Detection of Deleterious Genotypes in Multigenerational Studies. I. Disruptions in Individual Arabidopsis Actin Genes

Corresponding author: Richard B. Meagher, Department of Genetics, Life Sciences Bldg., University of Georgia, Athens, GA 30602-7223, meagher{at}bscr.uga.edu (E-mail).

Communicating editor: E. MEYEROWITZ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant actins are involved in numerous cytoskeletal processes effecting plant development, including cell division plane determination, cell elongation, and cell wall deposition. Arabidopsis thaliana has five ancient subclasses of actin with distinct patterns of spatial and temporal expression. To test their functional roles, we identified insertion mutants in three Arabidopsis actin genes, ACT2, ACT4, and ACT7, representing three subclasses. Adult plants homozygous for the act2-1, act4-1, and act7-1 mutant alleles appear to be robust, morphologically normal, and fully fertile. However, when grown as populations descended from a single heterozygous parent, all three mutant alleles were found at extremely low frequencies relative to the wild-type in the F₂ generation. Thus, all three mutant alleles appear to be deleterious. The act2-1 mutant allele was found at normal frequencies in the F₁, but at significantly lower frequencies than expected in the F₂ and F₃ generations. These data suggest that the homozygous act2-1/act2-1 mutant adult plants have a reduced fitness in the 2N sporophytic portion of the life cycle, consistent with the vegetative expression of ACT2. These data are interpreted in light of the extreme conservation of plant actin subclasses and genetic redundancy.

THE classic genetic approach to a problem involves isolating mutants with a predicted phenotype and then dissecting the biochemical or molecular genetic defect. However, for many genes, predicting the phenotype of a null mutation can be an impossible undertaking. The diverse Arabidopsis actin gene family (MCDOWELL et al. 1996B ) is a perfect case in point. Arabidopsis contains eight functional actin genes with several distinct but overlapping tissue-specific expression patterns. To identify insertions in target genes of interest, like actin, sequence-based screens are used to identify mutations—without any prior knowledge of phenotype (BALLINGER and BENZER 1989 ; ZWAAL et al. 1993 ; MCKINNEY et al. 1995 ). However, many of the mutations isolated in highly conserved and seemingly essential genes via such reverse genetic approaches do not result in easily identifiable phenotypes (see DISCUSSION). For genes to be highly conserved, natural selection must have acted on some resulting phenotypic trait even if the selective forces are very small (OHTA 1992 ).

The possible reduction of fitness of these nonlethal mutations is seldom considered in the current literature. In fact, mutations that are not lethal or do not produce an obvious morphological phenotype are often used to categorize genes or gene functions as nonessential or redundant (see DISCUSSION). These categorizations have hindered our understanding of the evolution and function of such genes. Undoubtedly, it is often quite difficult or time consuming to determine if homozygous wild-type individuals in a population have a survival and/or reproductive advantage over individuals with one or two copies of a particular mutant allele. Even genotypes with significantly reduced fitness may take several generations to be measurably reduced in a population. Nevertheless, without multigenerational data on allele frequencies, it is not appropriate to label genes as truly redundant or nonessential. Although several generations are brief on an evolutionary time scale, a mutation that is measurably reduced in frequency in a large population in a few generations is definitive proof that a gene is essential. In this article, we use multigenerational data to examine the fitnesses of a set of mutant actin alleles in Arabidopsis. The high fidelity of inbreeding in Arabidopsis (SNAPE and LAWRENCE 1971 ; ABBOTT and GAMES 1989 ) and its small physical size make it ideally suited for laboratory analysis of allele frequencies over multiple generations.

Quantitative evolutionary studies and gene expression studies provided strong reason to believe that the conservation of most of the eight functional Arabidopsis actin genes might be essential to the survival of the species (MCDOWELL et al. 1996B ). There are two ancient classes of actins, vegetative and reproductive, that have been preserved for at least 350–400 million years (my). These two classes differ by only about 6% at the amino acid level, and most amino acid changes are conservative ones. The actin genes can be further subdivided into five subclasses that have not shared common ancestry with their closest subfamily for 150 to 300 my and differ from each other in 2–4% of their amino acids. Three of the subclasses contain a pair of closely-related actin genes. The two members of these pairs diverged from a common ancestral gene only 30–50 my ago. For example, the ACT2/ACT8 pair and the ACT4/ACT12 pair are each highly divergent in synonymous nucleotide site substitutions, differing in 56 and 86% of these synonymous nucleotide site positions, respectively. However, ACT2 differs from ACT8 and ACT4 differs from ACT12 by only a single, conservative amino acid change (0.2% divergence). Thus, it is clear that the amino acid sequence of each of the functional Arabidopsis actins is under strong selective constraint.

Further evidence for the conservation of the actins comes from their patterns of gene expression. Each of the five subclasses is strongly expressed in a different subset of Arabidopsis tissues and organs during development (AN et al. 1996A ; AN et al. 1996B ; HUANG et al. 1996 ; MCDOWELL et al. 1996A ; HUANG et al. 1997 ). Although there is an overlap of expression in some tissues and organs (e.g., three subclasses are strongly expressed late in pollen development), these expression patterns often complement each other and at least one gene appears to be expressed in every tissue and stage of development. Furthermore, when the Arabidopsis actin regulatory sequences are used to express reporter genes in representatives of distant plant families, many of these patterns of expression are conserved (A. VITULE and R. MEAGHER, unpublished results). Based on these data and the fact that actin is an essential protein in nearly all eukaryotic cells, it is reasonable to assume that actin gene regulation is also conserved.

