SBF-SEM allows classification of axenic promastigotes into seven characteristic cell cycle stages

Promastigote form Leishmania mexicana cells from an asynchronous exponential phase in vitro culture were imaged by SBF-SEM, generating a dataset containing 750 individual slices (Z-slice = 100 nm) (Movie 1). We analysed a total of 41 promastigote cells whose cell body and flagella lay within the imaged dataset, representing seven different stages of the cell cycle (see below). The cell body membrane, flagellum, nucleus, mitochondrion with kinetoplast, Golgi, contractile vacuole, paraflagellar rod, lipid bodies, acidocalcisomes and glycosomes were segmented for each individual whole cell (Fig 1A and 1B, S1 and S2 Movies). Organelles were identified as per previous descriptions in transmission electron microscopy [14,19,20]. The volume and surface area was calculated from the segmentation of each cellular structure and organelle and counts were made of multi-copy organelles, shown below and in S1 Table. The endoplasmic reticulum (ER) was segmented in each slice of the dataset to illustrate positioning within each cell, but due to its undulating nature these fragments are not connected across sections to produce a 3D model due to the limitations of Z-resolution. The sub-pellicular microtubules were also not modelled as they are separated by only 20 nm thus were only visible when orientated parallel to the Z axis.

Download:

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

Fig 1. Distinct ultrastructure features were used to define seven cell cycle stages.

(A) 3D reconstructed Stage 1 promastigote cell showing the spatial organisation of organelles within the cell body. Organelles are modelled using the following colours: cell body (opaque green), flagellum (pink), nucleus (blue), kinetoplast (blue), mitochondrion (yellow), acidocalcisomes (dark purple), lipid bodies (red), glycosomes (pale green), contractile vacuole (dark blue), Golgi (purple), ER (pink), PFR (teal), flagellar pocket (deep red). Scale bar = 5 μm. (B) SBF-SEM data slice of a Stage 1 promastigote cell highlighting the ultrastructural features used to identify organelles. Scale bar = 2 μm. (C-D) 3D reconstruction and SBF-SEM data slice of seven distinct cell cycle stages representing the entire promastigote cell cycle. Organelles are modelled using the following colours: cell body (opaque green), old flagellum (pink), new flagellum (orange), nucleus (blue), kinetoplast (blue). (Stage 1) Cell with a single flagellum, kinetoplast and nucleus. (Stage 2) New flagellum (white arrow) wholly within the flagellar pocket. (Stage 3) Cell with a new flagellum that has emerged from the anterior cell tip (white arrow). (Stage 4) Cell with an elongated (dividing) kinetoplast and a single nucleus. (Stage 5) Cell with 2 kinetoplasts and a dividing nucleus. (Stage 6) Cell with two separate kinetoplasts and nuclei, with a division fold forming. (Stage 7) Cells with cytokinesis furrow ingression progressing from the anterior end of the cell (white arrow). Scale bar = 2 μm. (E) The approximate timings using ergodic principles of the seven cell cycle stages during the cell cycle. n = 500. (F) Dot plot showing cell body length across the seven cell cycle stages defined. Stage 7 was split into two categories: nascent daughter cell inheriting the old flagellum (square data points) and nascent daughter cell inheriting the new flagellum (triangle data points). Dot plot showing (G) cell body volume, (H) cell body surface area, (I) old flagellum length and (J) new flagellum length across the seven cell cycle stages defined. Error bars show ±standard deviation. Total number of cells per stage: 6 cells in Stage 1, 9 cells in Stage 2, 4 cells in Stage 3, 6 cells in Stage 4, 6 cells in Stage 5, 6 cells in Stage 6 and 4 cells in Stage 7.

https://doi.org/10.1371/journal.ppat.1012054.g001

Using the asynchronous exponential phase culture means the cells are randomly sampled from all stages of the cell cycle [11,21]. We defined seven distinct cell cycle stages by identifying the number of kinetoplasts and nuclei and the length of the new flagellum, which will enable us to examine changes in both cell and organelle morphogenesis through the cell cycle (Fig 1C and 1D). Stage 1–3 have a single kinetoplast and nucleus and are distinguished by the flagella: Stage 1) Single flagellum (S1 Movie); Stage 2) Two flagella, with the new flagellum (NF) restricted to the flagellar pocket (Fig 1C and 1D; arrow—NF); Stage 3) New flagellum that has emerged from the flagellar pocket (Fig 1C and 1D; arrow—NF); Stage 4–7 are distinguished by kinetoplast, nucleus and cell body division stage: Stage 4) Single elongated (dividing) kinetoplast and a single nucleus (Fig 1C and 1D; arrow–kinetoplast); Stage 5) Two kinetoplasts and a dividing nucleus (Fig 1C and 1D; arrow–nucleus); Stage 6) Two kinetoplasts and two nuclei, with a cell body division fold (Movie 2); Stage 7) Cell body cytokinesis furrow from the anterior end of the cell towards the posterior end (Fig 1C and 1D; arrow—furrow).

For a population in exponential growth it is possible to calculate time spent in each cell cycle stage, assuming that all cells are undergoing the same replicative cycle (Fig 1E) [21]. Here, we represent timings as proportion of time (in units, u) through the ~8 h cell cycle [4]. The new flagellum begins to grow within the flagellar pocket in Stage 2 at 0.3 u and emerges from the flagellar pocket in Stage 3 at 0.53 u. The kinetoplast shows a more elongated form, indicative of the earliest stages of reorganisation for division (Stage 4), at 0.76 u followed by a late mitotic nucleus (Stage 5) characterised by two distinct nuclear lobes linked by a cytoplasmic mitotic nuclear bridge at 0.8 u with segregation of the kinetoplast and nucleus (Stage 6) occurring at 0.81 u.

