Comparative Evaluation of Capsular Polysaccharide-Specific IgM and IgG Antibodies and F(ab′)₂ and Fab Fragments as Delivery Vehicles for Radioimmunotherapy of Fungal Infection

Current antimicrobial therapies are relatively inefficient in immunosuppressed patients, such as cancer patients or HIV-infected individuals. This problem combined with an increasing prevalence of diseases caused by highly resistant microorganisms creates an urgent need for new approaches to treatment of infectious diseases. Passive antibody therapy is a potentially useful therapeutic and preventive strategy against a variety of infectious diseases (1). The specificity of the antigen-antibody interaction provides an attractive option for delivering microbicidal agents to sites of infection. Recently, we showed the feasibility of radioimmunotherapy as an anti-infective therapy by treating murine cryptococcosis with a monoclonal antibody (mAb) to the Cryptococcus neoformans (CN) capsular glucuronoxylomannan labeled with 213-bismuth (²¹³Bi) or 188-rhenium (¹⁸⁸Re; refs. 2, 3). Subsequently, we showed the applicability of radioimmunotherapy to other fungal and bacterial infections (4, 5). Radioimmunotherapy also proved to be effective both in vitro and in vivo against HIV-1 chronically or acutely infected human cells (6).

The use of radioimmunotherapy for tumor therapy has been studied for 2 decades and there is a wealth of knowledge derived from both laboratory and clinical studies about the delivery of radionuclides by mAbs and their fragments, both murine and humanized, to cancer cells, surrounding tissue, and major organs (7, 8). In contrast, the application of radioimmunotherapy to infectious diseases is in its infancy, with experience limited to animal models. Most of the information on the issues associated with the application of mAbs as delivery vehicles for radioimmunotherapy of infectious diseases has been inferred from the literature on radioimmunotherapy of cancer as well as from the reports on the use of microorganism-specific mAbs for scintigraphic imaging of infection. Although some of this information is applicable to the development of radioimmunotherapy as an anti-infective modality, fundamental differences between neoplastic and infectious processes necessitate different approaches and considerations.

As a part of an on-going effort to translate the radioimmunotherapy of infection to the clinic, we have carried out a comparative evaluation of capsular polysaccharide-specific antibodies with IgG1 and IgM isotypes as well as of F(ab′)₂ and Fab fragments as potential delivery vehicles for radioimmunotherapy of the human pathogenic fungus CN.

Cryptococcus neoformans. CN strains H99 (C. neoformans var. grubii) and 24067 (C. neoformans var. neoformans) were obtained from Dr. John Perfect (Duke University, Durham, NC) and the American Type Culture Collection, respectively. These two strains were chosen because of their clinical relevance as representative of the most common serotypes encountered in clinical practice in temperate regions. The cells were grown as in ref. 9.

CN-specific mAbs, fragments, and radiolabeling of mAb. Glucuronoxylomannan-binding murine mAb 18B7 (IgG1) and 13F1 (IgM) were produced as in refs. 10, 11. Isotype-matching control mAb MOPC21 (IgG1) and TEPC (IgM) were acquired from MP Biochemicals. F(ab′)₂ and Fab fragments of 18B7 mAb were obtained by digestion using immobilized ficin from a commercial kit [ImmunoPure IgG1 Fab and F(ab′)₂ preparation kit, Pierce]. Fab fragments were obtained by digestion of 18B7 for 3 h at 37°C on a ficin column in the presence of 20 mg/mL cysteine. F(ab′)₂ fragments were prepared by digestion of 18B7 on the ficin column for 24 h at 37°C in the presence of 2 mg/mL cysteine. Following the digestions, the fragments were eluted from the column and purified by passage over a column of immobilized protein A to remove any components that contained the Fc portion of the antibody. The molecular weight of the obtained fragments was analyzed by SDS-PAGE and fast protein liquid chromatography. The protein concentration was determined by the method of Lowry et al. (12) using a kit from Bio-Rad. ¹¹¹In was purchased from Isotex; ¹⁸⁸Re was eluted from ¹⁸⁸Re/¹⁸⁸W generator (Oak Ridge National Laboratory). Whole IgG and IgM mAbs were radiolabeled with ¹⁸⁸Re “directly ” as described (2). For radiolabeling of whole 18B7 and of its F(ab′)₂ and Fab fragments with ¹¹¹In, the antibodies were incubated overnight at room temperature in bicarbonate buffer at pH 8.5 with 1.5 molar excess of bifunctional chelating agent N-[2-amino-3-(p-isothiocyanatophenyl)propy1]-trans-cyclohexane-1,2-diamine-N,N′,N′′,N‴,N‴-pentaacetic acid (CHXA"). Before incubation with CHXA", the thiol groups on Fab were blocked by the addition of 22 mmol/L iodoacetamide. The average number of CHXA" molecules per antibody was determined to be 1.0 to 1.5 by the Yttrium-Arsenazo III spectrophotometric method (13). The antibodies were purified from excess CHXA" on Centricon microconcentrators and radiolabeled with ¹¹¹In as in ref. 2.

