Introduction

Ovarian cancer (OC) remains the most lethal gynaecological malignancy worldwide with a 5-year survival rate below 50% [1]. High grade serous carcinoma (HGSC), accounting for over 90% of OC cases, is an aggressive and heterogenous disease that is characterised by high mortality, late diagnosis and propensity for recurrence due to chemoresistance [2, 3]. The first line of treatment includes a combination of debulking surgery, and a platinum and/taxane-based chemotherapy regimen [4]. While around 70% of patients respond to the treatment, the majority of these subsequently unfortunately experience cancer recurrence due to a lack of effective maintenance therapies and the drug-resistant nature of the tumors [5]. Therefore, there is an urgent need to develop new therapeutics to improve the prognosis of OC, especially HGSC.

HGSC spreads predominantly in the peritoneal cavity whereby exfoliated tumor cells travel via ascites fluid to secondary sites [6]. Ascites, present in more than one third of OC patients, acts as a reservoir for various cellular components including tumor cells, cancer associated fibroblasts, mesothelial cells, and immune cells [7]. Several studies have observed that immune cell infiltration into the tumor microenvironment, where malignant ascites accumulates, is correlated with good prognosis of OC [8, 9]. For instance, the presence of CD3 + T cells in the tumor is correlated with increased survival in HGSC patients [10]. Studies also indicate the importance of natural killer (NK) cells, although the precise mechanism of how NK cells can enhance patient prognosis is still to be evaluated [11,12,13].

NK cells, effector lymphocytes of the innate system, play a pivotal role in inducing cytolytic activity against aberrant cells caused by viral infections and cancer [14]. The ability of NK cells to exhibit antitumor effects without priming or prior activation is advantageous compared to other immune cells, which are restricted by patient-specific major histocompatibility complex molecules to elicit an immune response [15, 16]. Therefore, NK cell-based adoptive cellular immunotherapy has become increasingly popular [17]. NK cell activation is tightly controlled by a range of activating and inhibitory receptors that interact with respective ligands expressed on target cells [18]. Once activated, NK cells can exert cytolytic effects through various effector mechanisms such as antibody dependent cellular cytotoxicity, cytotoxic granules, and secretion of inflammatory cytokine or chemokines [19,20,21]. These diverse functions of NK cells in immune surveillance may allow a potentially important therapeutic strategy for cancer immunotherapy. Indeed, the presence of NK cells within the immune repertoire of ascites is reported to be positively correlated to better outcomes in OC patients [15]. However, the killing capacity of NK cells can be mitigated by several immune evasion strategies of the cancer itself, including the establishment of an immunosuppressive milieu, which promotes cancer cell survival, metastasis and invasion [22].

Podocalyxin (PODXL) is a sialomucin normally expressed on the surface of various cells including kidney glomeruli, epithelial and endothelial cells, mesothelium, and hematopoietic progenitor cells [23]. High PODXL expression has been associated with aggressive tumor phenotypes and poor prognosis in several cancers including breast, pancreatic and ovarian cancer [23,24,25,26,27]. In OCs, PODXL is more likely to be expressed in HGSC (87%) compared to other subtypes, and its surface expression is more associated with late stage HGSC and a significant decrease in disease-free survival [25, 28].

Functionally, we have reported that in HGSC, PODXL promotes the formation of compact and chemoresistant cancer spheroids [24]. In both HGSC-derived cell lines and ascites-derived primary HGSC cells, cancer spheroids with higher PODXL expression are more compact and less fragile to fragmentation than those of lower PODXL levels [24]; furthermore, the former were more resilient to chemotherapy drugs as they showed higher cell proliferation following treatment [24]. These results suggest that PODXL increases HGSC spheroid survival [24].

It is still unknown how PODXL bestows HGSC spheroids with such an advantage, but we believe it is interrelated to the fundamental role of PODXL in promoting epithelial barrier functions, which has been recently revealed in endometrial epithelial cells where PODXL promotes an impermeable epithelial barrier [29]. We therefore hypothesized that PODXL may play a role in protecting cancer spheroids from the surroundings by hindering penetration/action of drugs and immune cells such as NK cells.

