Discussion

Heterogeneity of BetA and SeMet-BetA

The initial step and primary requirement for structure determination of a protein is to obtain it in sufficient and pure amounts. BetA and SeMet-BetA could be purified by recombinant technology using the Strep-II affinity tag, which led to a pure protein in sufficient amounts for structural studies. If the Strep-II affinity tag works for a certain system, it is preferable to the commonly used His-tag, because negatively charged cytoplasmic proteins or those containing histidine clusters are observed to bind non-specifically to the NiNTA column material. By comparison, interaction of the StrepII affinity tag with the Streptactin material is much more specific.

The purification of membrane proteins requires detergents for solubilisation, and subsequent detergent exchange is often necessary in order to obtain 3D crystals.

Initially, a detergent was selected for its efficiency in the extraction process and in keeping the bioactivity of the membrane protein. Yet, this does not always lead to well-diffracting 3D crystals. It has been observed that changes in the alkyl chain of the detergent have an influence on 3D crystallisation (Michel, 2003; Palma et al., 1999; Shinzawa-Itoh et al., 1995). Extraction and further purification results in formation of diverse protein-detergent or protein-detergent-lipid complexes.

However, the exact amount and composition of detergent and lipids being bound to the membrane protein is not easily assignable.

Variations in lipid and/or detergent content and different BetA and SeMet-BetA oligomers were detected by TLC, SEC and CN-/BN-PAGE (Figure 14, Figure 18, Figure 19, Figure 20, Figure 21 and Figure 22). The variable stoichiometry of either bound lipid, #-DDM or Cymal-5 might be one reason why the purification of a monodisperse sample was not feasible. Furthermore, a membrane protein that shows a tendency for 2D and 3D crystal growth might tend to form unspecific oligomers or aggregates during purification. Interestingly, the molecular weight ratios for higher BetP oligomers in the 4-12% CN-/BN-PAGE are all divisible by three. This implies that the different BetP oligomers must arise from trimer-trimer interactions. Another interesting feature observed in the 4-12% CN-/BN-gels is the double trimer band of BetA. Both trimers may have a qualitative or quantitative difference in lipid and detergent composition. Why this is only visible for the trimer fraction may be

explained by the high concentration of trimer fraction in the sample applied on the CN-/BN gels. Specific lipid mass-spectrometry (Ejsing et al., 2006; Tsai et al., 2007) analysis of the two different trimers, which can be cut out of the CN-/BN-gel, would give more information on the defined composition of bound detergent and lipid.

As determined by TLC (Figure 22) lipids remain bound in some BetA samples throughout the entire purification. In some cases, phospholipids have been shown to be important for stabilising solubilised membrane proteins (Hunte, 2005).

Interestingly, the 2D crystal quality of BetP was observed to be very dependent on the lipid environment (Morbach and Kramer, 2005; Schiller et al., 2006; Tsai, 2008;

Tsai et al., 2007; Tsai and Ziegler, 2005). However, addition of various lipids at the stage of affinity column purification did not result in better 3D BetP crystals (Ressl, 2006).

Lower temperature results in higher monodispersity?

Lower growth temperature during heterologous expression in E. coli cells positively influenced the monodispersity of BetA, as judged by SEC profiles in (Figure 14 and Figure 24). The profiles showed a single peak in the trimer range when BetA was purified from an overnight expression at 24°C. On the other hand, SEC profiles were very diverse when BetA derived from a three hour expression at 37°C.

There was no difference in growth medium but in the induction point. The 24°C overnight BetA expression was induced below an OD600 of 1, whereas induction for the three hour expression at 37°C was chosen at an OD600 of ~1.8.

It is known that E. coli cells vary their membrane lipid composition (Cronan, 1968), which is dependent on growth temperature (de Siervo, 1969). The major phospholipid classes of E. coli (phosphatidyl ethanolamine (PE), cardiolipin (CL) and phosphatidyl glycerol (PG)) were quantified by phosphate assays at different stages of the growth cycle at 27° and 37°C (de Siervo, 1969). The major finding was a decrease of lipid phosphorus from cultures grown at 37°C and an accumulation of phospholipids in cultures grown at 27°C. Complex quantitative changes in patterns of the PE, PG and CL lipids were observed between these two temperatures as well (de Siervo, 1969). In the stationary growth phase at 27°C, when E. coli cells are usually harvested, PG is replaced by phosphatidyl glycerol-phosphate (PGP), which introduces additional negative charges to the membrane. Studies on chill activation of

BetP revealed a change in lipid composition of C. glutamicum cells upon changed growth temperatures as well (Ozcan et al., 2007). These studies showed that an increase of growth temperature resulted in a decrease of PG. As the activity of BetP is highly dependent on the lipid composition of the membrane (Rubenhagen et al., 2000), lower growth temperature during heterologous expression in E. coli cells might result in a optimal lipid composition and hence protect fold, stability and function of BetP during purification. In this work SEC profiles were monodisperse and crystal growths derived more readily from low temperature expression than of samples from high temperature expression. Phospholipids thus play an essential role in membrane protein folding and structure stabilisation.

BetP biosynthesis at 37°C may be reconsidered because longer expression time at lower temperature may promote better insertion and folding of BetP in the membrane. Furthermore, lowering the temperature to at least 27°C can result in an accumulation of important phospholipids in the membrane. Further careful analysis is necessary to find the effect of expression time, temperature and membrane lipid composition on the resulting BetP sample quality.