Multiple Introductions of the Old World Begomovirus Tomato yellow leaf curl virus into the New World

Abstract

A phylogenetic analysis of three genomic regions revealed that Tomato yellow leaf curl virus (TYLCV) from western North America is distinct from TYLCV isolated in eastern North America and the Caribbean. This analysis supports a second introduction of this Old World begomovirus into the New World, most likely from Asia.

Tomato yellow leaf curl virus (TYLCV) is an emergent, monopartite begomovirus (of the family Geminiviridae) that causes perhaps the most devastating epidemics in tomato agriculture (32). TYLCV was first identified in Israel in 1939 (32) and was isolated and sequenced in 1988 (isolate TYLCV-IL [7, 29]). Molecular typing of outbreaks has shown that viruses closely related to TYLCV-IL have spread worldwide (19, 27, 32, 46).

TYLCV was first introduced into the New World, either through the Dominican Republic in the early 1990s (26, 33) or through Cuba (19, 36), possibly in a shipment of tomato seedlings from Israel (34). TYLCV subsequently spread to other Caribbean islands and to North America (24). Previously published phylogenies have supported the monophyly of New World TYLCV (8, 17, 23, 25, 28, 43, 48), suggesting that all New World TYLCV isolates were derived from a single introduction of a TYLCV-IL-like virus into the New World (13, 17, 23, 28, 48). However, recently obtained isolates of TYLCV from western North America (WNA; from Sinaloa, Mexico, and California) are not as closely related to other New World TYLCV isolates (5). Herein, we test whether these WNA isolates could have descended from a second, independent introduction of TYLCV into the New World.

New World sequences of TYLCV were downloaded from GenBank (Table 1 [names following current nomenclature {9}]). BLASTN (1) was used to identify sequences closely related to the WNA TYLCV, and two recent isolates of TYLCV from Asia that have 99% identity to TYLCV-US:TX and TYLCV-MX:Cul were included in our analysis (Table 1). TYLCV-IL, which is closely related to the New World TYLCV (13, 17, 23, 28, 48), was also included. Sequences were aligned by eye with SE-AL (http://evolve.zoo.ox.ac.uk). Most of the available sequences were partial genome sequences, some covering less than 10% of the TYLCV genome (Table 1). To include all of the extant sequences from the New World, we identified three adjacent, nonoverlapping regions of the TYLCV genome for phylogenetic analysis, for which different numbers of sequences were available: ∼255 nucleotides (nt) of the intergenic region, which is known to be the most variable portion of the genome (10, 11, 30); 61 noncoding nt from the intergenic region with two-thirds of the pre-coat protein-coding region (V2), ∼360 nt; and a portion of the coat protein (CP)-coding region, ∼538 nt. Together, these regions covered ∼40% of the TYLCV genome.

TABLE 1.

New World (and sister taxa found on other continents) isolates of TYLCV used in analyses^a

Origin	Name	Yr(s) of isolation	Source plant	GenBank accession no.	Sequence length (nt)	Sequence data for indicated region			Source or referencec
						Int (255 nt)	V2 (360 nt)	CP (538 nt)
Antigua	TYLCV-AG:StJ	NA	Tomato	EF028239	545			•	Direct submission
	TYLCV-AG:StP	NA	Tomato	EF028240	530	•	Direct submission
Cuba	TYLCV-CU	1995-1997	NA	AJ223505	2,781	•	•	•	Direct submission; E. Bejarano, personal communication
	TYLCV-CU2	NA	Tomato	U65089	777	•	37
	TYLCV-CU3	2001	Pepper	AF414089	630	•	36
	TYLCV-CU4	2003	Squash	DQ207810	449	•	20
Dominican Republic	TYLCV-DO	1994	Tomato	AF024715	2,781	•	•	•	24, 39
Guadeloupe	TYLCV-GP	2001	Tomato	AY319645	542		•		44
	TYLCV-GP2	2001	Tomato	AY319646	966	•	44
Jamaica	TYLCV-JM:2T	1994-1996b	Tomato	U84146	315	•			Direct submission
	TYLCV-JM:10T	1994-1996b	Tomato	U84147	312	•	Direct submission
	TYLCV-JM:52	1994-1996b	Pepper	U88889	252	•	Direct submission
Mexico	TYLCV-MX:Yuc	1996-1997	Tomato	AF168709	753	•	•		2
	TYLCV-MX:Cul	2005	Tomato	DQ631892	2,781	•	•	•	5
	TYLCV-MX:Sin	NA	Tomato	DQ377367	742	•	•	Direct submission
	TYLCV-MX:Sin2	NA	NA	EF523478	2,781	•	•	•	Direct submission
Puerto Rico	TYLCV-PR	2001	Tomato	AY134494	2,781	•	•	•	3
United States	TYLCV-Flo	1997	Tomato	AY530931	2,781	•	•	•	34, 47
	TYLCV-US:Flo	1997	Tomato	AF260331	1,320	•	•	•	48
	TYLCV-US:SC	2005	Tomato	DQ139329	806	•	•	•	18
	TYLCV-US:TX	2006	Tomato	EF110890	2,752	•	•	•	16
	TYLCV-US:CA	2007	Tomato	EF539831	2,781	•	•	•	Direct submission
Israel	TYLCV-IL	1988	Tomato	X15656	2,787	•	•	•	29
China	TYLCV-CN:SH2	2006	Tomato	AM282874	2,781	•	•	•	46
Japan	TYLCV-JA:Tosa	2004	Tomato	AB192965	2,781	•	•	•	43

^aNA, not available; Int, intergenic region; •, sequence used in this alignment.

