KNOX2 Genes Regulate the Haploid-to-Diploid Morphological Transition in Land Plants

See allHide authors and affiliations

Science  01 Mar 2013:
Vol. 339, Issue 6123, pp. 1067-1070
DOI: 10.1126/science.1230082

A Mossy Veil

Land plants have two distinct generations: a haploid gametophyte responsible for producing the gametes, and a diploid sporophyte, that in most land plants is the dominant form observed. However, in mosses, a basal land plant, the primary biomass is composed of the haploid gametophyte. Sakakibara et al. (p. 1067; see the Perspective by Friedman) analyzed loss-of-function mutants of KNOX2 (class 2 KNOTTED1-LIKE HOMEOBOX) genes in the moss Physcomitrella patens. Mutant plants exhibit apospory, where fertilization apparently occurs, but the normal sporophyte phase is bypassed and instead a diploid structure resembling a haploid gametophyte is produced. The results suggest that KNOX2 regulates the alternation of generations by suppressing the haploid body plan in the diploid phase.


Unlike animals, land plants undergo an alternation of generations, producing multicellular bodies in both haploid (1n: gametophyte) and diploid (2n: sporophyte) generations. Plant body plans in each generation are regulated by distinct developmental programs initiated at either meiosis or fertilization, respectively. In mosses, the haploid gametophyte generation is dominant, whereas in vascular plants—including ferns, gymnosperms, and angiosperms—the diploid sporophyte generation is dominant. Deletion of the class 2 KNOTTED1-LIKE HOMEOBOX (KNOX2) transcription factors in the moss Physcomitrella patens results in the development of gametophyte bodies from diploid embryos without meiosis. Thus, KNOX2 acts to prevent the haploid-specific body plan from developing in the diploid plant body, indicating a critical role for the evolution of KNOX2 in establishing an alternation of generations in land plants.

Plants have a life cycle characterized by alternation between two generations, haploid (gametophyte) and diploid (sporophyte), where each phase develops a multicellular body (1, 2). The gametophyte produces gametes—sperm (or pollen) and egg cells—and may be the dominant photosynthetic generation, as in liverworts, mosses, and hornworts. The sporophyte produces haploid spores via meiosis and is the dominant photosynthetic generation in the vascular plants. The alternation of generations in land plants results in the possibility that tissue differentiation in each generation is governed by different genetic programs, initiated by either fertilization (haploid to diploid) or meiosis (diploid to haploid). Land plants probably evolved from a freshwater algal ancestor, with a life cycle similar to extant charophycean algae, the algal taxa most closely related to land plants (35). These algae have multicellular haploid bodies that produce gametes, and after fertilization, the diploid zygote directly undergoes meiosis to produce haploid spores. The multicellular diploid sporophyte generation of land plants is thought to have evolved through a delay in zygotic meiosis, with extensive mitotic cell divisions between the formation of the zygote and the production of spores via meiosis (1). Thus, the diploid sporophyte generation of land plants evolved from an ancestral single-celled diploid generation.

The genetic mechanisms regulating the multicellular diploid generation of land plants are presently unknown, but their origin may lie in the genetic machinery regulating the sexual cycle of land plant ancestors. Chlorophyta are a green algal clade with a sister relationship to charophycean algae plus land plants (4). In Chlamydomonas, a unicellular chlorophyte, the haploid-to-diploid transition is regulated by a pair of related TALE (three–amino acid length extension) class homeodomain protein–encoding genes (68). The two gamete types, "plus" and "minus," express two homeodomain genes, GSP1 (encoding a BELL-related TALE homeodomain protein) and GSM1 [encoding a KNOTTED1-LIKE HOMEOBOX (KNOX)–related TALE homeodomain protein], respectively (7). On their own, each of the two encoded proteins is cytoplasmic, but upon gamete fusion, the proteins physically interact and translocate into the nucleus to regulate zygotic gene expression (7). Loss-of-function GSP1 alleles result in a failure to activate the diploid genetic program with zygotes reentering the haploid program, whereas gain-of-function alleles can ectopically activate the diploid program during the haploid phase of the life cycle (6, 8). Lee et al. have proposed that KNOX-TALE genes in land plants are candidates for regulating aspects of the alternation of generations (7).

