INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Maize (Zea mays) is one of the most important food crops in the world. However, like most other cereals, its nutritional value for monogastric animals is low because the major storage proteins of the seed, the prolamins or zeins, are devoid of several essential amino acids, Lys being the most limiting (Nelson, 1969). Mertz et al. (1964) showed that the opaque2 (o2) mutation can nearly double the Lys content of the endosperm, compared with wild type. The O2 gene encodes a transcriptional activator that controls expression of several zein genes, especially those encoding the 22-kD -zeins (Kodrzycki et al., 1989; Schmidt et al., 1990). The o2 mutation typically reduces -zein content by one-half and enhances the synthesis of a number of non-zein proteins (Damerval and deVienne, 1993; Habben et al., 1993). Both effects contribute to the higher Lys content of the mutant endosperm (Moro et al., 1996).

Understanding the basis for the higher Lys content of o2 endosperm could provide an approach for selecting maize genotypes with better protein quality. Habben et al. (1995) showed that the protein synthesis factor elongation factor 1 (eEF1A) can be significantly increased in o2 mutants, and its concentration is highly correlated with the Lys content of the endosperm. Moro et al. (1996) analyzed a diverse set of normal and o2 maize inbreds with extensive variability in zein, non-zein, Lys, and eEF1A content and found a consistently high correlation (r = 0.9) between eEF1A and Lys content. This correlation exists even though eEF1A itself accounts for only about 1% of the endosperm protein and 2.3% of the endosperm Lys content (Sun et al., 1997). Thus, there appears to be a stochiometric relationship between eEF1A and the other major proteins that contribute to the Lys content of the endosperm.

eEF1A appears to be a multifunctional protein. It is one of the components of EF1, the protein synthesis factor that binds aminoacyl-tRNAs to the ribosome during the process of protein synthesis (Browning, 1996), but it also appears to have several other activities. eEF1A is associated with the centromere and mitotic apparatus of sea urchin (Stronglyocentrotus purpuratus) eggs (Kuriyama et al., 1990; Ohta et al., 1990), the endoplasmic reticulum (ER) of Chinese hamster (Cricetulus grisens) fibroblast cells (Hayashi et al., 1989), and the plasma membrane of carrot (Daucus carota) suspension cells (Yang et al., 1993). It is capable of in vitro interactions with a number of proteins, including actin (Yang et al., 1990; Sun et al., 1997), tubulin (Durso and Cyr, 1994), and calmodulin (Kaur and Ruben, 1994). In maize endosperm, eEF1A is associated with a network of F actin surrounding the rough ER (RER) at sites where protein bodies are forming (Clore et al., 1996). In this case, eEF1A appears to be part of a cytoskeletal network involved in zein biosynthesis (Stankovic et al., 2000).

It is unclear whether the diverse set of biological activities of eEF1A results from one or more isoforms of the protein. eEF1A is subject to several types of posttranslational modifications, including methylation of Lys residues (Hiatt et al., 1982), addition ofphosphoglycerylethanolaminetoGluresidues (Whiteheart et al., 1989), and phosphorylation (Venema et al., 1991). In higher eukaryotes, eEF1A is typically encoded by a multigene family. In maize, there are 10 to 15 genes, five of which are expressed in the endosperm (Carneiro et al., 1999). Thus, the different biological activities ascribed to eEF1A could result from one or more posttranslational modifications of the protein or expression of different eEF1A genes.

Although the biological significance of the variation in eEF1A content in maize endosperm is unclear, the phenotypic variability that exists provides an opportunity to investigate the genetic basis of these differences and the potential to use eEF1A selection to create genotypes with higher Lys content. The development of simple sequence repeats (SSRs) for maize greatly facilitates genetic mapping studies because the procedure is PCR-based and requires small amounts of DNA. SSRs are short repeating units of 1 to 5 nucleotides that are dispersed throughout the genome (Tautz and Renz, 1984; Wang et al., 1994). They serve as highly reproducible, codominant genetic markers, making their application to genetic linkage analysis straightforward (for review, see Powell et al., 1996; Senior et al., 1996). SSRs are relatively abundant in the maize genome; currently there are more than 1,500 mapped SSRs in public maize genome databases (http://www.agron.missouri.edu/cgi-bin/sybgwmdb).

