Targeted hyperthermia therapy (THT) using gold nanorods remodels the tumor microenvironment to sensitize murine microsatellite-stable colorectal cancer to immune checkpoint blockade

Abstract

Microsatellite-stable (MSS) colorectal cancers are refractory to immune checkpoint blockade (ICB) due to limited baseline inflammation, inefficient antigen presentation, and impaired T cell priming. We investigated whether nanomaterial-enabled targeted hyperthermia therapy (THT), based on intratumoral gold nanorods (GNRs) that convert near-infrared (NIR) light into localized heat, could remodel this microenvironment and enhance responsiveness to ICB. CT26 colorectal tumors were established subcutaneously in BALB/c mice. Intratumorally delivered GNRs (Sona Nanotech Inc.^™) were activated by NIR light to induce hyperthermia (42–48 °C) for 5 min. Mice received one or two THT treatments and were subsequently treated with anti-PD-1 alone or in combination with anti-CTLA-4. Tumor progression and immune responses were evaluated for up to 49 days using flow cytometry, bulk RNA sequencing, qPCR, and with tumor growth and thermal analyses. THT achieved reproducible intratumoral heating and transiently slowed MSS CT26 tumor growth. Immune profiling demonstrated expansion of adaptive immune populations, accompanied by transcriptional induction of chemokine, complement, matrix-remodeling, stress-response, and antigen-processing pathways. This thermally conditioned immune landscape improved responsiveness to anti-PD-1 therapy, with THT combined with anti-PD-1 enhancing tumor control and prolonging survival. Reduced pulmonary gp70 levels in THT-plus-anti-PD-1 cohorts suggested diminished lung metastasis. A second THT treatment further strengthened adaptive immune responses, promoted focused TCRβ clonal remodeling, and improved tumor control, achieving complete regression in a subset of anti-PD-1-treated mice. Collectively, these findings demonstrate that GNR-enabled targeted hyperthermia can reprogram immunologically cold MSS colorectal tumors, enhance antigen visibility and T-cell activation, and restore sensitivity to ICB.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04310-8.

Background

With nearly 2 million new colorectal cancer (CRC) cases and over 900,000 new deaths worldwide in 2020, CRC remains the third most common malignancy, and the second leading cause of cancer mortality after lung cancer [1, 2]. By 2040, global CRC incidence and mortality are projected to nearly double, underscoring CRC as one of the fastest-growing contributors to the global cancer burden [2]. Early-stage CRC is often managed effectively with surgery and adjuvant chemoradiation, yielding 5-year survival rates of 70–90% [3]. However, a substantial proportion of patients ultimately relapse. For those who present with or progress to metastatic disease, outcomes remain poor, with 5-year survival rarely exceeding 14% [4, 5].

While immunotherapy has transformed the treatment landscape in several solid tumors, its impact in CRC has been largely confined to the minority of tumors with mismatch-repair deficiency (dMMR) or microsatellite instability (MSI). These tumors harbor elevated mutational burdens, abundant neoantigens, and pre-existing lymphocytic inflammation, accounting for their sensitivity to immune checkpoint blockade (ICB) [6]. In contrast, more than 85% of CRCs retain proficient mismatch repair (pMMR) and are microsatellite-stable (MSS), a phenotype associated with sparse T-cell infiltration, impaired antigen presentation, high stromal and myeloid content, and a strongly immunosuppressive tumor microenvironment (TME) [6, 7]. They also exhibit Wnt and β-catenin-driven immune exclusion, impaired IFN-γ signaling, and constitutive stromal TGF-β activity, all of which block dendritic cell recruitment and prevent productive T-cell priming [8–12]. As a result, these immunologically “cold” tumors respond poorly to blockade of programmed cell death 1 (PD-1), programmed cell death ligand 1 (PD-L1), and cytotoxic T-lymphocyte antigen 4 (CTLA-4) [13]. Collectively, these challenges underscore the need for strategies capable of remodeling the TME and restoring responsiveness to immunotherapy.

CT26 tumors closely mirror key features of immune-excluded MSS colorectal cancer and therefore provide a stringent, reproducible model for evaluating thermal immune reprogramming and checkpoint sensitization. In addition to retaining an MSS phenotype, CT26 cells are KRAS-driven, globally undifferentiated, and defined by Cdkn2a loss, Mapk1/Met amplification, and stem-associated transcriptional programs [14]. CT26 also lacks MHC class II and Ciita expression, contains abundant myeloid-derived suppressive cells and tumor-associated macrophages which limit CD4⁺ T-cell priming, and exhibits minimal baseline cytotoxic T-cell infiltration despite intact MHC-I [13–16]. Consistent with its immune-excluded state, CT26 is highly refractory to PD-1 and CTLA-4 blockade, making it a widely used model of immunotherapy-resistant CRC [17–20].

Targeted hyperthermia therapy (THT) has emerged as a promising approach to overcome these barriers. Local heating within a controlled therapeutic window (42–48 °C) can induce immunogenic cell death (ICD) and enhance antigen release. It can also upregulate MHC expression, activate dendritic cells, and promote effector T-cell priming [21–23]. THT integrates gold nanorods (GNRs) with near-infrared (NIR) light, enabling spatially confined, tumor-specific heating with minimal off-target injury. GNR-mediated hyperthermia increases damage-associated molecular patterns, activates heat shock proteins (HSPs), and triggers intrinsic apoptotic signaling, collectively promoting ICD [24]. It also promotes infiltration of cytotoxic lymphocytes and macrophages, enhances type I IFN signaling through cGAS-STING activation, and amplifies the efficacy of cytokine therapies and ICB [24–28].

Our recent preclinical study in melanoma and breast cancer demonstrates that combining THT with immunotherapies such as IL-2 cytokine treatment substantially improves tumor control relative to treatment alone [29]. These findings support the concept that thermal conditioning can convert poorly immunogenic tumors into more responsive ones. Based on this rationale, we hypothesized that THT would amplify antitumor immunity and improve therapeutic outcomes in the immune-excluded CT26 CRC model.

