Second-Hand Chloroplasts and the Case of the Disappearing Nucleus Jeffrey D

Proc. Natl. Acad. Sci. USA Vol. 93, pp. 7432-7435, July 1996 Commentary

Second-hand chloroplasts and the case of the disappearing nucleus Jeffrey D. Palmer and Charles F. Delwiche Department of Biology, Indiana University, Bloomington, IN 47405

The basic facts of chloroplast and mitochondrial endosymbio- the chlorarachniophyte and cryptomonad nucleomorphs most sis are by now the stuff of textbooks. Early in eukaryotic commonly group together, sometimes strongly so, and this evolution, an anaerobic protist engulfed a respiring bacterium, seemingly monophyletic nucleomorph assemblage usually af- probably an a-proteobacterium (1, 2), and the resulting part- filiates with red algae (12, 16). Thus, taken at face value, the nership between symbiont (bacterium turned mitochondrion) rRNA trees suggest that all* nucleomorphs arose from a and host gave rise to most, perhaps all, extant eukaryotes. common secondary symbiosis, involving a red alga, followed by Subsequently, a respiring eukaryote acquired a second bacte- differential evolution of photosynthetic pigments and appara- rium, unquestionably a cyanobacterium (3, 4), which became tus. the chloroplast, thereby endowing a subset of eukaryotes with The key element in the apparent resolution of this puzzle by photosynthetic powers. Massive relocation of gene function Van de Peer et al. (8) is their recently developed "substitution from organelle to nucleus ensued, by both direct transfer and rate calibration" method (17) for estimating the true evolu- indirect substitution, so that the great majority of organellar tionary distance between small subunit rRNA sequences (the proteins are now encoded by nuclear genes (1, 2, 3, 5). Current actual number of substitutions) from raw sequence dissimilar- evidence (1-4) suggests that all chloroplasts probably trace ity (the observed number of substitutions). Application of this back to a single primary (prokaryotic/eukaryotic) endosym- method to a neighbor-joining analysis of 106 rRNA sequences biont event, as do all mitochondria. from the so-called "crown" of the eukaryotic phylogenetic tree Less well appreciated is the fascinating realm of secondary splits apart the two nucleomorph lineages, strongly placing the endosymbiosis, in which one eukaryote engulfs another eu- chlorarachniophytes in the expected position (i.e., with green karyote and permanently retains part of its prey as a degen- algae) and weakly grouping the cryptomonads with red algae erate endosymbiont. Although no eukaryotes are known to (8). The authors attribute the previous clustering of both have obtained their mitochondria secondarily, a rich diversity nucleomorph types, often together with red algae, to the of eukaryotes have acquired their plastids via a multitude of relative inability of the variety of methods used to handle the independent, but strikingly parallel, secondary symbioses (Fig. well-known tendency of long branches to artefactually cluster 1). Two papers in this issue investigate the phylogeny and not only together but also deeply (17-19). molecular biology of chlorarachniophytes, a group of unicel- The same pattern of relationships (chlorarachniophytes with lular amoeboid algae whose secondarily acquired nucleus, green algae, cryptomonads with reds) observed for nucleo- termed the nucleomorph, is located between the inner and morph/nuclear rRNA sequences (8) is also seen in phyloge- outer pair of membranes surrounding its chloroplast (Fig. 1). netic analyses of plastid-encoded rbcL (20) and rRNA (8, 20) The study by Van de Peer et al. (8) effectively settles a sequences. In the latter case, however, a different type of phylogenetic conundrum concerning the origin of the phylogenetic artefact must be overcome. The chlorarachnio- chlorarachniophyte's algal endosymbiont, while that by Gilson phyte plastid rRNA