In this study we aimed to investigate whether PODXL levels in HGSC spheroids influence NK cell infiltration and spheroid destruction. We co-cultured HGSC spheroids with primary human NK cells isolated from peripheral blood mononuclear cells (PBMCs) and examined their infiltration into the spheroid and related cytotoxicity. We first used a cell line model of HGSC spheroids employing Kuramochi cells, which express the highest level of PODXL among known HGSC cell lines [24]. To examine the importance of PODXL, we compared spheroids of control and PODXL knockout (PODXL-KO) Kuramochi cells that we have previously created using CRISPR/Cas9 gene editing [24]. We then validated the data in primary cancer spheroids derived from ascites of HGSC patients that express high and low levels of PODXL. Collectively, our results suggest that PODXL protects HGSC spheroids from infiltration and cytotoxicity effects of NK cells.

Methods

Culture of Kuramochi and primary HGSC cancer cells

Kuramochi cell line was purchased from CellBank Australia (NSW, Australia) and cultured in RPMI 1640 Medium + GlutaMAX supplement (Thermo Fisher Scientific, MA, USA, #61870036). Control (transfected with empty vectors) and PODXL knockout (KO) Kuramochi cells were generated as previously described [24], and maintained with 1 µg/ml puromycin (Sigma-Aldrich, #P8833) that was added into the culture media. Ascites-derived primary HGSC cells were isolated as previously described [24], and maintained in a 1:1 ratio of Medium 199 (Thermo Fisher Scientific, #11150-059) and MCDB131 (Thermo Fisher Scientific, #10372-019). Primary cells were screened for CA125 to confirm OC cancer origin and were used between passages 4–7. All media were supplemented with 10% (for Kuramochi cells) or 15% (for primary cells) fetal bovine serum (FBS, Thermo Fisher Scientific) and 1% antibiotic–antimycotic (Thermo Fisher Scientific, #15240062); all cells were cultured at 37 °C under 5% CO2.

Isolation of primary human NK cells and evaluation by flow cytometry

NK cells were isolated from PBMCs derived from whole blood of 4 healthy female volunteer donors (aged 21–28), which was provided by the Australian Red Cross with ethics approval by RMIT College of Human Ethics Advisory Network (#28056). All work was conducted according to the Declaration of Helsinki Principles and the Australian National Health and Medical Research Council (NHMRC) Code of Practice. Signed informed consent was obtained from all donors before the study.

PBMCs were isolated from fresh blood by a density gradient centrifugation using Lymphoprep (StemCell Technologies, BC, Canada, #07801) whereby 10 ml of Lymphoprep was added to a 50 ml tube then overlayed with 25 ml of diluted blood (diluted 1:1 in incomplete (i) RPMI). The tube was then centrifuged at 800 xg for 20 min at 23 °C with brake off. The interface containing mononuclear cells was transferred into a fresh 50 ml tube containing iRPMI, centrifuged at 450 xg for 4 min at 23 °C with brake on, washed once, frozen in freezing media (10% DMSO in FBS) and stored at ≤−150 °C until further use.

At the time of NK cell use, cryopreserved PBMCs were thawed and NK cells were isolated using an NK Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany, #130092657), LS Column (Miltenyi Biotec, #130122729,) and QuadropMACS™ Separator (Miltenyi Biotec, #130090976) according to the manufacturers’ protocol. The enrichment of NK cells was evaluated by staining for viability with Live/dead NearIR (Thermo Fisher Scientific, #L34975) followed by anti-CD3 fluorescein isothiocyanate (FITC) (Miltenyi Biotec, #130113690) and anti-CD56 phycoerythrin (PE) (Miltenyi Biotec, #130113874) and analysis on an LSR Fortessa X-20 flow cytometer (BD Biosciences, USA). NK cells were defined as live cells that were CD3- and CD56+ (Supplementary Fig. 1) and on average 93% of the enriched cells were confirmed to be NK cells.

Isolated NK cells were activated by culturing for 72 h at 37 °C under 5% CO2 in RPMI 1640 Medium + GlutaMAX supplement (Thermo Fisher Scientific, MA, USA, #61870036), supplemented with 10% FBS, 1% penicillin-streptomycin, and 10ng/ml IL-15 (Stem cells, #78031.1). They were stained with carboxyfluoroscein succinimidyl ester (CFSE, Thermo Fisher Scientific, #C34570) for 10 min (diluted 1:1000 in PBS) as per manufacturer’s protocol before being co-cultured with cancer spheroids.

