The ABC transporter ABCB1 plays an important role in the disposition of xenobiotics. Embryos of most species express high levels of this transporter in early development as a protective mechanism, but its native substrates are not known. Here, we used larvae of the sea urchin Strongylocentrotus purpuratus to characterize the early life expression and role of Sp-ABCB1a, a homolog of ABCB1. The results indicate that while Sp-ABCB1a is initially expressed ubiquitously, it becomes enriched in the developing gut. Using optimized CRISPR/Cas9 gene editing methods to achieve high editing efficiency in the F0 generation, we generated ABCB1a crispant embryos with significantly reduced transporter efflux activity. When infected with the opportunistic pathogen Vibrio diazotrophicus, Sp-ABCB1a crispant larvae demonstrated significantly stronger gut inflammation, immunocyte migration and cytokine Sp-IL-17 induction, as compared with infected control larvae. The results suggest an ancestral function of ABCB1 in host–microbial interactions, with implications for the survival of invertebrate larvae in the marine microbial environment.
Successful development depends on the precise regulation of small molecule movement across cell membranes. This is especially important for free-living ‘orphan’ embryos and larvae, which lack systemic circulation or specialized organs for xenobiotic metabolism, yet develop in direct contact with bioactive small molecules in their environment (Bard, 2000; Hamdoun and Epel, 2007). The marine ecosystem is particularly rich in these compounds, which are produced by diverse marine phytoplankton and bacteria (Turner and Tester, 1997; Pohnert et al., 2002; Adolph, 2004; Steele et al., 2011; Calle, 2017). How planktonic organisms adapt to these microbial small molecules is important to understand in light of rapid environmental change (Walworth et al., 2020).
An ancient and conserved suite of small molecule transporters (SMTs) are known to act as master regulators of small-molecule signaling and protection (Bard, 2000; Dean and Annilo, 2005; Nigam, 2015; Rosenthal et al., 2019). These include the ATP-binding cassette (ABC) transporters, which handle a variety of substrates. One ABC transporter of interest is ABCB1 [aka P-glycoprotein (P-gp), encoded by MDR1]. ABCB1 was first described for its role in multiple drug resistance in cell lines (Danø, 1973; Juliano and Ling, 1976; Gros et al., 1986). ABCB1 is a plasma membrane transporter, with broad substrate specificity, thus functioning to eliminate a diverse range of hydrophobic endobiotics and xenobiotics (Szakács et al., 2006; Aller et al., 2009).
Homologs of ABCB1 are expressed in embryos and larvae of many species including zebrafish (Fischer et al., 2013), zebra mussels (Faria et al., 2010), oysters (Shi et al., 2015), water fleas (Campos et al., 2014) and sea urchins (Shipp and Hamdoun, 2012). While this protein has been demonstrated to protect early embryos against anthropogenic toxicants (Bošnjak et al., 2009; Fischer et al., 2013; Hamdoun et al., 2004), its natural substrates are not well established. Uncovering these endogenous ligands is important for understanding the evolution of this protein.
In the marine environment, a major source of bioactive small molecules is derived from bacteria that can be commensal, symbiotic or pathogenic to their animal hosts (Piel, 2009; Wilson et al., 2014; Calle, 2017; Smith et al., 2018). A number of prior observations support the hypothesis that ABCB1 may play an important protective role in host–microbe interactions in marine animals. Marine bacteria produce brominated compounds that are structurally similar to anthropogenic pollutant congeners known to bind mammalian ABCB1 (Teuten, 2005; Agarwal et al., 2014; Nicklisch et al., 2016; Le et al., 2020). Early embryos of the marine worm Urechis caupo, which develop in sediments rich in bacteria and metabolic byproducts, express higher levels of ABCB1 than those found in drug-resistant cancer cell lines (Toomey and Epel, 1993; Toomey et al., 1996). In later life stages of marine invertebrates, the intestinal epithelium becomes the tissue most exposed to microbial products (McFall-Ngai, 2002; McFall-Ngai et al., 2013). However, the function of ABC transporters remains underexplored in this context.
The goal of this study was to investigate the role of ABCB1 in host–microbe interactions in larvae of the purple sea urchin (Strongylocentrotus purpuratus). The sea urchin homolog of ABCB1/P-gp, Sp-ABCB1a, is an apically localized transporter that is expressed at relatively high levels throughout development (Gökirmak et al., 2012; Shipp and Hamdoun, 2012). In the early embryo, Sp-ABCB1a can reduce the accumulation of cytotoxic small molecules such as vinblastine (Hamdoun et al., 2004). The role of Sp-ABCB1a against endogenous marine or microbial-derived substrates in larvae is unknown.
Strongylocentrotuspurpuratus larvae develop guts that contain three distinct functional compartments by 72 h post-fertilization (hpf) (Annunziata et al., 2019). Larvae can begin feeding at 3 days post-fertilization (dpf). Exposure to opportunistic pathogens, such as Vibrio diazotrophicus, causes robust and reproducible gut inflammation (Ho et al., 2016). During the course of infection, immunocytes called pigment cells migrate to the inflamed gut, and coordinated molecular responses occur across the animal. This exposure assay has been adapted to model several aspects of host–microbe interactions (Buckley et al., 2017; Schuh et al., 2020). Importantly, transcript levels of ABCB1 increase in Vibrio-exposed larvae (Buckley et al., 2017; Ho et al., 2016), suggesting that this protein may be important for host defenses.
