Non-invasive imaging of CD4+ T cells in humanized mice

Abstract

Antibody-based Positron-Emission Tomography (immunoPET) with radiotracers that recognize specific cells of the immune system provides an opportunity to monitor immune cell trafficking at the organismal scale. We previously reported the visualization of human CD8+ T cells, including CD8+ tumor infiltrating lymphocytes (TILs), in mice using a humanized CD8-targeted minibody. Given the important role of CD4+ T cells in adaptive immune responses of health and disease including infections, tumors, and autoimmunity we explored immunoPET using an anti-human-CD4 minibody. We assessed the ability of [⁶⁴Cu]Cu-NOTA-IAB41 to bind to various CD4+ T cell subsets in vitro. We also determined the effect of the CD4-targeted minibody on CD4+ T cell abundance, proliferation, and activation state in vitro. We subsequently evaluated the ability of the radiotracer to visualize CD4+ T cells in T cell rich organs and orthotopic brain tumors in vivo. For the latter, we injected the [⁶⁴Cu]Cu-NOTA-IAB41 radiotracer into humanized mice that harbored intracranial patient-derived glioblastoma (GBM) xenografts and performed in vivo PET, ex vivo autoradiography, and anti-CD4 IHC on serial brain sections. [⁶⁴Cu]Cu-NOTA-IAB41 specifically detects human CD4+ T cells without impacting their abundance, proliferation, and activation. In humanized mice [⁶⁴Cu]Cu-NOTA-IAB41 can visualize various peripheral tissues in addition to orthotopically implanted GBM tumors. [⁶⁴Cu]Cu-NOTA-IAB41 is able to visualize human CD4+ T cells in humanized mice and can provide non-invasive quantification of CD4+ T cell distribution on the organismal scale.

INTRODUCTION

Antibodies targeting immune checkpoints induce durable tumor regression and prolonged survival in subsets of patients with primary and brain resident metastatic melanoma and non-small cell lung cancer (1–3). However, many patients do not respond. Further development of effective immunotherapeutic candidates will require a deeper understanding of the effect of these drugs on immune cell activation and in vivo immune cell localization (4). This is particularly relevant for CD8+ and CD4+ T lymphocytes which migrate between the tumor and lymphatic organs to develop effective anti-tumor immunity (5).

Compared to CD8+ T cells, the contribution of specific CD4+ T cell subsets toward the anti-tumor immune response is less well characterized (6). Studies in mouse cancer models demonstrated that CD4+ T cells were necessary to maintain a CD8+ anti-tumor immune response (7,8). Further, clinical vaccine studies suggest that CD4+ T cells can mediate the development and maintenance of anti-tumor immunity (9,10). CD4+ T cells are polyfunctional cells which differentiate into distinct subtypes with different implications for anti-tumor immunity. Upon activation naïve CD4+ T cells can differentiate into conventional helper T cell (T_H) subtypes or non-conventional regulatory T cells (T_regs). Conventional CD4+ T cells drive anti-tumor CD8+ T cell effector function, provide CD40 ligand signaling to induce B cell antibody production, and are capable of directly inducing tumor cell lysis (11). Collectively, T_regs are a subset distinct from the conventional CD4+ effector lineage and are known to mediate tumor-associated immunosuppression (12).

The need to develop non-invasive diagnostic platforms to monitor immune cell distribution is particularly urgent for patients with glioblastoma (GBM), the most common primary malignant brain tumor in adults. Antibodies targeting the PD-1 immune checkpoint in malignant glioma have failed to show clinical activity in Phase 3 trials despite evidence for T cell activation in GBM tumor biopsies (13,14). There is growing evidence implicating CD4+ T cells in the anti-tumor response to PD-1/PD-L1 immune checkpoint blockade. For example, CD4+ T cells express PD-1 in lung cancer patients and the depletion of CD4+ T cells in an orthotopic mouse model of lung cancer reduced anti-PD1 therapeutic efficacy (15). Further, CD4+ T cells were required for anti-PD-1 response in mouse models of melanoma (16). Determination of CD4+ T cell abundance in the GBM tumor microenvironment and of CD4+ T cell distribution on the organismal scale may hence provide insight into immunotherapeutic resistance in GBM patients. However, repeated access to tumor tissue samples is difficult and the tools do not exist to measure immune cell densities in brain tumors throughout treatment course. There is a need for non-invasive immune cell monitoring tools to enable a thorough and rigorous study of immunotherapeutic resistance and response in GBM and other brain tumors. Non-invasive visualization of CD4+ T cells circumvents invasive biopsy and enables the study of CD4+ T cell kinetics in brain tumors over time.

