Journal of Chromatography B

Visceral leishmaniasis, a disease caused by Leishmania infantum chagasi , represents a major public health problem in many areas of the world. However, there is currently no vaccine for human use. The aim of this work was to purify the 503 antigen of Leishmania i. chagasi directly from unclariﬁed Escherichia coli feedstock through expanded bed adsorption (EBA) chromatography. Batch experiments were performed to optimize the adsorption and elution conditions of the antigen onto a STREAMLINE TM Chelating resin using two central composite rotatable designs (CCRD). The results showed that the optimal binding conditions of the 503 antigen were pH 8.0 in the presence of 2.4 M NaCl. For the elution of the target protein, the optimized conditions included the presence of 600.0 mM imidazole. The adsorption isothermal data of the 503 antigen were ﬁtted to the Langmuir adsorption isotherm. The EBA experiment successfully recovered 59.2% of the 503 antigen from the unclariﬁed E. coli homogenate with a puriﬁcation factor of 6.0.


Introduction
Visceral leishmaniasis (VL) is a potentially fatal disease caused in the Americas by infection with the protozoa Leishmania infantum chagasi.The disease is transmitted to humans via the bite of female sandflies of the genus Lutzomyia [1,2].Infection leads to a wide spectrum of clinical outcomes, ranging from asymptomatic infection to active disease characterized by fevers, cachexia, hepatosplenomegaly and immunosuppression.Without treatment, most symptomatic patients die [3].Antileishmanial drug treatment is problematic because leishmanicidal drugs are toxic and costly and drug resistance is emerging, particularly in developing countries.Furthermore, there is no effective and safe vaccine against VL approved for human use [4,5].
The protozoa of the genus Leishmania show two distinct morphological forms, namely promastigote and amastigote, in insect and mammalian hosts, respectively [6].Consequently, distinct host immune responses occur to antigens expressed during the intracellular amastigote stage, which are difficult to produce in bulk [7].The 503 antigen, a protein with 100% identity to the elongation factor 1␥ (EF-1␥) of L. infantum, was identified through screening a cDNA library from the intracellular amastigote form of Leishmania i. chagasi [8].This antigen constitutes a potential target for the immune response during mammalian infection, suggesting that the 503 antigen can be utilized in the development of diagnostic agents for serological tests or of vaccines against VL [8].
In contrast, the production of proteins by genetically engineered microorganisms, yeasts and animal cells has become a very important technique for the preparation of bioproducts, but these need to be highly pure for most applications [9].The initial protein recovery is usually performed from solutions containing cells or cell debris and cannot be applied to packed-bed chromatography, which leads to a process with multiple steps.The packed bed acts as a depth filter, causing the accumulation of cells and cell debris and consequent clogging of the column [10].Centrifugation and filtration are the two most widely used techniques for the removal of biomass or cell debris.However, these clarification methods possess several disadvantages, such as being time consuming, resulting in substantial protein loss, and having high operational and capital costs [11,12].
To minimize such limitations, expanded bed adsorption (EBA) has emerged as an innovative and special chromatography technique for the recovery, separation and purification of target biomolecules directly from complex unclarified feedstock without elimination of suspended solids [13,14].EBA combines the processes of clarification, concentration and initial purification into a one-step operation and thus provides increased process economy, increased yield, shorter overall processing times, reduced work cost, and reduced running costs and capital expenditure [15][16][17].EBA has been used successfully with a number of different feedstocks, such as bacterial [18,19], yeast [12,20], and mammalian cell fermentation broths [21,22], as well as others [16,23,24], for the recovery of important biomolecules.
In this context, the main goal of this study was to purify the 503 antigen of Leishmania i. chagasi directly from unclarified feedstock of Escherichia coli M15 using EBA chromatography.

Materials
The STREAMLINE TM Chelating resin was purchased from GE Healthcare (Uppsala, Sweden).It is composed of highly cross-linked 6% agarose, which entraps an inert quartz core to yield a mean particle density of 1.2 g/mL and a distribution of size of 100-300 m (data from the manufacturer's manual).A glass custom-made column (30.0 cm × 2.6 cm I.D.) was fitted with an adjustable piston to minimize the headspace over the fluidized bed.A ruler was placed on the column wall to record the bed height.The feedstock was loaded into the column by a peristaltic pump (model Perimax 12, Spetec).

