Abstract
The immunogenicity of therapeutic Abs is a concern as anti-drug Abs may impact negatively on the pharmacodynamics and safety profile of Ab drugs. The factors governing induction of anti-drug Abs are not fully understood. In this study, we describe a model based on mouse-human chimeric Abs for the study of Ab immunogenicity in vivo. Six chimeric Abs containing human V regions and mouse C regions were generated from six human anti-Rhesus D Abs and the Ag-binding characteristics of the parental human Abs were retained. Analysis of the immune response toward the individual chimeric Abs revealed the induction of anti-variable domain Abs including anti-idiotypic Abs against some of these, thereby demonstrating the applicability of the model for studying anti-drug Ab responses in vivo. Immunization of BALB/c, C57, and outbred NMRI mice with a polyclonal composition consisting of all six chimeric Abs demonstrated that the immunogenicity of the individual Abs was haplotype dependent. Chimeric Abs, which were nonimmunogenic when administered individually, did not become immunogenic as part of the polyclonal composition, implying the absence of epitope spreading. Ex vivo Ab-binding studies established a clear correlation between the level of immunogenicity of the Abs comprised in the composition and the impact on the pharmacology of the Abs. These analyses demonstrate that under these conditions this polyclonal Ab composition was generally less susceptible to blocking Abs than the respective mAbs.
The therapeutic potential of Abs has been recognized for more than a century and today Ab-based therapeutics find widespread use in the clinic. Originally, so-called passive Ab therapy was based on polyclonal Abs (pAb)3 derived from animal or human serum (1, 2). However, technological developments over the last three decades have allowed the generation of more and more refined Ab therapeutics; especially, the introduction of mAb technologies has broadened the clinical application of Ab therapeutics. The majority of Abs in clinical use are either mouse-human chimeric Abs consisting of human C regions and a mouse V region, or so-called humanized Abs where the mouse CDR have been grafted into a human Ab framework (3). However, the majority of Abs in clinical development are humanized Abs and fully human Abs generated by various cloning technologies or by the use of mice transgenic for parts of the human Ab repertoire (4).
One of the major driving forces in the development of Ab therapeutics has been to decrease their immunogenicity. The initial mouse mAb elicited blocking Ab responses in humans which significantly reduced their bioavailability in nearly all treated patients and the clinical usefulness of such Abs. Substituting the mouse C regions with human C regions in chimeric Abs greatly reduced the immunogenicity of these Abs. However, the induction of neutralizing Abs against chimeric Abs remains a well-known phenomenon, despite a varying degree of impact on different chimeric Abs (5, 6). As expected, immunogenicity is observed especially after repeated long-term administration of the Abs in chronic disorders. Hence, up to 60% of patients treated with the chimeric anti-TNF-α Ab Infliximab develop neutralizing Abs which are associated with an increased risk of infusion-related reactions and reduced treatment response (7, 8). Also humanized Abs, containing <5% mouse sequence, may elicit immune responses in >50% of the treated patients (5, 6, 9, 10).
Anti-drug Ab responses have also been reported in patients treated with fully human Abs (5). Thus, despite the minimization and even complete removal of xenogenic sequences in the therapeutic Abs, the induction of Abs potentially interfering with the pharmacological action of the Abs cannot be excluded. A potential reason for the inability to eliminate immunogenicity even in fully human Abs is the fact that CDR differ significantly between human beings, which combined with haplotype differences and differences in the TCR repertoire may result in certain CDR sequences being immunogenic in particular individuals (11). The multifactorial nature of this makes it difficult to predict the immunogenicity of a given therapeutic Ab in an outbred patient population before development of the Ab.
Technologies for cloning and manufacturing recombinant human pAb now exist (4, 12). Therapeutics based on recombinant pAb are attractive, because they possess the specificity of mAb, but offer the pharmacological benefits of targeting more than one Ag epitope, e.g., on pathogens or malignant cells. Also, it can be speculated that pAb therapeutics would be less susceptible to neutralizing Abs than mAb. Hence, the extent of the immunogenicity of the individual Abs comprised in such a pAb-based drug would differ between individuals treated, and the likelihood of all sequences being immunogenic in that individual would be small. Thus, in any individual, the presence of blocking Abs against one of the Ab species would only have a limited impact on efficacy of a recombinant pAb preparation compared with mAb and likely lead to a reduced side effect profile.
