Development of a RPLC-UV Method for Monitoring Uncleaved HIV-1 Envelope Glycoprotein

Abstract

One of the HIV-1 vaccine design efforts has focused on developing a recombinant HIV-1 trimeric envelope glycoprotein (Env) as an immunogen to induce broadly neutralizing antibodies. A native-like immunogen, the BG505.DS.SOSIP.664 gp140 (Env) construct has been well-characterized as a vaccine candidate. This vaccine candidate comprises of three identical gp120 and truncated gp41 subunits that form into a trimer of heterodimers. During production, recombinant Env is expressed as a gp140 precursor polypeptide in which a furin cleavable site is engineered to generate a heterodimer of gp120 and gp41 subunits. Each heterodimer is connected by an intermolecular disulfide bond, and three heterodimers form into a trimer. Furin cleavage is an important factor to mimic native-like HIV-1 Env conformations and is needed to help induce an immune response. Therefore, it is critical to monitor cleavage for ensuring functionality of the Env vaccine product. In this paper, a new RPLC-UV method coupled with reduction was developed to routinely determine the percentage of uncleaved gp140 relative to the cleaved gp120 and gp41 subunits. Baseline separation was achieved among the gp120, gp41 and uncleaved gp140 peaks, thus enabling relative quantification of uncleaved gp140. Overall, this RPLC-UV approach has been successfully applied to support Env vaccine candidate developments.

Graphical Abstract:

We demonstrate a quantitative and sensitive RPLC-UV strategy coupled with reduction to routinely monitor furin cleavage efficiency of recombinant envelope glycoprotein constructs during HIV-1 vaccine development and manufacturing.

Introduction:

The HIV-1 vaccine program is aimed at providing immune protection by inducing virus neutralizing antibodies. Since the trimeric envelope glycoprotein (Env) is the target for broadly neutralizing antibodies (bNAbs), significant progress has been made toward developing a recombinant Env trimer that mimics the native structure found on the HIV-1 surface protein.^1–5 Many factors like proteolytic cleavage, trimer assembly and glycosylation are integral for making a native-like and potent Env-based immunogen.^1–2,6–8 Therefore, an analytical method is needed to assess proteolytic cleavage efficiency of the Env vaccine.

The BG505.DS.SOSIP.664 gp140 trimer has been designed with specific mutations to enhance its stability, while still closely resembling native Env with multiple epitopes to induce bNAbs.^1–3 During Env protein expression, furin protease, which recognizes a hexa-arginine (RRRRRR or R₆) motif located between the gp120 and gp41 subunits, cleaves the full-length gp140 (Env) protein in order to generate gp120 and gp41 heterodimers.^1–2,9 After cleavage, the two subunits stay associated together via an intermolecular disulfide bond, and three heterodimers self-assemble into a trimer.^1–3 Furin cleavage impacts protein folding, overall structure and function of the HIV-1 Env trimer.^7–10 Cleaved gp140 trimers render a compact structure with glycosylation similar to the native HIV-1 Env spike, which is important for the induction of bNAbs, whereas uncleaved gp140 trimers lack these key properties.⁷ Therefore, any uncleaved gp140 is treated as a product impurity during manufacturing since these molecules are defective as a vaccine.

SDS-PAGE methods coupled with reduction are widely used to monitor protein cleavage through gel bands. Cleaved gp140 can reduce to its gp120 and gp41 subunits, while any remaining gp140 band is assumed to be uncleaved gp140 (Env).^3,6,8 SDS-PAGE however has limitations based on its sensitivity and reproducibility for quantification. In addition, the gp41 subunit is usually poorly resolved and shows multiple gel bands.¹¹ Therefore, a RPLC-UV method was developed for routine biopharmaceutical analysis and was demonstrated to be sensitive, accurate and robust. This communication reports a newly developed RPLC-UV method, quality assessment of the method, and the successful application of the method for two HIV-1 Env vaccine candidates.