sequences falsely cluster with similarly and McFadden (9) reveals a startling array of compaction AT-rich sequences of non-green algae unless one uses meth- mechanisms employed by the highly miniaturized nucleo- ods, such as transversion analysis (8) or LOGDET transforma- morph genome of the chlorarachniophyte. tion (20), which are less sensitive to base compositional bias. In sum, the application of analytical methods that correct for Puzzles of Chlorarachniophyte Ancestry rate or compositional heterogeneity in rRNAyields a satisfying picture of highly congruent evidence from photosynthetic The chlorarachniophyte is in essence a cell within a cell. The pigments and three molecular data sets. As long suspected (6, algal endosymbiont is highly reduced, comprising a single 15), the chlorarachniophyte endosymbiont is indeed a green plastid, a tiny nucleus (the nucleomorph), a vestigial cytoplasm alga and clearly represents an independent secondary symbi- (the space between the two inner and two outer "plastid" osis from that which gave rise to cryptomonads, whose endo- membranes) containing a few ribosomes, and a remnant symbiont is, it now seems (21), a red alga. plasma membrane (i.e., the third plastid membrane; the fourth is believed to be derived from the vacuolar membrane of the Unequal Rates: The Achilles Heel of Molecular Phylogeny host cell) (10). In keeping with its chimeric origin, the chlorarachniophyte contains two phylogenetically unrelated The significance of the two papers by Van de Peer et al. (8, 17) sets of cytoplasmic ribosomal RNAs (rRNAs), one encoded by extends well beyond the question ofnucleomorph phylogeny to the nucleomorph and the other by the host cell nucleus the general and vexing problem of unequal rates of molecular (10-12). The above description also fits a second group of evolution and their confounding effects on attempts to recon- algae, the cryptomonads (13-15). However, based on their struct phylogeny (17-19). Despite a recent ambitious attempt photosynthetic pigments, these two groups have generally been (22-24) to estimate divergence times across the entirety of thought (6, 15) to be derived from independent secondary life's panoply by using molecular clock-based analyses of many symbioses, the chlorarachniophytes from uptake of a green poorly-sampled proteins, it is clear from inspection of any alga (both have chlorophyll b) and the cryptomonads from red well-sampled molecule, be it rRNA or protein, that rates of algae (both have phycobilins, although cryptomonads also sequence change are often highly variable, and in unpredict- have chlorophyll c). It is therefore puzzling that previously able ways (i.e., a group that changes rapidly for one molecule published phylogenies of cytoplasmic rRNA fail to group the can be slow for another, and vice-versa) (25-31). And perhaps chlorarachniophyte nucleomorph with green algae; instead, even more vexing than the overall rate variation is change in 7432 Downloaded by guest on September 28, 2021 Commentary: Palmer and Delwiche Proc. Natl. Acad. Sci. USA 93 (1996) 7433

Secondary Plastid

Green Algae (& Plants) Chlorarachniophytes Chromophytes Red Algae Cryptomonads Haptophytes Glaucocystophytes Euglenoids Apicomplexans? Dinoflagellates? FIG. 1. Origin and spread of chloroplasts by primary and secondary endosymbiosis, respectively. Groups of eukaryotes with primary or secondary plastids are listed below the appropriate cell type. Note the variable membrane status of the secondary plastids shown in the right most part of the figure: some have four bounding membranes (as shown), while others have three or possibly fewer membranes (see text). Also, in many cases there is a direct continuity (not shown) between the outermost plastid membrane (which often contains ribosomes on its cytoplasmic side) and the outer membrane of the nuclear envelope (6, 7). the mode of evolution within a lineage, for