A method for facile production of variable lymphocyte receptors using SHuffle Escherichia coli

Abstract

Variable lymphocyte receptors (VLRs) are the antigen receptors of jawless vertebrates such as lamprey. VLRs are of growing biotechnological interest for their ability to bind certain antigenic targets with higher affinity than traditional immunoglobulins. However, VLRs are disulfide‐bonded proteins that are often challenging to produce requiring genetic modifications, fusion partners, non‐scalable host cell lines or inclusion body formation and refolding. As a potential VLR expression platform option, the SHuffle Escherichia coli strain has been genetically altered to allow cytoplasmic disulfide bond formation by mutations to thioredoxin reductase (trxB) and glutathione reductase (gor) to create an oxidative cytoplasm. Furthermore, the SHuffle strain expresses disulfide bond isomerase DsbC in the cytoplasm to promote correct disulfide bond pairing. Here, we demonstrate that the SHuffle strain can produce high yield VLRs with titers ranging from 2 to 32 mg of VLR per liter of SHuffle culture. Three VLRs (P1C10, RBC36, VLRA.R2.1) were expressed in SHuffle E. coli and the products were compared directly to those generated using the Rosetta E. coli strain. All VLRs were validated for correct sequence, purity, and activity. For all VLRs, SHuffle E. coli produced 2–9 times more soluble VLRs than Rosetta E. coli. Furthermore, the soluble protein fraction was 2–6 times greater in SHuffle E. coli than Rosetta E. coli for all VLRs. Overall, these results suggest that the E. coli SHuffle strain is a convenient and effective expression system for producing large amounts of VLRs.

1. INTRODUCTION

Variable lymphocyte receptors (VLRs) are a class of leucine rich repeat (LRR) proteins that are of interest in protein engineering research. As antigen receptors derived from the adaptive immune systems of jawless vertebrates like hagfish and lamprey, ¹ , ² , ³ VLRs share similar antigen‐targeting capabilities as that of immunoglobulin (Ig) antibodies. ⁴ , ⁵ Unlike mammalian host species of traditional Ig‐antibodies, lamprey diverged from mammals over 500 million years ago, giving VLRs an advantage of recognizing antigenic targets that mammalian antibodies struggle to recognize. ⁶ Notably, VLRs that are capable of targeting the brain vasculature, ⁶ carbohydrates, ⁵ , ⁷ cell markers, ⁸ and many other conserved mammalian proteins ⁹ , ¹⁰ , ¹¹ have been identified.

As a result of growing interest in their antigen‐targeting capabilities, recombinant VLRs have been produced in a variety of expression systems including Saccharomyces cerevisiae, ⁴ , ⁶ , ¹² , ¹³ , ¹⁴ HEK293 mammalian cells, ¹ , ⁶ , ⁸ , ¹⁰ , ¹² Escherichia coli, ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ and Hi5 insect cells. ⁷ , ⁹ While these expression systems have proven capable of producing VLRs, existing production methods present several challenges. In their typical recombinant form, VLRs exist as monomeric, solenoid‐shaped, LRR proteins with four disulfide bonds. To improve production of recombinant VLRs, various strategies have been utilized to express properly folded VLRs. Yeast Surface Display (YSD) technology has been employed for successful high‐throughput screening of VLRs tethered to the surface of yeast and for soluble expression in S. cerevisiae, but secreted protein yields are limited. ⁶ , ¹² VLRs also have been engineered with folding partners, including Fc domains, for improved expression in HEK293 cells, but these approaches require further downstream processing to release the unaltered VLRs. ⁶ , ⁸ , ¹² Furthermore, modifications to the amino‐terminus of VLRs have been used to improve expression in E. coli. ¹⁵ , ¹⁶ When larger amounts of VLR are desired without genetic changes to the protein, VLRs can be produced by stable transfection in high‐density bioreactors, ¹ baculovirus transfection systems, ¹⁸ and inclusion body isolation and refolding. ¹³ , ¹⁷ However, many of these strategies are notoriously labor intensive, produce low yields of properly folded protein, do not work for all VLRs, and do not address the need for genetically unmodified VLR production at larger scales. ¹⁹ , ²⁰ , ²¹

While E. coli is generally a preferred host for large‐scale recombinant protein expression due to its accessibility, ease of use, affordability, and ability to overexpress large quantities of recombinant protein, ²² standard E. coli production strains like the Rosetta strain are not always ideal for proteins like VLRs that possess disulfide bonds. E. coli is maladapted for disulfide bond formation due to the naturally reducing environment of the cytoplasm. ²³ Recombinant proteins are often manufactured in the cytoplasm of E. coli, but disulfide bonds are only able to form in the oxidative environment of the E. coli periplasm. ²³ Thus, any recombinant proteins that require disulfide bond formation must be shuttled to the periplasm via chaperones and leader sequences such as pelB. ²⁴ Unfortunately, since proteins are translated in the cytoplasm, many proteins begin to fold before reaching the periplasm. ²⁴ , ²⁵ , ²⁶ When proteins prematurely fold in the cytoplasm, insoluble inclusion bodies can be formed. ²⁴ , ²⁵ , ²⁶ For proteins that remain unfolded until they are shuttled to the periplasm, oftentimes only a small percentage will adopt the correct disulfide bond formation due to the large flux of overexpressed proteins and the requirement for oxidative protein folding. ¹³ Specifically, with respect to VLRs, inclusion bodies are often formed in standard E. coli strains in the absence of genetic modification or folding partners as described above. ¹³ , ¹⁷