One of the most notable morphological features of Leishmania is the change in cell body length during the cell cycle (Fig 1F). The overall change in cell length agrees with our previous light microscopy analysis [4]. Despite a reduction in cell body length, the cell volume increased from Stages 1–5 and then remained relatively constant through Stages 6–7 (Fig 1G). The universal presence of sub-pellicular microtubules at even spacing means cell surface area is proportional to the total amount of microtubules/tubulin needed for the sub-pellicular array. The cell surface area increased steadily across the cell cycle and therefore so did the amount of tubulin in the sub-pellicular array (Fig 1H). This shows that the cell volume and surface area increases whilst the cell body length decreases as the cell undergoes large cellular rearrangements in preparation for cytokinesis.

New flagellum does not rotate around the old flagellum

In L. mexicana promastigotes the single flagellum remains assembled throughout the cell cycle with the new flagellum growing alongside, as is typical of trypanosomatid parasites [4,7]. Consistent with previous light microscopy [4], the old flagellum length was variable but was always longer than the new flagellum (Fig 1I). By the end of the cell cycle, the new flagellum length (Fig 1J) did not reach that of the old, with flagellum growth therefore continuing after cytokinesis.

In T. brucei trypomastigotes the new flagellum extends into the existing flagellar pocket on the anterior side of the existing flagellum. The basal body and new flagellum then rotate around the old flagellum to position the new flagellum towards the posterior of the cell. This rotation occurs prior to the new flagellum extending out of the flagellar pocket and flagellar pocket duplication [22]. Here, we investigated if this rotation occurs during new flagellum growth in L. mexicana. The axoneme, paraflagellar rod and flagellar pocket were segmented in cells (Fig 2A). The paraflagellar rod is always positioned alongside microtubule doublets 4–7 of the axoneme [23] and the paraflagellar rod is present on axonemes within the flagellar pocket neck of L. mexicana [6]. Therefore, we used this feature as a positional marker. The new flagellum was always (n = 25) positioned next to the side of the old flagellum with the PFR (Stage 2—Fig 2B; Stage 3 –Fig 2C) indicating that new flagellum rotation does not occur in Leishmania and suggests that this phenomenon is restricted to parasites with a trypomastigote morphology [7,22]. This raised a further question regarding the timing of flagellar pocket duplication relative to the elongation of the new flagellum. The new flagellum extended into the existing flagellar pocket bulbous domain in Stage 2 (Fig 2B). In Stage 3, the new flagellum exited the existing flagellar pocket, but flagellar pocket duplication had not occurred (Fig 2C). The first morphological indicator of flagellar pocket duplication was seen in Stage 4 cells, with a ridge of cytoplasm invading the bulbous domain of the existing flagellar pocket between the old and new flagellum (Fig 2D; arrow). This cytoplasmic ridge has previously been observed in T. brucei flagellar pocket duplication [22]. Prior to the ridge forming, the old and the new flagellum share the same flagellar pocket exit point (Fig 2E; Stage 2–3). Whilst the ridge is forming from the base of the flagellar pocket, the exit point of the flagellar pocket was still shared by the two flagella (Fig 2E; Stage 4). Two separate exit points for each flagellum were first observed in Stage 5 cells, indicating completion of flagellar pocket duplication by this cell cycle stage (Fig 2E; Stage 5). This sharing and separation between the new and old flagellum exit points can also be seen by whole mount negative staining of the cytoskeleton of dividing cells (Fig 2F and 2G). This shows that flagellar pocket duplication occurs in two stages with the division of the bulbous domain by the cytoplasmic ridge followed by the division of the flagellar pocket neck (Fig 2D) [24].

Download:

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

Fig 2. New flagellum rotation around the old flagellum does not occur.

(A-C) SBF-SEM data slice (scale bar = 1 μm) and two 3D reconstructed models (scale bar = 2 μm) of a Stage 1, 2 and 3 cell illustrating that there is no rotation of the new flagellum (orange) around the old flagellum (opaque pink) during new flagellum growth within and extending out of the flagellar pocket (FP—opaque red) The old flagellum paraflagellar rod (PFR–dark blue) is used as an orientation marker demonstrating the new flagellum is always on the right-hand side of the old flagellum PFR. Mature basal body (MBB), pro basal body (PBB), kinetoplast (K–blue). (D) SBF-SEM data slice showing a cytoplasmic ridge (arrow) within an existing flagellar pocket. Scale bar = 1 μm. (E) Longitudinal SBF-SEM data slice of Stage 2–5 cells showing the flagellar pocket exit (white dashed line). Cross sections of 3D reconstructed Stage 2–5 cell illustrating that the old flagellum (pink) and new flagellum (orange) share the same flagellar pocket (FP) exit (white dashed line) until Stage 5 post cytoplasmic ridge has formed and two separate flagellar pocket exits are present indicating flagellar pocket duplication is complete. Stage 4 shows the positioning of the cytoplasmic ridge (arrow) within the existing flagellar pocket, but the two flagella still share the same exit point. Scale bar = 1 μm. (F) Whole cell negative stain mounts of a cell with two flagella sharing the same flagellar pocket exit point (arrow). Scale bar = 10 μm. (G) Whole cell negative stain mounts of a cell with two flagella with two separate flagellar pocket exit points (arrow). Scale bar = 1 μm. (H) Transmission electron image of a flagellum with a cap at the distal tip (arrowhead). Scale bar = 500 nm. (I) Serial electron tomogram of the L. mexicana flagellar tip in longitudinal and cross-section. There is one annulus that sits within the outer doublets (arrow) and a second annulus that caps the outer doublets (arrowhead). (J) Whole mount detergent extracted L. mexicana cell. Higher magnification micrographs of the flagellum distal tip showing the capping structure (arrowhead). Scale bar = 5 μm. Inset scale bar = 200 nm.

https://doi.org/10.1371/journal.ppat.1012054.g002

The flagellum is assembled at its distal tip, with material transported to the site of assembly by the intraflagellar transport system [25,26]. Regulatory mechanisms controlling, for example axoneme stability, are therefore likely to operate at the distal tip of the flagellum [27]. We observed an elaborate cap structure at the flagellum distal tip that has previously been identified in other trypanosomatid parasites including L. major (Fig 2H–2J) [28]. The cap structure consisted of two annuli, one associated with the central pair that sits inside the outer doublets that are capped by a second annulus. The annulus associated with the central pair is readily observable in whole mount cytoskeletons (Fig 2J). This elaborate cap structure is only present on the old flagellum and is therefore added post-assembly, with a potential role in beat regulation and axoneme stability [27–29].

