Mass Spectrometry Imaging of Fibroblasts and the Fibroblast Extracellular Matrix Niche: Promise and Challenge

Abstract

Introduction:

Fibroblasts maintain tissue and organ homeostasis through output of extracellular matrix that affects nearby cell signaling within the stroma. Altered fibroblast signaling contributes to many disease states and the extracellular matrix secreted by fibroblasts has been used to stratify patient by outcome, recurrence and therapeutic resistance. Recent advances in imaging mass spectrometry allow access to single cell fibroblasts and their ECM niche within clinically relevant tissue samples.

Areas covered:

We review biological and technical challenges as well as new solutions to proteomic access of fibroblast expression within the complex tissue microenvironment. Review topics cover conventional proteomic methods for single fibroblast analysis and current approaches to accessing single fibroblast proteomes by imaging mass spectrometry approaches. Strategies to target and evaluate the single cell stroma proteome on the basis of cell signaling are presented.

Expert opinion:

The promise of defining proteomic signatures from fibroblasts and their extracellular matrix niches is the discovery of new disease markers and the ability to refine therapeutic treatments. Several imaging mass spectrometry approaches exist to define the fibroblast in the setting of pathological changes from clinically acquired samples. Continued technology advances are needed to access and understand the stromal proteome and apply testing to the clinic.

1.0. Introduction

1.1. The Fibroblast in Health & Tissue Homeostasis

Fibroblasts are specialized cells that work to maintain homeostasis of the connective tissue that surrounds all cells and forms the unique structure of each organ^1–5. A primary function of the fibroblast is to synthesize, excrete and regulate the localized extracellular matrix (ECM), a network of glycoproteins, proteoglycans, and proteases that serves as a scaffold for all other cell types within the tissue^6,7. In healthy tissue, fibroblasts exist in a quiescent or in an activated, contractile state throughout the tissue structure. Periodic activation increases outputs of localized ECM to regulate health and disease through inflammation & the immune response, wound healing, and differentiation^8–10. Activated fibroblasts work to recruit inflammatory and immune cells to site specific locations through selective ECM degradation and output influencing cell migration, cell adhesion, cell-ECM and cell-cell recognition. Significantly, fibroblast post-translational modification of the ECM forms the basis for tissue organization and homeostasis^11–14. Collagens are the primary proteins of the ECMs and have a multitude of site specific PTMs that regulate and activate cell signaling^11,15,16. Major collagen PTMs include hydroxylation of proline and lysine, and glycosylation; these PTMs control basic collagen organization, stabilizes the scaffold structure and work on control of cells moving through the ECM^16–20. Primary cell-ECM recognition occurs through site specific recognition of PTMs by tyrosine kinase discoidin domain receptors (DDRs) and integrins^21–26. Induction of cell signaling through DDR and integrin PTM modulated sites alters pathways such as Src, ERK1/2, Notch, MAPK, contributing to differentiation potential^21,26–30. Fibroblast expression programs are linked to their anatomic location to regulate organ growth and differentiation throughout a lifetime^1,2,31. Fetal fibroblast gene programs differ greatly from adult fibroblast programs and are characterized by different localized needs of an organ at a particular time during growth & development^3,6,7,32. Healthy fibroblast programming thus maintains tissue homeostasis through development, growth and healing by time-and location-dependent signaling through self-regulated ECM.

