Increases in Tumor N-Glycan Polylactosamines Associated with Advanced HER2-Positive and Triple-Negative Breast Cancer Tissues

Abstract

Purpose:

Using a recently developed matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDIIMS) method, human breast cancer formalin-fixed paraffin-embedded (FFPE) tissue sections and tissue microarrays (TMA) are evaluated for N-linked glycan distribution in the tumor microenvironment.

Experimental design:

Tissue sections representing multiple human epidermal growth factor receptor 2 (HER2) receptor– positive and triple-negative breast cancers (TNBC) in both TMA and FFPE slide format are processed for high resolution N-glycan MALDI-IMS. An additional FFPE tissue cohort of primary and metastatic breast tumors from the same donors are also evaluated.

Results:

The cumulative N-glycan MALDI-IMS analysis of breast cancer FFPE tissues and TMAs indicate the distribution of specific glycan structural classes to stromal, necrotic, and tumor regions. A series of high-mannose, branched and fucosylated glycans are detected predominantly within tumor regions. Additionally, a series of polylactosamine glycans are detected in advanced HER2+, TNBC, and metastatic breast cancer tissues. Comparison of tumor N-glycan species detected in paired primary and metastatic tissues indicate minimal changes between the two conditions.

Conclusions and clinical relevance:

The prevalence of tumor-associated polylactosamine glycans in primary and metastatic breast cancer tissues indicates new mechanistic insights into the development and progression of breast cancers. The presence of these glycans could be targeted for therapeutic strategies and further evaluation as potential prognostic biomarkers.

1. Introduction

Breast cancer remains one of the top causes of death among women worldwide.^[1] Breast cancer is a heterogeneous disease that incorporates several distinct subtypes with remarkably different biological characteristics and clinical behavior.^[2] Because of this, breast cancers are primarily classified based on the expression status of hormone receptors and human epidermal growth factor receptor 2 (HER2). Currently, four immunohistochemistry stains are used to classify breast cancer: estrogen receptor (ER), progesterone receptor (PR), HER2 receptor, and Ki-67. Based on the presence of these markers, breast cancer can be characterized as luminal-A (ER and/or PR-positive, HER2-negative with low proliferative activity), luminal-B (ER and/or PR-positive, HER2-negative with high proliferative activity), HER2-positive and triple-negative (TN).^[3] Among these subtypes, HER2-positive breast cancer is common in younger women and has a natural history characterized by poor prognosis, high rate of recurrence, and mortality without appropriate treatment. While targeted therapies exist for breast cancers that are positive for ER, PR, and HER2 (such as endocrine and HER2-targeted therapies), these are ineffective in patients with triple-negative breast cancer (TNBC).^[4] TNBCs represent 10–15% of all breast cancers, and these tumors are characterized by occurrence in a younger patient population, high proliferative activity, and a relatively poor outcome even if treated with aggressive multi-agent chemotherapy.^[4,5] TNBC is typically associated with high risk of early meta stasis and a higher risk of death within 5 years of diagnosis.^[6] It is well recognized that TNBC is a heterogeneous disease^[7] with different associated genetic mutations; however, there is currently no targeted therapy for this disease.

The glycocalyx is the dense outer layer of carbohydrates on the cell surface and is comprised of multiple types of oligomeric and protein-linked complex carbohydrates such as: N- and O-linked glycoproteins, proteoglycans, glycosaminoglycans, and glycolipids.^[8] Changes in glycosylation have been implicated in a variety of disease states including cancer. Alterations in glycan structures have been confirmed to affect the invasiveness and metastatic ability of a cell.^[9–13] Changes in asparagine-linked (N-linked) glycosylation are of particular interest because of their effect on cellular mobility, signaling, metastatic capability, and immune properties.^[12] Recently, our lab pioneered a method using MALDI imaging mass spectrometry to profile the distributions of N-linked glycans within clinical and experimental formalin-fixed paraffin-embedded (FFPE) tissues.^[14] Following release of tissue N-glycans by digestion with peptide N-glycosidase F (PNGase F), the relative abundance and spatial localization of 50 or more detected glycans can be directly correlated to the histopathology of the tissue. In cancer tissues, this has allowed the mapping and structural grouping of glycan classes to specific regions of tumor, stroma, and inflammation.^{[9,10,13,15–18]}

