**1. Introduction**

Honey is a highly concentrated mixture of mainly two dissolved sugars, fructose and glucose, plus at least 22 other composite sugars [1] and 70 other compounds including proteins, vitamins, minerals, organic acids, aromatic compounds, and various derivatives of chlorophyil [2]. Many more honey components may remain undiscovered. Therefore, identifying potential fraudulent honeys by component analyses may be di fficult unless specific breakdown products (metabolites) can be identified such as substances like hydroxymethylfurfural, or 5-hydroxymethyl-2-furaldehyde (C6H6O3) (HMF). HMF can be selectively produced from keto-hexose—notably from D-fructose [3] and other acidic media containing dissolved monosaccharides [4]. Normally, HMF is present in honey in trace amounts [5]. The rate of HMF formation in foods depends on environmental temperature, the type of sugar, pH, and the concentration of divalent cations in the medium [6,7]. Excessive heating or inappropriate storage conditions can increase HMF levels, which are recognized as a marker of quality deterioration for a wide range of foods containing carbohydrates [8]. Inappropriate heat processing of honey a ffects honey fermentation and reduces honey quality [9]. In fresh honey, HMF can occur at concentrations as high as 15 mg HMF/kg, but it normally occurs at levels between 0.06–0.2 mg HMF/kg [5]. For the most part, HMF is naturally present in honey, and at low concentrations (e.g., ~100–500 ppm) it does not reduce honey quality; it could thus be used as an identifier of a honey's origin and quality. The Codex Alimentarius of the World Health Organization (WHO) and the European Union (EU Directive 110/2001) have defined a maximum HMF quantity level in heat-treated honeys (40 mg HMF/kg) above which honey quality begins to deteriorate. HMF concentration increases above 20 ◦C. Temperatures inside a hive normally exceed 20 ◦C (~28–30 ◦C) and in summer can reach as high as 40 ◦C or more, when

the concentration of HMF can reach 10 mg/kg of honey [10], a level one-third of that known to be harmless to bees (30 mg/kg) [11]. Although these high levels of HMF are considered nontoxic to bees, few studies actually confirm a safe level of HMF in honey bee colonies [12]. Concentrations of HMF <10–15 mg/kg in honey pose little risk to honey bees, but toxic concentrations of HMF seems to induce lethal intestinal tract ulceration [12]. About 150 mg HMF/kg of commercially acid-hydrolyzed inverted sugar syrup can cause 50% bee mortality within 16 days [11]. The HMF concentration in inverted syrup for feeding bees should not exceed 20 mg/kg, as in most honeys [13].

There is no standard limit value of HMF in bee nourishment. Approximately 250 ppm HMF in the honey bee diet is considered toxic [14]. High concentrations of HMF in stored honey could represent a factor in the early death of bees and in the extinction of honey bee colonies [15]. It is therefore important to understand the potential adverse e ffects of high HMF doses on honey bees. Thus, the objective of this study was to determine the toxic e ffects and mode of action of HMF on caged bees fed in laboratory assays. We also used an immunohistochemical assay to examine the impact of HMF toxicity on the cellular death of epithelial cells lining the worker midgut.

#### **2. Materials and Methods**

## *2.1. Toxicological Tests*

Toxicological tests on the autochthonous Carniolan honey bee *Apis mellifera carnica* occurred in laboratory conditions. Worker bees were maintained in incubators set at a near-constant 28 ◦C and at ~65% relative humidity (RH), according to normal temperatures outside the brood nest [16,17]. This was to simulate the environment outside the brood nest, where we normally install sugar patties in honey bee colonies.

Plastic ~8 cm (H) × ~12 cm (Dia.) "cake" boxes originally designed to hold up to 100 CDs were repurposed as bee containment units (test chambers) by drilling ~80 circular ventilation holes, each ~2 mm wide, into the top cover. Similar experimental plastic box approaches have been used in previous experiments [18–20]. Two additional holes of 12 mm diameter were added as placeholders for plastic feeding tubes, which provided bees with Apifonda sugar candy and water, respectively. Apifonda contains sucrose, glucose syrup, and invert sugar syrup and corresponds to the nutritional value of 0.9 kg crystalline sugar (Südzucker Sugar, Mannheim, Germany). To simulate a colony habitat, each containment unit contained a 4 cm × 5 cm piece of wax foundation. We divided the bees into five groups. Each group of 70 bees was placed into its own individual containment unit. We provided thoroughly homogenized Apifonda as bee nourishment, into which was added HMF (5-hydroxymethyl-2-furaldehyde, Sigma Aldrich) at concentrations of 0 mg HMF/kg candy (control), 100 mg HMF/kg, 500 mg HMF/kg, 1000 mg HMF/kg, and 1500 mg HMF/kg. We took the capped brood from three clinically healthy honey bee colonies a day before starting the experiment and placed combs into the incubator at 34.5 ◦C and 65% RH. The next day, newly emerged 0–24 h old bees were placed into the containment units. Each HMF-treated group was replicated five times. We recorded the daily food consumption and daily mortality rates of the confined worker bees.