no method is yet crown group reassignments, then, assuming their rRNA genes available that can tolerate such changes. Rather, Van de Peer's are not derived by lateral transfer (which seems a safe assump- method takes into account varying rates of change at different tion), these organisms will pose challenging test cases for the sites within a sequence, but retains the assumption that the efficacy of all methods, including Van de Peer's (8, 17), that mode of evolution (i.e., base composition and pattern of attempt to correct for rate heterogeneity in rRNA evolution. site-to-site rate variation) is uniform across a tree. The importance of testing the reigning rRNA-based, biphasic Van de Peer et al. (8, 17) also apply their substitution rate theory of eukaryotic evolution by both reanalyzing current calibration method to another group of notably rapidly chang- data using better methods and seeking new forms of evidence ing rRNA sequences, from the malarial parasite genus Plas- from other molecules cannot be overstressed. modium, and for the first time obtain reasonably strong The Smallest Nuclear Genome: A Marvel of Compaction support for its placement with other hematozoan members of the parasitic phylum Apicomplexa. In contrast-and what Previous work from McFadden's group (11, 39) showed that really underscores the challenges posed by rapidly evolving the chlorarachniophyte nucleomorph genome, at 380 kb, is the sequences-consider a recent study by Kumar and Rzhetsky (32). Despite also using a method designed to take into account site-to-site rate variation in rRNA sequences, their The "crown" of analyses place the Plasmodium lineage entirely outside the eukaryotic rRNA diversity entire "crown" group of eukaryotes, including not only phylum Apicomplexa but also the superphylum Alveolata (which includes ciliates and dinoflagellates)! Thus, Van de Peer's method for dealing with site-to-site rate variation seems particularly promising, and we therefore look forward to its application to the putatively deeper divergences of eukaryotes. The current, rRNA-based view of eukaryotic evolution is that of an early period, from about 2.5 to 1.0 billion years ago, of fairly gradual evolution during which a number of protistan lineages successively emerged, followed by a nearly simultaneous divergence about 1.0 billion years ago of the "crown" taxa (animals, plants, fungi, alveolates, and several other protistan groups) (33, 34). As Fig. 2 shows, the crown rRNAs form a fairly tight-knit group, with the successively deeper rRNA sequences representing strikingly long and seemingly isolated branches. Might some, perhaps many, of these putatively basal rRNA branches represent instead rapidly changing lineages that have fallen artefactually deeply (e.g., outside the crown) due to long-branch-attraction prob- lems (as did Plasmodium in ref. 32)? Two prime candidates, ripe for reexamination using better are the slime and analytical methods, molds Dictyostelium FIG. 2. Phylogenetic relationships of eukaryotes based on small Physarum. For, despite their moderately deep and separate subunit rRNA -sequences. Shown is the eukaryotic portion of a placements in all rRNA trees (refs. 33 and 34; Fig. 2), these two maximum likelihood analysis of rRNAs representing all of life, with slime molds are closely related to animals, fungi, and other branch lengths proportional to inferred evolutionary distance (re- crown taxa for essentially all proteins that have been examined drawn from figure 2 of ref. 35). The crown taxa shown represent the (25-27, 36-38). Another group for which recent, albeit scant- following groups: apicomplexans (Babesia), ciliates (Paramecium), red ier, protein data challenge the conventional rRNA perspective algae (Porphyra), chromophytes (Costaria), oomyctes (Achyla), green are the microsporidians, which are obligately intracellular, algae and plants (Zea), cryptomonads (Cryptomonas), fungi (Copri- amitochondrial These form a in all nus), and animals (Homo). Other crown groups relevant to this paper parasites. deep lineage are chlorarachniophytes, glaucocystophytes, haptophytes, and rRNA trees (ref. 34; Fig. 2), yet features of their ,B-tubulin (28) dinoflagellates (see Fig. 1). The shadowed, arrowed taxa represent and elongation factor EF-la (29) proteins provocatively sug- those groups (slime molds in blue, microsporidia in violet), suggested gest a relatively recent, specifically fungal origin. If additional by protein sequence data to be more derived phylogenetically than as data for slime molds and microsporidians corroborate their shown here based on rRNA (see text). Downloaded by guest on September 28, 2021 7434 Commentary: Palmer and Delwiche Proc. Natl. Acad. Sci. USA 93 (1996) smallest nuclear genome known, consisting of but three linear morph) at an early stage (3, 5) following primary plastid chromosomes, each containing a subtelomeric rRNA operon endosymbiosis, and then again from the nucleomorph to the at both ends. Gilson and McFadden (9) now explore for the host nucleus following secondary symbiosis. This initial snap- first time issues relevant to the coding capacity, gene organi- shot of the nucleomorph genome suggests that it is only a zation, and expression of this vestigial nucleus. Their results tertiary player in quartering the genes for the chlorarachnio- are provocative in two ways. First, they reveal an unprece- phyte plastid, well behind both the primary player (the nuclear dented (for a nuclear genome) degree of genetic compaction. genome) and the secondary player (the plastid genome). The exceptionally short spacers (averaging only 65 bp) be- Retention of a vestigial genome that largely encodes genetic- tween the eight newly sequenced coding regions put an obvious machinery factors for self-perpetuation has heretofore been squeeze on the many cis-acting transcriptional sequences seen only in the degenerate plastid genomes of permanently normally found between eukaryotic genes. This squeeze is nonphotosynthetic plants and "algae," which contain roughly eased by a second condition: two of the nucleomorph protein 95% genetic-machinery genes and 5% "raison d'etre" genes genes are shown to be cotranscribed, and others may be as well (45-47). The nucleomorph genome represents even greater (9). Whether this dicistronic RNA serves as message for both genetic and biochemical waste in two ways; not only does it proteins, as in classical bacterial operons, or is processed (e.g., contain -5-10 times as many genes as these rudimentary by cleavage, polyadenylylation, and trans-splicing), as in the plastid genomes, but it also entails the cost of maintenance of polycistronic "operons" commonly found in trypanosomes and its entire host compartment, the nucleomorph itself. the nematode Caenorhabditis elegans (40), remains to be determined. A third and equally unusual result of compaction Parallel Evolution Writ Large is overlapping genes; the overlap in the nucleomorph is notable in that the mRNA for the ribosomal protein gene S13 contains On a gross scale, the independently derived cryptomonad an antisense copy of a structural RNA (for the spliceosomal nucleomorph has undergone a similar degree of reductive RNA U6) located wholly within its 3' untranslated region. evolution. Its genome also exists as three tiny chromosomes, In all three respects-very short intergenic spacers, operons, which sum to only 550-660 kb (48, 49). As with the chlorarach- and overlapping genes-the organization of the chlorarachniophyte nucleomorph, only one of the first 10 or so genes niophyte nucleomorph genome is decidedly more prokaryotic sequenced specifies a plastid protein, the rest again being than eukaryotic. But the fourth element of genetic compaction concerned simply with nucleomorph maintenance (U. Maier, (9) resoundingly reminds us