Multidimensional plasticity jointly contributes to rapid acclimation to environmental challenges during biological invasions

Abstract

Rapid plastic response to environmental changes, which involves extremely complex underlying mechanisms, is crucial for organismal survival during many ecological and evolutionary processes such as those in global change and biological invasions. Gene expression is among the most studied molecular plasticity, while co- or posttranscriptional mechanisms are still largely unexplored. Using a model invasive ascidian Ciona savignyi, we studied multidimensional short-term plasticity in response to hyper- and hyposalinity stresses, covering the physiological adjustment, gene expression, alternative splicing (AS), and alternative polyadenylation (APA) regulations. Our results demonstrated that rapid plastic response varied with environmental context, timescales, and molecular regulatory levels. Gene expression, AS, and APA regulations independently acted on different gene sets and corresponding biological functions, highlighting their nonredundant roles in rapid environmental adaptation. Stress-induced gene expression changes illustrated the use of a strategy of accumulating free amino acids under high salinity and losing/reducing them during low salinity to maintain the osmotic homoeostasis. Genes with more exons were inclined to use AS regulations, and isoform switches in functional genes such as SLC2a5 and Cyb5r3 resulted in enhanced transporting activities by up-regulating the isoforms with more transmembrane regions. The extensive 3′-untranslated region (3′UTR) shortening through APA was induced by both salinity stresses, and APA regulation predominated transcriptomic changes at some stages of stress response. The findings here provide evidence for complex plastic mechanisms to environmental changes, and thereby highlight the importance of systemically integrating different levels of regulatory mechanisms in studying initial plasticity in evolutionary trajectories.

INTRODUCTION

In the past several decades, biogeographical patterns of many species have been rapidly reshaped globally with the intensification of human activities (Wiens 2016; Sandoval-Castillo et al. 2020), posing a challenge to the survival of organisms. Compared with long-term evolutionary adaptation across many generations, rapid responses to a novel or changing environment during the lifetime (i.e., rapid acclimation) are initially plastic, and these environmentally induced rapid plastic responses can buffer the adverse effects of local environments to facilitate organismal survival and population persistence (Sandoval-Castillo et al. 2020). However, highly dynamic and complex plastic responses have been reported when taking different plastic traits of interest, environmental contexts, and timescales into consideration (Metzger and Schulte 2018; Fox et al. 2019; Huang and Zhan 2021). Thus, untangling the complexity of plastic mechanisms during the response to environmental changes is crucial for understanding the organismal performance in many ecological and evolutionary scenarios, such as those in global change and biological invasions.

Biological invasions provide contemporary “natural experiments” for studying complex mechanisms of plastic response to environmental changes (Zhan et al. 2015; Santi et al. 2020). During invasions, range expansions can occur at extremely large geographical scales, rapidly spanning distinct environmental regimes over relative short timescales (Zhan et al. 2010, 2015). When invasive species are initially introduced to novel environments, rapid plastic response is firstly stimulated to mitigate environmental stresses, and subsequently might participate in adaptive evolution after successful establishment. For example, the plastic response of producing more clonal propagules to water availability in the invasive sunflower Helianthus tuberosus facilitated the plant survival and persistence when introduced to riparian habitats (Bock et al. 2018). Marine invasive species are often introduced into novel habitats by human activity-mediated vectors such as ballast water and hull fouling during shipping (Ruiz and Carlton 2003; Lin et al. 2020). During the transport stage of trans-oceanic shipping, the drastic salinity shifts can be ∼15 ‰ in ballast tanks during a several-day voyage (Briski et al. 2013), and such fluctuation ranges can be even larger as fouling communities are transported among ports (Bereza and Shenkar 2022), thus posing huge osmotic challenges to marine invasive species. The rapid plastic response of maintaining osmotic homoeostasis is crucial for the survival of marine invaders and subsequent successful invasions to environmentally distinct habitats (Jeffries et al. 2019; Posavi et al. 2020). However, most studies often focused on a single or specific level (e.g., Huang et al. 2017; Li et al. 2020; Chen et al. 2022), ignoring the intrinsic integrity and complexity of plastic response mechanisms.

Gene expression is among the most well-known molecular plastic traits, mainly because transcriptional regulations, which link genotypes and phenotypes, can be immediately induced to cope with environmental stresses during many ecological and evolutionary processes (Fu et al. 2021; Josephs 2021). Many studies have examined the roles of gene expression plasticity in organismal acclimation and adaptation to various environmental changes. For instance, the higher gene expression plasticity under acute thermal stresses was associated with enhanced thermal tolerance of a highly invasive inland siversides (Menidia beryllina) when compared to endangered native delta smelt (Hypomesus transpacificus), potentially conferring competitive advantages on invasive fish under warming scenarios (Komoroske et al. 2021). Gene expression change, however, is not the only intermediate regulatory process linking genotypes and phenotypic traits. Alternative splicing (AS), as an important co- or post-transcriptional mechanism of generating multiple mRNA isoforms from a single gene, has received relatively little attention despite its importance in adaptation (Salisbury et al. 2021; Singh and Ahi 2022; Verta and Jacobs 2022). Several recent studies have demonstrated the contribution of AS to phenotypic novelty such as the caste differentiation of bumble bee (Price et al. 2018) and parallel ecological adaptation of ecotype-specific feeding morphology in a salmonid fish (Jacobs and Elmer 2021). Furthermore, genome-wide AS patterns could be rapidly reprogrammed at a short timescale to cope with environmental changes (Shalgi et al. 2014; Suresh et al. 2020; Tian et al. 2020), making AS a good candidate molecular plastic trait for studying initial plasticity. In addition to AS, alternative polyadenylation (APA) is another post-transcriptional mechanism of generating multiple transcript isoforms with different lengths of 3′-untranslated regions (3′UTR) from a single gene, mainly through differential usage of polyadenylation sites (PAS). APA isoforms with different 3′UTR length can present quite different mRNA characteristics such as distinct mRNA stability, cellular localization, and even coding sequences (Sadek et al. 2019), resulting in protein diversity or different phenotypic outcomes. Although accumulating evidence illustrates that APA should be involved in various biological processes (Zheng et al. 2018; Pereira-Castro and Moreira 2021), it remains largely unexplored and even overlooked when investigating the mechanisms of rapid response to environmental stresses.

