Human Neural Stem Cell Biodistribution and Predicted Tumor Coverage by a Diffusible Therapeutic in a Mouse Glioma Model

Abstract

Engineered neural stem cells (NSCs) intrinsically migrating to brain tumors offer a promising mechanism for local therapeutic delivery. However, difficulties in quantitative assessments of NSC migration and in estimates of tumor coverage by diffusible therapeutics have impeded development and refinement of NSC-based therapies. To address this need, we developed techniques by which conventional serial-sectioned formalin-fixed paraffin-embedded (FFPE) brains can be analyzed in their entirety across multiple test animals. We considered a conventional human glioblastoma model: U251 glioma cells orthotopically engrafted in immunodeficient mice receiving intracerebral (i.c.) or intravenous (i.v.) administrations of NSCs expressing a diffusible enzyme to locally catalyze chemotherapeutic formation. NSC migration to tumor sites was dose-dependent, reaching 50%-60% of total administered NSCs for the i.c route and 1.5% for the i.v. route. Curiously, the most efficient NSC homing was seen with smaller NSC doses, implying existence of rate-limiting process active during administration and/or migration. Predicted tumor exposure to a diffusing therapeutic (assuming a 50 µm radius of action) could reach greater than 50% of the entire tumor volume for i.c. and 25% for i.v. administration. Within individual sections, coverage of tumor area could be as high as 100% for i.c. and 70% for i.v. routes. Greater estimated therapeutic coverage was observed for larger tumors and for larger tumor regions in individual sections. Overall, we have demonstrated a framework within which investigators may rationally evaluate NSC migration to, and integration into, brain tumors, and therefore enhance understanding of mechanisms that both promote and limit this therapeutic modality. Stem Cells Translational Medicine 2017;6:1522-1532.

Keywords: Anticancer therapy; Biodistribution; Diffusible therapeutic; Glioblastoma; Glioma; Mouse model; Neural stem cells.

Figure 1

Distributions of iron‐labeled hCE1m6‐F3 neural stem cells (NSCs) administered i.c. into the frontal lobe of U251 glioma‐bearing mice. (A): Low‐power image of a hematoxylin‐eosin (H+E)‐stained mouse brain section. (right upper) expanded area from A showing Prussian blue‐stained hCE1m6‐NSCs (pararosaniline counterstain) at the tumor mass and an invasive tumor nodule (inset); (right lower) DAB‐visualized (brown) eGFP‐expressing U251 glioma cells of the same tumor mass from the next slide (hematoxylin counterstain). Scale bar = 50 µm. (B): Distribution of hCE1m6.NSCs at tumor sites after i.c. (a1, a2) or i.v. (b1, b2) administration. NSCs are marked in red and tumor cells are colored according to their distance from the NSCs (yellow for ≤50 µm, light green for ≤100 µm, and dark green for >100 µm). For i.c. NSC administration, the arrow indicates apparent migration of a subset of NSCs from the injection site to the tumor. Scale bar = 250 µm. (C): (Ci–Cviii) Paired images of consecutive sections (feraheme‐Prussian blue and eGFP‐DAB) showing distributions of hCE1m6‐F3 NSCs within and around an engrafted U251 tumor at increasing depths below the pial surface (indicated) and thus in different anatomical locations within the host brain. Note the disseminating U251 glioma cells in many sections, and lateral expansion in the external capsule along with the presence of hCE1m6‐F3 NSCs at greater depths. Positions of Ctx, EC, and Str are indicated in panels A, Ba1 and Cvi. Estimated per cent tumor coverage by a secreted therapeutic (50 μm effective radius) determined as described (Fig. 2 and text) is also indicated. Images in A and C show tumors at 7 days engraftment and four days after NSC administration; tumors in B are at 14 days engraftment and 4 days post‐NSC injection. Abbreviations: Ctx, cortex; EC, external capsule; Str, striatum.

Figure 2

Distribution hCE1m6‐F3 neural stem cells (NSCs) and representation of predicted tumor coverage by a diffusing chemotherapeutic. (A1, A2): Consecutive brain sections stained for hCE1m6‐F3 NSCs (above, Prussian blue for feraheme) and U251 glioma cells (below, DAB immunostaining of eGFP). (B): Segmentation of cells of interest by color deconvolution (see Methods) with hCE1m6.F3NSCs shown in red and U251 glioma cells in dark green. Small tumor nodules were included in analyses by virtue of this segmentation method, which was based on expression of human nestin by all tumor cells rather than by delineation of the tumor mass which will be limited by tumor morphology. (C): Computed radii (circle 25 µm above and 50 µm below) centered on each cell. (D): Estimated tumor coverage by a secreted drug (light green) centered on each pixel at the edge of each stem cell (red) for a particular radius of action from each stem cell.