Binding of mAbs 18B7 and 13F1 to H99 and 24067 cells.¹⁸⁸Re-18B7 was added in different amounts to 24067 or H99 cells (2 × 10⁶ per tube). After incubation for 1 h at 37°C, the tubes were counted in a gamma counter, the cells were collected by centrifugation, and the pellets were counted again. Scatchard analysis was used to compute the mAb binding constant and binding sites per cell as in ref. 14. In a separate series of experiments, the kinetics of mAbs 18B7 and 13F1 binding to 24067 cells was assessed. For this purpose, aliquots of cell suspension were removed after 0-, 10-, 20-, and 30-min and 1-h incubation time and centrifuged to separate the cell pellet from the supernatant, and the cell pellets were counted in a gamma counter.

In vitro activity of ¹⁸⁸Re-labeled 18B7 and 13F1 against CN. CN (10⁵ cells) of 24067 and H99 strains was placed in microcentrifuge tubes in 0.1 mL PBS. Serially diluted ¹⁸⁸Re-18B7 or ¹⁸⁸Re-13F1 in 0.5 mL PBS was added to obtain the desired concentrations of radioactivity. After 60 min of incubation at 37°C, CN cells were collected by centrifugation, washed with PBS to remove nonbound radioactivity, suspended in 1 mL PBS, and incubated at 4°C for 48 h on a rocker. The cells did not divide during the incubation and maintenance periods because they were maintained in PBS, which lacks nutrients. Following the incubation, 10³ cells were removed from each tube, diluted with PBS, and plated to determine colony-forming units (1 colony = 1 colony-forming unit). Experiments were done in duplicate.

Cellular dosimetry. To compare the doses delivered by ¹⁸⁸Re-labeled 18B7 and 13F1 to CN cells during radioimmunotherapy experiments, we used algorithm that we developed for radioimmunotherapy of fungal cells (4). Briefly, in this approach, the mean absorbed dose to the cell from cellular radioactivity is given by the following equation:

where à is the cellular cumulated activity, bj is the branching ratio of the jth radionuclide in the decay series, and Sj(C←CS) is the cellular S value (absorbed dose to the cell per unit cumulated activity) for the jth radionuclide localized on the cell surface (CS) of the cell (C). The cumulated activity à can be written as à = Ã_I + Ã_M + Ã_CF, where Ã_I, Ã_M, and Ã_CF are the cellular cumulated activities during the periods of incubation for cellular uptake of radioactivity, maintenance at 4°C, and colony formation, respectively.

Biodistribution and scintigraphic imaging of ¹⁸⁸Re- and ¹¹¹In-labeled mAbs and fragments in A/JCr mice systemically infected with CN. All animal experiments were carried out in accordance with the guidelines of the Albert Einstein College of Medicine Institute for Animal Studies. In the first series of experiments, we did biodistribution measurements of ¹⁸⁸Re-18B7 IgG1 and ¹⁸⁸Re-13F1 IgM and of their respective isotype-matching controls ¹⁸⁸Re-MOPC21 and ¹⁸⁸Re-TEPC. Twelve groups of four A/JCr female mice (National Cancer Institute, Frederick, MD) were infected i.v. with 10⁵ 24067 CN. Approximately 24 h after infection, each mouse received i.p. 20 μCi (5 μg) of the following mAbs: groups 1 to 3, ¹⁸⁸Re-18B7; groups 4 to 6, ¹⁸⁸Re-MOPC21; groups 7 to 9, ¹⁸⁸Re-13F1; and groups 10 to 12, ¹⁸⁸Re-TEPC. At 24, 48, and 72 h after injection of radioactive mAb for ¹⁸⁸Re-18B7 and ¹⁸⁸Re-MOPC21 mAbs and at 4, 24, and 48 h for ¹⁸⁸Re-13F1 and ¹⁸⁸Re-TEPC mAbs, the groups of mice were sequentially humanely sacrificed, and their major organs were removed, weighed, and counted in a gamma counter.