To address these issues, researchers have developed engineered E. coli strains adapted for disulfide bond formation. Origami and SHuffle strains are two unique E. coli strains capable of cytoplasmic production of recombinant proteins with disulfide bonds. ²³ Both Origami, a BL21(DE3) E. coli derivative, and SHuffle, a K12 E. coli derivative, have a modified disulfide bond formation pathway that diverts disulfide bond formation to the cytoplasm. ²³ In brief, thioredoxin reductase (trxB) and glutathione reductase (gor) are suppressed in Origami and SHuffle E. coli to create an oxidative cytoplasmic state to enable the formation of disulfide bonds, as opposed to the naturally occurring reductive environment of wild‐type E. coli cytoplasm. ²³ However, unique to the SHuffle strain, disulfide bond isomerase DsbC is expressed in the cytoplasm to aid disulfide bond isomerization and form the native disulfide linkages in proteins that have multiple non‐consecutive disulfide bonds, such as VLRs. ²³ In this work, we describe a new approach for VLR expression using the SHuffle E. coli strain. We demonstrate large‐scale production of correctly folded, soluble VLRs. Furthermore, the proteins bound their known binding partners, confirming the correct fold of the VLRs.

2. MATERIALS AND METHODS

2.1. Construction of VLR genes in pET vectors

VLR genes were constructed in pET vectors for expression in E. coli (Table S1). Each VLR was engineered as two separate constructs. The first construct had an amino‐terminus six‐histidine (NT‐6xHis) epitope tag with a recombinant tobacco etch virus (TEV) cleavage site. The second construct had a carboxyl‐terminus six‐histidine (CT‐6xHis) epitope tag (Figure 1). RBC36 NT‐6xHis, RBC36 CT‐6xHis, VLRA.R2.1 NT‐6xHis and VLRA.R2.1 CT‐6xHis were purchased in pET‐30a(+) from Genscript. P1C10 NT‐6xHis was purchased in pET‐28a(+) from Genscript. P1C10 CT‐6xHis was purchased in pET‐21b(+) from Genscript. Plasmids were expanded in DH5α chemically competent E. coli by the heat shock transformation method. ²⁷ pET30 and pET28 plasmids were plated on LB agar containing 50 μg/mL kanamycin antibiotic as a selection marker, and pET21 plasmid was plated on LB agar containing 100 μg/mL ampicillin antibiotic as a selection marker for overnight incubation at 37°C. Three colonies were picked from each plate and each colony suspended in 3 mL of LB media containing the appropriate antibiotic selection in disposable inoculation tubes (Thermo Scientific #150268). pET30 and pET28 plasmids were suspended in LB media containing 50 μg/mL kanamycin antibiotic, and pET21 plasmid was suspended in LB media containing 100 μg/mL ampicillin. Cultures were grown overnight at 37°C with 250 RPM shaking. Plasmids were purified using the ZymoPURE Plasmid Miniprep Kit (#D4208T). One milliliter of bacterial culture was collected in a 1.5 mL tube and centrifuged at maximum speed for 15–20 s in a microcentrifuge. Supernatant was discarded, and the ZymoPURE Plasmid DNA Purification Centrifugation Protocol was followed at room temperature. The final elution of purified plasmid was done with sterile filtered ultrapure water.

FIGURE 1.

Variable lymphocyte receptors expression constructs in pET vectors with T7 promoters. (a) NT‐6xHis constructs were designed with a 6xHis tag and tobacco etch virus (TEV) cleavage site at the N‐terminus of the VLR. (b) CT‐6xHis constructs were designed with a 6xHis tag at the C‐terminus of the VLR.

2.2. pET transformation in expression hosts

Rosetta 2(DE3) Competent E. coli (Novagen #71397), SHuffle T7 Express Competent E. coli (New England Biolabs #C3029J), and Origami B(DE3) Competent E. coli (Novagen #70837) were transformed in triplicate (n = 3) by the heat shock method with each pET plasmid. ²⁷ The Origami B(DE3) Competent E. coli were transformed only with P1C10 CT‐6xHis in pET‐21b(+) for selection by Ampicillin since Origami B is Kanamycin resistant and not selectable for the other plasmids. For each transformation 50 μL of E. coli was thawed on ice. One microliter of purified plasmid DNA was added to the E. coli transformation tube. The E. coli transformation tube was gently flicked two times to mix the E. coli and DNA. The transformation tubes were incubated on ice for 30 min followed by 30 s at 42°C in a pre‐warmed water bath. Reaction tubes were moved directly from the water bath to ice for a 2‐minute incubation. Eighty microliter of room temperature SOC Medium was added to each tube. The E. coli transformations were incubated at 37°C for 1 h with 250 RPM shaking. Each biological replicate was plated on LB agar with the appropriate antibiotic selection and incubated overnight at 37°C. pET30 and pET28 plasmids were plated on LB agar containing 50 μg/mL kanamycin antibiotic as a selection marker, and pET21 plasmid was plated on LB agar containing 100 μg/mL ampicillin antibiotic as a selection marker.

