Introduction

Reef-building corals (Scleractinia) are declining due to heat stress caused by global warming1. When corals are under heat stress, the symbiotic relationship with symbiotic dinoflagellates (family Symbiodiniaceae) breaks down and result in coral death1,2 Acropora is one of the most abundant genera in the Indo-Pacific region, and includes more than 100 species3,4. Acropora species are vulnerable to heat stress; for example, almost all colonies of Acropora except for juveniles at the study site in Okinawa, Japan, could not survive the 1998 mass-bleaching event5.

Knowledge of what biological responses occur in corals under environmental stress can help us to predict what will happen to corals in the future as a result of environmental changes. Information on molecules associated with biological responses to environmental stress is important for uncovering the biological mechanisms of stress responses in corals.

Heat shock proteins (HPS) are molecular chaperones that protect cells by stabilizing the folding of damaged proteins6. Sacsin belongs to the HSP40 (40 kDa molecular mass) family that acts as the HSP70 (70 kDa molecular mass) family of co-chaperones7,8 whereby some of Hsp40s may recruit an Hsp70 to a specific set of substrates9.

In corals, several cases of the stress responses of sacsin have been reported. Under experimentally elevated temperatures, sacsin was present in Pocillopora acuta, although it was absent in the controls10. A sacsin-like gene was highly expressed at branches compared with branch tips in A. palmata, may be a response to high concentrations of reactive oxygen species (ROS) in the symbiont-rich regions of branches11. Exposure to an immunogen (polyinosinic:polycytidylic acid) affects sacsin-like gene expression in A. millepora12.

Genes related to local adaptation of corals may have been maintained over long periods of time because dispersing larvae may experience strong selection in each generation when they settle in a heterogenous environment with different stress factors13. Indeed, a genome scan for local adaptation by balancing selection in A. millepora pointed to a region containing a sacsin-like gene (annotated as Amillepora11972)13. The origin of two divergent haplogroups of this sacsin-like gene was expected to predate the divergence of A. digitifera and A. tenuis, because sacsin-like genes in A. digitifera and A. tenuis genome assembly were related to a single haplogroup13. However, since only sequences belonging to one haplogroup each were analyzed from these two species (A. digitifera and A. tenuis), it was not clear whether the haplogroups of the sacsin-like gene have persisted in only A. millepora or also in other lineages in the genus Acropora. According to heat shock protein function and balancing selection, two divergent haplogroups in a sacsin-like gene may also be associated with heat response in natural populations of A. millepora13. If a sacsin-like gene has played a role in local adaptation, the two divergent haplogroups may have persisted in lineages of Acropora by balancing selection, because the larvae of other members of this genus also disperse and settle on different environments with favorable haplogroups.

In this study, we analyzed nucleotide sequence of the sacsin-like gene from two species from the genus Acropora (A. digitifera and A. tenuis), and two species from the genus Montipora (M. informis and M. aequituberculata) which is a sister clade of Acropora, and a genomic region including a sacsin-like gene from A. digitifera. We revealed that two haplogroups of the sacsin-like gene exist in Acropora and these haplogroups have persisted before the divergence of the genera Acropora and Montipora (119 Ma14).

Methods

Verification of coding sequence (CDS) of a sacsin-like gene

The genomic assembly version 2.01 of A. millepora13 was downloaded from the web site (https://przeworskilab.com/acropora-millepora-genome/). RNA-seq reads of one A. millepora (accession: SRR2086157) were mapped (similarity fraction > 0.8, length fraction > 0.9) to the genome assembly using CLC Genomics Workbench (https://www.qiagenbioinformatics.com/). Mapping results were checked manualy and the coding sequence (CDS) of the sacsin-like gene was determined. To identify conserved domains in the sacsin-like gene, the CDS was translated to amino acids and used as the query for an homology search by Phyre2 (Protein Homology/analogY Recognition Engine V 2.0)15.

