The PCR primers used in this study are shown in Table 1 . To determine the presence of infectious IHHNV and to test its replication level in challenged shrimp, a long-amp IHHNV detection method was used to detect a 3665 base-region of IHHNV (approximately 92% of the whole genome and excluding its hairpin ends). A Long-Amp™ Taq PCR mix (New England Biolab, USA) was used. The PCR reaction consisted of Long-Amp Taq PCR reaction mix, 0.4 µM of forward and reverse primers (98F/3762R), 1U Long-Amp™ Taq polymerase, and either 100 ng DNA before digestion or 2 ng DNA post-enzyme digestion. The PCR cycle was started with initial denaturation at 94°C for 30 s then followed by 35 cycles of 94°C for 20 s, 55°C for 30 s, 72°C for 2.5 min and final extension at 72°C for 10 min.

Quantitative PCR by droplet digital PCR (IHHNV-ddPCR) was used to check the number of viral copies in the crude IHHNV stock and the number of IHHNV-cvcDNA in the circular DNA preparation. The ddPCR reaction was prepared by using EvaGreen™ ddPCR supermix (Bio-Rad, USA) which consisted of 1X ddPCR mix, 0.2 µM of forward and reverse primers (309NF/309NR), and either 1 µl of diluted crude viral stock (at 10^-7 dilution) or 1 ng of the circular DNA preparation as the template. The ddPCR amplification cycle was set according to the manufacturer’s protocol by adjusting the annealing temperature to 56°C. After the complete PCR cycles, the reactions were analyzed by fluorescent signal using a ddPCR plate reader. The absolute amount of target DNA copy per reaction was calculated based on Pearson’s correlation method using QuantaSoft™ ddPCR analysis software (Bio-Rad, USA). PCR reactions for each individual sample were performed in duplicate.

The short-amp IHHNV-PCR method (19) was used to check for infectious IHHNV sequences in DNA extracts and in infected shrimp. As an internal control gene for linear, chromosomal DNA, primers specific to shrimp elongation factor 1 alpha (EF-1α) gene were used to give an amplicon of 122 bp (18). PCR amplicons were analyzed by 1.5% agarose gel electrophoresis followed by visualization of ethidium bromide staining by UV light.

Frozen black tiger shrimp (Penaeus monodon) were checked for IHHNV infection using the long amp-IHHNV detection method. The pleopods from 5 IHHNV-positive shrimp were collected, pooled, homogenized and dissolved in cold 1X PBS pH 7.4. The tissue homogenate was centrifuged at 8,000 rpm to remove cell debris before it was subjected to filtration through a 0.2 µm membrane filter. The filtrate was collected and aliquoted into small tubes and referred to as “crude IHHNV stock”. A similar protocol was applied to 5 IHHNV-negative shrimp (P. monodon) samples from the same batch of frozen shrimp. The crude IHHNV stock was subsequently used in the challenge tests with P. vannamei where IHHNV infection and replication was confirmed by PCR. The crude IHHNV stock was stored at -80°C for further experiments. To check the virus titer, crude IHHNV stock was serially diluted and subjected to quantification by the IHHNV-ddPCR.

The sequence of the long-amp IHHNV amplicon in the DNA extract from the crude IHHNV stock was determined and compared with previously reported IHHNV sequences in the Genbank database using the UPGMA method and MEGA X software (https://www.megasoftware.net/). Results revealed that the crude IHHNV stock contained a type of IHHNV closely related to isolates previously reported from Thailand (AY362547 and AY102034) and Taiwan (AY355307) (see Supplementary Figure S1 ).

After receiving the NGS sequencing results for the cvcDNA preparations from IHHNV-infected shrimp, the DNA extracts and the cvcDNA preparations from the IHHNV-negative and IHHNV-positive P. monodon were tested by PCR (17) for the presence of an IHHNV-EVE corresponding to the GenBank reference sequence DQ228358.

