Rapid affinity purification of intracellular organelles using a twin strep tag

ABSTRACT

Cells are internally organized into compartmentalized organelles that execute specialized functions. To understand the functions of individual organelles and their regulations, it is critical to resolve the compositions of individual organelles, which relies on a rapid and efficient isolation method for specific organellar populations. Here, we introduce a robust affinity purification method for rapid isolation of intracellular organelles (e.g. lysosomes, mitochondria and peroxisomes) by taking advantage of the extraordinarily high affinity between the twin strep tag and streptavidin variants. With this method, we can isolate desired organelles with high purity and yield in 3 min from the post-nuclear supernatant of mammalian cells or less than 8 min for the whole purification process. Using lysosomes as an example, we show that the rapid procedure is especially useful for studying transient and fast cellular activities, such as organelle-initiated signaling and organellar contents of small-molecular metabolites. Therefore, our method offers a powerful tool to dissect spatiotemporal regulation and functions of intracellular organelles.

Highlighted Article: A method for rapid, high purity, isolation of lysosomes, mitochondria and peroxisomes from cell lysate was developed based on twin strep tag binding to Strep-Tactin XT magnetic beads.

INTRODUCTION

Eukaryotic cells are compartmentalized into distinct membrane-enclosed organelles. Each organelle carries hundreds to thousands of proteins, lipids and metabolites, and performs a unique set of cellular functions. For example, lysosomes are the major degradation compartments responsible for the clearance of unnecessary materials inside the cells (Kurz et al., 2008). Mitochondria are the major organelles for the ATP generation required to support cellular synthetic pathways (Chandel, 2014). Peroxisomes are the primary vesicles that catabolize long-chain fatty acids and regulate the balance of oxidization (Lodhi and Semenkovich, 2014; Smith and Aitchison, 2013). In addition to their well-known classic functions, recent studies have also revealed that some organelles are directly involved in cell signaling. For example, mechanistic target of rapamycin complex 1 (mTORC1) is recruited to and activated on the lysosomal surface by sensing the abundance of nutrients in the lumen, such as amino acids and cholesterol (Castellano et al., 2017; Zoncu et al., 2011). Similarly, mitochondria can also function as a signaling organelle (Chandel, 2014). For example, cytochrome c released from the mitochondria initiates cell death (Bhola and Letai, 2016; Burke, 2017; Liu et al., 1996). Another example is AKAP family proteins, which anchor and regulate the activities of protein kinase A and other signaling enzymes on the outer membrane of mitochondria (Chandel, 2014; Esseltine and Scott, 2013).

With rapid technical advancements, profiling the global levels of RNA, protein, lipids and metabolites has become common in current biomedical research. However, most of these large-scale profiling studies do not provide spatial information (Uhlen et al., 2015), thus cannot explain how different organelles regulate their highly compartmentalized cellular functions. The ability of measuring the compositions of specific organellar populations and their changes in response to stimuli would provide a powerful tool to understand the functions of these organelles.

Isolation of different organelles is traditionally accomplished by subcellular fractionation with differential centrifugation or multi-step density gradient ultracentrifugation (de Araujo and Huber, 2007; Foster et al., 2006; Frezza et al., 2007; Graham, 2001a,b,c; Michelsen and von Hagen, 2009). However, most subcellular fractionation approaches bear some intrinsic drawbacks. For example, the heterogeneous nature in the density of any given organellar population makes it difficult to obtain a type of organelle without contamination from the others. In addition, the concentration of a desired organellar population collected from multiple fractions is often relatively low, making some downstream analyses challenging. Moreover, to the best of our knowledge, the subcellular fractionation methods usually take more than an hour (Frezza et al., 2007; Graham, 2001a,b,c), which may lead to changes in the compositions of organelles, especially the signaling molecules associated with the cytoplasmic leaflet of the organelles and some labile small-molecule metabolites. Besides fractionation, specific methods have also been developed for the purification of certain organelles. For example, lysosomes can be isolated by magnets after being loaded with iron oxide-conjugated dextrans (Rofe and Pryor, 2016). However, depending on the duration of loading and chasing, dextrans are enriched to different degrees in various endosome populations and lysosomes (Humphries et al., 2011). Moreover, long-term accumulation of non-degradable dextran may have some unexpected effects on lysosomal functions (Kurz et al., 2008). Some recent studies have shown the successful purification of mitochondria and lysosomes by using beads conjugated to antibodies against an endogenous mitochondrial or lysosomal-resident protein (Franko et al., 2013; Michelsen and von Hagen, 2009), or against an epitope tag fused to these resident proteins (Abu-Remaileh et al., 2017; Ahier et al., 2018; Chen et al., 2016; Zoncu et al., 2011). Antibody affinity purification is fast and specific; thus it overcomes some drawbacks of the traditional approaches. However, antibody affinity purification requires a substantial amount of antibody. In addition, the elution of functional organelles is achieved via the competition by a high concentration of epitope peptides, which is usually not very efficient.

