RESEARCH ARTICLE – Pharmaceutical Biotechnology

Dose Levels in Particulate-Containing Formulations Impact Anti-drug Antibody Responses to Murine Monoclonal Antibody in Mice MALIHEH SHOMALI,1 SULTAN TANRIVERDI,2 ANGELIKA J. FREITAG,3 JULIA ENGERT,3 GERHARD WINTER,3 MICHAEL SIEDLER,2 ZEHRA KAYMAKCALAN,2 JOHN F. CARPENTER,4 THEODORE W. RANDOLPH1 1 Department of Chemical and Biological Engineering, Center for Pharmaceutical Biotechnology, University of Colorado, Boulder, Colorado 80303 2 AbbVie Bioresearch Center, Worcester, Massachusetts 01605 3 Department of Pharmacy, Pharmaceutical Technology and Biopharmaceutics, Ludwig-Maximilians-University Munich, Munich D-81377, Germany 4 Department of Pharmaceutical Sciences, Center for Pharmaceutical Biotechnology, University of Colorado Anschutz Medical Campus, Aurora, Colorado 80045

Received 8 August 2014; revised 29 January 2015; accepted 30 January 2015 Published online 3 March 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.24413 ABSTRACT: Dosage levels and particulate contents of therapeutic protein formulations are potential factors that impact immunogenicity of protein therapeutics. Here, we evaluated the effect of dose levels on the immunogenicity of protein particulates formed by adsorbing a  murine monoclonal IgG2c/␬ antibody (mAb1) onto silicone oil microdroplets, glass, or aluminum hydroxide (Alhydrogel ) microparticles. Immune responses to these particulate-containing preparations were compared against responses to solutions of mAb1 that had been ultracentrifuged to minimize particle levels. Formulations containing 5 or 500 ␮g of adsorbed mAb1 were administered subcutaneously to C57BL/6J or BALB/c mice. Antidrug antibodies (ADAs) were detected using an isotype-specific enzyme-linked immunosorbent assay (ELISA) method or a chemiluminescence method. Sera from BALB/c mice showed greater ADA responses to administration of particles at the 5-␮g dose level than at the 500-␮g dose level. In sera from C57BL/6J mice, ADA levels detected by ELISA were independent of the particle dose levels tested. ADAs were not detected in sera from C57BL/6J mice performing the chemiluminescence technique. In conclusion, mice  administered formulations of a murine antibody adsorbed onto silicone oil microdroplets, glass microparticles, or Alhydrogel showed greater ADA responses that those that received particle-free mAb1 preparations, and responses were greater for formulations containing C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci 104:1610–1621, 2015 lower doses of antibody.  Keywords: Protein formulation; immunology; adsorption; aggregates; vaccine adjuvants; anti-drug antibodies; tolerance R

R

INTRODUCTION Antigen dose has been shown to impact the resulting immune response. It has been well known for decades that tolerance to an antigen may develop in response to low doses (e.g., low-zone tolerance) or relatively high doses (e.g., high-zone tolerance). For example, in early studies, Dresser1 observed that higher doses of bovine gamma-globulin (BGG) induced tolerance in CBA mice. However, the minimum dose that is required for tolerance appears to vary from one antigen to another. For instance, doses as low as 50–200 :g of aggregate-free BGG induced high-zone tolerance in CBA mice.1 In another study, ABC mice injected subcutaneously with bovine serum albumin (BSA) developed tolerance when the dose level was either very low or very high.2 Tolerance was obtained by injection of doses of about 10 :g BSA (low-dose tolerance) or by injection of about Abbreviations used: mAb, monoclonal antibody; ADA, antidrug antibody; ELISA, enzyme-linked immunosorbent assay; BGG, bovine gamma-globulin; PBS, phosphate-buffered saline. Correspondence to: Theodore W. Randolph (Telephone: +303-492-4776; Fax: +303-492-4341; E-mail: [email protected]) This article contains supplementary material available from the authors upon request or via the Internet at http://onlinelibrary.wiley.com/. Journal of Pharmaceutical Sciences, Vol. 104, 1610–1621 (2015)  C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association

1610

10–100 mg BSA (high-dose tolerance). However, an intermediate dose (1 mg) induced an immune response. More recently, Braun et al.3 showed that the immunogenicity of aggregated human INF-" in wild-type or transgenic mice increased directly with the dosing level. Conversely, in a clinical research study of infliximab and adalimumab in Crohn’s disease patients, lower doses were reported to induce higher antidrug antibody (ADA) responses than higher doses.4 Hermeling et al.5 were also observed a clear interferon-alpha aggregate dose–ADA response effect in the low-microgram dose range in transgenic immune tolerant mice. Unwanted immune responses to therapeutic proteins have been reported for decades.6,7 Protein aggregates have been shown to play a critical role in such adverse immunogenicity. For example, it was shown that aggregated human gamma globulin (HGG) had a greater potential to elicit immune responses in rats, dogs, or mice than its monomeric form.8–11 In the these studies, ultracentrifugation of protein samples, which removes protein aggregates and particulates, was reported as an effective means for reducing the level of immunogenicity. The potential of aggregates to contribute to immunogenicity was further demonstrated in recent studies by administering heated or agitated INF-", INF-$, or rhGH into wild-type or transgenic mice.12–15 In another recent study, it was reported that aggregates caused by agitation and heating of human IgG1