Here, we targeted Sp-ABCB1a by optimized CRISPR/Cas9 mutagenesis and challenged larvae with V. diazotrophicus. Notably, ABCB1 transporter knockouts in other animals do not exhibit obvious morphological phenotypes unless challenged with drugs or other toxicants (Schinkel et al., 1994). Therefore, we developed a methodology to first screen for the effectiveness of synthetic single guide RNAs (sgRNAs) to Sp-ABCB1a in the F0 generation (referred to as crispants), using bioinformatic tools and live-animal transporter substrate assays. We next exposed validated crispants to V. diazotrophicus, and found that crispant larvae exhibit significantly higher levels of inflammation than infected wild-type larvae. Our results shed light on potential bacterial substrates of ABCB1 and support an ancestral role for transporters in host–pathogen interactions. This work suggest that ABCB1 may have important implications for the adaptation of marine invertebrate larvae to changing microbial environments.
MATERIALS AND METHODS
Animals and culturing
Purple sea urchins, Strongylocentrotus purpuratus (Stimpson 1857), were collected off the coast of San Diego, CA, USA, and kept at 14°C in flow-through seawater tanks. Sea urchin embryos were reared in 0.22 µm-filtered local sea-water (FSW) as previously described (Hamdoun et al., 2004). For larval-stage experiments, cultures were diluted to 3–5 larvae ml−1 FSW at 3 dpf and cultured in suspension with stirring paddles rotating at 15 rpm. Starting at 5 dpf, larvae were fed Rhodomonas lens at 3000 algal cells ml−1 every 2 days.
Probe generation and whole-mount in situ hybridization
Primers used to amplify whole-mount in situ hybridization probe cDNAs are listed in Table S1. Primer pairs contain a T7 promoter site incorporated into either the forward (for sense probe transcription) or reverse primer (for antisense probe transcription). A probe template of ∼1.4 kb was amplified from ABC transporter plasmids with the full-length coding sequence of Sp-ABCB1a (GenBank ID JQ390048; Addgene ID: 34939; Gökirmak et al., 2012) using Phusion high-fidelity DNA polymerase (NEB, Ipswich, MA, USA) and purified by PCR-column purification (Qiagen, Germantown, MD, USA). RNA antisense probes were transcribed with T7 synthesis and digoxigenin (DIG) labeling kits (Roche, Indianapolis, IN, USA) according to manufacturer protocols. Our Sp-ABCB1a probe spans coding sequence in exons 2–12, starting 15 nucleotides downstream from the first translated ATG.
Whole-mount in situ hybridization was performed as previously described (Shipp and Hamdoun, 2012) with minor modifications. Briefly, embryos were hybridized in hybridization buffer [50% formamide, 5× saline-sodium citrate (SSC), 5 mmol l−1 EDTA, 0.1% Tween-20, 2× Denhardt's solution, 50 µg ml−1 heparin, 500 µg ml−1 yeast tRNA] overnight at 62°C, at a final probe concentration of 0.5–1 ng µl−1. Samples were washed and then incubated with anti-DIG-AP antibody (Roche, Indianapolis, IN, USA) overnight at 4°C for alkaline phosphatase staining the following day. Stained samples were imaged on an AxioImager M2 microscope (Zeiss) using a 20×, 0.9 NA objective and color camera.
Synthetic sgRNA design and microinjection
Target sites were identified in sea urchin transporters using a commercial CRISPR design tool (Synthego, Redwood City, CA, USA) with the S. purpuratus reference genome (v3.1.38). In order to maximize the likelihood of disrupting the encoded protein, sites were selected to target an N-terminal region of coding sequence or the nucleotide binding domain (NBD) of each transporter, which are required for ATP-driven active transport. The four target sites were checked for potential off-target sites by performing BLAST analysis against the sea urchin genome. The Sp-nodal sgRNA was synthesized from published sequence (Lin and Su, 2016). For enhanced stability, all sgRNAs were synthesized with 2′-O-methyl 3′ phosphorothioate modifications in the first and last three nucleotides (Hendel et al., 2015). All sgRNAs were synthesized by Synthego (Table 1): 1.5 nmol of each sgRNA was resuspended in 15 µl nuclease-free water and stored in aliquots at −80°C.
Cas9 mRNA was synthesized from a Streptococcuspyogenes pCS2-nCas9n plasmid with the Thermo Fisher mMESSAGE mMACHINE SP6 Transcription Kit according to the manufacturer's protocol. pCS2-nCas9n was a gift from Wenbiao Chen (Jao et al., 2013) (Addgene plasmid 47929, RRID:Addgene_47929). For microinjections of one-cell sea urchin embryos, each sgRNA was mixed with capped nls-Cas9-nls mRNA at final concentrations of 150 and 750 ng µl−1, respectively. These were the optimal concentrations as determined by Lin and Su (2016).
Sp-ABCB1a MASO design and morphant analysis
An ATG-blocking Sp-ABCB1a morpholino antisense oligonucleotide (MASO) was designed in the 5′ UTR region of Sp-ABCB1a, 33 base pairs upstream of the translation start site (Table S1). The Sp-ABCB1a MASO and a negative control MASO (Table S1) were microinjected into one-cell sea urchin embryos at a concentration of 125 ng µl−1.
