Determination of the membrane topology of lemur tyrosine kinase 2 (LMTK2) by fluorescence protease protection

Alexander Nixon, Ying Jia, Carl White, Neil A. Bradbury


Lemur tyrosine kinase 2 (LMTK2) is a novel membrane-anchored kinase reported to be involved in several normal and pathophysiological conditions, including endosomal membrane recycling, prostate cancer, and neurodegeneration. In this study, we have investigated the topology and orientation of LMTK2 within cellular membranes utilizing fluorescence protease protection. Appending the green fluorescent protein to either the amino or carboxyl terminus of LMTK2, we were able to determine which side of intracellular membrane these regions were located. Our results indicate that LMTK2 is an integral membrane protein in which both the amino and carboxyl termini are exposed to the cytoplasm. Moreover, this topology places the kinase active site within the cytoplasm.

  • LMTK2
  • BREK
  • topology
  • membrane protein
  • fluorescence

lemur tyrosine kinase 2 (lmtk2), also known as brain-enriched kinase (BREK), apoptosis-associated tyrosine kinase (AATYK-2), cyclin-dependent kinase 5 (cdk5)/p35-regulated kinase, kinase/phosphatase inhibitor 2 (KPI-2), KIAA1079, and Cprk, is a member of the lemur membrane associated kinase family (9, 20). Homology analyses proposed that LMTK2 was a tyrosine kinase; however, several studies have now shown that LMTK2 selectively phosphorylates serine and threonine residues (9, 21). Although the functions of LMTK2 are not completely understood, initial studies have implicated LMTK2 in intracellular transport and trafficking, endosomal recycling (4, 8, 9), and neurodegeneration (1, 14). Moreover, LMTK2 has recently been identified as a major susceptibility gene for prostate cancer (6, 7), although the role LMTK2 plays in prostate cancer initiation and progression remains elusive. LMTK2 knockout mice are viable; however, initial studies have shown male infertility due to defects in spermatogenesis (10); a thorough analysis of the LMTK2 knockout mouse remains to be performed. While several groups are beginning to determine the function(s) of this novel membrane-associated kinase, little attention has been directed towards any structural information. Such information is important, since the correct topology and orientation of integral membrane proteins is essential for their proper function. LMTK2 is a membrane protein, yet it is not highly expressed in the plasma membrane but rather is present on intracellular membranes particularly those of the endosomal and recycling pathways. Hydropathy analysis and early studies on LMTK2 suggest that LMTK2 has a short soluble amino-terminal domain followed by two hydrophobic transmembrane (TM) helices between residues 11–29 and 46–53 (20) and a large soluble carboxyl domain harboring the kinase catalytic site. However, at present there are no formal data addressing the topology and orientation of LMTK2 within living cells.

Knowledge of the topology of many integral membrane proteins is limited due to the time and effort required using approaches such as limited proteolysis, cysteine scanning, and glycosylation mapping. Computational analyses of the relative hydrophobicity of various polypeptide regions of a protein can predict membrane spanning domains typical of integral membrane proteins, yet such algorithms are not always reliable (16) nor do they provide information as to whether the amino or carboxyl domains of a protein face the cytoplasm, organelle lumen, or cell exterior. To determine the topology and orientation of LMTK2, we utilized the novel technique of fluorescence protease protection (FPP) as described by Lorenz et al. (13). This approach relies on the selective permeabilization of the plasma membrane by digitonin. Digitonin is a glycoside toxin obtained from the plant Digitalis purpurea, which, when added to cells, intercalates into cholesterol-rich membrane regions, causing plasma membranes to become permeabilized (17). Under such conditions, cytosolic contents can diffuse out of the cell and small exogenous molecules like trypsin can diffuse into the cell. Since the detergent effects of digitonin are restricted to cholesterol-rich membranes (i.e., the plasma membrane), intracellular organelle membranes, which have significantly lower concentrations of cholesterol, are unaffected under conditions of low level digitonin (12). Our current studies employed FPP to determine the topology and orientation of LMTK2 within living cells.



