Detection of recombinant breakpoint in the genome of human enterovirus E11 strain associated with a fatal nosocomial outbreak

Serum and biopsy samples were collected from four patients involved in the nosocomial outbreak: Pt0 (the index case), Pt1, Pt2 and Pt3 (Table 1). As Pt1 was the first one to develop severe symptoms and acute hepatitis after the autopsy, a liver biopsy was extracted after mechanical homogenization and underwent EBV, CMV, HSV1-2, HBV, and HHV8 DNA and Enterovirus, Flavivirus, Arenavirus and Hantavirus RNA amplification with commercial and in house real-time PCR or RT-PCR. Only the Enterovirus test resulted positive with RealTime RT-PCR (RQ ENTEROVIRUS, AB analitica, Padova, Italy). The serum of Pt2 and Pt3, and the embedded paraffin liver biopsy were analyzed for Enterovirus diagnosis with all specimens resulting positive.

Vero E6 cells (ATCC® number CRL-1586™) were maintained in Modified Eagle Medium (MEM) supplemented with 10% heat-inactivated fetal calf serum (FCS) at 37 °C in a humidified atmosphere of 5% CO₂. For viral culture assay, Vero E6 cells were infected with clinical samples of infected patients. Specifically, three Enterovirus isolates were isolated from the native liver biopsy of patient 1 (Iso_pt1_L) and from the serum of patient 2 and patient 3 (Iso_pt2_S and Iso_pt3_S respectively).

The Viral RNA was extracted from cell supernatants and clinical samples using RNeasy mini kit (QIAGEN), RNA from the embedded paraffin liver was extracted performing a pre-treatment with EZ1® RNA Tissue Mini Kit (Qiagen; Hilden, Germany) according to the manufacturer’s instructions. The RNA amplification of partial VP1 and UTR gene and full genome sequencing was performed using Qiagen OneStep RT-PCR kit (QIAGEN) according to the manufacturer’s instructions. The amplicons were sequenced using ABI prism 3130 × 1 GENETIC ANALYZER DNA SEQUENCER.

The typing was performed using the Enterovirus Genotyping Tool (https://​www.​rivm.​nl/​mpf/​typingtool/​enterovirus/​) [15]. Species assignment was obtained on the basis of the partial sequence of 5′ Untranslated Region (UTR) while VP1 partial sequence was used for serotype assignment (Fig. 1). The 5′UTR partial region was amplified using primers described by Nicholson et al. [16] while amplification of a portion of VP1 gene was obtained using primers AN88-AN89, by Nix et al. [17].

The RNA genome of the three Enterovirus isolates (Iso_pt1_L, Iso_pt2_S and Iso_pt3_S) and of the one derived from the native liver biopsy of patient 1 (Pt1_L) were amplified in 23 overlapping reverse transcriptase-PCR amplicons, using One-Step RT-PCR Kit (QIAGEN; Hilden, Germany) and sequenced with the Sanger method (oligonucleotide sequences are supplied in Additional file 1). The nucleotide sequences were edited and aligned by using Clustal W [19] and BioEdit v 7.0.5.3 software, to reconstruct the full-length RNA genomes. Sequencing of nucleic acid derived from specimens of infected patients were approved by the Institutional Ethics Committee of the National Institute for Infectious Diseases, INMI, “L. Spallanzani” (Issue n. 9/2020).

The phylogenetic analysis was carried out comparing the full-length RNA sequence of the E-11 described here, ECHO11_INMI (Accession n: KX527626.1) with 121 Enterovirus prototype sequences retrieved from the Picornavirus homepage [18], and CV-B1 sequence (Accession n: MG845887.1); the accession numbers of whole sequences are provided in Additional file 2. Three different alignment datasets were analyzed separately to highlight the phylogenetic relationships: one of the full-length genome sequences; one of the portion coding for the P1 region of the polyprotein; and the last one of the P3 region of the polyprotein. The phylogenetic tree of each alignment dataset was inferred using the Neighbor-Joining method [20], with 1000 bootstrap replicates. The rate variation among sites was modeled with a gamma distribution (shape parameter = 1). All ambiguous positions were removed for each sequence pair (pairwise deletion option). There were a total of 7620 positions in the final dataset of full genome, 2598 positions in the final dataset of P1, 345 positions in the final dataset of P3. Evolutionary analysis was performed by using MEGA X [21].