Genomic DNA extraction and PCR
To determine Cas9 cutting efficiency, genomic DNA (gDNA) was extracted from control (wild-type, WT) and injected embryos at 48 hpf. Briefly, 50–100 embryos were pipetted into 100 µl of homogenization buffer (100 mmol l−1 Tris-Cl pH 8.5, 200 mmol l−1 NaCl, 0.2% SDS, 5 mmol l−1 EDTA, 100 µg ml−1 proteinase K) and incubated at 55°C for 2 h. Proteinase K was heat inactivated at 95°C for 10 min before the addition of 1 µl of RNase A (10 µg ml−1) and incubating at 37°C for 30 min. Samples were then mixed 1:1 with a phenol:chloroform:isoamyl alcohol (25:24:1) solution, mixed, vortexed and centrifuged to phase separate gDNA in the aqueous phase. The extraction was repeated once and then the aqueous phase was precipitated overnight in 1/10th volume of 3 mol l−1 sodium acetate solution (pH 5.2) and 2.5 volumes of 100% ethanol. Following overnight precipitation, gDNA was centrifuged, washed once with 70% ethanol, resuspended in nuclease-free water and stored at −20°C until use. Target regions were amplified using PrimeSTAR® Max DNA Polymerase (Takara Bio USA, Mountain View, CA, USA) and gene-specific primers (Table S1). In preparation for Sanger sequencing, PCR amplicons were re-amplified with M13-tailed primers (Table S1).
T7 endonuclease I assay of CRISPR-induced genomic mutations
T7 Endonuclease I (NEB) was used to detect the genomic mutations created by CRISPR/Cas9 mutagenesis as per the manufacturer's protocol, with minor modifications. Briefly, 500 ng of PCR-amplified genomic fragments were denatured and re-annealed in T7E1 buffer (50 mmol l−1 NaCl, 10 mmol l−1 Tris-HCl, 10 mmol l−1 MgCl2, 1 mmol l−1 DTT, pH 7.9 at 25°C) at a total reaction volume of 19.5 µl, and incubated with 5 U of T7E1 for 25 min at 37°C. The enzymatic reaction was terminated using 1.5 µl 0.25 mol l−1 EDTA and run on a 2% agarose gel to resolve the cleaved products. T7E1 negative reactions included the same quantity of re-annealed PCR amplicons and underwent the same protocol but without addition of the enzyme. PCR primers are listed in Table S1.
Inference of CRISPR Edits analysis of mutation frequency
PCR-amplified fragments encompassing sgRNA target sites in both controls and crispants were Sanger sequenced using M13 sequencing primers (Eurofins Genomics, Louisville, KY, USA; Table S1). Raw sequencing (.ab1) files were analyzed using Synthego's Inference of CRISPR Edits (ICE), a commercial bioinformatic tool, similar to TIDE (Tracking of Indels by DEcomposition), that analyzes the discordance between WT and crispant sequences near the sgRNA target site (Hsiau et al., 2019 preprint). ICE predicts the Cas9 cutting efficiency and mutant alleles present in genetically modified samples.
Assays of ABCB1 efflux transport activity
ABC transporter activity was assessed by efflux assays using fluorescent substrates and inhibitors as previously described (Gökirmak et al., 2012). PSC833, calcein-AM (CAM) and Bodipy-Verapamil (Bver) were purchased from Sigma-Aldrich (St Louis, MO, USA), Biotium (Fremont, CA, USA) and Invitrogen (Waltham, MA, USA), respectively. All substrates and inhibitors were rehydrated in dimethyl sulfoxide (DMSO) and stored at −20°C. All stocks were diluted such that the final DMSO concentration did not exceed 0.5%.
Activity of transporters was assessed in sea urchin embryos at the late gastrula stage, approximately 48 hpf, or early larval stage (72 hpf). Injected crispants/morphants and un-injected embryos were incubated with CAM and Bver at final concentrations of 250 and 50 nmol l−1, respectively, in FSW at 14°C for 60 min. Embryos incubated with Bver were washed 3 times with FSW after incubation, to remove Bver remaining in the water, and immediately imaged. For inhibitor assays, embryos were incubated in 1 µmol l−1 PSC833 along with substrates. Embryos were imaged on a Zeiss LSM 700 confocal microscope using a 20× Plan-Apo, 0.8 NA air objective. All embryos exposed to the same substrate were imaged with identical confocal settings (pinhole size, gain, laser power, zoom and scan speed).
Average intracellular fluorescence intensity was measured using Fiji (Schindelin et al., 2012). For each treatment, at least 10 embryos per injection were imaged from at least three independent batches of embryos. Within each batch, fluorescence intensity was measured for WT, 1 µmol l−1 PSC833-inhibited and Cas9/sgRNA-injected or morpholino-injected embryos. Incubation with 1 µmol l−1 PSC833 inhibits most ABCB1-like efflux activity (Hamdoun et al., 2004); therefore, fluorescence intensity of WT and crispant embryos was compared with that of PCS833 inhibited embryos. Raw fluorescence intensity values of each individual crispant or WT embryo were divided by the mean of the raw intensity values of the PSC833-inhibited embryos to calculate the percentage substrate accumulation relative to inhibited controls. Then, the percentage substrate accumulation of each individual crispant was divided by the mean percentage substrate accumulation of the WT embryos to calculate fold-changes in accumulation.
To quantify SpABCB1a's contribution to substrate efflux, the mean fluorescence intensity of WT embryos was subtracted from all raw fluorescence intensity values of SpABCB1a crispants and PSC833-inhibited embryos. The relative contribution of SpABCB1a in CAM and Bver efflux was then calculated by comparing the fluorescence intensity of crispants with that of the PSC833-inhibited embryos.
Microbial challenge assay
To test the effects of targeted mutagenesis of Sp-ABCB1a on larval gut physiology and immune responses, 10 dpf Sp-ABCB1a sgRNA E13-22 crispants and sibling-matched control (WT) larvae were exposed to Vibrio diazotrophicus (Guerinot et al., 1982). Efficient ABCB1a CRISPR was validated by T7E1 and CAM efflux assays in all batches before proceeding with microbial challenges. Vibrio cultures were grown overnight to log phase in Marine Broth 2216 (Difco, BD Sciences) at 15°C, 250 rpm. Bacteria were washed 3 times in FSW and counted in a Petroff–Hauser counting chamber. Larvae were exposed to 107 ml−1Vibrio in FSW or control FSW alone, as in Ho et al. (2016).
