Our previous studies found sex was not a biological variable for RFA effects (42). This study exclusively used male mice for consistency. We expect the same results in female mice.

Endotoxin-free ovalbumin (OVA) and AddaVax were purchased from Invivogen (San Diego, CA). Fluorescence-conjugated antibodies were purchased from Biolegend (San Diego, CA) or otherwise specified. Adenosine 5′-diphosphate (ADP)-Agarose (A2810-5ML) was obtained from Sigma-Aldrich (St. Louis, MO). Goat polyclonal antibody against mouse/rat MyD88 (AF3109) was purchased from R & D Systems (Minneapolis, MN). Rabbit polyclonal antibody against human/mouse/rat TIRAP (PA5-88657) and Alexa Fluor 647-conjugated OVA (AF647-OVA, O34784) were obtained from Thermo Fisher Scientific (Waltham, MA). FITC-conjugated anti-HSP70 antibody (C92F3A-5, ab61907) and recombinant mouse HSP70 (ab113187) were purchased from Abcam (Cambridge, MA). HSP70 ELISA kit (ADI-EKS-715) was purchased from Enzo Life Sciences (Farmingdale, NY).

C57BL/6 mice (6–8 weeks old, male) were purchased from Jackson Laboratory (Bar Harbor, ME). TLR2 knockout (KO) (004650), TLR4 KO (029015) and Myd88 KO mice (009088) were obtained from Jackson Laboratory (Bar Harbor, ME) and self-bred for use in this study. Heterozygous HSP70 KO mice were originally obtained from the Mutant Mouse Resource & Research Centers (MMRRC) at University of Missouri and self-bred to obtain HSP70 KO mice for use in this study. Animals were housed in facilities of University of Rhode Island (URI) and anesthetized by intraperitoneal injection of a mixture of 80 mg/kg Ketamine and 10 mg/kg Xylazine for hair removal, RF treatment, and immunization. In most of the experiments, hair on the lateral back skin was removed by shaving followed by topical application of hair removal lotion (Nair) one day before experiment. Mice were euthanized by CO₂ inhalation. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Rhode Island (AN1516-004) and conducted in accordance with National and Institutional Guidelines and Regulations. Animal experiments were reported in accordance with the ARRIVE guidelines.

A handheld fractional bipolar RF device (~1 MHz, high-energy setting, Norlanya Technology Co., Hong Kong, China) equipped with 12 × 12 array of microelectrode in 2 × 2 cm² was used to treat mouse skin for 1–2 min without causing visible or histological skin damages as indicated in our previous study (40). Before treatment, a thin layer of ultrasound coupling medium was applied on the skin surface and RF device was firmly pressed to enable treatment tips to have a tight contact with the skin. For sham treatment, ultrasound coupling medium was applied and RF device was firmly pressed except the device was not activated.

Lateral back skin of C57BL/6 mice were exposed to RF or sham treatment. RF or sham-treated skin (12 × 12 mm²) was dissected at indicated time points followed by digestion in collagenase D (0.2%) and dispase (0.6 U/mL) for preparation of single-cell suspensions as in our previous report (43). Skin cells were then stained with fixable viability dye eFluor 450 (65-0863-14, Thermo Fisher Scientific) and then fluorescence-conjugated CD11b (clone M1/70), CD11c (clone N418). Cells were then fixed and permeabilized, and further stained with fluorescence-conjugated anti-Hsp70 (C92F3A-5) followed by flow cytometry analysis in BD FACSVerse. Flow cytometry data were analyzed using FlowJo™ software (version 10).