Co-culture of cancer spheroids with NK cells

Kuramochi and primary HGSC cancer spheroids were formed in Costar ultra-low attachment round bottom 96 well plates (Merck, Darmstadt, Germany, #CLS7007-24EA) by culturing 2,500 cells per well in complete medium for 3 days as previously reported [24]. The resulting spheroids were then mixed with fluorescently labelled NK cells that were prepared as described above (at a ratio of 3:1), and co-cultured for 24, 48 and 72 h respectively. NK cell infiltration into spheroids and cancer cell cytotoxicity were assessed as described below.

Analysis of NK cell infiltration and the impact on the spheroid size and cell number

At each time point of assessment, brightfield images of the co-cultured spheroids were taken using the Nikon eclipse TS100 microscope equipped with a Nikon DS-Fi1 camera (Tokyo, Japan); thereafter confocal images of co-cultured spheroids were taken with an A1R confocal microscope (Nikon, Japan) to identify the fluorescently labelled NK cells. The co-cultured spheroids were then washed with PBS using a 10 µl pipette tip until all attached NK cells on the outside of the spheroids are completely removed, after which brightfield images of the washed spheroids were taken.

To analyse NK cell penetration into the spheroid, total NK cell fluorescence intensity within the co-cultured, unwashed spheroids was quantified (as washing may lead to loss of some NK cells). To do this, the outline of each spheroid was determined on the washed spheroid then overlayed onto the confocal image of the unwashed spheroid, the total NK cell fluorescence intensity within the spheroid were then determined using the ImageJ software version 1.53c (NIH, USA). The final data were expressed as total fluorescence after subtracting background fluorescence. The experiment was repeated 4 times with NK cells of 4 different donors (n = 4). For each experiment with NK cells of one donor, 3 individual spheroids were quantified and their average was used as the data. This applies to all other experiments concerning quantification following co-culture with NK cells.

The volume of spheroids following the PBS wash was also determined at each time point. To do this, the diameter of each spheroid was measured using the average length value of 4 different angles using the “straight line” function of the ImageJ software, and the spheroid volume was calculated using the formula of 4/3πr3. The number and viability of live cells contained within the washed spheroids were also analysed. To do this, the washed spheroids were trypsinized and dissociated into single cells, they were then mixed with trypan blue (Thermo Fisher Scientific) and analysed with a Countess 3 automated cell counter (Thermo Fisher Scientific).

Analysis of NK cell-induced cytotoxicity within spheroids

Spheroids were formed and co-cultured with fluorescently labelled NK cells for 24, 48 and 72 h respectively as described above, they were then incubated for 1 h with CellEvent Caspase-3/7 red detection reagent (Thermo Fisher Scientific, #C10430) (diluted 1:100 in complete media); after which the spheroids were first imaged using an A1R confocal microscope (Nikon, Japan), then dissociated into single cells by gentle pipetting and assessed for fluorescence at ~ 502/530 nm (excitation/emission) on a CLARIOstar® Plus plate reader (BMG LABTECH, Ortenberg, Baden-Württemberg, Germany). As primary HGSC spheroids were too small and compact to dissociate by manual pipetting, primary spheroids were measured on the plate reader without dissociation.

Analysis of Ki-67 positive cancer cells within the co-cultured spheroids

To assess cancer cell proliferation within spheroids following co-culturing with NK cells, spheroids were co-cultured with NK cells as described above, they were then resuspended into single cells using trypsin, pipetted onto a droplet of Histogel (Epredia, MI, USA, #HG- ID="EN6">4000-012), then smeared onto a microscope slide. After the gel was air dried, cells were fixed with 4% paraformaldehyde for 30 min, permeabilised with 0.1% Triton X-100 for 10 min, then blocked with 1% BSA for 2 h. Cells were then incubated first with anti-Ki-67 rabbit antibodies (Abcam, Cambridge, UK, #ab16667; 1:250 dilution in 1% BSA) or rabbit IgG (Dako, #X0936; diluted to 4 ug/ml in 1% BSA) at 4 °C overnight, then with rabbit-anti mouse Alexa Fluor 568 antibodies (Thermo Fisher Scientific, #a10042; 1:200 dilution in 1% BSA) for 2 h. Nuclei was counterstained for 10 min with DAPI (Sigma Aldrich, #D9542; diluted in PBS to 0.5 µg/ml). A drop of fluorescent mounting agent (Dako, #S3023) was added to the slide, a coverslip was mounted, and cells were analysed using a BX60 fluorescent microscope (Olympus, Japan). The percentage of Ki67 positive cells over total number of live cells were calculated.