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

RESEARCH ARTICLE – Pharmaceutical Biotechnology

and IgG2 increased the innate and late-stage T-cell immune responses in cultured peripheral blood monocytes.16 Also, a dosedependent presentation of antigenic peptides of the surface of antigen presenting cells was observed when the cells were treated with increasing doses of subvisible particles composed of a therapeutic protein.17 Most recently, Ahmadi et al.18 used in vitro experiments with human CD4+ T cells and dendritic cells to show that minute amounts of subvisible aggregates in formulations of the monoclonal antibody trastuzumab resulted in potent CD4+ T cell proliferation and strong cytokine and costimulatory signals from the dendritic cells. In addition to particles assembled directly from protein molecules, protein-containing particulates can form because of protein adsorption to foreign materials from container closure, delivery devices, or manufacturing processes.19–21 Such microparticulate contaminants include glass, stainless steel, tungsten, and silicone oil.19–22 The current study focuses on glass microparticles, which may be shed from glass vials or syringes, and microdroplets of silicone oil, a common lubricant in prefilled syringes. The immune response to an administered antigen also depends on recipient genetic factors. In the case of murine models, the strain of mouse used in a study may impact the resulting immune response.12 For example, BALB/c mice were less easily tolerized to ultracentrifuged human or BGG than were C57BL/6J mice.23–25 Injection of even very small amounts of ultracentrifuged HGG (as low as 50 :g) induced tolerance in C57BL/6J mice, whereas tolerizing doses in BALB/c mice were as high as 10 mg.25 It was suggested that BALB/c mice may process trace amounts of aggregates remaining in the samples more efficiently than C57BL/6J mice.25 Interestingly, when trace amounts of aggregates were removed by salt fractionation, mice from both strains became unresponsive to small doses of HGG. Similar results were observed when single doses of ultracentrifuged BGG were administered to BALB/c and DBA/2 mice.24 In that study, DBA/2 mice became tolerant at doses of 0.2 mg BGG, whereas BALB/c mice required more than 20 mg to induce tolerance. In the current study, we used a murine monoclonal antibody (mAb1) as a model therapeutic protein and administered it subcutaneously to either C57BL/6J or BALB/c mice. mAb1 is a murine monoclonal antibody of the IgG2c isotype that binds mouse tumor necrosis factor (TNF). mAb1 is syngeneic to C57BL/6J mice, because it was generated in this strain. In contrast, BALB/c mice do not produce immunoglobulins of the IgG2c isotype,26,27 which makes mAb1 allogeneic to this strain. Therefore, we expected a stronger immune response to mAb1 in BALB/c than in C57BL/6J mice. In a previous study, we showed that 50 :g doses of mAb1 were more immunogenic in BALB/c and C57BL/6J mice when the antibody was adsorbed onto microparticles of glass or   Alhydrogel .28 Alhydrogel is a microparticulate adjuvant that is commonly used in vaccine formulations. It is known to be effective in provoking humoral immunity by inducing an antibody (Th2) response in the form of IgG1.29–31 Proteins can  be adsorbed to Alhydrogel , producing particulates roughly 29 3–4.5 :m in size. Here, we explored the effect of mAb1 dose levels on immune responses in BALB/c and C57BL/6J mice to mAb1 adsorbed to  glass microparticles, Alhydrogel , or silicone oil microdroplets. We hypothesized that, because of tolerization effects, a higher dose (500 :g) of mAb1 would provoke lower immune responses R

R

1611

 R

than a lower dose (5 :g). Although Alhydrogel is not found in commercial formulations of therapeutic antibodies, we included it in some formulations of mAb1 in order to allow comparison of the effects of glass microparticles or silicone oil microdroplets with those of a particle type known to stimulate immune responses in both humans and in mice of these strains. The immune response was monitored by measuring titers of various anti-mAb1 antibody isotypes produced in mice after subcutaneous injection of various formulations. Anti-mAb1 antibodies were analyzed by isotype, as described previously.28,32 Isotype analysis may provide more sensitive ADA detection,32 especially for cases where the drug itself is an antibody. Furthermore, the isotype of an ADA response may be reflective of the balance between Th1 and Th2 type responses. Titers of antimAb1 antibodies were measured with two different methods, an enzyme-linked immunosorbent assay (ELISA) and the Meso  Scale Discovery (MSD)-chemiluminescence assay. Solid-phase ELISA-based assays have long been used and are the most common ADA detection assays. However, recently developed liquidphase assay techniques such as MSD-chemiluminescence technology may (in some cases) offer a higher sensitivity to detect ADA. In addition during the course of the study, the levels of mAb1 in the murine sera resulting from the various injections were measured to determine the pharmacokinetic (PK) profiles. R

MATERIALS AND METHODS Materials All chemicals used in this work were of reagent grade or higher quality. Sterile water for injection was used, and all materials used for injection were USP grade. mAb1, a mouse monoclonal antibody (IgG2c/6, 145 kD) against mouse TNF-", and mouse TNF-" were provided by AbbVie Bioresearch Center (Worcester, Massachusetts). HRP-goat antimouse IgM, IgG1, IgG2b, and IgG2c were purchased from Jackson ImmunoResearch Laboratories Inc. (West Grove, Pennsylvania). HRPrabbit antimouse IgG3 was purchased from Fitzgerald Industries International Inc. (Acton, Massachusetts). Silicone oil, Dow Corning 360 1000 cSt Medical grade, was purchased from  Dow Corning Corporation (Midland, Michigan), Alhydrogel was purchased from Brenntag Biosector (Frederikssund, Denmark), L-histidine was from RPI (Prospect, Illinois), sucrose was from Sigma–Aldrich (St. Louis, Missouri), citric acid, trisodium salt dehydrate were from ACROS ORGANICS (Fair Lawn, New Jersey), 3,3 ,5,5 -Tetramethylbenzidine (TMB) was from Thermo Scientific (Rockford, Illinois), and sulfuric acid was from Mallinckrodt (Hazelwood, Missouri). Other chemical reagents were purchased from Fisher Scientific (Pittsburgh, Pennsylvania), including sodium chloride, polysorbate 20, and phosphate-buffered saline (PBS; 10x solution, DNase-RNaseand protease-free, 1.37 M sodium chloride, 0.027 M potassium chloride, and 0.119 M phosphate buffer). R

R

R

DOI 10.1002/jps.24413

mAb1 Stock Solution A stock solution of mAb1 was received at a concentration of 24 mg/mL in a 15-mM histidine buffer, pH 6.0. Size-exclusion chromatography analysis showed that the stock solution contained 97.3% mAb1 monomer, 2.1% low-molecular-weight species, and 0.6% high-molecular-weight species. The stock Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

1612

RESEARCH ARTICLE – Pharmaceutical Biotechnology

solution was aliquotted into 1-mL tubes under aseptic conditions and stored in a freezer at −80°C until further use. Preparation of Ground Glass Microparticles Glass microparticles were prepared from vials (5cc, Type 1 glass, USP/PhEur, nontreated; Schott Inc., Syracuse, New York) as described previously.20 Preparation of mAb1 Adsorbed on Glass Microparticles The diameter-based size distribution measured by laser diffraction ranged from 1 to 2 :m for glass particles. The .potential of glass particles in 20 mM histidine buffer and adsorbed mAb1  on glass particle media was measured with a Zetasizer NanoZS (Malvern Instruments Ltd., Malvern, UK) and reported to be −41.70 ± 1.76 and +11.77 ± 0.62 mV, respectively. In a previous in vitro study of the reversibility of the adsorption of mAb1 to glass particles in PBS buffer, it was reported that at least 30% of the injected mAb1 would remain adsorbed on glass particles once administered in vivo.28 The secondary and tertiary structure of the adsorbed mAb1 were found to be minimally disturbed by infrared and fluorescence spectroscopy, respectively.28 The mass of glass microparticles required to adsorb mAb1 in monolayer coverage was calculated from the microparticle specific surface area (4.06 ± 0.03 m2 g−1 ) and the saturation value of the mAb1 adsorption isotherm (2.5 mg mAb1/m2 ; see Supplemental Fig. S1), as described previously.28 On the basis of this calculation, a “low-dose” formulation was prepared that contained 2.5 mg/mL glass microparticles, 25 :g/mL mAb1, and 8% (w/v) sucrose in 20 mM histidine, pH 5.7. At this mAb1–glass ratio, comparison with the previously measured adsorption isotherm suggests that essentially all of the protein adsorbs to the glass microparticles. Aliquots of the formulation, each with a volume of 4 mL, were prepared in 15-mL falcon tubes and rotated end-over-end at 8 rpm for 30 min at room temperature. “High-dose” formulations contained 2.5 mg/mL mAb1, but were otherwise identical. Thus, it was anticipated that in the high-dose formulations the added glass microparticles would provide sufficient surface area to adsorb only approximately 1% of the mAb1. Samples were freshly prepared on the day of each injection, and were tested for endotoxin levels using the LAL endotoxin assay according to the protocol in the kit (LONZA, Walkersville, Maryland). Formulations contained endotoxin levels33 lower than 0.5 EU/mL. R