All of the abovementioned molecular plastic changes, including gene expression, AS, and APA isoform switch, integrally affect the quantity and quality of mRNA and could potentially contribute to phenotypic variation in changing environments. It has been proposed that different types of plasticity should act on distinct environmental changes and different timescales (Metzger and Schulte 2018; Fox et al. 2019). Therefore, in order to untangle the complexity of initial plastic changes, it is necessary to investigate the regulatory interplay among these mechanisms, as well as their relative importance during responses to different environmental challenges. As such, we used a model marine invasive species, the solitary ascidian Ciona savignyi, to study its initial plastic response mechanisms to salinity changes during invasions. Ciona savignyi is usually considered to be a northern Asian native, but now has been widely recorded globally such as on the coasts of North America, and even New Zealand and Argentina in the Southern Hemisphere (Smith et al. 2010; Zhan et al. 2015; Fofonoff et al. 2018). During the transport stage, ascidian individuals can encounter drastic salinity shifts, and studies show that they can survive a wide range of salinity from 18% to 40% (Lambert and Lambert 2003; Fofonoff et al. 2018). As an osmoconformer animal, the osmolarity of extra- and intracellular fluids can change with the external environment (Sokolov and Sokolova 2019), but the rapid regulation process has yet to be fully determined.

In order to systematically elucidate the underlying plastic mechanisms of the salinity stress response, we used C. savignyi to conduct time-course hyper- and hypo-salinity challenges to investigate the enzyme activity, gene expression, AS, and APA plasticity (Supplemental Fig. S1A). We aim to study the interplay of different plastic mechanisms at transcriptional and post-transcriptional regulatory levels and further identify key candidate genes and functional pathways responsible for rapid salinity acclimation.

RESULTS

Physiological response to salinity stresses

We detected significant and dynamic changes of four physiological indicators after salinity challenges. Specifically, the content of malondialdehyde (MDA) was rapidly accumulated soon after 1 h of high salinity (HS) and low salinity (LS) challenges, maintained at a significantly high level at 24 h of HS but significantly decreased at 24 h of LS, and finally returned to the control level at 48 h after HS and LS (Fig. 1A). As for the antioxidant enzymes, the catalase (CAT) activity significantly decreased throughout the whole stress treatment process (Fig. 1B), while the superoxide dismutase (SOD) activity was significantly induced at 1 h of HS and LS, remained rising at 24 h of HS, but shifted to a repressed status after 48 h of HS and LS (Fig. 1C). The Na⁺/K⁺ ATPase activity was significantly induced across all three time points under LS, but did not increase until 48 h after HS (Fig. 1D). These results illustrated that the antioxidant system and ion transport regulation were actively involved in the physiological response to salinity stresses.

FIGURE 1.

Time course physiological response to high and low salinity stresses in Ciona savignyi. Four basic physiological indices include the oxidant stress indicated by MDA (A), antioxidant activity of CAT (B) and SOD (C), and ATP-dependent ion pump activity of Na⁺/K⁺ ATPase (D). The asterisk (*) indicates the statistical difference (P < 0.05) between the treated and corresponding control groups. The lower and upper boundaries of boxes show the 25/75 percentile, horizontal lines across boxes indicate the median, and whiskers extending above and below boxes represent the maximum and minimum values.

Gene expression response to salinity stresses

An average of 19.68 million clean reads with high sequence quality per sample were obtained, and the mapping rate per library ranged from 76% to 87% with an average of 81.72%, resulting in at least 10 million mapped reads per sample for subsequent analyses (Supplemental Table S1). Based on all 12,172 annotated genes in the reference genome, the overall transcriptomic profile from principal component analysis (PCA) analysis showed that, although variation among individuals could not be explained by salinity changes or durations alone, the low salinity-treated ascidians, particularly for those at 24 and 48 h, were more separated from the unstressed samples than high salinity-challenged ascidians (Supplemental Fig. S1B), indicating LS-induced stronger transcriptomic reprogramming.

After salinity decreased from 30‰ to 20‰, 101, 1172, and 651 genes underwent significant gene expression changes after 1, 24, and 48 h, respectively, while 241, 103, and 264 genes significantly changed their expression when salinity increased from 30‰ to 40‰ (Fig. 2A). The Mann–Whitney U-test (GO-MWU) analysis was performed to explore the functional representation of salinity stress response genes. The immediate response to LS and HS at 1 h was dominated by up-regulating signal transduction and potassium ion transport functions but a repressing biosynthetic process (Supplemental Figs. S2A, S3A). By 24 h, the vesicle-mediated transport process was significantly up-regulated under an LS challenge, and notably, the neurotransmitter transport process showed an opposite direction of gene expression regulation to LS and HS (Supplemental Figs. S2B, S3B). After 48 h of HS, the initial down-regulation of the ribonucleoprotein complex assembly process switched to up-regulation (Supplemental Fig. S2C), but no biological processes were significantly enriched at LS48.

FIGURE 2.

