Neuroplasticity in Honey Bee Brains: An Enhanced Micro-Computed Tomography Protocol for Precise Mushroom Body Volume Measurement

Background In insect brains, mushroom bodies are associated with memory and learning behavior. It has been demonstrated that the volume of the mushroom bodies in the brain of a worker honey bee changes during the adult stage. Changes in mushroom body volume imply high neuroplasticity in the brains and may be related to the age polyethism of honey bees. A suitable volume measurement method is needed to understand the correlation between behavioral changes and mushroom body volume changes in honey bees.


Introduction
It has been demonstrated in several studies that the mushroom bodies (MBs) volume of honey bees (Apis mellifera) changes with age (Durst et al., 1994;Farris et al., 2001;Krofczik et al., 2008;Withers et al., 1995Withers et al., , 1993)).These studies used traditional histology methods to generate two-dimensional images of honey bee brains and calculated the volume of the MBs.However, such methods usually include dissection, slicing and dehydration, which may cause structural changes and harm the brain tissue (Metscher, 2009b;Silva et al., 2015).Fixation can improve the mechanical strength of tissue, thereby minimizing damage and distortion during histological processing (Thavarajah et al., 2012).Previous studies on mammalian brain tissue have tended to use transcardial perfusion as a standard method for tissue fixation (Kasukurthi et al., 2009).Through the vascular network, the fixative quickly reached every region of the brain.Perfusion fixation prevents serious anoxia in the brain tissue and minimizes mechanical damage during the subsequent dissection (Gage et al., 2012;Kasukurthi et al., 2009).When studying insects, researchers tend to dissect the brains quickly before submerging them in fixative due to the open circulatory system.However, fresh insect brains are more vulnerable to damage during the dissection process than fixed mammalian brains.As a result, physical damage caused by dissection or slicing affects the shape and volume of insect brains.
New imaging technology may offer a chance to prevent these defects.Microcomputed tomography (micro-CT) provides a noninvasive method to generate threedimensional (3D) models of biological samples (Keklikoglou et al., 2021;Ritman, J o u r n a l P r e -p r o o f 2011; Schambach et al., 2010).With extremely high resolution, micro-CT is suitable for morphological studies of small animals, such as insects (De Paula et al., 2022;Herhold et al., 2023;Keklikoglou et al., 2021;Killiny and Brodersen, 2022;Kundrata et al., 2020;Martin-Vega et al., 2018;Shaha et al., 2013;Sombke et al., 2015;Zhao et al., 2020).Several studies have used micro-CT to study the morphology of the brains of different bees, including honey bees (Alba-Tercedor and Alba-Alejandre, 2019; Greco et al., 2012;Ribi et al., 2008), bumble bees (Rother et al., 2021;Smith et al., 2016), and mason bees (Alba-Tercedor and Bartomeus, 2016).Micro-CT images of the bumble bee brain were used to calculate the volume of each brain region and to establish the standard bumble bee brain (Rother et al., 2021;Smith et al., 2016).These studies have expanded the application of micro-CT in studying insect brain.
Previous studies often use air-dried, critical-point-dried, or chemically dried samples for micro-CT (Alba-Tercedor and Alba-Alejandre, 2019;Alba-Tercedor and Bartomeus, 2016;De Paula et al., 2022;Killiny and Brodersen, 2022;Sombke et al., 2015;Zhao et al., 2020).Because air has a much lower radiodensity than water, the contrast between air and tissue would be higher than that between water and tissue; thus, a dried sample usually yields a better CT image than a wet sample.However, airdried or critical-point-dried samples may still cause soft tissue shrinkage or distortion (Mensa et al., 2022;Rivera-Quiroz and Miller, 2022).Sombke et al. (2015) compared micro-CT images of fly brains that had soaked in ethanol, critical point dried, and chemically dried using hexamethyldisilazane. Critical-point-dried and chemically dried samples showed a higher contrast and signal-to-noise ratio but had shrinkage or disruption of nervous tissue.Shrinkage or disruption may change the morphology of the brain and affect brain volume calculation.An embedding protocol that minimizes tissue distortion and still yields high-contrast images is essential to illustrate the morphology or calculate the precise volume of insect brains.
This study presents a successful embedding protocol for honey bee brain micro-CT that prevents brain shrinkage and yields high-resolution CT images.The images allowed the identification of different neuropils in the brain, and the samples could be preserved for at least five months.Using clear, high-contrast boundaries between different brain regions in the CT image, we built 3D models of MBs and calculated their volumes.The results revealed that the MB volume changed with worker age, even within the nurse bee stage.

