Collection and rearing of the host animal (Metopograpsus thukuhar Owen, 1839) had been described previously (Hsiao et al., 2016). Adult animals with a carapace width larger than 2 cm were used for the experiments described in the present study. Animals that are assigned as ‘non-parasitized’ are those not bearing any externa and tested negative by a polymerase chain reaction (PCR) amplifying the parasite 18S rRNA gene in the host thoracic ganglia; those tested positive by the 18S rRNA gene-PCR, but not bearing any externa, are assigned as ‘internally parasitized’ (Hsiao et al., 2016). Animals that bear externae on the abdomen are assigned as ‘externally parasitized’. To minimize possible interactions among the externae on the same hosts, data described in this study were those derived from the hosts bearing one single externa only.

To monitor changes in the color of the externa and to determine the mean duration time of each developmental stage (see Figure 1 ), experimental animals were individually reared and checked at least twice a day for external changes of the externa and (for internally parasitized hosts) emergence of the externa. The duration time (in day) each developmental stage of the externa lasts is, for the Creamy stage, defined as from the day of emergence (day 1) to the day before reaching the Orange stage, and for the other stages (i.e., Orange, Yellow, and Brown stages), the day reaching the stage to the day before entering the next stage. As externae emerged from internally parasitized hosts reared individually can only develop to the Orange stage but fail to develop beyond that stage, the mean duration time for the Orange, Yellow, and Brown stages were determined using host animals already bearing externa when collected from the field.

Animals were anesthetized with cold seawater before experimental procedures and sampling. Hemolymph was withdrawn from the anesthetized animals using syringes coupled to a 29-gauge needle, diluted 1:1 (V:V) with an anti-coagulant (0.01 M EDTA in Pantin’s crab saline, pH 7.6, Pantin, 1934), centrifuged at 16,000g for 20 min at 4°C. The resulting supernatant was collected, mixed 1:1 (V:V) with an extraction buffer (0.02M Tris-HCl, pH 7.6) containing a cocktail of serine and cysteine protease inhibitors (cOmplete™, Roche), and stored at -80°C until used for protein gel electrophoresis. Other tissues, including the gonad, thoracic ganglia, hepatopancreas, and if present, the externae, were promptly dissected out under stereoscopes (E24, Leica) using fine iris scissors and forceps (Fine Science Tools Inc.).

Depending on the purposes of the experiment, tissues were fixed in 10% neutral formalin at room temperature (for histological processing) or preserved in RNAlater (Invitrogen) at -80°C (for quantification of gene expression) until subsequent processing, or immediately homogenized in the extraction buffer containing proteinase inhibitors, and centrifuged at 16,000g for 20 min at 4°C, followed by collection of the supernatants and preservation at -80°C until used for protein gel electrophoresis.

In addition, to calculate the relative tissue weight (the tissue somatic index), host animals and freshly dissected gonad and hepatopancreas were weighed (AB104-S, Mettler Toledor). The tissue weight is expressed as a percentage of body weight, i.e., the tissue somatic index = (tissue weight in g/body weight in g) X 100.

Animals used for histological study were field-collected animals reared and checked as described in 2.1. Externa and host tissues were dissected from animals with desired stages of externa development (see below). Tissue processing, sectioning, and staining were performed at the Rapid Science Co. Ltd (Taiwan). Briefly, tissues were fixed in 10% neutral formalin for 48 hours, dehydrated through an alcohol series (70, 80, 95, 100%), cleared through a xylene series (25, 50, 75, 100%) using an automated vacuum processor (ASP300 S, Leica), embedded in paraffin (HistoCore Arcadia, Leica), sectioned longitudinally at 3μm (Biocut, Leica), de-paraffinated, rehydrated, stained with hematoxylin/eosin Y, and coverslipped using a workstation including Autostainer XL ST5010, Transfer Station TS5015, and Coverslipper CV5030 (Leica).

Histological observations and measurements were done and imagines taken under light microscopes (DM 500, Leica) with a software platform Application Suite X (Leica). Histological observations described in the present study were based on the examination of serial tissue sections derived from at least 5 animals/developmental stage (see Figure 1 ), chosen from various portions of the stage. For the measurement of the diameter of parasite’s oocytes in the externae at the beginning of each developmental stage, the diameter of 100 most developmentally advanced oocytes in randomly selected visual fields/animal was measured and the mean oocyte diameter calculated.

