Spatial separation and bidirectional trafficking of proteins using a multi-functional reporter

Background The ability to specifically label proteins within living cells can provide information about their dynamics and function. To study a membrane protein, we fused a multi-functional reporter protein, HaloTag®, to the extracellular domain of a truncated integrin. Results Using the HaloTag technology, we could study the localization, trafficking and processing of an integrin-HaloTag fusion, which we showed had cellular dynamics consistent with native integrins. By labeling live cells with different fluorescent impermeable and permeable ligands, we showed spatial separation of plasma membrane and internal pools of the integrin-HaloTag fusion, and followed these protein pools over time to study bi-directional trafficking. In addition to combining the HaloTag reporter protein with different fluorophores, we also employed an affinity tag to achieve cell capture. Conclusion The HaloTag technology was used successfully to study expression, trafficking, spatial separation and real-time translocation of an integrin-HaloTag fusion, thereby demonstrating that this technology can be a powerful tool to investigate membrane protein biology in live cells.


Background
Membrane proteins are encoded by over 25% of all sequenced open reading frames and constitute the majority of known drug targets [1]. Therefore, tools providing a greater understanding of membrane proteins may benefit cell biology research and pharmacological development [2][3][4][5]. The advance of methods for labeling proteins by genetic fusion is expanding the understanding of protein function in complex intracellular environments (see recent reviews) [6][7][8]. Current reporter proteins such as carrier proteins (i.e. peptidyl PCP or acyl ACP), tetra-cysteine tags (i.e. Fluorescein and Resorufin Arsenical Helical binders), O 6 -alkylguanine-DNA alkyltransferase (AGT), photoactivatable proteins and others (reviewed by Chapman et al) allow more flexibility than originally available with GFP [5,[9][10][11][12]. However, visualization of multiple pools of the same protein through space and time can still be technically challenging and new options could only benefit this expanding field.
The multifunctional HaloTag ® technology complements existing methods and provides new options to study spa-tial and temporal changes in different pools of a single membrane protein. In addition, it can be used to study protein topology and post-translational modification and to capture cells. The HaloTag technology is based on the formation of a covalent bond between the HaloTag reporter protein and synthetic ligands [13]. The HaloTag reporter protein is an engineered catalytically inactive derivative of a bacterial hydrolase (Figure 1a). The synthetic ligands contain two crucial components: 1) a common reactive linker that forms a covalent bond with the HaloTag protein, and 2) a functional reporter such as a fluorescent dye or an affinity handle such as biotin ( Figure  1b). HaloTag ligands have the same chloroalkane reactive linker, but differences in the functional reporter and distance of the reporter from the linker create an interchangeable labeling technology. For instance, the HaloTag ® TMR ligand is a cell permeable red-emitting ligand, but unlike some red fluorescent proteins like DsRed, it does not require tetramerization (though directed evolution has since created a monomeric red fluorescent protein) [14,15]. The green-emitting HaloTag ® Alexa Fluor ® 488 and PEG-Biotin are cell impermeable ligands. The interchangeability of a broad range of ligands permits a variety of functional studies of fusion proteins generated from a single genetic construct (Figure 1c).
We used an integrin model to assess the applicability of the HaloTag technology for observing the dynamics of membrane protein processing and trafficking. Integrins are membrane proteins that are central to cellular adhesion and migration, thereby involving them in development, inflammation, and disease [16][17][18]. Integrins are heterodimers of αand β-subunits, which typically have a large extracellular domain, a single transmembrane segment and a short cytoplasmic tail [16]. Integrins have been studied in living cells by fusing GFP to the intracellular cytoplasmic tail or to the transmembrane domain [19][20][21]. We fused the HaloTag reporter protein to the extracellular domain of a truncated human β1 integrin (β1Int-HaloTag, Figure 2a). By expressing the HaloTag reporter protein on the cell surface, we were able to use the multifunctional HaloTag technology to study a membrane protein in living cells and to capture cells.

