The fifth session of the workshop explored the development of tools and preclinical models for monitoring and optimizing the host’s pro-regenerative environment. The objectives of this session were to (1) explore recent advances in monitoring and imaging of the immune system as well as the potential implications of these new approaches for clinical translation of regenerative medicines and (2) discuss challenges and opportunities with regard to preclinical models for studying immune system involvement in response to regenerative medicine. The session was moderated by Sadik Kassim of Vor Biopharma.
TOOLS FOR IMMUNE PROFILING AND MONITORING
Garry Nolan, Rachford and Carlota Harris Professor in the Department of Pathology at Stanford University, discussed how new multiplex tools can be used for immune profiling and monitoring, with a particular focus on cancer. Tissue damage initiates wound repair, a type of disruption that provides insight into the rules that underpin tissue architecture, he commented. The immune system is intimately involved in wound repair, which is a highly dynamic process involving cellular rearrangement. Thus, a view of the immune system as a dynamic and mobile organ is an instructive starting point, he said. Multiple tools are now available, such as mass cytometry and imaging versions of single-cell analysis: multiplexing, ion beam imaging, and a new split-pool approach with potential to substantially reduce the cost of single-cell RNA and protein analysis and ATAC-Seq,1 Nolan described. At the most fundamental level, this work is about the tissue architecture as defined not only by the cell–cell interactions but also by the tissue context. Until recently, the tools and mathematics to understand those issues have not been available. “Tissue building blocks”—also called “tissue schematics”—can be used to elucidate the essential rules of the immune system’s operation in both normal and pathologic circumstances, which can then be used to extract meaning and (eventually) identify therapeutic targets, Nolan explained.
CODEX: CO-Detection by indEXing Tool for Multiplex Imaging by Reannealing
Nolan and his team developed the CO-Detection by indEXing (CODEX) multiplexing microscopy tool, which is based on reannealing groups of oligo-fluorophores to precisely image tissues and characterize cell types (Goltsev et al., 2018). The CODEX approach is relatively straightforward and consists of several steps, Nolan explained. The first step is to stain
___________________
1 Assay for Transposase-Accessible Chromatin with high-throughput sequencing.
with 50–120 antibodies, each of which has a unique DNA tag. In groups of three, oligonucleotides, each of which is antisense to a DNA-labeled individual antibody and has a unique fluorophore label, are annealed, and then imaging is conducted at the desired level of resolution. Next, the oligonucleotides are removed by simple denaturation, and subsequent sets of oligo-fluorophores are annealed and imaged iteratively. The process is performed by a robot that moves the reagents and the oligonucleotides into different chambers where the chemistry occurs. Advantages of the CODEX approach include its integration with existing microscope platforms, making it broadly accessible to more researchers, and that it does not require coordinating with a tissue analysis core, he said.
In the current era, collecting data through the CODEX approach and analysis methods is relatively straightforward; rather, the primary challenges are data analysis and extraction of meaning from large datasets, Nolan remarked. In the context of cancer, the CODEX tool is more predictive due to the insight acquired into how cells are organized, which is more instructive than any individual set of markers identified using simpler modalities (Lu et al., 2019). For instance, adding multiplex immunohistochemistry to other datasets—such as gene expression profiling, programmed death-ligand 1 (PD-L1) immunohistochemistry, and tumor mutation burden—can yield a higher area under the curve (AUC) in ROC curve analysis (Lu et al., 2019).2 It may also be especially informative to combine a multiplexing microscopy tool like CODEX with traditional RNA sequencing analysis.
To date, the CODEX system has more than 130 validated DNA-bar-coded human antibodies (Bhate et al., 2021; Hickey et al., 2021; Phillips et al., 2021; Schurch et al., 2020). Nolan explained that they start by identifying high-level phenotyping markers that distinguish well-known cell types, as well as tissue-specific or intracellular markers, such as phospho-specific markers. These markers are used to further define both the cell phenotype and the activation state (e.g., if the cell is dividing, whether it has recently seen an antigen) based on whether those markers are positive or not in a given cell.
