The application of STEP-technology® for particle and protein dispersion detection studies in biopharmaceutical research.

Particle detection and analysis techniques are essential in biopharmaceutical industries to evaluate the quality of various parenteral formulations regarding product safety, product quality and to meet the regulations set by the authority agencies. Several particle analysis systems are available on the market, but for the operator, it is quite challenging to identify the suitable method to analyze the sample. At the same time these techniques are the basis to gain a better understanding in biophysical processes, e.g. protein interaction and aggregation processes. The STEP-Technology® (Space and Time resolved Extinction Profiles), as used in the analytical photocentrifuge LUMiSizer®, has been shown to be an effective and promising technique to investigate particle suspensions and emulsions in various fields. In this study, we evaluated the potentials and limitations of this technique for biopharmaceutical model samples. For a first experimental approach, we measured silica and polystyrene (PS) particle standard suspensions with given particle density and refractive index (RI). The concluding evaluation was performed using a variety of relevant data sets to demonstrate the significant influences of the particle density for the final particle size distribution (PSD). The most challenging property required for successful detection, turbidity, was stated and limits have been set based on the depicted absorbance value at 320 nm (A320 values). Furthermore, we produced chemically cross-linked protein particle suspensions to model physically "stable" protein aggregates. These results of LUMiSizer® analysis have been compared to the orthogonal methods of nanoparticle tracking analysis (NTA), dynamic light scattering (DLS) and micro-flow imaging (MFI). Sedimentation velocity distributions showed similar tendencies, but the PSDs and absolute size values could not be obtained. In conclusion, we could demonstrate some applications as well as limitations of this technique for biopharmaceutical samples. In comparison to orthogonal methods this technique is a great complementary approach if particle data e.g. density or refractive index can be determined.

Systematic investigation of the effect of lyophilizate collapse on pharmaceutically relevant proteins III: collapse during storage at elevated temperatures.

This study investigates the effect of lyophilizate collapse on the stability of pharmaceutical proteins. Recently, it was shown that collapse during freeze-drying has no major negative impact on protein stability during storage at elevated temperatures when compared to non-collapsed cakes [1,2]. In this part of the study, lyophilizates that collapsed during the freeze-drying process were compared to cakes that were initially non-collapsed but collapsed during subsequent storage under accelerated stress conditions. Collapsed and non-collapsed lyophilizates of identical formulation and comparable residual moisture levels, containing a monoclonal IgG antibody, were stored at 40 °C and 50 °C for up to 3 months. Protein stability was monitored using a comprehensive set of analytical techniques assessing the formation of soluble and insoluble aggregates as well as protein conformation. The properties of the freeze-dried cake, namely the glass transition temperature, excipient crystallinity, sucrose degradation, reconstitution behavior, and the residual moisture content, were analyzed as well. The incorporated protein was significantly better stabilized in cakes that collapsed during the freeze-drying process when compared to lyophilizates that collapsed during subsequent storage. This effect can be related to the onset of crystallization and hydrolysis of the stabilizer and non-enzymatic browning.

Structural prerequisites for endotoxic activity in the Limulus test as compared to cytokine production in mononuclear cells.

The structural prerequisites for lipopolysaccharide (LPS) and its partial structures for the activation of the Limulus clotting cascade (Limulus amebocyte lysate [LAL] test) are described and compared with the corresponding requirements for the activation of human immune cells such as mononuclear cells. A necessary, but not sufficient, structural motif for this is the presence of the 4(')-phosphate-diglucosamine backbone recognition structure ('epitope') in lipid A. High activity is only expressed by assemblies of endotoxins, but this is largely independent of the type of supramolecular aggregate structure. A particular conformation of the epitope within the lipid A assembly must be present, which is influenced by addition of further saccharide units to the lipid A moiety, but also reacts slightly to the acylation pattern. In contrast, the cytokine production of human immune cells induced by LPS sensitively depends on the type of its aggregate structure. In the case of a hexa-acylated bisphosphorylated lipid A structure, high activity is only observed with cubic inverted aggregates. Furthermore, addition of antimicrobial agents (such as polymyxin B) leads to a nearly complete inhibition of cytokine production, whereas the reduction in the Limulus assay is much lower. These data are important since a reliable determination of endotoxin concentrations, in particular with respect to its ability to elicit severe infections, is of high interest.

