Académique Documents
Professionnel Documents
Culture Documents
Introduction
Commonly in rheumatology practice, therapeutic deci- responses and as drug targets. Moreover, immune cells
sions need to be made in situations of uncertainty. Even might not only provide information about the status
for rheumatoid arthritis (RA), a well-defined rheu- quo of an immune response, but also about its history
matic disease with established treatment protocols, (for example, by measuring the frequency and specifi
the optimal treatment strategy after a patient has failed city of memory Tcells) and potentially about its future
first-line therapy with methotrexate (Box1) is uncertain. (for example, by measuring cellular responses to invitro
Currently, we cannot predict whether this patient is most stimulation). Other medical specialties, most notably
likely to respond to inhibition of TNF or IL6, to Bcell haematology, oncology and transplantation medi-
depletion, to Tcell co-stimulatory blockade or any other cine, have already demonstrated the broader utility of
pharmacological intervention. A different type of uncer- this approach. For example, flow cytometry analysis of
tainty is encountered in the setting of a poorly defined peripheral blood, lymph node and bone marrow (Box4)
inflammatory condition (Box2), for which data from ran- is routinely used to search for abnormal cell populations
domized controlled clinical trials are lacking and expert that are indicative of lymphoproliferative disorders, 3
opinions diverge. In both clinical scenarios, biomarkers recipients of bone marrow grafts are monitored by flow
would help in choosing the most appropriate treatment cytometry for engraftment and immune reconstitution,
Division of
strategy. This Review focuses on the development of cel- and commercial assays are available to screen heart trans-
Rheumatology, lular biomarkers in rheumatic diseases. We discuss tech- plant recipients for evidence of rejection by analysing
Immunology, and nological advances for the multidimensional profiling of peripheral blood cells.4 Personalized medicine and pre-
Allergy, Brigham and
Womens Hospital, immune cells and consider their utility as discovery tools cision medicine5,6 are concepts that emphasize the need
Smith Building, 1Jimmy and their potential use in clinical practice. to tailor therapies according to insight into the genetic,
Fund Way, Boston,
MA02115, USA (J.E.,
Biomarkers, defined as a characteristic that can cellular and molecular basis of the disease in each indi-
D.A.R., M.B.B.). be objectively measured as an indicator of normal or vidual patient. We are optimistic that this strategy can
Division of Genetics pathological biological processes, or as an indicator of be implemented successfully to improve outcomes for
(N.C.T.), Division of
Rheumatology, response to therapy,1 can be derived from different types patients with inflammatory rheumatic diseases and guide
Immunology and Allergy of data, including genetic polymorphisms, autoantibody treatment decisions in situations ofuncertainty.
(S.R.), Brigham and
Womens Hospital, New
profiles, cytokine levels or clinical parameters (Box3).2
Research Building In immune-mediated diseases, immune cells are particu- High-resolution marker analysis
(NRB), 77Avenue Louis larly promising from the biomarker perspective owing Flow cytometry
Pasteur, Boston,
MA02115, USA. to their central role both as orchestrators of immune Fluorescence-based flow cytometry (Figure1a) was
introduced approximately four decades ago and has
Correspondence to:
J.E. Competing interests been a useful tool for the phenotypic characterization
jermann@partners.org The authors declare no competing interests. and functional analysis of immune cell subsets. Progress
a b
Laser ICP
Ion clouds
MS
100
Normalized excitation/
FITC
PE 100
80
emission (%)
PerCPCy5.5
intensity (%)
Normalized
60 PeCy7
40
20
0 0
Atomic mass
0 400 500 600 700 800 900
Wavelength (nm)
Figure 1 | Cellular immunophenotyping by multiparameter single-cell analysis. a | Fluorescence-based flow cytometry. Cells
Nature Reviews | Rheumatology
are stained with monoclonal antibodies conjugated to various fluorescent dyes. A stream of single cells is guided through a
beam of monochromatic laser light, exciting the dye molecules to emit a characteristic spectrum of less energetic photons
with longer wavelengths. Mirrors and filters direct the emitted light to photomultiplier tubes for quantification. Scattered
laser light provides additional information about cell size and granularity that permits discrimination of the major subsets
of lymphocytes, monocytes and granulocytes. The emission spectra of fluorophores overlap; therefore, signals from one
fluorophore might be detected by more than one channel. This problem can be partially corrected by compensation, but
ultimately limits the number of parameters measurable per cell. b | Mass cytometry. Cells are labelled with antibodies
conjugated to rare earth metals not present in biological specimens. Individual cells are vaporized and ionized by an ICP
torch. The resulting ion cloud is passed into a mass spectrometer to detect and quantify antibody-derived metal isotopes.
Mass cytometry does not require compensation due to distinct mass peaks of the metal isotopes. Abbreviations: Cy,
cyanine; FITC, fluorescein isothiocyanate; ICP, inductively coupled plasma; MS, mass spectrometer; PE, phycoerythrin;
PerCp, peridinin chlorophyll.
cytometry. Transcriptomic profiling became feasible with systemic sclerosis and a subgroup of patients with RA,50
the development of microarray technology in the late suggesting partial overlap of the p athophysiology of
1990s and has been highly successful in the assessment these rheumaticdiseases.
of haematological malignancies, enabling the molecular Multiple studies have sought to identify gene expres-
classification of lymphomas and providing prognos- sion signatures that would predict the response of
tic value.44,45 The basic principle of cDNA microarray patients with RA to treatment with TNF inhibitors,
technology is depicted in Figure3a. rituximab or tocilizumab (reviewed elsewhere 51 ).
