Acute thermal stress elicits interactions between gene expression and alternative splicing in a fish of conservation concern

Transcriptomics provides a mechanistic understanding of an organism’s response to environmental challenges such as increasing temperatures, which can provide key insights into the threats posed by thermal challenges associated with urbanization and climate change. Differential gene expression and alternative splicing are two elements of the transcriptomic stress response that may work in tandem, but relatively few studies have investigated these interactions in fishes of conservation concern. We studied the imperilled redside dace (Clinostomus elongatus) as thermal stress is hypothesised to be an important cause of population declines. We tested the hypothesis that gene expression-splicing interactions contribute to the thermal stress response. Wild fish exposed to acute thermal stress were compared with both handling controls and fish sampled directly from a river. Liver tissue was sampled to study the transcriptomic stress response. Thermally stressed fish showed a prominent transcriptional response (estimated with mRNA transcript abundance) related to transcription regulation and responses to unfolded proteins, and prominent alternatively spliced genes related to gene expression regulation and metabolism. One splicing factor, prpf38b, was upregulated in the thermally stressed group compared to the other treatments. This splicing factor may have a role in the Jun/AP-1 cellular stress response, a pathway with wide-ranging and context-dependent effects. Given large gene interaction networks and the context-dependent nature of transcriptional responses, our results highlight the importance of understanding interactions between gene expression and splicing for understanding transcriptomic responses to thermal stress. Our results also reveal transcriptional pathways that can inform conservation breeding, translocation, and reintroduction programs for redside dace and other imperilled species by identifying appropriate source populations. SUMMARY STATEMENT Gene expression and alternative splicing interact in response to thermal stress in an imperilled fish, with implications for conservation and mechanisms of thermal tolerance in vertebrate ectotherms.

with fractional counts (--fraction), where input files were the new splice junction-specific .gtf 2 5 7 file, the super-clusters count file from Corset, and the aligned .bam files from STAR to generate 2 5 8 exon counts (Liao et al., 2013). DEU was tested for with DEXseq version 1.34.1 (Anders et al., 2 5 9 2012). As with tests for DGE, RIN scores were used but with a centered and scaled mean around by being fit to experimental treatments. Only exons with differential expression significant at an 2 6 4 q<0.05 were retained for downstream function analyses. Similar to DGE analyses, exons with 2 6 5 higher or lower expression in the thermal stress treatment compared to both the wild and handled 2 6 6 treatments (i.e., |log 2 -fold change|>0 compared to both wild and handled) were retained as 2 6 7 exhibiting 'thermal stress-unique' expression. These steps were performed following guidelines Transcript-Usage-on-a-non-model-organism). Component 2018 databases with q<0.05 were retained for further analyses. As with GO terms 2 7 2 represented by DGE, Revigo was used to explore non-redundant GO terms for exons that  To explicitly evaluate our hypothesis that alternative splicing is an important mechanism 2 7 7 used by redside dace responding to a thermal challenge, we focused on splicing factors uniquely Danio rerio database. Here, genes with significant DEU were identified as possibly important for 2 8 2 thermal stress response. putative transcripts and genes, 155,547 transcripts representing 59,755 genes were annotated 2 9 0 using Trinotate and associated programs after filtering for E-values <1 x 10 -6 and bit scores >50. Corset clustered transcripts from Trinity into 83,217 super-clusters representing 143,841 clusters. Of the 143,841 clusters from Corset irrespective of available annotations, 46,140 had 2 9 5 measurable expression in any single individual. Between the thermal stress group and handled 2 9 6 control, 1,531 clusters showed significant DGE, 786 with relatively higher expression in thermal 2 9 7 stress and 745 with relatively lower expression in thermal stress compared with the handled 2 9 8 control (Table 1; Supplementary Table S2). Between the thermal stress and wild group, 6,770 2 9 9 clusters showed significant DGE, 3,992 with relatively higher expression in thermal stress and   showing relatively higher expression in thermal stress compared to the handled control, and 218  For genes that showed thermal stress-specific expression (i.e., |log 2 -fold changes| > 0 3 2 3 compared to both other groups for thermal stress-specific expression, respectively), 579 were  Table S4). For GO terms related to genes 3 2 7 specific to thermal stress, 32 GO terms were identified among genes with positive expression (21 3 2 8 Biological Process terms, 11 Molecular Function terms) while no GO terms were identified for Revigo with the thermal stress-specific GO terms, 12 Biological Process GO terms and 8  Each of jun, jun-B, jun-D, ier2, myc, c-Fos, and fosB showed higher expression in the 3 4 6 thermal stress treatment than in the wild group, while only jun showed higher expression in the 3 4 7 thermal stress treatment compared to the handled control ( Figure 4). Among 143,841 clusters in the data, 31,042 had detectable exons, and 4,943 of these 3 5 1 clusters (~16%) had at least one exon that showed significant DEU between any two 3 5 2 experimental treatments (Table 2). These clusters with significant DEU were comprised of  Table   3 5 9 S6).

