© 2011 Nature America, Inc. All rights reserved.

ne w s and vie w s and expansion in spleen resulted mainly in systemic IgG humoral immunity. In addition, Ye et al.1 used CpG oligonucleotide as an adjuvant, whereas earlier studies (e.g., those reviewed in ref. 8) used nontoxic cholera toxin and E. coli labile toxin. CpG oligonucleotide probably induces a modest proinflammatory response through toll-like receptor 9 signaling, whereas nontoxic cholera toxin and E. coli labile toxin, which bind ganglioside receptors on epithelial cells, do not. Finally, nasal gD-Fc plus CpG oligonucleotide resulted in the development of inducible, bronchus-associated ­lymphoid ­tissue, which is often associated with inflammatory responses or immunity. Two major goals for developing many mucosal vaccines are (i) protection from infections that occur in different tissues to the site of vaccination, and (ii) induction of prolonged immunity (memory). Ye et al.1 achieve both benchmarks: mice given nasal gD-Fc were able to reduce HSV titers and survive a lethal vaginal challenge with virulent HSV even 6 months after vaccination, whereas control groups were not protected1. The study of Ye et al.1 raises several questions for future research. First, why was the CpG oligonucleotide adjuvant required? The authors propose that it may have prevented induction of mucosal tolerance. I would broaden this question to consider the relative activation of the mucosal and systemic immune systems by their vaccination protocol. Oral or intranasal antigen delivery with mucosal adjuvants induces both mucosal and systemic immunity, whereas delivery of antigen alone often induces mucosal tolerance8. Delivery of antigen through the FcRn pathway may subvert induction of mucosal responses (either immunity or tolerance) and result mainly in enhancement of systemic IgG antibody responses. The protocol of Ye et al.1 apparently diverted antigen away from the mucosal inductive site, nasal-associated lymphoid tissue, to a systemic induction site, mediastinal lymph nodes, resulting in weak SIgA responses (produced by the mucosal immune system) and robust IgG responses (produced by the systemic immune system). However, many infectious diseases, such as influenza, rotavirus or cholera, are best controlled by a combination of SIgA antibodies and IgG antibodies8. It will be important to determine whether nasal, FcRnmediated delivery, given with a mucosal adjuvant or vector can induce strong SIgA responses as well as strong systemic immunity. Although vaccine campaigns have eradicated smallpox and are closing in on polio, similar successes against diseases such as AIDS and tuberculosis are not in sight. The ideal vaccine would be one that is completely effective, avoids the need for a cold chain and could be given needle-free by unskilled caregivers. Sound 138

impossible? Consider that adult humans express FcRn in the gastrointestinal and respiratory tracts. Oral or intranasal delivery of an effective Fc-viral, Fc-bacterial or Fc-parasitic vaccine could dramatically alter the global reach of immunization efforts. It is time to ­consider combining current vaccine approaches with FcRn-mediated transport. COMPETING FINANCIAL INTERESTS The author declares no competing financial interests. 1. Ye, L. et al. Nat. Biotechnol. 29, 158–163 (2011).

2. Roopenian, D.C. & Akilesh, S. Nat. Rev. Immunol. 7, 715–725 (2007). 3. Baker, K. et al. Semin. Immunopathol. 31, 223–236 (2009). 4. Simister, N.E. & Mostov, K.E. Nature 337, 184–187 (1989). 5. Chaudhury, C. et al. J. Exp. Med. 197, 315–322 (2003). 6. Yoshida, M. et al. Immunity 20, 769–783 (2004). 7. Yoshida, M. et al. J. Clin. Invest. 116, 2142–2151 (2006). 8. Kiyono, H., Kunasawa, J., McGhee, J.R. & Mestecky, J. The Mucosal Immune System. in Fundamental Immunology (ed. Paul, W.E.) 983–1030 (Lippincott Williams & Wilkins, Philadelphia, 2008).