Armed with this technology, investigators have Interestingly, two studies found (concordantly) that an
conducted genome-wide gene expression profiling of interferon signature was associated with a poor response
peripheral blood cells in casecontrol studies of many to rituximab.52,53 Results were otherwise heterogeneous,
rheumatic diseases. The most convincing finding was with little reproducibility between studies, probably
the interferon signature in patients with systemic lupus because most of these studies were small, used diverging
erythematosus (SLE),46,47 which consists of a set of typeI response criteria and differed in sample preparation and
interferon-inducible genes that is highly expressed in choice of microarray platform. Furthermore, the analy-
individuals with SLE and is consistent with other lines of sis of whole blood or unfractionated peripheral blood
evidence that suggest that typeI interferon has an impor- mononuclear cells is confounded by the variable abun-
tant function in the pathogenesis of SLE.48,49 An inter- dance of neutrophils, monocytes and lymphocytes in
feron signature was also found in patients with subacute samples from different individuals, which substantially
cutaneous lupus, primary Sjgren syndrome, myositis, affects the gene expression patterns measured in these
FACS94 Intracellular cytokine staining7 5-colour flow CFSE invitro Intracellular staining of phosphorylated 34-marker mass
cytometry97 proliferation assay9 signalling molecules8 cytometry42
3-colour flow cytometry96 cDNA microarrays100 11-colour flow cytometry98 Single-cell RNA-seq61
1969 1977 1984 1988 1991 1995 1996 1997 1998 2001 2002 2006 2008 2009 2011
Technical advances
in flow cytometry
PCR with Real-time 18-colour flow cytometry 37
RNA-seq 58
Technical advances
1-colour flow thermostable quantitative
in gene expression
cytometry using mAb95 Taq polymerase99 PCR (Taqman)101 Tetramer staining of antigen-specific Tcells32 Nanostring64
analysis
Figure 2 | Timeline of technical advances. Advances in flow cytometry and gene expression analysis since 1969.
Nature Reviews | Rheumatology
Abbreviations: CFSE, carboxyfluorescein succinimidyl ester; FACS, fluorescence-activated cell sorting; mAb, monoclonal
antibody; RNA-seq, RNA sequencing.
samples.54 Many investigators have, therefore, started Advances in efficient library preparation from small
to analyse gene expression signatures in sorted cell amounts of RNA have made RNA-seq of individual cells
subsets. For example, in their study of peripheral blood possible.61 Microfluidic devices can separate individual
CD8+ Tcells in anti-neutrophil cytoplasmic antibody cells and automatically perform subsequent biochemi-
(ANCA)-associated vasculitis and SLE, McKinney etal.55 cal reactions in nanolitre-scale volumes, thereby enab
found a gene expression signature dominated by IL7- ling parallel processing of many samples and lowering
receptor-induced and TCR-induced genes that could the cost of consumable reagents.62 Furthermore, bar-
be used to distinguish a subset of patients with a worse coding cDNAs from individual cells permits multiple
prognosis; these findings were subsequently extended to cells to be pooled into a single sequencing reaction,
include patients with Crohn disease.56 Pratt etal.57 found thus further reducing the cost. Potential applications
a STAT3 gene expression signature in CD4+ Tcells of of single-cell sequencing include the analysis of indi-
patients with early arthritis that predicted the develop- vidual rare cells (for example, antigen-specific Tcells
ment of RA, particularly in anti-citrullinated protein identified by MHCpeptide tetramer staining) to detect
antibody (ACPA)-negative patients. Interestingly, the whether there is enrichment of abnormal splice vari-
strength of the gene expression signature correlated with ants or somatic mutations in these cells, or to analyse
the serum concentration of IL6, a cytokine that signals heterogeneous cell populations, as has been done with
throughSTAT3. dendritic cells.63 Single-cell RNA-seq might complement
Technological progress and the plummeting cost of the analysis of individual cells by high-dimensional flow
next-generation sequencing technologies have made or mass cytometry. RNA-seq is not limited by the avail-
gene expression analysis by RNA sequencing (RNA-seq) ability of specific antibodies and can probe the whole
feasible (Figure3b).58 RNA-seq has several advantages transcriptome. Together, these approaches might identify
over hybridization-based microarray technology. First, candidate cell populations or functions in small cohorts
sequencing itself does not depend on prior reference of patients that could then be analysed in larger studies
sequence information; as a result, RNA-seq can be used and, ultimately, with more readily available and less
to detect variant splice isoforms and sequence variations. expensive technologies in the clinic. One such techno
Furthermore, RNA-seq is more sensitive than micro logy to measure the expression of multiple genes in
array technology to rare transcripts, has a high dynamic parallel is the nCounteranalysis system (NanoString
range and is very accurate when compared with the gold Technologies, USA; Figure3c),64 which enables the
standard of quantitative PCR.59 Finally, RNA-seq has accurate measurement of several hundred transcripts
the potential to provide an absolute count of transcript over a large dynamicrange.65
abundance, whereas the signal intensity of microarrays
depends not only on abundanceof the target sequence Invitro assays of immune cell function
but also on the binding strength of thespecific oligo- Many invitro assays of immune cell function exist for the
nucleotide sequence. These features ofRNA-seq could measurement of cell proliferation, viability, changes in
help to overcome a major drawbackof microa rray gene or protein expression, phagocytosis or cytotoxicity.
te chnologythe difficulty of combining data from Yet, functional cellular assays are mostly missing from
experiments that were performed at different times and the routine clinical laboratory. Poor reproducibility
in different laboratories thereby limiting the number between laboratories owing to a lack of standardized
of samples that could be included in a study. Genome- protocols is a major factor contributing to this status
wide association studies have derived great strength from quo. The need for more detailed reporting of experimen-
the analysis of very large cohorts of cases and controls tal protocols has been recognized as a first step toward
that were often genotyped using different technology assay standardization and has led to the development of
platforms. 60 RNA-seq might be the key to analysing the MIATA (Minimal Information About Tcell Assays)
genomewide gene expression profiles in similarly guidelines.66,67 Other hurdles for the introduction of
largecohorts. functional assays into clinical practice include the (often
Lyse cells
Purify RNA
Sequence fragments
Align reads to
Normalize genome and count Count and
fluorescent decode
signal fluorescent
intensity barcodes
Gene
Sample
Figure 3 | Methods for genome-wide and multiparameter gene expression analysis. cDNA microarrays: sample mRNA is
Nature
reverse-transcribed into fluorescently labelled cDNA, which is then hybridized to gene-specific Reviews | Rheumatology
oligonucleotide probes
arrayed on a solid matrix (the chip). After complementary sequences have bound, and unbound sample is washed away,
the chip is scanned. Fluorescence intensity detected at a specific location on the array provides a measurement of mRNA
abundance. RNA-sequencing: RNA is reverse-transcribed into short double-stranded cDNA fragments that are sequenced
using next-generation sequencers, mapped insilico to the reference genome and counted. The number of reads
representing a specific mRNA corresponds to the abundance of that mRNA in the sample. nCounteranalysis system
(NanoString Technologies, USA): mRNA molecules are hybridized in solution to two sequence-specific probes per gene
ofinteresta capture probe that anchors the hybrids to a solid support for analysis and a reporter probe carrying a
fluorescent barcode to identify the RNA. The number of hybrids with a specific barcode corresponds to the abundance
ofthat mRNA species in the sample.