We crossed several maize o2 inbreds that differ in the level of eEF1A protein (Moro et al., 1995). Self-pollinated ears from F₂ progeny were phenotyped with regard to eEF1A content and leaf DNA was used in conjunction with informative SSR markers to genotype the plants. Two quantitative trait loci (QTLs) that account for 25% of the variation in eEF1A content were identified. One of these is linked with a complex locus encoding the 22-kD -zeins on the short arm of chromosome 4, whereas the other is near the centromere on the long arm of chromosome 7.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

To investigate the genetic basis of the phenotypic variation in eEF1A content in maize endosperm, we created F₁ and F₂ progeny from two sets of inbreds that differ in eEF1A concentrations (Moro et al., 1996). As shown in Figure 1A, Oh51Ao2 contains more than twice the concentration of eEF1A as Oh545o2, whereas the eEF1A content in CM105o2 endosperm is somewhat less than 50% higher than that in Va99o2 (Fig. 1C). The level of eEF1A in the reciprocal F₁ crosses of Oh51Ao2 and Oh545o2 shows an incompletely dominant effect that relates to gene dosage: In Oh545o2 × Oh51Ao2, the eEF1A level is between the mean value of the two parents and the high parent, Oh51Ao2, whereas in Oh51Ao2 × Oh545o2 it is similar to the high parent. Because the level of eEF1A cannot be accurately measured in individual F₂ endosperms, ears of F₂ plants were self-pollinated and endosperm flour was prepared from a pool of 20 kernels taken from the central region of well-filled ears. For the cross of Oh51Ao2 and Oh545o2, 106 and 69 well-filled F₂ ears were harvested in the spring and fall seasons of 1996, respectively, and 148 F₂ ears were phenotyped for the cross of CM105o2 and Va99o2 in the spring of 1996. The level of eEF1A in the F2:3 progeny of both crosses showed continuous variation (compare with Fig. 1, A, B, and C) that ranged between the phenotypes of the parents. Because there was less phenotypic variability in the F₂ progeny of the CM105o2 × Va99o2 cross than the Oh51Ao2 × Oh545o2 cross, we focused on the latter for a QTL mapping study.

View larger version (63K): [in this window] [in a new window]   Figure 1.   Relative concentration of eEF1A protein in endosperms from F2 progeny of Oh51Ao2 × Oh545o2 and CM105o2 × Va99o2. The height of each bar indicates the relative content of eEF1A in flour samples of the designated parents; F1 and F2 progeny as determined by ELISA. The measure of eEF1A protein in an equal mixture of endosperm flour from the two parents was assigned a value of "1," and the eEF1A content in the progeny was indexed to this standard. The measurements were based on the average of two independent extractions from a pool of 20 kernels. A, Oh51o2 × Oh545o2, spring of 1996; B, Oh51o2 × Oh545o2, fall of 1996; C, CM105o2 × Va99o2, spring of 1996.

A linkage map of informative SSR markers was created for the Oh51Ao2 × Oh545o2 cross by testing approximately 300 SSR primer pairs on the parental DNAs. Approximately 70% of the SSRs were polymorphic, and 83 of the informative markers that are well distributed throughout the genome were used to create a linkage map (Fig. 2). The 83 polymorphic SSR markers cover a total of 1,402.4 cM of the maize genome with an average interval of 16.9 cM. Chromosome 1 had the lowest density of markers and averaged 24 cM between SSRs. The average interval between markers for the other chromosomes was very close to 16 cM, and these were all generally well spaced. It was difficult to identify polymorphic markers near the centromeres of chromosomes 1 and 5. The average interval between markers for chromosome 6 was 14 cM, but there were two regions of approximately 40 and 35 cM where no polymorphic SSRs could be identified.

View larger version (36K): [in this window] [in a new window]   Figure 2.   Ten linkage groups of the maize genome based on polymorphic SSR marker analysis of Oh51o2 and Oh545o2. The map consists of 83 informative markers identified from an analysis of over 300 SSRs. The markers cover 1,402 cM of the maize genome, with an average interval of 16.9 cM.

To simplify the QTL analysis, we genotyped 40 F₂ progeny that comprised the 20 highest and the 20 lowest eEF1A-containing individuals. By using Map Manager software (http://mcbio.med.buffalo.edu/mmQT.html), we identified several SSRs linked with the QTLs. All available flanking polymorphic SSRs subsequently were also tested for linkage. All the SSRs found to be linked with QTLs were used to genotype the entire F₂ population and interval mapping was performed.

The genotypic analysis of the F₂ population identified two QTLs that account for approximately 25% of the phenotypic variation for eEF1A content (Fig. 3, Table I). Interval mapping identified regions on chromosomes 4 and 7 that have LRS values of 14.0 and 17.7, respectively. To establish significant threshold values for the LRS, interval mapping was conducted with a free regression model, and permutation tests (1,000 shuffles) on individual chromosomes were done to establish the significant threshold value of LRS (Churchill and Doerge, 1994; Doerge and Churchill, 1996). This analysis established 12.1 and 11.2 as significant (95%) LRS values for the QTLs on chromosomes 4 and 7, respectively. As a consequence, both regions contain significant QTLs. The QTL on the short arm of chromosome 4 contributes 11% of the variance for eEF1A content, and its effect is primarily additive rather than dominant. The QTL on the long arm of chromosome 7 contributes 14% of the variance for eEF1A content, and it has primarily an additive effect.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Interval mapping of QTLs associated with eEF1A content on chromosomes 4 and 7. These plots are derived from interval mapping of loci associated with eEF1A content in the F2 population of Oh51Ao2 × Oh545o2. The solid curve illustrates the likelihood ratio statistic (LRS); SSR DNA markers are indicated on the left. The significant threshold of LRS resulting from the permutation test is 12.1 and 11.3 for the loci on chromosomes 4 and 7, respectively (shown with dashed lines).

