A quantitative study of the number and distribution of neoblasts in Dugesia lugubris (Planaria) with reference to size and ploidy

The role of the planarian neoblast as a totipotential stem cell has been discussed in the literature for well over a century (cf. Brøndsted’s excellent review, 1955). It remained a matter of strong debate until 1949 when Dubois demonstrated conclusively the migration of neoblasts (through regions depleted of their neoblasts by radiation) to the surface of a wound. She showed that the onset of regeneration was delayed until the neoblasts reached the wound area, and that once they had arrived regeneration took place at the normal rate with the neoblasts actively dividing in and just posterior to the blastema. Since then many authors (the Brøndsteds, Stéphan-Dubois, Pedersen, Lender, the Benazzis) have studied the histochemistry, the distribution, and the factors which influence differentiation of the planarian neoblast in several species. Only Brøndsted & Brøndsted (1961) have reported on the total number of neoblasts in a planarian. This paper reports observations on the number of neoblasts in diploid and triploid planarians, the relation of this number to the size of the animal and the distribution of neoblasts with respect to the animal’s long axis.

The animals used for this study were the diploid (A) and triploid (B) biotypes (Benazzi, 1957) of a naturally occurring autopolyploid series of Dugesia lugubris (O. Schmidt). Only animals which had hatched from cocoons laid in the laboratory were used, so that the size of each animal could be taken as a measure of its age (cf. Reynoldson, 1961), i.e. the cell counts for small animals were not influenced by possible variations due to shrinkage of starved animals.

The length of each planarian was measured by placing the living animal in a very shallow depth of water in a Petri dish directly on mm graph paper. At a magnification of × 10, the length of the animal was determined to the nearest 0·5 mm while it was swimming in a relaxed and extended condition.

Animals of various sizes were starved for 6 days before killing, in the relaxed and extended position, either with a drop of 2 % HNO₃ (Hyman, 1924), or by placing the animal in a small drop of water between a glass slide and a No. 0 coverglass prior to fixation in Susa or Formol-Zenker. Specimens were serially sectioned (frontally) at 10 μ on a rotary microtome. The material reported on was stained with 25 % Azure B in pH 4·0 Mcllvainie’s buffer and differentiated in tert-butanol at 40 °C for 18-22 h (18 h for Susa-fixed material, 22 h for Formol-Zenker).

Azure B is specific for DNA and RNA and does not fade (Flax & Himes, 1952). RNA specificity was tested by RNase preincubation, resulting in a total absence of blue or purple staining material (RNA). The RNA specificity of a stain is an important aide in the recognition of neoblasts (Benazzi-Lentati, 1943; Pedersen, 1959). The neoblast is easily recognized due to its peculiar morphology, i.e. a nucleus of ∼ 7 μ diameter surrounded by a narrow (∼ 1 μ) band of RNA-rich cytoplasm.

Neoblasts, or parts of neoblasts, were recognized by their staining properties and all of them were counted at × 400. Twenty per cent of the sections (two every 100 μ along the cephalo-caudal axis of each planarian) were counted for twelve triploids and seven diploids. Serial sections were made of twenty-eight triploids and eleven diploids, but since there was good agreement between the results obtained from animals of the same size (live length), only two or three specimens of a given ploidy were counted for each size.

In order to check the reliability of the counts, slides were chosen at random for recounting. For small animals (2 mm live length) the median variation for one section and the variation in the total number of neoblasts summed over the recounted sections were both less than 5 %, while for large animals (10 mm live length) the variations were 12-13 % and < 0-5 % respectively.

Measurements of nuclear diameters (the mean of two measurements, perpendicular to each other, per cell) were made at × 844 magnification with an ocular micrometer (Leitz) to the nearest 0·3 μ.

The area of every twentieth section was measured at a magnification of × 187·5 with a Leitz eye-piece micrometer. The area of the parenchyma plus gut was obtained by subtracting the area enclosed by the pharynx sheath from that of the entire section. These areas were graphed against position on the anterior-posterior axis of the animal and volumes were obtained by measuring the area under the curves.

Estimates of the number of neoblasts in a given interval along the anterior-posterior axis of a planarian, based on the 20 % sample taken, were not significantly different (P > 0·05) from those obtained by counting every section in that interval. The 20 % sample yields estimates of the number of neoblasts per slide within 10 % of the actual count, and summed over several slides the difference drops to ∼ 1 %. This is of the order of the variations found in recounts for large animals (10-12 mm), i.e. 12-13% for single sections and < 0-5 % for sums over several slides.