Here, we demonstrate that targeted, thermocouple-guided hyperthermia induces extensive local immune remodeling in the CT26 CRC model and markedly enhances the antitumor efficacy of PD-1 and CTLA-4 blockade. While photothermal therapy using gold nanomaterials has been widely explored, the optimal operational framework for repeated, temperature-controlled, sub-ablative hyperthermia designed to promote durable adaptive immune remodeling remains incompletely defined. We therefore directly compare single versus temporally spaced hyperthermia within a precisely controlled 42–48 °C intratumoral window and define how thermal dose and retreatment timing shape T-cell activation, TCR repertoire dynamics, and responsiveness to PD-1 blockade.

Methods

Animals

Female BALB/c mice (6 weeks old) were obtained from Charles River Laboratories (Montreal, Canada) and acclimated for one week at the Carleton Animal Care Facility (Dalhousie University, Halifax, Canada). Mice were housed in ventilated cages under a 12-h light/dark cycle at 22 °C and 55–60% humidity, with ad libitum access to standard chow and water. All procedures were approved by the Dalhousie University Committee on Laboratory Animals and adhered to the Canadian Council on Animal Care guidelines (Protocol #23–081).

CT26 colorectal cancer cell lines were obtained from ATCC and confirmed mycoplasma-free. They were maintained in RPMI-1640 containing 10% heat-inactivated FBS, 1% penicillin-streptomycin, and 2 mM L-glutamine, and were used within 10 passages.

CT26 tumor cells (5 × 10⁵ in 100 µL sterile PBS) were implanted subcutaneously into the right flank of isoflurane-anesthetized mice. The implantation site was shaved and disinfected immediately prior to injection. Tumor growth was monitored beginning on day 7 post-implantation. Length and width were measured using digital calipers, and height was assessed with a standardized ruler. Tumor volume was calculated at each timepoint using the ellipsoid formula: Volume = ½ × length × width × height.

Nanoparticle preparation

Gold nanorods used in this study were synthesized using a proprietary wet-chemical growth method as previously described [29]. Following synthesis, native surfactants were replaced with low-molecular- weight PEG to enhance stability and biocompatibility. Physicochemical characterization of this formulation has been previously reported [29] and is summarized here to define the material used in the present study. Transmission electron microscopy confirmed uniform rod morphology, with 79.9% rod-shaped particles (n = 319). Mean rod dimensions were 51.1 ± 10.8 nm (length) and 12.2 ± 1.9 nm (width), corresponding to an aspect ratio of 4.2 ± 0.9, comparable to commercial GNR formulations. Dynamic light scattering demonstrated a Z-average diameter of ~12.2 nm across PBS, 10 mM NaCl, and water, with polydispersity indices of 0.42–0.43. Intensity-based peaks at ~58–61 nm corresponded to rod length distribution, and volume-based analysis indicated > 99% of particles within the expected nanoscale range, consistent with minimal aggregation under physiological ionic conditions. Residual surfactants were quantified by RP-HPLC and were substantially below reported LD50 thresholds. Endotoxin levels were below the calculated limit of 4.66 EU/mg Au (USP-BET85 compliant).

THT

An 860-nm NIR diode laser (Sona Nanotech Inc.^™) capable of delivering up to 4,000 mA output was mounted at an adjustable height above the treatment platform. Laser height and beam diameter were calibrated using an irradiance lookup table to achieve a target irradiance of 1.0 ± 0.2 W/cm² at the tumor surface. For a representative 1.3 cm tumor, a laser height of 1.5 cm and current of ~1,800 mA produced ~0.82 W/cm².

Intratumoral temperature was continuously monitored using a calibrated type-T thermocouple inserted centrally within the tumor mass. A thermal imaging camera (HIKmicro) provided real-time skin surface temperature monitoring. A thermoelectric cooling system maintained laser diode stability at 25 °C prior to activation.

Mice received intratumoral GNRs (1 µg/mm³ tumor volume; typically, 50 µg in 25 µL PBS for a 50 mm³ tumor) immediately before irradiation. An approximate 5 mm layer of aloe vera gel was applied to the skin surface to reduce thermal injury.

Temperature feedback control

Hyperthermia was maintained within a 42–48 °C intratumoral window for 5 min using real-time temperature feedback and manual laser modulation. Laser output was activated via foot pedal and paused when tumor temperature approached 48 °C. Irradiation resumed once temperature declined to close to 43 °C. This on-off cycling maintained controlled sub-ablative hyperthermia while preventing overheating.

If tumor temperature fell below 42 °C during the exposure period, additional time was added to ensure a full 5 min within the target thermal window. Skin surface temperature was continuously monitored and maintained below 50 °C by reducing laser current, increasing cooling gel thickness, or both. Laser output was immediately paused if skin temperature approached 50 °C.

Control tumors (PBS-injected) underwent identical laser cycling parameters but did not achieve hyperthermic temperatures.

CEM43 thermal dose calculation

Thermal dose was quantified as cumulative equivalent minutes at 43 °C (CEM43) using continuous temperature recordings. Values were calculated using the Sapareto-Dewey equation [30], CEM43 = Σ (i = 1 to n) Δt_i · R^(43 − T_i), where T_i is the recorded temperature during interval Δt_i. Consistent with convention, R = 0.5 for T_i ≥ 43 °C and R = 0.25 for T_i < 43 °C. Temperature traces were processed in R software using custom scripts to calculate time in window, mean temperature, mean temperature in window, time to 42 °C, and CEM43.

In vitro THT

CT26 cells were seeded at 4 × 10⁴ cells per well in 24-well plates and allowed to adhere overnight. The following day, cells were incubated with GNRs (25 µL added to 500 µL complete RPMI media (10% FBS)) and exposed to 860 nm NIR irradiation for 5 min. Temperature was monitored using the HIKmicro thermal imaging camera to maintain hyperthermic conditions. Control cells received identical laser exposure without GNRs.

