Translational Immunology: Applying fundamental discoveries to human health and autoimmune diseases

Abstract

Studying the human immune system is challenging. These challenges stem from the complexity of the immune system itself, the heterogeneity of the immune system between individuals, and the many factors that lead to this heterogeneity including the influence of genetics, environment, and immune experience. Studies of the human immune system in the context of disease are increased in complexity as multiple combinations and variation in immune pathways can lead to a single disease. Thus, although individuals with a disease may share clinical features, the underlying disease mechanisms and resulting pathophysiology can be diverse among individuals with the same disease diagnosis. This has consequences for treatment of diseases as no single therapy will work for everyone, therapeutic efficacy varies among patients, and targeting a single immune pathway is rarely 100% effective. This review discusses how to address these challenges by identifying and managing the sources of variation, improving access to high quality well-curated biological samples by building cohorts, applying new technologies such as single cell omics and imaging technologies to interrogate samples and bringing to bear computational expertise in conjunction with immunologists and clinicians to interpret those results. The review has a focus on autoimmune disease, including rheumatoid arthritis, multiple sclerosis, systemic lupus erythematosus and type 1 diabetes, but its recommendations are also applicable to studies of other immune-mediated diseases.

Graphical Abstract

Studying the human immune system requires a toolkit that includes high quality, well-curated biological samples from carefully designed cohorts, and cutting-edge single cell approaches such as omics profiling, imaging techniques and antigen-specific assays. Data analysis and interpretation requires a research team combining expertise in bioinformatics, computational biology, immunology and clinical care. Created with BioRender.com.

Introduction

The goal of biomedical research is to understand biological systems so that we can determine the pathogenic mechanisms of disease and intervene in those processes. However, biology is complicated, and the immune system is no exception. To limit the complexity that is found in humans, scientists have relied heavily on model systems, in particular mice and immortalized cell lines, to tackle the fundamental questions of how the immune system works. These studies have yielded a deep understanding of how the components of the immune system function and have revealed genetic and biochemical mechanisms that underpin them. In addition, murine models of immune diseases have paved the way for the development of many immune therapies, from cytokine blocking biologics for the treatment of autoimmune and allergic diseases to immune checkpoint inhibitors for treatment of cancer. Yet, murine models fall short in providing us with the knowledge needed to understand the complex genetic, environmental, and experiential factors that influence the human immune system. The complexity of human immunology makes its study a daunting task, yet if we are to ultimately prevent and cure immune related diseases, it must be undertaken.

In this review, I will focus on why studying the human immune system is necessary, the hurdles that must be overcome to do it and how by reimagining the way we study the human immune system we can accelerate our ability to improve human health. Much of the discussion will focus on how this paradigm shift will impact the study of autoimmune diseases, but these same concepts and approaches can be applied to other immune diseases from allergy or infection to cancer.

The problem as illustrated through the lens of rheumatoid arthritis

Every person is unique. This is true in health as well as disease. Studies of healthy individuals have shown that immune variation between individuals is high, but that immune characteristics within an individual are stable [1]. It is also well-established that multiple factors influence an individual’s baseline immune landscape or “immunotype”. This has been elegantly shown in twin studies where shared immune features in monozygotic, but not dizygotic, twins demonstrate a clear role for genetics, conversely, features that are not shared in monozygotic twins indicate a role for environmental or stochastic factors in determining these specific features [2, 3]. Further study of shared immune phenotypes in dizygotic twins have underscored the importance of shared childhood environments in the development of an individual’s immune landscape [4]. Immune experience adds another layer to human immune heterogeneity as immunotype is influenced by many factors such as age, cytomegalovirus (CMV) status, and chronic inflammatory states [5]. Collectively these findings highlight the need to understand the breadth of immunotypes seen in health, the factors that drive differences across individuals, so that we can ultimately determine where individuals with autoimmunity deviate from immune health.