In this study, we have established a model for the in vivo study of factors governing the induction of neutralizing Abs. We have generated six unique mouse-human chimeric IgG1 Abs specific for the human polymorphic erythrocyte Ag rhesus D (RhD). The transfer of the human variable parts to the mouse IgG1 C region did not affect the binding properties in vitro but significantly reduced the immunogenicity of the Abs after administration to mice. In vivo experiments demonstrated that the immunogenicity of the V regions of the individual chimeric Abs varies with the recipient haplotype. Furthermore, chimeric Abs, which were nonimmunogenic when administered individually, did not become immunogenic as part of the polyclonal composition, implying the absence of epitope spreading. Ex vivo Ab-binding studies established a clear correlation between the level of immunogenicity of the Abs comprised in the composition and the impact on the pharmacology of the Abs. Hence, it appeared that under the conditions of the present study, pAb are less susceptible to Ab-mediated interference than individual mAb.
Materials and Methods
Human RhD-specific Abs
A range of human Abs specific for the polymorphic erythrocyte Ag RhD were isolated by standard phage display (13). Six of these Abs were chosen for the current study based on sequence diversity (Fig. 1⇓).
H chain V region sequences of the of the six human anti-RhD Abs. The CDR1, CDR2, and CDR3 regions and the human VH families they represent are indicated.
Generation of expression vectors encoding RhD-specific mouse-human chimeric Abs
The mouse Cκ region and the human Vκ regions were assembled by overlap PCR in two steps, to generate the six chimeric L chains; similar methodology as previously described by Horton et al. (14) was used. Overlap primers specific for each connection and end-primers specific for mouse Cκ and each human Vκ family are shown in Table I⇓.
Overlap primers, end primers, and primers for altering of restriction sites
The complete mouse IgG1 CH region was amplified from a previously cloned and verified genomic DNA fragment and provided with an XhoI site at the 5′-end and a SpeI site at the 3′-end. The assembled chimeric L chains and the mouse IgG1 CH region were transferred to the expression vectors encoding the human VH regions as NheI-NotI and SpeI-XhoI fragments, respectively. Complete sequence analysis (Agowa) of all six vectors constructs revealed no mutations.
Nucleotide sequence accession numbers
The VH- and VL-encoding sequences of the mouse-human chimeric Abs have been deposited in GenBank with accession no. DQ437872-DQ437883.
Expression and purification of mouse-human chimeric Abs
Using site-specific integration, the vector constructs encoding six mouse-human chimeric Abs were transfected into Chinese hamster ovary (CHO) FlpIn cells (Invitrogen Life Technologies) using Fugene 6 (Roche). Adherent CHO cells expressing mouse-human chimeric Abs were cultivated in serum-free ProCHO4-CDM medium (Cambrex). Supernatants were filtered (0.45-μm low protein binding filters) and the chimeric Abs were purified by affinity chromatography using either HiTrap rProtein A FF columns or HiTrap G HP columns (Amersham Biosciences), as described by Wiberg et al. (15). Eluting fractions were analyzed by SDS-PAGE with or without reducing agent and stained with Coomassie brilliant blue (Invitrogen Life Technologies).
