Introduction

The four brown midrib (bm) mutants of maize (Zea mays L.) have brown vascular tissue in the leaf and stem, which is related to a reduction in content and change in subunit composition of the cell wall polymer lignin (Kuc et al, 1968). Lignin is important for the survival of the plant because of its role in water transport, rigidity and defense against pests and pathogens. On the other hand, the presence of lignin has been associated with reduced forage quality (Cherney et al, 1991). The lignin monomers, p-coumaryl alcohol, coniferyl alcohol and sinapyl alcohol, are synthesized via the concerted action of the shikimic acid and phenylpropanoid pathways. After synthesis and transport to the cell wall these monomers undergo polymerization via an oxidative coupling mechanism. Recent progress in understanding lignin biosynthesis has been reviewed by Baucher et al (1998) and Whetten et al (1998).

The Brown midrib genes play a role in the biosynthesis of lignin, and analyzing the lignin of the bm mutants is helpful in understanding the role of the functional genes and the proteins they encode. The bm mutants offer the possibility to increase understanding of lignification and to help identify how changes in lignin affect other parts of plant metabolism. The bm1 mutant incorporates coniferaldehyde residues in the cell wall and stains more brightly in the Wiesner reaction (Provan et al, 1997; Halpin et al, 1998). The Wiesner reaction is based on the formation of a red chromophore resulting from the reaction between phloroglucinol and coniferaldehyde end groups under acidic conditions (Geiger and Fuggerer, 1979; Provan et al, 1997). The changes in lignin composition in the bm1 mutant are correlated with a reduced activity of the enzyme cinnamyl alcohol dehydrogenase (CAD) (Halpin et al, 1998). The bm2 mutant has a reduction in guaiacyl residues (Chabbert et al, 1994) and a disturbance in the tissue-specific patterns of lignification (Vermerris and Boon, 2001). The bm3 mutant accumulates 5-hydroxyguaiacyl residues in its lignin (Lapierre, 1993) as a result of a mutation in the gene encoding caffeic acid O-methyltransferase (Vignols et al, 1995). The bm4 mutant is the least characterized brown midrib mutant and little is currently known about its chemical composition. All four brown midrib mutations are considered to be completely recessive (Neuffer et al, 1997), since the brown midrib phenotype is only visible in plants that are homozygous for the mutant allele.

The bm mutations have been shown to affect flowering time (Vermerris and McIntyre, 1999). The bm2 mutation delayed flowering time, while the bm1 mutation tended to expedite it. There is additional evidence for a relationship between cell wall composition and flowering. Maize plants with the bm3 mutation were reported to flower later than their wild-type counterparts (Gentinetta et al, 1990) and a quantitative trait loci (QTL) affecting flowering time was identified near the bm4 locus (Koester et al, 1993). A survey of the Maize Genetics Database (www.agron.missouri.edu) reveals that QTL for flowering time map near all four maize brown midrib loci.

In trying to understand the relationship between lignin biosynthesis and flowering several possibilities need to be considered. Lignin content per se is an unlikely factor, since the bm1-bm2 double mutant, which has the lowest lignin content among the genotypes that were analyzed, did not show significant changes in flowering time (Vermerris and McIntyre, 1999). It is, however, possible that flowering time is influenced by lignin composition. As lignin plays a role in water transport, it is possible that certain alterations in lignin composition can mimic drought stress. The water potential is known to affect flowering dynamics. In addition to a reduction in plant height (Sari-Gorla et al, 1999), drought conditions tend to result in increased silking delay (Moss and Downey, 1971; Herrero and Johnson, 1981). The silking delay, or anthesis-silking interval (ASI), refers to the time that elapses between pollen shedding and the appearance of silks on the ear of the same plant. This is the result of a reduction in the time to anthesis, and an increase in the time to silking (Ribaut et al, 1996; Veldboom and Lee, 1996). Alterations in lignin composition that would simulate drought stress are therefore anticipated to decrease flowering time and increase silking delay.

Another possibility is that variation in flowering time is the result of changes in growth and development prior to flowering. For example, an accelerated growth rate would expedite the overall maturation of the plant, including flowering.

