High-level fed-batch production of recombinant type II cholesterol oxidase from Brevibacterium sterolicum and its validation for clinical diagnostic applications

Highlights

• •Scalable fed-batch production of recombinant type II cholesterol oxidase.
• •High-level production achieved in chemically defined minimal medium.
• •C-terminal His-tagged BCO outperforms N-terminal variant.
• •Validated for clinical diagnostics with excellent analytical performance.
• •Stable for at least one year at 4 °C with preserved activity.

Abstract

Cholesterol oxidase (CO, EC 1.1.3.6) is widely used in clinical diagnostics for the enzymatic determination of cholesterol in human serum and plasma. Among bacterial COs, the type II CO from Brevibacterium sterolicum (BCO), characterized by a covalently bound FAD cofactor, is particularly attractive due to its high stability and favorable kinetic properties. In this study, we developed an integrated bioprocess for recombinant BCO production in Escherichia coli. Two His-tagged variants were constructed and compared: a C-terminally tagged form (BCO–H) and an N-terminally tagged form (H-BCO). Expression was optimized in a chemically defined modified M9 medium and scaled up using a fed-batch strategy. Both variants reached high cell densities (OD₆₀₀ ≈ 50) and showed a > 20-fold increase in volumetric productivity compared to shake-flask cultures. BCO–H outperformed H-BCO, reaching a final productivity of 102 ± 1.5 KU/L. Single-step Ni-affinity purification yielded highly pure enzymes with specific activities comparable to commercial CO. Both enzymes were successfully incorporated into a commercial cholesterol diagnostic kit formulation and evaluated according to European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) quality specifications. The in-house formulations showed excellent linearity, high precision, low bias, and total error within desirable limits, with comparable or faster kinetics than the reference formulation. Importantly, analytical performance was fully preserved after one year of storage at 4 °C. Overall, this work establishes a scalable fed-batch production process for recombinant type II CO suitable for use in real clinical diagnostic settings, achieving the highest reported volumetric productivity to date.

Graphical abstract

1. Introduction

Cholesterol oxidase (EC 1.1.3.6, CO) is a bifunctional enzyme that catalyzes the oxidation of cholesterol and the subsequent isomerization of cholest-5-en-3-one, a transient intermediate, resulting in equimolar amounts of cholest-4-en-3-one and hydrogen peroxide as the final products of the reaction [1]. The CO is an enzyme of great commercial and analytical value, widely used for the enzymatic determination of cholesterol concentrations in human serum. The determination of serum cholesterol has become essential for the diagnosis and risk assessment of cardiovascular diseases such as atherosclerosis, coronary heart disease, and thrombosis. High levels of cholesterol and cholesteryl ester (hypercholesterolemia) have been directly linked to such conditions, while abnormally low levels (hypocholesterolemia) have been associated with cancer, depression, and respiratory diseases. Additionally, hypercholesterolemia has been implicated in the pathogenesis of Alzheimer’s disease through oxidative stress mechanisms [2]. Enzymatic assays employing CO are highly sensitive, specific, and simpler than conventional chemical methods, contributing to their widespread use. Since most cholesterol in serum exists in esterified form, a pre-treatment of the sample with cholesterol esterase is typically required to hydrolyze esters and release free cholesterol. CO then catalyzes the oxidation of cholesterol to cholest-4-en-3-one, producing hydrogen peroxide, which can subsequently be detected through oxidative coupling with a chromogenic dye and quantified spectrophotometrically [3,4].

CO is a flavoenzyme that exists in two forms, distinguished by the nature of the interaction between the flavin adenine dinucleotide (FAD) cofactor and the protein. In type I cholesterol oxidase, FAD is associated with the enzyme through noncovalent interactions, whereas in type II, the cofactor is covalently attached to the protein [1]. Both native and recombinant COs have been produced, purified, and characterized from a variety of bacterial species, including Streptomyces spp., Brevibacterium spp., Rhodococcus spp., Bacillus spp., and Burkholderia spp., Escherichia fergusonii, among others [[5], [6], [7]]. Particularly, the type II CO from Brevibacterium sterolicum (BCO) has attracted considerable attention due to its distinct functional properties that make it a suitable biotechnological tool. This enzyme consists of 561 amino acids in its mature form and is classified as a type II cholesterol oxidase because it contains a FAD cofactor covalently bound to His69. It demonstrates excellent kinetic parameters, and the stable cofactor interaction makes it a better choice than type I COs for applications where FAD dissociation compromises biocatalytic performance, such as immobilization-based processes including biosensor fabrication [[8], [9], [10]].

The BCO was previously overexpressed in E. coli by Volontè and col., [10] who demonstrated the successful expression of the mature and functional enzyme lacking the 52-amino-acid pre-sequence. Various cultivation and induction conditions were evaluated in shake-flask cultures, and batch bioreactor experiments were subsequently performed using rich media. These promising results provided a solid basis for further investigation and suggested that the improved overproduction of BCO would support its development as a biotool for biotechnological applications. In the present study we expressed and validated for diagnostic use two recombinant variants of BCO: one bearing and C-terminal His-tag (BCO–H) and the other a N-terminal His-tag (H-BCO). We developed a fed-batch bioreactor process using industrially relevant minimal media to enable scalable enzyme production and demonstrated that both enzymes were fully functional in a commercial diagnostic kit and remained stable for at least one year under these conditions. This work constitutes a significant advancement in the development of COs and their production processes for use as tools in biochemical diagnostic applications.

