequent. Thus, while LWS-variant integrations are fewer than the HWS-variant they contributed more to the high levels of expression in LWS individuals and certain F9 birds. An obvious interpretation is that selection for high body weight has been effective to purge or silence high expressing ALVE loci. Another possible explanation is that a previous ongoing infection would have produced novel integrations that led to the increased levels in the LWS line. For this to occur would require novel integration in the germ line in order to transmit to the next generations.
Our data from the F9 generation suggest that the actively expressed ALVE loci are causing reduced growth and that this effect is more pronounced in the females than in males. This pattern may be explained by a sex-specific response or because the effects by high ALVE expression are more penetrant for smaller birds and pullets are over-all smaller than males. ALVE integration is of interest for the poultry industry because the frequency of integration alters the responses to selection for economical traits [9, 10, 12–15]. The mechanism may be that integrations directly or indirectly disrupt other genes [8, 42]. However, in humans there are only rare examples where a recessive monogenic disorder is caused by HERV integration disrupting gene function. Alternatively, a high virus expression load such as in the LWS line may affect the growth indirectly. The activation of inflammatory cytokines such as the interferon-gamma, TNF-alpha, interleukin-1 and -6, their receptors and signalling pathway components are signatures of retrovirus infection [43, 44]. Such genes were not over-represented in the cDNA array analysis results . Factors that regulate retrovirus trans-cellular transport and budding are also regulated at high virus loads such as actin-related modulators including Rho-like factors and trans-golgi factors . Similar activation patterns have been seen after avian RSV infection of chick fibroblasts . The budding of enveloped RNA viruses, including HIV and other retroviruses, usurp a cellular pathway that is normally used to form vesicles and transport them into multi-vesicular bodies . Some of the differentially expressed genes observed in our previous study  while associated with alterations in neuronal plasticity are also regulated during acute and chronic retrovirus infections [45, 47]. These include vesicle trafficking systems such as the ARL/ARF factors and FKBP5 as well as the Nephroblastoma overexpressed gene (Nov). Nov was reported to decrease in fibroblasts after Rous sarcoma virus transformation  and we observed lower Nov expression in LWS chicken than in HWS chickens. Nov was initially identified as a cellular gene in chick nephroblastomas induced by the retrovirus myeloblastosis-associated virus . The identification of Nov as being differentially expressed between lines indicates that the expressed endogenous ALVE sequences may influence cellular gene expression and may therefore contribute to the selection response for growth.
Both HWS and LWS pullets showed delayed age of onset of egg production, and a considerable proportion of LWS females never mature [49, 50]. Delayed sexual maturity for LWS females were attributed to anorexia because it was possible to induce egg laying by force-feeding. Moreover, sexually matured LWS females were heavier at 56 days of age than those that did not show sexual maturation later in life . Other studies have also indicated a relationship between viral integration and traits related to reproduction. Gavora et al  reported that certain virus-producing ev -loci, ev-10 or 19 and 12, and silent gene ev -1 can affect egg productivity for layers. Also, the total number of ev -loci per genome was significantly related to body weight at first egg and mature body weight . The body weight of LWS juvenile females is related to that of sexually matured LWS females and sexual maturation might be related to the number of ALVE integration and the ALVE expression. Therefore, it is not surprising that high expression of ALVE is correlated to the low juvenile body weight in female chickens.
Quantitative trait locus (QTL) analysis has been performed after crossing the HWS and LWS lines and more than 13 growth-related QTLs were identified all with minor individual effects [16, 53] and a high degree of epistasis . Although the exact location of ALVE integrations remain to be defined, our results are consistent with the QTL data in that we present data that multiple proviral loci together contribute to one aspect of the phenotype, namely to the low weight of pullets.