Differential expression induced by high salinity (HS) and low salinity (LS) stresses. (A) The number of up- and down-regulated differentially expressed genes (DEGs) at 1, 24, and 48 h after two types of salinity treatments. (B) Venn diagram showing common differentially expressed genes (DEGs) and stress-specific DEGs. (C–J) GSEA plots showing significant enrichment of different gene sets (|NES| > 1 and FDR < 0.05), including up-regulation of SLC6 gene family at HS24 (C), HS48 (D), down-regulation of SLC6 gene family at LS24 (E) and LS48 (F), up-regulation of ion transporter at HS24 (G), down-regulation of peptidase activity genes at HS48 (H), as well as down-regulation of amino acid (AA) biosynthesis process related genes at LS24 (I) and LS48 (J).

Among 1644 LS-responsive and 573 HS-responsive differentially expressed genes (DEGs), only 210 DEGs simultaneously responded to HS and LS (Fig. 2B). To further clarify the differential response to HS and LS, the gene set enrichment analysis (GSEA) against several osmolyte solutes regulating processes revealed that solute carrier proteins 6 (SLC6) gene family members were significantly enriched among the up-regulated genes at HS24 and HS48 (Fig. 2C,D), while among the down-regulated genes at LS24 and LS48 (Fig. 2E,F), indicating free amino acid (FAA) accumulation and FAA loss responding to high salinity and low salinity stresses, respectively. Moreover, the ion transmembrane transporting process was significantly stimulated at HS24 while peptidase activity mediated proteolysis was inhibited at HS48 (Fig. 2G,H). The amino acid biosynthetic process was significantly enriched among down-regulated genes at LS24 and LS48 (Fig. 2I,J), potentially contributing to the decrease of cellular osmolyte solutes to cope with ambient low salinity.

Hub genes responding to salinity stresses

To identify key hub/driver genes in transcriptional regulatory networks of salinity stress response, the weighted gene correlation network analysis (WGCNA) was performed using all expressed genes, and each module represents a cluster of genes exhibiting coexpression profiles across all samples. A total of 13 modules were identified with module size ranging from 42 to 2161 genes. The relationship between a module eigengene (the first principal component of gene expression value) and salinity gradients showed that the blue module was the most negatively correlated with salinity gradients, while the light cyan module was positively correlated with salinity gradients (Supplemental Table S2). The overall gene expression in the blue module was significantly induced by low salinity stress but suppressed by high salinity stress (Fig. 3A). Furthermore, the gene expression was also significantly influenced by the stress duration and their interaction (i.e., stress duration × salinity) based on linear mixed model analysis (both P < 0.001), with higher gene expression of the blue module in the later stage of the salinity challenge (Supplemental Fig. S4). Hub genes (the top 30 genes with high intramodular connectivity) in this module contained transmembrane transport-related genes SLC49A3 and mup-4, cell growth related gene IGFBP, methyltransferase gene METTL27, two zinc metalloproteinase genes Nas-13 and Nas-16, as well as an adhesion molecule OTOA (Fig. 3C). All hub genes are listed in detail in Supplemental Table S3.

FIGURE 3.

Gene coexpression analysis to identify hub genes responsible for salinity challenges. Eigengene value distribution of blue (A) and light cyan (B) modules show the gene expression level in high salinity (HS), low salinity (LS), and control groups. The lower and upper boundaries of boxes show the 25/75 percentile, horizontal lines across boxes indicate the median, and whiskers extending above and below boxes represent the maximum and minimum values. Network visualization of gene interactions of the top 30 with highest connectivity is shown in blue (C) and light cyan (D) modules. Each node represents an expressed gene labeled with a gene symbol and its detailed information is included in Supplemental Table S3.

Contrary to the expression profile of the blue module, the gene expression in the light cyan module was significantly induced by high salinity stress but suppressed by low salinity stress (Fig. 3B), but was not influenced by the stress duration and their interaction. The hub genes in the light cyan module included three solute carrier proteins (SLC) family members SLC16A12, SLC49A3, and SLC5A8, two fucosyltransferase genes FUT3 and FUT6, oxidative stress response gene FANCD2, and several protein processing genes such as HUB1, SZT2, and C2dc3 (Supplemental Table S3; Fig. 3D). In addition to the annotated genes, many novel genes with unknown functions were also identified as hub genes in these two modules.

Differentially alternative spliced genes (DASG) and differentially expressed APA (DeAPA) genes in response to salinity stresses

Five types of alternative splicing (AS) events were detected using rMATS, including skipped exon (SE), retained intron (RI), alternative 5′ and 3′ splice site (A5′SS and A3′SS), and mutually exclusive exon (MXE). SE was the most dominant AS type in response to salinity stresses, accounting for 70.66% of all splicing events, followed by MXE (11.75%), A5′SS (10.38%), A3′SS (6.96%), and RI (0.24%) (Fig. 4A). Accordingly, the subsequent analyses mainly focused on salinity stress-induced SE changes. After salinity decreases, the inclusion level of 114, 147, and 170 exons changed significantly at 1, 24, and 48 h, affecting the isoform composition of 44, 101, and 77 genes, respectively, while the high salinity stress significantly induced the inclusion level changes (the percent spliced in, ΔPSI) of 76, 110, and 81 exons in 46, 48, and 54 genes accordingly (Fig. 4B). Among these differentially alternative splicing events (DASEs), the number of obtaining (ΔPSI > 0) and losing (ΔPSI < 0) a particular exon after salinity changes was comparable (Supplemental Fig. S5), and we did not observe a clear tendency of salinity stress–induced exon loss or retainment. The induced DASGs varied considerably over time, with only seven and three DASGs shared in all three time points after low salinity and high salinity stresses, respectively (Fig. 4C), indicating dynamic AS response to salinity challenges. Among the seven DASGs that persistently underwent AS regulation in response to low salinity stress, one noteworthy gene was annotated as ABC transmembrane type-1 domain-containing protein (Abcc1), which has the ATPase-coupled transmembrane transporter activity and functions in transporting various substrates. While the DDX46 gene was among the three common DASGs in response to the high salinity stress, which is a probable ATP-dependent RNA helicase and plays essential roles in the splicing process.