Assessing the binding reactivity of chimeric and human RhD-specific Abs
Binding activity of RhD-specific Abs was measured by indirect Coombs test with fluorescence-conjugated secondary Abs, followed by flow cytometry analysis. Hence, RhD-positive erythrocytes, 50 μl diluted 1/10 in PBS with 1% BSA (PBS/BSA) were added to round-bottom 96-well plates. The RhD+ erythrocytes were centrifuged at 2000 × g for 1 min and resuspended in 150 μl of PBS three times. The final pellet of washed erythrocytes was incubated with 50-μl serial dilutions of the RhD-specific mouse-human chimeric or the corresponding human RhD-specific Abs in PBS/BSA. After incubation for 1 h at room temperature (RT), the erythrocytes were washed four times in 150 μl of PBS and 50 μl of allophycocyanin-conjugated rat anti-mouse IgG1 Ab (BD Biosciences) or PE-conjugated goat anti-human IgG Ab (Beckman Coulter) diluted 1/200 in PBS/BSA was added to each well.
Flow cytometry
For flow cytometric analysis of Ab binding, the erythrocytes were washed three times in 150 μl of PBS after visual inspection and resuspended in 150 μl of PBS/BSA. Samples were acquired using a FACSCalibur flow cytometer (BD Bioscience) equipped with a high throughput screening unit. The mean fluorescence intensity (MFI) of allophycocyanin- or PE-conjugated secondary Abs was measured and analyzed by CellQuest software (BD Biosciences).
Immunization studies
Female BALB/c (H-2d), C57BL/6 (H-2b), and NMRI mice (H-2q), 8–10 wk of age, were from Bomholtgaard and were housed at the animal unit at the University of Copenhagen under standard conditions. All animal experiments were approved by the relevant committee. For evaluation of Ab immunogenicity, BALB/c mice were immunized s.c. in the neck with 20 μg of each of the individual monoclonal chimeric or human IgG1 Abs (3–5 mice/group). For immunization with polyclonal composition, the six chimeric Abs were present in equal amounts, 3.33 μg, to give a total of 20 μg per mouse (5–10 mice/group). The Ab solutions were mixed 1:1 with 50 μl of CFA (Sigma-Aldrich). A group of 5 mice were used as controls and immunized with 100 μl of PBS. After 2 wk, mice were boosted by s.c. administration of 20 μg of the relevant Ab in 100 μl of IFA (Sigma-Aldrich). After two additional weeks, all mice were sacrificed and bled by cardiac puncture. The blood was left to coagulate, centrifuged at 15,000 rpm for 10 min and the sera were collected.
In separate experiments, BALB/c, C57, and NMRI mice were immunized with a polyclonal mixture containing a total amount of 20 μg of Ab. Immunization and serum preparation was performed as described above.
Identification of anti-idiotypic and V-gene family-specific Abs
The reactivity of sera from BALB/c, NMRI, and C57 mice immunized with each of the six RhD-specific monoclonal chimeric or human Abs or the polyclonal composition was assessed in ELISA. Ninety-six-well ELISA plates (Corning) were coated overnight with 1 μg/ml of the relevant human RhD-specific Ab, mouse IgG1, or human IgG1 in 50 mM carbonate buffer (pH 9.4) and blocked 1 h at RT with PBS with 2% skim milk powder. After 1 h incubation at RT the plates were washed four times in PBS with 0.05% Tween 20 (PBST) before applying sera from immunized mice diluted 100, 300, 900, and 2700 times in PBST. After 1-h incubation, HRP-conjugated goat anti-mouse IgG/IgM Ab (Caltag Laboratories) diluted 1/3000 in PBST was applied and left to incubate 1 h at RT. After washing four times with PBST, the plates were developed with TMB Plus Substrate for HRP (KemEnTec) and the absorbance at 450 nm was determined. Selected sera were also tested for the presence of IgM, IgG1, IgG2a, and IgG2b Abs specific for the V region of human RhD1/chimeric 1AG1 Abs and a human VH2 Ab not specific for RhD.
Serum inhibition studies
Three sera from BALB/c mice immunized with 4DG1 or 6FG1 mAb or a mix of all six mouse-human chimeric Abs were tested as well as three sera from NMRI mice and C57 mice immunized with a mix of all six chimeric Abs.