2. Materials and methods

2.1. Bacterial strain and chemicals

E. coli BL21(DE3) strain and pET expression vectors were from Novagen (Merck KGaA; Darmstadt, Germany). Plasmids were purified using the Wizard Plus SV Miniprep DNA purification system (Promega, Madison, WI, USA). Kanamycin sulfate was from Gibco™ (Cat N°11815024), ampicillin from KLONAL Laboratorios (EFA.003/D) and Isopropyl β-D-1-thiogalactopyranoside (IPTG) from Chem-Impex Int'l. Inc (Wood Dale, IL, USA. Cat N°00194). Proteins were quantified using the Protein Assay Dye Reagent Concentrate (Bio-Rad; Hercules, CA, USA. Cat N°5000006). Culture media were prepared using ultrapure water and analytical grade chemicals. SDS-page gels and buffers were prepared using ultrapure water and molecular biology grade chemicals.

2.2. Expression plasmids

BCO–H and H-BCO were expressed from the plasmids pET25-BCO and pET50-BCO, respectively. These expression vectors contain the coding sequence of mature form of type II BCO (protein data: PDB: 1I19) (Fig. 1A and B). For H-BCO, the N-terminal His-tag and BCO coding sequence were commercially synthesized, codon-optimized for E. coli expression, and cloned into the pET50 vector (GenScript, https://www.genscript.com/). For BCO–H, the BCO coding sequence was similarly synthesized and codon-optimized for E. coli, then cloned into the pET25 vector to generate a C-terminal fusion with the vector-encoded His-tag (GenScript, https://www.genscript.com/). The resulting plasmids were transformed into E. coli BL21(DE3) for protein expression.

Fig. 1.

Amino acid sequence and structure of BCO variants expressed in this study. (a) N-terminal sequence of BCO, which is naturally synthesized as a 613-amino-acid precursor. Removal of the N-terminal pre-sequence (52 aa) yields the mature and fully active enzyme form (561 aa). (b) N- and C-terminal sequences of the BCO variants expressed in this study and their alignment with the native mature BCO sequence (PDB: 1I19). BCO–H: variant containing a His-tag at the C-terminus. H-BCO: variant containing a His-tag at the N-terminus. (c) 3D structure of BCO–H and H-BCO predicted by AlphaFold. The N- and C-termini, corresponding to the regions where the His-tag was added depending on the variant, are indicated.

2.3. Small-scale protein expression

Transformed E. coli cells were grown at 37 °C with shaking in 50 mL of LB or low-Mg M9 medium [11], supplemented with the indicated antibiotic [ampicillin (100 µg/mL) for BCO–H or kanamycin (25 µg/mL) for H-BCO]. When the cultures reached an OD₆₀₀ of 0.6–0.8, protein expression was induced by the addition of 0.4 mM IPTG, followed by overnight (ON) incubation at 20 °C with shaking. After the ON incubation, an aliquot of the cultures was centrifuged at 5000 rpm for 10 min. The resulting cell pellets were collected, resuspended in buffer (50 mM Tris–HCl pH 7.5, 500 mM NaCl) and lysed by sonication. Following centrifugation, the supernatants containing the soluble protein fraction were recovered. Enzyme activity assays (Section 2.6.2.) were performed on this extract, and protein production level was expressed as volumetric productivity (KU/L).

2.4. Bioreactor protein expression

Fed-batch experiments were carried out in a Biostat-A bioreactor (Sartorius Stedim, Goettingen, Germany), equipped with a 2-L stirred tank and integrated data acquisition and control systems. Cultivations were performed using modified low-Mg M9 medium following the procedure described by Godino et al. [11] with the corresponding modifications. The cultures were initiated by inoculating the bioreactor with an ON seed culture to achieve an initial OD₆₀₀ of 0.15 in 1 L of culture medium supplemented with ampicillin (BCO–H) or kanamycin (H-BCO). The cultures were grown in batch mode at 36.5 °C with shaking at 700 rpm and an airflow rate of 0.6 L min⁻¹ until reaching an OD₆₀₀ of approximately 10, corresponding to glucose depletion as indicated by a sharp rise in dissolved oxygen (DO). At this point, the process was shifted to fed-batch mode, induced with 0.4 mM IPTG, and the temperature was decreased to 20 °C. Feeding was carried out using an external peristaltic pump with a feeding solution containing 350 g/L glucose, 1 g/L MgSO₄·7H₂O, and 2.5 mL/L trace element solution [11], maintained at a constant flow rate of 5 mL/h for 20–23 h. The pH was automatically controlled at 7 by adding either 1.5 M H₃PO₄ or 25% (w/v) NH₄OH solutions. DO level was automatically maintained at 20% saturation by controlling both airflow and stirrer speed. Samples were taken periodically to determine residual glucose, cell growth and protein production level. The glucose was analyzed enzymatically by using the Glicemia kit (Wiener Laboratorios SAIC, Argentina). Cell growth was determined by measuring culture optical density at 600 nm (OD₆₀₀). Protein production level was analyzed as described in previous section (Section 2.3.) and expressed as volumetric productivity (KU/L). At the end of the culture, the total cells were collected for recombinant protein purification.