2.3. VLR expression studies

Primary cultures of each VLR construct were prepared with appropriate antibiotics in both TB and LB media for initial expression tests. For LB cultures, one plate of colonies was scraped into 100 mL of selective media in 500 mL baffled shake flasks and grown at 37°C with 200 RPM shaking to an OD₆₀₀ of 0.7–0.9. The LB cultures were then cooled to 20°C, and IPTG was added to each LB culture at a final concentration of 1 mM. The induced LB cultures were grown overnight at 20°C. For TB cultures, one plate of colonies was scraped into 200 mL of selective media in 500 mL baffled shake flasks and grown at 37°C with 200 RPM shaking to an OD₆₀₀ of 0.7–0.9. The TB cultures were cooled to 20°C with 200 RPM shaking and grown for an additional 24 h. After 24 h, IPTG was added to each TB culture at a final concentration of 0.1 mM. The induced TB cultures were grown for an additional 18–24 h at 20°C. The 20°C induction temperature was chosen following initial expression studies that indicated this temperature produced a higher yield of soluble VLRs when compared to cultures induced at 16°C.

2.4. Insoluble and soluble sample preparation for sodium dodecyl‐sulfate polyacrylamide gel electrophoresis

Overnight cultures were harvested the next day by centrifugation. Fifteen milliliter samples were centrifuged at 3500×g for 10 min. The pellet was resuspended in 1.5 mL lysis buffer (50 mM sodium phosphate, 20 mM imidazole, 300 mM sodium chloride, 10% glycerol, pH 8). Samples were maintained on ice and sonicated with a microtip for two intervals of 15–20 s at 100% amplitude with 5 min of rest between each interval. A 500 μL sample of each lysate was collected for analysis. The 500 μL samples were centrifuged at 21, 330×g on a benchtop centrifuge for 2 min. All supernatant from each sample was collected as the “soluble” fraction and brought to 1.5 mL in ultrapure water. The pellet was collected as the “insoluble” fraction and resuspended in 1.5 mL of 8 M Urea. Soluble and insoluble samples were prepared for a reduced sodium dodecyl‐sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) NuPAGE 4%–12% bis‐Tris gel (Invitrogen #NP0323BOX). Reduced samples were prepared with 2 μL of soluble or insoluble sample, 2.5 μL NuPAGE LDS Sample Buffer (4×) (Invitrogen # NP0007), 1 μL NuPAGE Sample Reducing Agent (10x) (Invitrogen # NP0004), and 4.5 μL deionized water. Samples were heated at 70°C for 10 min. Samples were loaded at 10 μL per lane. Gels were run with NuPAGE MES SDS Running Buffer (Invitrogen # NP0002) at 200 V for 35 min. All gels were run at ambient temperature.

2.5. Western blot quantification of insoluble and soluble fractions

NuPAGE gels were transferred to nitrocellulose membranes with 0.2 μm pore size (Invitrogen #LC2000) for western blot analysis. ²⁸ The transfer was done on ice in cold NuPAGE Transfer Buffer (Invitrogen #NP0006) at 100 V for 1 h with stirring. After transfer, membranes were washed with deionized water four times for 5 min each with agitation. After washing, membranes were blocked with 5% BSA in Tris‐Buffered Saline with 0.1% Tween 20 (TBST) for 1 h at ambient temperature with gentle shaking. Following blocking, the membranes were incubated in primary antibody (Invitrogen 6x‐His Tag Monoclonal 4E3D10H2/E3, mouse IgG1) diluted 1:1000 in 5% BSA in TBST overnight at 2–8°C with gentle shaking. The following day membranes were washed three times in TBST for 10 min with agitation. After washing the membranes were incubated in the secondary antibody solution (LI‐COR IRDye 800CW Donkey Anti‐Mouse IgG # 926‐32212) in a 1:20,000 dilution in 5% BSA in TBST for 1 h with agitation at ambient temperature. After incubation with the secondary antibody, membranes were washed three times in TBST for 10 min with agitation. Membranes were scanned on a LI‐COR scanner using the 800 nm channel. Soluble and insoluble fractions were quantified by densitometric measurement of the signal at 800 nm with background subtraction for each biological replicate.

2.6. Scale‐up production of VLRs in SHuffle E. coli

SHuffle E. coli were transformed as described above. P1C10 was grown in TB media, and RBC36 and VLRA.R2.1 were grown in LB media with the antibiotic selection described above. Each construct was grown in 12 1 L cultures in 2 L baffled shake flasks with vented lids. Cultures were harvested at 7000×g for 15 min. Supernatant was discarded, and the cellular pellets were flash frozen in liquid nitrogen for storage at −80°C. Forty gram of frozen cell pellets were resuspended in 120 mL of lysis buffer (50 mM sodium phosphate, 20 mM imidazole, 300 mM sodium chloride, 10% glycerol, pH 8). Samples were kept on ice and sonicated with a preparatory tip for five intervals of 1 min each at 85% amplitude with 2 min of rest between each interval. Lysates were centrifuged at 45,000×g for 30 min to pellet insoluble material. Supernatants were collected for VLR purification.