Cellular Neighborhoods
Nolan introduced the model framework of cellular neighborhoods (CNs) to explain how the CODEX tool can be used to understand cancer and other pathologies. CNs are defined by a characteristic local composition of cell types. In neighborhood analysis, an analysis window passes across CODEX-derived images of tissue to assess the composition of cell
___________________
2 Receiver operating characteristic (ROC) curve analysis is a classical method of evaluating the predictive success of a classification model.
types present based on standardized cell determinations and established algorithms. If a set of cell types is found repeatedly across a unit, it is defined as a neighborhood, and tissues typically have 10–15 CNs (Bhate et al., 2021; Phillips et al., 2021; Schurch et al., 2020). These CNs typically coincide nicely with what pathologists are accustomed to seeing with traditional histology, Nolan noted. The next steps of analysis are to (1) examine how the different CNs interact with each other; (2) identify any rules by which they interact, not only as neighborhoods but as cells within the neighborhood; and (3) explore whether there is anything distinct about the interface between CNs.
A variety of mathematical approaches can be applied to determine whether there are standardized rules and interactions that seem to be obvious between cells (Schurch et al., 2020). The interactions themselves are not observed as such; rather, they are inferred based on repeated behavior or the repeated incidence of cell context, which implies the potential importance of the cell’s presence (or not), Nolan elaborated. When the data indicate an interaction exists, Nolan and his colleagues search for literature evidence to support the proposed interaction and apply advanced statistical methods (e.g., tensor analysis) to derive meaning from it. One of the objectives of this process is to determine whether there are CNs, or cell types within neighborhoods, that might predict survival outcomes or some other mechanistic determination, said Nolan. Importantly, this work extends beyond mapping to extract information about underlying dynamic behavior from a static image, he emphasized. The various types of inter-CN relationships give rise to the most compelling aspect of this research because the relationships may broaden understanding of how changes in the dynamic processes that underlie pathology relate to a positive or negative resolution, Nolan remarked.
Lymph Node Cellular Neighborhoods
To demonstrate how relational rules can be extracted from CN analysis, Nolan used the example of lymph nodes from the tonsil and spleen, because they provide a standard baseline of a nonactivated immune system. The lymph nodes across the human body are evolutionarily similar, yet the tissues have structural differences, he noted. Conducting a CN analysis on those lymph nodes revealed about ten different neighborhoods, identified based on the primary cell type (e.g., granulocytes, B cells, T cells, macrophages) that seems to be reflective of the neighborhood itself (Schurch et al., 2020). Although the overall set of CNs is similar across lymph node subtypes, within a particular CN, differences in the presence or absence of various cell types can be identified, thereby providing means to distinguish between the lymph node subtypes (Schurch et al., 2020).
To determine rules from cellular relationships, the next step is to consider each CN and identify how it relates to other neighborhoods by identifying common and unique adjacencies. These rules can be used to build a tree structure to visualize observable inter-CN relationships that seem to be repeated within individual lymph node types (Schurch et al., 2020). Some rules are common across the various lymph nodes. There are tissue-specific structures as well as evolutionarily conserved structures that seem to be present or necessary across all the different lymph node types, Nolan emphasized. These rules are then used to examine whether those same types of structures also occur surreptitiously or adventitiously in pathology tissue in the case of cancer, for example.
Case Study: Colorectal Tumor Immune Microenvironment
To illustrate how cellular neighborhoods can be used to study pathology, Nolan and his team conducted a case study of the colorectal tumor immune microenvironment (Schurch et al., 2020). For the study, they narrowed down a large cohort of colorectal patients to identify two extreme classes of disease that are associated with different patient outcomes, Nolan explained. One category included patients with a diffuse inflammatory infiltrate that appears disorganized and who tend to have poor outcomes. The second category was characterized by Crohn’s-like reactions, in which a follicle is visible, with an immune system infiltrate that seems organized; this group of patients is associated with better outcomes.
For each tumor, a pathologist examined the tumor–immune interface to distinguish different ecosystems that might represent the total ecosystem of the tumor, stated Nolan. Four samples from each tumor were analyzed in a tumor microarray to search for rules. Researchers stained with approximately 60 markers, determined the cell types, and conducted neighborhood analysis. The initial expectation was that the two classes of tumors would have very different neighborhoods; surprisingly, this did not prove to be the case, he said. Regardless of how the tumors were clustered, the same kinds of neighborhoods were observed; however, the organization of the neighborhoods was extremely different, he explained. In patients that do poorly, CNs exhibit fragmentation, suggesting that the immune system is not allowed to become organized. This insight may be valuable in the context of tissue regeneration, he added. For example, there may be instances of pathology in which the immune system is unable to form a coherent structure that would enable it to carry out its function.
