Background
Maize (
Zea mays L.) is one of the most important cereals in Sub-Saharan Africa and is grown in approximately 25 million hectares [
1]. Maize is consumed as a preferred calorie source by 95% of the population, at an average of 1075 kcal/capita/day, which represents more than 50% of the recommended daily intake [
2]. Maize production is destined for human consumption or animal feed at a proportion of 88 and 12%, respectively [
3,
4].
In 2011 maize lethal necrosis disease was first detected in Kenya [
5‐
7], and confirmed in several countries in East and Central Africa, specifically in Tanzania, Uganda [
8], Rwanda [
9] DR Congo [
10], Ethiopia and South Sudan [
11]. Corn lethal necrosis (CLN) was first described in the State of Kansas in 1978 [
12]. In their original descriptions, corn lethal necrosis and maize lethal necrosis defined the same disease. Herein we use maize lethal necrosis disease.
In Sub-Saharan Africa, smallholder farms account for approximately 80% of the farm land and employ 175 million people directly [
13,
14]. Small-scale farmers largely rely on maize, as a major source of energy and revenue [
15]. With yield losses ranging from 30 to 100% that lead to food shortages and contribute to hunger and malnutrition [
16], maize lethal necrosis is currently a threat to maize production and food security in Sub-Saharan Africa.
Maize lethal necrosis is caused by a synergistic co-infection of MCMV, a Machlomovirus in the family
Tombusviridae [
17], and specific members of the family
Potyviridae, such as SCMV [
12],
Wheat streak mosaic virus (WSMV) [
18], or JGMV [
19]. In maize lethal necrosis outbreaks, MCMV and SCMV is the most prevalent virus combination [
9,
10,
20]. In Rwanda,
Maize yellow mosaic virus (MaYMV), a polerovirus, was recently detected in maize plants showing symptoms similar to those caused by maize lethal necrosis [
21].
Typical maize lethal necrosis symptoms include severe yellowing and leaf drying from the edges, stunting and premature plant death, sterility in male plants, poor tasseling, lack of or only a few grains in the cob, malformed or rotten cobs [
7,
19]. In farmer’s fields in Kenya, we detected plants showing bright yellow stripes with green edges, which deviate from typical maize lethal necrosis symptoms. Additionally, symptomatic plants often tested negative for SCMV by ELISA, as described by others [
19,
21,
22].
Maize lethal necrosis continues to spread rampantly and is a major concern to maize stakeholders [
5] including small and large-scale farmers, commercial seed sector, millers, transporters, policy makers, local and international communities. These raises several questions such as why is maize lethal necrosis still difficult to manage and what strategies can farmers implement?
Natural and engineered genetic resistance provide a successful approach to managing viral diseases [
23]. With respect to natural genetic resistance, massive screens of commercial hybrids and thousands of maize lines reported high levels of susceptibility. Only few lines were moderately resistant [
24,
25]. Several efforts are underway to identify and characterize maize resistance to MCMV [
26] and SCMV [
27].
We hypothesized that uncharacterized viruses synergistically interact with MCMV to cause maize lethal necrosis, and there is genetic variation between SCMV and MCMV in East Africa compared to the rest of the world. To test these hypotheses, we collected samples from symptomatic and asymptomatic maize leaves in sixteen counties in Kenya. Cultivated and wild sorghum [Sorghum bicolor (L.) Moench] and napier grass (Pennisetum purpureum S.) were also included to determine their potential as alternate hosts. Viruses present were identified by metagenomics using next-generation sequencing of total RNA and bioinformatics. Viral presence was determined for each individual sample using de-novo assembled contigs.
After de-novo assembly, complete and partial genomes were obtained for MCMV, SCMV, Maize yellow dwarf virus-RMV (MYDV-RMV) and Maize streak virus (MSV). Partial genomes were assembled for other four potyviruses and one polerovirus. A geographic analysis showed the wide distribution of MCMV, SCMV, MYDV-RMV and MSV infecting maize and sorghum in Kenya. A large number (30/68) of the samples analyzed had a combination of four viruses: MCMV, SCMV, MYDV-RMV, and MSV. Only one sample had MCMV in the absence of other viruses. All the other samples (67/68) had MCMV plus one, two, three, or four other viruses. Phylogenetic analyses of near complete genome nucleotide sequences showed that MCMV, MSV and MYDV-RMV in Kenya are similar to isolates from East Africa. In contrast, SCMV from Kenya exhibits the largest genetic variation and distance with respect to isolates from others parts of the world, including East Africa. These results provide a solid foundation to develop virus diagnostic protocols, management strategies, and raise the possibility of a synergistic interaction between MCMV and a polerovirus to cause maize lethal necrosis.