We performed a chi-square test on the linkage of SSR markers flanking the two QTLs among the high- and low-eEF1A genotypes (Table II). Among the 20 high eEF1A individuals, the phi026 locus on chromosome 4 had a chi-square value of 10.6 (P < 0.01), and the phi072 locus had a chi-square value of 4.8 (P < 0.1). As a consequence, phi026 cannot be discounted as a significantly linked flanking marker, although phi072 does not appear to be as tightly linked. Based on the chi-square test, neither of these SSRs is significantly linked with the trait among the low-eEF1A individuals. The phi114 locus and the umc1666 loci on chromosome 7 have chi-square values of 9.9 (P < 0.01) among the 20 low-eEF1A individuals, and neither of these loci is significantly linked among the high eEF1A individuals. Based on this analysis, the QTL on chromosome 4 contains an allele from Oh51Ao2 that is responsible for the high eEF1A content, whereas the QTL on chromosome 7 contains an allele from Oh545o2 that is responsible for the high eEF1A content.

The QTL on the short arm of chromosome 4 is linked with a cluster of 22-kD -zein genes (Llaca and Messing, 1998), and the QTL near the centromere of chromosome 7 is near the locus encoding the 27-kD -zein (Benner et al., 1989). To assess the significance of these linkages, we analyzed the zein proteins from endosperms of Oh51A+ and o2 and Oh545+ and o2. Figure 4A illustrates a Coomassie Blue-stained gel showing the relative amount of - and -zeins in these inbreds. As is common, both o2 mutants show a significant reduction in -zein protein compared with their wild-type counterparts, with a noticeable reduction in the level of 22-kD -zeins (Moro et al., 1995). However, compared with the wild type, there is much more of a reduction in -zein synthesis in Oh545o2 than Oh51Ao2. There appears to be a slight increase in the synthesis of 27-kD -zein in Oh51Ao2 compared with Oh51A+, but the relative level of 27-kD -zein is higher in Oh545+ compared with Oh545o2.

View larger version (47K): [in this window] [in a new window]   Figure 4.   SDS-PAGE and ELISA measurement of - and -zein content in Oh51Ao2, Oh545o2, and their F2 progeny with extremely high and low concentrations of eEF1A. A, For each lane, total zein extracts were prepared from 750 µg of endosperm flour of Oh51A+ and o2 and Oh545+ and o2 and separated by 12.5% (w/v) SDS-PAGE. Mr standards are shown on the right. To measure the amount of - and -zein, samples were prepared from endosperm flour obtained from 20 kernels of the ear from an F2 plant. Equal amounts of flour were mixed from the 10 high-eEF1A or the 10 low-eEF1A individuals to create the high- and low-eEF1A pools; triplicate measurements of zein proteins were made for each pooled sample. The -zein (B) and -zein (C) contents were determined by ELISA as described in "Materials and Methods."

Because Coomassie Blue-stained gels only indicate the relative levels of - and -zeins, we performed an ELISA to obtain quantitative measurements of these proteins. In this case, we compared Oh51Ao2 and Oh545o2 and pooled flour samples from the 10 high-eEF1A and the 10 low-eEF1A individuals. Figure 4B shows that the relative level of -zein proteins is nearly 10 times greater in Oh51Ao2 than Oh545o2. In addition, the -zein level is about 50% greater in the flour from the pool of high-eEF1A individuals than the pool of low-eEF1A individuals. It is interesting that this trend is reversed for -zein content (Fig. 4C). There is approximately 40% more -zein in Oh545o2 than Oh51Ao2, but there appeared to be an insignificant difference in the level of -zeins in the high- and low-eEF1A pools.

Because of the similarity between -zeins and -zeins, our antisera do not distinguish between polypeptides within these structural classes. Therefore, to more specifically assay zein gene expression in Oh51Ao2 and Oh545o2, we analyzed the level of - and -zein RNAs in developing endosperms of the two inbreds. The northern blot in Figure 5 shows there is a higher level of 27-kD -zein transcripts in Oh545o2 than Oh51Ao2 at both 15 and 20 d after pollination (DAP); the phosphoimager measurement indicated 0.5 times more radioactivity at 15 DAP and 2.5 times more radioactivity at 20 DAP. However, the transcription of -zein RNAs is dramatically lower in Oh545o2 compared with Oh51Ao2. The 22-kD -zein RNAs were nearly undetectable in Oh545o2 at 15 and 20 DAP. Small amounts of these transcripts were measured in Oh51Ao2 at 15 DAP, and significantly more RNA was present by 20 DAP. Phosphoimager measurements showed 20 times more radioactivity hybridizing at 15 DAP and 35 times more radioactivity hybridizing at 20 DAP in Oh51Ao2 compared with Oh545o2. Transcripts of 19D -zeins, a distinct sequence homology group (Marks et al., 1985), were detectable in Oh545o2 by 20 DAP, but these RNAs accumulated to a much higher level in Oh51Ao2. The phosphoimager measurement with this probe showed 130 times more radioactivity at 15 DAP and 14 times more radioactivity at 20 DAP in Oh51Ao2. Thus, these data corroborate significantly higher levels of -zein gene expression, especially for the 22-kD -zeins, in the Oh51Ao2 parent, but a higher level of 27-kD -zein gene expression in the Oh545o2 parent.