Nolan and his colleagues initially focused on whether they could relate any kind of cell-type observations to cancer outcome. In short, the presence or absence of certain cell types in the tissue was not reflective of outcome. However, investigating the presence or absence of certain cell types within
certain definable neighborhoods suddenly revealed an entire litany of relationships that were positively or negatively predictive of patient outcome. Thus, cell type matters, but cell type in a particular context matters more, Nolan emphasized.
The final step in this process is to organize the observations into relationships to determine how the cell types are organized relative to each other. Specifically, the interfaces between neighborhoods and the changes in the cell types of those neighborhoods reflect the biology, so the upregulation of cells at the interface is just as important as the presence or absence of the cell type, explained Nolan. The CN interfaces provide information about the biology that enables a specialist with domain knowledge to determine what they mean locally. Furthermore, this organizational information has prognostic value and can become reflective of a prediction and an outcome, Nolan said. Local CNs and the relationships between them can be used to predict which type of tumor a patient has, which may influence therapeutic options, he suggested. In the context of regenerative medicine, mapping the organization of pathologies could inform the development of novel therapeutics to enhance tissue regeneration processes.
ENGINEERED IMMUNITY AS A MODEL FOR REGENERATIVE MEDICINE
Michel Sadelain, Stephen and Barbara Friedman Chair and Director of the Center for Cell Engineering at the Memorial Sloan Kettering Cancer Center, discussed therapeutic tools with potential applications in the field of regenerative medicine.
Mastering T-Cell Responses
Novel immunotherapeutic tools emerged with the rise of engineered T cells as cancer drugs. T cells engineered with a chimeric antigen receptor can delay tumor progression and, importantly, be a curative modality, Sadelain remarked. Realizing this curative potential requires mastering several fundamental components of biology: (1) how to harness T-cell specificity, (2) how to support the functional persistence of T cells, and (3) how to achieve the potency needed to overcome tumors and the tumor microenvironment.
T cells form in the thymus, where they acquire their T-cell receptors (Janeway et al., 2001). The T-cell receptor is generated through the recombination of hundreds of genes that can form a vast number of potential T-cell receptors, Sadelain explained. The repertoire of receptors is pruned as it is formed to obey the laws of immune tolerance, with the average adult human harboring an estimated 20 million different specificities (Chen et al., 2017).
The medical field has long attempted to direct the T-cell repertoire through a range of approaches. The most well-established strategy is active immunization, which involves trying to amplify a subset of T cells that would protect against a future infectious disease. A more recent approach, harnessed specifically in the domain of cancer, is known as checkpoint blockade immunotherapy, which uses antibodies to relieve inhibition on tumor-reactive T cells. A third approach, chimeric antigen receptor (CAR) T-cell engineering, employs genetic engineering to instruct T-cell functionality and is the primary focus of Sadelain’s work.
Assembling Chimeric Antigen Receptors for Cell Therapy
Sadelain provided an overview of the process by which CARs, or synthetic T-cell receptors, are assembled for cell therapy (Riviere and Sadelain, 2017). Incorporating knowledge from tumor biology and the immunology of T cells, the approach is heavily predicated on genetic engineering and takes advantage of an expanding array of tools to modify T cells with synthetic receptors. CARs substitute for physiologic antigen receptors (i.e., T-cell receptors) to instruct the function of engineered T cells, he explained. Rather than engage human leukocyte antigen (HLA) peptide complexes as the normal adaptive immune system does, CARs engage cell surface molecules like CD19.3 Given initial skepticism about the potential value of this form of immunotherapy, academic investigators had to develop their own cell-manufacturing sciences to introduce genes into T cells and regulate them up to FDA standards, Sadelain said. At the outset of the manufacturing process, T cells were typically retrieved from a patient through apheresis. The cells were then genetically modified as they were activated and slightly expanded, quality controlled, released, and infused back to the recipient (Hollyman et al., 2009).
Clinical History and Impact of CD19 CAR Therapy
The CD19 paradigm demonstrated the therapeutic potential of CAR immunotherapy in the clinic, Sadelain observed. The two first CARs for immunotherapy were approved by the FDA in 2017 to treat certain B-cell malignancies: CD28 CAR (axicabtagene ciloleucel) and 4-1BB CAR (tisagenlecleucel). Both prototypical CARs are specific for the molecule CD19, which had been previously identified as a potential target for CARs (Brentjens et al., 2003). Both CARs comprise an array of signaling domains that are not normally found on the same molecule. The molecular structures
___________________
3 CD19 is a transmembrane protein expressed by normal B cells until differentiation and by most B-cell malignancies (Hollyman et al., 2009).