FIGURE 4.

Differentially alternative splicing and alternative polyadenylation profiles in response to salinity stresses. (A) Average percentages of five main types of alternative splicing events between all pairwise comparisons of treatment and control groups detected by rMATS. (B) The number of differentially alternative splicing genes (DASE) and differentially alternative spliced genes (DASG) in each treatment-control comparison group. (C) Venn diagram showing DASGs at different durations under low and high salinity stresses, respectively. (D) The number of differentially alternative polyadenylation (DeAPA) genes with shortening and lengthening 3′UTR after salinity stress.

With APAtrap, we identified 7545 genes with alternative poly(A) sites among all samples, accounting for 62% of all genes in the reference genome. After salinity changes, 1179 and 808 DeAPA genes significantly changed their poly(A) sites usage in response to low salinity and high salinity stresses, respectively, with 403 genes commonly responding to both stresses. At all three time points of low salinity or high salinity stresses, we detected more salinity stress-induced shortening genes than lengthening genes (Fig. 4D), indicating a global 3′UTR shortening profile caused by preferential usage of proximal poly(A) sites. To further test whether salinity stress-induced 3′UTR shortening was related with enhanced gene expression abundance, we compared the gene expression changes of shortening and lengthening gene sets, but we did not detect a significantly different gene expression response (Supplemental Fig. S6).

Stress-induced isoform switch events

Two significant isoform switch events were identified using 3D RNA-seq APP - SLC2a5 (Gene ID: ENSCSAVG00000011114, solute carrier family 2, facilitated glucose transporter member 5) and Cyb5r3 (Gene ID: ENSCSAVG00000011164, NADH-cytochrome b5 reductase 3) (Fig. 5). Two alternative transcript isoforms were transcribed from the SLC2a5 gene, which functions as a fructose transporter in salt uptake, isoform 1 (transcript ID: ENSCSAVT00000019133) with 10 exons and isoform 2 (transcript ID: ENSCSAVT00000019134) with nine exons (Fig. 5A). Under normal conditions, isoform 2 was the dominant transcript, but its expression was down-regulated and the ratio of isoform 1 increased and became the dominant transcript under both high salinity and low salinity stresses (Fig. 5B). By comparing the transcript and deduced protein structures, we found that the second exon of isoform 1 was skipped in isoform 2, leading to one more transmembrane region predicted in the deduced protein sequence from isoform 1 than that from isoform 2 (Fig. 5C). Such a change could potentially affect the capacity of fructose transportation and play a role in the osmatic regulation under salinity stresses.

FIGURE 5.

Two isoform switch events occurring in SLC2a5 and Cyb5r3 genes. (A) The gene structure of two transcript isoforms of SLC2a5 gene. (B) The expression changes of two transcript isoforms of SLC2a5 gene across all experimental groups. (C) Transmembrane region prediction of two transcript isoforms of SLC2a5 gene. (D) The gene structure of two transcript isoforms of Cyb5r3 gene. (E) The expression changes of two transcript isoforms of Cyb5r3 gene across all the experimental groups. (F) Transmembrane region prediction of two transcript isoforms of Cyb5r3 gene.

Another significant isoform switch event occurred in the Cyb5r3 gene, leading to two alternative transcripts with different 5′ coding regions (Fig. 5D): isoform 1 (transcript ID: ENSCSAVT00000019225) and isoform2 (transcript ID: ENSCSAVT00000019226). Under control conditions and high salinity stress, isoform 1 was the dominant transcript, while it was down-regulated under low salinity stress, making isoform 2 as the dominant one (Fig. 5E). By predicting the transmembrane region of deduced proteins, the appearance of a transmembrane region in isoform 2 made it a transmembrane protein locating at the endoplasmic reticulum membrane or mitochondrion outer membrane (Fig. 5F), while the lack of a transmembrane region in isoform 1 made it a soluble protein which mainly functions in cytoplasm.

Complementary regulations of AS, APA, and gene expression mechanisms

Overall, the number of genes undergoing AS regulation was far fewer than the genes with mRNA abundance changes in response to salinity shifts, with all splicing abundance ratios (SAR) lower than 100 (Fig. 6A). The number of genes undergoing APA regulation far exceeded that of DEGs at LT1 and HS24 (polyadenylation abundance ratio PAR > 100, Fig. 6A), suggesting the dominant role of the APA mechanism in transcriptomic plasticity at those time points. By comparing the genes involved in gene expression, AS and APA regulations in response to salinity stresses, we found that a relatively low proportion of shared genes coordinately acted on more than two regulatory layers, with only four and zero genes shared by all three regulations under low salinity and high salinity stresses, respectively (Fig. 6B,C). Most of the DASGs and DeAPA genes were not differentially expressed, and most of DeAPA genes were not differentially alternatively spliced, indicating relatively independent regulations among AS, APA, and gene expression plasticity. To test whether gene structure affected the choice of different regulatory mechanisms, we found that the genes with more exons tended to adopt AS changes to respond to salinity challenges (Fig. 6D,E). Different regulatory mechanisms acted on varied genes and corresponding biological functions (Supplemental Fig. S7); for example, ion transmembrane transport-related genes changed gene expression to cope with decreased salinity, and genes responding to stimulus were subjected to AS regulation, while organic substance transport-related genes adopted APA regulation (Supplemental Fig. S7). Although the three response mechanisms affected different genes, there exist some interactions among them; for example, the genes associated with mRNA poly(A) tail shortening were alternatively spliced, and RNA splicing relevant genes were under both gene expression and APA regulations in response to low salinity stress (Supplemental Fig. S7A).