Purification of HSP70 from mouse skin referred to a published protocol with slight modification (7). RF or sham-treated mouse skin was homogenized in 1.5 ml hypotonic buffer (10 mM NaHCO₃, 0.5 mM PMSF, pH 7.1) followed by centrifugation at 12,700 rpm for 10 min. Supernatants were collected and changed to buffer D (20 mM Tris-acetate, 20 mM NaCl, 15 mM β-mercaptoethanol, 3 mM MgCl₂, 0.5 mM PMSF, pH 7.5) using PD-10 column (GE Healthcare Life Sciences, Marlborough, MA). Samples were then concentrated with Centrifugal Filter Unit (Amicon^®, 10 kDa cutoff) to 0.5 mL and applied to an ADP-Agarose column equilibrated with buffer D. The column was washed with buffer D containing 0.5 M NaCl and then buffer D. The column was then incubated with 0.5 mL buffer D containing 3 mM ADP at room temperature for 30 min. Finally, column was eluted with 1mL buffer D containing 3 mM ADP. HSP70 (RF) or HSc70 (sham)-rich elutes were resolved on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by silver staining via a commercial kit (24612, Thermo Fisher Scientific).

HSP70 (RF) or HSc70 (sham)-rich elutes were resolved on SDS-PAGE followed by transferring to polyvinylidene difluoride (PVDF) membrane. After blocking with 5% non-fat milk, PVDF membrane was incubated with mouse anti-HSP70 antibodies (1:1000, MA1-10889, Invitrogen) (no cross-reactivity with HSc70) or rabbit anti-HSc70 polyclonal antibodies (1:500, AB1427, Abcam) (no cross-reactivity with HSP70) at 4˚C overnight. After washing in Tris-buffered saline (TBS) containing 0.05% Tween 20 (TBST), PVDF membrane was incubated with HRP-conjugated anti-mouse (1:5000, NA931, GE Healthcare Life Sciences) or anti-rabbit secondary antibodies (1:1000, 7074P2, Cell Signaling Technology) at room temperature for 1 h. After washing in TBST, PVDF membrane was incubated with Pierce ECL Western Blotting Substrate (32109, Thermo Fisher Scientific). PVDF membrane was imaged under Imager (Thermo Fisher Scientific).

Skin was homogenized in RIPA buffer for exploration of cytoplasmic protein interactions or membrane lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% NP40, 0.05% CHAPS, 1 mM PMSF, and protease inhibitor) for exploration of protein interactions involving membrane proteins as recommended (44). The same amounts of total proteins were incubated with anti-OVA (ab181688, Abcam), -TIRAP (PA5-88657, Thermo Fisher Scientfic), or -IRAK4 antibodies (MA515883, Thermo Fisher Scientific) at 4 °C for 1 h. Protein A/G agarose was then added and incubated at 4 °C overnight. Supernatants were removed after centrifugation. Protein A/G agarose was then washed with RIPA or membrane lysis buffer 3 times. Sample buffers were added, boiled, and centrifuged. Supernatants were subjected to SDS-PAGE and IB detection of HSP70 (MAB1663, R & D Systems), TLR4 (sc-293072, Santa Cruz), and IRAK1 (Ab238, Abcam).

The lateral back skin of C57BL/6 mice was exposed to RF or sham treatment. RF or sham-treated skin (12 × 12 mm²) was dissected right after treatment, cut into ~1 mm wide slices, and cultured in 200 μl RPMI 1640 complete medium at 37 °C in a 5% CO₂ incubator for 18 hr. Culture medium was then harvested, centrifuged, and supernatants were used for measurement of HSP70 levels. Skin tissues were harvested at the same time, homogenized in T-PER buffer, and centrifuged. Supernatants were used for measurement of skin HSP70 levels.

