SARS-CoV-2 coinfection in immunocompromised host leads to the generation of recombinant strain

Abstract

Objectives

Recombination related to coinfection is a huge driving force in determining the virus genetic variability, particularly in conditions of partial immune control, leading to prolonged infection. Here, we characterized a distinctive mutational pattern, highly suggestive of Delta-Omicron double infection, in a lymphoma patient.

Methods

The specimen was characterized through a combined approach, analyzing the results of deep sequencing in primary sample, viral culture, and plaque assay.

Results

Bioinformatic analysis on the sequences deriving from the primary sample supports the hypothesis of a double viral population within the host. Plaque assay on viral culture led to the isolation of a recombinant strain deriving from Delta and Omicron lineages, named XS, which virtually replaced its parent lineages within a single viral propagation.

Conclusion

It is impossible to establish whether the recombination event happened within the host or in vitro; however, it is important to monitor co-infections, especially in the exceptional intrahost environment of patients who are immunocompromised, as strong driving forces of viral evolution.

Background

Since late 2020, SARS-CoV-2 has clearly demonstrated its capacity to generate new variants, marked by the emergence of sets of mutations that impact virus characteristics, including transmissibility and antigenicity.

SARS-CoV-2 evolution is an intricate process related to the different mechanisms on the molecular, organism, and population scale. The development of point mutations has played a big role in the emergence of new variants [1]; on the other hand, the recombination between closely related genotypes occurs readily due to the high sequence identity and may result in the emergence of new strains [2]. Coronaviruses have an intrinsically high intratypic recombination rate (approximately 25%) across the genome. To allow for homologous recombination, coinfection of genetically different viruses must occur in the same host cell. The crossover sites may occur anywhere, but the selection pressure can lead them to cluster in certain hotspots [3,4].

Favorable conditions for coinfection—and subsequent recombination—spike in periods of coexistence of two major lineages. The most recent one in our geographic area happened between October 11, 2021 and March 27, 2022, when Omicron succeeded Delta as the predominant lineage but the two variants co-circulated for a time. Coinfections have been reported multiple times [5], [6], [7], more recently involving Delta and Omicron [8,9]. Delta-Omicron recombinants have also been reported [10], [11], [12], [13].

Recombinant viruses were initially identified only through bioinformatic tools, but they have now been isolated in culture as well, which allows the investigation of their epidemic potential. Most often, this has been done on patients who were presumably infected with a recombinant strain to begin with [10,12,13]. However, Burel et al. were able to monitor a coinfection between B.1.160 and Alpha for 14 months, until its evolution in a recombinant strain, and culture it [14].

The origin of variants is still a matter of speculation. Several hypotheses take zoonotic origin, selective pressure during treatment with antiviral drugs, monoclonal antibodies, or convalescent plasma into consideration and a few studies point to the significance of the intrahost environment of patients who are immunocompromised to explain the evolution of immune escape variants [15,16]. Individuals who are immunocompromised are more likely to be long carriers, which increases the likelihood of subsequent coinfection and recombination events.

Because homologous recombination related to coinfection and conditions of partial immune control are strong driving forces of viral evolution, it is very important to monitor such instances. Here, we describe the composite approach we used to accurately characterize a peculiar SARS-CoV-2 sequence, suggestive of a double viral population, combining bioinformatic tools and plaque assay on viral culture.

The case

A male patient, aged 47 years, diagnosed with stage IVa nodular sclerosing non-Hodgkin lymphoma and diabetes, was admitted to the hospital on January 14, 2022 due to severe respiratory distress. The point-of-care testing for SARS-CoV-2 was positive, and later confirmed by our laboratory on January 21, 2022. Patient death was recorded 17 days after admission (January 31, 2022).