substantial) degree of manual sample handling by human analytes using a multiplex bead immunoassay. Rather
operators and logistical issues pertaining to sample than looking at antigen-specific responses, this study
acquisition, storage and processing. Each of these factors focused on nonspecific stimulation with ligands of
can introduce batch effects and introduce day-to-day microbial origin, cytokines or a combination of anti-CD3
variability even within one laboratory. Nevertheless, as and anti-CD28 antibodies as a polyclonal Tcell stimu-
the clinical utility of the IFN-release assay (IGRA) lus. Minimal sample handling was required owing to the
demonstrates, these obstacles are not insurmountable. use of a novel semi-closed culture system (Figure4b).
Approximately a decade ago, the IGRA was introduced The 25 healthy volunteers recruited for this study had
to identify patients infected with Mycobacterium tuber mostly identical responses, except for two participants
culosis; its two variants (Figure4a), an ELISPOT assay 68 whose blood cells did not secrete IL1 in response to
(marketed as TSPOT.TB, Immunotec, UK) and an any of the stimuli, suggesting that this kind of immune
ELISA-based method (marketed as QuantiFERONTB profiling is both reliable and sensitive enough to detect
Gold, Quest Diagnostics, USA),69 are now established in aberrant responses. A drawback of the method is that
the clinic. bulk stimulation of whole blood activates a spectrum of
A conceptually different approach, taken by Duffy cell types, each of which could contribute differentially to
etal.,70 was the stimulation of whole blood with a panel the global response measured. Some results might, there-
of 27 stimuli followed by measuring the secretion of 32 fore, be difficult to interpret given the lack of cell-count
ELISpot Remove
(T-SPOT.TB) Plasma cells
PBMCs
RBCs
Supernatant
Sedimented
cells
Figure 4 | Invitro assays of cell function. a | Mycobacterium tuberculosis IFN-release assay. InNature Reviews
the ELISA | Rheumatology
method (marketed
as QuantiFERONTB Gold In-Tube Test; Quest Diagnostics, USA), whole blood is stimulated with M.tuberculosis proteins, and
the amount of IFN secreted into the supernatant is quantified by ELISA.69 In the ELISPOT method (marketed as TSPOT.TB;
Oxford Immunotec, UK),68 PBMCs are prepared by density gradient centrifugation. A defined number of cells is then stimulated
with M. tuberculosis protein for 24h on plates coated with anti-IFN antibodies. Antigen-responsive cells secreteIFN, which
binds to these antibodies and is, after removal of the cells, subsequently detected by a second labelled anti-IFN antibody.
The number of spots on the plate corresponds to the number of IFN+ cells in the sample. b | Whole blood multiparameter
stimulation assay.70 Whole blood is added to a battery of TruCulturetubes (Myriad RBM, USA) prefilled with standardized
amounts of individual stimuli. The tubes are incubated on a simple heating block. After 24h, the supernatant is separated
from the cells by centrifugation and analytes of interest are measured in the supernatant using a bead-based multiplex ELISA.
The number of data points generated per sample is equal to the number of stimuli multiplied by the number of analytes
measured in the supernatant. Abbreviations: ELISA, enzyme-linked immunosorbent assay; ELISPOT, enzyme-linked
Immunospot; M.tb., Mycobacterium tuberculosis; PBMC, peripheral blood mononuclear cell; RBC, red blood cell.
normalization and the inability to distinguish primary so-called big data and led to the development of tech-
from secondary effects with a single readout after 24h. niques for quality control and data normalization as well
Nonetheless, approaches like this might have utility in as novel methods for data analysis and visualization.7274
the clinic. The pattern of the response to antigen-specific One commonly used method for analysing microarray
stimulation in diseases for which candidate antigens data is cluster analysis and the generation of heat maps
have been identified (for example, citrullinated proteins to visualize results (Figure5a);75 cluster analysis, for
in RA), or polyclonal stimulation in diseases without example, was used for the identification of the interferon
known autoimmune targets, might identify dominant signature in SLE. Principal component analysis (PCA)76
pathways in individual patients, thereby providing guid- and similar dimensional reduction methods are other
ance for therapeutic decisions. In the future, functional strategies to summarize microarray data and detect pat-
studies of this kind might use integrated microfluidic terns in large data sets. These approaches enable high-
devices, a lab on a chip,71 that require little starting mat dimensional data from individual samples to be plotted
erial and enable the automated sampling and analysis of along 2D axes in a way that preserves the relative distance
supernatant at multiple time points. between the samples. Application of these methods often
reveals relationships between subsets of samples that are
Big data not obvious to the eye.
