Synthesis and characterization of BSA@Cu2−xS NPs
BSA was determined as the template to synthesize BSA@Cu2−xS NPs due to its excellent biocompatibility and stability. The ratio between BSA and copper ions can be flexibly adjusted to obtain variable NPs sizes. To investigate the size differences for the biocompatibility and toxicity of BSA@Cu2−xS NPs, we synthesized large- and small-sized NPs, named LNPs and SNPs, respectively. The transmission electron microscopy (TEM) image in Fig. 1A depicts the amorphous morphology and the sizes of LNPs (17.8 nm) or SNPs (2.8 nm). The hydrodynamic sizes of LNPs and SNPs were 28.2 nm (PDI at 0.24) and 10.1 nm (PDI at 0.22) in water, respectively, with no significant size changes in other media (Fig. 1B). Zeta potential analysis shows values of − 34.9 mV and − 34.6 mV for LNPs and SNPs, indicating their similar surface properties with high stability (< − 30 mV). BSA@Cu2−xS NPs were stable in an aqueous solution for 7 consecutive days, with no significant change in size (Additional file 1: Fig. S1C). X-ray photoelectron spectroscopy (XPS) was utilized to explore the valence states of copper elements in BSA@Cu2−xS NPs, and the results revealed that Cu+ and Cu2+ were both available in LNPs and SNPs (Fig. 1C).
We further investigated the NIR absorbance and photothermal properties of BSA@Cu2−xS NPs and NPs of two sizes were compared. The UV–vis–NIR absorbance spectra depict that both NPs have strong absorption in the NIR range, especially in NIR-II (> 900 nm) (Fig. 1D). It was noted that LNPs had a stronger absorption in NIR-II than SNPs at the same copper concentration. The photothermal properties of BSA@Cu2−xS NPs were measured by photothermal heating curves and recorded by infrared thermography. The results show a concentration-dependent photothermal effect for both BSA@Cu2−xS NPs, although LNPs exhibit better photothermal characteristics than small-sized NPs (Fig. 1E and Additional file 1: Fig. S1A). The heating and cooling curves used for the calculation of photothermal conversion efficiency (PCE) are similar for LNPs (42.7%) and SNPs (44.9%) based upon an energy balance model previously described (Fig. 1F) [11, 29]. To investigate the photothermal stability of the BSA@Cu2−xS NPs, five consecutive “ON–OFF” cycles of LNPs or SNPs under a 1064 nm laser (1 W/cm2) were conducted. The temperature changes of all five cycles had no significant differences, suggesting the excellent stability of BSA@Cu2−xS NPs (Additional file 1: Fig. S1B). Other than photothermal properties, both BSA@Cu2−xS NPs showed a photoacoustic response (Additional file 1: Fig. S1D). In summary, the above results indicate that the photothermal heating ability and photoacoustic signal of LNPs are higher than those of SNPs, while PCEs are similar for both BSA@Cu2−xS NPs. It should be emphasized that the PCEs of our synthesized BSA@Cu2−xS NPs are higher than those of existing NIR-II agents .
Pharmacokinetics and biodistribution of LNPs and SNPs in SD rats
To compare the size differences for the pharmacokinetic profile of BSA@Cu2−xS NPs in SD rats, blood was collected, and the copper element was quantified by ICP-MS at different time points after a single intravenous (IV) dose via the tail vein. The blood distribution half-life (t1/2α) was 5.3 min and 13.8 min for LNPs and SNPs, respectively, while the blood terminal elimination half-life (t1/2β) of LNPs and SNPs was 124.12 h and 83.85 h, respectively, which suggests the rapid tissue distribution and delayed metabolism/elimination of LNPs compared with SNPs (Fig. 2A). All the other pharmacokinetic parameters are listed in detail in Additional file 1: Table S2, with the lower value of the area under the blood activity time curve, the higher volume of distribution, and the higher mean residence time of LNPs, confirming their quick accumulation and slow removal in the tissues compared to SNPs.
