Stereology and computer assisted three-dimensional reconstruction as tools to study probiotic effects of Aeromonas hydrophila on the digestive tract of germ-free Artemia franciscana nauplii


Wim Van den Broeck, Department of Morphology, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, B-9820 Merelbeke, Belgium. E-mail:


Aims:  Validation of stereology and three-dimensional reconstruction for monitoring the probiotic effect of Aeromonas hydrophila on the gut development of germ-free Artemia franciscana nauplii.

Methods and Results:  Germ-free Artemia nauplii were cultured using Baker’s yeast and dead Aer. hydrophila. Live Aer. hydrophila were added on the first day to the treatment group. The gut length and volume were monitored on days two and four using stereology and three-dimensional reconstruction. Both methods showed comparable results. Stereology was least labour intensive to estimate volumes, while three-dimensional reconstructions rendered architectural and topographical data of the gut. Moreover, a positive effect of probiotic bacterium, Aer. hydrophila is likely.

Conclusion:  Slight increment in the growth of the digestive tract of A. franciscana nauplii exerted by probiotic bacteria could be detected using stereology and three-dimensional reconstruction.

Significance and Impact of the Study:  The gnotobiotic Artemia rearing system is unique to investigate the effects of micro-organisms on the development of nauplii. However, in the base of this model system, only survival counts and length measurements exist as monitoring tools. Therefore, additional tools such as stereology and three-dimensional reconstruction are prerequisite to obtain more powerful analysis.


Multicellular organisms have developed various mechanisms to support complex and dynamic consortia of micro-organisms during their life cycle (Rawls et al. 2004). A better understanding of several biological functions in aquatic organisms such as host–microbe interactions can only be achieved by studying organisms that are cultured in fully controlled conditions (Marques et al. 2005). Within this context, the importance of a gnotobiotic model system with strong tools emerges. Therefore, in this study, the possibilities of using stereology and computer-based three-dimensional visualization as monitoring tools were first analysed. Second, the differences in the digestive tract of control germ-free Artemia nauplii compared to gnotobiotic nauplii fed with a probiotic bacterial strain Aeromonas hydrophila were investigated. Artemia franciscana nauplii were used as the animal model, because Artemia is the most widely used live feed in larviculture of marine fish and shellfish (Yeong et al. 2007). An important step forward in the development of such a gnotobiotic model was made more than twenty years ago when Sorgeloos et al. (1986) and later Marques et al. (2004a) described a successful method to obtain sterile Artemia cysts and nauplii, called gnotobiotic Artemia rearing system (GART). Several scientific studies based on GART have been performed on the probiotic effects of bacteria (Marques et al. 2005), the effects of different immunostimulants (Marques et al. 2004b; Soltanian et al. 2007), and heat shock proteins (Yeong et al. 2007). For the assessment of this model system, survival counts, individual length measurements and total length of Artemia already exist as monitoring tools. In addition, the current study intends to strengthen GART with highly performing monitoring tools in order to detect the slight differences in the length and the volume of the digestive tract segments of the control and treatment groups.

In this study, the midgut of Artemia nauplii was focused, as this segment is the functional site for absorption, storage and secretion (Schrehardt 1987), as well as the hindgut as it performs a mechanical function.

Materials and methods

Culturing and harvesting of germ-free yeast

Wild-type (WT) strain of baker’s yeast (Saccharomyces cerevisiae) [BY4741 (genotype, Mata his3Δ1 leu2Δ0 met15Δ0 ura3Δ0)], kindly provided by the European S. cerevisiae Archive for Functional Analysis (University of Frankfurt, Frankfurt, Germany), was used as live feed for Artemia. It was cultured germ free in minimal yeast nitrogen-based agar. Harvested yeast cells were counted by following the methodology described by Marques et al. (2006) and Gunasekara et al. (2010). Yeast suspensions were stored at 4°C and used to feed Artemia until the end of the experiment.