At 24 and 48 hours post-exposure (hpe), a subset of treated larvae was randomly chosen for analysis from master cultures of n>200 larvae. Larvae were immobilized under a coverslip anchored with double-sided tape (0.1 mm thick, Scotch 665, 3M, Maplewood, MN, USA). Larvae were imaged on an inverted Zeiss LSM 700 using a 20× Plan-Apo 0.8 NA air objective, and scored for pigment cell migration as in Ho et al. (2016), and midgut (stomach) epithelial phenotypes as in Schuh et al. (2020). Briefly, any pigment cell that had completely emerged out of the epithelium, was in transit through the blastocoel or was localized at the gut was scored as migratory. Midgut epithelial diameters were measured in Fiji (Schindelin et al., 2012) in a Z-slice where the largest midgut cross-section was in sharp focus. The mean number of migrated pigment cells and gut epithelial diameter were calculated for each condition. Each individual experiment was performed with sibling larvae from a single batch. Six independent batches were pooled for statistical analysis.
Quantitative real-time PCR (qPCR) of immune genes
Total RNA was extracted from 10 dpf WT and ABCB1a E13-22 crispant larvae at 0, 6, 12, 24 and 48 hpe to V. diazotrophicus. For this, 20–60 larvae were collected per time point directly into lysis buffer for RNA purification and column-based DNase treatment with RNeasy Micro columns (Qiagen), as per the manufacturer's protocol. Two other independent batches of 20–30 larvae were extracted with Trizol (Invitrogen) followed by RNAeasy Micro column clean up. All RNA samples were eluted in 15 µl nuclease-free water. First-strand cDNA was synthesized from 50 ng of each RNA sample by using anchored 1.5 µmol l−1 oligo-dT primers (Integrated DNA Technologies, Coralville, IA, USA), 0.5 mmol l−1 dNTPs (Fermentas, Glen Burnie, MD, USA), 200 U M-MLV Reverse Transcriptase (Promega, Madison, WI, USA) and 25 U RNasin (Promega) at final volume of 25 µl. All cDNA samples were diluted 1:5 with nuclease-free water and used for qPCR. Measurements were made on duplicate samples using an AriaMx Real Time PCR system (Agilent Technologies) with EVA QPCR SuperMix kit reactions (Biochain), as previously described (Shipp and Hamdoun, 2012). Gene expression levels were normalized to 18S rRNA measurements. Relative gene expression data are reported as fold-differences between uninfected larvae (0 hpe) and Vibrio-exposed larvae at 6, 12, 24 and 48 hpe time points, calculated from six internal replicates from three pooled independent batches. Sp-IL17-1, Sp-IL17-4 (Buckley et al., 2017) and Sp-18S primers are listed in Table S1.
The Shapiro–Wilks test was used to determine whether data were normally distributed (Razali and Wah, 2011). CRISPR cutting efficiency data, as determined by ICE, were normally distributed and statistical significance was determined by running an unpaired, two-tailed Student's t-test. Transporter efflux activity assay data were not normally distributed; therefore, statistical significance was determined by running a two-tailed Mann–Whitney U-test. Gut and pigment cell data from microbial infection assays were not normally distributed and thus were analyzed by one-way ANOVA, and differences between groups were compared by Kruskal–Wallis and post hoc Dunn's tests with a significance threshold of P≤0.05. For qPCR data, Student t-tests or Mann–Whitney tests were performed to determine significance between conditions at each time point (P≤0.05). Error bars in all figures represent the standard error of the mean.
Sea urchin ABCB1a expression is enriched in gut tissues of larvae
To define the spatial patterns of Sp-ABCB1a expression, and thus its potential locations of action within the embryo, we first localized its expression. Consistent with prior analyses (Shipp and Hamdoun, 2012; Tu et al., 2014), we found that Sp-ABCB1a transcripts were present in the unfertilized egg and remained ubiquitous throughout the blastula stages (Fig. 1A–C). By the gastrula stage, the gut tube is morphologically uniform but fated to form three future digestive compartments (Annunziata et al., 2014): the foregut (esophagus), the midgut (stomach; which is the primary site for digestion and absorption in sea urchin larvae; Burke, 1981; Annunziata et al., 2019), and the hindgut (intestine). Near the end of gastrulation, Sp-ABCB1a remains expressed in the ectoderm, but becomes enriched in the hindgut and midgut (Fig. 1D). By larval stages, Sp-ABCB1a expression is mostly enriched in the stomach and intestine (Fig. 1E,F).
Efficiency of CRISPR/Cas9 mutagenesis in crispants verified using T7E1 and ICE
We next targeted Sp-ABCB1a with CRISPR/Cas9. A challenge to using this approach in the F0 generation is the potential for confounding effects, such as inefficient cutting or mosaicism (Burger et al., 2016; Hoshijima et al., 2019). Furthermore, the knockout of protective genes such as ABCB1a may not produce conspicuous morphological phenotypes until the animals are challenged (Schinkel et al., 1994), making it essential to streamline the effectiveness of the CRISPR-mediated perturbation prior to downstream challenges.