To further understand the ADME profile of BSA@Cu2−xS NPs, the biodistribution of LNPs and SNPs in major organs at different time points was investigated through copper quantification. The accumulated amount of LNPs and SNPs in the major organs was liver > spleen > kidney > lung > heart. In contrast, no copper content was detected in the heart (Fig. 2B). It should be noted that LNPs accumulated in the liver was significantly higher than that of SNPs, and LNPs remained in the liver with a large amount, while SNPs were mostly removed from the liver to the basal level 168 h (7 days) after the single IV dose. The urine and feces were collected consecutively during the 7 days after administration. It seems that a minority of SNPs (5.6% of the total amount administered) were excreted rapidly from urine, and most of the SNPs (59.34%) were cleared in the feces within 1 week, while only 1.25% and 38.7% of LNPs were removed by urine and feces, respectively (Fig. 2C). These results highly illustrated that BSA@Cu2−xS NPs were mainly excreted through the hepatobiliary pathway, with more removal for SNPs than LNPs.
Gross pathology findings for the subacute studies of LNPs and SNPs in SD rats
The dosage regimen for photothermal therapy of NIR agents currently available is usually designed by a single-dose administration in clinical practice . According to ICH M3 and S3 guidelines, the rats were IV administered BSA@Cu2−xS NPs daily for 14 consecutive days, aiming to delineate the general toxicity profile that can support the investigative new drug application (IND) of copper sulfide-based NPs. Our preliminary studies have shown that the maximum tolerated dose of BSA@Cu2−xS NPs was 8 mg/kg, showing fatigue, anorexia, hypoactivity and blood trails under the nostrils of the treated rats. As a result, we selected 2, 5 and 8 mg/kg as the low (LNPs and SNPs at 2 mg/kg, abbreviated as L2 and S2, respectively), medium (L5 and S5) and high doses (L8 and S8), respectively. The body weight (BW) was recorded during the subacute toxicity study, showing a relative decrease in BW in a dose-dependent manner (Fig. 3A). A marked decrease in BW within three days was found in the S8 group, but this stress response seemed to be tolerated 3 days later by BW recovery. In comparison, the BW in the L8 group remained suppressed during the entire repeated dosing period.
BSA@Cu2−xS NPs administration was discontinued on day 14 followed by an extra 28-day recovery period (L8-R or S8-R group, total 42 days of observation) to investigate the reversibility of the NPs. Interestingly, rat BW in the S8-R group rapidly increased after the cessation of NPs injection and fully recovered to the same weight as the control group at day 42, while the rat BW in the L8-R group only showed a mild increase when LNPs were discontinued (Fig. 3B).
The weights of the internal organs were also measured, and the organ coefficient data showed no relative weight increase in the liver, kidney, lung or heart, while the spleen weight significantly increased and could be recovered in the S8-R group but not in the L8-R group, again supporting the toxicity reversibility of small-sized BSA@Cu2−xS NPs (Additional file 1: Fig. S2). Since the liver was determined to be the main organ for NPs distribution, the livers were collected and observed for gross pathology findings. As shown in Fig. 3C, the liver color of the rats treated with LNPs (L8) and SNPs (S8) for 14 days changed from normal bright red to dark green (darker in the L8 group than in the S8 group), which was close to the color of the BSA@Cu2−xS NPs, implying a large amount of NPs accumulation in the liver. However, it is optimistic to find that the rat liver in S8-R group was recovered to a normal and bright red color after 28 days of the recovery period, highly suggesting the efficient clearance of SNPs and reversible health status of SNPs-treated liver, in comparison, the liver color of L8-R group remained dark, mostly because of the slow clearance of LNPs in the liver.
Hematology and blood chemistry studies for LNPs and SNPs
Intravenous blood was collected for hematological and biochemical tests from the rats treated daily with BSA@Cu2−xS NPs for 14 days. Other than the fluctuation of white blood cells from LNPs-treated rats showing statistically significant differences from controls, most hematological indicators showed no distinct changes, and they were all within the normal ranges for both NPs (Additional file 1: Fig. S3). In comparison, blood biochemical studies showed a differential profile for LNPs and SNPs that were highly dependent upon administration duration. Specifically, the serum levels of ALT, AST, TBA and LDH in the S8 group were increased, and the ALB level decreased significantly at day 1, suggesting the rapid stress response of rat liver when subjected to a high dose of SNPs. After 14 consecutive days of administration, the levels of ALT, AST, TBA and LDH all increased in a dose-dependent manner by both LNPs and SNPs, while the ALB level was decreased by LNPs only, suggesting the functional alteration of the liver uniquely by LNPs (Fig. 4A).