2.7. Ni‐NTA purification of VLRs

VLRs were purified using peristaltic pump chromatography. Cleared supernatant was loaded onto a column packed with HisPur Ni‐NTA resin (Thermo Scientific #88223). Ten milliliter of resin was used for each set of 12 L cultures. After two washes with wash buffer (20 mM sodium phosphate, 300 mM sodium chloride, 25 mM imidazole, pH 8), protein was eluted with a 0–300 mM gradient of imidazole in elution buffer (20 mM sodium phosphate, 300 mM sodium chloride, 300 mM imidazole, pH 8). Fractions were collected according to peak absorbances at A₂₈₀ using a peristaltic pump with a 2 mL/minute flow rate. Fractions were assessed using SDS‐PAGE, and fractions containing pure protein were pooled together for buffer exchange. All VLRs were dialyzed into 10 mM TRIS, 200 mM NaCl pH 8. Following dialysis, VLRs were concentrated using Amicon Ultra‐15 Centrifugal Filter Units with molecular weight cutoff of 3 kDa (Amicon #UFC900396).

2.8. Size exclusion chromatography

Size exclusion chromatography (SEC) of purified VLRs was performed using an ÄKTA pure 25 M for evaluation of purity. Samples were separated using a Superdex 200 Increase 10/300 GL column (Cytiva # 28990944) with 10 mM TRIS, 200 mM NaCl pH 8 mobile phase. 0.2 mL samples were injected and eluted with a flowrate of 0.75 mL/min. Spectra were analyzed in Unicorn 7.

2.9. ECM binding assay and K _D calculation of P1C10

bEnd.3 (ATCC #CRL2299) culture was stored in liquid nitrogen prior to use. Immediately prior to use, the culture vial was thawed in a 37°C water bath. Using aseptic technique, the culture was transferred to conical tube with 9 mL prewarmed complete culture medium and centrifuged at 125×g for 5 min. The cell pellet was resuspended in 9 mL of prewarmed complete medium. The culture was transferred to a T75 culture flask and incubated at 37°C 5% CO₂ with medium renewal every 48 h. When cells reached 80%–90% confluence, 1000–2000 cells were seeded per well in a 96‐well flat bottom plate (Thermo Scientific #165305) and incubated at 37°C 5% CO₂ for 48–72 h until 100% confluent. Media was removed, and the cells were incubated twice with either versene for 30 min at 37°C or PBS + 1% Triton‐X‐100 for 5 min at 4°C to remove cellular debris and maintain ECM. Removal of cellular debris was confirmed via microscopy. Cells were incubated with DNAse for 30 min at ambient temperature. Finally, cells were washed three times with PBS, and a bicinchoninic acid (BCA) assay (Thermo Scientific #23225) was performed to confirm ECM proteins remained in the wells. Next, VLRs were diluted in assay diluent (1% BSA + 1.5% goat serum in PBS), and 100 μL of sample at saturating concentration of 100 μM was added per well and incubated for 1 h at ambient temperature. Following five washes, the plate was incubated for 1 h at ambient temperature with mouse anti‐6xHis diluted to 100 ng/mL in assay diluent (6x‐His Tag Monoclonal Antibody 4E3D10H2/E3, Thermo Scientific #MA1‐135). The plate was washed five times, labeled with goat anti‐mouse diluted to 1 μg/mL in assay diluent (Goat anti‐Mouse IgG (H + L) Highly Cross‐Adsorbed Secondary Antibody, Alexa Fluor Plus 488, Thermo Scientific # A32723), and incubated for an additional 1 h at ambient temperature. Following the final incubation, the plate was washed five times and read at excitation 493 nm and emission 518 nm on a BioTek SynergyH1 plate reader. All samples were assayed in four replicates.

2.10. Isothermal titration calorimetry and K _D calculation of VLRA.R2.1 and RBC36

A VP‐ITC microcalorimeter (Microcalorimetry, Inc.) was used to generate binding data for VLRA.R2.1 and RBC36. Prior to isothermal titration calorimetry (ITC), VLRA.R2.1 and RBC36 NT‐6xHis constructs were incubated with TEV protease for 48 h at 4°C to remove the 6xHis epitope tags. After passage over Ni‐NTA column to remove protease and uncleaved protein, the TEV‐cleaved constructs were subsequently dialyzed into 10 mM TRIS, 200 mM NaCl pH 8 dialysis buffer. The same dialysis buffer was used to dialyze the ITC injectants, hen egg lysozyme (HEL) (Roche #10837059001) and H‐trisaccharide (Sigma #sc‐300,786). For VLRA.R2.1, 0.35 mM HEL was loaded into the injector and titrated at 3 μL injection volumes into the cell that contained 0.033 mM VLRA.R2.1 at 25°C. For RBC36, 1.05 mM H‐trisaccharide was loaded in the injector and titrated at 3 μL injection volumes into the cell that contained 0.11 mM RBC36 at 10°C. All injections had a 20 s duration, 180 s spacing, and 307 RPM stir speed. K _D values were calculated using nonlinear least squares fits of ITC data with a single‐site binding model in Origin 7.0 software.

