Human tumor infiltrating lymphocytes cooperatively regulate prostate tumor growth in a humanized mouse model

Human lymphocytes infiltrate and control tumor growth in huPBL-NSG animals

According to the experimental paradigm, animals were immune reconstituted with a combination
of human PBL and DC alone, in combination with the subcutaneous (s.c.) implantation
of PC3 cells, or with PC3 cells alone. PBL were prepared from peripheral blood mononuclear
cells (PBMC) by depleting monocytes and natural killer cells (expressing CD14 and/or
CD16) and activated T cells (CD25+) using monoclonal antibodies (mAbs) as described
previously 15]. The resulting PBL consisted primarily of T cells (CD3+/CD56-; 72.7-90.2%), NKT cells
(CD3+/CD56+; 2.8-9.8%), B cells (CD3-/CD20+; 3.5-8.0%), and a few NK cells (CD3-/CD56+;
1.0-2.2%). Monocyte-derived DC were CD14-negative and expressed high levels of class
I and II major histocompatibility molecules and co-stimulatory molecules 15].

As shown in Figure 1A, the cross-sectional diameters of tumors recovered from huPBL-NSG mice were significantly
smaller than tumors from NSG mice that were not immune reconstituted (average 620?±?233 mm³ vs. 2792?±?711 mm³; p??0.05). When the rates of tumor growth over time were examined (Figure 1B), the difference between these two groups did not occur until at least two weeks
after implantation, corresponding to the time at which functional immune reconstitution
occurs in this model 15]. Immunohistochemistry (IHC) clearly demonstrated the presence of Prostate Stem Cell
Antigen (PSCA) positive tumor in both cases and the presence of CD45+ human leukocytes
infiltrating tumors harvested from huPBL-NSG animals (Figure 1C). As such, the presence of infiltrating human cells correlated with a reduction
in the rate of tumor growth.

Figure 1. Human PBL infiltrate and control tumor growth in huPBL-NSG animals. Six-8-week- old NSG mice were injected with saline alone or with human PBL (1×10⁷) by intraperitoneal injection to produce huPBL-NSG animals. Mice were simultaneously
implanted with 2×10⁶ human prostate tumor cells (PC3 cell line) by s.c. injection and tumor growth was
monitored over a 4 week period. (A) Final tumor volume (mm³) as measured percutaneously in the animal and corresponding images of tumors dissected
from NSG and huPBL-NSG animals at 4 weeks. (B) Tumor growth curves for NSG and huPBL-NSG animals. (C) Recovered tumors were formalin-fixed, paraffin-embedded,sectioned, and stained with
antibodies against either human PSCA (top panel, brown stain), to demonstrate tumor
cells, or human CD45 (bottom panel, brown stain), to demonstrate infiltrating human
leukocytes, and counterstained with Hematoxylin (blue stain). Magnification, x100.
Values represent mean of three animals per group?±?SD. *p??0.05. Representative experiment,
n?=?5 replicate experiments.

TIL recovered from huPBL-NSG animals exhibit a unique phenotype distinct from splenocytes
and similar to primary human TIL recovered from patients with prostate cancer

The distribution and features of leukocytes that infiltrate human tumors are distinct
from those present in peripheral blood or infiltrating normal tissues. Spleen cells
and TIL were therefore recovered at 4 weeks after implantation and examined by flow
cytometry. A representative example is presented in Figure 2A and documents the key features of TIL in this model. CD8+ T cells always predominated
in TIL and the CD4:CD8 T cell ratio in TIL was always lower than that observed in
spleen. While the splenic CD4:CD8 ratio was variable it ranged from 1.8-3.5 fold higher
than that in the corresponding TIL (p??0.005 by paired T test, n?=?5 donors). In
addition, TIL consistently exhibited a high percentage of cells exhibiting the CD8+/CD56+
NKT cell phenotype (34.02?±?9.53%) while NKT cells represented only a small minority
population within the spleen (5.76?±?3.24%) and the difference was statistically significant
(p??0.00005, n?=?6 experiments using different donors). In addition, the majority
of TIL, regardless of whether they were CD4+, CD8+, or CD8+/CD56+ demonstrated the
CD69 activation marker, while expression of this marker by splenic T cells was limited.
B cells were rarely identified in TIL (0.5%) nor were CD14+ monocytes (0.3%), the
later findings possibly reflecting the depletion of this population from the PBL used
for implantation. While the cell populations and markers were distinctively different
between tumor and spleen in tumor-bearing huPBL-NSG animals, the CD4/CD8 ratio and
the presence of NKT cells did not differ between the spleens of control huPBL-NSG
animals and those bearing PC3 tumors (data not shown). As such, TIL in this model
exhibited a unique composition and features. In order to assess whether this recapitulated
the features of native TIL we also performed FACS analysis on TIL obtained from patients
with primary prostate cancer. As demonstrated in Figure 2B, TIL from prostate cancer also featured a CD8+ predominance with a high percentage
of CD8?+?CD56+ NKT and expression of the CD69 activation marker. Unfortunately, given
the absence of matched spleen samples from patients, the features in human spleen
from patients with prostate cancer cannot be directly commented on or compared – one
of the important reasons for developing the humanized mouse model.