that this miniaturized genome is personal communication). Whether the fine-scale organiza- truly of nuclear descent. For, despite the evident pressures to tion of nucleomorph genes is as remarkably compact in streamline its genome, the nucleomorph protein genes are as cryptomonads as in chlorarachniophytes remains to be seen. densely packed with spliceosomal introns (5.5 per kb of coding There is a second pattern of parallel evolution in plastid sequence) as the most intron-rich genomes known, including secondary symbiosis. This represents the end result of the our own (41). However, the 12 nucleomorph introns charac- extensive transfer of gene function from nucleomorph to host terized thus far are remarkably tiny and astonishingly invariant nucleus seen in chlorarachniophytes and cryptomonads- in size: one is 18 bp in length, one is 20 bp, and the other 10 namely, the complete disappearance of the nucleomorph (Fig. are 19 bp (7). The previous record for smallest spliceosomal 1). At least four groups of eukaryotes seem to have each intron was 20 bp, in the ciliate Paramecium tetraurelia, but this independently acquired their plastids by secondary symbiosis was merely the tail of a Gaussian spread from 20 to 33 bp and then subsequently lost their nucleomorph entirely (3, 4, 6). (average of 26 bp) (42). How such precisely tiny and featureless In the absence of nucleomorphs, evidence for secondary introns (aside from the canonical 5' GT and 3' AG, the introns symbioses, as well as their multiple occurrences, comes from are 88% A+T, and neither they nor flanking exon sequences consideration of both the number of plastid-bounding mem- show any consensus) are properly recognized and removed by branes and phylogenies derived from nuclear and plastid the nucleomorph splicing apparatus remains to be determined. genes. Two of these anucleomorphic groups, chromophytes The presence of numerous but tiny introns in the highly (brown algae, diatoms, and several other types of algae) and compact chlorarachniophyte nucleomorph markedly extends haptophytes, contain four membranes around their plastids the lesson already learned from the streamlined genome of the (just as in chlorarachniophytes and cryptomonads) and prob- pufferfish (43)-that in some situations, wholesale shrinkage ably acquired their plastids from a red alga (3, 4, 6). Eugleno- of a genome's set of introns is more likely than their actual phytes, with three plastid-bounding membranes, contain a elimination (44). green-algal plastid (4, 5, 6, 20), while the recently discovered The second set of surprises from Gilson and McFadden's (3, 46) plastid in the nonphotosynthetic, parasitic phylum study (9) concerns the coding function of the nucleomorph Apicomplexa, with an unsettled number of membranes (50), is genome, which seems largely self-perpetuating: 10 of the 11 tentatively also of green algal origin (C.F.D., R. J. M. Wilson, genes identified thus far appear to function only in the D. S. Roos, and J.D.P., unpublished data). maintenance and expression of the nucleomorph's own genetic As if six independent cases of ancient, long-standing sec- apparatus. These include genes for three rRNAs, three spli- ondary plastid symbiosis weren't enough, dinoflagellates rep- ceosomal components, two cytoplasmic ribosomal proteins, a resent a group with secondary symbiosis gone wild. Our subunit of a eukaryotic RNA polymerase, and an RNA understanding here is more primitive because gene sequences helicase. Only one putatively chloroplast-targeted protein, the are not yet available from any dinoflagellate plastids. Pho- catalytic subunit of the ATP-dependent Clp protease, is topigment and ultrastructural data suggest that "typical" (i.e., encoded by the 15 kb of nucleomorph DNA sequenced thus far peridinin-containing) dinoflagellates (with three plastid mem- (9, 39). Assuming that this initial sampling is not biased and