1.2. The Fibroblast in Disease

Abnormally prolonged activation or incorrect activation of fibroblasts results in disease; the heterogeneity of fibroblast subtypes forms a major complication in treatment of disease^2–6. Disease-activated fibroblasts secrete dysfunctional and/or excess ECM producing scarring or fibrosis that impairs organ function and diverts normal cell signaling^33–35. The heterogeneity in ECM organization produced by fibroblasts within the tissue microenvironment is demonstrated by Figure 1. Disease activation led by cell signaling patterns including TGFB1, αSMA, PDGFRB, and FAP that appear to be induced dependent on localization within the tissue microenvironment and the identity of nearby cells (reviewed in⁴ and³⁶). Localization of fibroblasts within the tissue microstructure and surrounding cellular neighbors has a strong influence on acquiring aberrant phenotypic programming and inducing subtype populations of fibroblasts^4,37–39. The emergence of disease activation of fibroblasts is still poorly defined and is considered a maladaptation of the wound healing process that is reversible or irreversible likely due to epigenetic regulation^40,4. In cancer, the cancer-associated fibroblasts (CAFs) work within the stroma to promote tumor growth and progression under the control of the tumor program. CAF activation has a continual role throughout processes of desmoplasia, hyperplasia, inflammation and tumor growth and influences therapeutic resistance^41–47. CAFs are contributors to production of reactive stroma, a dysfunctional program of repair including growth factors, immune cells, cytokines led by excess ECM that co-evolves from high risk status to cancer progression^14,35,48,49. The ECM composition, particularly collagen types, secreted by CAFs can induce aberrant epithelial mesenchymal transition and stimulate metastasis^4,50–52. CAFs play a complex role in forming localized metastatic niches by inducing nearby cells to secrete growth factors that promote migration and invasion of cancer cells⁵³. It is the emergent fibroblast subtypes, derived from localized signaling patterns, that critically contribute to disease progression. It is hypothesized that within the heterogenous population of fibroblast subtypes there exists some subtypes with the ability to halt disease^2,4,5.

Figure 1.

Heterogeneity in stromal patterning linked to fibroblast cells within a single tissue section. Tissue is a section of triple negative breast cancer stained by hematoxylin and eosin (H&E). The pink eosin stain highlights collagen stroma. Purple stain is for cell nuclei. A-D highlight specific areas where stromal organization changes by directionality, packing density (high density, darker pink; low density light pink, demonstrated in (A). Comparison of A and D demonstrates changes from “wavy” collagen stroma (A) to linearly packed collagen stroma (D).

1.3. The Analytical Challenge in Analysis of the Fibroblast and the Fibroblast Niche

Specifically how fibroblast influence and are influenced by neighboring cell signaling, subtypes in various disease staging, and information on the secreted “good” or “bad” ECM composition remains largely undefined. A main complication is that fibroblasts exist within a complex tissue microenvironment. By cell culture, it is difficult to capture all aspects of signaling that influence fibroblast heterogeneity, such as the complex interplay of activation states, ECM composition changes within the fibroblast niche, paracrine signaling from nearby inflammatory and immune cells, and the biomechanics of the localized environment. Fibroblasts in different organs have been identified by a number of markers; no single marker exists that can identify all fibroblast types^2,4,39. Of critical importance is the location of the fibroblast relative to the pathology. Fibroblasts appear to acquire a different phenotype based on distance from emerging neoplasia and this phenotype may contribute to and be further altered by advanced progression of the lesion^38,54,55. Understanding how the ECM composition and organization reciprocally influences the fibroblast phenotype is particularly challenging. The ECM is an insoluble scaffold primarily composed of the biologic polymers, collagen and elastin, embedded with soluble ECM proteins that regulate ECM density, organization, and crosslinking⁵⁶. Organization is largely affected by fibroblast specific post-translational modifications modulated at precise locations on collagen protein structures; therefore, it is critical to understand proteomic regulation of the ECM niche. Chemokine gradients facilitated by fibroblast ECM composition and organization recruit and direct immune cell types through the stroma. However, immune cells secrete their own program of ECM and metalloproteases to remodel the tissue microenvironment^36,57,58. The fibroblast activation status and the immune program response thus exist as feedback systems to each other ultimately contributing to disease outcome. A main challenge in defining the proteomics of fibroblast biology is due to limitations of technology capable of reporting heterogeneity of subtypes within a complex tissue microenvironment and sensitive enough to address single cell niches. This review covers recent proteomics work on the fibroblast and the fibroblast niche down to a single cell level, with an emphasis on recent advances in LC-MS/MS and imaging mass spectrometry.

2.0. Defining the Fibroblast by LC-MS/MS Proteomics

A plethora of proteomic approaches have been used to report on the proteome of the fibroblast and extensive reviews exist covering conventional proteomics of the ECM^59–64. To contextualize tissue imaging proteomics discussed later, we first highlight general approaches, toolsets, and advances using liquid chromatographic tandem mass spectrometry (LC- MS/MS) proteomics to investigate the fibroblasts in vitro. We further present an overview of laser capture microdissection (LCM) coupled to LC-MS/MS that has been used to investigate fibroblast types directly from tissue, highlighting studies reported within the last 5 years.