View larger version (63K): [in this window] [in a new window]   Figure 5.   RNA-blot analysis showing - and -zein transcripts in developing endosperm of Oh545o2 and Oh51Ao2. Total RNA was extracted from developing endosperm at 15 and 20 DAP, and 10 µg was separated by formaldehyde gel electrophoresis. After blotting on nylon membranes, the 19D -, 22 -, and 27 -zein transcripts were detected by hybridization with 32P-labeled cDNA probes. Sample loading was standardized with an 18S rDNA probe.

The reduction in -zein synthesis in o2 mutants is associated with a 2- to 4-fold decrease in protein body size compared with the wild type (Geetha et al., 1991). Because of the contrasting differences in -zein gene expression and protein levels in Oh545o2 and Oh51Ao2, it was of interest to compare the relative sizes of protein bodies in these inbreds. Protein bodies were isolated from homogenates of 20-DAP endosperm by Suc gradient centrifugation (Habben et al., 1993) and fixed and embedded for transmission electron microscopy (TEM) analysis as previously described (Lending et al., 1988). In addition, cross sections of 20-DAP kernels were observed by scanning electron microscopy (SEM) to examine the size variation of intact protein bodies.

Figure 5 shows representative SEMs of protein bodies from Oh51Ao2 and Oh545o2 and the mean diameter of protein bodies isolated from 20-DAP endosperm homogenates. The SEM analysis examined tissue from the side of the kernel, beneath the subaleurone, because this region has the highest concentration of protein bodies (Lending and Larkins, 1989). The TEM analysis examined a random mixture of protein bodies from the entire endosperm. SEM showed that although there was some variation in protein body sizes in each inbred, they tended to be fairly uniform and smaller than those in the wild type, which averaged about 1 to 2 µM in diameter (data not shown). Based on measurements of approximately 500 protein bodies from each genotype, the mean diameter of protein bodies in Oh51Ao2 is 0.32 µM, whereas the mean diameter in Oh545o2 is 0.38 µM. Because the protein bodies are essentially spherical, these measurements indicate the protein bodies in Oh51Ao2 have approximately 40% less volume, on average, than those in Oh545o2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

The negative pleiotropic effects of the o2 mutant, such as reduced kernel density, lower yield, and greater susceptibility to insect and mechanical damage, significantly limited its widespread utilization in maize breeding programs. However, the creation of modified o2 mutants, so-called quality protein maize, which manifest a normal kernel phenotype while maintaining an elevated Lys content (Gevers and Lake, 1992; Villegas et al., 1992), has increased interest in the agronomic development of this mutant. However, optimum development of quality protein maize requires selection of genotypes with an even higher Lys content.

The relationship between the eEF1A concentration and the Lys content of maize endosperm provides the basis for a simple and inexpensive method to screen maize germplasm for genotypes with high levels of Lys-containing proteins (Habben et al., 1995; Moro et al., 1996). However, the ELISA procedure for measuring eEF1A requires a significant amount of time and repetitive sample preparation to evaluate breeding materials. As a consequence, it would be valuable to identify QTLs that influence eEF1A content, so the corresponding loci can be selected by breeding programs aimed at developing higher protein quality maize genotypes. In this way, valuable alleles could be transferred to other inbreds by recurrent backcrossing and marker-assisted selection, thus greatly reducing the breeding time (Frisch et al., 1999).

SSR genetic markers have been widely used for genome mapping, but they have only recently been developed for maize (Senior and Heun, 1993; Chin et al., 1996). Since the initiation of this research project, more than 1,500 maize SSRs were identified and made publicly available, and consequently we were able to rely exclusively on SSRs to create a uniform linkage map for the Oh51Ao2 × Oh545o2 cross.

View larger version (92K): [in this window] [in a new window]   Figure 6.   SEM and TEM analysis of protein bodies in 20-DAP endosperm of Oh51Ao2 and Oh545o2. Representative SEM images show protein bodies (p) and starch granules (s) in 20-DAP endosperms of Oh51Ao2 (A) and Oh545o2 (B). Both micrographs were taken at 4,000 magnification (bar in B = 2.5 m). Diameters of approximately 600 protein bodies were measured in five transmission electron micrographs of Oh51Ao2 (slashed bar) and Oh545o2 (solid bar) and analyzed for size distribution (C) and mean diameter; SEs are indicated (D).