In this study the role of the Bm1 locus on growth and development and on cell wall composition was examined by comparing near-isogenic lines with zero, one or two bm1 mutant alleles. These experiments were designed to begin to understand the nature of the relationship between cell wall biogenesis and flowering.

Materials and methods

Plant material

The bm1 seed was derived from a seed stock that also contained the linked mutations anthocyaninless2 (a2) and purple aleuronel (pr1). This seed stock was kindly provided by Dr RL Phillips (University of Minnesota, St Paul, MN, USA) and represented BC6F1 seed in an A619 background. A bm1 near-isogenic line (NIL) in an A619 background was produced from this stock by additional backcrossing. The recessive mutations a2 and pr1 affect the color of the seed. The a2-bm1-pr1 BC6F1 seeds were yellow. Ten plants grown from these seeds were backcrossed to A619 (female), which resulted in all purple BC7F1 kernels. Thirty BC7F1 plants (heterozygous for all three loci) were grown from this seed. All 30 plants were self pollinated, resulting in seeds that were segregating for kernel color (yellow, purple, and purple seeds with brown pericarps). The yellow seeds were either a2-pr1 or wild-type. In order to identify bm1 mutants flanked by the wild-type alleles of A1 and Pr1, 200 yellow BC7F2 seeds (from three BC7F1 plants) were planted. The bm1 mutants among them were identified, self pollinated and crossed to inbred line A619. The majority of the bm1 plants produced colored seeds from one or both of these crosses, but five bm1 plants produced all yellow seeds from both the self pollination and the cross to A619, and were bm1 NILs. The seed produced from the self pollination was used for the analysis of bm1 homozygotes, whereas the seeds from the cross to A619 were used as bm1 heterozygotes. The same seed sources were used in all experiments.

Field design

Twenty seeds of each of the three genotypes, wild type, bm1/+, and bm1, were planted in a Latin square the summer of 1999 in Clayton, NC (Latin square I). Fifty-two plants reached anthesis. Two Latin squares containing the same three genotypes were planted in the summer of 2000 in Throckmorton, IN (Latin square IIA, B) where 48 and 50 plants, respectively, reached anthesis. The plants were irrigated and fertilized according to standard nursery procedures.

Data collection

Flowering time (anthesis) and silking time were recorded by daily survey of individual plants and measured in days after planting. Flowering was defined as the emergence of pollen-shedding anthers from the tassel, and silking was defined as the emergence of the first silks from the ear. The ASI was defined as the time difference (in days) between flowering time and silking time. The height at anthesis was recorded for each individual plant. The development stage for each individual plant in Latin Square IIA and B was recorded every other day using the method described in Special Report no. 48 (Iowa State University Cooperative Extension Service, 1997). Staging data were recorded beginning at stage V2. Two data collection days were missed, one due to pesticide treatment and another due to inclement weather.

Statistical analysis

The flowering time, silking time, and time to each developmental stage were modeled using survival analysis as described by Vermerris and McIntyre (1999). In a survival analysis the time to an event is modeled without assumption of normality for the outcome. In the case of flowering time and time to developmental stages the distribution was often skewed. In order to carry out an analysis of variance (ANOVA), the data were log-transformed and then the log of the time-to-event was modeled. The ASI was normally distributed and was modeled using an ANOVA. All statistical analyses were performed using the SAS software package (SAS Institute, Cary, NC, USA).

Fourier-transform infrared spectroscopy

Fourier transform infrared (FT-IR) spectroscopy was performed in order to compare the chemical composition of the cell walls of the three genotypes. FT-IR spectroscopy is a method that relies on the absorption of energy from infrared light as a result of changes in the vibration energy of chemical bonds. These vibrations can either involve the distance between the atoms (stretch) or the angle between the atoms (bend), but can only be detected with FT-IR spectroscopy if they also involve a net change in the dipole moment. The difference in signal between the transmitted and detected light at different frequencies can be represented in a spectrum, and thus represents information on the chemical composition of the sample. Various functional groups have characteristic absorbances and their presence can be used for diagnostic purposes. FT-IR spectra of ligno-cellulosic materials are typically complex, because of overlap in signal between various functional groups (Faix, 1992; Séné et al, 1994; Stewart et al, 1997).

