If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
1 Equal contribution 3 Present address: North Carolina State University, Department of Chemical and Biomolecular Engineering, Raleigh, NC. 2 Present address: 3M Company, St. Paul, MN.
Automation of low endotoxin “miniprep” scale plasmid purification.
•
Automated buffer exchange expands available user applications post traditional affinity and ion exchange workflows.
•
Combined protein purification workflow leaves proteins in assay- and formulation-ready buffers of end user's choice.
Abstract
The continued acceleration of time-to-market product development and rising demand for biotherapeutics have hastened the need for higher throughput within the biopharmaceutical industry. Automated liquid handlers (ALH) are increasingly popular due to flexible programming that enables processing of multiple samples with an array of functions. This flexibility is useful in streamlining research that requires chromatographic procedures to achieve product purity for downstream analysis. However, purification of biologics often requires additional off-deck buffer exchange steps due to undesirable elution conditions such as high acid or high salt content. Expanding the capability of ALHs to perform purification in sequence with buffer exchange would, therefore, increase workflow efficiency by eliminating the need for manual intervention, thus expediting sample preparation. Here we demonstrate two different automated purifications using pipet-based dispersive solid-phase extraction (dSPE). The first is an affinity purification of His-tagged proteins from bacterial lysate. The second is an anion-exchange purification of plasmid DNA. Both methods are followed by buffer exchange performed by an ALH. Percent recoveries for the three purified recombinant proteins ranged from 51 ± 1.2 to 86 ± 10%. The yields were inversely correlated to starting sample load and protein molecular weight. Yields for plasmid purification ranged between 11.4 ± 0.8 and 13.7 ± 0.9 µg, with the largest plasmid providing the highest yield. Both programs were rapid, with protein purification taking <80 min and plasmid purification <60 min. Our results demonstrate that high-quality, ready-to-use biologics can be obtained rapidly from a crude sample after two separate chromatographic processes without manual intervention.
Introduction
As the quantity and breadth of biological agents used as biotherapeutics continues to expand, utilization of high-throughput technology offers a means to address concomitant escalating labor requirements [
]. Such technology has been put to excellent use in the genomics revolution wherein high-throughput cloning, cell sorting, library arrays, sequencing, and miniaturized platforms can run in parallel [
]. Protein and nucleic acid purification have lagged in comparison, largely due to limited purification strategies and mandatory polishing steps prior to downstream biophysical characterization [
]. Common purification strategies rely on either loose resin in spin columns or filter plates that require centrifugation or a vacuum manifold, or packed resins used with liquid chromatography systems. Filter plates on vacuum manifolds can result in uneven flow or channeling across different wells as a vacuum is pulled across the entire plate, not through individual wells. Magnetic bead-based purifications are more amendable for high throughput but concerns with the cost of resin and scalability limit its use. An alternative is a pipet-based approach where a resin is loosely packed within the hollow chamber, and air displacement is used to aspirate and dispense aqueous solutions through the pipet into the resin [
Pipette tip solid-phase extraction and gas chromatography – mass spectrometry for the determination of methamphetamine and amphetamine in human whole blood.
Simultaneous determination of methamphetamine and amphetamine in human urine using pipette tip solid-phase extraction and gas chromatography–mass spectrometry.
Pipette tip solid-phase extraction and ultra-performance liquid chromatography/mass spectrometry based rapid analysis of picrosides from Picrorhiza scrophulariiflora.
]. Combining this functionalized pipet tip and a dispersive solid phase extraction (dSPE) approach with an automated liquid handler (ALH) allows for the purification of up to 96 samples simultaneously. We have recently employed this technology for phosphopeptide enrichment on an ALH [
]. Recently, we established an automated platform for high-throughput sample preparation for mass-spectrometry (MS)-based analysis in the form of the multi-attribute method (MAM) [
]. Introduced in these studies were the SizeX IMCStips that functioned as a high-throughput buffer exchange tool that could be performed without the need for a vacuum manifold, off-deck centrifugation, or high-pressure chromatography. Buffer exchange via size exclusion chromatography is a common step in the purification of various biological agents that serves either to remove an agent that may interfere with a downstream process or place the biological agent in buffer conditions more appropriate for biophysical characterization. Current size exclusion chromatography methods are limited to pre-packed filter plates for automated high-throughput applications and require centrifugation. Manual buffer exchange via spin columns and cartridges has been shown to be efficient and reproducible, but limited in throughput [
]. Moreover, manual intervention introduces the possibility for human error and requires additional equipment. Instead of incorporating a centrifuge into the protein purification workflow, we exploited the zero-pressure, compressed O-ring expansion (CO-RE) available on the Hamilton Microlab STAR, to run desalting on-deck, providing purified and desalted proteins. There are several appealing aspects of such an automated buffer exchange regarding downstream compatibility: removal of contaminants such as imidazole or detergent that may interfere with downstream assays [
In this study, we present two fully automated workflows on the Hamilton Microlab STAR for purification of up to 96 samples of protein or plasmid DNA. Importantly, both workflows incorporate automated buffer exchange as a final step, fully enabling further integration for downstream processing in a time efficient manner (Fig. 1). Two downstream applications are demonstrated: namely, biochemical assays following protein purification and sequencing following plasmid purification. The workflows for protein purification and plasmid purification totaled less than 80 and 60 min, respectively. The results reported herein represent a substantial time savings over manual purification while demonstrating a highly reproducible method with minimal human interaction. We view this extremely customizable workflow as a means to assuage the ever-growing demand for high-throughput biological purification in the pharmaceutical industry.
