Retention of ⁶⁴Cu-FLFLF, a Formyl Peptide Receptor 1-Specific PET Probe, Correlates with Macrophage and Neutrophil Abundance in Lung Granulomas from Cynomolgus Macaques

Abstract

Neutrophilic inflammation correlates with severe tuberculosis (TB), a disease caused by Mycobacterium tuberculosis (Mtb). Granulomas are lesions that form in TB, and a PET probe for following neutrophil recruitment to granulomas could predict disease progression. We tested the formyl peptide receptor 1 (FPR1)-targeting peptide FLFLF in Mtb-infected macaques. Preliminary studies in mice demonstrated specificity for neutrophils. In macaques, ⁶⁴Cu-FLFLF was retained in lung granulomas and analysis of lung granulomas identified positive correlations between ⁶⁴Cu-FLFLF and neutrophil and macrophage numbers (R² = 0.8681 and 0.7643, respectively), and weaker correlations for T cells and B cells (R² = 0.5744 and 0.5908, respectively), suggesting that multiple cell types drive ⁶⁴Cu-FLFLF avidity. By PET/CT imaging, we found that granulomas retained ⁶⁴Cu-FLFLF but with less avidity than the glucose analog ¹⁸F-FDG. These studies suggest that neutrophil-specific probes have potential PET/CT applications in TB, but important issues need to be addressed before they can be used in nonhuman primates and humans.

Graphical Abstract

INTRODUCTION

Neutrophils play an indispensable role in immune responses against microbes and are among the first cells to be recruited to sites of microbial infection or tissue damage. Neutrophils are best known for their ability to phagocytose pathogenic organisms and kill them with antimicrobial peptides and reactive oxygen species, but they also secrete proteases including elastase and MMPs and produce extracellular traps (NETs) to entrap microbes. These responses are tightly regulated, and in the proper context, neutrophils perform functions that are critical for protection and resolution. In contrast, dysregulated neutrophil responses can exacerbate disease and damage uninvolved tissues.^1,2 This relationship between protection and pathology makes it important understand the relationship between neutrophils and tuberculosis (TB), a disease caused by the bacterium Mycobacterium tuberculosis (Mtb). Mtb infection causes granuloma formation in infected tissues, and while there is evidence that neutrophils can be protective in TB,^3–5 most studies, in mice,^6,7 nonhuman primates (NHPs),^8,9 and humans,^10–13 suggest that neutrophilic inflammation is pathogenic. Moreover, work in NHP granulomas has demonstrated that neutrophil numbers correlate with 2-deoxy-2-(¹⁸F)fluoro-d-glucose (¹⁸F-FDG; a glucose analog used as a PET probe to measure metabolically active cells and a proxy for inflammation^14–16) retention and elevated bacterial burdens in granulomas⁸ suggesting that neutrophils are among the cell types that contribute to pathology at the site of infection during active TB.

The relationship between neutrophil responses and exacerbated TB has generated substantial interest in developing probes to visualize neutrophil dynamics and treatment responses in tissues. Formyl peptide receptor (FPR) is a potential target for a neutrophil-specific PET probe. There are three FPR isoforms (FPR1, FPR2, and FPR3), and although this family of G protein coupled receptors is highly conserved in mammals, there is substantial variability between isoforms with respect to their amino acid sequences and substrate specificities.^17–20 FPR1 is the best-described isoform and mediates neutrophil chemotaxis toward short (3–5 amino acid) N-formylated peptides produced by bacteria^21,22 and released by mitochondria after tissue damage.^23,24 FPR2 has significant amino acid similarity with FPR1 but greater affinity for longer ligands,²⁵ formylated peptides produced by Staphylococcus aureus and Listeria monocytogenes,²⁶ eicosanoids, and acute phase proteins.^27,28 FPR1 and FPR2 expression is most closely associated with myeloid cells including neutrophils, monocytes, and macrophages but is also expressed by nonmyeloid cells including lymphocytes, epithelial cells, endothelial cells, and hepatocytes.²⁹ FPR3 is poorly characterized, and although it is expressed by most human myeloid cells other than neutrophils,^17,30 its functions and ligands remain largely unknown.¹⁷ Taken together, FPR1’s characteristics, including cell surface expression, well described ligand repertoire, and expression patterns, make this FPR isoform a potential option for a neutrophil-targeted PET probe.

Molecular imaging approaches based on cell-specific targets have been used for noninvasive monitoring of cell recruitment and function to otherwise difficult-to-sample tissues. This has been done with SPECT imaging of macrophages in murine models of rheumatoid arthritis,³¹ SPECT/CT imaging of atherosclerotic plaques in mice,³² including ex vivo optical imaging of activated macrophages in atherosclerotic mouse aortas,³³ and identification of TSPO-expressing macrophages in Mtb-infected mice by PET and SPECT imaging.³⁴ Less work has been done on neutrophil-focused imaging projects, and although much of this work uses intravital microscopic imaging techniques,³⁵ a ⁶⁸Ga-based citrate-coated nanoradiotracer based approach has been used for PET imaging of neutrophils after LPS-induced lung inflammation in mice,³⁶ thus demonstrating the feasibility of imaging strategies for in vivo neutrophil imaging. Studies with the FPR1 ligand N-cinnamoyl-F-_(D)L-F-_(D)L-F (FLFLF)³⁷ suggest this agent has potential for multimodal neutrophil imaging.³⁸ This probe has been used for imaging human neutrophils with fluorescence microscopy^39,40 and work in mouse models of inflammation,⁴¹ aortic aneurism,⁴² lung ischemia-reperfusion injury,⁴³ and intervertebral disk herniation,⁴⁴ demonstrate that technetium-99m labeled FLFLF may have applications in SPECT imaging. In MRI applications, gadolinium-labeled FLFLF has been tested in mice as a neutrophil-targeted contrast agent.⁴⁵ Neutrophilic inflammation occurs in microbial infections with protective and pathologic implications and the application of FPR1 targeting in this area were demonstrated by Locke et al.³⁷ with the use of ⁶⁴Cu-FLFLF for PET imaging to visualize neutrophilic recruitment to the lungs of mice infected with Klebsiella pneumonia. Considered together, these studies suggest that FLFLF may be valuable as a PET imaging agent for other infectious diseases that have not been evaluated in a larger animal model with human-like physiology.