A reverse transcription-PCR assay was used to examine vitellogenin (Vg) gene expression in the hepatopancreas, ovary, or both in the parasitized and non-parasitized crabs. Total RNA was extracted (Direct-zol RNA Miniprep Kits, Zymo Research) from the tissues and reverse transcribed (HiScript I™ First Strand cDNA Synthesis Kit, Bionovas) according to the manufacturers′ protocols. A primer pair (Met-vg-qF1and Met-vg-qR1) was designed to amplify a 136–bp DNA fragment of the host Vg (accession number OP556484); another pair of primers (Met-beta-actin-F and Met-beta-actin-R) was designed to amplify a 193–bp DNA fragment of the host actin, which serves as the internal control (accession number OP556483) ( Supplementary Table 1 ). PCR reactions contained 0.5 μl of the reverse transcription (RT) reaction, 0.2 mM each of dNTPs, 0.2μM each of primers, 1.25 U of GoTaq^® G2 Flexi DNA Polymerase (Promega), and 1X reaction buffer (10 mM Tris–HCl, pH 9.0, 50 mM KCl, 0.1% Triton X-100, 2 mM MgCl2). The final volume was adjusted to 25μl with sterile distilled water. PCRs were performed using a DNA Thermal Cycler (MJ Research) under the following parameters: an initial denaturation (5 min, 94°C), 30 cycles of denaturation (30 s, 94°C), annealing (30 s, 54°C), and extension (30 s, 72°C), followed by a final extension (3 min, 72°C). After PCR, an aliquot of the reaction was separated on a 2.0% agarose gel and visualized with GelStar^® (Cambrex). Reagents and procedures used for purification and auto-sequencing of the PCR products, as well as software resources for sequence analyses, had been described by Chen et al. (2007).

Vitellogenin gene expression in host tissues (the hepatopancreas and ovary) and in the externa was quantified using reverse transcription real-time PCR assays. RNA extraction and treatment, and reverse transcription reaction were performed on individual tissues as described above. For real-time PCR, cDNA templates were amplified in a 20-μl reaction using a commercially available kit (QuantiNova^® SYBR^® Green PCR Kit, QIAGEN) according to the manufacture’s protocol on a Rotor-Gene Q thermal cycler (QIAGEN). Reactions were performed under the following conditions: an initial denaturation (10 min, 94°C), 50 cycles of denaturation (10 s, 94°C), annealing (7 s, 59°C), and extension (16 s, 72°C); the rate of temperature change was 20°C/s. The primers pairs for the host Vg and actin are the same as given above. The primers pair for the parasite Vg is Sac-vg-F1/Sac-vg-R1, which was designed to amplify a 176–bp DNA fragment of the parasite Vg (accession number OP556485), whereas the primer pair for the parasite actin is EBA68842F3/EBA68842R4 designed to amplify a 238–bp DNA fragment of the parasite actin (accession number OP556486) ( Supplementary Table 1 ).

PCR efficiency and specificity were validated as previously described (Chen et al., 2007). Briefly, sample titration curve (log dilution factor vs. threshold cycle number, Ct) for each amplicon was plotted and its slope used for calculating amplification efficiency according to Ginzinger (2002). All primers used amplified at least 90%. To ensure amplification specificity, melting curve analysis was performed from 58 to 95°C after completion of each PCR. The comparative threshold cycle method, also known as 2^{-ΔΔ C t} method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008), was used to determine transcript levels of the target genes (Vg). Briefly, transcript levels of target genes (Vg) were normalized to the corresponding endogenous reference genes (actin) according to the following equation: X_N (normalized expression levels of target genes) = 2^{-Δ C t}; ΔC _t = (C _t,t - C _t,r), where C _t,t and C _t,r represent C _t of target and reference genes, respectively. Values of X_N were used for statistical analysis. Further, the normalized levels of target genes in each stage were expressed relative to those in a calibrator, an arbitrarily designated stage, using the equation: X_N,C (normalized and calibrated expression levels) = 2^{-ΔΔ C t}; ΔΔCt = ΔC _t,a – ΔC _t,b where ΔC _t,a and ΔC _t,b represents X_N of individual stage and calibrator stage, respectively.

