Long-term viability through selective permeability

The cell-selective permeability of a retrievable microporous device encapsulating human therapeutic cells and implanted in the intraperitoneal space of mice provides long-term protection to the transplanted cells by dampening foreign-body responses and preventing the immune rejection of the graft.

In graft transplantation, lasting viability and functionality are essential requirements. Systemic immune suppression avoids graft rejection by the host immune system at the expense of indiscriminatingly impairing protective immune responses. An alternative approach is to physically shield the engrafted cells from host leukocytes by encapsulating them in a biocompatible device1,2. The architecture of the encapsulating device defines its protective properties: a capsule that is fragile or too widely spaced allows the entrance of graft-reactive immune cells; conversely, a device with insufficient permeability will impede graft vascularization and its adequate oxygenation, and may affect the efficient sensing of endogenous messengers and limit the available concentration of therapeutic factors released by the engrafted cells. Cell-encapsulating implants should also allow for proper fixation to host tissue, ensure retrievability and minimize the foreign-body response (FBR) to the graft. The FBR is characterized by the formation of a fibrotic tissue layer around the implant that further impairs bidirectional exchange of nutrients, oxygen and signalling factors2. Therefore, the design of implantable cell-encapsulation devices entails a fine balance between device permeability and long-term viability and functionality. Reporting in Nature Biomedical Engineering, Daniel Anderson and colleagues now describe an implantable and retrievable device with a chemically modified surface that limits the FBR and improves the survival of encapsulated xenogeneic cells in immunocompetent mice bearing the device in the intraperitoneal space3.

Anderson and colleagues’ device consists of a silicone base covered with a microporous polycarbonate membrane (Fig. 1). In devices implanted in mice, the researchers tested the effects of different pore sizes by determining the extent of leukocyte infiltration into grafted devices encapsulating xenogeneic human cells that recombinantly expressed mouse erythropoietin (EPO). Pore sizes of 3 μm resulted in the escape of the human cells from the graft and in an influx of xenoreactive leukocytes that led to graft loss. Devices with 0.8-μm pores excluded macrophages and T cells, which aided the survival of the encapsulated therapeutic cells. Intriguingly, 1-μm pores prevented the entry of T cells but not of macrophages, and preserved graft viability. In view of the roles of macrophages on graft repair and protection4, the discrimination of T cells and macrophages by a pore-size threshold — which may vary across species — has broader implications for regenerative medicine.

Fig. 1: A retrievable implant for the long-term encapsulation and survival of therapeutic xenogeneic cells.

The device consists of a microfabricated body sealed with a microporous polymeric membrane, allowing for the exchange of macromolecules but preventing the infiltration of some immune cells (in particular, T cells). Figure adapted with permission from ref. 3, Springer Nature Limited.

To minimize the fibrotic response and to enhance the biocompatibility of the device, Anderson and co-workers also tested different device coatings. Tetrahydropyran phenyl triazole (THPT) suppressed pro-inflammatory cytokine responses, the accumulation of extracellular matrix protein, and cellular overgrowth. Compared with implanted uncoated devices, a THPT-coated device containing EPO-producing human cells and intraperitoneally implanted in immunocompetent mice significantly increased serum EPO levels and haematocrit, which decreased on implant removal. The researchers also show that the coated devices allowed for the on-demand release of EPO (via a cell line engineered for doxycycline-inducible EPO secretion) for 130 days after transplantation. Moreover, devices loaded with rat pancreatic islets and intraperitoneally transplanted into mice deficient in pancreatic beta cells rapidly and durably normalized blood-glucose levels for devices coated with THPT; conversely, uncoated devices gradually lost function.