In contrast to Chlorophyta, land plant genomes encode two subfamilies of KNOX genes, class 1 (KNOX1) and class 2 (KNOX2), resulting from a gene duplication in the lineage leading to land plants (9). KNOX1 genes are regulators of sporophytic (diploid) meristematic genes, but the functions of KNOX2 genes are unknown (10). We used translational fusions with the β-glucuronidase (GUS) reporter gene to investigate the spatio-temporal expression patterns of the two Physcomitrella KNOX2 genes, MKN1 (AF285148) and MKN6 (XM_001765523) (11, 12), throughout both gametophyte and sporophyte generations (fig. S1) (13, 14). In the gametophyte generation, GUS activity was not detected for either MKN1 or MKN6 in the protonemata (the early filamentous growth phase) or the gametophores (leafy shoots from which reproductive structures develop) (Fig. 1, A, B, K, and L, and fig. S2), similar to Physcomitrella KNOX1 genes (11). In all of the haploid tissue we examined, we detected GUS activity in only the egg mother cell, the surrounding archegonial tissues, and mature eggs before fertilization (Fig. 1, C, D, M, N, and V).

Fig. 1

Expression patterns of Physcomitrella KNOX2 genes. Blue staining documents GUS activity derived from translational fusions into endogenous KNOX loci. (A to J) MKN1-GUS#1, (K to U) MKN6-GUS#5, and (V to AA) MKN2-GUS#1 (KNOX1) lines. Protonemata and gametophores of the gametophyte (A and B, K and L), archegonial tissues with egg mother cells (C and M, arrows), and unfertilized egg cells (D and N, arrows) of gametophyte generation. MKN1-GUS expression in embryos (E), sporogenous precursor cells (F), sporogenous tissues (G and H), seta (I), and stomatal cells (J, arrowheads) in sporophytes. MKN6-GUS expression in the zygote (O, arrow), young embryos (P to R), older developing sporophytes (S and T), and a mature sporangium (U). MKN2-GUS expression in the egg mother cell (V, arrow), the zygote (W, arrow), and apical regions of young (X to Z) and older developing sporophytes (AA). Sporophytes were removed from archegonia and gametophores in (E) to (I), (P) to (U), and (X) to (AA). Scale bars, 100 μm (A, G to I, K, T, U, and AA); 1 mm (B and L); 50 μm (C, D, M, N, and Z); and 20 μm (E, F, J, O to S, and V to Y). Ploidy level is indicated in the upper right corner of each panel.

In the sporophyte generation, we first detected a signal in the MKN6-GUS line in the zygote (Fig. 1O) and later throughout young embryos (Fig. 1P), with the signal diminishing in the apical region during apical cell divisions (Fig. 1Q). Subsequently, we identified GUS activity throughout the embryo (Fig. 1R); at later stages of sporophyte development, we detected a signal in a complex pattern in several tissues, including the seta and foot (Fig. 1, S to U). In contrast, the MKN1-GUS signal was more restricted during sporophyte development (Fig. 1E), with signal in the endothecium, a tissue that gives rise to sporogenous precursor cells (Fig. 1, F and G), and later in sporogenous tissues (Fig. 1H) and a few tissues of the mature sporophyte (Fig. 1, I and J). These sporophytic patterns of Physcomitrella KNOX2 expression are initially overlapping with those of KNOX1 in the zygote (Fig. 1W), then complementary in the young embryo (Fig. 1X), and again overlapping in sporogenous tissues of developing sporophyte (Fig. 1, Y to AA) (11), such that all sporophytic meristematic cells expressed at least one KNOX gene.