The inhibition assays were performed as follows: mouse sera (30 μl) diluted five times followed by 2-fold dilutions were mixed and left to incubate 1 h at RT with 30 μl of (60 ng/ml) RhD4, RhD6, or a mix of all six RhD-specific Abs. Fifty microliters of this mixture was then transferred to round-bottom 96-well plates containing washed pellets of RhD+ erythrocytes and incubated as described above. Abs bound to erythrocytes were detected by PE-conjugated goat anti-human IgG Ab followed by flow cytometry as described above. The level of inhibition was calculated as: binding percent = ((MFI in the absence of mouse serum − MFI in the presence of mouse serum)/(MFI in the absence of mouse serum)) × 100.
Results
Construction of mouse-human chimeric Abs
To dissect the mouse immune response against the V and C regions of human Abs, a panel of mouse-human chimeric Abs was generated from six human RhD-specific Abs. The six mouse-human chimeric L chains were assembled by PCR and inserted into expression vectors containing the human VH regions. Restriction sites flanking the mouse IgG1 CH region were engineered before insertion of the sequences into the intermediate expression vectors containing the mouse-human chimeric L chains and the human VH region to complete the generation of expression vectors encoding RhD-specific mouse-human chimeric Abs (Fig. 2⇓). All modifications of the expression vectors were verified by restriction digest analysis and sequence analysis. The individual expression vectors were transfected into CHO cells, expressed, and the six generated mouse-human chimeric Ab were purified as described in Materials and Methods.
Ab expression vectors. A, Parental expression vector containing only human Ab sequences. B, Intermediate expression vector containing mouse-human chimeric L chain and human H chain. C, Final expression vector containing complete mouse-human chimeric Ab sequences. AMP, ampicillin resistance gene; pUC origin, a pUC origin of replication; NotI, SacI, HpaI, PmlI, XhoI, and SpeI, restriction enzyme sites; promoter A/promoter B, head-to-head promoter cassette including leader sequences; FRT site, a FRT recombination site; neomycin. Gene for neomycin resistance.
Verification of structure and binding characteristics of mouse-human chimeric Abs
The six parental Abs span a 5-fold difference in binding strength for RhD+ erythrocytes and are all capable of agglutinating RhD+ erythrocytes in the presence of secondary Abs (data not shown). To verify that the strategy used here resulted in functional mouse-human chimeric IgG1 Abs (Table II⇓), we tested the ability of these Abs to agglutinate human RhD+ erythrocytes in the presence of secondary Abs using an indirect Coombs test. Secondary anti-mouse IgG1 Abs induced erythrocyte agglutination whereas anti-mouse IgG2b Abs did not. No agglutination was observed when RhD− erythrocytes or isotype control Abs were used (data not shown). We then addressed whether the replacement of human IgG1 C region with the mouse IgG1 C region would influence the binding pattern of the Abs. Fig. 3⇓ shows the binding curves of the individual chimeric Ab in comparison to their human counterpart. The binding of A1G1, 2BG1, and 3CG1 was identical to that of the parental Abs whereas the titer of 4DG1, 5EG1, and 6FG1 was 2- to 3-fold less than that of the parental Ab. These data demonstrate that replacement of the human IgG1 C region only affected the Ab binding minimally.
Chimeric RhD-specific Abs retain the binding capacity of the parental Abs. The binding of the parental human Abs (•) and the corresponding chimeric Abs (○) to RhD+ erythrocytes was assessed using PE-conjugated goat anti-human IgG or allophycocyanin-conjugated rat anti-mouse IgG1 and flow cytometry. Ab binding is normalized to the highest mean fluorescence intensity of the individual Abs. A, Binding of 1AG1 and RhD1 to RhD+ erythrocytes; B, 2BG1 and RhD2; C, 3CG1 and RhD3; D, 4DG1 and RhD4; E, 5EG1 and RhD5; and F, 6FG1 and RhD6. One representative experiment of three is shown.
Characteristics of mouse-human chimeric anti-RhD Abs
We further assessed the Ab-binding capacity of the chimeric Abs by competition with a different RhD-specific human VH3 Ab, harboring V regions different from the six chimeric Abs. These results showed that all six chimeric Abs inhibit binding of this VH3 Ab to RhD in a dose-dependent manner. The inhibitory capacity of the individual Abs correlated with their own binding to RhD (data not shown). This is consistent with the notion that most RhD Ab epitopes are overlapping only allowing the binding of one Ab per RhD molecule (16, 17).