Contractile vacuole and Golgi re-position during the cell cycle

Our data confirmed there was a single contractile vacuole and Golgi located close to the flagellar pocket in Stage 1 cells (Fig 3A), supporting our previous work [8]. During the cell cycle, one additional contractile vacuole and Golgi were observed. A second contractile vacuole was first visible in Stage 2 and a second Golgi in Stage 3. (Fig 3A). The appearance of the second contractile vacuole along with the new flagellum are therefore the earliest ultrastructural changes we identified. The second contractile vacuole and Golgi initially appeared alongside the existing organelles (Fig 3B). However, during Stage 4, a single contractile vacuole and Golgi were located next to each separate flagellar pocket bulbous domain, such that there was a pairing of contractile vacuole and Golgi to be inherited by each daughter cell (Fig 3A). We were unable to identify which contractile vacuole/Golgi pair re-positioned with the new flagellum in Stages 2–4 because flagella pocket division had not yet occurred. However, there was a clear difference in volume between the old and new flagellum-associated contractile vacuole in Stage 5 (Fig 3C). Overall, this shows that as the cytoplasmic ridge ingresses to segregate the flagellar pocket bulbous domains, each bulbous domain acquires a dedicated contractile vacuole/Golgi pair, indicating that each bulbous domain has the potential to be endo/exocytically active at this point.

Download:

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

Fig 3. Increase in multi-copy organelle volumes are seen in early cell cycle stages.

(A) 3D reconstructed Stage 1–4 cells highlighting the duplication timing and movement of the Golgi (purple) and contractile vacuole (dark blue) as the flagellar pocket duplicates (opaque red). Scale bar = 1 μm. (B) SBF-SEM data slices of the appearance of an additional contractile vacuole (CV) and Golgi next to a CV and Golgi in a Stage 3 cell. Scale bar = 500 nm. (C) Dot plot graph showing contractile vacuole volumes across the seven cell cycle stages defined. Error bars show ±standard deviation; n = 41. (D-F) 3D reconstructed Stage 1 and Stage 6 cell showing the positioning of the lipid bodies (red), glycosomes (green) and acidocalcisomes (purple) within the cell body. White dotted line along the Stage 6 cell indicates where the division fold lies and the separation of the multi-copy organelles away from this fold in the cell membrane. Scale bar = 2 μm. Dot plots showing (G-I) the number of lipid bodies, glycosomes and acidocalcisomes per cell and (J-L) the total volume of lipid bodies, glycosomes and acidocalcisomes per cell across the seven cell cycle stages defined. n = 41. Key for 3C, 3G-L: Whole cells (circle data points), nascent daughter cell inheriting the old flagellum (square data points) and nascent daughter cell inheriting the new flagellum (triangle data points). Error bars show ±standard deviation. Total number of cells per stage: 6 cells in Stage 1, 9 cells in Stage 2, 4 cells in Stage 3, 6 cells in Stage 4, 6 cells in Stage 5, 6 cells in Stage 6 and 4 cells in Stage 7.

https://doi.org/10.1371/journal.ppat.1012054.g003

Multi-copy organelles increase in number and overall volume during the cell cycle

Multi-copy organelles such as lipid bodies, glycosomes and acidocalcisomes also need to duplicate through the cell cycle. In Stage 1 cells there were 6 ± 3.6, lipid bodies, 46 ± 17.3 acidocalcisomes and 37± 8.8 glycosomes (n = 6 cells) distributed throughout the cell body (Fig 3D–3I). Between Stages 1–4 as the cell increases in size there was an increase in these multi-copy organelles (Fig 3G–3I). The number of organelles plateaued around Stage 5 and by Stage 6, there were two nascent daughter cells, and these organelles were positioned away from the division fold segregated into the regions that would become the two daughter cells (Fig 3D–3F; dashed line–division fold) and each organelle was evenly distributed (Fig 3G–3I). The total volume of each multi-copy organelle in each cell type was also calculated and between Stages 1 and 2 decreased slightly before increasing from Stage 2 to 4 (Fig 3J–3L).

The mitotic nuclear bridge is devoid of nuclear pores during closed mitosis

Next, we examined division of the kinetoplast and nucleus (Fig 1C and 1D). The volume of the kinetoplast increased between Stages 1–4, with division into two separate kinetoplast disks completed by Stage 5 (Figs 1C and 4A). Nuclear volume and surface area increased up to Stage 5, after which the two nuclei were the same size (Fig 4B and 4C). Stage 5 cells were characterised by a mitotic nucleus with a fully intact nuclear envelope and two distinct lobes connected by a narrow nuclear envelope-bound bridge (Fig 1C and 1D; arrow) and the late mitotic nucleus was orientated perpendicular to the anterior-posterior axis of the cell (Fig 1C and 1D; Stage 5). Interestingly, in one cell with two nuclei, we observed membrane bound structures that traced a line from one nucleus to another that could represent the disassembly of the nuclear envelope bound bridge (Fig 4D; arrows). The kinetoplast volume and nuclear volume and surface area of Stage 2 cells was larger than that of Stage 1 cells, indicating that both kinetoplast and nucleus S-phase had likely begun in Stage 2 cells, along with new flagellum assembly and second contractile vacuole appearance.

Download:

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

Fig 4. Nuclear pores are excluded from the nuclear bridge during mitosis.