Statistical analyses

GraphPad Prism version 10 (San Diego, CA) was used for statistical analysis. Paired t-test was applied wherever appropriate, and data were expressed as mean ± standard deviation (SD); P ≤ 0.05 was considered significant.

Results

PODXL-KO Kuramochi spheroids are more vulnerable to NK cell infiltration than controls

We first employed spheroids formed with the Kuramochi line as a model of HGSC cells [24]. To study how PODXL levels may influence spheroid susceptibility to NK cells, we compared spheroids formed with control and PODXL-KO Kuramochi cells that we have previously engineered [24]. These spheroids were co-cultured with human NK cells, which were isolated from PBMCs of healthy female donors (n = 4) and labelled with CFSE fluorescent dye. The co-culture was examined after 24, 48 and 72 h. Representative brightfield and confocal images of spheroids immediately after the co-culture are shown in Fig. 1A. For both types of spheroids, a “ring” of NK cells was present around each spheroid, which was more obvious on the confocal image (Fig. 1A-a and b, both panels). The “thickness” of these “NK cell rings” around did not change significantly over time in control spheroids (Fig. 1A-b, top panel), but they became increasingly thick and more dispersed towards the spheroid centre with increasing time in co-culture in the PODXL-KO spheroids (Fig. 1A-b, bottom panel). When these co-cultured spheroids were washed with PBS to remove un-infiltrated NK cells that surrounded the spheroids, the size of control and PODXL-KO spheroids also appeared to differ (Fig. 1A-c, both panels). We thus further examined NK cell infiltration into spheroids and the impact of the co-culture on spheroid size and cell number. To analyse NK cell infiltration, we determined the total fluorescence readings of NK cells that were contained inside the spheroid after the co-culture but before PBS wash, reasoning that the wash might cause some NK cells to leak out of the spheroids. To achieve this, the outline of each spheroid was determined on the washed spheroid then overlayed onto the confocal image of the un-washed spheroid (Fig. 1B, top panel), and the total NK cell fluorescence within the spheroid outlines were quantified (Fig. 1B, bar graph). While no difference was apparent at 24 h, significantly more NK cells were present inside the PODXL-KO spheroids as compared to the control spheroids at both 48 h and 72 h of co-culture, indicating PODXL-KO spheroids sustained more NK cell infiltration (Fig. 1B).

Fig. 1
figure 1

Co-culture of control and PODXL-KO Kuramochi spheroids with human NK cells. A Representative images of spheroids co-cultured with NK cells for 24, 48 and 72 h respectively. Top panel: control spheroids. Bottom panel: PODXL-KO (KO) spheroids. For each panel: a and b, immediately after the co-culture; c, after PBS wash to remove NK cells still present outside the spheroids; a and c, brightfield; b, confocal images of NK cells (fluorescently stained in green). B Analysis of NK cell infiltration into spheroids. Images of b in A) (NK cells in green) overlaid with the outlines of spheroids as yellow circles derived from images of c in A). Bar graph, total NK cell fluorescence present inside the spheroids. C Analysis of spheroid size following PBS wash. Images of c in A) presented together with yellow circle outlines shown in B). Bar graph, spheroid volume. D Analysis of live cells present within the spheroid after the wash. Data presented as percentage of live cells of untreated counterpart spheroids at each time point. Scale bar: 50 µm. Data as mean ± SD, n=4. *P < 0.05, ** P<0.01, ****P < 0.0001

Full size image

To examine the impact of NK cell infiltration on spheroid size, we used the images of the washed spheroids (Fig. 1C, images also show spheroid outlines used in Fig. 1B) and measured the diameters of these outlines and calculated the spheroid volume (Fig. 1C, bar graph). No difference was obvious at 24 h, but at later timepoints control spheroids appeared to increase in size while PODXL-KO spheroid did not (Fig. 1C, bar graph). Consequently, the PODXL-KO spheroids were significantly smaller than that of control spheroids at 72 h (Fig. 1C, bar graph). In the absence of NK cells, at each time point, spheroids of both the control and PODXL-KO did not differ much in size (Supplementary Fig. 2 A).