2.1. Proteomics of Fibroblasts in vitro

A primary way of defining the fibroblast proteome has been in cell culture due to the ability to grow large numbers of cells for experimental manipulation, with recognition of the limitations of fibroblast cell culture. Proteomic analysis of cultured cells isolated from tissue or as cell lines are paired with microscopy information and/or cell assays to understand how proteomic fluctuation results in a functional change in cell behavior or signaling. Measuring changes in both the fibroblast and the secreted ECM yields a comprehensive understanding of cell activation and feedback mechanisms due to the ECM. Combinations of buffers for decellularization and/or enzymatic approaches are used to analyze cells and ECM separately and improve biological depth towards understanding the cell-ECM response (refer to section 3.7). An example of this is removal of the cells from the ECM substrate through decellularization with tryptic digestion is on the cellular fraction, while combinations of trypsin, collagenase, dispase, and elastase are used on the ECM substrate^65–67. Quantitative information on fibroblast signaling is gained using label-free approaches or by stable isotopic labeling in cell culture (SILAC); this allows both an understanding of stromal cell regulation and ECM turnover facilitated by the stromal cells. Combining labeling approaches with treatments in culture or co-culture yields time resolved signaling and turnover facilitated by fibroblasts^68–71. Additional studies have advanced to leveraging multiplexed transcriptomics and quantitative proteomics to understand a systems view of fibroblast signaling in disease^72–74. Recent work has used fluorescence-activated cell sorting to isolate specific single cells before doing proteomics with RNA analysis at a single cell level⁷⁵. Once mainstreamed, multiplexed single cell technologies should be especially useful for defining fibroblast subtypes^76,77.

2.2. LCM LC-MS/MS Analysis of The Fibroblast in Tissue

Laser capture microdissection (LCM) coupled with LC-MS/MS is a way to investigate fibroblast signaling within the complex tissue microenvironment. Here, fibroblasts are dissected from tissue defined by marker and location; the surrounding stroma may be measured separately. The tissues may be fresh frozen, paraformaldehyde-fixed or formalin fixed, paraffin embedded (FFPE). The use of FFPE tissue is essential to translational studies as most tissue used for clinical prognosis or diagnosis are stored as FFPE. Tissue is at minimum stained with the common pathology stain Hematoxylin & Eosin (H&E) to identify cell types by morphological features within a specific environment such as tumor or normal adjacent^78,79. Combinations of immunohistochemistry stains have been used to identify specific fibroblast types for cell-type resolved proteomics⁸⁰. Isolation of fibroblasts is typically done on tissue sections of 8–12 μm thickness by capturing regions ranging from 100,000 – 500,000 μm² with cellular counts of 4,000–25,000^78–80. ECM proteins may be captured from the tumor or non-tumor stromal niche and used to define compartmentalized fibroblast regions. For example, comparing tumor and stromal fibroblasts has been used to identify stromal expression of N-methyl transferase (NNMT) from 5,000–25,000 cells; inhibition of NNMT was subsequently found to reverse the cancer-activated fibroblast phenotype⁷⁸. This highlights that the fibroblast type can be reversed in tissue and that proteomic mining of stromal compartments may be used to direct new strategies for disease treatment. A recognized direction for all LC-MS/MS based proteomics of the fibroblast is the need to move to true single cell proteomics as part of studies covering multiple levels of molecular expression^81,82. It is especially technologically challenging to achieve these goals derived from a complex tissue microenvironment and one solution might be the use of imaging mass spectrometry.

3.0. Imaging Mass Spectrometry

Imaging mass spectrometry (IMS) is a mature technology^83–85 developed for molecular profiling and co-registration of multiple analyte distributions in tissue, and has been extensively reviewed^86–90. Imaging mass spectrometry (IMS) allows sampling of the proteome from tissue without removal of regions and homogenization, creating a molecular patterning in the context of the complex tissue microenvironment. For fibroblast analysis, this means that molecular patterns from fibroblasts can be visualized in situ comparative to the ECM niche and nearby cells. Recent IMS studies applied to the characterization and organization of the ECM proteome and the fibroblast niche have demonstrated the utility of the approach^91–94. A main factor in targeting a single cell in tissue by IMS is the ability of the ion source to focus to single cell or single cell niche diameters (≤5 μm or ≤15 μm diameters respectively). There are several ion sources that can target down to the single cell level, including nano-desorption electrospray ionization (nano-DESI)⁹⁵, matrix-assisted laser desorption ionization (MALDI)^96–99, secondary ion mass spectrometry (SIMS)¹⁰⁰, multiplexed ion beam imaging (MIBI)¹⁰¹ and laser ablation inductively coupled plasma mass spectrometry¹⁰². Here, we focus on advances in MALDI imaging mass spectrometry that allow targeting the proteome from the fibroblast niche.