Systematic investigation of the effect of lyophilizate collapse on pharmaceutically relevant proteins I: stability after freeze-drying.

The objective of this work was to investigate the effect of cake collapse during freeze-drying on the stability of protein lyophilizates containing a monoclonal IgG(1)-antibody or a second pharmaceutically relevant protein, referred to as PA01. In addition, L-lactic dehydrogenase was investigated because of its well-documented sensitivity towards freeze-drying stresses. Collapse was induced by two different means. First, by varying the ratio of the crystalline bulking agent mannitol to the amorphous stabilizer sucrose, different extents of collapsed cakes were generated. Second, formulations were freeze-dried using an aggressive collapse-cycle and a conventional freeze-drying protocol and collapsed and noncollapsed cakes of identical formulation were produced. Lyophilizates were analyzed using a comprehensive set of analytical techniques to monitor protein stability in terms of formation of soluble and insoluble aggregates, the biological activity and the conformational stability. The stability of excipients, namely the glass transition temperature, crystallinity, reconstitution behavior, and the residual moisture content was analyzed as well. In addition, the extent of collapse was quantified using the decrease of the specific surface area (SSA). Collapsed cakes had comparable residual moisture levels to noncollapsed lyophilizates. Reconstitution times were not increased. Protein stability was not relevantly different between collapsed and noncollapsed cakes.

Phase diagrams of monoacylated amide-linked disaccharide glycolipids.

A series of monoacylated glycolipids with even-numbered acyl chain lengths ranging from saturated C11 to C15 and an unsaturated C17:1 fatty acid connected by an amide in linkage to the disaccharide head groups maltose, melibiose and lactose were synthesized. The structural polymorphism of the glycolipids was investigated using Fourier-transform infrared spectroscopy and differential scanning calorimetry for the detection of the gel to liquid-crystalline acyl chain melting behaviour and small-angle X-ray scattering for the elucidation of the physical structure of the lipid aggregates. Also, the phase morphology was studied by polarizing microscopy in contact preparations. The data clearly show the existence of uni- and multilamellar structures. Although only one acyl chain is present, there is no evidence for the existence of micelles - of spherical or of cylindrical (H(I)) type - or of interdigitated phases. The preference for lamellar phases seems to be correlated with the intrinsic high conformational order of the amide linkage of these compounds which inhibits the formation of highly curved structures.

Stabilization of IgG1 in spray-dried powders for inhalation.

The protein stabilizing capabilities of spray-dried IgG1/mannitol formulations were evaluated. The storage stability was tested at different residual moisture levels prepared by vacuum-drying or equilibration prior to storage. Vacuum-drying at 32 degrees C/0.1mbar for 24h reduced the moisture level below 1%, constituting an optimal basis for improved storage stability. The crystalline IgG1/mannitol powders with a weight ratio of 20/80 up to 40/60 failed to prevent the antibody aggregation as assessed by size exclusion chromatography during storage. Ratios of 60/40 up to 80/20 IgG1/mannitol provided superior stability of the antibody and the powders could be produced with high yields. The lower the residual moisture, the better was the stabilizing capability. An amount of 20% mannitol provided the best stabilization. Storage stability of 60/40, 70/30, and 80/20 IgG1/mannitol formulations over one year was adequate at 2-8 degrees C and 25 degrees C. Closed storage (sealed in vials) at 40 degrees C/75% RH and open storage at 25 degrees C/60% RH revealed that the stability still required optimization. The lower the protein content, the better was the powder flowability. The aerodynamic properties of powders spray-dried with 10% solids content were inadequate, as the particle size ranged between 5.1 and 7.2 microm and the fine particle fraction accounted for only 4-11%. Reduction of the solids content to 2.5% did improve the aerodynamic properties as the mass mean aerodynamic diameter was reduced to 3.6 microm and the fine particle fraction was increased to about 14%. The reduction of the solids content did not influence the storage stability significantly. Also spray-drying at higher temperatures had no significant impact on the storage stability, despite a higher tendency to form amorphous systems. In order to improve the storage stability and to maintain the good flowability of 70/30 IgG1/mannitol powder or to keep the storage stability but to improve the flowability of the 80/20 IgG1/mannitol powder, mannitol was partially substituted by a second excipient such as trehalose, sucrose, glycine, lactose, lactosucrose, or dextran 1. Differences in the stabilizing capability were noticeable upon closed storage at 40 degrees C/75% RH and open powder storage. Protein stabilization was improved by the addition of glycine but trehalose and sucrose were most effective in preventing aggregation, which can be primarily attributed to the water replacement properties of the sugars. The addition of another excipient, isoleucine had positive effects on both flowability and protein stability.