Hek293 cells were purchased from ATCC (Manassas, VA). DMEM and penicillin-streptomycin were from GIBCO (Grand Island, NY). FBS was from Hyclone (Logan, UT). Digitonin, LMTK2 antibodies, green fluorescent protein (GFP) antibodies, and proteinase K were from Sigma-Aldrich (St. Louis, MO). GFP plasmids were from Clontech (Mountain View, CA). Nucleofection solutions were from Amaxa (Gaithersburg, MD). Antibodies used were LMTK2 (rabbit anti-LMTK2; Sigma no. SAB4500900), Myc (mouse mAb anti-cMyc; Invitrogen no. R950-25), Na/K-ATPase (mouse mAb anti-Na+/K+-ATPase α-1 subunit; Developmental Studies Hybridoma Bank, University of Iowa, no. a6F), PP2A (rabbit anti-PP2A C subunit; Invitrogen no. 406001), GFP (mouse anti-GFP; Invitrogen no. A11120). Fluorescent secondary antibodies for immunoblotting were from Licor (Lincoln, NE) and for immunostaining were from (Invitrogen). Protease inhibitors were from Roche (cOmplete, EDTA-free protease inhibitor cocktail, no. 11836170001; Indianapolis IN). Proteinase K was obtained from Promega (Madison, WI). All other reagents were from Sigma.

Cell culture and transfection.

Hek293 cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin in a humidified atmosphere of 5% CO2-95% air at 37°C. Transient transfection of cells was performed by nucleofection (Amaxa) using manufacturer's protocols for Hek293 cells.

Preparation of soluble and membrane fractions.

Hek293 cells transfected with appropriate LMTK2 constructs were washed once with PBS and scraped in ice-cold buffer containing 10 mM Tris·HCl pH 8.0, 1 mM EDTA, and 1 mM EGTA with protease inhibitors (4 μg/ml leupeptin, 8 μg/ml aprotinin, 200 μg/ml Pefabloc, 484 μg/ml benzamidine, and 14 μg/ml E64). Cell were homogenized in a Dounce-type homogenizer, and the homogenates were clarified by centrifugation at 1,000 g for 10 min at 4°C. The pellet was homogenized again, and the combined supernatants were centrifuged at 4,000 g for 10 min at 4°C. The supernatants were mixed with an equal volume of the same buffer containing 0.25 M sucrose and centrifuged at 100,000 g for 1 h at 4°C. The supernatants were precipitated using 10% trichloroacetic acid, and the membrane pellets were solubilized in lysis buffer containing 1% NP-40. For carbonate extraction studies, samples of the 100,000-g membrane pellet were resuspended in 0.1 M sodium carbonate (pH 11.5) on ice for 30 min. Samples were then subject to centrifugation (100,000 g for 1 h at 4°C). Carbonate supernatants and pellets were then prepared as above. Protein concentrations were measured and an equal amount of protein resolved by SDS-PAGE and analyzed by Western blot.

FPP assay.

Hek293 cells transiently transfected with appropriate plasmids were plated onto poly-lysine-coated coverslips. Twenty-four hours following transfection, cells were washed in KHM buffer (110 mM potassium acetate, 20 mM HEPES pH 7.4, and 2 mM Mg Cl2) at room temperature. Coverslips were mounted on a microscope stage and imaging started during the pre-permeabilization stage. Cells were then perfused with KHM buffer containing 20 μM digitonin for 1 min and then in KHM buffer containing 50 μg/ml proteinase K. Live and fixed cells were imaged using a Visitech VT-Infinity 2D confocal system attached to an Olympus IX-71 microscope. For each condition, multiple coverslips were imaged (≥30 cells per coverslip) under identical settings and fluorescence intensity was determined using SimplePC software (Hamamatsu, Pittsburgh, PA).

Biochemical protease protection assay.

Equal numbers of Hek293 cells were grown on six-well plates (Corning-Costar) and wells transfected in duplicate with 4 μg of relevant plasmid. Twenty-four hours posttransfection, cells from were washed in KHM buffer (110 mM potassium acetate, 20 mM HEPES pH 7.2, and 2 mM MgCl2) and then treated with KHM buffer alone or KHM buffer containing 20 μM digitonin for 1 min at room temperature. Following permeabilization, cells were washed in KHM buffer and challenged with 50 μg/ml proteinase K in KHM buffer for 5 min at room temperature. Cells were washed in PBS containing protease inhibitors and extracted in SDS sample buffer. Equal aliquots of the samples were subject to SDS-PAGE and Western blot analysis using anti-GFP antibodies. Detection of primary antibody binding was determined by application of IRDye 800-conjugated secondary antibodies (LI-COR Biosciences) using an Odyssey infrared imaging system (Li-COR Biosciences).


LMTK2 is a membrane protein.