Results
The β1Int-HaloTag fusion protein was well tolerated by multiple cell types, including mammalian cell lines and human neural stem cells [22]. Immunocytochemistry with β1 integrin and HaloTag antibodies showed that the β1Int-HaloTag fusion protein was expressed at the cell membrane in a similar pattern to endogenous β1 integrin ( Figure 2b). Fixed cells were non-permeabilized to show that the HaloTag reporter protein was localized on the cell surface.
To study membrane proteins in live cells with the HaloTag technology, we developed a fluorescent ligand which should not cross the cell membrane. To make this ligand cell impermeable, we added a negatively charged dye to the standard activated linker. To confirm that this novel ligand, HaloTag ® Alexa Fluor ® 488 (HaloTag 488), was cell impermeable, we labeled cells expressing HaloTag on the surface or only inside. Live cell imaging showed that the HaloTag 488 ligand specifically labeled the cell-surface HaloTag protein in cells stably expressing β1Int-HaloTag, but did not label the intracellular protein in cells stably expressing HaloTag fused to a nuclear localization sequence [23] (Figure 2c, d). This confirms that the novel ligand is cell impermeable and that the surface HaloTag protein fused to integrin can functionally bind ligands.
To reveal the β1Int-HaloTag protein topology and subcellular distribution, we used HaloTag ligands with a modified fluorescence protease protection (FPP) assay [24]. The FPP assay, described by Lorenz (2006), determines the topology and localization of proteins in living cells by monitoring trypsin-induced destruction of GFP attached to a protein of interest. We separately labeled surface and internal protein pools of β1Int-HaloTag in live cells with the cell impermeable fluorescent ligand, HaloTag 488, followed by the cell permeable fluorescent ligand, Halo-Tag TMR (Figure 3a). Spatial separation of protein pools is depicted by a green rim around a red interior ( Figure  3b). Trypsin exposure to live cells stripped the external HaloTag 488 ligand over time, but preserved the internal HaloTag TMR ligand (Figure 3a, b). This result shows that the HaloTag protein fused to integrin was orientated on the surface of the cell membrane, and that the multi-functional HaloTag technology can be used to determine topology of membrane proteins. In some instances, β1Int-HaloTag labeled on the surface with the HaloTag 488 ligand was internalized before trypsin addition. Unlike the surface exposed protein removed by trypsin, this recycled protein showed fluorescence protease protection (Additional Figure 1).
To reveal the β1Int-HaloTag protein subcellular localization, we combined the HaloTag technology and the permeabilization agent digitonin [24]. Cells were cotransfected with β1Int-HaloTag and GFP, and then labeled with the HaloTag TMR ligand (Figure 3c). Coexpression of the proteins is shown by the yellow overlay ( Figure 3d). Digitonin treatment permeabilized the membrane, which allowed freely floating GFP to diffuse out of the cell over time while internally bound β1Int-HaloTag was retained in the cell, presumably by cellular transport and recycling machinery (Figure 3c, d). This result shows that the HaloTag fusion did not affect normal processing of a membrane bound protein, and that the HaloTag tech- nology can be used to distinguish free floating and membrane bound proteins.