In the present study, we assessed the N-glycosylation tissue glycome in genetically subtyped breast cancer tissues. A majority of previous glycosylation studies in breast cancer have focused on serum O-glycan/mucin biomarkers or the use of cell line models.^[19] Few studies have been reported for analysis of the breast cancer tissue N-glycome in clinical samples, and most have relied on use of lectin detection of branched glycan structures.^[20] Using MALDI imaging mass spectrometry approaches, the goal of the study was to identify specific N-glycans associated with advanced HER2+ and TNBC subtypes. Multiple clinical tissue cohorts and tissue microarrays were evaluated. This resulted in the characterization of multiple polylactosamine (polyLacNAc) N-glycans and other branched N-glycans associated with metastatic breast cancer.

2. Experimental Section

Trifluoroacetic acid and alpha-cyano-4-hydroxycinnamic acid (CHCA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade methanol, ethanol, acetonitrile, xylene, and water were obtained from Fisher Scientific (Pittsburgh, PA, USA). Tissue Tack positively charged microscope slides were purchased from Polysciences, Inc. (Warrington, PA, USA). Citraconic anhydride for antigen retrieval was from Thermo Scientific (Bellefonte, PA, USA). Recombinant PNGaseF was obtained from the laboratory of Dr. Anand Mehta (Charleston, SC, USA), as described.^[14] Hematoxylin and eosin (H&E) stains were obtained from Cancer Diagnostics (Durham, NC, USA).

2.1. Clinical Breast Cancer FFPE Tissues

Tissue microarrays (TMAs) representing TNBCs (n = 41 patients), Her2+ cancers (n = 40), and controls were provided by the Institute of Pathology, University of Heidelberg, Germany, with the help of the Tissue Biobank of the National Center for Tumor Diseases according to the regulations of the local ethics committee(Reference No:#2270).All cases of the HER2+cohort were invasive breast carcinomas with Her2/neu 3+ scores; 18/40 patients had ER expression greater than 80% and 5/40 patients presented PR expression greater than 80%. The triple-negative cohorts included 37/41 invasive breast carcinomas, 2/41 invasive lobular breast cancer, 2/41 metaplastic breast carcinoma, and 5 breast cases with non-tumor tissue as control. Several source tissue slices of HER2+ and triple-negative FFPE tumor cores from the aforementioned TMAs were also analyzed (Figure S3, Supporting Information). For these full block slices analyzed, triple-negative tissues were grade III invasive ductal carcinomas classified as pT4b according to the American Joint Committee on Cancer (AJCC) 7th edition. HER2+ tissues were also grade III invasive ductal carcinomas with HER2/neu 3+ scores.

TNBC tissues from Thomas Jefferson University were collected as part of a clinical study evaluating the effects of fasting 18 h prior to lumpectomy. Seven matched tissue pairs were evaluated, with the first sample from each pair being a biopsy sample and the second being a lumpectomy sample.

A collection of 26 paired samples from patients diagnosed with breast cancer between 1999 and 2010, who experienced loco-regional and/or distant metastasis, were collected at Ospedale Policlinico San Martino within a study aimed to characterize breast cancer subtypes using proteomic immunohistochemistry techniques. All samples were re-evaluated at the moment of analysis by a pathologist: 18 were considered suitable for IMS analysis. Among them, 10 patients were metastatic (at liver, colon, lung, pleura, bone, skin, and lymph nodes and 5 patients relapsed more than once), while 8 presented with local recurrence. Five samples were TNBCs, seven samples were luminal-A breast cancers, and five samples were luminal-B breast cancers. More detailed information can be found in Table 1. All tissues used were de-identified to the analytical investigators and determined to be not human research classifications by the respective institutional review boards.

Table 1.