Kyte-Doolittle hydrophobicity analysis of LMTK2 predicts two likely transmembrane spanning helices at the amino-terminal domain corresponding to amino acids 11–29 and 46–63 (Fig. 1A), amino acid residues initially suggested by Wang and Brautigan (20). Topologies and potential orientations for LMTK2 are presented (Fig. 1B). Two possibilities are shown in which both the amino- and carboxyl-terminal domains are on the same side of the membrane, linked with a membrane anchor and short hydrophilic loop on the opposite side of the membrane; at present there are no formal data discriminating between these possibilities. Since LMTK2 has two predicted transmembrane helices, we anticipated that LMTK2 should be a membrane-bound protein. To test this, full-length myc-tagged LMTK2 was expressed in HEK293 cells and soluble and membrane fractions prepared from these cells. Full-length LMTK2 was recovered entirely in the membrane fraction and did not appear in the soluble fraction and appeared as a single dominant band of ∼210-kDa (Fig. 2A), consistent with previously published data (20). A low level of endogenous LMTK2 was also detected in membrane, but not cytosolic, fractions as evidenced by interaction with anti-LMTK2 antibodies but not anti-myc antibodies. The catalytic subunit of the sodium-potassium ATPase was used as a plasma membrane marker, and protein phosphatase 2A catalytic subunit was used as a cytosolic protein marker. To confirm that LMTK2 was indeed an integral rather than a peripheral membrane protein, microsomes (100,000 g pellet) were subject to extraction with 0.1 M sodium carbonate (pH 11.5). Results presented in Fig. 2B show that LMTK2 fails to be extracted by high salt (carbonate supernatant) and remains with the membrane pellet (carbonate pellet) confirming that LMTK2 is a bona fide integral membrane protein. Immunofluorescence imaging of LMTK2 expressed in Hek293 cells also displayed a discrete punctate localization, rather than a diffuse cytosolic appearance (Fig. 2C). Deletion of amino acids 1–70 (LMTK2ΔTM) would be predicted to delete the transmembrane domains of LMTK2, yielding a soluble fragment. Indeed, fractionation of HEK293 cells expressing LMTK2ΔTM revealed the presence of LMTK2ΔTM in the soluble fraction (Fig. 2D). The predicted loss in mass of ∼8 kDa in the ΔTM construct compared with full-length LMTK2 was difficult to resolve under our conditions.

Fig. 1.

Hydropathy plot and proposed topology of lemur tyrosine kinase 2 (LMTK2). A: hydropathy analysis of LMTK2 was performed according to the method of Kyte and Doolittle (11). There are two predicted transmembrane (TM) helices between amino acids 11–29 and 46–63. B: potential topologies and orientation of LMTK2 in endosomal membranes, in which the amino and carboxyl termini are both in the cytosol (i) or within an endosomal lumen (ii). LMTK2 is named after the Lemur, a Madagasgar primate with a long tail, since the Lemur family of kinases have long carboxyl tails (21). Out of the total 1,503 amino acids, 1,440 amino acids are proposed to comprise the COOH-terminal tail. For ease of illustration, the carboxyl tail is not drawn to scale. Kinase domain is shown in yellow.

Fig. 2.

Expression and distribution of LMTK2. A: Hek293 cells were transfected with either empty expression vectors or vectors containing full length myc-tagged LMTK2. Soluble (S) and membrane (M) fractions were prepared as described in materials and methods. Equal amounts of proteins were loaded on SDS-PAGE gels and analyzed by immunoblotting. Fractions were also probed with anti-LMTK2, anti-myc, anti-sodium/potassium ATPase (membrane protein marker), and anti-protein phosphatase 2A catalytic subunit (PP2Ac; cytosolic protein marker) antibodies. B: microsomal fractions (100,000-g pellets were resuspended in either 0.25 M sucrose buffer pH 7.4) or 0.1 M sodium carbonate buffer (pH 11.5) for 30 min on ice. Samples were again subject to centrifugation (100,000 g for 1 h at 4°C) before solubilization in SDS-sample bufffer and resolution by SDS-PAGE (CS, carbonate wash supernatant; CM, carbonate wash membrane pellet). C: Hek293 cells expressing exogenous LMTK2 were fixed and imaged using rabbit anti-LMTK2 antibodies, followed by Cy3-conjugated goat anti-rabbit secondary antibodies. Bar = 10 μm. D: Hek293 cells were transfected with either full-length LMTK2 or LMTK2 lacking the predicted NH2-terminal membrane spanning domain (ΔTM-HA). Soluble and membrane fractions were prepared and subject to immunoblot analysis using either anti-LMTK2 antibodies.

Establishment of FPP assay.