Among several different analytical methods available to characterize cell wall composition FT-IR spectroscopy was selected, because it is uniquely suitable to assay quantitative variation between cell wall samples without chemical extraction of particular cell wall fractions. Furthermore, FT-IR spectroscopy can also detect changes in interactions between cell wall compounds even if the actual concentrations of particular compounds are not altered (Séné et al, 1994). Furthermore, the analysis requires only small amounts of tissue.

When the tassel had just emerged (VT stage), vascular tissue from the midrib of leaf 6 was dissected with a scalpel and cut in sections of 3–5 mm. The sections were soaked in water (2 × 30 min) to remove water-soluble compounds, and in acetone (2 × 30 min) to remove unbound phenolic compounds. Samples of 2.0, 3.5 and 5.0 mg of dissected wild-type tissue were ground, mixed with 250 mg dry KBr powder and pressed into a thin pellet. The FT-IR spectrum was recorded using a Spectrum GX FT-IR spectrometer (Perkin Elmer; Norwalk, CT, USA). A pellet of 250 mg KBr without maize tissue was used as a blank. Nitrogen gas was used in the sample chamber and detector to warrant a consistent atmosphere. A total of 64 scans were recorded between 4000 and 370 cm−1 at a resolution of 2 cm−1. In the spectrum of the 2.0 mg sample the peak resolution was not as good as in the spectrum of the 3.5 mg sample. The 5.0 mg sample produced a spectrum that looked saturated. The comparisons between genotypes were therefore made with samples of 3.5 mg.

The midribs of two plants of each genotype were analyzed in duplicate. The data were acquired using Spectrum v. 3.02 software (Perkin Elmer) and analyzed with the Grams v.6.0 software package (Galactic Industries, Salem, NH, USA). Comparisons between spectra (in the range between 1800 and 750 cm−1) of the three genotypes were based on spectra that were baseline corrected and area-normalized. An ANOVA was used to evaluate overall differences in the absorbance at several diagnostic frequencies.

Wiesner reactions

Wiesner reactions were performed by incubating dissected midribs in 0.2 ml 1% (w/v) phloroglucinol in ethanol:concentrated HCl (3:1). After 10 min incubation the supernatant was removed and the samples were rinsed in water.

Results

Maize plants heterozygous at the Bm1 locus flower earlier than their homozygous counterparts

Table 1 shows the mean number of days until each developmental stage, flowering time, and silking time for Latin Square IIA and IIB. Mean flowering time in Latin square I for A619, bm1 and bm1/+ was 74.8, 73.7, and 71.5 days after planting, respectively, with a standard deviation of 2.77, 1.16, and 1.36 days. Survival analyses were performed to compare flowering time and silking time of the three genotypes. Hazard ratios appeared to be proportional, allowing the use of the semi-parametric Cox-proportional hazards model. The parametric models using a log-logistic, log-normal, and generalized gamma distribution were also considered. Table 2 summarizes the results from the flowering and silking time analysis. Based on the log-likelihood values the three parametric models perform equally well. The P-values for the test of equality of genotypes (bm1 compared with wild type, and bm1/+ compared with wild type) using the different survival models, and the ANOVA results for log-transformed flowering time are also listed in Table 2. From these results it is clear that the heterozygous plants are flowering earlier than the wild-type plants. It is noteworthy that this is irrespective of the model that was chosen. Additional contrasts were conducted to test whether the bm1 heterozygotes also flowered earlier than the bm1 homozygotes. In most models this difference was not statistically significant (data not shown). A graphic representation of the predicted survival curves shows a similar pattern for all models. Figure 1 shows the predicted survival curves of the wild-type, bm1 and bm1/+ plants from the three replicates using the log-logistic survival model. In all three replicates the predicted survival curves show that bm1 heterozygote flowers earlier than the bm1 mutant and the bm1 mutant flowers earlier than the wild type. The difference between the bm1 mutant and the wild type is statistically significant in Latin Square IIA. According to some models the silking time of the bm1/+ plants is significantly earlier than the wild-type plants in Latin square IIB (P-values 0.01–0.051), but not in Latin square IIA (P-values 0.41–0.78). The silking time of the bm1 mutants is not significantly different from the wild-type plants in either of the two replicates. Despite the lack of statistically significant differences, silking time follows the same basic pattern as the flowering time.

