American lobster postlarvae alter gene regulation in response to ocean warming and acidification

Abstract Anthropogenic carbon emissions released into the atmosphere is driving rapid, concurrent increases in temperature and acidity across the world's oceans. Disentangling the interactive effects of warming and acidification on vulnerable life stages is important to our understanding of responses of marine species to climate change. This study evaluates the interactive effects of these stressors on the acute response of gene expression of postlarval American lobster (Homarus americanus), a species whose geographic range is warming and acidifying faster than most of the world's oceans. In the context of our experiment, we found two especially noteworthy results: First, although physiological end points have consistently been shown to be more responsive to warming in similar experimental designs, our study found gene regulation to be considerably more responsive to elevated pCO2. Furthermore, the combined effect of both stressors on gene regulation was significantly greater than either stressor alone. Using a full factorial experimental design, lobsters were raised in control and elevated pCO2 concentrations (400 ppm and 1,200 ppm) and temperatures (16°C and 19°C). A transcriptome was assembled from an identified 414,517 unique transcripts. Overall, 1,108 transcripts were differentially expressed across treatments, several of which were related to stress response and shell formation. When temperature alone was elevated (19°C), larvae downregulated genes related to cuticle development; when pCO2 alone was elevated (1,200 ppm), larvae upregulated chitinase as well as genes related to stress response and immune function. The joint effects of end‐century stressors (19°C, 1,200 ppm) resulted in the upregulation of those same genes, as well as cellulase, the downregulation of calcified cuticle proteins, and a greater upregulation of genes related to immune response and function. These results indicate that changes in gene expression in larval lobster provide a mechanism to respond to stressors resulting from a rapidly changing environment.


| INTRODUC TI ON
The release of anthropogenic carbon emissions into the earth's atmosphere has resulted in large-scale changes in oceanic temperature and pH that have an impact on marine ecosystems and fisheries (IPCC, 2019). IPCC scenarios project pCO 2 -atm will continue to climb resulting in an average temperature increase of 0.6-2.0°C and a pH drop of 0.06-0.32 units, by the end of the century (IPCC, 2013).
These rapid, concurrent stressors are causing concern for the survival of some marine organisms and resilience of key fisheries (Gledhill et al., 2015).
The impacts of ocean acidification (OA) and ocean warming (OW), as single and joint stressors on marine organisms, are complex and species-specific (Fabry et al., 2008;Kroeker et al., 2013;Kurihara, 2008;Whitman & Pörtner, 2013). As genomic analysis techniques increase in efficiency, transcriptomic analyses have become important metrics for quantifying the expression of stress-related genes (Evans & Hofmann, 2012), and enable the examination of a broad range of genetic responses to environmental change on organisms (Harms et al., 2014;Todgham & Hofmann, 2009). Not only can these techniques increase our understanding of the scope of the organismal response, but they have the potential to detect molecular compensation for environmental stress that may otherwise go undetected using more traditional physiological studies (Gracey, 2007).
The American lobster (Homarus americanus) is the single most valuable fishery in North America and has particular socio-economic importance in the Gulf of Maine (National Marine Fisheries Service, 2020). It also has a long history as a well-studied organism in physiology, development, and ecology (Factor, 1995). The American lobster's range extends from Northeastern Canada to the mid-Atlantic states of the United States, where sea surface temperatures during their larval seasons range from below 12°C in the north to over 20°C at the southern extent of its range and in shallow regions (Oulette et al., 2003;Quinn et al., 2013). The entirety of this area falls within the Northwest Atlantic and Gulf of Maine (GoM), where sea surface temperature is warming at a faster rate than the majority of the world's ocean (0.026°C per year since 1980) with a predicted increase ranging 0.03-0.05°C per year (Balch et al., 2012;LeBris et al., 2017;Pershing et al., 2015;Thomas et al., 2017). The region is also particularly susceptible to higher rates of acidification, due to depressed buffering capacity from freshwater inputs of rivers and incoming, relatively fresh currents, resulting in a low aragonite saturation level (Fabry et al., 2008;Gledhill et al., 2015;Salisbury et al., 2008). These conditions place urgent concern and interest in how this species will respond to climate-related stressors, particularly for pelagic life stages in the upper water column. The center of the American lobster fishery has already demonstrated a northward range shift in response to the rapidly warming temperatures within this region, and there is some concern this pattern may continue, affecting fisheries to the south (LeBris et al., 2018).
Here, we evaluate the joint effects of elevated pCO 2 and sea surface warming associated with end-century projected oceanic conditions using next-generation RNA sequencing (RNA-seq) to monitor gene expression changes in early life stage American lobster (Homarus americanus). Postlarval H. americanus have demonstrated a shift in gene regulatory response under increasing temperature treatments representative of end-century SST warming (Harrington et al., 2020). However, to our knowledge, this is the first report of the impact of changes in pCO 2 and temperature, as joint stressors, on gene expression in the early life stages Homarus americanus, and one of the few examining these two stressors on planktonic decapods as a group (Walther et al., 2010(Walther et al., , 2011Harms et al., 2014;Small et al., 2015;Waller et al., 2017).
Surprisingly, little is known about how this species will respond to end-century pCO 2 and temperatures as concurrent stressors, but it is suggested that molting through several key developmental pelagic and benthic life stages in a single season makes this species especially vulnerable to these oceanic changes (Gledhill et al., 2015;Kurihara et al., 2007;Pörtner & Farrell, 2008;Waller et al., 2017).
Physiology and metabolic demand differ between life stages, indicating stage-specific vulnerability and developmental success are closely tied to environmental conditions; as a result, few individuals survive these biological bottlenecks to settlement stage (Hines et al., 2014;Small et al., 2015;Waller et al., 2017). Thus, the final pelagic stage of this organism, the postlarvae, serves as an ideal and relevant organism for this study, since it must complete all larval ontogenetic stages within the water column to reach the stage of recruitment.