Toxicity reversibility is a highly recognized study that must be submitted to regulatory agencies to demonstrate that drug toxicity can be reversed and managed by the discontinuation of the drug. NPs administration was discontinued for surveillance of blood chemistry parameters at the end of the 28-day recovery period. The ALT level after 14 days of LNPs treatment at 8 mg/kg (70.8 U/L for L8) continued to increase to 143.9 U/L after recovery for 28 days (L8-R group), while the ALT value decreased from 119.0 to 81.3 U/L for small-sized NPs (Fig. 4B). With a similar trend, the AST level in the L8-R group remained 2.6-fold higher than that in the control group after the recovery period, while the AST value in the S8-R group decreased from a significantly high level (a 5.1-fold increase over the control for the S8 group) to almost the control value. The failure of recovery of total bile acid (TBA) also occurred with LNPs but not SNPs. These optimistic data strongly indicate that SNPs toxicity can be fully reversed by NPs discontinuation, while LNPs exhibit delayed toxicity even after the cessation of administration. It should be noted that neither LNP nor SNP would cause renal function injury by comparing the serum level of CREA and UA in control and dosing groups during the 14 consecutive days of administration (Additional file 1: Fig. S4).
Histopathological features of the liver by LNPs and SNPs
To further delineate the hepatotoxicity of large-sized (LNPs) and small-sized (SNPs) BSA@Cu2−xS NPs, the livers from the dosing and recovery groups were H&E stained and observed. In the control samples, the hepatocytes were radially arranged connecting the portal veins, while this cord shape became obscure by the low dose of LNPs (L2). At higher doses of LNPs (L5), several pathological alterations were evident, including hepatic sinusoid expansion, hepatocyte polarity disorder and focal lymphocyte infiltration (*). It is noted that the area for cell infiltration was enlarged in the L8 group, suggesting severe inflammation by the high dose of LNPs (Fig. 5A). LNPs at 5 and 8 mg/kg (L5 and L8) were found to be deposited in Kupffer cells, showing brown staining of the cells (arrows). LNPs may also deposit (arrowhead) in the cells of the inflammatory site, with the amounts of NPs highly correlating with the degree of lymphocyte infiltration.
In comparison, the pathological changes in the liver caused by SNPs showed a distinguished pattern. The low dose of SNPs (S2) introduced slight and focal hepatocyte swelling, while the hepatic cell cords remained organized. With the increase in the dose at 5 mg/kg (S5), cell infiltration sites with a small cluster of lymphocytes (*) were sporadically observed. Focal necrosis became evident (red arrow) in the highest dosing group (S8), along with the formation of fibrous connective tissue (#) (Fig. 5A). SNPs deposition in the liver was minimal in the low (S2) and medium (S5) dosing groups, while SNPs dispersively accumulated (arrowhead) in the fibrotic sites of the livers when rats were treated at a high dose (S8). Therefore, unlike LNPs, which promote a high degree of inflammatory infiltration and intra-Kupffer accumulation of NPs, SNPs are prone to hepatocyte necrosis and hepatic fibrosis.
The liver from rats treated with BSA@Cu2−xS NPs showed an optimistic recovery in SNPs compared with LNPs. Specifically, the lymphocyte infiltration area became further enlarged despite the discontinuation of LNPs, and LNPs (shown as brown deposition) remained accumulated within the inflammatory area during the 28-day recovery period (Fig. 5B). In comparison, it is interesting that hepatocyte necrosis and fibrosis caused by SNPs at 8 mg/kg were fully repaired at the end of the recovery period, showing the normal microstructure of the liver (Fig. 5B). In addition, no SNPs deposition was found in the histological sections of the S8-R group.
Based upon the above observation, dose-dependent liver injury was found in both LNPs and SNPs after 14 consecutive days of administration, and the histopathological profile of LNPs appeared to be more severe than that of SNPs. The damaged sites could have a full recovery in the livers of SNPs-treated rats, while the livers of LNPs-treated rats remained inflammatory with NPs retention. In comparison, the other major organs did not have distinct toxicological changes other than reversible spleen toxicity (Additional file 1: Fig. S5).
Uptake studies of liver and cells by LNPs and SNPs
Previous DMPK and toxicity studies demonstrated that hepatotoxicity may be associated with the retention of BSA@Cu2−xS NPs in the liver. Therefore, we quantified the copper contents of the livers from the dosing group (L8 and S8) and recovery group (L8-R and S8-R) by ICP-OES. The copper element was accumulated in the rat livers of the S8 groups at a high level of 760 μg/g after SNPs treatment for 14 consecutive days, while the copper content was significantly reduced to 128 μg/g (S8-R), namely, a 6.0-fold reduction at the end of the recovery period (Fig. 6A). In contrast, large-sized NPs were accumulated in the liver with a higher amount (960 μg/g) than SNPs after 2 weeks of the dosing period, and the remaining Cu element in the L8-R group after 28 days of recovery was 358 μg/g, which was 2.8-fold higher than that in the S8-R group.