Dead bees exposed to the control and HMF concentrations were counted and numbers were deducted from the initial bee population (*n* = 70). The survival rate was calculated as 100% minus mortality per control or treatment groups. Data analyses were performed using ANOVA (analysis of variance) in Statgraphic [21]. Mean bee survival rates were compared among the treatment groups with a one-way ANOVA and mean separation was accomplished with Tukey tests.

## *2.2. Immunohistochemical Analyses*

We established 5 groups of 50 bees. Brood combs were obtained from clinically healthy honey bee colonies in the same way as described for the toxicological tests. Newly emerged 0–24 h old bees were placed into the containment units. Bees in the first group received no HMF added to the Apifonda candy. Group 2 received Apifonda containing 100 mg/kg of HMF. Group 3 received

Apifonda containing 500 mg/kg of HMF. Group 4 received Apifonda containing 1000 mg/kg of HMF. Group 5 received Apifonda containing 1500 mg /kg of HMF. Three bees from each of these treatment groups were randomly sampled at 5 day intervals: on day 5, 10, 15, and 20. Sampled bees were anesthetized by subjecting them to cold for about 10 min, and their midgut was removed. Midguts were fixed in 10% neutral buffered formalin, dehydrated in ethanol, and embedded in paraffin wax, which was sliced into 5 μm sections that were then de-paraffinized and processed following the instructions provided with the In Situ Cell Death Detection Kit AP' (ISCDDK) (Roche, Mannheim, Germany). The EnVision System alkaline phosphatase kit (Dako) was used to obtain a red-colored precipitate in the sections treated with the ISCDDK assay reagents. The sections were counterstained with hematoxylin. TUNEL-positive cells possessed red nuclei, which indicate the reaction products of cell death. TUNEL-negative nuclei of healthy, intact cells appeared blue. A control labeling of midgut tissue was accomplished by substituting the deoxynucleotidyl transferase (TdT) enzyme with phosphate-buffered saline (PBS) during the TUNEL reaction. Sections were mounted in Faramount aqueous mounting medium (Dako). Slide contents were analyzed and digitally photo-documented with a bright field light microscope at 400× magnification. We repeated this immunohistochemical experiment twice.

#### *2.3. Semi-Quantitative Analysis of Cell Death*

TUNEL-labeled tissue slides were used for the quantification of cell type and cell death using ISCDDK. For each experimental group of bees, approximately 300 total cells from each individual (three bees at different collection times per group) were counted in random fields on different slides. The results were expressed as the proportion of cells with positive staining. To confirm reproducibility, 25% of the slides were chosen randomly and scored twice [22,23].

#### *2.4. Nosema Ceranae Spore Counts*

To quantify *N. ceranae* infection in sampled bees, we temporarily stored dead bees from separate containment units in a freezer. All dead bees in consecutive days from experimental cages were sampled and examined for *N. ceranae* spores. Nosema spores on adult bees were potentially derived from combs where bees emerged; we assumed that spores were equally distributed between bees. Altogether, 350 bees were examined microscopically for *N. cerane* spores across all treatment groups. Species determination of *N. ceranae* in these frozen bee samples was confirmed using multiplex PCR [24]. Spore counts were made using a Bürker hemocytometer and a Zeiss light microscope under phase contrast (Axioskop 2 Plus, Zeiss, Jena, Germany). Spore samples were extracted from a bee's detached abdomen by grinding the tissue in 1 mL water with a mortar and pestle. A drop (~60 μL) of this homogenous suspension was applied to a hemocytometer and counting was conducted a few minutes later when spores fully settled. The average number of *N. ceranae* spores was calculated after counting the number of spores in all four outer squares divided by four for each dissected bee.