3.1. An Overview of Tissue Imaging Mass Spectrometry

An overall goal in imaging mass spectrometry is to achieve single cell imaging with clinical benefit; this has been a driving force in technological development. A main challenge is the ability to define single cell heterogeneity within a tissue section through sample preparation, instrumentation, and data analytics. The overall approach to IMS of tissue is summarized in Figure 2. All IMS studies of tissues are paired with microscopy data, at minimum using a serial section to stain for hematoxylin and eosin (H&E). The use of a stained section allows for identification and visualization of cellular localization, stroma and morphology. Several novel IMS approaches are being developed to analyze molecular patterns from stained tissue sections, discussed later. Protein targeted strategies for tissue imaging are typically followed by scraping the tissue into solution for chromatographic sequencing^{87,91,92,94,103}. This produces a patient-specific database of peptides or intact proteins that is then linked through high mass accuracy back to the imaging studies. Regionalized expression of a specific peptide can be further confirmed by on-tissue fragmentation^92,104.

Figure 2.

The matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry workflow. A) Stained tissue sections (hematoxylin and eosin shown) are prepared by spraying enzymes and matrix (B) onto tissue sections. C) Imaging scanning uses a laser to collect data points at discrete x and y coordinates. D) Each data point is linked to a single pixel. E) A total average spectrum contains all signal across the tissue. F) Specific peaks can be explored to see changes in peak intensity across the tissue by image patterns.

Tissue microarrays contain small cores (200–500 μm diameter) of tissue that has been defined by a pathologists as representative of a patient’s tumor, normal adjacent tissue, and/or normal tissue. Cores are combined to form an array from up to 600+ patients for high clinical relevance of molecular signatures. Tissue arrays are uniquely accessible by MALDI IMS compared to LC-MS/MS and have been used by several groups to report diagnostic signatures (reviewed¹⁰⁵). New work has shown that in addition to tissue microarrays, IMS can be used to scan arrays of cells and biofluids^106–108; this increases the translational potential of IMS data as researchers may move from tissue studies to cell signaling to a clinical biofluid assay.

3.2. Mass Spectrometry Imaging Sample Preparation

For tissue type samples, sample preparation starts out similarly to immunohistochemistry studies: thin sections of fresh frozen (FF) or formalin-fixed, paraffin embedded (FFPE) tissue are mounted onto microscope slides and thoroughly washed with organic solvents to remove suppressing molecules; FFPE tissues are antigen retrieved to break formalin bonds caused by formalin fixative tissue preservation. FFPE tissues are especially useful for probing proteomic data as the processing to create the FFPE tissue block flushes out metabolites and lipids which can greatly increase ion suppression. Automatic sprayers^109–111 are used to apply enzymes and matrix, minimizing analytical “hotspots” and significantly increasing technical reproducibility. For MALDI IMS, automated sprayers facilitate the homogenous formation of small ≤1 micron sized matrix crystals that enhances sensitivity of imaging detection down to a single cell^99,112. The nearly universal use of modern automated sprayers has made the preparation of samples a uniform, robust process and has facilitated large multicenter studies by MALDI IMS^113–115. Different enzymes may be used to target specific parts of the protein structure. For proteomics, enzymes such as trypsin¹¹⁶ (tryptic peptides; largely cellular targeting), PNGase F¹¹⁷ (N-glycans), collagenase¹¹⁸ (targets digestion of collagens and other ECM proteins) and elastase¹¹⁹ are sprayed as a fine, nearly dry mist onto the tissue section. Digestion proceeds by incubation in a closed humidity chamber. This microlevel digestion, created by spraying the nearly dry enzyme onto the tissue, produces a highly localized proteomic signature that is measured by systematically scanning the tissue section using an imaging mass spectrometer. Each x,y coordinate where the tissue is sampled is called a data pixel and each data pixel reports a mass spectrum. Each spectrum contains hundreds to thousands of peak intensities from analytes. Selecting an ion peak generates a heat map of that ion across the tissue, correlating to tissue features. The data can be analyzed collectively as “big data” (e.g., principal components analysis, heatmap clustering, pathways mapping, gene ontology) with additional information gained on localization of each molecule relative to the tissue pathology.

