ARTICLE
https://doi.org/10.1038/s41467-022-28766-y OPEN
Antibody escape and global spread of SARS-CoV-2
lineage A.27
Tamara Kaleta1,22, Lisa Kern1,22, Samuel Leandro Hong 2,22, Martin Hölzer3, Georg Kochs 1, Julius Beer1,
Daniel Schnepf 1, Martin Schwemmle 1, Nena Bollen2, Philipp Kolb 1, Magdalena Huber1, Svenja Ulferts 4,
Sebastian Weigang 1, Gytis Dudas 5, Alice Wittig 3,6, Lena Jaki1, Abdou Padane7, Adamou Lagare8,
Mounerou Salou9, Egon Anderson Ozer 10, Ndodo Nnaemeka11, John Kofi Odoom12, Robert Rutayisire13,14,
Alia Benkahla15, Chantal Akoua-Koffi16, Abdoul-Salam Ouedraogo17, Etienne Simon-Lorière 18,
Vincent Enouf19, Stefan Kröger 20, Sébastien Calvignac-Spencer 21, Guy Baele 2, Marcus Panning 1✉ &
Jonas Fuchs 1✉
In spring 2021, an increasing number of infections was observed caused by the hitherto rarely
described SARS-CoV-2 variant A.27 in south-west Germany. From December 2020 to June
2021 this lineage has been detected in 31 countries. Phylogeographic analyses of
A.27 sequences obtained from national and international databases reveal a global spread of
this lineage through multiple introductions from its inferred origin in Western Africa. Variant
A.27 is characterized by a mutational pattern in the spike gene that includes the L18F, L452R
and N501Y spike amino acid substitutions found in various variants of concern but lacks the
globally dominant D614G. Neutralization assays demonstrate an escape of A.27 from con-
valescent and vaccine-elicited antibody-mediated immunity. Moreover, the therapeutic
monoclonal antibody Bamlanivimab and partially the REGN-COV2 cocktail fail to block
infection by A.27. Our data emphasize the need for continued global monitoring of novel
lineages because of the independent evolution of new escape mutations.
1 Institute of Virology, Freiburg University Medical Center, Faculty of Medicine, University of Freiburg, Freiburg, Germany. 2 Department of Microbiology,
Immunology and Transplantation, Rega Institute, KU Leuven, Herestraat 49, 3000 Leuven, Belgium. 3Methodology and Research Infrastructure,
Bioinformatics, Robert Koch Institute, Berlin, Germany. 4 Institute of Experimental and Clinical Pharmacology and Toxicology, Freiburg University Medical
Center, Faculty of Medicine, University of Freiburg, Freiburg, Germany. 5 Gothenburg Global Biodiversity Centre, Carl Skottsbergs gata 22B, 413 19
Gothenburg, Sweden. 6 Hasso Plattner Institute, Digital Engineering Faculty, University of Potsdam, 14482 Potsdam, Germany. 7 Institut de Recherche en
Santé, de Surveillance Epidémiologique et de Formation (IRESSEF), Rufisque, Senegal. 8 Centre de Recherche Médicale et Sanitaire (CERMES), 634 Bld de la
Nation, BP:10887YN034 Niamey, Niger. 9 Laboratoire de Biologie Moléculaire et d’Immunologie, Département des Sciences Fondamentales, Université de
Lomé, Lomé, Togo. 10 Center for Pathogen Genomics and Microbial Evolution, Institute for Global Health, Northwestern University Feinberg School of
Medicine, Chicago, IL 60611, USA. 11 National Reference Laboratory, Nigeria Centre for Disease Control, Abuja, Nigeria. 12 Department of Virology, Noguchi
Memorial Institute for Medical Research, University of Ghana, Legon, Accra, Ghana. 13 Rwanda National Joint Task Force COVID-19, Rwanda Biomedical
Centre, Ministry of Health, Kigali, Rwanda. 14 National Reference Laboratory, Rwanda Biomedical Center, Kigali, Rwanda. 15 Laboratory of BioInformatics
bioMathematics, and bioStatistics (BIMS), Institut Pasteur de Tunis, University of Tunis El Manar, Tunis, Tunisia. 16 Laboratoire Central, Centre Hospitalier et
Universitaire de Bouaké, Bouaké, Côte d’Ivoire. 17 Department of Bacteriology and Virology, Souro Sanou University Hospital, Bobo-Dioulasso, Burkina Faso.
18 G5 Evolutionary Genomics of RNA Viruses, Department of Virology, Institut Pasteur, Paris, France. 19 National Reference Center for Respiratory Viruses,
Institut Pasteur, Paris, France. 20Department of Infectious Disease Epidemiology, Robert Koch Institute, 13353 Berlin, Germany. 21 Epidemiology of Highly
Pathogenic Microorganisms, Robert Koch Institute, 13353 Berlin, Germany. 22These authors contributed equally: Tamara Kaleta, Lisa Kern, Samuel
Leandro Hong. ✉email: marcus.panning@uniklinik-freiburg.de; jonas.fuchs@uniklinik-freiburg.de
NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 1
1234567890():,;
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
The continuing pandemic spread of SARS-CoV-2, the cau- E484K substitution in the RBD has been detected in multiplesative agent of coronavirus disease 2019 (COVID-19), has countries. The E484K amino acid change is also found in othera devastating global impact on life, health care systems and VOCs/VOIs and has been shown to reduce antibody
economies by causing significant morbidity and mortality in the neutralization16. A prominent amino acid change in the S protein
human population. SARS-CoV-2 is an enveloped, positive-sense of the B.1.617.2 variant is L452R that is also found in various
single-stranded RNA virus and infects host cells via binding of other lineages. This mutation was shown to enhance infectivity
the viral spike glycoprotein (S) to the angiotensin-converting in vitro and decrease neutralization by sera of COVID-19 patients
enzyme 2 (ACE2) receptor and proteolytic activation through and vaccinees17,18.
cellular proteases1,2. The mature S protein is cleaved into two Here, we describe the detection, inferred origin and phenotypic
subunits S1 and S2 and organized as a homotrimer in the viral characteristics of SARS-CoV-2 lineage A.27, which was primarily
particle3. While S1 forms a globular structure essential for ACE2 identified in Germany19 and France20 in spring 2021. This variant
binding, S2 mediates membrane fusion. Both the receptor-binding emerged in late 2020 and spread to over 30 countries. Through
domain (RBD) and the N-terminal domain (NTD)4 are targeted travel history-aware phylogeographic reconstruction, we estimate
by neutralizing antibodies in sera of convalescent and vaccinated Western Africa as the likely origin of this lineage, from where it
individuals5,6. Thus, multiple RBD-specific monoclonal anti- spread to other regions. A.27 is characterized by a mutational
bodies (mAb) are assessed in clinical trials or are approved to profile including L18F, L452R and N501Y in the S protein that
treat COVID-19, including Bamlanivimab (LY-Cov-555) in combines genetic changes found in various VOCs/VOIs, while
combination with Etesevimab (LY-COV016)7 and the REGN- lacking the D614G substitution present in most other lineages.
COV2 mAb cocktail (REGN10933 and REGN10987)8. Our data demonstrate that A.27 can partially escape the neu-
Early in the pandemic, SARS-CoV-2 acquired the S D614G tralization by sera of vaccinees and recovered COVID-19 patients,
substitution that has been associated with increased transmissi- and by therapeutic mAbs. This study emphasizes the importance
bility and set the genetic foundation for the large number of B.1 of continued molecular surveillance to quickly detect antibody
derived lineages9,10. As the pandemic progressed the genomic escape variants that threaten global vaccination efforts.
diversity of SARS-CoV-2 increased significantly and several var-
iants of concern (VOCs) and variants of interest (VOIs) emerged.
These variants may be associated with higher transmissibility, can Results
lead to more severe disease and/or significantly escape from Detection of SARS-CoV-2 lineage A.27. In early 2021, sequen-
antibody-mediated immunity, thereby reducing the effectiveness cing of material from SARS-CoV-2 infected individuals in Ger-
of available vaccines and treatments with mAbs11–13. Prominent many increased with about one third of sequences generated in
examples are the B.1.1.7 (Alpha) and B.1.617.2 (Delta) variants the south-western state Baden-Wuerttemberg (BW) located at the
that dominated global infections in late 2020 and 2021. These French border (Fig. 1a/b). Within the molecular surveillance
variants are characterized by specific patterns of concerning S program of the Robert Koch Institute (RKI, Public Health Insti-
mutations: apart from the D614G substitution, lineage B.1.1.7 has tute Germany), 851/178.264 (0.48%) sequences were classified as
the N501Y amino acid substitution associated with increased lineage A.27 from 18th of January to 1st of June 2021. The A.27
affinity to ACE214,15 and two deletions in the NTD, among other lineage was initially defined in January 2021 following an out-
21
changes. Moreover, a B.1.1.7 sub-lineage with an additional break in Mayotte , an overseas region of France located in the
Indian ocean between the coast of Mozambique and Madagascar.
a Sequences generated in Germany in 2021 b c
SH
d
BW
Overall SARS-CoV-2
sequences
20,000
40,000
A.27
5 50 500
Fig. 1 Detection of lineage A.27 in Germany. a Map of Germany displaying the cumulative number of all SARS-CoV-2 sequences (colors) and
A.27 sequences (circles) generated in different federal states between January and June 2021. BW—Baden-Wuerttemberg, SH—Schleswig-Holstein. b, c
Temporal distribution of sequences for each calendar week for (b) all or (c) A.27 sequences in Germany and the two federal states BW and SH. d World
map displaying the cumulative A.27 sequences obtained from the GISAID and RKI databases. Source data are provided as a Source Data file.