One popular protein purification strategy is fusing the proteins of interest to the Strep II tag (WSHPQFEK), which mimics the strong interaction between biotin and streptavidin (Kimple et al., 2013; Schmidt and Skerra, 2007). Strep II-tagged proteins can be efficiently eluted by a low concentration of biotin derivatives. The small Strep II tag is biologically inert, and the binding between the tagged proteins and streptavidin beads can take place under physiological conditions. In addition to protein purification, the readily reversible interaction has allowed the use of Strep II tag for the purification of live antigen-specific T cells (Liu et al., 2016). The recently generated streptavidin variant, Strep-Tactin XT, further increases the binding affinity between two tandem Strep tags (herein denoted the ‘twin strep tag’) (Schmidt et al., 2013; Yeliseev et al., 2017). As compared to micromolar-to-nanomolar dissociation constant (K_d) for most commercial epitope antibodies, such as FLAG, Myc and HA (Einhauer and Jungbauer, 2001; LaCava et al., 2015; Schiweck et al., 1997; Wegner et al., 2002), twin strep tag features a nanomolar-to-picomolar K_d towards Strep-Tactin XT while preserving reversibility of binding (Yeliseev et al., 2017). In addition, streptavidin is far more stable than antibodies as it is resistant to almost any protease and to extreme pH (Bayer et al., 1990; Kimple et al., 2013), which helps to reduce the waste of beads and obtain consistent results.

In the current study, we developed a new organelle isolation approach using the twin strep tag. We show rapid and efficient purification of lysosomes, mitochondria and peroxisomes using streptavidin magnetic beads that bind to twin strep tag fused to the cytoplasmic tail of a resident protein for lysosomes, or the targeting signals of mitochondria and peroxisomes. Furthermore, using lysosomes as an example, we demonstrate that this method can be used to monitor the transient lysosomal association of signaling protein complexes as well as small-molecule metabolites.

RESULTS

Design of a new affinity lysosome purification approach using twin strep tag

Many applications involved in organelle purification, such as evaluation of signaling events and measurement of small-molecule metabolites, require rapid recovery as well as maintenance of organelles in the physiological conditions during purification and elution. To test the use of twin strep tag in organelle purification, we fused the twin strep tag to the C-terminus of monomeric GFP (mGFP)-fused LAMP1, a lysosome-resident protein (hereafter denoted Lyso-2Strep) (Fig. 1A). The cytoplasmic orientation of the twin strep would allow subsequent purification of lysosomes using streptavidin beads (Fig. 1B). All lysosomal membrane proteins are synthesized in the rough endoplasmic reticulum and transported to trans-Golgi network before they are delivered to lysosomes (Braulke and Bonifacino, 2009). The expression of Lyso-2Strep is under the control of a tetracycline-inducible promoter, which offers the option of turning off the transcription a few hours before lysosome isolation. This allows lysosomal delivery of newly translated Lyso-2Strep protein, ensuring that the majority of molecules are delivered to lysosomes. The inclusion of the inducible promotor also allows the control of the expression level of Lyso-2Strep, thus minimalizing unexpected effects caused by overexpression of the exogenous Lyso-2Strep. We generated stable HeLa cells that inducibly express Lyso-2Strep in response to doxycycline after lentiviral infection and puromycin selection. Lyso-2Strep correctly localized to lysosomes, as demonstrated by its colocalization with the lysosomal marker LAMP2 (Fig. 1C). This indicates that addition of the twin strep tag does not interfere with lysosomal targeting of LAMP1 and Lyso-2Strep can be used for affinity purification of lysosomes.

Fig. 1.

Design of Lyso-2Strep for affinity purification of lysosomes. (A) Schematic diagram of Lyso-2Strep. The expression of the fusion protein, LAMP1–mGFP–twin strep (Lyso-2Strep), is under the control of a tetracycline-inducible promoter. (B) Workflow of organelle purification with Lyso-2Strep. Cells expressing Lyso-2Strep are rapidly harvested and homogenized. Post-nuclear supernatant (PNS) is incubated with streptavidin magnetic beads for a short period of time and analyzed after three washes. (C) Fluorescence images of the GFP signal of Lyso-2Strep (green) and immunofluorescence staining of LAMP2, a lysosomal marker (red), of HeLa cells stably expressing Lyso-2Strep. Lyso-2Strep expression was induced by the addition of 1 µg/ml doxycycline to the cell culture 1 day before immunostaining and imaging. Scale bar: 20 μm.

Determination of the critical factors for lysosome purification using twin strep tag

We started to perform lysosome purification using the magnetic Strep-Tactin beads from IBA Lifesciences (cat. #6-5510-050). However, the recovery of lysosomes was extremely low. We recognized that the beads we used had a diameter of 25 μm, a commonly used size for protein purification. We reasoned that the size of the beads might affect the binding capacity of lysosomes, and therefore tested the efficiency of lysosome isolation using streptavidin beads of different sizes (i.e. 50 nm, 1 μm, 5 μm, and 25 μm diameters). To avoid variations in the bead materials, the quality of streptavidin and conjugation efficiency, we purchased these beads from the same company (Creative Diagnostics). We incubated equal amounts of the post-nuclear supernatant (PNS) with beads of the same binding capacity but different sizes. We found that the 1 μm diameter beads recovered the most lysosomes, while other beads of either larger or smaller sizes exhibited significantly lower lysosomal recovery (Fig. 2A). This result suggests that beads with a diameter of ∼1 μm are critical to the efficient recovery of lysosomes using affinity purification.

Fig. 2.