Culturing, harvesting and killing of bacteria

The Aer. hydrophila strain (LVS3) selected for the experiment is a rod shaped, motile, gram negative, Aeromonadaceae bacterial strain. LVS3 was cultured, harvested, and the density was determined according to the methodology described by Marques et al. (2005, 2006) and Gunasekara et al. (2010). As LVS3 cells were used for feeding, they were killed by autoclaving at 120°C for 20 min. Bacterial suspensions were stored at 4°C and used to feed Artemia until the end of the experiment.

Culturing and feeding of gnotobiotic Artemia

The experiment was carried out using A. franciscana originating from the Great Salt Lake, Utah. Artemia cysts were hydrated and decapsulated following the procedures described by Defoirdt et al. (2005) in order to obtain sterile cysts and subsequently sterile nauplii. Sterilization of necessary equipment, decapsulation of cysts and setting of culture tubes for hatching were carried out according to the procedures described by Gunasekara et al. (2010). After about 20–24 h, 30 nauplii were transferred to fresh 50-ml sterile tubes containing 30 ml filtered autoclaved instant ocean. Half of the tubes received living WT yeast and dead LVS3 (control germ-free group) as daily feed, whereas the other half received the same daily feed, which was supplemented with live LVS3 on the first day only (treatment gnotobiotic group). All tubes were placed on a rotor turning at 4 rev min−1. Daily feeding was carried out ad libitum. As such 62·45 μg WT strain (ash-free dry-weight content) and 624·54 μg dead bacteria were administered per culture tube during the 4 days of the study period as described by Marques et al. (2004a). The treatment group was supplemented with 42·97 μg of live bacteria per culture tube on the first day.

The control and the treatment groups were tested with four replicates for each. Sampling was performed at days 2 and 4.

Verification of axenity

The axenity of the decapsulated cysts, water from Artemia culture and dead LVS3 suspension (after autoclaving) was verified using bacteriological plating. Absence or presence of bacterial growth was monitored after 5 days of incubation at 28°C of 100 μl culture medium or feed spread plated on marine agar (n = 2).

Processing the specimens

Nauplii were fixed, prestained and further processed for histological sections following the procedures described by Gunasekara et al. (2010). Per treatment and per day, two nauplii were cut into serial transverse sections of 5 μm thickness using a microtome (Microm HM360; Prosan, Merelbeke, Belgium). All histological sections were stained with haematoxylin and eosin [Eosin yellow (C.I. 45380), VWR International bvba/sprl, Leuven, Belgium].

The histological sections were photographed using a motorized microscope (Olympus BX 61; Olympus Belgium, Aartselaar, Belgium) linked to a digital camera (Olympus DP 50; Olympus Belgium).


The volumes of the midgut (MGV) and hindgut (HGV) of two nauplii per group were estimated on two-dimensional histological serial images by applying the Cavalieri method (Howard and Reed 1998; Casteleyn et al. 2007). Ten uniform random transverse sections of the mid- and hindgut of each serially sectioned specimen, separated by a fixed interval (T), were selected randomly by randomizing the position of the first section in the interval (Fig. 1) (Casteleyn et al. 2010). A point grid with a known fixed area associated with each cross point (a/p) was uniform randomly placed on the histological sections using the software Cell*F (Olympus Belgium) (Fig. 2). The gut epithelium together with its brush border (when present) and the underlying muscle layer were taken into account. The grid points (p) which hit these tissues were counted and the total area of the tissues per section (Ai: area of the ith section) were calculated by multiplying the area per point (a/p) with the total number of points counted per section (Pi: number of points hitting the tissues of the digestive tract on the ith section).

Figure 1.

 Three-dimensional image of the midgut of an Artemia nauplius, embedded in a paraffin block. S4–S10 represent the selected sections, a fixed slice interval T apart (from four to ten) for stereology.

Figure 2.

 Transverse section through the midgut of a 2-day-old Artemia nauplius showing the points of the grid hitting the tissue of interest that are counted (crosses). The squares represent the area associated with each grid point.