Dot plots showing (A) kinetoplast disk membrane volume, n = 56, (B) nucleus volume and (C) nucleus surface area per cell across the seven cell cycle stages defined. n = 52. (D) SBF-SEM dataset slices and 3D reconstruction of two nuclei (blue) and membrane bound structures (green—arrow). Scale bar = 1 μm. (E) SBF-SEM data slice of a Stage 1 nucleus showing the positioning of nuclear pores (arrows) within the nuclear envelope. Scale bar = 500 nm. (F) 3D reconstructed Stage 1 nucleus showing the distribution of nuclear pores (pink—white arrow) throughout the nuclear envelope. Scale bar = 500 nm. (G) Dot plot showing the number of nuclear pores (NP) per nucleus. n = 50. (H) Dot plot showing the nuclear pore (NP) density per nucleus. n = 50. (I) 3D reconstructed nuclei at different stages of mitosis illustrating the absence of nuclear pores (pink) along the mitotic nuclear bridge (arrow) as the nucleus progresses through mitosis. Scale bar = 500 nm. (J) 3D reconstruction of the ER (pink—white arrow) positioning in a Stage 1 and 5 cell. Scale bar = 2 μm. (K) SBF-SEM data slice showing the ER connected to the nuclear envelope (dashed box) in a Stage 1 cell. Scale bar = 1 μm. (L) SBF-SEM data slice showing the ER close to the sub-pellicular microtubules (dashed boxes). Scale bar = 1 μm. (M-N) Serial section tomograms of the ER (arrow) intercalating between the sub-pellicular microtubules (MTs). Scale bar = 50 nm. (O) 3D reconstructed Stage 1–7 cells showing the complex mitochondrion network throughout the cell cycle (yellow). Scale bar = 1 μm. Dot plots showing (P) the mitochondrion volume and (Q) mitochondrion surface area across the seven cell cycle stages defined. Stage 6 and 7 are split into 3 and 2 categories, respectively: cells with the mitochondrion not yet segregated (circle data points), nascent daughter cell inheriting the old flagellum (square data points) and nascent daughter cell inheriting the new flagellum (triangle data points). n = 50. (R) 3D reconstructed Stage 5 cell showing the branching of the mitochondrion network (yellow). Scale bar = 1 μm. Key for 4A-C, 4F, 4G, 4K: Whole cells (circle data points), nascent daughter cell inheriting the old flagellum (square data points) and nascent daughter cell inheriting the new flagellum (triangle data points). Error bars show ±standard deviation. Total number of cells per stage: 6 cells in Stage 1, 9 cells in Stage 2, 4 cells in Stage 3, 6 cells in Stage 4, 6 cells in Stage 5, 6 cells in Stage 6 and 4 cells in Stage 7.

https://doi.org/10.1371/journal.ppat.1012054.g004

Our data were of sufficient resolution to examine the number of nuclear pores in the nuclear envelope through the cell cycle (Fig 4E; arrows). Nuclear pores were distributed across the nuclear envelope and their number increased across the cell cycle from 27±3.5 in Stage 1 to 41±13.9 in Stage 5 cell (Fig 4F and 4G). However, given the average pore width of ~80 nm (n = 6 Stage 1 cells) and our slice thickness of 100 nm, our nuclear pore counts are likely underestimates of the total number. The average nuclear pore density (NP per nucleus surface area) was not constant throughout the cell cycle (Fig 4H). Nuclear pore density increased from 2.5±0. 6 per μm² in Stage 1 to 3±0.3 per μm² in Stage 2 (n = 15 cells) but then decreased to 1.8±0.62 per μm² during late mitosis in Stage 5 (n = 6 cells). Nuclear pore density then increased to 3.1±0.82 μm² in Stage 6 (n = 6 cells) at which point nuclear pore density was similar between the two daughter nuclei.

Interestingly, the nuclear bridge of the mitotic nucleus was relatively devoid of nuclear pores (Fig 4I). Given the surface area of this bridge region was 1.8 ± 2 μm² (n = 6 Stage 5 cells) and the average number of nuclear pores per membrane surface area was ~2±0.62 per μm² at Stage 5, we would expect to see ~4 pores within the bridge and in the 6 bridges examined we only saw 3 nuclear pores in total (p = 0.01046, Mann-Whitney U test). This suggests there is a potential exclusion mechanism to ensure the nuclear bridge is devoid of nuclear pores or that nuclear pores are quickly disassembled at the centre of the dividing nucleus, just before this region narrows down and becomes a noticeably ‘bridge-like’.

Mitochondrion retains a complex structure throughout the cell cycle

The ER and mitochondrion are large networks that must grow and segregate during the cell cycle. The ER network was more concentrated along the periphery of the cell body and extended along its length, with clear reticulated regions connecting to the nuclear envelope (Fig 4J and 4K). In addition, we observed regions of the ER appearing to contact the cell membrane and/or the sub-pellicular microtubules (Fig 4L). We investigated these ER extensions in more detail using serial section tomography (Fig 4M and 4N). In these tomograms the ER intercalated between the sub-pellicular microtubules over an extended region (n = 7 cells) (Fig 4M and 4N). This tight associated of the ER and the sub-pellicular microtubules may play a role in ensuring ER inheritance during cell division. A similar association between the ER and the microtubules of the microtubule quartet is seen in T. brucei and this association is also likely important for segregating the ER during cell division [30].

The single mitochondrion forms a network of interconnected loops and branches (Fig 4O), with the volume and surface area of the mitochondrion increasing from Stage 1 to 5, peaking at Stage 5 after replication and segregation of the kDNA was complete (Fig 4P and 4Q). Mitochondrial segregation was accompanied by a reorganisation of the network in Stage 5 cells. An enlarged region of the mitochondrion was positioned along the anterior-posterior division fold, with projections of the mitochondrial network extending into the two regions of the cell body that will form the daughter cells (Fig 4R). In our dataset, the mitochondrion was the last organelle to divide with only 50% (n = 6 cells) of Stage 6 cells containing two independent mitochondria. Similar late division of the mitochondrion was also seen in T. brucei [18,31]. Post-segregation in Stage 6 cells, the mitochondrion volume within the two nascent daughter cells was similar and increased in Stage 7 to a similar volume seen in Stage 1 cells.