We report in this article on the evolutionary dynamics of Arabidopsis plant populations with mutations in actin genes from three different subclasses. Adult plants homozygous for any of these mutant alleles appear normal in morphology and temporal development. This is inconsistent with the predicted essential functions for and conservation of the Arabidopsis actin genes. Therefore, two series of experiments were undertaken to examine actin mutant allele frequencies in Arabidopsis populations descended from single heterozygous actin mutant parents. In the first series, the F₂ progeny showed greatly reduced mutant allele frequencies for all three genes, suggesting that mutations in all three actin genes are deleterious. In the second series of experiments, the act2-1 allele frequency was monitored at each of three subsequent generations. Again, the act2-1 mutant allele was greatly reduced in frequency relative to the wild-type ACT2 allele, starting with the F₂ and continuing in the F₃ generations. These data deviated significantly from the expectations under selective neutrality. In a companion article, the act2-1 data are analyzed more completely using a mathematical model that delimits the evolutionary dynamics under and several selection parameters in pure-selfing populations (ASMUSSEN et al. 1998 ).

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant lines with mutant actin alleles:
For our first set of experiments we identified and characterized mutants in three distinct actin subclasses. A T-DNA insertion mutant library was prepared in the WS ecotype of Arabidopsis thaliana using a binary plasmid with a 17-kbp transferred region. The identification and characterization of the T-DNA insertions in two actin alleles, act2-1 and act4-1, are described in MCKINNEY et al. 1995 . A similar screening and characterization approach was used to identify a T-DNA insertion in the act7-1 allele (L. U. GILLILAND, E. C. MCKINNEY, and R. B. MEAGHER, unpublished results). The act7-1 allele was obtained from a library of T-DNA transformed Arabidopsis thaliana WS lines generously supplied by David Bouchez (Institute Nationale de la Recherche Agronomique, Versailles, France). One copy of the transferred T-DNA region in this mutant library is 7.3 kbp. In this screening process, pairs of oligonucleotide primers, near the border regions of the insertion and within the target gene, were used in the polymerase chain reaction (PCR) to amplify the junction sequences. The junction between a T-DNA insertion and the neighboring actin gene was detected by Southern blotting and hybridization with an actin probe. The identity and nature of the disruption can be rapidly determined using nested sets of target primers and DNA sequencing of these products. Each mutant was first identified in a pool of 100 lines, then in a pool of 10 lines, and then from an isolated line that had been stored as seed. Based on designating the heterozygous seeds obtained from the original transformation event, the F₀ generation, the individual transformed plant lines for all three mutants were selfed through two more generations to produce the F₂ generation. Our first series of experiments examined the genotypic frequencies of adult plants from the F₂ generation seeds.

Plant lines for examining the act2-1 allele over multiple generations:
In the first series of experiments described above only F₂ generation seeds and plants were analyzed. Furthermore, the original F₀ heterozygous act2-1 and act4-1 plant lines contained extraneous T-DNA elements not associated with the actin allele (MCKINNEY et al. 1995 ). In order to eliminate any potential problems with these other elements and to collect multigenerational data, a second more complete series of experiments was conducted on the act2-1 allele. A heterozygous ACT2/act2-1 plant was selfed and screened for progeny that had lost the two superfluous T-DNA insertions by segregation but were still heterozygous for the act2-1 allele. No T-DNA insertions other than the one at the act2-1 site were detected by PCR or Southern blot in these progeny (MCKINNEY et al. 1995 ). In the resulting offspring, kanamycin resistance and the act2-1 allele always segregate together. Because these plants might have had other collateral DNA damage due to the transformation process, an act2-1/act2-1 line was backcrossed once into the wild-type WS ecotype. The genotype of heterozygous plants was determined by PCR assays (above). These heterozygous F₀ parents were selfed to produce the F₁, F₂, and F₃ generation used in multigenerational studies.

Plant growth conditions:
Seeds were sown at moderate (20–30 seeds) or high density (160–180 seeds) in 16.5 x 12-cm flats on soil and germinated at 22° with top watering every day for 5 days after vernalization at 4° for 48 hr. Plants were then grown at 22° with 12-hr day length at low light intensity (200–300 microEinsteins). We typically observed 50–80% germination of seeds and established similar percentages of adult plants for each seed density. Adult plants were watered from below until the inflorescence and siliques were fully developed (~10 wk) and then water was withheld. Although individual adults were often assayed at each generation for genotype, seeds were harvested from a flat in bulk, and a random seed sample was used for the subsequent generation.