In the second series of experiments, we compared the biodistribution of ¹¹¹In-labeled 18B7 mAb and its F(ab′)₂ and Fab fragments in the same animal model of CN infection. Approximately 24 h after infection of six groups of three A/JCr female mice, each mouse received i.p. 20 μCi (5 μg) of the following mAbs: groups 1 to 2, ¹¹¹In-18B7; groups 3 to 4, ¹¹¹In-F(ab′)₂; and groups 5 to 6, ¹¹¹In-Fab. Twenty-four and 48 h after injection of radioactive mAb, the mice were anesthetized with isoflurane and scintigraphically imaged on a gamma camera equipped with pinhole collimator (Siemens) and ICON image processing software. After imaging, the mice were humanely sacrificed and biodistribution was done as above.

Statistical analysis. The nonparametric Wilcoxon rank sum test was used to compare uptake of radioactivity in the organs in biodistribution studies. Student's t test for unpaired data was used to analyze differences in the number of colony-forming units between differently treated groups during in vitro radioimmunotherapy studies. Differences were considered statistically significant when P values were <0.05. The Scatchard binding and kinetics data were analyzed by linear regression (Prism software, GraphPad).

Binding of ¹⁸⁸Re-18B7 and ¹⁸⁸Re-13F1 to different strains of CN. We measured pronounced differences in the abilities of ¹⁸⁸Re-18B7 IgG1 and ¹⁸⁸Re-13F1 IgM to bind to both 24067 and H99 CN strains (Fig. 1A and B). The binding constants for ¹⁸⁸Re-13F1 to 24067 and H99 CN strains were approximately 31 and 7 times lower, respectively, than those for ¹⁸⁸Re-18B7 (Table 1). When the kinetics of ¹⁸⁸Re-18B7 and ¹⁸⁸Re-13F1 mAbs binding to 24067 cells was evaluated, the cellular uptake of radioactivity was linear in time up to 30 min for both mAbs (Fig. 1C), at which time saturation was reached. Linear regression analysis of the kinetics data yielded k constants, min⁻¹, of 8.5 × 10⁻⁸ and 3.12 × 10⁻⁹ for ¹⁸⁸Re-18B7 and ¹⁸⁸Re-13F1, respectively. These values were subsequently used in the cellular dosimetry calculations.

In vitro killing of 24067 and H99 CN cells by ¹⁸⁸Re-18B7 and ¹⁸⁸Re-13F1 and cellular dosimetry results. Treatment of 24067 and H99 CN cells with ¹⁸⁸Re-18B7 resulted in substantial killing of the cells; approximately 90% and 30%, respectively, were killed by the 1 μCi dose (Fig. 2), which delivered 55 cGy to the cells according to the cellular dosimetry calculations. In contrast, no killing with ¹⁸⁸Re-13F1 of either 24067 or H99 CN was observed in the range of activities studied (0-4.5 μCi), with the 1 μCi dose delivering 2 cGy to the treated cells.