To address this need, we utilized an anti-human-CD4 immuno-Positron Emission Tomography (PET) tracer. The ideal CD4 immunoPET reagent for brain tumor patients would enable high target-to-background ratio, rapid imaging turn around, and would not display unwanted biological activities. Prior to our work, a mouse-specific cys-diabody-based immunoPET tracer non-invasively detected mouse CD4+ T cells in mouse models of colitis (17) and hematopoietic stem cell expansion (18). However, the cys-diabody induced the depletion of CD4+ T cells which may limit the clinical applicability of this radiotracer (19). Another mouse specific CD4-targeted F(ab)’2 immunoPET tracer detected mouse CD4+ T cells following anti-PD-1 checkpoint blockade in syngeneic skin, mast cell, colon, kidney, and breast cancer models at 24 hours post radiotracer injection (20). Both the CD4-targeted cys-diabody and F(ab)’2 radiotracers are mouse specific, limiting clinical translation. To address this gap and develop a radiotracer for use in patients, we evaluated a humanized CD4-targeted immunoPET tracer to detect human CD4+ T cells. We opted to evaluate a human specific anti-CD4 minibody as the serum half-life of a minibody enables high tumor-to-background ratios within 24 hours of radiotracer injection and the truncated Fc region renders the Fc region of the minibody biologically inert. In summary, a human CD4-targeted minibody-based approach satisfies the criteria for optimal immunoPET imaging.

Several prior studies have used humanized immune system (HIS) mouse models to evaluate imaging of human immune cells using novel PET tracers, but none of these studies explored imaging of brain tumors (21–24). Our group recently reported HIS mouse models of GBM to evaluate the capacity of a human CD8-targeted minibody to detect tumor infiltrating lymphocytes (TILs) in the brain tumor microenvironment (TME) (25). In our current study, we seek to expand immunoPET to other immune cell types and, specifically, to evaluate the ability of a CD4-targeted minibody to detect CD4+ T cells in peripheral organs as well as brain TILs.

MATERIALS AND METHODS

Radiolabeling and Synthesis of [⁶⁴Cu]Cu-NOTA-IAB41

The radiotracer synthesis and radiolabeling approach is depicted in Figure 1A. An 80-fold excess of 2-S-(4-Isothiocyanatobenzyl)-1,4,7-triazacyclononane-1,4,7-triacetic acid (p-SCN-Bn-NOTA) (Macrocyclics, CAT# B-605) was incubated with the humanized anti-CD4 minibody (IAB41, ImaginAb) in metal-free PBS at pH 8.8 for 4 hours at 4°C. NOTA-IAB41 was subsequently purified using Zeba spin column (Fisher Scientific, Molecular Weight Cut Off 7 kDa). [⁶⁴Cu]CuCl₂ was obtained from the University of Washington St. Louis. NOTA-IAB41 (0.2 mg) was diluted to a total reaction volume of 100 μL metal-free 250 mM ammonium acetate and incubated with 74 MBq (2 mCi) of ⁶⁴Cu at 42°C for 30 min at pH 5.5 and purified using Zeba spin column into metal-free PBS. Radiochemical purity and stability in human serum were assessed using thin later chromatography using a 50 mM EDTA (pH 5) eluent. Specific activity of [⁶⁴Cu]Cu-NOTA-IAB41 ranged between 296–444 MBq/mg (8–12 mCi/mg).

Figure 1. [⁶⁴Cu]Cu-NOTA-IAB41 synthesis and radiolabeling .

(A) Synthesis and radiolabeling workflow of [⁶⁴Cu]Cu-NOTA-IAB41 (25). (B) Radiolabeling and purification of [⁶⁴Cu]Cu-NOTA-IAB41 quantified by iTLC. CPM = counts per minute. (C) Stability of [⁶⁴Cu]Cu-NOTA-IAB41 in human serum quantified by iTLC. (D) Recombinant human CD4 in vitro bead-based immunoreactivity assay. Excess NOTA-IAB41 = 100-fold excess. For all panels error bars are the SD. **** = P ≤ 0.0001.

Animal Models

All animal experiments were performed in accordance with and with approval of the Institutional Animal Care and Use Committee (IACUC) at Memorial Sloan Kettering Cancer Center (MSKCC) (New York, NY). To generate humanized immune system (HIS) mice, 10⁷ HLA-typed peripheral blood mononuclear cells (PBMCs) from individual donors (StemCell Technologies, CAT# 70025) were injected into the tail vein of a 6 to 8-week-old female CIEA NOG mice (Taconic, RRID; ISMR_TAC:NOG). GBM patient derived xenografts (PDX) 160721-1 and 160615-1 were derived from recurrent World Health Organization (WHO) grade IV GBMs as previously described (25). GBM PDX-bearing HIS mice were generated using methods previously described (25). No Mycoplasma testing was performed. All experiments were performed on PDX passages 4–5. Three weeks following intracranial tumor implantation, mice were injected with PBMCs as described above.

Protein Binding Assay

The protein binding assay was conducted using previously described methods (26) and used recombinant human CD4 protein (Sino Biological, CAT# 10400-H08H-B-20).

Flow Cytometry

For each flow cytometry experiment pooled cells from all samples were used to generate fluorescence minus one (FMO) and unstained controls. Ultracomp eBeads (ThermoFisher Scientific, CAT# 01-2222-41) were stained to generate single color controls. Analysis was performed using FlowJo (Version 10, RRID:SCR_008520). Flow cytometry was performed on the BD LSRFortessa.