Proteins in the hemolymph and ovary of host crabs and in the externa of the parasite were separated using polyacrylamide gel electrophoresis (PAGE) performed under non-dissociating conditions (Hames, 1990). Aliquots of the samples were removed and assayed for protein content using a Bio-Rad Protein Assay kit (#500-0006) and bovine serum albumin (BSA) as standard. For electrophoresis, aliquots of the samples were mixed with the sample buffer (0.125 M Tris, 20% glycerol, 0.002% bromophenol blue) and 200 μg total proteins/sample were loaded onto the wells of the gels. For the stacking gel, the total monomer concentration (%T) was 4.0%, and the ratio of crosslinker to monomer (%C) was 2.6%. Corresponding values for the separating gel were 7.0% and 2.6%. Electrophoresis was conducted on 8 x 7-cm slab gels at constant voltage (100 V, 0.025 M Tris-0.192 M glycine, pH 8.3, Mighty Small Vertical Slab Unit, Hoefer). Proteins were stained using Coomassie Brilliant Blue R-250 (Bio-Rad) (Hames, 1990).

Protein bands predicted to be the hemolymph Vg (VgH) of the host, and the yolk protein (vitellin, Vn) of the host (VnH) and the parasite (VnP) (see Supplementary Data Sheet 1 ) were excised from stained gels and subjected to mass spectrometric analysis at the Mithra Biotechnology Inc. Briefly, the excised gels were reduced, alkylated, and trypsin-digested as previously described (Chang et al., 2010). The digested samples were subjected to mass spectrometric analysis using a liquid chromatography-mass spectrometry system that includes a Q Exactive mass spectrometer coupled to an Ultimate 3000 RSLCnano system (Thermo Scientific) with a C18 column (Acclaim PepMap RSLC, 75 μm x 150 mm, 2 μm, 100 Å). Full MS scan was performed with the range of m/z 300-2,000, and the selected ions from MS scan were subjected to fragmentation to obtain MS/MS spectra for peptide sequencing. The resulting set of the sequences of the tryptic peptide fragments from each protein samples were blasted against the deduced Vg amino acid sequence of the Chinese mitten crab Eriocheir sinensis (accession number: MH621336), that of the parasite (P. planus) Vg (accession number: OP556485), or both, using the function of LOCAL BLAST provided by the BioEdit (v7.0.3.5) with the expectation value set at 1. The matching fragments (the hits) were then mapped onto the vitellogenin, which the fragments were blasted against.

Data are expressed as mean values ± S.D. Differences in the measured parameters among hosts of various parasitization status or among externae of various developmental stages were analyzed using one-way ANOVA (with post hoc Tukey’s pairwise comparison) (SigmaStat v. 3.5). Correlation analysis of Vg gene expression levels in the externa and host tissues was performed using the Pearson’s Product-Moment Correlation (SigmaStat v. 3.5).

Development of the externa, the female reproductive organ of parasitic barnacles, is described for Polyascus planus that parasitizes its brachyuran host, the thukuhar shore-crab Metopograpsus thukuhar. It is divided into four stages, according to the developmental characteristics (see descriptions below) and named based on the distinct coloration of the externa, i.e., the Creamy, Orange, Yellow, and Brown stages ( Figure 1 ), with the mean duration time in each stage being 8.6 ± 3.9 (n = 5), 10.5 ± 4.2 (n = 6), 10.8 ± 1.3 (n = 8), 3.9 ± 0.4 (n = 8) days, respectively. The mean diameter of the developmentally most advanced germ cells in the parasite ovaries at the beginning of each stage of the externa development is, from Creamy to Brown stage, 22.6 ± 3.4, 20.0 ± 3.6, 121.4 ± 20.0, 153.1 ± 22.1 μm (n= 3 each stage), with the diameter significantly enlarges for 6 times during the transition from the Orange to Yellow stage (p < 0.001).

Histological sections of a virgin externa are shown in Figure 2 . Its mantle opening, situated anteriorly of the tissue, is still obstructed by cuticular plugs. The nascent male receptacle, opposite to the mantle opening, is composed exclusively of female cells, without sign of penetration of male larvae. A pair of receptacle ducts provide passages between the male receptacle and male cavity ( Figure 2A ). Inside the visceral mass, the ovary is composed of ovarian lobules and inter-lobular tissues, both of which radiate from the region where the colleteric gland is located ( Figures 2A, B ). Enclosed within the lobules are oogonia with a basophilic nucleus and large white cytoplasm ( Figure 2C ). The inter-lobular tissues are mainly packed with muscle cells that are characterized by having an elongated nucleus and eosinophilic cytoplasm; follicular cells, with a round nucleus and scanty cytoplasm, are also present in inter-lobular tissues ( Figure 2C ). Narrow lumen of the atrium of the colleteric glands is lined by a single layer of epithelial cells ( Figure 2B ).