Materials and Methods:

Reagents

BG505.DS.SOSIP.664 (gp140) Env material was produced in a CHO cell line without furin co-expression at the Vaccine Production Lab in NIH (Gaithersburg, MD) and was used for RPLC-UV method development. Another construct, ConC-FP8v2 RnS-3mut-2G-SOSIP.664, was also produced in a CHO cell line, but was co-transfected with a furin-containing plasmid to enhance cleavage. Purified trimeric Env material was formulated in either 1x PBS or final formulation buffer. Reagents included high-purity LC-MS acetonitrile and ammonium bicarbonate that were purchased from JT Baker (Phillipsburg, NJ). LC grade trifluoroacetic acid (TFA) was purchased from Millipore. LC-MS grade water was purchased from Omni-Solv (VWR, Radnor, PA) and dithiothreitol (DTT) was purchased from G-Bioscience (St. Louis, MO). Beta-mercaptoethanol (BME) was purchased from Thermo Fisher Scientific (Waltham, MA). Handling DTT or BME is recommended in a fume hood. Calcium chloride was purchased from Alfa Aesar by Thermo Fisher Scientific (Waltham, MA). Furin was purchased from New England Biolabs (Ipswich, MA).

RPLC-UV Method

Env was diluted to 0.5 mg/mL with 50 mM ammonium bicarbonate prior to RPLC-UV analysis. For the reduced sample preparation, Env was denatured and reduced with 100 mM DTT (final concentration) for 5 min at 90°C. An Acquity H-class Bio UPLC system (Waters, MA) with a UV detector was used for analyses. Water and acetonitrile containing 0.1% TFA were used for the mobile phases. Separation of cleaved gp120, gp41 and uncleaved gp140 was achieved at a flow rate of 0.2 mL/min on a UPLC BEH C4 column (2.1 × 100 mm, 1.7 μm) at 80°C, with a gradient of (time - % organic): 0 min – 10%, 2.0 min – 10%, 2.1 min – 30%, 7.0 min – 40%, 10.0 min – 90%, 15.0 min – 90%, followed by two wash steps and re-equilibration. A 2 μL injection volume was applied. Protein elution was monitored using UV detection at 220 nm. Empower 3 software (Waters) was used for LC data acquisition and processing. Percentage of uncleaved gp140 was determined using Equation S1 described in the supplementary section.

Furin enzyme spiking study for uncleaved gp140 peak identification:

The Env samples were diluted to 0.5 mg/mL with 50 mM ammonium bicarbonate, pH 7.8 in 1 mM calcium chloride (diluent). Various amounts (0 μL, 0.1 μL, 0.2 μL, 0.5 μL, 1 μL, 5 μL, and 10 μL) of furin enzyme (2 units/μL) were added to 50 μL of Env (0.5 mg/mL) and incubated for 6 hours at room temperature. As a negative control, 5 μL of furin was added to 50 μL of diluent. After furin incubation, each sample was divided in half by volume for non-reduced and reduced Env analyses. For reduced Env analysis, the furin-treated samples were treated with 100 mM DTT (final concentration) and incubated at 90°C for 5 minutes. For the non-reduced Env analysis, the furin-treated samples were analyzed directly by RPLC-UV.

Method accuracy assessment:

Purified Env materials with low (2.93%) and high (25.6%) percent uncleaved gp140 levels were used to evaluate the accuracy of the method. Accuracy was inferred by mixing the two samples (0.5 mg/mL) at different volumetric ratios (i.e., 0:2, 1:1, 2:0 v/v) to obtain a range of percent uncleaved gp140 in the Env samples. The experimental percentage of uncleaved gp140 was divided by the theoretical percent value to determine recovery.

Results and Discussion

RPLC-UV Method Development

A scheme of the possible Env species after denaturation and reduction is illustrated in Fig. 1. The trimer of heterodimers is held together by non-covalent interactions, and single heterodimers are present under denatured conditions. After reduction with DTT, cleaved gp140 dissociates into the gp120 and gp41 subunits while any uncleaved gp140 remains as a residual impurity.

Fig. 1.

Scheme of uncleaved and cleaved gp140 heterodimers. After reduction with DTT, cleaved gp140 dissociates into the gp120 and gp41 subunits while uncleaved gp140 cannot due to the hexa-arginine residue linker holding the two subunits together.