We further analysed the total live cell numbers remaining inside the spheroids (Fig. 1D). Within the first 24 h of NK cell co-culture, both spheroids showed a steep decline in cell counts. However thereafter, cell numbers gradually bounced back in control spheroids but declined in PODXL-KO spheroids, resulting in significantly more cells in control as compared to PODXL-KO spheroids at both 48 and 72 h (Fig. 1D). These data are consistent with the spheroid size differences shown in Fig. 1C.

PODXL-KO spheroids show higher NK cell-induced cytotoxicity than controls

Next, we assessed how NK cell-induced cytotoxicity differed between PODXL-KO spheroids and controls. Spheroids were co-cultured with CFSE-labelled NK cells for 24, 48 and 72 h respectively as above, then caspase-3/7 activity within spheroids was measured (Fig. 2). Caspase-3/7 activity was apparent in both types of spheroids, but the signal was stronger in PODXL-KO as compared to control spheroids (Fig. 2A-b and c, both panels). Furthermore, more overlap was seen between caspase-3/7 activity (red) with NK cell location (green) inside the spheroid in PODXL-KO (Fig. 2A-b, bottom panel) as compared control spheroids (Fig. 2A-b, top panel).

Fig. 2
figure 2

Analysis of caspase-3/7 activity within control and PODXL-KO Kuramochi spheroids following co-culture with human NK cells. A Representative images of spheroids co-cultured with NK cells for 24, 48 and 72 h respectively then analysed for caspase-3/7 activity. Top panel: control spheroids. Bottom panel: PODXL-KO (KO) spheroids. For each panel: a, brightfield images of spheroids co-cultured with NK cells; b and c, confocal imaging of caspase-3/7 activity (red) overlaid with (b) or without (c) NK cells (green). B and C Analysis of caspase-3/7 activity. B) Images of c in A) overlaid with spheroid outlines as yellow circles. C) Quantification of caspase-3/7 activity. Data presented as total fluorescence reading. Mean ± SD, n=4. ** P<0.01

Full size image

To further analyse the caspase-3/7 activity inside the spheroids, confocal images of caspase-3/7 activity shown in Fig. 2A was overlayed with spheroid outlines as done for Fig. 1C, which indicated that the red signals were contained inside the spheroids but more intensely in PODXL-KO compared to control spheroids (Fig. 2B). We then measured the total red fluorescence readings inside these spheroids using a plate reader (Fig. 2C). In the absence of NK cells, caspase-3/7 was negligible in the spheroids of both groups (Supplementary Fig. 2B). When co-cultured with NK cells, no difference was seen at 24 h, however with increasing time, the caspase-3/7 activity remained relatively stable in PODXL-KO spheroids but decreased in control spheroids. As a result, PODXL-KO spheroids displayed significantly higher caspase-3/7 activity than control spheroids, especially at 72 h (Fig. 2C). These results indicate that PODXL-KO spheroids endured higher rates of NK cell-induced cytotoxicity overtime than control spheroids.

We next examined the proliferative capacity of the surviving cells within the spheroids by assessing proliferation marker Ki67 at 48 h after co-culture with NK cells (Fig. 3). The control spheroids displayed more Ki67-positive cells than PODXL-KO spheroids (Fig. 3A). Quantification showed that 56% of the cells in the control but only 32% of the PODXL-KO spheroids were Ki67 positive (Fig. 3B). In the absence of NK cells, spheroids of both groups consistently showed over 70–90% Ki67 positivity (Supplementary Fig. 2 C). Thus, Ki67 positivity was inversely correlated to caspase-3/7 activities in these spheroids (Figs. 2 and 3).