The number of neoblasts in any one section, or in any part of a serially sectioned planarian comprising more than one section, has been taken to be the number of neoblasts counted in that section or the sum of the number of neoblasts in each of the relevant sections. Corrections for overcounting due to some cells appearing in more than one section will be considered under Results, section D, and in the Appendix.

The relationship between the length of a planarian and its total number of neoblasts is shown in Fig. 1. The data for diploids and triploids were handled separately and the best line for each set of data was fitted by the method of least squares. Graphing the data on Cartesian and on semi-logarithmic coordinate scales showed strong non-linearity, whereas if both variables were transformed logarithmically, linearity was evident. The equations for the two regression lines are :

Analysis of variance (Bennett & Franklin, 1954) for linearity of regression yielded F values of 369 for the diploids (F_1,10,0.005 = 22·8) and 534 for the triploids (F_1,5,0.005 5 = 12·8), indicating that the probability of obtaining such a degree of linearity by chance is much less than 0·005.

The significance of the differences in slope and intercept of the two regression lines were examined by means of the t-test, yielding a t value for the slope difference of t₁₅ = 0·2173 (P > 0·7) and a t value for the intercept difference of t₁₅ = 1·2373 (P > 0·1). Thus the number of neoblasts in a diploid planarian of a given size is not significantly different from that in a triploid of the same size.

In order to be able to compare the antero-posterior distribution of neoblasts in animals of various sizes over a wide range of neoblast numbers and planarian lengths, both variables had to be normalized. The independent variable, length, was normalized by taking class intervals of 10 % of the total (fixed) length. The dependent variable was normalized by setting the area under the distribution curve for each animal equal to one, i.e. for each class interval the proportion of the animal’s total neoblast count appearing in that interval was recorded. The agreement of distribution curves between animals of the same size and ploidy was good, as is shown, e.g. by the data for 12·0 mm triploids, in Table 1. The next row gives for each interval the mean of the proportions of neoblasts found in that interval; and the following row, the standard deviations of those means. The two maxima, one in the third and fourth intervals and the other in the eighth interval, with an intercalary relative minimum in the sixth and seventh intervals, are characteristic for the triploid Dugesia lugubris.

In Fig. 2 we see that the normalized distribution curves for 2·0, 3·0, 5·0, 8·0 and 12·0 mm long triploids are nearly superimposable, the relative minimum corresponding to the position occupied by the pharynx occurring in the sixth and seventh intervals for all sizes. In contradistinction to the near superimposability of the triploid distribution curves for all sizes, the diploids show a distinct progressions in the position of the pharynx as the animals increase in size (Fig. 3). The pharynx moves from the seventh interval in the 2·0 mm animal (freshly hatched) to the sixth interval in the 5·0 mm animal (immature), to the fifth interval in the 10·0 mm animal (sexually mature adult). The pattern to pre- and post-pharyngeal maxima with a pharyngeal relative minimum, however, is also found for the diploid.

In order to estimate the possibility that the distribution differences or similarities for various sized animals might be artifacts due to different degrees of shrinkage for different sized animals during fixation and processing, Olmstead & Tukey’s corner test for association (1947) was employed. No significant association was found (P > 0·1) between size of animal when alive and the ratio of fixed/live length (i.e. degree of contraction) for diploids only, triploids only, or the combined data. The mean degree of contraction was 51 % with a standard error of ± 1-5 %. Thus it is unlikely that the distributions reported above were influenced by the histological methods employed.

The method of Abercrombie (1946) was employed to correct for the possibility of counting a cell in more than one serial section. However, as Abercrombie assumed that errors due to the difference between cell diameter and cell diameter as measured in sections are negligible, a more rigorous formula has been derived and is discussed in the Appendix. For the conditions reported, this correction amounts to only 1 %, but for other conditions, this correction need not be negligible. The value of Abercrombie’s correction factor for diploids was found to be the same as that for triploids. Thus the relative positions of the points for diploids and triploids (Fig. 1) remain unchanged, a simple shift of scale giving the true number (lower scale) instead of the absolute count.

The relationship between the length of a triploid planarian and its volume (minus that enclosed by the pharynx sheath) is shown in Fig. 4. The line was fitted by the method of least squares. Analysis of variance shows significant linearity (P < 0·05). The regression equation