2.5. Recombinant protein purification

The collected cells from bioreactor cultures were resuspended in binding buffer (50 mM Tris–HCl pH 7.5, 500 mM NaCl) and lysed by high pressure homogenization (Avestin Emulsiflex C3). The supernatants of centrifuged lysates were loaded onto High Affinity Ni-Charged Resin FF Prepacked Column (GenScript Cat. No. L00683–51). The unbound proteins were removed by washing with 40 column bed volumes of binding buffer containing increased concentrations of imidazole (0, 2, 5 and 10 mM). His-tagged BCOs were eluted using 400 mM imidazole. Purified recombinant proteins were quantified by the Bradford method using Bio-Rad Protein Assay Dye and analyzed by SDS-PAGE (12%) [11,12]. The purified His-tagged BCOs were conformationally and functionally characterized as described in Section 2.6. and were validated for application in biochemical diagnostic kits (Section 2.7.).

2.6. Conformational and functional characterization

2.6.1. Conformational analysis by circular dichroism (CD)

CD studies were performed using a Jasco J-1500 spectro-polarimeter equipped with a Peltier temperature control system. Far-UV CD spectra were collected in a 0.1-cm path length quartz cuvette (Hellma) with a scan speed of 50 nm.min^-1 a response time of 2 s, and a bandwidth of 1 nm, in a range of 195–260 nm, using a protein concentration of 4.5 μM. All spectra are an average of at least six scans. All spectra were corrected for solvent contribution. Background subtraction was performed using protein-free buffer CD spectra measurements. CD values were converted to mean residue ellipticity (MRE) in units of degree.cm².dmol^-1 using Jasco Software. Prism 5 (GraphPad, San Diego, CA) and Origin (OriginLab Corporation, Northampton, MA) were used for plotting and curve fitting in the CD experiments. We estimated the secondary structure content using the algorithm implemented in CD Multivariate SSE (Jasco).

2.6.2. Enzyme activity and kinetic analysis

CO catalyzes the oxidation of cholesterol to cholestenone and H₂O₂. The activity of cholesterol oxidase was determined by quantifying the H₂O₂ produced through a coupled reaction with horseradish peroxidase and aminopyrine. The formation of the resulting quinoneimine dye was measured spectrophotometrically. One unit of cholesterol oxidase is defined as the amount of enzyme that catalyzes the formation of 1 µmol of H₂O₂ per minute at 37° [13,14]. Kinetic parameters (K_m and K_cat) for cholesterol were estimated by fitting the initial reaction rates measured at different substrate concentrations to the Michaelis–Menten equation using nonlinear regression in Origin (OriginLab Corporation, Northampton, MA). Values are reported as the mean ± SE of three independent experiments.

2.7. Validation for biochemical diagnostic applications

Validation was conducted according to the quality requirements established by the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM), based on the minimal criteria for biological variation, within the context of a commercial diagnostic kit (Colestat, Wiener Laboratorios SAIC, Rosario, Argentina; https://www.wiener-lab.com.ar/). Equivalent formulations to the commercial kit (Col_Ref) were prepared using our BCO–H and H-BCO and designated Col_BCO–H and Col_H-BCO, respectively. These formulations were comparatively evaluated against the commercial reference kit following the protocols established by the Clinical & Laboratory Standards Institute (CLSI) for the assessment of analytical performance.

All cholesterol determinations during the evaluation of analytical parameters were performed according to the manufacturer's instructions, using a CMD600i Wiener lab. automated clinical analyzer (Colestat kit prospectus [15]).

The analytical parameters evaluated included:

2.7.1. Linearity

Linearity assessment was performed following the guidelines of the CLSI EP6-A protocol [16]. A linearity panel was prepared by mixing a pool of fresh serum with high total cholesterol concentration and a pool of fresh serum with low total cholesterol concentration to generate 11 equidistant total cholesterol concentration levels. Each cholesterol concentration level was measured in quintuplicate. Data analysis was performed using Analyse-it software version 6.15, applying a weighted least squares (WLS) linear regression model with intercept. The acceptance criterion was set at a linearity deviation of up to ±6.2 (minimum bias of EFML criteria, 2024). The assay must demonstrate linearity up to at least 400 mg/dL.

2.7.2. Detection capability

The limit of detection (LOD) and limit of quantification (LOQ) were assessed following the CLSI EP17-A2 protocol [17]. A low total cholesterol concentration pool was subjected to four serial 1:2 dilutions, resulting in five levels of very low total cholesterol concentration. Cholesterol quantification within this low concentration range was performed in quintuplicate over five consecutive days. Additionally, a sample consisting of physiological saline solution (analyte concentration zero) was analyzed daily in duplicate. Data was processed using Analyse-it for Excel v6.15 software, applying the Profile and Detection Capability functions within the Precision Module.