We took advantage of our recently developed reservoir powder patches topically applied onto ablative fractional laser (AFL) treated skin to extract tissue fluids for quantification of extracellular HSP70 levels (45). Briefly, mannitol powder-coated cylindrical reservoir patches (8mm in diameter and 5mm in depth) were prepared by loading an excessive volume of mannitol powder over the blank reservoir patches followed by centrifugation at 20,000 rpm for 90 min. Powder above the surface of the reservoir patches was removed by a scalpel. Approximately 200 mg of mannitol could be coated per reservoir patch. Powder reservoir patches were kept in a desiccator before use. Lateral back skin of mice was exposed to RF or sham treatment. RF and sham-treated skin was then subjected to AFL treatment (5mJ energy, 10% coverage) to generate skin microchannels. Powder reservoir patches were then topically applied and further sealed with Tegaderm film to form an air-tight system. A narrow bandage was used to keep the entire system in position. Reservoir patches were removed 24 h later. At this moment, powder mannitol became a wet mass due to absorption of interstitial fluid through laser-generated skin microchannels. The wet mannitol was transferred to an Eppendorf tube and completely dissolved in 0.5 mL phosphate-buffered saline (PBS). Sample volume was reduced to ~100 µl via Centrifugal Filter Unit (Amicon^®, 3 kDa cutoff) for measurement of HSP70 levels.

HSP70 levels in tissue homogenates or extracted interstitial fluids were quantified by HSP70 High Sensitivity ELISA kit (ADI-EKS-715, Enzo Life Sciences), which specifically detects inducible HSP70. Briefly, diluted samples were added to ELISA plates coated with monoclonal antibody specific for inducible murine HSP70 and incubated at room temperature for 2 h. After washing, polyclonal antibody specific for inducible murine HSP70 was added and incubated at room temperature for 1 h. After washing, HRP-conjugated secondary antibodies were added and incubated at room temperature for 1 h. TMB substrate was added after plate washing and incubated at room temperature for 30 min. Stop solution was lastly added and plates were read at 450nm in a microplate reader (Molecular Device).

Nuclear fractions of RFA, Sham, and LPS-treated skin were extracted with a commercial tissue nuclear extraction kit (Abcam, AB219177). In brief, skin was cut into small pieces followed by homogenization in 1 mL cold Cytoplasm Extraction Buffer supplemented with protease inhibitor and DTT. After thorough homogenization, skin homogenates were transferred to an Eppendorf tube and incubated on ice for 10 min. Skin homogenates were vortexed and centrifuged at 1,000 g for 3 min. Supernatants (the cytoplasmic portion) were carefully and completely removed. Pellet was resuspended in 750 µL Nuclear Extraction Buffer supplemented with protease inhibitor and DTT and incubated on ice for 15 min. Tubes were vortexed every 5 min during the incubation followed by centrifugation at 5,000 g for 3 min. Supernatants were collected for use as soluble nuclear fraction in our studies.

Serum antibody titer was measured by enzyme-linked immunosorbent assay (ELISA). In brief, 96-well ELISA plates were coated with 10 µg/ml for OVA at 4˚C overnight. After blocking with 5% non-fat milk, 2-serial dilutions of immune sera were added and incubated at room temperature for 90 min. After washing in PBS supplemented with 0.05% Tween 20 (PBST), HRP-conjugated sheep anti-mouse IgG secondary antibodies (1:5,000, NA931, GE Healthcare Life Sciences) were added and incubated at room temperature for 1 hr. After washing in PBST, TMB substrates were added and reactions were stopped by addition of 3M H₂SO₄. Optical absorbance (OD_{450/630 nm}) was read in a microplate reader (Molecular Device). Serum antibody titers were defined as the reciprocal dilution factor that resulted in OD_{450/630 nm} that was ~3 times higher than the background value.

Skin was homogenized in RIPA buffer supplemented with protease inhibitor and PMSF (1 mM). Supernatants were collected after centrifugation. Total protein levels were measured by BCA assay. IL-6 levels were measured by a commercial ELISA kit (431304, BioLegend, San Diego, CA). Briefly, ELISA plates were coated with capture antibodies at 4°C overnight. Plates were washed and then blocked at room temperature for 1 h. Samples were added and incubated at room temperature for 2 h. Plates were washed and detection antibodies were added and incubated at room temperature for 1 h. Plates were washed and avidin-HRP was added and incubated at room temperature for 30 min. After washing, substrate was added and reactions were stopped 20 min later. Plates were read at OD_{450/570 nm}.

