Fig. S1. Effect of physical stress on vitellogenin, and indication of excess of 40 kDa vitellogenin compared with 150 kDa vitellogenin fragment. (A) The effect of physical stress, i.e. sonication and vortexing, on full-length honeybee vitellogenin (180 kDa) was tested for investigating the possibility that the abdominal 40 kDa vitellogenin is a by-product of the sample treatment. The test was performed by vortexing and sonicating a hemolymph sample − containing 180 kDa vitellogenin only − similar to the treatment of the abdominal extract: 30 l TBS buffer containing 0.3 l hemolymph was added to 1.5 ml 20 mmol l⁻¹ HEPES buffer (pH 7) with 0.2 mol l⁻¹ NaCl and protease inhibitors (minipills; Roche). The sample was strongly vortexed and then sonicated (10 watts, 5 s, 10 times) with a Sonics Vibra Cell ultrasonic processor and a 2 mm probe on ice. The control sample (C) was not vortexed or sonicated, but was stored on ice during the treatments. No difference between the treated sample and the control can be seen. There is no indication of physical stress-induced decomposition of honeybee vitellogenin. (B) An estimation of higher intensity of the 40 kDa gel band is based on SDS-PAGE gels with fat body extracts and standard proteins of known concentration (see Materials and methods). The 150 and 40 kDa vitellogenin bands are indicated with an asterisk in the fat body protein extract (fb), and the standard bands of 150 and 50 kDa are indicated. The Precision Plus Protein Standard (Bio-Rad) contains 0.21 g l⁻¹ and 0.75 g l⁻¹ of 150 kDa and 50 kDa standard proteins, respectively. The concentrations of the other standard proteins are: 0.36 g l⁻¹ (250 kDa), 0.14 g l⁻¹ (100 kDa), 0.66 g l⁻¹ (75 kDa), 0.22 g l⁻¹ (37 kDa) and 0.8 g l⁻¹ (25 kDa).

Fig. S2. Structural dynamics of honeybee vitellogenin model at the N terminus. Root mean squared deviations of the major secondary structure elements in honeybee vitellogenin relative to inital minimized model (based on lamprey lipovitellin, PDB-ID: 1LSH) against time are shown. The calculations are based upon 10 ns long all-atom molecular dynamics simulation.

Fig. S3. Alignment of insect specific loops of vitellogenin. The alignment comprises the loop sequences missing in the honey bee vitellogenin N-sheet model (V152-E161 and F190-G208). The uppermost line of numbers corresponds to the honey bee vitellogenin sequence. Orange color refers to ultra-conserved amino acid residues. Conserved glycine and proline residues of the loops are shown as blue, and conserved charged residues are shown as red. If a species has multiple vitellogenin genes, only vitellogenin-1 is presented, if not stated otherwise after the species name. The first 21 species are insects and the last 9 species are vertebrates.