Somatic polyploidization and cellular proliferation drive body size evolution in nematodes

We studied free-living terrestrial nematodes belonging to three families: Rhabditidae, Cephalobidae, and Panagrolaimidae in the order Rhabditida. The wild-type C. elegans N2 strain used in this study was derived from the Caenorhabditis Genetics Center (Minneapolis) in 1996 and has been kept in continuous culture in the Leroi laboratory since; it is designated RX008 to distinguish it from other N2s. Other species were Acrobeloides nanus (DF5047), Acrobeloides maximus (DF5048), Panagrellus redivivus (MT8872), Panagrolaimus rigidus (AF36), Rhabditoides regina (DF5012), Pellioditis sp. (EM434), Pellioditis typica (DF5025), Rhabditella octopleura (DF5044), Oscheius myriophila (BW290), Oscheius dolichuroides (DF5018), and Oscheius sp. (DF5000). The following mutant C. elegans strains were used: LGIII: daf-4 (m63), sma-2 (e502), dpy-2 (e8); all are loss-of-function mutations (25–27). All strains were cultured on agar plates seeded with Escherichia coli (OP50) and incubated at 20°C. (28).

Growth curves were determined for each strain from worms grown individually on 5-cm Petri dishes. Worms were measured at 8-h intervals after hatching until maturity and then every 24 h until death (sample size n = 25). Images were captured at ×50 (Wild dissecting microscope) by using a video camera (JVC KY-F50E) attached to a Power Macintosh running nih image 1.62. Length and area of each image were determined by using object image 1.62; volume was estimated assuming cylindrical body shape. Maximum volume was estimated from a logistic function fitted to each growth curve by least-squares nonlinear regression. Age at maturity was taken as age at final molt as scored by vulval eversion. Maximum body size of mutant strains was determined 96 h after hatching (n = 20).

For hypodermal cell counts, young adult worms were anesthetized with 1% sodium azide, mounted as for cell lineage determination (11), and viewed at ×1000 under differential interference contrast optics with a Nikon Eclipse E600 microscope. Images were captured with a CV-M300 camera attached to a Power Macintosh running nih image 1.62 and reconstructed by using Adobe PhotoShop 4.0. For comparisons among species, all nuclei, except neuronal nuclei, between the posterior pharyngeal bulb and anus were counted; for comparisons among mutants, all nuclei between the mouth and tail-tip, except neuronal nuclei, were counted. Bilateral symmetry was assumed and counts are on one side of each animal only. n = 10–13 for each species and mutant strain.

DNA content was determined by microdensitometry. Upon completing growth, worms were fixed in Carnoy's solution for a minimum of 24 h, stained in a 0.007 mg/ml solution of 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) (28), and viewed with a Leitz epifluorescence microscope. Images of nuclei were collected by using a CV-M300 video camera, and a Scion LG3 framegrabber mounted in a Power Macintosh running scion image 1.62a. On average, 17 nuclei per worm were measured; n = 6–10 for each species; n = 16–22 for each mutant strain.

Haploid genome sizes were estimated, using DAPI-based densitometry, from sperm taken from freeze-fractured nematodes. For the two Acrobeloides species, which are parthenogenetic and so have no sperm, we inferred a haploid value from a variety of neuronal cells associated with the pharynx, the tail, and the ventral cord. These are the smallest nuclei within the organism and were therefore assumed to be diploid, as they are in C. elegans (16). DNA contents were converted from pixels to megabases (Mb) by taking the haploid genome size of C. elegans, 97 Mb, as a standard (29). We confirmed the known linear relationship between DNA content and DAPI fluorescence (30) by using haploid and diploid cells from C. elegans N2 and from 4N C. elegans, SP346 (31). Estimates of DNA contents for the sperm of different species and the hypodermal nuclei of different C. elegans mutants were done twice; the correlation between blocks was 95% and 96%, showing the high repeatability of the technique and justifying the pooling of blocks.

Confocal images were taken with an MRC 600 and reconstructed by using VoxelView on a Silicon Graphics workstation.

The phylogeny is based on 18S rDNA data (15, 32, 33). Phylogenetic contrasts on log-transformed data were obtained from caic 2.0, assuming equal branch lengths (34, 35), and relationships between variables were tested by regression through the origin. Basic statistics were calculated by using jmp 3.2.2 (SAS Institute) or Excel 98 (Microsoft); unless stated otherwise, all analyses are on log-transformed data.