Having established that LMTK2 is an integral membrane protein, we sought to determine its topology using FPP The likely topology of LMTK2 in organelle membranes would correspond to a short cytosolically oriented NH2 terminus, two membrane spanning domains linked by a short hydrophilic region within the organelle lumen and a long cytoplasmically oriented COOH terminus. This topology, although predicted, requires formal confirmation as alternative arrangements, including the NH2 and COOH termini facing the organelle lumen, are also valid possibilities. In these assays, we employed an imaging-based protocol whereby Hek293 cells were transfected with LMTK2 tagged at either the NH2 or COOH terminus with GFP. This approach is based on a well-characterized method previously utilized to monitor membrane protein topology (13). Following digitonin permeabilization of the plasma membrane, the media are switched to one containing proteinase K. From Fig. 1B, it can be reasoned that if the topology of LMTK2 was that shown in Fig. 1Bi, then both NH2- and COOH-terminal fluorescent tags signals should be degraded upon protease exposure. If the topology were as in Fig. 1Bii, then neither of the termini signals should be abrogated. To verify that digitonin was able to permeabilize the plasma membrane efficiently to allow entry of proteinase K, we added digitonin (20 μM) to cells expressing GFP. Since GFP is freely soluble in the cytosol and nucleoplasm, GFP reflects the behavior of soluble cytoplasmic components. Permeabilization of the plasma membrane was evident as loss of GFP signal, initially from the cytosol and subsequently from the nucleoplasm following addition of digitonin (Fig. 3A). Under these same conditions, the integrity of intracellular membranes was not compromised, as we did not detect leakage of soluble proteins from the lumens of the endoplasmic reitculum, Golgi, or endosomal compartments (Fig. 3B).

Fig. 3.

Digitonin permeabilizes the plasma membrane without affecting organelle membranes. A: digitonin treatment permeabilizes the plasma membrane and releases initially cytosolic and later nucleoplasmic GFP. Hek293 cells expressing GFP were exposed to digitonin (20 μM). Images were taken before and after Digitonin permeabilization at the indicated time points. B: intracellular organelles remain intact following digitonin permeabilization. Hek293 cells expressing lumenal marker proteins for the endoplasmic reticulum (ER; pDsRed2-ER vector), endosomes (pAcGFP1-Endo vector), and mitochondria (pDsRed2-Mito vector) were incubated for 10 min with 50 μM digitonin in KHM buffer. No leakage of soluble marker proteins was observed under these conditions.

FPP assay for LMTK2.

Initial studies were performed to confirm that appending a GFP moiety to the NH2 or COOH termini of LMTK2 did not alter the membrane targeting of the constructs. Figure 4A shows that while GFP was found in the soluble fraction, appending GFP to either the NH2- or COOH-terminal region of LMTK2 caused a redistribution of the GFP signal to a membrane fraction. Cellular GFP fluorescence was monitored by image analysis. Upon 20-μM digitonin treatment, the fluorescence signal associated with both chimeras remained stable (Fig. 4B). However, upon addition of proteinase K, the fluorescent signal from both the NH2- and COOH-terminal tagged constructs was ablated significantly reduced (Fig. 4; P < 0.05 for difference ± protease treatment for each chimera). Therefore, the data are consistent with the NH2 or COOH facing the cytosol and exposed to protease activity. As a control, we examined the mitochondrial marker Mito tagged with DsRed. This protein, pDsRed2-Mito (Clontech) is a fusion protein of DsRed and the targeting sequence from subunit VIII of human cytochrome c oxidase, which targets the construct to the mitochondrial matrix (18). In cells expressing pDsRed2-Mito, the marker was stable during digitonin treatment. Upon addition of proteinase K, the signal persisted, as expected for a fluorophore located within the mitochondrial matrix and not in the cytoplasm (Fig. 4B).

Fig. 4.

Quantitative analysis of fluorescence intensities of NH2- and COOH-terminal-tagged LMTK2. LMTK2-tagged constructs expressed in Hek293 cells were subject to fluorescence protease protection assays. A: representative Western blot showing that LMTK2 has the ability to traffic GFP from a soluble fraction to a membrane fraction and that appending GFP to LMTK2 does not alter the membrane targeting of LMTK2. B: representative image showing fluorescence from NH2-terminal GFP-LMTK2 in the presence of digitonin alone (left) or digitonin and proteinase K (right). Bar = 10 μm. Fluorescent regions of interest were quantified following digitonin treatment and protease treatment. Background-subtracted mean fluorescence intensities are shown in the graphs. C: fluorescent intensities of the digitonin permeabilized cells were set to 100%. Error bars indicate SD, which is given for means of n ≥ 30 cells per condition. *P < 0.05 for difference ± protease treatment by Student's unpaired t-test.