Because the two Physcomitrella KNOX2 genes are closely related paralogs (9, 11) and their expression patterns partially overlap, we constructed single- and double-deletion mutants (fig. S3). Consistent with their expression patterns, the deletion of both KNOX2 genes did not affect the phenotype of the gametophytic (1n) protonema and gametophore morphologies (Fig. 2, A to D). However, no mature sporangia developed in the mkn1mkn6 deletion line (Fig. 2, E and F, and table S1), with development of sporophytes suspended at the 4-week embryo stage (Fig. 2, G and H, and table S1). The mkn1mkn6 embryos often had protruded filamentous structures, similar in morphology to protonemata or rhizoids, cell types normally produced only in the haploid generation (Fig. 2H). To further characterize the defects in sporophyte development, we cultured mkn1mkn6 and wild-type (WT) embryos on plate media without plant hormones. Notably, the filamentous structures that emerged from mkn1mkn6 sporophytes continued to grow, forming protonemal tissues resembling that of a WT gametophyte within 1 week (Fig. 2J and Table 1). Subsequently, gametophore buds developed from the mutant protonemal filaments approximately 2 weeks after incubation (Fig. 2K), again, similar to a WT gametophyte (Fig. 2C). Gametophores produced via mkn1mkn6 embryos formed gametangia at a similar frequency as the wild type, with the antheridia and archegonia producing functional sperms and eggs, respectively (table S2). Self-fertilization resulted in embryos arrested at the same stage as the original mkn1mkn6 lines. However, the growth of gametophyte-like bodies derived from mkn1mkn6 embryos is not indistinguishable from WT embryos, with a reduced frequency of gametophore initiation and possible differences in sizes of protonemal filament cells noted (fig. S4 and tables S3 and S4). Thus, some developmental epigenetic modifications may not be properly reset in protonemata that differentiated directly from embryonic tissues. Consistent with their expression patterns, the phenotype of single-gene knockouts of mkn6 was similar to the double mutant, whereas mkn1 single mutants showed no dramatic phenotype, indicating that MKN6 expression (Fig. 1, E and P to R) is critical to maintain sporophyte identity (Table 1 and table S1).

Fig. 2

Embryogenesis in KNOX2 deletion lines is suspended with haploid-like tissues produced in the diploid generation. (A to D) Gametophytes of the wild type (A and C) and double-deletion line mkn1mkn6#12 (B and D). Protonemata (A and B) and gametophyte tissues 2 weeks after culturing of protonemata (C and D) of WT and mkn1mkn6 lines. (E to F) Haploid gametophores 8 weeks after gametangia induction of the wild type (E) and the mkn1mkn6 deletion line (F). Arrows in (E) indicate mature diploid sporangia. The inset in (E) depicts a maturing WT sporophyte 6 weeks after gametangia induction. (G) Wild-type sporophyte 4 weeks after gametangia induction. (H) Sporophyte in the mkn1mkn6 deletion line 6 weeks after gametangia induction. Protruding filamentous tissues were observed in 50% (19 of 38) of embryos (arrow). (I to K) Embryo culture after removal from archegonia in the wild type (I) and mkn1mkn6 deletion lines (J and K). (I) Wild-type embryos on the first day (left) and after 1 week (right) in culture. (J) mkn1mkn6 embryos on the first day (left) and after 1 week (right) in culture. (K) Protonemata with gametophores derived from mkn1mkn6 embryos after 16 days in culture (arrow). Scale bars, 100 μm (A, B, and G to J); 1 mm (C, D, and K); 5 mm (E and F); and 500 μm [inset in (E)]. Ploidy level is indicated in the upper right corner of each panel.

Table 1

Protonemal tissue formation in mkn6 and mkn1mkn6 embryos after 1 week in culture. n represents the number of embryos examined. Intact embryos were isolated from archegonia and incubated on BCD plates (14) including 1 mM CaCl2 and 0.8% agar for 1 week.

View this table:

The irregular transition from sporophyte to gametophyte without meiosis is called apospory. In mosses, apospory can be induced, with protonemata and leafy gametophores (gametophytic tissues) developing from setae (sporophytic tissues) that have been severed (15). The opposite irregular transition from gametophyte to sporophyte without fertilization is called apogamy. Apogamy and apospory have been observed to occasionally take place under both natural environmental and experimental conditions in nonseed plants (16, 17). As the number of chromosomes in the sporophyte (2n) is double that of the gametophyte (1n), an aposporously produced gametophyte should have 2n chromosomes (18).

To determine the ploidy of protonemata developing from mkn1mkn6 embryos, we measured the DNA content of cell nuclei. Chloronemal cells, including apical cells, of Physcomitrella spend most of the cell cycle in the G2/M phase (19); therefore, the amount of nuclear DNA in the haploid (1n) controls—WT and mkn1mkn6 chloronemal apical cells—is at a 2C level. Importantly, when we analyzed apical cells of protonemata growing from mkn1mkn6 sporophytes, the average nuclear intensity was close to double (4C) that of the controls (Fig. 3 and fig. S5). These results indicate that deletion of Physcomitrella KNOX2 genes results in apospory, a haploid phenotype developing from a diploid body. This phenotype, combined with the expression pattern of these genes throughout sporophyte development (Fig. 1), indicates that KNOX2 transcription factors function to repress the haploid generation developmental genetic program during the diploid generation of sporophyte development.