Preparation of Silicone Oil Emulsions

measured in triplicate using a Coulter LS230 laser diffraction instrument (Beckman Coulter Inc., Miami, Florida). The total concentration of silicone oil in the emulsion was determined by infrared spectroscopy using a Bomem MB-series spectrometer (ABB, Quebec, Canada) after extraction into heptane, as described previously.22 Silicone oil has the characteristic absorbance at 1260 cm−1 ; therefore, the area under the peak between 1280 and 1240 cm−1 of the infrared absorbance curve was measured and compared with the area under the peak of a standard curve (Supplemental Fig. S2). The concentration value was used in combination with the measured droplet size distribution to calculate the specific silicone oil–water interfacial area. Previously, it was shown that microdroplet size distributions within these silicone oil emulsions are stable for a month22 ; in the present study, emulsions were used to generate formulations for injection (see below) within a week after the emulsion was prepared. Preparation of mAb1 Adsorbed on Silicone Oil Microdroplets On each of the days for injection, “low-dose” formulations of mAb1 adsorbed to suspensions of silicone oil microdroplets were prepared by mixing silicone oil emulsion and buffered mAb1 solutions together in appropriate ratios to yield aqueous suspensions containing 5 :g/mL silicone oil emulsified with 25 :g/mL mAb1, 8% (w/v) sucrose, and 20 mM histidine buffer, pH 5.7. The amount of silicone oil emulsion in the “low-dose” formulation was sufficient to adsorb essentially all of the mAb1 in the formulation, based on the emulsion’s measured specific surface area (200 m2 /mg silicone oil on the day of preparation) and the amount of mAb1 required to saturate the oil–water interface (ca 2.5 mg/m2 , see Supplemental Fig. S3). Size-exclusion chromatography analysis of samples after removal of silicone oil-bound mAb1 by centrifugation showed that the unbound mAb1 was monomeric. “High-dose” formulations were identical except for the concentration of mAb1, which was 2.5 mg/mL. Four milliliter volume of the “low-dose” and “high-dose” formulations was prepared in 15-mL falcon tubes and rotated endover-end at 8 rpm for 30 min at room temperature. Samples were tested for endotoxin levels33 using the LAL endotoxin assay according to the protocol in the kit (LONZA), and found to have endotoxin levels lower than 2.0 EU/mL. Preparation of Samples Containing Adjuvant On each of the days for injection, “low-dose” formulations were  prepared by adding 150 :g of Alhydrogel (15 :L of a 10-mg/mL stock solution) to 4 mL of a solution containing 25 :g/mL mAb1, 8% (w/v) sucrose, and 20 mM histidine, pH 5.7. “High-dose” formulations were identical except for the concentration of mAb1, which was 2.5 mg/mL mAb1. The solutions were prepared in 15-mL falcon tubes and rotated end-over-end at 8 rpm for 30 min at room temperature. Samples were freshly prepared on the day of each injection, and were tested for endotoxin levels using the LAL endotoxin assay according to the protocol in the kit (LONZA). Formulations contained endotoxin levels33 lower than 0.5 EU/mL. R

To prepare emulsions of silicone oil microdroplets, 5% (v/v) silicone oil were first mixed with water for injection in the stainless steel cylinder of a Virtishear mechanical homogenizer (VirTis Company, Warminster, Pennsylvania). The mixture was mixed at room temperature for 20 min at 5000 rpm. Immediately thereafter, the mixture was passed five times through a highpressure homogenizer (Emulsiflex C5 homogenizer; Avestin Inc., Ottawa, Ontario, Canada) at a pressure of 50 MPa. The final emulsion, which contained approximately 1% (v/v) silicone oil as microdroplets, was collected in a glass bottle and stored at room temperature until further use.22 The .potential of silicone oil droplets and adsorbed mAb1 on  silicone oil were measured with a Zetasizer Nano-ZS (Malvern Instruments Ltd., Malvern, UK), yielding . potential values of −65.97 ± 0.29 and +6.91 ± 0.85 mV, respectively. The size distribution of silicone oil microdroplets in the emulsion was R

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

Animal Protocol The protocol for the animal experiments was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Colorado at Boulder under ID #09-05-RAN-02. Female BALB/c or C57BL/6J mice at the age of 8 weeks were DOI 10.1002/jps.24413

RESEARCH ARTICLE – Pharmaceutical Biotechnology

purchased from Charles Rivers Laboratories (Wilmington, Massachusetts) and kept three mice per cage with available food and water ad libitum. Before the study began, mice were allowed to acclimate for a week. The initial weights of all mice were in the range of 17–22 g. Injection and Serum Collection Schedule On days 1, 6, 13, 20, 27, and 41, groups of six mice were first weighed, and then blood was collected from the retro-orbital venous sinus. Blood samples were collected in centrifuge tubes and kept on ice during the procedure. Later, blood samples were centrifuged for 10 min at 3000g, and the resulting cellfree serum samples were aliquoted and stored at −80°C until thawed and used for analysis. On days 4 and 18, groups of mice were injected subcutaneously at the back of the neck with 200 :L of placebo (20 mM histidine buffer plus 8%, w/v, sucrose, pH 5.7), or 200 :L samples containing either low (5 :g) or high (500 :g) doses of native mAb1, mAb1 adsorbed to glass microparticles, mAb1 adsorbed  to silicone oil microdroplets, or mAb1 adsorbed to Alhydrogel . A few mice died at the 1st week of immunization caused some groups include four to five mice. The mice in each group were injected with the same syringe and needle (BD 1 mL plastic syringe and BD 25 G1 needle). R

Determination of Anti-mAb1 Antibodies Level Using ELISA The levels of antibodies in the mouse sera with specificity to mAb1 were quantified by indirect ELISA assay as described previously.28