Fig. 1A. Representation of the operation of the two types of tips used in this study: affinity purification via dispersive solid phase extraction (dSPE) and buffer exchange via unidirectional flow across a size exclusion resin. B. Generalized workflow presented in this work. C. Deck layout for affinity purification followed by buffer exchange via SizeX150 on the Hamilton Microlab STAR system.
Sodium chloride, isopropyl β-D-1-thiogalactopyranoside (IPTG), N-(2-hydroxyethyl)piperazine-N’-[2-ethanesulfonic acid] (HEPES), phosphate buffered saline (PBS), B-PER Complete Bacterial Protein Extraction Reagent, and imidazole were purchased from Thermo Fisher Scientific (Waltham, MA) and used without further purification. Plasmid+ media was purchased from VWR (Radnor, PA) and used as received. Triton X-114, 4-methylumbelliferone (4-MU), 4-methylumbelliferyl sulfate potassium salt (4-MUS), and kanamycin sulfate were purchased from Millipore Sigma (Burlington, MA) and used without further purification. Ni2+-Sepharose 6 Fast Flow (Sepharose 6FF) resin and was purchased from Cytiva (Uppsala, Sweden), diluted, and stored in 20% ethanol at 4°C. Eppendorf 96-well deep-well plates were purchased from VWR. Micro-chromatography tips used in this study were purchased as empty tips from Hamilton Company (Reno, NV) and packed with resins of interest at Integrated Micro-Chromatography Systems (IMCS) (Irmo, SC). microPure LE IMCStips (P/N 04T-H6R33-1-25-96) were packed at IMCS and used for plasmid purification.
Recombinant proteins
A plasmid for expression of DasherGFP (GFP) was purchased from ATUM (Newark, CA) and transfected into NEB T7Express (New England BioLabs (NEB), Ipswich, MA) Escherichia coli according to NEB's transformation protocol for chemically competent cells. Briefly, 50 µL of competent cells were thawed on ice and added to 2 µL (5 ng) of plasmid DNA. The mixture was placed on ice for 30 min and then shocked at 42°C for 30 s. Cells were then placed on ice for five minutes and then brought to 1 mL with Luria broth (LB) (Thermo Fisher Scientific). Cells were then grown for 60 min at 37°C and 250 rpm in an Innova 44 incubator shaker (New Brunswick Scientific, Edison, NJ). After growth, 50 µL of cells were spread onto a 50 µg/mL kanamycin LB plate and grown overnight at 37°C. Individual colonies were picked and grown as 1 mL LB starter cultures containing 50 µg/mL kanamycin. 100 µL of starter culture was used to inoculate production cultures (250 mL – 2 L). Cells were induced with 1 mM IPTG unless otherwise noted and lysed using a CF2 series homogenizer (Pressure BioSciences Inc., South Easton, MA). His-tagged GFP (27.4 kDa) was purified on an immobilized metal affinity chromatography (IMAC) Sepharose 6FF resin chelated with cobalt. Captured protein was washed with 1X PBS, 15 mM imidazole, pH 7.4 and eluted with PBS containing 300 mM imidazole, pH 7.4. Eluted proteins were dialyzed overnight against 1X PBS, pH 7.4. Arylsulfatase (ArSulf) (61 kDa) and beta-glucuronidase (βGlc) (284 kDa) are commercially available products (IMCS) with respective product numbers 04-PaS-010 and 04-E1F-005.
Automated buffer exchange optimization
SizeX150 tips were assembled at IMCS. Inert resin utilized for size exclusion chromatography was packed into 1 mL wide-bore Hamilton tips in 10 mM Tris, pH 8 and 20% ethanol. All three proteins were diluted to and used at 4 mg/mL. Extinction coefficients (ε280) for the three proteins, GFP, βGlc, and ArSulf are 25.12, 140.57, and 102.79 mM−1 cm−1, respectively. These values were calculated based on amino acid composition [
]. Buffer exchange results were from four technical replicates. Protein concentrations in eluates were measured by Abs280 on a NanoDrop 2000 from Thermo Fischer Scientific. Recovered volumes were measured using a handheld Eppendorf pipet.
Desalting efficiency was determined using imidazole in 1X PBS. A standard curve of imidazole was established across a concentration range from 1 to 200 mM using Abs230 on a NanoDrop 2000. A standard of 300 mM imidazole in 50 mM HEPES, pH 8 was loaded on SizeX150, and the eluate was measured on NanoDrop. To address desalting efficiency, the absorbance of the eluted material from the SizeX150 IMCStips was measured at 230 nm. Absorbance values below the limit of quantitation (< 1 mM imidazole) represented a > 99.7% effective for desalting such molecules.
