Tracking SARS-COV-2 variants using Nanopore sequencing in Ukraine in 2021

The use of real-time genomic epidemiology has enabled the tracking of the global spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), informing evidence-based public health decision making. Ukraine has experienced four waves of the Coronavirus Disease 2019 (COVID-19) between spring 2020 and spring 2022. However, insufficient capacity for local genetic sequencing limited the potential application of SARS-CoV-2 genomic surveillance for public health response in the country. Herein, we report local sequencing of 103 SARS-CoV-2 genomes from patient samples collected in Kyiv in July-December 2021 using Oxford Nanopore technology. Together with other published Ukrainian SARS-CoV-2 genomes, our data suggest that the third wave of the epidemic in Ukraine (June-December 2021) was dominated by the Delta Variant of Concern (VOC). Our phylogeographic analysis revealed that in summer 2021 Delta VOC was introduced into Ukraine from multiple locations worldwide, with most introductions coming from Central and Eastern European countries. The wide geographic range of Delta introductions coincides with increased volume of travel to Ukraine particularly from locations outside of Europe in summer 2021. This study highlights the need to urgently integrate affordable and easily scaled pathogen sequencing technologies in locations with less developed genomic infrastructure, in order to support local public health decision making.

www.nature.com/scientificreports/ COVID-19 pandemic would be severe in Ukraine 15 . The war will likely exacerbate the effects of the pandemic by disrupting healthcare in Ukraine, and displacing the population 11 . In summer 2021, Ukraine was one of the few European countries to open its borders to international travelers, requiring a negative SARS-CoV-2 test pre-departure and purchase of private medical insurance for entry for foreign citizens 16 . With the aim of encouraging international tourism and supporting rapid economic recovery, businesses such as restaurants, bars, and nightclubs were permitted to operate at full capacity 17 . Over 4.2 million international tourists travelled to Ukraine in 2021 compared to 3.4 million in 2020, with higher proportion of foreign citizens arriving from Asian, African, and North American countries 18 . Ukraine has a recent history of outbreaks of infections such as polio and measles, 19,20 , drug-resistant tuberculosis 21,22 , and numerous veterinary pathogens 23 , which are sustainably controlled in other countries. Furthermore, despite a high HIV and hepatitis C virus prevalence, infectious disease genomic epidemiology is currently not integrated into public health decision making in Ukraine 24,25 . Unlike other European countries, this is due to the unavailability of large-scale genetic sequencing capacity in Ukraine. As a result, over the course of the SARS-CoV-2 pandemic thus far, sequencing activities were only minimally supported by the government, and SARS-CoV-2 genomic epidemiology has not contributed to the public health response in Ukraine. To date, only one study on the phylodynamic analysis of the SARS-CoV-2 sequences obtained from Ukrainian patients has been reported, describing the first four months of the pandemic. It showed that the initial rise in the number of COVID-19 cases was likely led by multiple independent introductions of the virus into Ukraine prior to border closures 26 , similar to the findings of the importation and circulation of SARS-CoV-2 lineages in the UK 7 . For the first time, our study aimed to build local and sustainable SARS-CoV-2 sequencing capacity and genomic epidemiology expertise in Kyiv, Ukraine using the Oxford Nanopore Technology (ONT) MinION sequencing platform. We generated and analysed the most recent SARS-CoV-2 genetic sequences obtained in Ukraine, and described patterns of Delta VOC introductions in summer 2021, integrating our findings with global surveillance data, thus providing an evidence-based framework for future public health decisions in Ukraine.