Formation of the division fold begins prior to mitotic nuclear bridge formation and is retained post cytokinesis

The shape of the cell body is defined by the subpellicular microtubule array and the plasticity of this array enables the cell body to undergo morphological changes during the cell cycle [3,4]. The first step cytokinesis is the in-folding of the cell membrane to form a division fold, as previously seen in T. brucei [32]. This was seen in Stage 4 and later cells with the in-folding of the cell surface occurring centrally along the anterior to posterior axis of the cell (Fig 5A). Fold formation occurred symmetrically via an in-folding from both sides of the cell body (Fig 5A and 5B; arrows). Furrow formation is first visible in Stage 6, and is more pronounced in Stage 7 cells, progressing unidirectionally from between the two flagellar pockets at the anterior end of the cell towards the posterior. Furrow progression is accompanied by the twisting of the two daughter cells away from each other, generating the characteristic posterior cell tip to tip of late cytokinetic cells (Fig 5C and 5D).

Download:

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

Fig 5. Sub-pellicular microtubule organisation from whole cell electron tomography reveals a cytokinesis-derived seam.

(A) 3D reconstructed Stage 5 cell showing the division fold occurring bi-directionally and along the anterior-posterior axis of the cell body. Scale bar = 1 μm. (B) SBF-SEM data slice of a Stage 5 cell showing the in-pushing of the cell body membrane to form the division furrow (arrows). Scale bar = 1 μm. (C) 3D reconstructed Stage 6 and 7 cell showing the twisting (arrows) of the old flagellum (pink) and the new flagellum (orange) away from each other as the furrow ingresses towards the posterior end of the cell. Scale bar = 1 μm. (D) 3D reconstructed Stage 7 cell showing the double indentation of the cell membrane to form the ridge along the anterior-posterior axis. Scale bar = 1 μm. (E) 3D reconstructed Stage 1 cell showing the presence of the ridge post abscission. Scale bar = 1 μm. (F) 3D reconstructed sub-pellicular microtubules of Stage 1 and 2 cells, from serial tomograms. In the two Stage 1 cells, a unique ‘seam’ of converging microtubules is visible (dashed circle). Scale bar = 1 μm. (G) Box whisker plot of sub-pellicular microtubule length distribution in each cell (Stage, S) analysed in 5F. Boxes represent the 25th, 50th and 75th percentiles and whiskers represent the 5th and 95th percentile. (H) Bar graph of number of sub-pellicular microtubules identified in each cell (Stage, S) analysed in 5F. This does not include the flagellar axoneme, microtubule quartet, or pocket and lysosome associated microtubules. Data points represent estimated minimum possible number of microtubules, from maximum cell radius. (I) Quantitative schematic of the microtubule organisation in each cell analysed in 5F, produced by ‘unwrapping’ the sub-pellicular microtubules around a line running along the midline of the long axis of the cell from anterior to posterior. In each plot, each line represents a microtubule and where it starts and ends along the cell midline, plotted, from top to bottom, clockwise around the cell looking posterior to anterior. The unique seams in the two Stage 1 cells are visible as neighbouring microtubules with ends far from the cell posterior. A problematic region for microtubule tracing near the Stage 1 cell posterior is indicted. (J) Plot of radial position, in 30 degree bins, where microtubules end in the Stage 1 and Stage 2 cells.

https://doi.org/10.1371/journal.ppat.1012054.g005

We noted that Stage 1 cells also appeared to contain a fold, which we speculated might be from a previous cytokinesis event. Stage 1 cells do not have a circular cross section and instead have a characteristic set of indentations with a central ridge along one side of the cell body that is generated by the resolution of the cytokinesis fold and furrow (Fig 5D and 5E; arrow). This fold was not present in Stage 2 cells, suggesting that as the cell body elongates microtubule rearrangement occurs to generate a cell with a circular cross-section and resolve the cytokinesis fold. As SBF-SEM does not have the sufficient resolution to visualise and reliably track sub-pellicular microtubules, we reconstructed the sub-pellicular cortex of whole cells by serial section electron tomography to reconstruct whole cells at ~2 nm resolution. Reconstructed whole cells required a large number of sequential dual-axis tomograms (S2 Fig). Thus, we focused on examining the transition from Stage 1 to Stage 2, to investigate the microtubule rearrangements that lead to ridge disappearance. We reconstructed two Stage 1 cells that did not contain a new flagellum and one Stage 2 cell where there was a short new flagellum within the flagellar pocket (Figs 5 and S2). Tracing of the microtubules from the serial section tomograms revealed a parallel array of evenly spaced microtubules, with no discontinuities such as microtubule crossovers or double layers of microtubules (Fig 5F). The microtubules lay in a weakly right-handed helical configuration, with a pitch of ~20 μm (Fig 5F). Consistent with our data and others, the cell body length increased from Stage 1 to Stage 2. Intriguingly microtubule number decreased as the cell elongated, but the overall length of the microtubules increased (Fig 5G and 5H).

The whole cell microtubule models clearly showed the characteristic ridge and double indentation matching that formed by resolution of the division fold and furrow seen by SBF-SEM, indicating that the underlying microtubules were part of its formation. Within the ridge portion of the double indentation, we identified a line of microtubule ends (Fig 5F). To better visualise the microtubule organisation, we ‘unwrapped’ the microtubules around the midline of the cell along its long axis, producing quantitative schematics of where microtubules start and end within the reconstructed cell (Fig 5I). A cluster of anomalously short microtubules starting near the cell anterior but not reaching far to the cell posterior was evident (Fig 5I and 5J; seam). This was visible as a converging line of microtubule ends in the 3D microtubule organisation of the Stage 1 cells (Fig 5F and 5J). As the only discontinuity in an overall otherwise homogenous array we referred to this as a ‘seam’ in Stage 1 cells, which was resolved by Stage 2. Importantly this seam is at the correct position to represent a ‘scar’ of the line of furrow ingression from the previous cell cycle.