A better understanding of neutrophil dynamics in TB may improve the ability to predict risks for developing active TB and for monitoring treatment responses. The absence of a neutrophil-specific probe that is suitable for PET imaging in human TB is a primary obstacle to this objective. To address this area of need, we evaluated the suitability of the FPR1-targeting probe FLFLF as a tool for following neutrophil dynamics in TB inMtb-infected cynomolgus macaques, a highly translatable model that replicates the pathology and disease course of human Mtb infection.^46,47 We found that FPR1 was present on the surface of peripheral blood neutrophils and confirmed that granulomas retained more probe than nondiseased lung. The ⁶⁴Cu-FLFLF probe showed avidity for lung granulomas by PET/CT, but with limited sensitivity, high levels of background in the lungs, and significant liver uptake. We found that FPR1 was expressed by neutrophils in lung and granulomas but also by granuloma macrophages and lung epithelial surfaces. Given the similarities between NHP and humans, these results highlight the challenges associated with using FPR-1 targeted probes for monitoring neutrophilic inflammation human TB.

MATERIALS AND METHODS

Reagents.

All chemicals were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO), unless otherwise specified. Aqueous solutions were prepared using ultrapure water (resistivity, 18 M). Rink amide 4-methylbenzhydrylamine resin (loading, 0.77 mmol/g) and all Fmoc protected amino acids were purchased from Chem-Impex International, Inc. (Wood Dale, IL). Fmoc-PEG₁₂ carboxylic acid was purchased from ChemPep (Wellington, FL. Cy3 carboxylic acid was purchased from Lumiprobe (Hallandale Beach, FL). Cinnamic acid was purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO). The chemical structures of the relevant probes, linkers, and conjugates are shown in Figure 1. Analytical and semipreparative reversed-phase high-performance liquid chromatography (HPLC) were performed on a Waters 1525 Binary HPLC pump (Milford, MA) with a Waters 2489 UV/visible detector and a model 106 Bioscan radioactivity detector (Bioscan inc., Washington, DC). Nonradioactive HPLC samples were analyzed on an analytical Jupiter C18 column and purified on a semipreparative Jupiter C18 column (Phenomenex, Torrance, CA). Radiochemistry reaction progress and purity were monitored on a Jupiter C18 column (Phenomenex, Torrance, CA). Radioactive samples were counted using either an automated Packard Cobra II gamma counter (Packard, Ramsey, MN) or a PerkinElmer 2470 WIZARD² Automatic Gamma Counter (Waltham, MA). PET/CT data were acquired using a Siemens Inveon Preclinical PET/CT (Siemens Medical Solutions, Knoxville, TN).

Figure 1.

Structure of cinnamoyl-F_(D)LF_(D)LFK-PEG₁₂-CB-TE1A1P and cinnamoyl-F_(D)LF_(D)LFK-PEG₁₂-Cy3; R= CB-TE1A1P or Cy3.

FLFLF Synthesis.

Cinnamoyl-F_(D)LF_(D)LF (0.1 mmol) was synthesized using Fmoc/tBu strategy on a Rink amide resin (loading, 0.77 mmol/g). First, Fmoc-Lys(Dde)–OH was attached to the resin, and then the F_(D)LF_(D)LF was assembled. Cinnamic acid was coupled to the free NH₂ terminal after Fmoc deprotection of the last amino acid. All couplings were accomplished in DMF using PyBOP (3 equiv, 0.3 mmol) as a coupling reagent and N,N-Diisopropylethylamine (DIEA) (9 equiv, 0.9 mmol). Couplings were monitored with the Kaiser color test. The peptide was thoroughly washed with DMF, DCM, and diethyl ether. After synthesis, peptide-resin (50 mg) was resuspended in DMF, the Dde protecting group was cleaved with 2% hydrazine in DMF (3 × 15 min), and the beads were washed with DMF (6 × 1 min). PEG linker Fmoc-PEG_12-COOH (3 equiv) was added to the beads and reacted for 2 h under agitation. The Fmoc was then deprotected with 25% piperidine in DMF and CB-TE1A1P carboxylic acid (3 equiv), or Cy3 carboxylic acid (3 equiv) solubilized in DMF was coupled under standard conditions for 3 h. The beads were thoroughly washed with DMF, DCM, and diethyl ether and dried under a vacuum. fMLF peptide (0.1 mmol) was synthesized using standard Fmoc/tBu strategy on a rink amide resin (loading, 0.77 mmol/g) with pyBOP as activator in the presence of DIEA. N-terminal formulation was done using formic acid (10 equiv) and DIC (20 equiv) for 1 h at room temperature. Cleavage of the peptides from the resin was achieved using a mixture of TFA (95%), triisopropylsilane (2.5%), and water (2.5%) for 3 h at room temperature. TFA was then evaporated; the crude peptides were precipitated with cold ether and purified on a semipreparative Jupiter C18 column (300 Å pore size, 5 μm particle size, 250 × 10 mm) using a Waters HPLC system. Solvent A was H₂O/0.1% TFA, and solvent B was CH₃CN/0.01% TFA. Elution was conducted at a flow rate of 3 mL/min, and detection was performed at 214 nm. Peptides were purified using a stepwise gradient from 20% to 100% B over 20 min for FLFLF peptides and 40% to 100% for fMLF peptide. The compounds were characterized by analytical HPLC on a Jupiter C18 column (300 Å pore size, 5 μm particle size, 150 × 4.6 mm) and verified for identity by electrospray mass spectroscopy (calculated MW 1903.2 Da, measured MW 1902.8 Da). The purity of all compounds was determined to be >98%.