Discussion
Maize lethal necrosis disease is caused by the synergistic co-infection of MCMV and a member of the
Potyviridae. Synergism has been confirmed for SCMV [
6,
12], WSMV [
18], and JGMV [
19]. Recently, the polerovirus
Maize yellow mosaic virus (MaYMV) was detected in maize plants showing lethal necrosis-like symptoms in Rwanda [
21]. In the analysis described here, the polerovirus
Maize yellow dwarf virus (MYDV-RMV) was found to be widely distributed in Kenya (Figs.
1b and
6), and the polerovirus
Barley virus G was detected in 11 of the 68 samples analyzed (Fig.
1d). MYDV-RMV was always found as part of a complex that included MCMV and SCMV, or MCMV, SCMV and MSV. The wide distribution of poleroviruses infecting maize in Rwanda [
21] and in Kenya (Fig.
1b) suggests the possibility of a synergistic interaction between MCMV and a polerovirus to cause maize lethal necrosis, and may contribute to the variation on virus-induced symptoms observed in the field (Fig.
1a).
The molecular mechanisms of viral synergism in maize lethal necrosis remain to be determined. One model is that maize lethal necrosis is mediated by silencing suppressors encoded by the co-infecting viruses. In support of this model, the synergistic interaction between potyviruses and
Potato virus X (PVX) and
Cucumber mosaic virus (CMV) is mediated by silencing suppression activity of potyviral HC-Pro [
42,
43]. Consistent with this model, SCMV and WSMV encode RNA silencing suppressors HC-Pro and P1, respectively [
44,
45]. Several, poleroviruses, including MaYMV encode PO, a strong RNA silencing suppressor [
46]. These observations are consistent with a role for maize-infecting poleroviruses in maize lethal necrosis.
However, no silencing suppressor has been described for MCMV [
47] or MSV (a Mastrevirus) [
48]. Interestingly, in
Wheat dwarf virus (a Mastrevirus) replication-associated proteins are silencing suppressors [
49], which suggest that MSV harbors silencing suppressor proteins. Further investigation is needed to determine the role of silencing suppression, and the contribution of poleroviruses and MSV to maize lethal necrosis.
There is ambiguity with respect to the scientific name given to poleroviruses infecting maize. In 2013, the first species was named
Maize yellow dwarf virus (MYDV-RMV) [
50]. Two different isolates from China were named
Maize yellow mosaic virus (MaYMV) [
46] and
Maize yellow dwarf virus-RMV2 (MYDV-RMV2) [
51], while an isolate infecting sugarcane in Nigeria was named
Maize yellow mosaic virus (MaYMV) [
52], and an isolate infecting maize in Kenya was renamed as
Maize yellow dwarf virus-RMV (MF974579.2). Three near genome length polerovirus contigs from maize and one from sorghum described here were most closely related (99% similarity) to
Maize yellow dwarf virus-RMV (Figs.
6b and
8b). However, one near genome length contig from maize was most closely related to
Maize yellow mosaic virus (MaYMV) (Fig.
8b), the most prevalent virus infecting maize in Rwanda [
21]. These observations suggest that a complex of closely related poleroviruses infect both maize, sorghum, and possibly other species in East Africa.
In East Africa, ELISA [
19,
21,
22] and RT-PCR [
9] procedures have provided inconsistent detection of SCMV. Sequencing analysis described here and before [
6] show that the SCMV present in Kenya (Figs.
3 and
4) and in Rwanda [
9] is distantly related to isolates from other parts of the world (Fig.
7b). Interestingly, our results showed that most of the variation occurs between the C terminus of NIb and the N terminus of the coat protein (Figs.
3a and
4). Both nucleotide and amino acid variation was observed in all twenty Kenya samples that provided near complete genome contigs (Fig.
4 and Additional file
7: Figure S3). Based on this variation, Kenya samples, and isolates from other parts of the world, were divided into three groups. Nucleotide substitutions that resulted in several amino acid substitutions in both NIb and the coat protein was the most frequent event (group 1, 11 samples) (Figs.
3a and
4b). However, in the other nine samples, in-frame deletions resulted in a 13-amino acid deletion at the C terminus of NIb (group 2, 6 samples) (Fig.
4b), or in a 15-amino acid deletion distributed between the C terminus of NIb and the N terminus of the coat protein (group 3, 3 samples) (Fig.
4b).
In members of the Potyviridae, NIb is required for virus replication, while the coat protein participates in virion assembly, cell-to-cell and systemic movement [
53]. The effect of amino acid substitutions and deletions at the C terminus of NIb and at the N terminus of the coat protein on virus pathogenicity remain to be determined. Presence of these deletions in SCMV isolates from other parts of the world suggest that viruses harboring these deletions are pathogenic. Consistent with this hypothesis, in
Wheat streak virus (Family Potyviridae, genus Tritimovirus), a genetic analysis using an infectious clone showed that deletions at the N terminus of the coat protein are tolerated and mutants cause more severe symptoms than the wild type virus in several hosts [
54,
55]. Alternatively, in the absence of co-infecting viruses, in SCMV deletions between the C terminus of NIb and the N terminus of the coat protein may be lethal.