PCR screening for genotype:
DNA to be used for screening genotype was prepared either by the method of MCKINNEY et al. 1995 or by the rapid alkali DNA screening method of KLIMYUK et al. 1993 , modified as follows. A young leaf disk was clipped using the cap of a sterile Eppendorf tube. The tissue was ground in 82 µl of 0.25 M NaOH and placed in boiling water for 30 sec. This solution was neutralized with 80 µl of 0.25 M HCl and buffered with 40 µl of 0.5 M Tris-HCl, pH 8.0, and 0.25% Nonidet P-40. The sample was boiled for 2 min and immediately used for PCR or placed at -20° or 4° for storage. All stored samples were boiled again for 2 min immediately prior to use in PCR. Then 1–2 µl of the boiled tissue slurry containing visible pieces of tissue was used as the source of DNA template in a 50-µl PCR reaction. Oligonucleotides used in the PCR reactions for ACT2 and ACT4 are described in MCKINNEY et al. 1995 and those for ACT7 are described in Table 1. The act2-1 allele was detected using an outward facing oligonucleotide from the T-DNA right border sequence (RB16843S) paired with a sense oligonucleotide (AAc2-I485S) that primes in intron L, which is located within the 5' untranslated region (UTR) of ACT2 upstream of the insertion. The wild-type ACT2 allele was detected using AAc2-I485S paired with an antisense oligonucleotide (AAc2-ATGN30) located on the junction of the intron in the 5' UTR and exon 1, a region destroyed by the insertion of the T-DNA into act2-1. The act4-1 allele was detected using an outward facing oligonucleotide from the T-DNA left border sequence (LB102A) paired with a sense oligonucleotide (ACT4-240S) that primes in exon 3 upstream of the insertion. The wild-type ACT4 allele was detected using ACT4-240S paired with an antisense oligonucleotide from the 3' UTR (ACT4-3'252A). The act7-1 allele was detected using a primer in the T-DNA left border sequence (GKBLB114A) paired with a sense oligonucleotide that primes in the third intron of ACT7 (AAc7-N3S). The wild-type ACT7 allele was detected using ACT7-N3S paired with AAc7-3'N2, which is located in the 3' UTR of exon 4. PCR reactions were carried out as described in MCKINNEY et al. 1995 except that the annealing reactions were carried out at 52° for pairs of unique oligonucleotides and 42° for reactions involving degenerate oligonucleotides, and the 72° extension reaction was shortened to 30–50 sec. This method appears to work most reproducibly for PCR products less than 700 bp in size. Individual plants being assayed in a flat were marked with toothpicks. Reactions scored as negative for one allele and positive for the other (i.e., they were homozygous for one allele or the other) were repeated twice more to confirm the presence of one allele and absence of the other.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation and characterization of Arabidopsis actin mutants:
T-DNA insertion mutant alleles were characterized in three highly divergent Arabidopsis actin genes, act2-1, act7-1, and act4-1 (see MATERIALS AND METHODS and Figure 1). ACT2 is strongly expressed in nearly all mature vegetative tissues of Arabidopsis (AN et al. 1996B ), accounting for more than 50% of the actin mRNA in roots and leaves, and remains high in older tissues. The ACT7 gene is expressed primarily in early developmental stages of nearly all vegetative tissues, including hypocotyl and seed coat, where ACT2 is not expressed. ACT7 also responds to stimulation by a variety of phytohormones that affect growth and development (MCDOWELL et al. 1996A ). The expression of ACT4 is restricted primarily to late stages in pollen development (HUANG et al. 1996 ). Therefore, based on the requirement for all eukaryotic cells and tissues for actin one might have predicted each of these genes to be essential.

View larger version (56K): In this window In a new window Download PPT slide   Figure 1. —Actin gene map and screening strategies for wild-type and T-DNA insertion alleles. (A) Map of a typical plant actin gene with the sites of T-DNA insertions in act2-1, act4-1, and act7-1 alleles indicated. The five actin exons are labeled L, 1, 2, 3, 4. The ACT2 gene lacks intron 1, fusing exons 1 and 2 indicated in this diagram. Numbers above the gene indicate codon positions in the 377-codon actin gene. Ivs, intervening sequence; L, mRNA leader; LB, left border; RB, right border; UTR, untranslated region of transcript; pA, polyadenylation sites. (B) The rapid PCR assay for the ACT2 allele is shown for eight F2 lines (MATERIALS AND METHODS). The PCR products were resolved on 1.5% agarose, stained with ethidium bromide, and their fluorescence image photographed. The ACT2 product of 492 bp was produced using an upstream sense primer in IvsL and antisense primer located in exon 1 and is scored (+ or -) below the image. A positive control assay for the ACT2 allele was made on two different ecotypes, WS and RLD. The size standards were prepared from AluI digested pBR322 plasmid. (C) The rapid PCR assay for the act2-1 allele is shown for the same eight F2 lines shown in B. The act2-1 product of 511 bp was produced from an upstream sense primer and an outward facing right border primer (RB) and is scored (+ or -) above the image. Genotypes were identified by comparing the data in B and C (+/- = A1A1, -/+ = A2A2, +/+ = A1A2). Data for the whole population are summarized in Table 3.

All three mutations have the potential to disrupt actin gene expression, as shown in Figure 1A. The act2-1 allele contains a T-DNA insertion a few hundred nucleotides downstream from the start of transcription and five nucleotides before the start codon of the ACT2 gene (MCKINNEY et al. 1995 ). The insertion event deleted the splice acceptor site of the first intron-exon junction, including 12 nucleotides of intron L, which lies in the 5' UTR of the mRNA and four noncoding nucleotides of exon 1. The insertion at the act7-1 allele lies in exon 4 of the ACT7 gene, 16 base pairs after the stop codon (L. U. GILLILAND, E. C. MCKINNEY and R. B. MEAGHER, unpublished results). This insertion has the potential to effect mRNA polyadenylation, transport, stability, and/or translation. The T-DNA insertion in the act4-1 allele lies after codon 281 in the third coding exon and disrupts the last quarter of the 377-codon open reading frame (MCKINNEY et al. 1995 ). Consistent with the observation that the actin protein is highly conserved both in sequence and length, act4-1 could not produce a functional actin protein. The T-DNA insertions in all three actin alleles are flanked on either side by two identical right border sequences (act2-1) or by two identical left border sequences (act4-1, act7-1), but no internal T-DNA junctions sequences were detected (MCKINNEY et al. 1995 ). Therefore, it is likely that portions of two T-DNA elements are inserted in each allele.

