Nematode strains and husbandry

The C. briggsae wild isolate strains AF16 and HK104 were obtained from the Caenorhabditis Genetics Center. ZZY00020, which is AF16 containing a GFP transgene [myo-2::GFP], was obtained from Z. Zhao [29]. Populations were maintained in 20° incubators on nematode growth medium (NGM) agar plates seeded with cultures of the Escherichia coli strain OP50 according to standard practice [38]. A digital peristaltic pump was used to dispense agar medium, and a digital dispensing pipettor was used to dispense volumes of OP50 culture onto plates. These measures were taken to ensure that environmental conditions (agar thickness; food availability) were as consistent as possible from plate to plate.

Hybrid backcrosses

P0 generation AF16 hermaphrodites were self-sperm depleted by moving them onto new NGM plates daily until no self-offspring were produced in a 24 hour period, rendering a hermaphrodite functionally self-sterile (“pseudofemale.”) Each pseudofemale was mated to 3–5 HK104 males. After 24 h, if mating occurred, then the P0 pseudofemale was producing F1 offspring. All of the P0 males and the P0 pseudofemale were then removed from the plate. At approximately 48 h, L4 stage F1 hermaphrodites (virgins) were isolated on separate plates, self-sperm depleted, and then backcrossed to new HK104 males to produce BC1 offspring.

Developmental profiling and micrography

Synchronized F2 individuals were collected by simultaneous embryo production from multiple adult AF16-HK104 F1 hybrid hermaphrodites, multiple adult AF16 hermaphrodites (control), and multiple adult HK104 hermaphrodites (control). Each of the three populations was housed on its own NGM agar plate for two hours at the same time, during which the adults laid fertilized embryos onto the agar substrate. At the end of two hours, all of the adults were removed. Embryos present on each of the NGM plates were thus the same developmental age (within two hours). Embryos were singled onto new NGM plates and micrographed at 24, 36, 48, 52, 58, and 72 hours post-laying (hpl). To avoid potential artifacts (e.g. change in nematode length, harming the nematode) that could be introduced by anesthetizing and mounting worms on glass slides repeatedly during their development, we micrographed each nematode on its NGM agar plate at room temperature using a Zeiss AxioCam MRm digital camera attached to a Zeiss Discovery.v8 dissecting stereoscope at 8x magnification.

The micrographs were manually analyzed with ImageJ [39] using the segmented line tool to measure the anterior-posterior length (in arbitrary units) of each individual at each timepoint. Arbitrary units were then converted to metric lengths by using ImageJ to measure distances on a micrograph of a hemocytometer grid. We then used these data to calculate the micrometer per arbitrary unit ratio, and we multiplied this value by the arbitrary unit length from each micrograph. Using nematode total length to assess development regardless of developmental stage, which is the standard metric for reporting development, is intended to circumvent concerns about qualitative assessment of nematode developmental stage [40].

Developmental delay phenotyping

AF16-HK104 F2 hybrid developmental delay has been described as presence of L2/L3 larvae at 48 hours, while wild-type siblings are L4 larvae at that point [37]. Here, we described a worm as delayed if at 58 hpl it remained an L4 larva, based on visual identification of L4.3–L4.7 stage vulval morphology [41].

Genotyping

Each individual nematode was isolated in a PCR tube containing 38 μL of single-worm lysis buffer (50 mM KCl, 10 mM TRIS pH 8.3, 2.5 mM MgCl2, 0.45% nonidet P-40, 0.45% Tween-20, 0.01% gelatin) and 2 μL of 20 mg/mL proteinase K. Each worm was then frozen at -80° and digested for 60 m at 60°. The proteinase K was inactivated by incubation at 95° for 15 m, and then the single worm lysate was immediately used as PCR template.

To genotype nuclear alleles, an amplified fragment length polymorphism (AFLP) analysis was used. Nuclear genome insertion-deletion (indel) markers that distinguish AF16 and HK104 alleles were amplified by polymerase chain reaction (PCR) using published primer sequences [22]. Each 10 μL reaction contained 0.25 U OneTaq DNA polymerase (New England Biolabs), 1x reaction buffer, 0.2 mM each dNTP, either 5 ng of purified DNA template (at 5 ng/μL) or 1 μL of single worm lysate or 1 μL of water, and 1 μM each of the forward and reverse primers. All amplifications occurred in a Bio-Rad T100 thermal cycler using the following cycle sequence: 95° 2 min, 30 cycles of 95° 1 min, 45° 1 min, 68° 2 min, followed by a final 68° 5 min extension.