Total DNA extract from pleopods of the IHHNV-infected P. monodon samples was subjected to circular DNA isolation as previously described (10). Briefly, total shrimp DNA was prepared using a DNA extraction kit (Qiagen, USA) and DNA concentration was determined by NanoDrop spectrophotometer (Thermo Scientific, USA). Extraction of circular DNA from total DNA was performed by enzymatic digestion of linear DNA using the plasmid-safe DNase (PS-DNase) (Lucigen^®, Epicentre, UK) that “digests linear dsDNA and, with lower efficiency, closed-circular and linear ssDNA to deoxynucleotides at slightly alkaline pH”. The circular DNA extraction protocol is described in Supplementary Figure S2 . PS-DNase digestion was carried out for 4 days with addition of fresh PS-DNase every 24 hours (4 times). This protocol was used independently with 2 sub-samples from the pooled total DNA extracted from the 5 IHHNV-positive P. monodon samples, but only once with a sub-sample of total DNA extracted from the 5 IHNNV-negative P. monodon. During the extraction process, XhoI enzyme in the protocol was used to cut shrimp chromosomal DNA into smaller fragments to accelerate DNA digestion by PS-DNase. The enzymes XhoI was chosen because it has no cutting site in shrimp mitochondrial DNA or in the IHHNV genome. This was determined by checking for no cutting sites in the P. monodon mitochondrial DNA sequence using GenBank accession no. AF217843 as a reference and restriction mapping was evaluated by the online software RestrictionMapper version 3 (http://www.restrictionmapper.org/). The lack of a cutting site was confirmed when the result in Figure 3 showed that mtDNA remained after enzyme digestion. If there were IHHNV-cvcDNA entities that contained portions of host DNA, they might be cut by this enzyme and be lost during circular DNA preparation. After digestion and extraction, the quantity of putative circular DNA was determined by Qubit fluorometer (Invitrogen, USA).

To confirm the presence of circular DNA and lack of linear chromosomal DNA in the circular DNA preparation, PCR tests were carried out using 1) elongation factor 1 alpha (EF-1α) primers PmEf-1α-F/PmEf-1α-R that yielded a 122 bp amplicon as a representative of chromosomal linear DNA and 2) a shrimp mitochondrial DNA (mtDNA) PCR detection method using primers PmmtDNA-F/PmmtDNA-R that yielded a 150 bp amplicon representing circular DNA. The presence of amplicons for 2 targets in the pre-digested DNA extract and absence of EF-1α amplicons but presence of the mtDNA amplicons in the post-digestion extract would confirm the success of circular DNA preparation. The ddPCR reaction with 309NF/309NR primers was used to quantify the IHHNV fragments in the pre-digestion DNA extract (presumed to contain non-circular IHHNV genome + IHHNV-cvcDNA) and post-digestion DNA extract (presumed to contain only IHHNV-cvcDNA).

The concentration of purified circular DNA extract from IHHNV-infected shrimp was quantified by Qubit fluorometer (Invitrogen, USA). To obtain a sufficient concentration for sequencing, the circular DNA (40 ng) was subjected to rolling circle amplification (RCA) using Repli-G midi kit (Qiagen). RCA is a technique used for multiple amplification of circular DNA to obtain sufficient amplicons for next-generation nucleotide sequencing (NGS) (20). The RCA amplified products were verified by gel electrophoresis and capillary electrophoresis to determine DNA quality before sending the DNA for sequencing using the 150 bp pair-end sequencing method (Ilumina sequencing) by Novogene Co. Ltd., Hong Kong. Random fragmentation of the RCA product was by sonication. The resulting DNA fragments were end polished, A-tailed and ligated with the full-length adapters of Illumina sequencing. This was followed by further PCR amplification with P5 and indexed P7 oligos. The PCR products for the final construction of libraries were purified using the AMPure XP system. Then libraries were checked for size distribution by Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA), and quantified by real-time PCR (to meet the criteria of 3 nM). For data analysis, the raw reads of nucleotide sequences were de novo assembled and compared to IHHNV reference genomes in databases. The 21-nt clean reads was mapped with two IHHNV reference genome sequences, i.e., DQ228358 and AF218266 by bowtie2 and then average counts for all reads from 21-nt position along the virus reference sequences were plotted. The plot of mean reads data was constructed by ggplot in R program (https://www.r-project.org/).