Biodistribution of ¹⁸⁸Re-18B7 and ¹⁸⁸Re-13F1 in A/JCr mice systemically infected with 24067 CN. The animal model used in biodistribution studies used i.v. infection of A/JCr mice with CN. This model is the same as was used in the radioimmunotherapy survival experiments (2) and is suitable for the evaluation of experimental therapies because the dissemination of infection leads to rapid death in untreated mice. Figure 3A-C shows the percentage injected dose per gram (% ID/g) in CN-infected A/JCr mice injected with ¹⁸⁸Re-18B7 or isotype-matching control ¹⁸⁸Re-MOPC21. At all three time intervals (24, 48, and 72 h), the uptake of radioactivity in the liver and spleen of infected mice given ¹⁸⁸Re-18B7 was higher (P < 0.05) than in control group, which might be the consequence of the deposition of antibody-antigen complex (15). In contrast, the activity in the blood of mice given ¹⁸⁸Re-MOPC21 was significantly higher than in mice given ¹⁸⁸Re-18B7 (P < 0.05). For ¹⁸⁸Re-18B7, the highest uptake in the lungs was at 24 h (1.2% ID/g), decreasing to 0.8% and 0.4% ID/g at 48 and 72 h, respectively. The uptake in the lungs in the ¹⁸⁸Re-18B7 group was twice that of ¹⁸⁸Re-MOPC21 injected mice at 24 and 48 h and equalized at 72 h, which can be attributed to the shedding of capsular polysaccharide antigen together with the bound antibody. ¹⁸⁸Re-MOPC21 showed elevated kidney uptake at 48 h, which by 72 h became almost equal to that measured in the ¹⁸⁸Re-18B7 group. This biodistribution pattern is unlikely to be caused by instability of ¹⁸⁸Re radiolabel, as the stomach uptake, indicative of the presence of free ¹⁸⁸Re, was low at all times studied for all groups. The differences in the blood clearance between capsule-specific 18B7 IgG and the isotype-matching control are almost certainly due to the presence of polysaccharide antigen in the blood, which results in fast clearance of the specific antibody from the blood as a result of antigen-antibody complex formation and clearance by the reticuloendothelial system. This is confirmed by our previous studies that compared the distribution of specific 18B7 in infection-free and infected mice to that of control MOPC21 in infected mice (2). In those studies, the blood clearance of 18B7 in infection-free animals was closer to that of MOPC21 than to 18B7 in infected mice.

The biodistribution of ¹⁸⁸Re-13F1 IgM and its isotype-matching control was done in a similar manner except that the first-time measurement was done at 4 h. The rationale for evaluating earlier time intervals for IgM than for IgG1 is that in mice IgM and IgG have serum half-life of 2 and 8 days, respectively (16). The biodistribution patterns shown in Fig. 4A-C showed that ¹⁸⁸Re-13F1 was unable to bind specifically to CN cells in vivo: at all time points, the uptake of ¹⁸⁸Re-13F1 in the lungs and the brain (the target organs for disseminated CN infection) was significantly (P < 0.05) lower for ¹⁸⁸Re-13F1 than for the control ¹⁸⁸Re-TEPC.

Biodistribution of ¹¹¹In-labeled 18B7 and its F(ab′)₂ and Fab fragments. When comparing the in vivo targeting of CN infection with F(ab′)₂ and Fab fragments versus whole antibody, the criteria for evaluating those agents would be uptake in infected and noninfected organs. In vitro Fab fragments manifest weaker binding to CN cells than whole mAbs or F(ab′)₂ because of lower valence, but in vivo predictions are more complicated because there are two kinds of polysaccharide antigen: on the CN cells in the infected organs and soluble polysaccharide in the blood. It is conceivable that Fab fragments bind less strongly to the soluble polysaccharide in the blood because of the lower valence yet reach infected cells in the target organs faster than whole mAb or F(ab′)₂ because of their smaller size. Thus, the comparative biodistribution studies were warranted. We carried out biodistribution and imaging studies with ¹¹¹In-labeled 18B7 and its F(ab′)₂ and Fab fragments (Figs. 5 and 6). At 24 h, the uptake in the lungs was 4.7% ID/g for whole ¹¹¹In-18B7, 1.4% ID/g for F(ab′)₂ fragments, and 0.2% ID/g for Fab fragments (P = 0.03). At 48 h, this trend changed and lung uptake of ¹¹¹In-F(ab′)₂ fragments reached 2.1% ID/g, whereas uptake of whole ¹¹¹In-18B7 decreased to 0.2% ID/g. This could be explained by the fact that the blood uptake of ¹¹¹In-F(ab′)₂ remained practically constant at ∼1% ID/g, allowing for continuous uptake of ¹¹¹In-F(ab′)₂ into the infected lungs, whereas the blood uptake of the whole ¹¹¹In-18B7 fell from 2.5% ID/g at 24 h to 0.3% ID/g. Both lung and blood uptake of Fab remained low (0.2-0.3% ID/g) from 24 to 48 h. Liver uptake was several-fold higher at both 24 and 48 h for ¹¹¹In-18B7 (22% and 35% ID/g, respectively) and ¹¹¹In-F(ab′)₂ (20% and 10% ID/g) when compared with ¹¹¹In-Fab (2.5% and 2.4% ID/g), respectively. At both 24 and 48 h, the kidney uptake of F(ab′)₂ (12% and 12.5% ID/g) and Fab (22% and 11% ID/g) fragments was significantly higher (P < 0.05), due to their smaller size, when compared with whole 18B7 mAb (1% and 0.8% ID/g).