Overview of HaloTag ® Technology
Membrane proteins, such as integrins, are typically trafficked through secretory and endocytic pathways [25][26][27]. Permeabilization experiments showed β1Int-HaloTag was retained in the cell, but to confirm that the β1Int-HaloTag fusion protein was associated with intracellular transport and recycling organelles, we combined fluorescent Halo-Tag ligands to label live cells with fixed cell immunocytochemistry. To assess protein delivery to the membrane, live cells expressing β1Int-HaloTag were labeled with the HaloTag TMR ligand, and then processed for immunocytochemistry to visualize the endoplasmic reticulum (ER), ER-intermediate golgi complex (ERIGC), or golgi ( Figure   4a). To assess protein recycling from the membrane, live cells expressing β1Int-HaloTag were labeled with the HaloTag 488 ligand, and then processed for immunocytochemistry to visualize the early and late endosomes (Figure 4b). Co-localization of the β1Int-HaloTag protein with cellular transport machinery is shown by the yellow overlay, and suggests that fusing HaloTag to the truncated integrin does not alter normal protein flow through secretory and endocytic pathways.
We used the HaloTag technology and immunocytochemistry to also show that the β1Int-HaloTag protein co-localized with expected membrane proteins, such as cadherin and transferrin receptors [20,28,29]. To assess protein colocalization at the membrane, live cells expressing β1Int-Targeting HaloTag protein to the cell surface Revealing protein topology and subcellular localization Cell images were generated with Olympus FV500 confocal microscope in sequential mode using appropriate filter sets.
Tag were labeled with the HaloTag 488 ligand and a transferrin Alexa Fluor ® 594 conjugate ( Figure 4c). Colocalization of the β1Int-HaloTag protein with the transferrin receptor is shown by the yellow overlay, and suggests that the β1Int-HaloTag fusion not only internalizes through the proper cellular machinery, but also co-internalizes with expected membrane proteins.
Mature integrins at the cell membrane are glycosylated, and the internal pool is partially glycosylated as it traffics through the secretory pathway [30]. To confirm that the We also used the HaloTag technology to follow protein modification over time. Cells expressing β1Int-HaloTag were labeled sequentially with HaloTag 488 and TMR ligands, and lysate was then collected immediately or up to 12 hours after labeling ( Figure 5b). As expected, lysate from cells labeled with the HaloTag TMR ligand alone showed two protein pools (lane 1) and lysate from cells labeled with the HaloTag 488 ligand alone showed only the higher molecular weight protein pool (lane 2). Substantiating figure 5a, lysate from cells sequentially labeled with both ligands showed separation of the two protein pools at early time points (lanes 3-5). However, over time the lower band for the red internal pool shifted up, presumably as this protein arrived at the membrane in a glycosylated form (lanes 6-9) [31]. Additionally, the upper band for the green surface pool disappeared, presumably as this protein was endocytosed and degraded. The overlaid SDS-PAGE shows that the HaloTag technology can be used to track different protein pools and monitor posttranslational modifications over time.
Finally, we used the HaloTag technology and live cell imaging to visualize spatial separation and real-time translocation of β1Int-HaloTag. Cells expressing β1Int-HaloTag were labeled sequentially with HaloTag 488 and TMR ligands. Live cell imaging showed two distinct protein pools, with the surface protein labeled specifically with the HaloTag 488 ligand and the intracellular protein labeled with the HaloTag TMR ligand (Figure 5c). Reimaging 12 hours after labeling shows that the initial red cytoplasmic pool has moved to the membrane and the initial green surface pool has internalized (Figure 5d). Timelapse imaging shows real-time translocation after labeling, which begins with the green surface pool internalizing at 1 hour, continues with the red internal pool trafficking to the surface, and ends with the red surface pool internalizing (Additional Figure 2 Live cells were labeled with HaloTag PEG-Biotin ligand and then captured on streptavidin coated plates. Luciferase assay results show the specific capture of β1Int-HaloTag-expressing cells compared to HaloTag-expressing control cells (Figure 6a). Specific capture of β1Int-Halo-Tag-expressing cells was blocked when streptavidin coated plates were pre-coated with HaloTag PEG-Biotin. In addition to the in vitro luciferase assay, live cell imaging confirms that β1Int-HaloTag-expressing cells can be specifically captured using the HaloTag technology and that captured cells survive. Live cells were labeled with HaloTag PEG-Biotin ligand, captured on streptavidin coated plates, and then replated for live cell imaging. Labeling plated cells with the HaloTag TMR ligand shows the survival of specifically captured β1Int-HaloTagexpressing cells compared to control cells (Figure 6b).