3. RESULTS

3.1. SHuffle E. coli produce soluble VLRs

We first set out to produce a VLR known as P1C10, which has been shown to have brain drug delivery capabilities by targeting pathologically‐exposed brain extracellular matrix (ECM). ¹² As such, it was desired to produce P1C10 at large‐scale for downstream characterization and use. Previously, P1C10 had been expressed at small scale via yeast (S. cerevisiae) and as Fc‐fusions in mammalian culture (HEK293), both with limited yields. ¹² To expand production capacity, we attempted to produce P1C10 in a standard E. coli protein production strain, Rosetta BL21 (DE3), via inclusion body refolding approaches following previously published production strategies. ¹³ , ¹⁷ The Rosetta BL21(DE3) derivative is one of the most common bacterial strains used for recombinant protein production because in comparative studies with other standard E. coli strains, the Rosetta derivative consistently produces the highest recombinant protein titer, lowest byproducts, and the fewest proteases. ²⁹ However, despite exploring several refolding conditions, purified inclusion bodies from the Rosetta expression system were never successfully refolded where the desired P1C10 monomers were the dominant species (Table S2 and Figure S1). Thus, we decided to explore SHuffle E. coli as an alternative bacterial expression system to produce soluble, properly folded P1C10. Moving forward, we also wanted to validate any of our expression approaches and production workflow by evaluating two additional VLRs known as RBC36 and VLRA.R2.1. RBC36 is a well‐characterized VLR with known binding to H‐trisaccharide, and RBC36 has been produced in insect cells ⁷ and yeast ⁶ , ¹² as well as a dimeric Fc fusion in mammalian cells. ⁶ VLRA.R2.1 binds to Hen Egg Lysozyme (HEL), and it has been produced from refolding inclusion bodies expressed in E. coli. ¹³ Thus, the set of three VLRs offer a representative sample set to validate a new VLR production strategy.

As an initial comparison, VLRs were expressed in both SHuffle and Rosetta bacterial strains using a cytoplasmic expression construct under the control of the T7 Promoter (Figure 1). Two different constructs with 6xHis tags included at either the N‐terminus (NT) or C‐terminus (CT) of the constructs for purification and detection were evaluated for each VLR. Both NT‐6xHis and CT‐6xHis constructs were assessed since protein solubility can be dependent on the location of the epitope tag. ³⁰ The NT‐6xHis constructs also included a TEV protease cleavage site for downstream uses that may require tagless VLRs. All cultures were induced at 20°C, and bacteria recovered 18–24 h after IPTG induction. Soluble and insoluble bacterial cell fractions were collected from each of the six VLR constructs for both the Rosetta and SHuffle strains to quantify the relative amounts of soluble VLR produced and the fraction of total VLR protein that was soluble. After bacterial lysis, the cell pellets were used to quantify insoluble VLR, and the lysate supernatant was used to quantify soluble VLR (see materials and methods for details). Western blots were quantified for each VLR construct (Figure 2a–f). Total soluble VLR production was uniformly higher in SHuffle E. coli compared to Rosetta E. coli for all six constructs (Figure 2g–i). P1C10 NT‐6xHis and CT‐6xHis constructs yielded 7.3‐fold and 3.9‐fold, respectively, higher soluble protein amounts in SHuffle compared to Rosetta E. coli. RBC36 NT‐6xHis and CT‐6xHis constructs yielded 9.4‐fold and 5.7‐fold, respectively, higher soluble protein amounts in the SHuffle strain. Finally, VLRA.R2.1 NT‐6xHis and CT‐6xHis constructs yielded 2‐fold and 4.6‐fold, respectively, higher soluble protein amounts in SHuffle compared to Rosetta E. coli. When comparing the yields for NT‐6xHis and CT‐6xHis, the P1C10 CT‐6XHis construct produced slightly more soluble protein in the SHuffle strain (1.1‐fold increase) compared to its NT‐6xHis counterpart, but this improvement did not extend to RBC36 or VLR2.1.

FIGURE 2.

Comparison of soluble variable lymphocyte receptor (VLR) yields using the SHuffle and Rosetta bacterial strains. (a–f) Western blot comparison for 6 VLR constructs. Bacterial lysates from Rosetta and SHuffle cultures were separated into insoluble and soluble fractions via centrifugation. Samples were run on reduced 4%–12% Bis‐Tris gels, transferred to a nitrocellulose membrane, and probed with anti‐6xHis antibody. (g–i) Densitometric quantification of the of total soluble VLR. Data were normalized to Rosetta NT‐6xHis for each VLR. (j–l) Densitometric quantification of the fraction soluble was assessed by dividing the total soluble material by the sum of the insoluble and soluble material. Data in all panels represent the mean and standard deviation of three independent bacterial transformants (n = 3). Data are expressed as mean ± SD and statistical significance was determined by a one‐way ANOVA with Tukey's post‐hoc test. * represents p < 0.05, ** represents p < 0.01, *** represents p < 0.001, all comparisons with no p‐value indicated were not statistically significant.