A common theme of fluorescence-based cytometry, mass Gene expression analysis by RNA-seq poses additional
cytometry, genome-wide analysis of gene expression and analytical challenges in that substantial data processing
multiplex functional assays is the generation of very large is required to convert the massive amounts of raw data
data sets. Transcriptomic analysis, starting with micro- generated by automated sequencing machines into gene
arrays, was a pioneering technology for dealing with expression values. This process requires sophisticated
a b c
Cell ID
IFN- IFN-
7 9 17 4 12 16 22 18 15 1011 1 2 3 13 19 20 6 8 5 21 14 23
29
2
3
38
5
13
7
8
9
40
10
12
14
15
44 TNF TNF
16
32
17
18
48
27
19
20
21
Gene
22
1
23
24
6
25
47
49
26
28 IL-2 IL-2
30
4
31
11
33
34
50
35
37
39
41
42
43
36
45
3 cells
46 385 cells 1 cell
Figure 5 | Analysing and displaying large complex data sets. Human memory CD4+ Tcells were Nature
stimulated | Rheumatology
Reviewsfor 24h with
anti-CD3 and anti-CD28 beads. After 24h of incubation, cells were either sorted for single-cell RNA-seq or analyzed by
32-parameter mass cytometry. a | Cluster analysis of single-cell RNA-seq data. 50 differentially expressed genes are
depicted (green, high expression; red, low expression). A cluster algorithm has ordered cells (columns) and genes (rows)
byexpression, revealing shared and distinct transcriptional patterns. Cellular hierarchy is represented by a dendrogram.
b | SPADE for mass cytometry data. With 32markers, SPADE groups cells into nodes on the basis of panel-wide similarity
of marker expression. Node size represents the number of cells in the cluster. Nodes are joined in a minimal-spanning tree
that represents phenotypic similarity; the degree of dissimilarity between two nodes increases with the number of nodes
separating them. Colour depicts expression of an individual marker across all nodes. IFN, TNF and IL2 are shown as
examples. c | viSNE map for mass cytometry data. Adimensionality reduction algorithm (tSNE) projects the 32parameter
data set onto a 2D plane. Each dot represents an individual cell and phenotypic similarity across the marker panel is
represented by distance. Colour represents expression ofindividual markers. Abbreviations: RNA-seq, RNA sequencing;
SPADE, spanning-tree progression analysis of densitynormalized events; viSNE, visualization of the tdistributed Stochastic
Neighbor Embedding algorithm.
data handling procedures. Next-generation sequencers include markers expressed across a spectrum of cell
generate millions of short (~100bp) reads. The first crit- types. To this end, we and others have explored the utility
ical step in analysing RNA-seq data is the mapping of of automated gating methods.8082 SPADE (Spanning-tree
reads to specific genes using software implementations Progression Analysis of Density-normalized Events)83
of sequence alignment algorithms.77,78 The second step is a radical departure from 2D gating, as cells are clus-
is to infer expression values, which in essence means to tered in ndimensional space according to similarity in
count the number of reads per gene. Mapped reads can expression across all n measured parameters. For data
also be used to identify novel transcripts. The optimal visualization, the cluster network is flattened into a
strategies for RNA-seq analysis of specific applications 2D dendrogram (Figure5b). Since the introduction of
are still being defined.79 SPADE, several alternative approaches have been devel-
Flow cytometry data have traditionally been analysed oped for the identification of novel cellular subsets and
by manual gating of 2D dot plots. This approach is suf- the analysis of rare cell types, including viSNE (visualiza-
ficient to visualize the production of two cytokines by a tion of the tdistributed Stochastic Neighbor Embedding
cell population of interest, but cannot adequately repre- algorithm), which preserves individual cellular events
sent the data when three or more cytokines are measured and represents similarity across all nmarkers as distance
simultaneously. Moreover, although biological evidence between cells in a 2D plot (Figure5c).84,85
supports the use of a CD3 gate to identify Tcells or a Although many computational strategies have been
CD19 gate for Bcells, for example, manual gating developed, additional research in this area is needed.
becomes increasingly difficult and subjective when high- RNA-seq data analysis continues to evolve as new
dimensional data sets are analysed, data sets that might sequencing protocols emerge, and mass cytometry data
analysis is an active area of research as investigators professionals. Substantial training and learning as well as
consider its application in disease-specific casecontrol the development of appropriate clinical decision support
studies. The large-scale application of mass cytometry tools will be required to empower rheumatologists to
will require the development of effective normalization deal with big data.1
strategies86 and of automated methods to infer and correct
for variations in cell proportions in different clinical Conclusions
samples.87 Clearly, a paramount challenge is to translate The clinical vignettes in Box1 and Box2 highlight two
statistical findings into biological meaning with pathway common situations in which uncertainty exists regard-
analyses, integrating with biological databases, and ing the optimal treatment strategy for patients with a
automated cross-referencing with biomedicalcontext. rheumatic disease. How might immune-cell profiling
contribute to the management of these patients? In clini
Translation into clinical practice cal scenario1 (Box1), several validated options exist
We have described existing and emerging techniques for treating this patient with RA;92 yet, for any of these
for the high-dimensional profiling of immune cells in interventions a fraction of treated patients will have an
research. By using either of two basic strategies, these inadequate response, either due to inter-individual dif-
investigations could enable the development of clini- ferences in pharmacokinetics and pharmacodynamics,
cally useful biomarkers. The first strategy relies on data or because of differences in the disease process itself.
reduction and the identification (by applying appropriate Mechanistic biomarkers3 that capture the disease state
filters) of a single parameter or small number of para in the individual patient and predict the response to
meters that can be measured using standard techniques. therapy would be extremely useful. Potential cellular bio
The search for prognostic markers in colorectal cancer markers include the frequency of immune cell subsets,
is a good example of this approach. A combination of their gene expression signatures, or quality and mag-
gene expression analysis, tissue microarrays and flow nitude of response to antigen-specific stimulation (for
cytometric profiling of colorectal cancers showed that example, using citrullinated peptides). The new technol-
the extent and localization of infiltrating memory Tcells, ogies introduced in this Review might help to identify
in particular CD8+ Tcells, correlated with a lack of early candidate biomarkers (for example, using multiparam-
metastatic invasion and a better overall prognosis.88,89 eter cytometry) or to improve the quality of existing
From the initial multidimensional analysis, a simple approaches (for example, by replacing m icroarrays
immunohistochemical score was developed that requires withRNA-seq).