Three plant populations segregating for both the wild-type and actin mutant alleles were analyzed to determine the impact of genotype on phenotype. The F₂ generation seeds for each of the three mutant alleles descending from the three original heterozygous F₀ transgenic plants were planted on soil at moderate density (15 seeds/100 cm²) and leaves were sampled four weeks later. A PCR screening strategy was used (MATERIALS AND METHODS) to rapidly determine the genotype of large numbers of these F₂ plants from very small tissue samples. Separate combinations of PCR primers identified the intact wild-type actin gene or the fusion between the actin gene and T-DNA associated with the mutant allele. All three possible genotypes were identified in the F₂ populations for each actin mutation, based on the presence of a PCR product for the wild-type allele, and/or the mutant allele, as shown for several of the act2-1 segregating plants in Figure 1B and Figure C, respectively. The individual plant lines (lines 35–43 are shown) were positive either for ACT2 alone (homozygous wild-type; lines 36, 37, 39, 40, 41, 43), act2-1 alone (homozygous mutant; line 42), or both (heterozygous; line 38). We were surprised to find viable plants homozygous for each mutant allele (act2-1/act2-1, act7-1/act7-1, act4-1/act4-1), considering that each of these genes is well conserved in protein coding sequence. Based on a visual inspection of adult plants under a dissecting microscope, the leaves, stems, flowers, siliques, and seeds of all three homozygous mutant genotypes appeared morphologically and developmentally indistinguishable from homozygous wild-type plants. The siliques of all three homozygous actin mutants had approximately the same distribution of size and seed set as wild type.

The actin alleles appear to be deleterious mutations:
In our first series of experiments genotypic frequencies were determined for large numbers of these F₂ generation plants. The seed transformation method used to generate T-DNA insertion mutants resulted in F₀ generation seed that is heterozygous for the T-DNA insertion. Because Arabidopsis normally undergoes efficient self- fertilization (ABBOTT and GAMES 1989 ) with negligible outcrossing even under closely spaced laboratory growth conditions (SNAPE and LAWRENCE 1971 ), the number of plants that are heterozygous for these insertions should be reduced by half in each subsequent generation in the absence of selection or any other evolutionary forces. Thus, based on normal Mendelian segregation, pure selfing, and no selection for any one genotype over another, the wild-type (A₁A₁), heterozygous (A₁A₂), and homozygous mutant (A₂A₂) plants should be found in the relative frequencies of 3:2:3 in the F₂ generation. The results of analyzing the F₂ generation plants, containing the act2-1, act4-1, and act7-1 alleles, are shown in Table 2. The genotypic frequencies in all three F₂ populations had highly significant deviations from the expected 3:2:3 segregation ratio. In particular, homozygous mutant plants (A₂A₂) are found at significantly lower frequencies (0.221, 0.155, and 0.068, respectively) than the expected value of 0.375. The homozygous mutant and homozygous wild-type genotypes should be found at equal frequencies (1:1 ratio), but in fact the ratios of their frequencies in these three populations were only 0.57, 0.27, and 0.18, respectively. Based on ² analysis, segregation ratios that are this skewed for the three genotypes should be observed rarely if there was no selection, with P values on the order of 10^-2 to 10^-9 for each allele. These data suggest that these actin insertion alleles are deleterious mutations, with a reduced fitness relative to the wild-type allele.

Prompted by the low frequencies of individuals homozygous for the mutant allele in the F₂ generation for all three mutants, a second more detailed series of experiments was initiated to ascertain the onset and severity of selection against the act2-1 allele. Individual F₀ progeny were confirmed as heterozygous for the act2-1 allele by PCR analysis. The frequency of the three possible genotypes in the next three selfing generations were as shown in Table 3 and Figure 2 (A₁A₁, A₁A₂, A₂A₂, where A₁ = ACT2 and A₂ = act2-1). To perform this experiment, the F₁ seeds produced from an individual F₀ generation heterozygote were collected and sown at high density (85 seeds/100 cm²) on soil and allowed to germinate. After several weeks, a single leaf was clipped from those plants with four or more true leaves and subjected to PCR analysis for genotype. The F₁ generation genotypic frequencies did not significantly deviate from the expected Mendelian ratios with no selection: ² analysis reveals that this amount of deviation would occur more than half the time by chance alone (P = 0.63, Table 3). The F₁ plants produced F₂ generation seeds, which were collected in bulk. A total of 170 F₂ seeds were sown at high density (85 seeds/100 cm²) on soil and allowed to germinate. After several weeks, over 100 F₂ plants were sampled and analyzed for genotype. In the absence of selection, each class of homozygous plants would be equally represented [i.e., freq(A₁A₁) = freq(A₂A₂)]. However, as shown in Table 3 and Figure 2, significantly fewer act2-1/act2-1 and more ACT2/ACT2 individuals were found in the F₂ generation than would be expected strictly by chance alone. Similarly, the F₃ generation contains far fewer act2-1 homozygous plants than expected. The probabilities of obtaining this distribution of wild-type and mutant alleles in the F₂ and F₃ generations simply by chance were less than 10^-2 (P = 0.014 and P = 0.0036, respectively). Again the results deviated dramatically from the expected results without selection. The frequency of the act2-1 allele is plotted across these four generations in Figure 3. It is clear that the act2-1 allele frequency is decreasing with time starting with the F₂ generation, while the wild-type allele frequency is increasing proportionally. Note that the ACT2 and act2-1 allele frequencies in the F₂ generation agree quite well with those obtained in the first series of experiments (see open triangles in Figure 3), demonstrating the reproducibility of these skewed allelic ratios in spite of the differences in the execution of the two series of experiments.

View larger version (22K): In this window In a new window Download PPT slide   Figure 2. —Frequency of genotypes in a self-fertilizing population. The frequencies of the homozygous wild-type (ACT2/ACT2 = A1A1) and homozygous mutant (act2-1/act2-1 = A2A2) plants diverged from neutral expectations beginning with the F2 generation. The data for this figure are provided in Table 3, describing the second series of experiments. Expected values (broken lines) were calculated for four generations using Mendelian ratios for 100% self-fertilizing individuals and assuming no selection differences between genotypes or alleles. Thus, in each generation, homozygous individuals produce only homozygous individuals and heterozygous individuals are expected to yield progeny in a ratio of 1 A1A1: 2 A1A2: 1 A2A2.