Peripheral Blood Mononuclear Cell (PBMC) Flow Cytometry

Human PBMCs were stained with BV421-CD4 (BioLegend, OKT4, CAT# 304222, RRID: AB_2562134), PE/Cy7-CD25 (BioLegend, M-A251, CAT# 356108, RRID: AB_2561975), PE-CD45RA (BioLegend, HI100, CAT# 304108, RRID: AB_314412), PerCP/Cy5.5-CD45RO (BioLegend, UCHL1, CAT# 304222, RRID: AB_2174124), BUV395-CD3 (BD Biosciences, UCHT1, CAT# 563548, RRID: AB_2744387), APC-FOXP3 (Thermo Scientific, PCH101, CAT# 17-4776-42, RRID: AB_1603280), FITC-Ki67 (BD, B56, CAT# 556026, RRID: AB_396302). T cells activation with phorbol-12-myristate 13-acetate/ionomycin (PMA/iono) was performed according to the manufacturer’s protocol (BioLegend, CAT# 423301).

Spleen Flow Cytometry

Spleen single cell suspensions were prepared using methods previously described (25). Splenocytes were stained with BV711-CD4 (Biolegend, SK3, CAT# 344648, RRID: AB_2734350), BUV395-CD3 (BD Biosciences, UCHT1, CAT# 563548, RRID: AB_2744387), and 7-AAD viability staining solution (eBiosciences, CAT# 00-6993-50).

Whole Brain Flow Cytometry

Following dissection, whole brains were mechanically homogenized and passed through a 100-μM sterile filter. Cells were treated with a CNS-Specific Percoll Isotonic solution (Percoll – Cytiva) to remove myelin. Brain cells were stained with BV421-CD4 (BioLegend, OKT4, CAT# 304222, RRID: AB_2562134), PE/Cy7-CD25 (BioLegend, M-A251, CAT# 356108, RRID: AB_2561975), PE-CD45RA (BioLegend, HI100, CAT# 304108, RRID: AB_314412), PerCP/Cy5.5-CD45RO (BioLegend, UCHL1, CAT# 304222, RRID: AB_2174124), BUV395-CD3 (BD Biosciences, UCHT1, CAT# 563548, RRID: AB_2744387), APC-FOXP3 (Thermo Scientific, PCH101, CAT# 17-4776-42, RRID: AB_1603280) and LIVE/DEAD Fixable Green Dead Cell Stain Kit (ThermoFisher Scientific, CAT# L23101P).

Biodistribution Study

Mice were administered [⁶⁴Cu]Cu-NOTA-IAB41 via tail vein and scarified at indicated time points by CO₂ asphyxiation. Organs were collected and levels of radioactivity for each organ were measured using a PerkinElmer γ-counter. Normalized tracer uptake, which is expressed by percentage of injected dose per gram of body weight (%ID/g), was calculated as the amount of radioactivity divided by the organ mass and decay-corrected injected dose using mass standards of [⁶⁴Cu]Cu-NOTA-IAB41 injectate.

Fluorescence Activated Cell Sorting (FACs)

CD4+ cells were isolated from bulk PBMCs using the CD4+ T Cell Isolation Kit with MicroBeads for column-based magnetic activated cell sorting (MACS)(Miltenyi Biotech, CAT# 130-096-533). MACS was performed according to the manufacturers protocol. Cells were stained with PE/Cy7-CD25 (BioLegend, M-A251, CAT# 356108, RRID: AB_2561975), PE-CD45RA (BioLegend, HI100, CAT# 304108, RRID: AB_314412), PerCP/Cy5.5-CD45RO (BioLegend, UCHL1, CAT# 304222, RRID: AB_2174124), BUV395-CD3 (BD Biosciences, UCHT1, CAT# 563548, RRID: AB_2744387), and LIVE/DEAD Green Dead Cell Stain Kit (ThermoFisher Scientific, CAT# L23101). Pooled cells from all groups were used to generate the LIVE/DEAD single-color control, all fluorescence minus one (FMO) controls, and unstained controls. Ultracomp eBeads (ThermoFisher Scientific, CAT# 01-2222-41) were stained to generate single color controls. Analysis was performed using FlowJo (Version 10, RRID: SCR_008520). FACs gating strategy is detailed in Supplemental Figure 1.

Positron Emission Tomography (PET) and Computed Tomography (CT) Imaging

PET and CT imaging and image quantification were conducted using methods previously described (25). Mice received MRI scans 2–3 days prior to PET/CT imaging. We quantified the %ID/g in the tumor. Tumor region of interest (ROI) was determined by MRI. The ROI size for non-tumor bearing mice in the striatum was determined by the average tumor size in the tumor-bearing cohort (17.5 mm³). ROI placement was guided by the MRI for tumor-bearing mice.

Magnetic Resonance Imaging (MRI)

MRI imaging was conducted using methods previously described (25).