The recombination analysis was carried out by using the software Recombination Detection Program version 4 (RDP4) [22]. To identify the best sequences for the recombination detection analysis, we selected non redundant sequences between the prototype reported on the specific Picornaviridae database (http://​www.​picornaviridae.​com/​). The final data set was aligned using the MUSCLE program [23] and the alignment was manually controlled. Moreover, the closest specific sequences were added, these are identified by BLAST tool on genome, blasting the sliding windows of 2 Kb with a step of 2 kb.

RDP4 software can identify the specific positions of recombination breakpoint by automatically using multiple methods of analysis (GENECONV [24], CHIMAERA [25], Maximum Chi Square [26], BOOTSCANning [27], Sister Scanning [28], the 3SEQ [29], VisRD [30] and the BURT methods. In RDP4, we chose to check every possible point of recombination along the E11 genome, viewing a window size of 100 nt and using default parameters.

To investigate if the infection was caused by a single recombinant enterovirus variant or by a coinfection of two different serotypes, we designed two sets of primers mapping E-11 and Coxsackiesvirus of B species serotypes, upstream and downstream of the recombination site respectively. The sequences of the primers and the position are reported in Table 2.

The E-11 primers were designed comparing eleven full genome sequences of E-11 retrieved from GenBank. The CV-B specific primers were designed, using an alignment of 17 CV-B sequences retrieved from GenBank.

Three isolates were obtained from the clinical samples of the patients involved in the small outbreak. One was obtained from the hepatic biopsy of patient 1, here referred to as Iso_pt1_L, while the others were obtained from the serums of patient 2 and patient 3, named Iso_pt2_S and Iso_pt3_S respectively (see Table 3).

The Enterovirus strains involved in this small outbreak were typed, basing on VP1 partial sequences and all resulted belonging to E-11. Moreover, the phylogenetic analysis performed comparing the VP1 sequences with a set of E-11 strains retrieved from GenBank shows that our strains strictly correlated and segregated in a unique and well separated clade which belongs to the D5 genotype of E-11 according to Li et al. [31], thus confirming the epidemiological link between the Enterovirus infections described here. (Additional file 3).

To further characterize the RNA genome, the EVs isolated in cell culture (Iso_pt1_L, Iso_pt2_S and Iso_pt3_S) and the virus from the liver biopsy of patient 1 (Pt1_L) were entirely sequenced. All nucleotide sequences were then aligned using Clustal W and the amino acid sequences of the polyproteins were compared. As expected, all viral strains revealed a high identity at nucleotide level (median = 99.87%, ranging from 99.83 to 100%); two schematic tables of the nucleotides and amino acid differences respectively found, are provided in Additional file 4. Moreover, comparing the Liver Enterovirus strain (Pt1_L) with the corresponding Liver isolate (Iso_pt1_L), no amino acid differences were observed, while the sequence of Iso_pt2_S compared to Enterovirus strains of pt1, shows one amino acid substitution, C1677 L, located in the P3 region of the polyprotein. Specifically, it maps in the region encoding for the protein complex of 3BCD that is the precursor of the non-structural proteins 3B (Vpg primers for RNA transcription), 3C (protease) and 3D (viral polymerase). Similarly, we compared the amino acid sequence of Iso_pt3_S with those of pt1 (Iso_pt1_L and pt1_L) and we found the following four substitutions: D478G localized in the P1 portion of the polyprotein encoding for capsid proteins; T1898A, G2100V and E2101T, located in the portion encoding for the viral RNA polymerase 3D. Furthermore, comparing the amino acid sequence of the virus described here with 35 polyprotein sequences of other E-11 strains downloaded from Genbank, we found 25 amino acid substitutions (V101I, E115D, I310V, V572A, T651V, E895D, Q1031H, C1033S, L1938F, I1084V, T1122A, Q1185H, S1213E, S1365N, T1418S, S1429N, N1536S, P1539L, A1533S, V1558I, L1298I, T1868N, I2135V and D2140N). To date, no biological significance associated with any of these substitutions have been described.

