CRM imaging of alginate microbeads: validation, technical aspects, and quantitative analysis
Alginate microbeads represent one of the most popular microsphere designs applied in the cell encapsulation field1, which can be ascribed to their well-documented immunoprotective character in various animal models and also to the overall simplicity of this one-component system. The preparation method allows for certain design flexibility, i.e., the localization of alginate in the microbeads can be influenced by the experimental conditions. The currently available microbead designs range from completely homogeneous ones to microbeads with a high degree of heterogeneity where alginate is concentrated in the outermost layer of the microbead. In this work, alginate microbeads were used as the main model microsphere for tailoring the CRM methodology to the needs of the microencapsulation field. Besides the acquisition of both qualitative and quantitative data on the spatial distribution of alginate, this included also validation of the method and addressing various technical aspects, such as fixing the position of the microbead during the analysis, verification of the calibration method, data processing (normalization and deconvolution of probing confocal volume), and choosing the optimal laser wavelength.
The workflow diagram in Supplementary Fig. S2 depicts the individual steps involved in the CRM data acquisition and processing. To localize alginate in the microbeads, the symmetric stretching band ν(COO−) of both free and complexed carboxylate residues at about 1415 cm−1 was selected in the Raman spectrum of sodium alginate (provided in Supplementary Fig. S4) because its intensity does not depend strongly on the ratio of mannuronic to guluronic acid residues34. The position of this band was reported to shift to higher wavenumbers by up to 20 cm−1 depending on the Ca2+ concentration used to form the alginate hydrogel35,36. In this work, this band shifted less significantly, up to about 10 cm−1 (Supplementary Fig. S10). In practical terms, CRM imaging is complicated by the necessity to maintain a constant position of microspheres during data acquisition that can take up to 60 minutes. We solved this problem by designing a custom holder, manufactured through the laser ablation technique, featuring bowl-shaped wells to accommodate individual microspheres (for details see Supplementary Fig. S3).
In order to validate the CRM data, CLSM was employed as it is currently the standard method for visualizing the spatial distribution of polymeric components in hydrogel microspheres used for cell encapsulation18,20,37,38. As mentioned above, CLSM requires sample labeling. This was achieved through the reaction of alginate with fluorescein amine. Subsequently, the labeled alginate was used for preparation of two types of alginate microbeads, i.e., homogeneous and heterogeneous, that were then analyzed by both CLSM and CRM to obtain relative concentration profiles of alginate (Fig. 1c,g). The obtained profiles overlap, which confirms that the CRM method provides results comparable to CLSM in the qualitative analysis.
Both CLSM and CRM analyses of hydrogel microspheres are affected by a certain error originating from the changes in signal attenuation as the laser beam passes through different regions of the microsphere during the scan along the equatorial microbead cross-section (the x-axis). The signal attenuation is caused mainly by energy absorption when the laser beam propagates through hydrogel layers of different density (different local concentration of polymers) and thickness. In CRM, however, these fluctuations can be compensated by normalization of the Raman signal intensity to the signal intensity of water24. The bending band of water at around 1640 cm−1 was selected for the normalization because it is insignificantly sensitive to factors influencing the hydrogen bonding interactions (e.g., temperature and ionic strength)39. This method results in a more accurate representation of polymer spatial distribution as compared to using the fluorescence signal in CLSM where the signal intensity loss cannot be compensated. In addition, corrections related to the deconvolution of the probing confocal volume can be applied in the CRM analysis. In Supplementary Fig. S9, this approach is demonstrated and the CRM and CLSM methods are contrasted using radial intensity profiles (a scan along the z-axis) of a highly heterogeneous alginate microbead where, due to the high hydrogel density in the outer regions, the differences in signal attenuation are expected to be most significant.