FIGURE 6.

Comparisons of genes underlying three different regulatory mechanisms including gene expression, alternative splicing, and alternative polyadenylation. (A) SAR (splicing abundance ratio) and PAR (polyadenylation abundance ratio) in different treatment groups, indicating the number of alternative splicing changes or poly(A) changes relative to DEGs. (B) Venn diagram showing DEG, DASG, and DeAPA under LS and HS stress (C). (D) Exon number distributions of the genes regulated by three different mechanisms under LS and HS stress (E), respectively. Asterisk (*) indicates the statistical difference (P < 0.05), “ns” indicates nonsignificant.

DISCUSSION

Exploring the complex mechanisms underpinning environment-induced plastic response is an important and prerequisite step for understanding organismal performance in rapidly changing environments. Post-transcriptional processes such as AS and APA regulations have been widely overlooked in organismal response to environmental changes (Salisbury et al. 2021; Verta and Jacobs 2022). In this study, we used a model marine invasive species, Ciona savignyi, to investigate multiple layers of plastic response to salinity shifts, which simulated the real salinity changes during several days’ trans-oceanic voyages of the invasion process (Klein et al. 2009; Bereza and Shenkar 2022). Our results showed that different layers of plastic response were rapidly and dynamically induced to cope with salinity changes. Based on the number of affected genes, initial plasticity was dominated by gene expression plasticity at most time points under high salinity and low salinity stresses, except for APA plasticity as the dominant mechanism at LS1 and HS24. We also found that gene expression, AS, and APA plastic response acted on different genes and biological functions, playing complementary roles in rapid salinity acclimation. Interestingly, the choice of target genes by different mechanisms was related with gene structures, for example, genes under AS regulation tended to have more exons. Moreover, we observed a partly different plastic response induced by low salinity and high salinity stresses, and identified key biological functions and candidate genes involved in salinity acclimation. Altogether, the results obtained here reveal the complex interplays among different initial plastic responses to the time course of salinity challenges, providing novel insights into the roles of complex molecular plasticity toward adaptive evolution.

Interplay of multidimensional plasticity under salinity stresses

Environmental changes often trigger oxidative damages to organisms, and the antioxidant response is an important form of physiological plasticity to eliminate the excess oxidation substances (Mittler 2002; Jimenez et al. 2015). The rapid accumulation of MDA content after 1 h of salinity challenge indicated C. savignyi was under oxidative stress, while its subsequent recovery reflected the activation effect of SOD enzyme activity. Although both SOD and CAT are generally considered as the first line of the antioxidant defense system (Zhang et al. 2020), we observed significantly increased SOD activity while CAT activity decreased under salinity stress, and this contrasting response pattern was also reported by the blue crab Portunus trituberculatus under ammonia exposure (Meng et al. 2021). This may be caused by a different sensitivity of SOD and CAT to environmental changes; for example, CAT activity was seriously inhibited under osmotic pressure, leaving SOD as the dominant antioxidant enzyme. Subsequently, we further explore the underlying mechanism of these physiological responses by investigating their gene expression changes; however, none of them (SOD, CAT, and NKA) was identified as DEGs. Therefore, the direct interplay between physiological and gene expression plasticity was still hard to elaborate owing to the lack of several key intermediate processes such as protein translation or modification.

Interestingly, gene expression, AS, and APA plastic response affected different genes and biological functions, indicating their complementary regulatory roles in rapid salinity acclimation. The relatively independent roles of gene expression and AS were also reported in the parallel adaptive evolution of salmonid fish (Jacobs and Elmer 2021), environmentally determined phenotypes of pea aphid (Grantham and Brisson 2018), acute copper stress response of Daphnia (Suresh et al. 2020), and short-term maintaining mineral nutrient homeostasis of rice (Dong et al. 2018). Consistent with our results, there were more genes undergoing gene expression changes than AS changes in most studies (Shalgi et al. 2014; Grantham and Brisson 2018; Suresh et al. 2020; Tian et al. 2020; Huang and Zhan 2021). However, a few studies demonstrated contrasting results with more or comparable genes undergoing AS regulation (Singh et al. 2017; Dong et al. 2018; Jacobs and Elmer 2021), indicating that their relative importance varied among species, types of environmental stresses, and challenge durations. Additionally, when inferring their relative importance of gene expression and AS regulatory mechanisms based on the number of affected genes, the discrepancy derived from different AS analysis methods should also be considered. For example, rMATS and other events-based methods seem to be more conservative than the DEXseq tool on the detection of differential exon usage, resulting in less DASGs accordingly (Jacobs and Elmer 2021). We reperformed AS analysis using DEXseq and compared the results with that of rMATS, and found quite a different number of DASGs between these two methods (Supplemental Fig. S8A). Much more DASGs obtained from the DEXseq tool would lead to most of the SAR values greater than 100 (Supplemental Fig. S8B), and further could change our previous inference of the importance of gene expression and AS. Given that it is difficult to assess which method is more superior than the others, it should be noted that the method selection can have significant influence on AS analysis. Compared with AS, studies on APA plastic response to environmental changes and its relationship with gene expression or AS have just emerged in recent years. Several studies demonstrated that the 3′UTR landscape was reshaped under environmental stresses through the APA mechanism (Zheng et al. 2018; Sadek et al. 2019; Ye et al. 2019). Consistent with findings here, those studies detected dominant shortening profiles of 3′UTRs. Such environmentally induced 3′UTR shortening can lead to a higher gene expression level because shortened 3′UTR might lack some miRNA target sites and further escape from mRNA degradation processes (Tian and Manley 2017; Ye et al. 2019). However, we did not detect such a relationship between 3′UTR shortening and higher gene expression, and the low overlap between DEGs and DeAPA genes also indicated independent roles of these two mechanisms. Moreover, the number of DeAPA genes far exceeded DEGs at LS1 and HS24, suggesting important roles of the APA mechanism in specific phases of salinity acclimation. In addition, even though compelling evidence showed AS and APA as coordinated mechanisms (Nazim et al. 2018), our results, particularly the extremely low overlap between the affected genes, support independent roles of AS and APA during the plastic response to salinity challenges.