Clinical information detailing the primary and metastasis/recurrence tissues presented in Figure 6

Primary Breast Cancer				1 Relapse/Metastasis				2° Relapse/Metastasis				3° Relapse/Metastasis	3+ Polylactosamines Present			Recurrance (years)	Status
Site	Type	Hr	HER2	site	type	ER	HER2	site	Type	HR	HER2		Primary	Metastasis	2° Metastasis
L Breast	IDC	neg	neg	lung		neg	pos	supraclav node		neg	neg		yes	yes	yes	4	dead
L Breast	IDC			breast	IDC	pos	neg	n/a					yes	yes	n/a	3	dead
R Breast	IDC	pos	neg	armpit	in situ 90%	pos	neg	n/a					yes	yes	n/a	5	alive
L Breast	ILC	neg	neg	breast relapse	ILC	pos	neg	breast skin	ILC	pos	neg	yes	yes	yes	yes	7	dead
L Breast		pos	neg	chest wall		pos	neg	n/a					yes	yes	n/a	7	dead
L Breast	IDC	neg	neg	breast		neg	pos	n/a					yes	no	n/a	1	x
L Breast	IDC	pos	neg	armpit lymph node	IDC	pos	neg	n/a					yes	yes	n/a	2	dead
R Breast	IDC	neg	neg	2nd breast	IDC	neg	neg	n/a					yes	yes	n/a	2	alive
R Breast	IDC	neg	neg	Breast + supraclavicular node	IDC	pos	neg	n/a					no	no	n/a	10	dead
L Breast		pos	neg	supraclavicular node		pos	pos	chest node		pos	pos	yes		yes	yes	3	dead
L Breast	IDC	pos	neg	breast skin		pos	neg	breast skin	IDC	pos	2+	yes	no	yes	yes	2	dead
L Breast	IDC	pos	pos	breast		neg	pos	n/a					no	no	n/a	2	dead
L Breast	IDC	pos	neg	armpit	IDC	pos	neg	n/a					yes	yes	n/a	10	alive
L Breast		pos	neg			pos	neg	n/a					yes	no	n/a	7	dead
R Breast	IDC	pos	pos	breast		neg	pos	n/a				yes	yes		n/a	3	alive
L Breast	IDC	pos	neg	cervical skin	ILC	pos	neg	n/a					no	no	n/a	12	dead
R Breast	IDC	pos	neg	armpit		pos	neg	n/a					yes	yes	n/a	3	alive
L Breast	ILC	pos	neg	breast skin	ILC	pos	neg	breast skin	ILC	pos	neg	yes	yes	yes	yes	5	alive

2.2. Tissue Preparation and PNGaseF Digestion

FFPE tissues were sectioned at 5 μm and mounted onto charged slides (Polysciences, Inc.). Tissue slides were dewaxed and rehydrated for antigen retrieval using citraconic anhydride buffer (25 μL citraconic anhydride [Thermo Fisher Scientific, Waltham, MA], 2 μL 12 m HCl, 50 mL HPLC-grade water, pH 3.0–3.5), as previously described.^[18] Recombinant PNGaseF enzyme (0.1 μg μL⁻¹) was applied to desiccated, antigen-retrieved slides using a TMSprayer (HTX Technologies LLC., Chapel Hill, NC).Enzyme was sprayed onto the slide at a rate of 25 μL min⁻¹ for 15 passes at 45 °C at a velocity of 1200 and a 0 mm offset. Slides were placed in a humidity chamber at 37 °C for 2 h. After incubation, slides were desiccatedand7mgmL⁻¹ CHCAmatrixin50%acetonitrile0.1% trifluoroacetic acid (TFA) was applied using the TMSprayer at 0.1 mL min⁻¹ for 10 passes at 80 °C at a velocity of 1300 and a 1 mm offset. Slides were stored in a desiccator until matrix-assisted laser desorption/ionization Fourier transform ion cyclotron resonance mass spectrometry (MALDI-FTICR MS) analysis.