2 NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
Most of the German A.27 sequences originated in BW (81.4%), West Africa and multiple separate introduction events in
and in the northern state of Schleswig-Holstein (SH) (6.8%) Germany and France, leading to the different large German and
(Fig. 1a). In the beginning of 2021, the frequency of A.27 cases French clades (Fig. 2c and Supplementary Fig. 4). Other African
steadily increased, reaching up to 6% of all sequences generated in regions could have been the source for seeding introductions into
BW and SH (Fig. 1c). Afterwards, the relative detection rate of Europe, such as Mayotte seeding into France, although this was
this variant decreased while the frequency of VOC B.1.1.7 con- not consistently supported (Supplementary Data 1). However, we
sistently increased (Supplementary Fig. 1). From December 2020 also inferred introductions from France to the Benelux Union and
to June 2021, A.27 was reported in 31 countries. Apart from the Northern Africa, as well as introductions into Asia-Pacific
German sequences, 535 A.27 sequences had been deposited in the (APAC) from the Benelux Union. Although no A.27 sequences
GISAID database22. The earliest sequences were reported in were available to us from Southern Africa, our travel history-
Denmark (Europe) in December 2020 and the Western African aware phylogeographic analysis showed consistent and strong
countries Senegal, Burkina Faso and Togo. Interestingly, the support for a spread to this region from Western Europe, which
majority of A.27 sequences deposited in GISAID between January in turn led to an introduction into regions of the Benelux Union.
and April 2021 originated from France (n= 263) indicating a This bi-directional exchange of lineages between African regions
similarly rapid spread in France compared to Germany (Fig. 1d). and the European continent can also be observed for Northern
and Central Africa. This is in contrast with Western and Eastern
Africa, which were seen as exclusive sources of lineages in Europe
Phylogeographic analyses reveal West Africa as the likely ori- in our reconstruction. In conclusion, A.27 likely originated in
gin of A.27. To further characterize the global spread of A.27 and Western Africa in late 2020, from where it spread to multiple
to estimate the potential origin of this lineage, we performed countries around the globe resulting in large clusters in Germany
maximum-likelihood (ML) phylogenetic and Bayesian phylo- and France in spring 2021.
geographic analyses incorporating available travel data of 12 A.27
infected patients (Supplementary Table 1). A preliminary
unrooted ML phylogenetic analysis of 1386 complete A.27 gen- Proportion of hospitalized patients infected with A.27 or
omes and a representative set of 2516 sequences from an Africa- B.1.1.7 in Germany. As part of the German molecular surveil-
focused Nextstrain build, capturing the global SARS-CoV-2 lance program, sequences uploaded to the RKI are linked to case-
diversity, showed suf cient temporal signal in a root-to-tip based data reported by local public health authorities into thefi
regression analysis. Therefore, we estimated a global time- electronic surveillance system for infectious disease. We com-
calibrated phylogeny, which dated its most recent common pared available patient data of 329 sequenced A.27 and 56,453
ancestor (tMRCA) to the second half of November 2020. The patient specimens of VOC B.1.1.7 between January and June of
predicted evolutionary rate of 7.60e−04 substitutions per site and 2021. Interestingly, in the set of sequences that were randomlyselected and limited to not fully vaccinated patients, those
year was in line with previous estimates for SARS-CoV-223. We
subsequently performed a more targeted ML phylogenetic ana- infected with A.27 (n= 100) were on average 5.6 years older thanB.1.1.7 (n= 17.512) infected individuals (Fig. 3a). There was no
lysis on a subtree of the inferred global phylogeny that contained
all A.27 and 25 non-A.27 genomes. The analysis of this gender difference in terms of A.27 patients compared to B.1.1.7patients (Fig. 3b). Accordingly, we compared the proportion of
subset also had a sufficient temporal signal and a time calibrated
tree showed that A.27 is monophyletic and further diversi ed in hospitalized B.1.1.7 and A.27 infected patients. We found nofi significant differences in the proportion of hospitalization, 8.9%
early 2021 (Supplementary Fig. 2). Location-specific clusters
within Germany or France indicated possible independent for B.1.1.7 and 6.2% for A.27, respectively (Fig. 3c). Analysis ofA.27 and B.1.1.7 infections shows, that A.27 infection also pre-
introductions of A.27 into Europe.
Subsampled travel history-aware Bayesian phylogeographic ferentially occur in older individuals with an increasing risk forhospital admissions by age.
analyses of the A.27 subtree (Supplementary Fig. 2) using BEAST
1.10.524 estimated the ancestral origin of the A.27 lineage in
Western Africa (Fig. 2a and Supplementary Fig. 3). The A.27 is characterized by a mutation profile similar to current
subsampling was performed to take the sampling bias toward VOCs and VOIs. We characterized the mutational profile of A.27
the high proportion of sequences from Germany and France into based on 1386 full genome A.27 sequences. The nucleotide pro-
account (Fig. 2b). This analysis predicted a tMRCA for A.27 in files were determined in comparison to Wuhan-Hu-1 using
late September 2020 (95% Highest Posterior Density interval covSonar (https://gitlab.com/s.fuchs/covsonar) and aligned to
(HPD) ranging between mid-August 2020 and late October 2020) each other (Supplementary Fig. 5). Lineage-specific mutations
with an evolutionary rate of 8.15e−04 (95% HPD: [7.04e−04; were defined as mutations present in ≥75% of all sequences
9.33e−04]) substitutions per site and year. The earliest introduc- (Table 1). The A.27 lineage is characterized by 26 mutations
tion of A.27 into Germany likely occurred in the third week of including seven non-synonymous mutations in the S gene, a
November 2020 (95% HPD covering the second half of frameshift in ORF3a and a deletion in ORF8. The frameshift in
November 2020), while the introduction into France happened ORF3a is located at the C-terminus in a region that has so far not
slightly earlier around the beginning of November 2020 (95% been resolved in available cryo-electron microscopy analyses25
HPD between early October 2020 and mid-November 2020). This and leads to a 14 amino-acid truncated protein. The deletion in
places the estimated tMRCAs a few months before the first ORF8 also resides at the C-terminal end and translates into a
confirmed cases in both countries (January 4th and January 6th of deletion of an aspartate and phenylalanine involved in the sta-
2021 for France and Germany, respectively). Therefore, we bilization of the ORF8 dimerization interface26 (Supplementary
estimate that A.27 was introduced into Europe between 6 and Fig. 6). Three of the seven S substitutions, L18F, L452R, and
8 weeks after its common tMRCA. N501Y, are of particular interest (Fig. 4a–c). The L18F amino acid
We estimate that A.27 was initially transmitted within Western substitution is found within the first of five loops of the NTD
Africa (Fig. 2a and Supplementary Fig. 3) before spreading to supersite5,27 and the L452R and N501Y mutations are located in
other regions. The spread from West Africa was estimated by the the receptor-binding motif (RBM) which interacts with the
expected number of transitions between all regions (Markov human ACE2 protein1,4,28. Both regions are targets for neu-
jumps). We confirmed a large number of introductions out of tralizing antibodies27,29,30 and L18F and L452R have been
NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 3
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
a b 4035
30
25
20
15
10
5
0
South Africa
Mali
Denmark
Senegal
Rwanda
Togo
Burkina Faso
United Kingdom
Netherlands
USA
Indonesia
France
Nigeria
Switzerland
Germany
Mayotte
Belgium
Cote d’Ivoire
Turkey
Cameroon
Spain
Benin
Tunisia
Slovenia
Ireland
Italy
Luxembourg
Sweden
Austria
India
Australia
Greece
Pakistan
Jan Feb Mar Apr May Jun
2021 2021 2021 2021 2021 2021
c Benelux APAC
France
Northern Europe
Northern Africa
Eastern Africa Western Africa
Middle Africa North America
Western Africa Eastern Africa
North America
APAC Germany
Benelux
Germany Western Europe
France Eastern Africa
Southern Europe Northern AfricaNorthern Europe Benelux
Eastern Europe Eastern Europe
Northern Europe Southern Europe
Middle Africa Southern Europe
Western Europe Southern Africa
Middle Africa
France
Oct Nov Dec Jan Feb Mar Apr May Jun Eastern Europe
Germany
2020 2021 Northern Africa
Western Europe Southern Africa
Fig. 2 Origins of SARS-CoV-2 lineage A.27. a Region-annotated phylogeny, showing an inferred origin of the A.27 lineage in Western Africa with
subsequent spread to most of the other regions where A.27 was detected. Smaller white circles represent posterior support >0.5, whereas bigger black
circles represent posterior support >0.95. b Known locations, dates and individual travel histories of A.27 cases. Rows show the collection dates of
genomes on the bottom, as well as the frequency of genomes as a bar plot at the top. The origin and destination is shown for travel cases: genomes with
associated travel history are outlined with the color corresponding to the origin location, and are connected to this origin location with a smaller dot. c
Sankey plot showing the number of transitions between locations through the estimated number of Markov jumps. The thickness of the lines are
proportional to the number of Markov jumps from the location to the left into the location on the right, conditional on the corresponding Bayes factor being
higher than 3, pointing to strong support for an inferred origin of the A.27 lineage in Western Africa. Input XML files of the phylogeographic analysis are
supplied in the Supplementary Data 1.
previously associated with antibody escape and L452R addition- from a patient living in the Black Forest area in South Germany
ally with increased infectivity17,18,31. Furthermore, N501Y was and who was treated at the University Medical Center of Frei-
suggested to enhance the binding affinity to ACE214,15. These burg. Virus isolation was performed on VeroE6 cells and followed
three mutations are also found in multiple VOCs and VOIs by one cell culture passage to produce high titre stocks. We
(Fig. 4c). L18F is present in the VOCs B.1.351 and P.1, the L452R performed whole genome sequencing of the initial patient
substitution is found in high frequencies in B.1.617.2 and related material and the derived virus stocks to analyze if the virus had
AY lineages, and N501Y is known from B.1.1.7, B.1.351 and P.1. acquired cell culture adaptations during isolation. Analysis of the
One of the hallmarks of A.27 is the absence of the S D614G variant frequencies found in the respective samples showed a
substitution present in the globally dominating B.1-derieved ~60% variant frequency for the G11083T substitution in the
lineages, indicating an independent acquisition of the other spike ORF1ab after isolation. Although likely selected during isolation,
mutations. this mutation was already present in low frequencies in the
patient material (Fig.5a). Moreover, the virus isolate exhibited
The A.27 Black Forest isolate is attenuated in vivo. To char- high genomic stability throughout the isolation process and all
acterize the biological features of the A.27 lineage, we isolated this lineage-defining mutations were confirmed. The A.27 Black
variant from an oropharyngeal swab. The sample was derived Forest isolate reached similar titres in both VeroE6 and Calu3
4 NATURE COMMUNICATIONS |        ( 2022)1 3:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
total genomes
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
a b c
Fig. 3 Metadata analysis of patients infected with A.27.Metadata of confirmed A.27 and B.1.1.7 infected patients was acquired from the RKI and GISAID.
a Scatter plot of the age distribution of patients infected with B.1.1.7 or A.27 patients. Displayed is the mean age. Statistical analysis was performed with a
two-sided t-test (**p < 0.01). b Gender distribution and (c) hospitalization rate of B.1.1.7 and A.27 patients. Statistics were performed with a two-sided
Fisher’s exact test (ns— not significant). Source data are provided as a Source Data file.