Twenty-four hours post-treatment, cells were harvested and stained for viability dye, Annexin V, surface calreticulin, and surface HSP70 using flow cytometry (as described below). Data were analyzed as the percentage of positive live cells.

Treatment groups and dosing

Following THT on day 0, tumor size was balanced between treatment groups and dosed with immunotherapy beginning on day 1. Anti-PD-1 (200 µg; Bio X Cell) was administered intraperitoneally (i.p.) or intratumorally (i.t.) on days 1, 5, 9, 13, and 17. For CTLA-4/PD-1 combination therapy, mice received i.t. injections of 200 µg anti-CTLA-4 (clone 9H10) on days 1 and 5, followed by anti-PD-1 on days 9, 13, and 17. All injections were performed under brief isoflurane anesthesia. Data shown represent pooled results from four independent in vivo experiments performed at different time points using identical tumor implantation procedures, treatment protocols, and outcome assessments. Group sizes were not intentionally imbalanced and varied due to the sequential inclusion of additional treatment arms, technical constraints related to tissue availability for downstream analyses, and ethical considerations aimed at minimizing animal use.

Statistical analyses were conducted using tests that do not assume equal group sizes, with exact n values reported in all figure legends. To assess immune changes at 48 h post-THT, we performed an independent experiment containing 6 control (no THT) and 9 THT-treated mice with no additional treatments.

Flow cytometry

All animals were euthanized upon reaching predefined ethical endpoints. For downstream immune profiling, tumors were subsequently selected based on the time at which endpoint was reached, with only tumors collected between days 8 and 15 following THT included in flow cytometry analyses. This post hoc selection criterion was applied uniformly across all treatment groups to reduce variability related to post-treatment timing and tumor progression stage. Because THT delayed tumor progression in a subset of animals, fewer THT-treated tumors reached endpoint within this defined window, resulting in a smaller sample size for immune profiling despite adequate initial cohort sizes.

Tumors were enzymatically dissociated using the Miltenyi Biotec mouse tumor dissociation kit and gentleMACS^™ Dissociator. Single-cell suspensions were stained in BD Horizon^™ Brilliant Stain Buffer using the fluorophore-conjugated antibodies listed in Table 1. Data were acquired on a BD FACSCelesta^™ and analyzed with FlowJo^™ v10. Gating strategies are shown in Supplementary Figure S3.

Table 1.

Flow cytometry panel

Panel	Antigen	Fluorophore	Company
Lymphoid	CD45	PE-Cy7	BD
Lymphoid	CD3	APC	BioLegend
Lymphoid	CD4	APC-Cy7	BD
Lymphoid	CD8	FITC	BioLegend
Lymphoid	CD25	BV650	BD
Lymphoid	FOXP3	PE	BD
Lymphoid	CD27	BV421	BD
Lymphoid	CTLA-4	AF700	BD
Lymphoid	CD44	BV510	BD
Lymphoid	CD62L	BV786	BD
Lymphoid	NK1.1	PE-Cy5.5	BD
Lymphoid	PD-1	BV605	BD
Myeloid/ICD	CD45	PE-Cy7	BD
Myeloid/ICD	F4/80	AF700	BD
Myeloid/ICD	CD11b	BV786	BD
Myeloid/ICD	CD206	BV650	BD
Myeloid/ICD	CD86	APC-Cy7	BD
Myeloid/ICD	MHC II Ia/IE	BV421	BD
Myeloid/ICD	PD-L1(CD274)	PE	BD
Myeloid/ICD	MHC I	PerCP-Cy5.5	BD
Myeloid/ICD	Annexin	BV510	BD
Myeloid/ICD	Calreticulin	APC	Abcam
Myeloid/ICD	Hsp70	FITC	Abcam
Myeloid/ICD	Fixable Viability Dye	BV605	BD

RNA sequencing and analysis

Tumors used for RNA sequencing were selected from the same cohort and using the same inclusion criteria described for flow cytometry analyses, with only tumors collected between days 8 and 15 following THT included. These samples represent matched tumors from animals euthanized at predefined ethical endpoints.

Tumors were excised and snap-frozen in liquid nitrogen, and total RNA was isolated using the PureLink^™ RNA Mini Kit (Invitrogen). RNA integrity and concentration were assessed by a NanoDrop^™ (Thermo Fisher Scientific) and Agilent Bioanalyzer. Libraries were generated and sequenced at The Centre for Applied Genomics (TCAG, SickKids) on an Illumina NovaSeq 6000 platform. Raw reads were supplied in ORA-compressed FASTQ format and decompressed using DRAGEN ORA Decompression v2.7 [31].

Read quality was assessed with FASTQC, and adapters and low-quality bases were removed using Trim Galore!. Trimmed reads were aligned to the mouse reference genome (GRCm39) using STAR [32, 33], and sorted BAM files were generated with SAMtools [34]. Gene-level counts were obtained with featureCounts [35]. Differential expression, clustering, and pathway enrichment analyses were performed using iDEP, Enrichr, and custom R scripts (R v4.5.2) [36, 37].

TCRβ repertoire analysis was performed on bulk RNA sequencing data using MiXCR with default murine TCRβ settings [38]. Reads were aligned to the mouse TCRβ germline reference, and clonotypes were defined by a unique CDR3 amino acid sequence with assigned TRBV and TRBJ genes. Clonotype abundances were normalized to total TCRβ reads per sample and expressed as read fractions. Downstream analyses, including Shannon diversity, clonal dominance, V-gene usage, inter-sample repertoire overlap, and identification of clonotypes amplified after single and double THT, were performed using custom R scripts.