We used a selective genotyping strategy to increase the efficiency of mapping QTLs influencing eEF1A content (Lander and Botstein, 1989; Darvasi and Soller, 1992; Nandi et al., 1997). This decreased the time and expense of mapping, and it also helped resolve the problem of bias estimation for linked QTLs caused by selective genotyping (Lin and Ritland, 1996). This QTL mapping procedure (Haley and Knott, 1992) has the advantages of likelihood of odds (LOD) mapping (Lander and Botstein, 1989), but with more speed and simplicity of analysis (Kearsey and Farquhar, 1998). The QTLs associated with variation in eEF1A content, which are on the short arm of chromosome 4 and the long arm of chromosome 7, have LRS values of 14 and 17.7, respectively. These values are higher than the threshold of significance given by the permutation test (Fig. 3). When the LRS value is converted to an LOD score by dividing with a factor of 4.61 (Lander and Botstein, 1989), the LOD value for these QTLs is 3 and 3.8, respectively. These values are greater than or equal to the LOD value threshold 2 to 3, which was suggested as significant for QTLs by Lander and Botstein (1989).

The two QTLs we identified account for approximately 25% of the variability for eEF1A content in this population. Although this is only a portion of the phenotypic variability, our results compare favorably with many other QTL analyses. Based on an analysis of 176 mapping studies, QTLs commonly account for only about 50% of the phenotypic variability (Kearsey and Farquhar, 1998). In our case, there are several reasons why we could not identify QTLs for additional phenotypic variability. First, the 83 polymorphic SSRs we identified did not effectively cover some regions of the maize genome. Thus, some loci could have been overlooked in our analysis. Second, the size of our mapping population would not have allowed the detection of minor QTLs. We planted more than 200 F₂ kernels from the Oh51Ao2 × Oh545o2 cross, but only 106 well-filled ears were recovered for DNA and eEF1A analysis. There is also a limitation to our mapping data due to environmental effects. It would be useful to reanalyze this population after growing it multiple seasons, but this would entail significant effort. We are currently developing recombinant inbred lines from the F₂ progeny of the Oh51Ao2 × Oh545o2 cross, and these materials can eventually be used to recalculate the effect of each QTL and possibly identify minor QTLs. Although each parent is homozygous for a "high" QTL allele at one locus and a "low" QTL allele at the other, the F₂ individuals did not show significant transgressive segregation for eEF1A content (deVicente and Tanksley, 1993). One explanation for this result is a negative interaction between the QTLs, which is possible considering they could both be related to storage protein synthesis, and hence compete for common substrates. It is also possible the level of eEF1A in Oh51Ao2 is near a maximum for this tissue. Oh51Ao2 had the highest concentration of eEF1A among the nearly 100 inbred lines we characterized (Moro et al., 1996). However, additional data are required to evaluate this relationship.

One value of QTL mapping is the potential to identify genes responsible for a trait of interest. The strategy of identifying candidate genes by QTL mapping has been used in maize (Pelleschi et al., 1999), Arabidopsis (Swarup et al., 1999), humans (Nicolaides et al., 1997), and cattle (Parmentier et al., 1999). Neither of the loci we identified correspond to regions where maize eEF1A genes were mapped (Carneiro et al., 1999). As a consequence, variation in eEF1A alleles does not appear to directly explain the phenotypic variation in eEF1A protein. It is interesting that the two QTLs we found are associated with loci encoding zein storage proteins, and it is possible that differences in zein gene expression and protein body formation influence the level of eEF1A. The QTL on chromosome 4 is linked with a cluster of -zein genes that is 2.5 cM away from the glyceraldehyde-3-phosphate dehydrogenase 1 locus from which the phi026 marker was developed (Fig. 2). The QTL on the long arm of chromosome 7 is near the 27-kD -zein locus; however, we were unable to identify an SSR marker that anchors this gene. As a consequence, it was of interest to examine the pattern of expression of the 22-kD -zein and 27-kD -zein genes in the high- (Oh51Ao2) and low- (Oh545o2) eEF1A parents.