When studying autoimmune disease in humans, we must consider both the complexity of the pathogenic mechanisms that lead to disease and the immune heterogeneity between individuals with disease (Figure 1). The breadth of immune heterogeneity when viewed in the context of a complex disease like autoimmune disease means that there are multiple combinations and variation in immune pathways that can lead to a single disease. Thus, although individuals with a disease may share clinical features, the underlying disease mechanisms and resulting pathophysiology can be diverse among individuals with the same disease diagnosis. This is exemplified by the natural history of these diseases, the heterogeneity in the response to therapeutic interventions, and the complexity of the genetics of autoimmune diseases.

Figure 1. Genes, immune experience and stochastic factors drive immune heterogeneity.

A single clinical diagnosis can be caused by a combination of factors. The result is that individuals with the same clinical diagnosis are immunologically as well as clinically heterogeneous. Dissecting this heterogeneity will allow us to categorize individual patients and provide improved targeted therapies.

Rheumatoid arthritis (RA) is an example of an autoimmune disease with complexity and heterogeneity. RA, an inflammatory disease characterized by pain and swelling in the joints, is more common in women than men with a 3:1 ratio and can occur at any age, but increases in frequency with age [6]. Many, but not all, individuals with RA have anti-citrullinated protein antibodies (ACPA) and/or rheumatoid factor (RF) [7]. In addition, some individuals may have mild disease, whereas others have severe disease resulting in progressive disability due to joint damage [6]. Genetic factors are known to influence risk for RA. The strongest association is with HLA Class II alleles, for example, the HLA DRB1*0401 increases risk 6-fold [8] but genome-wide association studies (GWAS) have also identified more than 100 genetic variants that confer a modest level of risk [9, 10]. Many of these variants are in genes known to be important in the immune response, including PTPN22, which influences intracellular signaling in immune cells, CTLA4, an immune checkpoint inhibitor, and IL6R, TYK2, STAT4, all key components of cytokine signaling pathways regulating the immune response [11]. Interestingly, risk variants in HLA-DR4 and PTPN22 are linked to the development of ACPA, indicating a mechanism by which they promote autoimmunity in RA. However, genetic heritability only accounts for 60% of disease risk in RA [12–14], indicating a role for environmental factors as well. One environmental factor linked to RA is smoking, which can increase risk two-fold [15]. Also noteworthy is that genetic and environmental factors may interact to promote disease. For example, smoking enhances the combined effect of risk variants in HLA-DR4 and PTPN22 [16]. Lastly, the microbiome has also been implicated as a contributing factor to pathogenesis of RA, adding yet another level of disease complexity and individual heterogeneity [17–19].

Importantly, the impact of disease complexity overlaid with immune heterogeneity results in real-world problems in the treatment of patients with autoimmune disease. No single therapy works for everyone, therapeutic efficacy varies among patients, and targeting a single immune pathway is rarely 100% effective. Again, turning to RA, 30–40% of treatment naïve patients do not have adequate responses to current therapies [20–25]. This is despite tremendous advances in the treatment of RA and the growing numbers of types of drugs available in the clinic. The struggles that we face in treating RA likely derive, in part, from a multiplicity of disease endotypes, each driven by different combinations of immune pathways, thus requiring interventions that specifically target the culprit pathway(s). Moreover, due to the chronic nature of autoimmune diseases, time may also alter the pathogenic features of an individual’s disease. Due to the multiplicity of pathways involved, RA requires us to consider addressing these factors through sequential or combination therapies. Furthermore, like RA, other autoimmune diseases demonstrate similar complexity and heterogeneity with regards to pathogenic mechanisms. Multiple sclerosis (MS) and systemic lupus erythematosus (SLE) have a similar number ( >100) of genetic markers associated with disease [26–30], and multiple environmental factors have been linked to disease risk, including geographic location for MS [31] and air pollution for SLE [32], as well as the bacteria and viruses such as Epstein-Barr virus (EBV) [33, 34]. Similar to RA, there is heterogeneity in the response to immune interventions across the spectrum of autoimmune disease as exemplified in the recent trial of anifrolumab, a monoclonal antibody against type 1 interferon receptor subunit 1 in SLE, where only 47.8% of treated subjects showed a response [35]. Thus, to tackle the real-world problems of treating and ultimately curing autoimmune diseases in humans, we need to embrace the complexity and heterogeneity in order to develop approaches that will allow us to define the pathways and interactions that promote disease within and across individuals.

