Selectively Sensitizing Malignant Cells to Photothermal Therapy Using a CD44Targeting Heat Shock Protein 72 Depletion Nanosystem Shouju Wang,†,‡ Ying Tian,†,‡ Wei Tian,† Jing Sun,† Shuang Zhao,† Ying Liu,† Chunyan Wang,† Yuxia Tang,†,‡ Xingqun Ma,§ Zhaogang Teng,†,‡ and Guangming Lu*,†,‡ †
Department of Medical Imaging, Jinling Hospital, School of Medicine, Nanjing University, Nanjing 210002, P.R. China State Key Laboratory of Analytical Chemistry for Life Science, School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210093, P.R. China § PLA Cancer Center of Nanjing Bayi Hospital, Nanjing 210002, P.R. China ‡
S Supporting Information *
ABSTRACT: Selectively enhance the therapeutic efficacy to malignancy is one of the most important issues for photothermal therapy (PTT). However, most solid tumors, such as triple negative breast cancer (TNBC), do not have identifiable surface markers to distinguish themselves from normal cells, thus it is challenging to selectively identify and eliminate those malignances by PTT. In this report, we hypothesized that, by targeting CD44 (one TNBC-overexpressed surface molecule) and depleting heat shock protein 72 (HSP72, one malignancy-specific-overexpressed thermotolerance-related chaperone) subsequently, the TNBC could be selectively sensitized to PTT and improve the accuracy of treatment. To this end, a rationally designed nanosystem gold nanostar (GNS)/siRNA against HSP72 (siHSP72)/ hyaluronic acid (HA) was successfully constructed using a layer-by-layer method. Hydrodynamic diameter and zeta potential analysis demonstrated the formation of GNS/siHSP72/HA having a particle size of 73.2 ± 3.8 nm and a negative surface charge of −18.3 ± 1.6 mV. The CD44-targeting ability of GNS/siHSP72/HA was confirmed by the flow cytometer, confocal microscopic imaging, and competitive binding analysis. The HSP72 silencing efficacy of GNS/siHSP72/HA was ∼95% in complete culture medium. By targeting CD44 and depleting HSP72 sequentially, GNS/siHSP72/HA could selectively sensitize TNBC cells to hyperthermia and enhance the therapeutic efficacy to TNBC with minimal side effect both in vitro and in vivo. Other advantages of GNS/siHSP72/HA included easy synthesis, robust siRNA loading capacity, endosome/lysosome escaping ability, high photothermal conversion efficacy and superior hemo- and biocompatibility. KEYWORDS: photothermal therapy, triple negative breast cancer, gene therapy, heat shock protein, thermotolerance
P
malignant cells do not have identifiable surface markers to distinguish themselves from normal cells.15−17 In most cases, the receptors overexpressed by malignant cells, such as CD44, are also present in abundance on some normal cells, resulting in false positives and limited accuracy.18,19 Therefore, a rational design of nanosystem, which could target molecules like CD44 first and then exclusively increase the thermosensitivity of malignancy among those targeted cells, may benefit the precision of PTT through a two-step selection strategy.
hotothermal therapy (PTT) holds great promise in treating a wide range of malignant disease.1−3 Compared to conventional whole-body hyperthermia therapy, one major advantage of PTT is its ability to generate high temperature at desired sites with externally tunable control.4−8 Ideally, PTT is supposed to increase the temperature of tumor and eliminate malignant cells with minimal damage to adjacent normal cells/tissues.9−11 However, the precision of PTT is impaired in practice, which is to a large degree due to the nonefficient selective delivery of plasmonic nanoparticles to tumors. One common way to increase delivery selectivity is active targeting strategy, which decorates nanoparticles with ligands that can target tumor-specific cell-surface markers.12−14 However, in reality, such strategy can be difficult because most © 2016 American Chemical Society
Received: June 12, 2016 Accepted: August 30, 2016 Published: August 30, 2016 8578
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
www.acsnano.org
Article
ACS Nano
Scheme 1. (a) Schematic Representation for the Preparation of Layer-by-Layer Assembly of GNS/siHSP72/HA and (b) GNS/ siHSP72/HA First Target CD44 Positive Cells via Their Outermost HA Layer and Then Deplete the Overexpressed HSP72 in Malignant Cellsa
a By targeting CD44 and depleting HSP72 sequentially, these nanosystems should only selectively sensitize CD44+/HSP72+ TNBC cells to PTT, thus improve the therapeutic index of PTT.