Radiolabeling Chemistry.

⁶⁴CuCl₂ (t_1/2 = 12.7 h, β⁺; 17.8%, E_β+max = 656 keV, β⁻, 38.4%, E_β-max = 573 keV) was obtained from Washington University (St. Louis, MO) and University of Wisconsin (Madison, WI). The FLFLFK-PEG_12-CB-TE1A1P (1–10 μg) was mixed with ⁶⁴CuCl₂ (37–370 MBq) with a final volume of 100–500 μL in 0.3 M ammonium acetate buffer (pH = 6.5) and ethanol (1/1), Tween-20 (1%, 1–5 μL). The mixture was incubated at 50 °C for 1 h, and after incubation, the radiochemical purity of the ⁶⁴Cu-FLFLFK-PEG₁₂-CB-TE1A1P (herein referred to as ⁶⁴Cu-FLFLF) was monitored by radio-HPLC with a Jupiter C18 column (300 Å pore size, 5 μm particle size, 150 × 4.6 mm) or a TLC plate reader. HPLC was performed using a stepwise gradient from 20% to 100% B over 15 min. Solvent A was H₂O/0.1% TFA, and solvent B was CH₃CN/0.01% TFA. The radiochemical purity was >97%.

Binding Assay.

Cell experiments were performed to determine the binding affinity of ⁶⁴Cu-labeled cFLFLFK-PEG₁₂-CB-TE1A1P in FPR1 transfected 293T cells. Cells were seeded in poly-d-lysine coated 24-well plates (150000 cells per well) 24 h prior to the experiment. Before the experiment, cells were washed with 1 mL of HBSS twice and 0.5 mL of binding media (HBSS with 0.1% BSA, 1 mM Ca²⁺, and 1 mM Mg²⁺) was added to each well. Ten μg of fMLF peptide was then added to half of the wells as block to determine in vitro nonspecific binding, followed by ⁶⁴Cu-labeled cFLFLFK-PEG_12-CB-TE1A1P in increasing concentrations (0.2–80 nM). The samples were incubated for 4 h on ice. After incubation, the radioactive media was removed. Cell pellets were rinsed with ice cold binding buffer (1 mL) twice and dissolved in 0.5% SDS solution. The radioactivity in each fraction was measured in a well counter (Packard Cobra II gamma counter). The protein content of each cell lysate sample was determined (BCA Protein Assay Kit, Pierce). The measured radioactivity associated with the cells was normalized to the amount of cell protein present (cpm/mg protein). The K_d and Bmax were calculated using PRISM (GraphPad; San Diego, CA) software.

Animal Ethics Statement.

All animal studies were performed according to the Guide for the Care and Use of Laboratory Animals under the auspices of Division of Laboratory Animal Resources (DLAR) of the University of Pittsburgh. This research was approved by Institutional Animal Care and Use Committee, which adheres to national guidelines established in the Animal Welfare Act (7 U.S.C. Sections 2131–2159) and the Guide for the Care and Use of Laboratory Animals (8th Edition) as mandated by the U.S. Public Health Service Policy.

Small Animal PET/CT, Cell Isolation, and Analysis of the Murine Paw Inflammation Model.

A murine paw inflammation model^48,49 was used to test FLFLF avidity for inflamed tissue. Female ICR (n = 7, 6–8 weeks old) mice purchased from Taconic Laboratories (Albany, NY) were injected subcutaneously in the right footpads with Complete Freund’s Adjuvants (CFA; 50 μL/footpad) to induce local paw inflammation. After 24 h, the mice were injected intravenously via the lateral tail vein with ⁶⁴Cu-FLFLF (7.7–8.6 MBq) (n = 2). Mice were anaesthetized with 2% isoflurane, and small animal PET/CT was performed with static images were collected for 15 min at 6 and 18 h post injection. PET and CT images were coregistered with Inveon Research Workstation (IRW) software (Siemens Medical Solutions, Knoxville, TN). PET images were reconstructed with the Ordered-Subsets Expectation Maximization 3D (OSEM-3D).

Flow Cytometry on Mouse Samples.

Cy3-FLFLF was used to assess this peptide’s avidity for cells in tissue. At 24 h post CFA injection, 50 μg of Cy3-FLFLF was injected intravenously into the tail vein. At 4 h post probe injection, mice were euthanized and both paws were dissected. To obtain cells for flow cytometry, tissues were incubated with a mixture of collagenase D (2 mg/mL) and DNase-I (1/100, stock solution at 1 mg/mL) in DMEM/FBS (2%) for 30 min at 37 °C. Tissues were then collected and passed through a 70 μm mesh filter to create a single cell suspension. Red blood cells were lysed with ACK buffer (Thermo Fisher, Waltham, MA) and the remaining cells were washed with PBS. Cells were then labeled with fluorochrome-conjugated antimouse antibodies including anti-Ly6G (clone 1A8); anti-MHCII (clone M5/114.15.2; BioL-egend, San Diego, CA); anti-F4/80 (clone 6F12); anti-CD11b (clone M1/70); anti-CD3 (clone 17A2); for 30 min at 4 °C in staining buffer (BD bioscience). Unless otherwise indicated, all antibodies were purchased from BD Bioscience (San Jose, CA). Cells were analyzed by flow cytometry using a LSRFortessa (BD bioscience). Gating for neutrophils, macrophages, and T-cells was done first on the FSC/SSC characteristics of the cells and then by antigen expression. Neutrophils were identified as Ly6G +CD11b+ cells, macrophages were identified as F4/80+MHCII +, and T-cells were identified by CD3+ expression.