Results from physiological studies on larval and early juvenile
Homarus congeners suggest a range of responses to end-century acidification and warming as joint stressors (Agnalt et al., 2013;Keppel et al., 2012;Menu-Courey et al., 2019;Rato et al., 2017;Ries et al., 2009;Small et al., 2015;Waller et al., 2017). Elevated temperature and pCO 2 can interact to cause changes in behavior, carapace length, carbon content, and development time; however, reports are conflicting in some cases (Small et al., 2015;Waller et al., 2017). The few studies to date that have tested thermal and pCO 2 effects together under similar experimental designs suggest that larval or postlarval physiological and behavioral response to predicted end-century warming may be greater than that of pCO 2 alone (Agnalt et al., 2013;Small et al., 2015;Waller et al., 2017).
The primary goal of this study was to examine the gene regulatory response of the postlarval Homarus americanus to end-century projected sea surface temperature and pCO 2 both as independent and joint stressors, and to better understand organismal response on a molecular level to predicted OW and OA. Based on previous studies using a similar experimental design (Small et al., 2015;Waller et al., 2017), we hypothesized that the gene regulatory response to temperature would be more pronounced than to increased pCO 2 which is consistent with the prior physiological observations of larvae and postlarvae noted above.
Utilizing next-generation sequencing, our results suggest strong direct responses to pCO 2 and greater interactive effects of pCO 2 and temperature on genes associated with immune functions and shell formation.