Figure 1. (a, b) Hydrodynamic diameters and zeta potentials of GNS-PEI assembled at various concentration of PEI. (c, d) Hydrodynamic diameters and zeta potentials of GNS/siHSP72 assembled at various gold:siHSP72 mass ratios. The quantitative results represent as mean ± standard deviation (n = 3).
Heat shock protein 72 (HSP72) plays a key role in the cellular sensitivity to hyperthermia.20,21 HSP72 is a fundamental member of the highly conserved heat shock proteins 70 (HSP70) family of molecular chaperones.22 Unlike other
HSP70 members who are constitutively expressed at relatively high level, the expression of HSP72 is very low in most normal cells. However, in tumors from different origins, such as colon, liver, prostate, cervix, and breast, abundant HSP72 presents in 8579
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
ACS Nano
Figure 2. (a, b) Agarose gel electrophoresis of GNS/siHSP72 at various gold:siHSP72 mass ratios and corresponding quantitative analysis of retardation efficiency. (c, d) Western blotting assays showing HSP72 expression level in MDA-MB-231 cells transfected by GNS/siHSP72 at various gold:siHSP72 mass ratios and corresponding quantitative analysis. MDA-MB-231 cells transfected by Lipo2000 with scRNA and siHSP72 are used as the reference. The quantitative results represent as mean ± standard deviation (n = 3). The statistical difference is indicated by asterisks. N.S. means no significant difference.
cytoplasm to protect malignancies from apoptotic stimuli.23−25 Previous studies showed HSP72 inhibition could significantly enhance the tumor response to hyperthermia, indicating HSP72 is a promising target to regulate the sensitivity of malignant cells to hyperthermia-based therapy.26−28 However, the benefit of HSP72 depletion to PTT has been scarcely reported. In this study, we introduced a plasmonic-nanoparticle-based siRNA delivery system to selectively sensitize malignant cells to PTT. Triple negative breast cancer (TNBC) was chosen as a model due to its high mobility and lack of well-defined surface markers.29 The rationally designed nanosystem (GNS/ siHSP72/HA) consisted of three subunits of (1) gold nanostars (GNS), which generate heat upon near-infrared laser irradiation,8,30,31 (2) siRNA against HSP72 (siHSP72), which is carried on the surface of GNS though electric force, and (3) hyaluronic acid (HA), one ligand of CD44, functioning as a TNBC-targeting motif (Scheme 1a).6,32 Note that although CD44 is a promising marker for TNBC, it is also elevated expressed in some normal tissues, such as human breast epithelium. Thus, HA alone cannot distinguish TNBC from adjacent normal breast tissues thoroughly.18,19 Our results showed that GNS/siHSP72/HA could deliver siHSP72 into CD44 positive cells, selectively sensitize TNBC cells by downregulating overexpressed HSP72, and enhance the photothermal therapeutic efficacy to TNBC with minimal damage to normal cells (Scheme 1b). By targeting CD44 and depleting HSP72 subsequently, the therapeutic focus of PTT was successfully narrowed from all irradiated cells to CD44 positive cells to TNBC cells.
