The distinctive roles of five different ARC genes in the chloroplast division process in Arabidopsis


*For correspondence (fax +1904 432860; e-mail


ARC (accumulation and replication of chloroplasts) genes control different aspects of the chloroplast division process in higher plants. In order to establish the hierarchy of the ARC genes in the chloroplast division process and to provide evidence for their specific roles, double mutants were constructed between arc11, arc6, arc5, arc3 and arc1 in all combinations and phenotypically analysed. arc11 is a new nuclear recessive mutant with 29 chloroplasts compared with 120 in wild type. All the phenotypes of the double mutants are unambiguous. ARC1 down-regulates proplastid division but is on a separate pathway from ARC3, ARC5, ARC6 and ARC11. ARC6 initiates both proplastid and chloroplast division. ARC3 controls the rate of chloroplast expansion and ARC11 the central positioning of the final division plane in chloroplast division. ARC5 facilitates separation of the two daughter chloroplasts. ARC5 maps to chromosome 3 and ARC11 and ARC6 map approximately 60 cM apart on chromosome 5.


The division of young chloroplasts in expanding mesophyll cells ensures the presence of the full complement of chloroplasts in mature leaf cells. In the Arabidopsis leaf, cells mature from the base to the tip of the leaf (Pyke et al. 1991). Earlier proplastid division in mitotic cells keeps pace with cell division but it is the subsequent post-mitotic divisions of young developing chloroplasts in young mesophyll cells which assure the size of the population of mature chloroplasts and photosynthetic competence. The sequence of ultrastructural changes characteristic of chloroplast division in higher plants is well documented (Boffey 1992; Leech & Pyke 1988; Leech et al. 1981; Orros & Possingham 1989; Possingham et al. 1988; Whatley 1988). Chloroplast division always occurs when the young chloroplasts have grown to about 50% of their final volume and contain numerous thylakoid membranes and small granal stacks (Ellis & Leech 1985). The chloroplasts become increasingly centrally constricted, twist around the isthmus and the two equally sized daughter chloroplasts finally separate. Following three rounds of chloroplast division in Arabidopsis thaliana ecotype Landsberg erecta (Ler) the mature mesophyll cells contain a mean of 120 chloroplasts (Pyke & Leech 1992). The sequence of morphological changes leading to the development of the mature chloroplast complement based on our observations of developing mesophyll cells of Arabidopsis ecotypes is illustrated in Fig. 1.

Figure 1.

Diagram illustrating normal chloroplast development and chloroplast division in Arabidopsis thaliana (Ler) leaf development.

The proplastids in meristematic mesophyll cells undergo a series of divisions ensuring that the young post-mitotic cells contain at least 15 proplastids. During cell expansion the young chloroplasts also expand and begin to form a central constriction which becomes increasingly narrow until separation of two equally sized daughter chloroplasts occurs. Chloroplasts undergo three rounds of division so that the mature mesophyll cell has a complement of 120 chloroplasts which are all of similar size and shape.

To identify the eukaryotic genes controlling the different phases of chloroplast division, we have used the strategy of isolating and analysing mutants of chloroplast division in Arabidopsis. In these arc (accumulation and replication of chloroplasts) mutants the mesophyll chloroplasts differ considerably from wild type in number, size and shape (for a review see Pyke 1997). All the arc mutant phenotypes are stable and result from single nuclear recessive mutations, i.e. they follow normal Mendelian inheritance patterns in reciprocal backcrosses.

Eleven independent nuclear ARC genes involved in the control of chloroplast division have been identified so far. In view of their distinctive mutant phenotypes it seems likely that these ARC genes have several very different roles in the chloroplast division process. In order to investigate the genetic control of chloroplast division we used our arc mutants to examine the interactions and epistatic relationships of the ARC genes. Our approach was to construct and identify double arc mutants and analyse their chloroplast phenotypes. Five arc mutants with distinctive heritable phenotypes were chosen. arc6 has the most dramatic alteration in chloroplast number with a mean of only two very large chloroplasts (Pyke et al. 1994) forming large sheets covering the internal cell surface (Robertson et al. 1995). arc1 is unique because more chloroplasts are present per cell plan area (32/1000 μm2) than in wild type (25/1000 μm2) (Pyke & Leech 1992, 1994). In arc3 a few abnormally large chloroplasts are present (Pyke & Leech 1992; Pyke & Leech 1994) and in the arc5 mutant chloroplast fission is arrested at the final stage and the 13 chloroplasts are large and dumb-bell shaped (Pyke & Leech 1994; Robertson et al. 1996). arc11 is a recently isolated mutant, described here for the first time. In arc11 there are two chloroplast populations, one equivalent in size range to wild-type chloroplasts and a second population of larger chloroplasts.