Table 1 Means and standard deviations (SD) for the time to each developmental stage, flowering time, and silking time (in days after planting) in Latin squares IIA and IIB. The asterisks indicate missing data
Full size table
Table 2 Comparison of different models to estimate differences in flowering time and silking time for the bm1/bm1 and bm1/+ plants compared with the wild-type plants
Full size table
Figure 1
figure 1

Predicted probability of not flowering using the log-logistic survival model for the wild-type plants (solid line), bm1/bm1 plants (dashed line), and bm1/+ plants (dot-dashed line) in (a) Latin square I, (b) Latin square IIA and (c) Latin square IIB. Time is indicated in days after planting (dap).

Full size image

The bm1 heterozygote has a distinct cell wall composition

Figure 2 shows dissected midribs of wild-type, bm1/+ and bm1 plants. The bm1 samples have a distinct brown color relative to the yellow color of the wild-type samples. The bm1/+ samples look similar to the wild-type samples. After incubation of the samples with the Wiesner reagent the bm1 samples are stained more intensely than the wild-type and bm1/+ samples. This indicates that the cell wall of the bm1 mutant contains higher levels of coniferaldehyde residues. The results from the Wiesner reaction are consistent with the notion that the bm1 mutation is recessive (Neuffer et al, 1997).

Figure 2
figure 2

Dissected midribs from leaf 6 of (a) wild-type, (b) bm1/+ and (c) bm1/bm1 plants prior to (top panel) and after (bottom panel) staining with the Wiesner reagent. The midribs of the bm1 mutant are amber-brown, unlike the midribs of the bm1 heterozygote. The midribs of the bm1 mutant display a more intense staining with the Wiesner reagent, whereas the midribs from the wild type and bm1 heterozygote are similar. The bar indicates a length of 1 cm.

Full size image

The fact that the bm1 heterozygotes flower earlier than both homozygotes suggests a dosage effect. The Wiesner reaction is primarily a qualitative assay, and despite lack of variation in color prior to and after staining with the Wiesner reagent between the wild type and bm1 heterozygotes, the bm1 heterozygotes may have a different cell wall composition. A chemical analysis of the cell walls was conducted to ascertain whether the cell walls were indeed different among the three genotypes. Figure 3 displays the baseline-corrected, area-normalized and averaged FT-IR spectra of the cell wall samples from the three genotypes in the range between 1800 and 750 cm−1. It is illustrative to compare the overall spectra of the wild type and bm1 mutant first. These two genotypes have been shown to differ in cell wall composition (Provan et al, 1997; Halpin et al, 1998) and therefore provide an indication of the magnitude of changes that can be expected. It is apparent that there is only limited variation between those two spectra, which indicates that the overall cell wall composition and architecture of the bm1 mutant is not drastically different from that of the wild type.

Figure 3
figure 3

Averages of baseline-corrected, area-normalized FT-IR spectra of dissected midribs of wild type, bm1/+ and bm1/bm1 plants.

Full size image

Several differences between the three spectra are apparent. The peak at 1665 cm−1 represents C=O stretching in the carbonyl moiety of conjugated aldehydes or ketones (Mackay et al, 1997; Stewart et al, 1997) and based on an ANOVA the difference in absorbance that is observed for the three genotypes is statistically significant (P = 0.0004). The peaks at 1595 and 1510 cm−1 are indicative of lignin and hydroxycinnamic acids (Stewart et al, 1997). The value of A1510 of the bm1 mutant is considerably lower compared with the wild type and bm1 heterozygote and there is a statistically significant difference between genotypes (P = 0.0007). A high ratio of A1665: A1510 and A1595: A1510 is indicative of the incorporation of coniferaldehyde residues in the cell wall (Mackay et al, 1997; Stewart et al, 1997). Both ratios are higher in the bm1 mutant, confirming the results from Wiesner reactions (Figure 2). The ANOVA comparison indicates that the three genotypes are different (P < 0.0001). The peak at 1250 cm−1 reflects C-O stretching of phenolics (Séné et al, 1994). The difference between the genotypes at this frequency is significant (P = 0.02) and the bm1 heterozygote appears to have an absorbance that is distinct from the two homozygotes. The large peak with a maximum at 1048 cm−1 is due to the presence of polysaccharides (Séné et al, 1994; Stewart et al, 1997). Each of the three genotypes appears to have a distinct absorbance, and the overall difference between the three genotypes is statistically significant (P = 0.003). The spectrum of the bm1 heterozygote is similar to that of the wild-type at some frequencies (for example 1510 cm−1), but more similar to that of the bm1 mutant at other frequencies (for example 1150 cm−1; absorbance different from wild type (P = 0.003)). At several other frequencies (for example 1250 and 1048 cm−1) the spectrum of the bm1 heterozygote appears to show distinct absorbances. Therefore, based on these FT-IR data there is evidence that the bm1 heterozygote displays a chemical phenotype that is distinct from both homozygotes.