include an activation domain and one or more costimulatory domains that further amplify the signal and impart a different functional profile to the genetically enhanced cells (Maher et al., 2002), Sadelain elaborated. Three clinical trials that were conducted at three separate academic medical centers established the therapeutic value of the first CD19 CARs by demonstrating rapid and complete eradication of refractory leukemia by 1928z CAR T cells (Brentjens et al., 2013; Couzin-Frankel, 2013; June and Sadelain, 2018). Sadelain highlighted several major clinical impacts of CD19 CAR therapy, the first gene therapy to be approved in the United States and the first engineered T cell to be approved worldwide (see Box 7-1). Perhaps most importantly, the success of the CD19 treatment paradigm convinced the pharmaceutical and biotechnology sectors to contemplate manufacturing cells as drugs, moving beyond chemicals to pave the path forward for novel cellular therapeutics, he said.
Today, work on developing these types of therapies has expanded to the extent that 700 CAR trials are listed at ClinicalTrials.Gov as of March 2021, with 40 percent of those targeting CD19 (Globerson Levin et al., 2021). This percentage is likely to decrease over time as new targets are brought to bear, Sadelain predicted. Yet, many researchers still choose to include CD19 in initial tests of new manufacturing strategies, CAR designs, molecules, or combinations of molecules. He added that the vast majority (about 95 percent) of interventional CAR clinical trials are being conducted in the United States and China (MacKay et al., 2020).
Therapeutic Potential of Senolytic CAR T Cells in Regenerative Medicine
In addition to providing a paradigm for the development of CAR T cells, advances in CD19 CAR therapy have potential to open new avenues
for other cell therapies, said Sadelain. These include therapies targeting other hematological malignancies (e.g., myeloma, acute myeloid leukemia), solid tumors, severe infections, and autoimmunity or even inducing tolerance in organ transplantation.
CAR T cells also have prospective applications in regenerative medicine, Sadelain said. Specifically, while investigating the potential to use CAR T cells to remove senescent cells, he and his team conducted a search for surface molecules induced by senescence and identified the urokinase plasminogen activator receptor (uPAR) as a means to target senescent cells (Amor et al., 2020). Since CARs engage molecules (e.g., proteins, carbohydrates, glycolipids) on the cell surface, the researchers examined the surface profiles of cells induced into senescence through a variety of stress triggers and found that the protein uPAR is a biomarker of senescence. In various murine models that induce senescence and fibrosis in the liver, a single infusion of senolytic CAR T cells targeted to uPAR not only removed the senescent cells but also restored tissue homeostasis in senescence-associated fibrosis (Amor et al., 2020; Wagner and Gil, 2020).
Senolytic CAR T cells thus represent a novel tool to leverage the engineering of immune cells—including T cells but also natural killer cells, macrophages, or perhaps neutrophils—by targeting them through genetic engineering, assessing their function on clearance of senescent cells, and evaluating their ability to facilitate or induce organ regeneration, Sadelain suggested. Regenerative medicine can reciprocally offer new prospects for the development of tools and therapies to the field of T-cell engineering, he noted. For instance, CAR therapy typically relies on autologous T cells collected from patients, but investigators are interested in alternatives like collecting cells from healthy volunteers or using pluripotent stem cells. The field of regenerative medicine has generated knowledge about working with these cell types that could also serve the development of CAR therapies.
Generation of CAR T Cell In Vitro from Pluripotent Stem Cells
In 2013, Sadelain and his group provided a first proof of concept that the CAR T cells could be generated in vitro starting from pluripotent stem cells (Themeli et al., 2013). They were able to reprogram T cells to a pluripotent state or ground-state level, introduce therein a CAR that was constitutively expressed, and then re-induce lymphoid populations from these pluripotent stem cells. They have since made considerable progress in defining the requirements for these differentiations, he noted.
The induction of T cells from pluripotent stem cells is not straightforward because T cells are not uniform and acquire different properties that define distinct lineages or commitments in their fate, explained Sadelain. Furthermore, this differentiation is dictated by the T-cell receptor that is
uniquely acquired in every clonotype. Despite these complexities, methods like single-cell RNA studies can now elucidate the sequence of steps that the pluripotent stem cell undergoes until it becomes a CD8 T lymphocyte. Today, T cells can be induced to both express a CAR and no longer harbor their endogenous T-cell receptor; these cells look more like physiological CD8 T cells than gamma-delta T cells or other immune cells, Sadelain described. To further direct the fates of T cells and their corresponding therapeutic profiles, the researchers have successfully induced CD19-specific CAR T cells that display a phenotype similar to effector memory T cells. These cells have exhibited therapeutic effectiveness in models of leukemia.