| Experimental design
The The experiment was designed as a full factorial with two temperatures and two levels of pCO 2 , representing four distinct treatment groups: control, elevated temperature, elevated pCO 2 , and elevated temperature and pCO 2 . Temperatures were replicated from previous larval studies on H. americanus (Waller et al., 2017). All tanks were held in a temperature-controlled room at 16°C (±0.1°C), representative of the average summer sea surface temperature during larval season in Midcoast Maine (Mackenzie, 1988;Quinn & Rochette, 2015). The elevated temperature treatment (19°C; ±0.5°C) was achieved using Hydor submersible aquarium heaters, representing an end-century increase of 3°C (IPCC, 2013). This temperature (19°C) is also just below the thermal threshold recorded for lobsters in the region, where the mortality rate increases with temperature (Mackenzie, 1988).
The two pCO 2 treatments (400 ppm and 1,200 ppm) were created by mixing pure CO 2 with CO 2 -stripped, compressed air to create predetermined concentrations of gasses (Waller et al., 2017).
Each of the four treatment combinations was maintained in triplicate for a total of 12 tanks. Salinity, temperature, and pH were monitored daily. Salinity was measured using an Oakton SALT meter, and pH and temperature were monitored using a Thermo Orion 3-star benchtop pH probe, calibrated using NIST buffers. To calculate the carbonate chemistry of the water, total pH (pH t ) was measured spectrophotometrically (Hitachi U-310 dual-beam, Hitachi, USA) using the pH-sensitive indicator dye m-cresol purple (Sigma-Aldrich) following SOP (standard operating procedure 6b: Dickson et al., 2007;  ], Ω Ar , Ω Ca ) using the CO2SYS2.1 system (Lewis et al., 1998).

| RNA extraction
RNA was extracted following the method of  and RNA-seq performed on a total of 11 animals drawn from the four treatment combinations (n = 3 in all treatments except the control 16°C, 400 ppm treatment, where n = 2). Briefly, the preserved postlarvae were individually homogenized in 1 ml Trizol (TH electric homogenizer; OMNI International), placed in a chloroform/Trizol mixture (200 µl chloroform/ml of Trizol), and incubated for 3 min at room temperature. Samples were then centrifuged at 12,000 g at 4°C (15 min), and the collected supernatant was added to an equal volume of 100% ethanol. RNA was extracted with a RNeasy kit (Qiagen) with an on-column DNasel digestion, and quantified using a NanoDrop1000 spectrophotometer (Thermo Fisher Scientific F I G U R E 1 Heat map depicting expression of the 1,108 differentially expressed genes subject to single and joint stressor treatments, relative to control treatment of 400 ppm, 16°C. This represents only 0.27% of all transcripts identified. Colors represent absolute value of log2 fold change. Red represents a downregulation relative to the control, whereas green represents upregulation. More than half of these genes (55%) were functionally annotated and subject to further analysis (Figures 2-6)

| Bioinformatics
The are based on a negative binomial distribution and are both recommended for small sample sizes, DESeq2 uses a geometric normalization method, whereas edgeR calculates a weighted mean of log-ratios for normalization (Dillies et al., 2013;Nguyen et al., 2018).
The integration of both tools in our methodology produces a more robust gene expression analysis (Nguyen et al., 2018). Although edgeR resulted in a more conservative number of statistically significant transcripts, 75.3% of those significantly F I G U R E 4 Comparison of number of annotated, statistically significant (p < .05), differentially expressed transcripts identified by DESeq2 and edgeR analysis within elevated temperature (a), elevated pCO 2 (b), and joint elevated temperature and pCO 2 treatments (c) differentially regulated transcripts were also identified with DESeq2. We found 7% of the differentially expressed DESeq2identified transcripts to be shared between both analyses for the elevated temperature treatment (Figure 4a). There were no differentially expressed transcripts within the pCO 2 treatment using edgeR ( Figure 4b). Conversely, there were 30% of the differentially expressed DESeq2 identified transcripts shared with edgeR within the joint stressor treatment, representing 71.9% of all identified differentially regulated transcripts via edgeR analysis (Figure 4c).

Analysis of annotated genes revealed that DESeq2 and edgeR
were in 100% agreement with regard to the direction of gene regulation. While DESeq2 consistently found lower differential expression,