Automated affinity purification
Different amounts of purified polyhistidine-tagged proteins (Dasher GFP, beta-glucuronidase, and arylsulfatase) were added to 300 µL of empty vector bacterial lysates and diluted with 1X PBS to achieve His-tagged protein quantities of 0.2, 0.4, 0.6 and 0.8 mg in 825 µL of sample volume. Protein concentration of the bacterial lysate background was measured by the Pierce Coomassie Protein Assay Kit (Thermo Fisher Scientific) using bovine serum albumin as a standard. The protein concentration of the background lysate solution was set to 3 mg/mL. IMCStips containing 25 µL of Cytiva's Ni-Sepharose 6FF (Ni6FF) resin in 1 mL tip volume were used for the affinity purification workflow. Storage buffer was first dispensed and then tips were equilibrated in 50 mM HEPES, pH 7.5, 500 mM NaCl (1000 µL). Experiments were run in triplicate for each of the three protein samples and used 40 aspiration and dispense cycles for sample binding. After sample binding, tips were washed successively in 1000 µL of equilibration buffer, 1000 µL of the same HEPES buffer with 15 mM imidazole, and 1000 µL of equilibration buffer for five aspiration and dispense cycles each. Bound material was then eluted from the tip in 155 µL of 1X PBS and 500 mM imidazole over ten aspiration and dispense cycles. Protein concentrations in eluted samples were measured using a NanoDrop 2000 and calculated extinction coefficients.
Combined affinity and buffer exchange workflow for protein purification
Cell lysates and protein preparations were performed similarly to affinity purifications. Combined workflow with affinity and desalting was performed with eight replicates. The elution (150 µL) was transferred to equilibrated SizeX150 and used according to the optimized settings established previously with a single elution step. Protein concentrations in the now desalted eluates were measured by either Abs280 nm on the NanoDrop 2000 (ArSulf and βGlc) or fluorescence on a SpectraMax M5 (Molecular Devices, San Jose, CA) (GFP). Recovered volumes were measured using a handheld pipet. Cross-contamination testing and functionality of purified ArSulf enzyme was tested with activity assays utilizing p-nitrocatechol sulfate (pNCS) as a substrate [
Use of combined affinity and buffer exchange workflow to test the effect of time post-induction on protein production
Example purification of ArSulf from bacterial cultures was performed similarly to the cell lysate model system described above with several exceptions. NEB T7Express cells containing the plasmid for expression of ArSulf were grown overnight on an LB plate containing kanamycin (50 µg/mL). Three colonies were picked and grown overnight in 2 mL of LB containing 50 µg/mL kanamycin at 37°C, 250 rpm. One mL of overnight culture was inoculated into 110 mL of LB containing kanamycin (50 µg/mL) and grown at 37°C, 250 rpm until the culture reached an optical density at 600 nm (OD600) of 0.6 as measured by an Ultrospec 10 Cell Density Meter (Biochrom US, Holliston, MA). Cultures were brought to room temperature and induced with a final concentration of 0.4 mM IPTG. Cells were aliquoted out into 2 mL fraction over 24-well Duetz plates (Adolf Kühner AG, Basel, Switzerland) for each of the three cultures. Plates were grown at room temperature, 250 rpm for a total of 24 h. Three samples (1.5 mL) were taken from each of the three plates at 0, 2, 4, 6, 9, 12, 18, and 24 h after induction. OD600 was also taken at each time point. Cells were spun down at 14,000 x g for 5 min at 20°C, the supernatant was decanted, and pellets were stored at -20°C for 24 h. Cells were thawed and lysed by shaking cell pellets with 450 µL of B-PER Complete at room temperature for 30 min. After lysis, cells were pelleted at 14,000 x g, transferred to a 96-well deep-well plate and brought to 900 µL with 10 mM Tris, pH 8.0, 100 mM NaCl, 10 mM imidazole. All 72 samples were processed simultaneously using the combined affinity and buffer exchange workflow described above with several exceptions. Tips were conditioned with 10 mM Tris, pH 8.0, 100 mM NaCl, 10 mM imidazole and all wash steps were performed with the same buffer. Bound ArSulf was eluted with 10 mM Tris, pH 8.0, 100 mM NaCl, 500 mM imidazole. Purified protein was buffer exchanged into PBS as previously described.