In the context of advancing regenerative medicine, Sadelain emphasized a key lesson from the development of CD19 CAR therapy. The field of cell therapy is no longer limited to isolating and expanding naturally occurring T cells. Instead, it is feasible to begin considering design and manufacturing approaches for (1) overcoming immune tolerance, (2) dictating to T cells which antigens to recognize, and (3) controlling their effector functions and durability in vivo. In closing, Sadelain expressed his hope that these tools will be of use to advance regenerative medicine.
BASIC IMMUNOLOGY TO GUIDE REGENERATIVE THERAPEUTIC DESIGN
Kaitlyn Sadtler, Earl Stadtman Tenure-Track Investigator and chief of the Section on Immunoengineering at the National Institute of Biomedical Imaging and Bioengineering, discussed how basic immunology can guide regenerative therapeutic design, noting that it is now possible to apply an understanding of mechanistic biology to guide the design of biomaterials.
Immunoengineering in Human Health and Disease
The various functions of the immune system in human health and disease give rise to the potential of immunoengineering (i.e., engineering the immune system) to promote health, Sadtler remarked. In addition to playing a critical role in defense against pathogens, the immune system recognizes and responds to implanted materials from medical devices (e.g., breast implant, hip replacement) through the foreign body response, she explained. In the context of wound healing, the immune system determines whether scar tissue is formed or functional tissue is regenerated. Therefore, engineering the immune system can have broad impact across a variety of fields including tissue engineering and regenerative medicine, but also cancer therapeutics, autoimmunity, medical device design, and others, she said.
Scaffold Immune Microenvironment
The tumor immunology field developed the concept of an immune microenvironment, which has now been adopted by other fields, Sadtler remarked. The immune microenvironment of a biomaterial scaffold can alter the proliferation and differentiation of stem and progenitor cells. Her laboratory and others are looking more deeply into the immune response to the injury and to the material implantation. Both the location of an injury and the type of implanted material can affect immune cell recruitment, activation, and polarization, creating a varied repertoire of different cells and signaling molecules that can then ultimately interact with stem cells, explained Sadtler. In a pro-regenerative environment, this can lead to functional tissue development. Alternatively, an adverse environment could cause pathologic outcomes, such as excessive inflammation or fibrotic scarring.
Immune Cell Polarization
Polarization of immune cells is the process by which components of the immune system (e.g., macrophages, T cells) functionalize to home in on the right response to the challenge at hand, Sadtler explained. Importantly, the appropriate response in one situation may not be appropriate in another situation. For example, “good” type 1 inflammation can clear virally infected cells to prevent the spread of infection whereas “bad” type 1 inflammation can prevent proper healing, she elaborated. Similarly, “good” type 2 inflammation can promote extracellular matrix deposition and muscle wound healing, but “bad” type 2 inflammation can result in pathogenic scarring and thick fibrotic scar tissue deposition.
Biologically Oriented Therapeutic Development
Developing specific, targeted therapeutics depends upon an understanding of basic biology and relying on biology as the foundation from which to create rationally designed materials from the bottom up, Sadtler suggested. A prime example of engineering the immune system based on understanding biology is checkpoint blockade immunotherapy, she noted. By understanding the basic biology of how T cells regulate their responses to prevent overreaction, engineers and biologists created a therapeutic that could block these interactions in patients with tumors that pathologically upregulated the suppressive responses. In the context of wound healing, understanding the biology of how the body responds to biomaterial implants could be used to develop targeted therapeutics. Sadtler emphasized that such work requires multidisciplinary effort that integrates knowledge
across fields (e.g., wound healing, developmental biology, stem cell biology, immunology, materials science, bioengineering, chemistry).
Modeling the Poles of Innate Immune Responses with Representative Materials
To evaluate immune responses to implanted biomaterials, Sadtler and her team model opposing “poles” of immune responses using two representative materials—one pro-fibrotic and one pro-regenerative—implanted in a mouse model. The pro-fibrotic material is polyethylene, which is highly hydrophobic and nondegradable, and produces constant inflammation and fibrosis. The pro-regenerative material is extracellular matrix (ECM), which is biologically derived and degradable, and used clinically in hernia repair, dural repair, and diabetic ulcers. ECM scaffolds have even shown promise in complex trauma repair like muscle injury, she noted. In comparing the two poles of immune responses, there is a clear shift among granulocytes from a neutrophil-dominant phenotype in a pro-fibrotic or pro-inflammatory material to an eosinophil-dominant response in the ECM material (Sadtler et al., 2019). Eosinophils (Siglec-F+ cells) are present in type 2 immune responses, such as allergies, and neutrophils (Ly6G+ cells) are characteristic of viral and bacterial infections, she added.