The relative concentration profiles represent useful information that can be correlated with the conditions used for microbead preparation. Nevertheless, especially with regard to transplantation of encapsulated cells, there is critical need for data on the absolute local polymer concentration in hydrogel microspheres. We therefore constructed calibration curves (Supplementary Fig. S7) and used them to transform the relative Raman intensity profiles into absolute alginate concentration profiles. As mentioned by Heinemann et al.24, preparation of homogeneous alginate microbeads of precisely defined alginate concentration is not experimentally feasible. This is mainly due to the unpredictable swelling or shrinking of the microbeads, which changes the local alginate concentration, and also because of technical reasons such as the very high viscosity of more concentrated alginate solutions. For this reason, we also resorted to using alginate solutions for calibration purposes. In the concentration range where the experimental reasons permitted, we successfully validated this approach by CRM analysis of homogeneous alginate hydrogel slabs with defined alginate concentration that were prepared by the internal gelling method31. The real concentration of alginate in the slabs was quantified by following their swelling kinetics (Supplementary Table S1). As seen in Supplementary Fig. S8, these values were within the predicted concentration interval determined by the quantitative CRM analysis.
The quantitative CRM analysis of the heterogeneous microbeads introduced above showed that the alginate concentration was about 12 wt.% at the surface, decreasing steeply to 0.5–1 wt.% in the microbead core (Fig. 1d). Figure 1h reveals that the absolute alginate concentration across the homogeneous microbeads was approximately 2 wt.%. Knowledge of the local alginate concentration in various alginate microbead designs is crucial as previous studies have confirmed that the fate of immobilized cells, characterized by their metabolic and secretory activities13,40, survival41, and differentiation42, strongly depends on the alginate concentration in the cell environment. For instance, Stabler et al. showed that the increase in concentration of high G alginate in solution used for preparation of microbeads from 1 to 2 wt.% causes significant retardation of metabolic and secretory activities in immobilized βTC3 cells due to the inhibition of their growth40. Similarly, BHK cells were shown not to survive in an alginate hydrogel made of 2 wt.% high G alginate41. However, due to the lack of suitable analytical methods, such studies correlated the cell performance only with nominal concentrations of alginate in the original solutions and not with true concentrations in the immediate proximity of the cells. Quantitative mapping of local alginate concentration offered by CRM can provide considerably more accurate data and help thus avoid conditions in preparation of microbeads that can lead to temporary or permanent deterioration of viability and functionality of encapsulated cells. On a general note, CRM can also bring the necessary insight into the dynamics of the crosslinked network formation and subsequent re-arrangement triggered by external stimuli in alginate-based hydrogels and similar materials.
To confirm that the quantitative CRM analysis is not influenced by the choice of the laser type, we analyzed selected microbeads using two different laser wavelengths, 532 nm and 785 nm. As shown in Supplementary Fig. S1, for the particular hydrogel microbeads tested here, comparable concentration profiles were obtained at both the wavelengths. In general, the 532 nm wavelength offers higher spatial resolution and is potentially less damaging to sensitive samples as lower average probing power can be used. However, we opted for the 785 nm wavelength as it is less susceptible to autofluorescence, which might be significant when analyzing explanted microspheres where protein adsorption cannot be ruled out. The CRM data presented in this paper were thus acquired using the 785 nm laser.
Impact of environment on microsphere structure
Perhaps the most important area where CRM imaging can be applied is the visualization of changes to polymer spatial distribution in non-covalently crosslinked microspheres upon their exposure to different environments, especially in vivo conditions. As was previously mentioned, immunoprotective properties of the hydrogel microsphere are closely linked to the localization of polymeric components in the microsphere volume. Major changes to the polymer spatial distribution can thus compromise the protection that the microspheres provide to the encapsulated cells. For this reason, we employed CRM to determine how the structure of hydrogel microspheres responds to the changes of the environment. Two mainstream microsphere designs were studied: (i) heterogeneous, single-component, ionotropically gelled alginate microbeads, and (ii) three-component, polyelectrolyte complex-based microcapsules. In the following text, results obtained for each of the microsphere type will be discussed separately, which will form a basis for the subsequent comparison of the two designs.