Materials and methods
J o u r n a l P r e -p r o o f

Honey bee sample collection
Honey bees (Apis mellifera) were raised on the campus of National Taiwan University, Taipei, Taiwan.In March 2022, a comb with age-synchronized capped worker brood cells was moved into an incubator at 33-34 °.After the bees emerged, they were marked with acrylic paint on their thoraxes and then released back to the colony.The marked honey bees were collected at ages 0, 6, and 15 days.A feeder with 40% sucrose water was placed 10 meters away from the hive to collect foragers.
Sixteen honey bee workers were used in this study.Four honey bee heads were used for micro-CT scanning for each group (ages 0, 6, and 15 days and foragers).

Micro-computed tomography sample preparation
Honey bees were anesthetized at 4°C and decapitated.Then, the mouthparts were removed using sharp blades, and the staining procedure was performed.The heads were fixed in 70% ethanol solution with 2.5% glutaraldehyde for one day at 4°C.
After fixation, the heads were stained in 70% ethanol solution with 1% phosphotungstic acid (PTA) for 25 days at room temperature and then dehydrated in an ascending ethanol series (70%, 80%, 90%, and two repetitions at 100%).After dehydration, the samples were transferred to 100% xylene for 1 day, then transferred to 1:1 xylene Permount™ (Fisher Chemical™ Permount™ Mounting Medium) solution for 1 day, finally soaked in 100% Permount™ in a 0.2 ml polymerase chain reaction tube (PCR tube).The PCR tube containing Permount™ and honey bee head was then scanned by micro-CT.
To compare Permount™ embedding with other methods, several foragers were fixed and stained through the same protocol as described above.After staining, the head was transferred to 70% ethanol for 7 days instead of an ascending ethanol series.
Then, the 70% ethanol was discarded, and the head was quickly wiped dry on the surface and placed in an empty 0.2 ml PCR tube with 70% ethanol-moistened cotton for the first CT scan.After the first CT scan, the PCR tube was opened for 6 hours to air dry the sample.After air drying, the sample was scanned by CT for a second time.

Micro-computed tomography scanning
A Bruker SkyScan 2211 (at the Industrial Technology Research Institute, Taiwan) was used to scan the bee heads at 1.5 μm/pixel with 180 degrees.The voltage was 100 kVp, and the current was 260 μA at 5.5 watts with a 0.25 mm Al filter.Image reconstruction was performed using GPUNRecon (Bruker micro-CT, Kontich, Belgium) with ring artifact correction = 15 and beam hardening correction (%) = 20.
Reconstructed cross-sections were realigned, and the region of interest was J o u r n a l P r e -p r o o f determined using Dataviewer (Bruker micro-CT, Kontich, Belgium).Volume rendering 3D visualization was displayed by CTVox (Bruker micro-CT, Kontich, Belgium).

Honey bee MB volume calculation
Micro-CT cross-section images were analyzed using the open-source software 3D Slicer (Fedorov et al., 2012).The brain (supraesophageal and suboesophageal ganglia)

Results
As In addition to the brain, other soft tissues in the head could also be observed.
The head muscles and glands (corpora cardiaca, corpora allata, pharyngeal glands, and salivary glands) were well preserved and could be identified.
J o u r n a l P r e -p r o o f We also prepared samples fixed, stained, and then CT scanned the samples before and after air drying.Fig. 3 (A) shows the CT images of a Permount™-embedded sample.Figs. 3 (B) and (C) are CT images of a honey bee head before and after air drying.After air drying, the brains were visibly shrunken, and the retina was ruptured.
The brain condition was not suitable for volume calculation.Before air drying, the shapes of the brains were well preserved, but the images showed lower contrast inside the brains compared to Permount™-embedded samples.It would be more challenging to generate 3D models of the brain with a low-contrast image.
J o u r n a l P r e -p r o o f The brain and MBs volumes of sixteen honey bees at different ages were estimated.The head widths, brain volume, absolute/normalized volume of MBs, and standard deviations are shown in Table 1.The average head widths were 4.12 ± 0.08 mm for 0-day-old bees, 4.17 ± 0.02 mm for 6-day-old bees, 4.04 ± 0.05 mm for 15day-old bees, and 4.04 ± 0.10 mm for foragers.The head widths at different ages were not significantly different (Kruskal-Wallis test, p>0.05).We also compared whether there were volume differences between the left and right MBs, and there was no significant difference (Wilcoxon signed ranks test, p>0.05).For all honey bees in this study, the volume difference between the left and right MBs was less than 4.5%.The Kruskal-Wallis test showed a significant difference in normalized MBs volumes among bees at different ages.Further comparison of normalized MBs volumes between different ages was conducted with the pairwise test (Fig. 7).Significant differences appeared between 0-day-old and 15-day-old bees (p=0.031),0-day-old bees and foragers (p<0.01), and 6-day-old bees and foragers (p=0.021).
J o u r n a l P r e -p r o o f