Figure 2. Phenotype of human lymphocytes recovered from the spleen and tumor of tumor-bearing
huPBL-NSG and from primary human prostate cancers. Single cell suspensions were prepared from the spleen and tumor recovered from huPBL-NSG
mice at 4 weeks (A) and from a primary human prostate cancer removed at the time of prostatectomy (B). Cells were stained with an antibody cocktail, and then analyzed by flow cytometry
for the expression of CD4 vs CD8 (top panel, gated on all cells expressing human CD4
and/or CD8); CD4 vs CD56 (middle panel, gate on all CD3+ cells); and CD4 vs CD69 (bottom
panel, gated on all cells expressing human CD4 and/or CD8). The CD8+/CD56+ NKT population
is identified by CD4-/CD56+ T cells. Representative experiment, n?=?3 experiments.

The human cytokine and chemokine profile in huPBL-NSG animals is altered by the presence
of tumor

In contrast to the complexity of human serum, the human proteins present in the serum
of huPBL-NSG animals can only originate from the cells that have been implanted and
those expanded during the period of engraftment. As such, we hypothesized that serum
measurements from these animals would directly reflect their immune reconstitution
and its interaction with tumor (Figure 3). Sera were collected at 4 weeks from animals that were implanted with human PBL
alone, those receiving both PBL and tumor, and animals that were implanted with tumor
alone. hIFN-?, hIL-10 and hRANTES were all present in huPBL-NSG animals and in huPBL-NSG
animals that had been implanted with tumor, but not in animals that received PC3 cells
alone. In contrast, while hTNF-? was detected at a low level in one of the huPBL-NSG
animals it was detected in all of the huPBL-NSG animals that also had tumor implanted.
This difference was not due to the tumor itself, which released no detectable hTNF-?
when implanted alone, suggesting an immune cell-tumor interaction. The pro-inflammatory
factor IL-8 appeared to be made primarily by the tumor and while IL-6 was released
by both immune cells and tumor, its levels were more uniform and robust in reconstituted
animals bearing PC3 tumors, again suggestive of an immune cell-tumor interaction.
As the sample sizes were small and the distribution of cytokine levels were not always
uniform, the mean values of each cytokine for each replicate animal are presented
in Figure 3 in lieu of formal statistical analysis. These results suggest that the tumor-bearing
huPBL-NSG model provides a sensitive platform for investigating the human cytokine
and chemokine interactions relevant to the generation and maintenance of TIL.

Figure 3. Profile of human cytokines and chemokines in the serum of NSG and huPBL-NSG animals
in the presence or absence of PC3 tumor implants. Sera were collected from NSG animals 4 weeks after implantation of human PBL and/or
PC3 tumor cells and assayed for the presence of human IFN-?, TNF-?, IL-8, IL-6, IL-10
and RANTES by SearchLight multiplex assay. Representative experiment with each bar
representing results from one animal, n?=?3 experiments.

Human effector memory T lymphocytes (T-em) are increased in the TIL recovered from
tumor-bearing huPBL-NSG animals

A more detailed analysis of T cell subsets was undertaken to evaluate for the presence
of naïve (T-naïve, CD45RA^hi/CD127^hi), central memory T cells (T-cm, CD45RA^lo/CD127^hi) and T-em (CD45RA^lo/CD127^lo). As demonstrated in Figure 4, the majority of splenic CD4+ T cells in this model are split between T-cm and T-em,
however there is always a recognizable T-naive population. In contrast, the T-em phenotype
overwhelmingly predominates in TIL where the T-naive population was essentially absent
and cells exhibiting a T-cm phenotype were limited. This overall pattern was similar
for both the CD4+ and CD8+ T cells but most striking with respect to the CD8+ subset.
While the difference between spleen and TIL subsets was numerically small, especially
for the CD4+ population, it was highly reproducible and statistically significant
when all mice that had been reconstituted with same PBL donor were compared (p??0.01
for CD4+ T cells; p??0.0001 for CD8+ T cells) and this pattern was significant in
all experiments, with all donors when normalized by comparing the T-em: T-cm ratio
(p 0.05 for both CD4+ T cells and CD8+ T cells). When primary tumors from patients
were analyzed, the overwhelming majority of CD3+ T cells were CD45RA-negative (data
not shown), which is consistent with the finding from our tumor-bearing huPBL-NSG
model. In contrast, the majority of CD3+ T cells recovered from primary human tumors
expressed CD127, more consistent with a T-cm rather than T-em phenotype. This likely
reflects the sub-acute nature of the tumor-related interaction that occurs in this
model, which is temporally different than the interaction between primary human tumors
and infiltrating T cells in vivo, which occurs over a much longer period of time.

