Cytometry by time-of-flight workflow. (A) Suspension CyTOF consists of the following steps: (1) cells are selected, (2) antibodies are conjugated with a transition element isotope, (3) the cells are incubated with the antibodies conjugated with a transition element isotope, (4) the cells with an epitope-specific antibody conjugated to a transition element isotope reporter are vaporized, (5) vaporized cells enter the time-of-flight analyzer, where the isotopes are quantified, and (6) data are analyzed by a computer software. (B) Imaging mass cytometry consists of the following steps: (1) tissue is labeled with a pre-selected panel of antibodies conjugated with stable isotopes, (2) the tissue is inserted into an ablation chamber and a camera (not shown) acquires images from the tissue slide, (3) the laser beam ablates a spot of tissue stained with the metal-tagged antibodies, forming aerosol plumes, (4) plumes of tissue matter are directed by a flow of inert gas (Argon) into a coupling tube that delivers them to an inductively coupled plasma ion source, where they are vaporized, atomized, and ionized, (5) the metal-isotope ion cloud passes into the TOF mass spectrometer for analysis of isotope abundance, and (6) data are analyzed using computer software. Abbreviations: Ab, antibody; Ar, argon.

5. Use of Cytometry by Time-of-Flight for Characterization of Multiple Macrophage Phenotypes in the Hepatic Microenvironment

In a study dissecting the cancer-immune landscape in HCC using CyTOF, it was observed that TAMs play an essential role in the configuration of an immunosuppressive gradient across the tumor, non-tumor, and peripheral blood microenvironments ^[14]. Interestingly, the study also identified a chemotactic gradient in the tissue microenvironment, composed of the chemokines CCL20 and CXCL10, attracting TAMs and resident natural killer (NK) cells. It was observed that both types of cells induced an immunosuppressive microenvironment via the expression of high levels of IL-10 by TAMs and low levels of granzyme B by NK cells. Furthermore, it is showed that HBV patients who developed HCC had increased expression of both IL-10 by TAMs and the exhaustion markers PD-1, CD152, and Lag-3 by tumor-infiltrating lymphocytes, as compared to non-tumor-infiltrating lymphocytes. The authors claimed that this observation suggests that HCC patients treated with either macrophage-targeted therapy or immune checkpoint inhibitors will not have an illicit off-target immune response in the remaining non-cancerous tissue infected with HBV ^[14]. What appears certainly, and others, is that targeting the TAM chemokine axis can enhance the effect of immune checkpoint inhibitors by recruiting and activating more CD8⁺ and CD4⁺ T cells, which increases the number of tumor-infiltrating lymphocytes and makes the microenvironment less immunosuppressive ^{[15][16][17][18]}. The topology of the tissue microenvironment was analyzed by Sheng et al. ^[11] using IMC since suspension CyTOF cannot adequately assess intratumoral heterogeneity in situ, in the context of intact architecture or spatial context ^[11][19]. IMC provides in situ information regarding the tissue architecture and neighboring cell interactions by limiting the tissue analysis to ROIs. The topological analysis of Sheng et al. ^[11] revealed three important observations. First, the tissue microenvironment of HCC is divided into regional cellular functional units that impact the prognosis of patients. Second, CD80⁺/86⁺ infiltrating macrophages activated CD8⁺ T cells via the CD28 signaling pathway. In contrast to CD80⁺/86⁺ infiltrating macrophages, Kupffer cells positive for the PD-L1 marker inactivated CD8⁺ T cells by dephosphorylating CD28. Third, in a mouse model, depletion of Kupffer cells augmented the number of infiltrating macrophages, reduced the tumor growth, and improved the efficacy of an anti-PD-1 inhibitor.