Table 1 Consensus mutations of lineage A.27 (n= 1386). A.27 infected cells indicating that the missing C-terminal amino
acids do not impact its localization. This is in line with previous
results showing a comparable cellular localization of wild type
75% of 90% of gene amino acid ORF3a and a C-terminal deletion mutant of ORF3a34. A previous
sequences sequences substitution
study showed that ORF3a represents an important virulence
A361G A361G ORF1a synonymous factor in human ACE2 transgenic (hACE2) mice35. Therefore, we
C1122T C1122T ORF1a P286L hypothesized that the ORF3a frameshift in the A.27 isolate could
C2509T C2509T ORF1a synonymous
C8782T C8782T ORF1a synonymous lead to an attenuated phenotype in vivo. To test this hypothesis,
A9204G A9204G ORF1a D2980G we compared the pathogenicity of B.1, B.1.1.7 and A.27 isolates in
A11217G A11217G ORF1a N3651S hACE2 transgenic mice
36. Mice were infected with an intranasal
C16293T – ORF1b synonymous inoculum containing 132 plaque forming units (pfu) per virus
C16466T C16466T ORF1b P5401L variant and weight loss and survival were monitored. B.1 and
A18366G A18366G ORF1b synonymous B.1.1.7 infected mice showed pronounced weight loss and all mice
A20262G A20262G ORF1b synonymous reached humane endpoints between 6 and 7 days post infection
C21614T C21614T S L18F (Fig. 5e/f). Intriguingly, 75% of A.27 infected mice only tran-
G22468T G22468T S synonymous siently lost weight and recovered from the infection demon-
T22917G T22917G S L452R strating that A.27 is severely attenuated in vivo compared to the
A23063T A23063T S N501Y B.1 and B.1.1.7 isolates, possibly due to the deletion in ORF3a.
C23520T – S A653V
C23525T C23525T S H655Y
G23948T G23948T S D796Y A.27 escaped neutralization by patient sera and therapeutic
G25218T G25218T S G1219V antibodies. Lineage A.27 has two mutations in its RBD that
T25541C T25541C ORF3a V50A translate to L452R and N501Y. Previous binding studies of RBD
del:26160:8 – ORF3a del257/258fsX6 mutants with mAb and sera showed decreased binding for both
C27247T C27247T ORF6 synonymous mutations6,31 (Supplementary Fig. 7a/b). To estimate the effect of
T28144C T28144C ORF8 L84S the mutations found in the S gene of A.27 in the context of virus
del:28248:6 del:28248:6 ORF8 del:119/120 neutralization, serial dilutions of sera from convalescent COVID-
A28273T A28273T NCR synonymous 19 patients (Fig.6a and Supplementary Fig. 8) or BioNTech
G28878A G28878A N S202N
G29742A NCR synonymous BNT162b2 vaccinees (Fig. 6b and Supplementary Fig. 9) were–
analyzed by plaque reduction assays. The potential escape was
assessed by comparing the A.27 Black Forest isolate to the pro-
totypic B.1 isolate that harbors the D614G mutation. The neu-
cells, comparable with a prototypic B.1 isolate (Muc-IMB-1)32 tralizing titres resulting in 50% plaque reduction (NT50) of sera
that only harbors the S D614G substitution in its viral genome from convalescent COVID-19 patients and from vaccinees were
and four different VOCs (Fig. 5b/c). A clear exception was the significantly reduced, on average, two- to three-fold lower against
B.1.351 variant, which showed a 100-fold reduced viral titre the A.27 isolate compared to the B.1 isolate. Notably, the resis-
3 days post infection compared with A.27 on Calu3 cells (Fig. 5c) tance of A.27 toward antibody neutralization was similar in either
as previously reported33. Furthermore, cells infected with the B.1 group as there were no significant differences in the NT50 values
and A.27 isolates were analyzed by confocal immunofluorescence between convalescent and vaccinee sera (Fig. 6c). Furthermore,
microscopy (Fig. 5d). Both virus isolates showed a diffuse cyto- we assessed the escape of A.27, B.1 and four different VOC iso-
solic accumulation of the S and N proteins 8 h post infection. The lates from the neutralizing capacity of the mAbs LY-COV55537
frameshift in ORF3a in the A.27 isolate might result in an altered and the REGN-COV2 cocktail (REGN10933, REGN10987)8
cellular localization of this viral protein. Therefore, we addition- which can be used to treat COVID-19 patients. In strong contrast
ally stained for ORF3a, but found no differences between B.1 and to B.1 and B.1.1.7, the A.27 as well as the B.1.351, B.1.617.2 and
NATURE COMMUNICATIONS |         (2022)1 3:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 5
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
a SARS-CoV-2 spike (A.27)
S1 S2
FP TM
S1/S2 cleavage
NTD RBD RBM HR1 HR2 CT
b
L18F (not in structure)
A653V N501Y
H655Y 90°
L452R
D796Y NTD
RBD
RBM
G1219V (not in structure)
c >0% 100 %
not detected Mutation frequency
Fig. 4 A.27 shows a mutational pattern in its viral spike gene that resembles other VOCs and VOIs. a Schematic overview of the SARS-CoV-2 spike
gene and the non-synonymous mutations found in more than 75% of A.27 sequences (n= 1386) compared to the S of Wuhan-Hu-1 sequence
(NC_045512.2). NTD–N-terminal domain, RBD–receptor-binding domain, RBM–receptor-binding motif, FP–fusion peptide, HR1/2–heptad repeat region,
TM–transmembrane domain, CT–C-terminal domain. b The spike protein of SARS-CoV-2 (PDB accession number: 6vxx) with the mutations displayed in
(a) is shown in surface presentation. c Comparison of spike mutations present in A.27 and the different VOCs and VOIs. The mutation frequencies of the
different lineages were downloaded from outbreak.info70 (accessed on 2021-07-13) and the A.27 mutation frequencies replaced with the frequencies
calculated based on the 1386 sequences used in this study. The heatmap was visualized with the R pheatmap package. Source data are provided as a
Source Data file.
P.1 isolates completely escaped the neutralizing effect of LY- an overall broad and strong neutralizing capacity while LY-
COV555 (Fig. 6d). For REGN10933, B.1.351 and P.1 displayed a COV555 failed to neutralize most variants (Table 2). Besides
pronounced escape (Fig. 6e). Furthermore, the neutralizing neutralization, a prime function of antigen-bound (complexed)
capacity of REGN10987 against A.27 and B.1.617.2 was slightly IgG is the activation of Fc-gamma receptors (FcγRs) present on
reduced (Fig. 6f). The observed differences for the individual various immune cells such as monocyte-derived cells and Natural
REGN-COV2 antibodies could be compensated by a 1:1 Killer (NK) cells. One of the most potent antiviral immune
combination of both antibodies, mimicking the actual treatment mechanisms mediated by the Fc-part of complexed IgG (Fcγ) is
regimen8 (Fig. 6g). Based on the NT50 values, REGN10987 showed antibody-dependent cellular cytotoxicity elicited by NK cells
6 NATURE COMMUNICATIONS |    (     2022)1 3:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
L18F 13
305
319
L452R
N501Y
541
A653V
H655Y
D796Y 788
806
912
984
1163
G1219V 1213
1237
1273
VOC VOI
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
a ORF1a/b S ORF3a ORF6 ORF8 N
effect
96 98 97 94 98 96 97 11 59 97 26 97 97 98 98 99 98 97 99 91 98 98 98 97 98 84 86 99 97 0 97 99 98 97 96 virusstock (P2)
95 97 96 97 97 96 97 10 58 97 22 97 96 97 97 98 98 98 99 94 98 97 96 96 97 73 86 98 96 12 96 99 98 96 96 virus isolate (P1)
94 97 96 96 97 94 96 0 10 97 0 96 96 96 96 99 100 96 98 89 97 96 97 95 98 87 86 98 97 0 95 99 98 97 97 patient swab
variant frequency (%)
non−coding syn non-syn deletion frame shift
0 100
b VeroE6 c Calu3 d
mock B.1 A.27
N
S
e f
F-Actin DAPI viral protein
Fig. 5 A.27 Black Forest isolate displays all lineage-defining mutations and is attenuated in vivo. a Variant frequency plot (https://github.com/jonas-
fuchs/SARS-CoV-2-analyses) of the variant frequencies detected by next-generation sequencing in the oropharyngeal swab of a patient from the Black
Forest area in Germany infected with A.27, after virus isolation (passage 1/P1) and virus cultivation (passage 2/P2) on VeroE6 cells. The variant
frequencies in comparison to Wuhan-Hu-1 (NC_045512.2) are plotted as a heatmap with the respective frequencies indicated as %. Mutations present in
75% of A.27 sequences are marked bold. b, c Viral growth kinetics of the A.27/Freiburg isolate (P2) in comparison to the virus isolates B.1, B.1.1.7, B.1.351,
P.1 and B.1.617.2 in (b) VeroE6 or (c) Calu3 cells. Confluent cells were infected (moi of 0.001) and the cell supernatant was harvested after 8, 24, 48 and
72 h post infection. Viral titres were determined by plaque assay on VeroE6 cells. The log-transformed titres are shown as means ± SD from three
independent experiments. Significance was determined in comparison to A.27 via a two-way ANOVA with a Tukey´s multiple comparison test, *p < 0.05,
**p < 0.01. d Confocal fluorescence microscopy analysis of B.1 or A.27 infected VeroE6 cells (moi 0.1) 8 h post infection. Fixed cells were stained with
SARS-CoV-2 N, S and ORF3a specific antibodies (red). In addition, F-actin (phalloidin, white) and nuclear DNA (DAPI, blue) were detected. Shown are
representative pictures of two independent experiments. Scale bars indicate 10 µm. e, f Weight loss (e) and survival (f) of hACE2 transgenic mice infected
intranasally with 132 pfu of the A.27 (n= 7), B.1 (n= 5) or B.1.1.7 (n= 7) isolates were monitored daily (mean ± SEM). Significance in weight loss was
determined in comparison to B.1 via a two-way ANOVA with a Tukey´s multiple comparison test, *p < 0.05, ***p < 0.001. Significance for the survival was
calculated with a Log-rank (Mantel–Cox) test (ns—not significant, ***p < 0.001). Source data are provided as a Source Data file.
expressing FcγRIII/CD16. Therefore, we assessed the potential of and neutralizing capacity. Considering that all mAbs were able to
the above mAbs to activate CD16 in a cell-based FcγR-activation activate CD16, this showed that residual binding of the therapeutic
reporter assay. Inactivated virions from different strains were antibodies to the antigen could still result in CD16 activation.
titrated and immobilized on ELISA plates and incubated with the Taken together, these data argue for a pronounced escape of A.27
respective mAbs at a fixed concentration. CD16 reporter cells were from neutralizing antibodies similar to other VOCs.