Therefore, to most efficiently target Sp-ABCB1a, we utilized synthetic, chemically modified sgRNAs which have been shown to be more stable than in vitro transcribed sgRNAs (Hendel et al., 2015). We used several approaches to estimate the Cas9-mediated cutting efficiency of different sgRNAs. The first, the T7E1 assay, is a biochemical assay that employs an endonuclease that recognizes and cleaves mismatches in DNA heteroduplexes (Mashal et al., 1995). While this is a longstanding method for detecting larger indels following CRISPR/Cas9 mutagenesis (Hwang et al., 2013; Ota et al., 2014; Thomas et al., 2014), it is not quantitative, and can produce false positives. Thus, we also used a Sanger sequencing-based CRISPR analysis computational tool, ICE, to fully characterize the T7E1 positive sgRNAs. ICE estimates the editing efficiency by comparing sequence degeneracy in crispant sequence reads near the sgRNA target site relative to the WT sequence reads (Hsiau et al., 2019 preprint; Ying and Beronja, 2020).
Determining the efficiency of synthetic sgRNA E1-157 to nodal
We first grounded our approach in the comparison of synthetic sgRNAs with a previously described sgRNA targeting the S. purpuratusnodal gene (sgRNA E1-157; Fig. S1A; Table 1). This sgRNA was shown to cause a radialized embryo phenotype, indicative of Sp-nodal disruption, in 76% of injected embryos (Lin and Su, 2016). Of 418 Nodal crispant embryos we generated using the synthetic version of sgRNA E1-157, roughly 81% were clearly radialized (Fig. S1B), slightly higher than the level observed by in vitro transcribed guides in Lin and Su, 2016.
Next, we analyzed these crispants by T7E1 and ICE assays to establish the baseline for efficient perturbation. T7E1 gels of ‘mixed’ embryos (i.e. containing both embryos showing the radialized phenotype and those that looked normal) clearly revealed digested products (Fig. S1C) and depletion of control bands, indicative of effective CRISPR-induced mutagenesis. When measured by ICE, 68.7±10.1% of reads from these ‘mixed’ batches of embryos contained indels (Fig. S1D). When only radialized embryos were collected and analyzed by ICE, the corresponding indel percentage increased to 90.3±3.3% (Fig. S1D).
Efficiency of sgRNAs targeting ABCB1a and ABCB1b
Using our synthetic nodal sgRNA (E1-157) results as a baseline, we next sought to determine the efficiency of synthetic sgRNAs designed to target Sp-ABCB1a. The Sp-ABCB1a gene model spans 50.5 kb (Scaffold 19, genome v5) that contains 27–29 exons (Fig. 2A). We also targeted Sp-ABCB1b, which encodes a protein that does not efflux the fluorescent small molecules transported by Sp-ABCB1a (Gökirmak et al., 2012), and thus serves as a suitable negative control. Sp-ABCB1b is approximately 63 kb (Scaffold 16, genome v5) and contains 24–25 exons (Fig. 2B).
In silico prediction tools were used to prioritize the potential target sites that should maximize efficient, on-target editing with Cas9. This yielded eight gRNAs (Table 1) that targeted either the early N-terminal domain or the first nucleotide (i.e. ATP) binding domain (NBD). T7E1 assays indicated that out of these eight guides, two sgRNAs targeting the first NBD of Sp-ABCB1a (B1a E12-82 and B1a E13-22) and one sgRNA targeting an N-terminal region of Sp-ABCB1b (B1b E4-33) induced strong Cas9-mediated endonuclease activity (Fig. 2C). The remaining sgRNAs exhibited minimal or no cutting activity (Fig. S2). ICE analysis of the three T7E1-positive sgRNAs demonstrated an average indel percentage of 72.7±14.7% for B1a E13-22, 21.5±8.8% for B1a E12-82 and 60.3±3.3% for B1b E4-33 sgRNAs (Fig. 2D). Example indels for each gene are shown in Fig. 2E. Based on these results, we prioritized B1a E13-22 and B1b E4-33 crispants for subsequent work.
ABCB1a crispants have decreased efflux activity
As expected from previous studies in ABCB1 knockout mice (Schinkel et al., 1994), both ABCB1a E13-22 and ABCB1b E4-33 crispants did not have outward morphological defects (Fig. S3). To determine whether these crispants exhibited changes in transporter activity, 72 hpf embryos were incubated with either Bver or CAM, both known fluorescent substrates of Sp-ABCB1a (Gökirmak et al., 2012), and embryos were assayed for substrate accumulation (Fig. 3A–D). The 72 hpf time point was chosen to allow for complete depletion of maternal mRNA and/or protein (Shipp and Hamdoun, 2012).
ABCB1a crispants exhibited increased accumulation of both CAM and Bver as compared with control (WT) embryos (Fig. 3A), while there were no changes in accumulation of substrates in ABCB1b crispants (Fig. 3B). ABCB1a crispants exhibited 3.0±0.3-fold and 1.9±0.1-fold increases in accumulation of CAM and Bver, respectively, which was significantly more than ABCB1b crispants (Fig. 3C,E; P<0.001). ABCB1b crispants did not exhibit significant increases in the accumulation of CAM and Bver, relative to control embryos (P>0.050).
As Sp-ABCB1a is not the only transporter known to efflux Bver and CAM from sea urchin embryos, we sought to determine the relative contribution of ABCB1a to the efflux of these substrates. We therefore compared the level of substrate accumulation in ABCB1a crispants with that of WT embryos incubated with PSC833 (Fig. 3D–F), a broad inhibitor of all P-gp transporters (Smith et al., 2000; Gökirmak et al., 2014; Morita and Terada, 2014). Crispants exhibited 23.9±2.8% and 39.3±4.6% accumulation of Bver and CAM, respectively, relative to PSC833-inhibited controls (Fig. 3F; P<0.001). There were no significant differences between WT and ABCB1a crispants when both were exposed to PSC833 (Fig. S4). However, PSC8333-inhibited crispants accumulated significantly more CAM (P<0.010) and Bver (P<0.001) than non-inhibited crispants (Fig. S4). This suggests that there exist related transporters that are non-specifically inhibited by PSC833, but are not disrupted by ABCB1a sgRNAs.