In this paper we analyse the 10 double mutant phenotypes constructed from arc1, arc3-1, arc5, arc6-1 and arc11 in all combinations and compare them with the parental single mutant phenotypes. The interactions of the different ARC genes are described and a model for the independence and order of ARC gene action proposed.


The arc11 mutant

The arc11 phenotype is of great interest since all arc11 mesophyll cells have a heterogeneous population of chloroplasts as seen in Fig. 2(b) compared with the homogeneous wild-type population (Fig. 2a). Forty to 50% of the chloroplasts in arc11 cells are within the wild-type size range whilst 50–60% are larger than wild type; some chloroplasts are 10 times larger than wild type (Fig. 2c). The smallest mesophyll cells already show this heterogeneity of size among the population of young chloroplasts. The ultrastructural analysis of arc11 chloroplasts confirms that they vary greatly in size and are frequently larger than wild type (Fig. 2e,f). The arrangement of the appressed and non-appressed chloroplast thylakoid membranes and the density of the stroma in arc11 resemble very closely those of normal wild-type chloroplasts (Fig. 2e,f). The growth, vigour and fertility of the arc11 plant is normal under optimal growth conditions and the whole plant phenotype is very similar to wild type, except that arc11 begins flowering approximately 3 days earlier.

Figure 2.

Chloroplasts in the mesophyll cells of wild-type Arabidopsis thaliana (Ler) and in mesophyll cells of the arc11 mutant.

(a,b) Photomicrographs of isolated leaf mesophyll cells from wild-type (Ler) and arc11, respectively. The insert in (b) illustrates arc11 chloroplasts in division.

(c) Frequency distribution of chloroplast size, measured as plan area, for chloroplasts from fully expanded leaves of wild-type (Ler) and arc11. Percentages of chloroplasts of different sizes as measured from 143 wild-type chloroplasts (–––) and from 100 arc11 chloroplasts (–-). The chloroplast plan area bin size (i.e. the interval of plan area size used in the histogram) is 25 μm2.

(d) The relationship between chloroplast number per mesophyll cell and mesophyll cell plan area for wild-type (Ler) (–––) and arc11 (–-). Each data point represents the measurements from one cell in which the cell area was measured and the chloroplast number counted. Regression values (r2) are 0.935 (wild type) and 0.663 (arc11).

(e,f) Electron micrographs of wild-type (Ler) and arc11 mesophyll cell chloroplasts, respectively. Note the extremes in chloroplast size in arc11 mesophyll cells compared to the uniform size of chloroplasts in wild-type cells. arc11 mesophyll chloroplast ultrastructure is very similar to wild-type with normal thylakoids and grana. ap, appressed membrane; nap, non-appressed membrane.

Chloroplast division is limited in arc11 mesophyll cells, and apparently occurs only once since fully expanded arc11 mesophyll cells have a mean of 29 chloroplasts (Fig. 2d). In young (1000 μm2) post-mitotic cells of arc11 the dividing chloroplasts are asymmetric giving the appearance of ‘budding’ (Fig. 2b insert). In mature arc11 cells it appears that separation after chloroplast division yields one large and one small daughter plastid.

arc11 is a mutation of a novel, independent ARC locus. The unique nature of the ARC11 locus was revealed by F1 complementation analysis between arc11 and the other 10 arc mutants.

The arc11 mutant phenotype is stably inherited and in reciprocal backcrosses segregates as a monogenic nuclear recessive trait in a normal Mendelian manner. ARC11 is 30 cm south of a transposed Ac on chromosome V (see Experimental procedures).

Construction of double mutants

We constructed mutants homozygous recessive at two arc loci to study the extent of ARC gene interaction and to determine the hierarchy of ARC gene control in the chloroplast division process.

Identification of double mutants in crosses giving no novel phenotypes in the F2 generation.

In crosses between arc3 with arc6, arc5 with arc6, arc11 with arc6, arc11 with arc3, arc11 with arc5 and arc3 with arc5, the F2 plants segregated for wild type and parental types only but no novel F2 phenotypes were found. To identify the double mutant plants in these populations, F2 seedlings with each parental phenotype were allowed to self fertilise. The progeny from one of the parental F2s exhibited a 3:1 segregation pattern. The F3 segregation ratios were as follows:

arc3×arc6, 37 arc 3 : 11 arc62 0.11, P > 0.7)

arc5×arc6, 59 arc 5 : 18 arc62 0.11, P > 0.7)

arc11×arc6, 38 arc11 : 10 arc62 0.44, P > 0.5)

arc11×arc3, 35 arc 3 : 13 arc112 0.11, P > 0.7)

arc11×arc5, 42 arc11 : 13 arc52 0.06, P > 0.8)

arc3×arc5, 44 arc 5 : 11 arc32 0.73, P > 0.3).