Modern sample preparation for imaging mass spectrometry applies chemicals, enzymes and matrix using automated sprayers. Automatic sprayers^109–111 minimize analytical “hotspots” and significantly increase technical reproducibility. For MALDI IMS, automated sprayers facilitate the homogenous formation of small ≤1 micron sized matrix crystals that enhances sensitivity of imaging detection down to a single cell^99,112. The nearly universal use of modern automated sprayers has made the preparation of samples a uniform, robust process and has facilitated large multicenter studies by MALDI IMS^113–115.

3.4. State of the Art Mass Spectrometry Imaging Instrumentation

Dedicated research has resulted in multiple advances in ion sources towards achieving goals of cellular imaging across tissues^98,120,121. Commercially available MALDI IMS instruments have advanced to allow flexible tuning of the laser spot to diameters of 5–15 μm achieve high spatial resolution sampling down to single cell dimensions. At the same time, instrumentation has increased in speed of acquisition. Several imaging mass spectrometers exist that can sample at around >20 pixels per second, rapidly capturing the complex tissue microenvironment at cellular spatial resolutions^122–127.

Key advances in instrumentation such as post source ionization and ion mobility have increased the sensitivity and specificity of analytes detected by MALDI IMS. Post source ionization by a secondary laser, termed MALDI-2, greatly increases sensitivity of many analytes^128–132. In MALDI-2, the originating MALDI plume is synchronously intercepted by second laser operating at wavelengths shorter than 310 nm and pulsed to increase ionization within the gas phase plume; increased ionization likely occurs through additional resonance transfer dependent on the type of matrix¹³⁰. Ion mobility is used in many IMS studies to leverage collisional cross section measurements for increased specificity of ion detection within the complex tissue microenvironment^133–138. A surge of new work done on ion mobility instrumentation has contributed to increasing the specificity of detection within the complex tissue microenvironment¹³⁹. In one advance, cyclic ion mobility propels ions through cyclical pathing using sequential and short-lived or “traveling” waves of pulsed voltages. Increasing the number of cyclical passes increases ion mobility resolution to better differentiate between isobaric masses based on collisional cross section^140,141. In trapped ion mobility spectrometry (TIMS), a gas flow is used to separate trapped ions within an electrical field; ions elute from the trap dependent on collisional cross section within the gas flow^142–144. Combinations of high speed, focused laser diameter and TIMS have allowed specific and high spatial resolution in tissue and organs up to whole body sections from animal models of disease^126,127. MALDI-2 has been combined with TIMS for high sensitivity, high specificity imaging studies^129,131,145. Additional research has leveraged higher magnetic field Fourier transform resonance mass spectrometry to increase accuracy, sensitivity, and dynamic range for increased molecular detail^146,147.

3.5. Current Mass Spectrometry Imaging Data Analysis

The complexity of data from imaging analysis of the tissue microenvironment has created a significant field of research dedicated to computational approaches for IMS data, discussed in^148–150. It is nearly impossible within the short space of this review to give an accounting of the massive growth and progress that have been essential to evolving the field of imaging mass spectrometry towards single cell analysis. Topics in this burgeoning field of research include approaches to data annotation, peak picking, alignment and calibration^151–155, image segmentation¹⁵⁶, image sharpening^157,158, numerous approaches for sophisticated classification of molecular signatures^150,159,160, pipeline creation for imaging analyses with fusion & co-registration of microscopy data^161–164, and data fusion of multiplexed imaging techniques^165–168. Several open-sourced platforms exist to support analysis of imaging data^169–171. Three-dimensional data approaches exist to understand dimensional distribution of molecular patterning within tissues or cell clusters^172–175. IMS modalities^176,177 and/or other imaging modalities including medical imaging^{163,175,178–181} have been computationally combined for a comprehensive understanding of how cellular processes and their surrounding molecular diffusion patterns drive disease. Such research has recently produced new toolsets that have been applied to investigate cell populations within tissue sections^150,182,183. The current frontier of cell based IMS research utilizes artificial intelligence^184,185 and deep learning¹⁸⁶ for rapid interpretation and classification of heterogenous cell populations within tissue.