Fig. 1 shows the phylogenetic relationships of these species and demonstrates that body size has evolved repeatedly within several genera. Table 1 shows that haploid genome size, hypodermal nuclear number, and their degree of somatic polyploidization have also evolved repeatedly. We have not studied enough species, nor is the phylogeny of the Rhabditida sufficiently well known, for us to be able to determine the ancestral states for any of these traits or clades.

The number of nuclei visible in the adult hypodermal syncytia ranges from 65 in C. elegans to 246 in Panagrellus redivivus; lineaging of several species has confirmed that all these cells, except for a few visible in the hatchling, are the product of the lateral seam cells, H, V, and T (12, 14). Haploid genome size, as measured by DNA densitometry, also varies 3-fold, with Oscheius sp. (DF5000) having the smallest genome, 72.4 (±3.9) Mb, and the two Pellioditis species having the largest at 258.9 (±14.9) Mb and 240.3 (±9.9) Mb (means and 95% confidence interval).

Independent of haploid genome size, the ploidy of hypodermal nuclei is also evolutionarily variable, and is most likely the result of varying numbers of rounds of endoreduplication. As discussed above, Hedgecock and White (16) showed that in C. elegans lateral seam cell descendants endoreduplicate once shortly after fusion with the hypodermal syncytium, giving 4C nuclei (C being the haploid DNA content). They did not follow endoreduplication beyond the late L4. We have found that by early adulthood (45 h) hypodermal nuclei have a mean ploidy of 8.0C (±0.8) and, by late adulthood (211 h) when the worm is fully grown, a mean ploidy of 10.7C (±0.7). This result implies that some, but not all, nuclei undergo two postlarval rounds of endoreduplication; however, some nuclei of even higher ploidy were observed. All species show at least some degree of polyploidization of hypodermal nuclei, but the extent varies from 3.9C (±0.3) in Rhabditella octopleura to 16.6C (±3.4) in Oscheius myriophila. In all species somatic polyploidization was taken to be due to endoreduplication, as no obvious chromosomal condensation was seen. The adult ploidy of hypodermal nuclei can differ dramatically even among congeners; Oscheius sp. (DF5000), for example, has a ploidy of 9.2C (±1.3), nearly half that of O. myriophila. Fig. 2 shows a confocal reconstruction of the nuclei of O. myriophila and O. sp. (DF5000). The difference in the final DNA content of the hypodermal nuclei is a consequence of difference between these species in both haploid genome size and the number of rounds of endoreduplication that the nuclei of these species undergo (Table 1). A characteristic of our data is that by late adulthood the endoreduplicate nuclei do not fall into clear polyploid series (2C, 4C, 8C, etc.). This could be because of measurement error, partial endoreduplication by the selective loss or amplification of some genome regions, or partial endoreduplication of the entire genome. The repeatability of our technique argues against measurement error (see above). As regards the second and third possibilities, we can exclude only chromosomal diminution of the sort found in ascarids because diploid larval nuclei have been detected in all species studied here (data not shown); chromosomal diminution has, in any case, not been found in C. elegans or in any other free-living nematode so far (36).

Given this evolutionary variation in cell number, genome size, and ploidy, we now ask which of these variables explains the evolution of body size.

Fig. 3 shows the relationship across species (or independent contrasts) between body size and the number of hypodermal nuclei. Across species, hypodermal volume is not significantly correlated with nucleus number (r = 0.44; P = 0.15) but is across independent contrasts (P = 0.02). Examination of the data shows why this relationship is weak. Pellioditis sp. (EM434) with 140 nuclei is 13 times larger than Panagrolaimus rigidus with 202. Similar discrepancies between nucleus number and body size are found at lower taxonomic levels as well. O. dolichuroides with 108 nuclei is larger than O. sp. (DF5000) with 82, but smaller than O. myriophila, also with 82. We conclude that body size is partly independent of the number of hypodermal nuclei.