2.7.3. Precision

Precision was evaluated using the CLSI EP15-A protocol [18]. Cholesterol was measured over 5 days, with one run per day and five replicates per run (5 × 1 × 5; N = 25), using Wiener lab. commercial controls: Standatrol® SE Level 1 (normal range) and Level 2 (pathological range) [19]; and a pooled serum sample with a cholesterol concentration close to a medical decision level (230 mg/dL). Data and statistical analysis were performed using Analyse-it for Excel v6.15 software, within the Precision Module. Imprecision values [expressed as Analytical Coefficient of Variation (CVa)] were compared with the EFML (2024) quality requirement (desirable CVa: 2.6%).

2.7.4. Method comparison and bias evaluation

Method comparison was performed according to the CLSI EP09c protocol [20]. Systematic differences between methods (Col_Ref vs Col_BCO–H and Col_Ref vs Col_H-BCO) were assessed using a representative set of 50 native patient samples (including serum and plasma samples) covering the clinically common measurement range. Samples were analyzed in parallel by both methods and regression analysis was carried out with Analyse-it v6.15 software using the nonparametric Passing–Bablok regression. Bias was calculated at clinically relevant concentrations and compared with EFML (2024) quality requirements (desirable Bias: 4%). Total error (TEa) was determined by combining CVa with bias [TEa%=∣Bias%∣+z × CV%, z = 1.65 (95% one-sided confidence level)] and compared with the EFML (2024) desirable limit of 8.3%.

2.7.5. Kinetics

The colorimetric reaction product was monitored by measuring absorbance at 505 nm every 18 s for 10 min at 37 °C. The Wiener lab. Calibrador A Plus calibrator [21] was used as the cholesterol source. All measurements were performed in triplicate.

2.7.6. Stability

An accelerated stability assay was carried out using a protocol adapted from a previously described method [11]. Both the commercial kit and the kits containing our BCO variants were incubated at 37 °C for 57 days. Kit performance was periodically evaluated using the Calibrador A Plus calibrator as the cholesterol source and performed under manual conditions (rather than using the automated Wiener analyzer) following the protocol specified for the commercial kit. Moreover, kits were stored at 4 °C for one year to evaluate their stability under recommended storage conditions. After this period, their performance was assessed through cholesterol measurements using 59 serum samples and 32 plasma samples. The results were then compared, using Passing–Bablok regression, with those obtained using the reference reagent to determine potential changes in analytical performance.

3. Results

3.1. Small-scale expression analysis of BCO variants

Two variants of the mature form of BCO were overexpressed: (1) BCO–H, fused to a C-terminal His-tag, and (2) H-BCO, fused to an N-terminal His-tag (Fig. 1B and C). Expression was carried out in E. coli BL21 (DE3) at 20 °C using IPTG as the inducer, and a small-scale comparative study was performed in a rich complex medium (LB) and a modified minimal defined medium (M9), the latter considered more suitable for industrial-scale production. The M9 medium outperformed LB for both variants, yielding volumetric productivities of 4.5 ± 0.4 KU/L for BCO–H and 3.0 ± 0.3 KU/L for H-BCO (Fig. 2A). Maximum biomass accumulation was also observed in M9 for both variants [OD₆₀₀ = 4.8 ± 0.3 (BCO–H) and 3.9 ± 0.5 (H-BCO) in induced cultures] (Fig. 2B). These results provided the basis for using the modified M9 medium in scale-up studies in bioreactor, as described in the following section.

Fig. 2.

Productivity of BCO–H and H-BCO and cell biomass accumulation in LB and low-Mg M9 culture media. (a) Volumetric productivity (KU/L). (b) Cell density (OD₆₀₀) at the end of cultivation process. Values represent means ± SE of four independent experiments. * indicates significant differences (p < 0.05) according to Student’s t-test.

3.2. Bioreactor expression of BCO variants

The bioreactor cultivations were performed using modified M9 medium under fed-batch conditions, as described in Section 2.4. Cell growth (OD600) and volumetric productivity (KU/L, measured in the soluble fraction) were monitored over time during the cultivations of BCO–H (Fig. 3A) and H-BCO (Fig. 3B). Both variants reached high cell densities, with final OD600 values of approximately 50. This point was considered the end of the process, as critical operational parameters remained stable up to that stage: DO ≥ 20%, glucose concentration ≤ 2 g/L, pH between 6.8 and 7.0, and controlled foam levels. Under these experimental conditions, end-point productivities reached 102 ± 1.5 KU/L for BCO–H (Fig. 3A) and 63 ± 3 KU/L for H-BCO (Fig. 3B). Despite similar final cell densities, BCO–H consistently outperformed H-BCO in terms of volumetric productivity. These results are consistent with the small-scale experiments, in which BCO–H had already been identified as the more productive variant (Fig. 2A).

Fig. 3.