Quantitative PCR for lung RNA

Lung RNA was reverse transcribed to complementary DNA (cDNA) using the iScript^™ cDNA Synthesis Kit (Bio-Rad) according to the manufacturer’s instructions. Primers for GAPDH (forward: 5’ ACGGATTTGGTCGTATTGGG 3’, reverse: 5’ CGCTCCTGGAAGATGGTGAT 3’) and gp70 (forward: 5’ AAAGTGACACATGCCCACAA 3’ and reverse: 5’ CCCCAAGAGGCACAATAGAA 3’) were used. qPCR reactions were prepared using SYBR^™ Green PCR Master Mix (Invitrogen) according to the manufacturer’s protocol. Relative gene expression was normalized to GAPDH and quantified using the 2^−ΔΔCt method [39].

Statistics

Statistics on tumor volume and weight, temperature calculations, flow cytometry, survival, and qPCR were performed using GraphPad Prism 10.4.1 (GraphPad Software). For comparisons between two groups, a nonparametric Mann-Whitney U test was used. For experiments involving three or more groups, one-way ANOVA was performed followed by Tukey’s post-hoc multiple comparisons test. A p-value < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation (SD). Exact n values for each experiment are reported in the corresponding figure legends.

Results

Single treatment with targeted hyperthermia enables precise intratumoral heating, reducing tumor growth

CT26 tumors (average tumor size of 55.78 ± 51.9 mm³) received i.t. injections of PBS or GNRs (50 µg), followed by NIR treatment (Fig. 1A). NIR exposure using an on/off pulsed laser cycle reliably maintained intratumoral temperatures in the hyperthermic therapeutic window of 42–48 °C for 5 min (299.4 ± 29.6 s) in the GNR-bearing tumors (Fig. 1B; Supplementary Fig. S1A-B). Thermal monitoring prompted brief pauses in laser output as the surface temperature approached 50 °C, minimizing skin injury while preserving stable intratumoral heating.

Fig. 1.

Targeted hyperthermia intratumoral heating and impact on tumor size. A, Experimental setup for tumor establishment, treatment, and THT induction. B, Mean internal tumor temperatures of control and THT treatment groups during treatment (control n = 35, THT n = 58). C, Tumor volumes over time of mice treated with control (n = 14) and a single THT treatment (n = 12). D, Endpoint weights of control and THT tumors. Frequency of apoptosis V exposure (E), calreticulin exposure (F), and HSP70 exposure (G) in live CT26 cells 24 h post THT. Frequency of calreticulin exposure (H) and CD206⁺ of Ly6G- F4/80⁺ (I) in tumors 48 h post THT. (mean ± SD; *p < 0.05, **p < 0.01)

GNR-treated tumors displayed highly reproducible heating profiles, reaching 42 °C within 71.27 ± 52.95 s and maintaining a tightly regulated window temperature of 45.17 ± 0.84 °C (Fig. 1B; Supplementary Fig. S1C-D). In contrast, PBS-treated tumors exhibited sub-therapeutic heating, with a mean temperature of 30.76 ± 3.40 °C and no entry into the targeted range (Fig. 1B; Supplementary Fig. S1A and S1E). Thermal dose measurements (CEM43) confirmed substantially elevated hyperthermic exposure in the THT group (Supplementary Fig. S1F). These results demonstrate that THT mediated through NIR activation of GNRs provides tightly controlled, tumor-restricted heating in vivo.

During the first four days after heating, both groups displayed similar growth kinetics. However, by day 4, THT-treated tumors were significantly smaller than controls (p = 0.028) (Fig. 1C). THT tumors eventually resumed growth; yet, endpoint tumor weights were significantly lower in GNR-treated mice than PBS-treated controls (p = 0.0128) (Fig. 1D).

To determine whether THT directly induces tumor cell stress pathways, CT26 cells were exposed in vitro to GNR-mediated hyperthermia under conditions matched to the in vivo thermal window. Compared with control cells, THT significantly increased Annexin V positivity at 24 h, indicating induction of apoptotic cell death (Fig. 1E). Surface exposure of calreticulin, a hallmark of ICD, was also significantly increased following THT (Fig. 1F). A modest but non-significant increase in surface HSP70 was observed (Fig. 1G).

Together, these findings demonstrate that controlled sub-ablative THT directly induces tumor cell-intrinsic stress responses associated with ICD, providing a mechanistic basis for subsequent immune remodeling observed in vivo.

Analysis of tumors 48 h after treatment revealed sustained immunogenic effects in vivo. Calreticulin exposure on live tumor cells was significantly increased in THT-treated tumors relative to controls (Fig. 1H), indicating persistence of pro-phagocytic stress signaling following localized hyperthermia. THT also increased the proportion of CD206⁺ macrophages (CD206⁺ Ly6G⁻ F4/80⁺) at 48 h, demonstrating a strong trend toward significance (p = 0.0879) (Fig. 1I), consistent with early macrophage remodeling in response to thermal injury and damage-associated signaling. Notably, CD11c⁺ myeloid populations and dendritic cell frequencies were not significantly altered at this time point (Supplementary Fig. S1G-H), and neutrophil abundance remained unchanged between groups (Supplementary Fig. S1I). These findings suggest that early immune effects of THT primarily reflect phenotypic reprogramming of resident myeloid cells rather than large-scale immune cell recruitment.

Two THT treatments maintain heating precision and produce more durable tumor control

To determine whether an additional THT treatment enhances antitumor activity, mice received a second i.t. GNR injection and NIR light exposure, 5 days after the first treatment. This timepoint was chosen based on the initiation of growth in our single-treatment experiment. Both heating treatments rapidly achieved and sustained the therapeutic temperature range, with overlapping heating curves and comparable temperature metrics (Fig. 2A; Supplementary Fig. S2A-C).

Fig. 2.