MASO-mediated knockdown of Sp-ABCB1a recapitulates B1a E13-22 crispant substrate accumulation phenotypes
To validate the efflux phenotypes generated by our CRISPR experiments, we used an Sp-ABCB1a-specific ATG-blocking MASO to knockdown expression at the translation level. Embryos were injected with either the ABCB1a MASO or a control MASO (Table S1) and incubated with Bver and CAM at the gastrula stage (Fig. 4A,B). We focused on this developmental stage to balance between the effectiveness of MASOs, which can be depleted in later development, and turnover of maternal ABCB1a protein, which occurs at the hatching blastula stage (Shipp and Hamdoun, 2012). ABCB1a morphants exhibited 4.3±0.2- and 2.5±0.2-fold increases in CAM and Bver accumulation (Fig. 4C; P<0.001) relative to controls, and consistent with the increased accumulation of both ABCB1 substrates in crispants.
Sp-ABCB1a crispants exhibit increased gut inflammation upon exposure to pathogenic Vibrio bacteria
Having established a reliable protocol for the generation of ABCB1a crispants, we next sought to determine how Sp-ABCB1a may function to protect against microbes. Given that Sp-ABCB1a is expressed strongly in feeding larval guts (Fig. 1F) and responds transcriptionally to bacterial infection, we hypothesized that Sp-ABCB1a could play a prominent role in protecting larvae from the opportunistic pathogen V. diazotrophicus. To examine this potential role, we measured two cellular parameters of the immune response following bacterial exposure (Ho et al., 2016): gut epithelial width and pigment (immune) cell migration towards the gut at 24 and 48 hpe (Fig. 5A).
At 10 dpf, and prior to pathogenic bacteria exposure (0 hpe), crispant larvae exhibited a slightly larger gut epithelial width compared with control (WT) animals (Figs 5A and 6A; P<0.0013), suggesting a possible role for ABCB1 in the efflux of algal metabolites, or for maintaining gut barrier integrity. However, there was no difference in pigment cell location between WT and crispant larvae at 10 dpf (Fig. 6B,D; P=0.1899), indicating relatively low levels of immune activation.
Next, food (R. lens) was removed and larvae were exposed to V. diazotrophicus for 48 h (Fig. 5A). Both crispants and controls exhibited the stereotypical immune response to Vibrio infection, in which the midgut epithelium thickens and pigment cells are recruited to the inflamed gut (Fig. 5B–D). However, ABCB1a crispants exhibited significantly larger midgut epithelial widths (Fig. 6A,C; P<0.0001) and increased numbers of migratory and gut-associated pigment cells (Fig. 6B,D; P<0.0001) compared with controls, at both 24 and 48 hpe. ABCB1a crispants cultured in seawater with neither Vibrio nor R. lens showed no significant increases in pigment cell migration relative to controls over the same time period (Fig. 6D; P=0.2433). Interestingly, the width of gut epithelia in these animals also returned to normal in the absence of bacteria (Figs 5E and 6C; P>0.9999), suggesting that the initial differences observed prior to Vibrio exposure were a transient response to the R. lens cultures used to feed the larvae.
Sp-ABCB1a crispants exhibit a higher level of Sp-IL-17 expression compared with controls
Given the observed responses of crispant larvae to Vibrio infection, we tested whether they also experienced amplified immune gene expression (Fig. 6E; Fig. S5). We focused on conserved IL-17 cytokines known to be activated in the gut epithelium after Vibrio exposure (Buckley et al., 2017). ABCB1a crispants infected with Vibrio exhibited higher levels of both Sp-IL-17-1 and Sp-IL-17-4 cytokine activation compared with Vibrio-infected control larvae (Fig. 6E; Fig. S5). At 10 dpf and prior to the Vibrio exposure (0 hpe), there was no significant difference in the relative expression of Sp-IL-17-1 (P=0.2073) or Sp-IL-17-4 (P=0.8182) between crispants and control larvae (Fig. S5A,F). While Sp-IL-17-1 expression peaked and attenuated quickly in the infected control larvae (Fig. S5B–E), crispant larvae exposed to Vibrio had significantly sustained Sp-IL-17-1 activation at 24 hpe (Fig. S5D; P=0.0262) and 48 hpe (Fig. 6E; Fig. S5E; P=0.0500). By 48 hpe, infected control larvae exhibited a 1.7±0.3-fold increase in Sp-IL-17-1 expression compared with un-infected control larvae, whereas expression in Vibrio-infected ABCB1 crispant larvae remained higher, with a 8.9±2.3-fold change relative to un-infected crispants (Fig. 6E; Fig. S5E).
Crispant larvae also had a 21.3±8.9-fold increase in Sp-IL-17-4 expression, compared with a 2.6±1.4-fold increase in control larvae at 48 hpe (Fig. S5J; P=0.0411), consistent with previous reports that Sp-IL-17-4 is downstream of Sp-IL-17-1 signaling (Buckley et al., 2017). In sum, these results indicate significantly enhanced activation and later dysregulation of gut epithelial cytokine signaling in the absence of Sp-ABCB1a.
CRISPR/Cas9 genome editing is quickly becoming a reliable mutagenesis technique in marine animals including sea anemones, diatoms and lampreys (Square et al., 2015; Nymark et al., 2017; Nakanishi and Martindale, 2018). A contribution of this study is to outline how CRISPR/Cas9-mediated gene editing can be optimized towards understanding the function of transporter genes in the sea urchin larva, which may not have obvious morphological phenotypes prior to xenobiotic challenge. In this case, we used a modified approach to dissect the function of Sp-ABCB1a and revealed a potential endogenous role in host–microbe interactions.