Primers were designed based on the cvcDNA sequencing result to prove that the annotated circular DNA sequences obtained were present in the original cvcDNA sample preparations. Since the sequences and contigs obtained from NGS were reported as linear nucleotide chains, it was necessary to confirm the presence of circular forms in the original circular-DNA extract. This was done by outward facing primers designed specifically from the 3’ and 5’ ends of the linear dsDNA fragments such that PCR amplification would produce amplicons only from a matching cvcDNA sequence, but not from a matching linear fragment. Specifically, either 3’ primer IHHNV3031F or IHHNV3766F was used together with a ring-closing reverse 5’ primer IHHNV128R ( Figure 2 ). This is a standard protocol to test for circular dsDNA in de novo sequencing of dsDNA viruses (21). Single step PCR was performed using a One-Taq™ PCR reaction kit (NEB, USA) with 30 amplification cycles and 2 ng of original circular-DNA extract as the DNA template. The general protocol for PCR was 94°C for 5 min followed by PCR for 30 cycles of 94°C for 15 s, 55°C for 15 s, 72°C for 30 s and then 72°C for 5 min. The PCR products were determined by 1.5% agarose gel electrophoresis and ethidium bromide staining. The amplicon band was cut and purified from the agarose gel before cloning into pGEM-T easy vector (Promega, USA). The plasmids were transformed into E. coli DH5α. The plasmids prepared from 8 clones (no. 1-8) were selected and subjected to digestion with EcoRI enzyme to check for the variation of inserted IHHNV fragments. Four of these products from clones no. 2, 4, 6, and 7 were sent for sequencing using T7/SP6 primer (Macrogen, Korea).

A batch of juvenile Penaeus vannamei (2-3 g body weight, n=50) was obtained from the shrimp demonstration farm in Chachoengsao province and maintained in the laboratory in a 500L tank containing artificial seawater (15 ppt salinity) for 2-days with continuous aeration and water temperature between 28-30°C. The shrimp were fed with commercial feed pellets at 5% body weight daily until starting experiments. Prior to experiments, a sub-sample of 3 arbitrarily selected shrimp was tested by PCR for the absence of IHHNV using the short-amp IHHNV detection method. This would later be compared with the negative test results expected from the negative control group shrimp at the end of the IHHNV challenge experiment. The shrimp were divided into 3 groups; the PBS injection group (negative control, n=5), the IHHNV injection group (positive control, n=5), and the test group injected with 100 ng circular DNA extract + crude IHHNV stock (n=10). The crude IHHNV stock was diluted with 1X cold PBS pH 7.4 to obtain 1 x 10⁷ copies/50 µl to inject individual shrimp in the IHHNV-injected positive control group and cvcDNA test group. In the test group, the diluted virus was mixed with the circular DNA preparation containing putative IHHNV-cvcDNA before intramuscular injection into individual shrimp, while the negative control group was injected with 50 µl PBS only. At day 5 post-injection, shrimp pleopods were collected from individual shrimp and then subjected to total DNA extraction followed by total DNA assessment by NanoDrop spectrophotometer (Thermo scientific, USA). Then, 100 ng was used as the template for PCR analysis for IHHNV replication level using the long-amp IHHNV method.

PCR intensities were determined using the Gel Doc™ EZ Gel Documentation System (Bio-Rad, USA). The relative ratio of virus level was calculated corresponding to the internal control gene expression (Ef-1α). Differences in IHHNV replication were determined by calculating the mean relative Ef-1α amplicon band intensity in the agarose gels followed by appropriate adjustment of the IHHNV band intensities before comparison of by One-Way ANOVA. Differences were considered significant at p ≤0.05. Data analyses and graph preparations were carried out using GraphPad Prism version 7.0 (https://www.graphpad.com/scientific-software/prism/).

After total DNA extracted from IHHNV-infected and non-infected shrimp (P. monodon) from the same source were exposed to PS-DNase digestion for 4 days, the EF-1α amplicon linear DNA marker could no longer be detected, in contrast to the untreated control. This indicated that all linear DNA had been digested. At the same time, the positive-control, circular-mtDNA could still be detected, indicating that circular DNA constructs could survive the exonuclease treatment ( Figure 3 ). Note that the band intensity for the circular-mtDNA amplicon was less than that from initial DNA preparation prior PS-DNase digestion. This may have resulted from the PCR conditions that used different concentrations of DNA template or to loss of DNA during the precipitation and recovery step or possibly to partial digestion of the circular DNA itself. However, this did not negate it as a circular-DNA marker.