Autoradiography

Following transcardial perfusion with PBS 24 hours post injection of [⁶⁴Cu]Cu-NOTA-IAB41, mouse brains were embedded in optimal cutting temperature (OCT) compound Slides with 3 × 10 μM striatal (tumor-containing) sections were captured IHC. Slides with 3 × 40 μM striatal (tumor-containing), hippocampal, and cerebellar sections were taken for autoradiography and subsequent hematoxylin & eosin (H&E) staining. The 3 × 40 μM slides were exposed to the autoradiography film for 10 half-lives. The film was scanned using a Typhoon phosphoimager. FIJI (version 2.0, RRID:SCR_002285) was used to visualize the autoradiography images.

Immunohistochemistry (IHC)

IHC was performed using 5 μM thick coronal sections of striatum from formalin-fixed, paraffin embedded mouse brain tissue. Sections were stained using the Ventana Bench-mark ULTRA IHC automated staining system (Ventana Medical Systems, Roche-AstraZeneca). Immune cells were visualized using anti-human-CD4 (Abcam, EPR6855, ab133616, RRID: AB_2819211, 170 μg/ml). IHC was imaged using HALO software (version 3.0.311.228, Indica Lab). For optimal cutting temperature-embedded (OCT) sections, the detection of human-CD4 by IHC was performed at Molecular Cytology Core Facility, Memorial Sloan Kettering Cancer Center (New York, NY) using Discovery XT Processor (Ventana Medical Systems, Roche-AstraZeneca). The anti-human-CD4 antibody (Ventana-Roche, cat#790-4423) was used in 1:5 prediluted concentration. slides were counterstained with hematoxylin and coverslipped with Permount (Fisher Scientific).

Statistical Analysis

Statistical analyses were performed using GraphPad Prism (version 8). All data presented were analyzed using unpaired, two-tailed Student’s t-tests and differences >95% confidence level were considered significant.

Data Availability

The data generated in this study are available within the article and its supplementary data files.

RESULTS

[⁶⁴Cu]Cu-NOTA-IAB41 synthesis and radiolabeling

The anti-CD4 minibody IAB41 consists of two scFv regions bound to a truncated Fc region (CH3 domain-alone). The small size of the minibody (~80 kDa) enables a shorter circulation half-life relative to full length antibodies (27). We radiolabeled the minibody with ⁶⁴Cu, a positron emitter whose half-life (12.7 hours) is well matched to that of the minibody. The IAB41 minibody was incubated with an 80-fold excess of p-SCN-Bn-NOTA chelator in metal-free PBS to generate NOTA-IAB41. NOTA-IAB41 was subsequently incubated with ⁶⁴Cu at pH 5.5 in 250 mM metal-free ammonium acetate and purified with >99% radiochemical purity quantified by instant thin layer chromatography (iTLC) (Fig. 1A–1B). We also evaluated the stability of [⁶⁴Cu]Cu-NOTA-IAB41 in human serum by iTLC and found [⁶⁴Cu]Cu-NOTA-IAB41 maintained its stability at 4, 24, and 48 hours (Fig. 1C).

In order to ensure NOTA-IAB41 maintained its capacity to bind human CD4, we performed an immunoreactivity assay. The immunoreactive fraction (%Target Binding Fraction = [Counts per Minute (CPM)_Beads] / [CPM_Beads + CPM_Supernatent + CPM_Wash1 + CPM_Wash2]) of [⁶⁴Cu]Cu-NOTA-IAB41 was 70.7% (± 1.4 SD, n=3). Further, the target binding fraction was significantly higher than the non-CD4 containing control (0.8%, ± 0.3 SD, n=3, P<0.0001) and was outcompeted by a 100-fold excess of unlabeled NOTA-IAB41 demonstrating CD4 specificity (17.9%, ± 3.3 SD, n=3, P<0.0001, Fig. 1D).

The CD4-targeted minibody detects all CD4+ T cell subsets

CD4+ T cells can be categorized into three phenotypic states which perform distinct functions (Fig. 2A). Naïve CD4+ T cells have not yet encountered their cognate antigen. Upon antigen encounter, CD4+ T cells become activated and differentiate into functionally distinct CD4 T cell subsets; conventional CD4+ T cells (which include the T_H subtypes), referred to as ‘activated’ or CD4+ T_regs (6,28). T_regs are functionally distinct from conventional CD4+ T cells and are associated with immunosuppression (12).

Figure 2. [⁶⁴Cu]Cu-NOTA-IAB41 detects CD4+ T cell subtypes, but neither depletes nor activates CD4+ T cells in vitro.

(A) Schematic depicting CD4+ T cell subtypes analyzed. (B) [⁶⁴Cu]Cu-NOTA-IAB41 bound FACS sorted CD4+ T cell subtypes, quantified by γ-counting. CPM = counts per minute. (C) Bulk PBMC flow cytometry quantifying the mean fluorescence intensity (MFI) for three CD4+ T cell subtypes. Dose dependent effect of NOTA-IAB41 on (D) proportion of CD4+ T cells (E) Ki67 expression (F) proportion of naïve CD4+ T cells (G) proportion of activated conventional CD4+ T cells (H) and CD4+ regulatory T cells measured by flow cytometry. PMA/iono = phorbol 12-myristate 13-acetate/ionomycin. For all panels error bars are the SD.