3.6. Mass Spectrometry Imaging Acquisition Strategies Applicable to The Single Cell Fibroblast in The Setting of a Complex Tissue Microenvironment

A schematic of approaches that can be used towards imaging the single cell fibroblast niche is shown in Figure 3. In a general imaging approach, the entire tissue is scanned by systematically stepping through x, y coordinates. This is currently the general approach to tissue imaging MS that is done and molecular signatures that can be compared within one tissue section based on histological features, e.g., tumor, tumor margins, and normal adjacent. Understanding the comparative pathology involves working with a pathologist and combining the imaging data with pathology stains such as hematoxylin and eosin (H&E). Here, a pathologist marks the pathology either on the same tissue section that will be used for imaging MS^91,187 or on a serial section. This strategy allows relative quantitative analysis of patient-specific tissue features that may also be combined in large patient cohorts. The higher speed, higher spatial resolution MS instruments allow investigation of gradient patterns from different cellular regions. In contrast to laser capture microdissection where regionalized cells are removed, gradient patterns from different cellular regions can be added based on user selection of a cell types and compared at any point once the data is collected. In this strategy, one should consider that acquiring high spatial resolution data over an entire data sets requires significantly more computational resources to analyze, particularly when studies involve large clinical cohorts.

Figure 3.

A schematic of approaches that can be used towards imaging or profiling the fibroblast proteomic niche from global tissue patterns to single cell profiles. The fibroblast proteome is accessed for imaging analysis using enzyme specificity, examples include trypsin (cellular content), collagenase (extracellular matrix proteins) and PNGase F (N-linked glycans).

Combined immunohistochemistry (IHC) and imaging MS strategies allow evaluation of image patterns relative to the cell marker used in IHC. Tissues are stained for markers by IHC and IMS is done directly on the same tissue section, with staining and digital capture completed before removal of the coverslip and preparation for IMS¹⁸⁸. An advantage of this approach is that it directly ties the cell signaling to the tissue pathology. Figure 4 demonstrates this concept with a staining for for Prolyl 4-Hydroxylase Subunit Alpha 3 (P4HA3), a key enzyme in collagen synthesis that catalyzes formation of 4-hydroxyproline, essential in driving triple helical collagen formation. Combining the imaging patterns with P4HA3 staining demonstrates that very specific hydroxylated prolines are regulated localized to the tumor margins and stroma. Besides investigating whole tissue patterns relative to staining, the instrument may be focused to high spatial resolution (≤10 μm) in target regions of staining to acquire gradient patterns relative to the IHC marker. The advantage of acquiring small areas with high spatial resolution is that it increases the capability to have large data sets focused on a particular cell type without the need for significant computational resources. Additional studies have recently been done to combine lectin staining with N-glycan imaging¹⁸⁹.

Figure 4. Integrated immunohistochemistry with MALDI IMS with staining and imaging done on the same tissue section.

A) staining for prolyl-4-hydroxylase 3 (P4HA3) that catalyzes collagen hydroxylated proline (HYP). Collagen HYP ultimately controls how collagen organizes within the tissue microenvironment and presents cell-recognition sites. B) Collagen targeted imaging done on the same section after P4HA3 staining. The peptide mapped belongs to COL1A1, amino acids 405–420. Site specific probabilities of HYP are given in parentheses. Notably, proline 418 is not hydroxylated, whereas proline 410 and 413 are hydroxylated. The combined staining and imaging experiments represent the potential for disease and cell specific understanding of collagen regulation. C) Location of the peptide (amino acids 405–420) maps to COL1A1 triple helical structure.

