Efficient Production and Purification of Recombinant Murine Kindlin-3 from Insect Cells for Biophysical Studies

Kindlins are essential coactivators, with talin, of the cell surface receptors integrins and also participate in integrin outside-in signalling, and the control of gene transcription in the cell nucleus. The kindlins are ~75 kDa multidomain proteins and bind to an NPxY motif and upstream T/S cluster of the integrin β-subunit cytoplasmic tail. The hematopoietically-important kindlin isoform, kindlin-3, is critical for platelet aggregation during thrombus formation, leukocyte rolling in response to infection and inflammation and osteoclast podocyte formation in bone resorption. Kindlin-3's role in these processes has resulted in extensive cellular and physiological studies. However, there is a need for an efficient method of acquiring high quality milligram quantities of the protein for further studies. We have developed a protocol, here described, for the efficient expression and purification of recombinant murine kindlin-3 by use of a baculovirus-driven expression system in Sf9 cells yielding sufficient amounts of high purity full-length protein to allow its biophysical characterization. The same approach could be taken in the study of the other mammalian kindlin isoforms.


Introduction
Proteins of the kindlin family are a crucial component of focal adhesion assembly, and therefore essential for complex life. Kindlins, of which there are 3 isoforms in mammals (kindlin-1, kindlin-2, and kindlin-3), are considered coactivators of the extracellular receptors integrins alongside talin 1 . Integrin-mediated cell adhesion connects the cell surface to the extracellular matrix (ECM) in higher eukaryotes. It is a critical and common process in a plethora of physiological phenomena, which include tissue integrity, embryogenesis, bone metabolism, hemostasis, and immunity. Integrin-mediated cell adhesion is activated through inside-out signal transduction via the binding of talin and kindlin to the integrin β-subunit cytoplasmic tails (CTs) at their conserved NPxY motifs. The biomedical importance of kindlin proteins extends however as far as the nucleus, where kindlin-2 has been shown in several recent reports to be involved in transcriptional control 2,3 .
Kindlins are multidomain proteins of approximately 75 kDa, hallmarked by the possession of a bipartite C-terminal FERM (4.1 band, ezrin, radixin, moesin) domain, which is interrupted by a pleckstrin homology (PH) domain at the center of its F2 subdomain 4,5 . Studies of the kindlin-2 and kindlin-3 PH domains showed that it binds to the lipid second messengers phosphatidylinositol-(3,4,5)-trisphosphate and phosphatidylinositol-(4,5)-bisphosphate [6][7][8] . However, studies of the kindlin-1 PH domain show that it binds to PtdIns(3,4,5)P 3 with a much lower affinity, which can be explained in kindlin-1 by an isoform-specific salt bridge preventing lipids from binding 9 . In addition, there is an ~100 amino acid loop inserted into the F1 domain of the kindlins that is predicted to be unfolded but binds to phosphatidylserine in the inner leaflet of the plasma membrane 10,11 . The kindlin FERM domain is considered homologous to the talin FERM domain, although the talin FERM domain does not possess a pleckstrin homology domain. Both kindlins and talin interact with NPxY motifs on the integrin β-tails via the F3 region of their FERM domain, but kindlin binds to the membrane distal motif, while talin targets the membrane proximal one [12][13][14][15][16] . Kindlins and talin both in addition possess an N-terminal F0 domain with a ubiquitin-like fold that is not found in other FERM proteins 11,17 . Studies on the F0 domain of kindlin-2 have shown that it independently binds to phosphatidylinositol-(4,5)-bisphosphate enriched membranes 17 .
The kindlins exhibit paralogue-specific tissue expression patterns and nonredundant physiological functions. Kindlin-1 is primarily expressed in the epidermis, but also to a lesser extent the colon, stomach, and kidneys; kindlin-2 is ubiquitously expressed but is concentrated in striated and smooth muscle and is the only kindlin expressed in embryonic development 4 ; and kindlin-3 is expressed in the hematopoietic tissues with the highest concentration of kindlin-3 found in megakaryocytes 18 . However, more recent studies have suggested that functional protein is expressed in endothelial tissues as well 19 .