Bone marrow was isolated from femur and tibia of C57BL/6 mice and cultured in the presence of recombinant murine GM-CSF (15 ng/ml) and IL-4 (10 ng/ml) as in our previous report (46). Immature DCs were harvested on day 6 and seeded into at 10⁶ cells/mL in 96-well plates and then incubated with AF647-OVA in the presence of LPS, HSP70-rich elute, HSc70-rich elute or the same volume of medium. Cells were then harvested 20 hr later, stained with fllig;uorescence-conjugated anti-CD11c (N418), CD40(3/23), CD80 (16-10A1), and CD86 antibodies (GL-1) followed by flow cytometry analysis of AF647+ cells in CD11c+ cells and mean fluorescence intensity (MFI) of CD40, CD80, and CD86 in AF647+CD11c+ cells.

WT, HSP70 KO, TLR2 KO, TLR4 KO, and MyD88 KO mice were subjected to RFA or Sham treatment followed by ID injection of 2 µg AF647-OVA into RFA or sham-treated skin. Skin and draining lymph nodes (LNs) were dissected 20 hr later after euthanasia. Single-cell suspensions of the skin were prepared by collagenase D (0.2%) and dispase (0.6 U/mL) digestion. LNs were passed through 40 μm cell strainers to prepare single-cell suspensions. Skin and LN cells were then stained with fluorescence-conjugated anti-CD11c (clone N418), MHC II (clone M5/114.15.2), Langerin (clone 4C7), CD11b (clone M1/70), CD103 (clone 2E7), and CD80 antibodies (clone 16-10A1) followed by flow cytometry analysis.

Lateral back skin of C57BL/6 mice was intradermally injected with 5 µg LPS (LPS-EB Ultrapure, tlrl-3pelps, Invivogen) or subjected to RF or sham treatment. Skin was dissected 6 hr later and cryo-sectioned followed by PLA via a commercial kit (Duolink^® In Situ Red Starter Kit Goat/Rabbit, DUO92105, Sigma-Aldrich). In brief, frozen sections were recovered to room temperature, fixed in 3.7% paraformaldehyde, and permeabilized in PBS supplemented with 10% goat serum and 0.1% Triton-X 100. Sections were then blocked and incubated with goat anti-mouse MyD88 polyclonal antibodies and rabbit anti-mouse TIRAP polyclonal antibodies at 4˚C overnight. After washing, sections were incubated with Probe Minus and Plus from the kit at 37˚C for 1 hr. After washing, sections were subjected to ligation at 37˚C for 30 min and then amplification at 37˚C for 100 min. After final washing, sections were mounted with Duolink^® In Situ Mounting Media with DAPI mounting medium (DAPI) and cover slipped. Images were taken under a Nikon Ti2-E confocal microscope. The number of positive PLA signals in randomly selected skin area of 0.4 mm² was counted and compared between groups. Only PLA signals in close vicinity to cell nucleus were counted to avoid false signals.

Values were expressed as Mean ± SEM (standard error of mean). Student’s t-test was used to compare differences between groups and one-way ANOVA with Tukey’s Multiple Comparison test was used to compare differences for more than 2 groups except otherwise specified. P value was calculated by PRISM software (GraphPad, San Diego, CA) and considered significant if it was less than 0.05.

Our previous studies found that RFA could increase skin HSP70 levels 6 and 24 h after treatment (40). Here, we further explored the temporal and cell type-specific induction of HSP70 following RFA treatment. Skin cells were divided into three major types based on CD11c and F4/80 expression ( Supplementary Figure S1A ): DCs (CD11c⁺ F4/80^-), macrophages (F4/80⁺), and non-immune cells (CD11c^- F4/80^-). RFA rapidly induced HSP70 expression in DCs while showed a delayed effect to induce HSP70 expression in macrophages and non-immune cells. In support, the percentage of HSP70⁺ DCs rapidly increased as early as 0.5 h after RFA treatment, but the percentage of HSP70⁺ macrophages and non-immune cells showed no significant increase in the first 1.5 h after RFA treatment ( Figures 1A, B ). RFA also induced the most HSP70 expression in DCs. In support, the percentage of HSP70⁺ DCs reached a peak level of 49% at 6 h, while the percentage of HSP70⁺ macrophages and non-immune cells reached a peak level of 6-8% and 10-11% between 6–18 h, respectively ( Figure 1B ).

