Cellular therapies have emerged as a promising treatment for many otherwise untreatable diseases and disorders1–4
. However, widespread clinical implementation5–8
has been hampered partially because of poor long-term functionality and survival of therapeutic cells. In particular, it is not well understood whether graft failure may be the simple result of cell death following transplantation and, if so, when this occurs. A non-invasive imaging method that can probe cell viability would, therefore, speed up human translation of cell therapies. As of today, radionuclear imaging with 111
In-oxine-labeled cells is the only FDA-approved tracking method available in the clinic9
, but it cannot assess cell survival. This latter problem is common for all imaging techniques employing exogenous labeling agents that continue to display contrast when cells are dying, including magnetic resonance imaging (MRI) of superparamagnetic iron oxide (SPIO)-labeled cells10
. In contrast, reporter gene-based imaging relies on proteins that either accumulate or convert substrates, and ribosomal production occurs only in live cells. Reporter gene-based imaging is well established in the pre-clinical setting with luciferase-based bioluminescent imaging (BLI) being exceptionally robust11
. However, this technique is limited to small animals because of the light absorption and scattering by the tissue. PET is a clinical imaging modality providing a reporter gene-based approach that has recently been introduced into the clinic using the herpes simplex virus 1 thymidine kinase12
. However, even when humanized, such a xenogeneic (bacterial) protein raises clinical concerns of potential immunogenicity. Moreover, in order to achieve a stable, constitutive expression, lenti- or adenoviruses need to be used which also poses clinical concerns about overall safety.
Furthermore, the widespread use of clinical cell therapy has been hampered by graft immunorejection and the lack of cells that have the proper histocompatibility antigenic makeup. Microencapsulation has been proposed as a way to immunoprotect the graft by embedding them within a semi-permeable hydrogel (Supplementary Fig. S5
). This approach allows free diffusion of small molecules such as insulin, therapeutic growth factors and cytokines, nutrients, and metabolites, while blocking invading host immune effector cells and immunoglobulins. Microencapsulation has been used for cell therapy of liver failure13,14
, type I diabetes mellitus, and pancreatic carcinoma7
. By embedding contrast agents during synthesis, the engraftment of encapsulated cells has been tracked using X-ray/CT15–18
, and MR imaging16–19
. However, none of these techniques has been able to report on cell survival, and merely allow anatomical co-registration of engrafted cells together with real-time, image-guided delivery.
Chemical exchange saturation transfer (CEST) is an emerging MRI contrast mechanism20–23
based on the use of radiofrequency (RF) saturation pulses to detect agents containing protons that exchange rapidly with water. Importantly, the exchange rate, and thus the CEST contrast, can depend strongly on pH20,21,24
(). When the pH decreases from its normal cellular value (pH=7.3), the exchange rate (ksw
) decreases for base-catalyzed exchangeable protons, such as the guanidyl NH protons in L-arginine, leading to a decrease in CEST contrast. Cell death and inflammation are also associated with concurrent acidification of extracellular pH25–27
. We hypothesized that advanced biomaterials that can sense changes in pH may be used as nanosensors for probing cell viability.
Schematic showing the principles of in vivo detection of cell viability using LipoCEST microcapsules as pH nanosensors
Using L-arginine, a molecule with multiple exchangeable NH protons, as a pH-sensitive CEST contrast agent (), we present here an approach for non-invasive imaging of the viability of encapsulated cells. To this end, we synthesized arginine-rich “LipoCEST” microcapsules by incorporating L-arginine filled liposomes inside the capsule and protamine sulfate as an arginine-rich cross-linker in the alginate capsule coating. We demonstrate that apoptotic encapsulated human hepatocytes can be readily detected with CEST MRI in vivo, and validated this with conventional BLI using the luciferase reporter gene.
Two main criteria were used for the selection of marker molecules used as pH nanosensors: 1) the pH-dependent MRI CEST contrast should be sensitive to pH changes between 6.0–7.5, i.e. the physiological range; and 2) the nanosensor should be biocompatible for possible future clinical translation. Based on our previous studies28,29
L-arginine containing liposomes () are ideal, as they possess a strong pH-dependent CEST contrast within this pH range and L-arginine is FDA-approved. As we have shown previously for arginine peptides28
, their pH sensitivity is due to a reduction in the exchange rate (kSW
) of the guanidyl protons of L-arginine with water protons when pH decreases. shows the preparation and structure of microcapsules with the portions of the CEST probes containing exchangeable protons that generate CEST contrast labeled in red. In brief, alginate microcapsules loaded with L-arginine liposomes were prepared by mixing these liposomes with alginate and cells followed by gelation with Ba2+
ions, then crosslinking these beads with either clinical grade protamine sulfate (PS) or poly-L-lysine (PLL), and coating with a second layer of alginate30,31
(). This composition is similar to the formulations tested clinically5,32
. Light microscopy () showed a uniform distribution of cells within the microcapsules, which have a diameter of ~350 μm.