Starting from 2 µg of total DNA extracted from IHHNV-infected shrimp (P. monodon), there remained a putative circular DNA concentration in the range of 20-40 ng in the DNA extract after 4-day digestion (a reduction of 98 to 99%). Using digital droplet PCR to measure the quantity of IHHNV in the pre-digestion and post-digestion preparations ( Figure 4 ) revealed that the average pre-digestion IHHNV quantity of 3.0 x 10⁵ copies/ng DNA had dropped to 1.7 x 10³ copies/ng DNA. This constituted a residual of approximately 0.6% of the initial IHHNV-DNA quantity after linear DNA digestion (i.e., 99.4% reduction). From a repeated digestion with a parallel sample of the same IHHNV-infected shrimp DNA extract, the final IHHNV quantity in the putative cvcDNA preparation was 1.2 x 10³ copies/ng DNA and was not significantly different from the quantity in the first preparation (Student’s T-test, p = 0.37). IHHNV could not be detected in the negative control digests obtained from IHHNV-negative shrimp. In addition, exposing the putative cvcDNA extract to the restriction enzyme HpaI specific for IHHNV (but not active for shrimp mitochondrial DNA) resulted in a negative PCR test for IHHNV but a retained positive PCR test result for mtDNA ( Figure 5 ). This supported the contention that the IHHNV-positive PCR test result arose from IHHNV-cvcDNA. In summary, the results from Figures 4 , 5 suggested that the IHHNV copies detected in the DNA extract from the PS-DNase digestion mix from IHHNV-infected P. monodon consisted of residual IHHNV in the form of cvcDNA.

Before investing in the cost of NGS sequencing, experiments were conducted to determine if the IHHNV putative cvcDNA preparation could protect shrimp against infectious IHHNV in a laboratory challenge test. The details of IHHNV inoculum preparation from our P. monodon samples and testing of its infectivity in P. vannamei are given in Supplementary Figure S3 . On day 5 post-injection in the protection test, P. vannamei from all 3 groups were collected and the DNA was extracted and checked by PCR using the long amp-IHHNV detection method with EF-1α as the internal control ( Supplementary Figure S4 ). The relative intensities of amplicon bands adjusted by the mean average of EF-1α intensities were compared ( Figure 6 ). The 5 shrimp in the negative control group injected with PBS gave no PCR amplicons for IHHNV, while the positive control group of 5 shrimp injected with IHHNV gave a mean band intensity of 1.2. In contrast, the group of 10 shrimp injected simultaneously with IHHNV and 100 ng of putative IHHNV-cvcDNA gave a mean amplicon intensity of 0.2 that was significantly lower intensity (p<0.01) than the positive control by a One-Way ANOVA test. We considered this sufficiently encouraging to proceed with NGS sequencing.

The crude cvcDNA preparation (40 ng) was subjected to rolling circle amplification (RCA) using an REPLI-g mini kit (Qiagen) followed by quantification with a Qubit fluorometer revealing a total yield of 40 μg of RCA-amplified product. Subsequent agarose gel and capillary electrophoresis revealed that the majority of the RCA amplicons were of sizes larger than 11 kb ( Supplementary Figure S5 ). After sequencing of the RCA products, total reads of approximately 8 Mb nucleotides (8,363,236 bases) were obtained based on Illumina DNA sequencing with 99% effectiveness. A low sequencing error rate (0.03%) was found, as shown in Table 2 . De novo assembly of the raw read sequences gave 81,026 contigs. A detailed analysis of the assembly is shown in Table 3 .

Diagrams of the distribution of cvcDNA sequence reads related to IHHNV reference sequences were plotted. Alignment of the frequency reads in units of 21-nt bites against GenBank records for IHHNV revealed high homology to 2 accession numbers, i.e. GenBank accession no.AF218266 and GenBank accession no. DQ228358 ( Supplementary Figure S6 ). Short read DNA assembly based on the reference IHHNV sequences was carried out and the 3 longest contigs were selected for deeper analysis. Two of these (NODE_444 and NODE_1) matched the IHHNV genome accession no. AF218266 ( Supplementary Figure S7A ) from an extant, infectious type of IHHNV. The third contig (NODE_439_3766) matched accession no. DQ228358 which contains an ancient IHHNV sequence ( Supplementary Figure S7B ) that is inserted into the genome of some P. monodon specimens. Thus, it should now be called an endogenous viral element or EVE. It was discovered before use of the term EVE and was called “non-infectious” IHHNV (17). This was a completely unexpected and unpredicted result and was thus, not included in our experimental hypothesis for this study.