Determination of factors critical to lysosome purification. (A) Effect of bead sizes using beads from Creative Diagnostics. The same amount of PNS was incubated with streptavidin beads of different diameters as indicated for 5 min. The volume of the beads for each size was adjusted to have the same binding capacity. Lysosome abundance in the purified products was determined by western blotting for LAMP2, showing that 1 µm diameter beads are the most efficient. (B) Comparison of the effect of bead sizes on lysosomal purification efficiency using beads from Thermo Fisher Scientific. Western blot analyses of LAMP2 from PNS and products of purification of Lyso-2Strep using 25 µm and 1 µm streptavidin beads. (C) Purity of the lysosome preparation. Western blot analyses of PNS and products after purification with Lyso-2Strep using 1 µm streptavidin beads. Intracellular organelle markers used were: LAMP1 and LAMP2 (lysosomes, Lyso), SDHA and TOM20 (mitochondria, Mito), SLC27A2 and catalase (peroxisomes, Pex), PDI and SERCA (endoplasmic reticulum, ER), Golgin-97 and GOLM1 (Golgi), ATP1A1 and PMCA2 (plasma membrane, PM), S6K and ERK1/2 (cytosol, Cyt). The purified lysosomes are free of contamination from other organelles. (D) Intactness of the lysosome preparation. Fluorescence images of lysosomes isolated from HeLa cells expressing Lyso-2Strep (green), eluted from streptavidin beads with 20 mM biotin and then labeled with LysoTracker (red). Scale bar: 20 μm. (E) Effect of incubation time and bead abundance. The same amount of PNS (100 µl) was incubated with 30 µl or 60 µl of 1 µm diameter streptavidin beads (Thermo Fisher Scientific) for different time periods as indicated. The relative recovery efficiency of lysosomes was determined by comparing the density of LAMP2 to that of PNS on the same western blots.

By using magnetic beads from a different source (Thermo Fisher Scientific), we confirmed that beads with 1 μm diameter worked markedly better than those with 25 μm size (Fig. 2B). To test the purity of the isolated lysosomes, we performed western blot analysis for protein levels of several organelle markers (Fig. 2C). We detected LAMP1 and LAMP2 in the purified lysosomes, but not markers for mitochondria (SDHA and TOM20), endoplasmic reticulum [SERCA (detecting SERCA1/2/3, also known as ATP2A1, ATP2A2, and ATP2A3) and PDI (also known as H4BP)], Golgi [Golgin-97 (also known as GOLGA1) and GOLM1], peroxisomes (catalase and SLC27A2), plasma membrane [PMCA2 and ATP1A1 (sodium/potassium-transporting ATPase subunit alpha-1)], and cytosol [S6K, and ERK1 and ERK2 (ERK1/2, also known as MAPK3 and MAPK1, respectively)] (Fig. 2C), indicating no contamination by other organelles.

To test whether the purified lysosomes remain intact, we eluted the bound products from the streptavidin beads with biotin and then stained them with LysoTracker, which selectively labels acidic organelles (Chazotte, 2011a; Xiong and Zhu, 2016). The vast majority of the purified lysosomes were labeled by the LysoTracker, indicating that they remained intact and maintained a low luminal pH (Fig. 2D). Therefore, the affinity purification of lysosomes using Lyso-2Strep is a rapid, effective and specific approach that maintains the intactness of the organelles.

We then examined how incubation time and the amount of streptavidin beads affect lysosome recovery efficiency. Starting from the same amount of PNS (100 µl), the incubation with 30 µl beads for 2 min resulted in a yield of ∼60% of total lysosomes (Fig. 2E). Prolonging the incubation to up to 30 min did not obviously increase the recovery of lysosomes any further (Fig. 2E), suggesting that the binding capacity of the beads approached saturation very rapidly. The incubation time could be further shortened by using 60 µl beads, leading to a maximal recovery of lysosomes (60–70%) in as short as 30 s (Fig. 2E). The total isolation procedure, including cell collection and homogenization (<5 min), binding (0.5 min) and washes (2 min), could be finished in less than 8 min (Fig. 1B). These results demonstrate that Lyso-2Strep allows for exceedingly rapid isolation of lysosomes, a feature that can be extremely critical to some analyses, such as for analyzing proteins bound to lysosomal surfaces and small-molecule metabolites in the lysosomal lumen.

Examination of nutrient-dependent regulation of the subcellular location of mTORC1 using purified lysosomes

Recent studies have suggested that the lysosomal surface can be a platform for some important signaling pathways, such as the mTORC1 and TFEB pathways (Settembre et al., 2012; Zoncu et al., 2011). mTORC1 is the master regulator of cell metabolism and has been demonstrated to be transiently activated on the surface of lysosomes in responses to the increased levels of intracellular nutrients, such as amino acids (Xiong and Zhu, 2016; Zoncu et al., 2011). Since nutrient-regulated mTORC1 association with lysosomes is transient and involves only a minor fraction of the total cellular pool, directly analyzing the mTORC1 complex on the lysosomes rather than that in whole cells would offer insights into mTORC1 activation and regulation. We tested whether the purified lysosomes by Lyso-2Strep can capture the changes of the mTORC1 complex on lysosomes in HeLa cells cultured in normal, amino acid-deprived, and amino acid deprived and then refed growth media. Comparing to the normal culture medium, amino acid starvation led to an ∼60% decrease of the lysosome-associated mTORC1 components, mTOR and Raptor, which was restored by refeeding cells with amino acids (Fig. 3). Consistent with the changes in lysosomal association of the mTORC1 components, the level of mTOR phosphorylation at Ser2448, which indicates the activation status of mTORC1, on the isolated lysosomes, was also decreased by 60% after the 1 h amino acid starvation. Similarly, the decrease in mTOR phosphorylation was restored by amino acid replenishment (Fig. 3). These changes in mTORC1 on the isolated lysosomes are consistent with previous findings using a different approach of lysosome isolation (Zoncu et al., 2011). Therefore, lysosome isolation using Lyso-2Strep offers an extremely powerful approach to study mTORC1 and other signaling events occurring on the cytoplasmic leaflet of lysosomes.