Sliding window analysis of nucleotide diversity (Pi) among A. digitifera in a genomic region containing the sacsin-like gene

The publicly available genomic DNA reads of A. digitifera collected from the southern Ryukyu Archipelago located in southwestern Japan were used16. In total, 33 individuals with read mapping coverage over 9.5 × were selected. The accession numbers for genomic DNA reads are shown in Table S1. Sequence reads were mapped to the A. millepora genome assembly version 2.01 using CLC Genomics Workbench with the same mapping conditions mentioned above. Consensus sequences of a region containing the sacsin-like gene (chr7: 13,304,000–13,335,000) were extracted from each of all individuals using CLC Genomics Workbench with the following condition: minimum nucleotide count = 10; Noise threshold = 0.3. To remove a repetitive sequence region (chr7: 13,313,503–13,318,235) from the following analysis, these consensus sequences were divided into two parts, one upstream and one downstream, of the repetitive sequence. The consensus sequences were divided into alleles by a perl script (see Supplementary Material: 2 Description of the script) to import the sequences into DnaSP v517. Nucleotide diversity (Pi) was calculated with sliding window analysis (window length = 100 bp and step size = 25 bp) using DnaSP v517.

Identification and cloning of sacsin-like gene sequences

Genomic DNA was extracted from three A. digitifera and two A. tenuis individuals from our previous study18,19. Genomic DNA from single colonies of M. informis and M. aequituberculata were provided by Dr. Masayuki Hatta. Genomic DNA from each individual was used for templates of polymerase chain reaction (PCR).

A locus (chr 7: 13,325,506–13,326,780) containing the 1 kb region that was used for the phylogenetic analysis of the sacsin-like gene in a previous study13 was selected for sequence determination. Genomic sequences from the same Acropora individuals and from a M. aequiturberculata, M. efflorescens, M. spumosa, and a M. cactus (see Table S1) were used to design conserved primers (Table S2).

Approximately 1.2 kb was amplified using PrimeSTAR GXL DNA Polymerase (Takara, Shiga, Japan) with primers R10iF and R10i2R. The PCR conditions were as follows: denaturation step for 1 min 30 s at 93 °C, followed by 30 cycles of denaturation for 30 s at 93 °C, annealing for 30 s at 55 °C, and extension for 1 min 30 s at 72 °C.

All PCR products were cloned into pMD20 vector (Takara) using an In-Fusion HD Cloning Kit (Takara), and the sequences were verified using an Applied Biosystems Automated 3130xl Sequencer (Applied Biosystems). The PCR amplification and the PCR product cloning into a vector were performed once for each sample. To remove PCR errors, when the identical nucleotide sequences we determined from more than three independent clones, we used the sequences in our study. The nucleotide sequences were deposited in GenBank under accession numbers LC598237-LC598242 and LC598244-LC598248.

Determination of sacsin-like gene sequences from short reads of A. millepora

Genomic DNA reads of 12 A. millepora were downloaded from SRA and mapped to the genomic assembly version 2.01 of A. millepora13 with the same mapping conditions mentioned above. Mapped reads to the 1.2 kb locus used for PCR amplification were checked manually to select homozygous individuals (homozygous at the 111 sites that separate two haplogroups) of the haplogroups at this locus, and two consensus sequences were extracted from each of two individuals (HH16 and CS11) using CLC Genomics Workbench with the same conditions described above.

Examination of inversion on upstream and downstream sequence of the sacsin-like gene

To verify whether there is an inversion including the sacsin-like gene, upstream and downstream sequences of the genes were examined using the following PCR analysis. Genomic DNA of a heterozygous individual of haplogroups 1 and 2 (Adig3) of A. digitifera was used as a template for PCR. We hypothesized inversion boundaries on genomic regions where the nucleotide diversity (Pi) becomes low and designed primer sets upstream and downstream of the hypothetical inversion boundaries, respectively. These sets can amplify the target sequence when the template genomic DNA does not contain an inversion of the sacsin-like gene (Fig. S2a–c, primers in upper panels). If the template genomic DNA contains an inversion of the sacsin-like gene, the PCR product will be amplified by the primer set for amplification of the inverted region (Fig. S2a–c, primers in lower panels). Four primer sets are capable of amplifying the target sequence when the template genomic DNA does not contain an inversion of the sacsin-like gene, and six primer sets are capable of amplifying the target sequence when the template genomic DNA does not contain an inversion of the sacsin-like gene were used for PCR analysis (Table S3–4). To check the quality of genomic DNA, a primer set (34639_F1/34639_R1)19 was used to amplify elongation factor 1-alpha: LOC107334639 as a positive control. The PCR conditions were as follows: denaturation step for 1 min 30 s at 93 °C, followed by 35 cycles of denaturation for 30 s at 93 °C, annealing for 3 min (PCR#1,2,8 and a positive control) or 8 min (PCR#3–7,9 and 10) at 55 °C, and extension for 10 min at 72 °C. The results of PCR were confirmed by the electrophoresis patterns of PCR products.