HSP72 (GNS/siHSP72). Lastly, HA with a molecular weight of 10 kDa was assembled as the outermost layer (GNS/siHSP72/ HA). The key challenge involved in the assembly of nanosystem with the layer-by-layer approach is to identify the appropriate concentration of polyelectrolytes such that the polyelectrolytes could be coated without interparticle bridging.34,35 To successfully construct GNS/siHSP72, the optimized concentration of PEI was determined to be 10 mg/mL, and the minimum gold:siHSP72 mass ratio was 10:1. Figure 1a,b revealed that when coating in the lower concentrations of polyelectrolytes, the GNS would irreversibly aggregate, as evidenced by the significantly increased hydrodynamic diameters. At appropriated concentration of polyelectrolytes, the diameters of nanoparticles grew slightly after each step of coating, increasing from 63.6 ± 2.1 (GNS-MUA) to 65.1 ± 2.3 (GNS-PEI) to 68.7 ± 2.3 (GNS/siHSP72). The zeta potentials of nanoparticles during coating were also recorded. It was observed that the zeta potentials of GNS-PEI reached a plateau when incubating with PEI at concentrations higher than 10 mg/mL, which is in line with the change of their hydrodynamic diameter (Figure 1b). The zeta potentials of GNS/siRNA also rise with increased gold:siHSP72 mass ratio. At mass ratios higher than 10:1, the obtained nanoparticles exhibited positive zeta potentials, which allowed coating with the negatively charged HA (Figure 1d). To evaluate the siRNA binding ability, GNS-PEI was incubated with siHSP72 at mass ratios from 10:1 to 80:1 and analyzed by agarose gel electrophoresis. Figure 2a,b demonstrated that the siRNA retardation efficacy rise with increased gold:siHSP72 mass ratio, which is similar to the result of zeta potential observation (Figure 1d). When the mass ratio was higher than 20:1, the retardation efficacy was determined to be >98%, indicating complete siRNA binding via electrostatic interactions. To optimize the gold:siHSP72 mass ratio, MDA-MB-231 cells (TNBC cell line) were transfected by GNS/siHSP72 at
RESULTS Bare GNS were prepared by a surfactant-free method.33 To facilitate the coating of subsequent layers, MUA was deposited on the surface of GNS via forming Au−S bonds. Afterward, GNS-MUA were first coated with positively charged PEI (GNS-PEI) and then with double stranded siRNA against 8580
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
ACS Nano
Figure 3. (a, b) Hydrodynamic diameters and zeta potentials of GNS/siHSP72/HA assembled at various concentration of HA. (c) TEM images of GNS-MUA, GNS-PEI, GNS/siHSP72, and GNS/siHSP72/HA. Scale bars are 50 nm. (d) Normalized UV−vis spectra of GNS-MUA, GNS-PEI, GNS/siHSP72, and GNS/siHSP72/HA. (e) The temperature change of GNS-MUA, GNS-PEI, GNS/siHSP72, GNS/siHSP72/HA solutions (50 mg/L Au) and ultrapure water after an 808 nm laser irradiation at 1 W·cm−2 for 3 min. The quantitative results represent as mean ± standard deviation (n = 3). N.S. means no significant difference. (f) Thermal images showing the temperature of GNS-MUA, GNSPEI, GNS/siHSP72, GNS/siHSP72/HA solutions (50 mg/L Au) and ultrapure water during laser irradiation.
mass ratios from 10:1 to 80:1 in complete culture medium. The gene silencing efficacy was assessed by Western blotting assays and quantified by ImageJ. Cells transfected by Lipo2000 with 100 nM scrambled control siRNA (scRNA) and siHSP72 were used as negative and positive controls. As shown in Figure 2c,d, GNS/siHSP72 at gold:siHSP72 mass ratios between 20:1 and 80:1 significantly downregulated the HSP72 expression. The most effective mass ratio was 40:1, at which the silencing efficacy of GNS/siHSP72 was similar to that of Lipo2000. It is noted that GNS/siHSP72 with gold:siHSP72 mass ratios higher than 40:1 exhibited lowered gene silencing efficacy, probably because the siRNA content of each GNS/siHSP72 reduced with the increased gold:siHSP72 mass ratio, and thus lowered their siRNA delivery efficacy. To selectively sensitize and kill TNBC cells without severe off-target effect and other adverse events, the nanosystem should have the ability to deliver siHSP72 into TNBC cells efficiently.36,37 Despite the high siRNA loading capacity of PEIcoated nanoparticles, their delivery selectivity is very limited. Therefore, GNS/siHSP72 (mass ratio set at 40:1) was coated with negatively charged HA to target CD44, one major adhesion protein overexpressed by TNBC cells.18 Based on the measurement of hydrodynamic diameters and zeta potentials, the optimal concentration of HA for layer-by-layer assembly was determined to be 20 mg/mL (Figure 3a,b). The hydrodynamic diameter and zeta potential of GNS/siHSP72/ HA at this concentration of HA were 73.2 ± 3.8 nm and −18.3 ± 1.6 mV. The transmission electronic microscopy (TEM)