All amplicons were separated by electrophoresis in type III loading buffer modified by omitting xylene cyanol [42] on 2% agarose gels containing ethidium bromide. Electrophoresis was perfomed in 1x TRIS-acetate EDTA buffer. Promega 100 bp ladder (G2101) was loaded as a molecular weight standard. Digital micrographs of gels were obtained using an AlphaImager HP system (Alpha Innotech).

Developmental profiles of F2 hybrids

The presence of delay in F2 hybrids suggests that there might be a discrete point in time at which typical development either pauses or begins to lag. To characterize the temporal process of developmental delay, 66 F2 hybrid siblings, 22 AF16 and 22 HK104 individuals were micrographed across larval development. Fig 1 shows micrographs of eight example worms across development. S1 Table contains the length of each of the 110 individuals at each timepoint, and all of the original image files are available at 10.6084/m9.figshare.20085863.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Micrographs of wild-type and delayed individuals.

Micrographs of nematodes on NGM agar plates were acquired at the number of hours post-laying (hpl) listed above. Three example wild-type F2 and three developmentally delayed F2 are shown, along with one AF16 and one HK104 individual. Scale bar in the bottom-left panel: 0.5 mm.

https://doi.org/10.1371/journal.pone.0272843.g001

In general, wild-type F2s and wild isolates became L4 stage larvae between 36 and 48 hours post-laying (hpl), were still L4 larvae at 52 hpl, and developed into self-fertile adults by 58 hpl. In contrast, delayed F2s did not reach the L4 larval stage until between 48 and 52 hpl, were still L4 larvae at 58 hpl, and became self-fertile by 72 hpl.

The lengths of the P0 parental, wild-type F2 and delayed F2 populations are plotted in Fig 2. The AF16 and HK104 populations were significantly larger than the delayed F2 population at every timepoint (Table 1), and the wild-type F2 population was significantly larger than their delayed F2 sibling population at all timepoints except 24 hpl.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Development of F2 hybrids.

Nematode total length is plotted against the number of hours elapsed after each embryo was deposited by a hermaphrodite onto an NGM agar plate. The average length of delayed F2 nematodes is plotted as an X; the average length of wild-type F2 siblings is plotted as a dash (horizontal line). The average length of the combination of the AF16 and HK104 parental isolates is plotted as a circle. Vertical lines depict one standard deviation. Alternating background shading is used to distinguish data collected at 24, 36, 48, 52, 58, and 72 hours post laying, at which point every individual had reached adulthood. The data series at each timepoint have been jittered on the x-axis for clarity. Statistical analyses were conducted to identify significant differences between the parental and F2 wild-type populations, the parental and F2 delayed populations, and the F2 wild-type and delayed populations. Brackets indicate statistically different comparisons (Bonferroni-corrected Student’s T-tests, alpha 0.05; see Table 1 for p-values).

https://doi.org/10.1371/journal.pone.0272843.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Statistical comparisons of F2 hybrid lengths.

https://doi.org/10.1371/journal.pone.0272843.t001

Producing developmental delay with F1 backcrossing

If homozygosity for AF16 alleles on chromosome III is required to produce developmental delay, then backcrossing an F1 hybrid to HK104 should not produce any delayed offspring. Results from this cross have never been reported; we found that zero of 34 BC1 hybrids exhibited developmental delay, as expected. This finding supports the conclusion that homozygosity for AF16 is necessary to produce developmental delay, as previously suggested, and is consistent with presence of an HK104 toxin-antidote element [12, 19, 37].

F2 mapping

To collect F2 individuals for mapping the chromosome III interval associated with delay, several AF16-HK104 F1 hybrid hermaphrodites were selfed to produce 2,622 offspring, of which 515 were developmentally delayed (19.6%). 176 of these delayed F2 individuals were genotyped at one of eight AFLP loci on chromosome III. We used Bonferroni-corrected chi-square analysis to compare these genotypes with the Mendelian expectation in the F2 generation of a 1:2:1 genotype ratio (AF16 homozygous: heterozygous: HK104 homozygous). Delayed individuals exhibited significant deviation toward homozygosity for AF16 alleles at loci in the middle of chromosome III (Fig 3), with a maximum deviation of 88% homozygosity for AF16 at marker cb-m205. In these delayed F2 individuals, six markers revealed significant bias toward homozygous AF16 genotypes. The two markers with non-significant p-values are located near the chromosome III right telomere. These data confirm the presence of a delay locus on III around 3.572–5.649 Mbp.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. F2 genotypes across chromosome III.