Construction of phylogentic tree

An ortholog of the sacsin-like gene in the genus Astreopora was used for the outgroup. Astreopora locates as a sister clade of the clade that includes Acropora and Montipora14. An ortholog of the sacsin-like gene was searched by tblastx search20 using the CDS of the A. millepora sacsin-like gene as a query against all CDSs of Astreopora myriophthalma genome (GenBank: BLFK00000000.1). A CDS that was a top hit of the tblastx search20 was used as a query of a tblastx search against all CDSs of A. millepora genome to test the reciprocal best hit of the sacsin-like genes in A. myriophthalma and A. millepora.

Partial sequences of the sacsin-like gene were aligned with the sacsin-like gene sequence of A. myriophthalma (astr_s0055.g81.t1) using ClustalW in MEGA ver.721. A phylogenetic tree was constructed using the Maximum Likelihood method based on the Tamura 3-parameter model22 with 1000 bootstrap replications.

Results

First, we verified the positions of the CDS of the sacsin-like gene in the A. millepora genome using mapping RNA-seq data. Results indicated that the sacsin-like gene is composed of three exons on chromosome 7 (Fig. S1a).

In mapping this region, we observed four indels that did not match the mapping results of the short reads (Fig. S2b). We corrected these sites when we translated the CDS to amino acid sequence. The translated amino acid sequence of sacsin-like protein contained a conserved domain that is structurally related (> 90% confident in Phyre2) to the nucleotide-binding domain of Hsp90, a J domain and a HEPN domain (higher eukaryotes and prokaryotes nucleotide-binding domain).

In a previous study, the elevation of nucleotide diversity (Pi) in the sacsin-like gene compared with surrounding regions among A. millepora was reported13. To test whether such elevation of nucleotide diversity (Pi) is also observed among other Acropora species, we calculated Pi among A. digitifera in a region containing the sacsin-like gene. As in A. millepora, A.digitifera showed an increase in Pi between the Hsp90 and J domains in the sacsin-like gene (chr. 7; 13,325–13,330 Kb) compared to the surrounding region (Fig. 1), indicating diverged haplogroups of this region exist in A.digitifera.

Figure 1
figure 1

Gene structure and nucleotide diversity (Pi) of the sacsin-like gene. The predicted CDS structure of the sacsin-like gene on chromosome 7 is shown, as well as the flanking downstream gene region. The positions of conserved domains in the sacsin-like gene are shown under the CDS structure. Values of nucleotide diversity estimated in sliding windows of 100 bp with a step-size of 25 bp are shown. The bar labeled “Inversion Test Region” indicates the region where the PCR was performed for the inversion test. The bar “Tree” indicates the region used for the construction of the phylogenetic tree.

Full size image

We cloned and sequenced approximately 1.2 kb partial sequences of the sacsin-like gene from A. digitifera (n = 3), A. tenuis (n = 2), M. aequituberculata (n = 1), and M. informis (n = 1). One sequence was determined from a single individual of A. digitifera, A. tenuis and M. aequituberculata, and two different sequences were determined from two A. digitifera individuals, a single individual of A. tenuis and M. informis, respectively (Table S5). These results suggest that this sacsin-like gene is a single copy gene in these four species as with A. millepora13.