F I G U R E 5
Response of lobster postlarval exoskeleton formation transcripts to elevated temperature and pCO 2 by DESeq2 (orange) and edgeR (green) analyses. Depicted are Log2 fold change of 28 genes of interest at elevated temperature (a), elevated pCO 2 (b), joint elevated temperature & pCO 2 and (c) relative to gene expression under control conditions. Darkened colors indicate statistically significant outcomes (p < .05) more of these transcripts were found to be statistically significant (p < .05) using this analysis (Figures 5 and 6).
Genes involved in exoskeleton formation (cuticle proteins, chitin-binding proteins, and calcification peptides) were largely downregulated in response to OW and OA, with few exceptions ( Figure 5). In treatments where only temperature was elevated relative to control conditions, six of the 28 genes showed significant downregulation and one showed significant upregulation by at least one of the statistical analysis methods (Figure 5a). In treatments where only pCO 2 was elevated, DESeq2 showed significant upregulation for cuticle protein binding molecules, chitinase, and chitinase-like proteins, but edgeR resulted in no differential regulation for any of these genes (Figure 5b). In treatments where both temperature and pCO 2 were jointly elevated, both methods indicated significant downregulation of cuticle proteins, chitin-binding proteins, and calcification-associated peptides, and upregulation F I G U R E 6 Response of lobster postlarval immune function transcripts to elevated temperature and pCO 2 by DESeq2 (orange) and edgeR (green) analyses. Depicted are Log2 fold change of 42 transcripts of interest at elevated temperature (a), elevated pCO 2 (b), joint elevated temperature and pCO 2 treatment and (c) relative to gene expression under control conditions. Darkened colors indicate statistically significant outcomes (p < .05) of cuticle protein binding molecules, chitinase, and chitinase-like proteins (Figure 5c).
With respect to the differential expression of genes associated with the immune response, in the elevated temperature treatment, postlarvae downregulated HSP83, and upregulated one transcript for hemocyanin subunit 1 according to the DESeq2 (Figure 6a). In the elevated pCO 2 treatment, DESeq2 indicated significant upregulation of HSP70, 15 transcripts related to hemocyanin and its subunits, mannose-binding units, crustin, C-type lectin, and glutathione S-transferase, but edgeR detected no statistically significant (p < .05) change in these transcripts (Figure 6b). In the joint stressor treatment, DESeq2 indicated postlarvae upregulated more transcripts related to hemocyanin and its subunits, octopamine receptor, in addition to the genes differentially regulated within the elevated pCO 2 only treatment, whereas for edgeR, only 11 transcripts within this category were significantly differentially expressed (Figure 6c).