NODE_444 and NODE_1 showed 98 and 99% identity, respectively, to AF218266 ( Table 4 ) and 99% sequence identity to the IHHNV in the frozen P. monodon we used for preparation of the putative cvcDNA extract. These results were consistent with our prediction that IHHNV-cvcDNA would arise after IHHNV challenge. NODE_439_3766 showed 98% sequence identity to DQ228358.

Outward facing primers 3031F/128R designed to match the ends of the linear contigs NODE_1_3463 and Node_444_3736 obtained by NGS ( Figure 7A ) gave rise to PCR amplicons, confirming that the linear contigs obtained by NGS sequencing from Table 4 arose from cvcDNA. Failure to obtain amplicons would have indicated that the target sequence was a linear DNA fragment. The PCR results gave a single PCR amplicon of approximately 1,500 bp ( Figure 7B ). This confirmed that the contigs ID NODE_1-3463 and/or NODE_444-3736 were derived from closed-circular DNA forms.

The band of the 1,500 bp amplicon was purified and cloned. The plasmids from each clone were digested with EcoRI enzyme and the digestion results are shown in Figure 7C . There were variations in the amplicon sizes among the 8 selected clones indicating a mixture of amplicons. Four selected clones were subjected to plasmid sequencing. Approximately 600-800 bp were read using forward and reverse primers and the assembled sequences (approximately 1,000-1,500 bp) matched sequences in the IHHNV reference genome AF218266. They corresponded to the expected amplified regions based on the GenBank reference genome. The ring in Figure 7D represents a model cvcDNA of variable overall length with the orange portion indicating the contigs for NODE_1_3463 or NODE 444-3736 (or similar contigs that might contain the targets for primers IHHNV3031F/IHHNV128R at each end). In other words, the amplified sequences linked the two ends of each contig to form a circle and matched the expected corresponding sequences in the reference genome ( Figure 7E ).

Next we confirmed that contig no. NODE_439_3766 that matched the GenBank record DQ228358 also arose from cvcDNA. This contig must have arisen from an EVE in the genome of the experimental shrimp used, and it contained a portion of the host shrimp retrotransposon sequence to which the EVE is linked in the shrimp genome. To close the cvcDNA circle, we followed the same protocol that was followed in Section 3.4 for ID NODE_444-3736 and NODE_1-3463.

Primers IHHNV3766F/IHHNV128R designed to match the ends of NODE_439_3766 were used ( Figure 8A ). PCR results revealed an expected amplicon size of approximately 1,200 bp ( Figure 8B ). After PCR cloning, 8 positive clones showed a variety of amplicon sizes indicating several cvcDNA types ( Figure 8C ), similar to the phenomenon previously observed for ID NODE_444-3736 and ID NODE_1-3463, described in section 3.4 above. DNA sequence reads of 6 clones were of 2 general types, one showing sequence reads containing only IHHNV portions of the DQ228358 sequence (clone no. 1, data not shown) and others (clones no.2,7) containing the DQ228358 sequence joined to its retrotransposon part ( Figures 8D, E ). Again, based on the publications linked to reference sequence accession no. DQ228358, these cvcDNA must have originated from the IHHNV-EVE called DQ228358 and its linked host-transposon sequence. This was an unexpected finding providing evidence that cvcDNA can be generated from not only cognate viruses but also from EVE in the host genome itself.

Since we made pooled DNA extracts and cvcDNA preparations from one group of IHHNV-positive and one group of IHHNV-negative specimens (P. monodon) obtained from the same batch of frozen shrimp, it was of interest to determine whether the IHHNV-negative preparations were also positive for the presence of EVE-DQ228358. Subsequent PCR testing revealed that the DNA extract and cvcDNA preparation derived from the IHHNV-negative shrimp (P. monodon) gave positive PCR test results for the DQ228358 target sequence (data not shown). We did not pursue this further by NGS sequencing, but it suggested that the IHHNV-negative shrimp (P. monodon) carried EVE-DQ228358 and that IHHNV infection was not necessary for it to produce cvcDNA from it.