Kindlin-3 is of acute medical interest due to its important physiological role in the blood. It is critical for platelet aggregation and spreading during thrombus formation 20 . Kindlin-3 knock-out studies in mice revealed the crucial function of the protein in cell adhesion. KIND3 -/mice display distinct phenotypes, such as severe bleeding due to inactive platelet integrins, severe osteopetrosis, and impaired leukocyte adhesion 20,22 , resembling symptoms in humans lacking kindlin-3.
High-resolution structural data on the kindlins, to date, has been restricted to individual subdomains such as the pleckstrin homology (PH) domain of kindlin-1 9 and kindlin-2 26,27 and the F0 domain of kindlin-1 11 and kindlin-2 17 . Most of the subdomains of each kindlin polypeptide have however resisted cloning and structural analysis (Yates and Gilbert, unpublished observations), and studies of the full-length proteins have been hindered by the difficulty of expressing and purifying sufficient quantities using E. coli (unpublished observations and Harburger et al. 14 ). There is considerable medical interest in kindlin-3 and its function, alongside the other two family members, and recently we generated milligram quantities of it by recombinant expression in Spodoptera frugiperda cells driven by baculovirus infection 12 . We therefore here describe methods for the production of milligram quantities of recombinant mouse kindlin-3 in insect cell culture, suitable for extensive structural studies and biochemical analysis.
In this protocol we make use of an engineered knockout bacmid (BAC10 KO:1629 ) that is, alone, unable to produce viable virions 28 . The viral DNA is thus rescued by recombination with a transfer vector that in this case also includes the kindlin-3 gene (FERMT3) and results in the FERMT3 gene replacing the virus very late gene, which is highly expressed but redundant, resulting in a recombinant virus that expresses mouse kindlin-3 as part of the virus life cycle 28 . We identified this method for the production of kindlin-3 after attempts to express and purify it in other expression hosts proved prohibitively difficult (unpublished observations) but also due to the versatility of the pOPIN vector suite, which we used for cloning and that can deployed in many expression hosts 29 .

Protocol
This protocol assumes that the mouse kindlin-3 gene (FERMT3) has been successfully cloned into a vector downstream of the very late p10 promoter and that the vector possesses flanking baculovirus sequences to permit recombination with the bacmid BAC10 KO:1629 developed by I.M. Jones and colleagues 28 . For this protocol, the kindlin-3 gene was cloned into pOPINE 29 and the primers and cloning strategy used can be found described elsewhere 12 . The plasmid is engineered so that the FERMT3 gene (kindlin-3) is under the control of the p10 baculoviral promoter and the vector contains 5′ UTR/ORF603 and ORF 1629 and encodes a C-terminal His 6 -tag for downstream purification 14. Concentrate the purified kindlin-3-CHIS 6 using a centrifuge protein concentrator with a 50 kDa MWCO to ~15 mg/ml, as assessed spectrophotometrically using a calculated extinction coefficient (ε) of 109,320 M -1 cm -1 (assuming all cysteine residues are reduced). 15. For storage at -20 °C and long term storage at -80 °C, aliquot the protein into PCR tubes and flash-freeze the samples in liquid nitrogen. Alternatively, the protein can be used directly for investigation using a number of biochemical and biophysical techniques.