images also demonstrated the majority of products remained as single multibranched particles during the assembly (Figure 3c). Furthermore, the obtained GNS/siHSP72/HA remained stable for up to 4 days in saline supplemented with or without 10% fetal bovine serum (FBS), guaranteeing its feasibility for subsequent in vitro and in vivo experiments (Figure S1). To exclude the possibility that HA coating may replace the bound siHSP72, GNS/siHSP72 was stirred with or without 20 mg/mL HA up to 24 h, and analyzed by agarose gel electrophoresis. As revealed in Figure S2, no released siHSP72 was detected after coating with HA, indicating that the coated HA would not deteriorate the siRNA loading capacity of GNS/siHSP72. The photophysical properties of GNS/siHSP72/HA were then explored. UV−vis spectra showed that the surface plasmon resonance (SPR) bands of GNS-MUA, GNS-PEI, GNS/ siHSP72, and GNS/siHSP72/HA peaked at 769, 773, 775, and 780 nm (Figure 3d). The slight red shift of SPR bands between 2 and 5 nm after each step of coating suggested that GNS were assembled with PEI, siHSP72 and HA without severe aggregation. The thermal images were captured to analyze the photothermal conversion ability of GNS/siHSP72/ HA. Upon 808 nm laser irradiation at 1 W·cm−2 for 3 min, the temperature of GNS-MUA, GNS-PEI, GNS/siHSP72, and GNS/siHSP72/HA solutions (50 mg/L Au) climbed rapidly with similar rates. In contrast, no significant change of temperature was observed in ultrapure water (Figures 3e,f and S3). Collectively, these results indicated that the photo8581
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
ACS Nano
Figure 4. (a) Western blotting assays showing the HSP72 expression level in MDA-MB-231 cells transfected by GNS-PEI (mock), GNS/ siHSP72, and GNS/siHSP72/HA with presence or absence of free HA. (b) Corresponding quantitative analysis of Western blotting assays. The quantitative results represent as mean ± standard deviation (n = 3). The statistic difference is indicated by asterisks. (c) FACS distribution showing the fluorescent intensity of FAM among cells incubated with GNS/siHSP72FAM/HA for 0, 3, and 6 h. (d) Dark-field microscopic images of HEK293 cells, MCF 10A, and MDA-MB-231 cells incubated with GNS/siHSP72/HA for 6 h. (e) Confocal fluorescence microscopy of HEK 293 cells, MCF 10A, and MDA-MB-231 cells incubated with GNS/siHSP72FAM/HA for 6 h. Cell nucleus is labeled by DAPI in blue; siHSP72 is labeled by FAM in green; endosome/lysosomes are labeled by Lysotracker in red. Scale bars are 15 μm.
HEK 293 cells after 6 h of incubation (Figure S4a). These observations corroborated the result of competitive inhibition test, suggesting the receptor-mediated endocytosis was involved in the CD44-targeting ability of GNS/siHSP72/HA. It is important to notice that the internalization of GNS/siHSP72/ HA had no remarkable difference between MDA-MB-231 and MCF 10A cells because CD44 is overexpressed by both cells regardless of their malignant or benign nature. This result demonstrated that although GNS/siHSP72/HA could differentiate TNBC cells from those CD44 negative normal cells by targeting CD44, it still needs other tactics to further distinguish TNBC cells from those CD44 positive normal cells. To further demonstrate the specific cellular recognition for CD44 positive cell lines, free HA was added to compete for the cellular uptake of GNS/siHSP72/HA in MCF 10A and MDAMB-231 cells. The ICP-AES measurement showed the intracellular gold concentration in both cells lines was remarkably decreased with the presence of free HA. In addition, the degree of cellular internalization was also dramatically inhibited by incubation at 4 °C (Figure S4b). These results confirmed that the internalization of GNS/ siHSP72/HA is mainly through an energy-dependent, receptormediated endocytosis pathway. The dark-field and confocal laser scanning microscope images show the intracellular distribution of GNS/siHSP72/ HA 6 h postincubation. In line with the FACS data, the spotty scatterings from branched GNS were bright and strong in MCF 10A and MDA-MB-231 cells, but indistinct in HEK 293 cells (Figure 4d). Similarly, extensive spotty green fluorescence appeared in MCF 10A and MDA-MB-231 cells, but was undetectable in HEK 293 cells (Figure 4e). Encouragingly, most of the green fluorescence of siHSP72FAM in MCF 10A and MDA-MB-231 cells was not colocalized with the red fluorescence of LysoTracker, but presented in the cytoplasm.