Effect of two sequential THT treatments on tumor progression. A, Temperature profile over time of THT (1x) (n = 58) or THT (2x) (n = 28). B, Tumor volume over time of mice receiving THT (1x) (n = 12) and those receiving THT (2x) (n = 9). Endpoint tumor weights of PBS control (n = 14), THT (1x) (n = 12), and THT (2x) (n = 9) treated tumors. (mean ± SD)

Thermal dose analysis showed that both the first and second THT generated markedly higher CEM43 values than controls, although the second THT yielded a significantly lower CEM43 (p < 0.01) compared to the first (control = 0.0068 ± 0.019 °C, THT (1x) = 45.19 ± 44.86 °C, THT (2x) = 20.02 ± 12.36 °C; Supplementary Fig. S2D). Repeated thermal exposure produced a more sustained antitumor effect, as reflected by prolonged suppression of tumor volume following the second THT treatment (Fig. 2B).

A second targeted hyperthermia treatment preferentially enhances intratumoral T cell activation

Indicators of an innate response, including myeloid population analysis, demonstrated that the innate phase of THT-related immunity was typically complete by day 8. Macrophage frequencies, including M1- and M2-like subsets, were not markedly altered by either one or two THT treatments (Fig. 3A-C, Supplementary Fig. 3A-B). Similarly, a modest, nonsignificant increase in apoptotic cells observed after THT peaked within 24–48 h and largely resolved by day 8 (Fig. 3D). This is consistent with the known kinetics of moderate thermal injury [29]. Accordingly, to standardize our analysis of adaptive immune activation, tumors were harvested between days 8 and 15 following THT.

Fig. 3.

Double THT and intratumoral immune cell infiltration. A-K, Cell populations acquired using flow cytometry from control (n = 8), THT (1x) (n = 6), and THT (2x) (n = 5) tumor tissue samples. Frequency of parent cell is shown for (A) Annexin V+ live cells (a) M1 macrophages (b) M2 macrophages (c) PD-L1⁺ M2 macrophages (d) activated CD4⁺ T cells (e) CD4⁺central memory cells (f) PD-1⁺ CD4⁺ T cells (g) T regulatory cells (h) activated CD8⁺ T cells (i) CD8⁺central memory cells (j) PD-1⁺ CD8⁺ T cells (k). Shannon diversity index (L) and clonal dominance (M), Representative TCRβ clonotypes exhibiting stepwise amplification following single THT and further expansion after double THT (N-Q). (mean ± SD; *p < 0.05, **p < 0.001)

Activated CD4⁺ T cells were unchanged after a single treatment but increased significantly after two treatments (p = 0.0383 vs. control, p = 0.0283 vs. THT (1x); Fig. 3E). Central memory-like CD4⁺ T cells (CD44⁺CD62L⁺) were enriched after single THT (p = 0.0039) and double THT (p = 0.0140) treatments (Fig. 3F), and PD-1⁺ CD4⁺ T cells increased significantly after two THT treatments (p = 0.0103 vs. THT (1x); Fig. 3G). Regulatory T cells (Tregs; CD4⁺CD25⁺FoxP3⁺) were also significantly increased after two treatments (p = 0.0273 vs. control, p = 0.0047 vs. THT (1x); Fig. 3H).

Within the CD8 compartment, two THT treatments trended toward increased frequencies of activated and central memory-like CD8⁺ T cells relative to controls, but these differences did not reach statistical significance (Fig. 3I-J). PD-1 expression on CD8⁺ T cells was unchanged by THT (Fig. 3K).

To determine whether repeated THT treatments altered T cell clonality, we next assessed TCRβ repertoire structure. Two THT treatments resulted in a modest reduction in TCR repertoire diversity, as measured by the Shannon diversity index (Fig. 3L), accompanied by increased clonal dominance (Fig. 3M), consistent with progressive focusing of the adaptive immune response. Repeated THT also promoted convergent TRBV gene usage across tumors (Fig. 3N), suggesting non-random selection of antigen-responsive T cell populations. At the clonotype level, several TCRβ sequences exhibited stepwise amplification following single THT and further expansion after double THT (Fig. 3O-P), supporting antigen-driven clonal selection rather than nonspecific lymphocyte recruitment.

Collectively, these findings demonstrate that two sequential THT treatments elicit a stronger and more durable adaptive immune response than a single exposure, characterized by enhanced CD4⁺ T cell activation, memory differentiation, and progressive TCR repertoire focusing.

Targeted hyperthermia induces coordinated transcriptional remodeling enriched for inflammatory, metabolic, and immunoregulatory programs

Unsupervised k-means clustering identified four major expression patterns distinguishing control from THT-treated tumors (Fig. 4A). Genes that increased after THT were enriched for sensory, stromal, and neuromodulatory pathways (Cluster 1), consistent with heat-induced disruption of tumor stroma (Fig. 4B). Cluster 2 showed a subtle increase after a single treatment and a slight reduction after two treatments, with enrichment for antigen processing, hematopoietic activation, and NK-cell cytotoxicity (Fig. 4B). Cluster 3 contained genes that were unchanged after a single THT exposure but incrementally upregulated after two treatments, including metabolic and proteolytic remodeling programs such as complement activation and matrix turnover, consistent with cellular injury and tissue restructuring (Fig. 4B). Cluster 4 represented core inflammatory responses, including cytokine-cytokine receptor interactions and chemokine signaling, which were detectable after one treatment and further enhanced after two treatments (Fig. 4B).

Fig. 4.

Transcriptional profiles and pathway enrichment. A, Heatmap of samples displaying top 2000 differentially expressed genes organized into four clusters by K-mean clustering: control (n = 6) and THT (1x) (n = 6) or THT (2x) (n = 5) treated tumors. B, KEGG pathway enrichment analysis of genes from the four clusters identified in A. C, Principal component analysis demonstrating variance between control and THT-treated tumors. D, Differential expression analysis showing gene activity. E, Comparison of gene transcript expression between control and THT (1x) or THT (2x)-treated tumors. (mean ± SD)

Principal component analysis showed clear separation of control and THT-treated tumors (Fig. 4C). Differential gene expression analysis confirmed induction of inflammatory and injury-associated transcripts after one and two THT treatments (Fig. 4D-E). Chemotactic pathways were upregulated, including Cxcl3 and Ccr9, consistent with enhanced recruitment of innate and adaptive immune cells (Fig. 4F).