Sequencing was performed on the nasopharyngeal swab in the context of routine surveillance and monitoring for SARS-CoV-2 variants. Sequence analysis showed an unusually low number of mutations (N = 17) compared with the circulating lineages on our territory at the time, Delta (N ≈ 45) and Omicron (N > 60). In addition, several mutations were detected in a lower fraction of the viral population (variant fraction, 70-90%). All the analysis softwares used for lineage characterization yielded inconclusive results.

Because this mutational pattern was highly suggestive of a double viral population, the primary sample was re-tested to exclude sequencing errors or contaminations. At the same time, viral culture on the specimen was paired with plaque assay to attempt to isolate and characterize the two populations.

Methods

Sequencing

Whole genome sequencing was performed on the original sample using an amplicon-based approach. We implemented the CleanPlex SARS-CoV-2 Panel (Paragon Genomics, Inc., Hayward, CA, USA) for target enrichment and library preparation, which involves multiplex polymerase chain reaction (PCR) reactions. The sequencing step was conducted on a MiSeq platform (Illumina, Inc., San Diego, CA, USA).

Bioinformatic tools

The data analysis for the consensus sequence generation and mutation calling was performed according to the supplier's recommendations using SOPHiA-DDM-v4 (SOPHiA Genetics, Lausanne, Switzerland). The software operates a cut-off, which excludes from reporting all the mutations detected below 70%. The consensus sequences were submitted to Pangolin and NextClade for lineage assignment. In addition, the raw data from the primary sample was aligned and analyzed using Lasergene SeqMan Ultra software (DNASTAR Inc, Madison, WI, USA) to detect mutations below 70%. Each mutation identified was analyzed compared with the database of all samples sequenced in our laboratory to date, comprising 2668 Delta sequences, 1043 Omicron, and 1500 Alpha at the time of the analysis. The mutations were considered markers of a specific lineage if they were significantly present within it (>90% samples) and absent in all others (<10%).

Plaque assay

Viral isolates were propagated from the residual specimen on Vero E6 cell cultures (American Type Culture Collection CRL-1586), as recommended [17]. A total of 500 µl of viral transport media were used to infect a cell monolayer at confluency, allowing a 1-hour adsorption and a 72-hour incubation. Viral replication was then assessed by reverse transcription-PCR. Serial dilutions of viral isolate were cultured using 0.5% agarose added to the medium to obtain visible, immobilized focuses of infection (plaque assay). Each focus was then separately eluted, cultured, and sequenced, performing data analysis as previously described.

Characterization of coinfection and recombinant strain identification

The double analysis on the primary sample, collected on January 21, 2022, highlighted the presence of respectively 17 and 21 mutations compared with the reference. Most mutations are traceable either to Delta (italics) or Omicron (bold) lineages (Table 1 ), and several were detected in an unusually low fraction of the viral population (70-90%). The low number of mutations cannot be attributed to data loss as the genome coverage was 99.8% and 99.9%, respectively. The most likely explanation is a higher number of mutations with a variant fraction below 70%, which would be hidden by the software cut-off. Marker mutations of multiple lineages, low variant fractions, and fluctuating mutational patterns are all hallmarks of coinfections [18].

Table 1.

Sequence profiles from the analysis of separate aliquots of the same primary sample and of the viral isolates. Only mutations above 70% are reported. Most mutations are referable either to Delta (italics) or Omicron (bold). The mutational pattern in the primary sample is consistent with the presence of two separate viral lineages within the specimen. Conversely, in the viral isolates the ORF1ab portion is Delta-like (italics), the rest of the genome is Omicron-like (bold). This mutational pattern is consistent with a single viral population, deriving from the recombination of two separate lineages.