thermal conversion ability of GNS was maintained after the layer-by-layer assembling process. Once the siRNA loading capacity and photothermal conversion ability of GNS/siHSP72/HA were confirmed, a competitive inhibition test was carried out in the presence and absence of free HA to evaluate their CD44-targeting ability. Figure 4a,b showed that the presence of free HA significantly inhibited the gene silencing efficacy of GNS/siHSP72/HA from ∼95% to ∼15%. In contrast, the HSP72 expression in MDAMB-231 cells treated by GNS-PEI and GNS/siHSP72 showed no remarkable difference regardless of the presence or absence of free HA. This result implied that the HA layer granted GNS/ siHSP72/HA the ability to target CD44 and may help the nanosystem differentiate suspicious TNBC cells from those CD44 negative normal cells. To further illustrate the targeting ability of GNS/siHSP72/ HA and their intracellular fate, flow cytometry assays and confocal microscopy imaging were carried out to track the fluorescent signal from carboxyfluorescein tagged siHSP72 (siHSP72FAM). It is critical to investigate whether GNS/ siHSP72/HA could distinguish TNBC cells from normal cells by targeting CD44 solely, therefore, three cell lines were tested, including HEK 293, a CD44 negative normal human embryonic kidney cell line; MCF 10A, a CD44 positive normal human breast epithelial cell line; and MDA-MB-231, a CD44 positive TNBC cell line.18,32,38 The fluorescence-activated cell sorting (FACS) analysis demonstrated strong FAM fluorescence in CD44 positive MCF 10A and MDA-MB-231 cells within 3 h postincubation, but not in CD44 negative HEK 293 cells even 6 h postincubation (Figure 4c). Moreover, the inductively coupled plasma-atomic emission spectroscopy (ICP-AES) measurement showed that the intracellular gold concentration in MCF 10A and MDA-MB-231 cells was significantly higher than that in 8582
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
ACS Nano
Figure 5. (a−c) Viability assays showing the difference between viability of cells transfected by GNS/scRNA/HA and GNS/siHSP72/HA after a 5 min laser irradiation at various power density. The assays were performed in MD-MBA-231, HEK 293, and MCF 10A cells. (d) Viability assays showing the viability of HEK 293, MCF 10A, and MD-MBA-231 cells after incubation with various concentration of GNS/siHSP72/ HA. The quantitative results represent as mean ± standard deviation (n = 5). The statistic difference is indicated by asterisks.
cm−2. This therapeutic efficacy enhancement of GNS/ siHSP72/HA was not observed in MCF 10A and HEK 293 cells. Taking the Western blotting and viability assays results together, the most likely explanation could be summarized as following: GNS/siHSP72/HA target CD44 positive cells and release siHSP72 into the cytoplasm. To those normal cells, the released siHSP72 would bring no significant biological effects because the HSP72 expression is already at low level. Conversely, in TNBC cells with abundant HSP72, GNS/ siHSP72/HA could downregulate the expression of HSP72, sensitize TNBC cells to hyperthermia, and thus selectively enhance the photothermal therapeutic effect to TNBC during the subsequent laser irradiation. In this paradigm, GNS/ siHSP72/HA actually served as a “and-gate” nanosystem to identify TNBC cells as CD44/HSP72 double positive cells. Biocompatibility is another important issue for therapeutic nanoparticles. Therefore, HEK 293, MCF 10A, and MDA-MB231 cells were incubated with GNS/siHSP72/HA at various concentration for 48 h and followed by a viability assay. As shown in Figure 5d, cell death was not observed in HEK 293 and MCF 10A cells, indicating the superior biocompatibility of GNS/siHSP72/HA. Interestingly, modest toxicity of GNS/ siHSP72/HA to MDA-MB-231 cells was noticed. This growth inhibition effect of HSP72 downregulation was also reported by several previous studies on other malignant cells, 40−43 suggesting the elevated level of HSP72 is beneficial for malignant cells to adapt to apoptotic stress from their deregulated signaling pathways. To explore the potential applications of GNS/siHSP72/HA in vivo, the whole biodistribution of GNS/siHSP72 and GNS/ siHSP72/HA was evaluated by measuring gold concentration in
