Abstract

   Heat stress in poultry houses, especially in warm areas, is one of the
   main environmental factors that restrict the growth of broilers or
   laying performance of layers, suppresses the immune system, and
   deteriorates egg quality and feed conversion ratio. The molecular
   mechanisms underlying the response of chicken to acute heat stress
   (AHS) have not been comprehensively elucidated. Therefore, the main
   object of the current work was to investigate the liver gene expression
   profile of chickens under AHS in comparison with their corresponding
   control groups, using four RNA-seq datasets. The meta-analysis, GO and
   KEGG pathway enrichment, WGCNA, machine-learning, and eGWAS analyses
   were performed. The results revealed 77 meta-genes that were mainly
   related to protein biosynthesis, protein folding, and protein transport
   between cellular organelles. In other words, under AHS, the expression
   of genes involving in the structure of rough reticulum membrane and in
   the process of protein folding was adversely influenced. In addition,
   genes related to biological processes such as “response to unfolded
   proteins,” “response to reticulum stress” and “ERAD pathway” were
   differentially regulated. We introduce here a couple of genes such as
   HSPA5, SSR1, SDF2L1, and SEC23B, as the most significantly
   differentiated under AHS, which could be used as bio-signatures of AHS.
   Besides the mentioned genes, the main findings of the current work may
   shed light to the identification of the effects of AHS on gene
   expression profiling of domestic chicken as well as the adaptive
   response of chicken to environmental stresses.

   Keywords: heat stress, WGCNA, machine learning, eGWAS, chicken

Introduction

   Heat stress is one of the main concerns of the poultry industry,
   especially in warm areas, as it causes major economic losses in both
   layer and broiler farms. Intensive genetic selection in breeding
   programs has led to an increased growth rate and metabolism. However,
   the development of the chicken thermoregulatory system cannot be
   matched with the growth rate, making it difficult for industrial
   chickens to regulate their body heat as temperatures fluctuate
   ([28]Havenstein et al., 2003). In addition, chickens are sensitive to
   high temperatures due to the absence of sweat glands ([29]Loyau et al.,
   2013), feather cover, and high density in commercial rearing facilities
   ([30]Brugaletta et al., 2022). The most common negative effects of heat
   stress (HS) on chickens are on growth performance, egg production and
   quality ([31]Barrett et al., 2019; [32]Awad et al., 2020), feed intake,
   appetite hormone regulation ([33]Mazzoni et al., 2022), oxidative
   properties ([34]Altan et al., 2003), intestinal health, immune response
   ([35]Deng et al., 2012), body temperature ([36]Van Goor et al., 2015),
   and increased mortality ([37]Khosravinia, 2016). It is estimated that
   the HS could lead to annual economic losses of $128 to $165 million in
   the United States poultry industry ([38]St-Pierre et al., 2003).
   Selective breeding of HS-resistant chicken is a suitable scenario for
   producing a well adaptable strains ([39]Radwan, 2020). Therefore, the
   characterization of genetic biomarkers associated with resistance to HS
   will pave the way for selective breeding to generate the HS-resistant
   chickens. Transcriptome comparison may elucidate the genetic base of HS
   ([40]Sun et al., 2015). The liver has an important role in general
   metabolism, synthesis of bile and proteins, and maintenance of
   homeostasis ([41]Rui, 2014), and is more vulnerable to HS than other
   organs as HS triggers oxidative stress ([42]Lin et al., 2006). The
   liver is also impacted by the increased production of biochemical
   anti-oxidants ([43]Mahmoud and Edens, 2003). Previous studies have
   shown that some genes in liver tissue undergo significant expression
   modifications under HS as compared to the normal condition. For
   example, HSP70, HSP90 ([44]Radwan, 2020), HSP90B1, HSPA5 ([45]Sánchez
   et al., 2022), MX1, TLX1, HSPB9 ([46]Wang et al., 2020), HSP70, HSPA5
   ([47]Wang et al., 2021), ANGPTL4 ([48]Lan et al., 2016) have been
   introduced as biomarkers for HS in chickens. These genes have not been
   identified from a sole, comprehensive research but from multiple
   similar studies, each of which had identified only one or, at most, a
   couple of the mentioned genes. In other words, there is little
   consensus on the results of the studies that aim to address the same
   scientific question. Therefore, there should be a statistical
   methodology to merge the findings of multiple independent but similar
   studies. Meta-analysis is a quantitative and systematic method for
   combining the p-values obtained from the analysis of RNA-seq data from
   multiple related studies. Meta-analysis could overcome the issues that
   arise from the low number of biological replicates in the experiments,
   and could result to an improved statistical power due to the larger
   sample size that come from multiple datasets ([49]Tseng et al., 2012;
   [50]Rau et al., 2014). By meta-analyzing, the results of multiple small
   but related studies could be combined to attain a pooled estimate that
   is closest to the common truth. By relieving the sources of
   disagreement among the related results, meta-analysis of multiple
   studies makes interesting relationships come to light. The Fisher
   approach, which is usually implemented in the meta-analyses, has been
   proven to be an appropriate method for the combining the p-values, and
   is useful for the identification of differentially expressed genes
   (DEGs) and novel biomarkers ([51]Calduch-Giner et al., 2014; [52]Landry
   and Sirard, 2018; [53]Lindholm-Perry et al., 2020). Because of the
   complex nature of the biological systems in which many genes or
   biological agents interact with each other, there is an increased
   request for researches that aim to elucidate the complex interaction of
   genes that have been identified as biologically important. The study of
   gene co-expression networks helps to categorize genes with the same
   expression pattern. The interaction of genes could more easily be
   predicted by analyzing the modules than by analyzing the genes
   themselves ([54]Cho et al., 2012). Therefore, weighted gene
   co-expression network analysis (WGCNA) could be used as a desirable
   approach for discovering the co-expressed genes and nodes
   ([55]Ramayo-Caldas et al., 2018; [56]Wang et al., 2020; [57]Sánchez et
   al., 2022). In the present study, RNA-seq data from four different but
   related datasets were assessed to identify the DEGs, meta-genes,
   modules, and hub-genes in the chicken liver tissue under acute heat
   stress (AHS). Finally, we introduced the major genes associated with
   AHS.

Materials and methods

RNA-seq data collection from databases

   The Sequence Read Archive (SRA) repository of the National Center for
   Biotechnology Information ([58]https://www.ncbi.nlm.nih.gov/sra) was
   screened precisely to find the appropriate RNA-seq datasets that
   address the research question of the current work using the keywords
   “Gallus gallus,” “chicken,” “liver” and “acute heat stress”. We could
   find only four RNA-seq datasets in which the AHS was the main
   treatment. Detailed information of the four selected datasets can be
   found in [59]Table 1. In addition, the accession numbers of the used
   samples are reported in In [60]Supplementary Table S1 .

TABLE 1.

   Information of the used datasets for the analyses.
   Dataset accession number Group Number of runs Raw reads Alignment rate
   (%) References Breed Sex Age at sample collection (week) Duration of