One possible explanation for independent relationships of gene expression, AS, and APA mechanisms is that they are controlled by different regulatory machineries involving distinct cis-elements and trans-acting factors, such as promoters and transcription factors for gene expression, splice sites, and spliceosome for AS, and polyadenylation sites and cleavage factors for APA (Schaefke et al. 2018; Sadek et al. 2019). Meanwhile, we also found a certain extent of interactions among different regulatory layers; for example, RNA splicing machinery–related genes were differentially expressed, and APA machinery–related genes were alternatively spliced under low salinity stress. Despite their subtle relationship, the affected biological functions such as transmembrane transporting, stress response, biosynthesis or metabolic processes of amino acid and fatty acid have been widely reported in rapid salinity acclimation in other species (Maynard et al. 2018; Jeffries et al. 2019; Niu et al. 2020; Posavi et al. 2020). Therefore, AS and APA-based plasticity responses should provide additional layers of initial plasticity to gene expression, jointly contributing to a rapid response to salinity stresses. As mentioned above, the regulatory relationship analysis among different mechanisms might be largely dependent on the methods based on short-read RNA sequencing. Recent advances in full length transcriptome sequencing have great potential to overcome the inherent limitations of short-read assembling, and thus provide unprecedented approaches to identifying accurate AS and APA.

Contrast between plastic response to high and low salinities

Salinity fluctuates dramatically during trans-oceanic voyages (Ghabooli et al. 2016; Bereza and Shenkar 2022), posing environmental constraints for the survival and establishment of invasive species. As osmoconformers, ascidians must be able to adjust their intercellular osmolarity when ambient salinity changes (Sokolov and Sokolova 2019). Osmoregulatory mechanisms of osmoconformers have been extensively assessed in mollusks, and similar to results obtained in C. savignyi here, free amino acids (FAAs) were the preferential osmolytes to regulate osmolarity (Yancey 2005; Liu et al. 2018; Sokolov and Sokolova 2019; Pourmozaffar et al. 2020). The SLC6 gene family is responsible for transporting amino acids, nutrients, or other kinds of osmolytes (Bröer and Gether 2012), the expansion of which was found in the genome of C. savignyi (Ren et al. 2019). Our results showed that SLC6 genes were significantly up-regulated under high salinity stress and down-regulated under low salinity stress, as well as the down-regulation of the amino acid biosynthetic process under low salinity stress, consistently supporting that FAA, as the main osmolytes, accumulated to increase intracellular osmolarity to cope with ambient high salinity stress but decreased under low salinity stress. Besides, some studies argued that inorganic ions also participated in salinity acclimation of osmoconformers, but in fact the relative contributions of inorganic ions and FAA largely varied among species, tissue types, stress intensity, and duration time. For example, the Pacific oyster Crassostrea gigas transcriptionally triggered the FAA biosynthesis process to cope with salinity stresses (Zhao et al. 2016), while the white clam Amarilladsma mactroides mainly recruited Na⁺ and very low amounts of FAAs and the lagoon cockle Erodona mactroides only used Na⁺ and K⁺ as effectors for osmoregulation (Medeiros et al. 2020). We observed in this study that ion transmembrane transporting related genes were significantly up-regulated at HS48, indicating that their involvement in high salinity response might be environment-specific or time-dependent. However, more direct evidence such as the time-course changes of cellular osmolarity, ion (e.g., Na⁺ and K⁺) or FAA (e.g., taurine, glycine, alanine, and proline, etc.) concentration under salinity stresses should be provided to illustrate the dynamic osmoregulatory process in further studies.

Key candidate genes involved in response to salinity stresses

SLC genes are a group of over 400 membrane proteins to transport an extraordinarily diverse group of solutes including organic molecules, inorganic ions, and gas ammonia (He et al. 2009). Different SLC genes were found to be involved in response to salinity stresses, such as SLC39A6 and SLC5A9 in the tiger puffer Takifugu rubripes (Jiang et al. 2020), SLC6A6 in Cynoglossus semilaevis (Si et al. 2018), and SLC4A4 and SLC12A2 in hybrid tilapia (Su et al. 2020). Our present study found that SLC gene family members were consistently identified as hub genes on gene expression regulatory layer using WGCNA analysis, including SLC49A3, SLC5A8, and SLC16A12, indicating their important roles in salinity acclimation of ascidians. Moreover, we also identified a key isoform switch event that occurred on another SLC gene, SLC2a5. The SLC2a5 gene is a member of the GLUT transporter specific to fructose (Burant et al. 1992), and its expression change has been linked to several metabolic disorders and human cancers (Barone et al. 2009). A total of 12 transmembrane regions (TMs) are necessary for its normal function, and incomplete TMs might impair its transporting activities (Mueckler and Makepeace 2002). In the present study, we found that the shorter isoform of SLC2a5 with 11 TMs was constitutively expressed in the control group, while the longer isoform with 12 TMs was induced under both salinity stresses, indicating the facilitated fructose transporting function for osmoregulation under salinity stresses.

In conclusion, our results revealed that rapid plastic response patterns varied greatly with salinity shift context (HS and LS), short timescales (1, 24, and 48 h), and distinct regulatory mechanisms (gene expression, AS, and APA), and such multidimensional plastic responses involve different genes and biological functions. Considering that different layers of molecular plastic responses will be jointly translated to functional protein changes and further phenotypic diversity that natural selection can act upon, how to integrate all the different plastic responses or discern the dominant plastic response (Supplemental Fig. S9) can offer valuable insights into the predictability of organismal adaptation to environmental changes.