2.3. N-Glycan MALDI-FTIMS Analysis

Released N-linked glycan ions were detected using a Solarix dual source 7T FTICR mass spectrometer (Bruker Daltonics) (m/z 490–5000) with a SmartBeam II laser operating at 2000 Hz, and a laser spot size of 25 μm. A total of 200 laser shots were collected for each pixel using the smartwalk feature set to 25 μm with one scan per pixel. Time domain was set to 512K word with a mass range of 500–5000 m/z, resulting in a 1.2059 s transient with a calculated resolving power of 160 000 at m/z 400. Ion accumulation time was 0.1 s. Following MS analysis, data was loaded into FlexImaging Software focusing on the range m/z = 700–4000. FlexImaging 4.0 (Bruker Daltonics) was used to generate images of differentially expressed glycans normalized to total ion current. Observed glycans were searched against the glycan database generated using GlycoWorkbench.^[21] Indicated glycan structures were generated in GlycoWorkbench and represent compositionally correct structures determined by accurate mass, as well as previous structural characterizations using MALDI-FTICR or matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) instruments.^{[9,10,13,15–18,22]} Definitions of monosaccharide unit symbols used in glycan structural depictions can be found in Figure S1, Supporting Information. SCiLS 2016b (Bruker Daltonics) imaging software was also used for further analysis of glycan intensity comparisons and statistical evaluation, as follows. Using intensity values (normalized to total ion current) of both TMAs or tissue groupings, the area under the peak (AUP) intensity values for each glycan in the peak list was generated using SCiLS and exported as an excel file. For statistical comparisons, these AUP values were tested for normal distribution based on a Kolmogorov–Smirnov test. If data was normally distributed, a two sample t-test was used to determine the p-value. If data was not normally distributed, a Wilcoxon Rank Sum test was used.

2.4. Histopathological Correlation

After analysis, matrix was washed from slides using 100% ethanol for 5 min, followed by H&E staining (Cancer Diagnostics). Stained slides were analyzed and annotated for tumor localization by a pathologist for glycan map comparison and were visualized and imaged on a EVOS FL Auto microscope.

3. Results

While conducting a previous study characterizing N-glycans in regions of necrosis in TN breast tissues,^[23] it was observed that the complexity and total numbers of N-glycans detected in TN tissues were higher than other breast cancer tissue sub-types, as well as other tumor types. To better assess this, a panel of seven pre-surgery biopsy and lumpectomy TN FFPE tissue samples were processed for N-glycan imaging as described in Section 2, resulting in panels of glycans representing distinct structural regions in stroma, necrotic, and tumor (Figure 1). A list of the 64 N-glycans detected is provided in Table S1, Supporting Information, and includes a designation for whether or not the glycan specie was predominantly localized to tumor regions. Within the tumor region, highlighted in Figure 1a, a broad range of paucimannose and high mannose glycans were detected (Figure 1b), representing structures comprised of Man2–9/HexNAc2/Fuc0–1. In the stromal region, a common structural theme was present within the expressed N-glycans because the majority had bi-antennary structures with a single core-fucose attached to the first N-acetylglucosamine residue (Figure 1c), representing structures comprised of Hex3–5/HexNAc4/Fuc1/NeuAc0–2. In general, this glycan structural class represents the most abundant N-glycan intensities detected in tissues.^[13,23] Additionally, non-fucosylated bi- and tri-antennary glycans (representing Hex3–6/HexNAc4–5/NeuAc0–2) were specifically detected in regions of tumor necrosis (Figure 1d), as previously described.^[23] As the analysis of the N-glycome of this sample set continued, another interesting structural pattern emerged. A series of tetra-antennary glycans with different numbers of lactosamine (LacNAc; N-acetylglucosamine-galactose disaccharide) chains and a single fucose were present in all biopsy and lumpectomy samples (Figure 2). Structural compositions were coordinated with previous N-glycan MALDI imaging studies in human tissues^{[9,13,14,16–18,23]} and with an extensive MALDI-TOF and ESI mass spectrometry derived database of N-glycans from pancreatic cancer cell lines.^[22] Other variations of these LacNAc structures, which included similar structures without fucose or with two fucoses, were also detected but at less intensity and less consistency across TN samples. We had not previously identified this pattern of single fucose polylactosaminylated N-glycans in lower grade, non-TN breast cancer tissues. These larger N-glycan structures were not previously reported for N-glycan MALDI imaging of other breast cancer sub-types.^[23]

Figure 1.

Triple-negative matched biopsy (left columns) and lumpectomy (right columns) of breast cancer tissues with a) corresponding H&E stains with notated tumor regions (within black highlights), and representative glycans showing their distribution in b) tumor (m/z 1419.50700), c) stromal (m/z 1647.64910), and d) necrotic (m/z 1976.73570) regions of the tissues (images in SCiLS).

Figure 2.

Expression of polylactosamine N-glycan series in a) biopsy samples and b) matched lumpectomy samples. *Biopsy/lumpectomy pairs are matched horizontally.

