Protein purification and crystallization artifacts: The tale usually not told

Abstract

The misidentification of a protein sample, or contamination of a sample with the wrong protein, may be a potential reason for the non-reproducibility of experiments. This problem may occur in the process of heterologous overexpression and purification of recombinant proteins, as well as purification of proteins from natural sources. If the contaminated or misidentified sample is used for crystallization, in many cases the problem may not be detected until structures are determined. In the case of functional studies, the problem may not be detected for years. Here several procedures that can be successfully used for the identification of crystallized protein contaminants, including: (i) a lattice parameter search against known structures, (ii) sequence or fold identification from partially built models, and (iii) molecular replacement with common contaminants as search templates have been presented. A list of common contaminant structures to be used as alternative search models was provided. These methods were used to identify four cases of purification and crystallization artifacts. This report provides troubleshooting pointers for researchers facing difficulties in phasing or model building.

Keywords: YadF (carbonic anhydrase); YodA (metal-binding lipocalin); crystallization artifacts; protein purification artifacts; reproducibility.

Figure 1

Purity of protein samples used for crystallization estimated by SDS‐PAGE. Protein samples after Superdex200 size‐exclusion chromatography (SEC) were separated by polyacrylamide gel electrophoresis on a 12% SDS‐PAGE gel and Coomassie‐stained. As a reference, for A (line 1,2) and C (line 1) samples of the protein preparation prior to SEC are also shown. The numbers on the sides of each gel indicate the molecular weight (MW) of the markers (M) in kDa; the numbers above gels are fraction numbers from SEC. (A) The final samples obtained after the presumed purification of Sigma70 (20 kDa). All fractions were pooled together, concentrated, and used for crystallization as described in the text. The band between 20 and 25 kDA was eventually identified as YodA. (B) The final samples obtained after the purification of presumed GNAT (21 kDa—MW including the N‐terminal His‐tag).The band between 20 and 25 kDa was eventually identified as YodA (23 kDa). (C) The final samples obtained after the purification of MBP‐survivin SIX (MBP‐SIX) fusion protein (60 kDa). Fractions 2–5 were pooled together, concentrated, and used for crystallization. The amount of contamination in the 23 kDa band (eventually identified as YadF [25 kDa]) was incorrectly assumed to be too small to seriously affect the suitability of the sample for crystallization.

Figure 2

Overview of MR experiments run with diffraction data collected from crystallization artifacts: Sigma70 factor (YodA), GNAT (YodA), survivin SIX or xSix (YadF), and NYSGRC‐021790 (as a negative control). The MR method was applied to each test case, using as templates selected structures of purification artifacts listed in Table S2, Supporting Information. Because the identities of the artifact proteins were known before running MR experiments, only the structures of E. coli native proteins and two selected affinity/solubility tags were used as templates. All PDB deposits listed in Table S2, Supporting Information for a given purification artifact were tested—the R‐factors reported on the figure are mean of R‐factor values of MR experiments run for each template corresponding to a given artifact. The overall R‐factors used for success‐failure determination are calculated after 15 cycles of REFMAC refinement of the best MR solution (determined by MOLREP) for each template‐dataset pair. Successful identification of an artifact is marked with arrow and a failure of MOLREP to produce a MR solution is marked with a dot. For all templates, chain A was selected for MR.