Fig. 3
figure 3

Analysis of cell proliferation marker Ki67 in Kuramochi spheroids following co-culture with NK cells. Data of control and PODXL-KO (KO) spheroids co-cultured with NK cells for 48 h are presented. A Representative images of Ki67 immunostaining (red); blue, DAPI. Scale bar: 50 µm. B Quantification of Ki67 staining. Data presented as percentage of Ki67-positive cells over all live cells. Mean ± SD, n=4. ** P<0.01

Full size image

More NK cells infiltrate into primary HGSC spheroids that express lower levels of PODXL

We next investigated spheroids of ascites-derived primary cells obtained from HGSC patients. Due to the difficulty of culturing primary cells, we focused on cells from two patients that we previously found to express the highest and lowest levels of PODXL among a cohort [24], and named them as high-PODXL and low-PODXL cells [#1 and #6 respectively as shown in [24]. Spheroids formed with these primary cells were co-cultured with NK cells for 24, 48 and 72 h and analysed as for Kuramochi spheroids. Representative brightfield and confocal images of these spheroids immediately after the NK cell co-culture are shown in Fig. 4A. Here, grossly NK cells penetrated more into the centre of both high-PODXL and low-PODXL spheroids (Fig. 4A-a and b, both panels), rather than forming “rings” as seen in Kuramochi spheroids (Fig. 1A).

Fig. 4
figure 4

Co-culture of ascites-derived primary HGSC spheroids with human NK cells. A Representative images of spheroids of HGSC cells expressing high (High-PODXL) or low (Low-PODXL) levels of PODXL that were co-cultured with NK cells for 24, 48 and 72 h respectively. Top panel: High-PODXL spheroids. Bottom panel: Low-PODXL spheroids. For each panel: a and b, immediately after the co-culture; c, after PBS wash to remove NK cells still present outside the spheroids; a and c, brightfield; b, confocal images of NK cells (fluorescently stained in green). B Analysis of NK cell infiltration into spheroids. Images of b in A) (NK cells in green) overlaid with spheroid outlines as yellow circles derived from images of c in A). Bar graph, total NK cell fluorescence present inside the spheroids. C Analysis of spheroid size following the PBS wash. Images of c in A) are presented together with yellow circle outlines shown in B). Bar graph, spheroid volume. D Analysis of live cells present within the spheroid after the wash. Data presented as percentage of live cells of the untreated counterpart spheroid at each time point. Scale bar: 50 µm. Data as Mean ± SD, n=4. *P < 0.05, ** P<0.01

Full size image

When NK cell infiltration was examined more closely as done for Kuramochi spheroids (Fig. 4B), no difference was seen at 24 h; however, with increasing co-culture time, the total NK cell fluorescence intensity inside the spheroids decreased in high-PODXL but not in low-PODXL spheroids (Fig. 4B, bar graph), leading to significantly more NK cells present inside low-PODXL spheroids than high-PODXL spheroids especially at 72 h.

In the absence of NK cells, at each time point, spheroids of high-PODXL and low-PODXL primary cells did not differ much in size (Supplementary Fig. 3 A). In contrast, when co-cultured with NK cells, both groups showed a trend of size reduction over time, but low-PODXL spheroids were significantly smaller than high-PODXL spheroids (Fig. 4C, bar graph). Counting live cells present inside these spheroids revealed that for both high-PODXL and low-PODXL spheroids, the live cell numbers reduced sharply following the initial 24 h co-culture (Fig. 4D). At later timepoints, cell numbers started to bounce back somewhat but the increase was greater in high-PODXL as compared to low-PODXL spheroids, resulting in more cells in high-PODXL than low-PODXL spheroids overall (Fig. 4D). Collectively, these data of primary HGSC spheroids are consistent with the Kuramochi spheroids, showing that lower PODXL expression correlated with more NK cell infiltration and greater reductions in spheroid size and live cell numbers.

Primary HGSC spheroids expressing lower PODXL display higher NK cell-induced cytotoxicity

We also examined NK cell-induced cytotoxicity in primary HGSC spheroids (Fig. 5). Primary spheroids were co-cultured with NK cells as above, then assessed for caspase-3/7 activity. Due to limited availability of primary cells, and because studies of caspase-3/7 activity with the Kuramochi line showed little difference at 24 h (Fig. 2B and C), here we only examined the caspase-3/7 activity after 48 and 72 h co-culture of primary spheroids with NK cells (Fig. 5).