Overcoming the limitations posed by studying humans

How do we embrace complexity and heterogeneity in the study of human autoimmune diseases? First and foremost, we need to identify the sources of variation and then manage their impact through study design and data interpretation. The two main types of variation with the potential to influence the immune landscape in humans are related to sampling and intraindividual variation (Figure 2). Sampling-related variables have a direct impact on data quality and validity. The very act of obtaining a biological sample and exposing it to external factors is the first source of variation. Sample processing and storage can introduce further alterations and variation. The impact of these processes on potential immunologic readouts should be well understood and considered in study design. To minimize this variation within a given study, it is essential that all samples are handled in the same manner. Ideally, samples should be processed immediately by a dedicated lab with optimized and validated protocols. However, when this is not possible, sample types and assays that tolerate longer delays in processing should be selected and protocols for processing on site simplified to remove variability based on the site or the technician performing the isolation and storage. Importantly, for each study, the ideal protocol based on sample and assay type should be determined and, where possible, used across institutions and studies to allow for comparisons. Factors known to result in intraindividual variation should be considered as well. For example, time of day or year should be considered as a variable needing to be controlled, as there is growing recognition that diurnal variation and seasonality influence the immune landscape and responses within an individual [36–38]. At a minimum, these factors can be controlled for by tracking the time of day and date for each sample collected and considering them as variables during data analysis. However, in the ideal study design, the timing of collection would be consistent across individuals and if a longitudinal study, then at the same time of day for each time point from the same subject.

Figure 2. Designing broad scope versus in-depth studies of human immunology.

Considerations include study questions, subject selection, sample handling and data acquisition.

Subject specific features such as age, sex, race, genetics, and geographic location can lead to interindividual variation. Disease characteristics such as heterogeneity in clinical presentation and course are also sources of interindividual variation. Thus, these subject-specific variables must also be considered in the design and data capture of human studies. The majority of these data can be obtained through the electronic medical record. However, additional processes may need to be put in place to acquire more in-depth, point-of-care information on disease activity and health history. For these processes to be effective and sustainable, they need to be designed to minimize the burden on both the patient and the health care provider, while maximizing the value to the researchers. An overlooked set of subject-related variables that are of increasing importance and interest are socioeconomic and lifestyle related factors. There is a growing understanding that our immune system is influenced by diet, activity, stress, and sleep, factors that are also influenced by socioeconomic factors [39–43]. Again, as with all the variables mentioned above, tracking these variables and including them in the data analysis is critical for the generation of biologically-meaningful data that can be shared with the greater scientific community.

Human Immunology Toolkit for autoimmune diseases

Advances in understanding human immune diseases require improving access to high quality, well-curated biological samples by building cohorts, applying new technologies to interrogate samples, and bringing to bear computational expertise in conjunction with immunologists and clinicians to interpret those results.

Cohort development:

In the past 20 years, many biorepositories have been established and are rich resources for the study of human disease. Examples include the US Department of Defense Serum Repository, which was recently used to demonstrate that EBV infection increased the risk of developing MS [33] and the UK Biobank providing genetic and health information for numerous GWAS focused on autoimmune diseases [10, 44, 45]. Other biorepositories have focused on prospective longitudinal sampling in at-risk cohorts in order to understand the natural history of autoimmune disease, such as those focused on type 1 diabetes in at-risk children (Type 1 Diabetes TrialNet Pathway to Prevention Study [46, 47], TEDDY Study [48], and the BABYDIAB Study [49, 50]). These studies have given us the tools to predict who is high risk for T1D [51, 52], leading to the development of interventions to delay [53], and in the future to find therapies that will prevent, T1D. Prospective longitudinal at-risk cohorts have also been established for RA (SERA [54], SCREEN-RA [55]), SLE (SisSLE [56, 57]), and Crohn’s disease [58]. We need to continue to expand on this early success. As part of this we need to focus on inclusion of historically underrepresented groups. It is clear that disease characteristics, severity, and responses to therapy differ across ethnic, racial, and socioeconomic groups [59, 60]. An example of this is the high burden of MS in the US Black community that has been under recognized in the past [61]. Expanding the communities whom we study will allow us to understand more fully the breadth, immune mechanisms, and clinical features of immune diseases. We also need to prioritize tissue collection where possible, as this will allow us to develop better understanding of disease pathology at the site of inflammation. Ideally tissue collection should be paired with peripheral blood collection, so that we can determine biomarkers that can translate directly to the clinic. Studies of tissues have enriched our understanding of tissue resident cells [62], the patchy nature of islet inflammation in T1D [63–65], and the diversity of synovial inflammation in RA [66]. Development of additional longitudinal cohorts will also be vital for understanding how immunologic characteristics change over time and in the context of disease activity, immune experience, and environment. These longitudinal cohorts require a commitment to long term maintenance and follow-up.