Several immune-related transcripts, including genes associated with innate danger signaling, antigen presentation, and lymphoid remodeling, followed a graded pattern with the highest expression in tumors receiving two THT treatments (Fig. 4F). Notably, induction of antigen presentation related transcripts such as H2-M5 and Tarm1 is consistent with prior work demonstrating that thermal stress can partially restore antigen visibility in tumor models with limited baseline antigen presentation capacity [14]. These transcriptional changes are supported by protein level and cellular immune phenotyping by flow cytometry shown in Fig. 3. In contrast, pathways related to complement activation, cellular stress responses, and extracellular matrix remodeling are currently supported primarily at the transcriptional level and will require further targeted proteomic analyses to define their functional contribution and their relationship to single versus repeated hyperthermia treatments.

Targeted hyperthermia enhances the efficacy of immune checkpoint therapies

To evaluate whether hyperthermia enhances responsiveness to standard immunotherapy, THT was combined with anti-PD-1 or anti-CTLA-4, and tumor progression and survival were monitored (Fig. 5). Anti-PD-1 monotherapy was ineffective in preventing tumor growth (Fig. 5A-B). Combining THT with systemic anti-PD-1 significantly improved tumor suppression, with differences emerging as early as day 6 (day 6: p = 0.0453 vs. THT (1x); day 7: p = 0.0007 vs. control, p = 0.0146 vs. PD-1 i.p.; Fig. 5A). Adding a second THT treatment to systemic anti-PD-1 significantly prolonged the overall effect (Fig. 5B). Notably, in the group receiving two THT treatments plus systemic anti-PD-1, 4 of 9 tumors fully regressed (Fig. 5B). This effect was similar when anti-PD-1 was delivered intratumorally in comparison to systemically (Fig. 5C). Complete tumor regression was observed in 5 of 17 intratumorally treated mice.

Fig. 5.

Efficacy of checkpoint blockade on tumor volume and survival. A, Tumor volume of control (n = 14) vs. THT (1x) treatment with i.p. injected anti-PD-1 (THT (1x) PD1ip; n = 12) or without (THT (1x); n = 12; PD1ip alone n = 12). B, Tumor volume of control vs. THT (2x) with PD1ip treatment (THT (2x) PD1ip; n = 9) or without (THT (2x); n = 9). C, Tumor volume of control vs. THT (2x), THT (2x) PD1ip, PD1ip, with i.t. anti-PD-1 (THT (2x) PD1it; n = 17) or without (PD1it; n = 11). D, Probability of survival for controls and all treatments. *p < 0.05 vs. control, #p < 0.05 vs. THT (1x), ^p < 0.05 vs. PD-1 (i.p. or i.t.), †p < 0.05 vs. THT (2x)

Interestingly, substituting anti-CTLA-4 for PD-1 blockade in the first two doses of the combination therapy as a sequential strategy was significantly less effective than using PD-1 blockade only in all five doses. Although overall tumor trajectories resembled those of THT (2x) alone, 1 of 7 tumors achieved complete regression (Supplementary Fig. S4A).

Across monotherapy groups survival patterns were similar, with median survival ranging from approximately 10 to 15 days and few animals surviving beyond day 20 (Fig. 5D). In contrast, combining THT with immunotherapy extended survival across all groups. The greatest effect was observed in mice receiving two THT treatments plus systemic anti-PD-1, where several tumors fully regressed and 4 out of 9 animals survived to the study endpoint on day 48 (Fig. 5D).

To assess impacts on early metastatic spread, expression of the CT26 retroviral antigen gp70 was quantified in lungs. The gp70 levels were significantly reduced in mice treated with THT plus anti-PD-1 compared to control, suggesting diminished metastasis (p = 0.0479; Supplementary Fig. S4B).

Discussion

Our study provides strong evidence that targeted hyperthermia can overcome immune exclusion in a genomically complex, immune-cold MSS colorectal cancer model. We began by confirming that THT can be applied with precise spatial and thermal control in CT26 tumors. Using GNRs and NIR light, we achieved reproducible heating curves that consistently maintained the target 42–48 °C intratumoral window with minimal detectable off-tumor injury. This level of precision mirrors our previous observations in melanoma and breast cancer models and establishes that colorectal tumors can be thermally conditioned with comparable reliability [29].

At these sub-ablative temperatures, hyperthermia perturbs tumor cell structure and local physiology in ways that promote immune engagement. Comparison of single versus double hyperthermia exposures revealed important physiologic and immunologic adaptations to heating. Both treatments achieved precise intratumoral temperatures, yet the second exposure delivered a significantly lower CEM43, indicating a reduced capacity of the tumor to accumulate heat. This reduction aligns with heat-induced thermotolerance and treatment-associated perfusion changes established by the first exposure [40].

Consistent with direct tumor-cell intrinsic effects of controlled hyperthermia, single THT exposure induced hallmarks of immunogenic stress, including increased Annexin V positivity and surface calreticulin exposure in vitro, with sustained calreticulin upregulation observed in vivo at 48 h after treatment. These findings indicate that sub-ablative thermal exposure initiates ICD-associated signaling features, in addition to any vascular or metabolic perturbations that may accompany heating. This early immunogenic stress response provides a mechanistic link between localized thermal injury and the subsequent immune remodeling observed following repeated THT, supporting a temporal progression from early tumor stress and myeloid reprogramming to later adaptive immune activation.

Building on these early tumor-intrinsic stress signals, prior work has shown that the physiologic consequences of heating evolve over time and strongly influence subsequent responses. Griffin et al. showed that repeated hyperthermia initially induces vascular thermotolerance, leading to increased tumor blood flow and oxygenation during early repeat treatments spaced 12–48 h apart. However, continued daily hyperthermia exceeds vascular repair capacity, resulting in cumulative endothelial damage and progressive loss of perfusion [41]. Consistent with these dynamics, Simón et al. repeated photothermal therapy in CT26 tumors after only 24 h and observed inconsistent heating and minimal additional tumor control, likely due to maximal thermotolerance combined with impaired perfusion and nanoparticle delivery at this early timepoint [42].