Immunofluorescence and Confocal Microscopy.

Immunofluorescence microscopy was performed on CFA-injected mouse paws as previously indicated.⁵⁰ Mice were injected with Cy3-FLFLF, and mouse paws were harvested at 5 h postprobe injection. Frozen sections (10 μm) were made using a cryotome; sections were then fixed with 4% PFA–PBS for 5 min at room temperature, and blocking was performed with 2% goat serum. Sections were then incubated with anti-CD16/CD32 Fc block (BD Bioscience) for 20 min at room temperature and then stained with rat anti-Mouse Gr1-FITC (clone RB68C5; Abcam, Cambridge, MA) followed by antirat IgG-AlexaFluor488 (Thermo Fisher). Slides were washed in PBS, and coverslips were mounted using Prolong Diamond antifade mounting medium (Thermo Fisher). Slides were imaged with a Zeiss Apotome system (Carl Zeiss Microscopy, Thornwood, NJ) equipped with a Zeiss HPO PL APO 63x oil immersion lens.

Macaque Studies.

Four adult cynomolgus macaques (Macaca fascicularis) were purchased from Valley Biosystems (Sacramento, CA). These animals, designated as 15413, 17713, 18414, and 18514 were infected intrabronchially with 3–92 CFU of Mtb (Erdman strain) and followed clinically as previously described.^46,51 Information on animal infections and duration of time in the study are included in Table S1 and PET/CT scans and probes included in this study are indicated in Table S2. Blood from animals involved in other ongoing studies and not dedicated to this project was used for some experiments. Macaques dedicated to this study were humanely euthanized prior to necropsy as previously described⁵¹ at 25, 18, 17, and 16 weeks post infection for 15413, 17713, 18414, and 18514, respectively.

PET and CT Imaging in Mtb-Infected Macaques.

PET and CT scans were performed on a hybrid preclinical system linking a Siemens Focus 220 microPET system (Siemens Molecular Solutions, Knoxville, TN) and a helical 8-slice Neurologica Ceretom CT scanner (Neurologica Corp., Danvers, MA). CT images were acquired following injection with 9 mL of ISOVUE 300 CT contrast agent (Bracco Diagnostics Inc., Singen, Germany) with the following X-ray parameters: 140 kV, 2 mA, “extra sharp” image setting, 250 mm helical scan length yielding 190 slices with a total scan time of 25 s with respiratory breath hold. PET images were acquired with the following parameters after designating the specific isotope, 600 s acquisition time; lower level energy cutoff, 350 keV; upper level energy cutoff, 650 keV; timing window, 6 ns. Resulting PET listmode files were presented as histograms with the following parameters, 3D histogram with no smoothing; span, 3; ring difference, 47; global average deadtime correction. PET images were reconstructed using an OSEM3D (Ordered Subset Expectation Maximum-3 Dimension) algorithm with CT-based attenuation, ramp projection filter, and scatter correction yielding a 284-slice image. Images were analyzed using OsiriX DICOM viewer program (Pixmeo, Geneva, Switzerland). PET probe uptake values in granulomas are reported in partial volume corrected maximum SUV (Standardized Uptake Value). To calculate partial volume correction, a standard curve was created from analysis of a phantom containing spheres ranging in size from 3.9 to 30.3 mm filled with radioactive isotope.

Processing Cynomolgus Macaque Blood and Tissues.

Whole blood was obtained from 12 cynomolgus macaques that were infected for 14 days (20515, 20615, 20715, 20815, 20915, 21015), 29 days (19515, 19615, 19715, 19815), or 54 days (20115, 20215). We lysed the RBCs with RBC lysing buffer (BD Biosciences) and stained the cells for CD3 (clone SP34–2), CD11b (clone ICRF44), CD14 (clone M5E2), and CD20 (clone 2H7). Unless otherwise indicated, antibodies were purchased from BD Biosciences. Neutrophils were identified as being CD11b_highCD14_lowSSC_high, and monocytes were identified as being CD11_lowCD14_highSSC_low. Cells were run on a LSRFortessa flow cytometer (BD Biosciences) maintained by the University of Pittsburgh’s Unified Flow Core, and data were analyzed using FlowJo (BD Biosciences).

For quantitative analysis of ⁶⁴Cu-FLFLF uptake by macaque tissues, macaques 18414 and 18514 were infused with tracer and imaged via PET/CT 18 h prior to necropsy (see below). Animals were maximally bled, humanely euthanized, and tissues were excised as previously described, using the prenecropsy scan as a guide for recovering identified lesions.^16,52 Tissues included lung granulomas, extrapulmonary granulomas, lymph nodes, blood fractions, liver, spleen, and kidney, and the pectoralis muscle was used as a negative (no uptake) control. Excised tissues were placed into preweighed conical tubes containing 10% neutral buffered formalin, weighed again to determine the mass of the tissue, and transferred to new tubes for quantifying γ emission with a PerkinElmer 2480 WIZARD γ counter. Formalin-fixed tissues were retained for embedding in paraffin and sectioning for immunohistochemistry. γ emissions for each sample were normalized to the tissue’s mass and expressed as CPM/gram.