Then, the total volume (V) of the digestive tract of the examined nauplius was calculated by multiplying the sum of all areas by the section interval (T).


The precision of the volume estimation of each mid- and hindgut was obtained by calculating the coefficient of error (CE) for the Cavalieri method developed by Gundersen and Jensen (1987). The lower the value, the more precise the volume estimation is. Coefficient of error of 10% or less was considered as precise.

Additionally, the ratio of the digestive tract volume and the total body length (IL) was calculated.

Lengths of the midgut (MGL) and hindgut (HGL) were calculated using histological sections. MGL and HGL of each nauplius were calculated using the total number of sections of the mid- or hindgut (n) and the thickness of the sections (t).

Length of the gut segment = n × t

Three-dimensional (3D) reconstruction

Preparation of digital images

All serial histological sections of the same nauplii used for stereology were utilized for 3D reconstruction. All sections were photographed using the objective lens 20× (1·0 μm × 1·0 pixel size) or 40× (0·5 μm × 0·5 μm pixel size) depending on the size of the section.

All images of each nauplius were saved in jpeg format in two separate folders according to the magnification. The last image in each folder was obtained from the same section as the first picture of the next folder.

Loading, registration and segmentation of digital images for three-dimensional reconstruction

A plain text file for every folder was created to load the bricks of pictures in the stacked slice file format into the Amira 4.0.1 (Visage Imaging GmbH, Berlin, Germany) application, following the detailed information described by Cornillie et al. (2008).

All the slices within a brick were aligned, using an automated module, followed by manual correction whenever necessary (Fig. 3a, b). Thereafter, every pixel of each picture was assigned a grey-tone value based on channel 1 (green). Important structures on these images such as the contours of the nauplii, foregut, midgut and the hindgut of every four sections were labelled manually with the brush (for internal structures) and lasso (for external surfaces) tools in the segmentation editor. The sections in-between were subsequently labelled through the interpolation command.

Figure 3.

 (a, b) Alignment of two consecutive slices of an Artemia nauplius without using external references of landmarks. The colour image of the first section is superimposed to the negative colour image of the next slice.


The voxel size of the two bricks of labelled sections were down-sampled by a factor 2 in each direction (x, y and z) using the resampling module. Resampling facilitated saving memory enabling fast surface generation.

Surface generation

Surface of the body, foregut, midgut and the hindgut were generated using SurfaceGen module. Two separate 3D images were created from two bricks of labelled sections and constrained smoothing was performed. 3D surfaces of two bricks of each nauplius were aligned manually using the transform editor of the Amira program.

Length and volume measurement

Lengths of the total body, midgut and hindgut were measured on the generated 3D image (Fig. 4), whereas volumes of the segmented serial sections were measured using the tissue statistics of the Amira program. Finally, the results acquired by 3D measurements were compared with the results obtained by stereology.

Figure 4.

 Measurement of the lengths of the midgut and hindgut using the 3D measurement tool.

Statistical analysis

The ratio between the lengths of the digestive tract segments and the individual length, and the ratio between the volumes of the digestive tract segments and the individual length of Artemia nauplii subjected to both treatments were compared separately using a Tukey HSD test, using Statistica 7 (Tulsa, OK). Each statistical analysis was tested at a 0·05 level of probability.


Three-dimensional reconstructions facilitated the estimation of the topography and the evaluation of the architectural organization of the digestive tract of Artemia nauplii. The alimentary tract of A. franciscana nauplii is freely suspended in haemolymph and composed of three clearly distinguishable segments, i.e. the foregut, the midgut and the hindgut. The midgut and the hindgut are aligned, whereas the foregut is a hooked tube with a shorter part perpendicular to the midgut and the remaining longer part is nearly parallel to it. The fore- and hindgut are shorter and have a smaller diameter when compared to the midgut. Two lateral globular protrusions from the anterior part of the midgut, called gastric or hepatopancreatic caeca, can be identified (Fig. 4).