Figure 4. T- naïve, T-cm and T-em subsets in the spleen and tumor. Single cells were prepared from spleens (A) and tumors (B) recovered from huPBL-NSG mice 3 weeks after implantation and stained with antibodies
against human CD3, CD4, CD8, CD45RA and CD127. Cells were gated for expression of
CD3 and then analyzed by flow cytometry for the expression of other markers on the
CD3+ population. Representative experiment, n?=?5 experiments.

Despite the predominance of CD8+ TIL, CD4+ T cells play an essential role in tumor
regression in tumor-bearing huPBL-NSG animals

Activated CD8+ T cells and NKT cells with a T-em phenotype predominate in TIL recovered
from prostate tumors and in our tumor-bearing huPBL-NSG model. However, animal models
have routinely suggested that CD4+ T cell help is essential for effective anti-tumor
immunity. As a direct proof of principal, NSG mice were engrafted with either whole
PBL or with PBL from the same donor that had been depleted of either CD8+ T cells
[huPBL-NSG (?CD8)] or CD4+ T cells [huPBL-NSG (?CD4)]. As shown in Figure 5A, immune reconstitution with CD8-depleted PBL lead to the same control of tumor growth
as did the administration of whole PBL. In addition, NSG animals that had been reconstituted
with CD4-depleted PBL failed to control tumor growth (Figure 5B). Taken together, these results suggest that CD4+ T cells play an essential role
in the immune response to tumor growth even though the majority of cells that accumulate
within the tumor exhibit a CD8+ T-em phenotype. To better understand this phenomenon
we examined the phenotype of T cells in the spleens and tumors of these animals (Figure 6). As shown in Figure 6A/B, even though CD8+ T cells were extensively depleted prior to implantation (0.1%
CD8+), there was a substantial population of CD8+ T cells (11-15% of total T cells)
when cells were recovered from the spleens of reconstituted huPBL-NSG animals. Furthermore,
as shown in Figure 6C there was still a preferential accumulation of CD8+ T cells in the tumor where the
CD8+ population represented 25-35% of recovered T cells. As others have demonstrated
in a variety of models 16],17], the repopulation of single-positive CD8+/CD4- T cells likely represents the rapid
and marked homeostatic expansion of small numbers of contaminating CD8+ cells still
present in the implanted PBL. On the other hand, the presence of double-positive CD4+/CD8+
T cells may reflect the de novo expression of CD8 that can occur when CD4+ cells are stimulated to expand 18]. As such, while CD4+ cells may be important we cannot rule out a significant contribution
of the CD8+ T-em population with respect to controlling tumor growth. A related phenomenon
appeared to occur in huPBL-NSG (?CD4) animals (Figure 6 D-F). While CD4+/CD8- cells did not regenerate when CD4+ cells were depleted prior
to implantation, there was a striking increase of CD8+/CD4+ double-positive cells
in spleen (32.7%) and tumor (27.4%) as compared to animals that received whole PBL
where these double-positive cells only represented 8.2% and 9.9%, respectively. Moreover,
unlike our consistent observation that the % of single CD8+ T cells was always greater
in tumor than in spleen, there was no significant increase of CD8 T cells in tumor
(68.1%) vs. spleen (65.5%) in huPBL-NSG (?CD4) mice, suggesting that the failure to
expand tumor infiltrating CD8+ T cells contributed to the lapse in tumor control.
Collectively, these results suggest a significant interaction between CD4+ and CD8+
T cell subsets is required with an essential role for both in effectively mediating
anti-tumor responses.

Figure 5. Role of CD4+ and CD8+ lymphocyte subsets in controlling tumor growth. NSG mice were reconstituted with either 1×10⁷ human PBL (huPBL-NSG) or PBL from the same donor that had been depleted of CD8+ T
cells [huPBL-NSG (?CD8); A] or depleted of CD4+ T cells [huPBL-NSG (?CD4); B]. On the same day they received 2×10⁶ PC3 cells implanted by s.c. injection. Tumors were recovered after 3 weeks and individually
weighed. Values represent mean of three animals per group?±?SD. *p??0.05. Representative
experiment, n?=?3 experiments.

Figure 6. Spleen reconstitution and tumor infiltration in NSG animals receiving purified human
CD4+ or CD8+ T cells. HuPBL-NSG, huPBL-NSG (?CD8) and huPBL-NSG (?CD4) animals were established as detailed
in Figure 5 and implanted with 2×10⁶ PC3 cells by s.c. injection. A D: The different purified PBL populations, before implantation, were stained with specific
antibodies against human CD3, CD4 and CD8 followed by FACS analysis. Results are gated
on the human CD3+ cells. Whole spleens B E and tumors C F were then recovered after 3 weeks and single cell suspensions stained with specific
antibodies against human CD4 and CD8 followed by FACS analysis. Results were gated
on all cells expressing CD4 and/or CD8. Representative experiment, n?=?3 experiments.