Embedding with Permount™
Iodine and PTA are two often used chemical stains for micro-CT.According to previous studies, PTA staining showed a better image of brain tissue and caused less shrinkage than iodine staining (Lesciotto et al., 2020;Rother et al., 2021).However, PTA penetrates soft tissue very slowly, so it takes a longer staining time.In addition, PTA does not penetrate the exoskeleton, and an opening on the head is necessary (Gutierrez et al., 2018;Metscher, 2009a).Nevertheless, PTA does not overstain, and lengthening the staining time could properly stain samples (Lesciotto et al., 2020).In J o u r n a l P r e -p r o o f our experiment, PTA mainly entered the head through the cut on the mouthpart.Most samples were not completely stained after staining for 20 days, and Supplementary Fig. 1 shows a sample that was not completely stained.If stained for 25 days, more than 70% of the samples were properly and completely stained.
Micro-CT imaging is an advantage because it is a low-disturbance method, and the embedding procedure can prevent soft tissue distortion.Without embedding, the brain would be ruptured, and the retina would separate from the exoskeleton during air drying.Distortion can be prevented if samples are preserved in ethanol.However, the contrast of the CT image would be low.Permount™ Mounting Medium was used as the embedding medium due to its lower radiodensity of X-ray compared to water, ethanol, glycerol, and epoxy resin.The lower radiodensity of the embedding medium can increase the contrast between the tissue and the embedding medium.In Fig. 3, the Permount™ embedded sample showed a higher contrast image than the sample preserved in 70% ethanol.Since the neuropils are not physically separated from other tissues, a higher contrast between brain tissues is crucial to identify different neuropils.The clear boundaries of neuropils make it possible to generate 3D models and calculate the volume.
The high viscosity of Permount™ also prevents sample movement during the time-consuming CT scanning.Moreover, Permount™ is suitable for storing samples for long periods.One of the samples was scanned in December 2021 and rescanned in May 2022.These two images showed identical results, suggesting long-term sample preservation with Permount™ (Fig. 4).This makes it possible to rescan certain areas of the brain if necessary.
The radiodensity increases as the atomic number of the material increases (Spiers, 1946).The main ingredients of Permount™ are toluene and polyterpene hydrocarbon resin, which contain only carbon and hydrogen (Fisher Scientific, 2014).Compared to ethanol and water, which contain oxygen atoms, Permount™ has a lower effective atomic number and therefore has a lower X-ray radiodensity.We assume that if samples were preserved in xylene, which also consists entirely of hydrogen and carbon, the quality of CT images would be very similar to Permount™ embedding.
However, measures to avoid movement of samples would be necessary, and long-term storage has not been tested.Ribi et al. (2008) were the first to perform CT scans of embedded honey bee brains.As the head capsules were opened from the anterior and posterior, the J o u r n a l P r e -p r o o f hypopharyngeal and salivary glands and muscles were removed.In the present study, however, the muscles and glands were preserved along with the brain in the head capsule to reduce disturbance to the brain.CT images also showed clear images of the muscles and glands in the head capsules.The mouthparts, which are relatively far from the brain, were removed to make the staining process faster.Previous studies have shown that fixation and methanol dehydration may also cause severe shrinkage of nerve tissue (Buytaert et al., 2014).In our results, no cleavage in the brain was observed, and the soft tissue of compound eyes and ocelli attached to the exoskeleton indicated no shrinkage of the brain.The relatively complete exoskeleton and connective tissue may also support the brain and minimize shrinkage during sample preparation.

Bee MBs volume measurements
For MBs volume normalization, we used head widths, not brain volume, as the standard.In contrast to soft tissue, such as the brain, the head as a complete structure is covered with an exoskeleton and has less flexibility, and the size may not change significantly during development (Nijhout et al., 2014).Previous studies have already shown a significant change in MBs volume in honey bees, and it is very likely that the total brain volume also changes over time.Thus, we believe that the width of the head is more constant than the brain volume during the adult phase of a honey bee and can reflect MBs volume changes.According to previous studies, the estimated volume of forager MBs ranged from approximately 0.12 to 0.16 mm 3 (Durst et al., 1994;Fahrbach et al., 2003;Farris et al., 2001;Gowda and Gronenberg, 2019;Mares et al., 2005;Withers et al., 1995Withers et al., , 1993)).The average MBs volume measurement in this study, 0.158 mm 3 , is within this range but close to the upper limit of the earlier measurements.Various factors can affect volume measurement, including the age of the bees, sample preparation protocols, and individual differences.For example, the average honey bee head widths measured by Gowda and Gronenberg (2019) were approximately 3.7 mm, smaller than our result (4.09 mm).A larger head most likely indicates a larger brain.
However, the width of the head was not measured or reported in most studies.As honey bee worker size is influenced by the rearing environment (Chole et al., 2019), comparing our results with previous studies requires more information for accurate size comparisons.Nevertheless, our 3D models of MBs resemble the morphology of MBs in honey bee standard brains (Brandt et al., 2005).The standard deviation of the MBs volume appeared to be larger of foragers than in the other groups.This may be due to the forager group being composed of bees of different ages, as the volume of J o u r n a l P r e -p r o o f forager MBs also changes with age and flying experience (Farris et al., 2001).

Increase in mushroom body volume
Our results show that the MBs volumes of honey bee workers increased with age, which is consistent with previous studies (Durst et al., 1994;Withers et al., 1993).
Previous research found a significant difference in MBs volume between 1-day-old workers and foragers but not between nurse bees and foragers (Farris et al., 2001;Withers et al., 1993).Durst et al. (1994) showed a significant volume difference between newly emerged bees and 11-day-old bees but did not compare 11-day-old bees with foragers.MB volume increased before honey bees turned into foragers.
However, except for 1-day-old, 2-day-old, and 11-day-old bees, no other bees of a specified age were compared their MBs volumes (Durst et al., 1994;Farris et al., 2001;Withers et al., 1993).We cannot conclude from previous studies whether the MBs volume increases gradually or in other patterns.Our results showed a significant volume difference between 6-day-old bees and foragers but no significant difference between 15-day-old bees and foragers.In a comparison between 0-day-old bees to 15day-old bees, the volumes of MBs showed significant differences.According to this result, we assumed that the volume of MBs continuously increased during the nurse bee stage.We did not find a substantial difference between 0-day-old and 6-day-old bees or between 6-day-old and 15-day-old bees, indicating that the volume increase might be gradual.

Limitations and recommendations for further research
In this study, micro-CT showed neuropils of non-dissected and embedded honey bee heads.The feature of micro-CT that does not require dissection is conducive to the preservation of brain morphology.However, due to the lack of specific stains, micro-CT is still unable to completely replace traditional methods.Micro-CT generates grayscale images of the samples, where different shades indicate different tissue densities.Different tissues might exhibit very similar density.Additional information is needed to identify different tissues.In this study, this was not an issue because honey bees serve as a model species in neuroscience.However, it could pose a problem when applied to other species.Since PTA binds to various proteins and connective tissues, PTA staining can enhance tissue contrast in the brain, but cannot specify individual tissue types.When combined with immunohistochemistry, it is possible to stain and display specific tissues using micro-CT (Kozomara and Ford, 2020).
Despite the fact that our method for sample preparation does not require J o u r n a l P r e -p r o o f dissection, a cut on the mouthpart is still necessary for PTA staining.Theoretically, a larger opening on the capsule could speed up the staining process, but would increase the risk of affecting brain morphology.In this study, we aimed to keep the opening far away from the brain, so a long staining time was required.Based on images of incompletely stained samples (Supplementary Fig. 1), the area of the brain that was stained was on the side closer to the opening, rather than the outer areas of the brain.
We speculate that PTA may not be able to penetrate the membrane outside the brain and must enter the brain through the cut on the mouthparts.Thus, creating a small opening at the front of the head might not significantly improve staining.Yet, further testing is required to confirm this speculation.
Image segmentation usually takes a lot of time and effort especially for large datasets.The recent rapid advancements in artificial intelligence (AI) technology may offer a solution to the problem.Several studies have used artificial intelligence and deep learning techniques to segment CT images, and the authors reported significant time savings in the imaging process (Abdolmanafi et al., 2022;Hamwood et al., 2021;Losel et al., 2023;Saood and Hatem, 2021).With the help of AI, the time spent on segmentation can be significantly reduced.We cannot predict the future development of AI, but for now it is still important to produce images with good quality in order to reduce the error rate of automatic segmentation.