We also assayed the fraction of the total VLR produced that was soluble to compare the cytoplasmic folding efficiency of the Rosetta and SHuffle expression systems (Figure 2j–l). As expected, given the differences in the cytoplasmic folding environment, the SHuffle strain led to uniformly higher fractions of soluble VLR. The P1C10 NT‐6xHis and CT‐6xHis constructs had 2.4‐fold and 1.8‐fold, respectively, higher fractions of soluble material in the SHuffle strain compared to Rosetta. Similarly, RBC36 NT‐6xHis and CT‐6xHis constructs had a 5.8‐fold and 2.9‐fold increase in the fraction of soluble VLR in SHuffle compared to Rosetta. VLRA.2R.1 performed similarly with a higher fraction of soluble VLR in SHuffle compared to Rosetta for the NT‐6xHis and CT‐6xHis constructs with 1.6‐fold and 3.3‐fold increases, respectively. Taken together, the major improvements in soluble production were a result of the SHuffle strain with more minor effects of the 6xHis epitope tag location. In addition to the Rosetta and SHuffle strains, we tested the Origami strain for its ability to express VLR P1C10 given its similarity to SHuffle with an oxidizing cytoplasm and since it was previously used to produce a highly engineered VLR scaffold. ¹⁵ Western blot analysis of P1C10 CT‐6xHis showed no significant difference of total and fraction soluble VLR in SHuffle compared to Origami (Figure S2). However, since the SHuffle strain has DsbC for disulfide bond isomerization, this strain was chosen for our subsequent scale‐up.

3.2. VLRs produced in the SHuffle strain can be purified as soluble protein preparations

SHuffle‐produced VLRs were purified using Ni‐NTA resin (see materials and methods). VLR purity was assessed by reducing SDS‐PAGE with Coomassie staining (Figure 3). A major band for each VLR construct was observed at the correct molecular weight. Purified VLRs were then assessed by liquid chromatography‐mass spectrometry (LC–MS) to confirm that all purified VLRs were the correct proteins and identified peptides covered 99%–100% of the amino acid sequences for the VLRs (Table 1). Finally, purified VLRs were analyzed by SEC. All VLRs produced a dominant, symmetric peak with purity ranging from 65% to 96% (Figure 4a). Moreover, SEC chromatograms showed that the dominant VLR peaks for RBC36 and VLRA.R2.1 were at the expected retention time for the theoretical molecular weights of the monomers relative to the molecular weight standards (Figure 4b–d). P1C10 VLR constructs both eluted later than would be expected for monomers from the molecular weight standards. However, because the SDS‐PAGE analysis of P1C10 does not show any low molecular weight species, we attributed the longer retention time of P1C10 on the SEC column to hydrophobic interactions between P1C10 and the stationary phase of the SEC column which are known to lead to increased retention times in SEC. ³¹ To further confirm that the purified products were correctly disulfide‐bonded and did not form disulfide‐bonded multimers as observed in inclusion body refolding experiments (Figure S1), we evaluated P1C10 NT‐6xHis and RBC36 NT‐6xHis proteins by non‐reducing gel and data show that only a very small fraction of these proteins can be found as disulfide‐bonded dimers (Figure S3). In sum, the SHuffle‐produced VLRs can be purified as soluble protein products of uniform size.

FIGURE 3.

Reducing sodium dodecyl‐sulfate polyacrylamide gel electrophoresis assessment of purified, SHuffle‐produced variable lymphocyte receptors (VLRs). Ni‐NTA‐purified VLR preparations were resolved on a reducing 4%–12% bis‐Tris gel and visualized by Coomassie staining. Purified VLRs were loaded in duplicate lanes as depicted on the top of the gel. Molecular weight standards were loaded in lanes 1 and 14. Theoretical molecular weights for each construct are noted at the top of the gel.

TABLE 1.

Liquid chromatography‐mass spectrometry results for purified variable lymphocyte receptors (VLRs).

VLR	Location of 6xHis‐tag	Unique peptides	Unique spectra	Total spectra	Amino acid coverage (%)
P1C10	N‐Terminus	4	10	1130	100
	C‐Terminus	4	11	991	100
RBC36	N‐Terminus	32	166	1292	100
	C‐Terminus	31	148	1022	99
VLRA.R2.1	N‐Terminus	36	105	848	100
	C‐Terminus	37	107	858	99

FIGURE 4.

Analysis of purified SHuffle‐produced variable lymphocyte receptors (VLRs) by size exclusion chromatography (SEC). (a). The purity for each VLR construct was assessed by SEC by integrating the largest peak and comparing to the total area under the elution curve in panels (b)–(d). (b–d) Purified VLR preparations were separated on a Superdex 200 Increase 10/300 GL column. The SEC chromatograms for each VLR were overlaid with that for the molecular weight standard to compare the peak retention times relative to the standard.

3.3. VLR production using the SHuffle strain is scalable

For many applications, it is desirable to produce large amounts of VLR protein. Thus, once production of soluble VLRs in the SHuffle strain was verified, protein production was scaled‐up to generate bulk protein for downstream analyses. Twelve 2 L shake flasks with 1 L of VLR‐producing SHuffle cultures were used for each of the six VLRs. Following induction of protein expression, bacterial pellets containing the expressed VLR were recovered by centrifugation and flash frozen in liquid nitrogen for further processing. VLRs were purified as described above with HisPur Ni‐NTA purification. The yields of purified protein for each VLR are listed in (Table 2). The crude yield of purified VLR reported in Table 2 was adjusted to reflect the percentage of the protein in the major peak as quantified by SEC in Figure 4a. The yield of purified VLR ranged from 14 to 32 mg/L for VLRs P1C10 and RBC36 with lower levels of 2.1–4.1 mg/L for VLRA.R2.1. In addition, all proteins could be concentrated to ≥15 mg/mL for downstream evaluation.