1613

Determination of mAb1 Levels in Serum by ELISA mAb1 levels in the mouse sera were measured by an indirect ELISA assay. The plate wells were coated with 100 :L/well of mouse TNF-" in PBS at 100 ng/mL and incubated overnight (20–22 h) at 4°C. After washing three times with washing solution (0.05% polysorbate 20 in PBS), the plate wells were blocked with 200 :L/well 1% BSA in PBS for 2 h at room temperature, followed by another three cycles of washing. Six microliter of serum was diluted in 75 :L of PBS. Twenty-five microliter of diluted sera was pipetted in the first row of wells on the plate, and the remaining 50 :L were pipetted into the second row of wells and mixed with 50 :L of dilution buffer (14.4 mM citric acid, tri-sodium salt dehydrate, 1.1 M sodium chloride, and 0.1% polysorbate20). The diluted serum from the second row was serially diluted (1–3) in dilution buffer down the plate wells. Samples were incubated in dilution buffer for another 2 h at room temperature. Detection was carried out by application of HRP-conjugated goat antimouse IgG2c antibody. The plate wells were washed five times with washing solution, and then detecting antibody diluted 1:15,000 (50 :L/well) was added and incubated for 1.5 h while shaking at 600 rpm. After washing the plate wells with washing solution for five cycles, 1-step Ultra TMB (50 :L/well) was added as a substrate. The reaction was quenched after 10–15 min with 0.5 M sulfuric acid. Absorbance at 450 nm was measured using a MAXline microplate reader (Molecular Devices Corporation, Sunnyvale, California). A calibration curve was constructed for the sera from each mouse strain. Serum samples from preimmunized mice were spiked with 10 :g/mL mAb1, the samples were added to plate wells that were coated with mouse TNF-", and blocked with 1% BSA in PBS. Sera were then diluted serially in dilution buffer, and the ELISA protocol was performed as described above.

Determination of Anti-mAb1 Antibodies Level Using MSD Chemiluminescence Assay

Determination of mAb1 Levels in Serum by MSD-Chemiluminescence

A 60-:L aliquot of biotin-labeled mAb1 (1 :g/mL; AbbVie Bioresearch Center), 60 :L of Sulfo TAG-labeled mAb1 (1 :g/mL; AbbVie Bioresearch Center), and 60 :L of murine serum samples at various dilutions were mixed and incubated at 4°C overnight. MSD streptavidin plates were blocked with 3% MSD Blocker A in TTBS (Tris-buffered saline with addition of 0.1% Tween 20) for 1 h at room temperature with gentle shaking. Then, the plate wells were washed four times with TTBS. Seventyfive microliter of the incubated sample, biotinylated mAb1, and Sulfo-TAG-labeled mAb1 mix were transferred into the plate wells in duplicate and allowed to bind to them for 2 h with gentle shaking at room temperature. Afterwards, the plate wells were washed four times with TTBS, 150 :L of 2× MSD Read Buffer T was added to the wells and the plates were read on an MSD reader. Assay standards consisting of varying concentrations of polyclonal goat antimouse anti-IgG, heavy- and light-chain specific (cat: 1031-01; Southern Biotech, Birmingham, AL), in assay buffer were added to each plate well to generate a standard curve. Concentrations of mouse anti-mAb1 in samples were determined by interpolating signal against the standard curve using IDBS XLfit with a four parameter logistic fitting model. To check the plate-to-plate variability, three positive controls, high (200 ng/mL), medium (20 ng/mL), and low (2 ng/mL) made up with known goat antimouse anti-IgG concentrations in assay buffer were added to wells on each plate.

A solution containing 60 :L biotin-labeled mouse TNF-" (0.5 :g/mL; AbbVie Bioresearch Corporation-ABC, Worcester, Massachusetts) and 60 :L of diluted murine serum samples were mixed and incubated at 4°C overnight. Wells of MSD streptavidin plates were blocked with 3% MSD Blocker A in TTBS for 1 h at room temperature with gentle shaking and then washed four times with TTBS. Fifty microliter of the incubated sample and biotinylated mouse TNF-" mix were transferred into the wells of the plates in duplicate and allowed to bind to them for 1 h with gentle shaking at room temperature. Afterward, the wells were washed four times with TTBS, and 50 :L (0.25 :g/mL) of detecting antibody (polyclonal goat antimouse anti-Ig labeled with Sulfo-TAG, cat# R32AC-5) were added to the wells and incubated for 1 h with gentle shaking at room temperature. The plate wells were again washed four times with TTBS; 150 :L of 2X MSD Read Buffer T (cat: R92TC-1) were added to the wells and the plates were read on an MSD reader. Assay standards consisting of varying concentrations of mAb1 (AbbVie Bioresearch Center) in assay buffer were added to each well to generate a standard curve. Concentrations of mAb1 in samples were determined by interpolating signal against the standard curve using IDBS XLfit with a four parameter logistic fitting model. To check the plate-to-plate variability, three positive controls, high (200 ng/mL), medium (20 ng/mL), and low (2 ng/mL) were made up with known mAb1

DOI 10.1002/jps.24413

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

1614

RESEARCH ARTICLE – Pharmaceutical Biotechnology

concentrations in the assay buffer and added to each plate well. Murine IgG2c (AbbVie Bioresearch Center) was used as an isotype control. Data Analysis For results from the ELISA method, serum ADA levels were analyzed and reported as antibody titers of different immunoglobulin isotypes (IgM, IgG1, IgG2b, and IgG3), as described previously.28 The PK data determined by ELISA assays were measured for each serum, and the mean value of absorbance (at 450 nm) for each group was converted to mAb1 concentration in :g/mL mAb1 based on a previously determined calibration curve (Supplemental Fig. S4). The results of ADA or PK analysis as determined by the MSD-chemiluminescence method were reported as the mean concentrations of ADA (:g/mL) or mAb1 level (:g/mL) for each group. Statistical Analysis All statistical analyses were performed using SigmaPlot version 12 software. Nonparametric Mann–Whitney analysis was used to compare the isotypic anti-mAb1 antibody titer responses between the groups that received high and low doses of the various mAb1 formulations. Titers were compared on day 41. Nonresponders (serum samples with undetectable anti-mAb1 titers) were not included in statistical significance determination. Antibody titers following administration of 5 :g mAb1 adsorbed on glass, silicone oil microdroplets, or  on Alhydrogel were compared with titers from mice that had been administered 5 :g native mAb1. The same comparison was performed for the groups that received 500 :g mAb1. Also, the differences in antibody titers between the two dose levels were compared for each formulation. Differences in immune responses were considered to be significant when p values of 0.05 or less were obtained (Table 1). R

RESULTS Mice were inspected visually during the study. Nothing unusual was observed, and the mice appeared healthy throughout the study. The mice had normal weight gains. By day 41, BALB/c mice had gained up to 18% of their initial weights and C57BL/6J mice gained up to 50%. Table 1. The p Values for Anti-mAb1 Antibody Titers for Responding Mice Between High- and Low-Dose Levels (5 or 500 :g mAb1) for Each Formulation Calculated Using Nonparametric Mann–Whitney Analysis Formulation Mouse Strain

Isotype

Native mAb1

Adsorbed on Glass

Adsorbed on Al

Adsorbed on SiOil

BALB/c

IgM IgG1 IgG2b IgG3

0.394 0.041 0.132 0.485

0.394 0.394 0.699 0.699

0.394 0.002 0.310 0.699

0.394 0.026 0.310 0.485

C57BL/6J

IgM IgG1 IgG2b IgG3

1.0 1.0 1.0 0.015

0.394 0.699 0.015 0.937

0.937 0.041 0.699 1.0

0.937 1.0 0.699 0.394

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

Figure 1. The mean value of IgG levels (ADAs) against mAb1 in BALB/c mice determined by MSD-chemiluminescence following administration of 5 :g mAb1 (a) or 500 :g mAb1 (b) in native state (blue line with diamonds), mAb1 adsorbed on glass particles (red line with  R squares), mAb1 adsorbed on Alhydrogel (green line with triangles), and mAb1 adsorbed on silicone oil droplets (purple line with stars) on day 1 (preimmunization), 6 (after one injection), 13, 20 (after the booster injection), and 41 (end day; after 2-week recovery period).