Representative Results
The large scale expression of recombinant mouse kindlin-3 using baculovirus-infected Sf9 cells can take less than two weeks to achieve milligram quantities, as illustrated in the schematic Figure 1A and requires only a small quantity of plasmid DNA from a QIAprep Miniprep kit, for example. The generation of recombinant baculovirus is achieved by cotransfecting Sf9 cells with mouse kindlin-3 (FERMT3)-containing plasmid together with an engineered linearized bacmid (BAC10:KO 1629 ) and harvesting the newly formed virions after 5-7 days, as shown in Figure 1A. This method of baculovirus generation results in 100% recombinant viruses and essentially dispenses the need for plaque purification 28,30 . A representative small-scale (2 ml monolayer) culture will generate a recombinant virus-containing solution with an expected viral titer of 1 x 10 7 plaque forming units (pfu) per ml of culture. One can perform a plaque assay to determine the actual viral titer but this is perhaps too laborintensive when a high number of constructs are tested for structural studies. The success of the virus generation step can be assessed by using an eGFP-containing plasmid in parallel, or, for this kindlin-3 construct, the Sf9 monolayer can be resuspended in PBS and assessed by SDS-PAGE and western blotting, which typically demonstrates a clear band corresponding to a 75 kDa His-tagged protein, as shown in Figure 1B. The virus is subsequently amplified to generate sufficient quantities for large-scale (liter volumes) insect cell infection and recombinant protein isolation. The second passage virus (P2) is generated by infecting suspension cultures with an estimated MOI of 0.1 (see protocol). It is vital that the amplification Sf9 suspension culture occupies only a twentieth of the total flask volume. This additional aeration ensures that the resulting virus-containing media, harvested after 3 days (72 hr) post-infection, will generate sufficient quantities of kindlin-3 in the subsequent expression culture. The amplified virus stock (P2) is assumed to possess an expected viral titer of 2 x 10 8 pfu/ml based on a conservative estimate of 100 pfu/cell using a cell density of 2 x 10 6 cell/ml during the amplification 31 . Generally, for optimum baculovirus amplification and protein production, the Sf9 cells should be uniform in size and spherical, as shown in Figure 1C. Additionally, viral amplification and protein expression is maximal at 72 hr post-infection, and both are significantly reduced at 96 hr post-infection. 6 -tag using immobilized metal affinity chromatography, as shown in Figure 2. It is important to carry out the purification at 4 °C and that sufficient protease inhibitors have been added to prevent proteolysis. The partially-purified kindlin-3 is further purified to near homogeneity by ion exchange chromatography (IEC) using a heparin column, as shown in Figure 3. We employed the use of a heparin column instead of a conventional ionic exchange column, as we predicted that the large number of basic residues, including a poly-lysine stretch within the kindlin-3 F1 domain, would interact strongly with the negatively charged sulphate groups of the column. This strategy is particularly useful for DNA-and RNA-binding proteins with basic nucleic acid-binding patches, for example the RNA-binding terminal uridylyltransferase Cid1 32 . Finally kindlin-3 purity is "polished" by size exclusion chromatography to remove aggregates and achieve homogeneity, as shown in Figure 4. The buffers specified in the protocols are standard buffers frequently used in protein purification for structural analysis. Generally, phosphate-based buffers are avoided for the purification of proteins for structural studies, especially crystallization screening due to the formation of phosphate crystals in the crystallization drops (particularly for experiments at 4 °C). However, we performed a Thermofluor-based thermal shift assay to determine which buffers were stabilizing for kindlin-3, as shown in Figure 5. Briefly, a purified protein solution is diluted into buffers that cover a range of pH and sodium chloride concentrations, therefore forming a 2-dimensional screen. The melting of the protein is measured by observing the fluorescence from Sypro orange dye (molecular probes), that binds to hydrophobic residues within the folded protein core, over the temperature range 20-95 °C (293-368 K). The temperature mid-point at which the protein unfolds (transition temperature, T m ) was calculated using the Opticon Monitor software and is described elsewhere 33 . Kindlin-3 was observed to be stable at high sodium chloride concentrations (500 mM) within the pH range 7.0-9.0, with a consistent transition temperature (T m ) of 55 °C. It was also observed that the T m of Kindlin-3 was approximately 55 °C within the pH range 7.0-7.5 irrespective of sodium chloride concentration.

Once virus has been amplified and used to infect Sf9 cells in large scale experiments recombinant kindlin-3 is partially purified by virtue of its engineered C-terminal His
We found that there was limited proteolysis of kindlin-3 during the purification but SDS-PAGE analysis of highly concentrated protein (~15 mg/ ml) demonstrated limited contamination with two additional polypeptides of equal intensity. Oddly, however there is no indication of additional species by size exclusion chromatography, as shown in Figure 4. It is thus thought that the protein is nicked by proteases but remains folded and hydrodynamically indistinguishable from full-length protein. Interestingly, Calpain is a known protease that cleaves kindlin-3 at Tyrosine373, which is in the β1-β2 loop of the pleckstrin homology (PH) domain 34 and may explain our observations. Furthermore, western blotting reveals that one of the polypeptide doublets possesses a C-terminal His-tag and the apparent molecular weight of the doublet, when summed, equals 75 kDa, the same molecular weight as the native protein. The yield of purified recombinant kindlin-3 per liter of Sf9 cells (~2 g cell weight) is, at best, 5 mg.