Cell body volume occupied by lipid bodies was 6-fold higher in sand fly derived promastigotes

We wanted to extend our analysis to Leishmania promastigotes in the sand fly, as they experience the natural environment with different energetic and metabolic pressures to those in culture, which could be reflected in their 3D cellular organisation. We chose to examine parasites from a late stage infection at the stomodeal valve, as this gives a good number of parasites which are likely furthest removed from in vitro log phase promastigotes, giving the most extreme of comparisons. We analysed a 10-day post infection L. mexicana infected sand fly midgut SBF-SEM dataset containing 218 slices, which covered part of the stomodeal valve [33] (Fig 6A). Every cell examined (n = 89) was in Stage 1 based on our in vitro classification, as they only had one flagellum, suggesting that these cells were in stationary phase. In total, 10 random cells whose whole cell body and flagellum were laying within the dataset and their flagellum not attached to the stomodeal valve, were segmented and the volume and number of organelles—the flagellum, flagellar pocket, mitochondrion, acidocalcisomes and lipid bodies–were analysed (Fig 6B and 6C). Glycosomes, contractile vacuole, Golgi and ER were partially reconstructed in some cells but were not included in this analysis because of low contrast staining leading to lower resolution of the SBF-SEM dataset captured from a sand fly infection, compared with the in vitro promastigote dataset.

Download:

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

Fig 6. Culture and sand fly derived promastigotes differ in organelle volume and spatial organisation.

(A) SBF-SEM data slice showing the cross section of the sand fly stomodeal valve. Scale bar = 5 μm. (B) SBF-SEM data slice of sand fly derived cells highlighting the key organelles identified. Scale bar = 1 μm. (C) 3D model of leptomonad promastigote cells showing the spatial organisation of organelles within the cell body. Organelles are modelled using the following colours: cell body (green), flagellum (pink), nucleus (blue), kinetoplast (blue), mitochondrion (yellow), acidocalcisomes (dark purple), lipid bodies (red) and the flagellar pocket (opaque deep red). Scale bar = 1 μm. (D) Scatter plot of cell body length against flagellum length. Each data point represents one cell. n = 16. Dot plot showing (E) cell body volume (F) cell body surface area (G) nucleus volume and (H) cell body surface area to volume ratio between the sand fly derived cells and culture derived (Stage 1) cells. n = 16, t-Test. (I) 3D models of leptomonads showing the spatial organisation of acidocalcisomes (purple) and lipid bodies (red) throughout the cell body. Scale bar = 1 μm. Dot plot showing (J) number of acidocalcisomes per cell (K) total acidocalcisomes volume per cell and (L) ratio of acidocalcisomes volume in cell between the sand fly derived cells and culture derived (Stage 1) cells. n = 16. Dot plot showing (M) number of lipid bodies per cell (N) total lipid body volume per cell and (O) ratio of lipid body volume in cell between the sand fly derived cells and culture derived (Stage 1) cells. n = 16. Error bars show ±standard deviation. 6 Stage 1 culture derived cells and 10 sand fly derived cells were analysed.

https://doi.org/10.1371/journal.ppat.1012054.g006

Previous studies have defined different promastigote stages in sand fly infections using morphological criteria including cell body and flagellum length [34,35]. Using these criteria, we identified ten leptomonad promastigotes in our dataset ranging from 5.6–13.3 μm in cell body length, with flagella ranging from 5–21 μm in length (Fig 6C and 6D). The mitochondrion structure remained complex in the leptomonad, resembling that seen in the in vitro promastigotes (Figs 6C and 4M). Overall, the promastigotes found in the stomodeal valve of the sand fly were significantly smaller in cell body volume, surface area and nucleus volume, while cell body surface to volume ratio was significantly higher, in comparison to Stage 1 in vitro promastigotes (Fig 6E–6H). Acidocalcisomes and lipid bodies were also distributed throughout the cell body in the leptomonad promastigotes, similar to the culture promastigotes (s 6I, 3D-F). When we examined the numbers and volumes of acidocalcisomes and lipid bodies we also saw differences between the leptomonad and in vitro promastigotes. The number and total volume of acidocalcisomes was lower in the sand fly promastigote cells, representing 0.9% of the total cell body volume in comparison to 7% for the in vitro cells (Fig 6J–6L). In contrast to acidocalcisomes, lipid bodies increased in number and in total volume, representing 5% of the total volume of the cell body in comparison to the in vitro cells where lipid bodies only represented 0.8% (Fig 6M–6O). This suggests differences in metabolism between the in vitro and sand fly parasites, with an increase in lipid storage in the sand fly parasites that correlates with transcriptomic data [36].

Leishmania mexicana culture

L. mexicana (WHO strain MNYC/BZ/1962/M379) promastigotes were grown at 28°C in M199 medium with 10% foetal calf serum, 40 mM HEPES-HCl (pH 7.4), 26 mM NaHCO₃ and 5 μg/ml haemin.

Infection of L. mexicana in the sand fly

Sand fly L. mexicana infection was carried out as described in [55]. Female Lutzomyia longipalpis were fed heat-inactivated sheep blood containing log phase L. mexicana promastigotes at a concentration of 2 x 10⁶ cells/ml, through a chick-skin membrane. Blood engorged females were separated and maintained at 26°C and high humidity with free access to a 50% sugar solution, with a 14-h light/10-h dark photoperiod. They were dissected on day 10 after a blood meal, and the dissected guts were fixed for 24 h at 4°C in Karnovsky fixative (2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer (pH 6.9), transferred to 0.1 M cacodylate buffer with 2.7% glucose washing solution and kept at 4°C until further processing.