View larger version (15K): In this window In a new window Download PPT slide   Figure 3. —Frequency of alleles in a self-fertilizing population. The frequency of the act2-1 mutant allele (A2) decreased, with a corresponding increase in the wild-type ACT2 allele (A1) starting with the F2 generation. These data were in sharp contrast to the constant frequencies expected if no selection acted on the act2-1 allele (horizontal broken line). F1, F2, and F3 generational data were taken from the second series of experiments as shown in Table 3. (, ) F2 generation data for these alleles are shown in Table 2 from the first series of experiments.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Disruptions in plant actin genes are deleterious:
The lack of a lethal or obvious morphological phenotype for the three actin mutant alleles was somewhat surprising. These three actin genes belong to three ancient subclasses, have strong and distinct temporal and spatial expression patterns (AN et al. 1996B ; HUANG et al. 1996 ; MCDOWELL et al. 1996), and have extremely well-conserved protein coding regions (MCDOWELL et al. 1996B ). The initial goal of examining allele frequencies was to determine if conserved genes such as ACT2, ACT4, and ACT7 were of selective importance to the survival of Arabidopsis in spite of the lack of an overt physical phenotype. In our first set of experiments, we found that mutations in these three Arabidopsis actin gene family members were in fact deleterious to the plants containing these alleles. When populations of plants were grown at moderate to high density and all the genotypes were competing for resources such as nutrients and light, the mutant alleles were selected against. All three mutant alleles examined, act2-1, act4-1, and act7-1, were measurably reduced in frequency by the F₂ generation (Table 3). Mutant alleles that are measurably lowered in frequency in fewer than 10 generations have essentially the same overall effect as a lethal mutation, because their frequencies are rapidly reduced in the population. The large reduction in frequency of the three actin alleles in the F₂ generation plants suggests that these actin mutations would be rapidly lost from natural populations.

To provide the multigenerational perspective missing from most studies of gene redundancy, we initiated a second series of experiments, starting with a plant heterozygous for the ACT2 disruption (ACT2/act2-1) in a wild-type genetic background. The frequencies of the wild-type and mutant allele were followed for three generations. The number of heterozygotes was almost as expected in each generation (Figure 2), and since only heterozygotes can give rise to heterozygotes, one might predict that the heterozygotes are at a selective disadvantage or they would have increased relative to the homozygous mutants. Furthermore, no statistically significant loss of the act2-1 allele was detected in the F₁ generation (Figure 3), suggesting that the two alleles are transmitted at equal frequencies by heterozygous plants. However, a significant reduction in the act2-1 alleles relative to the ACT2 allele was detected in the F₂ and F₃ generations. In contrast to F₁ plants, the frequencies of the genotypes deviated significantly from Mendelian expectations without selection for these two generations. These data suggest that a phenotypic effect for the act2-1 mutation occurs during the 2N sporophytic portion of the plant life cycle, consistent with the vegetative expression pattern of ACT2. The simplest interpretation would be that the act2-1 allele is a deleterious recessive. Future analysis of mixed populations will be needed to confirm this view. The phenotype producing differences in survival or reproduction of these plants could be a subtle morphological defect not detected in our initial analysis. Alternatively, the effect could be physiological and require more detailed analysis at the cellular or molecular level. Because we had no simple quantitative way to interpret the deviations from Mendelian segregation ratios, a mathematical model for selfing populations was developed that dissects the fertility, viability, and meiotic drive parameters affecting allele frequencies. This model is described in a companion article (ASMUSSEN et al. 1998 ).

Functional redundancy:
Arabidopsis, Drosophila, and yeast systems are able to produce a wealth of phenotypes based on mutations of single genes, and many such mutant genes have been characterized at the sequence level. However, with the advent of reverse genetics technology and sequence-based methods of identifying or constructing mutants, more and more genes are identified first by sequence and only characterized later by genetics (BROOKFIELD 1997 ). For example, in the "simple" eukaryotic system of yeast, no detectable phenotype was found for a majority of disruptions of new genes identified as long open reading frames in yeast chromosome III (OLIVER et al. 1992 ). This indicates that some degree of redundancy is quite common. However, in most such yeast genetics experiments, mutants are scored as viable or not, and perhaps scored as slow growing relative to wild type, but no formal fitness measurements are made. This apparent redundancy may be acting on different levels and to different extents within the biological systems in which each gene is expressed. If we define "true" redundancy as the case where each gene can function with full fitness on its own, then true redundancy should be difficult to maintain over time (NOWAK et al. 1997 ). In contrast, genes with subtle differences can be maintained through minute fitness advantages for the apparently redundant wild-type condition (BROOKFIELD 1997 ). These fitness differences may be reflections of any of a number of functional requirements for two or more genes, like the need for a cumulative amount of products, for process fidelity, or for the individual divergent properties of the products (THOMAS 1993 ). Selection for the presence of duplicate genes could be nearly neutral, acting weakly over many generations (OHTA 1992 ) or contingent upon special environmental conditions (KIMURA 1991 ; MEAGHER 1995 ) and still preserve this genetic redundancy in the organism. In the case of actin, there should be strong selection to maintain cytoskeletal processes such as cytoplasmic streaming, positioning the division plane, and programmed cell wall development (MEAGHER 1991 ; MEAGHER and WILLIAMSON 1994 ).