Bioreactor production of BCO–H and H-BCO using fed-batch culture and low-Mg M9 medium. (a) y (b) Cell density (OD₆₀₀) and volumetric productivity (KU/L) during fed-batch culture of BCO–H or H-BCO, respectively. Arrow indicates induction point. (c) cell-specific productivity (KU·L⁻¹·OD₆₀₀⁻¹) of BCO–H and H-BCO during induction stage. * indicates significant differences (Student's t-test, p < 0.05) between BCO–H and H-BCO at the endpoint of the fed-batch culture. (d) SDS-PAGE analysis of BCO–H and H-BCO purification. Lane SP: total soluble protein extracts; Lane M: protein molecular weight marker (PageRuler Unstained Protein Ladder; Thermo Fisher); lanes E1 and E2: elution fractions (E1=1.34 µg of BCO–H or 1.6 µg of H-BCO, E2=0.68 µg of BCO–H or 0.8 µg of H-BCO). Bioreactor experiments were performed in two independent biological replicates. The data shown correspond to one representative biological replicate.

Fig. 3C shows cell-specific productivity (volumetric productivity normalized by OD600) during the induction phase. H-BCO cell-specific productivity reached 1.23 ± 0.06 KU·L⁻¹·OD₆₀₀⁻¹, stabilizing after 15 h post-induction. In line with volumetric productivity trends, BCO–H displayed a sustained increase, continuing to rise beyond 15 h and reaching approximately 2 KU·L⁻¹·OD₆₀₀⁻¹ at the end of the process.

Overall, the fed-batch strategy enhanced productivity of both variants by more than twenty-fold compared to small-scale experiments, with BCO–H standing out due to its superior performance.

3.3. Purification and characterization of BCO variants

BCO–H and H-BCO were purified from bioreactor cultures using single-step His-tag affinity chromatography. SDS-PAGE analysis of total soluble protein extracts revealed both enzyme variants as major bands of approximately 63 kDa (62.6 kDa for BCO–H and 63.4 kDa for H-BCO) (Fig. 3D). The soluble protein extracts were applied to a Ni-charged resin column, washed, and the His-tagged recombinant proteins were eluted with imidazole. A high-purity enzyme preparation was efficiently obtained (Fig. 3D), yielding 2 g/L of purified BCO–H and 1 g/L of purified H-BCO.

Far-UV CD spectra of the purified BCO–H and H-BCO variants displayed nearly identical spectral profiles (Fig. 4A and B). Both spectra exhibited the characteristic features of a mixed α/β protein, with a broad negative band centered around 215–218 nm. Secondary structure estimation further supported this observation. The BCO–H construct was estimated to contain 29.8% α-helix, 21.0% β-sheet, 10.1% turns, and 39.1% unordered structures, whereas the H-BCO construct contained 28.5% α-helix, 22.1% β-sheet, 10.2% turns, and 39.2% unordered structures. The differences between both constructs were below 2% for all structural elements.

Fig. 4.

Far-UV CD spectra of purified BCO variants. (A) Experimental far-UV CD spectra expressed as mean residue weight ellipticity (MRE). (B) Normalized CD spectra highlighting the similarity in spectral shape between the two protein variants.

The enzymatic activity of the purified proteins was evaluated as described in Section 2.6.2. BCO–H showed a specific activity of 52 ± 4 U/mg and H-BCO of 48 ± 2 U/mg, with no significant differences (Table 1). Moreover, both variants displayed specific activities comparable to those of commercial enzymes used as controls (Table 1). To further evaluate the effect of His-tag position on enzyme activity, kinetic parameters were determined using cholesterol as substrate. Both variants exhibited similar substrate affinities, with K_m values of 0.032 ± 0.0067 mM for BCO–H and 0.039 ± 0.0068 mM for H-BCO. Likewise, BCO–H exhibited a k_cat value of 61.7 ± 2.8 s⁻¹, whereas H-BCO exhibited a k_cat value of 53.7 ± 5.3 s⁻¹. None of these differences reached statistical significance (Student's t-test, p = 0.50 for K_m and p = 0.25 for k_cat).

Table 1.

Specific activity and volumetric productivity of BCOs produced in this study, commercial preparations, and recombinant enzymes reported in the literature.

Enzyme	Specific activity (purified protein)	Maximum volumetric productivity	Reference
BCO–H	52.1 ± 4.5.1 U/mg a	102 ± 1.5 KU/L	This study
H-BCO	48.4 ± 2.1 U/mg a	63 ± 3 KU/L	This study
Commercial CO (sigma C8868)	- Reported by the manufacturer: ≥50 U/mg protein. -Measure in this work: 70.5 ± 2.3 U/mg.	No data	link
Commercial CO (Toyobo COO-331)	-Reported by the manufacturer: ≥12 U/mg-solid. -Measure in this work: 23.4 ± 1.4 U/mg.	No data	link
BCO (C-terminal his-tag)	No data	25 KU/L	[10]
ChoB of Brevibacterium sp.	16 ± 0.3 U/mg (non-tagged) 12 ± 0.1 U/mg (N-terminal his-tag) 4 ± 0.3 U/mg (C-terminal his-tag)	3.35 KU/L (non-tagged) 0.046 KU/L (N-terminal his-tag) 0.014 KU/L (C-terminal his-tag) b	[24]
ChOA of Streptomyces sp. SACOO (N-terminal his-tag)	7.07 U/mg	0.157 KU/L c	[22]
ChO of Chromobacterium sp. DS1 (N-terminal his-tag)	14.3 U/mg	2.115 KU/L	[23]
PA157 cholesterol oxidase of Pseudomonas aeruginosa PA157 (non-tagged)	11.6 U/mg	0.522 KU/L d	[33]
ChONs of Nocardioides simplex VKM Ac-2033D (N-terminal his-tag)	22 ± 2 U/mg	No data	[25]

^aNo significant differences (Student's t-test, p = 0.46).