The gp120 subunit (53 kDa) is more hydrophilic in nature due to its extensive glycan shield (gp120 subunit contains 24 potential N-glycosylation sites), while the gp41 subunit (17 kDa) is more hydrophobic based on its amino acid sequence and lower glycosylation level (gp41 subunit contains only 4 N-glycosylation sites).^12–13 Given that the gp120 and gp41 subunits, as well as the uncleaved gp140 heterodimer, have different molecular hydrophobicity, a RPLC-UV approach was applied for separation and to monitor uncleaved gp140. Fig. 2 shows the profiles of the Env subunits under reduced and non-reduced conditions. Under non-reduced conditions, gp140 was observed at approximately 9 min, with a front shouldering peak (Fig. 2a). After DTT treatment, cleaved gp140 dissociated into gp120 and gp41 subunits while the peak remaining at 9 min was indicative of the uncleaved gp140 (Fig. 2b). Both of the gp120 and gp41 peaks were confirmed at 6.8 min and 9.3 min respectively in a separate report.¹³ Overall, baseline separation of uncleaved gp140 from the rest of the cleaved subunit peaks was achieved using the reduction protocol. As such, this workflow was implemented in the final assay.

Fig. 2.

RPLC-UV chromatograms of Env under (a) non-reduced and (b) reduced conditions.

Evaluation of Reducing Conditions

For accurate relative quantification of uncleaved gp140, complete reduction of cleaved gp140 was an important factor in the sample preparation step. Therefore, the reducing conditions prior to RPLC-UV analysis were evaluated and optimized. An initial screening was performed by incubating excess amounts of DTT or BME with the purified Env material for 5 min at 90°C. For example, when 100 mM DTT or 100 mM BME was incubated with Env, the average percentage of uncleaved gp140 was 15.6% and 16.9% (n = 2), respectively (Fig. S1). Since both reductions were effective, to be consistent with other previously reported SDS-PAGE analyses^8,11,14 and with the merit of lower toxicity, DTT was chosen as the reducing agent. Higher amounts of DTT were also screened, but the percentage of uncleaved gp140 remained similar (Table S1). The addition of urea in the denaturation and reduction step was further evaluated, but higher variability between sample preparations, as well as lower recovery of the Env subunits was observed. Therefore, incubation of Env with 100 mM DTT at 90°C for 5 min was chosen in the final sample preparation protocol.

Confirmation of Uncleaved gp140

Since cleavage at the gp120-gp41 junction is a result of furin enzyme recognition and proteolytic activities, we sought to confirm the identity of uncleaved gp140 in the RPLC-UV profile by treating the purified Env sample with exogenous furin in vitro. To that end, furin protease introduction helped confirm the identity of the uncleaved gp140 peak for the BG505.DS.SOSIP.664 construct in both reduced and non-reduced RPLC-UV analyses. Under reduced conditions, the percentage of uncleaved gp140 near 9 min decreased as the amount of furin added increased. Fig. 3 summarizes the result from this systematic study. Without adding furin, the percent of uncleaved gp140 was 15.6%. Incubation with 0.2 units of purified furin led to the percent of uncleaved gp140 dropping to 9.2% and adding more furin enzyme led to a further reduction of the percent uncleaved gp140 levels. However, a plateau of 2% uncleaved gp140 was reached when the added furin amount was ≥10 units, indicating that the remaining gp140 could not be further cleaved by soluble furin (i.e., a gp140 sub-population resistant to cleavage). This may be rationalized by the accessibility of furin to the gp120-gp41 junction.⁸ Improper folding/conformation of this resistant, uncleaved gp140 (Env) species may cause furin to be non-accessible to the site and might underpin this finding. From this furin-spiking study, it was clear that furin was able to cleave some of the remaining unprocessed Env precursor in vitro and enabled confirmation of the uncleaved gp140 RPLC-UV peak. Both portions of the uncleaved gp140 species (furin cleavable and furin non-cleavable) in BG505.DS.SOSIP.664 contributed to the total percent of uncleaved gp140 impurity in the final assay. Furthermore, the finding was similar using the non-reduced protocol prior to RPLC-UV analysis. Under non-reduced conditions, a front shoulder peak was observed before the major peak near 9 min (Fig. S2–a). The front peak of the Env gp140 species decreased as the amount of furin increased (Fig. S2–b), indicating the front shoulder peak was the remaining unprocessed gp140 impurity and that the major peak was cleaved, fully processed gp140. Overall, this study helped characterize uncleaved gp140 for this new RPLC-UV assay.