Next, we explored whether RFA-induced HSP70 could be purified by ADP-Agarose column due to the ability of HSP70 to bind ADP through its N-terminal nucleotide-binding domain (NBD) (7). In this study, silver staining was used to detect HSP70 after SDS-PAGE due to its relatively low amount. Also, recombinant murine HSP70 (full-length, inducible) and purified samples from HSP70 KO mice were included to facilitate the identification of inducible HSP70. As shown in Figure 2A , two bands with molecular weight slightly higher than 70 kDa were found in purified WT samples, while only one band with the higher molecular weight was found in purified HSP70 KO samples. Additionally, the lower-molecular-weight bands in purified WT samples matched the recombinant murine HSP70. These data hinted the lower-molecular-weight bands were inducible HSP70, while the higher-molecular-weight bands were constitutively expressed HSc70. More intense HSP70 band in RF group of WT mice hinted RF treatment induced its synthesis.

Western blotting was further used to confirm the above findings using anti-HSP70 and anti-HSc70 antibodies with no cross-reactivity to each other. As shown in Figure 2B , anti-HSP70 antibodies recognized the lower-molecular-weight bands in purified WT samples and also recombinant murine HSP70, while anti-HSc70 antibodies recognized the higher-molecular-weight bands in both WT and HSP70 KO samples. We also found that RF treatment significantly induced HSP70 expression in WT but not HSP70 KO mice ( Figures 2C, D ). Our results indicated RF-induced HSP70 could be separated from constitutively expressed HSc70 in SDS-PAGE despite their small difference in molecular weights as reported (47).

HSP70 was reported to export extracellularly via non-canonical pathways to modulate immune system function (48, 49). Next, we explored whether RFA-stimulated HSP70 could be exported into extracellular space. We first conducted ex vivo skin culture and measured HSP70 levels in the culture medium. As shown in Figure 3A , significantly higher HSP70 levels were detected in the culture medium of RFA-treated skin than that of Sham-treated skin (0.14 vs. 0.075ng). RFA-treated skin also showed significantly higher HSP70 levels than Sham-treated skin (4.4 vs. 2.6 ng) ( Figure 3B ). This result indicated RFA could stimulate HSP70 synthesis with significant release into extracellular space. Interestingly, low-level HSP70 could be detected in Sham-treated skin, in line with our previous report (40).

Next, we used another method recently developed in house to extract interstitial fluids for measurement of extracellular HSP70 levels. This method took advantage of ablative fractional laser (AFL)-generated skin microchannels and topically applied bulk mannitol powder to absorb interstitial fluids followed by dissolution for convenient measurement of secreted HSP70 levels (45). As shown in Figure 3C , HSP70 levels in interstitial fluids of RF-treated skin was much higher than that of sham-treated skin (0.1 vs. 0.004 ng). HSP70 levels were also significantly higher in RF-treated than sham-treated skin (5.7 vs. 2.6 ng, Figure 3D ). These data correlated well with the ex vivo skin culture data, supporting RF stimulated HSP70 synthesis with a significant extracellular release. Considering HSP70 may also be exported as a membrane-bound form, we further used cmHSP70 antibodies (kindly provided by Dr. Gabriele Multhoff) (18, 50) to specifically identify membrane bound HSP70 after RFA or Sham treatment. As shown in Supplementary Figure S3 , a small percentage (<1%) of skin cells showed positive membrane HSP70 binding after RFA treatment, which showed no significant difference from that after sham treatment. This result indicated extracellular HSP70 following RF treatment mainly existed as a soluble but not membrane-bound form. Interestingly, we failed to detect HSP70 in blood stream even after bilateral RF treatment (data not shown), hinting RF mainly induced localized HSP70 release.