To identify the best delivery vehicle for radioimmunotherapy of the human pathogenic fungus CN, we evaluated the pharmacokinetic and binding characteristics of capsular polysaccharide-specific antibodies with IgG1 and IgM isotypes and IgG1-derived F(ab′)₂ and Fab fragments.

One of the advantages of radioimmunotherapy of infection (17, 18) is that the antibodies used as delivery vehicles for radionuclides do not need to be protective as required in naked antibody therapy. In fact, the only requirement for an antibody to be a suitable reagent for radioimmunotherapy is that it binds to the targeted microbe and delivers the radionuclide to the vicinity of the microbial cell. As the binding of the organism-specific antibodies to the microbial cells in the host is the result of a complex interplay of antibody affinity, its molecular weight, and circulation time in the blood, we hypothesized that 13F1 IgM would have certain advantages as a potential carrier for the radioimmunotherapy of CN infection because of its higher valence, which might translate into higher affinity, and its shorter plasma half-life than IgG1, which will make it clear faster from the blood.

However, the binding experiments showed that the affinity constants of 13F1 for 24067 and H99 strains of CN were 31 and 7 times lower, respectively, than for 18B7. This lower binding of 13F1 IgM translated into its inability to kill CN cells during in vitro radioimmunotherapy experiments. In fact, the dose delivered by ¹⁸⁸Re-13F1 to CN, 2 cGy, was 28 times lower than dose delivered by the same activity of ¹⁸⁸Re-18B7. These results confirm our previous observations on the correlation between the efficacy of cell killing with the binding capacity of the mAb to the capsule (19). When given to CN-infected mice, ¹⁸⁸Re-18B7 showed specific binding to the target organs for CN infection, whereas ¹⁸⁸Re-13F1 was unable to bind specifically. High affinity of the mAb is of particular importance for the treatment of established cryptococcosis because patients with this disease often have large amounts of circulating capsular polysaccharide (cryptococcal antigen) that could interfere with radioimmunotherapy by binding radiolabeled mAb. However, in our previous studies, we found that radiolabeled 18B7 was therapeutic even in mice with serum antigen levels of ∼1.9 μg/mL as determined by Latex-Crypto antigen detection system (2), which is comparable with antigen levels found in patients. The most likely explanation for this effect is that the antibody has higher affinity to capsule and tissue polysaccharide than to soluble polysaccharide. Support for this explanation comes from the competition experiments whereby very large amounts of soluble polysaccharide were needed to inhibit antibody binding to a small amount of polystyrene-absorbed polysaccharide by ELISA (20). The concentration of polysaccharide in the CN capsule is ∼6 mg/mL (21), which is several thousand times higher than the concentration of soluble polysaccharide in the blood, thus providing much higher antigen density for better binding. The binding of specific mAb to the target organ is dependent on the dose of this mAb: pretreatment of infected mice with 1 mg unlabeled 18B7 before administration of 30 μg radiolabeled 18B7 reduced uptake in the target organs 2-fold (2). The problem of high levels of circulating antigen in patients can be counteracted by using a pretargeting approach with administration of unlabeled bispecific antibodies followed by radiolabeled hapten (22).

We investigated further the influence of affinity, molecular weight, and circulation time in the blood on the biodistribution patterns of mAbs in systemic CN infection by doing comparative biodistribution experiments of ¹¹¹In-labeled whole 18B7 and its F(ab′)₂ and Fab fragments. As in the case of 13F1 IgM, which proved unsuitable for in vivo delivery of radionuclides because of its low affinity for the target antigen, the importance of high affinity for the target antigen again became obvious for Fab fragments. Although whole 18B7 and its F(ab′)₂ showed comparable uptake in the lungs, which was higher for whole ¹¹¹In-18B7 at 24 h after administration and for ¹¹¹In-F(ab′)₂ at 48 h, the uptake of ¹¹¹In-Fab in the lungs was very low at both time intervals. In cancer radioimmunotherapy, Fab fragments show significantly lower % ID/g than intact mAb despite having a smaller molecular weight as a consequence of lower affinity resulting from losing the contribution of polyvalence to avidity. For example, at 48 h after injection, the tumor % ID/g for intact mAb CC49, which binds to TAG72 antigen in human colon carcinoma xenografts, and its F(ab')2 and Fab fragments, is 18, 13.9, and 2.7, respectively (23). The elevated kidney uptake of F(ab′)₂ and Fab, which is always observed in cancer radioimmunotherapy due to the lower molecular weight of the fragments in comparison with the whole mAb, also manifested itself in our study. However, it has been shown in preclinical studies and in patients that administration of positively charged amino acids effectively reduces kidney uptake of the fragments (24, 25).