Cohort selection and development and the approach to manage variation should be based on the study goals and the questions to be addressed (Figure 2). Large diverse populations and/or longitudinal cohorts are best for studies with broad scope addressing questions focused on identification of factors that influence disease development, biomarker discovery or understanding disease heterogeneity. Studies investigating the factors driving heterogeneity would require a cohort that reflects the breadth of individuals with a disease, and in this setting, well-matched control cohorts are vital. In contrast, fewer subjects but more selective recruitment based on features of disease, genetics and environmental factors are more appropriate for in-depth studies investigating specific groups of patients, disease mechanisms or the impact of interventions on these mechanisms. For example, the study of a single gene variant may require holding race and other genetic variants constant. Additionally, these in-depth studies provide an opportunity for more complex sampling protocols including collection of multiple sample types and larger sample volumes. With regards to choosing controls, again the study goals and questions are paramount since it has become increasingly evident that a healthy control may not always be the best option. Additionally, consideration must be given to which features should be matched between the controls and study subjects; age, sex, and race are the first choice but specific genetic variants and/or environmental factors may also be relevant. The type of cohort will also differ based on the scope of the study. A broad scope will require large diverse populations, excellent data management, and may require simplified and targeted sample acquisition across the cohort. An example of this is the NIH-sponsored All of Us Research Program with its goal of recruiting one million participants across races, ethnicities, age groups, US geographic regions, disease types, and socioeconomic status [67]. Whereas studies where the intent is to perform deep immune profiling will require well-curated, high-quality samples with linked genetic, immunophenotyping and functional analysis, and may have fewer subjects due to scope and cost. An example of this approach is the Milieu Interieur Project focused on defining the naturally-occurring variability of the human immune response in one thousand healthy individuals [68]. Specialized cohorts of individuals with rare or monogenic diseases are also useful to highlight specific immune pathways and processes. Additionally, cohorts of individuals treated with immune interventions can be a resource to understand immune perturbations effects in vivo; this has been a focus for many of the clinical trials conducted by both the Immune Tolerance Network and the Type 1 Diabetes TrialNet [69–71]. Lastly, Experimental Medicine Units (EMU) allow short mechanistic studies with in-depth immune monitoring and targeted immunologic endpoints [72–74].

Single Cell Technologies:

Recent advances in single cell technologies and computational methods for data analysis are moving human immunology from descriptive to mechanistic. Single cell profiling is providing unprecedented insight into the fundamentals of human immunology with the discovery of new cell types, characterization of cell-cell interactions, and lineage trajectories and is allowing us to characterize pathogenic cell types [75]. Notably, the information generated through single cell studies has the potential to identify rare pathogenic cell population or changes to cell populations that are predictive of development of disease and/or disease progression. This knowledge is essential for the development of therapies specifically targeting the cell populations and pathways involved in pathogenesis, as well as for the identification of biomarkers that predict disease progression and response to therapy.