6. Digital Spatial Profiling Workflow

Figure 3. Digital spatial profiling workflow

Digital spatial profiling workflow (A) The workflow involves these steps: (1) tissue of interest is biopsied (e.g., liver), (2) tissue is sectioned and fixed to a positively charged glass slide, (3) the slide is stained with up to four morphology markers and a nuclear stain, (4) probes are incubated with the tissue; (5) regions of interest (ROIs) are selected after incubation; (6) target-specific barcodes are liberated by the ultraviolet laser from the selected ROIs, (7) targets are quantified by the nCounter platform, and (8) data are analyzed using a combination of imaging analysis software and R scripts, several of which are provided by the company. (B) The top image represents an RNA probe that is composed of a target complementary sequence conjugated with a photocleavable linker that bonds oligonucleotide RNA sequence barcode. The bottom image represents a protein probe that is composed of a target antibody conjugated with a photocleavable linker that bonds an oligonucleotide RNA sequence barcode. (C) Representative image of a ROI within the liver lobule and (D) represents the respective CD163⁺ tissue segmentation. (E) Representative image of a ROI within the portal tract and (F) represents the respective CD163⁺ tissue segmentation. (G) Example of a volcano plot showing enriched liver protein expression by CD163⁺ macrophages in the liver lobule of HCV+ patients with cirrhosis who did and did not develop HCC. The plot was generated by probing protein expression from a liver biopsy. Abbreviations: Ab, antibody; ROI, region of interest; UV, ultraviolet. Scale bar = 100 µm.

7. Digital Spatial Profiling of Biomarkers and Fetal-like TAMs

DSP techniques have been used in an elegant study to more closely examine the association between the onco-fetal reprogramming of endothelial cells and TAMs in HCC. In the study, Sharma et al. ^[20] investigated the link between phenotypic features of malignant cells and early development programs. They observed that the HCC tissue microenvironment was similar to the ecosystem of fetal development, expressing fetal-associated endothelial cells and fetal-like TAMs. The biological implication of this observation is that malignant cells take advantage of fetal programs to generate a fetal microenvironment that is immunologically more tolerogenic. In this fetal microenvironment, tumor cells evade the immune system, which is more tolerogenic than the adult microenvironment. Also, the identification of fetal TAMs agrees with the prevailing view that, within the tissue microenvironment, the behavior of each TAM is not biologically equivalent. The authors proposed that these fetal programs are potential targets for therapy and may yield biomarkers to stratify patients into immune checkpoint inhibitor responders and non-responders.

8. Single-Cell RNA Sequencing Workflow

Figure 4. Single-cell RNA sequencing workflow.

Single-cell RNA sequencing workflow. ScRNA-seq is a technique that consists of the following steps: (1) tissue of interest is selected (e.g., liver), (2) the tissue is dissociated using enzymatic digestion, (3) isolated single cells are lysed, (4) RNA molecules of interest are purified and extracted from the total single-cell RNA, (5) a reverse transcription reaction is used to convert single-stranded RNA into complementary DNA, (6) complementary DNA is amplified using real-time polymerase chain reaction, (7) RNA-seq libraries are created by adding adapters and barcodes, (8) DNA is sequenced using a next generation sequencing platform, (9) the data generated by the NGS are trimmed, filtered and analyzed by computer software algorithms to yield single-cell expression profiles, and (10) cells are mapped to specific phenotypes and subpopulations of rare cell types. Abbreviations: cDNA, complementary DNA.

