Morphology and regeneration of Parhyale legs

Illustration of the 4 distal-most podomeres of an intact Parhyale T4 or T5 leg, including the merus, carpus and propodus (top), and of a leg amputated at the distal part of the carpus (bottom). In the amputated leg stump, the site where regeneration takes place is illustrated with a cartoon of the regenerating leg. Our live regeneration experiments focus on that region.

Live imaging capturing the phases of leg regeneration in Parhyale

Phases of leg regeneration observed by live imaging of nuclei labelled with H2B-mRFPruby, captured from Video 1 (dataset li48-t5). Proximal parts of the leg are to the left and the amputation site is on the right of each panel. (A) T5 leg imaged shortly after amputation, showing hemocytes adhering to the wound site (on the right of dashed line). (B) At 16 hpa, hemocytes have produced a melanized scab at the wound; epithelial cells are migrating below the scab. (C) At 32 hpa, the leg tissues have become detached from the scab (open arrow). (D) At 45 hpa, the new carpus-propodus boundary becomes visible (white arrow); dividing cells can be seen at the distal part of the leg stump (mitotic figures marked by circles). (E) At 57 hpa, the propodus-dactylus boundary becomes visible (black arrow); the carpus and propodus are well separated (white arrows). (F,G) In later stages, tissues in more proximal parts of the leg (to the left of the white arrows) retract, making space for the growing regenerating leg. hpa: hours post amputation. Scale bars, 20 µm.

Live imaging of Parhyale leg regeneration, documented across 22 time lapse recordings

Overview of 22 live image recordings of Parhyale T4 and T5 leg regeneration. For each recording, we indicate the span of the early phase of wound closure and proliferative quiescence prior to epithelial detachment from the scab (in blue), the early phases of cell proliferation and morphogenesis, up to the time when the first and second podomere boundaries becomes visible (in yellow and orange, respectively), and the late phases of regeneration, when the leg takes its final shape, cell proliferation gradually dies down and cells differentiate (in red). The total duration of this process varies from 3 to 10 days. Note that the fastest instances of regeneration occurred in young individuals (li51 and li52); the fastest, li51 was imaged at 29°C (marked by an asterisk). Datasets li-13 and li-16 were recorded until the molt; the other recordings were stopped before molting. Further details on each recording are given in Suppl. Data 1.

Temporal pattern of cell divisions in regenerating Parhyale legs

The number of cell divisions detected per time point are shown for 5 recordings of regenerating legs. The divisions were extracted from tracking data (for li13-t4) or detected using a semi-automated approach (see Methods). The detection of divisions is not exhaustive.

Evaluating image acquisition and post-processing based on cell tracking performance (A)

Elephant’s tracking performance was evaluated on the same image dataset rendered at different resolutions and image quality (datasets #1 to #5), as described in the Methods. To measure linking performance independently of detection, the linking metrics were determined for tracking performed on the spots of the ground truth data (’TRA on GT’). Dataset #1 corresponds to our standard image acquisition settings; improvements in z and t are indicated in green; lower xy resolution or numbers of averaged replicates are indicated in red. The best scores for each metric are highlighted in yellow. To account for variation in the training, three detection and three linking models were trained and evaluated independently on each dataset. The table shows the averaged scores of the three iterations. (B) Elephant’s tracking performance was evaluates on dataset #1 with and without post-processing by denoising or deconvolution (see Methods). The metrics are the same as above. Image denoising did not significantly improve cell tracking and deconvolution (under the setting tested) gave worse results.

Identification of in situ stained cells in live image recordings (A,A’)

Regenerated T5 leg at 189 hpa, in the last frame of a live recording (A) and after immunostaining with antibodies for Prospero and acetylated alpha tubulin (A’). (B,B’) Regenerated T5 leg imaged at 97 hpa, in the last frame of a live recording (B) and after HCR with probes for orthologue of spineless (mostly nuclear dots corresponding to nascent transcripts, B’). The same cells can be identified in the live recordings and in immunofluorescence or HCR stainings. Examples are highlighted by coloured circles. Note that not all corresponding cells are visible in these optical sections, due to slight differences in mounting and tissue distortion.