Anti-V domain and idiotypic Ab responses to individual chimeric Abs in vivo
Having verified that the chimeric Abs had retained the Ag-binding characteristics of the parental Abs, we assessed the immunogenicity of the individual chimeric Abs in vivo. Thus, BALB/c mice were immunized with human or chimeric Abs and the specific anti-Ab responses were measured by ELISA 4 wk later. As seen in Fig. 4⇓, the human RhD2 Ab gave rise to a prominent Ab response in BALB/c mice. As expected, this response was directed against the C regions as demonstrated by the strong serum reactivity against human control IgG1 Abs. In contrast, when mice were immunized with the chimeric counterpart, 2BG1 which contains the human RhD2 V region and mouse C region, the immunogenicity was completely abolished, even in mice boosted 2 wk later (Fig. 4⇓ and data not shown).
Elimination of C region immunogenicity with chimeric Abs. BALB/c mice were immunized with 20 μg of the human IgG1 Ab, RhD2, or the derived chimeric Ab 2BG1 and sera were collected 4 wk post 1. immunization. Sera from mice immunized with the human RhD2 Ab (□) or the chimeric counterpart 2BG1 (▪) were tested for reactivity against human non-RhD-specific IgG1 Abs and human RhD2 Abs in Ab-specific ELISA. All sera were tested at 100-fold dilution. Wells were coated with 1 μg/ml of the human RhD2 Ab, human non-RhD-specific IgG1 Abs or mouse non-RhD-specific IgG1 Ab. Bound Ab was detected by HRP-conjugated anti-mouse IgG and IgM Abs, diluted 1/3000. All values were normalized to the absorbance of the wells coated with mouse non-RhD-specific IgG1 Ab. Data represent the mean from three individual experiments; SEM values are indicated.
Fig. 5⇓ dissects the immunogenicity of the individual chimeric Abs 2 wk after immunization of BALB/c mice. The V regions of Abs 1AG1, 3CG1, 4DG1, and 6FG1 elicits an Ab response in BALB/c mice whereas 2BG1 and 5EG1 does not.
Identification of anti-variable domain Abs including anti-idiotypic Abs in BALB/c mice. Sera from BALB/c mice immunized with 20 μg of either chimeric Ab 1AG1 (A), 2BG1 (B), 3CG1 (C), 4DG1 (D), 5EG1 (E), or 6FG1 (F) were collected 2 wk postboosting with 20 μg of Ab (4 wk after 1 immunization). Using Ab-specific ELISA, the individual sera were tested for reactivity against the V region of chimeric Ab used for immunization, the V region of each of the five other chimeric Abs and human non-RhD-specific IgG1 Abs. All sera were tested at 100-fold dilutions. Wells were coated with 1 μg/ml of one of the six human RhD Abs, human non-RhD-specific IgG1 Ab or mouse non-RhD-specific IgG1 Ab. Bound Ab was detected by HRP-conjugated anti-mouse IgG and IgM Abs, diluted 1/3000. All values were normalized to the absorbance of the wells coated with mouse non-RhD-specific IgG1 Ab. Data represent the mean from three individual experiments; SEM values are indicated.
Ab 4DG1 is immunogenic in BALB/c mice, whereas 2BG1 is not. Because these Abs both are of the VH4 family but harbor different CDR sequences, the observed immunogenicity against 4DG1 was likely to be directed toward the Ab Id. Similar results were obtained when comparing the two chimeric VH3 Abs, 5EG1 and 6FG1. Immunogenicity of 5EG1 was negligible compared with the prominent response against 6FG1 in BALB/c mice (Fig. 5⇑). As 1AG1 is the only VH2 Ab among the six chimeric Abs, sera from BALB/c mice immunized with 1AG1 were tested for reactivity against another human VH2 Ab. This revealed negligible reactivity as compared with the reactivity against the 1AG1 V regions (data not shown) and it is thus likely that this response was also anti-idiotypic. Thus, replacing the human C region in this system allows the identification of anti-variable domain and presumably also idiotypic Ab responses in vivo.