Fig. 3.

RPLC-UV chromatograms of Env under reduced conditions after being spiked with varying amounts of furin. Env sample was incubated with 0 μL (blue), 0.2 units (orange), 0.4 units (green), 1 unit (black), 2 units (red), and 10 units (purple) of furin. The teal trace was a negative control of 10 units of furin without Env.

RPLC-UV Application for Two HIV-1 Env Constructs

Using the final method, the extent of endogenous gp120-gp41 cleavage for Env was examined and also helped confirm that furin co-expression can increase cleavage efficiency of secreted gp140 (Env) during the cell culture process. This concept was applied to BG505.DS.SOSIP.664 and ConC-FP8v2 RnS-3mut-2G-SOSIP.664 development materials. The BG505.DS.SOSIP.664 described herein was not co-expressed with furin during protein production. For the other Env construct though, co-expression of furin and ConC-FP8v2 RnS-3mut-2G-SOSIP.664 was implemented at the cell culture stage to enhance cleavage efficiency. Fig. 4 summarizes the RPLC-UV analysis of both BG505.DS.SOSIP.664 and ConC-FP8v2 RnS-3mut-2G-SOSIP.664 following DTT reduction. As stated earlier, the percent of uncleaved gp140 in the BG505.DS.SOSIP.664 development material was approximately 15% whereas for the second ConC-FP8v2 RnS-3mut-2G-SOSIP.664 construct, the percent of uncleaved gp140 impurity level was about 1%. This observation demonstrated the successful engineering design, in which nearly all furin recognition sites were cleaved for the ConC-FP8v2 RnS-3mut-2G-SOSIP.664 construct. Overall, Fig. 4 highlights that RPLC-UV analysis was capable of measuring uncleaved gp140 in different HIV-1 Env trimer constructs after DTT treatment. The gp140 peak was unambiguously identified, and the percentage of uncleaved gp140 impurity level was quantified. To this end, this assay has been routinely applied to measure the percentage of uncleaved gp140 impurity levels for different Env trimer lots/constructs throughout development and manufacturing.

Fig. 4.

RPLC-UV application for two different HIV-1 Env constructs: profiles of Env trimer constructs (a) BG505.DS.SOSIP.664 and (b) ConC-FP8v2 RnS-3mut-2G-SOSIP.664 after reduction by DTT.

Method Quality Assessment:

As part of method development, the assay quality was assessed for its precision, accuracy, and specificity using the BG505.DS.SOSIP.664 construct. An intermediate precision study was also performed. As demonstrated in Table S2, good reproducibility of the percent uncleaved gp140 was reported. Between two different LC systems, the average percentage of uncleaved gp140 was 14.6% and 14.7%. Using the same LC system, the average percentage of uncleaved gp140 was reported to be 14.6% on Day 1 and 15.3% on Day 2, with a relative standard deviation (%CV) of 2.6%. Overall, the %CV for precision of the assay with different sample preparations, days, instruments, analysts and reagents was <3% (Tables S2 and S3). Injection repeatability (n = 6) was also evaluated for a single Env trimer sample and found to have a %CV of 0.2%. Minimum fluctuations in the UV profile were also observed during the repeatability test.

Although the sample preparation was performed using a starting concentration of 0.5 mg/mL in the final protocol, a dilution linearity assessment determined the working sample concentration for Env trimer to be in the range of 0.2 mg/mL – 0.8 mg/mL. Specifically, sample concentrations of the BG505.DS.SOSIP.664 construct from 0.2 mg/mL and 0.8 mg/mL were tested in duplicate preparations, and the average percentage of uncleaved gp140 was determined to be 15.2% (CV < 1.5%, from 14 total data points). Fig. S3 illustrates good linearity of R² > 0.99 for the UV response versus concentration in this range. Therefore, the sample testing range was set to 0.2 mg/mL – 0.8 mg/mL for consistent measurements.