Since all CD4+ T cell subtypes (naïve, activated conventional T_H, and T_regs) have been observed in the context of tumors, we evaluated the ability of [⁶⁴Cu]Cu-NOTA-AB41 to detect these distinct CD4 T cell populations (6). We sorted CD4+ cells from bulk peripheral blood mononuclear cells (PBMCs) (PBMC Donor 1, Supplemental Table 1) by MicroBeads for column-based magnetic cell isolation (MACs) (Miltenyi Biotech). We then performed FACS for CD3+CD45RA+CD45RO- (naïve), CD3+CD45RA-CD45RO+CD25- (activated), or CD3+CD45RA-CD45RO+CD25+ (T_reg-like) cells (gating strategy found in Supplemental Fig. 1). We found that [⁶⁴Cu]Cu-NOTA-IAB41 was able to detect all CD4+ T cell subtypes (Fig. 2B).

We next performed flow cytometry on bulk PBMCs from the same donor to determine the relative expression of CD4 on each T cell subtype quantified by mean fluorescence intensity (MFI) (PBMC Donor 1, Supplemental Table 1). We found that naïve (CD3+CD45RA+CD45RO-, MFI = 10597.6, ± 766.3 SD, n=3), activated (CD3+CD45RA-CD45RO+CD25-, MFI = 10466.3, ± 666.5 SD, n=3) and T_reg-like cells (CD3+CD45RA-CD45RO+CD25+, MFI = 10356, ± 469 SD, n=3) had no significant difference in MFI, suggesting similar levels of CD4 expression within each cell population (Fig. 2C, gating strategy found in Supplemental Fig. 2). This reflects the non-significant difference in [⁶⁴Cu]Cu-NOTA-IAB41 binding across CD4+ T cell subtypes (Fig. 2B) indicating CD4 expression corresponds to the amount of activity bound.

The IAB41 minibody binds human CD4 but does not contain a functional Fc region, presumably rendering IAB41 biologically inert. However, Freise et al. did report a mouse CD4-targeted cys-diabody depletes mouse CD4+ T cells and alters their phenotypic status independent of a functional Fc region (19). To determine whether NOTA-IAB41 impacted the depletion, proliferation, or activation of CD4+ T cells, we assessed the dose response of NOTA-IAB41-treated PBMCs ranging from 0.001 μg/mL to 10 μg/mL for 48 hours (flow cytometry panel validation can be found in Supplemental Fig. 3). We found that the proportion of CD4+ T cells was constant across treatment groups, demonstrating that NOTA-IAB41 does not deplete CD4+ T cells in vitro (Fig. 2D, CD3+CD4+/CD3+). Further, NOTA-IAB41 does not alter the expression of the proliferation marker Ki67 on CD4+ T cells relative to vehicle control (Fig. 2E, CD3+CD4+Ki67+/CD3+CD4+). Phorobol myistate/Ionomycine treatment (PMA/iono) treatment induces non-specific T cell activation and proliferation and was used as a positive control for flow cytometry. Additionally, NOTA-IAB41 does not alter the proportion of naïve (CD3+CD4+FOXP3-CD45RA+CD45RO-/CD3+CD4+), activated conventional (CD3+CD4+FOXP3-CD45RA-CD45RO+/CD3+CD4+) CD4+ T cells, or T_regs (CD3+CD4+FOXP3+CD25+/CD3+CD4+), relative to the vehicle control (Fig. 2F–H, gating strategy can be found in Supplemental Fig. 4).

We subsequently repeated this assay with a second PBMC donor and found similar results (Supplemental Fig. 5, PBMC Donor 2, Supplemental Table 1). After determining that NOTA-IAB41 does not appear to alter CD4+ T cell persistence, proliferation, or activation in vitro, we assessed the minibody’s ability to image human CD4+ T cells in vivo.

[⁶⁴Cu]Cu-NOTA-IAB41 detects human CD4+ T cells in vivo

In order to evaluate the ability of [⁶⁴Cu]Cu-NOTA-IAB41 to visualize human CD4+ T cells we generated a PBMC HIS mouse model. NOD.Cg-Prkdc^scid Il2rg^tm1Sug/JicTac (CIEA NOG) mice were either intravenously injected with human PBMCs from a single donor (+ PBMCs) or a vehicle control (− PBMCs) and three weeks following, both groups were injected with [⁶⁴Cu]Cu-NOTA-IAB41 (PBMC Donor 3, Supplemental Table 1). Representative PET/CT images of + PBMC mice demonstrate uptake in the spleen, lung, liver, and kidney (Fig. 3A). Organs of interest were extracted following PET/CT imaging and the % injected dose (ID) per gram of tissue was quantified by ex vivo gamma counting. The uptake in + PBMC spleens (8.9% ID/g, ± 1.1 SD, n=4, P<0.0001) and lungs (2.2% ID/g, 0.7 ± SD, n=4, P=0.0450) was significantly higher than − PBMCs controls (Fig. 3B). Splenic and lung CD4+ T cell infiltrates have been previously reported in PBMC HIS mouse models as the mice develop graft-versus-host disease (29).

