The VH constructs C36, C47, and DP47a were a kind gift of Dr. Winter. The VHH RE3 was isolated by a phage display screen of a VHH immune library cloned into the pHEN4 phagemid (10; Olichon, unpublished data). The sequence corresponding to RE3 was subcloned into the pHEN6 vector and this transformed into XL Blue competent cells. Transformed cells were used to inoculate Terrific Brot. The culture was induced with 1 mM IPTG when the OD₆₀₀ reached 0.5 and grown overnight at 28°C. The pellet corresponding to a 1 L culture was recovered by centrifugation and initially resuspended in 5 mL of TES buffer (0.2 M TrisHCl, pH 8.0, 0.5 mM EDTA, 0.5 M sucrose) before freezing in liquid nitrogen. It was then thawed on ice and 5 mL TES plus 5× EDTA-free protease inhibitor cocktail (Roche) were added. After 30 min incubation on ice, osmotic shock was induced by supplementing the cell suspension with 15 mL of 1:4 diluted TES buffer. After further 30 min incubation on ice, the cells were centrifuged 20 min at 30,000 g and the supernatant was recovered. NaCl and imidazole were added to a final concentration of 250 mM and 10 mM, respectively, and the fraction was purified by immobilized metal affinity chromatography (IMAC) using a HiTrap Chelating column (Amersham) and FPLC. The eluted fraction was desalted in 50 mM TrisHCl, pH 8.0, 50 mM NaCl and 10% glycerol and checked by SDS-PAGE. Its concentration was estimated after measurement of the absorbance at 280 nm.

The gel filtration experiments were performed using a Superpose 12 column (GE Healthcare) coupled to an FPLC system.

Heat purification of both RE3 and VHs was performed by heating the supernatants recovered after centrifugation of the bacterial periplasmic fraction and adding of PEG 6000 (10% final volume). Samples were heated 15 min at 70°C and thereafter cooled on ice for 20 min before pelleting the denatured proteins 11. Thermotolerant proteins were recovered from the supernatant and analyzed by SDS-PAGE.

An ELISA test was performed according to 12 using both heat-treated and untreated RE3. A VHH directed against lysozyme was used as a control. VHHs were detected using an anti-His monoclonal antibody (Qiagen) and a secondary anti-mouse HRP conjugated antibody (Amersham).

The fluorimetric assay proposed by Nominé et al. 13 has proved to be a reliable and simpler method than size exclusion chromatography for estimating the aggregation degree of proteins in solution 14. It is the ratio between the adsorbance at 280 nm (light scattering due to soluble aggregates) and that at 340 nm (specific absorbance of the accessible aromatic groups). The measurements were performed using an AB2 Luminescence Spectrometer (Aminco Bowman). Samples were excited at 280 nm and the emission spectra between 260 and 400 nm were recorded.

Standard far-UV CD spectra of the recombinant antibodies were recorded at 20°C using a Jasco J-710 spectrophotometer and cuvettes of 1 mm pathlength. 15 runs were accumulated for each sample and three independent repeats were performed to confirm the results. The spectra of the unfolded proteins were determined after having increased the temperature to 95°C using a Peltier heater and those of the refolded proteins after a further incubation of 30 min at 20°C. Capped cuvettes were used to prevent sample evaporation.

The wavelength for which the difference of ellipticity value between the folded and unfolded states was maximum was selected for progressively heating the samples and to identify the melting temperature (Tm) of the protein. The temperature was increased at a rate of 30°C/hour from 15 to 95°C and the ellipticity values were recorded at every increase of 0.2°C.

The RE3 binder was first selected after panning a llama VHH phage display library using Ran-GTP, a protein involved in mitotic spindle assembly. It was sub-cloned in pHEN6 for expression in bacterial periplasm and showed both high affinity and specificity in enzyme-linked immunosorbant assay (ELISA) (Olichon, unpublished).

The first IMAC purifications resulted in a high yield of recombinant VHH. A stability test was performed using buffers at different pH and salt concentrations. The results indicated that the stability of the RE3 VHH was strongly salt dependent, since RE3 tends to polymerize and to form progressively large aggregates in the absence of NaCl. The polymerized forms were separated by SDS-PAGE in the absence of fresh DTT (Fig. 1). The addition of 100 mM NaCl was sufficient to prevent aggregation and to obtain monomeric VHH (Fig. 1). It can be speculated that NaCl prevents hydrophobic interactions favorable to the formation of intermolecular disulfide bonds. Using such conditions, 8 mg of VHH were purified from 1 L culture, and the low aggregation index indicated that the protein was monodispersed. In the presence of NaCl, the VHH did not aggregate even when kept at room temperature for several days (data not shown).

Protein aggregates tend to precipitate faster than monodispersed proteins when heated. Therefore, tolerance to heat treatments has been proposed as a way to selectively recover native proteins from a mixture in which aggregates are present 9. As an alternative to IMAC, RE3 was purified by heating the supernatant recovered after osmotic shock of the bacterial cells for 15 min at 75°C. No significant differences in purity and yield were detected when compared to the affinity purified VHH (Fig. 1). In addition, the binding efficiency of RE3 measured by ELISA was not inhibited by the heat treatment (Fig. 2). Four further llama VHHs were recovered monodispersed in the soluble fraction after heat purification (data not shown), confirming that thermotolerance is not an exclusive feature of RE3 (See additional file 1).

Purified RE3 was used to understand what structural modifications happen during heat treatment. Figure 3 shows the CD spectra of the control protein (IMAC purified), the protein heated 15 min at 75°C, and the same heat-treated protein after 30 min on ice, indicating that VHH underwent total denaturation during the heat treatment but quickly refolded. Both gel filtration and the fluorimetric analysis confirmed its monodispersed structure (Fig. 3). The melting temperature for RE3 was determined using the CD and calculated at 65.3°C.

These results confirm that the antibody was expressed in a stable and native form in the bacterial periplasm and that its native structure was preserved during affinity purification and recovered after denaturation. However, its stability and monodispersity were strictly dependent on salt concentrations.

Binders belonging to a specific class of VH antibodies 8 have been shown to possess a large variability in their capability to refold after heat denaturation. In particular, the authors noticed a strong correlation between heat tolerance and yields of soluble antibodies expressed recombinantly in bacteria. We thought that it would be interesting to compare the heat tolerance features of RE3 with those of the described VHs.

The heat tolerant VHs C36 and C47 as well as the heat sensitive DP47a were expressed and purified from bacteria periplasm. As expected, all three constructs were expressed and a recognizable band was detectable in each total lysate. However, only C36 and C47 were recovered soluble in the supernatant after centrifugation (Fig. 4A). Since phage displayed C47 refolded into an active structure after heat denaturation 8, purification by IMAC and by heat treatment were attempted. Similar yields of homogeneous C47 were recovered using both of the purification protocols (Fig. 4B).

The effect of the heat treatment was determined as described above for VHH. The secondary structure of the heat- and IMAC-purified C47 was analyzed by CD and the aggregation of the two samples evaluated at the fluorimeter. No significant difference was detectable (aggregation indexes of 0.12 and 0.14 for IMAC and heat-purified C47, respectively; CD data in Fig. 5), indicating that heat purification of thermotolerant proteins may represent an alternative to conventional affinity chromatography.