Immunogenicity of pAb compositions
We then analyzed the immunogenicity of the individual chimeric Abs when administered to BALB/c mice as constituents of a model pAb composition. Mice were immunized with an equal mixture of the six chimeric IgG1 Abs, and the serum reactivity against the individual Abs and the pAb composition was assessed 2 wk later. As seen from Fig. 6⇓, the Abs 1AG1, 4DG1, and 6FG1 all elicited anti-variable domain responses when comprised in the pAb composition. However, the chimeric Ab 3CG1 which gave rise to Ab responses when administered alone (Fig. 5⇑), did not elicit a response when comprised in the pAb composition (Fig. 6⇓). This absence of immunogenicity was not due to insufficient Ab concentrations as similar data were obtained when the Ab concentration was increased 6-fold to reflect the concentration used when the Abs were administered individually (data not shown). These data indicate masking of the antigenic determinants in 3CG1 by the other constituents in the pAb compositions and further indicated no evidence for epitope spreading. Hence, the immunogenicity of a single Ab was not significantly altered by being included in a mixture of Abs.
The immunogenicity of pAbs is haplotype-dependent. BALB/c (A), C57 (B), and NMRI (C) mice were immunized with 20 μg of pAb (▪), which is an equal mixture of the six chimeric Abs or PBS (▦). Sera were collected 2 wk postboosting with 20 μg of pAb. Using Ab-specific ELISA, the individual sera were tested for reactivity against the V region each of the six monoclonal chimeric Abs in the pAb composition and human non-RhD-specific IgG1 Abs (Hum IgG1). All sera were tested at 100-fold dilutions. Wells were coated with 1 μg/ml of one of the six human RhD Abs, human non-RhD-specific IgG1 Ab or mouse non-RhD-specific IgG1 Ab. Bound Ab was detected by HRP-conjugated anti-mouse IgG and IgM Abs, diluted 1/3000. All values were normalized to the absorbance of the wells coated with mouse non-RhD-specific IgG1 Ab. Data represent the mean from three individual experiments; SEM values are indicated.
Next, we investigated the magnitude and extent of immunogenicity of chimeric pAb in different mouse strains. Immunization of C57 mice only elicited a weak immune response toward the chimeric Abs 2 wk postimmunization (data not shown). However, when these mice were boosted, Ab responses were observed against all six chimeric Abs comprised in the pAb (Fig. 6⇑). In contrast, when immunizing outbred but haplotype identical NMRI mice with the pAb, only the 4DG1 Ab elicited a measurable Ab response. Thus, the extent and magnitude of the anti-variable domain Ab responses was dependent on the genetic background of the recipient.
The pAbs are less susceptible to blocking Abs than the mAbs
We wanted to assess the susceptibility of pAb to changes in pharmacological properties mediated by blocking anti-variable domain Abs. We therefore took advantage of the fact that the chimeric Ab 4DG1 was immunogenic in all three mouse strains tested and that the immunogenicity of the Ab comprised in the pAb composition differed among these strains. Thus, sera from mice immunized with different Abs were tested for their ability to compete with either human RhD4 or human pAb binding to RhD in vitro. Fig. 7⇓A shows that serum from BALB/c mice immunized with the chimeric Ab 4DG1 completely inhibits binding of the corresponding Ab RhD4 to RhD+ erythrocytes in a dose-dependent manner, whereas it had little or no effect on the binding of pAb comprising all the six anti-RhD Abs. These data were confirmed using the 6FG1 Ab (Fig. 7⇓B). Also, similar data were observed when using sera from NMRI mice immunized with the pAb (Fig. 7⇓D), in agreement with only 4DG1 being immunogenic in these mice. Three of the six Abs were immunogenic in BALB/c mice. Serum from mice immunized with pAb completely blocked the binding of RhD4 to RhD whereas it inhibited binding of the human pAb by ∼50–60% (Fig. 7⇓C). In agreement with the observation that all six Abs used here are immunogenic in C57 mice after repeated immunization, sera from these mice completely inhibited the binding of both human RhD4 and pAb to RhD+ erythrocytes (Fig. 7⇓E). These results demonstrated a clear correlation between the level of immunogenicity and pharmacological effect: it was necessary to have anti-variable domains Abs toward each single component to block a polyclonal composition. Hence, due to the reduced likelihood of inducing a full set of blocking Abs, we argue that pAb are generally less susceptible to inhibitory Ab responses than mAb.