E. coli cultivation for 503 antigen production
E. coli M15 strain expressing the His-tagged 503 antigen was kindly provided by Dr. Mary Wilson from the University of Iowa (Iowa, USA) [8].The clone was cultured in 2xTY medium (16 g/L tryptone, 10 g/L yeast extract, and 5 g/L NaCl, pH 7.0) supplemented with ampicillin (0.1 g/L) and kanamycin (0.025 g/L) at 37 • C and 400 rpm on a bioreactor (Biostat B., B. Braun Biotech International) with a working volume of 1.5 L. The expression of the recombinant protein was induced by the addition of lactose to the cultivation medium at a final concentration of 10 g/L when the optical density at 600 nm reached approximately 0.5 [25].Polypropylene glycol was used as an antifoam agent when necessary.
After cultivation, the cells were pelleted by centrifugation at 1500 × g and 4 • C for 30 min and subsequently re-suspended in lysis buffer (8.0 M Urea, 20.0 mM NaH 2 PO 4 , 0.5 M NaCl, and 10.0 mM imidazole, pH 8) to a biomass concentration of 5% or 10% (w/v).Cell disruption was performed as previously described by Vaz et al. [25].

Adsorption conditions
The influence of the independent variables pH and NaCl concentration on the adsorption of the 503 antigen was evaluated using a 2 2 central composite rotatable design (CCRD), which included four trials under the axial conditions and three repetitions at the central point for a total of 11 experiments [26,27].The real and coded values of the independent variables are shown in Table 1.
The adsorbent (chelated using 0.1 M Ni 2+ with a 5-mL total charge volume) was equilibrated with the binding buffer (20.0 mM NaH 2 PO 4 , 10.0 mM imidazole, 0.18-2.31M NaCl, pH 4.2-9.8)and then mixed with 5 mL of unclarified feedstock at 10% (w/v).The mixture was incubated at 25 • C in an orbital shaker for 2 h.The supernatant was collected from the settled adsorbent.The amount of unbound 503 antigen in the supernatant was quantified.The data were treated using the Statistica version 7.0 software (Statsoft, OK, USA).

Elution conditions
The influence of the factors pH, imidazole concentration and NaCl concentration on the elution of the 503 antigen was evaluated using a 2 3 CCRD that included six trials under the axial conditions and three repetitions at the central point for a total of 17 experiments [26,27].The real and coded values of the independent variables are shown in Table 2.
The optimization of the elution conditions of the bound 503 antigen was performed in the batch mode.A 0.1 M Ni 2+ -charged adsorbent was incubated with unclarified feedstock of the 503 antigen at 25 • C in an orbital shaker for 2 h.The supernatant was withdrawn, and the adsorbent was washed thoroughly with the binding buffer (20.0 mM NaH 2 PO 4 , 0.5 M NaCl, and 10.0 mM imidazole, pH 8).The adsorbed protein was eluted with sodium phosphate (20.0 mM) buffer at pH ranging from 5.3 to 8.7, which contained 0.7-2.3M NaCl and 31.0-619.0mM imidazole.The data were treated using the Statistica version 7.0 software.

Adsorption isotherm
The unclarified feedstock containing different concentrations of the 503 antigen was prepared in the binding buffer (20.0 mM NaH 2 PO 4 , 2.4 M NaCl, and 10.0 mM imidazole, pH 8).The unclarified feedstock was mixed with 0.25 g of pre-equilibrated Ni 2+ -charged adsorbent and incubated in an orbital shaker at 25 • C for 2 h to ensure equilibrium between the solid and liquid phases [28].The supernatant was then collected, and the amount of unbound antigen was quantified by a mass balance.The adsorption isotherm was fitted using the Origin Pro 8.0 software (OriginLab Corporation, MA, USA).

EBA chromatography operation
The 0.1 M Ni 2+ -treated adsorbent was manually loaded into the column.The loading of feedstock and the collection of fractions were performed in the upward mode at room temperature (25 ± 2 • C).

Bed expansion characteristics
A Ni 2+ -charged adsorbent was loaded into the column to a settled-bed height of approximately 5 cm.Binding buffer containing 0% and 5% (w/v) of unclarified feedstock was pumped into the column with an increasing linear flow velocity (0-250 cm/h), and the corresponding bed expansion was recorded [28].Bed expansion was allowed to stabilize for 10 min before a reading was taken.The degree of bed expansion was determined as the ratio between the expanded bed height (H) and the settled bed height (H 0 ).
The bed expansion was also characterized by the Richardson-Zaki coefficient (n) according to Eq. ( 1):  where U represents the superficial velocity, and ε and U t are the bed porosity and terminal velocity of settling resin, respectively.The bed porosity, ε, was calculated using Eq. ( 2): The settled bed porosity, ε 0 , was assumed to have a value of 0.4 [16,29].