Composition dependence of vesicle morphology and mixing properties in a bacterial model membrane system.

We have determined the mixing properties and lamellar organization of bacterial membrane mimetics composed of 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (POPE) and -phosphatidylglycerol (POPG) at various molar ratios applying differential scanning calorimetry, small and wide-angle X-ray scattering, as well as optical phase contrast microscopy. Combining the experimental thermodynamic data with a simulation of the liquidus and solidus lines, we were able to construct a phase diagram. Using this approach, we find that the lipids mix in all phases non-ideally in the thermodynamic sense. As expected, pure POPE assembles into multilamellar and pure POPG into unilamellar vesicles, respectively, which are stable within the studied temperature range. In contrast, mixtures of the two components form oligolamellar vesicles consisting of about three to five bilayers. The layers within these oligolamellar liposomes are positionally correlated within the gel phase, but become uncorrelated within the fluid phase exhibiting freely fluctuating bilayers, while the vesicles as a whole remain intact and do not break up into unilamellar forms. X-ray, as well as DSC data, respectively, reveal a miscibility gap due to a lateral phase segregation at POPG concentrations above about 70 mol%, similar to previously reported data on mixtures composed of disaturated PEs and PGs. Hence, the existence of a region of immiscibility is a general feature of PE/PG mixtures and the mixing properties are dominated by PE/PG headgroup interactions, but are largely independent of the composition of the hydrocarbon chains. This is in accordance with a recent theoretical prediction.

Structural polymorphism and endotoxic activity of synthetic phospholipid-like amphiphiles.

The physicochemical characteristics and in vitro biological activity of various synthetic hexaacyl phospholipid dimers were compared with the respective behavior of bacterial endotoxins (lipopolysaccharide, LPS). The structural variations of the synthetic amphiphiles include different stereochemical (R,S) configurations about their ester- and amide-linkages for the acyl chains and differences in the length of the serine backbone spacer. The temperature of the gel to liquid crystalline phase transition of the acyl chains (T(c)) lies between 10 and 15 degrees C for the compounds with the shortest backbone and decreases rapidly for the compounds with longer backbones. The phase transition enthalpies (8-16 kJ x mol(-1)) are considerably lower than those of lipid A from hexaacyl endotoxins (28-35 kJ x mol(-1)). In contrast, the dependence of T(c) on Mg(2+) and water content shows a behavior typical for endotoxins: a significant increase with increasing Mg(2+) and decreasing water concentrations. The aggregate structure is sensitively dependent not only on the length of the backbone spacer but also on the different stereochemical variations. It can be directly correlated with the biological activity of the compounds. Thus, as with natural lipid A, the capacity to induce cytokine production in mononuclear cells is directly related to the affinity to form nonlamellar cubic or inverted hexagonal H(II) aggregate structures. Together with the data on the transport and intercalation of the dimers into phospholipid liposomes mediated by the lipopolysaccharide-binding protein (LBP), our conformational concept of endotoxicity and cell activation can be applied to these non-LPS structures: endotoxically active compounds incorporate into membranes of immune cells and cause conformational changes at the site of signaling proteins such as Toll-like receptors or K(+)-channels due to their conical molecular shape.

Packing characteristics of a model system mimicking cytoplasmic bacterial membranes.