	Sequence 1			Sequence 2			Plaque assay
gene	protein	depth	var fraction %	protein	depth	var fraction (%)	gene	protein	depth	var fraction (%)	gene	protein	depth	var fraction (%)	gene	protein	depth	var fraction (%)
ORF1ab				Ile695Val	1702	73.6	ORF1ab	Ile695Val	1515	99.7	Spike	Ala67Val	2995	99.8	ORF3a	Thr64=	2786	99.8
	Phe924=	2101	99.7	Phe924=	1777	99.3		Phe924=	1871	99.5		His69_Val70del	3001	99.9	E gene	Thr9Ile	1556	100
				Pro2046Leu	1017	71.5		Gly934Val	1856	99.5		Thr95Ile	3931	99.9	M gene	Asp3Gly	2015	99.9
				Pro2287Ser	2874	74.1		Asn1076=	1771	99.9		Gly142_Tyr145delinsAsp	4445	99.9		Gln19Glu	2013	99.9
				Ala2529Val	1633	74.3		Ala1306Ser	3610	99.9		Asn211_Leu212delinsIle	1308	100		Ala63Thr	2158	99.9
	Thr3255Ile	1690	100	Thr3255Ile	1616	99.8		Tyr1873=	1737	99.8		Arg214_Asp215insGluProGlu	1305	97.5	ORF6	Arg20=	459	99.1
				Ala3645=	3997	72.2		Pro2046Leu	1217	99.4		Gly339Asp	2524	99.8	ORF7b	Leu18=	3204	99.8
				Thr3646Ala	3997	72.2		Pro2287Ser	4221	99.6		Ser371Pro	134	100	N gene	Pro13Leu	5706	99.9
	Leu3674_Gly3676del	744	100	Leu3674_Gly3676del	525	100		Ala2529Val	2973	99.7		Ser371Phe	134	100		Glu31_Ser33del	2189	99.7
	Val3689=	746	100	Val3689=	1199	99.7		Asp2907=	2428	99.6		Ser373Pro	134	100		Arg203Lys	1771	99.8
	Pro4715Leu	2651	99.7	Pro4715Leu	2368	99.9		Val2930Leu	522	99.2		Ser375Phe	134	100		Arg203=	1771	99.9
	Gly5063Ser	6005	73.2	Gly5063Ser	5473	75.3		Thr3255Ile	2100	99.9		Lys417Asn	584	100		Gly204Arg	1771	99.9
	Pro5401Leu	5085	74.3	Pro5401Leu	6446	76.2		Ala3645=	8379	99.8		Asn440Lys	4132	99.9
				Ala6319Val	471	71.5		Thr3646Ala	8379	99.8		Gly446Ser	4131	99.9
Spike	Ala67Val	1323	73.5					Val3689=	2176	99.9		Thr547Lys	2721	99.7
	His69_Val70del	1323	73.3					Pro4715Leu	2930	99.8		Asp614Gly	5187	99.9
	Thr95Ile	2156	99.9	Thr95Ile	2304	99.9		Gly5063Ser	5078	99.9		His655Tyr	4796	99.8
	Gly142_Tyr145delinsAsp	1211	100	Gly142_Tyr145delinsAsp	1181	100		Pro5401Leu	9979	99.9		Asn679Lys	3369	100
	Glu156_Arg158delinsGly	334	98.2	Glu156_Arg158delinsGly	252	100		Ala6319Val	2302	100		Pro681His	3368	99.8
	Leu452Arg	159	100	Leu452Arg	106	100						Ala701Val	231	99.6
	Thr478Lys	314	100	Thr478Lys	210	100						Asn764Lys	1232	99.5
	Asp614Gly	2780	99.9	Asp614Gly	3559	100						Asp796Tyr	408	100
	Asp950Asn	651	79.7	Asp950Asn	1003	90.1						Asn856Lys	3903	99.2
												Gln954His	1544	99.8
												Asn969Lys	872	100
												Leu981Phe	884	99.5
												Asp1146=	1428	99.7

ORF, open reading frame.

This hypothesis was confirmed through analysis with a second alignment software to categorize all the mutations below the initial cut-off. As expected, we found, across the whole viral genome, a very high number of mutations previously undetected and well below 70%, pertaining to both lineages. In two instances, we were able to identify the simultaneous presence of two marker mutations, respectively for Delta and Omicron, at the same genomic position (Tables 2 and 3 ). Once analyzed below 70%, the two runs yielded very similar results. In no case, however, we found patient-specific mutations.