There is no evidence for mimicked drought stress

In order to test whether the changes in cell wall composition mimicked drought stress and thus resulted in an increase in the anthesis-silking interval and/or a decrease in plant height, the silking time and plant height at anthesis were recorded. Table 3 lists the ASI and the mean plant height for the wild type, bm1 and bm1/+ plants. An ANOVA with ASI as the outcome did not provide evidence for a difference between genotypes. This indicates that the silking delay is not affected by the bm1 mutation, under the conditions tested. There was limited evidence for significant differences in height among the three genotypes. In Latin square IIA the differences in plant height at flowering between the wild-type plants and the bm1 heterozygotes were significantly different. The wild-type plants were the tallest, followed by the bm1 mutants and then the bm1 heterozygotes. The ranking of tallest to shortest genotype, however, was different between the two replicates. In Latin square IIB the bm1 mutants were the tallest followed closely by the wild types and then the bm1 heterozygotes. This difference was not significant, suggesting that the bm1 mutation does not have a significant effect on plant height under the conditions tested.

Table 3 Mean height at flowering and standard deviation (SD) of the wild-type, bm1/+ and bm/bm1 plants in Latin Squares IIA and IIB and the P-value for the comparison between height at flowering of the bm1/+ and bm1/bm1 plants with the wild-type plants
Full size table

The differences in ASI and plant height are not consistent across replications. This indicates that the change in cell wall composition does not result in a mimicked drought stress under the conditions tested.

Growth and development in the early stages are not affected by the bm1 mutation

The differences in flowering time may be the result of an overall change in growth and development of the plant. It is also possible that the flowering time is controlled directly. Under the latter scenario the different genotypes would grow at the same rate until tassel formation and anthesis. Comparing the time it takes a plant to reach a specific developmental stage can distinguish between these two possibilities. The time to each of the developmental stages was recorded every other day for each of the plants in Latin square IIA and IIB and compared using survival analyses at each stage. The results of this analysis indicate that until stage V7 there is no significant difference in growth and development between the three genotypes. During and after stage V8 the bm1/+ plants seem to grow at a higher rate. The current data, however, are not sufficient to attribute the earlier flowering of the bm1/+ plants solely to this accelerated growth in or after V8 for two reasons. First, the development of the plants is quite rapid at this stage and changes can occur daily. Second, a crucial day was missed due to pesticide treatment during the transition from V8. In order to determine whether growth after V7 is different, it will be necessary to record the growth and development daily.

Discussion

Like many mutations the maize brown midrib mutations have been considered recessive, referring to the fact that the mutant phenotype is only apparent in plants homozygous for the mutant allele. While this is true for the brown color of the vascular tissue in the bm1 mutant, we hypothesized the existence of a dosage effect at the Bm1 locus to explain the fact that in our previous study the segregating population (with wild-type plants and bm1 heterozygotes) could be distinguished from the wild-type population based on its flowering dynamics (Vermerris and McIntyre, 1999). The term dosage effect refers to the situation where the heterozygote has a phenotype that is distinct from both homozygotes. The data from the present study with bm1 heterozygotes provide evidence for a dosage effect. In the case of the Bm1 locus the heterozygote can be distinguished from the homozygotes, based on its flowering dynamics. In addition there is evidence that the composition of the cell wall is also affected in a dosage-related manner. The Bm1 locus now represents an interesting case. The bm1 mutant allele is completely recessive with respect to the brown color of the midrib, partially dominant with respect to cell wall composition and overdominant with respect to flowering time. Loci that show dosage effects are not uncommon in maize. Examples include mutations affecting the endosperm (Tsaftaris et al, 1981; Singletary et al, 1997), the Pl locus, encoding a transcription factor involved in anthocyanin biosynthesis (Hollick and Chandler, 1998), and the Rp8 locus, which affects rust resistance (Delaney et al, 1998). In those cases, however, there is only one, consistent phenotypic effect.