Quantitative comparisons between cell populations in granulomas and normalized tissue γ counts were performed as previously done.⁴⁹ Briefly, immunohistochemistry (IHC) on 5 μm-thick FFPE tissue sections was performed as previously described^53,54 where six granulomas were selected for imaging per animal: the three with the highest CPM and the three with the lowest CPM. Tissue sections were stained for the most abundant cells in granulomas including CD3 (T cells; Dako, Carpintera, CA), CD20 (B cells; Thermo Fisher), and calprotectin (clone Mac378, Thermo Fisher) neutrophils. IHC was done as previously indicated,^53–55 with anti-CD3 stained with a donkey antirabbit Cy5-labeled secondary antibody (Jackson ImmunoResearch, West Grove, PA) and followed with anti-CD20 and anticalprotectin antibodies labeled with Zenon labeling reagents (Thermo Fisher). Stained sections were mounted with DAPI-containing ProLong Gold mounting medium (Thermo Fisher) and imaged by confocal microscopy. Multiple overlapping microscopic fields encompassing the whole tissue section were imaged and combined into a single complete image with Image Composite Editor (Microsoft, Redmond, WA) and the marker-positive cells and nuclei were quantified with CellProfiler.⁵⁶ There is no single antigen defining all macrophages in primate granulomas,⁵³ and approximate macrophage numbers per granuloma were extrapolated by subtracting the number of T cells, B cells, and neutrophils from the total number of nuclei per granuloma.

Statistics.

Several variables (CPM, CPM/gram, cell/gram, and mass) were log₁₀-transformed and then tested for normality using the Shapiro-Wilk test. For normally distributed data, two independent group means were compared using a student’s t test. For non-normal data, the Mann–Whitney test was used to compare two independent group means and the Wilcoxon matched-pairs signed rank test was used to compare two dependent group means. Relationships between CPM/gram and cell/gram were tested using linear regression after the data were log₁₀ transformed. All statistical tests were two-sided with a type I error rate of 0.05. Analyses were conducted using GraphPad Prism version 8.0 (GraphPad Software, La Jolla, CA).

RESULTS

Preparation, Stability, and Function of ⁶⁴Cu-labeled FLFLF.

Optimization of the conditions for preparing ⁶⁴Cu-FLFLF was required to minimize aggregation. The final formulation included acetate buffer (0.3 M) pH 6.5 (50%), ethanol (50%), and Tween-20 (1%) at 50 °C for 1 h, which resulted in high specific activity labeling of 37 MBq/μg with less than 5% retention in the labeling vials and syringes. The stability of ⁶⁴Cu-FLFLF was determined in freshly prepared macaque serum at 37 °C (Figure S1). ⁶⁴Cu-FLFLF conjugate was highly stable with respect to enzymatic degradation, showing no decomposition or free copper out to 20 h incubation (Figure S1).

Cell Binding Assay.

In vitro binding assays were used to confirm that FLFLF bound FPR1 but did not activate cells. The functionality of the FPR1 receptor was validated by the ability of fMLF, a peptide agonist of FPR1, to bind and induce internalization the receptor (Figure 2A). Addition of FLFLF-PEG₁₂ did not lead to internalized receptor, indicating that this peptide was not inducing FPR1-mediated signaling in the cells. Saturation-binding assays demonstrated that ⁶⁴Cu-FLFLF-PEG₁₂ bound FPR1, with a K_d of 37 ± 11 nM and a B_max of 3300 ± 490 fmol/mg (Figure 2B). These data demonstrate that our FPR1-targeting peptide was functional in vitro and did not induce receptor internalization and signaling or activation. On the basis of these data, we moved forward with in vivo studies measuring the binding and specificity of FLFLF in mice and peripheral blood cells from macaques.

Figure 2.

FLFLF binds FPR1 but does not activate FPR1-expressing cells in vitro. (A) HEK293T cells were transformed to express GFP-FPR1 (green) to confirm that FLFLF treatment did not lead to nonspecific activation. Untreated controls, Formylmethionyl-leucyl-phenylalanine (FMLP)-treated positive controls, and FLFLF-treated cells were imaged to validate that the FPR1 construct was functional and internalized (arrows) upon FMLP treatment and that pegylated FLFLF did not lead to FPR1 signaling, as indicated by receptor internalization, before PEG₁₂-FLFLF was tested. DAPI-stained nuclei are indicated in blue. (B) FLFLF PEG₁₂ was tested for its binding affinity for FPR1.

FLFLF Is Retained in Mouse Tissues after Injection of Mycobacterial Antigens.

To assess FLFLF retention in inflamed tissues, we performed a pilot study where we injected Complete Freund’s Adjuvant in a mouse’s paw to induce inflammation⁵⁰ and then used intravenous Cy3-FLFLF administration to identify FLFLF retention in injected and control contralateral paws. Using flow cytometry, we found a trend toward neutrophils binding more Cy3-FLFLF than macrophages and T cells, and after subtracting for background, 89.1% of neutrophils bound Cy3-FLFLF vs 1.5% of macrophages and 1.7% of T cells were Cy3-FLFLF+ (Figure 3A). Cy3-FLFLF labeled neutrophils could be detected in tissues by immunofluorescence, although other non-Gr-1-stained cells were detectable (Figure 3B). We then injected ⁶⁴Cu-FLFLF into the mice and imaged these animals by PET/CT at 6 and 18 h post injection, we observed ⁶⁴Cu-FLFLF retention in the injected paw relative to the contralateral paw (Figure 3C). Taken together, these data show that FLFLF is bound by neutrophils and retained in inflamed tissues but also suggest that other cells at sites of inflammation can contribute to FLFLF retention.