Figure 3. [⁶⁴Cu]Cu-NOTA-IAB41 PET imaging of humanized non-tumor bearing mice.

(A) Representative PET/CT images of + PBMC and − PBMC CIEA NOG mice 4 hours following [⁶⁴Cu]Cu-NOTA-IAB41 tracer injection (n=4/condition). (B) Ex vivo biodistribution of [⁶⁴Cu]Cu-NOTA-IAB41 in humanized and non-humanized CIEA NOG mice 24 hours after tracer injection quantified by gamma counting (n=4/condition). CPM = counts per minute. (C) Quantification of splenic CD3+CD4+ T cells three weeks following humanization in + and − PBMC CIEA NOG mice. (D) Splenic uptake of [⁶⁴Cu]Cu-NOTA-IAB41 or [⁶⁴Cu]Cu-NOTA-IAB41 plus 5-fold excess cold NOTA-IAB41 3 weeks following humanization in + PBMC and − PBMC mice. For all panels * = P ≤ 0.05; *** = P ≤ 0.001; **** = P ≤ 0.0001. Error bars are the SD.

To determine whether [⁶⁴Cu]Cu-NOTA-IAB41 uptake in the spleen is a result of the presence of human CD4+ T cells, we extracted the spleen of + PBMC mice three weeks following humanization and found that 19.8% (CD3+CD4+/total cells) of splenocytes were CD4+ T cells (± 5.6 SD, n=3, Fig. 3C, gating strategy can be found in Supplementary Fig. 6). To determine whether [⁶⁴Cu]Cu-NOTA-IAB41 uptake in the lung represents organ infiltration by human CD4+ cells, we extracted the lungs of + PBMC and − PBMC mice three weeks post-humanization and performed IHC human CD4. CD4+ cells were present in the lung of + PBMC mice and absent in the lungs of − PBMC mice (n=3, Supplemental Fig. 7). We next evaluated whether [⁶⁴Cu]Cu-NOTA-IAB41 is specific to human CD4+ T cells in vivo we assessed whether a 5-fold excess of unlabeled NOTA-IAB41 would be sufficient to block [⁶⁴Cu]Cu-NOTA-IAB41. Three weeks following PBMC injection, + PBMC mice were injected with [⁶⁴Cu]Cu-NOTA-IAB41-alone or [⁶⁴Cu]Cu-NOTA-IAB41 + a 5-fold excess of unlabeled NOTA-IAB41. We also included a − PBMC control. We found that excess unlabeled NOTA-IAB41 was sufficient to significantly reduce the PET signal of [⁶⁴Cu]Cu-NOTA-IAB41 in the spleen (+ PBMCs 11.3 %ID/g, ± 0.7 SD, n=3; + PBMCs + excess unlabeled NOTA-CD4 4.9% ID/g, ± 1.0 SD, n=3, P=0.001), suggesting specificity of the tracer (Fig. 3D, PBMC Donor 4, Supplemental Table 1). The PET signal in the spleen of the + PBMC group was modestly higher than that of the prior imaging experiment (Fig. 3B). This is likely due to differences in splenic T cell burden as these two experiments were performed with distinct PBMC donors (25).

[⁶⁴Cu]Cu-NOTA-IAB41 images CD4+ brain tumor infiltrating lymphocytes

PET/CT images and ex vivo gamma counting demonstrated no uptake of [⁶⁴Cu]Cu-NOTA-IAB41 in non-diseased brain, suggesting low background suitable for use of [⁶⁴Cu]Cu-NOTA-IAB41 for neuroimaging (Fig. 3B). Our group has previously found that the CD8-targeted minibody IAB22M2C (ImaginAb) readily images CD8+ TILs in orthotopically transplanted GBM tumors in the brain (25). To determine whether the minibody-based PET probe [⁶⁴Cu]Cu-NOTA-IAB41 images CD4+ TILs in a humanized brain tumor model, we first evaluated whether the implantation of an orthotopic GBM was sufficient to induce the infiltration of human CD4+ T cells to the TME. To do so, we intracranially injected a GBM PDX into CIEA NOG mice and three weeks later we humanized the mice (+PBMCs) from a partial-HLA matched donor (workflow Fig. 4A). Patient-derived orthotopic GBM xenograft models recapitulate the molecular and histopathological features of GBM, including heterogeneous blood brain barrier permeability, and hence represent a suitable model system to evaluate radiotracer uptake (30,31). We found human CD4+ T cells infiltrate into the GBM 3 weeks following humanization by IHC (Fig. 4B, PDX 160721-1, PBMC Donor 3, Supplemental Table 1). We next determined the proportion of CD4+ T cell subtypes in the brain TME by flow cytometric analysis. Humanized GBM PDX 160721-1-bearing mice were prepared according to the workflow in Figure 3A. Whole brains were extracted for flow cytometry (gating strategy found in Supplemental Fig. 8). The majority of CD4+ T cells in the brain were either T_regs (CD3+CD4+FOXP3+CD25+ cells / CD3+CD4+ cells, 19.1% ± 13.0 SD, n=6) or activated conventional CD4+ T cells (CD3+CD4+FOXP3-CD45RA-CD45RO+ cells / CD3+CD4+ cells, 55.1%, ± 12.5 SD, n=6) with a minority of naïve CD4+ T cells (CD3+CD4+FOXP3-CD45RA+CD45RO- / CD3+CD4+ cells, 3.1%, ± 1.3 SD, n=6) (Fig. 4C, PBMC Donor 5, Supplemental Table 1). The ratio of activated conventional CD4+ T cells to T_regs is the inverse of reports in GBM patients. This disparity is likely due to the PBMC mouse models development of graft-versus-host disease (32,33). However, these cell types are key in anti-tumor immunity and their presence in the brain TME of the GBM HIS mouse model enables the evaluation of their detection in vivo.