Final concentrations of purified proteins were measured by NanoDrop. To test purity of the eluted sample, 3 µL of the sample, flow-through, or eluate were mixed with 3 µL Laemmli buffer (Bio-Rad, Hercules, CA) with 2-mercaptoethanol and loaded onto a 4-20% SDS-PAGE gel. Gels were run at 150 V for 60 min covering the length of the gel (6.7 cm). Purity of the eluted samples was measured using the ImageQuant software. Specific activity of the eluted protein was measured using 4-MUS. Briefly, 50 µL of a series of 4-MUS solutions (0-62.5 µM) in 100 mM Tris, pH 8.0 were mixed with 50 µL of a 10,000-fold dilution of the eluted ArSulf. A standard curve relating to concentration of 4-MU to relative fluorescence units (RFU) was constructed in 100 mM Tris, pH 8.0 by measuring the fluorescence (λEx:365 nm, λEm:445 nm) on a SpectraMax M2 (Molecular Devices). Samples were incubated at 25°C and the growth of the hydrolysis product (4-MU) was measured by fluorescence every 5 s for 100 s. Initial rates were measured by fitting the data to a linear regression. For Michaelis-Menten kinetics, initial rates were normalized to protein amount used in the assay.
Combined ion exchange and buffer exchange workflow for plasmid purification
Three proprietary plasmids of various sizes (3593 (pCRS240.3), 6258 (pCRS166), and 8484 (pCRS158) bp were synthesized for IMCS by Life Technologies Corporation (Carlsbad, CA). Plasmids contained a T7 promotor upstream of the multiple cloning site, a pUC19 origin of replication, and kanamycin resistance. Plasmids were transfected into NEB 5α competent E. coli cells and plated on kanamycin plates (50 µg/mL). Individual colonies for each of the plasmids were selected and grown in overnight cultures (2 mL) of Plasmid+ media at 37 °C for 16 h. Cultures were pelleted at 14,000 x g and supernatant was discarded. Pellets were resuspended in 300 µL of P1 buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 100 µg/mL RNaseA) and then lysed with 300 µL of P2 buffer (200 mM NaOH, 1% SDS). Cells were lysed for three minutes and then neutralized with 300 µL of P3 (3.0 M potassium acetate, pH 5.5). The solution was then centrifuged at 10,000 x g for 10 min and the supernatant (∼900 µL) was added to a 2.0 mL deep-well plate. A solution of 20% Triton X-114 totaling 10% of the total volume was added to the supernatant.
MicroPure LE IMCStips are an anion exchange resin-based product for low endotoxin plasmid purification. The vendor supplied buffers were used.
As in the combined affinity/buffer exchange workflow, eluted sample (100 µL) from microPure LE IMCStip was transferred to a SizeX100 IMCStip after SizeX100 was equilibrated in 1X TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0). Sample was eluted off SizeX100 with optimized settings and plasmid concentrations were determined using a NanoDrop 2000. Samples, with and without BamHI (NEB) digest, were analyzed for purity and supercoiled content on a 0.7% agarose gel (Bio-Rad, 200 V for 15 min). Furthermore, samples were tested for endotoxin content using a Pierce Chromogenic Endotoxin Quant Kit (Thermo Fisher Scientific).
Results
Automated affinity purification
Three different poly histidine-tagged proteins (GFP, ArSulf, and βGlc) were purified using dispersive solid phase extraction in 1 mL pipet tips filled with 25 µL of loose Ni6FF resin. Purifications were performed on an automated liquid handler, the Hamilton Microlab STAR system. Four different amounts (200, 400, 600, and 800 µg) of protein were into spiked bacterial cell lysate and PBS solution (825 µL final volume) for each of the three proteins tested. Yields after purification were quantified via NanoDrop and purity was analyzed using 4-20% gradient Mini-Protean TGX gel (Bio-Rad) and ImageQuant (Cytiva) software (Figs. 2A, S1). The SDS-PAGE gel was run at 200 V for 40 min. Purified GFP and ArSulf showed excellent purity (>95%) with band intensities correlating to measured protein amounts (Figs. 2A, SI, 1B). Yields for all three proteins followed a similar trend to that of GFP, such that purified material increased in conjunction with increasing starting material (Fig. 2B), but percent yields decreased slightly as starting loads approached 800 µg (Fig. 2C). At the lowest protein load, target capture for all three proteins was similar, with recoveries of 88 ± 1.0%, 95 ± 1.8%, and 90 ± 9.4% of spiked material for GFP, ArSulf and βGlc, respectively. As the concentration of starting material was increased, differences between proteins of different molecular weights became more apparent. At 800 µg initial loads, purification of the largest protein, βGlc, yielded only a 56 ± 0.6% recovery, while purification of the smallest protein, GFP, yielded 75 ± 2.1%.
Fig. 2The affinity purification of GFP using 25 µL Ni6FF resin bed in 1 mL IMCStips at four different protein quantities in bacterial lysate are shown. A. SDS-PAGE of the starting material (S), the flow-through (FT) which is the sample post processing with affinity tips, and eluate (E) across the four different starting amounts suggest proportional increase in yields. B. Quantified proteins in the elution fraction increase with sample loads. C. Percent recoveries decrease slightly with increasing starting amount, with 88% capture and recovery for the lowest protein amount and 75% recovery for the highest protein load.