Figure 3.

Testing FLFLF in inflamed mouse paws. (A) Mice were injected with Cy3-FLFLF in and Cy3-labeled cells in inflamed paws were identified by flow cytometry to show that neutrophils bound FLFLF in vivo. In the mouse pictured, both hind paws were injected to maximize the number of cells recovered for flow cytometry. (B) Histology of inflamed paw tissue from a mouse that was injected with Cy3-FLFLF and stained for Gr-1 (green), and nuclei (DAPI, blue) showing variable amounts of colocalization of Cy3-FLFLF and Gr-1. Scale bar represents 20 μm. (C) ⁶⁴Cu-FLFLF was imaged at 6 or 18 h post injection (n = 2 mice/time point; representative data shown).

Assessment of FPR1 as a Target for a Neutrophil-Based PET Probe in Macaques.

We used immunofluorescence microscopy and flow cytometry to confirm that FPR1 was expressed on macaque neutrophils. We stained RBC-lysed whole blood for calprotectin (also known as S100A8/9, used here as a marker for neutrophils⁵³) and FPR1. We noted that neutrophils costained for FPR1 and calprotectin while FPR1 was also occasionally observed on monocyte-like cells (solid arrow, Figure 4A) but not on lymphocyte-like cells (open arrowhead, Figure 4A). Using flow cytometry (Figure 4B), we found little FPR1 expression on B cells, T cells, and unclassified lymphocyte-sized CD3⁻CD20⁻ cells (Figure 4C). Neutrophils and monocytes expressed more FPR1 than B cells and T cells, and neutrophils and monocytes expressed more FPR1/cell than unclassified lymphocyte-sized CD3-CD20- cells (Figure 4C).

Figure 4.

FPR1 is expressed on neutrophils and monocytes in peripheral blood and by multiple cell types in lung tissue. (A) Immunofluorescence of peripheral blood cytospins showing calprotectin (green) positive cells with segmented nuclei (DAPI; blue) that costain with FPR1 (red). (B) Frequencies of FPR1-expressing peripheral blood assayed by flow cytometry (red) compared against a similarly stained population of cells without FPR1 staining (blue). Cells were gated according to surface marker expression and sidescatter (SSC) profiles to identify SSC(_low) lymphocyte-sized cells including CD20+ B cells, CD3+ T cells, and unclassified CD3-CD20- lymphocyte-sized cells, CD11_lowCD14_highSSC_low monocytes, and CD11b_highCD14_lowSSC_high neutrophils. (C) Comparison of peripheral blood FPR1 expression, defined as frequency of gated cells for each cell subset (left) or geometric mean fluorescence intensity (right). * = p < 0.05, ** = p < 0.005, *** = p < 0.0005, and **** = p < 0.0001; Kruskal–Wallis test. (D) FPR1 (green) is expressed in multiple cell types and sites in the lung. Ciliated epithelium and submucosa (top) and epithelioid macrophages (bottom) are also noted to express FPR1. Arrows indicate strongly FPR1-positive neutrophils (calprotectin; red) and arrowheads indicate neutrophils that express negligible or low levels of FPR1.

Peripheral blood is easily sampled, but results from blood may not replicate FPR1 expression in granulomas. To determine which cells and tissues express FPR1 in the lung, FFPE granuloma sections were stained for FPR1 and calprotectin (Figure 4D) for confocal microscopy. We identified FPR1-expressing neutrophils (Figure 4D, arrows) in granulomas but noted substantial variability in FPR1 among neutrophils with some strongly staining for FPR1 (Figure 4D, arrows) while others expressed little or no FPR1 (Figure 4D, arrowheads). We also noted FPR1 expression by epithelioid macrophages in granulomas, chondrocytes in hyaline cartilage, the luminal surface of ciliated bronchial epithelium, and in the lamina propria (Figure 4D, top right panel; Figure S2). When the blood and tissue-level data are considered, our results indicate that neutrophils express FPR1 in granulomas but suggest that other cell types and tissues may also contribute to the PET signal obtained with FPR1-targeted probes.

To quantitatively compare FLFLF’s specificity for granulomas vs other tissues, ⁶⁴Cu-FLFLF was injected into two cynomolgus macaques (18414 [red] and 18514 [blue]), and tissues were harvested after their last ⁶⁴Cu-FLFLF scan to measure their emissions on a γ counter.⁴⁹ We sampled tissues including blood, pectoralis muscle, uninvolved (nondiseased) liver, spleen, and kidney (Figure 5A). Overall, ⁶⁴Cu-FLFLF retention in blood and muscle was very low. We tested the fractions produced by Percoll gradient separation including the RBC pellet (mostly RBCs and neutrophils) and the lymphocyte- and monocyte-rich buffy coat and found that most of the FLFLF in peripheral blood remains in the plasma fraction rather than being bound to blood cells (Figure S3). Compared to blood, liver, spleen, and kidney retained substantially more probe, and these organs had high CPM/gram values.

Figure 5.

⁶⁴Cu-FLFLF retention differs by tissue and disease state. (A) γ counts per minute (CPM) were normalized to tissue mass for tissues from two animals, 18414 (red) and 18514 (blue), and the log-transformed data was plotted on a log₁₀ scale. Each point represents a different sample, pairwise comparisons by the Mann–Whitney test. The black horizontal lines indicate the overall median value whereas the red and blue horizontal lines in the lung, lung granulomas, lymph nodes, and lymph node granulomas indicate the median value for 18414 and 18514, respectively. (B) FFPE sections of inguinal (left), cranial hilar (center), and mainstem bronchus (right; green arrow in panel A) lymph nodes from monkey 18415 were stained for T cells (red), B cells (blue), and neutrophils (green) and to identify the basis for the mainstem bronchus lymph node’s high specific activity. Arrows indicate granulomas within the carinal hilar and mainstem bronchus lymph nodes.