Tracking the progenitors of spineless-expressing cells in the distal carpus

Snapshots of live recording of a regenerating Parhyale T5 leg at 0, 18, 60, 76, 85 and 97 hours post amputation (left) and corresponding illustrations of the same legs (right). Coloured circles highlight cell lineages that contribute to spineless-expressing cells in the distal part of the carpus; each colour highlights the lineage of a distinct progenitor cell. In the bottom-right panel, spineless-expressing nuclei (identified by HCR) are marked by filled circles, whereas spineless-non-expressing nuclei derived from the same progenitors are marked by open circles. The images show single optical sections as they appear in the Mastodon user interface; nuclei that are only partly captured in the current optical section appear as smaller circles. Distal parts of the leg oriented towards the right.

Testing the effects of scanning speed and averaging on image quality

Duplicate images were captured over a range of settings for scan speed (0.52 to 4.12 ms per pixel) and averaging (averaging 1 to 8 scans). Signal-to-noise and contrast ratios (SNR and CR, respectively) were determined as described in Ulman et al. 2017, using the Cell Tracking Challenge fiji plug-in. (A) Images captured with different settings, indicating the SNR and CR measured in each case. (B) Plots showing the effects of scan speed and averaging on SNR and CR. The points highlighted in grey correspond to settings that result in the same amount (duration) of light exposure per pixel.

Variation in the duration of early and late phases of regeneration in relation to the overall speed of leg regeneration

For each of the 22 live recordings presented in Figure 3 and Suppl. Data 1, we defined the early phase as the period prior to epithelial detachment from the scab (wound closure phase, depicted in blue in Figure 3), and the late phase as the period between epithelial detachment and the appearance of both podomere boundaries (morphogenesis phase, depicted in yellow and orange in Figure 3). The later phase, leading to differentiation (red in Figure 3), was not included in this analysis, as the end of that phase could not be defined by an objective landmark. (A,B) Graphs showing the duration of the early and late phases of leg regeneration in relation to the overall duration of early and late phases. (A’,B’) Graphs showing the relative duration of the early and late phases of leg regeneration in relation to the overall duration of early and late phases. (C) Overview of the relative duration of different phases of leg regeneration, colour coded as in Figure 3.

Apoptosis in legs that have not been subjected to live imaging

(A) Quantification of apoptotic nuclei in the carpus in T4 or T5 limbs, 3 days post amputation. The legs were amputated at the distal part of the carpus, fixed 3 days post amputation, stained with DAPI and imaged on a confocal microscope. The number of apoptotic nuclei was counted within a 28 µm deep stack from the surface of the carpus. (B) Optical section through a carpus at 3 days post amputation. Nuclei that are undergoing apoptosis are highlighted in red circles. The data for making this figure are provided in Suppl. Data 7.

Tracking efficiency in relation to imaging depth

Elephant’s performance in detecting nuclei at different depths was assessed in the datasets presented in Table 1A. Detection precision and recall (see Methods) were scored at z intervals of 5 µm. The data for making this figure are provided in Suppl. Data 8.

In situ staining of Parhyale legs by Hybridization Chain Reaction

(A) Leg segment at late embryonic stages (after cuticle deposition) stained with HCR probes for futsch (magenta) and nompA (green); the image also shows weak nuclear staining with DAPI and cuticle autofluorescence (grey). (B) Optical section through an adult leg stained with HCR probes for futsch, marking putative neurons in the interior of the leg (magenta, with weak autofluorescence in the cuticle), and nuclei stained with DAPI (grey). (C) Surface view of adult Parhyale leg stained with HCR probes for futsch (magenta) and nompA (green), and nuclei stained with DAPI (grey). Scale bars, 20 µm.

Spineless expression in regenerating Parhyale leg

(Top) T5 Parhyale leg at 97 hours post amputation (same leg as in Figure 6) stained with HCR probes for spineless (in green) and futsch (in red), and with DAPI (in blue). Nascent transcripts of spineless are visible in a band of cells at the distal end of the carpus (arrowhead), and in other cells in the carpus, propodus and dactylus. Autofluorescence is visible in the cuticle (in green) and in granular cells (in both red and green channels, appearing yellow). No futsch-stained neurons are visible in this optical splice. Scale bar, 20 µm. (Bottom) Topology of the lineage trees generating the spineless-positive cells tracked in Figure 6, and number of progenitors tracked with each tree topology.