detection of only three markers (CD3, CD8 and CD45R) Clinical scenario2 (Box2) was introduced to highlight
in two regions of a tumoura scoring system that might the emerging concept that inflammatory diseases can be
be implemented broadly in clinical practice.90 classified according to similarities in pathogenesis or
The second strategy involves the use of high- therapeutic responsiveness.93 TNF inhibitors, as the first
dimensional assays in clinical practice, as the pattern biologic agent in rheumatology practice, have proven
of multiple markers might be more informative than a their therapeutic efficacy for many diseases. However,
single marker or a simple ratio between two markers. responses diverge substantially for subsequently intro-
In addition to the technical issues already discussed, the duced biologic agents. For example, IL6 blockade,
implementation of this approach generates additional although effective for treating RA or Still disease, does
challenges. In a research setting, molecular and cellular not work for patients with ankylosing spondylitis, and
analyses are typically performed at a single institution, IL1 blockade is effective in treating autoinflammatory
often by a small number of technicians, and samples syndromes and systemic juvenile idiopathic arthritis, but
are batched into just a few runs. In the clinical setting, is only marginally effective in the treatment of RA.93 The
however, samples are processed in real-time on a daily realization that rheumatic diseases do not respond uni-
basis, and the data from individual patients are compared formly to the increasing number of therapeutic options
to a historical reference. This strategy, therefore, relies is of potential value for managing patients with rare
on the development of robust standardized protocols for conditions. The identification of immune signatures for
sample handling and analysis that enable reproducible clinical responsiveness to targeted therapies, as defined
results within and between institutions and a precise in randomized clinical trials of common diseases, might
definition of normal. The Human Immunology Project enable clinicians to treat patients with uncommon clini-
is one effort to generate this knowledge.39 cal presentations more effectively by comparing patient
Factors contributing to variability in flow cytometry data to these validated immune signatures. Immune
include the definition of cell subsets, choice of antibody cell profiles of peripheral blood samples are probably
clones and fluorochromes, cytometer settings, sample major components of such immune signatures, together
handling and analysis. LyoplatesTM (BD Biosciences, with genetic or serological markers and histopathology
USA) contain standardized antibody mixtures that findings; the latter dimensions of the disease process
improve reproducibility of results.91 Operator training are outside the scope of this Review. The potential
and the development of software tools for automated of immune cell profiling for predicti ng therapeutic
analysis are other strate gies for improving the reli- responses in rheumatic diseases has not yet been real-
ability of results. An additional problem to be solved is ized, but recent technological advances promise progress
the interpretation of multiparametric data by medical in the near future.
1. Robinson, W.H. etal. Mechanistic biomarkers Tcells and Thelper type17 cells in rheumatoid 36. Franz, B., May, K.F. Jr, Dranoff, G. &
for clinical decision making in rheumatic arthritis patients. Clin. Exp. Immunol. 171, Wucherpfennig, K. Exvivo characterization
diseases. Nat. Rev. Rheumatol. 9, 267276 237242 (2013). andisolation of rare memory Bcells with antigen
(2013). 20. Thiolat, A. etal. Interleukin6 receptor blockade tetramers. Blood 118, 348357 (2011).
2. Emery, P. & Drner, T. Optimising treatment in enhances CD39+ regulatory Tcell development 37. Chattopadhyay, P.K. etal. Quantum dot
rheumatoid arthritis: a review of potential in rheumatoid arthritis and in experimental semiconductor nanocrystals for
biological markers of response. Ann. Rheum. Dis. arthritis. Arthritis Rheumatol. 66, 273283 immunophenotyping by polychromatic flow
70, 20632070 (2011). (2014). cytometry. Nat. Med. 12, 972977 (2006).
3. Virgo, P.F. & Gibbs, G.J. Flow cytometry in 21. Schmidt, J.V., Su, G.H., Reddy, J.K., 38. Bendall, S.C., Nolan, G.P., Roederer, M.,
clinical pathology. Ann. Clin. Biochem. 49, 1728 Simon,M.C. & Bradfield, C.A. Characterization Chattopadhyay, P.K. A deep profilers guide
(2012). of a murine Ahr null allele: involvement of the tocytometry. Trends Immunol. 33, 323332
4. Kittleson, M.M. & Kobashigawa, J.A. Long-term Ahreceptor in hepatic growth and development. (2012).
care of the heart transplant recipient. Curr. Opin. Proc. Natl Acad. Sci. USA 93, 67316736 39. Maecker, H.T., McCoy, J.P. & Nussenblatt, R.
Organ Transplant 19, 515524 (2014). (1996). Standardizing immunophenotyping for the
5. Mirnezami, R., Nicholson, J. & Darzi, A. Preparing 22. Pawlik, A. etal. The expansion of CD4+CD28 Human Immunology Project. Nat. Rev. Immunol.
for precision medicine. N.Engl. J. Med. 366, Tcells in patients with rheumatoid arthritis. 12, 191200 (2012).
489491 (2012). Arthritis Res. Ther. 5, R210R213 (2003). 40. Newell, E.W. etal. Combinatorial tetramer
6. Hollebecque, A., Massard, C., & Soria, J.C. 23. Martens, P.B. Goronzy, J.J., Schaid, D. & staining and mass cytometry analysis facilitate
Implementing precision medicine initiatives in Weyand, C.M. Expansion of unusual CD4+ Tcell epitope mapping and characterization.
the clinic: a new paradigm in drug development. Tcells in severe rheumatoid arthritis. Nat.Biotechnol. 31, 623629 (2013).