Direct purification of the 503 antigen from unclarified feedstock
Approximately 27 mL of Ni 2+ -charged STREAMLINE TM Chelating adsorbent was loaded into the EBA column to a bed height of 5 cm.The bottom inlet was packed with glass beads (4 cm in height) to distribute the inlet flow.To avoid leakage of nickel ions throughout the purification of the 503 antigen, the adsorbent was first washed with 100 mL of elution buffer (20.0 mM NaH 2 PO 4 , 600.0 mM imidazole, and 1.0 M NaCl, pH 7) at 150 cm/h.The adsorbent was then equilibrated with binding buffer (20.0 mM NaH 2 PO 4 , 2.4 M NaCl, and 10.0 mM imidazole, pH 8).The adsorbent bed was allowed to expand and stabilize to a degree of 2. The unclarified feedstock (100 mL) was then loaded into the column at a linear flow velocity of 150 cm/h and then washed with 200 mL of washing buffer (20.0 mM NaH 2 PO 4 , 10.0 mM imidazole, 2.4 M NaCl, and 10% glycerol, pH 8).The elution of the 503 antigen was also performed in the upward mode with 100 mL of eluting buffer at 50 cm/h.Protein samples were collected in fractions (10 mL per fraction) and analyzed to obtain the amounts of total protein and 503 antigen.

Analytical methods
Fractions were collected and the concentration of total protein was determined according to the Lowry method, using bovine serum albumin (BSA) as standard [30].The 503 antigen was analyzed on a 12% sodium-dodecyl sulfate polyacrylamide gel (SDS-PAGE) under denaturing conditions, according to the protocol by Laemmli [31].The gels were stained with silver, and photographed to estimate the 503 antigen concentration by densitometry using the ImageJ software [32].

Calculations
The specific 503 antigen concentration, yield (%) and purification factor (PF) were defined according to Eqs. ( 3)-( 5), respectively.Specific 503 antigen concentration = Amount of recovered 503 antigen Amount of total protein Yield (%) = Amount of recovered 503 antigen Amount of 503 antigen in the feedstock × 100 Purification factor = Specific 503 antigen concentration Specific 503 antigen concentration in the feedstock (5)

Optimization of the binding and elution conditions in batch mode
The effects of pH (X 1 ), NaCl concentration (X 2 ) and their interactions on the binding capacity of the 503 antigen in the STREAMLINE TM Chelating resin were evaluated using a 2 2 central composite design.According to the experimental conditions used, the amount of 503 antigen adsorbed ranged from 0.127 to 0.268 mg/mL of adsorbent, as shown in Table 1.As observed, the highest value (0.268 mg of 503 antigen/mL of adsorbent) was obtained in Run 8.
Based on the compound central design, a second-order model (Eq.( 6)) describing the 503 antigen adsorption as a function of the pH and NaCl concentration was established.The pure error, which was calculated from the central points, was very low, i.e., approximately 0.03%, according to the total sum of squares, indicating good reproducibility of the experimental data.Based on the F-test results, the model was found to be significant because the calculated F-value (6.63) was 2.16-fold higher than the listed F (3.07).Thus, the coded model expressed by Eq. ( 6) was used to generate the response surfaces for 503 antigen adsorption (Fig. 1).According to this model, both terms (linear and quadratic) for the variable pH (p < 0.05) and the linear term for the variable NaCl concentration (p < 0.05) are significant.