Double mutants were confirmed by their lack of segregation in the subsequent F4 generation, and analysis of F1 progeny from crosses to parental lines confirmed that the double mutants were homozygous recessive at both mutant loci. F4 seedlings were used in the chloroplast and cell analysis of the double mutants.

Identification of double mutants in crosses giving novel phenotypes in the F2 generation. Among the F2 progeny of the crosses arc1 with arc6 and arc1 with arc11 four phenotypes were identified; wild type, maternal type, paternal type and a novel phenotype which had characteristics of both parents. F2 segregation data were consistent with a 9:3:3:1 ratio as follows:

arc1 × arc6, 43 wild type: 15 arc1 : 18 arc6 : 6 arc1 arc6

2 0.89, P > 0.8)

arc1 × arc11, 46 wild type : 13 arc1 : 17 arc11 : 4 arc1 arc11

2 0.756, P > 0.8).

Selected putative double mutant plants were pollinated by relevant homozygous parental lines to confirm their genotypes. Lack of segregation of phenotypes in the F3 generation further confirmed that the putative double mutants were indeed homozygous recessive at both mutant loci. F3 seedlings were used in the chloroplast and cell analysis of the double mutants.

Analysis of double mutants

The photomicrographs in Fig. 3(g–p) are representative of more than 1000 mesophyll cells from each double mutant. The characteristic shapes and relative sizes of the chloroplasts in the cells of each genotype are illustrated diagramatically in Fig. 4.

Figure 3.

Isolated leaf mesophyll cells from fully expanded leaves of wild-type Arabidopsis thaliana (Ler) and arc mutants viewed with Nomarski differential interference contrast optics.

(a) Wild-type (Ler); (b) arc6; (c) arc1; (d) arc11; (e) arc3; (f) arc5; (g) arc11 arc6; (h) arc3 arc6; (i) arc5 arc6; (j) arc1 arc6; (k) arc1 arc11; (l) arc1 arc3; (m) arc1 arc5; (n) arc11 arc3; (o) arc11 arc5; and (p) arc3 arc5. The cells illustrated are representative of over 1000 cells viewed for each double mutant genotype. The wild-type, arc1, arc3, arc1 arc3 and arc1 arc5 images are taken from Pyke & Leech (1994). Bar = 25 μm.

Figure 4.

Diagrammatic representations of chloroplast shape and size for wild-type Arabidopsis thaliana (Ler), for five single arc mutants and for 10 double arc mutants.

Each square contains a line drawing representing the characteristics of the chloroplasts of a single genotype. The x and y axes give the genotypes of the parents. Single arc mutants, i.e. parental chloroplast phenotypes, are represented in squares bound by two lines. Double mutant phenotypes are represented in squares bound by a single line. The number in each square is the mean number of chloroplasts per mesophyll cell for that mutant. Bar = 25 μm.

arc11 arc6, arc3 arc6 and arc5 arc6 double mutants.

Each of these double mutants had only one or two chloroplasts per mesophyll cell (Table 1, Fig. 3g–i) identical in size, distribution and appearance to arc6 chloroplasts (Fig. 3b). The double mutants segregated from the maternal phenotypes in F3 progeny. The relationship between the number of the chloroplasts per mesophyll cell and the size of the mesophyll cell across a range of cell sizes in all the double mutants is indistinguishable from that in the single arc6 mutant, as shown in Fig. 5(c–e).

Table 1.  Chloroplast number, chloroplast plan area and mesophyll cell plan area for populations of mesophyll cells from fully expanded first leaves of wild-type and arc mutants of Arabidopsis thaliana cv. Landsberg erectaa
 Chloroplasts Chloroplast plan areadMesophyll cell plan areaeChloroplasts
Genotypeb(number per mesophyll cell)c (μm2)(μm2)(number per 1000 μm2 mesophyll cell plan area)
  • a

    Data for arc1, arc3, arc5, arc1 arc3 and arc1 arc5 are taken from Pyke & Leech (1994).

  • b

    The maternal parent of the double mutant is given first.

  • c

    Determined from a regression of chloroplast number per cell on mesophyll cell plan area using the value for mean mesophyll cell plan area (see e).

  • d

    Mean of at least 100 chloroplasts from at least 30 different mesophyll cells.

  • e

    Mean of at least 150 cells per genotype.

  • The mean plan area of cells in which chloroplasts were measured was not significantly different from the mean plan area of the 150 cells.

  • SEM values are shown in parenthesesd,e.