branes) probably arose via a single, relatively ancient second- that the as-yet-uncharacterized plastid genome of chlorarach- ary symbiosis. However, several other dinoflagellates seem to niophytes is not unusually rich in genes (plastid genomes have relatively recently, permanently, and independently ac- typically encode only about 10-20% of plastid proteins; refs. quired plastids from a wide variety of eukaryotic algae, 2, 5), these data strongly suggest that most of the roughly 1000 including green algae, cryptomonads, chromophytes, and hap- different proteins found in plastids are now encoded by the tophytes (4, 6, 51). Since most of these endosymbionts are host nucleus of the chlorarachniophyte. Many of these genes themselves the product of secondary symbiosis, this means that are undoubtedly of cyanobacterial origin, and most of these many dinoflagellates represent a tertiary stage of plastid sym- would have been transferred not once but twice during evo- biosis (not illustrated in Fig. 1). The extent of gene transfer lution, first from the chloroplast to the nucleus (now nucleo- from tertiary symbiont to dinoflagellate nucleus remains to be Downloaded by guest on September 28, 2021 Commentary: Palmer and Delwiche Proc. Natl. Acad. Sci. USA 93 (1996) 7435 determined. In contrast, some dinoflagellate chloroplasts are 21. Cavalier-Smith, T., Couch, J. A., Thorsteinsen, K. E., Gilson, P., nonsymbiotic in origin; these "kleptochloroplasts" (51) are Deane, J., Hill, D. A. & McFadden, G. I. (1996) Eur. J. Phycol., captured by endocytotic uptake of foreign cytoplasm and in press. used as organelles. Stolen chloro- 22. Doolittle, R. F., Feng, D.-F., Tsang, S., Cho, G. & Little, E. temporarily photosynthetic (1996) Science 271, 470-477. plasts are used by a wide variety of eukaryotes, not only other 23. Morrell, V. (1996) Science, 271, 448. microbes, such as ciliates and foraminiferans, but also solar- 24. Mooers, A. 0. & Redfield, R. J. (1996) Nature (London) 379, powered, green sea slugs, which array their ingested chloro- 587-588. plasts into special surface cells (10, 52, 53). 25. Bhattacharya, D. & Ehlting, J. (1995) Arch. Protistenk. 145, Thus, starting from a notably singular (but see refs. 4 and 54 155-164. on Paulinella) and far-reaching event-the single primary 26. Baldauf, S. L. & Palmer J. D. (1993) Proc. Natl. Acad. Sci. USA symbiosis of cyanobacterium and eukaryote over one billion 90, 11558-11562. years ago-chloroplasts have been permanently acquired by a 27. Drouin, G., Moniz de Sa, M. & Zuker, M. (1995) J. Mol. Evol. 41, lineages through many independent, 841-849. wide variety of eukaryotic 28. Edlind, T. D., Li, J., Visvesvara, G. S., Vodkin, M. H., McLaugh- but often strikingly parallel, second- and third-hand symbioses, lin, G. L. & Katiyar, S. K. (1996) Mol. Phylogenet. Evol. 5, as well as transiently via ingestion of both whole algal cells (51, 359-367. 53) and their chloroplasts. The study by Gilson and McFadden 29. Kamaishi, T., Hashimoto; T., Nakamura, Y., Nakamura, F., (9) highlights the unusual molecular biology that has accom- Murata, S., Okada, N., Okamoto, K., Shimizu, M. & Hasegawa, panied endosymbiont nuclear degeneration in one such sec- M. (1996) J. Mol. Evol. 42, 257-263. ondary symbiosis, while recent studies on plastid protein 30. De Rijk, P., Van de Peer, Y., Van den Broeck, I. & De Wachter, targeting emphasize the complications imposed by the need to R. (1995) J. Mol. Evol. 41, 366-375. import proteins from host cytoplasm across three or four 31. Fitch, D. H. A., Bugaj-Gaweda, B. & Emmons, S. W. (1995) Mol. Biol. Evol. 12, 346-358. membranes (55). 32. Kumar, S. & Rzhetsky, A. (1996) J. Mol. Evol. 42, 183-193. We thank Geoff McFadden and Uwe Maier for providing 33. Knoll, A. H. (1992) Science 256, 622-627. unpublished results, and James Allen, Rupert De Wachter, 34. Sogin, M. L. (1994) in Early Life on Earth, Nobel Symposium No. Will Fischer, Geoff McFadden, and Kathleen Pryer for com- 84, ed. Bengtson, S. (Columbia Univ. Press, New York), pp. ments on the manuscript. 