Can haploid genome size or hypodermal ploidy explain more of the variation in body size than nucleus number? At first glance it appears not, for neither is significantly correlated with body size. The correlation between genome size and body size is not significant (across species r = 0.35; P = 0.3; contrasts P = 0.1), nor is that between mean hypodermal ploidy and body size (across species r = 0.5; P = 0.1; contrasts P = 0.1). However, simultaneous analysis of all three variables shows the influence of at least two on body size. A backward stepwise multiple regression of nucleus number, genome size, and ploidy across species shows a significant effect of nucleus number (P = 0.02), genome size (P = 0.03), and ploidy (P = 0.0013); no higher-order interactions are significant. A similar analysis of the contrasts (multiple regression through the origin) shows a significant effect of nucleus number (P = 0.0006), ploidy (P = 0.0002), but not genome size (P = 0.98); of the higher-order interactions only nucleus number × ploidy was significant (P = 0.004). These analyses concur on the importance of nucleus number and ploidy, but not on that of genome size. Another way of assessing the joint importance of nucleus number and ploidy is illustrated in Fig. 4, which shows that body size is strongly correlated with their product (across species, r = 0.79; P = 0.002; across contrasts P = 0.0001). We conclude that body size evolution in nematodes is associated either with alterations in lateral seam cell lineages so as to give various numbers of hypodermal nuclei or else with variation of the ploidy of existing nuclei by varying the number of rounds of endoreduplication, or both.

The partial association of endoreduplication with body size evolution suggests that variation in the extent of acellular syncytial growth may be an important mechanism of body size evolution. This can be demonstrated directly. The growth curve of any worm can be divided into larval and adult phases. Larval growth is due, at least in part, to cellular proliferation and fusion. But, as in C. elegans, hypodermal cell proliferation in all species studied here ceases at maturity. Three lines of evidence point to this. First, as in C. elegans, all nonneuronal lateral hypodermal cells fuse either with the hypodermal syncytium (hyp 7) or else with the seam syncytium formed at the L4 which, in turn, forms the lateral alae present in all species (data not shown). Second, mitotic chromosomal condensations have never been seen in the adult hypodermis of any species; instead all nuclei undergo at least some endoreduplication, suggesting that they have permanently exited the mitotic cell cycle. Finally, the lateral seam cell lineages (V cells) of 4 of the 12 species studied here have been directly lineaged (12), among them Panagrellus redivivus (15), a species that has a high nucleus number and is only a distant relative of C. elegans. In these 4 species, then, cell proliferation has been directly shown to end before the L4/adult molt. Given an absence of cell proliferation in the adult hypodermis, adult growth must be due entirely to the acellular growth of the syncytium. The percentage contribution of adult growth to final body size is shown in Table 1. Wild-type C. elegans, for example, has a maximum body size of 0.005 mm³, of which 42.8% occurs during embryogenesis and the four larval stages, whereas the remaining 57.2% occurs between maturity and death. In other species, the contribution of adult growth to final body size varies from less than 40% in the small species Acrobeloides nanus to nearly 90% in the large Pellioditis sp. (EM434). In general, the larger the species, the greater the contribution of adult growth to final body size. Regression shows that variation in adult growth explains nearly all of the variation among these species (86% across species; 85% contrast, not log transformed), larval growth explaining the remainder. Body size evolution in these nematodes occurs, then, mainly by altering the extent of adult, acellular, growth.

We examined three dwarfism mutants in C. elegans to determine the cellular basis of their small body size (Fig. 5). Two of these mutants, daf-4 and sma-2, encode proteins implicated in TGF-β signal transduction, respectively a type II serine/threonine kinase receptor and a Smad (25, 26); dpy-2, encodes a cuticle collagen (27). The sizes of adult daf-4 and sma-2 worms respectively are 27% and 13% of wild-type by volume, whereas dpy-2 is 40% of wild-type (Table 2). The mutants also differ in proportion, daf-4 and sma-2 being both shorter and thinner than wild-type, dpy-2 being mainly shorter. Hypodermal nuclear counts of the mutants showed little or no difference relative to wild-type (Table 2). This observation implies wild-type lateral seam cell lineages and suggests that the severe dwarfism of these mutants must be caused by a deficiency in cell or syncytial growth. We therefore examined hypodermal ploidy in these mutants. dpy-2 had hypodermal nuclei with a final wild-type ploidy of 9.5. daf-4 and sma-2, on the other hand, had mean final ploidies of 5.8 and 7.0. We conclude that dwarfism caused by disruption of the TGF-β pathway in C. elegans is, at least in part, associated with a deficiency of endoreduplication in the hypodermal nuclei, whereas dwarfism caused by a disruption of cuticle structure is not.