The percent of F2 genotypes homozygous AF16 at each marker in developmentally delayed F2 hybrids is plotted on the y-axis (“% AF16/AF16”). The total number of typed individuals (n) at each locus is provided above the plot, along with the p value (chi-square, Bonferroni-corrected) of the difference between expected and observed genotype values. “ns” indicates not significant p values. The gray line at 25% AF16 homozygosity denotes the Mendelian expectation for AF16 homozygosity in F2 hybrids. The gray shaded box indicates the two markers with maximum allele fraction deviation.

https://doi.org/10.1371/journal.pone.0272843.g003

Evaluating the accuracy of delay frequency observations

To evaluate the possibility that some F1 offspring of self-sperm depleted hermaphrodites crossed to males might be self-offspring, and thus influence the observed frequency of F2 developmental delay, we self-sperm depleted four HK104 hermaphrodites using the traditional approach (Materials and Methods). We then crossed each with ZZY00020 males, which are AF16 with a GFP transgene. In total, the four hermaphrodites produced 114 offspring after mating. To evaluate the provenance of each offspring, fluorescence microscopy was used to phenotype the presence of GFP. All 114 offspring were fluorescent, indicating that none of the 114 offspring were self-progeny of the HK104 parental hermaphrodite. Thus, even if the process of self-sperm depletion happens not to completely exhaust a hermaphrodite of self-sperm, the number of remaining self-sperm would not be large enough to substantially influence the observed frequency of developmental delay.

Developmental regulation in Caenorhabditis

Developmental delay is a widely reported phenomenon in hybrids [e.g. 43–48]. In this study, we have continued to refine our understanding of the phenotype and the genetic architecture of hybrid developmental delay in C. briggsae [12, 19, 36, 37]. Although no genetic variants segregating in nature have yet been suggested to cause postembryonic hybrid developmental incompatibilities in the close relative C. elegans, the process of postembryonic development has been more extensively studied there. In early studies, the lengths of nematodes were measured and correlated with larval molts. This demonstrated that wild-type growth does not involve severe pauses, including during molting [49]. Later studies showed that Caenorhabditis larval development is controlled by heterochronic genes [50]. Mutations in these genes can cause accelerated or delayed development [51]. Because developmental delay is a common effect of mutations in Caenorhabditis, future efforts to identify the genetic basis of AF16-HK104 hybrid developmental delay should benefit from the temporal description of developmental delay that we provide here. We have found that developmental delay is evident across much of larval development, with a significant decrease in total length between 36 and 72 hours post laying compared to wild-type siblings. Even delayed hybrids routinely become self-fertile adults and are no longer casually distinguishable by eye from their wild-type F2 siblings. We find that delay is easiest to detect by eye at 58 hpl, when wild-type F2 are adults and their delayed F2 siblings are still predominantly L4 larvae. Although we chose to use worm total length as a proxy for development [40], this is not the same approach as identifying developmental stages. Some genetic factors, and thus potentially some incompatibilities, can alter body size irrespective of developmental stage [52, 53]. Thus, it will be valuable in future to compare both size- and developmental stage-based phenotyping.

Genetic architecture of delay

Because the AF16 and HK104 wild isolates do not exhibit the developmental delay phenotype [37], delay might result from hybrid dysfunction involving a negative epistatic interaction between divergent alleles at two or more loci. Absence of delay in AF16-HK104 F1 hybrids indicates that this phenotype does not result from a dominant epistatic interaction between AF16 and HK104 alleles [37]. The presence of a maternal effect hybrid incompatibility [37] raised the possibility that a toxin-antidote (TA) system causes developmental delay. Such selfish genetic elements have been documented in Caenorhabditis, including the zeel-1/peel-1 system in C. elegans [9, 10] and recently in C. tropicalis and in C. briggsae, where this developmental delay phenotype has been reported to be caused by a TA element present in HK104 [12].

