2.1. Human versus murine tumors

There are important differences between the experimental syngeneic tumor models commonly used in mice and the spontaneously occurring tumors arising clinically. Thus, mouse implantable tumors models are characterized by a high tumor burden, minimal matrix, very rapid tumor growth and lack the prolonged initial phases of multistage tumor evolution present in humans when anti-tumoral mechanisms are activated [2]. These differences are thought to underlie the sobering finding that <8% of therapies successful in murine models are also successful in clinical trials [3]. Given the importance of Tregs in limiting antitumor immunity [1], consideration of differences between human and murine Tregs appears apposite.

2.2. Human versus murine Tregs

The phenotypic characteristics of Tregs in mice are usually established using cells from lymph nodes and spleen, whereas most human studies involve analyses of Tregs in blood [4]. While data concerning Tregs in mouse blood or human spleens are rarely reported, preventing any useful comparisons, data from analyses of human and murine lymph node Tregs are available. Surprisingly, human lymph node Tregs are not anergic in vitro and actively proliferate, in contrast to a decades-old axiom derived from murine studies that Tregs are anergic and non-dividing in vitro [5]. Moreover, laboratory mice are housed in pathogen-free conditions and usually studied as juveniles. As a result, Tregs in mice are typically classified as naïve cells that have recently exited the thymus and contain a relatively poor memory pool, whereas Tregs in humans are very heterogenous functionally and phenotypically contain a large proportion of effector or memory cells. Likewise, the human FOXP3 gene, but not murine Foxp3, encodes several alternately spliced isoforms with differing functions, and is upregulated to a modest extent upon activation of conventional T cells. As a result of this apples and oranges scenario, one cannot uncritically use data from murine studies to interpret the results of clinical studies.

2.3. Tregs lack a distinctive phenotype

There are no shortages of reports or categorical statements in the literature that expression of this or that marker is unique for Tregs or a particular Treg cell subset; a few examples of such supposedly Treg-specific markers are Helios, neuropilin, CD39 and TNFR2. These findings commonly arise in one of the following ways. Investigators often compare the phenotypes of resting cells and do not assess whether inflammatory stimuli or other events alter expression of the gene of interest; this is important since any such marker is often then applied to disease samples without any further validation. Another example occurs when investigators compare blood and tissue Tregs, without comparing Tregs at a particular tissue site with corresponding conventional T cells. The phenotypic characteristics of extravascular Tregs cells are often quite different from their intravascular counterparts but may not be unique when compared with conventional T cells isolated from the same site. Lastly, Treg expression of some genes appears to be species-specific (e.g. subsets of murine but not human Tregs show membrane expression of CD73, CD103, neuropilin/CD304, whereas subsets of human but not murine Tregs show membrane expression of MHC class II). Putting these considerations in context, we recently reported an analysis of Tregs isolated from the blood, lung, tumor and tumor-associated lymph nodes of 92 patients with non-small cell lung cancer using 35 markers previously suggested to be important for Treg function, to be upregulated in tumor Tregs and/or related with tumor T cells and Treg biology [6]. None of the 35 flow cytometric markers widely used to study Tregs was able to define a tumor-specific Treg population.

2.4. Status of the Treg-specific demethylation region (TSDR) cannot be used to enumerate ‘true’ Tregs

Much credence has been placed on assessment of demethylation of the TSDR/CNS2 site within the FOXP3 locus as guide to whether or not a FOXP3+ cell is a ‘true’ Treg (versus, for example, a transiently activated conventional T cell upregulating FOXP3 expression). However, we have shown that there is a decrease in TSDR demethylation in functional human FOXP3+ Tregs that occurs with age or upon exposure to certain drugs, including calcineurin inhibitors [6–8]. Failure to take the effects of patient age into account can lead to spurious assessments of the rates of development of inducible Tregs at the tumor site.

2.5. Suppressive functions of tumor Tregs

Most studies identify FOXP3 mRNA or protein expression and assume that the cells are Tregs and, without explicitly stating the point, functionally suppressive. However, activated conventional T cells can express FOXP3, and CD4+ CD25+ CD127^lowCTLA4+ Tregs, such as in kidney or liver transplant recipients receiving maintenance immunosuppression with calcineurin inhibitors, can express FOXP3 and yet showed decreased suppressive compared to healthy donor Tregs [7]. Indeed, there are relatively few studies of the suppressive capabilities of Tregs isolated from clinical tumor samples, and as with most clinical studies, such reports did not report FOXP3 expression within their isolated CD4+ CD25+ (or CD4+ CD25+ CD127^low) supposed Treg cells, rendering unreliable any direct comparison of Treg function between samples. As a result, our study is the first involving comparison of the function of Tregs isolated from tumors versus other sites, including blood and lymph nodes, in a relatively large group of samples, with use of best practices to ensure a Treg phenotype. We found lung lymph nodes and the tumor themselves had increased numbers of Tregs compared to other sites in the same patients, and that the tumor Tregs had significantly increased suppressive function compared to all other sites when assayed under standardized conditions [6,8].

2.6. New approaches to Treg studies

A common and practical problem with Treg studies is that insufficient cells may be available for functional assessment. One group chose to tackle this problem by isolating intratumoral Treg and subjecting them to limiting dilution and then expansion in vitro over several weeks [9]. We have made two new contributions that may facilitate studies by other investigators when faced with this situation.

First, use of the Prime Flow assay, which measures mRNA levels of genes of interest by flow cytometry as a result of the incorporation and amplification of fluorescent oligonucleotides by cells that can be co-labeled with conventional cell markers, allows evaluation of mRNA expression of target genes in human Tregs that are 100% pure. Use of this assay allowed us to show that intratumoral Tregs had significantly increased levels of FOXP3 mRNA and protein compared to Tregs from other sites in the same patients [6]. These elevated FOXP3 mRNA levels were associated with significantly increased expression of four transcription factors (EOS, IRF4, SATB1, and GATA1) that, experimentally, each promote the robust induction of a Treg-specific gene signature, including upregulated and downregulated genes previously shown to be FOXP3 independent [10]. Such studies illustrate how insights into intratumoral Tregs can be provided by novel techniques and can identify new targets for potential therapeutic development.

Second, we have shown that certain lysines in the Foxp3 sequence are essential for DNA binding, IL-2 suppression and optimal Treg function [11], and that acetylation of key lysine residues is necessary for Foxp3 dimerization and Treg function [12,13]. Polyclonal antibodies specific for acetylated Foxp3 residues of murine Tregs have been produced and we are currently characterizing corresponding monoclonal antibodies, with human FOXP3 targeting to follow. When cell numbers are limiting with regard to assays of Treg suppressive function, the flow cytometric application of such monoclonal antibodies may help more accurately predict the functional capability of FOXP3+ Treg cells than simple delineation of FOXP3 mRNA or protein expression.