Table 2.

Sequence profiles from the analysis of aliquot 1 from primary sample. Mutations below 70% are reported. Most mutations are referable either to Delta (italics) or Omicron (bold). The mutational pattern in the primary sample is consistent with the presence of two separate viral lineages within the specimen.

Sequence 1
gene	protein	depth	var fraction %	gene	protein	depth	var fraction %	gene	protein	depth	var fraction %
				Spike	Thr19Arg	26	800
ORF1ab	Ile695Val	69	2025		Ala67Val	73	1323	ORF3a	Ser26Leu	34	815
	Lys856Arg	35	1214		His69_Val70del	73	1323		Thr64=	66	1102
	Phe924=	100	2096		Thr95Ile	100	2156		Asp155Tyr	41	2114
	Gly934Val	58	2094		Gly142_Tyr145delinsAsp	1211	100
	Asn1076=	36	598		Glu156_Arg158delinsGly	334	98	E gene	Thr9Ile	60	602
	Ala1306Ser	40	2672		Gly339Asp	64	4471
	Ala1707=	66	739		Ser371Pro	67	109	M gene	Asp3Gly	62	2196
	Pro2046Leu	66	1148		Ser371Phe	67	109		Gln19Glu	63	1996
	Pro2287Ser	69	2533		Ser373Pro	67	109		Ala63Thr	56	1982
	Ala2529Val	68	1501		Ser375Phe	65	109		Ile82Thr	45	1983
	Ala2710Thr	37	696		Lys417Asn	58	952
	Asp2907=	64	2293		Leu452Arg	100	159	ORF6	Arg20=	48	591
	Thr3255Ile	100	1690		Thr478Lys	100	314
	Pro3395His	34	2457		Thr547Lys	50	3938	ORF7a	Thr120Ile	50	2243
	Ala3645=	68	3282		Asp614Gly	100	2780
	Thr3646Ala	68	3282		His655Tyr	49	4559	ORF7b	Leu18=	57	3203
	Leu3674_Gly3676del	744	100		Asn679Lys	53	2032		Thr120Ile	27	961
	Val3689=	100	746		Pro681Arga	47	949
	Ile3758Val	43	1940		Pro681Hisa	53	1079	N gene	Pro13Leu	45	7798
	Val4310=	36	3859		Asn764Lys	57	1805		Asp63Gly	53	1143
	Pro4715Leu	100	2651		Asn856Lys	62	3616		Arg203Meta	58	1949
	Asn4992=	29	2756		Asp950Asn	80	649		Arg203Lysa	42	1389
	Gly5063Ser	73	6005		Gln954His	20	650		Gly204Arg	42	3339
	Pro5401Leu	74	5086		Asn969Lys	58	1121		Gly215Cys	58	8679
	Ile5967Val	32	1998		Leu981Phe	57	1124		Asp377Tyr	51	4223
					Asp1146=	66	1679

^⁎In two instances, we were able to identify the simultaneous presence of two marker mutations, respectively for Delta and Omicron, at the same genomic position.

Table 3.

Sequence profiles from the analysis of aliquot 2 from primary sample. Mutations below 70% are reported. Most mutations are referable either to Delta (italics) or Omicron (bold). The mutational pattern in the primary sample is consistent with the presence of two separate viral lineages within the specimen.