then cultured on opsonized virions and IL-2 production was
measured as an indicator of receptor activation as described
previously38. Directly immobilized mAbs served as a positive Discussion
control and showed that all mAbs are equally able to activate Here, we report the epidemiology, inferred origin and phenotypic
CD16 (Fig. 6h). Incubation on opsonized A.27 and B.1.617.2 characteristics of SARS-CoV-2 lineage A.27, whose genome
virions resulted in reduced CD16 activation for LY-COV555 in contains several substitutions that have also been observed in
line with the neutralization data (Fig. 6i). However, for the other various VOCs and VOIs. The root of the pandemic lies in the
isolates there was no direct correlation between CD16 activation parental A lineage with the characteristic C8782T and T28144C
NATURE COMMUNICATIONS |    (     2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 7
ORF3a
G29742A
G28878A (S202N)
A28273T
del:28248:6 (del:119/120)
T28144C (L84S)
C28087T (A65V)
del:27386:1
C27247T
del:26160:8 (del257/258fsX6)
A25910G (D173G)
C25731T
T25541C (V50A)
G25218T (G1219V)
C23948T (D796Y)
C23525T (H655Y)
C23520T (A653V)
A23063T (N501Y)
T22917G (L452R)
G22468T
C21614T (L18F)
A20262G
A18366G
C16466T (P5406L)
C16293T
C11750T (L3829F)
A11217G (N3651S)
G11083T (L3606F)
G11082T (L3606)
A9204G (D2980G)
T8941G
C8782T
C2509T
A1461C (K399T)
C1122T (P286L)
A361G
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
a convalescent b vaccinees c
d LY-COV555 e REGN10933 f REGN10987 g REGN10933 + REGN10987
(REGN-COV2)
h i
Fig. 6 Neutralizing capacity of sera and therapeutic monoclonal antibodies against A.27. a Neutralizing activity of convalescent sera (n= 16) or (b) sera
from BioNTech BNT162b2 vaccinees (n= 14) against B.1 (gray) or A.27 (red). 100 pfu of each virus was incubated with serial twofold sera dilutions and
analyzed by plaque assay. The curve fits (light color) and the mean curve fit (dark color) were plotted (left panel) and neutralization titres calculated (right
panel). Statistics were performed with a paired, two-tailed t-test (**p < 0.01, ***p < 0.001). c The fold difference of the NT50 between B.1 and A.27 was
calculated for the analyzed convalescent sera (n= 16) and sera from vaccinees (n= 14). Shown are the individual values and the geometric mean. Statistics
were performed on log2-transformed values with a two-tailed t-test (ns—not significant). d–g Neutralizing capacity of the therapeutic antibodies (d) LY-
COV555, (e) REGN10933, (f) REGN10987 or (g) the combination of REGN10933 and REGN10987 in a 1:1 ratio. Serial tenfold dilutions of the monoclonal
antibodies were incubated with 100 pfu of the different viruses and analyzed by plaque assay. Plotted are the curve fits and mean ± SD of three independent
experiments. Statistics were calculated with a two-way ANOVA (Tukey’s multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001). h, i CD16 reporter
cell IL-2 production was quantified via anti-mIL-2 ELISA. REGN10987, REGN10933 or LY-COV555 were (h) titrated and immobilized or (i) inactivated
virions of different SARS-CoV-2 isolates were titrated, immobilized and opsonized with 20 ng/µl of IgG. In (e) the mean of two independent experiments is
displayed. The dotted horizontal line in (h) represents the ELISA background set to 1. For (i) the area under the curve was calculated from virion titration
(undiluted, 1:10,1:100). Shown are mean and standard deviation of three independent experiments. Statistics were performed with a one-way ANOVA
(Tukey’s multiple comparison test, ns—not significant, *p < 0.05, **p < 0.01). Source data are provided as a Source Data file.
mutations. However, lineages derived from A are rare, with only a The early detection in multiple countries in Western Africa in late
few thousand sequences reported worldwide. In the COVID-19 December 2020, an outbreak in January 2021 in Mayotte and
pandemic, infections are currently dominated by B.1-derived A.27 infections of Belgian military personnel returning from Mali
lineages harboring the prominent D614G mutation9. This makes pointed to a suspected origin in the African continent41,42.
the mutational pattern of lineage A.27 of particular interest as it Through carefully crafted phylogeographic analyses that exploit
shows the independent acquisition of the same mutations found individual travel histories of patients infected with A.27, we here
in B.1-derived VOCs and VOIs in a different genomic back- provide support for the origin of A.27 in Western Africa. After
ground. A similar pattern of concerning mutations in an emer- completion of our phylogenetic and phylogeographic analyses, on
ging A-derived lineage has so far only been described for the the 17th of September 2021, three additional A.27 genomes from
A.23/A.23.1 lineage first discovered in Uganda and Rwanda39,40. Burkina Faso appeared on GISAID with sampling dates on the 16
8 NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
Table 2 NT values of the different monoclonal antibodies B.1.1.7 cases, but a significant mean higher age. However, these50
against the tested variants. observations were clearly limited by the small amount of available
metadata. To investigate if both viruses had a comparable viru-
lence, we analyzed the growth of the A.27 isolate in cell culture as
LY-COV555 REGN10933 REGN10987 REGN-COV2 well as its pathogenicity in hACE2 transgenic mice. The A.27 and
A.27 > 10 µg/ml 0.12 µg/ml 0.05 µg/ml 0.05 µg/ml B.1.1.7 isolates demonstrated a comparable viral replication in cell
B.1 0.11 µg/ml 0.09 µg/ml 0.02 µg/ml 0.03 µg/ml culture but A.27 was significantly attenuated in mice compared to
B.1.1.7 0.06 µg/ml 0.02 µg/ml 0.01 µg/ml 0.02 µg/ml B.1 and B.1.1.7. This indicates that despite comparable viral
B.1.351 > 10 µg/ml 4.70 µg/ml 0.01 µg/ml 0.05 µg/ml
µ µ µ µ replication in cell culture, A.27 possesses features which decreaseP.1 > 10 g/ml >10 g/ml 0.01 g/ml 0.02 g/ml
B.1.617.2 > 10 µg/ml 0.01 µg/ml 0.04 µg/ml 0.02 µg/ml its pathogenicity in mice. Genomic changes like the frameshift in
ORF3a and the deletion in ORF8 might contribute to this phe-
notype. The lack of position 119/120 in the accessory protein
and the 19th of December 2020. They additionally confirm the ORF8 might lead to a decreased stability of the homodimer and a
early circulation and our inferred origin of A.27 in West Africa partial loss of function26. Interestingly, B.1.617.2 lacks the two
toward the end of 2020. The entire backbone of the A.27 phy- ORF8 amino acids 120/21, strongly indicating a convergent
logeny was estimated to be located in Western Africa, with the evolution for deletions in this region or the lack of a selective
virus consequently spreading from there to other, mostly Eur- pressure to maintain these amino acids. ORF8 seems to be dis-
opean, regions. From Western Africa, we observed that pensable for viral replication and deletions/stop codons could
A.27 spread to Europe through multiple separate introduction represent an adaptation to the human host45. ORF3a has been
events, eventually forming several large German and French shown to induce apoptosis and block autophagy34,46. Our
clades. Interestingly, the Markov jump analysis did not support immunofluorescence analysis suggests a similar cellular localiza-
bi-directional seeding events between the two adjacent countries, tion and expression of ORF3a in B.1 and A.27 infected cells.
which was further supported by the clear phylogenetic separation However, the frameshift in ORF3a might impair some of the
of the French and German clades. Note that due to the effect of proteins’ functions. An attenuation due to a crippled ORF3a
sampling bias, for example as a result of varying sequencing would be in line with a previous study showing that ORF3a and
efforts between countries, the major German and French clades ORF6 are the major contributors of viral pathogenesis in hACE2
do not necessarily signify that the spread of A.27 was largely transgenic mice35. Implications of these amino acid changes for
confined to these two countries. This was confirmed by the human disease are presently unclear. Functional characterizations
individual travel histories we collected, with two infected patients of mutations in accessory viral proteins of SARS-CoV-2 are
testing positive in the Netherlands after traveling back from South urgently needed to better understand their impact on virulence
Africa, although South Africa did not yet report any A.27 and pathogenicity of SARS-CoV-2 in humans.
genomes. Vaccinations are currently the major instrument to combat the
While the origin of most French A.27 clades also lay in Wes- COVID-19 pandemic and treatments with mAb are a potent
tern Africa, one of the larger French clades seemed to be more therapeutic option to treat COVID-19. As such, antibody escape
closely related to Eastern-African sequences, a fact also corro- mutations in circulating variants could further fuel the pandemic.
borated by the Markov jump analysis. This Eastern-African clade A.27 harbors multiple mutations in the viral S, the major target
corresponds to an outbreak in Mayotte, a French overseas terri- for neutralizing antibodies, raising the question whether mAb and
tory in the Indian Ocean, indicating the possibility that one or sera from COVID-19 patients or from vaccinees will protect from
more travel cases from Mayotte to France led to the introduction an A.27 infection. Our data suggest that A.27 can escape
of A.27 into France. Notably, only one out of five replicates (see antibody-mediated immunity. We observed a consistent decrease
Supplementary Data 1) of our phylogeographic analysis provides of the neutralizing capacity of sera from COVID-19 patients and
strong support (Bayes factor > 20) for A.27 being introduced into vaccinees. Moreover, A.27 completely escaped the neutralization
France via Mayotte, with the other four replicates showing of LY-COV555 and partially of REGN10987. Importantly,
positive support (Bayes factor > 3). Known travel cases from B.1.617.2 escaped these antibodies in a similar manner, suggesting
Mayotte to France of patients infected with A.27 could have led to that the L452R mutation present in both variants facilitates this
more conclusive evidence, but unfortunately, we were not able to escape47. This is in agreement with deep mutational scanning
obtain such individual travel histories for the purpose of our data that showed decreased binding of L452R and LY-COV55531.