Proposed redundancies among genes are often based on similarities in protein structure, overlapping expression patterns, and/or even the presence of alternate or parallel pathways. Cytoskeletal protein genes such as actin, tubulin, and many others are nearly always found in conserved gene families in multicellular organisms. Null mutations in Dictyostelium genes for actin cross-linking proteins and a myosin heavy chain all have very mild phenotypes (DE LOZANNE and SPUDICH 1987 ; WITKE et al. 1992 ). Mice lacking a vimentin (an intermediate filament) gene expressed at narrow windows in early development exhibit no obvious phenotype (COLUCCI-GUYON et al. 1994 ). Null mutants for one of only two ß-tubulin genes in Aspergillus are viable, and experiments demonstrate the gene products are functionally interchangeable (MAY 1989 ), even though functional differences between -tubulins are seen in Drosophila (HUTCHENS et al. 1997 ). Even in yeast, where only one actin gene is present in the genome, functional redundancy is found between actin modulating factors in terms of genetic suppression and synthetic lethality (AYSCOUGH and DRUBIN 1996 ). The functional distinction among gene family members may be their novel expression patterns. This is indicated from promoter-swapping experiments with homeobox genes and transcription factors, which show that despite diverged coding sequences, the essential differences in these genes reside in their regulation (LI and NOLL 1994 ; HANKS et al. 1995 ; WANG et al. 1996 ). Interestingly, RUDNICKI et al. 1992 found a related family member, Myf5, was upregulated when the transcription factor MyoD was inactivated. While no phenotype is observed for MyoD-inactivated mice, a decreased fitness for the null mutants is still observed in the form of a decreased juvenile frequency. In summary, while tantalizing evidence on genetic redundancy abounds, the biological and evolutionary implications of proposing redundancy, when an obvious phenotype is not observed, are only now being appreciated.

Challenges to the dissection of actin gene function:
Our current study could be strengthened by the analysis of many more independent and well-characterized mutant alleles. T-DNA insertions in nonessential genes that did not produce a deleterious phenotype would make convincing negative controls. They could be used to demonstrate, for example, that the resident kanamycin resistance gene, NPTII, carried by the T-DNA was not deleterious. A recent study demonstrates that there is no difference in the survival or seed set between field-grown wild-type and T-DNA-transformed Arabidopsis plants expressing NPTII (PURRINGTON and BERGELSON 1997 ). However, this study did not follow the plants in closely spaced, mixed populations for multiple generations, as those examined herein. Further analysis of a variety of mutant alleles could be used to demonstrate that copy number or size of the insertion did not have global effects on the survival of transgenic plants. Preliminary mapping data on the three actin alleles discussed herein suggest that they each contain portions of at least two T-DNA elements, but each showed a distinct loss of fitness. The act7-1 allele, with the smallest T-DNA element (see MATERIALS AND METHODS), appeared to be under the strongest negative selection. In addition, while the act2-1 mutation appeared to affect the sporophytic generation, one might expect the act4-1 mutation to affect the gametophytic generation, because ACT4 is a pollen-specific gene. More detailed information will be needed on the relative survival of mutant alleles in heterozygous plants to examine gametophytic effects. In addition, multiple independent actin alleles with the same deleterious effects in populations are needed to confirm the work presented herein and to further dissect actin function. Of greatest importance, mutants in the remaining five expressed actins are needed to further characterize the functional requirements for a plant actin gene family and explain the need for such strong conservation of actin sequence.

Assessing redundancy as an applied problem:
With the rapid development of crop plants engineered with multiple transgenes inserted at a variety of loci, there are very practical reasons for having a more detailed understanding of genetic redundancy. Robust transgenic crop plants with transgenes inserted into seemingly unimportant loci could have a significant drop in survival rate in the field when competing with weeds, fighting off pathogens, and responding to extremes in climate. Our present knowledge of genetic redundancy in most systems is insufficient to make molecular- or cell-biological predictions about the effects of losing ostensibly redundant alleles. Quantitative multigenerational studies on plant populations will undoubtedly help to explain why such large numbers of insertion mutations in well-conserved genes have no obvious morphological phenotype. The roles of fertility, viability, and meiotic drive need to be considered. Arabidopsis, with its wealth of mutants, streamlined genome, small physical size, habit of inbreeding, and short generation time, can play a leading role in the population studies necessary to dissect genetic redundancy.

	ACKNOWLEDGMENTS

We thank KENNETH FELDMAN, DAVID BOUCHEZ, and BEATRICE COURTAIL for providing DNA samples from their Arabidopsis T-DNA libraries, and MIKE ARNOLD and GAY GRAGSON for critiquing the manuscript. WYATT W. ANDERSON and SUE WESSLER provided helpful suggestions and encouragement to this research. This work was funded by grant support from the National Institutes of Health to R.B.M. and from the National Science Foundation to M.A.A.

Manuscript received January 27, 1998; Accepted for publication March 11, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ABBOTT, R. J. and M. F. GAMES, 1989  Population genetic structure and outcrossing rate of Arabidopsis thaliana (L.) Heynh. Heredity 62:411-418.

AN, Y.-Q., S. HUANG, J. M. MCDOWELL, E. C. MCKINNEY, and R. B. MEAGHER, 1996a  Conserved Expression of the Arabidopsis ACT1 and ACT3 actin subclass in organ primordia and mature pollen. Plant Cell 8:15-30[Abstract].

AN, Y.-Q., J. M. MCDOWELL, S. HUANG, E. C. MCKINNEY, and S. CHAMBLISS et al., 1996b  Strong, constitutive expression of the Arabidopsis ACT2/ACT8 actin subclass in vegetative tissues. Plant J. 10:107-121[Medline].

ASMUSSEN, M. A., L. U. GILLILAND, and R. B. MEAGHER, 1998  Detection of deleterious genotypes in multigenerational studies. II. Theoretical and experimental dynamics with selection and selfing. Genetics 149:727-737[Abstract/Free Full Text].