Other laboratories have also advanced the understanding of these phenotypes, said Sadtler. Ed Botchwey and his team, for example, have identified functionally diverse subpopulations of neutrophils that respond to tissue defects (Turner et al., 2020). In a project led by Josh Doloff, Bob Langer’s research group showed that macrophages—but not neutrophils—were required for fibrosis in the foreign body response and that a macrophage inhibitor could reduce fibrosis (Doloff et al., 2017). Finally, James Anderson and his colleagues published seminal work on the role of macrophages in fibrotic foreign body responses to biomaterials (Anderson et al., 2008).
Role of T Cells in Adaptive Immunity
Expanding their considerations from innate to adaptive immunity, Sadtler and her colleagues also examined the role of T cells in wound healing and found that interleukin 4 (IL-4) upregulation is dependent upon adaptive immune cells, especially CD4+ cells (Sadtler et al., 2016).4 In the context of muscle injury and pro-regenerative biomaterials, the loss of adaptive immune cells—specifically Th2 cells and the protein IL-4—leads to (1) a loss of a pro-regenerative macrophage phenotype and (2) a massive
___________________
4 CD4+ cells are a subtype of helper T cells (Sadtler et al., 2016).
imbalance in cell differentiation and muscle healing (Sadtler et al., 2016). In mouse models without adaptive immune cells, small, irregularly shaped muscle fibers and substantial intramuscular adipose deposits are observed after a muscle injury, described Sadtler. The phenotype can be rescued by supplementing mice that lack an adaptive immune system with wild-type T cells. However, if the supplemental T cells cannot polarize via the Th2 pathway, the pro-regenerative phenotype is not rescued; establishing Th2specific polarization is required, she emphasized.
Sadtler highlighted other efforts to explore the role of adaptive immunity in wound healing. For instance, Ajay Chawla’s research group found that eosinophil activation and type 2 immune signals were necessary for muscle regeneration after a cardiotoxin injury (Heredia et al., 2013). Steve Badylak and his colleagues demonstrated that ECM integration into a muscle defect correlated with the presence of M2 polarized macrophages (Badylak et al., 2008). Bryan Brown’s lab modified polymers to elute IL-4 to promote implant integration and minimize type 1 inflammation at the site of implant (Hachim et al., 2017). Dave Mooney and his team showed that IL-4–functionalized gold nanoparticles could promote the recovery of muscle after injury (Raimondo and Mooney, 2018).
Human Immune Responses to Trauma and Recovery: Learning from Clinical Data
While the context of mouse models can be useful, Sadtler emphasized the importance of learning from clinical models and patients to inform animal models of disease, and vice versa. For a large biomarkers study, her team is building a database of clinical data to evaluate the systemic immune status of patients at admission for various types of traumas. Researchers can use the database to consider demographic information that correlates with trauma outcome, identify proteins that might predict recovery, and inform the design of biomaterials for treatment of an injury. Other laboratories have also explored biomarkers and large datasets to better understand the injury environment, she added. Garry Nolan and his team used a computational approach to identify different cell types associated with surgical recovery (Gaudilliere et al., 2014). Robert Guldberg and his colleagues detected biomarkers that correlated with bone regeneration after trauma (Cheng et al., 2021), while Jennifer Elisseeff’s research group utilized single-cell sequencing to define new immune cell subpopulations in response to biomaterials (Sommerfeld et al., 2019).
Examining the Future of Immunoengineering and Regenerative Therapeutic Design
Sadtler underscored the importance of applying basic biology and integrating various types of data to guide regenerative therapeutic design. To create better therapeutics, it may be valuable to consider the existing approach to solving these fundamental problems, how studies are currently designed, and how to untangle the basic biology to learn from the immune system, she explained. Given animal models and a set of preliminary materials, it is possible to assess the tissue structure and the basic immunology of the materials as well as apply computational approaches to further study them. This information can be aggregated in a database of known responses and outcomes and applied to the clinical development of materials; that clinical knowledge can also augment development of preclinical models, she said. Eventually, the aim would be to start with quantitative modeling and perturb the computational system to eliminate approaches that are unlikely to be successful before testing them in vivo or in the clinic, Sadtler described.