antigen binding capacity
The response surface (Fig. 1) shows that there is a specific region (pH 7.0-9.0)that presents maximum values for the binding capacity of the 503 antigen.The pH plays an essential role in the binding of a target protein onto the adsorbent because it can easily influence the binding strength between functional groups on the resin and target protein [28].Similarly to our study, Tan et al. [11] reported that the optimal pH for the adsorption of a protein associated with the Newcastle disease virus (NDV) onto STREAMLINE TM Chelating adsorbent was pH 8.0.This may be due to increased binding of the recombinant protein to metal ions at a pH higher than the pKa of histidine residues (pKa 6-7).In addition, the concentration of hydrogen ions (H + ) increases under acidic pH conditions, inducing competitive adsorption between H + and Ni 2+ , which results in poor binding of the 503 antigen onto immobilized Ni 2+ at lower pH conditions [28].
The influence of the ionic strength on the adsorption was evaluated by adding NaCl to the adsorption buffer.The response surface shows that the predicted maximum binding capacity was reached when 2.4 M NaCl was included in the buffer.Other studies have reported that cells and cell debris carrying a negative charge interact with the cationic matrix [33,34].Because immobilized nickel ions have a positive charge, these ions may interact with cells and debris with a negative charge.However, the presence of 2.4 M NaCl in the adsorption buffer should have reduced the electrostatic interaction between immobilized metal ions and cells/debris carrying a negative charge.In contrast, the recombinant 503 antigen has six histidine residues (His 6 tag), which could enable strong interactions between the protein and the matrix.As a result, in this study, the sodium phosphate binding buffer was adjusted to pH 8 and 2.4 M NaCl was added prior to any experimental applications.
The effects of pH (X 1 ), NaCl concentration (X 2 ), imidazole concentration (X 3 ) and their interactions on the elution of the 503 antigen in STREAMLINE TM Chelating resin were evaluated using a 2 3 central composite design.According to the experimental conditions used, the amount of 503 antigen eluted ranged from 0.024 to 0.201 mg/mL of adsorbent, as shown in Table 2.The highest value (0.201 mg of 503 antigen/mL of adsorbent) was obtained in Run 14.
Based on the central composite design, a second-order model (Eq.( 7)) describing the 503 antigen elution as a function of pH, NaCl concentration and imidazole concentration was established.The pure error calculated from the central points was very low, i.e., approximately 0.008%, according to the total sum of squares, indicating good reproducibility of the experimental data.Based on the F-test results, the model is significant because the calculated F-value (14.57) was 4.27-fold higher than the listed F (3.41).According to this model, the linear and quadratic terms for imidazole are significant (p < 0.05).The amount of the eluted 503 antigen increased with an increase in the imidazole concentration, and as a result, the imidazole concentration in the elution buffer (sodium phosphate) was adjusted to 600.0 mM prior to any experimental applications in this study.Imidazole is commonly used as a competitive agent to displace proteins from metal ions because it is inexpensive and barely affects the biological properties of proteins [28,35].The statistical analysis showed that the pH and NaCl concentration did not affect the elution.Thus, a neutral pH (7.0) and a lower amount of NaCl (1.0 M) were used in the next steps in this study.

Equilibrium adsorption isotherm
The adsorption isotherm on the STREAMLINE TM Chelating adsorbent was determined by plotting the 503 antigen adsorbed per unit mass of adsorbent as a function of the equilibrium concentration of the 503 antigen in the binding buffer.The adsorption isotherm for proteins onto an adsorbent is typically modeled by the Langmuir isotherm equation, according to Eq. ( 8): where q is the amount of target protein adsorbed onto the resin, c represents the equilibrium concentration of the unbound target protein in the suspension, q max represents the maximum binding capacity, and k d is the dissociation constant, which can be derived from the linear regression of the Langmuir plot.As observed in Fig. 2, the data were fitted relatively well to the Langmuir isotherm (R 2 = 0.9807).The maximum adsorption capacity (q max ) of the 503 antigen to the STREAMLINE TM chelating resin was 1.95 mg of protein per g of adsorbent, and the dissociation constant (k d ) was 0.34 mg/mL.

Bed expansion characteristics
The degree of bed expansion is influenced by a number of variables, such as the superficial velocity, viscosity and density of the feedstock and certain physical properties of the adsorbent beads [10,36].The mode of bed expansion significantly affects the protein adsorption and yield.The bed expansion was evaluated using the binding buffer containing 0 and 5% (w/v) unclarified feedstock.
The results showed that the bed expansion increased linearly with an increase in the superficial velocity when both feedstocks were used, as observed in Fig. 3.The values of n derived from the linear regression of the Richardson-Zaki equation (Eq.( 1)) were 4.1 for the binding buffer and 4.4 for 5% unclarified feedstock, which is  close to the theoretical value of 4.8 that is normally used in the laminar flow regime [37].Thus, the bed was assumed to be stable upon expansion in the presence of biomass.As result, 5% (w/v) unclarified feedstock was applied at a linear flow velocity of 150 cm/h in the EBA experiment, in which the bed expansion degree was maintained at 2. According to Rosti et al. [38], bed expansion degrees of 2-3 give optimum conditions for protein binding by providing sufficient voidage for the passage of particles suspended in the feedstock because, at these particular conditions, the liquid flow through the bed and the contact efficiency is similar to that obtained with the packed bed.