Alginate microbeads used for immunoprotection of transplanted cells have been intentionally prepared with either higher5 or lower6,15 degree of heterogeneity. It is generally assumed that the immunoprotective properties are related to the localization of alginate in the microbead13. However, whether the initial spatial distribution of alginate is maintained when the microbeads are exposed to different environments, including the in vivo environment, remains currently unknown. Using preparation conditions similar to those previously utilized in clinical trials5,6, two types of heterogeneous alginate microbeads were prepared with markedly different initial degrees of heterogeneity. The preparation and storage medium was a D-mannitol solution, which ensured preservation of the microbead structure because anti-gelling sodium ions were absent. Before implantation, the microbeads were transferred to saline. In order to achieve more representative comparison of microbeads before implantation and after explantation, a small fraction of these microbeads was separated and incubated for 24 hours at 37 °C to simulate the potential body temperature influence. The remaining microbeads were implanted into the peritoneal cavity of nude mice for a period of 4 weeks, and stored in saline after explantation. To evaluate the effects of environment, the microbeads were analyzed by CRM after the three different stages: preparation (mannitol storage), incubation in saline at 37 °C for 24 h, and explantation (Fig. 2). Note that CaCl2 was added to all the storage media (2 mM final concentration) to mimic the physiological concentration of Ca2+ ions. This is of particular relevance in the case of explanted microbeads where CaCl2 helps conserve the attained alginate concentration profile prior to the CRM analysis.
Optical microscopy images of the more heterogeneous microbeads shortly after preparation reveal that the used gelling conditions afforded the expected core/shell morphology resulting from the gelling front movement from the microbead surface towards its core (Fig. 2a)29. A shell of about 40 μm thickness was formed, corresponding to the local alginate concentration of about 10 to 11 wt.%. This shell could be identified by both optical microscopy and CRM. Closer to the microbead core, CRM alone was capable of visualizing the regions of different local alginate concentration that reached a minimum of about 0.5 wt.% in the microbead core. Figure 2b depicts the microbeads incubated for 24 h at 37 °C in saline and indicates that the level of heterogeneity was significantly reduced compared to microbeads stored in D-mannitol, with a maximum alginate concentration at the microbead surface of about 6 wt.% and about 1 wt.% in the core. Such a change is not surprising and is ascribed to the influence of the non-gelling sodium ions causing a partial dissolution of the alginate hydrogel network due to the exchange of the crosslinking ions for non-gelling sodium ions20. The attained profile did not alter with time, indicating that Fig. 2b represents the equilibrium spatial distribution of alginate in a saline environment. Strikingly, in the microbeads exposed to the in vivo environment, the heterogeneity of alginate spatial distribution diminished almost completely, i.e., only minor differences in alginate concentration were observed, with up to 4 wt.% at the surface and 2–3 wt.% across the rest of the microbead (Fig. 2c).
The CRM profiles and optical microscopy images of alginate microbeads prepared with a lower degree of heterogeneity are shown in Fig. 2d–f. These microbeads were treated in the same way as the more heterogeneous microbeads above. While the optical microscopy images are virtually identical in the respective stages, the CRM profiles reveal a different picture, i.e., changing levels of heterogeneity. The span of alginate concentrations between the microbead surface and core was approximately 6 to 0.5 wt.%, 3.5 to 1.0 wt.%, and 3.0 to 1.5 wt.% for microbeads after preparation (D-mannitol storage), after 24 hours storage at 37 °C in saline, and after explantation, respectively. Notably, the final profiles for both the analyzed microbead designs, shown in Fig. 2c,f, are visually similar, confirming that the ultimate alginate spatial distribution in the microbeads exposed to the in vivo environment does not depend significantly on the used gelling conditions (i.e., the initial microbead heterogeneity). Note that the gelling process causes shrinking of the original alginate droplets, concentrating alginate in the resulting microbeads. This shrinkage is different for the two gelling methods used here, which explains the concentration and size differences visible for the explanted microbeads (Fig. 2c,f).