Accuracy of the percent uncleaved gp140 was also evaluated by mixing two Env developmental samples containing known percentages of uncleaved gp140 at high (25.6 %) and low (2.93%) levels. Between 2.93% and 25.6% uncleaved gp140, the average recovery (duplicate injections for each point) was determined to be 102% with the %CV <3%, as shown in Table S4. To this end, accurate results of uncleaved gp140 could be reported confidently down to ~3%. This quality assessment clearly demonstrated the ability of the assay to measure the percent uncleaved gp140 with high sensitivity.

Specificity of the assay was also tested to ensure that the sample preparation reagents, such as DTT or the buffer, did not interfere with the Env species. Negative controls of the final formulation buffer, 50 mM ammonium bicarbonate and 1x PBS were treated with 100 mM DTT at 90°C for 5 min. These reagents most likely eluted in the void volume and did not cause extraneous peaks or interfere with the Env analysis.

Other method characteristics, including hold-time stability of the sample after DTT reduction and diluent were also evaluated. Specifically, a sample was held at 8°C in the autosampler for up to 16 hours (overnight) prior to analysis. A 0.2% difference was observed between samples at hold times of T=0 and T=16 hrs. Therefore, degradation or reformation of the intermolecular disulfide bond between gp120 and gp41 was determined to be low risk. Additionally, a diluent study was performed to determine its effect on the reported uncleaved gp140 results. Env was diluted with 50 mM ammonium bicarbonate (pH 7.8), final formulation buffer (pH 7.2) or 1x PBS (pH 7.4) to a final concentration of 0.5 mg/mL (~7x dilution) prior to DTT treatment and subsequent RPLC-UV analysis. No change in the percentage of uncleaved gp140 was observed when any of these diluents were used. For the final protocol, 50 mM ammonium bicarbonate (pH 7.8) was selected as the diluent for sample preparation.

Lastly, to demonstrate the advantage of using RPLC-UV over SDS-PAGE analysis for quantitation of uncleaved gp140, the same Env trimer reference lot was subjected to SDS-PAGE analysis following 100 mM DTT treatment at 90°C for 5 min. The result of the SDS-PAGE analysis using Coomassie blue staining was similar to those reported previously.^3,11,14 Fig. S4–a shows non-reduced Env as a single gel band. Upon sample reduction, the gp140 band was mainly converted to its gp120 and gp41 subunits. The cleaved gp120 subunit migrated as a single gel band (slightly below the uncleaved gp140 remaining gel band) whereas the cleaved gp41 subunit was poorly resolved and under-represented. Due to the poorly detected gp41 gel band in the SDS-PAGE, the percentage of uncleaved gp140 was most likely over-estimated. By SDS-PAGE analysis, uncleaved gp140 was determined to be 17.9% or 15.7% using either 1 or 2 μg for sample loading; whereas, the RPLC-UV assay reported that 9.4% uncleaved gp140 remained in the sample (Fig S4–b) with excellent reproducibility.

Overall, both techniques can be complimentary for demonstrating furin cleavage. SDS-PAGE is a quick and inexpensive method for providing a semi-quantitative result while this new RPLC-UV method can provide increased resolution and reproducible measurement of uncleaved gp140 impurities. Especially in a quality control laboratory setting, this approach can serve as an excellent quantitative lot release method for furin cleavage monitoring. In summary, this work shows that this newly developed RPLC-UV method is a robust tool for measuring the percent uncleaved gp140 down to 3% with good accuracy, precision and sensitivity. This assay is simple and has been implemented for the routine analysis of Env vaccine candidates.

Conclusion

A new RPLC-UV method following DTT reduction was developed to monitor the cleavage efficiency of Env (gp140) constructs. Baseline separation of the reduced Env species (cleaved gp120, uncleaved gp140 and cleaved gp41) was achieved, allowing the accurate quantitation of the relative percentage of uncleaved gp140 by UV detection. Method qualities from precision, accuracy, specificity to linearity/working range were demonstrated. Compared with literature methods, this new method is accurate and sensitive with good specificity. Overall, this method has a significant impact, providing a reliable method for uncleaved gp140 in Env vaccine candidates.