Considerations for CRISPR/Cas9 mutagenesis of ABC transporters and analysis of phenotypes in the F0 generation
An attractive feature of the sea urchin as a model organism is the ability to rapidly generate large numbers of embryos, which can easily be genetically modified to produce F0 crispants. However, CRISPR/Cas9-mediated editing has highly variable efficiency in the F0 generation, dependent on numerous factors including: the choice of physical delivery method, molecular delivery of CRISPR components, sgRNA length and sequence, cell type and organism (Zhang et al., 2016; Lino et al., 2018). In sea urchins, this technique has only been used in a handful of studies to date (Lin and Su, 2016; Shevidi et al., 2017; Liu et al., 2019; Wessel et al., 2020; Yaguchi et al., 2020), all of which have targeted genes in which a mutation confers an easily visualized, and previously established, phenotype. In order to assess more cryptic phenotypes of Sp-ABCB1a, we screened for synthetic sgRNAs with relatively high efficiency in producing indels. Sp-ABCB1a sgRNAs with over 70% cutting efficiency proved effective at reducing transport function, generating up to a 3-fold increase in ABCB1a substrate accumulation (Fig. 2).
Compared with other mutagenesis studies in the sea urchin (Lin and Su, 2016), our gene editing efficiency appeared slightly improved by the use of chemically modified synthetic sgRNAs (81% versus standard 76% visible Sp-nodal phenotype; Fig. S1), which may have greater stability than in vitro transcribed guides (Hendel et al., 2015). Previous studies indicate that sgRNAs enriched with guanine also exhibit more stability (Huppert et al., 2008; Moreno-Mateos et al., 2015). Additionally, a cytosine at the −3 position (3′ relative to the protospacer adjacent motif, PAM) and an adenine or guanine at the −1 position are often correlated with higher sgRNA activity (Doench et al., 2014; Ren et al., 2014; Wong et al., 2015; Xu et al., 2015). However, the three most efficient sgRNAs used in our study (nodal E1-157, ABCB1a E13-22, ABCB1b E4-33) conferred relatively similar efficiencies despite containing varying amounts of guanine, and did not follow the −3/−1 trend (Table 1, Fig. 2D). These results could simply be an artifact of the smaller number of sgRNAs tested (8), or suggest that parameters for predicting sgRNA stability and efficiency can vary between model systems.
A conserved role for ABCB1 in host–microbe interactions in gut epithelia
This study demonstrated that Sp-ABCB1a is enriched in the gut tissues of sea urchin embryos and larvae (Fig. 1) and that, in the absence of Sp-ABCB1a, the larval gut experienced a transient increase in baseline epithelial inflammation after feeding on R. lens (Figs 5 and 6). Further exposure to the opportunistic pathogen V. diazotrophicus in crispants led to significantly increased gut pathology, stronger recruitment of pigment cells to the inflamed gut, and an elevated pro-inflammatory Sp-IL-17 cytokine response, as compared with infected control larvae (Fig. 6; Fig. S5). These observations strongly suggest a role for ABCB1a in managing gut microbial products.
It has been hypothesized that in mammals, the expression of ABC transporters is specifically necessary to reduce epithelial exposure to harmful products of enteric bacteria (Mercado-Lubo, 2010; Cario, 2017). Intriguingly, one of the ‘unanticipated’ phenotypes in the first ABCB1 knockout mouse was the observation that mice lacking this protein in gut epithelial cells spontaneously develop inflammatory bowel diseases such as colitis (Panwala et al., 1998). Mdr1−/− mice exhibit a pronounced thickening of the mucosa and immune cell infiltration in the gut. Importantly, the pathology is entirely dependent on the presence of the commensal microbes. In turn, challenges with pathogenic bacteria, such as Helicobacter, Listeria and Salmonella, accelerate the occurrence and severity of gut inflammation (Maggio-Price et al., 2002; Neudeck et al., 2004; Siccardi et al., 2008).
Our results in the sea urchin larva mirror those observed in knockout mammals and may extend to downstream immune signaling (Figs 5 and 6; Fig. S5). In both mice and sea urchins, IL-17 genes are activated by bacteria in the gut mucosa, and an autocrine IL-17 cascade controls downstream signaling and effector genes that recruit immune cells to the gut and help repair tissue damage from inflammation (Ramirez-Carrozzi et al., 2011; Song et al., 2011; Buckley et al., 2017). However, an inability to properly control IL-17 cytokine responses can lead to even more tissue damage. IL-17C becomes aberrantly induced in dextran sulfate sodium (DSS)-induced colitis models (Ramirez-Carrozzi et al., 2011), but to our knowledge, IL-17 responses have yet to be described in mdr1−/− mice. Our results are the first to suggest that in the absence of ABCB1, IL-17 signaling can be activated at greater levels and for longer periods of time following bacterial infection (Fig. 6E; Fig. S5).
Inflammatory phenotypes in Sp-ABCB1a crispants before and after bacterial exposure
We observed a transient gut epithelial phenotype in Sp-ABCB1a crispant larvae prior to Vibrio infection (Figs 5E and 6A). This could be interpreted as a potential direct function of Sp-ABCB1a in maintaining barrier integrity de novo, which has been demonstrated in the blood–testes barrier of mammals (Su et al., 2009; Su et al., 2011). However, larval guts appear normal in earlier development stages (Fig. S3), and the initial gut inflammatory phenotype was not maintained when larvae were cultured for a further 48 h in the absence of algae or Vibrio (Fig. 6C). Therefore, we suspect that the inflammation in crispants results mostly from the bacterial infection, rather than from a direct role of Sp-ABCB1a in maintaining barrier integrity.