The lengths and the volumes of the gut segments were measured using both three-dimensional reconstructions and stereology, and ratios were calculated in relation to the individual length. Using both methods, approximately similar results were obtained for the lengths and volumes of the segments of the gut and for the total body length measured on both days 2 and 4 (Figs 5 and 6).

Figure 5.

 Individual length, midgut length and hindgut length of Artemia nauplii obtained by 2D sections and 3D reconstruction (n = 2). (inline image) Total body length (using 2D sections); (inline image) total body length (using 3D reconstruction); (inline image) MGL (using 2D sections); (inline image) MGL (using 3D reconstruction); (inline image) HGL (using 2D sections) and (inline image) HGL (using 3D reconstruction).

Figure 6.

 Volumes of the midgut and hindgut of Artemia nauplii measured by stereology and 3D reconstruction (n = 2). (inline image) MGV (using stereology); (inline image) MGV (using 3D reconstruction); (inline image) HGV (using stereology) and (inline image) HGV (using 3D reconstruction).

Larger length and volume ratios (although not significant) between gut dimensions (mid- and hindgut) and total body length were observed in nauplii which received live Aer. hydrophila when compared to the nauplii which only received dead Aer. hydrophila (Table 1). However, this was not the case for the midgut volume/individual length (MGV/IL) of 4-day-old Artemia nauplii, as measured with the 3D reconstruction.

Table 1.   Ratios of midgut and hindgut dimensions and individual length of Artemia nauplii measured by stereology and 3D reconstruction (n = 2)
  1. Data are shown as mean ± SEM. Values in the same column with same superscripts are not significantly different (< 0·05).

Day 2Germ-free0·70 ± 0·01a0·73 ± 0·05a0·24 ± 0·01a0·25 ± 0·01a1221 ± 193a1271 ± 220a111 ± 9a128 ± 13a
Gnotobiotic0·72 ± 0·02a0·75 ± 0·04a0·26 ± 0·01a0·25 ± 0·00a1431 ± 189a1555 ± 314a122 ± 12a136 ± 23a
Day 4Germ-free0·70 ± 0·07b0·75 ± 0·01b0·25 ± 0·00b0·25 ± 0·01b1490 ± 52b1760 ± 81b150 ± 24b161 ± 1b
Gnotobiotic0·73 ± 0·02b0·75 ± 0·02b0·27 ± 0·01b0·26 ± 0·00b1596 ± 55b1614 ± 57b159 ± 3b166 ± 5b


In the present study, A. franciscana was used as an aquatic model to investigate host–microbe interactions for the following reasons. Artemia is an important live feed for commercial production of fish and shellfish larvae (Sorgeloos et al. 1986). Artemia nauplii can be cultured under germ-free and gnotobiotic conditions using various types of feed sources with simple experimental equipment (Verschuere et al. 1999, 2000) and the model has a short generation time (2–3 weeks), although Artemia can live for several months under optimal conditions (Van Stappen 1996). In addition, large quantities of cysts are available in some salt lakes (Bossier et al. 2004). Finally, their small body size facilitates simple culture systems. According to Marques et al. (2005), studies of host–microbe interactions using gnotobiotic aquatic animals should consider all possible interactions, including several performance indices of the host such as survival, growth or histological development.

Comparing the dimensions of the digestive tracts of germ-free and gnotobiotic Artemia nauplii, as measured by both monitoring tools, it was observed that the length of the nauplii was larger in the presence of the putative probiotic bacterium Aer. hydrophila. This finding is in agreement with the results obtained by Marques et al. (2005). Furthermore, larger ratios, although not significant, between the various gut segments and the individual length were seen both on days 2 and 4 (except for the MGV/IL at day four obtained by 3D reconstructions) in the group which received live probiotic bacteria. This indicates that the length of the gut is relatively larger in relation to body length in gnotobiotic nauplii than in germ-free nauplii.