The phase diagram of fully hydrated mixtures of dipalmitoylphosphatidylethanolamine and -phosphatidylglycerol was constructed and the coexistence lines of the solidus and liquidus curve calculated based on regular solution theory using two nonideality parameters for each of the phase to account for nonideal and nonsymmetric mixing. Both lipids show nonideal miscibility in the liquid-crystalline phase, while a region of immiscibility exists in the lamellar-gel phase between the mole fraction x(DPPE)=0.05-0.4. Two lines of three-phase coexistence around 35 and 40 degrees C reflects the presence of lipid domains predominantly composed of phosphatidylglycerol as well as of the mixed lipid system. This is reflected in the positive nonideality parameters of the gel phase obtained from the simulation of the phase diagram. Moreover, segregation of pure phosphatidylethanolamine domains was detected in mixtures x(DPPE)>0.9, which formed multilamellar liposomes, while unilamellarity was observed for the mixed lipid systems owing to the presence of the negatively charged phosphatidylglycerol. The packing constraints of these phospholipids, major components of cytoplasmic bacterial membranes, may be of importance in the interaction with various solutes like antimicrobial peptides, and were explained based on the nature of the headgroups and the molecular geometry of the phospholipids.

A Fourier transform infrared spectroscopic study of the interaction of alkaline earth cations with the negatively charged phospholipid 1, 2-dimyristoyl-sn-glycero-3-phosphoglycerol.

The interaction of aqueous phospholipid dispersions of negatively charged 1,2-dimyristoyl-sn-glycero-3-phosphoglycerol, sodium salt (DMPG) with the divalent cations Mg(2+), Ca(2+) and Sr(2+) at equimolar ratios in 100 mM NaCl at pH 7 was investigated by Fourier transform infrared spectroscopy. The binding of the three cations induces a crystalline-like gel phase with highly ordered and rigid all-trans acyl chains. These features are observed after storage below room temperature for 24 h. When the gel phase is heated after prolonged incubation at low temperature phase transitions into the liquid crystalline phase are observed at 58 degrees C for the DMPG:Sr(2+), 65 degrees C for the DMPG:Mg(2+), and 80 degrees C for the DMPG:Ca(2+) complex. By subsequent cooling from temperatures above T(m) these complexes retain the features of a liquid crystalline phase with disordered acyl chains until a metastable gel phase is formed at temperatures between 38 and 32 degrees C. This phase is characterized by predominantly all-trans acyl chains, arranged in a loosely packed hexagonal or distorted hexagonal subcell lattice. Reheating the DMPG:Sr(2+) samples after a storage time of 2 h at 4 degrees C results in the transition of the metastable gel to the liquid crystalline phase at 35 degrees C. This phase transition into the liquid crystalline state at 35 degrees C is also observed for the Mg(2+) complex. However, for DMPG:Mg(2+) at higher temperatures, a partial recrystallization of the acyl chains occurs and the high temperature phase transition at 65 degrees C is also detected. In contrast, DMPG:Ca(2+) exhibits only the phase transition at 80 degrees C from the crystalline gel into the fluid state upon reheating. Below 20 degrees C, the rate of conversion from the metastable gel to a thermodynamically stable, crystalline-like gel phase decreases in the order Ca(2+)&z. Gt;Mg(2+)>Sr(2+). This conversion into the crystalline gel phase is accompanied by a complete dehydration of the phosphate groups in DMPG:Mg(2+) and by a reorientation of the polar lipid head groups in DMPG:Ca(2+) and in DMPG:Sr(2+). The primary binding sites of the cations are the PO(2)(-) groups of the phosphodiester moiety. Our infrared spectroscopic results suggest a deep penetration of the divalent cations into the polar head group region of DMPG bilayers, whereby the ester carbonyl groups, located in the interfacial region of the bilayers, are indirectly affected by strong hydrogen bonding of immobilized water molecules. In the liquid crystalline phase, the interaction of all three cations with DMPG is weak, but still observable in the infrared spectra of the DMPG:Ca(2+) complex by a slight ordering effect induced in the acyl chains, when compared to pure DMPG liposomes.

Miscibility of phosphatidylethanolamine-phosphatidylglycerol mixtures as a function of pH and acyl chain length.