The administration of 20 mM histidine buffer did not induce significant immune response in mice of either strain. The weak responses that were detected in the group of mice injected with buffer are obviously nonspecific because they were not exposed to mAb1 (Supplemental Figs. S5 and S6). Analysis of ADA Titers and mAb1 PK in BALB/c Mice In BALB/c mice, a strong ADA response was detected by MSDchemiluminescence analysis of serum samples from mice that received 5 :g mAb1 doses (Fig. 1a). Formulations containing mAb1 without added particles were immunogenic, but the most immunogenic mAb1 formulations were those that contained silicone oil droplets. ADA responses in BALB/c mice following 500 :g injections (Fig. 1b) were greatly diminished when compared with those resulting from 5 :g doses. Some ADA developed after mice received the higher dose, but these responses occurred at a later time and at a lower level compared with those resulting from the 5-:g doses. However, the formulation that contained silicone oil droplets again induced the highest ADA responses. DOI 10.1002/jps.24413

RESEARCH ARTICLE – Pharmaceutical Biotechnology

1615

Figure 2. Concentration of mAb1 in the serum samples of BALB/c mice detected by MSD-chemiluminescence in the group received 5 :g (a) or received 500 :g (b) mAb1 formulation. The level detected by ELISA in the group received 5 :g (c) or received 500 :g (d) mAb1 formulation. Each line represents the mean value of mAb1 concentration in the sera of mice that received native mAb1 (blue line with diamonds), mAb1 adsorbed  R on glass microparticles (red line with squares), mAb1 adsorbed on Alhydrogel (green line with triangles), and mAb1 adsorbed on silicone oil droplets (purple line with stars). Error bars are the standard deviation of the mean values.

The level of circulating mAb1 determined by MSDchemiluminescence in the sera of BALB/c mice (Fig. 2a) that received 5 :g mAb1 doses increased sharply on day 6 after the first injection and then decreased rapidly to a very low level on day 13. No increase in circulating mAb1 was detected in any group of mice at time points after the second injection. The level of circulating mAb1in the BALB/c mice that received 500 :g mAb1 formulations was high on day 6 after the first injection (Fig. 2b). A week later on day 13, the mAb1 level in all formulations was moderate and similar to those observed after the second injection on day 27. On day 20, the mAb1 levels in all formulations rose to similar levels as of day 6, indicating the absence of significant levels of ADAs. Each of the mAb1 formulations tested in BALB/c mice generated a high IgG1 response (Fig. 3). IgG1 responses measured by ELISA were significantly higher in mice that received the lower dose of the various mAb1 formulations compared with those that received the higher dose (Table 1), except in mice that were administered mAb1 adsorbed on glass microparticles (p = 0.394). In the responding BALB/c mice, no differences in the final anti-mAb1 IgG2b titers determined by ELISA were observed following administration of low or high doses of any of the mAb1 formulation types (Supplemental Fig. S7). However, in the low-dose groups, IgG2b responses were generated quickly, and ADAs could be detected on day 13 (9 days after the first injection). In contrast, in high-dose groups, most mice needed a second injection to elicit a response. Likewise, for the weak IgG3 response that was detected by ELISA in BALB/c mice, no significant differences in titers were observed between mice that received the low or high doses (SupDOI 10.1002/jps.24413

plemental Fig. S8; Table 1). In addition, the number of BALB/c mice that generated IgG3 responses to 5 :g doses of mAb1 was almost the same as the number of mice responding to 500 :g doses (Table 2). The IgM responses detected in BALB/c mice also were not strong (Supplemental Fig. S9), but more mice responded to low-dose formulations than to high-dose formulations (Table 2). Pharmacokinetic data determined by ELISA from BALB/c mice that received 5 :g doses of mAb1 showed an initial increase in the circulating level of mAb1 (measured on day 6 2 days after the first injection), followed by a rapid decrease within a week (Fig. 2c). No increase in mAb1 level was observed after the second injection of 5 :g mAb1 to BALB/c mice. In the groups of BALB/c mice that received 500 :g doses of mAb1, the level of mAb1 increased after each injection as detected by ELISA on day 6 (2 days after the injection on day 4) and on day 20 (2 days after the second injection on day 18). Interestingly, the level detected on the last day of the study (day 41) was very low (Fig. 2d). Analysis of ADA Titers and mAb1 PK in C57BL/6J Mice In C57BL/6J mice (syngeneic strain), the IgM and IgG1 responses to mAb1 generally were not strong. IgM levels measured by ELISA in responding mice (Supplemental Fig. S10) did not differ significantly between the two dosing levels (Table 1). Similarly, IgG1 levels in the responding mice did not differ greatly between the different dosing levels (Supplemental Fig. S11) with one exception: the calculated p value between two groups that received low or high doses of mAb1 adsorbed  on Alhydrogel was 0.041 (Table 1). R

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

1616

RESEARCH ARTICLE – Pharmaceutical Biotechnology

Figure 3. Anti-mAb1 titers of IgG1 in BALB/c mice (by ELISA) following the administration of 5 :g (upper panel) or 500 :g (lower panel) (a)  R mAb1 in native state, (b) mAb1 adsorbed on glass particles, (c) mAb1 adsorbed on Alhydrogel , and (d) mAb1 adsorbed on silicone oil droplets on day 1 (preimmunization), 6 (after one injection), 13, 20 (after the booster injection), and 41 (end day; after 2-week recovery period). Each bar represents the titer of one mouse serum sample.

Table 2. Number of BALB/c Mice with Antibody Responses to Administration of 5 :g (Low Dose) or 500 :g (High Dose) mAb1 IgM

Native mAb1 mAb1 adsorbed on glass  R mAb1 adsorbed on Alhydrogel mAb1 adsorbed on silicone oil

IgG1

IgG2b

IgG3

Low

High

Low

High

Low

High

Low

High

2 2 2 2

0 1 0* 1

6 6 6 6

6 6 4* 6

6 6 4 6

6 5 4* 6

5 3 1 2

3 3 0* 3

*Mice were counted as responders if the level of IgG titer was higher than the level of titer on day 1. Each group had six mice, except for those marked with an asterisk, which had only four mice.