The different analytical techniques available for the analysis of plant cell walls all have their advantages and disadvantages. The use of FT-IR spectroscopy to analyze the cell wall composition of the three genotypes has the advantage that it is relatively fast and that it does not require the isolation of individual cell wall compounds. This means that not only differences in the composition of any of the cell wall compounds can be monitored, but also differences in the chemical interactions between different cell wall compounds. Based on the FT-IR analysis, specifically because of the higher ratios of A1665: A1510 and A1595: A1510, it is apparent that the cell walls of the bm1 homozygote contain higher levels of coniferaldehyde than the wild type and bm1 heterozygote. This agrees with the histochemical data from the Wiesner reactions and with observations on loblolly pine (Pinus taeda L.) (Mackay et al, 1997) and tobacco (Stewart et al, 1997) in which the expression of the gene encoding CAD had been reduced. In both instances this resulted in increased incorporation of coniferaldehyde in the cell wall.

Based on the FT-IR analysis we have evidence that the cell wall of the bm1 heterozygote is distinct from the cell wall of the two homozygotes. The exact chemical basis for these differences is unknown, however, because the identification of specific cell wall compounds is generally difficult with FT-IR spectroscopy. Wet chemical techniques for the determination of lignin and polysaccharide composition are more appropriate in that respect, and further studies are needed to obtain a more complete understanding of the changes in cell wall composition in the bm1 heterozygote.

Variation in flowering time in brown midrib mutants provides evidence for a relationship between cell wall composition and flowering time. At this point, the chemical basis for this relationship is unknown. It is possible that one of the intermediates in cell wall biogenesis directly affects plant development. In that case the FT-IR spectra may provide clues on how the cell wall composition and flowering time are linked. Since the bm1 heterozygotes flower earlier than either of the two homozygotes, correlations between wavenumbers at which the heterozygote has an absorbance distinct from the homozygotes may point towards cell wall compounds that are correlated with flowering time. A comparison of absorbances at these wavenumbers in other bm mutants can reveal if this is a consistent correlation, and analysis of the cell walls with additional chemical techniques may provide further clues towards unraveling the mechanism underlying the link between cell wall composition and flowering time.

We can exclude mimicked drought stress as a basis for early flowering under the conditions tested. As far as developmental effects are concerned, at this point we can only conclude that the basis for the early flowering of the bm1/+ plants is beyond the V7 stage. Up to V7 the three genotypes develop very similarly, but it is possible that during or after the V8 stage the bm1/+ plants display accelerated growth. This needs to be investigated in more detail.

The Bm1 locus affects the expression of the enzyme CAD. Interestingly, the Cad gene was shown to affect growth in loblolly pine, and the average effect of a mutant allele, cad-n1, included enhanced radial growth of the stem (Wu et al, 1999). It was suggested that the Cad locus could represent a QTL affecting stem growth in loblolly pine. The Cad locus may have a general effect on plant growth and development, although given the evolutionary distance between maize and pine, the biochemical mechanism underlying the growth effects may be different.

Finally, the fact that the bm1 heterozygote has a distinct chemical and physiological phenotype may offer an additional tool for the development of new varieties. The improved forage quality of bm1 corn (Cherney et al, 1991), is not associated with increased levels of coniferaldehyde (Grabber et al, 1998). Therefore, the wild-type levels of coniferaldehyde in the bm1 heterozygote do not automatically preclude the bm1 heterozygote from having a potentially higher forage quality, especially since the FT-IR spectrum indicates certain similarities to the bm1 mutant. It may therefore be worthwhile examining the forage quality of the bm1 heterozygote. In addition, other traits such as susceptibility to diseases, pests and lodging may be different in the bm1 heterozygote. If the effect on flowering time is consistent in other genetic backgrounds, the data presented in this study also offer the potential to manipulate flowering time by introducing a copy of the bm1 mutant allele in the line of interest.