Sequencing for the cream of the crop Jer-Ming Chia & Doreen Ware Sequencing technologies are poised to transform conventional strategies for crop improvement. Next-generation sequencing methods are beginning to find their way into plant genetics, promising substantial improvements in crop yields over the coming decades. Three recent papers in Nature Genetics1–3 highlight this trend, which should enable crop breeders to more quickly identify the best germplasm for introgression into elite crop varieties. One of the studies1 describes the construction and use of a haplotype map for genome-wide association studies (GWAS) in rice. In the other two studies, short sequencing reads are aligned to reference genomes to detect millions of singlenucleotide polymorphisms (SNPs) in soybean2 and maize3. Collectively, these reports underscore the potential of genome resequencing for identifying superior genetic variants to be used in crop breeding (Fig. 1). The conventional method for determining the genetic basis of crop traits is linkage mapping. Carefully bred progeny lines are generated from biparental crosses, and parental haplotypes that modify traits (quantitative trait loci) are identified by tracking the recombinant chromosomal segments. This approach has ­several limitations. Resolution is constrained by the number of recent recombination events in the mapping population, and the only diversity accessible for testing is that already ­represented in the population4. In human genetics studies, Jer-Ming Chia and Doreen Ware are at the Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, USA. Doreen Ware is with the US Department of Agriculture, Agricultural Research Service, USA. e-mail: [email protected]

linkage mapping has been largely supplanted by GWAS5. But the plant community has been slow to adopt this approach, primarily owing to the scarcity of the high-resolution variation maps needed for genome-wide tests. Working with the model plant Arabidopsis thaliana, Atwell et al.6 were among the first to demonstrate the potential of GWAS in plant genetics. Using microarray technology, they genotyped 250,000 SNPs across 191 accessions. By collecting extensive quantitative data related to 107 phenotypes from each accession and performing association tests using every SNP, they identified genetic loci associated with features as diverse as developmental processes, plant architecture, disease responses and metabolism. This study effectively evaluated genotype–­phenotype associations every 500 bp in this relatively small genome (125 Mb)—a much higher density than that of typical human GWAS. For plants with larger and more complex genomes, sufficiently dense microarrays needed for high-resolution GWAS are not available. This limitation has been overcome for major crops such as rice1, soybean2 and maize3,7 through the use of reference genome sequences and next-­generation sequencing technologies. By resequencing six elite maize varieties, selected for their commercial importance and genetic relationships, Lai et al.3 identified ~1,200,000 SNPs, about a third of which lie within genes. These polymorphisms add to the pool of genetic variation previously reported for maize7 and, in conjunction with limited linkage disequilibrium, will enable high-resolution association mapping in maize. Additionally, the authors documented extensive presence/

volume 29 number 2 FEBRUARY 2011 nature biotechnology

ne w s and vie w s Genotype by resequencing

Construct genotype matrix (m loci for n individuals) X1

© 2011 Nature America, Inc. All rights reserved.

Xn

Score genotype matrix against phenotype

Genomic selection

GWAS - MAS M2 M1

u=

u1 u2

M3 M4

--

un

Estimated breeding values

Associated markers

Genotype germplasm and select based on estimated breeding values or alleles at associated markers

Figure 1 Genome resequencing enables genomic selection and marker-assisted selection. Sequencing all individual plants in an association-mapping population reveals their genotypes at polymorphic sites, which can be used as markers of phenotype (represented here as plant height and fruit color). These genotypes may be used as predictors of performance for selecting from an existing germplasm population through either genomic selection or marker-assisted selection. In a genomic selection screen, all markers are used to construct a vector of genomic estimated breeding values. GWAS identifies markers significantly associated with traits for use in marker-assisted selection.