AYSCOUGH, K. and D. G. DRUBIN, 1996  Actin: general principles from the studies in yeast. Annu. Rev. Cell Dev. Biol. 12:129-160[Medline].

BALLINGER, D. G. and S. BENZER, 1989  Targeted gene mutations in Drosophila.. Proc. Natl. Acad. Sci. USA 86:9402-9406[Abstract/Free Full Text].

BROOKFIELD, J. F. Y., 1997 Genetic redundancy, pp. 137–155 in Advances in Genetics, edited by J. C. HALL, T. FRIEDMANN, J. C. DUNLAP and F. GIANNELLI. Academic Press, San Diego.

COLUCCI-GUYON, E., M.-M. PORTIER, I. DUNIA, D. PAULIN, and S. POURNIN et al., 1994  Mice lacking vimentin develop and reproduce without an obvious phenotype. Cell 79:679-694[Medline].

DE LOZANNE, A. and J. A. SPUDICH, 1987  Disruption of the Dictyostelium myosin heavy chain gene by homologous recombination. Science 236:1086-1091[Abstract/Free Full Text].

HANKS, M., W. WURST, L. ANSON-CARTWRIGHT, A. B. AUERBACH, and A. L. JOYNER, 1995  Rescue of the En-1 mutant phenotype by replacement of En-1 with En-2.. Science 269:679-682[Abstract/Free Full Text].

HUANG, S., Y.-Q. AN, J. M. MCDOWELL, E. C. MCKINNEY, and R. B. MEAGHER, 1996  The Arabidopsis ACT4/ACT12 actin gene subclass is strongly expressed in post–mitotic pollen. Plant J. 10:189-202[Medline].

HUANG, S., Y.-Q. AN, J. M. MCDOWELL, E. C. MCKINNEY, and R. B. MEAGHER, 1997  The Arabidopsis ACT11 actin gene is strongly expressed in tissues of the emerging inflorescence, developing ovules, and pollen. Plant Mol. Biol. 33:125-139[Medline].

HUTCHENS, J. A., H. D. HOYLE, F. R. TURNER, and E. C. RAFF, 1997  Structurally similar Drosophila -tubulins are functionally distinct in vivo. Mol. Biol. Cell 8:481-500[Abstract].

KIMURA, M., 1991  Recent development of the neutral theory viewed from the Wrightian tradition of theoretical population genetics. Proc. Natl. Acad. Sci. USA 88:5969-5973[Abstract/Free Full Text].

KLIMYUK, V., B. J. CARROLL, C. M. THOMAS, and J. D. G. JONES, 1993  Alkali treatment of rapid preparation of plant material for reliable PCR analysis. Plant J. 3:493-494[Medline].

LI, X. and M. NOLL, 1994  Evolution of distinct developmental functions of three Drosophila genes by aquistition of different cis-regulatory regions. Nature 367:83-87[Medline].

MAY, G. S., 1989  The highly divergent ß-tubulins of Asperigillus nidulans are functionally interchangeable. J. Cell Biol. 109:2267-2274[Abstract/Free Full Text].

MCDOWELL, J., Y.-Q. AN, E. C. MCKINNEY, S. HUANG, and R. B. MEAGHER, 1996a  Preferential expression of the Arabidopsis ACT7 actin gene in developing tissues. Plant Physiol. 111:699-711[Abstract].

MCDOWELL, J. M., S. HUANG, E. C. MCKINNEY, Y.-Q. AN, and R. B. MEAGHER, 1996b  Arabidopsis thaliana contains ten actin genes encoding six ancient protein subclasses. Genetics 142:587-602[Abstract].

MCKINNEY, E. C., N. ALI, A. TRAUT, K. A. FELDMANN, and D. A. BELOSTOTSKY et al., 1995  Sequence based identification of T-DNA insertion mutations in Arabidopsis: actin mutants act2-1 and act4-1.. Plant J. 8:613-622[Medline].

MEAGHER, R. B., 1991  Divergence and differential expression of actin gene families in higher plants. Int. Rev. Cytol. 125:139-163[Medline].

MEAGHER, R. B., 1995 The impact of historical contingency on gene phylogeny: Plant actin diversity, pp. 195–215 in Evolutionary Biology, edited by M. K. HECHT, R. J. MACINTYRE and M. T. CLEGG. Plenum Press, New York.

MEAGHER, R. B., and R. E. WILLIAMSON, 1994 The plant cytoskeleton, pp. 1049–1084 in Arabidopsis, edited by E. MEYEROWITZ and C. SOMERVILLE. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

NOWAK, M. A., C. BOERLIJIST, J. COOKE, and J. M. SMITH, 1997  Evolution of genetic redundancy. Nature 388:167-171[Medline].

OHTA, T., 1992  The nearly neutral theory of molecular evolution. Annu. Rev. Ecol. Syst. 23:263-286.

OLIVER, S. G., Q. J. M. VAN DER AART, M. L. AGOSTONI-CARBONE, M. AIGLE, and L. ALBERGHINA et al., 1992  The complete DNA sequence of yeast chromosome III. Nature 357:38-46[Medline].

PURRINGTON, C. B. and J. BERGELSON, 1997  Fitness consequences of genetically engineered herbicide and antibiotic resistance in Arabidopsis thaliana.. Genetics 145:807-814[Abstract].

RUDNICKI, M. A., T. BRAUN, S. HINUMA, and R. JAENISCH, 1992  Inactivation of MyoD in mice leads to upregulation of the myogenic HLH gene Myf-5 and results in apparently normal muscle development. Cell 71:383-390[Medline].

SNAPE, J. W. and M. J. LAWRENCE, 1971  The breeding system of Arabidopsis thaliana.. Heredity 27:299-302.