The O2 gene encodes a transcription factor that regulates -zein gene expression (Kodrzycki et al., 1989), particularly the genes encoding 22-kD polypeptides (Schmidt et al., 1992). We would expect to find low levels of -zein synthesis in Oh51Ao2 and Oh545o2 compared with their wild type counterparts (Fig. 4), but it was surprising to find such a large difference (10-fold) between the levels of -zein proteins in these inbreds. The ELISA assay we used to measure -zein protein did not allow us to estimate how much the 22-kD -zeins account for this difference; however, analysis of -zein RNA transcripts in developing endosperms showed a significantly lower level of 22-kD -zein gene expression in Oh545o2 compared with Oh51Ao2. There is a high level (90%) of sequence identity between genes encoding 22-kD -zeins (Marks et al., 1985), and the probe we used should cross-hybridize with all the 22-kD -zein RNAs encoded at the locus on chromosome 4 (Llaca and Messing, 1998). As a consequence, it appears that this locus is expressed 20 to nearly 40 times higher in Oh51Ao2 compared with Oh545o2. Although there also appears to be lower levels of 19-kD -zeins in Oh545o2 compared with Oh51Ao2, the difference is not as great as for the 22-kD -zeins. Nevertheless, we cannot be sure the differences we measured with the 19-kD -zein probe are representative because this probe would not be expected to cross-hybridize with transcripts of other subfamilies of 19-kD -zein genes (Marks et al., 1985).

Although there are significant differences in - and -zein protein and transcript levels in Oh51Ao2 and Oh545o2 and -zein content in the pools of high- and low-eEF1A F₂ progeny from their cross, whether or not the levels of these storage proteins cosegregate with the two alleles associated with each QTL remains to be determined. We have limited amounts of endosperm flour from the original F2:3 endosperm because these samples were also used to investigate segregation for free Lys levels (Wang and Larkins, 2001; Wang et al., 2001). We have developed 75 recombinant inbred lines from the F₂ progeny, and we plan to make ELISA measurements of the - and -zein proteins and level of eEF1A in these lines and determine their linkage with the QTL flanking markers we identified.

The mechanism by which protein bodies assemble is not fully understood, but it could relate to interactions between - and -zeins (Coleman and Larkins, 1999). -Zeins appear to initiate protein body assembly and provide the mechanism for ER retention of -zeins, whereas -zeins comprise most of the protein body filler (Lending and Larkins, 1989; Coleman et al., 1996). Our previous research showed a relationship between -zein content and protein body number (Geetha et al., 1991; Dannenhoffer et al., 1995). Nevertheless, we have no information regarding the nature of - and -zein interactions in a protein body, nor whether protein body formation requires stochiometric concentrations of these proteins. We observed a small, but significant, difference in expression of the 27-kD -zein gene in Oh545o2 and Oh51Ao2, which was detected at both the transcript and protein levels. Nevertheless, the level of -zein protein was not greatly different between the high- and low-eEF1A progeny pools. It could be significant that the QTL on the long arm of chromosome 7 inherited from the Oh545Ao2 parent is near the -zein locus. It is unfortunate that we have not been able to identify a SSR marker that anchors the -zein locus that would allow us to investigate this relationship in more detail.

One explanation for the relationship between eEF1A and maize endosperm Lys content is the observation that eEF1A appears to be associated with a cytoskeletal network that surrounds the RER at sites where protein bodies are forming (Clore et al., 1996). Components of the cytoskeleton would be expected to contain Lys, and their mass would significantly contribute to the total Lys content of the endosperm (Sun et al., 1997). Genotypes that generate a larger number of protein bodies, especially with a large surface to volume ratio as is the case in o2 mutants (Geetha et al., 1991), would be expected to develop a more extensive cytoskeleton and hence a higher Lys content. It is technically difficult to count and accurately measure the surface area of protein bodies and their associated rough ER (RER), and this task is confounded by the variation in protein body number in different cells and regions of the endosperm. The SEM and TEM analyses we performed provide an approximation of the average volume of the protein bodies in these inbreds. Based on this analysis, which showed that protein bodies in Oh51Ao2 are slightly smaller in diameter than those in Oh51Ao2, and the measurements showing a higher -zein content in Oh51Ao2, it appears there are more protein bodies and hence more RER surface area in Oh51Ao2 than Oh545o2. Although these results appear to contradict what we would have expected based on previous studies of - and -zein mutants (Geetha et al., 1991; Dannenhoffer et al., 1995), they simply underscore our ignorance of the mechanisms that determine protein body assembly. Because of the difference in protein body size, the similarity in protein body density, and the difference in total zein content (approximately 4 mg in Oh545o2 and 6 mg in Oh51Ao2), we calculated that Oh51Ao2 has approximately 80% more RER surface area than Oh545o2. This could explain a significant portion of the increased eEF1A content of Oh51Ao2. It will be possible to investigate these relationships in more detail using the recombinant inbred lines developed from this cross.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Materials

Two populations of F₂ plants were created from F₁ seeds obtained from the following crosses: Oh51Ao2 (high eEF1A content) and Oh545o2 (low eEF1A content) and CM105o2 (high eEF1A content) and Va99o2 (low eEF1A content). F₂ seeds of each cross were planted in the spring and fall of 1996 and the ears self-pollinated, harvested, and air-dried. Twenty F3 kernels from each F₂ plant (F2:3) were selected from the middle of well-filled parental and progeny ears for analysis of eEF1A content. The kernels were degermed and a mixed sample of the ground endosperms was prepared as described by Moro et al. (1996). Developing kernels were harvested at 15, 20, and 25 DAP from self-pollinated ears grown in the greenhouse at the University of Arizona Campus Agricultural Center.