Fig. 3.

Demonstration of mTORC1 activation on lysosomes using lysosomes isolated with Lyso-2Strep. Western blot analysis of mTOR, phosphorylated (p)-mTOR (S2448), LAMP2 and Raptor in the PNS (Input) and purified lysosomes (Lyso-Prep) of HeLa cells cultured in normal growth medium (Fed), amino acid-free medium for 1 h (Starved), or amino acid-free medium for 1 h followed by the normal growth medium for another hour (Refed). Note the decreases in mTOR, p-mTOR and Raptor in the Lyso-Prep of the starved samples. Representatives of three independent experiments with similar results.

Mass spectrometry analysis of lysosomal luminal amino acids using isolated lysosomes

Rapid isolation of lysosomes would allow us to analyze metabolic activity and metabolites in the lysosomal lumen. Lysosomes are directly involved in the metabolism of several macromolecules and their building blocks, such as proteins and amino acids. Metabolic activity in the cytosol is subject to regulation by the levels of amino acids in the lysosomal lumen through modulating the activity of key metabolism regulatory proteins, such as mTORC1 (Abu-Remaileh et al., 2017; Zoncu et al., 2011). As a proof of principle, we measured the levels of different amino acids in isolated lysosomes by mass spectrometry and compared them with those in cytosol. All 20 amino acids were detected in both lysosomes and cytosol. Among them, leucine and isoleucine constitute ∼50% of total amino acids in both cytosol and lysosomes. In contrast, the proportions of some amino acids are different between lysosomes and cytosol. For example, cysteine, arginine, aspartic acid and valine display markedly higher proportions in lysosomes than in cytosol, while phenylalanine, the second most abundant amino acid species, is ∼2-fold enriched in cytosol as compared to lysosomes. All other amino acid species exhibit similar proportions in cytosol and lysosomes (Fig. 4A–C). These findings suggest that under normal culture conditions, lysosomes in HeLa cells preferentially maintain a limited group of amino acid species, each representing polar, positively charged, negatively charged and neutral amino acids, respectively. The functional significance and underlying mechanism(s) warrant further investigation. These results highlight the importance of analyzing individual amino acid species and other metabolites in isolated lysosomes. Varying the metabolic conditions of the cells before lysosome purification will likely yield crucial insights into the mechanisms of regulation of cell metabolism.

Fig. 4.

Measurement of amino acid contents in the cytosol and lysosomes by LC-MS. Cytosol and lysosome preparations were made from HeLa cells stably expressing Lyso-2Strep as described in the Materials and Methods. Amino acid levels were determined by LC-MS. (A,B) Pie chart presentations of proportions of individual amino acids in cytosol (A) and lysosomes (B). (C) Same data as in A and B plotted as a bar graph showing comparisons between cytosol and lysosomes for the proportions of individual amino acids as means±s.d. of triplicate measurements. Note: a logarithmic scale and two breaks are used to accommodate the large span of data values between aspartic acid (0.01% in cytosol) and isoleucine/leucine (∼50% in cytosol).

Affinity purification of mitochondria and peroxisomes using twin strep tag

We also tested whether the same design can be used to purify other organelles, such as mitochondria and peroxisomes. Similar to what we undertook for Lyso-2Strep, we fused mGFP and the twin strep tag to the C-terminus of the mitochondrion-targeting sequence of the mitochondrial resident protein TOM20 and the peroxisome-targeting sequence of the peroxisomal-resident protein PEX-3 (Kapitein et al., 2010; Komatsu et al., 2010), to generate Mito-2Strep and Pex-2Strep for the purification of mitochondria and peroxisomes, respectively (Fig. 5A). HeLa cells stably expressing Mito-2Strep or Pex-2Strep were established by lentiviral infection followed by puromycin selection. After doxycycline induction for expression, both Mito-2Strep and Pex-2Strep showed intended localizations to mitochondria and peroxisomes, respectively, as demonstrated by their colocalization with the mitochondrion-specific dye MitoTracker and the peroxisomal-resident protein catalase (Fig. 5B,C). Furthermore, the isolated mitochondria were stained with MitoTracker Red or Tetramethylrhodamine ethyl ester (TMRE) (Fig. 5D,E). Given that the staining of MitoTracker Red and TMRE by mitochondria depends on mitochondrial membrane potential (Chazotte, 2011b), these results suggest that the purified mitochondria are intact. Importantly, TMRE staining of purified mitochondria was drastically reduced by a 30-min pretreatment with 50 µM CCCP, a mitochondrial oxidative phosphorylation uncoupler (Fig. 5E), suggesting that the purified mitochondria preserve the membrane potential.

Fig. 5.