The pAbs are generally less susceptible to blocking Abs than mAb. BALB/c mice were immunized with 20 μg of 4DG1 (A), 6FG1 (B), and the chimeric pAb (C), containing all six chimeric Abs in an equal mixture. NMRI mice (D) and C57 mice (E) similar received a dose of 20 μg of pAb. Sera were collected 2 wk after boosting with 20 μg of pAb. The sera were diluted 100 times and tested for the ability to inhibit the binding of 60 ng/ml parent human pAb (□), RhD4 (▪ in A and C–E), and RhD6 (▪ in B), to RhD+ erythrocytes. The pAb comprise all six human mAbs in an equal mixture at 60 ng/ml. Abs bound to RhD were detected by PE-conjugated goat anti-human IgG Abs using flow cytometry. All values were normalized to the binding of RhD-specific Abs without the presence of sera (100% binding). Data represent the mean from three independent experiments; SEM values are indicated.
Discussion
Despite efforts to minimize the immunogenicity of therapeutic and diagnostic Abs, induction of anti-drug Abs is a continuing concern as anti-drug Abs potentially interfere with the pharmacological effect of the Abs and increase the risk of infusion-related syndromes. Although insight into the factors governing the induction of anti-drug responses to Ab pharmaceuticals is accumulating, these factors are not fully understood (5, 11). Full understanding of these factors would probably require explorative immunogenicity studies in humans (11), which, however, would be complicated for ethical reasons. An alternative to human studies is to develop animal models allowing the dissection of Ab in vivo response to human Ab drugs. We have therefore developed a model based on mouse-human chimeric Abs for the study of mouse anti-human V region Ab response in vivo.
The substitution of the human C regions with mouse C regions completely eliminated the immunogenicity of the Ab C regions in vivo without major effect on the Ag-binding characteristics of the Abs. This allowed a more precise dissection of the characteristics of the Ab responses against the human V regions. In BALB/c mice, a comparison of differences in Ab responses against Abs of the same VH family primarily differing in the CDR, presumably allowed us to identify Abs specific for the Id. The BALB/c Ab response against the chimeric Abs 1AG1, 4DG1, and 6FG1 were most likely directed against the idiotypes as the elicited Abs did not cross-react with another Ab of the same VH family. These observations were based on administration of Abs in adjuvant to be able to characterize the nature of the immune response against mAbs vs pAbs in vivo. This is in contrast to administration of therapeutic Abs, where the utmost care is taken not to induce immune responses in the patients. Also the use of adjuvant might affect the elicited Ab repertoire. It would thus be interesting to assess the altitude and specificity of the immune response in mice which have received these mouse-human chimeric Abs in the absence of adjuvant in a more chronic setting. Such studies are currently ongoing.
The in vivo analysis of the pAb composition in BALB/c mice demonstrated a similar immunogenicity profile as when administrating the six Abs individually. Importantly, one of the immunogenic Abs, 3CG1, did not elicit immunogenicity when included in the mixture. This lack of immunogenicity was not due to the lower Ab contraction in the pAb composition, as 3CG1 did not become immunogenic when administrating a 6-fold higher concentration of the pAb to the mice. The reason for this epitope masking is currently not clear, but it is intriguing to speculate that this phenomenon might reduce the potential immunogenicity of some of the therapeutic Abs when comprised in a pAb composition. Also, these studies demonstrated the absence of epitope spreading (18, 19). Hence, Abs which are nonimmunogenic when administered individually did not become immunogenic as part of a pAb composition. This has important implications for the therapeutic use of pAb therapeutics as it allows a more rational assessment of the immunogenicity of a given composition based on an evaluation of the individual Ab species.