In histology-directed profiling by IMS¹⁹⁰, the laser is guided to a target cell for direct targeting of specific cell types within a tissue section. Contemporary literature describes this approach with staining done on serial sections and with accurate alignment of the stained and unstained section a significant task prior to cell targeting. Current approaches may use staining, either H&E for fibroblast morphology or immunohistochemistry to mark out specific fibroblasts in the tissue, with histology-directed cell targeting done on the identical tissue section. This workflow is demonstrated in Figure 5, where the tissue is stained for phosphatase and tensin homolog (PTEN), an oncogenic stroma regulator and powerful switch between oncogenesis (low PTEN) and oncosuppression (high PTEN)¹⁹¹. Fibroblast cell targeting is accomplished on the same tissue section to understand stromal proteome regulation associated with PTEN status. Current instrumentation allows defined modulation of the laser size so that the cell or a desired proportion of the niche surrounding the cell may be captured for profiling. Acquisition in this approach is very fast as a few hundred cell-specific MS spectrum are collected per tissue compared to the tens of thousands of data pixels making up large tissue images. The majority of the study time is spent marking cell types for targeting which can be easier when combined with an IHC stain on the identical tissue section. A significant advantage of histology-directed profiling compared to high spatial resolution across the entire tissue section is that fewer computational resources are required for analysis as compared to full tissue imaging. This highly refined approach to targeting the single cell fibroblast may be useful in determining new and specific markers for fibroblast subtypes. At a technological level, the imaging mass spectrometry platform presents a useful translational tool that can define fibroblast signaling gradients from single cell niches to nearby cells, either across a whole tissue section of by targeting specific fibroblast subtypes.

Figure 5. Histology directed workflow on immunostained tissue to understand the single cell fibroblast niche.

A) The tissue was stained for phosphatase and tensin homolog (PTEN) an oncogenic stroma regulator. The tumor displays heterogenous PTEN expression. B) Fibroblasts with low PTEN (high oncogenic signaling, n=50) and high PTEN (attenuated oncogenic signaling, n=50) were marked on the tissue. After preparing the same tissue prepared for ECM targeted proteomics, the MALDI laser is directed to specific fibroblast x,y coordinates for profiling, marked here by black or red circles. C) Example spectrum from targeting a single fibroblast with positive PTEN stain. D) Example spectrum from targeting a single fibroblast with no detectable PTEN stain. E) Example hydroxylated proline (HYP) modified COL1A2 peptide found with significantly higher expression in the PTEN positive fibroblasts compared to PTEN negative fibroblasts.

3.7. Mass Spectrometry Imaging Methods for Accessing Molecular Regulation of the Fibroblast Niche

We focus on enzymatic approaches that can specifically report protein signal from cells or from the surrounding extracellular matrix. Using automated sprayers that maintain spatial resolution down to cellular levels, any of the acquisition approaches outlined above may be combined with enzymatic access of the fibroblast proteome.

Trypsin is a primary enzyme used for IMS studies used for accessing cellular proteins and PTMs from FFPE or FF tissue reviewed in^87,192. ECM has been more challenging, particularly from FFPE clinically archived tissue, as these proteins are relatively inaccessible to trypsin digestion due to PTM modification of lysine or arginine. Certain ECM proteins such as collagens and elastin have lower lysine or arginine content than cellular proteins; this also presents challenges for tryptic access. MALDI IMS methods that specifically target the extracellular matrix from FFPE tissue have been developed⁹³. These approaches use collagenase blends and matrix metalloproteinases to specifically target and localize extracellular matrix regulation within the tissue microenviroment. The use of C3 on FFPE tissue reports from collagens and other ECM proteins (e.g., decorin, biglycan, vimentin, lumican, fibrinogen) that regulate the stromal response. This approach is unique because it also specifically targets triple helical regions of collagens that contain disease sensitive post-translational modifications. For tissues such as wedgectomies and lumpectomies, imaging is followed by using C3 in-solution and high-resolution accurate mass LC-MS/MS for global sequencing of patient-specific ECM proteomes. This strategy was reported that collagen regulation within the reactive stroma of radical proctectomies occurs as a gradient around the tumor⁹¹. High expression of specific hydroxylated proline sites from COL1A1 and COL3A1 were detected in tumor regions as compared to normal adjacent tissue. Patient-specific LC-MS/MS revealed that collagen signatures changed by regulation of type of collagen. ECM imaging of tissues has been done on tissue microarrays containing specific patient cohorts. In breast cancer, distinct ECM peptide signatures were linked to the ontology of progression including inflammatory cell infiltrate, by type of cancer and staging of cancer⁹⁴. In lung cancer, ECM peptides were able to define normal lung apart from low grade lung adenocarcinoma (LUAD) with striking regulation of collagen hydroxyproline⁹² in LUAD. Other enzymes, including MMP12 (elastase) have been demonstrated in imaging studies with significant regional regulation in colon cancer and in heart biology⁹³. Both ECM and cellular surfaces are highly glycosylated and glycosidases have been used to define localized glycans in disease^{189,193–195}, a potential additional source for further differentiating fibroblasts. New studies have demonstrated that these enzyme targeted approaches can be performed serially to define different types of molecular regulation from cells and ECM^187,196,197.