9. Decoding the Tumor Microenvironment of Hepatocellular Carcinoma One Cell at a Time

More studies using scRNA-seq are likely to reveal crucial aspects of the TME and TAM phenotypes, such as population heterogeneity and changes associated with chemotherapy, prognosis, and tumor progression. The tissue microenvironment of HCC is complex, as it is composed of multiple cellular phenotypes, such as TAMs, tumor infiltrating lymphocytes, stromal cells, malignant cells, cytokines and chemokines, and physical barriers, such as the extracellular matrix ^[21]. Single-cell resolution has revealed new paradigms in the crosstalk between macrophages and the microenvironment. For instance, ScRNA-seq was used to describe how some microenvironmental cues determine macrophage polarization and the fate of tissue fibrosis ^[22]. Another research group observed in a single-cell study, how transcriptional profiling of bone marrow macrophages identified which subpopulation was activated ^[23]. Furthermore, single-cell studies have shown that within the same tissue, subsets of immune cells respond differently to the same stimulus. Another study analyzed the immune system by using biopsies from a cohort of eight patients with HBV-associated HCC. They found a negative correlation between the proportion of M2 TAMs and the proportion of tumor-infiltrating T cells in scRNA-seq and deconvoluted bulk-cell RNA-seq datasets ^[21]. This result was confirmed by IHC of CD8⁺ T cells and CD163⁺ M2 TAMs in HCC biopsy specimens ^[21]. They also observed that expression of the immunosuppressive marker LAIR1 was enriched in TAM cell clusters expressing the cancer promoting M2 macrophage marker CD163 ^[21]. As a result of using scRNA-seq, certain TAM characteristics have been identified, which are potential therapeutic targets of immune checkpoint inhibitors.

10. Conclusions

Several cutting-edge platforms are unveiling new paradigms in the role TAMs play in the tissue microenvironment of HCC. The advantages and disadvantages of each platform are highlighted in Table 1. Despite their unique methods, there is a common pattern in the experimental data. For example, experiments using the platforms discussed support the critical role TAMs play in the tumor immune microenvironment and progression of HCC ^{[11][14][20][21][24]}. Through unique pathways, each TAM phenotype can create an immunosuppressive tissue microenvironment that favors the proliferation of malignant cells ^{[11][14][20][21][24]}. This strongly suggests that no single TAM-targeted therapy will be effective in every HCC case and that more than one phenotype of TAM may have to be targeted in each patient. Two important questions that remain unanswered are whether M2 macrophages favor the progression of the tumor because they become dysfunctional, or because malignant cells take advantage of the immunosuppressive microenvironment created by M2 macrophages. These questions are crucial because they will provide substantial evidence to support the use of macrophage targeted therapy as an adjuvant therapy in the treatment of HCC.

Table 1. Summary of strengths, weakness, opportunities, and threats for each platform.

Platform	Strengths	Weakness	Opportunities	Threats
MSI	Provides in situ visualization of multiple cell phenotypes, while preserving tissue architecture   Data can be used for spatial and nearest-neighbor analyses Allows data collection across entire tissue section	Limited amount of markers can be analyzed   Spectral overlap can hinder colocalization analysis and biomarker quantification	Presence of specific TAMs in the TME may allowed for personalized treatment Automated equipment for high throughput is available Can be incorporated into routine brightfield microscopy	Additional standardization is needed prior to clinical implementation   Most pathologist lack training with MSI microscopy
IMC	Allows for in situ visualization of more than 40 targets No spectral overlapping The use of non-biologic markers increases signal-to-noise ratio	Vaporization of cells and tissue ablation are irreversible steps	Provides high-dimensional analysis of various cellular features  Allows for stratification of patients into treatment responders and non-responders	Targets with low exppression or specific populations of cells may be challenging to detect  Complexity of data analysis
DSP	Spatial characterization of both RNA and protein expression using a limited amount of tissue Allows for morphological segmentation of a unique population of cells  Combines high-plex microscopy and spatial genomics	Analysis is limited to pre-selected proteins and RNA probes Gene and protein assays are evaluated on two sequential slide sections from a tissue block	Gene expression and protein profiling in a single ROI Single cell resolution is in development	High Cost Requires at least 150–200 cells per ROI for sufficient counts Rare target analysis is more costly and time consuming
ScRNA-seq	Single cell RNA resolution transcriptomics Accurate identification of specific phenotypes and subpopulations of rare cell types	Loss of tissue architecture  No spatial co-localization analysis	Characterization of cellular crosstalk at single cell resolution Profile potential therapeutic targets for rare phenotypes	Cell isolation method, number of cells per experiment, cost per cell and sensitivity vary between scRNA-seq platforms