FLFLF retention was assessed in uninvolved lung, lung granulomas, uninvolved lymph nodes, granulomatous lymph nodes, and extrapulmonary granulomas. Nondiseased lung had low levels of ⁶⁴Cu-FLFLF retention whereas lung granulomas retained significantly more activity. Normal and granulomatous lymph nodes retained more ⁶⁴Cu-FLFLF than normal lung or granulomas but were not significantly different from each other. The mainstem bronchus lymph node from monkey 18414 was a notable outlier and had a high level of activity (1566550 CPM/gram). FFPE sections from this lymph node were stained for CD3 (T cells), CD20 (B cells), and calprotectin (neutrophils) and compared to an uninvolved inguinal lymph node (110,151 CPM/gram) and a granuloma-containing hilar lymph node (213,492 CPM/gram). The uninvolved and granuloma-containing lymph nodes imaged here had relatively few neutrophils, whereas the mainstem bronchus lymph node contained a granuloma with large numbers of neutrophils (Figure 5B) suggesting the high ⁶⁴Cu-FLFLF signal in this tissue was attributable to an abundance of neutrophils. Liver granulomas from monkey 18514 had ⁶⁴Cu-FLFLF retention profiles that were similar to lung granulomas, but lower than uninvolved liver, suggesting either ⁶⁴Cu-FLFLF retention in normal liver is attributable to probe uptake by hepatocytes or liver-resident Kupffer cells have higher levels of FRP1 expression than the cells in granulomas. Taken together, our results suggest there are organ-specific patterns of FPR1 expression and ⁶⁴Cu-FLFLF retention but also indicate that in the lung, granulomas retain more ⁶⁴Cu-FLFLF than uninfected lung tissue.

Macrophage and Neutrophil Numbers Correlate with Lung Granuloma ⁶⁴Cu-FLFLF Retention.

Macaque granulomas are complex structures that contain multiple cell types.⁵⁷ To determine which cell populations are responsible for ⁶⁴Cu-FLFLF retention in lung granulomas, we did immunohistochemistry and image analysis on a subset of lesions from animals 18414 and 18514 to identify cell populations that best correlated with probe retention (Figure 6). We selected granulomas with different specific activities when normalized for mass but with similar masses and cell numbers as defined by the number of nuclei/section (Figure 6A). The granuloma were stained for CD20+, CD3+, and calprotectin, and the number of cells per section were quantified. Macrophages are phenotypically diverse⁴⁹ and cannot be assessed with any single marker; thus the remainder of non-B cell, T cell, and neutrophil nuclei were used as surrogates for approximate macrophage numbers. To measure the relationship between cell frequencies and ⁶⁴Cu-FLFLF retention, cell numbers were correlated against the log₁₀ of the CPM/gram for each cell type. We found significant positive correlations between all cell types and γ counts/min with the strongest associations between cells/gram and CPM/gram for macrophages and neutrophils (Figure 7B). These data indicate that multiple cell populations are associated with ⁶⁴Cu-FLFLF retention in granulomas but suggest that macrophage and neutrophil populations may be driving much of the PET/CT signal/granuloma for this probe.

Figure 6.

Granuloma macrophage and neutrophil numbers strongly correlate with⁶⁴Cu-FLFLF retention. (A) Three granulomas with the lowest and highest CPM/gram profiles were selected from monkeys 18415 (red) and 18514 (blue) and the basic parameters including emissions profile, mass, and number of cells (nuclei) were compared. Lines represent medians, t test p-values are reported. (B) Correlations between cells/gram and CPM/gram for tissues analyzed by immunohistochemistry. Dashed line represents the linear regression and F-test p-value and R² are reported. Each dot represents a granuloma.

Figure 7.

FPR1-targeted PET probes can identify granulomas in Mtb-infected macaques. (A) PET/CT profiles of ¹⁸F-FDG and ⁶⁴Cu-FLFLF probed cynomolgus macaques. Arrows indicate granulomas in 18415 while numerous granulomas can be seen in the left (L) lobes of monkey 18514. Left and right (R) sides of the PETCT scans and liver are indicated. (B) Longitudinal analysis of ⁶⁴Cu-FLFLF (purple) and ¹⁸F-FDG (green) uptake in four animals showing that FPR1 probed granulomas have significantly (p < 0.05) lower SUVs than FDG. Pairwise comparisons between adjacent time points by the Wilcoxon matched-pairs signed rank test. Each dot represents a granuloma and each line connects the same granuloma over time. ** p < 0.01, *** p < 0.001, **** p < 0.0001.

We injected Mtb-infected macaques with ⁶⁴Cu-FLFLF to assess the probe’s utility in PET/CT imaging and compared it against the well-characterized behavior of ¹⁸F-FDG^47,58,59 (Figure 8A). Consistent with previous studies,^{47,49,51,52 18}F-FDG-avid granulomas were readily identifiable above lung background while liver uptake was minimal (Figure 8A). Granulomas did retain ⁶⁴Cu-FLFLF and had avidity over background lung, but probe uptake was substantially lower than with ¹⁸F-FDG (Figure 7A). We also performed a longitudinal study comparing the performance of the different probes within an animal, but not between animals, and found that ⁶⁴Cu-FLFLF had significantly lower retention than ¹⁸F-FDG (Figure 7B) at all measured time points. Taken together, our data demonstrate that FPR1-targeted probes are retained by lung granulomas but with less avidity and less intergranuloma and intertime point range than ¹⁸F-FDG. Moreover, high levels of liver uptake (Figure 7A) and signal from non-neutrophil cell types that bind FLFLF in lung further complicate the use of this probe for detection of granulomas or definitive assessment of neutrophilic involvement in primate TB models.