Cartoon outlining the procedure for preparation of LipoCEST microcapsules
To optimize the pH nanosensor, we compared six formulations (n=3) with the two crosslinkers (PLL and PS) to form alginate-PLL-alginate (APLLA) and alginate-PS-alginate (APSA) capsules, respectively. The CEST contrast was measured at pH=7.4 following RF saturation pulses at 2 ppm (arginine protons) from the water resonance frequency. shows the measured CEST contrast, which is defined as MTRasym
, i.e. MTRasym
, where S−Δω
are the MRI signal intensities after saturation at −Δω and +Δω frequency offsets from the water proton frequency (set at 0 ppm). For both crosslinkers, the addition of liposomes to microcapsules enhanced the CEST contrast at 2 ppm ( and Supplementary Fig. S1
). Furthermore, PS provided much higher CEST contrast than PLL. This is not surprising, as the contrast at 2 ppm is primarily produced by the guanidyl protons of the arginine side chain, which are present on both L-arginine and arginine-rich PS28
but not in PLL.
Experimental data determining the CEST MRI contrast and stability for LipoCEST microcapsules as a function of formulation
It has previously been reported that different concentrations of cholesterol may affect the release profile of cargo molecules from liposomes embedded in alginate33
. We therefore evaluated different ratios of cholesterol used for preparing liposomes. APSA capsules with liposomes prepared with increased cholesterol content produced the highest CEST contrast ().
We then measured the in vitro stability of the CEST contrast for the two best formulations, Lipo70-APSA and Lipo50-APSA, over a period of one month at 37°C with daily replacement of saline. Lipo70-APSA showed a relatively constant contrast, with an overall decrease in MTRasym of ~0.06 over one month (). The rates of decrease (stabilities) for the two formulations were comparable and based on the magnitude of CEST contrast produced by Lipo70-APSA capsules, we selected this formulation for the remainder of the studies.
Next, we tested the sensitivity of Lipo70-APSA for probing pH values in the physiological range. The MTRasym decreased by 0.24 from pH 7.5 to 5.5. (), demonstrating that the sensor is mainly sensitive over the pH range of 6.0–7.5. The spatial selectivity of CEST contrast with pH changes was examined through the addition of 10 μL aliquots of 1 M hydrochloric acid () to the top of a sample containing homogeneously distributed capsules. The CEST contrast at 2 ppm (, pH 7.4) corresponded to individual capsules in (pH 7.4) and was constant. After addition of a total of 100 μL acid, the MTRasym map showed a clear reduction in CEST contrast within the top layer but not the bottom layer (, addition of acid). The top layer contrast was reduced proportionally to the total acid volume (), with a 33% decrease in contrast upon reduction of the pH from 7.4 to 6.9. These results indicate that LipoCEST capsules are sufficiently sensitive to be used as imaging probes for detecting pH changes in the physiological range.
Experimental data displaying the sensitivity of the MRI contrast for the LipoCEST capsules prepared to local pH
Capsule permeability was assessed through measurement of lateral diffusion of 10–500 kDa dextrans labeled with FITC34
(Supplementary Methods). All of the tested capsule formulations were permeable to 10–150 kDa dextran-FITC, semi-permeable to 250 kDa dextran-FITC, and impermeable to 500 kDa dextran-FITC (Supplementary Fig. S2
). Their mechanical strength was then tested by placing them under osmotic pressure and by agitating with glass beads35
. The APSA capsules had a higher mechanical strength than APLLA capsules. In general, the Lipo70-APSA appeared to possess a similar size selective permeability and favorable mechanical strength compared to APLLA (formulation used in clinical trials).
To evaluate whether LipoCEST-based imaging can report on the viability of grafted cells, encapsulated hepatocytes were imaged before and after the induction of apoptosis using staurosporine (STS). Apoptosis is one of the main causes for early cell death in transplanted cells36
. We observed a significant drop in the MTRasym
at 2 ppm at 12 hrs after induction of apoptosis (). The MTRasym
of the STS-treated capsules containing apoptotic cells was significantly lower than that of the untreated capsules containing live cells (n=3, P<0.01, ), resulting from the drop in pH as cell death occurred. The MTRasym
of the dead cells group did not decrease over time. In theory, this decrease in CEST contrast could also be due to the loss of L-arginine, however, this is highly unlikely based on our stability measurements (the MTRasym
at day 0 was the same as that at day 7; ).