Figure 3

Migration of hCE1m6‐F3 NSCs to tumor sites after i.c. or i.v. administration. Numbers of NSCs counted at tumor sites are shown in relation to (A) the numbers administered and (B) tumor volume. Doses of hCE1m6‐F3 NSCs were 10–200 × 10³ cells for i.c. and 0.01–2.0 × 10⁶ cells for i.v. routes. In this and subsequent figures, each dot represents data from an individual mouse. The dotted lines are linear regressions given to provide a visual reference to trends in the data, and have no other significance. In some graphs X‐axes are log₁₀ to provide for visual expansion of the data. (A) Comparison to total numbers of NSCs administered (a1, a3) Number of NSCs at tumor in relation to numbers of NSCs administered r ² = 0.81, p < .0001 for i.c.; r ² = 0.06, p = .28 for i.v. (a2, a4) Percentage of NSCs at tumor in relation to numbers of NSCs administered r ² = 0.20, p = .12 for i.c.; r ² = 0.22, p = .03 for i.v. (B) Comparison to total tumor volume (b1, b3) Number of NSCs at tumor in relation to tumor volume r = 0.26, p = .37 for i.c.; r = 0.43, p = .05 for i.v. (b2, b4) Percentage of total administered NSCs found at tumor in relation to tumor volume r = 0.35, p = .22 for i.c.; r = 0.34, p = .12 for i.v. Abbreviation: NSCs, neural stem cells.

Figure 4

Percentage of tumor coverage by an NSC‐secreted therapeutic. Coverage is estimated (see Methods) assuming a 50 µm radius of action for delivery i.v. or i.c. (A): Percentage tumor coverage in relation to numbers of NSCs administered (a1) r ² = 0.26, p = .07 for i.c.; (a2) r ² = 0.12, p = .12 for i.v. (B): Percentage tumor coverage in relation to numbers of NSCs at the tumor site (b1) r = 0.42, p = .14 for i.c.; (b2) r = 0.73, p = 1.07 × 10⁻⁴ for i.v. (C): Percentage tumor coverage in relation to tumor volume (c1) r = 0.60, p = .03 for i.c.; (c2) r = 0.24, p = .23 for i.v. Abbreviation: NSCs, neural stem cells.

Figure 5

Percentage tumor coverage evaluated on individual brain sections. All brain sections containing engrafted tumor cells in the cohort of 14 i.c.‐ and 22 i.v.‐injected brains were evaluated without regard to the presence or absence of neural stem cells (NSCs); n = 101 sections for i.c. and 161 for i.v. (A): Percentage tumor coverage in relation to numbers of NSCs administered (a1) r ² = 0.03, p = .07 for i.c.; (a2) r ² = 0.003, p = .49 for i.v. (B): Percentage tumor coverage in relation to tumor area in each section (b1) r = 0.44, p = 3.29 × 10⁻⁶ for i.c.; (b2) r = 0.41, p = 9.90 × 10⁻⁸ for i.v. Abbreviation: NSCs, neural stem cells.

Figure 6

Neural stem cell (NSC) density and its influence on tumor coverage. (A): Density of hCE1m6‐F3 NSCs in relation to tumor volume (a1) r = −0.35, p = .22 for i.c.; (a2) r = −0.002, p = .99 for i.v. (B): Percent tumor coverage in relation to density of hCE1m6‐F3 NSCs (b1) r = 0.02, p = .95 for i.c.; (b2) r = 0.66, p = .001 for i.v. Average percent tumor coverage for i.c. delivery is 23.8 ± 14.5% for the density range 1.9 × 10⁴–9.2 × 10⁴ NSCs/mm³, and for i.v. delivery is 15.8 ± 9.4% for the density range 1.1 × 10⁴–5.0 × 10⁴. Within this range of overlapping densities, p = .18 (two‐tailed two‐sample t test, unequal variance), indicating that there was no statistically significant difference in average percent coverage between the two routes of NSC administration. Abbreviation: NSCs, neural stem cells.

Figure 7

Nonuniform distribution of hCE1m6‐F3 neural stem cells (NSCs) and consequences for the efficiency of tumor coverage by a secreted therapeutic. hCE1m6‐F3 NSCs were often found in clusters and close to the tumor edge. As a quantitative measure of clustering, the Euclidean distance between each pair of NSCs was calculated from the x, y coordinates of each NSC (see Methods). The clustering index (CI) is defined for each section as 1 – [(number of clusters)/(total number of NSCs)], such that a higher CI indicates greater clustering. (A): Clustering indices (CIs) for each section in a cohort of 73 individual tumor‐bearing brain sections from 12 brains with i.c.‐administered NSCs. Sections contained between 2 and 390 NSCs. Shown for each brain are individual section CIs, and box plots where the box spans the 25th to the 75th percentile, the error bars indicate the 10th and 90th percentiles, and the horizontal bar marks the median, with CI values for each section are superimposed. (B): Relationship between the number of NSCs present at the tumor and the CI. Open symbols mark CI ≤ 0.5 for which there was considerable scatter. Filled symbols mark CI > 0.5 for which the CI appears to increase with the number of NSCs. (C): Relationship between the tumor coverage estimated for each section assuming a therapeutic radius of 50 µm and the theoretical maximum assuming homogeneous distribution of NSCs through the tumor. Symbols are as in (B); greater clustering (higher CI) appears associated with reduced efficiency. Abbreviation: NSCs, neural stem cells.