Polyprotein alignment showed that the 15-amino acid deletion observed in three Kenya samples (group 3) is present in one isolate from China and in the isolate reported in the original description of maize lethal necrosis in Kenya (JX286708.1) [
6] (Fig.
4b). Amino acid variation at the C terminus of NIb and at the N terminus of the coat protein in Kenya group 1 is similar to variation in two isolates from Rwanda [
9], two from Ethiopia and one from Mexico. Additionally, the 13-amino acid deletion observed in 6 samples from Kenya (group 2) is present in a SCMV isolate from Ethiopia (Fig.
4b). Furthermore, three complete genomes have been described from Ethiopia [
11]. In our analysis, they formed a clear cluster between the China and Mexico isolate (Fig.
7b). Interestingly, one isolate from Ethiopia harbors the 13-amino acid deletion described here for 6 Kenya samples (Fig.
4b, group 2). Cloning, sequencing, and restriction digestion analysis of SCMV infecting sugarcane in India [
56] and maize in Brazil [
57] showed that the N terminus of the coat protein is hypervariable. A similar analysis showed genetic diversity in SCMV coat protein sequence in Cameroon and Congo [
58]. These observations show that SCMV harbors a hypervariable region between NIb and the coat protein.
Variation at the C terminus of NIb and N terminus of the coat protein in SCMV could explain inconsistent detection of SCMV by ELISA [
19,
21,
22], and failure to detect SCMV in Rwanda [
9] (similar to group 1) by RT-PCR using primers designed for Kenya group 3. These observations highlight the need to raise antibodies against African isolates and universal primers to detect SCMV. Alternatively, or in addition, plants showing maize lethal necrosis symptoms could be infected by other potyviruses. In addition to SCMV, other potyviruses found in Kenya samples include were
Hubei Poty-like virus 1,
Scallion mosaic virus and
JGMV (Fig.
1d). Interestingly, JGMV in combination with MCMV, causes maize lethal necrosis [
19]. The role of other potyviruses in maize lethal necrosis remains to be determined.
Screening of germplasm and commercial hybrids for resistance to maize lethal necrosis has focused on MCMV and SCMV [
5,
24‐
26]. The widespread distribution of MYDV-RMV, MSV, and possibly JGMV (Fig.
2b) [
21] highlights the need to include other viruses in breeding programs seeking to develop virus-resistant cultivars or hybrids for East Africa.
Multiple sources of virus may contribute to maize lethal necrosis epidemic. Soil and seed transmission is possible for both MCMV and SCMV [
5,
59]. Additionally, both potyviruses and poleroviruses are transmitted by aphids [
50,
60]. MCMV is transmitted by several species of beetles in the family
Chysomelidae [
61] and by western flower thrips (
Frankliniella occidentallis) [
62]. Despite lacking visible viral symptoms, the three sorghum samples we analyzed (Fig.
1c) contained MCMV, SCMV, MSV and MYDV-RMV. Similarly, one asymptomatic napier grass sample contained MCMV (Fig.
2). Consistent with these observations, several grass species and sorghum cultivars were determined to be asymptomatic hosts for MCMV, SCMV and WSMV [
5,
59]. Thus, sorghum, napier grass and possible other grass species are virus reservoirs for insect vectors to spread the viruses to maize. Several factors, including genotype, plant age and days after infection at the time samples were collected, may contribute to the absence of symptoms in our sorghum and napier grass samples. Further experimentation is needed to determine the response of sorghum and napier grass to viruses that cause maize lethal necrosis and to understand their role as alternate hosts.
Although MCMV, SCMV, WSMV and JGMV are present, maize production is not reduced due to maize lethal necrosis in the United States [
63,
64]. After the initial detection in Kansas and Nebraska in the 1970’s [
65], maize lethal necrosis was managed by a combination of agronomic practices that included crop rotation, removal of alternate hosts, and use of hybrids tolerant to MCMV or SCMV [
65,
66]. Epidemiological models and field surveys show that growing maize continually results in an increase of virus inoculum [
5,
63]. Consistent with these observations, crop rotation could reduce the prevalence and delay infection [
63]. However, in East Africa, maize is grown year-round during two growing seasons, underscoring the need to develop integrated management strategies to slow the spread and damage caused by maize lethal necrosis. The strategy must include identification and deployment of virus tolerant germplasm, seed sanitation and distribution programs, identification and removal of alternates, and insect vector control, and the establishment of a systematic surveillance program. SCMV in Kenya are genetically different to isolates from other parts of the world (Fig.
4). Thus, phytosanitary regulations could be implemented on maize and sorghum grain imports. These measures require rapid and reliably diagnosis. Sequences described here provide a solid foundation to develop global, directed multiplex nucleic acid-based methods to diagnose MCMV, SCMV, MSV, MYDV-RMV and closely related viruses.