Lender & Gabriel (1960) have published a neoblast distribution curve for Dugesia lugubris, presumably of biotype B (triploid), the type commonly found in northern Europe (Benazzi, 1957), serially sectioned at 5 μ. They counted five sections at each of 23 sites along the antero-posterior axis of the animal. Regraphing their data, connecting the points with straight lines and integrating the area under the curve yields as an estimate of the total number (absolute count) of neoblasts in their 8 mm (measured when alive) animal, 2·00 x 10⁵ cells. Applying Abercrombie’s correction (5/(7 + 5) = 41 %) gives the corrected total as 8·2 x 10⁴ cells. This is within 13 % of our corrected number of neoblasts for an 8 mm triploid (7·3 x 10⁴). The distribution curves presented here differ from that of Lender & Gabriel in that instead of two maxima prepharyngeally, we find one major maximum prepharyngeally and another smaller maximum post-pharyngeally. The shape of our distribution curve is more like those of Brøndsted & Brøndsted (1961) for D. lacteum and E. torva. We did find a small relative maximum just posterior to the eyes, but this was swamped by the rising prepharyngeal maximum. The distribution curves presented here are also in agreement with the qualitative description of Pedersen (1959) for Planaria vitta. The fact that our estimate of the total number of neoblasts in Lender & Gabriel’s material agrees as well as it does with our own suggests that the differences in the distribution curves might be due to different fixation methods, as the Brøndsteds used Zenker, and formol-Zenker or Susa, whereas Lender & Gabriel used 80 % ethanol or Carnoy. Pedersen (1959) notes that Carnoy fixation alters the shape of the neoblast. Stéphan-Dubois (1961), using the methods of Lender & Gabriel, studied the neoblast distribution off), lacteum. The shape of her distribution curves (7-8 sections counted at each of six sites for three animals of ≈ 10-11 mm length) differs from that of the Brøndsteds for D. lacteum and from those of the present report and that of Lender & Gabriel for D. lugubris. The number of neoblasts per half-section, which Stéphan-Dubois publishes, when corrected with Abercrombie’s correction (from her photograph of D. lacteum neoblasts, the measured nuclear diameter is ≈ 5 μ) yields (in units of neoblasts per 10 μ section), 271 in the region of the brain, 397 post-cephallically but within the first 10 % interval, 385 in the third 10 % interval, 440 between the fifth and sixth intervals (region of the pharynx), 397 in the seventh interval (postpharyngeal) and 225 in the ninth interval. The differences between these counts and the present ones for corresponding positions in a 12 mm D. lugubris, do not differ by more than 25 % of the sum of the corrected number of neoblasts at each level for both species. Thus the counts of the Brøndsteds and those of Stéphan-Dubois do not agree with each other. This difference is difficult to reconcile in view of the fact that the total counts obtained by Brøndsted & Brøndsted for D. lacteum and E. torva (3·78 × 10⁴ and 3·13 × 10⁴, respectively) of 15 mm live length are almost an order of magnitude lower than the extrapolated value for D. lugubris (Fig. 1). Also, their method of correcting for double counting would tend to give values higher, rather than lower, than the actual number, thus diminishing, rather than enhancing, this difference.

The constancy of the relative position of the pharynx in triploids from 2 mm (just hatched) to 12 mm (cocoon-laying adult) in length as opposed to the shift forward in the diploids is unlikely to be a fixation artifact as planarians of all sizes (for the fixatives used) shrank to the same percentage of their live length. This implies a difference in mode of growth between diploids and triploids. It would seem that triploids grow uniformly, whereas diploids grow more from the post-pharyngeal than from the pharyngeal end. The diploid pharynx shift was also observed in animals not used for the neoblast counts.

The observation that the volume of D. lugubris B (excluding that of the pharynx, although its inclusion would not alter the conclusion) increases with length at a faster rate than does the total number of neoblasts implies that the overall density of neoblasts in a large animal is lower than that in a small animal. This could have some bearing on the problem of senescence in planarians. A large animal is physiologically ‘old’ with respect to regeneration ability (head frequency), whereas if it is starved or cut in half to regenerate, the reproportioned individual (after completion of morphallaxis) becomes physiologically ‘young’.

The distribution curves for fixed total volume were, in general, uniform in pattern, rising smoothly to a maximum in the fifth 10 % interval and then dropping caudally.

The distribution curves of the ratio of % neoblasts/ % volume for each 10 % interval along the antero-posterior axis were rather variable but had a general pattern. In the 5, 8 and 12 mm long triploids, the ratio for the head (1st 10 % unit) was 1·0, increasing with length from 1·14 (5 mm animal) to 1·75 (12 mm animal). The ratio decreases posteriorwards from the first interval, is a minimum in the fifth or sixth interval (0·72 = 0·78), then rises until the eighth interval, and finally falls for the 5 mm animal, levels off for the 12 mm animal and continues rising for the 8 mm animal.