The second type of microspheres whose structure was followed by the CRM imaging were multi-component hydrogel microcapsules, illustrating the utility of the CRM methodology in the analysis of complex systems43. Specifically, we employed a microcapsule made of three polymeric components: sodium alginate (SA), sodium cellulose sulfate (SCS) and poly(methylene-co-cyanoguanidine) (PMCG)26. Encapsulation of islets of Langerhans in this microcapsule has led to diabetes reversal in various animal models26,44,45, and recently showed biotolerability in a preclinical model of non-human primates46,47. The microcapsule is predominantly stabilized by two coexisting networks: (1) an ionotropic network between calcium or barium cations and alginate, and (2) a polyelectrolyte complex between SCS and PMCG26. The microcapsules were analyzed by CRM at two stages: prior to implantation (storage in saline) and after explantation from the intraperitoneal cavity of C57bl/6 mice 2 weeks post-implantation (Fig. 3) or nude mice 4 weeks post-implantation (Supplementary Fig. S15). Note that the microcapsule depicted in Fig. 3 was made using D-mannitol as an osmolyte in the first step of the microcapsule preparation, i.e., under conditions similar to those used for the preparation of alginate microbeads shown in Fig. 2 above. We see this as the key point for the side-by-side comparison of the two encapsulation designs. The following Raman bands were used for determining the spatial distribution of the individual polymeric components: 1415 rel. cm−1 for SA, 1070 rel. cm−1 for SCS, and 770 rel. cm−1 for PMCG (Supplementary Fig. S6). Validity of the obtained data was confirmed by analyzing microcapsules made of fluorescently labeled PMCG by both CRM and CLSM (Supplementary Fig. S14). In contrast to the alginate microbeads discussed above, data on absolute concentrations of the individual polymers cannot be easily extracted in the present multi-component system. Nevertheless, relative changes in the spatial distribution of the individual polymeric components can be assessed. CRM imaging can thus bring useful insight into the mechanism of hydrogel network formation and provide information on how the microcapsule structure responds to the changing environment.
Figure 3 shows the optical microscopy images of the studied microcapsules together with the obtained spatial distribution profiles for all the three polymeric components. The optical microscopy images imply that the SA-SCS/PMCG microcapsule consists of a core that is covered by a membrane of several tens of micrometers in thickness. In general, the CRM data show that the polymers are concentrated in the outermost region of the microcapsule. Inside the microcapsule, SA is the main polymeric component (around 35% relative concentration) while SCS and PMCG show only negligible presence here. In the microcapsule prior to the implantation, CRM reveals two distinct regions in the outer sphere (see the detailed Fig. 3c). We postulate that the outermost layer (ca 0–40 µm from the surface) roughly overlaps with the visually observable membrane. SA co-localizes with PMCG at the membrane and its relative concentration is highest here. The second layer (between ca 40 and 60 µm from the surface) appears to represent the outermost region of the core. Here, PMCG co-localizes with SCS, with both the polymers showing their concentration maxima. This co-localization indicates the formation of a strong polyelectrolyte complex between SCS and PMCG26, which agrees qualitatively with the data obtained previously by CLSM38. The two layers are separated with a region where all the three polymers show their local concentration minima. The microcapsule structure observed by CRM reflects well the main processes taking place during the 2-step microcapsule preparation. In the first step, a microbead stabilized predominantly by an alginate network is formed that contains trapped SCS. In the second step, the microbead is exposed to cationic PMCG, which results in the membrane formation. PMCG then diffuses further towards the core, but this process is apparently progressively slowed down by complexation with SCS. In this way, another SCS-PMCG polymer-rich region is formed, i.e., the second layer observed by CRM. Limited diffusion of high-molecular weight fractions of SCS throughout the hydrogel network from the core towards the microcapsule surface can also account for this observation.