Initial optimizations for SizeX150 used GFP at 4 mg/mL concentration to monitor buffer exchange with maximum protein recovery (Table S1). The buffer exchange program was partitioned into four parameters; sample volume, sample load volume, chaser volume, and chaser dispense volume (Table S1). Sample volume refers to the liquid sample containing the protein and salts prior to desalting. Sample load volume is the plunger movement to displace air, thereby loading the sample onto to the resin bed. Chaser volume corresponds to the volume of elution buffer loaded on top of the resin, and chaser dispense volume is the plunger movement to displace air to push the chaser through the column to elute off the protein. Sample and sample load volumes were fixed to 150 µL and 80 µL, respectively (Table S1), volumes that allowed for the sample volume to be fully absorbed into the column bed. Chaser volume was set at either 150 or 180 µL. Minimal improvements in recovery but appreciable increases in final sample volume, and thus, decreases in protein concentration, were observed (Table S2). When chaser volume was held at 150 µL and chaser dispense volume was adjusted from 150 to 170 and then to 200 µL, incremental increases in protein recovery from 76 ± 2 to 78 ± 3, and 83 ± 2% respectively were observed (Table S1, Fig. S2). This coincided with an increase in eluted volume and a reduction in protein concentration (Table S1). The starting concentration of 4 mg/mL was reduced to 3.15 mg/mL in both the 150 µL/200 µL and 180 µL/150 µL chaser volume/chaser dispense volume scenarios with the observed difference being a slightly higher protein recovery and elution volume in the former. These optimizations were then tested on the two larger proteins in the study (ArSulf, and βGlc) to similar results (Table 1). An increased chaser dispense volume resulted in slightly more dilute eluates but higher overall recoveries for both ArSulf (87 ± 3%) and βGlc (85 ± 1%).
Table 1SizeX150 buffer exchange settings for three proteins.
Combined affinity purification and desalting of proteins
The combined workflow comprised the two aforementioned chromatography programs into one method, Ni6FF resin (25 µL) in 1 mL pipet tip to purify three different polyhistidine tagged proteins (GFP, ArSulf and βGlc) followed by buffer exchange to remove the high concentration of imidazole in the elution buffer. In line with the prior affinity purification studies, lysates were fortified at four different amounts (200, 400, 600 and 800 µg) of protein. Overall recoveries mirrored earlier IMAC-only experiments. Total protein recovered, and thus yield, was highest for GFP at each initial load measured with the exception of the lowest spike amount (200 µg) (Fig. 4A). Yields for GFP purification gradually decreased as the target protein in lysate increased from 77 ± 4.7% at 200 µg to 68 ± 3.8% at 800 µg. Decreases in yield for ArSulf were similar, 73 ± 3.4% and 58 ± 10%, respectively. More striking was the decrease in βGlc yield (86 ± 10% to 51 ± 1.2%) over the same range. Percent recoveries of material from affinity purification followed by buffer exchange were similar for samples containing βGlc and GFP with recoveries relative to IMAC-only purification falling between 88 and 95%. ArSulf recoveries were lower and ranged from 77 to 83%. To verify there was no well-to-well contamination and the protein activity was retained, ArSulf samples were tested for specific enzyme activities after purification. Wells containing solution that was eluted from the combined workflow from lysate not spiked with ArSulf all showed no color development based on the pNCS assay. Each of the wells containing lysate spiked with ArSulf showed an increase in absorbance at 515 nm, consistent with the formation of the product 4-nitrocatechol.
To test the usefulness of this method under higher-throughput protein expression conditions, a series of purifications from 1.5 mL cultures grown in 24-well Deutz plates was performed. Using this workflow, an ample amount of purified protein amount after 9 h of induction (263 ± 5.8 - 279 ± 20 µg) was generated. This was more than sufficient for a multitude of experiments (Fig. S3), including activity assays and gel analysis. Gel analysis by ImageQuant software showed purity >95% for all samples post-induction (Fig. S3A). The protein concentration of the eluted sample ranged from 0.37 mg/mL at 2 h post-induction to 2.23 mg/mL at 12 h post-induction. Purification and buffer exchange of all 72 samples took a total of 78 min and subsequent assay preparation was performed on the deck in under three minutes. Total protein and activity peaked at 12 h after induction (Fig. S3C). Specific activity of the expressed protein toward 4-MUS did not vary greatly after 4 h of induction, with an average specific activity of 14.6 ± 1.4 nmol min−1 µg−1. Similarly, the KM was largely unchanged across all samples, with an average value of 2.9 ± 0.7 µM 4-MUS.
Combined anion exchange purification and desalting of pDNA
The combined workflow included two steps: purification of plasmid DNA (pDNA) from cellular lysate via anion exchange chromatography (AEX) and subsequent buffer exchange and desalting of the eluted pDNA (SI Fig. 4). The initial plasmid purification was performed on 2 mL Plasmid+ cultures containing NEB5α cells harboring one of three plasmids (pCRS158, pCRS166, or pCRS240.3) that had been pelleted and lysed via traditional alkaline lysis. Tips containing 25 µL of a strong anion exchange resin were then used to extract pDNA from a solution of Triton X-114 (2% v/v) and cleared lysate. The elution buffer used in this process, 50 mM Tris, 1500 mM NaCl, pH 8.5, 15% (v/v) ethanol, left the eluted plasmid in a buffer not compatible for downstream functions such as sequencing and transfection. As such, the eluate was then moved to the second automated process of buffer exchange and desalting via SizeX100 IMCStips. The result of this AEX/SEC-coupled purification was sequence-ready plasmid in under 60 min.