Very rapidly, within less than 2 days of emergence, as externa continues to develop, its color turns creamy. In this stage, vitellogenic oocytes, the cytoplasm of which begins accumulating eosinophilic yolk bodies, are visible inside the ovarian lobes; the muscle cells become elongated and their cytoplasm now intensely eosinophilic ( Figure 3A ). Penetration of male larvae into the male receptacles is already visible ( Figures 3B, C ). Colleteric glands begin to develop, with a central atrium and branching tubules lined with secretory epithelia ( Figure 3D ).

The Orange stage is characterized by marked histological changes that ultimately leads to full maturation of the externa toward the end of the stage. Initially, while vitellogenic oocytes continue to develop with bright orange yolk granules accumulating in intensely hematoxylin-stained cytoplasm, the most significant developmental event is an expansion of the interlobular tissues, with the muscle cells became more elongated and larger ( Figure 4A ). By the time the development passes halfway the stage, there is an abrupt and significant increase in the size of the externa, when the oocytes become fully mature with the interlobular tissues wrapping around individual oocytes ( Figure 4B ). Spermatogenesis is actively taking place with dividing spermatocytes lining the lumen of the male receptacle ( Figure 4C ). The atrium of the colleteric glands now has a large lumen lined with secretion ( Figure 4D ).

The transition to Yellow stage from the late Orange stage is rather rapid, with the color of the externa turning into distinctly bright yellow usually within a day. Fully mature oocytes ( Figure 5A ) become slightly larger than those in the late Orange stage (cf. Figure 4B ). Spermatozoa are now present in the wide lumen of the male receptacle ( Figure 5B ). Colleteric glands are also fully active with a thick basophilic epithelium and eosinophilic secretion lining the lumens of the atrium and the branching tubules ( Figure 5C ). Subsequently, oocytes entering the atrium of the colleteric gland could be seen ( Figure 5C ), followed by the presence of ova in the mantle cavity enclosed by ovisacs and fertilization taking place as balls of sperms are attached to the ovisacs ( Figures 5D, E ).

As embryos within the mantle cavity develop, color of the externa gradually turns brown. When the externa become uniformly dark brown, the mantle cavity is filled with developing embryos, with the eyespots, muscle tissues, and appendages now clearly differentiated ( Figures 5F, G ). Release of naupillar larvae begins, through a raised mantle opening ( Figure 5H ), usually within 1-2 days after externa becomes dark brown. Typically, the release lasts 3 to 5 days, with a single daily surge, during which more than 80% of the larvae of a brood was released.

Color of the externa gradually loses its brown hue with the release of larvae and becomes bright yellow again immediately after the daily surge of larva release ( Figure 5I ). The mantle cavity cuticle that lines the entire cavity molts shortly after completion of larval release, and the exuviae is extruded ( Figure 5J ) via the mantle opening. Shortly after the cuticle extrusion, a new round of oviposition begins, filling the mantle cavity in the following 1 – 2 days ( Figure 5K ), which is followed by fertilization, embryo development, and color changes (turning dark brown) as mentioned above. For each externa, multiple (usually 3-4) broods of larvae could be produced, and therefore, the externa cycles accordingly between the Yellow and Brown stages, before it becomes spent and eventually detached from the host (see Figure 1 ). After the spent externa is detached, molt cycle of the host is apparently re-initiated so that the crabs would usually molt 1 month or so after the detachment and emergence of the new externa occurred concurrently with the ecdysis of the host.

Effects of parasitism of P. planus on the development of the gonad and hepatopancreas of Metopograpsus thukuhar were examined. The gonadosomatic index indicates that the relative ovarian weight, but not the relative testicular weight, is significantly different among hosts of various parasitization status (i.e., non-parasitized hosts and hosts with externa of various developmental stages) (Ovary: F(4, 10) = 12.856, p < 0.001; Testis: F(4, 10) = 0.608, p = 0.666). The relative ovarian weight of parasitized hosts is significantly lower than that of non-parasitized animals ( Table 1 ). The hepatopancreas-somatic index is not significantly different among hosts of various parasitization status, irrespective of the sex of the host (Female: F(4, 10) = 1.894, p = 0.188; Male: F(4, 10) = 1.772, p = 0.205) ( Table 1 ).

