To the Editor: The severe acute respiratory syndrome (SARS) outbreak of 2002–03 and the subsequent implication of bats as reservoir hosts of the causative agent, a coronavirus (CoV), prompted numerous studies of bats and the viruses they harbor. A novel clade 2c betacoronavirus, termed Middle East respiratory syndrome (MERS)–CoV, was recently identified as the causative agent of a severe respiratory disease that is mainly affecting humans on the Arabian Peninsula (1). Extending on previous work (2), we described European Pipistrellus bat–derived CoVs that are closely related to MERS-CoV (3). We now report the identification of a South Africa bat derived CoV that has an even closer phylogenetic relationship with MERS-CoV.

During 2011–2012, fecal pellets were collected from 62 bats representing 13 different species in the KwaZulu-Natal and Western Cape Provinces of South Africa and stored in RNAlater solution (Life Technologies, Carlsbad, CA, USA). Details about the bat sample are available in the Technical Appendix. RNA was extracted by using the QIAamp Viral RNA Mini Kit (QIAGEN, Hilden, Germany). Screening for CoVs was done by nested reverse transcription PCR using broadly reactive oligonucleotide primers targeting a conserved region in the RNA-dependent RNA polymerase (RdRp) gene (online Technical Appendix). PCR results were positive for 5 (8%) of the 62 specimens. PCR amplicons for 4 positive specimens yielded alphacoronavirus sequences related to recently described bat alphacoronaviruses from South Africa (4). The other positive specimen, termed PML/2011, was from an adult female Neoromicia cf. zuluensis bat sampled in 2011; the specimen yielded a novel betacoronavirus (GenBank accession no. KC869678). Technical Appendix Figure 1 shows the distribution of this bat species.

To obtain better phylogenetic resolution, we extended the 398-nt RdRp fragment generated by the screening PCR to 816 nt, as described (5). PML/2011 differed from MERS-CoV by only 1 aa exchange (0.3%) in the translated 816-nt RdRp gene fragment. Thus, PML/2011 was much more related to MERS-CoV than any other known virus. The amino acid sequence of the next closest known relatives of MERS-CoV, from European Pipistrellus bats (3), differed from MERS-CoV by 1.8%. The amino acid sequences of viruses from Nycteris bats in Ghana (3) and the 2c prototype bat CoVs, HKU4 and HKU5, from China (6) differed by 5.5%–7.7% from MERS-CoV. The smaller 152- to 396-nt RdRp fragments of 2c bat CoVs from a Hypsugo savii bat in Spain (7), bat guano in Thailand (8), and a Nyctinomops bat in Mexico (9) showed no or only partial overlap with the 816-nt fragment generated in this study; thus, a direct comparison could not be done. However, in their respective RdRp fragments, these CoVs yielded amino acid sequence distances of 3.5%–8.0% and were thus probably more distant from MERS-CoV than the virus described here.

A Bayesian phylogenetic analysis of the 816-nt RdRp sequence confirmed the close relationship between PML/2011 and MERS-CoV (Figure). Their phylogenetic relatedness was as close as that of SARS-CoV and the most closely related bat coronavirus known, Rs672 from a Rhinolophus sinicus bat (Figure). Like PML/2011 and MERS-CoV, Rs672 and SARS-CoV showed only 1 aa exchange in the translated 816-nt RdRp fragment. To confirm this relatedness, we amplified and sequenced a short 269-nt sequence encompassing the 3′-terminus of the spike gene for PML/2011 (oligonucleotide primers available upon request from the authors). A partial spike gene–based phylogeny using this sequence yielded the same topology as that using the partial RdRp sequence (Technical Appendix Figure 2). Again, PML/2011 was most closely related to MERS-CoV, showing only a 10.9% aa sequence distance in this gene, which encodes the glycoprotein responsible for CoV attachment and cellular entry. This distance was less than the 13.3% aa sequence distance between MERS-CoV and the European Pipistrellus CoVs (3) and less than the 20.5%–27.3% aa sequence distance between MERS-CoV and HKU5 and between MERS-CoV and HKU4 (6) in the same sequence fragment.

Our results further support the hypothesis that, like human CoV-229E and SARS-CoV, ancestors of MERS-CoV might exist in Old World insectivorous bats belonging to the family Vespertilionidae, to which the genera Neoromicia and Pipistrellus belong (3). Knowledge of the close relatedness of PML/2011 and MERS-CoV, which contrasts with the more distant relatedness of CoVs in bats from the Americas and Asia, enables speculations of an African origin for bat reservoir hosts of MERS-CoV ancestors. This hypothesis is limited by a global sampling bias, the small sample size, and the single clade 2c betacoronavirus detection in this study. Still, a putative transfer of MERS-CoV ancestors from Africa to the Arabian Peninsula would parallel the transfer of other viruses (e.g., the exportation of Rift Valley fever virus from East Africa, which led to a severe outbreak in Saudi Arabia in 2000) (10).

Studies of Vespertilionidae bats and potential intermediate hosts (e.g., carnivores and ungulates, such as camels) are urgently needed to elucidate the emergence of MERS-CoV. Such studies should focus on the Arabian Peninsula and Africa.