4.0. The Clinical Potential of Defining the in-situ Fibroblast

From a clinical perspective, the fibroblast and associated ECM components, particularly collagen signatures, have been continually identified as biomarkers stratifying patients by disease type, stage or outcome^198–206. There is an ever-increasing need for therapies that target the fibroblast and fibrotic processes. A search at clinicaltrials.gov for “fibrosis” yielded 3,895 results, a nearly 50% increase since 2017²⁰⁷, demonstrating the increased need for higher molecular definition of the fibroblast. Therapeutic targeting of the disease-activated fibroblast requires finding more specific protein markers, which can now be obtained by defining the fibroblast subtypes present within the complex tissue environment. Many approved therapies target or use proteins that are post-translationally modified^208,209. The lack of information on fibroblast specific PTMs and their role in tissue regulation is clearly a missing piece in the puzzle in understanding therapeutic treatment of disease-related fibroblast types and their ECM output. There is a large need for antifibrotic therapies and currently 144 antifibrotic agents are in testing or have completed (clinicaltrials.gov, accessed 06/04/2021). However, the majority of anti-fibrotic agents used for therapies are non-specific and decrease multiple stromal proteins including those that may inhibit tumor spread^210–213. There is very little information on specific proteomic regulation of the collagen structure and this may be responsible for the confusing and conflicting results for antifibrotic therapies²⁰⁵. Defining more specific targeting, especially in the case of ECM PTM regulation, is required to restore a fibroblast back to functional healing versus disease mediation and scar formation. The clinical benefit of refining the proteomic signature of the fibroblast and the localized niche shows high potential for new and more specific fibroblast markers, markers for patient stratification, prognosis, diagnosis and therapeutic intervention.

5.0. Limitations and Needs of Current MS Technologies and Methodologies.

Limitations in measurements of the fibroblast and the ECM niche include complexity of sample preparation, challenges incorporating standardized controls to the imaging experiment, expertise required in operating the instrument, and obtaining rapidly interpretable readouts. Elegant work has been done to decrease sample preparation and achieve robust signal²¹⁴, yet overall, by both LC-MS/MS methods and MSI, sample preparation is still quite complex and generally take several hours to achieve, particularly for the single cell level. Simplified workflows that are robust and portable between laboratory settings are needed. The imaging instrumentation and experiments particularly requires a high level of expertise to assess system suitability for analysis and lacks ways to standardize acquisition and analysis during tissue or cell-targeted analysis. “Smart” diagnostics on standards or specific m/z could be implemented to help the user assess instrument performance for quality control and system suitability. Assessing internal standards could be done “on the fly” with capabilities for adding decision criteria (e.g., pass/fail, halt analysis) during each run. Although certain instruments address quality control to some minimal amount, at this point there is no imaging instrument capable of automated evaluation of standard or experimental suitability. Finally, results from an imaging experiment are not easily portable, requiring specialized software to view and do statistical comparison and this limits collaboration among groups. At the current time, it is likely that all components would need to be developed together to breakthrough to a pre-clinical or clinical use. This would include the simplified method, an instrument capable of smart assessment of performance with automated software scoring output on an interpretable scale.

6.0. Conclusions

The promise defining the proteomics of fibroblasts and their associated ECM output are the known significant effects on halting or promoting disease progression by proteomic regulation of cellularly localized gradients. These gradients are tissue specific and location specific, working to both self-regulate the fibroblast and feed into nearby cells to affect signaling processes. Main challenges to understanding single cell fibroblast function lie in determining stage and disease specific fibroblast subtype markers, defining how fibroblast subtypes contribute tissue homeostasis and disease based on location within the tissue microenvironment, and determining the structure-function effects of organ specific fibroblasts. The localized fibroblast derived ECM output is a largely untapped source of markers for disease status, particularly at the post-translational level and in the context of the complex tissue microenvironment. State-of the art imaging mass spectrometry approaches offer several solutions to answering understanding fibroblast subtypes and regulation within tissue relative to neighboring cell types and by disease staging. Continued advances in strategies and technology are needed to access and understand the tissue-specific fibroblast proteome. Ultimately, the ability to define the fibroblast proteome within the complex tissue microenvironment will contribute to new clinical tools for use in patient stratification, prognosis and diagnosis.