Resin embedding for serial-block face scanning electron microscopy (SBF-SEM)

Leishmania-infected sandfly midguts fixed as described above were washed and kept in 0.1 M cacodylate buffer (pH 6.9) at 4°C until post-fixation. L. mexicana promastigotes (C9T7) were grown to exponential phase (1 x 10⁷ cells/ml) as described above and then fixed in growth medium by the addition of glutaraldehyde for a final concentration of 2.5%. Samples were then spun and fixed in 2.5% glutaraldehyde in 0.1M phosphate buffer and washed in the same buffer immediately before post-fixation. After primary fixation with aldehydes, all samples were post-fixed in 1% osmium tetroxide and 1.5% potassium ferricyanide in 0.1 M buffer (phosphate or cacodylate, for in vitro promastigotes and midguts, respectively), for 1 hour in the dark. Samples were then washed in ddH₂O, incubated in freshly prepared 1% thiocarbohydrazide for 20 min in the dark, washed in ddH₂O, and incubated in 2% osmium tetroxide in ddH₂O for 30 min in the dark. Samples were washed again in ddH₂O and incubated in 1% uranyl acetate in ddH₂O overnight, at 4°C and in the dark. Samples were washed with ddH₂O, dehydrated in an ethanol series and embedded in TAAB 812 Hard resin (TAAB).

Serial-block face scanning electron microscopy

The tips of resin blocks containing samples were trimmed and mounted onto aluminium pins using conductive epoxy glue and silver dag, and then sputter coated with a layer (10–13 nm) of gold in an Agar Auto Sputter Coater (Agar). Before SBF-SEM imaging, ultrathin sections (70 nm) of the block face were examined in a Jeol JEM 1400 Flash transmission electron microscope (JEOL), to verify sample quality. Samples were then imaged in a Merlin VP compact high resolution scanning electron microscope (Zeiss) equipped with a 3View 2XP stage (Gatan-Ametek), an OnPoint back-scattered electron detector (Gatan-Ametek) and a focal charge compensation device (Zeiss). The following imaging conditions were used: 1.8kV, 20 μm aperture, 100% focal charge compensation, 5nm pixel size, 2–4 μs pixel time, 100nm (in vitro promastigotes) or 75nm (midguts) section thickness.

Serial-block face scanning electron microscopy segmentation and data analysis

Data were processed using the IMOD software package [56]. Briefly, image stacks were assembled, corrected (for z scaling and orientation) and aligned using eTOMO, and 3D models were produced using 3dmod. Whole individual promastigote cells in varying stages of the cell cycle were identified and trimmed from the dataset and converted into single.mrc files. Within our sand fly stomodeal valve SBF-SEM dataset, we were biased in our selection process as we only identified and analysed promastigote cells whose cell body and flagellum were entirely within the dataset. Incomplete sand fly derived promastigote cells and cells attached to the stomodeal valve (haptomonad promastigotes) were discarded from our analysis. Metacyclic promastigotes are defined based on the morphological criteria where the flagellum length is at least twice the length of the cell body (cell body length < 14 μm)[57]. Our sand fly stomodeal valve SBF-SEM dataset comprised of 218 slices at 75 nm thickness per slice therefore we were bias against finding complete metacyclic promastigotes whose cell body and flagellum were entirely in our dataset. Organelles were manually segmented as individual objects and identified based on specific ultrastructural characteristics (S1 Fig) and modelled following the outer membrane of each organelle. Source data can be found in S1 Table. The volume and surface area of organelles were automatically generated from the object’s surface rendering. The kinetoplast volume was a measurement of the kinetoplast membrane surrounding the kDNA. Segmentation of the Golgi was challenging due to dense packing and stain density of the Golgi stacks, therefore segmentation of the Golgi stacks were combined rather than individual stacks. The volume of the nucleus included the nucleolus. Nuclear pores were identified as gaps in the nuclear envelope. The mitotic nuclear bridge was characterised by the pinching of the nuclear cytoplasm to form two distinct nuclear lobes on either side of the mitotic nuclear bridge. The length of the cell body was measured using the Slicer Window to adjust the X, Y and Z axis of the ZaP Window so the whole of the posterior-anterior axis of the cell body could be viewed in one window and cell body length measured. The SBF-SEM data set shows the entire ultrastructure of the new flagellum inside the flagellar pocket therefore the length of the new flagellum was measured entirely rather than only the external portion of the new flagellum. The statistical analysis tests (Mann-Whitney test and t-Test) were calculated in Microsoft Excel.

Electron tomography sample preparation

L. mexicana promastigote cells were fixed in culture with 2.5% (v/v) glutaraldehyde (TAAB, Aldermaston, UK) for 5 min. Cells were then washed and resuspended in 200 mM phosphate buffer (pH 7.0) with 2.5% glutaraldehyde and 2.0% paraformaldehyde for 2 h. The pellet was washed, post-fixed with 1% osmium tetroxide (Amsbio, Abingdon, UK) for 2 h and stained en bloc with 2% uranyl acetate (Amsbio, Abingdon, UK) for 2 h, then dehydrated in an ethanol series and embedded in Agar 100 resin (Agar Scientific, Stansted, UK). Sections with a nominal thickness of 300 nm were cut using an ultramicrotome and a 35 DiATOME diamond knife and collected via a water bath onto slot electron microscopy grids with a Formvar membrane and stained with Reynolds lead citrate (TAAB, Aldermaston, UK) for 2 min. 48 serial sections were collected across two grids, from which a low magnification approximate map was made to identify whole cells lying near-perpendicular to the plane of sectioning and lying wholly within the sectioned volume (S2 Fig).

Electron tomography data capture

Transmission electron micrographs of tilt series for tomographic reconstruction were captured on a Tecnai TF-30 at 300 kV electron acceleration voltage using a 2048×2048 Gatan CCD, at a magnification of 23.2 Å/px. Images were captured at 1° steps from −65° to +65° using Serial EM, recording the location of each field of view using the navigator (coordinate recording) function and location map images. Where necessary, up to three overlapping fields of view were captured per section to cover the entire cell. The sample was then removed, the grid rotated 90° in the sample holder, the navigator coordinates remapped to account for sample rotation, the same fields of view identified assisted by the navigator function, then precise field locations identified by registration of a test image to 90° rotated copy of the map images. A second tilt series was then captured, with the tilt axis perpendicular to the first–with the exception of a few sections where electron beam-induced damage led to loss of that grid region prior to capture of the second axis.