In our study, treatments were intentionally spaced five days apart to coincide with the onset of renewed tumor growth and recovery of vascular function. At this interval, the second exposure expanded activated and central-memory CD4⁺ T cells, increased PD-1⁺ helper T cells, and amplified inflammatory, chemotactic, matrix-remodeling, and antigen-presentation programs. Supporting the importance of intermediate spacing, Wu et al. reported substantial tumor regression when nanoshell-mediated hyperthermia was repeated every 3–4 days, proposing that sequential heating enhances cumulative injury before complete cellular repair occurs [43]. Similarly, Hsiao et al. showed that repeated activation of a stable intratumoral heat source at four-day intervals produced progressively stronger therapeutic effects [44].

By days 8 to 15, the immunologic changes were largely confined to the adaptive compartment, consistent with a transition from early innate activation to sustained T cell-driven responses. Early innate effects were assessed at 48 h, where we observed increased tumor-cell calreticulin exposure and early macrophage phenotypic remodeling without significant changes in dendritic cells or neutrophils; later changes (days 8–15) were dominated by adaptive remodeling [28, 41]. Given that CT26 tumors typically lack spontaneous CD8⁺ T cell infiltration [17] and exhibit intrinsic defects in MHC II-mediated CD4⁺ T-cell priming [13, 40], the expansion of activated and central memory like CD4⁺ T cells observed after two hyperthermia treatments most likely reflects newly induced priming rather than amplification of pre-existing immunity.

This interpretation is consistent with established roles of extracellular HSPs as endogenous danger signals that activate antigen-presenting cells through inflammatory cytokine induction, costimulatory upregulation, and chaperoning of tumor-derived peptides to facilitate efficient MHC I cross-presentation [45]. Together, these processes enhance antigen visibility, sustain T-cell priming, and partially compensate for deficits in classical CD4⁺ T cell-dependent antigen presentation. Accordingly, the increased frequency of PD-1⁺ CD4⁺ T cells likely reflects active antigen engagement rather than terminal exhaustion, while expansion of Tregs is consistent with compensatory mechanisms accompanying heightened inflammatory signaling [46].

Treg expansion following double THT warrants careful consideration. Increased Treg frequency may represent a physiological counter-regulatory response to heightened inflammation and antigen release, but it also raises the possibility that thermal conditioning activates regulatory circuits that could constrain long-term durability. In our study, however, Treg expansion did not prevent significant tumor regression when THT was combined with PD-1 inhibition, indicating that effector activation remained functionally dominant under these conditions. Although several studies have reported reduced intratumoral Tregs or improved CD8:Treg ratios following hyperthermia or mild photothermal therapy, these differences likely reflect variation in thermal dose, retreatment interval, and timing of immune assessment [47–50]. Together, these findings suggest that THT amplifies both stimulatory and regulatory immune pathways, generating a dynamic equilibrium within the TME. Future studies should define the phenotype and suppressive capacity of post-THT Tregs and determine whether selective modulation strategies, including transient Treg depletion or CTLA-4 targeting, can further enhance therapeutic durability without compromising effective immune priming.

Consistent with antigen-driven immunity, TCRβ repertoire analysis revealed progressive focusing of the intratumoral T-cell response following repeated hyperthermia, with two THT treatments modestly reducing repertoire diversity and increasing clonal dominance. Several clonotypes exhibited stepwise amplification after single and double THT, supporting antigen-driven clonal selection rather than nonspecific lymphocyte recruitment. Collectively, these findings indicate that targeted hyperthermia selectively amplifies adaptive immune responses, an effect of particular relevance in MSS colorectal cancer, where failure to sustain productive T-cell activity is a major determinant of resistance to ICB.

Bulk RNA sequencing showed coordinated, dose-dependent induction of inflammatory, metabolic, and antigen-presentation pathways. Upregulation of Cxcl3 and Ccr9 suggests enhanced recruitment of neutrophils, monocytes, and lymphocytes into a microenvironment normally resistant to immune entry [51, 52]. Upregulation of complement C9 implies activation of terminal complement pathways that amplify tumor injury and immune surveillance [53]. Increased Ctsb and Timp1 suggest stromal remodeling that enhances immune access [54, 55]. CT26 normally expresses high levels of TGF-β, IL-10, ARG1, and iNOS, with minimal co-stimulation and absent mature dendritic cells [17], so induction of antigen-presentation and chemotactic pathways represents a meaningful shift toward a permissive immune microenvironment. This is further supported by the upregulation of antigen-processing genes such as H2-M5 and Tarm1, increasing visibility of tumor antigens [56]. The stronger transcriptional response after two THT treatments underscores thermal dose as a tunable determinant of immunogenic remodeling.

These immunologic changes translated into substantial therapeutic benefit when hyperthermia was combined with ICB. As expected, anti-PD-1 monotherapy had minimal activity in CT26, reflecting the profound immune exclusion that characterizes this model [57]. In contrast, THT markedly improved tumor control and survival, with the most pronounced effects observed when two hyperthermia treatments were paired with either systemic or intratumoral PD-1 blockade. The observed reduction in lung gp70 expression further suggests that thermal conditioning may also limit early metastasis [22, 58–60]. This capacity to sensitize tumors to immunotherapy aligns with prior observations that innate immune activation can synergize with checkpoint blockade in CT26. For example, class A 9 agonists enhance anti-PD-1 efficacy by depleting Tim4⁺ suppressive macrophages and reducing exhausted CD39⁺CD8⁺ T cell populations, leading to near-complete responses when combined with PD-1 inhibition [61, 62].