Conclusion
In conclusion, our procedure preserves brain morphology compared to dried samples and provides better image contrast compared to samples soaking in alcohol.The embedding procedure prevented tissue distortion and helped long-term storage.The method may also be applied to soft tissues found in other insects or arthropods.
Using the embedding protocol, we successfully calculated the volume of the MBs in honey bee brains and observed that the volume of MBs increased with age.

Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationship that could be construed as a potential conflict of interest.

Author Contributions
volume and MB volume were measured.Images were first processed with a Gaussian blur image filter (σ= 1.5) to reduce noise.Segmentation and smoothing functions were used to generate the 3D model of the brain and MBs.After 3D models were constructed, the segment statistic function was used to calculate the volume of honey bee brains and MBs.The head width was measured and used for volume normalization.MB volumes were normalized using the cube of the head width as a reference.The Kruskal-Wallis test and pairwise test (IBM SPSS Statistics 25) were used to compare head widths and normalized volumes between different groups.The Wilcoxon signed-rank test (IBM SPSS Statistics 25) was used to compare the normalized volumes between the left MB and right MB.
shown in Figs. 1 and 2 and supplementary video 1, the honey bee brain was visible in the sections of the 3D CT scan of the honey bee head.Images were volume rendered, and the opacity was adjusted by CTVox.Fig. 1 shows different sections of a forager's head, and Fig. 2 shows different sections of a 6-day-old worker's head.The anterior views of varying cross sections revealing the central complex, antenna lobe, corpora cardiaca, and corpora allata are shown in (A), (B), and (C), and the lateral views of sagittal sections are shown in (D).The images also show different neuropil regions, including laminas, medullas, lobules, antenna lobes, central complex and MBs.The brain regions were approximately symmetrical, with no visible cleavage or distortion.The retina of the compound eye tightly attached to the exoskeleton of the compound eye indicated no shrinkage or distortion of the brain.The resolution was high enough to observe the lips of the MBs and even the ommatidia of the compound eyes.

Fig. 1 .
Fig. 1.Micro-CT 3D visualization sections of a forager.Different CT sections of the same forager honey bee head.Images were volume rendered and opacity adjusted by CTVox.(A), (B), and (C) are anterior-view cross sections that show different brain structures, including the retina (Re), laminar (La), medulla (Me), lobula (Lo), mushroom body (MB), central complex (CX), antenna lobe (LA), muscle (Mu), ocellus (Oc), saliva gland (SG), pharyngeal gland (PG), corpus cardiacum (CC), and corpus allatum (CA).(D) shows a lateral view of a sagittal section.J o u r n a l P r e -p r o o f

Fig. 3 .
Fig. 3. Anterior and dorsal views of a Permount™-embedded honey bee head and a 70% ethanol-preserved honey bee head before and after air drying.(A) CT images of a Permount™-embedded sample as described in the main article.(B) and (C) are CT images of a honey bee head before and after air drying.

Fig. 7 .
Fig. 7.The normalized mushroom bodies volume of workers at different ages.The normalized volume is the ratio of the actual volume to the cube of the head width.The Kruskal-Wallis test indicated that bees of different ages had significantly different normalized mushroom bodies.Pairwise tests showed significant differences between 0-day-old and 15-day-old bees (p=0.031), between 0-day-old bees and foragers (p<0.01), and between 6-day-old bees and foragers (p=0.021).The error bars represent the standard deviation of each data point.Sixteen worker honey bees were used in this study, with 4 individuals in each group.
E.C. Yang and S.J. Fu designed the experiments and wrote the manuscript.S.J. Fu collected and analyzed the data.J o u r n a l P r e -p r o o f

Table 1 .
Mushroom body volumes of honey bees at different ages