DNA Extraction and PCR Analysis

Young leaves from F₂ plants were lyophilized with a speed vacuum dryer at 40°C. DNA was prepared by the hexadecyltrimethyl-ammonium bromide method (Shen et al., 1994) and diluted to a final concentration of 20 ng mL¹ for PCR reactions. SSR primers were synthesized by Life Technologies (Grand Island, NY) or obtained from Research Genetics (Huntsville, AL) or Pioneer Hi-Bred International (Johnston, IA). The primer sequences are available in the Maize Genome Database (http://nucleus.agron.missouri.edu/cgi-bin/ssrbin.pl). Selection of SSR primers was based on the Maize Microsatellite-RFLP consensus map and mapped SSRs described in the Maize Genome Database. If SSR primers did not yield polymorphic PCR products, other markers from the same region were tested. PCR reactions were initiated by denaturing the DNA at 95°C for 5 min, followed by 30 cycles of PCR as follows: 94°C, 1 min; 56°C, 1 min; and 72°C, 1.5 min. The final cycle was extended at 72°C for 5 min. Reactions were conducted in 0.2-mL thin-walled PCR tubes in a GeneAmp PCR System 9600 (Applied Biosystems, Foster City, CA). Each reaction contained 20 ng of maize (Zea mays) DNA, 1.5 µL of 10× PCR reaction buffer, 0.5 µL of 50 mM MgCl₂, 20 pmol of forward and reverse primers, and 0.25 units of Platinum Taq DNA polymerase (Life Technologies); a final volume of 15 µL was made with double distilled water. Following DNA amplification, the PCR products were separated by electrophoresis in 4% (w/v) agarose and visualized by staining with 0.01 µg of ethidium bromide per milliliter of gel (Chin et al., 1996).

Selective Genotyping and Interval Mapping

Linkage maps of the 10 maize chromosomes were created based on the genotypes of polymorphic markers from the F₂ population of Oh51Ao2 × Oh545o2. A selective genotyping strategy initially was used, based on the analysis of the 20 highest and 20 lowest eEF1A phenotypes (Lander and Botstein, 1989). A first round of simple interval mapping was performed to identify potential QTLs linked with eEF1A content. When QTLs with an LRS value larger than 10 (Haley and Knott, 1992) were detected in the high- and low-eEF1A samples, the remaining F₂ individuals were then genotyped with other flanking markers. In addition, all available SSRs between the flanking markers were tested, and the informative markers were used to genotype the entire population. Finally, the genetic distance between each marker was recalculated and the effect of the identified QTL was reevaluated based on the entire F₂ population (Darvasi and Soller, 1992). Permutation tests were performed to establish the significant threshold value of LRS for each chromosome (Churchill and Doerge, 1994; Doerge and Churchill, 1996).

Map Manager QTXb03 (http://mcbio.med.buffalo.edu/mmQT.html) was used to create linkage maps and a simple interval mapping method was used to detect QTLs. The order of SSR markers on maize chromosomes is known (http://www.agron.missouri.edu/cgi-bin/sybgwmdb/), so the distance between flanking markers could be calculated based on the genotype of the F₂ individuals. Because the nature of QTL gene action (i.e. additive, dominant, etc.) was unknown, a free regression model was used to perform interval mapping. Analyses for variance and regression were performed using the data analysis software package in Excel (Microsoft, Redmond, WA).

Estimation of eEF1A Content by ELISA

eEF1A content of maize endosperm flour was determined by ELISA, similar to that described by Habben et al. (1995) and Moro et al. (1996). Protein was extracted from duplicate samples of endosperm flour of 20 pooled F2:3 kernels as described by Wallace et al. (1990). Each extract was diluted 1,000-fold in carbonate coating buffer (CCB; Clark et al., 1986), and 50 µL of the sample was mixed with 100 µL of CCB in the well of an ELISA plate (Immulon2, Dynatech Laboratories, Inc., Chantilly, VA). After all the primary samples were loaded, a multichannel pipette was used to make four 3-fold dilutions into adjacent wells containing CCB. The protein was allowed to bind to the plate overnight at 4°C; subsequently, the wells were washed twice using Tris-buffered saline containing 0.05% (v/v) Tween 20. The rabbit eEF1A antiserum (Habben et al., 1995) was diluted (1:1,000) in TTBS and 100 µL added to each well. Following incubation for 4 h, the primary antibody was removed, the wells were washed twice with TTBS, and the secondary antibody, goat anti-rabbit IgG alkaline conjugate (Sigma Chemical Co., St. Louis) in TTBS was added and allowed to bind for 2 h. The dilution of the secondary antibody was 1:1,000. After removal of the secondary antibody, the wells were washed twice with TTBS and 200 µL of alkaline phosphatase substrate (Sigma), diluted in diethanolamine substrate buffer (Clark et al., 1986), was added. The color reaction was allowed to develop for 30 to 45 min, and the absorbency was read at 410 nm with an ELISA plate reader (MR700, Dynatech).