Omics Based Approaches:

There has been an explosive growth in the number of available single cell technologies giving us the ability to assess the following at the single cell level: cell surface proteome, transcriptome, chromatin accessibility, and DNA methylation, as well as the metabolome and lipidome [75]. The most recent advance has been the development of multimodal assays for simultaneous measurement of two or more data types [75–85](Table 1). These approaches have already yielded important insights. Combining single cell RNA sequencing (scRNA-seq) and genotype data has enabled mapping of expression quantitative trait loci (eQTL) to specific cell types in several autoimmune diseases, including SLE [86, 87]. scRNA-seq has discovered a novel cell type in the blood and joints of RA patients at the time of flare [88] and cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) has recently been used to define signatures of disease activity in SLE [89]. A recent study used a trimodal approach to map the immune landscape in the human tonsil and then showed that autoimmune-associated genetic variants were enriched in chromatin accessibility regions specific to germinal center-associated T and B cells [90].

Table 1.

Multimodal single cell profiling assays

Multimodal assays	Assays	References
Cell surface protein epitopes + transcriptome	CITE-seq	77
	REAP-seq	78
Transcriptome + chromatin accessibility	sci-CAR	79
	SNARE-seq	80
	SHARE-seq	81
Cell surface protein epitopes + intracellular proteins + chromatin accessibility + mitochondrial DNA variants	ASAP-seq	82
Cell surface protein epitopes + chromatin accessibility	ICICLE-seq	83
Cell surface protein epitopes + transcriptome + chromatin accessibility + mitochondrial DNA	DOGMA-seq	82
	TEA-seq	83
Cell surface protein epitopes + histone modifications	scCUT&Tag	84
Transcriptome + histone modifications	Paired-Tag	85

ASAP-seq: ATAC with select antigen profiling by sequencing; scATAC-seq: single cell assay for transposase-accessible chromatin using sequencing; CITE-seq: cellular indexing of transcriptomes and epitopes by sequencing; DOGMA-seq: done overfitting genomics methods acronym; ICICLE-seq: integrated cellular indexing of chromatin landscape and epitopes; Paired-Tag: parallel analysis of individual cells for RNA expression and DNA from targeted tagmentation by sequencing; REAP-seq: RNA expression and protein sequencing; scCUT&Tag: single cell cleavage under targets and tagmentation; sci-CAR: single-cell combinatorial indexing chromatin accessibility and mRNA; scRNA-seq: single cell RNA sequencing; SHARE-seq: simultaneous high-throughput ATAC and RNA expression with sequencing; SNARE-seq: single-nucleus chromatin accessibility and mRNA expression sequencing; TEA-seq: simultaneously measures transcriptomics (scRNA-seq), epitopes and chromatin accessibility (scATAC-seq)

Antigen-Specific Cell Assays:

The development of assays to identify and characterize antigen-specific T and B cells at the single cell level has also moved the field of human immunology forward. For T cells these approaches include MHC multimers [91–93] and the development of HLA agnostic approaches, such as the activation-induced marker assays [94, 95]. Similar approaches are being used to identify antigen-specific B cells [96, 97]. Importantly, these assays are being combined with single cell profiling techniques enabling an in-depth characterization of antigen-specific cells. This now allows us to identify specificity, clonal expansion, and fate mapping of antigen-specific cells [94, 98–102].

Imaging Techniques:

Imaging techniques are also important tools for studying human immunology. Whole body imaging such as magnetic resonance imaging (MRI), single photon computerized tomography (SPECT), positron emission tomography (PET) and magnetic particle imaging (MPI) are being deployed to track immune cells in vivo [103–105]. Each of these modalities rely on labelling the immune cell of interest with a tracer for visualization. To date, most of this work in humans has focused on molecular imaging of cytotoxic T cells using lineage tracers to assess the response to CAR T cell therapy or immune checkpoint inhibitor therapy in the setting of cancer [103, 105]. Whole body imaging has also been utilized in the setting of autoimmune disease but primarily in animal models. For example, SPECT imaging with nanobodies has been used to monitor macrophages in joint inflammation in the collagen-induced arthritis mouse model of RA [106, 107]. Exciting advances are also being made in the field of tissue imaging with new approaches such as spatial transcriptomics providing an opportunity to map the tissue at a depth not possible with traditional histology and immunohistochemistry. Spatial transcriptomics is unbiased as it profiles the whole transcriptome unlike immunohistochemistry or in situ hybridization, which both require a known target. Relevant to the autoimmunity field, spatial transcriptomics has provided new insight into the cell populations present in synovial tissue in RA during early disease, disease flares and remission [108–110]. It has also been utilized on postmortem brain tissue to track neurodegeneration in progressive MS [111].