Histological examination reveals that the ovary is heavily infiltrated by rootlets. Additionally, while the color of the ovary of mature parasitized hosts (i.e., black) appears similar to that of non-parasitized counterparts, histological examination indicates that most oocytes are arrested at the pre-vitellogenic stage. Degenerated oocytes are commonly observed ( Figure 6A ). In contrast, examination of the mature ovaries of non-parasitized crabs shows that the ovaries contain large and fully developed oocytes ( Figure 6B ).

Testicular sections on the other hand shows that spermatozoa in parasitized hosts are as abundant as in non-parasitized crabs. However, there are significant amounts of secretion in the testes of parasitized hosts, which are not observed in the testes of non-parasitized hosts ( Figures 6C, D ). Rootlets have never been observed infiltrating the testicular tissues of the parasitized hosts.

The hepatopancreas of both sexes of hosts is heavily infiltrated by rootlets, which appear in the inter-tubular tissues. Nevertheless, tissue integrity of the infiltrated hepatopancreas appears intact, as with the hepatopancreas of the non-parasitized animals, with typical cell types of the tubular epithelia of the hepatopancreas ( Figures 6E, F ).

Expression of Vg gene was examined in the ovary, hepatopancreas, or both in non-parasitized and parasitized crabs using a primer pair specific for host Vg. A DNA band of expected size (136 bp) was amplified in the hepatopancreas and ovary of non-parasitized females, as well as those of parasitized females ( Figure 7 ). Similarly, the 136-bp band was amplified in the hepatopancreas of parasitized males, but not in that of non-parasitized males ( Figure 7 ).

Vitellogenin gene expression in the hepatopancreas of female and male hosts with externa of different developmental stages was quantified. The expression levels are not significantly different among parasitization status in either male (F(2, 8) = 3.913, p = 0.065) or female (F(3, 12) = 1.112, p = 0.383) hosts ( Figures 8A, B ). Nor are the expression levels of Vg gene in the ovary of female hosts significantly different among parasitization status (F(2, 6) = 1.210, p = 0.362) ( Figure 8C ).

Expression of the parasite Vg gene in the externa of P. planus was also quantified. Data show that there is no statistical difference in the parasite Vg expression among the various developmental stages of the externa on male (F(2, 8) = 1.744, p = 0.235) or female (F(3, 12) = 0.489, p = 0.697) hosts ( Figure 9 ).

The expression levels of Vg in the hepatopancreas of male hosts are negatively correlated with those in the externa ( Figure 10A , correlation coefficient r = -0.595; p = 0.024), whereas the levels in the hepatopancreas and ovary of female hosts are not significantly correlated with those in the externa ( Figures 10B, C ; B: r = -0.012; p = 0.964; C: r = -0.107, p = 0.785).

Hemolymph Vg of the host (VgH), and the major yolk proteins of the host (VnH) and the parasite (VnP) isolated from tissues using non-dissociating electrophoresis ( Supplementary Data Sheet 1 ) were subjected to tryptic digestion and the tryptic fragments to mass spectrometric analysis for de novo sequencing. When the sequences of the tryptic fragments experimentally derived from these proteins are blasted against the deduced Vg sequence of the Chinese mitten crab Eriocheir sinensis (a brachyuran species that is phylogenetically close to the host species and its whose full-length Vg sequence is available), the number of hits (fragments that match in sequence to the E. sinensis) is 6, 11, and 11 for VgH, VnH, and VnP, respectively. Schematic representation of the matching fragments mapped onto the E. sinensis Vg is shown in Figure 11A . Tryptic fragments derived from VnP are also blasted against the deduced Vg sequence of the parasite, resulting in a total of 4 matching fragments ( Figure 11B ). No matching fragments are found when the fragments derived from VgH and VnH are blasted against the parasite Vg.

Sequences of the matching fragments are presented in Supplementary Table 2 , which shows that VnP and VgH share 4 identical tryptic fragments and that VnP and VnH 9 identical fragments (also see Figure 11A ). The 4 fragments mapped onto the deduced sequence of the parasite Vg ( Figure 11B ) are unique to VnP, not found in VgH or VnH (see Supplementary Table 2 ).