There are multiple approaches underway to integrate these data and engineer biomaterials for immune-guided regenerative therapeutics—for example, to prevent fibrosis of the medical devices used in reconstruction and to grow back new tissues using regenerative therapeutics, she explained. Another promising avenue is to combine quantitative modeling with patient data to develop individualized therapeutics, Sadtler said. It may be possible to predict the outcome of an intervention based on the patient’s biology and use in vitro platforms to evaluate therapeutic design and dosing on an individual-by-individual basis. In addition to precision medicine, which can be cost prohibitive, broad-reaching therapeutics are important for promoting human health, noted Sadtler. To drive efforts of bioengineering for human health, the National Institute of Biomedical Imaging and Bioengineering recently announced the launch of the Center for Biomedical Engineering and Technology Acceleration within its intramural research program. The initiative aims to drive innovative science through diverse people with diverse minds to catalyze the development of new technologies for human health.
DISCUSSION
Effect of Biomaterial on Immune Response
Kassim asked Sadtler to elaborate on the model she presented framing biomaterials as pro-regenerative versus pro-fibrotic. He asked if pro-regenerative biomaterial is always pro-regenerative, or whether it is context
specific. Sadtler replied that it is context specific, and there are also limitations based on material properties, such as mechanics. In unpublished studies, her group compared a subcutaneous injury, a muscle injury, and an intraperitoneal implantation. The intraperitoneal implant exhibited more B-cell recruitment than the other injuries did, indicating different immune responses to the same material, she explained. Furthermore, clinical experience shows that immune response differs based on the location where a material is placed (e.g., adjacent to fat versus muscle), and some organs can sustain a more robust immune response than others, she added.
Sadtler was asked whether immune responses change when progenitor stem cells are added to biomaterials before implantation in the host. She replied that her group has not investigated that directly because their work focuses primarily on endogenous repair, but she noted that stem cells themselves can be highly immunomodulatory. For instance, MSCs are known to alter the immune response associated with a material. Sadtler added that researchers at the University of Florida have studied allogeneic encapsulated islet cells to treat type 1 diabetes and found that encapsulation of the cells altered the immune response as well (Stabler et al., 2020). Cell-laden therapeutics should be evaluated to understand how the cells and their secretome might affect surrounding immune cells, Sadtler said.
CAR T Therapy: Conditioning and Mechanistic Functioning
Noting that lymphoid depletion and conditioning are important aspects of CAR therapy efficacy, Kassim asked Sadelain whether similar conditioning to create a receptive environment would be required for senolytic CAR treatment. Sadelain replied that conditioning—sometimes called lymphodepletion—is a key component. It is well established that T cells cannot simply be infused into an immunocompetent recipient to achieve tumor rejection; it requires some form of prior conditioning, such as a short pulse of chemotherapy. Reducing the number of host lymphocytes increases the chance that available cytokines will support the CAR T cell rather than host T cells, he said. Conditioning has many other effects, including on the endothelium and the gut, and may reduce tumor burden in some cases. Further study in humans and experimental models would be beneficial to understand optimal conditioning, Sadelain suggested.
Kassim then asked what is known about the mechanism of action of senolytic CARs. Sadelain responded that this question cannot be fully addressed yet within the nascent field of senescence, partially due to the heterogeneity of senescent cells. In terms of CAR T cells, re-engineering provides them with new means to engage a different range of targets. Accordingly, the engineered cells may not engage with antigen-presenting cells because they do not recognize HLA; rather, their action is more restricted to
the intended target. He added that with genetic engineering and other tools “we’re starting to bend the rules” to develop novel therapeutics.
Role of Costimulatory Domains in Modulating the Function of CAR T Cells
Sadelain was asked about differences in costimulatory domains to modulate the function of CAR T cells for immune suppression, specifically with respect to regulatory T cells compared to antitumor CAR T cells or senolytic CAR T cells. He explained that the costimulatory domain is critical in shaping and programming functions into an engineered T cell, the most classical being the two canonical CARs that use either the CD28 or the 41BB costimulatory domain. The CD28 domain creates an “explosive” effector cell endowed with maximal effector functions, but this comes at the expense of longevity of the cells, which rapidly proceed into terminal differentiation. Other costimulatory domains, 41BB being the prototype, program a weaker effector profile, but they enhance the persistence of those cells. Research is underway to find additional costimulatory molecules that may further fine-tune the properties of therapeutic T cells, he noted.