THOMAS, J. H., 1993  Thinking about genetic redundancy. TIG 9:395-399.

WANG, Y., P. N. J. SCHNEGELSBURG, J. DAUSMAN, and R. JAENISCH, 1996  Functional redundancy of the muscle-specific transcription factors Myf5 and myogenin. Nature 379:823-825[Medline].

WITKE, W., M. SCHLEICHER, and A. A. NOEGEL, 1992  Redundancy in the microfilament system: abnormal development of Dictyostelium cells lacking two F-actin cross-linking proteins. Cell 68:53-62[Medline].

ZWAAL, R. R., A. BROEKS, J. VAN MEURS, J. T. M. GROENIN, and R. H. A. PLASTERK, 1993  Target-selected gene inactivation in Caenorhabditis elegans by using a frozen transposon insertion mutant bank. Proc. Natl. Acad. Sci. USA 90:7431-7435[Abstract/Free Full Text].

This article has been cited by other articles:

				L. U. Gilliland, M. Magallanes-Lundback, C. Hemming, A. Supplee, M. Koornneef, L. Bentsink, and D. DellaPenna Colloquium Paper: Genetic basis for natural variation in seed vitamin E levels in Arabidopsis thaliana PNAS, December 5, 2006; 103(49): 18834 - 18841. [Abstract] [Full Text] [PDF]

				F. Roux, C. Camilleri, S. Giancola, D. Brunel, and X. Reboud Epistatic Interactions Among Herbicide Resistances in Arabidopsis thaliana: The Fitness Cost of Multiresistance Genetics, November 1, 2005; 171(3): 1277 - 1288. [Abstract] [Full Text] [PDF]

				N. Gass, T. Glagotskaia, S. Mellema, J. Stuurman, M. Barone, T. Mandel, U. Roessner-Tunali, and C. Kuhlemeier Pyruvate Decarboxylase Provides Growing Pollen Tubes with a Competitive Advantage in Petunia PLANT CELL, August 1, 2005; 17(8): 2355 - 2368. [Abstract] [Full Text] [PDF]

				M. K. Kandasamy, E. C. McKinney, R. B. Deal, and R. B. Meagher Arabidopsis ARP7 Is an Essential Actin-Related Protein Required for Normal Embryogenesis, Plant Architecture, and Floral Organ Abscission Plant Physiology, August 1, 2005; 138(4): 2019 - 2032. [Abstract] [Full Text] [PDF]

				T. Nishimura, E. Yokota, T. Wada, T. Shimmen, and K. Okada An Arabidopsis ACT2 Dominant-Negative Mutation, which Disturbs F-actin Polymerization, Reveals its Distinctive Function in Root Development Plant Cell Physiol., November 15, 2003; 44(11): 1131 - 1140. [Abstract] [Full Text] [PDF]

				L. U. Gilliland, M. K. Kandasamy, L. C. Pawloski, and R. B. Meagher Both Vegetative and Reproductive Actin Isovariants Complement the Stunted Root Hair Phenotype of the Arabidopsis act2-1 Mutation Plant Physiology, December 1, 2002; 130(4): 2199 - 2209. [Abstract] [Full Text] [PDF]

				C. Ringli, N. Baumberger, A. Diet, B. Frey, and B. Keller ACTIN2 Is Essential for Bulge Site Selection and Tip Growth during Root Hair Development of Arabidopsis Plant Physiology, August 1, 2002; 129(4): 1464 - 1472. [Abstract] [Full Text] [PDF]

				K. Mouline, A.-A. Very, F. Gaymard, J. Boucherez, G. Pilot, M. Devic, D. Bouchez, J.-B. Thibaud, and H. Sentenac Pollen tube development and competitive ability are impaired by disruption of a Shaker K+ channel in Arabidopsis Genes & Dev., February 1, 2002; 16(3): 339 - 350. [Abstract] [Full Text] [PDF]

				M. K. Kandasamy, L. U. Gilliland, E. C. McKinney, and R. B. Meagher One Plant Actin Isovariant, ACT7, Is Induced by Auxin and Required for Normal Callus Formation PLANT CELL, July 1, 2001; 13(7): 1541 - 1554. [Abstract] [Full Text] [PDF]

				K. S. Mysore, J. Nam, and S. B. Gelvin An Arabidopsis histone H2A mutant is deficient in Agrobacterium T-DNA integration PNAS, January 18, 2000; 97(2): 948 - 953. [Abstract] [Full Text] [PDF]

				D. B. Szymanski, M. D. Marks, and S. M. Wick Organized F-Actin Is Essential for Normal Trichome Morphogenesis in Arabidopsis PLANT CELL, December 1, 1999; 11(12): 2331 - 2348. [Abstract] [Full Text]

				R. B. Meagher, E. C. McKinney, and M. K. Kandasamy Isovariant Dynamics Expand and Buffer the Responses of Complex Systems: The Diverse Plant Actin Gene Family PLANT CELL, June 1, 1999; 11(6): 995 - 1006. [Full Text]

				D. Bouchez and H. Höfte Functional Genomics in Plants Plant Physiology, November 1, 1998; 118(3): 725 - 732. [Full Text]

				M. A. Asmussen, L. U. Gilliland, and R. B. Meagher Detection of Deleterious Genotypes in Multigenerational Studies. II. Theoretical and Experimental Dynamics with Selfing and Selection Genetics, June 1, 1998; 149(2): 727 - 737. [Abstract] [Full Text] [PDF]

				M. K. Kandasamy, E. C. McKinney, and R. B. Meagher Functional Nonequivalency of Actin Isovariants in Arabidopsis Mol. Biol. Cell, January 1, 2002; 13(1): 251 - 261. [Abstract] [Full Text] [PDF]