| D ISCUSS I ON
To our knowledge, this is the first study to examine how the joint The effects of elevated pCO 2 and temperature on the transcription level associated with cuticle formation and calcification were treatment-dependent. Elevated temperature resulted in downregulation of cuticle proteins and a calcification-associated peptide. Calcification-associated proteins act as acidic protein sites for nucleation of CaCO 3 during larval biomineralization (Addadi & Weiner, 1985;Faircloth & Shafer, 2007). Downregulation indicates a temperature cost to calcification, though fewer genes were differentially regulated than in other treatments. Warming has been shown to reduce shell integrity in the mussel Mytilus edulis when food limited, an effect attributed to a reallocation of energy away from biomineralization in order to address temperature-related increases in maintenance requirements (MacKenzie et al., 2014).
Under elevated pCO 2 , postlarval lobster upregulated chitinase and cuticle proteins associated with calcium binding, both of which are important components of exoskeleton development and remodeling. This result is consistent with previous findings in which larval lobster upregulate chitinase synthesis genes during ontogeny, presumably to support increased chitin synthesis and maintenance (Cohen, 2010;Hines et al., 2014). Juvenile lobsters also increase shell calcification under elevated pCO 2 conditions (Ries et al., 2009;Whitely, 2011). The combination of end-century temperature and pCO 2 has an additive effect on exoskeleton forming genes causing additional cuticle protein downregulation. These findings mirror other studies that have examined skeletal formation in marine invertebrates when exposed to elevated temperature and acidity. Downregulation of calcification-related genes was reported for pearl oysters (Pinctada fucata) after being exposed to end-century conditions (Liu et al., 2012). Larvae of the purple sea urchin (Strongylocentrotus purpuratus), however, exhibit inhibited skeletal growth under elevated pCO 2 , but not at elevated temperatures (Padilla-Gamino et al., 2013). Nevertheless, bivalve mollusks and urchins appear to have lower capacity to compensate for these elevated pCO 2 than the relatively small set of crustaceans evaluated to date (Kurihara, 2008;Wood et al., 2008).
Heat shock proteins (HSP) were downregulated in lobster larvae under elevated temperature alone, but were upregulated when exposed to elevated pCO 2, and an even greater upregulation when pCO 2 and temperature were elevated simultaneously. Heat shock proteins are molecular chaperones that are upregulated after exposure to stressful conditions to prevent improper folding or denaturation of proteins (Alberts et al., 2015;Flaherty et al., 1990;Kiang & Tsokos, 1998). Virtually, all organisms upregulate HSP expression as a method to alleviate physiologically stressful conditions (Evans & Hofmann, 2012). Thus, HSPs can modify an organism's thermal sensitivity and act as important biological stress markers (Tedeschi et al., 2015). When reared in elevated pCO 2 , or joint stressor condi- When larvae were exposed to 1,200 ppm pCO 2 at ambient temperatures, we observed the upregulation of 15 transcripts related to hemocyanin, mannose-binding proteins, crustin, and c-type lectin, all of which play roles in pathogen recognition and/or defense (Clark & Greenwood, 2016). These same genes were differentially expressed in greater numbers in the high temperature/high pCO 2 treatment, indicating an overall higher energy input to immune function when both stressors are present. Differential expression of transcripts coding for antilipopolysaccharide factors (ALFs) and their isoforms can indicate individual pathogens through differential expressions (Beale et al., 2008;, but none of these factors were identified as differentially regulated among any of our treatments, indicating that transcripts were nonspecific to known lobster pathogens, and therefore likely a response to treatment conditions. The effect of elevated temperature and pCO 2 on nonspecific immune response could have implications on the H. americanus antigen defense systems in future oceanic conditions.
Crustaceans, as a group, have shown relative resistance to end-century ocean acidification compared to other calcifying organisms (Whitely, 2011). This may be the result of a heightened capacity for ionoregulation, though the energetic trade-offs. The ramifications of this strategy, particularly with respect to suboptimal food conditions, are still not well understood (Gledhill et al., 2015;Wernberg et al., 2012;Whitely, 2011 Understanding the physiological and genetic responses to environmental change is critical to anticipate the effect of warming and acidification on lobster, and the information is needed to improve our ability to predict economic repercussions of climate change on the most valuable single-species fishery in North America. Characterizing the gene regulatory responses, in particular, can provide a mechanistic understanding of how the vulnerable stages of this species adapt to a rapidly changing environment. Pairing these techniques with whole-organism physiological and ecological studies will deepen our understanding and ability to anticipate response to environmental changes. As the Gulf of Maine continues to be one of the most rapidly warming coastal areas of the world, the American lobster stands as an icon for the urgency to understand how ocean change is impacting our living marine resources. This study exposed the larval and postlarval American lobster stages to a future climate scenario during the entirety of their planktonic development, and measured gene regulatory response within its final stage. The design of this study is similar in rearing methodology, treatment conditions, and exposure time to earlier studies that looked at the physiological response of the whole animal (e.g., growth rates, respiration rates, and swimming speed ;Small et al., 2015;Waller et al., 2017). The results of those studies generally found little effect of OA relative to responses to temperature.
In contrast, the results of this experiment investigated response on gene expression, and found a greater differential expression of genes under OA stress than temperature stress, alone, and an interactive effect when both stressors were present. The results of this experiment provide compelling evidence that compensatory mechanisms at the cellular level minimize the physiological and morphological changes measured in other studies to pCO 2 effects (Small et al., 2015;Waller et al., 2017) and suggests that postlarvae are more responsive to predicted end-century levels of elevated pCO 2 than previously assumed. The metabolic cost of these compensatory mechanisms is unknown. Experiments using limiting food conditions or sensitive physiological measurements to determine metabolic cost are warranted.
Lacking, though, is a broader understanding of how rapidly populations may be able to adapt to changing conditions. Comparisons of gene regulation among subpopulations along geographic environmental gradients may reveal population-level differences that could provide new insight into the process of local adaptation. In addition, multigenerational studies would provide a better understanding of the long-term effects, compensatory ability, and potential for adaptation to warming and acidification.