Figure 4. [⁶⁴Cu]Cu-NOTA-IAB41 images CD4+ tumor infiltrating lymphocytes in two humanized mouse models of GBM.

(A) Experimental timeline highlighting the order of orthotopic PDX engraftment, humanization with HLA donor matched PBMCs, imaging, and necropsy (top panel). (B) Visualization of tumor and CD4+ T cell infiltrate by H&E (top panel) and IHC (bottom panel) (n=3). (C) Quantification of whole brain flow cytometry reflecting the proportion of three CD4+ TIL subtypes in the humanized GBM PDX 160721–1, PBMC donor 4 (n=6). (D) Coregistered PET/CT images of + Tumor + PBMC (top panel), + Tumor − PBMC (middle panel), and − Tumor + PBMC (bottom panel) mice 24 hours following [⁶⁴Cu]Cu-NOTA-IAB41 tracer injection. − Tumor mice received a sham IC injection. (n=4/condition). (E-F) Quantification of tracer uptake for PDX model 160721–1 donor 4 (E) and PDX 160615–1, PBMC donor 5 (F). (G) H&E staining, IHC for human CD4, and autoradiography (n=3/condition) of PDX 160721–1. Sections were acquired 24 hours post injection with radiotracer. (H) Distribution of [⁶⁴Cu]Cu-NOTA-IAB41 in PDX 160615–1-bearing mouse brains. H&E and autoradiography (n = 3/condition) of PDX 160615–1. * = P ≤ 0.05;** = P ≤ 0.01. Error bars are the SD.

We next sought to determine whether [⁶⁴Cu]Cu-NOTA-IAB41 images CD4+ TILs in the brain. To do so we implanted CIEA NOG mice with PDX 160721-1 (+ Tumor) and subsequently humanized with PBMCs (+ PBMCs) (PBMC Donor 3, Supplemental Table 1). We also generated tumor-bearing non-humanized (+ Tumor − PBMCs) and non-tumor-bearing humanized (− Tumor + PBMCs) controls. Non-tumor-bearing controls received a sham intracranial injection. We observed the highest signal in the tumor-bearing humanized (+ Tumor + PBMCs) brain 24 hours following radiotracer injection with relatively little uptake in the non-tumor bearing and non-humanized controls (representative PET/CT images, Fig. 4D). Mean uptake in + Tumor + PBMC mice was 1.6% ID/g (± 0.5 SD, n=3) which was significantly higher than + Tumor − PBMC mice (0.6% ID/g, ± 0.1 SD, n=4, P=0.0086) and − Tumor + PBMC mice (0.6% ID/g, ± 0.1 SD, n=3, P=0.0220, Fig. 4E). We repeated this workflow with a second GBM PDX 160615-1 and again found significantly higher uptake in the + Tumor + PBMC group (1.8% ID/g, ± 0.5 SD, n=4) relative to the + Tumor − PBMCs (0.7% ID/g, ± 0.2 SD, P=0.0040) and the − Tumor + PBMCs groups (0.3% ID/g, ± 0.06 SD, n=3, P=0.0169, Fig. 4F; representative images of PDX 160615-1 can be found in Supplemental Fig. 9, PBMC Donor 4, Supplemental Table 1).

To determine whether [⁶⁴Cu]Cu-NOTA-IAB41 uptake is specific to brain parenchymal CD4+ TILs, we performed autoradiography and IHC. Following PET/CT imaging of PDX 160721-1 and PDX 160615-1, mice were perfused to clear the blood-pool from the brain and whole brain was embedded in optimal cutting temperature (OCT) compound. We acquired 40 μM sections for autoradiography and H&E. In addition, we took 10 μM serial sections for anti-human-CD4 IHC. We found that tracer uptake localizes to CD4+ T cell infiltrate in the brain of + Tumor + PBMC mice with little uptake in + Tumor − PBMC controls (PDX 160721-1, Fig. 4G; PDX 160615-1, Supplemental Fig. 10). Additionally, we acquired striatal (tumor-bearing), hippocampal, and cerebellar sections and found that tracer uptake localizes to tumor bearing, CD4+ T cell infiltrated brain sections (PDX 160615-1, Fig. 4H). These results corroborate our previous findings by PET, indicating [⁶⁴Cu]Cu-NOTA-IAB41 uptake is specific to human CD4+ T cell infiltrate.