Like the binding step of the affinity purification, initial binding of the plasmid to the resin in the plasmid purification step of the two-step process took place over a series of aspiration and dispense cycles. To optimize this step, the amount of bound material was measured after 2, 4, 6, 8, 10, 15, 20, and 25 binding cycles. Plasmid was observed to bind in an exponential fashion with a plateau being achieved at 20 binding cycles (Fig. S5A). At 20 binding cycles, 13.7 ± 0.9, 11.4 ± 0.8, and 12.5 ± 0.2 µg was purified for pCRS158, pCRS166, and pCRS240.3, respectively. Isolated plasmid appeared free of contaminants when viewed on a gel (Fig. S5B) with no detectable RNA and band intensity paralleling an increase in plasmid bound as a function of binding cycles.
To demonstrate repeatability and applicability over a range of plasmid sizes, eight samples of each plasmid culture were purified with the two-in-one procedure. Average recoveries were above 10 µg for all three plasmids tested (Table 2). The largest plasmid, pCRS158 (8484 bp), showed the highest recovered amount with 14.6 ± 1.3 µg, while the recoveries for pCRS166 and pCRS240.3 were 10.9 ± 1.1 µg and 11.3 ± 1.1 µg, respectively. The recovery for pCRS158 was significantly different from pCRS166 and pCRS240.3 when analyzed with one-way ANOVA (Fig. 4A). The yield difference between pCRS166 and 240.3 was not significant. Similarly, all three plasmid constructs retained concentrations above the target value. The concentration of pCRS158 was also the highest at 151.0 ± 13.4 ng/µL with pCRS166 and pCRS240.3 having similar values of 106.8 ± 14.7 ng/µL and 107.0 ± 10.6 ng/µL, respectively. Similar to recovery amount, significant differences were only observed between pCRS158 and the other two plasmid constructs (Fig. 4B). The post-SizeX100 volumes were similar with only significant differences between the largest and smallest constructs, pCRS158 (97.1 ± 5.6 µL) and pCRS240.3 (105.2 µL ± 5.4 µL) (Fig. 4C).
Table 2Plasmid recoveries and endotoxin contamination for buffer-exchanged samples for each of the three plasmids tested (n = 8).
To test overall purity and whether the addition of detergent was required, the pCRS158 and pCRS240.3 samples were purified with 2% (v/v) final volume of Triton X-114 while the pCRS166 samples were purified without the addition of detergent. All samples displayed similar A260/280 ratios (Table 2) of 1.88-1.89, suggesting the pDNA is pure. The A260/230 ratios were different between each of the samples with the greatest difference being observed in the non-endotoxin removal buffer sample (pCRS166). Endotoxin values for the pCRS166 sample were 3.36 ± 0.901 EU/µg plasmid while the values of the pCRS158 and 240.3, both treated with endotoxin removal buffer, were 0.116 ± 0.110 EU/µg and 0.065 ± 0.109 EU/µg, respectively (Table 2). This difference in values is similar to those reported in earlier studies [
]. All eight samples of pCRS240.3 were run as eluted from SizeX100 to show repeatability across samples (SI Fig. 6). The samples appeared to be primarily supercoiled and free of contaminations.
To demonstrate use for downstream processes, 4 mL of LB culture containing pCRS166 was grown overnight, pelleted, and then lysed. Plasmid DNA was then extracted from this culture using the two-in-one method outlined above except for the final buffer being 10 mM Tris, pH 8.8 buffer instead of water or TE buffer. The resulting plasmid solution (100 µL) had a concentration of 76 ng/µL and 25 µL of this solution was sent to Eurofins Genomics (Louisville, KY) for sequencing. Sequencing was successful (SI Figs. 7 and 8) with an average Phred quality score of 57.6 over the first 950 base pairs, demonstrating the applicability of this process for high-throughput production of sequence-grade plasmid DNA.
Discussion
As the use of biologics as therapeutic agents continues to expand, there is an ever-growing need to expedite and increase the efficiency of the initial stages of purification [
]. To address this concern, we have previously introduced protein purification on ALHs using dSPE and automated buffer exchange as part of a preparative method for product characterization. In this work, we have combined the two strategies to provide an extremely flexible workflow that can be widely applied to address biotherapeutic production bottlenecks.