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

DOI 10.1002/jps.24413

RESEARCH ARTICLE – Pharmaceutical Biotechnology

1617

Figure 4. Anti-mAb1 titer of IgG2b in C57BL/6J mice (by ELISA) to the administration of 5 :g (upper panel) or 500 :g (lower panel) (a) mAb1  R in native state, (b) mAb1 adsorbed on glass particles, (c) mAb1 adsorbed on Alhydrogel , and (d) mAb1 adsorbed on silicone oil droplets on day 1 (preimmunization), 6 (after one injection), 13, 20 (after the booster injection), and 41 (end day; after 2-week recovery period). Each bar represents the titer of one mouse serum sample.

The IgG2b response measured by ELISA in C57BL/6J mice was significantly stronger (p = 0.041) in the group of mice that received 5 :g adsorbed mAb1 on glass particles compared with the group of mice injected with the higher dose (Table 1; Fig. 4). IgG3 was the strongest response that was detected by ELISA in C57BL/6J mice. However, the number of responding C57BL/6J mice and the level of IgG3 response were not very different when the two dosing levels of injections were compared (Table 3; Fig. 5) with an exception for groups injected with native mAb1, wherein the response was much stronger in the high-dose group as compared with the low-dose group (p value of 0.015; see Table 1). No ADA response could be detected by MSDchemiluminescence analysis in serum from C57BL/6J DOI 10.1002/jps.24413

mice following 5 or 500 :g doses of mAb1, even when the  formulations contained added Alhydrogel , glass, or silicone oil microparticles. The level of circulating mAb1 by MSD-chemiluminescence in C57BL/6J mice that received 5 :g increased after each injection except when mAb1 was delivered in the silicone oil formulation (Fig. 6a). Following 500 :g/mL doses of mAb1 in C57BL/6J mice, circulating mAb1 levels were similar for all of the formulations (Fig. 6b). The level of circulating mAb1 measured by ELISA in C57BL/6J mice that received 5 :g (Fig. 6c) increased after each injection and decreased gradually toward day 41. The level of circulating mAb1 in the sera of mice that received adsorbed mAb1 on glass particles (Fig. 6c) decreased to a very low level R

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

1618

RESEARCH ARTICLE – Pharmaceutical Biotechnology

Table 3. Number of C57BL/6J Mice with Antibody Responses to Administration of 5 :g (Low Dose) or 500 :g (High Dose) mAb1 IgM

Native mAb1 Adsorbed on glass  R Adsorbed on Alhydrogel Adsorbed on silicone oil

IgG1

IgG2b

IgG3

Low

High

Low

High

Low

High

Low

High

0 1 1 1

0 3 1* 2

0 0 2 0

0 0 1* 0

0 4 1 0

0 2 2* 1

0 5 5 4

5 5 5* 4

*Mice were counted as responders if the level of IgG titer was higher than the level of titer on day 1. Each group had six mice, except for those marked with an asterisk, which had only five mice.

Figure 5. Anti-mAb1 titer of IgG3 in C57BL/6J mice (by ELISA) to the administration of 5 :g (upper panel) or 500 :g (lower panel) (a) mAb1  R in native state, (b) mAb1 adsorbed on glass particles, (c) mAb1 adsorbed on Alhydrogel , and (d) mAb1 adsorbed on silicone oil droplets on day 1 (preimmunization), 6 (after one injection), 13, 20 (after the booster injection), and 41 (end day; after 2-week recovery period). Each bar represents the titer of one mouse serum.

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

DOI 10.1002/jps.24413

RESEARCH ARTICLE – Pharmaceutical Biotechnology

1619

Figure 6. Concentration of mAb1 in the serum samples of C57BL/6J mice detected by MSD-chemiluminescence in the group that received 5 :g (a) or that received 500 :g (b) mAb1 formulation. The level detected by ELISA in the group that received 5 :g (c) or that received 500 :g (d) mAb1 formulation. Each line represents the mean value of mAb1 concentration in the sera of mice that received native mAb1 (blue line with  R diamonds), mAb1 adsorbed on glass microparticles (red line with squares), mAb1 adsorbed on Alhydrogel (green line with triangles), and mAb1 adsorbed on silicone oil droplets (purple line with stars). Error bars are the standard deviation of the mean values.

on day 41. The level of circulating mAb1 following administration of mAb1 adsorbed on silicone oil droplets was also very low during the course of the study. The level of circulating mAb1 in the groups of C57BL/6J mice (by ELISA) that received 500 :g doses of mAb1 formulations (Fig. 6d) decreased over time but remained above the limit of detection until last day of study, day 41. The detected mAb1 levels on day 41 were 37, 26, 80, and 5 :g/mL for mice injected with 500 :g doses of native mAb1 formulation, mAb1 adsorbed  on glass, mAb1 adsorbed on Alhydrogel , and mAb1 adsorbed on silicone oil droplets, respectively. Overall, mAb1 was more immunogenic when adsorbed to particles than when it was in its native, soluble form. In BALB/c mice, IgG1 responses were predominant. In contrast, in C57BL/6J mice, IgG2b and IgG3 responses were predominant. Anti-mAb1 antibody responses (especially IgG1) were greater in BALB/c mice receiving lower dose than the mice receiving higher dose. In C57BL/6J mice, the lower of the two doses elicited only a slightly higher ADA response (Supplemental Table S12). R

DISCUSSION ELISA Analyses of ADA Titers Overall, the immune responses to injections of mAb1 formulation (either 5 or 500 :g doses) were stronger in BALB/c mice than in C57BL/6J, and were dominated by the production of IgG1. This was in line with the result observed in our previous DOI 10.1002/jps.24413

study when mice of each strain were injected several times with 50 :g doses of mAb1.28 In BALB/c mice, the lower dose of mAb1 was more immunogenic than the higher dose. High- and low-dose tolerance effects were first demonstrated by Mitchison in 1964,34 but the dose level that is required for tolerances varies from one antigen to another. In our model system, it appears that injection of 500 :g mAb1 induces a tolerance in BALB/c mice. Determination of the mechanism of tolerance in this model system was beyond the scope of the current study; however, there have been several proposed mechanisms for inductance of high-zone tolerance. These include cathepsin-induced apoptosis of T cells35 as well as induction of CD4+ CD25+ suppressor T cells.36 The immune responses to mAb1 in the syngeneic C57BL/6J mice were not as strong as the responses detected in BALB/c mice, and the anti-mAb1 antibodies were produced mostly in the form of IgG3 and IgG2b isotypes, in contrast to the IgG1 response that was dominant in the BALB/c strain. It is likely that the target of the ADA response to mAb1 is different in the two mouse strains, with the response in BALB/c mice more likely directed against the allogenic Fc region, and the response in the syngeneic C57BL/6J mice being directed at neoepitopes formed by aggregation or adsorption to particles. Injecting C57BL/6J mice with either the lower or the higher dose of mAb1 produced similar level of immune responses. The concentrations  of particulate additives (glass, silicone oil, or Alhydrogel ) in both dosing formulations were equivalent. Because the adsorption isotherms for mAb1 are fairly steep, indicative of strong binding (see Supplementary Fig. S3 and previous report28 ), the R