Fig. 3

Ploidy of protonemal apical cells in WT and mkn1mkn6 deletion lines. Comparison of nuclear DNA content (as calculated from fluorescence of nuclei stained with 4′,6-diamidino-2-phenylindole) of interphase apical cells of chloronema in the WT gametophyte (white; n = 50 cells), the mkn1mkn6 deletion line gametophyte (black; n = 50), and interphase apical cells in the protonema-like tissues derived from the embryo of the mkn1mkn6 line (stripes; n = 50). Average values and standard deviations are shown. The asterisk indicates significant differences relative to WT chloronemata (P < 0.001, Mann-Whitney U test).

In contrast to apospory observed in KNOX2 mutants, inactivation of the Physcomitrella polycomb repressive complex 2 (PRC2) results in apogamy, the development of a sporophyte body plan in a haploid gametophyte. Mutations in either Physcomitrella patens CURLY LEAF (PpCLF) or Physcomitrella patens FERTILIZATION-INDEPENDENT ENDOSPERM (PpFIE), both of which are predicted to encode components of PRC2 based on sequence similarities, result in fertilization-independent, sporophyte-like bodies from branches of gametophytic protonemal filaments and the induction of sporophyte-specific gene expression (e.g., KNOX1) in gametophyte buds (20, 21). Thus, polycomb-mediated repressive chromatin modification throughout the genome is required to maintain gametophyte identity through repression of the sporophyte genetic program. In contrast to PRC2, a global regulator of facultative chromatin that acts to repress target genes, KNOX2 is a specific transcription factor family that regulates the alternation of generations in land plants.

In land plants, KNOX1 and KNOX2 proteins heterodimerize with proteins encoded by BELL-class TALE genes to affect gene expression (2224). Furthermore, KNOX expression in the Physcomitrella egg cell may function as KNOX expression in Chlamydomonas gametes. In the flowering plant Arabidopsis thaliana, the four KNOX2 genes (KNAT3, KNAT4, KNAT5, and KNAT7) are broadly expressed in the sporophyte body (2n) (25, 26). Intriguingly, loss-of-function alleles of knat3 partially suppress phenotypes induced by ectopic expression of a BELL gene, BLH1, in the embryo sac (1n) (27), indicating that KNAT3 is expressed in WT embryo sacs. Conversely, three of the four Physcomitrella BELL genes were primarily expressed in the antheridia, archegonial tissues, and/or the sporophyte (fig. S6). We speculate that gamete-specific expression of KNOX and BELL genes to initiate zygotic gene expression may be conserved more broadly throughout plants.

The KNOX1 and KNOX2 subfamilies arose after the divergence of the Chlorophyta and land plant lineages. The evolution of a multicellular diploid generation from an ancestral single-celled zygotic diploid generation required the maintenance of continued mitotic divisions after fertilization and repression of the haploid developmental program during development of the diploid generation. We hypothesize that in the common ancestor of land plants, KNOX1 genes acquired functions in the maintenance of sporophytic meristematic cells, as evidenced by loss-of-function phenotypes in both flowering plants (e.g., Arabidopsis) and a moss (Physcomitrella), and KNOX2 genes evolved to maintain diploid differentiation by suppression of the gametophytic development program. Thus, KNOX1/KNOX2 duplication facilitated the evolution of more complex gene regulatory networks to perform two critical roles in the diploid phase of the alternation of generations and was instrumental in the establishment of a multicellular diploid generation in land plants.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Tables S1 to S7

References (2835)

References and Notes

  1. Supplementary materials and methods are available on Science Online.
  2. Acknowledgments: We thank Y. Kabeya for technical assistance of flow cytometry; C. L. Volk Murayama for translation of references; D. R. Smyth, Y. Eshed, M. Shimamura, and T. Nishiyama for critical comments on the manuscript; and the Model Plant Research Facility of the National Institute for Basic Biology for experimental facilities. This research was partly supported by grants from the Ministry of Education, Culture, Sports, Science and Technology (MEXT); the Japan Society for the Promotion of Science (JSPS) KAKENHI grant (to K.S., Y.T., H.D, and M.H.); and the Australian Research Council grant FF0561326 (J.L.B.). K.S. was a JSPS research fellow. Sequence data associated with this manuscript have been submitted to the DNA Data Bank of Japan.
View Abstract


Navigate This Article