Discussion
The ability to analyze proteins in their native environment is critical to developing a detailed understanding of protein processing and trafficking. The study of protein trafficking is particularly timely and valuable considering the recently reported link between disrupted protein trafficking and certain disease states [32,33]. The study of protein glycosylation patterns is also relevant to disease states, as recently reported by Lyly et al for palmitoyl protein thioesterase I related to childhood encephalopathies [34].
We showed that the HaloTag technology can be used to study the expression, topology, glyscosylation, distribution, and translocation of a vital cellular protein. This was possible in multiple cell types, including mammalian cell lines like CHO, HeLa, HEK293 and U2OS and also in Spatial and temporal separation of proteins using HaloTag Technology  By fusing the HaloTag gene to the extracelluar domain of a truncated β1 integrin, we provided proof of concept that the HaloTag reporter protein can be expressed on the cell surface and can be used to study various aspects of membrane protein biology. In addition to integrin, we have successfully used the HaloTag technology to study the biology of other membrane proteins, including glycosylphosphatidylinositol (GPI) and the GABA A receptor (data not shown). While we can not rule out subtle effects of the HaloTag reporter protein on β1 integrin, we provide strong evidence that integrin function was not affected.
Results showing that β1Int-HaloTag localized in a similar pattern to endogenous integrin, trafficked and internalized through expected intracellular machinery and followed proper post-translational glycosylation suggest that the integrin-HaloTag fusion protein maintained cellular dynamics consistent with native integrins.
To specifically study surface-displayed HaloTag fused to membrane proteins, we developed a novel HaloTag Alexa 488 ligand that we showed is cell impermeable and functional. A previous report made an extracellular GFPintegrin fusion and showed impressive focal adhesion motility in live fibroblasts [21]. However, GFP was used primarily as a fluorescent marker of the fusion protein rather than the multifunctional HaloTag-integrin fusion we used to spatially and temporally visualize integrin, to determine integrin topology and post-translational modification and to capture live cells. In addition, GFP was fused directly to the transmembrane domain with complete removal of the extracellular domain. While we truncated the integrin in our β1Int-HaloTag construct, we intentionally retained portions of the extracellular and cytoplasmic sequences and the specificity determining loop (SDL) to ensure integrin traffics to the membrane [35,36].
In addition to using the HaloTag technology to study membrane protein biology, we have applied the HaloTag technology to sort cells by fluorescent activated cell sorting FACS [37]. We were able to successfully separate β1IntHT2-expressing cells labeled with HaloTag TMR or Alexa488 ligands from non-expressing cells (data not shown). As an alternative method for cells sensitive to FACS, the HaloTag technology can also be used to sort cells by panning [38]. We showed that the technology can be used to select cells by labeling surface-expressed Halo-Tag with the PEG-Biotin ligand. Labeled cells can then be captured on a streptavidin plate for in vitro assays or live cell imaging.

Los and Wood (2006) previously described the HaloTag technology simply for imaging intracellular proteins and
Cell capture using HaoTag Technolgy b RLU only using cell permeable ligands. By expressing HaloTag on the cell surface and developing a fluorescent cell impermeable ligand, we have greatly expanded this multifunctional technology to include visualizing different pools of a membrane protein over space and time, assessing post-translational modification, and capturing cells.
In addition, we combined the HaloTag technology with FPP and digitonin permeabilization to study protein topology and distribution.
The multifunctional HaloTag technology supports many experimental procedures including immunocytochemistry, fixed and live cell imaging, SDS page, FPP and capture. We successfully used this technology to determine protein topology and subcellular localization, to capture and sort cells, and to assess protein modification. In addition, we used different colored cell impermeable and permeable HaloTag ligands to show spatial separation of membrane and internal protein pools, and real-time translocation of these protein pools in live cells.

Conclusion
We have shown that the multifunctional HaloTag technology provides the ability to separate protein pools in space and time, and could be a powerful tool to examine the trafficking and cellular biology of membrane proteins such as integrins or other proteins of interest.

HaloTag ® protein and ligands
The HaloTag reporter protein is an engineered catalytically inactive derivative of a bacterial hydrolase [13]. Replacement of the catalytic base (His) with Phe renders the HaloTag protein inactive by impairing its ability to hydrolyze the ester intermediate, leading to the formation of a stable covalent bond. The synthetic ligands contain two crucial components: 1) a common reactive linker that forms a covalent bond with the HaloTag ® protein, and 2) a functional reporter such as a fluorescent dye or an affinity handle such as biotin. HaloTag ® ligands have the same chloroalkane reactive linker, but differ in the functional reporter and distance of the reporter from the linker.