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

1620

RESEARCH ARTICLE – Pharmaceutical Biotechnology

particle surfaces are essentially saturated at the concentration found in the low-dose samples, and further increases in concentration result in increased levels of free, unbound mAb1. Thus, the amount of adsorbed mAb1 present in both high- and low-dose samples was essentially equivalent, but the quantity of free, unadsorbed mAb1 was different. The apparent lack of a dose-dependent immune response to mAb1 in C57BL/6J might be explained if the response is largely because of mAb1 in particulate form, rather than the soluble, native form. In contrast, IgG1 responses were dose-dependent in the allogenic BALB/c mice, suggesting that they might have been responding predominantly to the unbound mAb1 in the various dosage forms. The results of the PK analysis in two mouse strains receiving 5 :g mAb1 indicated that the level of mAb1 in all formulations decreased faster in BALB/c than in C57BL/6J mice. This enhanced clearance is consistent with the greater ADA response detected in BALB/c mice. The same trend was observed in the groups of mice that received the 500-:g doses. Only very low levels of mAb1 could be detected in BALB/c mice by day 41, whereas in C57BL/6J mice that produced weaker ADA responses, the decrease in mAb1 level was much slower and relatively moderate levels could still be detected on day 41. MSD-Chemiluminescence Analysis of ADA Titers  R

Meso Scale Discovery -chemiluminescence analysis detected ADA responses in BALB/c mice that were consistent with those found by isotype analysis using standard ELISA techniques. But no ADAs were detected by MSD-chemiluminescence analysis in any groups of C57BL/6J mice. Interference form circulating drug might contribute to the inability of the MSDchemiluminescence technique to detect ADA in serum from C57BL/6J mice. In the ELISA-based assays, addition of an acid-dissociation step was found to improve ADA detection. MSD-chemiluminescence measured total antidrug IgG levels, whereas in the ELISA assays the immunoglobulin subclasses (IgM, IgG1, IgG2b, and IgG3) were separately measured. Each of the individual anti-mAb1 isotypes presumably exhibits a different affinity for mAb1, and the ELISA assays showed that mAb1 generated different isotype responses in each of the two mouse strains tested. Dose Response for Immune Responses to mAb1 Dose level is a factor that may influence the level of immune response to therapeutic proteins. The immunogenic and tolerogenic levels are not universal, but rather would be expected to depend on the drug identity. In the present study, mAb1 and various mAb1 aggregates at the lower of the two levels tested were slightly more immunogenic (in C57BL/6J mice) or much more immunogenic (in BALB/c mice) than at the higher level, an effect that can likely be ascribed to high-zone tolerization at the higher dose. Neither the minimum dose that was capable of inducing an ADA response, nor the minimum dose for inducing a tolerogenic response was determined in this study. mAb1 readily adsorbs to microparticles such as glass, silicone oil, or aluminum hydroxide, which act as adjuvants to increase the probability of an immune response. Although aggregated forms of several therapeutic proteins enhance immune responses in mice13,15,28,37–40 and aggregated antibodies stimulate immune responses in peripheral blood monocyte cultures,16,19 there are few published reports on any potential correlations between particulates in marketed drug products Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015

and levels of immune response in patients, and clinical trials to directly assess the immunogenic potential of particulate contaminants in patients would be unethical. Furthermore, the stronger ADA response that we generally observed in response to the lower dose of mAb1 in formulations containing microparticles complicates extrapolation of results in murine models to patients. Recently, however, a strong correlation was reported between levels of neutralizing antibodies in patients receiving various commercial interferon-$ formulations and the levels of micro- and nanoscopic particle contaminants that could be detected in the formulations.41 It should be noted that the formulations that we tested did not include surfactants. Most, but not all, commercial formulations of therapeutic proteins contain a nonionic surfactant. We have previously shown that addition of the nonionic surfactant polysorbate 20 to formulations of trastuzumab decreased the amount of trastuzumab that adsorbed to silicone oil–water interfaces.22 Thus, if antigen adsorption is critical for generating immune responses, it might be expected that surfactant addition might reduce immunogenicity. On the contrary, surfactants are commonly added to vaccine formulations,42 and addition of Tween to the squalene oil–water microemulsions that comprise the commercial vaccine adjuvant MF59 is critical to maintain the preparation’s adjuvanticity.43 Early studies showed that the nonionic surfactants Span 20, but not Tween 60, acted as adjuvants when BGG was injected into mice.44 Thus, addition of surfactants might alter some of the enhanced immune response we see in formulations containing glass microparticles or silicone oil micropdroplets, but we cannot predict the effects of such additions based on the current set of experiments.

CONCLUSIONS Immune responses in mice to a murine monoclonal antibody were generally higher at the lower of the two dose levels tested, possibly because of a high-zone tolerization effect at the higher dose. Immune responses were different in each of the two mouse strains tested, as evidenced by the different patterns and levels of ADA isotypes that were produced, as well as in the different PK profiles for mAb1 in the two strains. The monoclonal antibody that we tested readily adsorbed to microparticles such as glass or silicone oil, and addition of such microparticles increased the incidence of immune responses, analogous to the effect of particulate adjuvants in vaccine formulations.

ACKNOWLEDGMENTS We would like to thank AbbVie GmbH and Company KG, Ludwigshafen, Germany for supporting this project financially. Also, thanks to Professor Raul Torres (National Jewish Health, Denver, Colorado) for scientific discussions and suggestions in immunogenicity study. Zehra Kaymakcalan and Michael Siedler are employees of AbbVie Inc. and are AbbVie stockholders. Sultan Tanriverdi is a contract worker at AbbVie Inc. The University of Colorado and the Ludwig-Maximilians-University Munich received research funds from AbbVie (former Abbott Laboratories) to conduct the study. AbbVie (formerly Abbott Laboratories) provided financial support, provided the murine antibody used in this study, as well as resources to support the in vivo studies and the DOI 10.1002/jps.24413

RESEARCH ARTICLE – Pharmaceutical Biotechnology

bioanalytical characterization. Furthermore, Abbott authors were involved in study design, research, analysis, data collection, interpretation of data, reviewing, and approving the publication.