L.er12050 (0.5)4800 (135)25
arc621000 (34)3700 (62)0.5
arc11 arc611090 (37)3100 (68)0.3
arc3 arc621140 (28)3500 (75)0.6
arc5 arc621070 (35)3700 (75)0.5
arc19430 (0.5)2900 (68)32
arc1 arc69530 (23)2900 (53)3
arc1 arc117950 (2.5)2900 (90)27
arc1 arc326120 (5)4200 (192)6
arc1 arc54980 (5)4800 (145)10
arc1129110 (9)4200 (112)7
arc11 arc327190 (14)3700 (91)7
arc11 arc512160 (9)3500 (78)3
arc316290 (13)3900 (80)4
arc513310 (11)3500 (83)4
arc3 arc511260 (12)3400 (58)3
Figure 5.

The relationship between chloroplast number per mesophyll cell and mesophyll cell plan area for wild-type Arabidopsis thaliana (Ler), five single arc mutants and 10 mutants homozygous recessive at two arc loci.

(a) Wild-type (Ler); (b) arc1 (▿), arc3 (○), arc5 (□), arc6 (♦) and arc11 (▵) single mutants all in the Ler background. Before the construction of double mutants arc6 was crossed into the Ler background, (c) arc11 arc6; (d) arc3 arc6; (e) arc5 arc6; (f) arc1 arc6; (g) arc1 arc11; (h) arc1 arc3; (i) arc1 arc5; (j) arc11 arc3; (k) arc11 arc5; and (l) arc3 arc5. Each graph in (c–l) shows the data points for the double mutant (○) with the maternal (····) and paternal (–-) regression lines. Each data point represents the measurement from one cell in which the cell area was measured and the chloroplast number counted. At least 50 cells were analysed for each double mutant except for arc1 arc3 and arc1 arc5. Measurements for the arc1, arc3, arc5 single mutants and arc1 arc3 and arc1 arc5 double mutants were taken from Pyke & Leech (1994). The values on the y axes differ to accommodate the chloroplast numbers per cell in the single and double mutants.

The epistatic interaction between the mutant genes arc11, arc3 and arc5 with arc6 can be interpreted as indicating that ARC6 gene action is upstream of ARC11 and ARC3 and ARC5 gene action.

The arc1 arc6, arc1 arc11, arc1 arc3, and arc1 arc5 double mutants.

arc1 arc6 double mutants could be identified in F2 progeny by their novel chloroplast phenotype (Fig. 3j). The mean chloroplast number (9) per cell in arc1 arc6 is greater than the number in arc6 (2) but less than in arc1 (94) (Table 1, Fig. 5f). Chloroplast size in arc1 arc6 is also intermediate between the two parental lines: arc1 arc6 chloroplasts are half the plan area of the arc6 chloroplasts but more than 15 times larger than the arc1 chloroplasts (Table 1, Fig. 3c). Double mutant seedlings could be recognised by their pale and slightly twisted leaves indicative of the homozygous recessive arc1 and arc6 mutations, respectively.

arc1 arc11 mutants (Fig. 3k) were identified in the F2 generation by their pale leaf phenotype indicative of a homozygous arc1 mutation, but also by the persistence of the variable chloroplast size per cell indicative of the homozygous arc11 mutation. The double mutant cell has a chloroplast complement of 79, i.e. greater than arc11 (29) but less than arc1 (94) (Table 1). Mean chloroplast size in arc1 arc11 is intermediate between the two parents as is the relationship between number of chloroplasts per mesophyll cell and mesophyll cell size (Fig. 5g).

The arc1 arc3 and arc1 arc5 double mutants have novel phenotypes (Fig. 3l,m) in which the characteristics of both parents are seen: the chloroplast number per mesophyll cell is greater than in either arc3 or arc5 single mutants (Fig. 5h,i) but less than in the arc1 parent. Chloroplast size in these double mutants is also intermediate between the two parental mutants (Table 1).

These results suggest that the ARC1 gene acts independently of the ARC6, ARC11, ARC3 and ARC5 genes during the chloroplast division process.

The arc11 arc3 double mutant.

arc11 arc3 double mutants (Fig. 3n) segregated with the arc11 chloroplast number amongst the arc3 population in F3. arc11 arc3 chloroplasts resemble arc3 in the proportion of large chloroplasts with plan areas between 400 and 600 μm2 but resemble arc11 chloroplasts in number (27) and size range (25–675 μm2) (Table 1). The relationship between the number of chloroplasts and the size of the mesophyll cells in arc11 arc3 is also identical to arc11 (Fig. 5j).

The arc11 arc3 chloroplast phenotype is therefore similar but not identical to arc11, suggesting that ARC11 is partially epistatic to ARC3.

The arc11 arc5 double mutant.

The arc11 arc5 double mutants (Fig. 3o) segregated with the arc5 chloroplast number amongst the arc11 population in F3. In addition, only a proportion of the chloroplasts in arc11 arc5 had the characteristic ‘arc5 ’ dumb-bell shape indicative of arrest in the final stage of chloroplast division. arc11 arc5 has 12 chloroplasts (Table 1) compared with 13 in arc5 indicating that no chloroplast divisions have been completed. The relationship between the number of chloroplasts and the size of the mesophyll cell in arc11 arc5 is also identical to arc5 (Fig. 5k). The range of chloroplast size in the arc11 arc5 (20–450 μm2) resembles neither the arc5 (125–700 μm2) nor the arc11 (25–675 μm2) distribution.