DISCUSSION

FPR1-targeting agents have been developed for neutrophilic inflammation in mice,^37,38 and a similar approach for monitoring neutrophil recruitment to tuberculous granulomas or other TB pathologies would be valuable for assessing disease risk and treatment responses in people with TB. To this end, we tested ⁶⁴Cu-FLFLF as a PET agent in NHPs to determine whether this can be used to visualize neutrophils in lung granulomas and follow tissue-level neutrophil dynamics over the course of TB. A strength of our study is our use of an animal model that replicates the spectrum of disease and pathology seen in human TB and the use of a highly translational PET/CT imaging approach that has been thoroughly evaluated for nonhuman primate and human TB. Limitations in our study include the small sample sizes and absence of female macaques. Although we have not previously noted sex-related differences in male and female macaques to Mtb infection,⁶⁰ only using male macaques for the end point γ emission study and longitudinal imaging experiments is a weakness in this project. With these caveats in mind, we found this probe was taken up and retained by lung granulomas in Mtb-infected macaques, but it lacked the sensitivity or specificity needed to accurately quantify neutrophils. These results differ from work using ⁶⁴Cu-FLFLF in a mouse model of Klebsiella pneumoniae infection,³⁷ possibly reflecting differences in K. pneumonia infection, which leads to a neutrophil-rich pneumonia, while Mtb infection produces granulomas that are highly variable in size and neutrophil content. Moreover, our ability to assess the neutrophil contribution to probe avidity was confounded by likely FLFLF retention by macrophages, which has been previously noted with this probe,⁴³ and to a lesser extent, lymphocytes and other FPR1-expressing lung cells. These potential problems were compounded by the probe’s hydrophobicity and liver uptake. Even with these challenges, our results suggest that an optimized probe that targets an abundant cell surface receptor on neutrophils with more desirable pharmacokinetics could be used to follow neutrophil dynamics in TB.

¹⁸F-FDG uptake has been used as a proxy for neutrophilic inflammation in lung disease^15,61 and is used as a tool for diagnosing and monitoring TB progression in humans and NHPs.^{47,49,52,58,61–64} This combination of applications raises questions as to how well FDG correlates with neutrophil infiltration into granulomas and whether ¹⁸F-FDG retention is equivalent to granuloma neutrophil content. We found that granulomas retained significantly more ¹⁸F-FDG than ⁶⁴Cu-FLFLF with little intergranuloma variability in ⁶⁴Cu-FLFLF retention demonstrating a significant disconnect between granuloma glucose metabolism (FDG) and FPR1 expression. These findings suggest that FDG retention by granulomas is driven by multiple cell types, not just neutrophilic inflammation, and highlights a caveat that needs to be applied when considering how ¹⁸F-FDG PET data are related to specific cell types in TB granulomas.

Our results also highlight important considerations when designing cell-type specific probes and applying them to nonmurine systems including macaques and humans. First, while peripheral blood can be easily accessed and serially sampled, cellular activation states and protein expression on peripheral blood cells may be substantially different from the same cell in tissues or sites of infection. Similarly, tissue resident cells can express an unexpectedly broad and variable range of proteins and receptors that may confound the interpretation of results, and this problem is exacerbated by the immune events associated with microbial infection. Moreover, the overall abundance of a particular cell type needs to be factored into probe design. In our study, granulomas retained more ⁶⁴Cu-FLFLF than normal lung, but variability in the number of neutrophils per granuloma, coupled with the unexpectedly broad range of cells that express FPR1, limited our ability to interpret these results in lung. In comparison, the results seen in murine Klebsiella infection, which induces an inflammatory response that is much richer in neutrophils than most TB granulomas, suggest that FPR1-targeted probes may be best suited for monitoring highly neutrophilic lung infections. Finally, when using different animal models for probe development, it is important to consider each species’ unique physiologic traits when interpreting the results. The physiology of a mouse is significantly different than an NHP or human and these differences complicate interpretation of results.

A highly specific neutrophil probe would be useful for investigating relationships between neutrophil numbers or timing of neutrophilic inflammation at the site of infection and disease outcome and would enable treatment regimens to be fine-tuned to improve infection outcomes. Thus, neutrophils remain an important target for PET probe development in TB research. Our results indicate that ⁶⁴Cu-FLFLF is taken up in TB granulomas at a higher level than uninvolved lung, but this agent cannot accurately quantify neutrophils in NHPs with TB, and suggest that an agent with greater specificity is needed. Alternative approaches including intravascular addition of fluorochrome-labeled antibodies before necropsy, as has been done for macaques with TB,⁶⁵ have promise but are limited by an inability to visualize granuloma dynamics over the course of disease. This obstacle could be overcome by incubating isolated neutrophils with a PET tracer before returning them to circulation, as has been done with neutrophils labeled for SPECT imaging in nonsmall cell lung cancer patients.⁶⁶ This labor-intensive procedure carries a risk of exogenously activating the neutrophils, thus an approach using a neutrophil-specific probe would alleviate this concern. Future efforts in this area, identifying highly expressed neutrophil surface proteins or probes that can be selectively taken up by neutrophils and delivery systems with favorable physiologic profiles, will improve our understanding of how neutrophils influence TB progression at the local and systemic levels.