cell counting tech note 6340 A Comparison of the TC10™ Automated Cell Counter and the Hemocytometer for Quantifying Insect Hemocytes Joanna Randall, Yamini Tummala, and Kenneth Wilson stored at –20ºC until required. For each of the two methods, Lancaster Environment Centre, Lancaster University, Lancaster, hemolymph was thawed, diluted 1:1 in phosphate-buffered UK, LA1 4YQ, United Kingdom saline, pH 7.4, and placed on ice. The hemolymph was then Introduction stained in a 1:1 ratio with 0.4% trypan blue to distinguish Hemocytes are an essential component of the invertebrate viable from nonviable cells. innate immune system (Strand 2008). Their close similarities to vertebrate white blood cells enable their use as a tractable In the first method, cell counts were determined using model for immunological research. The quantification of the TC10 automated cell counter in accordance with the hemocytes in the hemolymph (invertebrate blood) is one way manufacturer’s instructions (Bio-Rad Laboratories, Inc.). of assessing the efficacy of the immune response because Duplicate 10 µl samples of stained hemolymph were the absolute number of hemocytes changes during the transferred to the two chambers of a Bio-Rad counting course of an infection (Bergin et al. 2003, Brayner et al. 2007), slide. The slide was inserted into the automated counter and the density of hemocytes in the hemolymph is strongly and readings of total and viable cell counts were generated correlated with the capacity to encapsulate or phagocytose within 30 seconds. In the second method, duplicate 10 µl parasites (Prevost and Eslin 1998, Kraaijeveld et al. 2001, samples of stained hemolymph were transferred to each side Wilson et al. 2003, Costa et al. 2005). Most researchers of the chamber in the hemocytometer. Hemocytes were then continue to use cell staining and manual microscopy counted in five nonadjacent squares on the hemocytometer techniques to enumerate the hemocytes in a given sample. grid under a light microscope, which took approximately However, given the laborious nature of this method (with 3 min/sample. a throughput often as low as 6–8 samples/hr), more rapid The TC10 cell counter was subsequently used to measure the techniques are required to increase our understanding of density of hemocytes in the hemolymph of S. littoralis larvae the role of hemocytes in invertebrate immunology. fed on artificial diets that varied in their relative amounts of Here, we tested the suitability of Bio-Rad’s TC10 automated protein and carbohydrate. The methodology was based on cell counter, a tool originally developed for assessing previously described experiments (Povey et al. 2009, Cotter mammalian cell counts, for counting insect hemocytes. et al. 2011). Briefly, a total of 200 larvae on the first day of their Hemocyte counts from the model insect species Spodoptera final instar were randomly allocated to one of 20 diets that littoralis (Lepidoptera: Noctuidae) were compared with varied in the relative percentages of both soluble protein and those obtained using a traditional tool for counting insect digestible carbohydrates, both varying from 2.8 to 52.5%, with hemocytes: a hemocytometer with improved Neubauer ruling the remainder of the diet comprising indigestible cellulose (see under light microscopy. We then tested the capacity of the Cotter et al. 2011 for a full diet description). After 48 hours, the TC10 cell counter to distinguish among cell counts in insects insects were bled, and 10 µl of hemolymph was added to fed on artificial diets varying in their protein and carbohydrate 10 µl of ETDA/glycerol (50:50) and stored at –20ºC until contents, which has previously been shown to influence required. Hemocyte counts were then obtained using the hemocyte counts in this species (Ponton et al. 2011). TC10 cell counter as described above. Materials and Methods Statistical Analysis The two sampling methods were first assessed independently Hemolymph samples from 30 sixth instar larvae were for repeatability using a series of linear models in the freeware collected into a precooled microcentrifuge tube by piercing package R version 2.15.0 (http://www.R-project.org) for the cuticle between the first and second prolegs with a the relationships between the duplicate hemocyte counts 25-gauge BD Microlance needle (Becton, Dickinson and for each sample (n = 30). All data (live, dead, and total cell Company). The samples were immediately frozen and counts) were normalized using a log10 transformation (count + 1) prior to analysis. The linear models were assessed by fitting each regression line through the origin. The slope of the A 7.0 0 regression lines was then tested for deviation from a slope of 6.80 unity, the expected slope if the paired counts were correlated 6.60 without bias. The mean numbers of each cell type (live, TC10 count 6.40 dead, and total) determined with the two counting methods 10 log (hemocytometer vs. TC10 cell counter) were then compared cells/ml 2, replicate 6.20 as described above. 6.00 6.00 6.20 6.40 6.60 6.80 7.00 log TC10 count replicate 1, cells/ml The effects of the two digestible macronutrients on live, dead, 10 and total hemocyte counts as measured by the TC10 cell B 7.50 counter were determined using three general linear models in R version 2.15.0. As potential explanatory terms, we tested 7.0 0 the total amount of digestible protein in the diet (%P), the total TC10 count 10 6.50 amount of digestible carbohydrate (%C), quadratic functions log replicate 2, cells/ml 2, replicate of the two terms (%P2 and %C2), and an interaction between P and C. The dependent variables (live, dead, and total 6.00 6.00 6.50 7.00 7.50 counts) were first transformed, when appropriate, to conform log10 TC10 count replicate 1, cells/ml to the normality assumption, and the minimal model was C 7.40 determined by stepwise deletion using F-tests. 7.20 Results and Discussion There was a high level of repeatability for the TC10 cell TC10 count 10 7.0 0 log counter method, as indicated by a strong correlation between cells/ml 2, replicate counts for the paired samples (Figure 1A–C; live cell count: 2 6.80 slope, b ± SE = 0.997 ± 0.002, r = 0.99; dead: b ± SE = 6. 8 0 7. 0 0 7. 20 7. 40 2 2 log TC10 count replicate 1, cells/ml 1.001 ± 0.002, r = 0.99; total: b ± SE = 1.000 ± 0.001, r = 10 1.00); the slopes of all the regression lines did not differ D 7.0 0 significantly from unity P( > 0.05). The hemocytometer also 6.80 showed a good correlation between sample replicates 6.60 (Figure 1D–F; live cell count: slope, b ± SE = 0.984 ± 0.003, r2 = 0.99; dead: b ± SE = 1.006 ± 0.002, r2 = 0.99; total: b ± 6.40 hemocytometer count hemocytometer replicate 2, cells/ml 2, replicate SE = 0.999 ± 0.002, r2 = 0.99); however, for both live and 10 6.20 log dead counts there was evidence of deviation from a slope of 6.00 one (live: t = 5.43, P < 0.001; dead: t = 5.3, P < 0.001), 6.00 6.20 6.40 6.60 6.80 7.00 29 29 log10 hemocytometer count replicate 1, cells/ml indicating that the hemocytometer method was slightly less E 7. 2 0 reliable for counting insect hemocytes than the TC10 cell counter. In particular, there was a slight reduction in the 7.0 0 number of live cells in the second replicate count compared 6.80 with the first, which may be attributable to the length of hemocytometer count hemocytometer time required to complete the second reading (up to cells/ml 2, replicate 6.60 10 20 min between samples) but which did not affect the log 6.40 total number of cells. 6.40 6.60 6.80 7.00 7.20 log10 hemocytometer count replicate 1, cells/ml When compared across methods, the mean counts of live, F 7. 3 0 dead, and total cells were also strongly correlated (Figure 7. 2 0 2A–C; live cell count: slope, b ± SE = 0.987 ± 0.004, r2 = 0.99; dead: b ± SE = 0.980 ± 0.004, r2 = 0.99; total: 7.10 2 b ± SE = 0.984 ± 0.002, r = 0.99), demonstrating that 7.0 0 hemocytometer count hemocytometer the Bio-Rad method is a suitable replacement for the cells/ml 2, replicate 10 6.90 hemocytometer for estimating hemocyte counts in an insect log 6.80 hemolymph sample. However, it should be noted that 6.80 6.90 7.00 7.10 7.20 7.30 log hemocytometer count replicate 1, cells/ml the slopes were significantly different from unity for all 10 Fig. 1. Repeatability of the two methods for counting hemocyte density three cell counts (t29 = >3.25, P < 0.003), which was due to the consistently lower cell counts obtained with in hemolymph from S. littoralis larvae. Live, dead, and total cell counts, respectively, are displayed in A–C, TC10 automated cell counter, and D–F, the hemocytometer compared with the TC10 hemocytometer. The dashed lines indicate a slope of unity and the solid lines automated cell counter. are the fitted regressions forced through the origin. All six regression analyses were statistically significant P( < 0.001). © 2013 Bio-Rad Laboratories, Inc. Bulletin 6340 To determine whether the TC10 cell counter was capable A 6 x 106 of distinguishing between hemocyte counts obtained from different experimental treatments, we used the cell counter 4 x 106 to estimate hemocyte densities in larvae fed on diets varying in the percentages of both digestible carbohydrate and 2 x 106 soluble protein.