A novel microfluidic device capable of maintaining functional thyroid carcinoma specimens ex vivo provides a new drug screening platform

Thyroid carcinoma (TC) is the most common malignant disease of the human endocrine system, accounting for ~ 2.1% of all cancer cases worldwide, and disproportionately affecting women [1]. The incidence of TC has increased by 139% since the early 1990’s in the United Kingdom, with a particular increase in the diagnosis of papillary TC (PTC; Cancer Research UK [2018]). PTC, along with follicular-TC (FTC), form the differentiated TC group (DTC), which represent ~ 90% of all thyroid malignancies [2]. Patients with DTC have a 10-year survival rate of 90%, underpinned by an efficacious treatment regimen, which includes thyroidectomy followed by remnant and metastatic ablation using radioactive iodine (¹³¹I).

Treatment, using ¹³¹I, capitalises on the ability of thyrocytes to absorb iodide (I⁻) for thyroid hormone biosynthesis. I⁻ is transported across the basolateral membrane of thyrocytes by the sodium iodide symporter (NIS) using the sodium gradient generated by Na⁺/K⁺-ATPase [3]. Due to the essential involvement of the NIS in thyroid hormone biosynthesis and thus human metabolism in general, expression of the NIS is tightly regulated. In thyroid cancer, thyroid peroxidase (TPO) activity, and thus organification of free iodide is reduced. Oxidised iodine possesses a significantly longer retention time than free iodide, leading to a considerably reduced effective half-life in the malignant compared with the benign thyroid (0.5–3 days vs. 3–7 days, respectively; [4]). Despite this, treatment of DTC using ¹³¹I remains an effective means of eradicating remnant and metastatic malignant foci when a high dose can be delivered to the target tissue [5]. However, in approximately 15% of DTC cases, basolateral expression of the NIS is lost, due to phenotypic de-differentiation of thyrocytes, rendering ¹³¹I therapy ineffective [6]. Within this cohort of patients, the use of ¹³¹I therapy is negated, as thyrocytes cannot concentrate ¹³¹I, leading to a marked reduction in patient survival (only 10% alive at 10-years, mean survival 3–5 years; [7]). Treatment of advanced, ¹³¹I refractory TC using traditional systemic therapies including doxorubicin has limited efficacy with a high toxicity rate [8]. Alternative therapy using relatively new-to-market tyrosine-kinase inhibitors (TKI), particularly sorafenib and lenvatinib, has garnered significant interest and has shown objective response rates with significantly improved progression-free survival [9, 10]. Currently it is not possible to predict accurately a lack of response to ¹³¹I prior to administration, thus patients have a delay before receiving effective adjuvant therapy [11]. Although it is well accepted that iodide uptake is typically reduced in malignant thyroid tissue relative to surrounding tissue [12], the molecular underpinnings have not yet been fully elucidated. A previous study by Min et al. [6] investigated the relationship between NIS expression and ¹³¹I uptake in recurrent TC using immunohistochemical (IHC) analysis. They found that although lesional NIS expression predicted ¹³¹I uptake with 100% specificity, the sensitivity of this method was just 50% (14 of 28 patients with positive ¹³¹I uptake had negative NIS staining. Similarly, Arturi et al. [13] found, in a smaller study using RT-PCR, that only 4 out of 8 (50%) primary DTC with distant metastases and negative ¹³¹I uptake lacked NIS gene expression in the primary lesion; thus indicating that assessment of NIS expression is certainly not an infallible means of evaluating ¹³¹I uptake. Furthermore, Saito et al. [14] unexpectedly detected increased NIS mRNA and protein levels in PTC tissue compared with normal thyroid tissue. This variability suggests that reduced iodide uptake is perhaps due to failure to target or retain NIS protein at the basolateral membrane [15], rather than abrogated NIS protein production.

The microfluidic system described herein uses precision cut tissue slices (PCTS) which maintain the multicellular 3D architecture of the thyroid tissue essential for intercellular communication, which is not well recapitulated in standard in vitro systems. Two previous studies have employed static culture of thyroid tissue explants: Russo et al. [16] investigated apoptosis subsequent to ¹³¹I treatment in 1mm³ tissue fragments, whereas Nagy et al. [17] examined the effect of various hormones and cytokines on the proliferation of 2mm³ thyroid fragments. The results from these studies demonstrated that thyroid tissue cultured ex vivo can respond to external stimuli, however clinical utility was not assessed.

The bespoke device described in this manuscript mimics the in vivo vasculature and lymphatics by allowing continual perfusion of explanted tissue, along with removal of waste products and allows precise control of drug delivery [18]. Future treatment of ex vivo thyroid tissue maintained in this device with ¹³¹I, holds potential for a more realistic estimation of ¹³¹I treatment success by assaying not only uptake, but resultant cell death. The aim of this study was to characterise a reproducible, easy to use, microfluidic system to maintain precision sliced ex vivo thyroid tissue for up to 96 h.

The morphology of both freshly resected and post-culture (96 h) ex vivo thyroid tissue sections was assessed by H&E staining and showed preservation of gross morphology, with retention of tissue cohesion (Fig. 2). The follicular structure, typical of in vivo thyroid tissue, was clearly displayed in both fresh and post-culture tissue samples. Inter-tumour variation in cellularity and follicle size is clear, however these differences are retained in tumour tissue throughout culture.

Culture effluent was sampled throughout the 96 h culture period and examined for the presence of two markers of cell death, LDH and DCP. An initial raised level of LDH release was observed within the first 2 h of culture setup (Fig. 3a), which reduced after 24 h and remained consistently low for the remaining 72 h, until tissue was exposed to 10% (v/v) lysis buffer, which led to an average increase in LDH release from 0.01 ± 0.002 to 0.304 ± 0.06 and 0.005 ± 0.001 to 0.35 ± 0.03 absorbance units mg^− 1 in malignant and benign thyroid tissue, respectively. No difference in LDH release was found between the malignant and the benign tissue. DCP release by cells in the culture system followed a profile similar to that of LDH release (Fig. 3b), with the level being relatively high in the first 2 h of culture before decreasing after 24 h. DCP release reached a minimum after 72 h, with only 41.53 ± 28.33 RLU mg^− 1 detectable. Concordant with LDH release, purposeful lysis of cultured thyroid tissue subsequent to 96 h culture induced a > 1000-fold increase in detectable DCP.

No significant difference in viability between fresh (27.3 ± 7.3%) and post-96 h cultured (24.6 ± 8.8%) malignant thyroid tissue samples was observed using PI staining to (Fig. 4; n = 3; p = 0.19). These data were consistent with results using trypan blue exclusion to represent cell viability, whereby the mean proportion of viable dissociated cells was 38.33 ± 7.3% in fresh thyroid tissue prior to culture and 34 ± 9.5% in tissue after a 96 h culture period (Fig. 4; n = 3; p = 0.37). The proportion of apoptotic cells in both fresh and post-culture (+ 96 h) malignant thyroid tissues was assessed by TUNEL staining (n = 9) and a small, non-significant, increase in apoptotic cells was observed in the post-culture tissue (5.59 ± 0.99%) compared with the fresh tissue (2.77 ± 0.89%; Fig. 4-c and d). Moreover, on-chip treatment of thyroid explants with etoposide (topoisomerase II inhibitor) and SP600125 (JNK inhibitor) caused a significant increase in cellular apoptosis after 96 h treatment, compared to untreated thyroid tissue (15.25 and 14.24% vs. 5.59%; p = 0.0025 and p = 0.0069 respectively; Fig. 4).

Cell proliferation within the thyroid tumour tissue was investigated by IHC. Two markers were used; Ki67, a nuclear protein expressed during all active phases of the cell cycle (non-senescent cells) and BrdU which is incorporated during s-phase of the cell cycle [24]. This analysis demonstrated that proliferating cells were detectable in both freshly resected tissue and post-culture tissue. The mean percentage of Ki-67 positive nuclei was 1.45 ± 0.33% and 1.32 ± 0.34% in fresh and post-culture tissue (n = 8), respectively; no significant difference in the quantity of proliferating cells between fresh and post-culture malignant thyroid tissue was observed (p = 0.32). BrdU, which was perfused through the culture system for the final 24 h culture on the device, was clearly visible as incorporated into the cell nuclei of proliferating cells post-culture, illustrating that on-chip cell proliferation was occurring (Fig. 5). Subsequent to a 24 h perfusion with 10 μM BrdU, 12.09 ± 4.38% cells stained positive for BrdU incorporation (n = 3). Thyroid transcription factor 1 (TTF1) is crucial for thyroidal organogenesis and differentiation which, along with two other factors, paired box gene 8 (PAX8) and TTF2 (FOXE1), governs the expression of thyroglobulin (TG), thyroperoxidase (TPO) and NIS [25]. The expression of TTF1 was characterised in fresh (0 h) and post-culture (+ 96 h) malignant thyroid tissue; the staining pattern of TTF1 was maintained throughout culture in the microfluidic device (Fig. 5).

As expression of the NIS is crucial for adequate response to treatment to I¹³¹, it was important to ensure that 96 h culture of malignant thyroid explants within our MF device did not affect NIS protein levels. NIS expression was investigated in both freshly resected and post-culture malignant thyroid tissue and showed that there was no statistically significant difference between the density of NIS protein bands in fresh and cultured thyroid tissue, normalised to β-actin (Fig. 6; n = 5), indicating that the expression of NIS was maintained throughout culture.

The functionality of cultured ex vivo thyroid tissue was investigated by its capability to synthesise and release the primary thyroid hormone, T4. Production of T4 occurs as a result of a cycle of biochemical reactions occurring exclusively within the thyroid. Culture effluent was analysed by ELISA for the presence of T4 (n = 7 malignant tissue; n = 7 benign tissue). Data indicate that T4 release by both malignant and benign thyroid explants follows a similar overall trend, with an initially raised level of T4 production over the first 24 h, stabilising after 48 h (Fig. 7). Interestingly, T4 release by benign thyroid explants was consistently higher at each timepoint in comparison to malignant tissue and average T4 output throughout the complete culture period was significantly higher in benign tissue (0.29 ± 0.09 μg/dl mg^− 1) than in malignant tissue (0.14 ± 0.03 μg/dl mg^− 1; Fig. 7; p = < 0.05).

The current study aimed to characterise the ability of a PCTS device to maintain viable thyroid tissue explants over a 96 h period with a view towards therapeutic interrogation. Human tissue explants from human head and neck tumours, intestine and mouse brain have been maintained using similar microfluidic culture systems [20, 26, 27] however this is, to the best of the authors’ knowledge, the first time ex vivo human thyroid tissue has been cultured in a robust, easy-to-use, platform. Tissue sections taken from both fresh and PCTS-cultured thyroid tissue analysed by haematoxylin and eosin and IHC staining demonstrated that morphology was preserved in tissue which had been cultured in the device for 96 h, validated by a consultant histopathologist. Results observed agree with those shown in similar studies from the group, whereby head and neck squamous cell carcinoma and full thickness intestinal tissue also demonstrated a preservation of tissue morphology after 72 h on-chip culture [18, 20]. The gross inter-patient morphology of tissue specimens was variable, and heterogeneity was visible within each tissue sample. Areas of typical thyroidal histology exhibiting follicular architecture often neighboured regions of densely arranged cells or fibrous stroma. This inter- and intra- patient heterogeneity was expected and was also presented in a study by Radu et al. [28], who speculated that the morphological variation in thyroid carcinoma is due to manifestation of diverse tumour cell clones under the influence of the same etio-pathogenic factors. It will therefore be necessary to run multiple PCTS chips from the same patient when interrogating the tissue with different treatment regimens.

The release of both LDH and DCP as markers of cell membrane integrity was assayed throughout the culture period; loss of cellular plasma membrane integrity is synonymous with loss of cellular viability [29] and microfluidic culture offers a unique means of evaluating this in real time. As anticipated, the release of these two markers followed similar patterns, with a higher level at the beginning of the culture period, attributed to the damage caused to the tissue during preparation, which decreased during the remainder of the experiment. The relatively low LDH/ DCP release once the tissue had recovered from the initial cellular damage during set-up, which has also been experienced in previous studies [30, 31], indicates cultured explants do not undergo further degradation due to factors such as sheer stress or lack of oxygen within the culture chamber. Deliberate lysis of cultured tissue resulted in an increase in both LDH and DCP, suggesting that a significant number of viable cells remain after 96 h in the PCTS device, which agrees with the morphological assessment. A similar pattern has been observed studies from other groups where stabilisation of LDH release was seen after four days in a four-organ-chip model [32] and after six days in a liver/skin co-culture [33]. The discrepancy in time taken for tissue to recover from the initial damage could potentially be attributable to either the difference in tissue type, the system used or the presence of malignancy in the current system.

Tissue dissociation using a combined mechanical and enzymatic method offered an alternative means of investigating the survival of individual cells constituting the gross thyroid explants, by monitoring the ability of propidium iodide and trypan blue to traverse the cellular membrane. Again, no statistically significant difference in mean cellular viability was observed between samples which had been dispersed prior to, or after PCTS-culture, further demonstrating that on-chip maintenance was not causing increased cell death. This was despite the fact that the overall viability of cells after dispersal from gross tumour material was unexpectedly low (< 40%), presumably due to injury caused by the enzymatic (collagenase IV, DNase II) and mechanical (mincing of gross tissue using scalpel blades) process of dissociation. The high level of cell death observed in the current study is in contrast to Li et al. [34] who provided a mean cell viability of 65.66% (±4.96) using trypan blue staining following disaggregation of murine pancreatic fragments using collagenase IV. As malignant tumour tissue is marked by regions of necrosis owing to rapid oncogene-driven proliferation not supported by an adequate vascular bed [35], it is a possibility that presence of necrotic cells contributed to the lower overall viability in the current study. The viability of cells dispersed from human head and neck cancer specimens was assessed by flow cytometry in a separate study and produced viabilities ranging from 24 to 92% [36], partially aligning with the results seen here.

To further characterise on-chip tissue kinetics, the proportion of thyroid cells undergoing apoptosis was elucidated through the labelling of 3-OH termini of DNA strand breaks in situ, produced due to endonuclease mediated DNA fragmentation, characteristic of DNA degradation and late-stage apoptosis [37]. The culture of thyroid explants within the microfluidic device showed no change in the proportion of apoptotic cells. Similar studies have, however, reported comparable levels of apoptosis in untreated specimens’ post-microfluidic culture. Cheah et al. [22], found an apoptotic fraction of < 10% within untreated head and neck squamous cell carcinoma (HNSCC), whilst Wang et al. [38] also saw < 10% apoptosis post-culture in brain organoids, compared to ~ 40% apoptosis in organoids subjected to static culture conditions. The increase in apoptosis observed in this study was not mirrored in other tests of viability such as PI/TB uptake and LDH/DCP leakage data as the latter assays rely on loss of membrane integrity, synonymous with necrotic cell death [38], to facilitate detection of leaked cellular components (LDH/DCP) or PI/TB labelled cells. It is feasible that, if cultured for an extended period of time, a decrease in cell viability would also be observed using PI/TB uptake and LDH/DCP leakage assays, due to the occurrence of cell death by alternative mechanisms such as secondary necrosis, characterised by progressive loss of plasma membrane integrity, through caspase-3-dependent cleavage of the GSDMD-like protein DFNA5 [38, 39]. Furthermore, on-chip treatment of thyroid explants with 20 μM etoposide and SP600125 demonstrates the devices’ utility as a drug screening system. Perfusion of thyroid explants with either agent for 96 h resulted in significantly increased levels of apoptosis as examined by TUNEL staining. Etoposide is a well-characterised inducer of apoptosis [40] whilst SP600125 is a multikinase inhibitor which has been shown to have both anti-proliferative and migratory effects in thyroid cell lines [41].

IHC analysis was used to investigate the ability of the cells within the cultured tissue to proliferate. The Ki67 antigen is present in all active stages of the cell cycle, except for G_0, in which cells are in a quiescent state [42]. Previous studies have attempted to use Ki67 labelling index to aid diagnosis in thyroid neoplasms, however it has shown little value in this setting due to the low labelling index in most thyroid lesions (< 1% in 57% PTC samples tested; [43]). In agreement, Ki67-positive brown stained cells were detectable at low levels (< 2%) in tissue sections in the current study after 96 h within the PCTS device indicating the presence of a small number of proliferating cells. Furthermore, comparison of the proliferating cell population in both fresh and post-culture thyroid tissue revealed no significant difference.

BrdU staining was also employed, having the added advantage of demonstrating definitive on-chip division as BrdU is incorporated into de novo DNA during the S-phase of the cell cycle as a substitute for thymidine and is detectable by IHC [44]. Cells with BrdU positive nuclei (12.09%) were evident in thyroid tissue samples which had been maintained for 96 h in the PCTS-culture device, the final 24 h of which included perfusion with 10 μM BrdU. Identification of BrdU-positive cells throughout the tissue both demonstrates the ability of BrdU to perfuse successfully through the tissue and moreover the presence of proliferating cells in the S-phase of the cell cycle. A comparable study by Sart et al. [45] investigated BrdU uptake in H4-II-EC3 rat hepatoma cells cultivated in both a 2D and 3D spheroid setting. The authors saw a near 2-fold decrease in BrdU positivity in 3D (~ 16%) versus 2D (~ 30%), similar to results observed in this study.

In addition to the viability of the tissue, it was important to characterise the differentiation of thyroid explants, in order to examine whether culture within the microfluidic device could affect the expression of certain genes crucial for successful treatment with radioiodine. A previous study conducted by Cirello et al. [46] which aimed to investigate the effects of the multikinase inhibitor SP600125 on multicellular spheroids derived from primary thyroid tissue, examined expression of the transcription factor TTF1 as a marker of differentiation. The authors observed maintenance of TTF1 expression in spheroids derived from primary thyroid tissue. Encouragingly, the pattern of TTF1 staining in post on-chip culture (+ 96 h) thyroid tissue was equivalent to that of fresh thyroid tissue, suggesting that culture for 96 h in the PCTS device did not markedly affect differentiation.

The NIS is an integral plasma membrane glycoprotein that facilitates the accumulation of I⁻ within the thyroid gland at a concentration up to 40 times higher than plasma [47]. Application of I¹³¹ for the targeted destruction of malignant thyroidal tissue has become one of the most effective radiation regimens available, causing fewer side effects than other cancer therapies [48]. Loss of NIS protein expression at the basolateral membrane of thyrocytes is a well-accepted precursor for reduced effectiveness of I¹³¹ therapy, as demonstrated in poorly differentiated and undifferentiated thyroid cancers [49, 50]. Therefore, it was necessary to assess the expression of NIS in thyroid explants which had been subjected to 96 h within the microfluidic device and elucidate whether its expression was altered. Western blot data revealed that there was no significant change in terms of NIS protein levels caused by culture within the PCTS device, indicating that the system is valid for the future testing of I¹³¹ sensitivity.

T4 is exclusively synthesised by thyrocytes as a pro-hormone, which becomes de-iodinated within target tissues to produce the more biologically potent T3 [51]. This natural phenomenon allowed the measure of T4 release from on-chip PCTS to be investigated as a measure of tissue functionality. Loss of, or reduced functional differentiation, occurs in all thyroid carcinomas leading to diminished expression of proteins central to thyroid hormone production [52]. In the current study, benign tissue produced a significantly higher volume of T4 than malignant tissue throughout the culture period, agreeing with the notion that malignant thyroid tissue has a decreased ability to produce hormone [11]. Detection of T4 within culture supernatant strongly suggests that thyroid hormone was synthesised de novo by on-chip thyroid tissue and that essential components such as I⁻ are reaching the thyrocytes. In the clinical setting, the thyroid hormone precursor TG is used as a specific marker of remnant tumour tissue after thyroid removal [53]. As our intention was to demonstrate thyroid tissue functionality, we decided to measure T4 levels as T4 synthesis relies on the success of additional mechanistic steps such as cellular I⁻ uptake and TG iodination.

Results reported in this work validate the use of a PCTS device for the maintenance of ex vivo thyroid tissue which can be used for customising drug treatment. It is proposed to use the system described to delineate sensitivity of patients suffering from thyroid carcinoma to treatment with ¹³¹I, enabling patient stratification and reducing incidence of inappropriate treatment, reducing patient morbidity, whilst saving NHS resources. This is the first study of its kind to demonstrate reliably the maintenance of viable ex vivo thyroid tissue for up to 96 h, with pre- and post-culture tissue showing high uniformity in terms of morphology, viability and functionality.