ᵇVolumetric productivity (KU/L) calculated from reported total enzymatic activities measured in 100 mL of crude extract (332, 4.5, and 1.4 U for non-tagged, N-terminal His-tagged, and C-terminal His-tagged variants, respectively).

ᶜVolumetric productivity (KU/L) calculated from the total enzymatic activity reported under optimized conditions (78.5 U per 500 mL culture).

ᵈVolumetric productivity (KU/L) calculated from reported purification data (2.34 mg purified enzyme with a specific activity of 11.6 U·mg⁻¹ obtained from 200 mL culture), corrected for a 26% purification yield to estimate total pre-purification activity.

3.4. Validation of BCO variants for human biochemical diagnostic applications

As described in the Methods Section 2.7., equivalent formulations to the commercial Colestat kit (Col_ref) were prepared using our in-house cholesterol oxidases (Col_BCO–H and Col_H-BCO) and comparatively evaluated to assess their analytical performance according to EFLM quality criteria. The Colestat kit is based on three reactions: in the first reaction (Eq. (1)), the esterified cholesterol present in the sample is hydrolyzed by cholesterol esterase (EC 3.1.1.13, ChE), releasing free cholesterol. In the second reaction (Eq. (2)), CO catalyzes the oxidation of cholesterol to cholest-4-en-3-one, producing hydrogen peroxide. The hydrogen peroxide generated is then used in the third reaction (Eq. (3)), catalyzed by peroxidase (EC 1.11.1.7, POD), to produce a chromogen. The color intensity of the resulting chromogen is directly proportional to the cholesterol concentration in the sample.

$$Cholesterylester+H_{2}0\stackrel{ChE}{\to }cholesterol+fattyacids$$ (1)

$$Cholesterol+0_2\stackrel{CO}{\to }cholest-4-en-3-one+H_2{0}_2$$ (2)

$$2H_2{0}_2+acceptor+4-amino-phenazone\stackrel{POD}{\to }redquinoneimine\left(chromogen\right)+4H_{2}0$$ (3)

3.4.1. Linearity and detection capability

Linearity was evaluated across a panel of 11 known cholesterol concentrations.

An excellent linear correlation (R² > 0.999) was observed between the expected and measured concentrations for both Col_BCO–H and Col_H-BCO kits (Fig. 5A).

Fig. 5.

Linearity comparison of in-house and commercial cholesterol oxidase kits. (a) Linear correlation between measured and expected cholesterol values (R²: Col_Ref = 0.999,885, Col_BCO–H = 0.999,914, Col_H-BCO = 0.999,915). Data represent the mean ± SD of five measurements obtained at each of the eleven cholesterol levels tested. (b), (c), and (d) show the linearity deviation plots for Col_Ref, Col_BCO–H, and Col_H-BCO, respectively. The solid black line indicates the theoretical value corresponding to a perfect linear fit (0% deviation). Black (Col_Ref), blue (Col_BCO–H), and pink (Col_H-BCO) squares represent the percentage difference between the mean measured values and the linear fit. The dashed black lines denote the acceptable linearity deviation limits (±6.2%).

The linearity deviation did not exceed ±6.2% remaining well within the acceptance limit defined by the EFLM (Fig. 5B-D). These results confirm that the method displays linear behavior throughout the studied interval, ensuring reliable measurements even at cholesterol concentrations above 400 mg/dL (Fig. 5).

Detection capability, expressed as LOD and LOQ, was also evaluated. As shown in Table 2, both Col_BCO–H and Col_H-BCO formulations exhibited high analytical sensitivity, sensitivity comparable to the commercial reference kit (Table 2).

Table 2.

Analytical performance of in-house vs. commercial cholesterol oxidase diagnostic kits.

Analytical parameter	ColRef	ColBCO–H	ColH-BCO	Desirable (EFLM)
LoD	1.1 mg/dL	0.8 mg/dL	1.8 mg/dL	——-
LoQ	6.0 mg/dL	4.3 mg/dL	5.8 mg/dL	——-
CVa	0.8–1%	0.7–1%	0.6–0.7%	2.6%
Bias	N/A	−0.5–0.9%	−3.1/− 3.6%	±4.0%
TEa	N/A	1.6–2.6%	4.2–4.6%	8.3%

N/A (not applicabble): Bias and TEa were calculated relative to the Col_Ref; thus, these parameters are not applicable to Col_Ref.

3.4.2. Precision and exactitude

Method precision was evaluated at three concentration levels using commercial control materials (Standatrol® SE1, SE2) and a pooled serum sample. As summarized in Table 2, both formulations containing the in-house enzymes exhibited precision performance equivalent to that of the commercial reference kit. The imprecision, expressed as CVa, was ≤1.0% across all tested levels, remaining well below the desirable analytical imprecision limit of 2.6% established by the EFLM (Table 2).