DISCUSSION

CD4+ T cells play an integral role in driving anti-tumor immunity and there is growing evidence suggesting the role of CD4+ T cells in brain tumor immune evasion. Immunosuppressive CD4+ T cells are found within the GBM TME while conventional CD4+ subsets are deprived from the TME (34). Higher grade gliomas are associated with increased T_reg infiltrate (35). Further, depletion of T_regs in the SMA-560 mouse glioma model resulted in enhanced tumor rejection and improved survival (36). However, conclusions have been limited by a lack of tools to non-invasively monitor CD4+ T cell kinetics in GBM patients throughout disease progression and treatment. Our data demonstrate the capability of [⁶⁴Cu]Cu-NOTA-IAB41 as a non-invasive imaging agent for human CD4+ T cells in brain tumors.

GBMs are thought to have low cytotoxic CD8+ T cell infiltrate and relatively high proportions of immunosuppressive T_regs, suggesting the applicability of non-invasive CD4-targeted imaging. In contrast to GBMs seen in the clinic, the density of CD4+ TILs seen in our PBMC humanized GBM model system reaches the upper limit of CD4+ T cells visualized by biopsy and there is a higher proportion of activated conventional CD4+ T cells relative to T_regs (37). This may be attributed to the inflammatory nature of the PBMC HIS mouse model or PBMC donor-tumor alloreactivity due to mismatch in minor HLA-loci (29). These limitations mitigate drawing conclusions regarding GBM immuno-biology. However, this model system is an excellent resource to evaluate the capacity of a CD4-targeted PET tracer to detect CD4+ T cells and their subtypes in a brain tumor and in peripheral organs. We were able to determine, using this model system, that [⁶⁴Cu]Cu-NOTA-IAB41 detects CD4+ T cells both in peripheral organs and in brain tumors and CD4 T cells of distinct lineages and activation states revealing its wide applicability for the detection of CD4+ T cells in various tissues in homeostasis and disease.

The application of non-invasive monitoring of CD4+ T cells in the brain spans beyond cancer. CD4+ T cells have been implicated in a number of autoimmune diseases including those with neuroinflammatory pathology. CD4+ T cells mitigate disease progression in experimental models of multiple sclerosis, a central nervous system autoimmune disorder (38). CD4+ T cell infiltrate into the brain parenchyma has also been implicated in the pathogenesis of neurodegenerative diseases Parkinson’s and Alzheimer’s (39–41). In addition to imaging CD4+ T cells in the brain. CD4-targeted immunoPET may have applications in peripheral immune disorders. Previous work by Friese et al. demonstrates the utility of CD4-targeted ImmunoPET in a mouse model of irritable bowel disorder (17). Additionally, the PBMC HIS mice used in our study model the development of graft-versus-host disease over time (29).

[⁶⁴Cu]Cu-NOTA-IAB41 detects human CD4+ T cells regardless of their activation and functional states – including naïve, activated conventional CD4+ T cells, and T_regs. [⁶⁴Cu]Cu-NOTA-IAB41 is also able to detect CD4+ TILs in orthotopic GBM PBMC HIS mouse models. Our PET tracer does not differentiate between CD4+ T cell subtypes, a potential limitation since CD4+ T cell subtypes play distinct roles in anti-tumor immune response and resistance (6). At the current time, however, the role of specific CD4+ T cell subsets in GBM anti-tumor response remains to be defined and our current strategy hence represents an inclusive first approach to quantify all CD4+ T cell in vivo. This could be particularly informative when used in conjugation with radiotracers directed against other immune cells, for example CD8+ T cells (25). As the technology of multiplexed PET evolves, one can envision the use of multiple tracers in-tandem to delineate cell subtypes without the need of biopsy (42). An additional limitation is the observed uptake of [⁶⁴Cu]Cu-NOTA-IAB41 in the liver and kidney of − PBMC mice and this signal likely represents non-specific uptake (2–4% ID/g) of [⁶⁴Cu]Cu-NOTA-IAB41. This non-specific uptake in the kidney and the liver is likely a result of excretion of the radiotracer, a limitation that may limit the utility of the tracer to image CD4+ T cell infiltration in these particular organs (25,27).

The anti-CD4 minibody IAB41 detects human CD4+ T cells without depleting or altering proliferation or phenotype. Further, [⁶⁴Cu]Cu-NOTA-IAB41 images CD4+ tumor infiltrating lymphocytes in orthotopic PBMC HIS mouse models of GBM. The clinical capacity of IAB41 immunoPET should be further investigated.