For A. millepora, we extracted a consensus sequence of the sacsin-like gene using NGS data of A. millepora, because we do not have genomic DNA of A. millepora. From the 12 samples (Table S1) used in the previous study13, three (HH16, TR02, and FR20) were predicted to have one homozygous haplotype of the sacsin-like gene, and one sample (CS11) was predicted to have the other homozygous haplotype. Consensus sequences were extracted from two individuals (HH16 and CS11) with different haplotypes.

We constructed a phylogenetic tree using 11 cloned sequences, two consensus sequences, and one sequence of the sacsin-like gene from the A. myriophthalma genome assembly as an outgroup. Sequences of the sacsin-like gene from Acropora species formed two clades (Fig. 2) and we refer to these two clades as Haplogroup 1 and 2.

Figure 2
figure 2

Phylogenetic tree of partial sequences of the sacsin-like gene.

Full size image

All sacsin-like gene sequences isolated from Montipora formed one clade and this clade clustered with Haplogroup 1 (Fig. 2). This suggests that two haplogroups split before the divergence of Acropora and Montipora (119 Ma14). No sequence from Haplogroup 2 was found in Montipora species.

Phylogenetic gene tree based on 1251 bp of a sacsin-like gene region. The tree was constructed using the Maximum Likelihood method based on the Tamura 3-parameter model. Bootstrap probabilities obtained by 1000 replicates shown next to each node. The scale bar represents 0.05 substitutions per site.

Chromosomal inversions are a common mechanism to avoid recombination and to maintain diverged alleles23,24. We tested for inversions within the sacsin-like genes using PCR. All primer sets for amplification of the sacsin locus without inversions amplified PCR products (Table S3, Fig. S3). In contrast, the primer sets for amplification of the locus with inversions either did not amplify PCR products or amplified PCR products of many different lengths, inferred to be non-specific amplification (Table S4, Fig. S3). These results suggest that, at least in the genomic regions tested in this study, two sacsin haplogroups do not have chromosomal inversions.

Discussion

Acropora corals produce dispersing larvae that may experience strong selection in each generation from stressors operating in their settled environments13. Alleles related to stress response in different environment may have been maintained over long periods by balancing selection. In this study, we showed that two divergent sacsin-like haplogroups persist in at least three Acropora species. Although we analyzed two Montipora species, no sequence from Haplogroup 2 was found in Montipora species. This may be due to two possibilities: either the individuals of Montipora species used for PCR analysis were homozygous for Haplogroup 1, or Haplogroup 2 is lost in Montipora species.

High temperature experiments with P. acuta suggested that sacsin may be involved in the high temperature stress response10. Sacsin is a member of the HSP40s family, and some Hsp40s may recruit an Hsp70 to a specific set of substrates9. The sacsin-like gene of A. millepora contained the nucleotide-binding domain of Hsp90, a J domain and a HEPN domain (higher eukaryotes and prokaryotes nucleotide-binding domain).This is consistent with the structure of human sacsin, except that human sacsin proteins have a ubiquitin domain25. Two divergent sequences of the sacsin-like gene may be involved in the recruitment of Hsp70 to different substrates under stressful conditions such as changes in sea water temperature. The ancestors of corals have experienced periods of high seawater temperatures26,27 during which two divergent haplogroups of a sacsin-like gene have persisted with some potentially adaptive functional role in different habitats (possibly different water temperatures) to play.

Chromosomal inversion can help maintain two divergent haplogroups. For example, alleles of a 4.5-Mb region which control different morphs in the ruff (Philomachus pugnax) have been maintained for more than 3.8 Myr and these alleles are located on an inverted region23. However, at least in the genomic regions tested in this study, two sacsin haplogroups do not have such inversion. As reported in A. millepora13, the sacsin-like gene may have been maintained by selection to respond to different environmental stressors. Although the genetic mechanism by which diverged haplogroups of sacsin-like genes is maintained remains unknown, the long-term persistence of haplogroups suggests that sacsin-like gene diversity may function in local adaptation to heterogeneous environments with different heat stress factors. The divergent haplotypes of the sacsin-like gene in Acropora corals will facilitate our understanding of the stress response of corals to environmental change.