Single-cell characterization of tumor immune landscapes in colorectal cancer humanized mice
Curation statements for this article:-
Curated by eLife
eLife Assessment
In this study, Bootsma and colleagues use single-cell RNA-seq to characterize immune cell populations in a colorectal human mouse model. Because these models have not yet been fully characterized from this standpoint, the data generated will be useful to the field. The study is technically solid and lays the foundation for subsequent mechanistic and functional work.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Colorectal cancer (CRC) displays inter-patient heterogeneity in molecular tumor features and immune cell composition, which influence therapy response. Humanized immune system (HIS) mouse models offer a promising in vivo model to study human tumor-immune interactions, yet their ability to recapitulate the CRC tumor immune microenvironment at single-cell resolution remains incompletely defined. Here, we performed single-cell RNA sequencing of systemic and tumor-infiltrating human immune cells in HIS mice bearing human CRC tumors and benchmarked these data against reference datasets of healthy human spleens and primary CRC tumors. Major immune lineages and transcriptional programs characteristic of the human systemic immune compartment were identified, and HIS mouse tumors developed complex, human-like immune infiltrates. Tumor-infiltrating immune cells comprised diverse T cell, myeloid, natural killer, and B cell populations, including exhausted T cell states marked by expression of PDCD1, TIGIT, HAVCR2, LAG3, and CTLA4. We further demonstrate CRC consensus molecular subtype-associated spatial differences in immune infiltration. Collectively, our findings support the use of HIS mice as a relevant model for studying CRC immune landscapes and preclinical evaluation of immunomodulatory therapies.
Article activity feed
-
eLife Assessment
In this study, Bootsma and colleagues use single-cell RNA-seq to characterize immune cell populations in a colorectal human mouse model. Because these models have not yet been fully characterized from this standpoint, the data generated will be useful to the field. The study is technically solid and lays the foundation for subsequent mechanistic and functional work.
-
Reviewer #1 (Public review):
In this manuscript, Bootsma and coworkers provide a thorough characterization of the transcriptomic features of the immune system and tumor immune microenvironment in humanized immune system (HIS) mouse models of colorectal cancer. The work is mainly descriptive, but the methodology is sound, and the data provided may help future work with this model system.
The main strength of this work is in its simple yet technically sound design. While the concordance between human and HIS mouse immunology appears to be qualitative rather than at a fine-grained compositional level, the clear and thorough characterization of the immune cell populations in this mouse makes this work useful for future studies on this model. A major limitation is the lack of characterization beyond a purely descriptive layer. There are also …
Reviewer #1 (Public review):
In this manuscript, Bootsma and coworkers provide a thorough characterization of the transcriptomic features of the immune system and tumor immune microenvironment in humanized immune system (HIS) mouse models of colorectal cancer. The work is mainly descriptive, but the methodology is sound, and the data provided may help future work with this model system.
The main strength of this work is in its simple yet technically sound design. While the concordance between human and HIS mouse immunology appears to be qualitative rather than at a fine-grained compositional level, the clear and thorough characterization of the immune cell populations in this mouse makes this work useful for future studies on this model. A major limitation is the lack of characterization beyond a purely descriptive layer. There are also a few minor concerns that need to be addressed.
-
Reviewer #2 (Public review):
Summary:
This study aims to demonstrate that humanized immune system (HIS) mouse models can recapitulate colorectal cancer (CRC) immunology for immune-oncology research. The rationale is convincing, and there is an unmet need to overcome the limitations of other genetically engineered or syngeneic mouse cancer models. The authors bring new knowledge to the field through this comprehensive single-cell study across both normal spleen and CRC tumors, benchmarked against publicly available scRNA-seq datasets. They show that general immune subsets found in humans are achieved in these HIS mice, including exhausted T cells and other activated T cell states, which they were able to identify through single-cell transcriptional profiling. The authors demonstrate the feasibility of generating HIS mice with …
Reviewer #2 (Public review):
Summary:
This study aims to demonstrate that humanized immune system (HIS) mouse models can recapitulate colorectal cancer (CRC) immunology for immune-oncology research. The rationale is convincing, and there is an unmet need to overcome the limitations of other genetically engineered or syngeneic mouse cancer models. The authors bring new knowledge to the field through this comprehensive single-cell study across both normal spleen and CRC tumors, benchmarked against publicly available scRNA-seq datasets. They show that general immune subsets found in humans are achieved in these HIS mice, including exhausted T cells and other activated T cell states, which they were able to identify through single-cell transcriptional profiling. The authors demonstrate the feasibility of generating HIS mice with subcutaneous CRC tumors that resemble features of human tumor immunology, but there is still room for improvement and additional analyses to expand on these findings.
Strengths:
The matching of HIS mouse data to human data was logical and a good way to benchmark the dataset. Additionally, the use of cell lines representing each of the consensus molecular subtypes (CMS) gave way for a more nuanced approach to studying the HIS mice as opposed to a singular representation of CRC. In general, the data generated from this study provide ample ground for further exploration and analysis.
Weaknesses:
Most of the analyses are compositional - e.g., percentages from dissociated tissues without environmental or spatial context. While establishing abundance and distributions of cell types is important to convince the field of the utility of these models, measuring the presence of these cell types alone may not be sufficient to explain their reliability as a concrete preclinical model.