MATERIALS AND METHODS

Organism collection and experimental design

Ciona savignyi individuals were sampled from the aquaculture scallop cages on the coast of Dalian, Liaoning Province, China (38°49′13″N, 121°24′20″E). All collected ascidians were immediately transported to the laboratory and then acclimated in aerated tanks with an ambient temperature of 15 ± 1°C and salinity of 30‰ (the measured field parameters) for 1 wk. Following acclimation, individuals were randomly assigned to one of the three following groups: control (C), high salinity (HS, 40‰), and low salinity (LS, 20‰), with all other environmental factors unchanged (Supplemental Fig. S1A). The target salinities were achieved by adding instant sea salt or dechlorinated tap water. Considering that pharynx covered with gill slits is the primary interface where ambient water current and metabolic waste directly enter and exit, the pharynx tissues of six replicate ascidians were sampled individually after 1, 24, and 48 h of stress exposure according to our previous study (Huang et al. 2017). All samples were immediately preserved in liquid nitrogen and then stored at −80°C.

Assessment of physiological indicators

We chose four basic physiological indices, including MDA indicating lipid peroxidation of membranes, the enzyme activity of SOD and CAT indicating antioxidant status, and the activity of Na⁺/K⁺ ATPase indicating ion transport regulation, to study the dynamic physiological response to salinity stresses, with six ascidian individuals as biological replicates in each treatment group. The four indices were measured and calculated according to the kit instructions (Jiancheng, Nanjing, China).

RNA extraction, cDNA library construction, and RNA sequencing

Total RNA was extracted from 50–100 mg pharynx tissue of each sample using TRIzol reagent (Ambion) and then treated with DNase (Promega) to remove potential DNA contamination. Three best RNA samples were selected based on RNA integrity (RNA integrity number ≥ 7) and purity, which were evaluated using 1.5% (w/v) agarose gel electrophoresis and Agilent 2100 Bioanalyzer (Agilent Technologies). A total of 27 ascidian samples (three treatments × three time points × three biological replicates) were used for cDNA library construction using the NEBNext Ultra RNA Library Prep Kit (New England Biolabs) and RNA sequencing on the Illunima Hiseq 4000 sequencing platform with the pair end of 150 bp strategy. All raw sequencing data was deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under the accession number SRP152910.

Data processing and DEGs identification

Raw reads were firstly processed to filter out the adaptor and low-quality sequence using Trimmomatic version 0.36 (Bolger et al. 2014), and the acquired clean reads were then mapped to the C. savignyi reference genome CSAV 2.0 using HISAT2 with the following parameters: -p 8 -N 1 -L 20 -i S, 1, 0.5 -D 25 -R 5 –mp 1, 0 -sp 3, 0 -x hisat2_index (Kim et al. 2015). The mapped reads were subsequently assembled to the transcripts and the abundance of gene expression was estimated using StringTie software with default settings (Pertea et al. 2015). The read count number of each gene was normalized using the relative log (rlog) transformation method. Genes with read count lower than 10 in more than 90% samples were excluded from further analyses. DEGs were identified by pairwise comparing gene expression between each environmental treatment group and its corresponding control group at the same time point using the R package DESeq2 (Love et al. 2014), including LS1 versus C1, LS24 versus C24, LS48 versus C48, HS1 versus C1, HS24 versus C24, HS48 versus C48. The genes with the fold change of expression level larger than two as well as the false discovery rate (FDR) lower than 0.05 were considered as potential DEGs. The visualization of variation among all samples was conducted by principal component analysis (PCA) using TMM-normalized TPM values as expression data for all expressed genes.

Weighted gene coexpression network analysis

To explore the potential regulatory relationships among all genes in response to salinity stresses, WGCNA was constructed using the WGCNA package in R (Langfelder and Horvath 2008). According to the tutorials for the WGCNA package, the rlog transformation normalized read count was used as gene expression data. The optimal soft threshold (power = 16) was selected for calculating an adjacency matrix based on the scale-free topology criterion using the pickSoftThreshold function of WGCNA. Coexpression modules were then constructed by a one-step network construction method. The Spearman correlation between module eigengene and salinity gradients was calculated, and significantly correlated modules (the absolute value of correlation coefficient > 0.5, P-value < 0.01) were considered as salinity stress response modules. The effect of stress duration and their interaction (i.e., stress duration × salinity) on the expression of salinity stress response modules was further examined by the linear mixed model (LMM) using the lmer package. The top 30 hub genes in the given module were selected by intramodular connectivity, and the interactions among those genes in salinity responsible modules were visualized by OmicShare tools (http://www.omicshare.com/tools).

Alternative splicing analysis

rMATS v.4.0.2 software was used to detect DASE of five different AS types, including SE, RI, A5′SS and A3′SS, and MXE. rMATS quantified the inclusion level of certain alternative splicing sites (the percent spliced in, PSI or φ) between treatment and corresponding control groups (ΔPSI or Δφ) (Shen et al. 2014). DASEs were defined as those with |ΔPSI| > 10% and adjusted P-value < 0.05. Splicing abundance ratio (SAR) was calculated by DASG (differentially alternative splicing gene)/DEG number × 100, indicating the extent to which AS regulation dominates the transcriptome changes during acclimation to salinity stresses (Habowski et al. 2020). Considering that rMATS detected DASEs at the exon level, the significant isoform switch (IS) events among treatment groups were detected using the 3D RNA-seq APP at the transcript isoform level (Guo et al. 2020). The gene structures of identified IS genes were illustrated using Gene Structure Display Server 2.0 (GSDS, http://gsds.cbi.pku.edu.cn/) by aligning the coding sequences with the corresponding genomic DNA sequences from the same genes. The potential transmembrane regions in proteins deduced from different transcript isoforms were predicted by TMHMM server v2.0 tool (http://www.cbs.dtu.dk/services/TMHMM/).