The range of protein concentrations for the ELISA assay was such that the relationship between absorbency and relative antigen concentration was linear; an analysis for regression was performed. The slope of the regression is proportional to the antigen content, and it was used to measure the relative concentration of eEF1A. The assay was standardized to the amount of eEF1A in an equal mixture of endosperm flour from the parental genotypes. The corresponding ELISA reading was given a value of "1" for the purpose of comparing progeny samples.

SDS-PAGE and ELISA Measurement of - and -Zein Content

Protein was isolated from endosperms of Oh545o2 and Oh51Ao2 as described by Wallace et al. (1990). Fifty milligrams of flour was extracted overnight with 1 mL of borate buffer at 37°C, and soluble proteins were partitioned into zeins and non-zeins with addition of absolute ethanol at 70% (w/v) final concentration. Zein proteins were concentrated by lyophilization, dissolved in Laemmli (1970) buffer, and stored at 20°C until used. The proteins were analyzed with 12.5% (w/v) SDS-PAGE and stained with Coomassie Blue.

Measurement of - and -zein content by ELISA was similar to the procedure described for eEF1A, with some modifications (Moro et al., 1995). Following the initial extraction from endosperm flour, the extract was diluted 1:5,000 rather than 1:10,000. Samples for -zein analysis (dilution and binding to wells) were diluted in CCB, whereas samples for -zein analysis were diluted in 40% (v/v) ethanol and 10% (w/v) acetic acid (Wallace et al., 1990). Rabbit anti-- and --zein sera (Wallace et al., 1990) were diluted 1:2,000 and 1:1,000, respectively, in TTBS.

RNA Extraction and Northern-Blot Analysis

Total RNA was extracted from 5 g of 15- and 20-DAP endosperm and analyzed as described by Liu and Zhu (1997). Ten micrograms of total RNA was separated by formaldehyde-agarose gel electrophoresis and blotted onto a nylon membrane. Ethidium bromide (1.5 mg) was added to each sample to visualize the RNA and estimate equivalent sample loading; an 18S ribosomal genomic DNA probe was used to assess equal concentrations of RNA in each sample. Clones encoding 19D , 22  (Marks et al., 1985), and 27 -zein cDNAs were labeled with ³²P-dCTP by random priming. Radioactivity hybridizing to the nylon membrane was detected by x-ray film exposure and measured with a phosphoimager.

Protein Body Isolation and Analysis

Protein bodies were isolated from 20-DAP kernels of Oh51Ao2 and Oh545o2 by Suc density gradient centrifugation and processed for TEM as previously described (Lending et al., 1988). In brief, protein bodies were recovered from the Suc gradient with a Pasteur pipette, diluted with distilled water, and concentrated by centrifugation in microfuge tubes to form pellets about 2 mm in diameter and 0.5 mm thick. These pellets were fixed in freshly prepared 4% (v/v) paraformaldehyde and 1% (w/v) glutaraldehyde in 50 mM potassium phosphate and 5 mM EGTA buffer (pH 7), overnight at 4°C. Fixative was diluted away in the same buffer by three washes for 10 min each. The pellets were fixed in 2% (v/v) osmium tetroxide for 2 h at 4°C. Dehydration of the pellets was carried out by passage through a series of ethanol concentrations, followed by infiltration with 50% (w/v) LR White resin in ethanol for 1 h at 4°C and then 100% (w/v) LR White resin for 8-16 h. Thin sections were cut and stained with 5% (v/v) aqueous uranyl acetate for 5 min, rinsed with distilled water three times, and then briefly stained with Reynold's lead citrate. Photographs were taken of representative areas of sections from five different grids for each genotype at 10,000× magnification. Negatives were enlarged and printed such that each photograph usually contained 100 to 130 protein bodies. Diameters of protein bodies were measured in five photographs for each genotype to estimate mean size.

For SEM, 0.5-mm-thick slices were excised midway between the crown and base of 25-DAP kernels. These sections were processed similar to those for TEM, up to the infiltration step. The kernel sections were incubated in 100% (v/v) hexamethyldisilazane overnight at 4°C, were cracked in liquid nitrogen, and then thawed in hexamethyldisilazane. The tissue blocks were air dried overnight, coated with 30-nm-thick gold pallidium with a Hummer 6.2 Sputtering System (Anatech LTD, Alexandria, VA) and examined with an ISI WB6 SEM (Anatech LTD) at 4,000× magnification at 10 kV. Representative images were photographed from the fifth to seventh starchy endosperm cell layers.