The appearance of the arc11 arc5 double mutant is consistent with ARC5 acting downstream of ARC11 during chloroplast division.

The arc3 arc5 double mutant.

In fully expanded leaves, chloroplast numbers in arc3 and arc5 are very similar impeding the identification of double mutants (Table 1 and Fig. 5l). However, in 16-day-old seedlings the appearance of arc3 and arc5 chloroplasts is very different and can be used to identify double mutants. At this stage arc3 chloroplasts are larger than wild type and have an amorphous shape (Fig. 3e), but most arc5 chloroplasts are arrested in chloroplast division and a central constriction can be clearly seen (Fig. 3f). arc3 arc5 mutants (Fig. 3p) segregated with the arc3 phenotype amongst the arc5 population in F3 after 16 days of growth. The double mutant had no dumb-bell shaped chloroplasts characteristic of arc5.

The chloroplast phenotype of arc3 arc5 suggests that ARC3 and ARC5 function in the same pathway and that ARC3 acts upstream of ARC5.

Mapping of the ARC loci

We mapped ARC5, ARC6 and ARC11 using SSLP analysis on a small population of F2 mutant plants. ARC5 maps to the top arm of chromosome 3 between nga162 (Bell & Ecker 1994) and AtDMC1 (Klimyuk & Jones 1997). ARC6 and ARC11 both map to chromosome 5 but at distant locations. ARC6 maps between ARMS marker m247 (Fabri & Schäffner 1994) and CAPS marker DFR (Konieczny & Ausubel 1993) on chromosome 5. ARC11 maps close to the SSLP microsatellite marker nga139 on chromosome 5. The Ac element maps close to nga151 approximately 30 cm north of the ARC11 locus leaving open the possibility that Ac is effecting the arc11 mutation. These map positions are shown in Fig. 6.

Figure 6.

Map positions of ARC5, ARC11 and ARC6.

Chromosome 3 shows the position of ARC5 and chromosome 5 shows the positions of ARC6, ARC11 and the Ac element. (Lister and Dean recombinant inbred map, updated November 1998). It is interesting to note that AtFtsZ1;1* (identical to part of clone MC015) maps to chromosome 5 distal to PDC2 and AtFtsZ2;1 (identical to part of the BAC clone F2H17) maps to chromosome 2 (http://genome-


We were able to recover healthy double mutants from all the crosses and the double mutant combinations were not deleterious to plant growth. The Mendelian ratios obtained in the F2 generation (arc1 × arc6, arc1 × arc11) and in the F3 generation (arc3 × arc6, arc5 × arc6, arc11 × arc6, arc11 × arc3, arc11 × arc5 and arc3 × arc5) confirm that the arc mutants are the result of mutations in independent nuclear recessive ARC genes. Based on our analysis of the double mutant chloroplast phenotypes and comparison with the single mutants we are able to determine the hierarchy of the five ARC genes in the control of the higher plant chloroplast division process. The ARC genes will be considered in the order in which they apparently operate in the proplastid and chloroplast division pathways as illustrated in Fig. 7.

Figure 7.

A diagram illustrating the suggested hierarchy of ARC gene action in higher plant chloroplast division.

This model is based on Fig. 1 The stages in the chloroplast division process affected by ARC6, ARC1, ARC11, ARC3 and ARC5 gene action were deduced from our analysis of the respective single mutants and double mutants. The effects of mutation at each ARC locus are indicated by broken lines. ARC1 is involved in the down-regulation of proplastid division but is in a separate pathway. arc1 leads to increased proplastid division. ARC6 is involved in the initiation of both proplastid and chloroplast division. arc6 completely suppresses proplastid and chloroplast division but allows extended expansion until the chloroplasts are 20 times larger than wild type. ARC11 is involved in the central positioning of the division constriction. In arc11 the constriction is assymetric. ARC3 controls chloroplast expansion. The abnormally rapid expansion of arc3 chloroplasts prevents chloroplast division. ARC5 facilitates the separation of the two daughter plastids and in arc5 the chloroplasts remain dumb-bell shaped and continue to expand.

The ARC6 gene acts upstream of ARC11, ARC3 and ARC5 since the arc11 arc6, arc3 arc6 and arc5 arc6 double mutants all have a chloroplast phenotype indistinguishable from arc6. Just as in the single arc6 mutants both proplastid and chloroplast division are arrested in all the double mutants with arc6. Therefore ARC6 acts pleiotropically during the initiation of both proplastid division and chloroplast division, possibly by encoding a promoter of chloroplast division.