Comparative Morphology with Multiplexing Tools
Kassim commented that the CODEX platform has been used to perform comparative morphology between mouse and man. He asked about any themes that have emerged from this work, such as similarities between the neighborhoods of mouse and human spleens—and about the extent to which mouse models can be used to derive fundamental observations about the human CNs. Nolan replied that his group has comparatively analyzed models from human and primate to mouse. Although there were individualized differences, at a global level, the CNs were relatively similar, so based on this work, they are beginning to use neighborhoods to define the functional correspondence of cell types across species, he said. He noted that these results follow the same premise as the tissue disorganization observed in certain kinds of cancers. Together, these studies suggest a fundamental rule of tissue organization; his group is working to determine other basic rules and how they may be altered to serve a pathologic function.
Composition of Cellular Neighborhoods within the Same Tumor Type
Nolan was asked whether he and his lab have observed any differences in the composition of CNs of the same tumor type in an individual patient. He said that they have not investigated this in patients but considered a related question in a mouse model of melanoma. If a lymph node
is analyzed before a tumor enters the tissue, there are no differences in the cell types. However, rearrangement of the cells is detectable, which indicates either (1) the tumor signals to the lymph node ahead of its arrival or (2) the immune system already recognizes the tumor. When the tumor enters the lymph node, the cellular organization again rearranges, an effect which Nolan likened to “scattering the barstools in the saloon.”
Cellular Architecture of Immune-Desert Tumor Types
Kassim inquired about features within immune-desert tumor types that may point to unique neighborhood architectures that correlate with absence of immune infiltration. Nolan responded that they have made such observations, which appear to be driven by the presence of the tumor itself, as if the tumor creates an immune exclusion zone. It appears that when a tumor inserts itself into the tissue, it deactivates positive aspects of the immune response organization such that it creates that immune-desert environment, he said.
Use of CODEX to Image 3D Tissues
Nolan was asked if the CODEX platform can be adapted to stain and image optically cleared three-dimensional tissues. He confirmed that it can, if the tissues are not too thick and the appropriate microscope is used. The primary challenge is that thicker tissues require more time to flush one set of oligo-fluorophores out and reanneal another set so the speed is more difficult to achieve. While some research applications may benefit from three-dimensional analysis, Nolan emphasized that the multiplexing microscopy tool in combination with the appropriate analysis methods can capture the biological dynamics of three dimensions with a simple two-dimensional slice. The next steps will be to move to a sample thickness of 50–100 microns, which would not require complex optics to process, he said.
Developing New Tools and Preclinical Models
Kassim asked each presenter for their perspectives on priorities in developing tools and preclinical models for both monitoring the performance of regenerative medicines and using them to create more effective therapies. In the context of early-stage translational science, Sadtler discussed the potential benefits of collecting more clinical data and conducting more thorough analyses of human responses to materials and trauma. The integration of knowledge across various fields working on relevant biology will be important moving forward, she predicted. For example, different fields have worked on immunology in various contexts such as scarring and
skin wound healing or type 2 immune responses to multicellular parasites. A lot can be gained from bringing together investigators from different specialties, she said.
Sadelain encouraged the development of a cell-based approach to treat conditions that are secondary to the accumulation or poor removal of senescent cells. In both the expanding field of senolytics and other disciplines, there is a tendency to focus on developing highly specific molecular interventions, and chemicals as drugs are enticing due to their convenience, he observed. However, a cell-based approach may be more suitable in the context of complex phenomena, including those explored during the workshop. In those types of phenomena, cells can either aggravate or resolve local immune responses, Sadelain explained. In comparison to chemical drugs, cells can perform multiple tasks: for example, removal, reprogramming or altering macrophages, recruiting immune cells like eosinophils, and giving cues to promote regeneration. He added that cell-based strategies mimic nature’s normal process of removing senescent cells through a cellular mechanism.
Nolan underscored the need to search for organizational rules that govern states of pro-resolution or anti-resolution, beyond characterization or description of the state. Given high-level rules, vulnerabilities could be exploited to alter the immune state, he suggested. This process begins at a high level, in examining cellular organizations; subsequently, RNA sequencing or other approaches can be applied to learn more detailed information about the underlying gene networks, for instance. This type of research has made the need for new classes of therapeutics evident. Although cancer has been a useful starting point for understanding wound repair, he highlighted that the CODEX methodology could be used with other models of wound repair or pathology to search for common rules that guide the immune response. These common rules and cellular complexes might elucidate which immune cells are necessary in a certain organization to carry out a particular function, he added.