To further evaluate how common the presence of these poly-LacNAc glycans are in TN tissues and in other breast cancer subtypes, two tissue microarrays representing paired tumor cores from TN (n = 40 subjects), HER2+ (n = 40 subjects),and 7 control non-tumor breast tissues were analyzed (Table S2, Supporting Information). Figure S2, Supporting Information, includes the H&E stain of the TN TMA and the HER2+ immunostaining of the HER2+TMA. In Figure 3, an example of the N-glycan intensities for an abundant core-fucosylated N-glycans at m/z 1809.6393 (Hex5/HexNAc4/Fuc1) (Figure 3a) in both TMAs is shown, as well as lower detection of a non-fucosylated N-glycan at m/z 1976.6666 (Hex5/HexNAc4/NeuAc1), commonly found in necrotic regions (Figure 3b).^[23] Intensities were determined by performing SCiLS analysis on the two TMAs and comparing the AUP for the average mean intensities for each tissue core. For the abundant glycan m/z 1809.6393, statistically the p-value differences were not significant m/z 1976.6666 glycan, when present, was(p = 0.46957), while the glycan at significantly higher (p = 1 × 10E-8) in TN tissues relative to HER2shown for each TMA in+. The intensities of three poly-LacNAc N-glycans are Figure 4. For the tissues analyzed, there was a trend of increased poly-LacNAc species for both species, with slight differences for single, di, and tri-LacNAc glycans. (Figure 4a, p = 0.03776 for HER2+; Figure 4b, p = 0.00032 for TN; and Figure 4c, p = 0.0008 for HER2+).

Figure 3.

SCiLS comparison of glycan expression a) Hex5dHex1HexNAc4 (m/z 1809.6393), b) Hex5HexNAc4NeuAc1 + 1 Na (m/z 1976.6666) in TMAs representing triple negative (top) and HER2+ (bottom) breast cancers, c) box plot comparing all area under the peak values for Hex5dHex1HexNAc4 (m/z 1809.6393), d) box plot comparing all area under the peak values for Hex5HexNAc4NeuAc1 + 1 Na (m/z 1976.6666).

Figure 4.

SCiLS comparison of polylactosamine glycan expression a) Hex8dHex1HexNAc7 m/z 2905.3345, b) Hex9dHex1HexNAc8 m/z 3270.5727, c) Hex10dHex1HexNAc9 m/z 3636.9087 in TMAs representing triple-negative breast cancers (top) and HER2+ breast cancers (bottom).

While both TMAs contained glycans typically seen within tumor and stromal tissue, there were several glycans that were differential between the sub-types. For HER2there was a significant increase in the tri-antennary (+ tissues, m/z 2174.7715; Hex6HexNAc5Fuc1) (Figure 5a) and tetra-antennary (m/z 2539.9037; Hex7HexNAc6Fuc1) (Figure 5b) glycans (p = 2.04E-9 and p = 1.06E-8, respectively). These glycans are precursors to more complex sialylated and fucosylated structures (Figure 5a,b) (Table S4, Supporting Information). For TN tissues, there was a notable increase in the levels of the high mannose Man8 (Hex8HexNAc2) and Man9 (Hex9HexNAc2) structures relative to the HER2+ (p = 7.80E-5 and p = 0.00027, respectively), and increases in structures with outer arm fucoses (m/z 2320.8294) (p = 0.01003) (Figures 5c–e). As the TMA cores only represent a small region of a given tissue, representative full tissue blocks were analyzed, as shown in Figure S3, Supporting Information, that support the observations in the TMAs.

Figure 5.

SCiLS comparison of glycan expression with corresponding box plot comparing all area under the peak values for a) Hex6dHex1HexNAc5 m/z 2174.7715. b) Hex7dHex1HexNAc6 m/z 2539.9037, c) Hex8HexNAc2 (Man8) m/z 1743.5810, d) Hex9HexNAc2 (Man9) m/z 1905.6338, and e) Hex6dHex2HexNAc5 m/z 2320.8294 in TMAs representing triple-negative (top) and HER2+ (bottom) breast cancers.