Electron tomography reconstruction

Tomographic reconstruction of individual tilt series were made using the “Build Tomogram” project function of eTomo, part of IMOD. Fiducialless reconstructions were made of each tilt axis, before aligning the tomogram from the second tilt axis to the first, and averaging into a dual axis tomogram. The second axis typically showed an electron beam-induced sample shrinkage of ~10%. For sections captured with multiple overlapping tomographic volumes, these were subsequently montaged using the “Blend Montages” project function of eTomo, using the field of view captured first as the reference volume for alignment. Finally, the sample sections within the tomographic volumes were flattened using the “Flatten Volume” tool of eTomo.

For alignment of serial tomograms, extremely precise local alignment was necessary to track sub-pellicular microtubules separated by only ~20 nm. The “Join Serial Tomograms” project function of eTomo was found to be inadequate: Subtle section warping, likely from a combination of variation in electron beam induced sample shrinkage and local mechanical deformation from ultramicrotomy depending on local sample properties, necessitated custom alignment. Matched reference points at the top and bottom of sequential tomographic volumes, typically microtubules in flagella, sub-pellicular microtubules in neighbouring cells and mitochondrial cristae, were marked across the tomogram. Delaunay triangulation was used to split the image into triangular regions, and the position of the bounding reference points of each triangular region in the two tomograms were used to calculate the necessary affine transformation to warp that region of one tomogram to align with the other. This warping was applied to each triangle in the plane parallel to the section at all depths (i.e. in x, y, without affecting z), using custom ImageJ macros.

To align all sections, a central reference tomographic volume was selected. In the tomograms above and below that section, reference points were identified, triangulated and the neighbouring tomogram warped to align it to the reference section. Working up and down, reference points to the next section were identified, triangulated and the neighbouring tomogram warped, etc. For each warping step, reference points from structures in neighbouring cells were strongly prioritised over structures in the cell of interest, to avoid biasing subsequent joining of traced sub-pellicular microtubules between sections. Finally, the tomographic volumes from sequential sections were stacked into a single volume for the whole cell, with an estimated 10 to 20 nm depth lost between sections from the action of the diamond knife.

Sub-pellicular microtubule modelling from tomographic volumes

Microtubules in the sub-pellicular array were traced and a 3D model generated using 3Dmod, part of IMOD. First, sub-pellicular microtubules were traced as open contours in the tomographic volumes from each serial section separately. Comprehensive identification of microtubules was confirmed by the expected regular ~20 nm spacing across the cell. Microtubule tracing was possible in all regions except a small volume of the anterior of cell 1, where a region of the sub-pellicular array lay perpendicular to the tilt axis and perpendicular to the imaging plane of the camera leading to insufficient z resolution to identify individual microtubules.

Traced microtubules were joined across serial sections based on both proximity, i.e. the microtubule is in the same place in the two sections, and trajectory, i.e. making that join does not introduce a sharp kink or bend in the microtubule. A consistent approach was used to avoid bias. First, the traces of microtubules which could be unambiguously identified in both sections were joined. In order of preference: 1) based on distinctive nearby non-sub-pellicular microtubule features, like nearby ER or mitochondrial structures. 2) based on distinctive folds or curves in the sub-pellicular array, particularly where there is a sharp bend with two or three protruding microtubules. 3) based on variation in microtubule spacing, where a distinctive narrow or wide spacing can be seen in both sections. Having made joins between sections at these unambiguous sites, this provides ‘landmark’ joins at many positions around cell perimeter. The intervening microtubules can then be joined. On the basis that microtubule starts and ends were rare within the tomographic volumes, joins were made which minimised the number of necessary microtubule starts and ends falling in the lost volume between sections.

To map microtubule lengths cell-wide, microtubules in each serial section tomogram needed to be correctly joined across the 7 to 11 300 nm serial sections, between which a small depth (~10–20 nm) is lost, ~5% of the cell volume. We therefore carefully considered our strategy for microtubule joining. Based on the cell circumference and even ~20 nm microtubule spacing, we expected ~100 to 150 microtubules around the widest point of each cell. As ~5% of the cell volume is lost between sections, microtubule starts and ends lost in this lost volume should be 20× rarer than starts and ends visible within sections, yet in-section starts and ends were rare. Therefore, microtubule starts and ends lost between sections must be extremely rare, implying the subpellicular microtubules must be long and the correct between-section joins should introduce a minimum number of microtubule starts and ends. Following this strategy, and using advanced section-section alignment, we build high confidence model of all microtubules in the three cells.

Therefore, where the same number of microtubules lay between these landmarks in both sections, this implies no microtubule starts/ends in the between section space and therefore join the traces of these microtubules, while monitoring for anomalously large disjoints. Where the number of microtubules between two landmarks differed between two sections, then a best effort from proximity and trajectory was made to identify which microtubule(s) started/ended within the section. Only very close to the cell anterior or posterior were cases identified where multiple nearby microtubules seemed to start/end within the between section space–these were more likely to be erroneously joined, however any error here does not affect the major conclusions.

Negative staining protocol

For preparation of whole-mount promastigote cytoskeleton samples, 8 ml of log phase promastigotes (1 × 10⁷ cells/ml) were harvested by centrifugation (800g for 10 min) and resuspended in 500 μl of M199. 8 μl of cell resuspension was added to a glow-discharged formvar and carbon-coated 200 mesh nickel grid and left for 2 min. Promastigotes attached on the formvar membrane were treated with 1% IGEPAL in PEME (0.1 M PIPES, pH 6.9, 2 mM EGTA, 1 mM MgSO₄, 0.1 mM EGTA) for 5 min, fixed with 2.5% glutaraldehyde in PEME for 10 min and stained with 1% aurothioglucose in ddH₂O. For preparation of cell body microtubule depolymerised cytoskeleton samples, samples were treated with 300 mM CaCl₂ in PEME for 2 min and washed twice with PEME after the IGEPAL treatment. The samples were observed using a Jeol JEM-1400 Flash transmission electron microscope operating at 120 kV and equipped with a OneView 16-megapixel camera (Gatan/Ametek, Pleasanton, CA).