travel history-aware phylogeographic analyses. Of note, we were Interestingly, the VOCs B.1.351 and P.1 escaped LY-COV555 and
also not able to obtain travel history for the earliest A.27 sample REGN10933 which was likely facilitated by the E484K mutation
in Denmark which we assume to be a travel case given that all of present in both variants. This suggests that L452R and E484K
the other early A.27 samples came from West Africa. lead to an escape from LY-COV55531 and to a partial resistance
Within Germany, particularly in BW, we observed a constant to either REGN10987 or REGN10933, respectively. The FcγR
increase in A.27 sequences over several weeks followed by a rapid activation assay showed that this escape is to some extend inde-
decline of A.27. Since the increase of B.1.1.7 started earlier, this pendent of the CD16 activation. However, reduced neutralization
decline occurred with the start of the third wave in Germany, can result in reduced CD16 activation as observed for LY-
which was mainly driven by B.1.1.7 and led to its dominance. COV555 and A.27 or B.1.617.2 indicating key immunological
This suggests a fitness advantage of A.27 in comparison to other mechanisms linked to opsonisation can be additionally impaired
lineages but a disadvantage compared to the B.1.1.7 lineage. A in non-neutralizing mAbs. Importantly, the combination of both
possible explanation for this intermediate fitness phenotype could REGN-COV2 antibodies sufficiently neutralized A.27 as well as
be the acquisition of the N501Y mutation in the absence of all tested VOCs, emphasizing that REGN-COV2 but not LY-
D614G. Both mutations are present in VOCs B.1.1.7, B.1.351 and COV555 should be used to treat COVID-19 patients suffering
P.1 and might have additive effects. N501Y increases the affinity from an infection with these variants. This also argues for using
of the viral S for ACE214 and D614G is thought to prevent pre- different mAb preparations when no clear clinical effect is
mature dissociation of the S trimer leading to a higher infectivity observed, as the lineage might harbor mutations that are refrac-
and transmissibility10,43,44. We detected a lower, but not a sig- tory to a particular preparation. Notably, NTD polymorphisms
nificant lower proportion of hospitalization of A.27 compared to might also decrease the neutralizing capacity of sera as multiple
NATURE COMMUNICATIONS |   (      2022)1 3:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 9
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
studies have detected NTD-specific antibodies in COVID-19 locations. We performed this subsampling procedure five times, to exclude the
patients48 and vaccinees49. L18F lies within the first of five loops possibility of accidentally sampling a highly unlikely scenario. This yielded final
of the NTD supersite5,27 and could contribute to the escape from datasets of between 560 and 565 sequences from 31 countries, on which we per-
50 formed travel history-aware phylogeographic reconstruction
24. However, estimat-
antibody-mediated immunity as previously suggested . ing transition rates between locations that have very few sequences may be subject
The present study analyzed the genomic profile and biological to poor mixing61. In order to avoid this issue, we aggregated certain locations into
features of the A.27 lineage which was primarily detected in larger regions (with the categorization based mainly on the UN geoschemes). For
France and Germany in spring 2021. Our phylogeographic ana- example, sequences from Denmark, Sweden, UK and Ireland were grouped toge-
ther as belonging to “Northern Europe”. We refer to Supplementary Table 3 for a
lyses that were able to exploit individual travel histories provided detailed description of which countries were grouped into which regions and how
evidence that these sequences stem from separate introduction many sequences were included in total per region. This process resulted in a total of
events out of Western Africa, which we estimate to be the geo- 14 regions being considered in the phylogeographic analysis: Asia-Pacific (APAC),
graphic origin of A.27. Importantly, our data further suggest that Benelux (Belgium, Netherlands and Luxembourg), Eastern Africa, Middle Africa,
A.27 is less susceptible to SARS-CoV-2-speci c antibodies and Southern Africa, Western Africa, North America, Eastern Europe, Southern Eur-fi ope, Western Europe, France and Germany. We decided not to group France and
that COVID-19 patients and vaccinees might not be fully pro- Germany into a larger region, given that they are countries of interest for this
tected against this variant. The presence of concerning S muta- study. For twelve sequences in our dataset that were sequenced in the Netherlands
tions in an A-derived lineage supports the notion that the and Belgium, we obtained travel information, indicating cases in which a patient
same escape mutations can appear in relatively distant genomic had travelled in the days preceding diagnosis. Two patients had returned fromSouth Africa to the Netherlands, two from Burkina Faso to Belgium (and one other
backgrounds with similar phenotypic consequences. Therefore, of whom a household member returned from Burkina Faso to Belgium and tested
global molecular surveillance has to continue to detect novel positive) and seven from Mali to Belgium (Fig. 2a and Supplementary Table 1).
variants and to support assessing their risk for the human With these regions and individual travel histories in place, we performed travel
population. history-aware discrete phylogeographic analysis
24,59 (using BEAST 1.10.558, while
employing the BEAGLE 3.2.0 high-performance computational library62 to
improve performance. We made use of Bayesian stochastic search variable selection
Methods to simultaneously determine which migration rates are zero depending on the
Data acquisition. To assess the prevalence of lineage A.27 in Germany, all SARS- evidence in the data and to efficiently infer the ancestral locations, in addition to
CoV-2 full-genome sequences that were submitted to the RKI (n= 851) in 2021 providing a Bayes factor test to identify significant non-zero migration rates. We
until the 1st of June (Supplementary Data 2) were classi ed using the PANGO also estimated the expected number of transitions (known as Markov jumps
63)
fi
lineage assignment (pangolin version: 3.0.3, pangoLEARN: 2021-05-27)51,52. The between all regions in the dataset. On the sequence data partition, we made use of a
relative frequency of A.27 and B.1.1.7 sequences in comparison to all submitted general time-reversible substitution model with estimated base frequencies and64
sequences was assessed for each federal state and each week of 2021. Sequencing among-site rate heterogeneity , along with a relaxed molecular clock model with
data was linked to patient metadata obtained from local health authorities as part an underlying lognormal distribution
65. We used the following prior specifications
of the genomic surveillance program of the RKI via the national electronic for these analyses: a non-parametric skygrid coalescent model (for which we66
reporting system for surveillance of noti able infectious diseases (SurvNet)53. employed Hamiltonian Monte Carlo estimation ), a gamma (shape= 0.001;fi
Anonymized data of the hospitalization status, sex and age were extracted. Fur- scale= 1000) prior on the skygrid precision parameter, dirichlet (α1, … αK= 1.0;
thermore, as of 2021-08-24 all additionally available 535 A.27 sequences and K equal to the number of states) priors on the transition rates for the GTR
associated metadata for samples taken until 2021-06-01 were downloaded from substitution model and the frequencies for the GTR nucleotide-substitution model,
GISAID (Supplementary Table 2). The Germany map was downloaded from an exponential (mean= 0.5) prior on the shape parameter of the discretized
https://gadm.org/maps/DEU.html and the maps were visualized using tmap54. gamma distribution to model among-site rate heterogeneity, a Poisson prior
Patient metadata of hospitalization status, sex and age was analyzed with (mean= 13) on the sum of non-zero rates between regions, a CTMC reference
GraphPad Prism. prior on the mean evolutionary rate
67 and an exponential (mean= 1/3) prior on its
standard deviation. For the travel history-aware phylogeographic model, we treated
the departure time of the patient as a random variable, conditioned on a normal
prior distribution with a mean of 10 days before sampling date (based on a mean
Phylogenetic analysis and time-calibrated phylogenetic tree reconstruction. incubation time of 5 days and a constant ascertainment period of 5 days between
To put the A.27 sequences within the global context, we combined all symptom onset and testing68) and a standard deviation of 3 days. We truncated the
851A.27 sequences from the RKI with 535A.27 sequences from GISAID, along with distribution to be positive (back-in-time), in order to avoid an infection time at a
2545 sequences from the Africa-focused Nextstrain build (https://nextstrain.org/ncov/ later date than the corresponding sampling time.
gisaid/africa)55 for a total of 3907 sequences after removing duplicate entries. We Each of these phylogeographic analysis replicates ran for a total of 560 million
limited our selection of sequences from GISAID and Nextstrain to those collected up to iterations, respectively, with the Markov chains being sampled every 50,000th
June 1st to remain consistent with the sampling period of the RKI sequences. We used iteration, in order to reach an effective sample size (ESS) for all relevant parameters
this selection of sequences to infer an unrooted phylogenetic tree using IQTREE2 of at least 200, as determined by Tracer 1.769. We used TreeAnnotator to construct
v2.1.056 under a GTR model with empirical frequencies and four-category FreeRate maximum clade credibility (MCC) trees for each replicate.
model of site heterogeneity, which was selected as the best fitting model using IQTREE’s
ModelTest functionality. The resulting phylogeny was then time calibrated using
TreeTime v.0.7.457, rooting the tree on the “Wuhan/Hu-1/2019” isolate, following the Analysis of A.27 lineage-defining mutations and lineage comparison. The
Nextstrain SARS-CoV-2 workflow (https://github.com/nextstrain/ncov) and assuming a nucleotide and amino acid profiles of the 1386 A.27 sequences were determined in
strict molecular clock and a skyline coalescent model. TreeTime detected one GISAID comparison to Wuhan-Hu-1 using covSonar (https://gitlab.com/s.fuchs/covsonar).
sequence (EPI_ISL_1586901) and three RKI sequences (IMS-10020-CVDP- To extract the nucleotide mutations and define INDELs the R package stringr was
DCAB86B5-00C8-496F-9B16-297546A77DF2, IMS-10122-CVDP-DF8FAF93-3173- utilized. The profiles were matched with the R package dplyr, mutations with a
40A6-85D8-B50274A72B20, IMS-10122-CVDP-5B313DE6-7B34-4BAA-81BD- frequency below 1% were excluded and the resulting matrix visualized with the R
DEEE937126EC) as outliers, which we subsequently removed from further downstream pheatmap package. We extracted mutations that were present in 75% of the
analyses. Such a relatively low number of outliers is to be expected as the Nextstrain mutation profiles and defined them as lineage-defining mutations. Furthermore,
workflow already performs a similar data cleaning step. We visualized the resulting the data produced by covSonar was utilized to compare the A.27 amino acid profile
phylogeny using baltic (https://github.com/evogytis/baltic). in the viral spike gene with different VOCs and VOIs. Here, the amino acid profile
was subset for the viral spike gene and the frequency of the mutations calculated
excluding again frequencies below 1%. The amino acid mutation frequencies in the
Travel history-aware phylogeographic reconstruction. From the full time- viral spike of A.27 and the different VOCs and VOIs was downloaded from
calibrated ML phylogeny, we selected the subtree containing all A.27 sequences, outbreak.info70 on 2021-07-13. Outbreak.info calculates these frequencies based on
along with 25 non-A27 ancestral sequences (n= 1383) to perform a more targeted all available sequence data from GISAID. The A.27 frequencies were replaced with
reconstruction to determine the geographic origin of the A.27 lineage. To this end, our calculated frequencies as they include data from GISAID and RKI and
we aimed to perform a Bayesian phylogeographic reconstruction using BEAST visualized the mutation frequencies in the viral spike again with the R pheatmap
v1.10.558. We note that over 50% of the sequences in the A.27 clade were collected package.
in Germany (n= 884), with France a close second (n= 263) (see Fig. 2b and
Supplementary Fig. 2). Such severe sampling bias is known to affect discrete
phylogeographic reconstruction59, leading to overconfidence in inferring over- Visualization of viral protein structures. The EM structure of the closed of the
sampled locations as ancestral in the phylogeny60. To mitigate this, we employed a trimeric SARS-CoV-2 spike protein (10.2210/pdb6vxx/pdb) and the dimeric 2.04 Å
subsampling scheme where we removed identical sequences and limited our crystal structure of ORF8 (10.2210/pdb7JTL/pdb) were downloaded from the
dataset to a maximum of 10 randomly selected sequences per week for these two protein data bank and visualized with UCSF ChimeraX version: 1.1 (2020-09-09).