absence variations, which have been proposed to contribute to hybrid vigor3,7,8. Many cultivated species have a narrow genetic base, and it has been recognized that sustainable agricultural production and trait improvement depends on harnessing the broader pool of natural variation9. The use of sequencing to discover variants promises to reveal much of the genetic diversity in wild species and landraces. This hitherto underutilized genetic pool can be screened to identify beneficial alleles lost during domestication that can be selectively introgressed into elite varieties. Lam et al. demonstrated the potential of this approach by resequencing 17 wild and 14 domesticated soybean varieties2. They identified genetic alleles that have not been found in domesticated varieties, with more wild-specific alleles than domesticated-specific ones (35% versus 5%). If any of these wild-specific alleles are associated with superior fitness, they could be re-introduced or introgressed through marker-assisted breeding. One limitation the authors highlight in soybean is the extensive linkage disequilibrium that limits the ability of GWAS to identify causal genes. Huang et al.1 combined genotyping by resequencing and phenotyping of rice to provide GWAS for a major crop. Most notably, their study demonstrates that costly, high-depth resequencing is not necessary for GWAS in a domesticated crop. By skim sequencing >500 varieties of rice landraces, each at a relatively low coverage (~1×), the authors identified ~3,600,000 SNPs and imputed the missing genotypes using a measure of relatedness among individuals. They used the newly discovered SNPs to drive the association mapping of 14 agronomically important traits, such as grain width, length and amylose content, and showed that the significant association signals are enriched for loci within or close to previously identified causal genes. In both this study and the A. thaliana6 association-mapping study, variations in known causal genes show up as being significantly associated with the corresponding traits. In most examples, however, the peak association signal is close to, but not within, the known causal allele or locus1,6. There are two main reasons for this. First, whereas the mixedmodel approach used in both studies controls for population structure and relatedness, it is likely that residual effects, confounded by linkage disequilibrium, remain. Furthermore, rare alleles and multiple small-effect quantitative trait loci that contribute to a phenotype may also be unaccounted for. Similar effects are the subject of much debate among human geneticists10. Notably, the rice and A. thaliana studies also demonstrate that the genetic architecture of many agronomic traits involves alleles that are found at relatively high frequency (>5%) in

nature biotechnology volume 29 number 2 FEBRUARY 2011

each population. These alleles explain a high proportion of phenotypic variance in plant studies (e.g., 6–68% in rice1), in contrast to GWAS in human populations. GWAS of human traits and diseases is motivated primarily by an interest in the underlying biology and in therapeutic applications. In plants, GWAS is likely to have its greatest impact in breeding programs. Whereas GWAS in humans is limited to naturally breeding populations, in plants it can be used to enhance agronomically important traits owing to the availability of structured association-mapping plant populations, generated using controlled crosses to provide a complete knowledge of family structure. Such study designs, also known as joint linkage-association mapping, combine the statistical power of linkage mapping with the resolution of association mapping and can transcend many of the limitations of human GWAS4,11. High-resolution knowledge of the genetic elements that contribute to a trait advances markerassisted selection by genotype alone. In addition, cost-effective high-density genotyping will enable the implementation of genomic selection approaches in crop plants similar to methods prevalent in animal breeding12. Although the breeding programs for all crops stand to benefit, we anticipate that the most pronounced gains will be made in perennial species, where selection by genotype instead of phenotype will dramatically shorten the selection cycles. Ultimately, advances in sequencing capacity will likely enable routine resequencing of every individual in an association-mapping population. This would make all rare alleles and variable genomic spaces accessible for quantitative trait loci testing, allowing a comprehensive understanding of the relative contributions made by genetics and the environment. With the imminent arrival of third-generation sequencing technologies, this reality may be but a few harvests away. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. 1. Huang, X. et al. Nat. Genet. 42, 961–967 (2010). 2. Lam, H.-M. et al. Nat. Genet. 42, 1053–1059 (2010). 3. Lai, J. et al. Nat. Genet. 42, 1027–1030 (2010). 4. Myles, S. et al. Plant Cell 21, 2194–2202 (2009). 5. McCarthy, M.I. et al. Nat. Rev. Genet. 9, 356–369 (2008). 6. Atwell, S. et al. Nature 465, 627–631 (2010). 7. Gore, M.A. et al. Science 326, 1115–1117 (2009). 8. Fu, H. & Dooner, H.K. Proc. Natl. Acad. Sci. USA 99, 9573–9578 (2002). 9. Tanksley, S.D. & McCouch, S.R. Science 277, 1063–1066 (1997). 10. Goldstein, D.B. N. Engl. J. Med. 360, 1696–1698 (2009). 11. McMullen, M.D. et al. Science 325, 737–740 (2009). 12. Jannink, J.-L., Lorenz, A.J. & Iwata, H. Brief. Funct. Genomics 9, 166–177 (2010).

139