Purification of mitochondria and peroxisomes using twin strep tag. (A) Schematic diagram of the design of Mito-2Strep and Pex-2Strep for affinity purification of mitochondria and peroxisomes, respectively. (B) Fluorescence images of the GFP signal of Mito-2Strep (green) and MitoTracker staining of mitochondria (red) of HeLa cells stably expressing Mito-2Strep. Mito-2Strep expression was induced by the addition of 1 µg/ml doxycycline to the cell culture 1 day before MitoTracker staining and imaging. (C) Fluorescence images of the GFP signal of Pex-2Strep (green) and immunofluorescent staining of catalase to label peroxisomes (red) of HeLa cells stably expressing Pex-2Strep. Pex-2Strep expression was induced by the addition of 1 µg/ml doxycycline to the cell culture one day before immunostaining and imaging. (D) MitoTracker staining of isolated mitochondria eluted from streptavidin beads with 10 mM biotin showing the uptake of MitoTracker (red) by the purified mitochondria (green). (E) TMRE staining of isolated mitochondria and its inhibition by a 30-min pretreatment with 50 µM CCCP. Mitochondria were eluted from streptavidin beads with 20 mM biotin. Scale bars: 20 µm.

We then checked the purity of the isolated mitochondria and peroxisomes by examining the presence of different organellar markers through western blotting. The isolated mitochondria and peroxisomes were free from the markers for cytosol and most other organelles except for the presence of some of the peroxisomal marker catalase (Fig. 6A) in mitochondria and the mitochondrial markers SDHA and TOM20 in peroxisomes (Fig. 6B). In contrast, another peroxisomal marker, very long-chain acyl-CoA synthetase (SLC27A2), was not detected in the purified mitochondria (Fig. 6A). The co-existence of mitochondrial and peroxisomal markers has also been reported before in mitochondria and peroxisomes isolated by subcellular fractionation (Sugiura et al., 2017) and affinity purification methods (Bayraktar et al., 2019; Chen et al., 2017). This may reflect the intrinsic interacting nature of the two organellar types, instead of unspecific contamination, for the following three reasons: (1) mitochondria and peroxisomes are physically tethered at their contact sites (Shai et al., 2016); (2) new peroxisomes are partially derived from mitochondria and therefore mitochondria and peroxisomes carry some common proteins (Sugiura et al., 2017); (3) SLC27A2 is absent in the purified mitochondria (Fig. 6A). Peroxisomal SLC27A2 is likely derived directly from endoplasmic reticulum because it was previously shown to localize to peroxisomes and endoplasmic reticulum but not mitochondria (Singh and Poulos, 1988; Steinberg et al., 1999).

Fig. 6.

Characterization of isolated mitochondria and peroxisomes and determination of factors critical to the organelle purification. (A,B) Western blot analyses for the purity of mitochondrial (A) and peroxisomal (B) preparations using intracellular organelle markers as in Fig. 2C. (C,D) Effect of bead diameter on the isolation of mitochondria (C) and peroxisomes (D). The same amount of PNS was incubated with streptavidin beads of indicated diameters for 5 min. The volume of the beads for each size was adjusted to have the same binding capacity. Mitochondrial and peroxisomal abundances in the purified samples were determined by western blotting for SDHA (C) and catalase (D), respectively, showing that 1 µm diameter beads are the most efficient for both organelles. (E,F) Effect of incubation time and bead abundance on the purification of mitochondria and peroxisomes. Western blot analyses for the abundance of SDHA (E) and catalase (F) in mitochondrial and peroxisomal preparations, respectively, obtained by incubation of 100 µl PNS with 30 µl or 60 µl of 1 µm streptavidin beads (Thermo Fisher Scientific) for different time periods as indicated. The relative recovery efficiencies of mitochondria and peroxisome were determined by comparing the densities of SDHA and catalase, respectively, in the purified organelles to that in PNS on the same Western blots.

As for the lysosome purification, we compared the capabilities of different sizes of streptavidin beads in the isolation of mitochondria and peroxisomes. Among the different bead diameters tested, including 50 nm, 1 μm, 5 μm, and 25 μm, the 1 μm diameter beads again exhibited the best performance in purifying both mitochondria (Fig. 6C) and peroxisomes (Fig. 6D). Under our experimental conditions, ∼60% of mitochondria and ∼75% of peroxisomes were recovered from the PNS (Fig. 6E,F) within 30 s of bead incubation. Doubling the amount of beads slightly increased the recovery of mitochondria during the shortest incubation time (30 s) without dramatically affecting that from longer (>1 min) time incubation (Fig. 6E,F).

DISCUSSION

A key to isolating organelles is to enrich them with high purity in a short period of time, so that molecules associated with or inside the organelles, especially signaling proteins and labile metabolites, can remain unchanged during purification. In the current study, we report a general strategy for organelle purification using the twin strep tag. The strong and specific interaction between the strep tag and streptavidin offers some distinct advantages. Under our experimental conditions, we can recover ∼60% of total lysosomes and mitochondria, and more than 75% of peroxisomes, with as short as a half minute incubation of PNS and the beads, which is comparatively shorter than the 3.5 min incubation required for antibody-based affinity purification using 3×HA beads (Chen et al., 2017). Another advantage of our approach is that the number of cells can be adjusted easily based on the need of subsequent application. For example, lysosomes purified from three million cells are sufficient for the western blot analyses of mTOR signaling components presented in Fig. 3 and 20 million cells are sufficient for lysosomal amino acids measurement shown in Fig. 4. In contrast, most density gradient ultracentrifugation methods require typically 200–500 million cells and a time duration of several hours (de Araujo and Huber, 2007; Michelsen and von Hagen, 2009). Furthermore, our method requires minimal equipment and experience, and can be performed in any biomedical laboratory. Finally, the twin strep purification is expected to cost less than antibody-based approaches, due to the high yield of organelle recovery shown in our study (thus fewer beads are needed) and excellent stability of streptavidin (Bayer et al., 1990; Kimple et al., 2013).