The in vivo studies also demonstrated that the immunogenicity of the Ab species in the pAb composition was highly dependent on the haplotype of the mouse. Thus, one Ab elicited immunity in NMRI mice whereas three Abs were immunogenic in BALB/c mice. None of the Abs were immunogenic in C57 mice after a single administration, however, all six Abs became immunogenic in these mice after repeated administration. This observation underscores the potential difficulty in predicting the immunogenicity of Ab therapeutics in a large outbred patient population. It can therefore be argued that, despite the effort to reduce immunogenicity of therapeutic Abs, because of the variations in Ab repertoires, TCR usage and tissue-type haplotypes of the human population, the risk of Ab immunogenicity cannot be completely removed (11). This observation also implies that to reduce immunogenicity in an outbred population, pAb compositions should be more diverse than the pAb used here, i.e., contain more Ab species harboring more different V regions.
Despite the effect of the Ag binding of mouse-human chimeric Abs, the present data may also suggest that use of this model when studying the in vivo pharmacology of human Ab lead candidates in experimental in vivo disease models during Ab drug development can reveal potential immunogenicity of particular lead candidates. Due to the absent RhD expression in mice, this model does not, however, take into account the potential effect of Ag binding on the immunogenicity of therapeutic Abs. In vivo Ag binding might influence the nature of the immune response against the chimeric Abs. Further studies to elucidate this using this model could include concomitant infusion of the chimeric Abs and RhD expressing erythrocytes or murine cell lines, or the use of RhD-transgenic mice.
The ex vivo studies of Ab binding established a clear correlation between the number of Ab species in the pAb composition being immunogenic and the effect on the pharmacological function of the composition. Thus, the binding of mAb was completely abolished by any immunogenic serum, whereas the inhibition of binding of the pAb was dependent on the number of Abs being immunogenic in the mouse strain from which the sera were derived. Therefore, under some conditions, pAb appear to be less susceptible to a decrease in the pharmacological effect mediated by anti-drug Abs. Together with the pharmacological advantages of pAb therapeutics (20, 21, 22, 23, 24, 25, 26), the potential decrease in immunogenicity and resulting susceptibility to blocking Abs make recombinant human pAb attractive candidates for development of new treatment modalities in a number of human disorders.
Acknowledgments
We thank Drs. Frank Hinnerfeldt and Kim Warming, Aalborg Hospital (Aalborg, Denmark), for making RhD+ blood available, Prof. Mogens Claesson, University of Copenhagen (Copenhagen, Denmark), for assistance with animal experiments, and Jesper Kastrup for help with FACS experiments. The expert technical assistance of Barbara T. Andersen, Elisabeth V. Andersen, Hanne Wagner, Maria Schmidt, and Yvonne B. Larsen is gratefully appreciated.
Disclosures
J. L. Klitgaard, V. W. Coljee, P. S. Andersen, L. K. Rasmussen, L. S. Nielsen, J. S. Haurum, and S. Bregenholt are either current or previous employees of Symphogen and hold options, warrants, or stock in the company.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 Current address: Department of Physics, Harvard University, Cambridge, MA, 02138.
↵2 Address correspondence and reprint requests to Dr. Søren Bregenholt, Symphogen, Elektrovej 375, DK-2800 Lyngby, Denmark. E-mail address: sb{at}symphogen.com
↵3 Abbreviations used in this paper: pAb, polyclonal Ab; RhD, rhesus D; RT, room temperature; MFI, mean fluorescence intensity; CHO, Chinese hamster ovary.
- Received January 5, 2006.
- Accepted June 29, 2006.
- Copyright © 2006 by The American Association of Immunologists