Lab cultures of the larval diet species R. lens are typically not axenic; thus, crispant animals may initially be responding to both xenobiotic algal compounds (Eufemia et al., 2000) and products from algae-associated bacteria (Mercado-Lubo, 2010; Dorrestein et al., 2014; Ganal-Vonarburg et al., 2020). Feeding is the primary driver of microbial colonization in sea urchin larvae in the wild and in the lab (Carrier and Reitzel, 2018; Schuh et al., 2020). Anecdotally, we have observed that crispants will survive to adulthood and reproduce under normal culturing conditions (unpublished results). This suggests that the function of ABCB1 is most critical for defense against pathogenic strains of bacteria, rather than against commensals.
Here, we hypothesize that Sp-ABCB1a transporter activity most likely aids in the elimination of Vibrio-derived compounds. A build-up of these factors in crispants may subsequently drive the trio of stronger inflammatory responses we have observed. Potential Sp-ABCB1a substrates could include a variety of molecules generated by bacteria. By 24 hpe, V. diazotrophicus will translocate out of the gut epithelium into the body cavity (Ho et al., 2016), indicating the utilization of virulence factors. These may include known zonula occludens toxins (Castillo et al., 2018a,b), and small diffusible molecules required for quorum sensing, niche establishment and virulence (Yang et al., 2011; Girard, 2019). Additional pathogenicity factors could be secreted extracellular proteases and toxins (Farto et al., 2006; Nottage and Birkbeck, 1986; Schuh et al., 2020) that are generated through Type-II secretion systems prominent in Vibrio species (Cianciotto and White, 2017). Future studies will identify and validate Sp-ABCB1a substrates derived from V. diazotrophicus and other opportunistic bacteria in sea urchin larvae.
Potential ecological implications of ABCB1-mediated host–microbe interactions
Cellular mechanisms of protection from xenobiotics often evolve in concert with their respective environment (Bard, 2000; Hamdoun and Epel, 2007; Li et al., 2007; Calla et al., 2017). The xenobiotic metabolism genes of insects have diversified and acquired novel functions in gut tissues, depending on the plant or microbe-derived substrates that are encountered in their diets (Berenbaum, 2002). In yeast, paralogs of ABC transporters rapidly evolve substrate specificity to different factors in the environment (Srikant et al., 2020). In Atlantic killifish, expanded and highly polymorphic aryl hydrocarbon receptor and cytochrome P450 gene families have enabled subpopulations of fish to adapt to highly polluted environments (Whitehead et al., 2012, 2017).
In the sea urchin, ABC transporter and related detoxification genes are likewise expanded, compared with vertebrate homologs (Goldstone et al., 2006; Hamdoun and Epel, 2007). The Sp-ABCB1a gene is also polymorphic (unpublished observations). This polymorphism could have implications for the ability of different subpopulations of larvae to handle local microbes, as these animals occupy a very wide geographic range, from the coasts of southern Alaska, USA, to Baja California, Mexico (Thompson et al., 2017; Carrier and Reitzel, 2018). Given that global warming and increased pollution are predicted to cause large-scale modulations of microbial communities in the world's oceans (Hellweger et al., 2014; Walworth et al., 2020), ABC transporters could make subpopulations of S. purpuratus more or less vulnerable to environmental microbial changes or related dysbiosis in the gut.
The question of what small molecule transporters ‘really do’ (Nigam, 2015) is one that has eluded the field for decades. While ABC transporters certainly play important roles in drug disposition (International Transporter Consortium et al., 2010), they did not evolve to make life difficult for pharmacologists. CRISPR/Cas9 genome editing has provided a possible solution to this problem, making it relatively easy to modify genomes and perturb protein function (Jinek et al., 2012). Our localization data and improved CRISPR/Cas9 methodology in the sea urchin larva have revealed a potentially conserved role for ABCB1 in microbial defense at the gut mucosa. Strongylocentrotuspurpuratus is becoming a robust model for understanding fundamental aspects of gut homeostasis and immunity at the cellular, molecular and biochemical level (Buckley and Rast, 2019; Carrier and Reitzel, 2020; Stumpp et al., 2020). Our work establishing reproducible echinoderm ABCB1 crispants will allow us to begin dissecting functional mechanisms of ABCB proteins in host–microbial settings, which are relevant for predicting how marine larvae may adapt to future ocean environments.
The authors thank Elliot Jackson, Drs Vic Vacquier and Bianca Brahamsha, and two anonymous JEB reviewers for their insightful critique and comments on the manuscript.
Conceptualization: T.F., C.S.S., H.V., H.D.R., A.H.; Methodology: T.F., C.S.S., H.V., A.H.; Software: A.H.; Validation: T.F., C.S.S., H.V.; Formal analysis: T.F., C.S.S., H.V., A.H.; Investigation: T.F., C.S.S., H.V., H.D.R., A.H.; Resources: A.H.; Data curation: T.F., C.S.S., H.V.; Writing - original draft: T.F., C.S.S.; Writing - review & editing: T.F., C.S.S., H.V., H.D.R., A.H.; Visualization: T.F., C.S.S., H.V., A.H.; Supervision: A.H.; Project administration: A.H.; Funding acquisition: C.S.S., A.H.
This work is supported by National Institutes of Health grants ES 027921 and ES 030318; National Science Foundation grant 1840844 to A.H.; and National Institutes of Health grant F32ES029843 to C.S.S. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.