TABLE 2.

Scaled‐up production of variable lymphocyte receptors (VLRs) in the SHuffle strain.

VLR	Location of 6xHis‐tag	Total culture volume (L)	Wet cell pellet mass (g)	Purified protein mass (mg)	Purified protein concentration (mg/ml)	Purified protein yield per culture (mg/L)
P1C10	N‐terminus	12	30	240	18	19
	C‐terminus	12	40	450	18	32
RBC36	N‐terminus	12	40	240	20	18
	C‐terminus	12	40	216	18	14
VLRA.R2.1	N‐terminus	12	40	40	16	2.1
	C‐terminus	12	40	75	15	4.1

3.4. SHuffle‐produced P1C10 is properly folded and active

The fidelity of purified P1C10 produced in the SHuffle strain was verified through binding to bEnd.3 cell produced ECM, the substrate which was used to identify the P1C10 VLR clone. ¹² In an ELISA format, both purified P1C10 NT‐6xHis and P1C10 CT‐6xHis were tested for binding to ECM derived from decellularized bEnd.3 cell cultures. RBC36 NT‐6xHis was used as a non‐ECM targeted negative control in this assay. Both P1C10 constructs exhibited binding to the bEnd.3 ECM when compared to the RBC36 NT‐6xHis negative control (Figure 5). Additionally, the purified P1C10 product was subsequently used for x‐ray structure determination (PDB ID: 9CJ0, ³² ). Thus, SHuffle‐produced P1C10 VLR was demonstrated to be correctly folded and active.

FIGURE 5.

bEND.3 ECM ELISA to confirm activity of P1C10 variable lymphocyte receptors (VLRs). P1C10 NT‐6xHis and P1C10 CT‐6xHis binding was assayed at saturating concentrations with decellularized bEnd.3 ECM. RBC36 NT‐6xHis was used as a non‐binding control. Data represent the mean ± SD of assay replicates (n = 4). Statistical significance was determined by a one‐way ANOVA with Tukey's post‐hoc test. *** represents p < 0.001 compared to both NT and CT P1C10 6xHis, all comparisons with no p value present were not statistically significant.

3.5. SHuffle‐produced RBC36 and VLRA.R2.1 are properly folded and active

Finally, to verify the fidelity of purified RBC36 and VLRA.R2.1, we performed isothermal titration calorimetry (ITC) with their known binding partners. RBC36 has been reported to bind H‐Trisaccharide with a published K _D of 8.3 μM. ⁷ For RBC36 ITC, H‐trisaccharide was injected into an RBC36 containing solution (Figure 6a), ⁷ and the resulting ITC data yielded a K_D of 2.6 μM for RBC36 produced in SHuffle E. coli (Figure 6b). VLRA.R2.1 is known the bind HEL with a published K_D of 7 nM by ITC. ³³ HEL was injected into a VLRA.R2.1 containing solution (Figure 6c), ³³ and the resulting ITC data yielded a K_D of 2 nM for VLRA.R2.1 (Figure 6d). These results indicate that the VLRA.R2.1 and RBC36 proteins produced and purified from SHuffle bacteria were correctly folded and active.

FIGURE 6.

Isothermal titration calorimetry (ITC) with SHuffle‐produced RBC36 and VLRA.R2.1 and their known binding partners to confirm activity. (a) H‐trisaccharide was titrated into TEV‐cleaved RBC36 (no 6xHis tag). (b) ITC data was fit to a nonlinear least squares binding model producing a K _D of 2.6 μM for RBC36. (c) Hen egg lysozyme (HEL) was titrated into TEV‐cleaved VLRA.R2.1 (no 6xHis tag). (d) ITC data was fit to a nonlinear least squares binding model producing a K _D of 2 nM for VLRA.R2.1.

4. DISCUSSION

VLRs are an emerging class of proteins useful in many biomedical applications. The moderate to large‐scale production of functional VLRs will be crucial for characterization and analysis of VLRs used in diagnostics and therapeutics. In the present study, we have demonstrated the ability to produce a purified yield of as much as 32 mg of purified VLR per liter of SHuffle E. coli culture. VLRs P1C10, RBC36, and VLRA.R2.1 were chosen for the expression studies in SHuffle as these are well‐characterized VLRs but are subject to challenging production methods such as production as fusion proteins, production in insect cells and inclusion body refolding. ⁷ , ¹² , ¹³