In the explanted microcapsule (Fig. 3f), a simpler concentration profile was observed with all three polymeric components showing their relative concentration maxima in the outermost layer of the microcapsule, i.e., the membrane. The second layer became much less prominent as illustrated by only small local maxima of SCS and PMCG relative concentrations. A slight shift (ca 10 µm) of this layer from the surface towards the microcapsule center was observed. Apparently, the in vivo environment also caused certain transfer of SA from the core towards the membrane, which is visualized by the decrease in its relative concentration in the core from about 35% to about 25%. Taking into account the considerable polymer mass transfer observed in alginate microbeads above, it is possible that the loosening of the alginate/divalent ions network in the in vivo environment accounts for the alginate mobility and also allows the spatial rearrangement of the remaining two polymers, especially the high-molecular weight SCS. The increase in the apparent heterogeneity in the alginate distribution profile can then be ascribed to the SA-PMCG interactions that, despite being weaker than SCS-PMCG interactions26, can still influence mobility of SA. The data qualitatively identical to those shown in Fig. 3 were obtained for SA-SCS/PMCG microcapsules prepared at slightly different gelling conditions using saline instead of D-mannitol in the first gelling step (Supplementary Fig. S15). These conditions led to slightly lower heterogeneity of the microcapsule structure that was maintained in vivo. In this case, SA-SCS/PMCG microcapsules were implanted intraperitoneally to nude mice and explanted 4 weeks post-implantation, i.e., the identical in vivo conditions were used as for alginate microbeads (Fig. 2).
The obtained CRM data show that alginate microbeads and multi-component microcapsules have considerably different structural stability in the in vivo environment. In general, both alginate microbeads and alginate-based microcapsules are composed of non-covalently crosslinked hydrogel networks stabilized predominantly by electrostatic interactions48. The number and position of crosslinks can fluctuate in response to environmental characteristics such as osmotic pressure and presence of ions and other molecules (e.g., proteins in the in vivo environment). For both the heterogeneous alginate microbead designs studied here, the heterogeneous character decays during storage in saline, reaching an equilibrium profile with certain heterogeneity retained, and almost completely diminishes when the microbeads are exposed to the in vivo environment. At present, it is not clear which characteristics or components of the in vivo environment are responsible for the observed partial dissolution and equilibration of the ionotropically stabilized alginate network. Nevertheless, the obtained data allow us to hypothesize that in all the previously performed in vivo studies involving alginate microbeads, the spatial distribution of alginate eventually became homogeneous regardless of the preparation conditions. From this point of view, it appears futile to attempt at specific control of the initial heterogeneity of alginate microbeads. Furthermore, it can be expected that the loss of microbead heterogeneity upon exposure to the in vivo environment leads to an increase in the molecular weight cut-off (MWCO) of the microbeads, possibly deteriorating immunoprotective properties of the related cell encapsulation systems. Since considerable loss of heterogeneity was observed even in the saline environment, it is not surprising that the previously reported MWCO values for alginate microbeads of vastly different (initial) heterogeneity, exposed to saline after preparation, were similar (250 to 350 kDa, protein equivalent)15,49. It follows that the heterogeneous character of alginate spatial distribution has to be conserved by means other than ionotropic gelling, e.g., by polyelectrolyte complexation with a polycation11,20 or by a combination of polyelectrolyte complexation with covalent crosslinking50,51. This view is supported by our CRM results on structural changes in the SA-SCS/PMCG microcapsules. In this case, partial dissolution of the ionotropic alginate network enabled post-complexation processes accompanied by migration of SCS and PMCG towards the microcapsule membrane, conserving thus the heterogeneous profile of the microcapsules. Consequently, from the structural point of view, SA-SCS/PMCG microcapsules qualify for a better control of immunoprotective properties than alginate microbeads.