Microscopy
Cells were imaged on a confocal microscope FV500 (Olympus, Japan) using a 488 nm Ar/Kr laser line or a 543-or 633-HeNe laser line. Scanning speed and laser intensity were adjusted to avoid photobleaching of the fluorophores and damage of the cells. The microscope was equipped with microenvironmental chamber to maintain physiological conditions.

Fluorescence protease protection
To reveal the β1Int-HaloTag protein topology and subcellular localization, we used the HaloTag technology with a modified fluorescence protease assay -a new technique that determines the topological distribution of proteins in living cells by monitoring trypsin-induced destruction of GFP attached to a protein of interest [24]. To assess protein topology, HEK293 cells stably expressing the β1Int-HaloTag protein were sequentially labeled with the Halo-Tag 488 and TMR ligands. Immediately following labeling and rinsing, live cells were imaged before trypsin addition and then imaged at 50 and 450 seconds following 4 mM trypsin exposure. For additional figure 1, live cells were imaged 1 hour following labeling and rinsing, thereby some β1Int-HaloTag that had labeled on the surface with the HaloTag 488 ligand would internalize before trypsin addition. To assess subcellular localization, HeLa cells coexpressing the β1Int-HaloTag protein and GFP (Promega) were labeled with the HaloTag TMR ligand. Immediately following labeling and rinsing, live cells were imaged before digitonin addition and then imaged at 360, 440 and 590 seconds following 80 µM digitonin treatment.

Cell capture
HeLa cells were transiently co-transfected with β1Int-HaloTag, or HaloTag as a control, and humanized renilla luciferase. After 48 hours, cells were labeled with the HaloTag PEG-Biotin ligand, then rinsed, and collected with 3 mM EDTA. Cells were placed into SA-coated 96 well plates (Pierce) for 30 minutes at 37°C, then wells were rinsed to remove non-captured cells. Captured cells were lysed and luciferase activity was measured using the renilla luciferase assay, per manufacturer protocol (Promega). Plates were read on a microplate luminometer (Veritas) and RLU (relative light units) were averaged over 24 wells, with standard error of the mean. To verify specific capture, SA-coated plates were pre-blocked with the HaloTag PEG-Biotin ligand (10 µM, for 30 minutes) then rinsed before cell addition.
For live cell imaging, HEK293 cells stably expressing β1Int-HaloTag, or wildtype HEK293 as a control, were labeled with the HaloTag PEG-Biotin ligand, then rinsed, and collected with 3 mM EDTA. Cells were placed into SAcoated 96 well plates (Pierce) for 30 minutes at 37°C, then wells were rinsed to remove non-captured cells. Captured cells were removed using trypsin-EDTA (Sigma), replated on glass slides (Nalge Nunc), and labeled after 24 hours with HaloTag TMR ligand and imaged.

Glycosylation and timcourse analysis
To assess the glycosylation and location of separate protein pools in cells using the HaloTag technology we used HEK293 cells stably expressing the β1Int-HaloTag protein. Cells were labeled with the HaloTag 488 ligand alone, the TMR HaloTag ligand alone, or sequentially. Cells were then rinsed, collected in 1 × PBS with a protease inhibitor cocktail (1:100, Sigma), and lysed by fractionation. After live cell labeling and rinsing, cells were collected either immediately for the glycosylation study or at times 0-12 hours for the time course study.
Lysate was incubated with either O-or N-glycanase, using the enzymatic deglycosylation kit, per manufacturer protocol (Prozyme).
Because the HaloTag ligand is held by a stable covalent bond, the fluorescently labeled β1Int-HaloTag protein can be boiled with sample buffer and resolved by SDS-PAGE without loss of the fluorescence signal. Lysate in sample buffer (1% SDS, 10% glycerol, and 1.0 mM β-mer-captoethanol, pH 6.8) was boiled for 5 minutes and then resolved on SDS-PAGE (4-20% gradient gels, Invitrogen), with the DyLight fluorescent protein molecular weight marker (Pierce). Gels were analyzed on a fluorescence imager Typhoon 9400 (Hitachi, Japan) at an E ex /E em appropriate for Alexa Fluor ® 488 and TMR.