A method comparison analysis between the candidate formulations and the commercial reference method was further assessed using Passing–Bablok regression with native patient samples covering the clinically relevant measurement range. Both formulations demonstrated excellent agreement with the reference kit. The regression equations were y = −3.156 + 1.023x for Col_BCO–H and y = −1.008 + 0.9735x for Col_H-BCO, indicating slopes close to unity and minimal systematic deviation (Fig. 6A and B, respectively). Bias analysis confirmed the absence of significant systematic error, with mean biases of ≤ 1% for Col_BCO–H and approximately −3% for Col_H-BCO, both within the desirable limit of ±4% established by the EFLM. The TEa remained well below the EFLM desirable limit of 8.3%, ranging between 1.6% and 2.6% for Col_BCO–H and 4.2% to 4.6% for Col_H-BCO (Table 2). Overall, these results confirm that both in-house formulations exhibit analytical performance comparable to the commercial reference kit, ensuring accurate and reliable cholesterol quantification across the evaluated range.

Fig. 6.

Passing–Bablok regression analysis plots. Comparison of cholesterol measurements (mg/dL) between in-house and reference kits using 50 serum and plasma samples. (a) Col_BCO–H vs Col_Ref. (b) Col_H-BCO vs Col_Ref. The green diagonal line is the line of identity, and the red line is the calculated Passing–Bablok line of regression.

3.4.3. Kinetics and stability

The kit reaction kinetics were evaluated by monitoring the formation of the reaction product (chromogen) over time for both in-house formulations in comparison with the reference kit. As shown in Fig. 7, all formulations exhibited a typical kinetic profile characterized by a rapid increase in product formation followed by a plateau corresponding to the reaction end-point. Both Col_BCO–H and Col_H-BCO formulations displayed kinetic behavior comparable to the reference formulation, achieving similar maximum absorbance values. However, Col_BCO–H and Col_H-BCO reached the reaction endpoint earlier than Col_Ref, indicating a faster reaction rate and a shorter response time. Once the end-point was reached (defined at 5 min according to the commercial kit specification), the absorbance signal remained stable for at least 10 min, confirming the temporal stability of the enzymatic system and the absence of product degradation or signal drift.

Fig. 7.

Kit reaction kinetics. Product formation is monitored as an increase in absorbance at 505 nm, corresponding to the chromogenic product generated at the end of the reaction (Eq. (3), Section 3.4.). Data represent the mean values at each time point ± SD.

Finally, stability was evaluated under both accelerated and real storage conditions. Under accelerated conditions (37 °C), despite minor fluctuations in reactivity associated with the manual assay format, the overall stability trends were consistent across all formulations. The kits containing the recombinant BCO variants produced in this study exhibited behavior comparable to that of the commercial kit throughout the incubation period, with similar reactivity profiles (Fig. 8A). To evaluate stability under real storage conditions, formulations were stored at 4 °C for one year, corresponding to the recommended storage conditions of the commercial kit. After this period, analytical performance was assessed using 59 serum and 32 plasma samples, and results were compared with the commercial reference. Passing–Bablok regression demonstrated excellent agreement with the reference method, with slopes close to unity and negligible systematic deviation (y = −0.8053 + 1.010x for Col_BCO–H; y = −1.981 + 1.015x for Col_H-BCO), confirming that analytical performance was preserved after one year of storage at 4 °C (Fig. 8B–D). Moreover, linearity, precision, and reaction kinetics, were also evaluated after storage period and yielded results equivalent to the reference formulation (data not shown).

Fig. 8.

Kit stability. (a) Accelerated stability assay. Kit reactivity was periodically measured over 57 days at 37 °C. Reactivity is expressed as a percentage relative to the initial measurement (day 0 = 100%). (b–d) Stability under real storage conditions. (b) Cholesterol was measured in representative patient samples covering a range of concentrations using kits stored for one year at 4 °C. (c) y (d) Passing–Bablok regression analysis plots comparing cholesterol measurements (mg/dL) obtained with Col_BCO–H (c) and Col_H-BCO (d) versus Col_Ref stored for one year at 4 °C.

4. Discussion

To the best of our knowledge, the study by Volontè et al [10] remains the only detailed scientific report describing a recombinant production system for type II cholesterol oxidase from B. sterolicum expressed in E. coli. In that work, BCO was produced in its mature form as a fusion protein with a C-terminal His-tag, yielding excellent results; however, to date, no data has been reported on N-terminal fusion constructs for this enzyme. Several studies on other CO, including commercially available enzymes, have reported the successful use of N-terminal tags [[22], [23], [24], [25], [26], [27]]. Moreover, the N-terminus of BCO appears to be exposed, structurally disordered, and distant from the active site and FAD cofactor (Fig. 1C), making it a potentially suitable site for tag fusion. Therefore, in this work production and comparative analysis of both C- and N-terminal His-tagged BCO variants were undertaken.