^bThe TYLCV from Jamaica could not have been isolated before the virus was first observed on the island, in 1994 (21).

^cDirect submission, information in GenBank annotation (submitted by the researchers who sequenced the given isolate) not published elsewhere.

MODELTEST (35) was used to select the most appropriate model of nucleotide substitution for phylogenetic analysis (intergenic region, HKY85; V2 region, JC+Γ; CP region, Tamura-Nei), which was performed with PAUP*, using maximum likelihood (ML) and tree-bisection-reconnection branch-swapping (41). To assess the reliability of specific groups, a neighbor-joining bootstrap analysis was employed (1,000 replicates under the ML substitution model). The bootstrap values for all nodes with more than 50% support were reported (Fig. 1).

FIG. 1.

Evolutionary relationships among New World TYLCV isolates based on 255 nt of the intergenic region (Int), 360 nt of the intergenic/V2-coding region (V2), and 538 nt of the CP-coding region. An Israeli (TYLCV-IL) and two Asian (TYLCV-CN:SH2 and TYLCV-JA:Tosa) isolates were included and are shown italicized and bolded. The ENAC sequences are shown in the darker shading; the WNA sequences are shown in the lighter shading. All branch lengths are scaled according to the numbers of nucleotide substitutions per site.

Our data set of TYLCV contains four Mexican sequences (Table 1), one of which is from eastern Mexico (Yucatán), near the Caribbean (2), with the other three from the important tomato-growing region of Sinaloa, in western Mexico (6). By contrast, most of the sequences isolated from the United States have been found near the Caribbean (e.g., Florida [34], Louisiana [45], Georgia [22], and Mississippi [15]), even though the virus has recently been identified in western states, such as California and Arizona (14, 38). In all three trees, the sequences from eastern Mexico, the eastern United States, and the Caribbean grouped together, forming an eastern North America and Caribbean (ENAC) clade, albeit with strong support in the CP gene only (bootstrap values of 51% [Fig. 1, Int], 50% [Fig. 1, V2], and 100% [Fig. 1, CP]). The western Mexican (Sinaloa) and western United States (California) sequences grouped together in the two protein-coding region trees (Fig. 1, V2 [91% bootstrap support] and CP). In the intergenic region (Fig. 1, Int), the four WNA viruses did not form a monophyletic clade, and this was likely due to a single informative base that TYLCV-US:CA shares with TYLCV-IL and the ENAC clade (a cytosine at nt 176; nt 188 in TYLCV-IL). Despite this probable parallelism, all four WNA viruses were part of well-supported clades with the Asian TYLCV sequences in all three trees (bootstrap values of 100% [Fig. 1, Int], 88% [Fig. 1, V2], and 98% [Fig. 1, CP]). Crucially, this Asian/WNA clade was reciprocally monophyletic with the ENAC clade and thereby provided strong support for the assumption that the WNA viruses are more distantly related to ENAC TYLCV and likely came from a second introduction of TYLCV into the New World. Similarly, this analysis also indicates that this second introduction most likely came to the Pacific coast of Mexico from Asia.

To further test the hypothesis of multiple entry of TYLCV into the New World, we performed a Shimodaira-Hasegawa test (full optimization, with PAUP* [40]). Accordingly, the ML trees for each genomic region, which support two entries of TYLCV into the New World, were compared to “model” trees in which all New World isolates were forced to be monophyletic, as expected under a single entry into the New World. In all cases, the single entry hypothesis was rejected, though the V2 region offered only marginal support (intergenic region P = 0.017; V2 P = 0.105; CP P = 0.041). These results contradict the conventional wisdom that TYLCV has spread through the New World from a single introduction (14).

Previous studies have noted that there is a large (>1,500-mile) separation between the Yucatán, where TYLCV was first introduced into Mexico in 1996 (2), and Sinaloa, where TYLCV was observed in 2005 (5). Hence, a second introduction of TYLCV explains why TYLCV would be found on both coasts of Mexico without having migrated through the bulk of Mexico.

After the confirmation of TYLCV in Sinaloa, TYLCV was found farther north in Mexico (14), in Texas (16), in Arizona (14), and in southern California (38). Our analyses further confirm that TYLCV-US:CA is most closely related to the western Mexican viruses (38) and therefore likely spread over the border from Mexico (Fig. 1). However, others describe an Arizonian isolate (sequence currently unavailable) that is more closely related to TYLCV-US:TX, which is part of the ENAC clade (Fig. 1), than to viruses from Mexico (14). We therefore conclude that the ENAC strain of TYLCV, which had spread to Louisiana by 2000 (45), has spread through Texas to Arizona but that the first TYLCV to be isolated from California is more closely related to other WNA TYLCV strains from Mexico. This is not surprising, as two other geminiviruses infecting tomato, Tomato leaf curl Sinaloa virus (12) and Chino del Tomate virus (4, 42), have previously spread from Sinaloa into the United States.

Though our results shed some light on the passage of TYLCV through the New World, a combination of recent common ancestry and the limited availability of whole-genome sequences (only a third of the New World TYLCV sequences in GenBank are of whole genomes) (Table 1) meant that it was not possible to fully resolve the phylogenetic history of all New World isolates. Further, as recombination is an important aspect of geminivirus evolution (10, 31), sequencing only a portion of the genome could be revealing only part of the New World TYLCV evolutionary history.