Our recent studies found HSP70 was crucial for RFA to enhance ID vaccine-induced immune responses (41, 51). Extracellular release of RFA-induced HSP70 might encounter intradermally injected antigen and facilitate its uptake by APCs. Considering a close association of HSP70 and antigenic peptides or vaccine antigens was necessary to elicit potent adaptive immune responses (3), we attempted to detect association of RFA-induced HSP70 and ID model antigen OVA via a highly sensitive proximity ligation assay (PLA) (52). PLA utilizes DNA probe-conjugated antibodies to study protein/protein interactions. A close proximity of the targets brings together DNA probes to form a circular DNA, which can be amplified by rolling cycle PCR for easy detection by fluorescent probes (52). As shown in Figure 4A , more HSP70/OVA PLA signals were found in RFA than Sham-treated skin of WT mice. Quantitative analysis found HSP70/OVA PLA signals were significantly increased after RFA treatment (22 vs. 8, p<0.001) ( Figure 4B ). As expected, HSP70/OVA PLA signals showed no significant difference between RFA and Sham groups in HSP70 KO mice ( Figure 4C ).

Immunoprecipitation (IP) and immunoblotting (IB) were then used to confirm the above findings. As shown in Figure 4D , more HSP70 was detected in RFA than Sham-treated skin in both mice (#1 & #2) after IP/IB. More HSP70 in tissue lysates of RFA than Sham-treated skin indicated RFA induced HSP70 synthesis ( Figure 4D ). Interestingly, significantly more OVA was also detected in RFA than Sham-treated skin ( Figure 4D ), hinting RFA might help to retain OVA at local injection sites. Furthermore, the relative HSP70 to OVA levels significantly increased by RFA treatment ( Supplementary Figure S5 ), supporting HSP70/OVA binding. We also conducted the same experiments in HSP70 KO mice and failed to detect HSP70 after IP/IB in either RFA or Sham-treated skin ( Figure 4E ). Our results indicate the ability of RFA-induced HSP70 to bind ID OVA in situ.

In situ HSP70/OVA association hinted RFA-induced HSP70 might facilitate antigen uptake. To corroborate this, WT and HSP70 KO mice were subjected to RFA or Sham treatment followed by ID injection of 2 µg AF647-OVA into RFA or Sham-treated skin. AF647-OVA⁺ DCs were evaluated 18 h later in skin and draining LNs. As shown in Figure 5A , RFA significantly increased percentage of AF647-OVA⁺ cells in DCs of WT mice (35 vs. 20%), while failed to increase percentage of AF647-OVA⁺ cells in DCs of HSP70 KO mice (13 vs. 11%). AF647-OVA levels in skin DCs were significantly increased after RFA treatment in WT but not HSP70 KO mice ( Figure 5B ). These results indicated HSP70 played a crucial role in RFA-enhanced antigen uptake in the skin. Skin DC subtype analysis found HSP70 mainly enhanced antigen uptake in dermal DC subsets (I, III, IV) but not epidermal Langerhans cells (DC subset II) ( Supplementary Figure S6 ).

Antigen uptake was further analyzed in draining LNs, in which DCs were divided into 3 subtypes (cDC, migDC, and pDC) based on the relative CD11c and MHC II expression as shown in our prior studies (40, 43). As shown in Figures 5C, D , RFA significantly increased the percentage of AF647-OVA+ cells in migDC of WT mice (7.0 vs. 9.8%). RFA also increased the percentage of AF647-OVA+ cells in cDC (11.5 vs. 13.4%) and pDC (6.9 vs. 7.5%) of WT mice, but such a difference didn’t reach statistically significant levels ( Figures 5E, F ). Interestingly, RFA slightly reduced percentages of AF647-OVA+ cells in migDC (8.9 vs. 6.6%), cDC (8.8 vs. 6.8%), or pDC (6.2 vs. 4.5%) in HSP70 KO mice ( Figures 5D–F ). These results strongly support crucial roles of HSP70 in RFA-enhanced antigen uptake in draining LNs in particular in migDC.