In contrast, the chloroplast phenotypes of double mutants constructed with arc1 are all novel and additive and appear to be intermediate between the two parental phenotypes, demonstrating that ARC1 acts independently of the other four ARC genes. In the case of the ARC1 locus it is not possible to conclude from the characteristics of the single arc1 mutant alone whether the arc1 lesion affects proplastid or chloroplast division or both processes. However, analysis of arc1 arc6, arc1 arc11, arc1 arc3 and arc1 arc5 double mutants provides clear evidence that arc1 considerably accelerates proplastid division only and has no effect on chloroplast division. The chloroplast number in all the double mutants with arc1 is enhanced above the paternal chloroplast number even for those mutants which contain a second mutant gene completely inhibiting the initiation or completion of chloroplast division. The difference between the chloroplast number in arc3 (16) and arc1 arc3 (26) is a measure of the acceleration of proplastid division which occurs in the presence of arc1. In addition, the effect of arc3 can be seen by the increase in the size of the chloroplasts in arc1 arc3 beyond the sizes of arc1 chloroplasts. It is particularly interesting that in the arc1 arc5 mutant the presence of arc5 seems to act pleiotropically and to have affected the expression of arc1 and to have further enhanced proplastid division so that about twice as many young plastids are present in arc1 arc5 mesophyll cells as in arc1 arc3.

ARC3 has an important role in the initiation of chloroplast division since in arc3 the chloroplast number (16) is the same as the final proplastid number, i.e. no chloroplast division occurs. Further understanding of the role of ARC3 comes from the analysis of the arc3 arc5 double mutant whose chloroplast number and phenotype is identical to that of arc3. The lesion in ARC3 completely prevents the initiation of all chloroplast division: there are no dumb-bell shaped profiles in arc3 arc5 as would be seen if division had proceeded and then been stopped, as in arc5. In arc3 arc5 the expansion of the chloroplasts continues unchecked until the proportion of the cell surface covered by chloroplasts becomes the same as in arc3. Ellis & Leech (1985) provided evidence that chloroplast division only occurs in young chloroplasts when they are less than a certain size. We suggest that ARC3 ensures that the young chloroplasts expand to this optimal size to coincide with the initiation of chloroplast division. As a result of the arc3 mutation, expansion of the young chloroplasts is unchecked and the chloroplasts are too large to be able to divide at the time chloroplast division would normally be initiated.

The mutations which give rise to arc3 and arc11 have in common that they disrupt the normal relationship between chloroplast division and chloroplast expansion although ARC3 and ARC11 have different roles in this co-ordination. In arc11, the young chloroplasts divide asymmetrically (Fig. 2b and insert) and apparently only once giving rise to daughter chloroplasts of different sizes. The appearance of the arc11 arc3 double mutant provides considerable further support for our interpretation of the role of ARC11. The arc11 arc3 double mutant has the arc11 chloroplast heterogeneity and number per cell (29) but the mean sizes of both the small and large chloroplasts are larger than in arc11. Thus, in arc11 arc3 a lesion in the ARC3 gene facilitates chloroplast expansion resulting in larger chloroplasts as in arc3 and the lesion in the ARC11 gene allows only one round of asymmetric chloroplast division resulting in the typical arc11 chloroplast number. Further evidence supporting our interpretation of the role of ARC11 comes from the examination of the arc11 arc5 double mutant. In arc11 arc5, chloroplasts capable of division in arc11 cannot complete this division because of the presence of arc5 and are seen as larger than wild-type dumb-bells typical of arc5. The presence of arc5 has prevented the final separation of the two daughter plastids. In arc11 arc5 there is definite interaction in the expression of the two mutant genes since the dumb-bell shaped chloroplasts appear symmetrical in the mature cells. We do not know the exact time during chloroplast development when ARC11 operates; ARC11 could function just before chloroplast division is initiated (as illustrated in Fig. 7) or at any time earlier during the chloroplast expansion phase.

The analyses of the double mutants arc1 arc5 and arc3 arc5 confirm that the ARC5 gene plays a very late role in the final stages of the chloroplast division process and effects the separation of the daughter plastids. Similar to arc5, in arc1 arc5 all the chloroplasts are arrested in the dumb-bell configuration and the daughter plastids never separate. In arc1 arc5 the chloroplast number is intermediate between the two parents confirming that ARC1 and ARC5 are on different pathways. The appearance of the arc3 arc5 double mutant confirms that ARC5 acts downstream of ARC3 on the same pathway.