Direct purification of the 503 antigen from unclarified feedstock
After the optimized conditions for binding and elution were determined, the optimal conditions were used to develop a purification strategy for the 503 antigen from unclarified E. coli feedstock.The Ni 2+ -STREAMLINE TM Chelating adsorbent was loaded into an EBA column and pre-equilibrated with binding buffer (pH 8) at a linear flow velocity of 150 cm/h prior to loading of the crude feedstock.
The resulting chromatogram and mass balance for the purification protocol are shown in Fig. 4 and Table 3, respectively.The loading of unclarified feedstock was performed after the expanded bed became stable at an expansion degree of 2. A linear flow velocity of 150 cm/h was applied throughout the purification process.The amounts of the 503 antigen in the flow-through increased to 0.34 mg/mL, and just 13.57mg of the antigen applied was lost.During the wash step, more than 10.53 mg of the target protein was lost in the column effluent, whereas the total protein lost was 134.60 mg.Glycerol (10%) was added to the washing buffer to remove residual particulates and debris entrapped in the fluidized adsorbent.Elution with buffer containing 600.0 mM imidazole resulted in a sharp peak, indicating that imidazole at this concentration is effective for the elution of the 503 antigen from the adsorbent with a recovery and purification factor of 59.2% and 6.0, respectively (Table 3).However, the mass balance, which is presented in Table 3, shows that the 503 antigen concentration and the total protein obtained were higher than the initial concentration, suggesting that the imidazole concentration used during elution likely influenced the protein quantification by the Lowry method.In fact, Molina et al. [39] observed that imidazole buffers can increase the bovine serum albumin (BSA) concentration when  the Lowry method is used.We highlight that the values determined for blank assays (imidazole interference) were subtracted from the total value determined for proteins.Fig. 5 shows the protein profile obtained using the EBA protocol and analyzed on a 12% polyacrylamide gel, which demonstrates that the eluted 503 antigen corresponds to a molecular weight of 56 kDa.The combination of expanded bed adsorption and immobilized metal affinity chromatography (IMAC) has been proven to be simple and efficient for purification processes, with recovery and purification factors ranging from 40% to 93.5% and 2.7 to 7.94, respectively [10,28,29,40,41].In this study, a recovery of 59% was obtained.It should be highlighted that a custom-made column with a bead glass bed as a distributor was used in this study.It is likely that the back-mixing produced and the preferential channel presented by this system once the adsorption was performed in the expanded mode reduced the antigen binding to the resin, resulting in a decrease in the recovery of the 503 antigen.In contrast, the purification factor was quite similar to those reported in the literature because the elution was performed with the resin bed in the packed mode and back-mixing effect was reduced.
Ho et al. [42] reported that the use of an expanded bed allows the combination of many of the conventional primary recovery steps, such as clarification, concentration, and initial purification, into a single process.Furthermore, a typical cycle of purification by EBA chromatography including equilibration (30 min), loading (30 min), washing (60 min) and elution (30 min) only took 2.5-3.5 h, making the process attractive for large-scale use [43].
Immobilized metal affinity chromatography (IMAC) is most widely used for the purification of histidine-tagged proteins and presents excellent selectivity, a large capacity and good robustness.However, the additional time required for charging the column with Ni 2+ ions and regenerating it with EDTA can have a negative effect on the productivity of IMAC protocols compared with simpler ion-exchange methods [35,44].
Based on the results presented in Table 3, the 503 antigen was eluted with a concentration of 0.212 g/L (21.16 mg for elution of 100 mL).Based on this result, in intermediate-scale production (milligram to gram level), the use of 5.0 L of unclarified feedstock of E. coli may produce approximately 1.0 g of the 503 antigen.
In conclusion, the results obtained in this study demonstrated that EBA provides a good alternative for the purification of the Histagged 503 antigen.This process allows direct adsorption of the target protein onto the adsorbent from the unclarified bacterial homogenate with a purification factor of 6.0.Therefore, this study demonstrated an easy, low-cost and labor-effective method for the purification of the 503 antigen of Leishmania i. chagasi using EBA.

Fig. 1 .
Fig. 1.Three-dimensional response surface contour plot showing the simultaneous effects of pH (X1) and NaCl concentration (X2) on 503 antigen adsorption in the batch mode.

Fig. 2 .
Fig. 2. Equilibrium adsorption isotherm for the 503 antigen.A series of different concentrations of the 503 antigen was prepared using unclarified feedstock and allowed to adsorb onto the STREAMLINE TM Chelating resin until equilibrium was achieved.

Fig. 4 .
Fig. 4. Chromatogram of the purification profile of the 503 antigen directly from unclarified feedstock of E. coli using EBA.Total protein ( ) and 503 antigen ( ).

Table 1
Matrix of the 2 2 central composite rotatable design showing the real and coded values (in parenthesis) for the response adsorption of the 503 antigen in batch mode.

Table 2
Matrix of the 2 3 central composite rotatable design showing the real and coded values (in parenthesis) for the response elution of the 503 antigen in batch mode.

Table 3
Summary of the purification of the 503 antigen directly from unclarified feedstock of E. coli M15 using EBA.