Fig. 5
figure 5

Analysis of caspase-3/7 activity following co-culture of ascites-derived primary HGSC spheroids with NK cells. A Representative images of spheroids expressing high (High-PODXL) or low (Low-PODXL) levels of PODXL co-cultured with NK cells for 24, 48 and 72 h respectively then analysed for caspase-3/7 activity. Top panel: High-PODXL spheroids. Bottom panel: Low-PODXL spheroids. For each panel: a, brightfield images of spheroids co-cultured with NK cells; b and c, confocal imaging of caspase-3/7 activity (red) overlaid with (b) or without (c) NK cells (green). B and C Analysis of caspase-3/7 activity. B) Images of c in A) overlaid with outlines of spheroids as yellow circles. C) Quantification of caspase-3/7 activity. Data presented as total fluorescence reading. Mean ± SD, n=4. *P < 0.05

Full size image

Caspase-3/7 activities were present in both groups of spheroids, however the overlap between NK cells location and caspase-3/7 activity appeared to be more towards the centre in low-PODXL than high-PODXL spheroids (Fig. 5A). In the absence of NK cells, caspase-3/7 was negligible in the spheroids of both groups (Supplementary Fig. 3B). The Closer examination of caspase-3/7 fluorescence presentation showed, visually (Fig. 5B) and quantitatively (Fig. 5C), that caspase-3/7 activity was higher in low-PODXL as compared to high-PODXL spheroids at both 48 h and 72 h.

We also assessed the proliferative capacity of the surviving cells within these primary spheroids by staining for Ki67 at 48 h (Fig. 6). Visually (Fig. 6A) and quantitively (Fig. 6B), fewer Ki67-positive cells were present in low-PODXL as compared to high-PODXL spheroids. Together, these results suggest an inverse correlation between PODXL levels and NK cell-induced cytotoxicity, consistent with the data observed with Kuramochi spheroids.

Fig. 6
figure 6

Analysis of cell proliferation marker Ki67 in primary cancer spheroids following co-culture with NK cells. Data of spheroids of expressing high (High-PODXL) or low (Low-PODXL) levels of PODXL co-cultured with NK cells for 48 h are presented. A Representative images of Ki67 immunostaining (red); blue, DAPI. Scale bar: 50 µm. B Quantification of Ki67 staining. Data presented as percentage of Ki67-positive cells over all live cells. Mean ± SD, n=4. **** P<0.0001

Full size image

Discussion

This study investigated whether PODXL levels in HGSC spheroids may influence NK cell infiltration and cytotoxicity. We first used a cell line model, employing spheroids formed with control and PODXL-KO Kuramochi cells. When these spheroids were co-cultured with human NK cells, PODXL-KO spheroids showed higher NK cell infiltration and greater NK cell-induced cytotoxicity than controls spheroids, resulting in smaller spheroids with fewer live and proliferative cells remaining in the former compared to the latter. We then validated the data in primary spheroids derived from ascites of HGSC patients expressing different levels of PODXL. Again, NK cell infiltration was greater and more severe cytotoxicity in primary HGSC spheroids that expressed lower than higher levels of PODXL, leading to worse spheroid destruction and less proliferative capacity in the former. Collectively, these data suggest that PODXL may play an important protective role in HGSC spheroids from NK cell infiltration and spheroid destruction.

Several prior studies have demonstrated that NK cells can destroy cancer cells when added to cell monolayers [11, 12, 30]. However, only a few studies have investigated the effects of NK cells using spheroid models which can mimic the in vivo cell architecture, morphology and cell-cell interactions of metastatic spheroids [31,32,33,34,35,36]. In particular, NK cells derived from human hematopoietic stem and progenitor cells were reported to actively infiltrate and kill OC spheroids formed with cell lines SKOV3, IGROV1, and OVCAR3 in a dose dependent manner [36]. They are consistent with our findings that NK cells can infiltrate and destroy HGSC spheroids. However, our studies considerably extended this line of research and investigated the potential role of a specific molecule, PODXL, whose expression has been positively associated with poor prognosis of HGSC patients.