Moreover, while the major immune cell types were present in both HIS mice and the human spleens from publicly available datasets, there were still discrepancies present that may not be fully explained by technical constraints alone, and a deeper look at these differences could be useful. The discussion provided an overview of the limitations of the model (high abundance of immature pro B cells, absence of human stromal components, limited myeloid lineage maturation), which was appreciated.
There was more left to be desired with the CMS group comparisons - it was interesting that cell composition did not differ between CMS1 and CMS4 subtypes but that the location of specific cell types mattered (e.g., increased CD4+ T cells and CD163+ macrophages at the tumor center in CMS4 tumors). This may ultimately be related to how tumors respond to treatments, and a better understanding of the contributions of spatial localization of immune cells in HIS mice could be beneficial. It would also be interesting to show the transcriptional profiles of the CMS2 and CMS3 tumors despite them being generally non-immunogenic, in case there are differences with human data. Like in human data, there is also clearly a lot of "inter-patient" variability among the HIS mice across cell types, which could potentially be a case for uneven benchmarking. Overall, this is a commendable effort, but it misses some key experiments that could provide even more compelling evidence that these HIS mice can be used as standalone models for the evaluation of immunomodulatory treatments in CRC.
-
Reviewer #3 (Public review):
Summary:
Bootsma, Saris and colleagues set out to determine how faithfully humanized immune system (HIS) mice reproduce the human immune landscape of colorectal cancer. They performed single-cell RNA sequencing of sorted human CD45+ cells from spleens of tumor-free HIS mice and from subcutaneous tumors derived from four CRC cell lines, and compared these with published data from healthy human spleen and primary human CRC. They additionally assessed whether infiltration differed between cell lines assigned to consensus molecular subtypes CMS1 and CMS4, and whether tumor-infiltrating T cells expressed exhaustion-associated genes.
Strengths:
(1) Single-cell characterization of both the systemic and tumor compartments of this widely used model is a practical and welcome contribution, since the field largely …
Reviewer #3 (Public review):
Summary:
Bootsma, Saris and colleagues set out to determine how faithfully humanized immune system (HIS) mice reproduce the human immune landscape of colorectal cancer. They performed single-cell RNA sequencing of sorted human CD45+ cells from spleens of tumor-free HIS mice and from subcutaneous tumors derived from four CRC cell lines, and compared these with published data from healthy human spleen and primary human CRC. They additionally assessed whether infiltration differed between cell lines assigned to consensus molecular subtypes CMS1 and CMS4, and whether tumor-infiltrating T cells expressed exhaustion-associated genes.
Strengths:
(1) Single-cell characterization of both the systemic and tumor compartments of this widely used model is a practical and welcome contribution, since the field largely relies on flow cytometry panels that resolve far fewer states.
(2) The authors are appropriately open about known deficiencies of the model, including impaired B cell maturation, poor myeloid reconstitution, and absent neutrophils.
(3) Combining transcriptomic profiling with immunohistochemistry is a sensible addition, since composition alone does not capture spatial organization.
(4) The demonstration that a diverse lymphoid and myeloid infiltrate develops in these tumors is useful to laboratories deciding whether this platform suits their question.
Weaknesses:
The interpretation of PDCD1, TIGIT, HAVCR2, LAG3, and CTLA4 expression as T cell exhaustion is not established by the data. These genes are induced by activation as well as by chronic stimulation, and no functional readout, protein-level confirmation, or analysis of transcription factors distinguishing exhausted from recently activated cells is provided. A direct internal control is available but unused: the study includes spleens from tumor-free HIS mice, and if the same signature is present there, it cannot be attributed to the tumor microenvironment. The authors' own explanation for the dominance of CD4 memory cells, lymphopenia-driven expansion during reconstitution, would also be expected to raise activation marker expression. Since the exhaustion phenotype is the basis for the claim that this model supports preclinical evaluation of checkpoint blockade, and no immunomodulatory treatment was administered here, this claim currently rests on the presence of target gene transcripts alone.
A further point concerns depth. With 5,633 cells across six mice, on average fewer than a thousand per animal, the frequencies reported for subsets below 2% should be treated with caution, and cell numbers per individual mouse are not reported, although the hashtag-based design makes them available.
The comparison with human reference datasets is a reasonable approach and the results are informative, but the differences in dissociation, sorting and processing between this study and the published datasets mean that the compositional comparisons should be read as indicative rather than quantitative. This is worth stating explicitly in the discussion.
Relatedly, the reported overrepresentation of T cells and the increased CD4 to CD8 ratio in HIS tumors are interpreted as model artifacts, but species differences in T cell composition of tumors are relevant context that is not considered. The relative contribution of CD4 T cells to human tumor infiltrates differs from that in murine tumor models, and this should be discussed and referenced when judging how far the observed ratio reflects a deficiency of the model.
The claim that CMS-associated immune features are recapitulated rests on immunohistochemistry comparing a single cell line per subtype, so the observed differences may equally reflect properties of those two lines rather than the subtype they represent.
Impact:
As a resource describing what is and is not present in this model, the study has practical value. Its broader claims about recapitulating the human colorectal tumor immune microenvironment, particularly regarding T cell exhaustion, should be read with the above caveats in mind.
-
-
-