10 NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
Cell culture. Virus isolation, cell culture and mouse infection experiments with (Genbank: NC_045512) using BWA-MEM v.0.7.1774. For datasets produced with the
SARS-CoV-2 were performed under Biosafety Level 3 protocols at the Institute of ARTIC v3 protocol, primer sequences were trimmed with ivar trim v1.9 (https://
Virology, Freiburg, approved by the Regierungspraesidium Tuebingen (No. 25-27/ andersen-lab.github.io/ivar/html/manualpage.html). Variants (SNPs and INDELs)
8973.10-18 and UNI.FRK.05.16-29). Adherent African green monkey kidney were called with the ultrasensitive variant caller LoFreq v2.1.575, demanding a
VeroE6 cells (ATCC CRL-1586™) and human lung Calu3 cells (ATCC HTB-55™) minimum base quality of 30 and a coverage of at least 20×. Afterwards, the called
were cultured in 1× Dulbecco’s modified Eagle medium (DMEM) containing 5% or variants were filtered based on a minimum variant frequency of 10 % and on the
10% fetal calf serum (FCS), respectively. To isolate SARS-CoV-2 from patient support of strand bias. The effects of the mutations were automatically annotated in
material, filtered throat swab samples of patients with previous SARS-CoV-2 A.27 the vcf files with SnpEff v.4.3.176. Finally, consensus sequences were constructed with
(EPI_ISL_3200835) or Delta variant B.1.617.2 infections (EPI_ISL_2535433) were bcftools v.1.1.077. Regions with low coverage (>20×) or variant frequencies between 30
inoculated on VeroE6 cells (2 × 106 cells) in 4 ml DMEM with 2% FCS and and 70% were masked with N. The final consensus sequences have been deposited in
incubated at 37 °C and 5% CO2 for 4–6 days until a cytopathic effect was visible. the GISAID database (www.gisaid.org).
The culture supernatant was cleared and stored at −80 °C. Virus titres were The clades of the reconstructed viral genomes were classified with the Pangolin
determined by plaque assay on VeroE6 cells. Furthermore, the following SARS- webserver (pangolin.cog-uk.io). An in-house R script was also used to plot the variant
CoV-2 isolates were used: Muc-IMB-1, lineage B.1 (EPI_ISL_406862 Germany/ frequencies that were detected by LoFreq as a heatmap (github.com/jonas-fuchs/SARS-
BavPat1/2020)32, kindly provided by Roman Woelfel, Bundeswehr Institute of CoV-2-analyses). This tool is also available on usegalaxy.eu (“Variant Frequency Plot”).
Microbiology; Alpha variant B.1.1.7 (EPI_ISL_751799) and Beta variant B.1.351
(hCoV-19/Germany/NW-RKI-I-0029/2020; ID: EPI_ISL_803957), provided by Immunofluorescence analysis. VeroE6 cells seeded on glass coverslips were either
Donata Hoffmann and Martin Beer, Friedrich-Loeffler-Institute, Riems; and infected with SARS-CoV-2 isolates at a moi of 0.1 or left uninfected. At 8 h post
Gamma variant P.1 (EPI_ISL_3980444) provided by Michael Schindler, Institute infection, cells were fixed in 4% paraformaldehyde in PBS, permeabilized with 0.3%
for Medical Virology and Epidemiology, Tuebingen. All virus stocks used for Triton X-100 and blocked in 10% FCS. SARS-CoV-2 N- (Rockland #200-401-A50,
experiments were inspected for mutations compared to the parental virus isolate 1:1000), S- (Rockland #600-401-MS8, 1:250) and ORF3a-specific primary anti-
using whole genome Illumina sequencing. For the analysis of viral growth, VeroE6 bodies (https://mrcppu-covid.bio/, 1:100) and AF568-labeled goat-anti-rabbit
or Calu3 cells were inoculated in six well plates with a moi of 0.001 and super- (Invitrogen, #A11011, 1:400) secondary antibody as well as AF488-labeled Phal-
natant collected at 8 h, 24 h, 48 h and 72 h post-infection. Viral titres were then loidin (Hypermol, #8813-01, 1:400) were used for staining. The coverslips were
determined by plaque assay on VeroE6 cells. BW5147 mouse thymoma cells embedded in Diamond Antifade Mountant with 4′,6-diamidino-2-phenylindole
(kindly provided by Ofer Mandelboim, Hadassah Hospital, Jerusalem, Israel) stably (DAPI) (ThermoFisher, #P36971). Fluorescence images were generated using a
express human FcγR ectodomains genetically fused to the CD3ξ signalling
38 5 5 LSM800 confocal laser-scanning microscope (Zeiss) equipped with a 63X, 1.4 NAmodule . Cells were maintained at 3 × 10 to 9 × 10 cells/ml in Roswell Park oil objective and Airyscan detector and processed with Zen blue software (Zeiss)
Memorial Institute medium (RPMI GlutaMAX, Gibco) supplemented with 10% and ImageJ/Fiji.
(vol/vol) FCS, sodium pyruvate (1×, Gibco), 100 U/ml penicillin-Streptomycin
(Gibco) β-mercaptoethanol (0.1 mM, Gibco). Cells were cultured at 37 °C, 5% CO2.
All cell lines were routinely tested for mycoplasma. Infection of K18-hACE2 transgenic mice. Transgenic (K18-hACE2)2Prlmn
mice36 were purchased from The Jackson Laboratory and bred locally. Hemizygous
8–12-week-old males were used in accordance with the guidelines of the Federation
Evaluation of the neutralizing capacity of sera and monoclonal antibodies. for Laboratory Animal Science Associations and the National Animal Welfare
Serological neutralization tests were performed with patient sera collected after Body. All experiments were in compliance with the German animal protection law
resolved infection with SARS-CoV-2 or sera of vaccinated individuals ~10–50 days and approved by the animal welfare committee of the Regierungspraesidium
post-vaccination with the second dose of the BNT162b2 mRNA vaccine (Pfizer/ Freiburg (permit G-20/91). Mice were anesthetized using isoflurane and infected
BioNTech). Neutralizing antibody titres were determined by a plaque reduction intranasally (i.n.) with virus dilutions in 40 µl PBS containing 0.1% BSA. Mice were
assay. Serial twofold dilutions of the sera were incubated for 1 h with 100 pfu of the monitored daily and euthanized if severe symptoms were observed or body weight
SARS-CoV-2 isolates. The serum-virus mixture was then used to infect VeroE6 for loss exceeded 25% of the initial weight.
90 min at room temperature. The inoculum was removed and the cells overlaid
with 0.6% oxoid agar for 48 h at 37 °C. Cells were fixed with 3.7% formaldehyde
and stained with crystal violet. The reduction in counted plaque numbers was Plotting and statistical analysis. All plots and statistics were generated with
determined in comparison to an untreated mock-infected control without serum. GraphPad Prism v8.4.2 or R studio (R version 4.0.2).
Neutralization titres of mAb were determined by incubation of the respective
SARS-CoV-2 isolates with tenfold dilutions of the individual antibodies Ethical statement. The project has been approved by the ethical committee of the
(101–10−4 µg/mL). Plaque reduction assay was performed as described above Albert-Ludwigs-Universität, Freiburg, Germany. Written informed consent was
replacing sera dilutions with serial dilution of the mAbs from concentration obtained from all participants and the study was conducted according to federal
101–10−4 µg/ml. To evaluate the neutralizing capacity and determine the neu- guidelines, local ethics committee regulations (Albert-Ludwigs-Universität, Frei-
tralizing titre 50, a non-linear fit least squares regression (constraints: 0 and 100) burg, Germany: No. F-2020-09-03-160428 and no. 322/20) and the Declaration of
was performed. For sera, the mean of each dilution for all sera was determined and Helsinki (1975). All routine virological laboratory testing of patient specimens
plotted to visualize the overall tendency. The fold difference was calculated by the (virus isolation and next-generation sequencing) was performed in the Diagnostic
quotient of the NT50 for B.1 and A.27 for the individual sera. Department of the Institute of Virology, University Medical Center, Freiburg
(Local ethics committee no. 1001913). Convalescent sera and sera of vaccinees were
obtained from the Hepatology-Gastroenterology-Biobank as part of the Freeze-
Fc receptor activation assay. Virus stocks were concentrated by ultra- Biobank Consortium at the University Medical Center Freiburg. Written informed
centrifugation at 100.000 g for 2 h and 4 °C and the pellet dissolved in PBS. The consent was obtained from all blood donors prior to inclusion.
concentrated virus stocks were inactivated with 0.1% β-propiolactone for 16 h at
4 °C followed by 2 h at 37 °C. IgG or the inactivated virions were titrated in PBS
and incubated on an ELISA plate for 1 h at 37 °C for coating. Plates were then Reporting summary. Further information on research design is available in the Nature
blocked in PBS with 10% FCS for 1 h at RT. Immobilized virions were opsonized by Research Reporting Summary linked to this article.
incubation with 20 ng/µl of the respective mAbs (IgG) for 2 h at RT, followed by
incubation with mouse BW5147 reporter cells stably expressing human FcR Data availability
ectodomains genetically fused to CD for 16 h in an incubator (37 °C, 5% CO2). All necessary data and information are given in the paper. Source data are provided with
Immobilized IgG was incubated with reporter cells directly. Secreted mIL-2 was
38,71 this paper. Input XML files of the phylogeographic analysis is supplied in thequantified via anti-mIL-2 sandwich ELISA as described previously . Supplementary Data 1. The sequence data were submitted to the GISAID data base and
are publicly available (Supplementary Table 2). Note, that due to sequencing or
Whole genome sequencing. cDNA was produced from extracted RNA of oro- reconstruction errors (e.g., causing frameshifts) not all A.27 genome sequences obtained
pharyngeal swab or cell culture supernatant samples using random hexamer primers from external laboratories could be uploaded to GISAID. However, all sequences and
and Superscript III (ThermoFisher) followed by a PCR tiling the entire SARS-CoV-2 metadata obtained from the RKI are also available via https://github.com/robert-koch-
genome (ARTIC V3 primer sets; https://github.com/artic-network/artic-ncov2019). institut/SARS-CoV-2-Sequenzdaten_aus_Deutschland, including also all A.27 sequences
This produced ~400 bp long, overlapping amplicons that were subsequently used to used in this study (Supplementary Data 2). Raw sequencing data have been submitted to
prepare the sequencing library. The amplicons were purified with AMPure magnetic the European Nucleotide Archive (https://www.ebi.ac.uk/ena/browser) under the study
beads (Beckman Coulter). Afterwards the QIAseq FX DNA Library Kit (Qiagen) was accession number: ERP134884. Source data are provided with this paper.
used to prepare indexed paired-end libraries for Illumina sequencing. Normalized and
pooled sequencing libraries were denatured with 0.2N NaOH. These libraries were
sequenced on an Illumina MiSeq using the 300-cycle MiSeq Reagent Kit v2. Code availability
The de-multiplexed raw reads were subjected to a custom Galaxy pipeline, which is The script to visualize the variant frequencies is publicly available (github.com/jonas-
based on bioinformatics pipelines on usegalaxy.eu72. The raw reads were pre-processed fuchs/SARS-CoV-2-analyses, v1.0) and implemented on usegalaxy.eu (Variant
with fastp v.0.20.173 and mapped to the SARS-CoV-2 Wuhan-Hu-1 reference genome Frequency Plot).
NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 11
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y
Received: 28 October 2021; Accepted: 10 February 2022; 30. Premkumar, L. et al. The receptor binding domain of the viral spike protein is
an immunodominant and highly specific target of antibodies in SARS-CoV-2
patients. Sci. Immunol. 5, eabc8413 (2020).
31. Starr, T. N., Greaney, A. J., Dingens, A. S. & Bloom, J. D. Complete map of
SARS-CoV-2 RBD mutations that escape the monoclonal antibody LY-
CoV555 and its cocktail with LY-CoV016. Cell Rep. Med. 2, 100255 (2021).
References 32. Böhmer, M. M. et al. Investigation of a COVID-19 outbreak in Germany
1. Wang, Q. et al. Structural and functional basis of SARS-CoV-2 entry by using resulting from a single travel-associated primary case: a case series. Lancet
human ACE2. Cell 181, 894–904 (2020). Infect. Dis. 20, 920–928 (2020).
2. Hoffmann, M. et al. SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 33. de Souza, G. A. P. et al. Emerging SARS-CoV-2 Genotypes Show Different
and is blocked by a clinically proven protease inhibitor. Cell 181, 271–280 Replication Patterns in Human Pulmonary and Intestinal Epithelial Cells.
(2020). Viruses 14, 23 (2022).
3. Ke, Z. et al. Structures and distributions of SARS-CoV-2 spike proteins on 34. Zhang, Y. et al. The SARS-CoV-2 protein ORF3a inhibits fusion of
intact virions. Nature 588, 498–502 (2020). autophagosomes with lysosomes. Cell Disco. 7, 1–12 (2021).
4. Lan, J. et al. Structure of the SARS-CoV-2 spike receptor-binding domain 35. Silvas, J. A. et al. Contribution of SARS-CoV-2 accessory proteins to viral
bound to the ACE2 receptor. Nature 581, 215–220 (2020). pathogenicity in K18 hACE2 transgenic mice. J. Virol. JVI-00402 (2021).
5. McCallum, M. et al. N-terminal domain antigenic mapping reveals a site of 36. Winkler, E. S. et al. SARS-CoV-2 infection of human ACE2-transgenic mice
vulnerability for SARS-CoV-2. Cell 184, 2332–2347.e16 (2021). causes severe lung inflammation and impaired function. Nat. Immunol. 21,
6. Greaney, A. J. et al. Comprehensive mapping of mutations in the SARS-CoV-2 1327–1335 (2020).
receptor-binding domain that affect recognition by polyclonal human plasma 37. Chen, P. et al. SARS-CoV-2 neutralizing antibody LY-CoV555 in outpatients
antibodies. Cell Host Microbe 29, 463–476 (2021). with Covid-19. N. Engl. J. Med. 384, 229–237 (2021).
7. Gottlieb, R. L. et al. Effect of bamlanivimab as monotherapy or in combination 38. Kolb, P. et al. Human cytomegalovirus antagonizes activation of Fcγ receptors
with etesevimab on viral load in patients with mild to moderate COVID-19: a by distinct and synergizing modes of IgG manipulation. Elife 10, e63877
randomized clinical trial. Jama 325, 632–644 (2021). (2021).
8. Weinreich, D. M. et al. REGN-COV2, a neutralizing antibody cocktail, in 39. Bugembe, D. L. et al. Emergence and spread of a SARS-CoV-2 lineage A
outpatients with Covid-19. N. Engl. J. Med 384, 238–251 (2021). variant (A. 23.1) with altered spike protein in Uganda. Nat. Microbiol. 6,
9. Korber, B. et al. Tracking changes in SARS-CoV-2 spike: evidence that D614G 1094–1101 (2021).
increases infectivity of the COVID-19 virus. Cell 182, 812–827 (2020). 40. Butera, Y. et al. Genomic sequencing of SARS-CoV-2 in Rwanda reveals the
10. Zhou, B. et al. SARS-CoV-2 spike D614G change enhances replication and importance of incoming travelers on lineage diversity. Nat. Commun. 12, 5705
transmission. Nature 592, 122–127 (2021). (2021).
11. Lustig, Y. et al. Neutralising capacity against Delta (B. 1.617. 2) and other 41. Anoh, E. A. et al. SARS-CoV-2 variants of concern, variants of interest and
variants of concern following Comirnaty (BNT162b2, BioNTech/P zer) lineage A. 27 are on the rise in Côte d’Ivoire. medRxiv https://doi.org/10.1101/fi
vaccination in health care workers, Israel. Eurosurveillance 26, 2100557 2021.05.06.21256282 (2021).
(2021). 42. Pirnay, J.-P. et al. Variant Analysis of SARS-CoV-2 Genomes from Belgian
12. Geers, D. et al. SARS-CoV-2 variants of concern partially escape humoral but Military Personnel Engaged in Overseas Missions and Operations. Viruses 13,
not T-cell responses in COVID-19 convalescent donors and vaccinees. Sci. 1359 (2021).
Immunol. 6, eabj1750 (2021). 43. Zhang, J. et al. Structural impact on SARS-CoV-2 spike protein by D614G
13. Bager, P. et al. Risk of hospitalisation associated with infection with SARS- substitution. Sci. (80-.) 372, 525–530 (2021).
CoV-2 lineage B. 1.1. 7 in Denmark: an observational cohort study. Lancet 44. Ozono, S. et al. SARS-CoV-2 D614G spike mutation increases entry efficiency
Infect. Dis. 21, 1507–1517 (2021). with enhanced ACE2-binding affinity. Nat. Commun. 12, 1–9 (2021).
14. Ali, F., Kasry, A. & Amin, M. The new SARS-CoV-2 strain shows a stronger 45. Pereira, F. Evolutionary dynamics of the SARS-CoV-2 ORF8 accessory gene.
binding af nity to ACE2 due to N501Y mutant. Med. Drug Disco. 10, 100086 Infect. Genet. Evol. 85, 104525 (2020).fi
(2021). 46. Ren, Y. et al. The ORF3a protein of SARS-CoV-2 induces apoptosis in cells.
15. Gu, H. et al. Adaptation of SARS-CoV-2 in BALB/c mice for testing vaccine Cell. Mol. Immunol. 17, 881–883 (2020).
efficacy. Sci. (80-.). 369, 1603–1607 (2020). 47. Hoffmann, M. et al. SARS-CoV-2 variant B. 1.617 is resistant to
16. Jangra, S. et al. SARS-CoV-2 spike E484K mutation reduces antibody Bamlanivimab and evades antibodies induced by infection and vaccination.
neutralisation. Lancet Microbe. 2, e283–e284 (2021). Cell Rep. 36, 109415 (2021).
17. Deng, X. et al. Transmission, infectivity, and neutralization of a spike L452R 48. Suryadevara, N. et al. Neutralizing and protective human monoclonal
SARS-CoV-2 variant. Cell. 184, 3426–3437.e8 (2021). antibodies recognizing the N-terminal domain of the SARS-CoV-2 spike
18. Li, Q. et al. The impact of mutations in SARS-CoV-2 spike on viral infectivity protein. Cell 184, 2316–2331 (2021).
and antigenicity. Cell 182, 1284–1294 (2020). 49. Amanat, F. et al. SARS-CoV-2 mRNA vaccination induces functionally
19. Mallm, J.-P. et al. Local emergence and decline of a SARS-CoV-2 variant with diverse antibodies to NTD, RBD, and S2. Cell 184, 3936–3948.e10 (2021).
mutations L452R and N501Y in the spike protein. 41, medRxiv https://doi.org/ 50. Wu, J. et al. The Antigenicity of Epidemic SARS-CoV-2 Variants in the
10.1101/2021.04.27.21254849 (2021). United Kingdom. Front. Immunol. 12, 2205 (2021).
20. Colson, P. et al. Spreading of a new SARS-CoV-2 N501Y spike variant in a 51. Rambaut, A. et al. A dynamic nomenclature proposal for SARS-CoV-2
new lineage. Clin. Microbiol. Infect. 27, 1352.e1–1352.e5 (2021). lineages to assist genomic epidemiology. Nat. Microbiol. 5, 1403–1407 (2020).
21. E., S.-L. Potential new lineage causing a cluster in Mayotte 2021 [194] https:// 52. O’Toole, Á. et al. Assignment of epidemiological lineages in an emerging
github.com/cov-lineages/pango-designation/issues/11. pandemic using the pangolin tool. Virus Evol. 7, veab064 (2021).
22. Shu, Y. & McCauley, J. GISAID: Global initiative on sharing all in uenza 53. Krause, G. et al. SurvNet electronic surveillance system for infectious diseasefl
data–from vision to reality. Eurosurveillance 22, 30494 (2017). outbreaks, Germany. Emerg. Infect. Dis. 13, 1548 (2007).
23. Duchene, S. et al. Temporal signal and the phylodynamic threshold of SARS- 54. Tennekes, M. tmap: Thematic Maps in R. J. Stat. Softw. 84, 1–39 (2018).
CoV-2. Virus Evol. 6, veaa061 (2020). 55. Hadfield, J. et al. Nextstrain: real-time tracking of pathogen evolution.