The ability of HSP70 to induce DC maturation has been controversial due to the potential contamination of LPS in HSP70 samples. In vivo induction of HSP70 by RFA introduces no foreign materials into the body and provides a unique opportunity to study the potential impact of in situ generated HSP70 on DC maturation. We analyzed this in the above studies when we evaluated the potential impact of RFA-induced HSP70 on antigen uptake. As shown in Figure 6A , RFA failed to increase MFI of CD40, CD80, or CD86 of skin DCs in WT or HSP70 KO mice. In draining LNs, RFA treatment failed to significantly change MFI of CD40 in cDC, migDC, or pDC in WT or HSP70 KO mice ( Figures 6B–D ). Interestingly, lack of HSP70 significantly increased MFI of CD80 in cDC, migDC, and pDC and MFI of CD86 in migDC and pDC in draining LNs of RFA group ( Figures 6B–D ). Furthermore, significantly lower expressions of CD80 and CD86 were observed in cDC and pDC in RFA group when compared to PBS group ( Figures 6B–D ). These results hinted RFA-induced HSP70 might suppress CD80 and CD86 expression in draining LNs of RFA group.

Besides in vivo analysis, we also incubated BMDCs with purified HSP70 from RFA (HSP70-rich elutes) or Sham-treated skin (HSc70-rich elutes). LPS was included for comparison. As shown in Figure 6E , HSP70-rich elutes slightly induced CD80 expression, while HSc70-rich elutes more potently enhanced CD80 and CD86 expression. As compared to HSc70 or HSP70-rich elutes, LPS more vigorously enhanced CD40, CD80, and CD86 expression on BMDCs. These results indicated the weak ability of HSP70 in stimulation of BMDC maturation in the culture. In this study, we also added AF647-OVA into the BMDC culture to explore whether HSc70 or HSP70-rich elutes could increase antigen uptake by using LPS as the positive control. As shown in Supplementary Figure S7 , HSc70 or HSP70-rich elutes failed to increase AF647-OVA uptake by BMDCs, while LPS vigorously enhanced AF647-OVA uptake. These results indicated purified HSP70 failed to enhance antigen uptake in BMDC culture. We also explored whether purified HSP70 possessed vaccine adjuvant effects by co-administration with OVA. As shown in Supplementary Figure S8 , HSc70 or HSP70-rich elutes failed to significantly increase serum anti-OVA antibody titer, while RFA significantly increased serum anti-OVA antibody titer. This result indicated the lack of vaccine adjuvant effects of purified HSP70, potentially due to the lack of antigen binding when simply mixing the two moieties together.

Our previous studies found MyD88 played a crucial role in RFA effects (40, 41). MyD88 is an essential adaptor downstream of most TLRs except TLR3 (53). Most TLRs directly recruit MyD88, while TLR2 and TLR4 first recruit TIRAP and then MyD88 (53). TLR3 and TLR4 first recruit TRAM and then TRIF (53). To explore whether other adaptors were activated by RFA, we detected TIRAP/MyD88 or TRAM/TRIF association following RFA, Sham, and LPS treatments via tissue PLA (52). As shown in Figures 7A, B , RFA significantly increased and LPS more significantly increased TIRAP/MyD88 PLA signals. Interestingly, RFA didn’t increase TRAM/TRIF PLA signals, while LPS significantly increased TRAM/TRIF PLA signals ( Figures 7C, D ). These results indicated RFA induced TIRAP/MyD88 but not TRAM/TRIF association.

Next, we used FITC-conjugated anti-MHC II antibodies to stain DCs and macrophages to explore the relative localization of TIRAP/MyD88 PLA signals. As shown in Figure 7E , we captured 3D pictures via Z-Stack and found the majority of RFA-induced PLA signals overlapped with MHC II ( Figure 7E ). Overall, more than 90% PLA signals overlapped with MHC II ( Figure 7F ). This result indicated the majority of RFA-induced TIRAP/MyD88 association occurred in DCs and macrophages, in line with the function of these cells in innate and adaptive immunity.