181-192. 35. Barns, S., Delwiche, C. F., Palmer, J. D. & Pace, N. R. (1996) 1. Gray, M. W. (1992) Int. Rev. Cytol. 141, 233-357. Proc. Natl. Acad. Sci. USA 93, 1144. 2. Gray, M. W. (1995) in The Molecular Biology ofPlant Mitochon- 36. Kuma, K., Nikoh, N., Iwabe, N. & Miyata, T. (1995) J. Mol. Evol. dria, eds. Levings, C. S., III, & Vasil, I. K. (Kluwer, Dordrecht, 41, 238-246. The Netherlands), pp. 635-659. 37. Loomis, W. F. & Smith, D. F. (1995) Experientia 51, 110-115. 3. Reith, M. (1995) Annu. Rev. Plant Physiol. Plant Mol. Biol. 46, 38. Klenk, H.-P., Zillig, W., Lanzendorfer, M., Grampp, B. & Palm, 549-575. P. (1995) Arch. Protistenk. 145, 221-230. 4. Bhattacharya, D. & Medlin, L. (1995) J. Phycol. 31, 489-498. 39. Gilson, P. & McFadden (1995) Chromosoma 103, 635-641. 5. Palmer, J. D. (1991) in Molecular Biology ofPlastids, eds. Bogo- 40. Spieth, J., Brooke, G., Kuersten, S., Lea, K. & Blumenthal, T. rad, L. & Vasil, I. K. (Academic, San Diego), pp. 5-53. (1993) Cell 73, 521-532. 6. Gibbs, S. P. (1993) in Origins of Plastids, ed. Lewin, R. A. 41. Palmer, J. D. & Logsdon, J. M., Jr. (1991) Curr. Opin. Genet. Dev. (Chapman & Hall, New York), pp. 107-121. 1, 470-477. 7. Whatley, J. M. (1993) in Origins of Plastids, ed. Lewin, R. A. 42. Russell, C. B., Fraga, D. & Hinrichsen, R. D. (1994) Nucleic (Chapman & Hall, New York), pp. 77-106. Acids Res. 22, 1221-1225. 8. Van de Peer, Y., Rensing, S. A., Maier, U.-G. & De Wachter, R. 43. Elgar, G., Sandford, R., Aparicio, S., Macrae, A., Venkatesh, B. (1996) Proc. Natl. Acad. Sci. USA 93, 4467. & Brenner, S. (1996) Trends Genet. 12, 145-150. 9. Gilson, P. R. & McFadden, G. I. (1996) Proc. Natl. Acad. Sci. 44. Fink, G. R. (1987) Cell 49, 5-6. USA 93, 4318. 45. Wolfe, K. H., Morden, C. W. & Palmer, J. D. (1992) Proc. Natl. 10. McFadden, G. & P. Gilson (1995) Trends Ecol. Evol. 10, 12-17. Acad. Sci. USA 89, 10648-10652. 11. McFadden, G. I., Gilson, P. R., Hofmann, C. J. B., Adcock, G. J. 46. Preiser, P., Williamson, D. H. & Wilson, R. J. M. (1995) Nucleic & Maier, U.-G. (1994) Proc. Natl. Acad. Sci. USA 91,3690-3694. Acids Res. 23, 4329-4336. 12. Cavalier-Smith, T., Allsopp, M. T. E. P. & Chao, E. E. (1994) 47. Gockel, G., Hachtel, W., Baier, S., Fliss, C. & Henke, M. (1994) Proc. Natl. Acad. Sci. USA 91, 11368-11372. Curr. Genet. 26, 256-262. 13. Douglas, S. E., Murphy, C. A., Spencer, D. F. & Gray, M. W. 48. Eschbach, S., Hofmann, C. J. B., Maier, U.-G., Sitte, P. & (1991) Nature (London) 350, 148-151. Hansmann, P. (1991) Nucleic Acids Res. 19, 1779-1781. 14. Gray, M. W. (1994) ASM News 60, 423-428. 49. McFadden, G. I., Gilson P. R. & Douglas, S. E. (1994) J. Cell Sci. 15. Hibberd, D. J. (1990) in Handbook ofProtoctista, eds. Margulis, 107, 649-657. L., Corliss, J. O., Melkonian, M. & Chapman, D. J. (Jones & 50. McFadden, G. I., Reith, M. E., Munholland, J., & Lang-Un- Bartlett, Boston), pp. 288-292. nasch, N. (1996) Nature (London) 381, 482. 16. Bhattacharya, D., Helmchen, T., Bibeau, C. & Melkonian, M. 51. Schnepf, E. (1993) in Origins of Plastids, ed. Lewin, R. A. (1995) Mol. Biol. Evol. 12, 415-420. (Chapman & Hall, New York), pp. 53-76. 17. Van de Peer, Y., Van der Auwera, G. & De Wachter, R. (1996) 52. Taylor, F. J. R. (1974) Taxon 23, 229-258. J. Mol. Evol, 42, 201-210. 53. Margulis, L. (1993) Symbiosis in Cell Evolution (Freeman, New 18. Felsenstein, J. (1978) Syst. Zool. 27, 401-409. York). 19. Olsen, G. J. (1987) Cold Spring Harbor Symp. Quant. Biol. 52, 54. Bhattacharya, D. Helmchen, T. & Melkonian, M. (1995) J. 825-837. Eukaryotic Microbiol. 42, 65-69. 20. McFadden, G. I., Gilson, P. R. & Waller, R. F. (1995) Arch. 55. Grossman, A. R., Bhaya, D., Apt., K. E. & Kehoe, D. M. (1995) Protistenk. 145, 231-239. Annu. Rev. Genet. 29, 231-288. Downloaded by guest on September 28, 2021