The method described in the current study can be of great use in understanding the structures and functions of intracellular organelles. First, it is known that there are quantitative and qualitative differences in protein levels in the same type of organelles from different cell types (Itzhak et al., 2017). The simplified procedures and reduced requirement for cell numbers in our approach make it possible to isolate and profile the composites of proteins, lipids and small molecule metabolites in parallel and in the same organellar population from different cell types. This kind of approach might be extremely important for cancer studies. There are overwhelmingly increasing collections of information on genomic, transcriptional and global proteomic alterations in cancer cells. However, how oncogenic alternations affect the compositions of intracellular organelles and how these changes relate to their specialized functions remain mysterious. Second, the ability to rapidly recover the desired organelles using the current method is essential for studying signal transduction on organellar surfaces and metabolism of labile metabolites within the organelles, which either co-exist with the organelle transiently or have fast turnover rates. Using lysosomes as an example, we demonstrated the feasibility of monitoring changes in mTORC1 signaling in response to nutrient (amino acid) availability and of quantifying the contents of individual amino acid species (Figs 3 and 4). Third, the genetically encoded twin strep tag can also be used in animals to assess the physiological functions of specific organelles in vivo. It has been shown that mitochondria from specific cell types can be quickly isolated with the use of a mitochondrion-targeted epitope tag from complex tissues without cell sorting, which increases the speed of isolation and allows better retention of the mitochondrial metabolite profile (Ahier et al., 2018; Bayraktar et al., 2019). We expect that the tools described here can further increase the capability of in vivo profiling of different organelles in different cells and tissues. Finally, the twin strep tag can be used in combination with other tags, such as 3×HA, for rapid isolation of different organelles in the same cells or tissues.

While most amino acid species showed similar proportions among the total amino acids in the lysosomal lumen and in the cytosol in our lysosomal amino acid measurement, a few amino acids showed preferential lysosomal accumulations. Our overall conclusion is consistent with a previous lysosomal amino acid measurement that also showed differential distribution of some amino acids in the cytoplasm and lysosomal lumen (Abu-Remaileh et al., 2017). However, there are some obvious differences between the present study and that of Abu-Remaileh et al. For example, the relative proportions of cysteine, alanine, aspartate, arginine and valine are significantly higher in the lysosomal lumen than in cytosol in HeLa cells in our study, whereas cystine is more abundant in lysosomal lumen of HEK 293 cells (Abu-Remaileh et al., 2017). The different distributions of amino acids in the lysosomal lumen in these two cell types may be due to different expression levels of various lysosomal amino acid transporters and different experimental conditions. However, the exact reasons remain to be further investigated.

One interesting finding from our study is that the efficiency of organelle isolation is dependent on the size of the magnetic beads. Among the beads of different diameters, 50 nm, 1 μm, 5 μm, and 25 μm, tested in the current study, we found that the 1 μm diameter beads performed the best for the isolation of lysosomes, mitochondria and peroxisomes. This result suggests that comparable sizes between the beads and organelles are critical to the success of affinity purification of organelles, likely due to the maximal occupation (and/or exclusion) of organelles on the beads. Lysosomes, mitochondria and perixosomes, have a diameter of 0.5–1 μm (Xu and Ren, 2015), 0.75–3 μm (Wiemerslage and Lee, 2016) and 0.1–1 μm (Lodhi and Semenkovich, 2014), respectively. It is possible that the isolation efficiency can be further increased with optimal sized beads for each organelle type. For experiments in which the maximal recovery is desired, we suggest performing a pilot experiment using bead sizes between 10 nm and 200 μm to identify the bead size that yields the highest recovery efficiency. It may also be possible to predict the most appropriate bead size for an organellar population by mathematical modeling of maximal surface interactions between the desired organelles and the beads. We performed most of our experiments using 1 μm magnetic streptavidin beads from Thermo Fisher Scientific. It is reported that the twin strep tag has a picomolar K_d towards Strep-Tactin XT, a mutant form of streptavidin (Yeliseev et al., 2017). Unfortunately, except for the 50 nm sized beads, Strep-Tactin XT beads smaller than 1 μm in diameter were not commercially available at the time of our study. We anticipate that both the yield and speed of purification will be further improved by Strep-Tactin XT beads of more appropriate sizes.

In summary, we describe a rapid and efficient method for the isolation of intact lysosomes, mitochondria and peroxisomes. We provide evidence that lysosomes purified by this method can be used to analyze signaling dynamics and metabolites. Predictably, this method should also be compatible for quick isolation of specific organellar populations from other mammalian tissues or non-mammalian cells or tissues, such as those of yeast, plant, C. elegans or Drosophila, although organism-specific organelle targeting sequence might be needed to target the twin strep fusion proteins to the desired organelles.