6 0
Exons that showed higher expression in the thermal stress treatment compared to both 3 6 1 controls were represented by 1,688 annotated genes, summarized in 72 GO terms (46 Biological 3 6 2 Process, 13 Molecular Function, and 12 Cellular Component GO terms) (Supplementary Table   3 6 3 S10). Using Revigo with annotated clusters containing exons showing higher thermal stress-  representing exons with lower expression in thermal stress were retained for visualization  Mitochondria-related enrichment terms of mitochondrion organization (GO:0007005),  One splicing factor, pre-mRNA-splicing factor 38B, prpf38b, was upregulated in the  interacting mechanisms used to mount a cellular response to thermal stress. We identified several 3 9 5 hundred differentially transcribed genes unique to thermal stress, and these presumably represent 3 9 6 the molecular mechanisms that redside dace use to respond to acute thermal stress. We also 3 9 7 identified alternative splicing-based responses to thermal stress that may provide a complementary mechanism for an acute thermal stress response. Consistent with the hypothesis 3 9 9 that differential gene expression influences differential exon usage, we identified differentially splicing factor (prpf38b) that was upregulated in the thermal stress-challenged fish, and its 4 0 5 increased expression was related to differential exon usage in downstream genes, representing a 4 0 6 possible stress response pathway that incorporates both alternative splicing and gene expression. By comparing thermal stress-challenged redside dace to handled and wild groups, we As a positive control, we used a set of seven early response genes (jun, jun-B, jun-D, 4 1 6 ier2, myc, c-Fos, and fosB) that would be expected to show a stress response to verify that 4 1 7 whole-organism acute stress was reflected in transcriptomic responses. This panel of genes was 4 1 8 more highly expressed in the thermal stress treatment relative to the wild group. None of the 4 1 9 seven genes in this panel showed differential expression between fish in the thermal stress with wild) confirms that a stress response associated with handling, transport, and confinement likely represent a temperature-specific stress response when compared to both other groups.

2 7
Transcription regulation was prominent among genes differentially expressed in the 4 2 8 thermal stress fish compared to both other groups, indicating that these genes likely play a role in 4 2 9 coping with acute thermal challenge. While the rate-limiting step for protein synthesis is often One concern with CTmax methodology is that it is based on rapid warming, which may are one mechanism of the transcriptomic response to acute thermal stress in the redside dace. analyzed alternative splicing (measured by differential exon usage) for its possible roles in the expression regulation in response to stress, such as heat (De Nadal et al., 2011), and of splicing 4 6 0 in transcription regulation more generally (Smith et al., 1989). Also prominent were metabolism-4 6 1 related enrichment terms among genes showing differential exon usage. Alternative splicing is 4 6 2 one mechanism that regulates cellular metabolism, such as by splicing factors being targets of 4 6 3 metabolic stress (Biamonti et al., 2018). Energy utilization was found to change in response to 4 6 4 warming acclimation in fish, with decreased aerobic scope but increased energy utilization  fishes, these processes may play important roles in the response to increasing temperatures. One of our main goals was to test the hypothesis that there are direct and interacting links 4 7 7 between patterns of alternative splicing and differential gene expression in response to thermal 4 7 8 stress. To do this, we carefully searched for splicing factors among the genes that were found to 4 7 9 be differentially expressed in thermally stressed fish relative to both control groups. One splicing 4 8 0 factor, prpf38b, fit those criteria. Because protein abundance and mRNA levels are often can lead to large changes in pathway flux (e.g., Hochachka and Somero, 2002), the small log 2 -4 8 3 fold change values we measured may be biologically important.

8 4
The splicing factor prpf38b may influence two important genes that are part of the because jun was more abundant in the thermal stress treatment, relative to both other groups.

9 5
Activation of c-Jun/AP-1 has been implicated in numerous, sometimes opposing context-4 9 6 dependent cellular stress responses (e.g., both inhibition and activation of apoptotic responses; 4 9 7 (Leppä and Bohmann, 1999). More broadly, our data linking prpf38b, rbm39, and jun illustrates 4 9 8 how the interplay between splicing and gene expression may be an essential element of the 4 9 9 redside dace thermal stress response.

0 0
Beyond rbm39 and jun specifically, prpf38b has been linked to the co-expression and 5 0 1 direct regulation of numerous other genes (Ouyang et al., 2021). Therefore, while jun may be 5 0 2 one regulatory element with far-reaching effects for cellular stress responses, prpf38b may have 5 0 3 effects beyond jun, as well. As a splicing factor that was uniquely differentially transcribed in the between the transcriptional mechanisms of differential gene expression and alternative splicing.

0 6
In the present data, the separate gene expression and splicing analyses present enrichment term prpf38b; see also Boyle et al., 2017;Davidson, 2010). Therefore, further connections between 5 1 0 the mechanisms likely exist but remain largely unexplored, possibly because of context- need to study these two mechanisms in tandem. Understanding the mechanisms of thermal tolerance, and how these vary among finding suggests that energy mobilization may be a fundamental factor that limits thermal tolerance. Consistent with this idea, improved nutrition has increased the thermal tolerance of other species (Hardison et al., 2021;Lee et al., 2016;Robinson et al., 2008). We speculate that 5 2 9 there may therefore be negative consequences for the ability of fishes to cope with thermal stress 5 3 0 in conjunction with other environmental factors that also increase energy demands (e.g., the 5 3 1 metabolic detoxification of pollutants; Du et al., 2018). Future work using transcriptomic 5 3 2 approaches will be useful for identifying shared pathways among these physiological processes and therefore for understanding the consequences of multiple stressors.

3 4
In addition to the shared patterns of whole organism physiology across species, cellular and Somero, 2020). Therefore, elements of the transcriptional response to thermal stress as 5 3 7 studied in redside dace here may be applied to understanding transcriptional mechanisms in 5 3 8 many cyprinids and freshwater fishes. While organisms in freshwater habitats often face 5 3 9 simultaneous stressors that limit the potential for ecological inferences in studies that use one  There is widespread interest in understanding patterns of inter-individual and inter-