Results
Epidemiology of COVID-19 in Ukraine between March 2020 and February 2022. According to the national COVID-19 tracking portal, Ukraine has experienced four waves of the epidemic: the first with a peak of 22,218 daily cases on November 26th, 2020, the second with a peak of 20,341 cases on April 4th, 2021, the third one with a peak of 27,377 on November 3rd, 2021, and the fourth one which peaked at 43,778 on February 3rd, 2022 (Fig. 1A) 27 . The distribution of cases by region was uneven, with the capital city of Kyiv, and major urban centers in Odesa, Kharkiv, Dnipro, and Lviv regions being most affected, with over 300,000 cases reported in each (Fig. 1B). Data from Donetsk and Lugansk regions are likely to be underestimated for the whole pandemic period due to incomplete data availability across these regions, whilst no data was available from Crimea, as reported by Ukrainian national public health organisations; no data was reported from Ukraine after February 22nd, 2022. SARS-CoV-2 genomic epidemiology in Ukraine. SARS-CoV-2 sequencing with MinION. Using portable ONT MinION device and SARS-CoV-2 ARTIC Network workflow, we locally sequenced and analysed a set of 166 clinical samples (nasopharyngeal swabs) obtained from individuals who tested positive for SARS-CoV-2 by qRT-PCR in July (N = 24) and in October-December (N = 142) 2021 in Kyiv, Ukraine. All samples obtained in July resulted in high coverage (≥ 90% relative to the reference Wuhan-Hu-1 strain, N = 24) SARS-CoV-2 genomes. Of the 142 samples from October-December, 79 resulted in high coverage (≥ 90%) genomes and 63 produced partial genomes (37 samples resulting in 50-90% coverage and 26 samples < 50% coverage). In total, 103 highcoverage SARS-CoV-2 genomes from the third epidemic wave were obtained and further analysed in this study. Of those, 58% were obtained from female patients (see Supplementary Table S1 for additional information on the sequences and patients from whom they were sampled). We identified two variants of concern (VOC) in our samples: the Delta VOC (B.1.617.2-like) (N = 100) and the Alpha VOC (B.1.1.7-like) (N = 3) variants; all Alpha sequences were obtained from the samples collected in July. Virus genome analysis identified all common amino acid (aa) mutations which are associated with the Alpha and Delta VOC (Supplementary Table S1 and Fig. 2). Additionally, one sample assigned as an Alpha variant had the E484K mutation (G23012A) in the receptor binding domain (RBD) of the spike (S) protein and the N501S mutation instead of the N501Y mutation commonly found in this position for Alpha variant (Fig. 2). More details on the sequence quality and identified mutations can be found in the Supplementary Text.

SARS-CoV-2 lineages circulating in Ukraine.
Next, we analysed SARS-CoV-2 lineage distribution using all available sequences (N = 883) obtained from patient samples collected in Ukraine in the GISAID database as of May 11th, 2022. Of those, 873 sequences collected between March 2020 and April 2022 were selected for further analysis of the SARS-CoV-2 lineage distribution in Ukraine (Supplementary Table S2); ten sequences were excluded due to an incomplete genome or lack of collection date information. The majority of selected sequences were sequenced locally in Ukraine (N = 520; 60%) using different sequencing platforms such as Ion Torrent, ONT MinION, and Illumina (MiSeq and iSeq100). Most (66%, N = 342) of the locally generated sequences were produced on the Ion Torrent platform, 14% (N = 73) were sequenced with the Illumina, and 20% (N = 105) with ONT MinION, of which 103 (98%) were sequenced in this study.
The SARS-CoV-2 lineage distribution in Ukraine consisted of five predominant lineages as designated by the PANGO nomenclature 28 Table S3). The 466 Delta sequences were assigned to 25 Delta VOC sub-lineages and 105 Omicron sequences were assigned to 13 Omicron VOC sub-lineages as described in Supplementary Table S3. Lineage prevalence varied through the epidemic in the country, consisting of five broad time frames: a "Pre-Wave" (March to 31 Aug 2020) and "Wave 1"  Delta diversity tree, but we also found 6 pairs, two clusters of five sequences, one cluster of three sequences, and one cluster of seven sequences ( Supplementary Fig. S3). The phylogeographic analysis indicated the highest number of introductions from Turkey (four individual introductions) and at least one importation event from each of the Baltic countries, Slovenia, Nigeria, Jordan, Russia, Finland, Czech Republic, and United Arab Emirates. The TreeTime mugration analysis was not able to resolve the ancestral location for ten separate introduction events.