Biochemical protease protection assay for LMTK2.

In related experiments, Hek293 cells were transfected with LMTK2 tagged at either the NH2 or COOH terminus with GFP and processed in a similar manner as described above. In these analyses, however, cellular protein was monitored by Western blotting with GFP specific antibodies rather than immunofluorescence microscopy. In cells expressing NH2 terminally tagged LMTK2 that were either untreated or Digitonin permeabilized but not proteinase K treated and immunoreactive band at ∼230 kDa was observed, corresponding to the fusion protein (Fig. 5, GFP-N). When these cells were challenged with proteinase K following digitonin permeabilization, all GFP immunoreactivity was lost, which would be expected if the NH2-terminal domain of LMTK2 was orientated toward the cytosol. In cells transfected with the COOH terminally tagged variant of LMTK2, GFP immunoreactivity was also lost upon treatment of Digitonin permeabilized cells with proteinase K. These results are consistent with the localization of both the NH2- or COOH-terminal domains of LMTK2 residing in the cytosol and are in agreement with the immunofluorescence microscopy data.

Fig. 5.

Biochemical protease protection assay reveals cytoplasmic NH2- or COOH termini. Hek293 cells expressing NH2- or COOH-terminal tagged LMTK2 were digitonin permeabilized followed by incubation ± proteinase K. Cells were lysed and proteins resolved by SDS-PAGE and GFP detected by immunoblotting with polyclonal anti-GFP antibody.


The correct topology and orientation of integral membrane proteins are essential for their proper function. Methods for evaluating membrane protein topology have included the engineering of epitope tags such as hemagglutinin or myc, into proteins followed by assessment of immunoreactivity before and after cell permeabilization and antibody labeling. Alternatively, limited proteolysis (22) and glycosylation mapping (3) have also been utilized. Here we employed the relatively simple technique of FPP (13) to reveal the topology and orientation of the novel kinase LMTK2. This method, unlike many biochemical approaches, does not require large amounts of target protein nor does it require the need for appropriate and available antibodies that are essential for topological assays relying on site-specific immunoreactivity. Although it is formally possible that the addition of a GFP moiety to a protein could affect the localization and function of the chimera, numerous studies have demonstrated that most GFP chimeras preserve the characteristic of the untagged protein (5).

LMTK2 has previously been shown to run on SDS-PAGE with an aberrantly high apparent molecular weight (9, 20). The calculated theoretical molecular mass of LMTK2 is 165 kDa (ExPASy Bioinformatics Resource Portal), yet routinely runs at a molecular mass in excess of 200 kDa on SDS-PAGE gels. Why LMTK2 should run so anomalously is not entirely clear. It is possible that LMTK2 may undergo some posttranslational modifications, or alternatively its low electrophoretic mobility could be attributed to the presence of its COOH-terminal proline-rich domains, since proline-rich proteins often migrate slower on SDS-PAGE. It is clear, however, that loss of the amino-terminal region encompassing the membrane spanning domain does not render the protein subject to an electrophoretic mobility consistent with its predicted molecular weight.

Although no amino tail interacting proteins have thus far been identified, it is clear that the amino tail of LMTK2 is also cytoplasmically oriented. Indeed, given the short sequence that constitutes the amino terminus, there is a low possibility for interacting partners. Nevertheless, the absence of amino terminal binding partners for LMTK2 has not been formally determined. Precisely what targets LMTK2 to organelles of the endosomal and recycling system exclusively is not known. Potential targeting sequences could be within the membrane spanning domains or through interactions with other proteins on either the cytoplasmic side of the organelle membrane or with endosome lumen proteins interacting with the lumen loop linking the two transmembrane regions of LMTK2. Why LMTK2 is targeted to endosomal membranes, and what the physiological consequences of altered targeting are not known. In conclusion, we have, for the first time, formally demonstrated the topology and orientation of LMTK2 within endosomal membranes. Clearly, further studies will be required to understand how this orientation facilitates the function of LMTK2.


This study was funded by National Heart, Lung, and Blood Institute Grant HL-102208 (to N. A. Bradbury).


No conflicts of interest, financial or otherwise, are declared by the author(s).


Author contributions: A.N., Y.J., and C.W. performed experiments; A.N., Y.J., and N.A.B. analyzed data; A.N., Y.J., and N.A.B. interpreted results of experiments; A.N. prepared figures; C.W. and N.A.B. conception and design of research; N.A.B. edited and revised manuscript; N.A.B. approved final version of manuscript.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
View Abstract