It is known from cancer radioimmunotherapy (8) that the same mAb labeled with different radionuclides often displays significant differences in biodistribution. When compared with each other, ¹⁸⁸Re- and ¹¹¹In-labeled whole 18B7 showed comparable uptake in the infected lungs, which was highest at 24 h. Significant differences were observed in blood clearance, with ¹⁸⁸Re-18B7 clearing from the blood approximately five times faster than ¹¹¹In-18B7, and in liver uptake. The liver uptake of ¹¹¹In-labeled 18B7 was very high (35% ID/g at 48 h) compared with <2% ID/g for ¹⁸⁸Re-labeled 18B7. This can be explained by the fact that ¹¹¹In and other trivalent metal radiolabels tend to accumulate in the liver as a result of mAb catabolism, whereas ¹⁸⁸Re, which is attached to the mAbs directly via SH groups, is converted by mAb catabolism into perrhenate anion, which is then rapidly excreted through the kidneys. Thus, when the whole antibody is considered as a delivery vehicle for radioimmunotherapy of fungal infection, ¹⁸⁸Re seems to be a safer radioisotope than trivalent radiometals such as ⁹⁰Y or ¹⁷⁷Lu, which will likely show distribution patterns comparable with those of ¹¹¹In-labeled 18B7. Alternatively, whole antibody may be labeled with the short-lived (half-life, 46 min) α-emitter ²¹³Bi, which has already shown its efficacy and safety in radioimmunotherapy of experimental cryptococcal infection (2, 3).

When whole ¹¹¹In-18B7 is compared with its ¹¹¹In-F(ab′)₂ fragments, the fragments show more stable uptake in the infected lungs and significantly less liver uptake, although even for the ¹¹¹In-F(ab′)₂ its 24- and 48-h liver uptake at 20% and 10% ID/g was approximately two and five times higher, respectively, than for ¹⁸⁸Re-labeled whole 18B7. It remains to be shown experimentally if labeling of F(ab′)₂ fragments with ¹⁸⁸Re will result in stable uptake in the infected organs with simultaneous reduction of the liver uptake. Although “direct” labeling of F(ab′)₂ fragments with ¹⁸⁸Re is impossible, chelating agents such as MAG3 could be of potential use. Labeling of F(ab′)₂ fragments with ¹⁸⁸Re might also offer decreased kidney toxicity because of its relatively short-half life (16.9 h) and nonresidualizing nature. ¹⁸⁸Re seems to be a safer radionuclide in terms of long-term kidney toxicity than ⁹⁰Y (26, 27).

In conclusion, comparative evaluation of two antibodies to cryptococcal polysaccharide with IgG and IgM isotypes and of F(ab′)₂ and Fab fragments of IgG mAb as potential delivery vehicles for radioimmunotherapy of cryptococcal infection strongly suggests that high affinity for the target antigen is a prerequisite for successful targeting of infection in vivo and that prior in vitro determination of affinity may predict the in vivo behavior of the mAbs. Based on this criteria, whole IgG mAb and its F(ab′)₂ fragments proved to be better choices for delivery of radionuclides than the IgM mAb used in this work or than Fab fragments. Whole IgG labeled with the nonresidualizing radionuclide ¹⁸⁸Re also showed less uptake in the liver and kidneys than ¹¹¹In-labeled whole mAb or F(ab′)₂ fragments, which is an important safety consideration for radioimmunotherapy of infection.

We thank Dr. M.W. Brechbiel (National Cancer Institute, NIH, Bethesda, MD) for kindly providing CHXA-DTPA ligand.