Herein, we describe an intein-based system for conjugation of QDs to target proteins in vivo. This approach has several advantages over existing methodologies that make it truly unique, including i) site-specificity (N- or C-terminus), ii) low-intrinsic reactivity towards endogenous proteins which do not contain the intein motif required for splicing, thus eliminating mis-targeting of the QDs, iii) versatility conferred by the ability to target QDs to a single protein within any cellular compartment or molecular complex and iv) the ability to target spectrally resolvable QDs to multiple protein targets simultaneously without cross reactivity.
We have previously shown site-specific conjugation of QDs to the C-terminus of target proteins by using the naturally-split DnaE intein . However, C-terminal protein labelling with QDs can in some cases, interfere with protein localization and/or biological function, as can C-terminal fusion of fluorescent proteins [49–51]. This is due to interference with protein sorting or targeting signals located at the C-terminus of proteins, such as two common ER retrieval signals, the dilysine motif and the tetrapeptide KDEL, as well as the type 1 peroxisomal targeting signal peptide SKL . A C-terminal tag or marker could also disrupt signals for the incorporation of lipid anchors. For example, many members of the Ras superfamily carry sequences that signal the attachment of lipid anchors at their C-termini . A class of plasma membrane proteins, including cell adhesion molecules or receptors have a glycosylphosphatidylinositol (GPI) linker . The molecular signals engaging the lipid modification enzyme complexes reside at the C-terminus of these proteins and would definitely be disrupted by the addition of a fluorescent protein or QD. We therefore took advantage of the artificially split Ssp DnaB intein originally described by Sun, W. et al. , for site-specific conjugation of QDs to the N-terminus of target proteins. Ssp DnaB intein has been split artificially at a site (S1) proximal to the N- terminal, producing an N-terminal piece of only 11 aa in length and a C-terminal piece of 144 aa in length . This novel artificially split intein is quite useful due to the ease of chemical peptide synthesis and due to the fact that such short peptides are not prone to misfolding. We used the S1 split intein for site-specific conjugation of QDs to the N-terminus of a model target protein in vivo, namely mem-EGFP, and have shown that QD-memEGFP conjugates localize to the cell membrane and can be monitored in real time within the developing Xenopus embryo (Figure 5). Thus, the ability to target QDs to the N-terminus of proteins is very helpful for bioimaging studies aiming at determining protein localization and function, given that there are numerous proteins bearing C-terminal post-translational modifications or a C-terminal critical domain whose function would be impeded if a bulky QD was conjugated at the C-terminus.
We have also demonstrated, using this methodology, that Quantum Dots can be targeted via paxillin to focal adhesions, a specific molecular complex, for the first time. Focal Adhesions (FAs) are comprised of α and β integrin heterodimers that form a bridge between the intracellular actin cytoskeleton and the extracellular matrix (ECM) . While the extracellular domain of integrins binds directly to ECM proteins, the cytoplasmic tail is linked to the actin cytoskeleton via signalling and adapter proteins, such as focal adhesion kinase (FAK), vinculin, talin and paxillin . FAs play a crucial role in cell adhesion, spreading and motility by regulating various signal transduction pathways leading to rearrangement of the actin cytoskeleton [53, 54]. We have demonstrated that QDs can be efficiently targeted to focal adhesions via paxillin without altering protein localization and/or function. In fact Paxillin-QD conjugates retained full functionality as indicated by their ability to i) translocate to focal adhesions at the cell membrane (Figure 2A) and ii) associate with newly formed focal adhesion complexes and be released once the complexes were disassembled (Figure 3). This is an inherent advantage of QDs over fluorescent proteins since the former are conjugated to target proteins post-translationally and do not therefore interfere with protein folding and tertiary structure.
A useful additional application of this intein-based methodology is the simultaneous and specific conjugation of QDs to multiple proteins targets in vivo. Although fluorescent proteins already provide a straightforward solution to this problem . QD-conjugation methods are attractive complements given the superior optical properties of QDs over fluorescent proteins . Double in vivo labeling becomes possible with our system due to the existence of orthogonal pairs of split inteins that do not cross splice and therefore allow different protein targets to be simultaneously and specifically tagged with spectrally resolvable QDs within the cell. Such orthogonal split-intein combinations include Ssp DnaE and Sce VMA, Ssp DnaB and Sce VMA, Ssp DnaB and Mxe GyrA  to mention but a few and now Ssp DnaE and Ssp DnaB. In fact, given the large number of characterized split inteins, the number of individual targets that can be simultaneously tagged is only limited by the number of QDs that can be spectrally distinguished. Moreover, the fact that the trans-splicing reactions proceed with an identical molecular mechanism ensures similar reaction rates for QD-conjugation that would aid the comparison of the properties of the two proteins-otherwise the first protein of interest is already redistributing while the second protein is not yet sufficiently labelled. We have shown in this work that Ssp DnaE and Ssp DnaB inteins do not cross splice and may therefore be used to specifically target spectrally resolvable QDs to different proteins simultaneously in vivo (Figure 7).
Despite the successful conjugation of QDs to both the N and C terminus of target proteins, the current methodology and the materials used have certain limitations that need to be noted. We have observed that a pool of QDs remains in the cytosol, even when the target protein is in excess. This was expected in the case of paxillin, a cytosolic protein occasionally localized to the focal adhesion complexes on the cell membrane, but came as a surprise in the case of memEGFP, a protein expected to be exclusively localized on the cell membrane. An unwanted result of the presence of free QDs in the cytosol was the reduction of signal to noise ratio. These QDs are most likely not conjugated to the target protein due to the following two reasons. Firstly the commercially available solution of Streptavidin-coated QDs used in these experiments, contains both streptavidin-conjugated and free QDs (see Figure 6). This implies that even if the splicing reaction is 100% efficient, a portion of free QDs is still present in the cell. Secondly, in the Xenopus model, translation begins after the Midblastula Transition (~12 hours post injection). By that time, a portion of the streptavidin-conjugated QDs may have lost the streptavidin or the intein peptide (due to proteolytic degradation). This, in effect, generates additional free QDs, which will remain in the cytosol, thus reducing the apparent conjugation efficiency. Given that as the embryo develops, the amount of conjugated QDs is progressively reduced and given the target proteins' degradation rate, it is important to note that the time frame for imaging can be quite small. In addition, the presence of free QDs in the cytosol greatly impedes visualization of target proteins that do not localize to a specific organelle or structure in the cell, even early on. These limitations raise the need for i) commercially available QDs capable of retaining their conjugated biomolecule longer and ii) improved methodologies to ensure that the starting material consist of 100% conjugated QDs.
Our present results indicate efficient, covalent and site-specific in vivo-fusion of QDs to either the N- or C- terminus of a target protein within any cellular compartment or molecular complex. This methodology is notable due to its potential diagnostic and therapeutic applications, as it makes the targeting of nanostructures and nanodevices to different intracellular compartments and signalling complexes a viable possibility. Furthermore, this method is unique in that it facilitates QD conjugation to multiple target proteins, as long as orthogonal intein pairs are used. The number of potential applications for double (or multiple) in vivo labelling is quite large. Most obviously, protein localizations of two or more species can be followed simultaneously and protein-protein interactions may be explored using QDs suited for FRET experiments. In conclusion the intein-mediated approach for simultaneous, in vivo, site-specific (N- and C-terminus) conjugation of Quantum Dots to multiple protein targets, should serve as a powerful tool for bioimaging applications.