Curr. Opin. Oncol. 26, 340346 (2014). ArthritisRheum. 40, 11061114 (1997). 41. Bodenmiller, B. etal. Multiplexed mass
7. Sander, B., Andersson, J. & Andersson, U. 24. Scarsi, M., Ziglioli, T. & Air, P. Decreased cytometry profiling of cellular states perturbed
Assessment of cytokines by circulating CD28-negative Tcells in patients with by small-molecule regulators. Nat. Biotechnol.
immunofluorescence and the paraformaldehyde- rheumatoid arthritis treated with abatacept are 30, 858867 (2012).
saponin procedure. Immunol. Rev. 119, 6593 correlated with clinical response. J.Rheumatol. 42. Bendall, S.C. etal. Single-cell mass cytometry of
(1991). 37, 911916 (2010). differential immune and drug responses across
8. Perez, O.D. & Nolan, G.P. Simultaneous 25. Scarsi, M., Ziglioli, T. & Air, P. Baseline numbers a human hematopoietic continuum. Science
measurement of multiple active kinase states of circulating CD28-negative Tcells may predict 332, 687696 (2011).
using polychromatic flow cytometry. clinical response to abatacept in patients with 43. Gaudillire, B. etal. Clinical recovery from
Nat.Biotechnol. 20, 155162 (2002). rheumatoid arthritis. J.Rheumatol. 38, surgery correlates with single-cell immune
9. Wells, A.D., Gudmundsdottir, H. & Turka, L.A. 21052111 (2011). signatures. Sci. Transl. Med. 6, 255ra131
Following the fate of individual Tcells throughout 26. Dass, S. etal. Highly sensitive Bcell analysis (2014).
activation and clonal expansion. Signals from predicts response to rituximab therapy in 44. Golub, T.R. etal. Molecular classification of
Tcell receptor and CD28 differentially regulate rheumatoid arthritis. Arthritis Rheum. 58, cancer: class discovery and class prediction
the induction and duration of a proliferative 29932999 (2008). bygene expression monitoring. Science 286,
response. J.Clin. Invest. 100, 31733183 27. Brezinschek, H.P., Rainer, F., Brickmann, K. 531537 (1999).
(1997). &Graninger, W.B. Blymphocyte-typing for 45. Alizadeh, A.A. etal. Distinct types of diffuse
10. Gregersen, P.K., Silver, J. & Winchester, R.J. prediction of clinical response to rituximab. large Bcell lymphoma identified by gene
Theshared epitope hypothesis. An approach Arthritis Res. Ther. 14, R161 (2012). expression profiling. Nature 403, 503511
tounderstanding the molecular genetics of 28. Sellam, J. etal. Blood memory Bcells are (2000).
susceptibility to rheumatoid arthritis. disturbed and predict the response to rituximab 46. Bennett, L. etal. Interferon and granulopoiesis
ArthritisRheum. 30, 12051213 (1987). in patients with rheumatoid arthritis. signatures in systemic lupus erythematosus
11. Raychaudhuri, S. etal. Five amino acids in three ArthritisRheum. 63, 36923701 (2011). blood. J.Exp. Med. 197, 711723 (2003).
HLA proteins explain most of the association 29. Vital, E.M. etal. Management of nonresponse 47. Baechler, E.C. etal. Interferon-inducible gene
between MHC and seropositive rheumatoid torituximab in rheumatoid arthritis: predictors expression signature in peripheral blood cells
arthritis. Nat. Genet. 44, 291296 (2012). and outcome of re-treatment. Arthritis Rheum. ofpatients with severe lupus. Proc. Natl Acad.
12. Pitzalis, C., Kingsley, G., Murphy, J. & Panayi, G. 62, 12731279 (2010). Sci. USA 100, 26102615 (2003).
Abnormal distribution of the helper-inducer and 30. Daien, C.I. etal. High levels of memory Bcells 48. Ermann, J. & Bermas, B.L. The biology behind
suppressor-inducer Tlymphocyte subsets in the are associated with response to a first tumor the new therapies for SLE. Int. J.Clin. Pract. 61,
rheumatoid joint. Clin. Immunol. Immunopathol. necrosis factor inhibitor in patients with 21132119 (2007).
45, 252258 (1987). rheumatoid arthritis in a longitudinal 49. Crow, M.K. TypeI interferon in the
13. Ichikawa, Y., Shimizu, H., Yoshida, M. & prospective study. Arthritis Res. Ther. 16, R95 pathogenesis of lupus. J.Immunol. 192,
Arimori,S. Activation antigens expressed (2014). 54595468 (2014).
onTcells of the peripheral blood in Sjgrens 31. Vital, E.M., Dass, S., Buch, M.H., Rawstron,A.C. 50. Rnnblom, L. & Eloranta, M.L. The interferon
syndrome and rheumatoid arthritis. Clin. Exp. & Emery, P. An extra dose ofrituximab improves signature in autoimmune diseases. Curr. Opin.
Rheumatol. 8, 243249 (1990). clinical response in rheumatoid arthritis Rheumatol. 25, 248253 (2013).
14. Berner, B., Wolf, G., Hummel, K.M., Mller, G.A. patientswith initial incomplete Bcell depletion: 51. Burska, A.N. etal. Gene expression analysis
& Reuss-Borst, M.A. Increased expression arandomised controlled trial. Ann. Rheum. Dis. inRA: towards personalized medicine.
ofCD40 ligand (CD154) on CD4+ Tcells as a http://dx.doi.org/10.1136/annrheumdis-2013- PharmacogenomicsJ. 14, 93106 (2014).
marker of disease activity in rheumatoid 204544. 52. Thurlings, R.M. etal. Relationship between
arthritis. Ann. Rheum. Dis. 59, 190195 (2000). 32. Altman, J.D. etal. Phenotypic analysis of thetypeI interferon signature and the response
15. Niu, Q., Cai, B., Huang, Z.C., Shi, Y.Y. & antigen-specific Tlymphocytes. Science 274, to rituximab in rheumatoid arthritis patients.