Both His-tagged variants were successfully produced at both shake-flask and bioreactor scale under fed-batch operation. At the bioreactor level, we reached final biomass concentrations close to OD₆₀₀ 50 and volumetric productivities of 102 ± 1.5 KU/L for BCO–H and 63 ± 3 KU/L for H-BCO (Table 1), representing more than a 20-fold increase compared with our shake-flask experiments and a four-fold improvement over previously reported BCO production by Volontè et al [10]. In that study, BCO production was carried out using complex media (Terrific Broth) under batch cultivation as best condition, achieving volumetric productivities around 25 KU/L (Table 1). In contrast, our process was developed in a modified M9 medium, which is compatible with industrial manufacturing demands due to its lower cost, defined composition, and reduced lot-to-lot variability [28]. Furthermore, the volumetric productivities obtained in this study are competitive with those reported for other recombinant cholesterol oxidases produced in heterologous hosts (Table 1). Although direct comparison between studies is limited by differences in enzyme origin, assay conditions, and process scale, our data indicate that the combination of a minimal medium and a fed-batch strategy is highly effective for BCO production and aligns with industrial requirements for robust and scalable processes.

Single-step Ni-affinity chromatography yielded highly pure preparations of both BCO variants, with 2 g/L of purified BCO–H and 1 g/L of H-BCO, and specific activities of 52 ± 4 U/mg and 48 ± 2 U/mg, respectively. These activity values are similar to, or higher than, those reported by the manufacturers of the commercial CO preparations used as controls and recombinant CO reported in scientific literature (Table 1). Far-UV CD analysis revealed nearly identical secondary structure contents for both BCO variants. Likewise, both variants exhibited similar kinetic parameters, with no statistically significant differences in either K_m or k_cat. Together, these findings indicate that the position of the His-tag does not substantially affect the global conformation, substrate affinity, or catalytic properties of BCO.

The main goal of this work was to demonstrate the suitability of BCO–H and H-BCO produced herein for routine clinical use in cholesterol assays. Both BCO enzyme variants showed analytical performance comparable to a commercial formulation and remained stable for up to one year under industrial storage conditions, in compliance with CLSI protocols and EFLM biological variation–derived quality specifications. This study represents the first characterization of these enzymes in a real diagnostic setting.

Our small-scale and bioreactor experiments consistently demonstrated that the BCO–H achieved higher volumetric and cell-specific productivity than the H-BCO, despite both variants reaching similar final biomass levels. Interestingly, the purified enzymes exhibited comparable specific activities, kinetic parameters and performance in diagnostic kit, suggesting that His-tag orientation mainly influences expression efficiency and/or folding during biosynthesis rather than the intrinsic catalytic properties of the correctly folded protein. Similar observations have been reported indicating that the position of His-tag can significantly influence recombinant protein expression levels and overall production yields. For several recombinant His-tagged proteins, constructs bearing a C-terminal His-tag have been shown to result in higher expression levels and increased yields of purified protein. Other studies indicate that N-terminal tagging can be advantageous, as it may facilitate translation initiation and simplify purification procedures, leading to higher protein production levels [[29], [30], [31], [32]]. Therefore, the impact of His-tag orientation appears to be dependent on the target protein, implying that the optimal tag position is protein-specific rather than universally applicable.

For diagnostic applications, the higher productivity of BCO–H translates directly into increased enzyme availability for reagent formulation at a lower production cost. Therefore, BCO–H emerges as the preferred variant for large-scale manufacturing, while H-BCO remains a valid alternative in cases where different immobilization strategies or fusion partners might benefit from an N-terminal tag. Notably, both variants are fully functional, and their parallel evaluation remains important for future developments involving advanced functionalization or directed immobilization strategies, where tag orientation may differentially impact performance.

In conclusion, we developed an integrated approach for the recombinant production of type II cholesterol oxidase from B. sterolicum in E. coli and its direct implementation in a commercially available diagnostic kit for cholesterol determination. Our results demonstrate that a fed-batch process in a chemically defined medium enables high-level production of two His-tagged BCO variants, while preserving enzymatic performance comparable to reference commercial enzymes. Moreover, incorporation of these in-house enzymes into the Colestat diagnostic kit yielded analytical performance fully compliant with current EFLM quality specifications for cholesterol determination in human serum and plasma.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study included organ or tissue donors.This study includes human biological material and consent was obtained by donors, or their next of kin or legal representatives, for use in this study and for publication of the article. The samples used in this research were not sourced from executed prisoners or prisoners of conscience.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. Ethics committee approval was not required under relevant laws and institutional guidelines. The study exclusively used previously collected human serum and plasma samples that had been obtained with informed consent authorizing their use for basic in vitro research and publication of the resulting scientific findings. The samples were stored and provided in a fully anonymized and non-linkable manner, such that the identity of the individuals from whom the samples were obtained could not be determined or recovered by the investigators and therefore did not represent a risk to the individuals.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used ChatGPT solely to assist in enhancing its clarity and readability. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

CRediT authorship contribution statement

Agustina Godino: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Funding acquisition, Conceptualization. Marilla Amaranto: Writing – review & editing, Methodology, Investigation. Mauricio Rassetto: Methodology, Investigation. María Soledad Vila: Methodology. Gabriela Altamirano: Methodology. Magalí Soria: Methodology. Silvina Rosa Salinas: Methodology, Visualization, Writing – review & editing. José Luis Barra: Writing – review & editing, Supervision, Resources, Investigation, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.