Because triple-negative and HER2+ breast cancers are typically associated with the worst clinical outcomes, the presence of the poly-LacNAc glycans in breast cancer metastases or relapses, regardless of genetic subtype, were evaluated in a set of 18 matched primary breast cancer tumors and their associated local or distant recurrence lesions. Cases of invasive ductal carcinoma and invasive lobular carcinoma with varying receptor classifications were represented in this set (see Table 1). Locations of associated secondary lesions were in the breast, chest wall, lymph nodes, lungs, liver, and skin. Following N-glycan MALDIIMS, data analysis focused on the presence of poly-LacNAc glycans, as summarized in Figure 6 (representative images of matched primary and metastasis in Figure S4, Supporting Information) for the presence of at least three poly-LacNAc glycans in each tissue. In 12 of these sample pairs, poly-LacNAc expression was identified in both the primary tumor and the associated relapses. In two pairs, polyLacNAcs were found only in the primary tumor, while in one pair poly-LacNAc expression was found in only the metastatic lesion. In cases where there was primary and secondary metastasis, all tissues contained poly-LacNAcs (Table 1).

Figure 6.

Graph showing the presence of polylactosamine glycans in matched breast cancer primary and metastasis samples.

4. Discussion and Future Directions

The cumulative N-glycan MALDI-IMS analysis represents a comprehensive assessment of the tumor N-glycome for advanced breast cancers, linked with histopathology localizations. For TN tumors, there were increased numbers of detected Man8/Man9 glycans relative to HER2+ tissues, indicative of less mannosidase processing during biosynthesis, resulting in glycoproteins normally destined for lysosomes to be secreted or transported to the cell surface. It is clear that the presence of poly-LacNAc structures are associated with more aggressive breast cancers (metastasis or local recurrences) independent of genetic as demonstrated in the primary and relative recurrence pairs. In Figure 2, it is also demonstrated that poly-LacNAc structures detected in lumpectomy samples could also be detected in biopsy specimens.

The presence of poly-LacNAc structures are associated with the activity of the glycosyltransferase responsible for β1,6 branching that adds sugars to complex bi-antennary precursors (i.e., branching), termed GnT-V (β1,6 N-acetylglucosaminyltransferase V). The tetra-antennary glycan detected at m/z 2539 and more elevated in HER2+ tumors is a typical example of a tumor-associated branched glycan and is the precursor for the addition of poly-LacNAc residues. The activity of GnT-V has long been associated with breast cancer metastasis, derived largely from previous lectin histochemistry studies, using leukocytic phytohemagglutinin.^[20] Lectin staining is limiting as it only recognizes a particular carbohydrate structural motif, while the MALDI-IMS approach can directly detect each N-glycan. For example, 40% of the glycans detected in the TN breast cancer tissues represent a branched glycan, and 15 were present in tumor. The actual number of branched tumor glycans present is likely much larger, as molecules above 4400 m/z are difficult to detect with the MALDI-FTICR method used. Additionally, there are many possible isomeric structural forms of the detected glycans to be determined. While the MALDI-FTICR provides highly accurate mass resolution, it can only identify glycan compositions, and not the anomeric linkages or sites of attachment and extension on a tetra-antennary structure. An additional structural limitation of the MALDI-FTICR method is that multi-sialylated glycans are not well detected by MALDI-FTICR without chemical modification and stabilization prior to analysis.^[16] Additional studies are ongoing to evaluate stabilized multi-sialylated glycans and orthogonal analysis of released N-glycans by tandem mass spectrometry and HPLC approaches. Overall, the different complex glycan structures reported herein represent an initial characterization of metastatic breast cancer tissues, and the numbers of possible breast tumor–associated branched, sialylated, poly-fucosylated, and LacNAc-extended N-glycans will continue to increase.