The first step in the combined automated procedure was an affinity enrichment of poly-his tagged proteins from crude lysate. The loosely packed Ni6FF resin behaved similarly to previous work [
]. Efficiency of protein purification correlated to protein size and load. The largest protein, βGlc, had the lowest yield at the highest protein load (56±0.6%). This is most likely a result of the large particle size and reduced effective diffusivity of the protein. Steric bulk may have led to lower availability of binding sites and the reduced diffusivity of the protein may have made binding to porous binding sites slower. In agreement with this, yields at the highest proteins loads for ArSulf were higher than βGlc, and yields for GFP were higher than ArSulf. Initial purification recoveries ranged from 88 ± 1.0 to 75 ± 2.1% for GFP loads of 200-800 µg. Actual protein remaining in solution was less than the difference between eluted protein and initial load. Evidence for protein remaining in the dead volume of the tip was observed with the fluorescent protein, likely a result of the extra-particle volume of the loose resin and relatively low six-column volume elution. If overall recovery is most important for this step, a second elution can be added to reduce the protein retained. As the goal was to maintain a final volume of 150-160 µL for loading onto a subsequent column, the elution parameter was kept at 155 µL. Based upon previous studies, the time spent in the binding phase (40 cycles) for these concentrations was likely excessive [
]. Owing to the flexibility of the program, the number of cycles could easily be altered either to save overall processing time or bind more protein, contingent upon the goals of the end user.
Following affinity purification, samples were buffer exchanged via SizeX150. SizeX tips function akin to size-exclusion purification on a fast protein liquid chromatography system: sample is loaded on top of a flat column bed and positive pressure is used to push the liquid through a packed bed. In this case, the pressure is derived from the plunger movement of the STAR pipetting channel. Optimization of these plunger movements was performed to balance recovery, volume, desalting effectiveness, and concentration. By maintaining the sample dispense parameter at 80 µL, protein loaded on top of the SizeX150 moved minimally into the resin bed. The minimal difference in gain by keeping the chaser dispense at 150 µL and changing the chaser volume to 180 µL from 150 µL was offset by a drop in concentration of the product when compared to chaser dispense and volume settings of 170 µL and 150 µL, respectively. Moreover, while a larger chaser volume may be more forgiving when given a smaller protein such as GFP, the larger proteins in this study would travel faster through the SizeX150, causing more protein to be lost in the breakthrough fraction. Choice of 170 µL over 200 µL as the chaser dispense value, despite the higher overall recovery of the larger volume, was due to the dilution effect imposed by such conditions. This is likely due to a broadening of the protein distribution as it traverses the column. A larger chaser dispense more effectively captures the protein but does so by catching the tail end of the distribution, diluting the final product. Similar to the affinity protocol, these values are customizable. If overall product recovery is paramount, chaser dispense can be further tuned to capture the full protein load while still excluding small molecules such as imidazole. Importantly, when even the highest elution volume conditions (150 µL chaser volume, 200 µL chaser dispense) were used, imidazole concentrations remained below the limit of detection (1 mM).
When combined, the affinity and buffer exchange workflows provided clean, ready to use protein in under 80 min. Recoveries were slightly above the product of the two individual steps. As expected, overall yield reflected the trend seen in individual affinity purifications. While yields were 15-20% lower than in IMAC-only purification, exchange into a desired buffer has already been performed immediately after the initial purification step, presenting several benefits over dialysis or spin-filter workflows. Moreover, quantitative recovery is also unlikely from other buffer exchange methods without additional dilution.
The example of purification of ArSulf from lysate demonstrated the rapid nature with which product could be purified and tested. Even with small, 1.5 mL cultures, we were able to produce an ample supply of protein for analytical tests including gel electrophoresis and activity assays of 4-MUS hydrolysis. Protein expression and total activity increased for the first 12 h after induction. After this period, further growth showed no additional improvement in yield or activity. Similar experiments testing for IPTG concentration, temperature of incubation, or shaker speed could be performed across a host of variants simultaneously. Generally, this workflow could be applied for a variety of optimization tasks. For example, growth profiles and corresponding activity or protein expression could be monitored as a function of different minimal media additives, or a series of variants could be screened for activity. The highest concentration eluate in this experiment was 2.23 mg/mL. This was achieved from only 1.5 mL of culture. The model system experiments showed that capacity of the tips is at least 150% that shown in the experimental system (Fig. 3). By scaling the culture amount, greater concentrations of eluted product could be achieved. With high protein concentrations and the ability to rapidly buffer exchanged into a series of different buffer compositions, this workflow would be ideal for low-volume formulation studies.
Fig. 3The combined chromatography methods of affinity purification and buffer exchange on automated liquid handler can be used for high-throughput workflows. A. Recoveries of histidine labeled proteins (GFP, ArSulf, βGlc) which were added into bacterial lysates and purified with Ni-IMAC in pipet format followed by SizeX150 desalting. B. The overall recoveries of purified protein from the combination Ni-IMAC and SizeX150 method paralleled the amount of protein spiked into the lysate. C. SDS PAGE of the proteins at different quantities mirror the NanoDrop and fluorescence quantifications. D. Purified ArSulf were active after undergoing the two step purifications on the liquid handler, and there was no detectable crossover between wells or tips.