We have examined the mixing properties of phosphatidylethanolamine (PE) and phosphatidylglycerol (PG), the major components of many bacterial membranes. The phase transition behavior of dilute aqueous suspensions of PE:PG mixtures with different chain lengths (n = 14, 16) in 0.1 M NaCl at pH 7 and pH 2 was investigated by differential scanning calorimetry (DSC). The DSC curves were simulated using an approach which takes into account the broadening of the phase transition in addition to symmetric, non-ideal mixing in the gel and the liquid-crystalline phase. Based on the temperatures for onset and end of "melting" obtained by the simulations, the phase diagrams were constructed and then refined using a regular solution model with non-symmetric mixing in both phases. The mixing properties of PE:PG mixtures were analyzed as a function of pH and acyl chain length. In almost all cases, non-symmetric mixing behavior was observed, i.e. the non-ideality parameters are different for bilayers with low PG content compared to bilayers with high PG content. For equimolar mixtures at pH 7, when PG is negatively charged, the non-ideality parameters are negative for both phases, indicating preferential formation of mixed pairs. This mixed pair formation is more pronounced for the gel phase. At pH 2, when PG is partly protonated, the non-ideality parameter is less negative and the formation of mixed pairs is reduced compared to pH 7. The formation of PE:PG mixed pairs at pH 7 might be of benefit to a bacterial membrane, because it prevents demixing of lipid components with a concomitant destabilization of the membrane.

Synthesis, control of substitution pattern and phase transitions of 2,3-di-O-methylcellulose.

An improved heterogeneous procedure has been found for the regioselective introduction of trityl and 4-methoxytrityl groups at the primary positions of cellulose. The 6-O-tritylcelluloses produced were completely methylated by MeI-NaOH in Me2SO solution. The trityl groups were then completely removed to afford 2,3-di-O-methylcellulose without significant degradation of the polymer. 1H and 13C NMR spectroscopy and degradation analysis showed less than 5% deviation from the regular substitution pattern. Under optimum reaction conditions, almost perfectly regular cellulose derivatives could be obtained. Small changes in the substitution pattern had a strong effect on the phase transitions of the O-methylcelluloses in water. It was shown by DSC for the first time that perfect 2,3-di-O-methylcellulose does not undergo phase separation at elevated temperatures.

Miscibility of phospholipids with identical headgroups and acyl chain lengths differing by two methylene units: effects of headgroup structure and headgroup charge.

We have investigated the influence of the chemical structure and charge of the hydrophillic headgroup on the miscibility of saturated phospholipids with acyl chain lengths differing by two methylene units, namely DMPA/DPPA, DMPC/DPPC, DMPE/DPPE and DMPG/DPPG (0.1 M NaCl). All four mixtures were analysed by DSC at pH 7. To study the influence of a change in headgroup charge, we additionally investigated DMPA/DPPA mixtures at pH 4 and 12, and DMPG/DPPG mixtures at pH 2. The experimental DSC thermograms were fitted using methods described before [Johann et al., Biophys. J. 71 (1996), 3215-3228] to obtain the temperatures of onset and end of melting and first approximations for the non-ideality parameters as a function of composition. The resulting phase diagrams were then fitted using a four non-ideality parameter model for non-ideal, non-symmetric mixing in both phases. The phase diagram of the system DMPG/DPPG has a lens-like shape, the non-ideality parameters rhog and rhol for the gel and the liquid-crystalline phase, respectively, are zero, indicating ideal mixing in both phases. For the other mixtures, differences in miscibility are observed depending on the structure of the headgroup. At pH 7, rhog > rhol, i.e., the miscibility in the liquid-crystalline phase is more ideal than in the gel state. All rhog values are positive and the sequence for rhog observed is PA>PE>PC>PG. Partial protonation of PA at pH 4 or complete deprotonation at pH 12 leads to negative non-ideality parameters for both phases, indicating a preference for mixed pair formation. Protonation of PG in DMPG/DPPG mixtures at pH 2 leads to positive non-ideality parameters for both phases, indicating a tendency for demixing. The results show, that the miscibility of phospholipids with identical headgroups but chain lengths differing by two methylene groups is dependent on headgroup structure and on headgroup charge.