Alternative polyadenylation (APA) analysis

APAtrap was used to detect potential APA sites and differential APA events between treatment and corresponding control groups at the same time point (Ye et al. 2018). Differentially expressed APA (DeAPA) genes were defined according to the following three parameters: percentage difference (PD, the difference of APA sites usage between two samples) greater than 0.20, FDR lower than 0.05, as well as the positive or negative r value (Pearson product-moment correlation coefficient; in the present study the positive r value indicates shorter 3′UTR, whereas the negative r value indicates longer 3′UTR of the treatment group compared with the corresponding control group). Analogous to SAR, the polyadenylation abundance ratio (PAR) was calculated by DeAPA/DEG number × 100, indicating the extent to which APA regulation dominates the gene expression changes (Habowski et al. 2020).

Gene functional enrichment analysis

Three different analyses were conducted to explore the biological functions of salinity responsive genes. For the genes changing their expression level to cope with salinity shifts, we used Mann–Whitney U-test (GO-MWU) to screen the significantly up- or down-regulated GO categories (https://github.com/z0on/GO_MWU), based on the rank of log P-values from the above DEseq2 analysis (Wright et al. 2015). In order to assess the roles of several key biological processes regulating cellular osmolyte solutes for osmoconformers (Pourmozaffar et al. 2020), such as solute carrier proteins 6 family (SLC6) related to FAA transport, ion transmembrane transport, amino acid biosynthesis, proteolysis which were selected according to GO annotation (Supplemental Table S4), GSEA was performed on gene expression data with GSEA v.4.1.0 program (Subramanian et al. 2005), using the preranked method with a log₂ fold change value as the gene ranking statistic. Gene lists with the absolute value of normalized enrichment score (NES) > 1.0 and FDR < 0.05 were considered significant. For the functional comparison among DEGs, DASGs, and deAPA genes, Gene ontology (GO) functional enrichment analysis was conducted by Hypergeometric test using online Omicshare CloudTools (http://www.omicshare.com/tools). Removing redundancy of the significantly enriched GO terms was then conducted by online tool REVIGO (http://revigo.irb.hr).

DATA DEPOSITION

All raw sequencing data were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under accession number SRP152910.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.

MEET THE FIRST AUTHOR

Xuena Huang.

Meet the First Author(s) is an editorial feature within RNA, in which the first author(s) of research-based papers in each issue have the opportunity to introduce themselves and their work to readers of RNA and the RNA research community. Xuena Huang is the first author of this paper, “Multidimensional plasticity jointly contributes to rapid acclimation to environmental challenges during biological invasions.” Xuena is currently an assistant research fellow at the Research Center for Eco-environmental Science, Chinese Academy of Science, with a research focus on molecular response mechanisms of invasive species to rapid environmental changes.

What are the major results described in your paper and how do they impact this branch of the field?

Compared to well-studied gene expression regulatory mechanisms, the contribution of co- or post-transcriptional mechanisms to the environmental stress response is still largely unexplored. Here, we simultaneously studied gene expression, alternative splicing, and alternative polyadenylation response mechanisms to short-term acute salinity changes in the invasive ascidian Ciona savignyi. Our results revealed their largely independent roles by acting on different genes and biological functions, and their relative importance in the environmental response process varied with stress type and durations. In addition, we further identified key candidate genes involved in osmolyte solutes regulation. These findings highlight the nonredundant roles of co- or posttranscriptional mechanisms and the complex regulatory relationship among different mechanisms in rapid plastic response to environmental changes, and going forward, in evolutionary adaptation.

What led you to study RNA or this aspect of RNA science?

When I joined Dr. Aibin Zhan's Laboratory as a PhD student, I started to learn about the underlying mechanisms of contemporary adaptation to distinct environments using invasive ascidians as a model. I am particularly interested in short-term acclimation to acute environmental changes during the transport stage of biological invasions, and the roles of these environmentally induced plastic changes in long-term adaptation. RNA regulations, which link genotypes and phenotypes, comprise multiple layers of mechanisms such as mRNA/noncoding RNA expression, alternative splicing, alternative polyadenylation, etc., and can be immediately induced by environmental changes. In addition to acute environmental stimulus, RNA regulations are also determined by underlying DNA sequences, which makes RNA regulation an ideal target to demonstrate the short-term plastic responses and long-term adaptive evolution.

During the course of these experiments, were there any surprising results or particular difficulties that altered your thinking and subsequent focus?

Different alternative splicing mechanisms were induced to cope with different environmental stresses. Our previous study revealed a clear tendency of exon loss induced by temperature stresses, while comparable exon loss and exon retain events were observed under salinity stresses in this paper. For alternative polyadenylation mechanisms, we were excited to find extensive 3′UTR shortening profiles under salinity stresses, and it was expected that 3′UTR shortening could lead to higher gene expression because shortened 3′UTR might lack some miRNA target sites and further escape from mRNA degradation processes. However, we did not detect such a relationship, and the biological consequences of 3′UTR shortening profiles still need to be studied.

What are your subsequent near- or long-term career plans?

On one hand, to overcome the inherent limitations of short-read RNA sequencing and assembling, I will use full length transcriptome sequencing methodology to study the roles of alternative splicing and alternative polyadenylation in environmental adaptation. On the other hand, after thoroughly clarifying different transcriptional or co-/-post transcriptional response mechanisms to acute environmental changes within one generation, we want to know whether this environmentally induced plasticity changes will be passed on to the next generation and further affect the long-term adaptive evolution process.