24. Lemey, P. et al. Accommodating individual travel history and unsampled Bioinformatics 34, 4121–4123 (2018).
diversity in Bayesian phylogeographic inference of SARS-CoV-2. Nat. 56. Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for
Commun. 11, 1–14 (2020). phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534
25. Kern, D. M. et al. Cryo-EM structure of SARS-CoV-2 ORF3a in lipid (2020).
nanodiscs. Nat. Struct. Mol. Biol. 28, 573–582 (2021). 57. Sagulenko, P., Puller, V. & Neher, R. A. TreeTime: Maximum-likelihood
26. Flower, T. G. et al. Structure of SARS-CoV-2 ORF8, a rapidly evolving phylodynamic analysis. Virus Evol. 4, vex042 (2018).
immune evasion protein. Proc. Natl Acad. Sci. 118, e2021785118 (2021). 58. Suchard, M. A. et al. Bayesian phylogenetic and phylodynamic data
27. Cerutti, G. et al. Potent SARS-CoV-2 neutralizing antibodies directed against integration using BEAST 1.10. Virus Evol. 4, vey016 (2018).
spike N-terminal domain target a single supersite. Cell Host Microbe. 29, 59. Lemey, P., Rambaut, A., Drummond, A. J. & Suchard, M. A. Bayesian
819–833 (2021). phylogeography finds its roots. PLoS Comput. Biol. 5, e1000520 (2009).
28. Yan, R. et al. Structural basis for the recognition of SARS-CoV-2 by full-length 60. De Maio, N., Wu, C.-H., O’Reilly, K. M. & Wilson, D. New routes to
human ACE2. Sci. (80-.) 367, 1444–1448 (2020). phylogeography: a Bayesian structured coalescent approximation. PLoS Genet.
29. Brouwer, P. J. M. et al. Potent neutralizing antibodies from COVID-19 11, e1005421 (2015).
patients de ne multiple targets of vulnerability. Sci. (80-.) 369, 643–650 61. Lynch, S. M. Evaluating Markov Chain Monte Carlo Algorithms and Modelfi
(2020). Fit. in Introduction to Applied Bayesian Statistics and Estimation for Social
Scientists 131–164 (Springer, 2007).
12 NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28766-y ARTICLE
62. Ayres, D. L. et al. BEAGLE 3: improved performance, scaling, and usability for Dr. Halatoko Wemboo INH (Institut National d’Hygiene, Lomé, Togo), Dagnra
a high-performance computing library for statistical phylogenetics. Syst. Biol. Anoumou Yaotsè (Biolim Université de Lomé, Lome, Togo), Ilhem Boutiba, Riadh
68, 1052–1061 (2019). Gouider, Sameh Trabelsi, and Henda Triki (Laboratory of BioInformatics, bioMathe-
63. Minin, V. N. & Suchard, M. A. Counting labeled transitions in continuous- matics and bioStatistics (BIMS)) and Dr. Justin Lee (CDC Atlanta). We furthermore like
time Markov models of evolution. J. Math. Biol. 56, 391–412 (2008). to acknowledge the excellent technical assistance of Valentina Wagner and Annette
64. Yang, Z. Maximum likelihood phylogenetic estimation from DNA sequences Ohnemus. The authors are grateful to Zsolt Ruzsics, Walter Haas and Otto Haller for
with variable rates over sites: approximate methods. J. Mol. Evol. 39, 306–314 helpful comments on the paper. This work was funded by the Deutsche For-
(1994). schungsgemeinschaft (DFG, German Research Foundation, grant number PA 2274/4-1),
65. Drummond, A. J., Ho, S. Y. W., Phillips, M. J. & Rambaut, A. Relaxed by the Bundesministerium fuer Bildung und Forschung (BMBF) through the Deutsches
phylogenetics and dating with confidence. PLoS Biol. 4, e88 (2006). Zentrum fuer Luft- und Raumfahrt, Germany to M.P. and M.S. (DLR, grant number
66. Baele, G., Gill, M. S., Lemey, P. & Suchard, M. A. Hamiltonian Monte Carlo 01KI2077) and to A.S.O. and C.A.K. (ANDEMIA; grant number 01KA1606). S.L.H.
sampling to estimate past population dynamics using the skygrid coalescent acknowledges support from the Research Foundation - Flanders (“Fonds voor Weten-
model in a Bayesian phylogenetics framework. Wellcome Open Res. 5, (2020). schappelijk Onderzoek - Vlaanderen,” G0D5117N). G.B. acknowledges support from the
67. Ferreira, M. A. R. & Suchard, M. A. Bayesian analysis of elapsed times in Internal Funds KU Leuven (Grant No. C14/18/094). G.B. and N.B. acknowledge support
continuous‐time Markov chains. Can. J. Stat. 36, 355–368 (2008). from the Research Foundation - Flanders (“Fonds voor Wetenschappelijk Onderzoek -
68. Lauer, S. A. et al. The incubation period of coronavirus disease 2019 (COVID-19) Vlaanderen,” G0E1420N, G098321N). The funders had no role in the study design, data
from publicly reported confirmed cases: estimation and application. Ann. Intern. analysis, data interpretation, and in the writing of this report.
Med. 172, 577–582 (2020).
69. Rambaut, A., Drummond, A. J., Xie, D., Baele, G. & Suchard, M. A. Posterior
summarization in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, 901 Author contributions
(2018). J.F., M.H., M.P., L.K., and T.K. designed the study and contributed to experiment design
70. Alaa Abdel Latif, Julia L. et al. Hughes, and the C. for V. S. B. Outbreak.info, and data interpretation. M.H., J.F., S.L.H., N.B., S.C., S.K., and G.B. collected sequence
(https://outbreak.info/compare-lineages). and associated metadata. S.L.H., N.B. and G.B. performed phylogeographic analyses. J.F.,
71. Corrales-Aguilar, E. et al. A novel assay for detecting virus-speci c antibodies L.K., M.H., S.K., S.C., and A.W. performed statistical analyses of patient metadata orfi
triggering activation of Fcγ receptors. J. Immunol. Methods 387, 21–35 (2013). analyzed next-generation sequencing data. A.P., A.L., M.Sa., N.N., J.K.O., R.R., A.B.,
72. Kumar, A., Bangash, A. H. & Gruening, B. Community Research Amid C.A.K., A.O., E.S.L., V.E., and E.A.O. provided sequencing data of SARS-CoV-2. T.K.,
COVID-19 Pandemic: Genomics Analysis of SARS-CoV-2 over Public L.K., J.B., D.S., S.U., S.W., G.K., P.K., M.H.U., and L.J. performed experiments and
GALAXY server. 2020050343 https://doi.org/10.20944/ analysed the data. J.F., L.K., T.K., P.K., S.L.H., and N.B. wrote the paper. J.F., G.D., and
preprints202005.0343.v1 (2020). S.L.H. visualized the data. M.P., M.Sc., and G.K. were involved in funding acquisition.
73. Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: an ultra-fast all-in-one FASTQ
preprocessor. Bioinformatics 34, i884–i890 (2018). Funding
74. Li, H. & Durbin, R. Fast and accurate short read alignment with Open Access funding enabled and organized by Projekt DEAL.
Burrows–Wheeler transform. bioinformatics 25, 1754–1760 (2009).
75. Wilm, A. et al. LoFreq: a sequence-quality aware, ultra-sensitive variant caller for
uncovering cell-population heterogeneity from high-throughput sequencing Competing interests
datasets. Nucleic Acids Res. (2012) https://doi.org/10.1093/nar/gks918. The authors declare no competing interests.
76. Cingolani, P. et al. A program for annotating and predicting the effects of
single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila Additional information
melanogaster strain w1118; iso-2; iso-3. Fly. (Austin) 6, 80–92 (2012). Supplementary information The online version contains supplementary material
77. Li, H. et al. The sequence alignment/map format and SAMtools. available at https://doi.org/10.1038/s41467-022-28766-y.
Bioinformatics 25, 2078–2079 (2009).
Correspondence and requests for materials should be addressed to Marcus Panning or
Jonas Fuchs.
Acknowledgements
We thank Roman Woelfel (Bundeswehr Institute of Microbiology) for providing the B.1 Peer review information Nature Communications thanks the anonymous reviewer(s) for
(Muc-IMB-1) isolate; Donata Hoffmann and Martin Beer (Friedrich-Loeffler-Institut, their contribution to the peer review of this work. Peer reviewer reports are available.
Insel Riems) for providing the B.1.1.7 and B.1.351 isolates, Michael Schindler (Institute
for Medical Virology and Epidemiology, Tuebingen) for providing the P.1 isolate; Reprints and permission information is available at http://www.nature.com/reprints
Markus Hoffmann (Goettingen) for the Calu-3 cells, Todd Giardiello (Rockland
Immunochemicals PA) for providing anti-N and anti-S specific rabbit antisera and James Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
Hastie (MRC Protein Phosphorylation and Ubiquitylation Unit, College of Life Sciences, published maps and institutional affiliations.
University of Dundee) for providing polyclonal sheep antibody targeting ORF3a (see
https://mrcppu-covid.bio/). We gratefully acknowledge the authors from the originating
laboratories responsible for obtaining the specimens and the submitting laboratories Open Access This article is licensed under a Creative Commons
where genetic sequence data were generated and shared via the DESH hub of the RKI Attribution 4.0 International License, which permits use, sharing,
and GISAID (Supplementary Data 2 and Supplementary table 3). We acknowledge the adaptation, distribution and reproduction in any medium or format, as long as you give
contribution of all local and state public health authorities, laboratories, and health appropriate credit to the original author(s) and the source, provide a link to the Creative
workforce who have submitted COVID-19 case-based data to the German notification Commons license, and indicate if changes were made. The images or other third party
system. We would like to thank Bas B. Oude Munnink (Department of Viroscience,
material in this article are included in the article’s Creative Commons license, unless
WHO Collaborating Centre for Arbovirus and Viral Hemorrhagic Fever Reference and
indicated otherwise in a credit line to the material. If material is not included in the
Research, Rotterdam, Netherlands) for providing travel history information of A.27
article’s Creative Commons license and your intended use is not permitted by statutory
cases. We are furthermore grateful for the sequencing efforts from our colleagues Judd F.
regulation or exceeds the permitted use, you will need to obtain permission directly from
Hultquist, Ramon Lorenzo-Redondo and Lacy M. Simons (Center for Pathogen Geno-
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
mics and Microbial Evolution, Institute for Global Health, Northwestern University
Feinberg School of Medicine, Chicago, IL 60611, USA), Olubusuyi M. Adewumi licenses/by/4.0/.
(Department of Virology, College of Medicine, University of Ibadan, Ibadan, Nigeria),
© The Author(s) 2022
NATURE COMMUNICATIONS |         (2022) 13:1152 | https://doi.org/10.1038/s41467-022-28766-y | www.nature.com/naturecommunications 13