This work provides strong mechanistic and translational support for thermal immunotherapy, but several limitations should be acknowledged. First, CT26 was studied here as a heterotopic, subcutaneous tumor model, which does not fully recapitulate the anatomical, stromal, or vascular features of orthotopic colorectal tumors. Similarly, all studies were performed in female mice, and sex-specific differences in immune or thermal responses may influence outcomes. The observed expansion of Tregs may also limit therapeutic durability, suggesting that integrating Treg-targeting strategies could further enhance antitumor activity. Finally, CT26 represents a single genetic background. Additional studies across other MSS colorectal cancer models and human tumors with more complex stromal architecture will be important to define the generalizability of THT-induced immune remodeling. In addition to immune activation, hyperthermia may also influence tumor biology through non-immune mechanisms, including improved vascular perfusion, which could contribute to treatment responses. Future studies incorporating orthotopic models and perfusion metrics will help clarify the relative contribution of these effects.

Clinical translation of GNR-mediated THT also presents several challenges that warrant consideration. Although gold nanomaterials have demonstrated favorable short-term tolerability in preclinical models, comprehensive evaluation of long-term biodistribution, clearance kinetics, and chronic organ effects remains essential. Prior systemic toxicology studies of this GNR formulation in rats demonstrated tolerability at doses exceeding those used here, with predominant accumulation in liver and spleen and evidence of hepatobiliary and renal elimination [63]. In the present study, intratumoral microdosing substantially limits systemic exposure relative to intravenous administration, which may mitigate long-term toxicity risk. Nonetheless, extended histopathologic assessment and longitudinal immune monitoring will be necessary to fully define safety parameters for human application.

Tumor accessibility requires attention as well, when considering practical implementation. While superficial or endoscopically accessible lesions are well-suited for image-guided intratumoral delivery, treatment of deep-seated or metastatic lesions may require more technically complex approaches to achieve spatially-confined energy deposition. Addressing these technical and regulatory considerations will be critical for advancing GNR-mediated THT into early-phase clinical trials.

Beyond these operational complexities, optimizing the immunological outcomes remains key. Gold nanomaterial-mediated photothermal therapy has been widely investigated, and multiple prior studies have demonstrated tumor control and emerging immune modulation [64–66]. However, in most reports, thermal delivery is treated primarily as a cytotoxic intervention, with limited interrogation of how precisely controlled, sub-ablative temperature windows and re-treatment intervals shape adaptive immune programming. Collectively, our data indicate that the immunologic consequences of photothermal therapy depend not only on nanoparticle platform, but critically on precise thermal delivery parameters that remain comparatively underexplored.

Many widely used GNR formulations are synthesized with cetyltrimethylammonium bromide (CTAB) and require repeated purification to improve biocompatibility. Incomplete CTAB removal can contribute to membrane damage, dose-independent cytotoxicity, and inflammatory artifacts [67]. In this study, we used a CTAB-free formulation at the point of use, selected to improve intratumoral tolerability. This is particularly important in immunologic studies, where surfactant-associated membrane injury or inflammatory artifacts could confound interpretation of heat-induced immune remodeling.

More broadly, other gold photothermal materials (e.g., gold-silica nanoshells, nanostars, silica-coated GNRs) can achieve strong photothermal conversion and antitumor activity, including in clinically translated nanoshell platforms. However, these systems differ in hydrodynamic size, intratumoral dispersion, and heat dissipation kinetics, which can influence spatial heating gradients and thus immune conditioning versus ablation [68–70]. Importantly, we observed consistent heating kinetics and CEM43 values across treatment sessions, supporting material stability and batch reproducibility under in vivo conditions, a prerequisite for interpretable thermal-immune coupling.

Conclusions

In all, our data demonstrate that reproducible thermal delivery induces robust immune activation and enhances responsiveness to immunotherapy, supporting targeted hyperthermia as a strategy to overcome immune resistance in MSS colorectal cancer. GNR-enabled hyperthermia provides a precise and adaptable approach to increase antigen visibility, expand tumor infiltrating lymphocytes, and remodel chemotactic and stromal pathways. Collectively, these findings support thermal conditioning as a sensitizing strategy for immune checkpoint blockade and provide a rationale for advancing hyperthermia-based immunotherapy combinations into clinical evaluation for patients with refractory MSS colorectal cancer.

Abbreviations

cDNA

Complementary DNA

CEM43

Cumulative equivalent minutes at 43 °C

CRC

Colorectal cancer

CTAB

Cetyltrimethylammonium bromide

CTLA-4

Cytotoxic T-lymphocyte antigen 4

dMMR

Mismatch-repair deficiency

GNR

Gold nanorod

HSP

Heat shock protein

ICB

Immune checkpoint blockade

ICD

Immunogenic cell death

MSI

Microsatellite instability

MSS

Microsatellite stable

NIR

Near infrared

PD-1

Programmed cell death 1

PD-L1

Programmed cell death ligand 1

pMMR

Proficient mismatch repair

SD

Standard deviation

THT

Targeted hyperthermia therapy

TME

Tumor microenvironment

Treg

Regulatory T cells

Author contributions

CG provided scientific oversight across preclinical work. BK and JJ wrote the manuscript. BK, GS, KNC, CD, NPC, KJC, CG, and EBN collected and analyzed data, prepared figures, and created the graphical abstract. MRH, AE, and JSM provided subject matter expertise.

Funding

We wish to thank The Sidney Crosby Foundation, Dr. Mark Johnston, the Department of Surgery at Dalhousie University, the Gibran and Jamile Ramia QEII Health Sciences Centre Chair in Surgical Oncology Research Foundation, and the DMRF Capital Equipment Grant for their generous contributions, all of which supported this research.

Data availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Declarations

Ethics approval

The animal study was approved by the University Committee on Laboratory Animals at Dalhousie University and the Canadian Council on Animal Care. The study was conducted in accordance with the local legislation and institutional requirements.

Consent for publication

Not applicable.

Competing interests

CAG is Chief Medical Director for Sona Nanotech Inc.^™. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.