In Kuramochi spheroids, the levels of PODXL made a significant difference in spheroid susceptibility to NK cell infiltration and spheroid destruction. We saw more NK cell penetration, bigger drops in spheroid volume, and a larger reduction in live cell numbers in PODXL-KO spheroids than the control over times during co-culture with NK cells. As NK cell infiltration has been reported to be associated with cancer cell apoptosis inside the spheroid [31, 32, 36], we also examined apoptosis using caspase-3/7 activity as a marker. Indeed, we observed higher caspase-3/7 activities in PODXL-KO spheroids than controls, which positively correlated with their differences in NK cell penetration. Further analysis of proliferative capacity of cells remaining in the spheroid showed that more Ki67-positive cells were present in the control than PODXL-KO spheroids, indicating that the control spheroids are more actively proliferating and recovering following NK cell attack than PODXL-KO spheroids. Similar observations were also made with ascites-derived primary HGSC spheroids following co-culture with NK cells; spheroids expressing lower PODXL exhibited more NK cell infiltration, higher apoptosis and less proliferation than spheroids with higher PODXL. To our knowledge, this is the first study to utilise ascites-derived patient cancer spheroids to examine NK cell infiltration as well as the importance of PODXL. A recent study has reported that a glycopeptide epitope on the extracellular domain of PODXL, which is expressed only on cancer cells, is correlated with poor immune cell infiltration in HGSC tumors [37]. Although it is unclear how this glycopeptide epitope dictates immune infiltration, this study largely supports our investigation that PODXL may play a role in impeding immune cell penetration and thus, removal of PODXL may present a treatment opportunity to increase anti-cancer immune responses in HGSC [38, 39].

NK cell receptors recognise and destroy malignant cells through activating or inhibitory signals to induce cytotoxicity [40]. As an immune evasion strategy, cancer cells can modulate the expression of corresponding activating or inhibitory ligands to escape NK cell attack [41]. Studies in MCF7 breast cancer cells have demonstrated that PODXL expression on cancer cells can downregulate the NK cell activating receptors such as NKG2D, NKp30, and NKp44 to influence NK cell activity [42]. Many cancers are also known to shed NKG2D ligands as part of the tumor immune evasion strategy [43]. In a study using CaSki and SiHa cervical cancer cell line spheroids, co-culture with NK cells showed an accumulation of soluble NKG2D ligands (MICA, MICB and ULBP2) in the supernatant of co-cultures, which paralleled the loss of ligands from the cell surface [31]. This suggests that cancer spheroids can continuously shed cellular ligands of NKG2D [31]. We therefore speculate that PODXL could play a role in evasion of NK cell killing through facilitating the shedding of NK cell activating ligands on HGSC cells, although this needs further investigation.

It is well known that age is a strong risk factor of OC especially in those over the age of 65 [44], however, mechanisms remain unclear. Since physiological aging of NK cells are known to be associated with NK cell immunosenescence, a phenomenon that is linked to decreased NK cell activity and increased incidences of infections [45], in this study we used NK cells of young (aged 21–28) and healthy females to ensure their activity was not compromised by age. Future studies are warranted to investigate whether the age of NK cells also matters, and whether older women have physiologically older NK cells that are less potent for control of cancer cells. If this is true, it may provide novel insights into the understanding of why age increases the risk of OC. It would also rationalise why autologous NK cell therapy can be less effective [46, 47], as allogeneic NK cells from younger donors may lead to more favourable outcomes in older patients.

A limitation of this study was the limited amount of ascites-derived primary cells that we were able to acquire and maintain in culture. Furthermore, as HGSC is a heterogenous disease, the physiological differences observed in the primary cells may be influenced by other factors due to patient variation. Therefore, further studies with a larger cohort of patients would help further establish the correlations between PODXL expression and patient response to NK cell infiltration and cytotoxicity.

In summary, our data suggests that PODXL plays an important, protective role in hindering NK cell infiltration and NK cell-mediated destruction of HGSC cancer spheroids, and that lowering the levels of PODXL may sensitise HGSC to NK cells. Consequently, strategies to downregulate PODXL in HGSC patients with high PODXL expression may assist them to fight the cancer through their own NK cells and/or through adoptive immunotherapy using donor NK cells.