Of the three VLRs tested, VLRA.R2.1 was the only VLR previously produced in E. coli. Previously, VLRA.R2.1 was produced in BL21‐CodonPlus (DE3)‐RIL E. coli as an inclusion body. ¹³ The BL21‐CodonPlus (DE3)‐RIL E. coli production method required inclusion body separation, washing, resuspension, solubilization and refolding. ¹³ Following refolding, the protein required dialysis, concentration and two forms of purification. ¹³ The labor‐intensive example for VLRA.R2.1 demonstrates the utility of the SHuffle strain for VLR production. In contrast to the inclusion body expression approach, VLRA.R2.1 can be produced simply and purified from the SHuffle strain. Interestingly, VLRA.R2.1 had the lowest yield of the three VLRs produced, 2.1–4.1 mg of protein per liter of culture compared to 14–32 mg of protein per liter of culture for the other VLRs. This difference could be due to the biophysical differences in VLR isoforms, VLRB and VLRA. RBC36 and P1C10 are both in the VLRB family of proteins, which are secreted from B cell‐like lymphocytes. ³⁴ In contrast, VLRAs are cell surface receptors that are displayed on the surface of T Cell‐like lymphocytes. ¹³ , ³⁴ By analogy, native T Cell Receptors (TCRs) are known to be unstable with poor solubility when removed from the T Cell environment as they have evolved as transmembrane proteins that rely on the T Cell membrane for stabilization. ³⁵ Thus, it is also possible that VLRAs may similarly be less soluble when removed from the cell surface. Perhaps because of the instability of VLRAs, there are far fewer solved protein structures of VLRAs compared to VLRBs. The ability to produce soluble VLRAs in SHuffle strain will likely help enable future work to better understand this type of VLR.

Origami E. coli previously was used to express an engineered VLR. The engineered VLR, a scaffold VLR protein termed “repebody,” was engineered for production in Origami E. coli. ¹⁵ The repebody VLR required N‐terminal domain engineering using the internalin‐B cap to promote solubility when produced in the Origami strain. Following the redesign, the yield of the repebody in Origami E. coli was 2 mg of protein per liter of culture. ¹⁵ Here, with different VLR proteins, the SHuffle strain demonstrated higher yields even in the absence of engineering the constructs. However, when we directly compared the production capacity of the SHuffle and Origami strains with the same P1C10 CT‐6xHis VLR, total soluble yields were the same. Thus, while the SHuffle strain has a putative advantage as the result of cytoplasmic DsbC expression to help promote correct disulfide pairing, both strains could be used successfully to produce VLRs.

The location of epitope tags can affect solubility, and these effects have been demonstrated to be sequence‐dependent. ³⁰ However, for the three VLRs expressed in E. coli, the location of the 6xHis epitope tag had only a minor effect on soluble yields of P1C10 and no effect for RBC36 or VLRA.R2.1 in the SHuffle strain. The VLR solubility being less affected by the location of the epitope tag when produced in SHuffle further supports the use of the SHuffle strain for VLR expression.

Finally, we compared the VLR yields to those reported for disulfide bonded IgGs that were also expressed using the SHuffle strain. Production titers for IgGs have been reported to be as low as 0.1 mg of protein per culture liter, with other antibodies such as Humira reported yields as high as 168 or 427 mg/L when co‐expressed with PDI‐GPx7 fusion in high density fermentation. ³⁶ As other examples, Trastuzumab (3.7 mg/L), Cetuximab (4.5 mg/L) and Ranibizumab (10 mg/mL) were also produced in the SHuffle strain. ³⁶ Interestingly, Tastuzumab also was produced in SHuffle as an scFv. ³⁷ The scFv production was compared between BL21(DE3) and SHuffle. ³⁷ While the BL21(DE3) strain produced the Trastuzumab scFv at 214 mg/L compared to the SHuffle yield of 147 mg/L, a solubility assessment showed that the SHuffle strain fraction of soluble yield was approximately twice that of the BL21(DE3) soluble fraction, making the SHuffle production more suitable for soluble Trastuzumab scFv production. ³⁷ Therefore, compared to mammalian IgGs and scFvs, the VLRs tested here were produced in similar ranges.

5. CONCLUSION

In conclusion, we have demonstrated the utility of SHuffle bacteria for the facile expression and scale‐up of VLRs. Importantly, we were able to produce three different VLRs without the need to engineer VLR mutations to assist production, add fusion partners, or refold from insoluble inclusion bodies. As such, this study provides a method to produce high quality VLRs using the simple, scalable SHuffle bacterial host.

AUTHOR CONTRIBUTIONS

Elizabeth A. Appelt: Conceptualization; methodology; investigation; validation; formal analysis; visualization; writing – review and editing; writing – original draft. James B. Thoden: Conceptualization; methodology; investigation; validation. Candis Dancy: Investigation. Hannah D. Bachmeier: Investigation. Seth A. Gehrke: Investigation. Melanie A. Staffenson: Investigation; writing – review and editing. Ivan Rayment: Conceptualization; supervision. Moriah E. Katt: Investigation; validation; writing – review and editing; resources. Hazel M. Holden: Resources; supervision. Eric V. Shusta: Conceptualization; methodology; resources; writing – review and editing; project administration; funding acquisition.

FUNDING INFORMATION

This work was supported in part by funding from the National Institutes of Health awarded to Eric V. Shusta (NIH grant numbers NS099158 and NS118028) as well as funding from the UW–Madison Biotechnology Training Program supported by grant NIH 5T32 GM135066, and M. Katt start‐up funds from West Virginia University.

CONFLICT OF INTEREST STATEMENT

Eric V. Shusta has filed a US patent application regarding uses for the P1C10 VLR.

Supporting information

Appendix S1: Supplementary Information.

Appelt EA, Thoden JB, Dancy C, et al. A method for facile production of variable lymphocyte receptors using SHuffle Escherichia coli . Biotechnol. Prog. 2025;41(3):e3530. doi: 10.1002/btpr.3530

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.