Sequence 2
gene	protein	depth	var fraction %	gene	protein	depth	var fraction %	gene	protein	depth	var fraction %
ORF1ab	Ile695Val	74	1702	Spike	Thr19Arg	1024	30	ORF3a	Ser26Leu	974	39
	Lys856Arg	33	1561		Ala67Val	504	55		Thr64=	1294	62
	Phe924=	100	1765		His69_Val70del	504	55		Asp155Tyr	1839	44
	Gly934Val	61	1773		Thr95Ile	2304	100
	Asn1076=	30	725		Gly142_Tyr145delinsAsp	1181	100	E gene	Thr9Ile	788	53
	Ala1306Ser	34	2851		Glu156_Arg158delinsGly	252	100
	Ala1707=	72	922		Arg214_Asp215insGluProGlu	540	67	M gene	Asp3Gly	1619	62
	Pro2046Leu	72	1017		Gly339Asp	4855	64		Gln19Glu	1415	60
	Pro2287Ser	74	2874		Ser371Pro	170	65		Ala63Thr	1681	53
	Ala2529Val	74	1633		Ser371Phe	170	65		Ile82Thr	1681	47
	Ala2710Thr	27	897		Ser373Pro	170	65
	Asp2907=	69	2072		Ser375Phe	170	65	ORF6	Arg20=	334	51
	Thr3255Ile	100	1616		Lys417Asn	508	50
	Pro3395His	27	2849		Leu452Arg	106	100	ORF7a	Thr120Ile	2315	54
	Ala3645=	72	3997		Thr478Lys	210	100
	Thr3646Ala	72	3997		Thr547Lys	3698	41	ORF7b	Leu18=	3123	52
	Leu3674_Gly3676del	100	525		Asp614Gly	3559	100		Thr120Ile	811	30
	Val3689=	100	1199		His655Tyr	5317	45
	Ile3758Val	39	2381		Asn679Lys	2604	45	N gene	Pro13Leu	8973	46
	Val4310=	32	4634		Pro681Arga	1438	55		Asp63Gly	1574	50
	Pro4715Leu	100	2368		Pro681Hisa	1161	45		Arg203Meta	1711	55
	Asn4992=	29	2079		Asn764Lys	1804	52		Arg203Lysa	1378	45
	Gly5063Ser	75	5473		Asn856Lys	4261	65		Gly204Arg	3096	45
	Pro5401Leu	76	6446		Asp950Asn	1003	90		Gly215Cys	10837	57
	Ile5967Val	28	2662		Asn969Lys	933	53		Asp377Tyr	5012	48
	Ala6319Val	72	470		Leu981Phe	938	53
					Asp1146=	1276	65

^⁎In two instances, we were able to identify the simultaneous presence of two marker mutations, respectively for Delta and Omicron, at the same genomic position.

Conversely, the sequences of the initial viral propagation and of eight separate plaques of infection all yielded next-to-identical results, summarized with a single sequence illustrated in Table 1 (EPI_ISL_12870564) and Figure 1 . Variant fractions for all detected mutations are nearing 100%, a strong indicator of a single viral population. Furthermore, the open reading frame (ORF1ab) portion is generally consistent with a Delta lineage and specifically bears the marker mutation for AY.4 (Ala2529Val) [19]; the rest of the genome is comparable to BA.1.

Figure 1.

Schematic figure representing the recombinant structure with respect to the different lineages. NSP, nonstructural protein; ORF, open reading frame.

These results are compatible with a recombinant strain deriving from Delta and Omicron lineages. The sequence analysis with NextClade offered further confirmation, illustrating a clear breakpoint between ORF1ab and spike (approximate breakpoint site: 20418-21618).

The sequence was initially classified as XF, which caused a small cluster in the United Kingdom in February 2022 [20,21], but it has now been regrouped as XS by the lineage assignment softwares. Both XF and XS are recombinant strains deriving from AY.4 and BA.1, differing in the position of the breakpoint site. The first XS sequence has been deposited on the Global Initiative on Sharing Avian Influenza Data (GISAID) on February 02, 2022, coming from North America, as all sequences currently considered XS on GISAID (n = 61). This number may be underestimated, as sequences coming from recombinant strains are often difficult to assign and require much longer investigation.

Discussion

The ability of SARS-CoV-2 to generate new variants is an intricate process determined by the interplay among different mechanisms on the molecular, organism, and population scale. Although the development of point mutations has played a big role, recombination is a huge driving force in determining the virus genetic variability. To allow for homologous recombination, coinfection of genetically different viruses must occur in the same host cell [22].