MATERIALS AND METHODS

General reagents and antibodies

Streptavidin beads were purchased from Thermo Fisher Scientific (Waltham, MA, cat. #88817) and Creative Diagnostic (Shirley, NY; 50 nm, cat. #WHM G066; 1 µm, cat. #WHM-S103; 5 µm, cat. #WHM-S108; 25 µm, cat. #WHM-S177), while Tactin beads were purchased from IBA biosciences (Gottingen, Germany, cat. #6-5510-050). Beads were washed with phosphate-buffered saline (PBS) before use. LysoTracker was purchased from Thermo Fisher Scientific (Waltham, MA). Antibodies for S6K (cat. #9202, 1:1000), SDHA (cat. #5839, 1:1000), catalase (cat. #12980, 1:1000), PDI (cat. #3501, 1:1000), Golgin-97 (cat. #13192, 1:1000) and ERK1/2 (cat. #4696, 1:1000) were from Cell Signaling Technology (Danvers, MA); antibody for SERCA (cat. #SC-271669, 1:1000) was from Santa Cruz Biotechnology (Dallas, TX); antibody for LAMP1 (cat. #L1418, 1:2500) was from Sigma-Aldrich (St Louis, MO); antibodies for PMCA2 (cat. #19678-1-AP, 1:1000), ATP1A1 (cat. #14418-1-AP, 1:1000), GOLM1 (15126-1-AP, 1:1000), SLC27A2 (14048-1-AP, 1:1000) and TOM20 (cat. #11802-1-AP, 1:10,000) were from ProteinTech (Rosemont, IL); and antibody for LAMP2 (cat. #9840-01, 1:2500) was from the Developmental Studies Hybridoma Bank (Iowa City, IA). Dylight 800-conjugated goat anti-mouse-IgG (cat. #SA5-10176, 1:5000) and Dylight 680-conjugated goat anti-rabbit-IgG (cat. #35568, 1:5000) were from Thermo Fisher Scientific. All restriction enzymes and Q5^® High-Fidelity DNA Polymerase for PCR were from New England Biolabs (Ipswich, MA).

Molecular cloning

LAMP1, mGFP and the twin strep tag, were amplified by PCR from LAMP-CFP-FKBP (Komatsu et al., 2010), mGFP-PASS (Lu et al., 2016; Zhang et al., 2014), and AAVS1_Puro_PGK1_3xFLAG_Twin_Strep (Addgene #68375; Dalvai et al., 2015), respectively, and fused by PCR to generate the LAMP1–mGFP–2strep fusion protein. The Lyso-strep is then cloned into LT3G-mGFP-PASS (Lu et al., 2016) derived from L3GEPIR (Fellmann et al., 2013) that contains the third generation tetracycline-inducible promoter, to generate LT3G–LAMP1–mGFP–2strep (Lyso-2strep). The cDNA encoding amino acids 1–34 of TOM20 was amplified by PCR from TOM20-CFP-FRB (Komatsu et al., 2010) and then cut with NheI and BamHI at the artificially introduced sites. To generate LT3G–Mito–mGFP–2strep (Mito-2Strep), LT3G–LAMP1–mGFP–2strep was cut with NheI and BamHI, and the LAMP1 fragment was replaced with the TOM20 fragment. The cDNA encoding amino acids 1–42 of PEX3 was amplified by PCR from pβactin-PEX3-mRFP (Kapitein et al., 2010) and then cut with NheI and BamHI at the artificially introduced sites. To generate LT3G–Pex–mGFP–2strep (Pex-2Strep), the LAMP1 fragment of LT3G–LAMP1–mGFP–2strep was removed by NheI and BamHI digestion, and then replaced with the cut PEX3 fragment. The full sequences and maps of these constructs will be shared upon request.

Cell culture, lentivirus production and transduction

HeLa cells from ATCC (Manassas, VA) were maintained in Dulbecco's modified Eagle's medium (DMEM) plus 10% fetal bovine serum (FBS) (Thermo Fisher Scientific). Cells were passaged less than five passages before being used for the current study. For amino acid starvation, cells were treated with amino acid-free DMEM (US Biological) supplemented with 10% FBS that has been dialyzed with Slide-A-Lyzer™ Dialysis Cassettes with 3.5 kDa cut-off (Thermo Fisher Scientific). Lentiviruses were collected from TLA-293T cells (Thermo Fisher Scientific) co-transfected with the lentiviral vector (Lyso-2Strep, Mito-2Strep, or Pex-2Strep), pCMV-dR8.2 and pMD2.G using Lipofectamine and Plus reagent (Thermo Fisher Scientific) as described previously (He et al., 2017; Wang et al., 2017). At 2 days after infection with lentiviruses, HeLa cells were selected with puromycin (1 μg/ml) for stable expression.