The high uptake and slow clearance of copper elements by LNPs in the liver along with pathological findings imply the relative selectivity of LNPs by Kupffer cells, which are known as a specific type of macrophage responsible for the cellular uptake of exogenous substances/particles and hepatic inflammation [31, 32]. To further prove this hypothesis, primary rat hepatocytes and Kupffer cells were isolated and purified from rat liver and incubated with LNPs or SNPs for 6 h, aiming to investigate their uptake differences. After incubation of BSA@Cu2−xS NPs with rat hepatocytes, the intracellular Cu content in LNPs (0.76 μg/mg cells)-treated cell samples was comparable to that of cells treated with SNPs (0.56 μg/mg cells), as shown in Fig. 6B. To recapitulate the actual microstructure in the liver, primary rat hepatocytes were cultured into three-dimensional hepatocyte spheroids according to our in-house method. Approximately 1000 hepatocyte spheroids equivalent to 1 million cells were obtained for incubation with LNPs and SNPs at different time points followed by ICP-MS detection of Cu content. As shown in Fig. 6C, the uptake of BSA@Cu2−xS NPs by hepatocyte spheroids increased in a time-dependent manner, and the amount of SNPs uptake by hepatocyte spheroids was generally > two folds (2.8-fold at 24 h) that of LNPs, highly indicating the preferred SNPs uptake by hepatocytes. In contrast, Kupffer cells were prone to the uptake of larger-sized NPs, showing a much higher (3.1-fold) Cu content in LNPs (1.14 μg/mg cells) than SNPs (0.37 μg/mg cells) (Fig. 6D). Therefore, we herein suggest that hepatocytes in the liver may take up small-sized BSA@Cu2−xS NPs, potentially introducing rapid clearance, while Kupffer cells selectively take up large-sized NPs and retain them for a long period.
RNA sequencing analysis
To further illustrate the signaling pathways involved in the toxicity of BSA@Cu2−xS NPs and to compare the effects of sizes, differentially expressed genes (DEGs) were quantified by high-throughput RNA sequencing using the livers of the rats subjected to repeated dosing for 14 consecutive days. Pairwise comparison of the libraries between control (nontreated) and BSA@Cu2−xS NPs was performed after the calculation of gene expression abundance using the RSEM package. A total of 330 genes were upregulated and 178 genes were downregulated by LNPs at 8 mg/kg, while the number of upregulated or downregulated genes by SNPs was 464 and 283, respectively. Among these genes, 353 were shared by LNPs and SNPs, and SNPs showed unique features with more nonoverlapping genes and signaling pathways than LNPs (Fig. 7A). For more details on the expression levels of DEGs, please refer to Additional file 1: Table S3.
To further explore the signaling pathways changed by BSA@Cu2−xS NPs that were different in size, the Kyoto Encyclopedia of Genes and Genomes (KEGG) database was used to map and enrich these genes into bundles of pathways, and the top 20 KEGG pathways that significantly influenced LNPs and SNPs are listed (ranked by p-value). It seems that the inflammation modules (NF-kappa B signaling pathway and TNF signaling pathway, etc.) and lipid/drug metabolism-related pathways (such as the PPAR signaling pathway and cholesterol metabolism pathway, etc.) accounted for the major profile of BSA@Cu2−xS NPs-induced liver toxicity (Fig. 7B). It should also be noted that, unlike the SNPs, LNPs showed a significant alteration of phagosomal genes (total 18 genes), highlighting the unique behavior of active cellular uptake in the liver by LNPs that may be primarily due to the function of Kupffer cells.
Reversibility of toxicity responses at the molecular level by qPCR
Based upon the results from the RNA sequencing and KEGG pathway analysis, we selected the representative genes in the abovementioned four pathways highly relevant to liver toxicity and metabolism to investigate whether these pathways/genes can be recovered to the basal level after 28 days of cessation of NPs administration. Additionally, since copper metabolism may be highly involved in the toxicity of BSA@Cu2−xS NPs, we also grouped the genes known for the binding and transportation of copper elements to investigate their reversibility.