This observation suggested that siHSP72 delivered by GNS/ siHSP72/HA could escape from the endosome/lysosome into cytoplasm, which is a critical prerequisite to effectively silence mRNA. This endosome/lysosome escaping ability was probably due to the strong proton sponge effect of PEI.39 Since HSP72 is specifically overexpressed by malignant cells,40 we hypothesized that depleting overexpressed HSP72 could selectively sensitize malignant cells to hyperthermia, with little effect to normal cells since their expression of HSP72 is already low. In this way, the therapeutic focus of GNS/ siHSP72/HA-mediated PTT could be further narrowed from CD44 positive cells to TNBC cells. To test this hypothesis, we first determined the HSP72 expression level in HEK 293, MCF 10A, MDA-MB-231, and MCF 7 cells (another breast cancer cell line). The Western blotting assay demonstrated HSP72 level was specifically elevated in malignant cell lines, i.e., MDAMB-231 and MCF 7, but almost undetectable in benign cell lines, i.e., HEK 293 and MCF 10A (Figure S5). Then HEK 293, MCF 10A, and MDA-MB-231 cells were transfected by GNS/ siHSP72/HA, irradiated by an 808 nm laser at various power density for 5 min, and followed by a viability assay. Cells transfected by GNS/scRNA/HA were used as reference. The viability assay results are summarized in Figure 5. Powerdensity-dependent therapeutic efficacy was evident in CD44 positive MCF 10A and MDA-MB-231 cells. In contrast, unaffected viability was observed in CD44 negative HEK 293 cells. This result was consisted with the CD44-mediated cellular internalization of GNS/siHSP72/HA. More importantly, only in MDA-MB-231 cells, the viability of cells treated with GNS/ siHSP72/HA was significantly lower than that of GNS/scRNA/ HA after laser irradiation at power density from 0.75 to 1 W· 8583
DOI: 10.1021/acsnano.6b03874 ACS Nano 2016, 10, 8578−8590
Article
ACS Nano
Figure 6. (a) Tumor/muscle (T/M) ratios of gold concentration in MDA-MB-231/Luc bearing mice model after intravenous injection of GNS/siHSP72 and GNS/siHSP72/HA. (b) The temperature change of tumors during laser irradiation (5 min, 1 W·cm−2) in different groups of mice. The quantitative results represent as mean ± standard deviation (n = 3). (c) Thermal images showing the temperature of tumors from different groups at the end of laser irradiation.
tumors and major organs (including heart, liver, spleen, lung, kidney, and tumor-adjacent muscle tissues). Compared with GNS/siHSP72, GNS/siHSP72/HA showed improved in vivo pharmacokinetics with remarkably increased accumulation in tumors and decreased mononuclear phagocyte system uptake in liver, spleen, and lung (Figure S6). The tumor-to-muscle ratios of GNS/siHSP72/HA were 1.63, 2.29, and 2.15 for at 6, 24, and 48 h postinjection, which were significantly higher than those of GNS/siHSP72 group (Figure 6a). The remarkable difference between these two groups in biodistribution clearly demonstrated the targeting ability of GNS/siRNA/HA in vivo. To evaluate the therapeutic effect of GNS/siHSP72/HA in vivo, MDA-MB-231/Luc bearing mice were randomly divided into five groups and intravenously injected PBS, GNS/scRNA, GNS/siHSP72, GNS/scRNA/HA, and GNS/siHSP72/HA. Since it typically takes 24 h for the siRNA to knockdown the expression of protein, the time to perform PTT was chosen at 48 h postinjection to avoid potential deactivation of siHSP72 from heat. During treatment, the tumors were irradiated by an 808 nm laser for 5 min at 1 W·cm−2. Thermal images were taken to assess the tumor temperature during irradiation. Figure 6b,c demonstrated that tumors from GNS/scRNA and GNS/ siHSP72 groups had a temperature rise of