The polyLacNAc structure is a fundamental linear structure comprising the repeating N-acetyllactosamine unit (Galβ1–4GlcNAcβ1–3) commonly attached to glycoproteins and glycolipids.^[24] While the precise role of polyLacNAcs is still under investigation, several studies have been performed to elucidate the role of this structure in cancers.^[25,26] It is known that the structure has an important function in cell–cell and cell–matrix interaction, allowing it to potentially play a major role in metastasis. This could partially be due to the finding that polyLacNAcs are directly and indirectly able to control various molecular functions through their interactions with glycans and endogenous lectins such as receptor proteins. These structures are able to cross-link between glycoprotein receptors and endogenous lectins, resulting in restricted movement of the glycoprotein to the cell surface.^[26] In assessing their role in immune responses, when polyLacNAc chains were absent from glycoprotein acceptors or glycolipids, this resulted in the unwarranted activation of lymphocytes in response to external stimulation. However, when these polyLacNAc chains were present, they appeared to have a suppressive effect on lymphocyte activation, indicating that when these chains are present they are able to block immune response by promoting receptor aggregation.^[26] This mechanism can be specifically important in TNBC, as these subtypes of breast cancer are morphologically characterized by a brisk lymphocytic stromal reaction at the invasion front.^[27] Not only have polyLacNAcs been shown to reduce lymphocyte activation, their expression has also been shown to inhibit natural killer (NK) cells. PolyLacNAcs are able to decrease the interaction of tumor cells with natural killer cells (NK cells) by inhibiting the cytotoxic activity of NK cells as a result of galectin-3 (gal-3) binding on the core 2 of O-glycans of mucin-1 (MUC1).^[28] Thus, glycans and specifically polyLacNAcs may play an important role in enabling tumor cells to evade the host immune response.

In studies on lung metastasis, it was found that polyLacNAcs play a crucial role in the initial arrest of organ endothelium and in the subsequent extravasation events. Because polyLacNAcs have a high affinity for binding to gal-3, a lectin highly expressed on lung tissue,^[28] it is thought that this interaction plays a major role in the formation of lung metastasis.^[29] In melanoma mouse models, it was determined that the binding of polyLacNAc and gal-3 resulted in lung metastasis, and when this interaction was inhibited, it resulted in the blocking of cancer cell adhesion and metastasis.^[30] In this regard, it is interesting that TNBC shows increased rates of metastases to brain and lung.^[4] In another study, overexpression of GnT-V in cancer cells resulted in enhanced epidermal growth factor (EGF) and transforming growth factor beta (TGFβ) signaling upon binding of gal-3 and polyLacNAc.^[31] Gal-3 and polyLacNAc binding results in the formation of a molecular lattice that prevents glycoprotein endocytosis which causes activated receptors to remain at the cell surface where they can continue to promote signaling and cell growth.^[29–31] While the overexpression of GnT-V (the mannosyl (alpha-1,6-)-glycoprotein beta-1,6-N-acetyl-glucosaminyltransferase [Mgat5] gene) resulted in increased EGF signaling due to increased gal-3/polyLacNAc binding, EGFR signaling was significantly reduced when Mgat5 was knocked out, due to a decrease in gal-3/polyLacNAcs binding that resulted in the rapid internalization of EGFR from the cell surface.^[31] Of note, TN breast cancers demonstrate high EGFR expression in about 45–75% of cases.^[32]

In summary, we show the presence of multiple polyLacNAc N-glycans in HER2+, triple-negative, and metastatic breast cancers through the use of MALDI FT-ICR IMS. In addition to extending the numbers of glycans to be detected following sialic acid derivatizations, these glycan tissue maps will be used to begin analysis of the glycoprotein and glycopeptide carriers, specifically those modified with high mannose, β1,6 branched and LacNAc structures. Peptidomic and N-glycan MALDI-FTICR IMS analysis of a large tissue cohort of HER2−/ER+/PR+ disease is also ongoing. N-glycan correlate associated with immune infiltrating cells is an additional area for further investigation. Combined with other emerging glycomic and extracellular matrix methods, we expect continued tissue analysis by MALDI-FTICR IMS approaches of breast cancer genetic subtypes to continue to delineate the complex interactions of the glycocalyx with the stroma, immune, and tumor microenvironment.

Clinical Relevance.

The study represents a novel analysis and comparison of tumor N-glycan distributions in different genetic sub-types of breast cancer and in primary/metastatic tissue pairs. There are different distributions of specific N-glycans between HER2-positive and TNBC, likely reflective of underlying glucose metabolism pathways associated with these genetic sub-types. There were not major changes in N-linked glycosylation patterns detected between primary and metastatic tissues from the same donors, implying that the molecular events required for metastasis in the primary tumors was sufficient formigration and colonization. The presence of polylactosamine glycans in these tissues and the primary and metastatic breast cancer tissues highlights the structural complexity of N-linked glycosylation. As these structures are known to affect interaction of tumor cells with the immune system and other tumor microenvironment components, these findings provide newmechanistic insights into the development and progression of breast cancers. The presence of these glycans could be targeted for therapeutic strategies and further evaluation as potential prognostic biomarkers.