Reynoldson (1961) found that newly hatched D. lugubris (2·5 mm long) grow, when fed once a week under laboratory conditions, at a rate of 2·5 mm per 4 weeks, and shrink, on starvation at a summer temperature, at a rate of 2·0 mm per four weeks. As Lender & Gabriel (1961) state that in 2-months-starved D. lugubris (reduced to two thirds of their length) the shrinkage is not due to the loss of the neoblasts, it would be interesting to see quantitative data as to whether or not the number of neoblasts in a starved animal is, in fact, the same as the number that it would have had, had it not been starved.

1. Ce travail est consacré à des observations qui ont été effectuées sur le nombre des néoblastes chez des planaires diploïdes et triploïdes, sur la relation entre ce nombre et la taille de l’animal, ainsi que sur la distribution des néoblastes le long de l’axe principal du corps.

2. Le nombre des néoblastes est fonction de la longeur de l’animal chez le diploïdes comme chez les triploïdes, mais il n’y a pas de différence significative entre les deux biotypes. L’ensemble des faits observés donne la relation suivante : log (nombre des néoblastes) = 1,90 log (longueur)+ 3,47.

3. La distribution des néoblastes le long de l’axe principal de la planaire a deux valeurs maximales, l’une en avant, l’autre en arrière du minimum intermédiaire qui se situe au niveau de la région pharyngienne de l’animal.

4. La position relative du pharynx ne change pas au cours de la croissance de l’animal triploïde; au contraire, chez le diploïde, le pharynx subit un glissement progressif vers la région céphalique. Comme l’analyse statistique n’a donné aucune différence significative en ce qui concerne les méthodes histologiques utilisées pour les deux biotypes, cette observation est interprétée comme une démonstration de modes de croissance différents chez les diploïdes et les triploïdes.

5. Le volume de la planaire (à l’état fixé) est lié avec la longueur de l’animal (vivant) par la relation suivante: log (volume) = 2,62 log (longueur) + 5,70. La pente de cette équation présente une différence hautement significative avec celle de la courbe de la variation du nombre des néoblastes en fonction de la longueur. Donc des animaux plus gros ont une densité de néoblastes moindre, quoiqu’ils en possèdent plus, en valeur absolue, que les animaux plus petits.

6. On peut supposer que cette dernière observation a une incidence sur le problème de la sénescence chez les Planaires.

7. Le comptage d’objets sur coupes sériées est sujet à une surestimation qui peut être corrigée par la méthode d’Abercrombie. Cependant, cette méthode est basée sur une présomption qui n’est pas toujours valable. C’est pourquoi une formule de correction plus rigoureuse est proposée.

8. Cette formule a été appliquée à la mesure du diamètre nucléaire des néoblastes (fixés). Il n’y a pas de différence significative entre les diamètres nucléaires des néoblastes triploïdes et diploïdes.

The author wishes to acknowledge the skilled technical assistance of Miss Judith Whelan.

The L in Abercrombie’s formula is the mean true length (or diameter for spherical objects), which is different from the mean measured length when length is measured in sections. He implicitly assumes that all objects are cut and therefore takes the mean measured diameter as the mean chord of a circle or π / 4 times the true diameter. Although he argues that the error of density overestimation due to use of the mean measured diameter, in practice, seldom exceeds 10 %, the author thinks that it may be useful to consider a more rigorously derived formula for obtaining the mean true diameter from the mean measured diameter, valid for any object length and section thickness. This would serve the twofold purpose of enabling one to obtain mean true diameters (or lengths) from measurements based on sectioned material (if the objects are randomly distributed relative to the section thickness), and would allow one to use Abercrombie’s formula without introducing errors, albeit small ones, due to poor estimation of object length.

Consider a section of thickness T taken through a block of substance containing objects of diameter D_t (Fig. 5). All objects with centres lying between points A and D will be counted as lying in section T and will thus contribute to the mean measured diameter. Those objects with centres between B and C will be uncut and each will contribute a length D_t (the true diameter) to the mean measured diameter (D_m). The proportion of objects counted in the section T, having centres between B and C, is BC/AD or (T-D_t/T+D_t). All objects appearing in section T with centres not between B and C will be cut and each will contribute, on average, (π /4) D_t to the mean measured diameter (mean chord of a circle = π /4 × diameter). The proportion of cut objects is 1 - [the proportion of uncut objects], so we can write

Measurements of nuclear diameters (taken as the average of two measurements at right-angles to each other, for each cell) were made for both diploid and triploid neoblasts. A sample of twenty-five diploid, and thirty-eight triploid neoblasts produced mean measured diameters of 5·6 ±0·8 and 5·8 ± 1·1, corresponding to mean true diameters of 6·8 (6·3-7·2, 95 % fiducial limits) and 7·0 (6·5-7·6, 95 % fiducial limits) /μ respectively, for fixed cells, which is clearly not a significant difference.