On the basis of the analysis of single and double arc mutants it is clear that ARC6 is involved in the initiation of a series of divisions in the wild-type proplastids therefore the post-mitotic cells contain 15 young chloroplasts. The chloroplasts then expand and reach an optimal size for chloroplast division. The co-ordination of chloroplast division with chloroplast expansion is mediated by ARC3 with ARC11 also being involved. ARC6 is pleiotropic and is also involved in the initiation of the chloroplast division process. During division, chloroplasts become centrally constricted and the position of the narrow isthmus involves ARC11. The final separation into two equally sized daughter chloroplasts involves ARC5. The chloroplasts continue to expand and further rounds of division occur. The full wild-type chloroplast complement of 120 chloroplasts in the Landsberg erecta ecotype of Arabidopsis is attained after three complete rounds of chloroplast division. ARC1 down-regulates proplastid division in an independent pathway to ARC6, ARC11, ARC3 and ARC5.

We used only one of four allelic arc6 mutants and only one of two allelic arc3 mutants in this study. The four arc6 mutants are equally extreme and their chloroplast phenotypes are apparently identical: the two arc3 mutants are also similar. We isolated only one mutant each of arc5, arc1 and arc11 despite screening more than 20 000 individuals. arc5-1 is a strong allele since every single chloroplast is halted at the very last stage of chloroplast division. arc1-1 and arc11-1 also have chloroplast phenotypes greatly altered from wild type but are still stable. Even more severe alleles of arc1, arc3, arc5, arc6 and arc11 than the ones we have isolated may well be lethal. It is particularly interesting that no genetic mutants have been found for the stages between the initial (arc3) and final (arc5) stages of chloroplast division. It is theoretically possible that these intervening stages occur completely spontaneously without genetic control because of the physical characteristics of the chloroplast envelope (Leech et al. 1981).

Recent work has identified plant nuclear homologues of the prokaryotic cell division gene FtsZ in Arabidopsis (Osteryoung & Vierling 1995; Osteryoung et al. 1998) and in the moss Physcomitrella (Strepp et al. 1998). FtsZ, a cytoskeletal protein related to tubulin, is a vital component of the prokaryotic cell division machinery. In Escherichia coli, FtsZ is recruited to the equator of a dividing cell to form a ring around the division site. All FtsZ genes currently identified in plant species show sequence homology to one of two distinct groups; those which contain a chloroplast transit peptide sequence (FtsZ1) and those which do not (FtsZ2). The Arabidopsis FtsZ protein AtFtsZ1-1 has been shown to be imported into chloroplasts (Osteryoung & Vierling 1995). Antisense repression of Arabidopsis FtsZ genes from either of these two families decreases chloroplast numbers in the mesophyll cells (Osteryoung et al. 1998). At present there is no evidence to suggest that either of the AtFtsZ genes are homologues of any known ARC loci.

Our analysis of double arc mutants has shown that normal chloroplast division in Arabidopsis is a complex process and involves the integrated action of several independent unlinked nuclear genes. The rapidly increasing availability of polymorphic markers on the Arabidopsis chromosomes will facilitate the fine mapping of the ARC loci and the isolation of the ARC genes.

Experimental procedures

Plant material

Wild-type Arabidopsis thaliana plants Landsberg erecta (Ler) and arc mutant plants were grown in controlled conditions as described previously (Pyke & Leech 1991). The arc1, arc3 and arc5 mutants were isolated from an EMS mutagenised population of Arabidopsis thaliana (Ler) (Lehle seeds, Tucson, AZ, USA) (Pyke & Leech 1992; Pyke & Leech 1994; Robertson et al. 1996). arc6 was not tagged but was isolated from a T-DNA mutagenised population of Arabidopsis thaliana WS (Feldmann & Marks 1987; Feldmann 1991; Pyke et al. 1994; Robertson et al. 1995) obtained from the Nottingham Arabidopsis Stock Centre (University of Nottingham, Nottingham, UK; stock no. N3115). The arc11 mutant was isolated from a population of Arabidopsis thaliana (Ler) mutagenised with the NaeI deleted Ac element (Dean et al. 1992; Lawson et al. 1994). Twelve progeny from one full green seedling for each of 350 lines were analysed for mutant arc phenotypes. Two of the 12 seedlings analysed from line 122 (transformant 02213–3) displayed the arc11 phenotype. Seeds of all the genotypes used in this paper can be obtained from the Nottingham Arabidopsis Stock Centre (

Double mutant construction

Double mutants were constructed by crossing homozygous arc1, arc3, arc5, arc6 and arc11 plants in all combinations. All F1 progeny were phenotypically wild type and allowed to self fertilise. F2 and F3 seeds were sown in 12 × 8 gridded arrays. Tissue was sampled from the apical half of fully expanded first leaves of each seedling and placed into 3.5% (v/v) glutaraldehyde in 12 × 8 arrays of a corresponding microtiter plate (Pyke & Leech 1991) and chloroplast number per cell, chloroplast size and mesophyll cell size determined. Putative double mutants were allowed to grow, self fertilise and set seed.