The involvement of TIRAP hinted its unique upstream TLR (TLR2 or TLR4) might be activated by RFA. The potential association of TLR4/TIRAP or TLR2/TIRAP was then explored via IP (anti-TIRAP)/IB (anti-TLR2 or TLR4). The underlying roles of HSP70 in potential RFA-induced TLR2 and TLR4 activation were also explored. To this end, WT and HSP70 KO mice were treated with RFA or Sham or intradermally injected with LPS or PBS. As shown in Figure 8A , RFA and LPS weakly induced TLR4/TIRAP association in WT mice, while vigorously induced TLR4/TIRAP association in HSP70 KO mice. This result hinted a strong suppression of TLR4/TIRAP association by HSP70. Interestingly, we failed to detect TLR2/TIRAP association with the same method (data not shown), hinting RFA might mainly induce TLR4/TIRAP/MyD88 pathway. TLR4/MyD88 activation has been reported to recruit IRAK4, which auto-phosphorylates and then recruit IRAK1 (54). Subsequently, sequential phosphorylation of Thr209 and Thr387 of IRAK1 leads to hyperphosphorylation of the proline-, serine-, and threonine-rich ProST region followed by dissociation from Myddosome and association with TRAF6 (55). To this end, we detected possible association of IRAK4 and IRAK1 via IP (anti-IRAK4) and IB (anti-IRAK1) in WT and HSP70 KO mice. As shown in Figure 8B , RFA and LPS weakly induced IRAK4/IRAK1 association in WT mice, while vigorously induced IRAK4/IRAK1 association in HSP70 KO mice. This result indicated strong suppression of IRAK4/IRAK1 association by HSP70. In our study, we found IRAK1 showed a higher than normal molecular weight (>100 kDa), hinting a hyperphosphorylation status. We also directly detected phosphorylated p65 in nucleus, the key event of NFκB activation downstream of TLR4/MyD88 activation. We included various KO mice to explore whether HSP70, TLR2, TLR4, MyD88 played a role in this process. As shown in Figure 8C , RFA induced nuclear transmigration of phosphorylated p65 in WT mice. This process was profoundly enhanced in HSP70 KO mice, hinting HSP70 strongly suppresses this process. Lack of TLR4 and MyD88 but not TLR2 abrogated this process, hinting the crucial roles of TLR4 and MyD88 but not TLR2 in activation of phosphorylated p65 nuclear transmigration. Our previous studies found RFA could induce temporary IL-6 gene expression (40). Considering IL-6 gene expression is controlled by NFκB (56), we also measured skin IL-6 levels 6 h after RFA, Sham, or LPS treatment. As shown in Figure 8D , RFA induced low levels of IL-6 expression in WT mice (1.07±0.52 pg/µg) and about 3-fold higher IL-6 expression in HSP70 KO mice (3.10±0.75 pg/µg). Skin IL-6 levels were significantly higher in RFA than Sham group in HSP70 KO mice but not WT mice, indicative of a suppressive role of HSP70 in IL-6 expression. Similar to WT mice, RFA induced weak IL-6 expression in TLR2 KO, TLR4 KO, and MyD88 KO mice ( Figure 8D ). As expected, LPS more significantly increased skin IL-6 levels than RFA in WT, HSP70 KO, and TLR2 KO mice and failed to increase skin IL-6 levels in TLR4 KO and MyD88 KO mice considering LPS stimulates TLR4/MyD88-mediated NFκB activation (57). Significantly increased skin IL-6 levels after RFA or LPS treatment in HSP70 KO mice as compared to WT mice hinted HSP70-mediated suppression was shared between RFA and LPS activators. Taken together, our data indicated RFA-induced TLR4/MyD88/NFκB pathway was strongly inhibited by inducible HSP70.