Here, we describe the characterization of a peculiar SARS-CoV-2 sequence found in an immunocompromised patient, suggestive of coinfection. A more accurate bioinformatic analysis on the sequences deriving from the primary sample supports the hypothesis of a double viral population within the host. On the other hand, the sequencing of separate focuses of infection in vitro highlighted identical mosaic structures. The result is a recombinant SARS-CoV-2 strain derived from the combination of AY.4 (Delta) and BA.1 (Omicron), currently categorized as XS, derived from the coexistence of the two lineages.

It would be very interesting to establish whether the recombination event happened within the host or in vitro. This could be done in two ways: first, through the identification of sequencing reads containing markers for both lineages and second, through the generation of PCR products overlapping the putative recombination site. Neither of these methods are feasible in our context because the last Delta marker was identified at position 20418 and the first Omicron marker at 21618; there are no reads long enough to contain both. As for the detection of recombinant PCR products, it is obvious from Tables 2 and 3 that there is a very high presence of parent lineages in the primary sample, as indicated from the balanced percentage of markers of both lineages at the same genomic position; in this context, a negative result would be no indication of a later recombination event because it could very well stem from a low percentage of recombinant virus in an interfering environment.

Both Delta-Omicron coinfections and recombinants have now been reported and/or isolated multiple times [8], [9], [10], [11], [12], [13],23]. Recombinant strains are examined accurately for their epidemic potential and ability to escape neutralization as they have shown resistance to monoclonal antibodies, such as Sotrovimab [12], whereas the parent lineages are not. However, it is very difficult to monitor the exact moment of the strain generation. At present, and to the best of our knowledge, only Burel and colleagues were able to monitor a coinfection until its evolution in a recombinant strain over the course of 14 months [14] and culture it.

Our report aims to expand the body of work on the subject. Given the very short time span between first sequencing and patient death, there is a lack of sequential sampling providing more detailed information on viral evolution, which is the main weakness of the study. On the other hand, this also raises the question of a potential rapid development of recombinants under the right environmental conditions.

The generation of mutated strains in hosts who are immunocompromised is very well characterized as linked to their higher likelihood to be long carriers, which in turn increases the chance of subsequent coinfection and mutation events [16,[24], [25], [26]. This is especially related to the variants created through the accumulation of point mutations, while it only takes one mutational step to generate a single breakpoint recombinant. It is worth mentioning that contexts of partial immune control favor evolutionary jumps not only through very long infections that cannot be overcome, but also acting as selective pressure [16,24,25].

Furthermore, the region between ORF1ab and the Spike gene is a very frequent breakpoint site not only in Delta-Omicron recombinants, (usually with ORF1ab Delta region and an Omicron region encompassing Spike's receptor binding domain and C-terminal regions, [10], [11], [12], [13],23]) but dating as far back as Alpha recombinants [5]. This has been linked to the phenomenon of template switching by viral polymerase during normal transcription, where the polymerase pauses at a transcription-regulatory sequence after transcribing the last open reading frame of one subgenomic RNA and switches to a similar regulatory sequence, omitting a looped-out region of the template RNA, which contains at least ORF1ab in the case of SARS-CoV-2 [5]. In the context of coinfections, the availability of alternative template RNA molecules provides an environment that is highly conducive to homologous recombination.

This study expands on SARS-CoV-2 recombinants and especially on the advantages of pairing sequencing and bioinformatic analysis with culture to monitor and characterize coinfections and any newly generated strain. Despite our impossibility to pinpoint the time of recombination, it is worth noting the speed with which XS emerged and substituted its parent lineages in vitro. Considering the combination of favorable conditions for a recombinant strain to be generated in relatively short times, this study further stresses the necessity of monitoring patients who are immunocompromised carefully, especially in contexts of co-circulation between the different lineages.

Declaration of competing interest

The authors have no competing interests to declare.