Fig. 4Associated plasmid recovery values for plasmids of various sizes (n = 8). A. The amount of plasmid recovered in µg varied based on the size of the plasmid construct. Recoveries for pCRS158 were significantly higher than those for pCRS166 and pCRS240.3. B. The final concentration of plasmid in the elution well varied by plasmid construct. The concentration of pCRS158 was significantly higher than that of either pCRS166 or pCRS240.3. C. The volume eluted from the SizeX100 tip varied slightly by construct. The volume of the pCRS158 eluate was significantly lower than in pCRS240.3. An “*” denotes statistical significance (p-value < 0.05) and “ns” denotes no statistical significance.
The rapid conversion from affinity purification elution to buffer exchanged product may also serve to maintain protein integrity. In cases such as monoclonal antibodies (mAb), where elution from Protein A/G-based resins is performed with an acidic solution, susceptibility to aggregation and degradation upon exposure to acidic conditions imparts a time-sensitive component to the follow-up step [
]. Interest in the use of more sensitive immunoglobulins (IgM) or isotypes such as IgG2 and IgG4, has prompted researchers to develop purification techniques with milder elution conditions [
]. In early-stage research and development, where the number of constructs is high but the amount of each is low, the inability to maintain protein fidelity represents a significant barrier. Rapid transfer into a milder solution (∼10 min) without protein dilution offers a path to minimize protein structure alterations on a high-throughput scale.
Additionally, each individual sample's final buffer can be altered. This could be especially useful in preparation of samples for thermal stability or assay interference studies. The reproducible yields given the same initial sample coupled with an on-deck spectrophotometer could be utilized to rapidly produce a 96-well plate containing normalized protein solutions in 96 different buffer conditions from a bulk sample of cell lysate. The relatively high final concentration achievable (≥ 4 mg/mL) presents an ideal starting point for product formulation studies.
We demonstrated that coupling a buffer exchange step with anion exchange chromatography enables the generation of low-endotoxin plasmid DNA samples from overnight E. coli cultures. While purification strategies for plasmid DNA are plentiful at the “mini-prep” scale in the form of silica-based spin columns and spin plates, these purifications often contain high levels of lipopolysaccharides (LPS) which trigger the immune response by activating the Toll-like receptors (TLR4) and MD2 complex, reducing the effectiveness of transfection in several cell lines [
]. The yields achieved in this work are sufficient for low-volume transient transfections (∼6-10 µg), and additional transfection enhancers can be added during the buffer exchange step. This enables the platform to be used for general transfection use and expression, as well as serving as a tool for screening potential transfection enhancers. Purified samples were also shown to be suitable for immediate sequencing without further adulteration. Overall plasmid yields were sufficient for downstream purposes but represented ∼70% those of silica spin-column preparations (data not shown). There was little difference between plasmids of different sizes, and Triton X-114 appeared to have little effect on the amount of plasmid purified. In agreement with previous literature, we found that the inclusion of Triton X-114 did have the effect of reducing the endotoxin content [
]. Moreover, this purification can be used for up to 96 samples simultaneously. Spin-column and spin-plate solutions vary in throughput from 24 samples in 30 min to 96 samples in 120 min and require manual intervention. By incorporating a buffer exchange step, we have enabled the use of an anion exchange resin for purification of ≥10 µg of low-endotoxin pDNA/sample in under 60 min.
Conclusions
Purification of a biological agent represents one of the earliest and most ubiquitous steps in biotherapeutic characterization and discovery. The work presented here utilizes a zero-pressure pipet pick-up step and precise air displacement controls on the Hamilton Microlab STAR to introduce the concept of tip-based buffer exchange. This workflow was paired with two common modalities, affinity enrichment and anion exchange chromatography, to enable downstream-ready protein and plasmid DNA purifications, respectively. The introduction of an automated means for buffer exchange helps alleviate a major bottleneck in the generation of biotherapeutics by empowering scientists to ready their purified product for subsequent processes without manual intervention.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Competing Interest
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: L. Andrew Lee reports a relationship with Integrated Micro-Chromatography Systems, Inc. that includes: board membership, employment, and equity or stocks. Patrick A. Kates reports a relationship with Integrated Micro-Chromatography Systems, Inc. that includes: employment. Jordan N. Cook reports a relationship with Integrated Micro-Chromatography Systems, Inc. that includes: employment. Ryan Ghan is employed by Hamilton Company, which manufactures and sells the MicroLab STAR liquid-handling platform.
Acknowledgments
We thank Megan Capel for her rendering of Fig. 1A and assistance with figure design. We are also grateful to Dr. Pamela Quizon for her assistance with figure design and valuable discussions.
Pipette tip solid-phase extraction and gas chromatography – mass spectrometry for the determination of methamphetamine and amphetamine in human whole blood.
Simultaneous determination of methamphetamine and amphetamine in human urine using pipette tip solid-phase extraction and gas chromatography–mass spectrometry.
Pipette tip solid-phase extraction and ultra-performance liquid chromatography/mass spectrometry based rapid analysis of picrosides from Picrorhiza scrophulariiflora.
00005-5&id=gr1.jpg){kind=link}
00005-5&id=gr2.jpg){kind=link}
00005-5&id=gr3.jpg){kind=link}
00005-5&id=gr4.jpg){kind=link}