Here, we have interrogated the genetic architecture of F2 hybrid developmental delay, which involves homozygosity for AF16 alleles at a locus on chromosome III [12, 19, 37]. The broad chromosome III interval has never been systematically mapped. Initial mapping efforts genotyped two loci on chromosome III, at 10.1 and 12.2 Mbp [19]. In the same study, patterns of marker transmission ratio distortion suggested that the delay locus is in the central recombination domain of III, which spans 6.1 Mbp (from 4.7–10.8 Mbp) on the chromosome assembly and thus comprises almost half of the chromosome [19]. The most recently published mapping effort, which confirmed linkage of the delay locus on III, genotyped a single marker at 4.0 Mbp [12]. Thus, the current understanding is that chromosome III harbors an HK104 maternal effect TA element that causes developmental delay in offspring homozygous AF16 at the TA locus.

Using eight genetic markers spanning III, we find here that the region of strongest association exists between 3.572 and 5.649 Mbp (Fig 3, gray box), with the caveat that an association peak could exist beyond the boundaries of this interval, adjacent to the flanking markers. At the markers defining the minimal interval, 88% of genotypes from delayed individuals were AF16/AF16. At both markers, no delayed worm contained an HK104/HK104 genotype. Thus, developmental delay is often, but not always, associated with AF16 homozygosity at these two chromosome III loci. This interval also contains the PCR genotyping assay at 4.0 Mbp that was previously used to map delay, at which 1 of 41 delayed individuals did not contain a homozygous AF16 genotype [12]. Genetic mapping is notoriously difficult in the central recombination domains in C. elegans and C. briggsae, where crossing-over rarely occurs [19, 20, 54]. Thus, absence of a perfect genotype-phenotype correlation could be interpreted in at least two not mutually exclusive ways. First, we might not have genotyped a marker perfectly associated with the delay phenotype. Also, it is conceivable that multiple genetic interactions between AF16 and HK104 alleles could result in similar developmental delay phenotypes.

The possibility that hybrid developmental delay exhibits incomplete penetrance should also be considered. Mendelian principles suggest that a TA element present on HK104 chromosome III should produce 25% delay when an AF16-HK104 F1 hermaphrodite self-fertilizes, because all the embryos will contain the HK104 allele-derived toxin, and 25% of the F2 offspring should be homozygous AF16 on III and thus not produce any antidote. For the C. briggsae developmental delay phenotype, lower delay frequencies of around 20% have been reported by two other independent research groups [12, 37], and the cause for this lower frequency is not known. We report here the largest dataset measuring the developmental delay frequency produced by selfing AF16-HK104 F1 hybrids: our observation of 19.6% delay among 2,622 F2 individuals is statistically significantly different from 25% expected for a completely penetrant TA element (chi-square, p < 3E-10).

Potential causes of incomplete penetrance include the presence of modifier loci and environmental effects [55]. It is conceivable that an unknown modifier locus attenuates the effect of the chromosome III TA genotype. For example, if a dominant HK104 allele necessary for producing developmental delay exists on another autosome, then when 25% of self-offspring from an AF16-HK104 F1 are homozygous AF16 on III, 75% of those F2 (19.75% of all F2) would be homozygous HK104 or heterozygous at the unlinked modifier locus. It would be valuable to conduct genome-wide mapping in F2 populations to identify potential modifier loci. With respect to potential environmental influences, future efforts might explore whether hybrid developmental delay is temperature-sensitive, as has been suggested for another TA element in Caenorhabditis that differs in expressivity depending on temperature [12] and also in Tribolium castaneum, where both genetic background and temperature influence the time of hybrid death caused by a Medea element [14]. Reduced penetrance of other incompatibility factors, such as Wolbachia bacteria, has also been observed in Drosophila, possibly as a result of ongoing genetic co-evolution [56–58].

An experimental artifact could also potentially account for some deviation of the observed delay frequency from the expected frequency for a TA element. Any residual self-sperm in the pseudofemales used for crosses would reduce the observed delay frequency [12]. Our experimental results that zero of 114 progeny of a cross between ZZY00020 males and HK104 self-sperm depleted hermaphrodites were fluorescent suggest that even if rare self-sperm are retained by functionally self-sterile hermaphrodites, not enough self-sperm is present to cause the magnitude of effect necessary to bias the experimental measurement of developmental delay frequency. Also, in every case where we used self-sperm depleted P0 hermaphrodites to generate AF16-HK104 F1s, each of those selfed F1 hermaphrodites always produced some delayed F2 offspring. Thus, none of the P0 pseudofemales we used produced unexpected F1 self-offspring, because F1 self-offspring would not have generated F2 hybrid developmental delay.