Nonideal mixing and phase separation in phosphatidylcholine-phosphatidic acid mixtures as a function of acyl chain length and pH.

The miscibilities of phosphatidic acids (PAs) and phosphatidylcholines (PCs) with different chain lengths (n = 14, 16) at pH 4, pH 7, and pH 12 were examined by differential scanning calorimetry. Simulation of heat capacity curves was performed using a new approach that incorporates changes of cooperativity of the transition in addition to nonideal mixing in the gel and the liquid-crystalline phase as a function of composition. From the simulations of the heat capacity curves, first estimates for the nonideality parameters for nonideal mixing as a function of composition were obtained, and phase diagrams were constructed using temperatures for onset and end of melting, which were corrected for the broadening effect caused by a decrease in cooperativity. In all cases the composition dependence of the nonideality parameters indicated nonsymmetrical mixing behavior. The phase diagrams were therefore further refined by simulations of the coexistence curves using a four-parameter approximation to account for nonideal and nonsymmetrical mixing in the gel and the liquid-crystalline phase. The mixing behavior was studied at three different pH values to investigate how changes in headgroup charge of the PA influences the miscibility. The experiments showed that at pH 7, where the PA component is negatively charged, the nonideality parameters are in most cases negative, indicating that electrostatic effects favor a mixing of the two components. Partial protonation of the PA component at pH 4 leads to strong changes in miscibility; the nonideality parameters for the liquid-crystalline phase are now in most cases positive, indicating clustering of like molecules. The phase diagram for 1,2-dimyristoyl-sn-glycero-3-phosphatidic acid:1,2-dipalmitoyl-sn-glycero-3-phosphorylcholine mixtures at pH 4 indicates that a fluid-fluid immiscibility is likely. The results show that a decrease in ionization of PAs can induce large changes in mixing behavior. This occurs because of a reduction in electrostatic repulsion between PA headgroups and a concomitant increase in attractive hydrogen bonding interactions.

New approaches to the simulation of heat-capacity curves and phase diagrams of pseudobinary phospholipid mixtures.

A simulation program using least-squares minimization was developed to calculate and fit heat capacity (cp) curves to experimental thermograms of dilute aqueous dispersions of phospholipid mixtures determined by high-sensitivity differential scanning calorimetry. We analyzed cp curves and phase diagrams of the pseudobinary aqueous lipid systems 1,2-dimyristoyl-sn-glycero-3-phosphatidylglycerol/ 1,2-dipalmitoyl-sn-glycero-3phosphatidylcholine (DMPG/DPPC) and 1,2-dimyristoyl-sn-glycero-3-phosphatidic acid/1, 2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DMPA/DPPC) at pH 7. The simulation of the cp curves is based on regular solution theory using two nonideality parameters rho g and rho l for symmetric nonideal mixing in the gel and the liquid-crystalline phases. The broadening of the cp curves owing to limited cooperativity is incorporated into the simulation by convolution of the cp curves calculated for infinite cooperativity with a broadening function derived from a simple two-state transition model with the cooperative unit size n = delta HVH/delta Hcal as an adjustable parameter. The nonideality parameters and the cooperative unit size turn out to be functions of composition. In a second step, phase diagrams were calculated and fitted to the experimental data by use of regular solution theory with four different model assumptions. The best fits were obtained with a four-parameter model based on nonsymmetric, nonideal mixing in both phases. The simulations of the phase diagrams show that the absolute values of the nonideality parameters can be changed in a certain range without large effects on the shape of the phase diagram as long as the difference of the nonideality parameters for rho g for the gel and rho l for the liquid-crystalline phase remains constant. The miscibility in DMPG/DPPC and DMPA/DPPC mixtures differs remarkably because, for DMPG/DPPC, delta rho = rho l -rho g is negative, whereas for DMPA/DPPC this difference is positive. For DMPA/DPPC, this difference is interpreted as being caused by a negative rho g value, indicating complex formation of unlike molecules in the gel phase.