Discussion
Ukraine has experienced four COVID-19 epidemic waves between the identification of the first case on February 29th 2020 and April 2022. Compared to other European countries, the first three epidemic waves in Ukraine lagged by several months as Ukraine did not experience many cases in spring 2020, likely due to a combination of a timely lockdown and a restriction of movement policy 29,30 . The border closure was applied to both foreign citizens and Ukrainian nationals in March 2020, three weeks after the first case was diagnosed in the country. However, lower numbers of registered cases in Ukraine may also be attributed to many infections likely being undiagnosed due to limited testing and diagnostic capacity in spring 2020. When most of the world was approaching the second wave of the epidemic in September 2020, the number of COVID-19 cases in Ukraine started to grow quickly, resulting in the first epidemic wave in the country. Just as the epidemic waves in Ukraine followed a similar trajectory as in Western European countries, but with a few months delay, they have also been dominated by the same VOCs. As such, the second wave of the epidemic in Ukraine, which started in February 2021, was dominated by the Alpha VOC, and the third wave, which started in Ukraine in June 2021, by the Delta VOC, which remained dominant amongst the European Economic Area countries until September 2021 31 and in Ukraine until December 2021. The fourth COVID-19 wave in Ukraine has been driven by Omicron BA.1 infections that likely peaked in February 2022 and were followed by an Omicron BA.2 wavelet. However, the true dynamics of the Omicron wave sub-lineages is impossible to determine due to the paucity of published genomics data from Ukraine since the Russian invasion on February 24th, 2022, with concomitant disruption of healthcare including disease surveillance, and large numbers of internally displaced people.
Unlike the strict measures in the beginning of the epidemic, no travel restrictions were imposed in summer 2021, and increased tourist flow likely contributed to the third wave of the epidemic in June-December 2021. Our www.nature.com/scientificreports/ phylogeographic analysis, which focused on Delta sequences available in Ukraine between June and September 2021, showed that the Delta variant has been introduced into Ukraine from multiple geographic locations. While this analysis suggested Delta VOC sub-lineages were introduced from nearby countries, such as Turkey, Russia, and Eastern European nations, we found evidence for introduction of sub-lineages from more distant locations in Africa and the Middle East. Interestingly, in summer 2021, among the top-10 Delta case countries, introductions came only from two, from Turkey and US (see Methods). Such vast geographic spread of virus introductions into the country is likely the result of the increase in the tourist flow and the diversity of countries of origin of tourists travelling to Ukraine, as well as absence of COVID-19 restrictions in the country in summer 2021. Importantly, although we tried to find a balanced approach to sub-sampling of the global Delta variant diversity for our phylogeographic analysis, such inference is complicated by uneven testing and sequencing efforts across the globe, and by the fact that identical sequences can be found in the many countries considered in the analysis. Thus, results of our phylogeographic analysis should be interpreted with caution and only refer to the patterns observed in summer 2021 in the beginning of the third wave of the epidemic in Ukraine.
Large-scale genomic sequencing of SARS-CoV-2 has enabled global surveillance of pandemic dynamics but was yet to become part of the public health response in Ukraine at the time that it was invaded by Russia. In 2021, for the first time, we utilised ONT MinION technology in combination with ARTIC Network protocols to locally sequence 103 SARS-CoV-2 genomes collected from patient samples in Kyiv, Ukraine, to enable development of genomic epidemiology expertise and sequencing capacity in Ukraine. While most of the genomes sequenced with MinION were complete and of high-quality, some of the sequences produced in this study had lower than optimal genome coverage (< 90%), likely due to poor sample quality or low amplification efficiency where the amplicons are not equally amplified across all regions of the genome. The ARTIC V3 primer set was used for the 24 samples collected in July 2021, which has been previously reported to be associated with the amplicon 72 (S gene) dropout in the Delta variant sequences as observed here. The second major amplicon dropout reported for these sequences (amplicon 64 in ORF1ab) was likely due to a similar issue in primer compatibility. This prompted the development of the ARTIC V4 primer set compatible with the novel mutations 32 . Thus, ARTIC V4 primers were used to sequence samples collected in October-December 2021, which improved the coverage of Delta VOC and helped to significantly reduce the amplicon 64 and 72 dropouts.
Genomic epidemiology integration within public health response allows tracking of variants and mutations that potentially affect vaccine efficacy. For example, the E484K mutation in the S protein, an escape mutation that shows evidence of impacting the immune response and likely reducing vaccine effectiveness 33 , was identified in one Alpha variant sequence in our analysis. This mutation is rare for this lineage and is usually present in other VOC (Beta, Gamma, Zeta, and Eta) 34 . As of May 11th, 2021, the E484K mutation was identified in 3,786 (0.32%) of all Alpha sequences in GISAID, of which 0.23% were from Europe predominantly in France, Germany, Sweden and Denmark. The N501S mutation in S, another mutation found in one of our Alpha variant sequences, was also reported to increase receptor-binding domain (RBD) stability and RBD-ACE2 (angiotensin converting enzyme 2) binding affinity compared to the original Wuhan-Hu-1 isolate but was predicted to have a neutral effect on protein 35 . Furthermore, the same sample sequence had the Y144-deletion in the S protein associated with antibody escape 36 . Tracking of the prevalence of these mutations in Ukrainian SARS-CoV-2 viral population is of immense importance for the country's vaccine strategy going forward, but is likely to be of low priority at the time that the country is experiencing an on-going war and humanitarian crisis 11 .
Vaccination in Ukraine against COVID-19 is among the lowest in Europe, with 15.2 million people fully vaccinated (34.3% of the population), 35.7% partially vaccinated, and 1.7% given a booster dose as of February 27th, 2021 3 . Without domestic vaccine manufacturing, Ukraine has relied on donations and external sourcing, for example, the WHO COVAX program that supplied 4.2 million doses total of AstraZeneca AZD1222 ChAdOx1, Pfizer-BioNTech BNT162b2 mRNA, and Moderna mRNA-1273 vaccines 37,38 . Vaccination began in February 2021, with half of the vaccine doses having been adminisered primarily to adults aged 20-49, and the remainder to those over 50 + years, with a small proportion of teenagers.
The Delta variant, which dominated the third wave of the epidemic in Ukraine, is characterized by increased transmissibility and virulence, and marginally lower vaccine efficacy (VE) of multiple vaccines against symptomatic COVID-19 4,39 . In combination with waning immunity against SARS-CoV-2 4,40,41 , the WHO recommended booster vaccination to enhance protection against the Delta VOC 37 . Despite the evidence that Delta variant was the dominant lineage in the third COVID-19 wave in Ukraine (Fig. 3A), only a negligible number of people in Ukraine (9,515, 0.02% of the population) received a booster vaccine dose as of December 31st, 2021, the end of the third wave of the epidemic in Ukraine. This was likely due to the limited availability of VOC surveillance to provide evidence of the virus strains circulating patterns, as well as access to more vaccine doses in this lowermiddle income country.
While it is unlikely that genomic surveillance can be scaled up in Ukraine at the time of a crisis such as the ongoing war, our findings suggest that there is a pressing need in lower-and middle-income countries for input from genomic epidemiology to inform public health policy of the threat of emerging variants. In Ukraine, availability of national genomic surveillance earlier in the COVID-19 epidemic may have influenced better interventions, such as alternative timing of lifting of local restrictions and better regulation of international travel, thus slowing the spread of the VOCs in Ukraine. After the war, parallel to other efforts in restoration of health care systems in Ukraine, future investment in establishing a sustainable and local genomic surveillance program for SARS-CoV-2 and other prevalent pathogens will ensure evidence-based public health decisionmaking in Ukraine.