Isolation of lysosomes, mitochondria and peroxisomes

Approximately 6×10⁶ HeLa cells stably expressing Lyso-2Strep, Mito-2Strep or PEX-2Strep, were seeded on a 150-mm cell culture dish overnight and 1 μg/ml of doxycycline was added at the time of seeding. On the day of experiment, medium was replaced with fresh medium free of doxycycline ∼3 h before the experiment. All organelle isolation procedures were performed in a cold room with ice-cold reagents. Cells were washed twice with PBS, collected in 1 ml of potassium phosphate-buffered saline (KPBS; 136 mM KCl, 10 mM KH₂PO₄, pH 7.3) with a cell lifter, and then transferred into a 1.5 ml centrifuge tube. The cells were centrifuged for 1 min at 1000 g and then resuspended in 1 ml of KPBS. Resuspended cells were homogenized in a 2 ml glass tissue grinder (VWR, Radnor, PA) with 30 gentle and continuous strokes. The homogenates were centrifuged at 1000 g for 2 min and 800 μl of the post-nuclear supernatant (PNS) was added to 150 μl or 300 μl of prewashed streptavidin-conjugated magnetic beads (cat. #88817, Thermo Fisher Scientific) in a 1.5 ml centrifuge tube. The PNS and beads were gently mixed and incubated in a tube rotator for 0.5–30 min to allow binding. After incubation, the beads were collected with a magnetic stand (cat. #12321D, Thermo Fisher Scientific). The supernatant was discarded and beads resuspended in 1 ml KPBS and then transferred to a new 1.5 ml tube. The beads were washed two more times by resuspending in 1 ml KPBS and pelleting with the magnetic stand. For imaging, the organelles were eluted from the beads by incubation with 100 μl of 20 mM biotin in KPBS for 10 min, followed by collecting the beads with the magnetic stand and saving the supernatant.

Western blotting

Isolated organelles on the beads or PNS were mixed with an equal volume of 2×SDS sampling buffer. Protein samples were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Membranes were blocked with 1% casein in Tris-buffered saline (TBS, 50 mM Tris-HCl, 150 mM NaCl, pH 7.5) and probed with the indicated primary antibodies diluted in 1% casein in TBS supplemented with 0.1% Tween-20 (TBST), followed by fluorescently labeled secondary antibodies. The fluorescent signals were detected with the Li-COR Odyssey infrared imaging system from Li-COR Biotechnology (Lincoln, NE). The signal intensities of bands were analyzed with Image Studio Lite (Li-COR). The median value of blank space around each band was set as the background and subtracted from the corresponding band intensities. Western blots are representatives of at least three biological replicates from independent experiments with similar results.

Fluorescence microscopy

HeLa cells expressing the organelle probes were stained with antibodies specific for individual organelles, and visualized with a Nikon A1 confocal microscope. Cells, purified lysosomes or mitochondria, were incubated with 1 μM LysoTracker (Thermo Fisher Scientific), MitoTracker (Cell Signaling Technology) or TMRE (Sigma-Aldrich) for 15 min at room temperature before imaging. For CCCP treatment, after the third wash during mitochondrial isolation, beads were resuspended in 500 µl KPBS containing 2 mM sodium pyruvate and 50 µM CCCP and incubated at room temperature for 30 min. TMRE was then added to the solution to a final concentration of 1 µM and incubated at room temperature for 15 min. After incubation, beads were washed twice in 500 µl KPBS containing 2 mM sodium pyruvate and 50 µM CCCP. Mitochondria not treated with CCCP were processed in parallel for the same duration and temperature exposure. The bound mitochondria were eluted from beads in 100 µl 20 mM biotin in KPBS containing 2 mM sodium pyruvate for 10 min at room temperature. Fluorescence images are representatives of at least three independent experiments with similar results.

Sample preparation for mass spectrometry analysis of amino acids

After the final wash, beads bound with lysosomes were incubated with 60 µl of 50% methanol on ice for 5 min. Beads were then collected with a magnetic stand and eluates were transferred to a clean glass vial (VWR). For cytosolic amino acids, the cytosol was prepared by centrifugation of PNS at 16,000 g for 3 min at 4°C to precipitate all organelles. Then 40 µl of the supernatant was mixed with 40 µl 100% liquid chromatography and mass spectrometry (LC-MS) grade methanol. The samples for LC-MS analysis were prepared by spiking 5 µl of isotopic labeled standard mix into each sample.

Reagents and internal standards for mass spectrometry

High-performance liquid chromatography (HPLC)-grade acetonitrile, methanol and water were procured from Burdick & Jackson (Morristown, NJ). Mass spectrometry-grade formic acid was purchased from Sigma-Aldrich. Calibration solution containing multiple calibrants in a solution of acetonitrile, trifluroacetic acid and water was purchased from Agilent Technologies (Santa Clara, CA). Metabolite standards and internal standards, including N-acetyl aspartic acid-d3, tryptophan-15N2, sarcosine-d3, glutamic acid-d5, thymine-d4, gibberellic acid, trans-zeatine, jasmonic acid, 15N anthranilic acid and testosterone-d3, were purchased from Sigma-Aldrich.

Identification of amino acids

Analysis of metabolites was performed at the Metabolomics Core of Baylor College of Medicine (Houston, TX). Amino acids were identified by Zorbax eclipse XDB C-18 chromatography column (Agilent Technologies) using 0.1% formic acid (buffer A) and 0.1% formic acid in acetonitrile (buffer B). The samples were analyzed on 6490 triple quadrupole mass spectrometer coupled with 1290 series HPLC system equipped with a degasser, binary pump, thermostatted auto sampler and column oven (Agilent Technologies). Data analysis was carried out by using Agilent Mass Hunter workstation software. All the identified amino acids were normalized to the levels of spiked isotope labeled standard. LC-MS analysis was performed in MRM mode. Source parameters used were as follows: gas temperature, 250°C; gas flow, 14 l/min; nebulizer gas pressure, 20 psi; sheath gas temperature, 350°C; sheath gas flow, 12 l/min; capillary voltage, 3000 V positive and 3000 V negative; nozzle voltage, 1500 V positive and 1500 V negative. Approximately 8–11 data points were acquired per detected amino acid.