In vitro experimental data verifying that LipoCEST capsules can report on apoptosis through MRI contrast
Immunofluorescent staining () revealed that STS-treated capsules (top panel) contained mostly apoptotic cells while non-treated capsules (middle panel) contained predominantly live cells, and with STS-treated cells representing dead cells (bottom panel). Since we observed no significant decrease in contrast in capsules without the addition of STS, we conclude that the decrease in CEST must be derived from cells undergoing apoptosis. Moreover, STS-treated capsules containing a different numbers of cells (Supplementary Fig. S4
) showed that the magnitude of the decrease in MTRasym
is positively associated with the number of dead cells as quantified using BLI (transduction of cells with luciferase; Supplementary Fig. S3a
). In response to a decrease in average radiance, the relative CEST contrast decreased, which corresponded to about 1×105
apoptotic cells. To validate that co-encapsulation with LipoCEST agents did not affect cell survival, BL images were collected. As similar to APLLA, LipoCEST capsules were non-cytotoxic to encapsulated hepatocytes as observed in culture for 30 days (Supplementary Fig. S3
), which showed these capsules were suitable to pursue in vivo
2,500 LipoCEST capsules containing Luc
-transfected hepatocytes were subcutaneously (s.c.) transplanted into the flank region of immunocompetent Balb/C mice. Multiple transplantation routes and sites37
have been considered for delivery of therapeutic cells, including the peritoneal cavity37
, portal vein38
. We chose the s.c. delivery route because this site has several advantages, including a superficial location allowing collection of high quality bioluminescence data42,43
, a well-defined anatomical region of interest for CEST imaging, and a relatively limited supply of nutrients and oxygen that will cause cell death over time.
CEST and BL images were collected for three groups of mice at 6 hours (day 0), one day, one week and two weeks after transplantation (). These groups consisted of LipoCEST capsules () without cells (−Cells), LipoCEST capsules loaded with hepatocytes transplanted in mice receiving immunosuppression (+Cells/+IS), and LipoCEST capsules loaded with hepatocytes transplanted in non-immunosuppressed mice (+Cells/−IS). The group without cells displayed a stable CEST contrast (0.23±0.01; n=4), in good agreement with our in vitro
data (0.28±0.02; ) over two weeks. The CEST contrast of the two groups containing encapsulated hepatocytes decreased on day 1. Importantly, the decrease in contrast of the +Cells/+IS group was less severe than the +Cells/−IS group (), corresponding to a smaller decrease in BL signal (). This pattern was consistently observed for all animals within these groups. The CEST contrast for both the +Cells/+IS and +Cells/−IS group decreased by about ~14% on day 1, confirmed by an order of magnitude decrease in BL signal (). This initial drop in cell viability is commonly observed in cell transplantation studies44
. In the +Cells/+IS group, the CEST contrast remained approximately constant during the remainder of the 14-day study, in agreement with the BL signal hovering around ~106
/sr during this time period (). A significant loss in contrast was observed in the +Cells/−IS group on days 7 and 14 (28% and 33%, respectively), which was associated with a significantly lower BL signal on days 7 and 14 as compared to day 0 (n=8; P<0.01), and as compared to the +Cells/+IS group (). The CEST contrast the +Cells/−IS group correlated with the viability as measured by BLI, with P<0.05, r=0.81 ( and Supplementary Fig. S6
). Based on our permeability measurements, the LipoCEST capsules possessed a cut-off pore size that would still allow small molecules to invade the encapsulated cells45
, and hence less viable hepatocytes were found in the +Cells/−IS group (radiance ~104
/sr). In the +Cells/+IS group, immunosuppression46,47
alleviated this and resulted in BL signal one order of magnitude larger than the +Cells/−IS group. These observations demonstrate that the CEST method is sensitive enough to observe the different rates of cell death in these two groups of animals, although the nature of the contrast mechanism between the BLI and CEST is different. According to the BLI data (Supplementary Fig. S3a,b
), we estimate that the death of 3×105
hepatocytes (out of 5×105
transplanted) will result in a 33% drop in CEST contrast, indicating a significant sensitivity of this approach in vivo
(Supplementary Methods). In addition, we studied the CEST contrast changes at 28 days after transplantation in three groups of animals: +Cells/+IS, +Cells/−IS, and +DeadCells/+IS groups (Supplementary Fig. S9
). The +DeadCells/+IS group had capsules containing cells that were treated with STS before encapsulation, and exhibited no BLI signal. Similar to the in vitro