The baculoviridae infect insect cells and for recombinant protein expression the baculovirus used in this work is based upon the Autographa californica nuclear polyhedrosis virus (AcNPV). In nature the AcNPV, which infects the Autographa californica (alfalfa lopper) insect larvae, requires the polyhedrin protein to form occlusions whereby the virons are encapsulated in a crystalline protein matrix thus providing the necessary protection for their release. In cultured cells the formation of occlusion bodies is not required for replication and is therefore dispensable. In the case of expressing foreign proteins the polyhedrin protein gene can be replaced in a recombinant AcNPV with the gene for the protein of interest. AcNPV can infect other Lepidopteron species and for the purposes of recombinant protein expression army worm Spodoptera frugiperda pupal ovary cells are used. In the approach outlined here, the AcNPV bacmid (BAC10) is engineered so that an essential viral gene, ORF1629, is inactivated by the insertion of chloramphenicol acetyl transferase resulting in a knock-out bacmid (BAC10:KO 1629 ), such that it is unable to form infectious baculovirions 28 . Cotransfection of Sf9 cells with linearized BAC10:KO 1629 and the FERMT3-containing transfer vector repairs the inactive ORF1629, through transposition, resulting in a viable genome that has also incorporated the FERMT3 gene under the control of the polyhedrin promoter 28 .
We describe a purification protocol for the isolation of highly pure recombinant mouse kindlin-3 via a three step chromatographic approach. The methods used here could easily be applied to other His-tagged proteins. We employed an ion exchange step to further purify kindlin-3 but we believe that this is a further pseudo-affinity step as kindlin-3 possesses a large number of basic residues, including a poly-lysine stretch within its F1 domain. Additionally, kindlin-3 is considered to bind to and interact with the cytoplasmic face of the plasma membrane, where it functions, and therefore we predict that the clustering of basic residues will allow the protein to counter the negatively charged membrane.
The buffers described in the purification protocols are considered standard and are frequently used in structural biology. The thermofluor assay ( Figure 5) demonstrates that kindlin-3 is stable in most buffer conditions above pH 6.0. This was particularly useful and important for informing our experiments when studying kinldin-3:β1 A tail interaction by NMR, which yielded excellent spectra at pH 6.1 with low concentrations of NaCl 12 .
Before any biophysical study can be undertaken, it is important to demonstrate that the purified protein of interest is indeed correctly folded and is functionally active. In a previous publication we demonstrated that the recombinant kindlin-3 expressed and purified using this method was a monomer and monodispersed in solution, as assessed by size-exclusion chromatography, dynamic light scattering, analytical ultracentrifugation and small angle X-ray scattering, and was also capable of binding and recognizing the membrane-distal NPxY and upstream Serine/Threonine cluster of β 1A cytoplasmic tails 14 , thus confirming that it behaves like the native protein, which is in line with previous cellular and physiological studies 14,20,22 . The use of a thermal stability assay is an additional way of suggesting the proper folding of a protein of interest, as incorrectly folded protein will result in a high fluorescence background due to exposed hydrophobic residues.
The kindlin family of proteins has been the focus of much attention since their unexpected role as essential coactivators of integrins in vivo was discovered. This has triggered much effort to express them recombinantly and resolve their structures. To date limited success has been reported in expressing milligram quantities of full length recombinant protein but we have here described the use of a baculovirus system that permits large scale expression at levels where structural studies become feasible. By generating large quantities of recombinant kindlin-3 we anticipate that this will aid further studies of this protein. The baculovirus-driven method and purification workflow described here for recombinant murine kindlin-3 could also be used to express and purify the other kindlin isoforms, which are also difficult to express and also possess polylysine stretches, and may be further adapted for other cytoplasmic proteins, such as nucleic acid binding proteins, that fail to express in bacterial strains.

Disclosures
The authors have nothing to disclose.