Wang,L.L. Disturbed TH17/TREG balance in 9496 (1996). Arthritis Rheum. 62, 36073614 (2010).
patients with rheumatoid arthritis. 33. Snir, O. etal. Identification and functional 53. Raterman, H.G. etal. The interferon typeI
Rheumatol.Int. 32, 27312736 (2012). characterization of Tcells reactive to signature towards prediction of non-response
16. Wang, W. etal. The TH17/TREG imbalance and citrullinated vimentin in HLA-DRB1*0401- torituximab in rheumatoid arthritis patients.
cytokine environment in peripheral blood of positive humanized mice and rheumatoid Arthritis Res. Ther. 14, R95 (2012).
patients with rheumatoid arthritis. arthritis patients. Arthritis Rheum. 63, 54. Palmer, C., Diehn, M., Alizadeh, A.A. &
Rheumatol.Int. 32, 887893 (2012). 28732883 (2011). Brown,P.O. Cell-type specific gene expression
17. Cooles, F.A. etal. TREG cells in rheumatoid 34. Scally, S.W. etal. A molecular basis for the profiles of leukocytes in human peripheral blood.
arthritis: an update. Curr. Rheumatol. Rep. 15, association of the HLA-DRB1 locus, citrullination, BMC Genomics 7, 115 (2006).
352 (2013). and rheumatoid arthritis. J.Exp. Med. 210, 55. McKinney, E.F. etal. A CD8+ Tcell transcription
18. Samson, M. etal. Brief report: inhibition of 25692582 (2013). signature predicts prognosis in autoimmune
interleukin6 function corrects TH17/TREG cell 35. James, E. etal. Citrulline specific TH1 cells are disease. Nat. Med. 16, 586591 (2010).
imbalance in patients with rheumatoid arthritis. increased in rheumatoid arthritis and their 56. Lee, J.C. etal. Gene expression profiling of
Arthritis Rheum. 64, 24992503 (2012). frequency is influenced by disease duration and CD8+ Tcells predicts prognosis in patients
19. Pesce, B. etal. Effect of interleukin6 receptor therapy. Arthritis Rheumatol. 66, 17121722 withCrohn disease and ulcerative colitis.
blockade on the balance between regulatory (2014). J.Clin. Invest. 121, 41704179 (2011).
57. Pratt, A.G. etal. A CD4 Tcell gene signature sporulation time series. Pac. Symp. Biocomput. 95. Williams, A.F. Galfr, G. & Milstein, C. Analysis
forearly rheumatoid arthritis implicates 2000, 455466 (2000). of cell surfaces by xenogeneic myeloma-hybrid
interleukin6mediated STAT3 signalling, 77. Langmead, B., Trapnell, C., Pop, M. & antibodies: differentiation antigens of rat
particularly in anti-citrullinated peptide antibody- Salzberg,S.L. Ultrafast and memory-efficient lymphocytes. Cell 12, 663673 (1977).
negative disease. Ann. Rheum. Dis. 71, alignment of short DNA sequences to the human 96. Parks, D.R., Hardy, R.R. & Herzenberg, L.A.
13741381 (2012). genome. Genome Biol. 10, R25 (2009). Three-color immunofluorescence analysis of
58. Mortazavi, A., Williams, B.A., McCue, K., 78. Trapnell, C. etal. Transcript assembly and mouse Blymphocyte subpopulations. Cytometry
Schaeffer, L. & Wold, B. Mapping and quantifying quantification by RNA-Seq reveals unannotated 5, 159168 (1984).
mammalian transcriptomes by RNASeq. transcripts and isoform switching during cell 97. Roederer, M. etal. Heterogeneous calcium flux
Nat.Methods 5, 621628 (2008). differentiation. Nat. Biotechnol. 28, 511515 in peripheral Tcell subsets revealed by five-color
59. Wang, Z., Gerstein, M. & Snyder, M. RNA-Seq: (2010). flow cytometry using log-ratio circuitry. Cytometry
arevolutionary tool for transcriptomics. Nat. Rev. 79. Kratz, A. & Carninci, P. The devil in the details of 21, 187196 (1995).
Genet. 10, 5763 (2009). RNAseq. Nat. Biotechnol. 32, 882884 (2014). 98. De Rosa, S.C., Herzenberg, L.A.,
60. Okada, Y. etal. Genetics of rheumatoid arthritis 80. Pyne, S. etal. Automated high-dimensional flow Herzenberg,L.A. & Roederer, M. 11color,
contributes to biology and drug discovery. Nature cytometric data analysis. Proc. Natl Acad. 13parameter flow cytometry: identification of
506, 376381 (2014). Sci.USA 106, 85198524 (2009). human naive Tcells by phenotype, function, and
61. Tang, F. etal. mRNA-Seq whole-transcriptome 81. Hu, X. etal. Application of user-guided Tcell receptor diversity. Nat. Med. 7, 245248
analysis of a single cell. Nat. Methods 6, automated cytometric data analysis to large- (2001).
377382 (2009). scale immunoprofiling of invariant natural killer 99. Saiki, R.K. etal. Primer-directed enzymatic
62. Wu, A.R. etal. Quantitative assessment Tcells. Proc. Natl Acad. Sci. USA 110, amplification of DNA with a thermostable DNA
ofsingle-cell RNA-sequencing methods. 1903019035 (2013). polymerase. Science 239, 487491 (1988).
Nat.Methods 11, 4146 (2014). 82. Aghaeepour, N. etal. Critical assessment 100. Schena, M., Shalon, D., Davis, R.W. &
63. Shalek, A.K. etal. Single-cell RNA-seq reveals ofautomated flow cytometry data analysis Brown,P.O. Quantitative monitoring of gene
dynamic paracrine control of cellular variation. techniques. Nat. Methods 10, 228238 (2013). expression patterns with a complementary DNA
Nature 510, 363369 (2014). 83. Qiu, P. etal. Extracting a cellular hierarchy from microarray. Science 270, 467470 (1995).
64. Geiss, G.K. etal. Direct multiplexed high-dimensional cytometry data with SPADE. 101. Heid, C.A., Stevens, J., Livak, K.J. &
measurement of gene expression with color- Nat. Biotechnol. 29, 886891 (2011). Williams,P.M. Real time quantitative PCR.
coded probe pairs. Nat. Biotechnol. 26, 317325 84. Amir, E.D. etal. viSNE enables visualization of Genome Res. 6, 986994 (1996).