studies, the CEST contrast of the +DeadCells/+IS group did not change over time. The CEST contrast of the +Cells/−IS group was significantly lower than that of the +Cells/+IS group at day 28 (P<0.05; Supplementary Fig. S9a
). Moreover, in the +Cells/+IS group, we found that the CEST signal had a significant decrease at day 28 (P<0.01; Supplementary Fig. 9b,e
) as compared to day 0, which corresponds to a significant decrease in BLI signal at day 28 as compared to day 0 (P<0.001; Supplementary Fig. S9d,e
In vivo CEST and BL imaging of LipoCEST capsules containing hepatocytes
Co-registration of CEST images and conventional anatomical images allowed a further evaluation of the temporal-spatial differences that were observed between the groups. On day 14, CEST contrast at the center was 34% lower than that of the edge of the region containing LipoCEST capsules for mice in the +Cells/−IS group. The MT-weighted images (Supplementary Fig. S7
) indicated that there were approximately the same numbers of capsules in these two regions. To investigate whether this local drop in contrast was associated with local changes in cell viability, we collected 3D bioluminescent tomographic (BLT) images (Supplementary Fig. S8
). BLT detected a lower number of viable cells in the center of the capsule transplantation site for the +Cells/−IS mice, which was not observed in the +Cell/+IS mice. This corroborated with CEST results and a reduced availability of nutrients and oxygen in the center region is a likely cause.
Histology was performed to assess the inflammatory and immune responses of the host towards transplanted capsules. Haematoxylin and Eosin (H&E) staining showed distinctive differences among the three groups of mice for both the total number of infiltrating cells and cells surrounding each capsule (), which shows the foreign body reaction (FBR). The FBR for the LipoCEST capsules without cells was negligible (, left). The number of infiltrating cells was highest in the +Cells/−IS (, right) group followed by the +Cells/+IS group (, middle), and then the −Cells group. Cell infiltration could be enhanced by the extensive apoptosis in the +Cells/−IS group (). Small proinflammatory molecules can pass through the capsule in both +Cells groups, and the immunosuppression in the +Cells/+IS group reduces inflammation and cell infiltration. There were about 5 times more remaining cells found in capsules of the +Cells/+IS group as compared to the +Cells/−IS group. The histological findings underscore the BLI findings with significant differences in the number of viable encapsulated cells between the groups.
Taken together, these results indicate that LipoCEST capsules have potential as an indirect local marker for sensing cell viability, with the CEST contrast decreasing upon cell death over time. S.c. transplanted xenogeneic hepatocytes survived for two weeks but eventually died due to the incomplete immunoprotection of alginate and the presence of infiltrated cells creating hostile microenvironments, therefore immunosuppression was effective in prolonging cell survival. Immunosuppression has been used to sustain the survival of cell grafts in patients46
, and transient immunosuppression has been investigated as an auxiliary component of encapsulated cell therapy to improve cell survival47
. We demonstrate that LipoCEST nanosensors are sensitive enough to detect cell death caused by incomplete immunoprotection of hydrogels in cell therapies.
We can envisage two main limitations when using these CEST pH-nanosensors: 1) A complex interpretation of signal changes due to the endogenous processes occurring in vivo that could influence the CEST signals and 2) the speed of the CEST imaging protocol we have tested. Based on our experiments, we have concluded that cell death is a major source of the decrease in the CEST contrast. However, other processes could potentially contribute to the change in CEST contrast. For example, the CEST contrast might also reduce over an extended period of time as a result of biodegradation of the LipoCEST capsules. Second, for these initial studies we chose to image only one slice with our protocol, which might not represent the whole capsule region. With the ongoing improvements in pulse sequences and MRI hardware, we expect that multi-slice CEST imaging could be achieved in the near future which would provide more complete information on the entire capsule region.
A new nanosensor-based imaging platform has been developed for local sensing of pH changes in the microenvironment associated with cell death. The CEST MRI approach allows non-invasive, real-time imaging of transplanted cells without the limits of signal penetration depth encountered in optical (BL) imaging studies. Moreover, cell viability patterns can be evaluated within their anatomical context. These alginate encapsulated cells are currently being tested in clinical trials. We expect that this technology is translatable to the clinic based on the addition of biodegradable CEST liposomes and the common availability of MRI scanners.