REFERENCES 1. Dresser DW. 1962. Specific inhibition of antibody production. I. Protein overloading paralysis. Immunology 5:161–168. 2. Mitchison NA. 1968. The dosage requirements for immunological paralysis by soluble proteins. Immunology 15(4):509–530. 3. Braun A, et al. 1997. Protein aggregates seem to play a key role among the parameters influencing the antigenicity of interferon alpha in normal and transgenic mice. Pharm Res 14(10):1472–1478. 4. Anderson PJ. 2005. Tumor necrosis factor inhibitors: Clinical implications of their different immunogenicity profiles. Semin Arthritis Rheum 34(5):19–22. 5. Hermeling S, et al. 2006. Antibody response to aggregated human interferon alpha2b in wild-type and transgenic immune tolerant mice depends on type and level aggregation. J Pharm Sci 95:1084–1096. 6. Kessler M, Goldsmith D, Schellekens H. 2006. Immunogenicity of biopharmaceuticals. Nephrol Dial Transplant 21:v9–v12. 7. Schellekens H. 2003. Immunogenicity of therapeutic proteins. Nephrol Dial Transplant 18(7):1257–1259. 8. Christian CL. 1960. Studies of aggregated gamma-globulin II. Effect in vivo. J Immunol 84:117–121. 9. Dresser D, Gowland G. 1964. Immunological paralysis induced in adult rabbits by small amount of a protein antigen. Nature 203:733– 736. 10. Gamble C. 1966. The role of aggregates in the primary immune response of mice to human gamma globulin. Int Arch Allergy 30(5):446– 455. 11. Ring J, et al. 1977. Anaphylactoid reactions due to non-immune complex serum protein aggregates. Monogr Allergy 12:27–35. 12. Hermeling S, et al. 2005. Development of a transgenic mouse model immune tolerant for human interferon beta. Pharm Res 22(6):847–851. 13. Fradkin AH, Carpenter JF, Randolph, TW. 2009. Immunogenicity of aggregates of recombinant human growth hormone in mouse models. J Pharm Sci 98(9):3247–3264. 14. Seefeldt MB, et al. 2009. Application of high hydrostatic pressure to dissociate aggregates and refold proteins. Curr Pharm Biotechnol 10(4):447–455. 15. van Beers MMC, et al. 2010. Aggregated recombinant human interferon beta induces antibodies but no memory in immune-tolerant transgenic mice. Pharm Res 27(9):1812–1824. 16. Joubert MK, et al. 2012. Highly aggregated antibody therapeutics can enhance the in vitro innate and late-stage T-cell immune responses. J Biol Chem 287(30):25266–25279. 17. Rombach-Riegraf V, et al. 2014. Aggregation of human recombinant monoclonal antibodies influences the capacity of dendritic cells to stimulate adaptive T-cell responses in vitro. PLoS One 9(1):e86322. 18. Ahmadi M, et al. 2014. Small amounts of sub-visible aggregates enhance the immunogenic potential of monoclonal antibody therapeutics. Pharm Res.[Epub ahead of print] 19. Chi EY, et al. 2005. Heterogeneous nucleation-controlled particulate formation of recombinant human platelet-activating factor acetylhydrolase in pharmaceutical formulation. J Pharm Sci 94(2):256–274. 20. Bee JS, et al. 2009. Monoclonal antibody interactions with microand nanoparticles: Adsorption, aggregation, and accelerated stress studies. J Pharm Sci 98(9):3218–3238. 21. Tyagi AK, et al. 2009. IgG particle formation during filling pump operation: A case study of heterogeneous nucleation on stainless steel nanoparticles. J Pharm Sci 98(1):94–104.

DOI 10.1002/jps.24413

1621

22. Ludwig DB, et al. 2010. Protein adsorption and excipient effects on kinetic stability of silicone oil emulsions. J Pharm Sci 99(4):1721– 1733. 23. Fujiwara M. 1977. Cellular mechanisms of the resistance to the induction of immunological tolerance. Immunology 32(5):783–791. 24. Das S, Leskowitz S. 1974. The cellular basis for tolerance or immunity to bovine-gamma-globulin in mice. J Immunol 112(1):107– 114. 25. Golub ES, Weigle WO. 1969. Studies on the induction of immunologic unresponsiveness: III. Antigen form and mouse strain variation. J Immunol 102(2):389–396. 26. Martin RM, Brady JL, Lew AM. 1998. The need for IgG2c specific antiserum when isotyping antibodies from C57BL/6 and NOD mice. J Immunol Methods 212(2):187–192. ¨ a¨ I, et al. 1981. The four subclasses of IgG can be isolated 27. Seppal from mouse serum by using protein A-Sepharose. Scand J Immunol 14(4):335–342. 28. Shomali M, et al. 2014. Antibody responses in mice to particles formed from adsorption of a murine monoclonal antibody onto glass microparticles. J Pharm Sci 103(1):78–89. 29. Wilson-Welder JH, et al. 2009. Vaccine adjuvants: Current challenges and future approaches. J Pharm Sci 98:1278–1316. 30. HogenEsch H, et al. 2002. Mechanisms of stimulation of the immune response by aluminum adjuvants. Vaccine 20:S34–S39 31. Petrovsky N, Aguilar JC. 2004. Vaccine adjuvants: Current state and future trends. Immunol Cell Biol 82:488–496. 32. Freitag AJ, et al. 2015. Investigation of the immunogenicity of different types of aggregates of a murine monoclonal antibody in mice. Pharm Res 32(2):430–444. 33. Malyala P, Singh, M. 2008. Endotoxin limits in formulations for preclinical research. J Pharm Sci 97(6):2041–2044. 34. Mitchison NA. 1964. Induction of immunological paralysis in two zones of dosages. Proc R Soc Lond B 161:275–292. 35. Michallet M-C, et al. 2004. Cathepsin-dependent apoptosis triggered by supraoptimal activation of T lymphocytes: A possible mechanism of high dose tolerance. J Immunol 172(9):5405–5414. 36. Apostolou I, Von Boehmer H. 2004. In vivo instruction of suppressor commitment in naive T cells. J Exp Med 199(10):1401– 1408. 37. Fradkin AH, Carpenter JF, Randolph TW. 2011. Glass particles as an adjuvant: A model for adverse immunogenicity of therapeutic proteins. J Pharm Sci 100(11):4953–4964. 38. van Beers MMC, et al. 2012. Immunogenicity of recombinant human interferon beta interacting with particles of glass, metal, and polystyrene. J Pharm Sci 101(1):187–199. 39. Christie M, et al. 2014. Recombinant murine growth hormone particles are more immunogenic with intravenous than subcutaneous administration. J Pharm Sci 103(1):128–139. 40. Filipe V, et al. 2012. Immunogenicity of different stressed IgG monoclonal antibody formulations in immune tolerant transgenic mice. mAbs 4:6:740–752. 41. Barnard JG, et al. 2013. Characterization and quantitation of aggregates and particles in interferon-$ products: Potential links between product quality attributes and immunogenicity. J Pharm Sci 102(3):915–928. 42. Ascarateil S, Dupuis L. 2006. Surfactants in vaccine adjuvants: Description and perspectives. Vaccine 24 Suppl 2:S2–83–85. 43. Calabro S, Tritto E, Pezzotti A, Taccone M, Muzzi A, Bertholet S, De Gregorio E, O’Hagan DT, Baudner B, Seubert A. 2013. The adjuvant effect of MF59 is due to the oil-in-water emulsion formulation, none of the individual components induce a comparable adjuvant effect. Vaccine 31(33):3363–3369. 44. Dresser DW. 1968. An assay for adjuvanticity. Clin Exp Immunol 3(9):877–888.

Shomali et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:1610–1621, 2015