SARS-CoV-2 genome assembly.
Genomes were assembled using the ARTIC bioinformatics pipeline with a 20X minimum coverage across all genomic regions. First, basecalling was performed with guppy 5.0.11 using the dna_r9.4.1_450bps_sup.cfg model followed by demultiplexing with guppy 5.0.11 for barcode kits EXP-NBD104, EXP-NBD114 and EXP-NBD196. Then read filtering was performed with Artic Guppyplex allowing for read length between 400 and 700 nucleotides. Finally, ARTIC Minion nanopolish pipeline was run with the option to normalise to 200. All sequences obtained are available on GISAID EpiCov database (www. gisaid. org). The accession numbers of the sequences obtained in this study are presented in Supplementary Table S1.

Phylogenetic analysis.
We downloaded all complete SARS-CoV-2 genomes from Ukraine from GISAID database on May 11th, 2022. All Ukrainian sequences, from GISAID and generated in this study, were aligned using MAFFT. We masked all problematic sites in the alignment as suggested by De Maio et al. (https:// virol ogical. org/t/ maski ng-strat egies-for-sars-cov-2-align ments/ 480). PANGO lineage assignments were determined using the Pango v.4.0.6 nomenclature tool 28 . Mutation calling and sequence quality were analysed using Nextclade (https:// clades. nexts train. org/) and Geneious Prime 2021.2.2 software (Biomatters, New Zealand). Relative distribution of dominant lineages and mutation heatmap was made using R Studio with the ggplot2 package. A maximum likelihood phylogenetic tree was then built using only high-quality genomes with IQTREE2 program under the general time reversible model allowing for rate heterogeneity among sites and proportion of invariable sites (GTR + G + I).
Phylogeographic analysis. To establish the number and the source country of the Delta variant introductions into Ukraine in summer 2021, we have downloaded all complete high-quality with available sampling date Delta variant sequences sampled between January 1st and September 30th available on GISAID on September 30th (N = 310,007, including N = 35 from Ukraine). We then down-sampled the number of sequences in all location to 12, except for the countries with top-10 number of Delta variant cases in the world (USA, UK, France, Germany, Denmark, India, Canada, Turkey, Sweden, and Switzerland) 42 , countries that share border with Ukraine (Russia, Moldova, Romania, Hungary, Poland, Slovakia, Belarus), and other post-Soviet countries, including the three Baltic nations (Estonia, Latvia, Lithuania), Georgia, and Uzbekistan (no Delta sequences were available from the other post-Soviet countries in Central Asia and Transcaucasia); we included 36 sequences from each of these locations in the analysis. We then combined the down-sampled dataset and the 21 Delta sequences produced in this study in August 2021 and applied the mugration model implemented in TreeTime 43 to estimate the number and originating locations of Delta variant introductions in Ukraine. The mugration model assumes countries of origin as discreet states and considers the virus spread between these states as a general time-reversible process.