Analysis of chloroplast number, chloroplast size and mesophyll cell size

Chloroplasts in individual isolated fixed mesophyll cells were counted using Nomarski differential interference contrast optics (Nikon Optiphot, Nikon, UK). Mesophyll cell plan areas (Pyke & Leech 1991) and individual chloroplast plan areas (Pyke & Leech 1992) were measured directly from the microscope using an image analysis system (Seescan Imaging Ltd, Cambridge, UK). Fully expanded first leaves were always used.

Ultrastructural analysis

For ultrastructural analysis, first leaves from Ler wild type and arc11 were harvested 16 days after sowing and examined after fixation and embedding in Spurr's resin as described previously for Arabidopsis (Pyke et al. 1994).

Mapping of the ARC loci

To map the position of ARC5, an F2 population originating from a cross between arc5 (Ler) and Columbia was used. Genomic DNA for PCR was isolated from leaves of 300 F2 mutant seedlings (Dellaporta et al. 1983). SSLPs (Bell & Ecker 1994) and the CAPS marker AtDMC1 (Klimyuk & Jones 1997) and the recombinant inbred map (Dean & Lister 1998) were used.

ARC6 was mapped using the Arabidopsis RFLP Mapping Set, ARMS (Fabri & Schäffner 1994), on a population of F2 mutant progeny from a cross between arc6 (WS) and Ler. Genomic DNA was isolated from 22 F2 mutant progeny by CTAB extraction (Dean et al. 1992), digested with EcoR1 and Southern blotted. Nineteen of the 25 ARMS markers showed polymorphisms between WS and Ler (Rutherford 1996). Eight of these markers were used as probes in the preliminary mapping of ARC6. The CAPS marker DFR (Konieczny & Ausubel 1993) was used to verify the map position of ARC6.

The identification of the arc11 mutant in a population of Arabidopsis thaliana carrying a transposed Ac element potentially offered a good opportunity to isolate the first ARC gene. To establish the tagged status of arc11 we analysed the co-segregation of the arc11 mutant phenotype with the Ac element in mutant siblings from the F2 of a backcross between arc11 (Ler) and Columbia. Genomic DNA was digested with SspI, separated, Southern blotted and probed with the 904 bp HindIII-EcoRI fragment of Ac (Dean et al. 1992) labelled with 50 μCi 32P-dCTP. Ac was found in 59 of the 66 individuals tested. The segregation of the Ac element away from the ARC11 locus can be explained by excision of Ac leaving a mutagenic footprint in the ARC11 gene (Bancroft et al. 1993). Plant DNA sequences flanking the Ac were isolated by IPCR. Genomic DNA from arc11 was double digested using BstYI and BclI and self-ligated. After an initial 5 min at 94°C the conditions for the amplification were as follows: 30 sec at 94°C; 30 sec at 55°C and 3 min at 72°C. The cycle was repeated 34 times. The primer pairs used for the amplification of the plant DNA flanking the 5′ and 3′ regions of the Ac were B34 and D74 (Briza et al. 1995) and DL6 (Long et al. 1993) and D71 (Briza et al. 1995), respectively. Primers designed from the 150 bp 5′ and 1.1 kb 3′ IPCR products were used to amplify over the region of Ac excision in the arc11 individuals lacking the Ac element. Sequence analysis of the resulting PCR fragments showed that the arc11 plants lacking Ac had the wild-type sequence, so the Ac element is not located within the ARC11 gene and cannot be used to isolate the gene.

To map the position of ARC11, genomic DNA from F2 mutant seedlings originating from a backcross between arc11 (Ler) and Columbia were used with SSLP markers (Bell & Ecker 1994). To map the position of the Ac element CIC (Creusot et al. 1995) and Ecker (Ecker 1990) YAC libraries were probed with the 1.1 kb 3′´IPCR fragment. Hybridisation to the YAC filters was carried out as described by Schmidt et al. (1992) except that the washing conditions were 3 × SSC, 0.1% SDS for 10 min followed by 0.1 × SSC, 0.1% SDS for 10 min. Three YAC clones anchored to nga151 on chromosome 5 were identified.


We would like to acknowledge the assistance of Kevin Pyke in some of the early analyses of arc11. We thank Colin Abbot and Alison Sutcliffe for growing the plants, Vicki Hird for isolating the arc3arc5 double mutant and Joanne Walker for help with the mapping of the ARC5 locus. We are also grateful to Ottoline Leyser for many very helpful discussions. The work was carried out under an Agricultural and Food Research Council (UK) Plant Molecular Biology II Link grant LR87/528 and a Biotechnology and Biological Sciences Research Council (UK) grant 87/PO4906 awarded to R.M.L.