Larger dimensions of the digestive tract of Artemia nauplii which received live bacteria can be explained by cell kinetics. The epithelial cells and underlying muscle cells in the group which received the live bacteria may have a higher cell-proliferation rate. In parallel to our finding, Banasaz et al. (2001, 2002) and Abrams et al. (1963) found that beneficial bacteria increased mitotic rates in the intestine of rats. Also in accordance with this finding, Uribe et al. (1997) found reduced rates of proliferation in the epithelium of the intestinal crypts of Lieberkühn in germ-free mice compared with their conventionally raised counterparts. Rawls et al. (2004) described a similar situation in zebrafish. As the responses of the host to the presence of bacteria are diverse, an immunohistochemical assay is necessary to study the effects of bacteria on mitosis and apoptosis (Rekecki et al. 2009).

In modern biomedical research, stereological methods are of major importance to gain accurate quantitative information about cells, tissues, organs or organisms (Dorph-Petersen et al. 2001) and about the architecture of three-dimensional objects (Weibel 1979; Howard and Reed 1998). In this study, samples were fixed for a short period of time to minimize the possible shrinkage effect, and subsequent processing was carried out directly after fixation. No correction was made for the shrinkage, because an equal shrinkage was assumed in all specimens undergoing a similar fixation and tissue processing. However, it is advisable to correct the length and volume measurements for shrinkage when absolute values are needed (Casteleyn et al. 2010).

Modern computer-based 3D reconstruction using Amira software made it possible to evaluate and compare the architecture and topography of different structures (Cornillie et al. 2008). In the present study, 3D reconstruction resulted in useful information about the structure of the alimentary tract, which could hardly be provided by separate histological sections.

Three-dimensional reconstruction has several advantages. First, 3D images may offer more complete spatial information of an object, whereas separate 2D sections do not contain information on gradients in the axial direction (i.e. perpendicular to the image plane) (Kubínováet al. 2004). Second, users can control the degree of transparency of each structure so that overlapping structures can be viewed easily (Wang et al. 2006). Next, 3D structures provide an enormous amount of information, while the attractive nature of the models supports curiosity and long-term memorization compared with simple 2D illustrations (Ruthensteiner and Heß 2008). Finally, 3D reconstructions facilitate the efficient analysis, presentation of the microanatomy of small specimens and better demonstrate the complexity of organs in such specimens in accurate dimensions, positions and relations (Neusser et al. 2006, 2009).

While reconstructing serial sections is a difficult method of analysis (Fiala 2005), manual slice alignment and structure labelling are time-consuming (Ruthensteiner and Heß 2008), modern reconstruction computer programs, such as Amira used in the present study, largely overcome these problems.

Although stereology and 3D reconstructions render comparable results, some caution is necessary. In stereology, volumes and lengths are estimated from a data set of individual 2D sections, whereas in the applied 3D reconstructions, only the volume was measured using similar principles (tissue statistics). Lengths were measured directly from the 3D image that was calculated from the aforementioned data set. During this surface generation, processes such as smoothing and compaction may have biased the final results. However, the length measurements rendered by the applied 3D reconstructions reflected the real values of lengths of nauplii, while the lengths obtained by 2D sections can be less accurate when the nauplii are not oriented perfect perpendicular to the cutting plane.

Based on our findings, it can be concluded that both stereology and 3D reconstruction are valuable tools to study the development of Artemia nauplii raised under germ-free and gnotobiotic culture conditions.


This study was supported by a Special Research Grant, (Bijzonder Onderzoeksfonds, BOF, grant numbers B/07289/02; 05B01906) of Ghent University, Belgium awarded to the first author and by the Foundation for Scientific Research-Flanders (FWO) (Project: Probiont induced functional responses in aquatic organisms, G.0491.08). Excellent technical assistance of Lobke De Bels and Bart De Pauw is thankfully acknowledged.