Anderson and co-authors used xenogeneic transplants. However, allogeneic stem-cell-derived products (for instance, therapeutics for autoimmune diseases such as type 1 diabetes) are accompanied by alloreactive responses that are less acute. The potential influence of an ongoing autoimmune response on therapeutic efficacy should therefore be taken into consideration; although pore-size constraints governing immune-cell entry may be consistent regardless of the ongoing immune-rejection events, ensuing soluble inflammatory mediators may lead to different outcomes. Likewise, the observation that infiltrating macrophages, which were not phenotypically or functionally characterized in the authors’ study, do not cause harm to the graft should be confirmed under graft-rejection conditions. Also, the authors observed that the implanted devices experienced a diminished FBR. But as the nature and magnitude of the FBR are known to depend on tissue location5, whether similar positive effects would be observed upon transplantation into anatomical regions other than the peritoneum remains untested. Furthermore, permissiveness to angiogenesis and the level of exposure to prolonged hypoxia is also site-dependent, and immunological ‘danger’ signals induced by the implantation procedure may also contribute to the course of rejection and to the immunopathological pathways involved. As such, they may differ for alternative transplantation sites.

One main benefit of macroencapsulation with respect to microencapsulation (the encapsulation of individual cells or of small cell clusters) is the possibility of graft retrievability. The durable polymeric materials in Anderson and co-authors’ macrodevices should facilitate the removal of an aberrant graft, and ensure that undifferentiated and thus potentially teratogenic cells cannot escape the device. This is an advantage over encapsulation strategies based on alginate hydrogels6, which are more prone to rupture and hence to cell escape. A distinct benefit of microencapsulation, however, is the reduced risk of hypoxia and the more direct interaction of therapeutic cells with their surroundings. A macroencapsulation device for implantation in human patients with type 1 diabetes would need to be of a manageable size while also supporting an islet mass capable of restoring glucose homeostasis with acceptable kinetics of insulin secretion. The authors’ device takes these constraints into account by keeping the reservoir depth shallow (150 μm; comparable to inter-capillary distances) and by allowing for 375 islet equivalents into each device implanted in mice. Yet during allogeneic islet transplantation a patient typically receives at least 10,000 islet equivalents per kilogram of body weight, extracted from two donor pancreases; to achieve insulin independence, the patients often require two transplants. To be scaled for use in humans, the authors’ device would need to pack the cells more densely (thus reducing oxygen availability to the cells) to keep the device footprint low, which will unfavourably affect both the viability and functionality of the cells.

Type 1 diabetes is a challenging disease to treat via cell transplantation, owing to the high number of insulin-secreting beta cells and high levels of oxygen needed, and to a low tolerance for delays in the sensing of signals from the host and in the responses to them. In indications requiring a smaller graft, such as in haemophilia and in growth hormone deficiency, Anderson and co-authors’ technology may be more readily applicable. As potent therapeutic cells become available and strategies for local immunomodulation become clinically feasible, the technical requirements for a clinical retrievable macroencapsulation device may become less stringent. And, from a patient perspective, the clinical usability of a macroencapsulation device also depends on disease burden; for example, living with type 1 diabetes involves predicting the body’s behaviour, dosing, counteracting outcomes and a fear of hypoglycaemic events. A device with long-term functionality may lessen these burdens.


  1. 1.

    Desai, T. & Shea, L. D. Nat. Rev. Drug Discov.16, 338–350 (2017).

    CAS  Article  Google Scholar 

  2. 2.

    Henry, R. R. et al. Diabetes67 (Suppl. 1), 138–OR (2018).

  3. 3.

    Bose, S. et al. Nat. Biomed. Eng. (2020).

  4. 4.

    Salehi, S. & Reed, E. F. Curr. Opin. Organ Transplant.20, 446–453 (2015).

    CAS  Article  Google Scholar 

  5. 5.

    Anderson, J. M., Rodriguez, A. & Chang, D. T. Semin. Immunol.20, 86–100 (2008).

    CAS  Article  Google Scholar 

  6. 6.

    An, D. et al. Proc. Natl Acad. Sci USA115, E263–E272 (2018).

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Matthias von Herrath.

Ethics declarations

Competing interests

All authors are employed by Novo Nordisk, hold shares in the company and are involved in the company’s stem cell research and development.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Coppieters, K., Winkel, L. & von Herrath, M. Long-term viability through selective permeability. Nat Biomed Eng 4, 763–764 (2020).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing