Jun 03, 2023
Alto rendimento, etiqueta
Volume de comunicações da natureza
Nature Communications volume 13, Número do artigo: 3385 (2022) Citar este artigo
6841 acessos
11 Citações
110 Altmétrico
Detalhes das métricas
Aglomerados extremamente raros de células tumorais circulantes (CTC) são cada vez mais apreciados como precursores altamente metastáticos e praticamente inexplorados. As tecnologias são projetadas principalmente para detectar CTCs únicos e muitas vezes falham em levar em conta a fragilidade dos clusters ou em alavancar marcadores específicos de cluster para maior sensibilidade. Enquanto isso, as poucas tecnologias voltadas para clusters CTC carecem de escalabilidade. Aqui, apresentamos o Cluster-Wells, que combina a velocidade e a praticidade da filtração por membrana com a triagem sensível e determinística oferecida pelos chips microfluídicos. Os mais de 100.000 micropoços nos Cluster-Wells prendem fisicamente os clusters de CTC em sangue total não processado, isolando suavemente praticamente todos os clusters a uma taxa de transferência de >25 mL/h e permitem que os clusters viáveis sejam recuperados do dispositivo. Usando o Cluster-Wells, isolamos clusters de CTC variando de 2 a 100+ células de pacientes com câncer de próstata e ovário e analisamos um subconjunto usando sequenciamento de RNA. O isolamento rotineiro de aglomerados de CTC democratizará a pesquisa sobre sua utilidade no tratamento do câncer.
Células tumorais circulantes únicas (CTCs) coletadas da corrente sanguínea de pacientes com câncer fornecem informações valiosas sobre o estágio da doença1, permitem prognóstico e diagnóstico minimamente invasivos2,3,4, aprimoram nossa compreensão da metástase no nível celular5,6 e oferecem a potencial para melhorar o manejo clínico do câncer7,8. Além dessas CTCs únicas, os agregados de CTC que permanecem aderidos em circulação têm sido de grande interesse científico e clínico desde a década de 1950. Embora esses aglomerados de CTC, ou microêmbolos tumorais circulantes, sejam extremamente raros (estimados em apenas 2 a 5% de todas as CTCs), eles são desproporcionalmente eficientes em semear metástases. Estima-se que sua propensão metastática seja 100 vezes maior do que a de CTCs únicas9,10,11, com base em sua menor taxa de apoptose e redução dos atributos de sobrevivência prolongada9,12. Além disso, descobriu-se que um subconjunto de clusters de CTC derivados de pacientes inclui células imunes do hospedeiro13,14, destacando a utilidade desses precursores metastáticos para lançar luz sobre as interações tumor-sistema imunológico e seu papel na metástase. Por exemplo, clusters de CTC que transportam neutrófilos demonstraram ter potencial metastático aumentado em pacientes com câncer de mama avançado14, onde as CTCs associadas a neutrófilos demonstram níveis mais altos de expressão da proteína marcadora de proliferação (Ki67) e de genes associados à progressão do ciclo celular. Estudos clínicos apoiaram os resultados dessas investigações biológicas, descobrindo que a presença de agrupamentos de CTC está associada a menor sobrevida livre de progressão e sobrevida geral do paciente15. O aumento do estudo de clusters CTC, então, oferece uma grande promessa para melhorar a compreensão e gestão da metástase.
Até o momento, o isolamento confiável e eficiente de clusters CTC viáveis tem sido limitado porque a sensibilidade e a especificidade das tecnologias de isolamento CTC são calibradas principalmente para detecção de célula única16,17,18,19. Técnicas de microfiltração, por exemplo, são amplamente empregadas como ensaios de CTC devido à sua operação rápida e direta16,19,20. No entanto, os aglomerados de CTC sob pressão fisiológica podem se reorganizar como estruturas semelhantes a cadeias de arquivo único e atravessar constrições tão pequenas quanto 5 μm11, o que sugere que os agregados provavelmente podem passar pelos poros do filtro, dadas as pressões muito mais altas empregadas na filtração. Além disso, forças de cisalhamento mais altas experimentadas durante a microfiltração podem danificar os aglomerados de CTC ou dissociá-los em células únicas21, prejudicando o enriquecimento eficiente. Por outro lado, os sistemas de enriquecimento baseados em anticorpos, que há muito são usados para isolamento de CTCs individuais17,22,23, podem detectar apenas uma subpopulação selecionada de grupos de CTC dentro da população heterogênea de CTC devido à sua dependência de antígenos de membrana específicos13 ,24. Além disso, a proporção menor de área de superfície para volume de clusters CTC afeta negativamente as eficiências de imunocaptura de tecnologias baseadas em anticorpos25, tornando-as particularmente ineficientes para o enriquecimento de clusters CTC. Mais promissores a esse respeito são os recentes chips microfluídicos direcionados especificamente para clusters CTC, que podem atingir sensibilidades relativamente mais altas. Infelizmente, eles o fazem em taxas de processamento clinicamente impraticáveis21,26 ou sob ameaça de danos aos aglomerados devido às altas taxas de fluxo em canais estreitos27, uma preocupação especialmente relevante para grandes aglomerados, que foram relatados como ocasionalmente compostos por até dezenas de células tumorais28.
4) shots of each cluster, ensuring none of the clusters were missed and (2) capture multiple different conformations within the field of view, increasing the accuracy of cluster size estimation. We validated (see "Methods": measurement of the device sensitivity) the optimized characterization setup by operating the 2-channel microfluidic interface in a loop without the device attached to confirm there were no cell loss in the system (Supplementary Table 1) and also by ensuring the cluster counts obtained using our setup matched with the direct counts of the captured clusters on the device (Supplementary Fig. 4)./p>500 mL/h, spiked cell clusters dissociated in the device, as evidenced by the mismatch observed between the size distributions of spiked and processed cell populations (Supplementary Fig. 5)./p>2× (87% vs 37%) the efficiency of the Cluster-Chip (Supplementary Fig. 8). Even when compared to the reported release efficiency of a Cluster-Chip operated at 4 °C to reduce non-specific cell adhesion to the microfluidic chip, the Cluster-Wells demonstrated greater efficiency (87% retrieval for the Cluster Wells vs 80% for the Cluster-Chip)./p>150 cells, with the latter found in a sample from an ovarian cancer patient (Fig. 4a, iii). Besides raising questions on the physiological circulation of CTC clusters, the surprisingly large size of the isolated ovarian cancer CTC cluster points to a potential drawback of microfluidic chips, even if they are designed with CTC clusters in mind: a CTC cluster of this size would have likely clogged narrow microfluidic channels or split into smaller pieces if forced through. Some of the isolated CTC clusters were also found to contain leukocytes in both ovarian and prostate cancer samples (Fig. 4a, ii and b, iii), an observation consistent with previous reports on breast cancer CTC clusters14. To confirm physical cohesion between cells in these leukocyte-associated CTC clusters, such clusters were purposely expelled from microwells and then recaptured./p>50 rpm) of CD44. It has been previously reported that CD44 homophilic interactions and subsequent CD44–PAK2 interactions mediate tumor cell aggregation42 and improve stemness, survival, and metastatic progression43,44. Similarly, all CTC clusters expressed high levels of TIMP1, JUN, and FOS, which are known to stimulate cell proliferation, inhibit apoptosis, and regulate angiogenesis45,46. Furthermore, we have performed overrepresentation enrichment analysis for the "TOMLINS_PROSTATE_CANCER_UP" gene set from the MSigDB database at the Broad Institute, which lists the genes that are upregulated in prostate cancer compared to benign tissue47. As observed from the heatmap plot, all CTC clusters as well as prostate cancer cell lines expressed high levels of these genes consistent with their prostate tumor origin. We have also performed a similar analysis for the "CHANDRAN_METASTASIS_TOP50_UP" gene set from the MSigDB database, which includes the genes upregulated in metastatic prostate cancer tumors compared to the primary tissue48. Upregulation of the genes associated with metastasis in all clusters is in agreement with the enhanced metastatic potential of CTC clusters9,10. Among the set, we observed the highest expression levels in HSPD1 and HSP90AA1 genes that are members of heat shock proteins (HSPs), playing important roles in cancer development and invasion, progression, metastasis, and drug resistance in various cancer types49,50. Also, all CTC clusters expressed high levels of G3BP1 which has been correlated with the malignant degree of the tumor51 and observed to be most abundant in castration-resistant prostate cancer (CRPC)52, which agrees with the clinical diagnosis of both Patient-1 and Patient-2 (Supplementary Table 2)./p>150 cells in the samples tested, two observations seemingly at odds with each other and which have potential implications for the circulation of CTC clusters in the body. The frequency of CTC clusters in the peripheral blood of ovarian cancer patients may present valuable opportunities for further study to better understand hematogenous dissemination of ovarian cancer55 as well as opportunities for early detection of disease onset and tumor recurrence—two major problem areas for ovarian cancer, since the disease is asymptomatic early in its progression/recurrence56./p>250 mL/h, respectively. Smaller diameter devices were mounted to a commercially available 13 mm diameter filter holder (GE Healthcare Life Sciences) with a size matching hollow PDMS layers, which limit the fluid flow to the region of interest with a certain diameter and prevent stagnant flow region formation inside the filter holder. Before introducing the sample, the experimental setup (Fig. 2b) was primed with pure ethanol, which was followed by a PBS wash. Then, the setup was incubated with 3% bovine serum albumin (BSA) for 1 h to minimize non-specific cell adhesion. The Hoechst dye stained cells (see separate section on cell culture and preparation) were spiked into whole blood and run through the experimental setup using a syringe pump (Harvard Apparatus Infuse/Withdraw PHD Ultra) at the withdrawal mode. The blood with spiked cell clusters traversed through input channel (50 μm in height, 500 μm in width) of 2-channel microfluidic interface, device/filter holder assembly, and lastly, output channel (50 μm in height, 500 μm in width) of the microfluidic interface. The input and output microfluidic channels were simultaneously video recorded at 50 frames per second in fluorescent (DAPI) channel using an inverted fluorescence microscope (Eclipse Ti, Nikon, Melville, NY) for tracking the cells entering and exiting the analytical version of the Cluster-Wells, respectively. Then, the recorded video was processed by a custom-built software (Visual Studio, 2017), and the count of clusters entering and leaving the device was obtained with respect to number of cells within each cluster, which is used for the calculation of the capture efficiency./p>24). Trimmed reads were mapped to the hg38 build of the human genome using STAR mapper (PMID: 23104886) and transcripts were quantified by mapping to the GenCODE.v24 annotation version of the human transcriptome. For these prostate CTC clusters, a median of 83.15 M reads were input to STAR mapper (range 53.7–129.79 M), and a median of 88.8% reads mapped uniquely to the human transcriptome (range 80.57–91.59%). A total of 16,412 transcripts were detected with at least 10 mapped reads in one sample and used for DESeq2 analysis in R (PMID: 25516281). A total of 12,149 mapped Ensembl genes ranked by DESeq2 log2foldchange were used as input for GSEA analysis (PMID: 17644558), using the WebGestalt tool (PMID: 28472511). Enriched gene sets were analyzed using the Cancer Hallmark 50 gene sets and the KEGG Pathway gene sets. t-SNE analysis was performed in R using the M3C package with seed = 123 and perplexity = 1. Total read counts were normalized to FPKM values using Cufflinks (PMID: 22383036). A total of 26,391 transcripts had an FPKM > 1 in at least one sample, and 10,885 transcripts had a median FPKM > 1. To be used in the heatmap plot, reads per million (RPM) count for the genes was generated. Lastly, the heatmap plot was generated using ClustVis online tool./p>3.0.CO;2-9" data-track-action="article reference" href="https://doi.org/10.1002%2F%28SICI%291521-4095%28199908%2911%3A11%3C946%3A%3AAID-ADMA946%3E3.0.CO%3B2-9" aria-label="Article reference 32" data-doi="10.1002/(SICI)1521-4095(199908)11:113.0.CO;2-9"Article CAS Google Scholar /p>