(2008). high dimensional single-cell data and reveals 102. de Hair, M.J. etal. Features of the synovium
65. Kulkarni, M.M. Digital multiplexed gene phenotypic heterogeneity of leukemia. ofindividuals at risk of developing rheumatoid
expression analysis using the NanoString Nat.Biotechnol. 31, 545552 (2013). arthritis: implications for understanding
nCounter system. Curr. Protoc. Mol. Biol. http:// 85. Shekhar, K., Brodin, P., Davis, M.M. & preclinical rheumatoid arthritis.
dx.doi.org/10.1002/0471142727. Chakraborty, A.K. Automatic Classification of ArthritisRheumatol. 66, 513522 (2014).
mb25b10s94. Cellular Expression by Nonlinear Stochastic 103. Pitzalis, C., Kelly, S. & Humby, F. New learnings
66. Janetzki, S. etal. MIATA-minimal information Embedding (ACCENSE). Proc. Natl Acad. Sci. USA on the pathophysiology of RA from synovial
about Tcell assays. Immunity 31, 527528 111, 202207 (2014). biopsies. Curr. Opin. Rheumatol. 25, 334344
(2009). 86. Finck, R. etal. Normalization of mass cytometry (2013).
67. Britten, C.M. etal. Tcell assays and MIATA: data with bead standards. CytometryA 83, 104. Zhu, J. etal. Immune surveillance by CD8+
theessential minimum for maximum impact. 483494 (2013). skin-resident Tcells in human herpes virus
Immunity 37, 12 (2012). 87. Bruggner, R.V., Bodenmiller, B., Dill, D.L., infection. Nature 497, 494497 (2013).
68. Lalvani, A. etal. Rapid effector function in CD8+ Tibshirani, R.J. & Nolan, G.P. Automated 105. Schlapbach, C. etal. Human TH9 cells are skin-
memory Tcells. J.Exp. Med. 186, 859865 identification of stratifying signatures in cellular tropic and have autocrine and paracrine
(1997). subpopulations. Proc. Natl Acad. Sci. USA 111, proinflammatory capacity. Sci. Transl. Med. 6,
69. Mori, T. etal. Specific detection of tuberculosis E2770E2777 (2014). 219ra8 (2014).
infection: an interferonbased assay using new 88. Pags, F. etal. Effector memory Tcells, early 106. Sherlock, J.P. etal. IL23 induces
antigens. Am.J. Respir. Crit. Care Med. 170, metastasis, and survival in colorectal cancer. spondyloarthropathy by acting on RORt+
5964 (2004). N.Engl. J. Med. 353, 26542666 (2005). CD3+CD4CD8 entheseal resident Tcells.
70. Duffy, D. etal. Functional analysis via 89. Galon, J. etal. Type, density, and location of Nat.Med. 18, 10691076 (2012).
standardized whole-blood stimulation systems immune cells within human colorectal tumors 107. Gerlag, D.M. & Tak, P.P. How to perform and
defines the boundaries of a healthy immune predict clinical outcome. Science 313, analyse synovial biopsies. Best Pract. Res. Clin.
response to complex stimuli. Immunity 40, 19601964 (2006). Rheumatol. 27, 195207 (2013).
436450 (2014). 90. Galon, J. etal. Towards the introduction of the 108. Kelly, S. etal. Ultrasound-guided synovial biopsy:
71. Chen, W. etal. Emerging microfluidic tools for Immunoscore in the classification of malignant a safe, well-tolerated and reliable technique for
functional cellular immunophenotyping: a new tumours. J.Pathol. 232, 199209 (2014). obtaining high-quality synovial tissue from both
potential paradigm for immune status 91. Nomura, L., Maino, V.C. & Maecker, H.T. large and small joints in early arthritis patients.
characterization. Front. Oncol. 3, 98 (2013). Standardization and optimization of Ann. Rheum. Dis. 74, 611617 (2015).
72. Altman, R.B. & Raychaudhuri, S. Whole-genome multiparameter intracellular cytokine staining.
expression analysis: challenges beyond CytometryA 73, 984991 (2008). Acknowledgements
clustering. Curr. Opin. Struct. Biol. 11, 340347 92. Singh, P.P., Smith, V.L., Karakousis, P.C. J.E. is supported by an NIH grant
(2001). &Schorey, J.S. Exosomes isolated from R03AR06635701A1 and a Disease Targeted
73. Irizarry, R.A. etal. Exploration, normalization, and mycobacteria-infected mice or cultured Research Pilot Grant from the Rheumatology
summaries of high density oligonucleotide array macrophages can recruit and activate immune Research Foundation. D.A.R. is supported by an NIH
probe level data. Biostatistics 4, 249264 (2003). cells invitro and invivo. J.Immunol. 189, training grant T32 5T32AR007530. M.B.B. is
74. Mootha, V.K. etal. PGC1responsive genes 777785 (2012). supported by NIH grant 1UH2AR06769401. S.R.
involved in oxidative phosphorylation are 93. Schett, G., Elewaut, D., McInnes, I.B., issupported by NIH grants 1U01HG0070033,
coordinately downregulated in human diabetes. Dayer,J.M. & Neurath, M.F. How cytokine 1R01AR063759-01A1, 5U01GM092691-04,
Nat. Genet. 34, 267273 (2003). networks fuel inflammation: toward a cytokine- 1UH2AR067677-01, and 1R01AR065183-01.
75. Eisen, M.B., Spellman, P.T., Brown. P.O. based disease taxonomy. Nat. Med. 19,
&Botstein, D. Cluster analysis and display 822824 (2013). Author contributions
ofgenome-wide expression patterns. Proc. Natl 94. Hulett, H.R., Bonner, W.A., Barrett, J. & J.E., D.A.R. and N.C.T. researched data for the article.
Acad. Sci. USA 95, 1486314868 (1998). Herzenberg, L.A. Cell sorting: automated S.R., J.E., D.A.R. and N.C.T. substantially contributed
76. Raychaudhuri, S., Stuart. J.M. & Altman. R.B. separation of mammalian cells as a function to discussion of content and writing the manuscript.
Principal components analysis to summarize ofintracellular fluorescence. Science 166, All authors contributed to review/editing of the
microarray experiments: application to 747749 (1969). manuscript before submission.