Despite the typing of the VP1 gene assigned our virus to an E-11, the phylogenetic analysis of the full-genome sequence of ECHO11_INMI against a set of 123 sequences retrieved from the Picornavirus homepage, showed an unexpected pattern: ECHO11_INMI strain did not segregate with E-11 serotypes, but with a CV-B1 (Acc N: MG845887) (Fig. 2). This finding led us to hypothesize that ECHO11_INMI could be a chimeric strain, maybe originating from a recombination event between an E-11 (Acc N: AY167103) and CV-B1 (Acc N: MG845887).

Unfortunately, we were not able to sequence the entire genome of the virus infecting Pt 0 (index case) due to the low quantity of virus in residual material, as the liver bioptic sample (Pt0_L) was fixed in paraffin which badly conserved the nucleic acid. However, we were able to obtain two amplicons of the sequence by RT-PCR that localized upstream and downstream from the recombination breakpoint respectively. Both amplicons were sequenced and show a high identity (median value = 99.78% and 98.17% of upstream and downstream fragment, respectively) both with the sequences of the three isolates and the virus detected in the liver of patient 1 (Pt1_L), therefore suggesting that the recombination had already occurred in the virus infecting the source (Pt0).

The phylogenetic tree of the full genome sequences (data not shown) shows that ECHO11_INMI, segregates with the CV-B1 (MG845887.1). To better define the phylogenetic relationships, we analyzed separately the P1, containing the VP1 gene, and the P3 region as it locates close to 3′ end of the genome and far from P1 (Fig. 1). The phylogenetic tree of the P1 region (Fig. 2) shows that our ECHO11_INMI sequence clusters together with all E-11 types while CV-B1 (MG845887.1) segregates with CV-B1 type in a separate clade. The P1 region contains the sequence coding for the capsid proteins, VP4, VP2, VP3 and VP1. In particular, it is well known that VP1 is the most antigenic protein and its sequence is used for the typing of Enterovirus genus as it has been shown to correlate very well with the classical serotype [32]. Indeed, phylogenetic studies on VP1 sequences of the genus have clearly shown that strains of the same serotype always cluster together [33].

Therefore, this result confirms that our sequence belongs to the E-11 type. The last tree (Fig. 2), constructed on the basis of the P3 region of the genome, shows that the ECHO11_INMI segregates close to the CV-B1 (MG845887.1). This result is consistent with the hypothesis that recombination occurred in the P2 region of the genome, between P1 and P3.

To confirm the presence of a recombination breakpoint in our strains, we performed a recombination detection analysis, using RDP4 software (Fig. 3). Specifically, the analysis recognized that ECHO11_INMI was a chimeric strain of E-11 (AY167103) and CV-B1 (MG845887); it also identified the breakpoint of recombination between nucleotide 4083 and 4201 of ECHO11_INMI sequence without gap, with 99% certainty (p values 5.259*10⁻²⁴, as reported by RDP4).

The recombination site is located in the region encoding for P2 of the polyprotein that is the precursor of three non-structural proteins involved in the replication process: 2Apro, 2B and 2C.

To confirm that the virus in our samples is a new variant originating from a recombination event between an E-11 and a CV-B1, and to exclude the hypothesis of a co-infection with both viruses, we designed four sets of primers. Two of them were specific for E-11 serotype, targeting respectively the region upstream and downstream of the breakpoint (E11 2F-E11 2R, E11 3F-E11 3R); in the same way we designed two sets of primers specific for CV-B serotypes, targeting respectively the region upstream and downstream from the recombination site (CVB 2F-CVB 2R, CVB 3F-CVB 3R). Figure 1 shows a schematic representation of the EVs genome and details of experimental design for RT-PCR amplification of both E-11 and a CV-B1.

Pt1_S and all isolates were tested with all sets of primers described and we obtained similar results (Additional File 5). The amplification resulted positive with the set E112F-E112R, that targets the region upstream of the recombination breakpoint, and with the set CVB 3F-CVB 3R, that targets the region downstream of the breakpoint, instead the other sets of primers, CVB 2F-CVB 2R and E11 3F-E11 3R, that map at 5′ and 3′ of the genome, respectively, gave negative results (Fig. 4).

These findings are consistent with the recombination hypothesis and confirm the presence of only the recombinant variant in the examined samples; in addition the presence of the same pattern of amplification obtained by the analysis of the virus in serum sample and the one isolated from cell culture, revealed that the recombination was not generated by the isolation procedure.