Given the plethora of single cell assays, it is an exciting time for human immunology. However, deciding which assay to use can be challenging and is again dependent on the study goals and questions of interest. Other considerations include cost, these are expensive technologies, and how to analyze the data, cutting-edge bioinformatics expertise is essential to make biological sense of the wealth of data generated. Furthermore, although single cell profiling can be useful for discovery, it should always be complemented with functional assays. This is especially important because studying the immune response to perturbations may reveal changes not seen in the resting state. This was the observed in T1D where the TruCulture whole blood stimulation assay [112] revealed type 1 interferon (IFN-1) hyper-responsiveness that was not detected at baseline [113]. Again, the choice of functional assay and stimuli of interest is dependent on the question to be addressed but the assays can be in vitro or directly ex vivo and can test a range of responses from activation in response to antigen, cytokines, and interactions with ligands such as toll-like receptor (TLR) ligands. Lastly, validation is also a requirement for a well-designed study of human immunology, including independent validation cohorts and lab-based assays assessing loss- and gain-of-function. Further validation can also be achieved by mining publicly available datasets.

Concluding remarks

In the past, human immunology studies were given second class status compared to studies using model systems due to the difficulty in proving mechanism and the variability of data generated from human samples. Today we have the tools to overcome many of these past challenges and can now produce high quality data and manipulate human cells and systems in ways that lead to an understanding of disease mechanisms that was not possible in the past. To continue to advance our understanding of human immunology in the next decade, we will need: 1) long term commitments to funding, 2) collaborations across areas of expertise and institutions, and 3) ability to integrate and analyze large data sets. With this toolkit we can now address the questions most salient to patients and their caregivers. Who is at risk for a disease, how can we prevent it, and, if someone has a newly diagnosed disease, what will be the most effective therapy for them to control or reverse the disease process.

Abbreviations:

ACPA

anti-citrullinated protein antibodies

ASAP-seq

ATAC with select antigen profiling by sequencing

scATAC-seq

single cell assay for transposase-accessible chromatin using sequencing

BCR

B cell receptor

CAR-T

chimeric antigen receptor T cell

CITE-seq

cellular indexing of transcriptomes and epitopes by sequencing

CTLA-4

cytotoxic T lymphocyte associated protein 4

DOGMA-seq

done overfitting genomics methods acronym

EMU

experimental medicine unit

eQTL

expression quantitative trait loci

GWAS

genome-wide association study

HLA

human leukocyte antigen

ICICLE-seq

integrated cellular indexing of chromatin landscape and epitopes

IFN-1

type 1 interferon

MHC

major histocompatibility complex

MS

multiple sclerosis

mtDNA

mitochondrial DNA

Paired-Tag

parallel analysis of individual cells for RNA expression and DNA from targeted tagmentation by sequencing

PTPN22

protein tyrosine phosphatase non-receptor type 22

RA

rheumatoid arthritis

REAP-seq

RNA expression and protein sequencing

scCUT&Tag

single cell cleavage under targets and tagmentation

sci-CAR

single-cell combinatorial indexing chromatin accessibility and mRNA

scRNA-seq

single cell RNA sequencing

SHARE-seq

simultaneous high-throughput ATAC and RNA expression with sequencing

SLE

systemic lupus erythematosus

SNARE-seq

single-nucleus chromatin accessibility and mRNA expression sequencing

TCR

T cell receptor

TLR

toll-like receptor

TEA-seq

simultaneously measures transcriptomics (scRNA-seq), epitopes and chromatin accessibility (scATAC-seq)

T1D

type 1 diabetes

TYK2

tyrosine kinase 2

Data availability statement:

Data sharing is not applicable to this article as no new data were created or analyzed in this study.