Inflammatory genes such as TNF-α and IL-6 in the rat livers were significantly upregulated by SNPs, while these two genes were fully recovered to the background level after the discontinuation of SNPs. However, these two inflammatory factors were continuously expressed after the recovery period of LNPs (L8-R), showing a comparable level to that of the treated samples (L8) (Fig. 8A). We also investigated the RNA level of IL-1β, suggesting the primary response of inflammation, but did not find an increase in this gene by either LNPs or SNPs (Additional file 1: Fig. S6A). Interestingly, IL-1β was significantly increased after the discontinuation of LNPs (L8-R), but it remained silent in the S8-R group, implying a delayed toxicity effect of LNPs mostly due to their accumulation in Kupffer cells. With a similar trend, the levels of FABP4 and LPL, two key genes in PPAR pathways related to cellular intermediary metabolism and inflammation, were highly expressed but slid back with a more attenuation effect by SNPs. Specifically, the FABP4 level was massively elevated (49- and 74-fold by LNPs and SNPs, respectively) but was reduced to a low level with SNPs samples closer to the background. The mRNA of LPL by LNPs remained unchanged after the recovery period, but LPL had an ~ 50% cut after a 28-day cessation of SNPs (Fig. 8B).
CYP7A1 (main pathway) and CYP7B1 (alternative pathway) are two enzymes that metabolize cholesterol into bile acids. FXR is a nuclear receptor that maintains bile acid homeostasis, and when bile acids increase, FXR is activated and negatively affects CYP7A1 through SHP to reduce the synthesis of bile acids . As expected, our results revealed that BSA@Cu2−xS NPs had an inhibitory effect on FXR after 14 consecutive days of dosing and introduced SHP downregulation, which in turn upregulated CYP7A1/CYP7B1 (Fig. 8C). Again, SNPs showed much better reversibility of CYP7A1/CYP7B1 by approaching the background level, while their expression remained after the cessation of LNPs (Fig. 8C). Alternatively, drug metabolism-related enzymes, including PXR, CYP3A2 and CYP2E1, were decreased by both BSA@Cu2−xS NPs, with a similar trend of recovery after discontinuation of NPs (Fig. 8D). We also investigated the dosing and recovery profiles of the major transporter genes by BSA@Cu2−xS NPs. The gene expression of the bile acid efflux transporter proteins BSEP and MRP2 located in the apical membrane did not change significantly, and the level of NTCP located in the basolateral membrane did not change (Additional file 1: Fig. S6B). In comparison, the gene expression of bile acid efflux transporter proteins MRP3 and MRP4 located in the basolateral membrane was upregulated, which could be major contributors to the increased blood level of bile acid, as shown in Fig. 4A. These two proteins recovered to the basal level with no significant difference for either NPs.
Copper metabolism- and transportation-related genes were selected to explore their unique role in BSA@Cu2−xS NPs toxicity. Metallothioneins (MTs) are small cysteine-rich proteins for metal binding that play a critical role in metal transport, storage and detoxification [34, 35]. The MT expression level was markedly increased 84-fold and 47-fold by LNPs and SNPs, respectively, in the rat liver at the end of the 14-day dosing period (Fig. 8E). However, the MT level in the L8-R group remained at a high level after the recovery period, implying copper accumulation from LNPs in the liver. In contrast, MT expression completely returned to a background level after the discontinuation of SNPs, indicating that trace levels of copper had a minimal adverse effect on the rat liver after the recovery period. Other than MTs, BSA@Cu2−xS NPs of both sizes did not affect the expression of the ceruloplasmin (CP) and ATP7B genes, which suggests that excessive copper elements in the NPs did not activate copper transporters, and the mechanisms can be delineated in our future work (Additional file 1: Fig. S6C).
In conclusion, our qPCR analysis of five groups of genes indicated that LNPs and SNPs had differential adverse effects on inflammation, lipid metabolism, cholesterol and bile acid metabolism, drug metabolism, and metal homeostasis. More importantly, small-sized BSA@Cu2−xS NPs showed a much better recovery or reversible toxicity than large-sized NPs. The liver can be restored to a normal level concerning lipid/drug metabolism/transport, inflammation homeostasis and other typical functions after the discontinuation of SNPs, while the liver failed to recover after the cessation of LNPs, causing prolonged and delayed liver damage that may be likely due to the accumulated and imbalanced level of copper in Kupffer cells in the liver.