POSH, expressed by the gene sh3rf1, is a scaffold molecule that contains four SH3 domains, a Rac binding domain, and multiple polyproline domains able to bind several signaling molecules important for T cell development and differentiation. We have recently demonstrated, using novel inhibitors, a role for the SH3 domain 3 (SH3.3) of POSH in regulating the activation of JNK and differentiation of T cells in vitro (7, 23). To develop a deeper understanding of the role of the full POSH molecule in vivo, we generated conditional knockout mice strains, POSH^fl/fl CD4-Cre and POSH^fl/fl GzmB-Cre, using the IMPC Sh3rf1^{tm1a(EUCOMM)Hmgu} construct ( Figure 1A , see also methods). These mice contained Stop^fl/fl Rosa26 eGFP cassettes to act as a reporter for POSH deficient T cells. The mice were maintained as homozygous for the POSH^fl/fl gene, homozygous for eGFP, and Cre positive (denoted as POSH^fl/fl CD4-Cre, POSH^fl/fl GzmB-Cre). POSH^fl/fl control strains that were Cre negative and Stop^fl/fl Rosa26 eGFP homozygous were also generated and are denoted as POSH^fl/fl or “Control”. POSH deficient GFP+ CD8 T cells are referred to as POSH cKO in the rest of the paper.

To confirm the deletion of POSH, CD8 T cells from POSH^fl/fl and POSH^fl/fl CD4-Cre mice were sorted by flow cytometry. RT-PCR confirmed that in POSH^fl/fl CD4-Cre mice, exon 7 is removed within CD8 T cells at the genetic level, resulting in a 300-base pair band ( Figure 1B ). At the protein level, POSH is undetectable in POSH^fl/fl CD4-Cre CD8 T cells as compared to POSH^fl/fl ( Figure 1C ). Thus, POSH is effectively knocked out in POSH^fl/fl CD4-Cre CD8 T cells.

POSH has been linked to JNK, Akt, and NF-κB signaling, and crosstalk between these pathways has been implicated in T cell fate in the thymus. JNK is important for negative selection in developing thymocytes and for effector function in mature cells (4, 5). Additionally, NF-κB differentially contributes to T cell development depending on the maturation stage, with CD8 T cell development being more dependent on NF-κB than CD4 (2, 8, 31). Akt and NF-κB interact to support survival and proliferation after TCRβ selection (3, 8).

To define the role of POSH in T cell development, the phenotype of thymocytes from the thymi of polyclonal POSH^fl/fl CD4-Cre mice was examined by flow cytometry. POSH is deleted in these cells after TCRβ selection as they transition from CD4^-, CD8^- double-negative (DN) thymocytes to cells expressing both CD4 and CD8, becoming CD4⁺, CD8⁺ double-positive (DP) thymocytes (32). Thus, our gating strategy excluded DN thymocytes and we compared the DP, CD8 single-positive (CD8SP), and CD4 single-positive (CD4SP) populations in the thymi of POSH^fl/fl CD4-Cre mice to those in C57BL/6 (B6) and POSH^fl/fl control mice ( Figure 2A ). While there are subtle differences in the number and frequency of DP, CD8SP, and CD4SP thymocytes, overall, these populations are similar between POSH^fl/fl CD4-Cre mice and controls. We additionally examined the maturation stages of single-positive thymocytes. Post-selection SP thymocytes must transition through the semi-mature (SM), mature 1 (M1), and mature 2 (M2) stages of maturation, which can be identified based on the expression of CD69 and MHC-I. This maturation process has been shown to be important for enabling cells to mount proliferative responses and activate effector functions upon antigen recognition before leaving the thymus (9). We found that medullary POSH cKO SP thymocytes appeared to mature normally compared to wild-type (WT) SPs from B6 and POSH^fl/fl controls ( Figure 2B ). However, when comparing the frequency of GFP⁺ POSH cKO thymocytes to GFP^- WT thymocytes from the same mouse, there were subtle but significant differences in the frequencies of SM and M2 cells (9), although there was no complete block in development ( Figure 2C ). Together, these results suggest that POSH, at least by itself, does not play a major role in T cell development or maturation.

We next analyzed the phenotype of GFP⁺ POSH cKO CD8 T cells in the periphery of POSH^fl/fl CD4-Cre mice. There were no significant differences in the frequency or total number of CD8 T cells in the spleen ( Figure 3A ). However, within POSH^fl/fl CD4-Cre mice, only 75% of all splenic CD8 T cells are GFP⁺ POSH cKO T cells, with the remaining 25% being GFP^- WT CD8 T cells ( Figure 3B ). To determine which phenotype of POSH cKO cells was missing or lost, we examined the frequency of POSH cKO naïve and antigen-experienced CD8 T cells in these mice ( Figure 3C ). We found that approximately 85–90% of naïve CD8 T cells (CD44^- CD62L⁺) were POSH cKO ( Figure 3D ). Similarly, about 90% of T central memory (Tcm) (CD44⁺ CD62L⁺) CD8 T cells were POSH cKO ( Figure 3D ). However, within the T effector (Teff)/effector memory (Tem) CD8 population (CD44⁺ CD62L^-), only 40% were POSH cKO ( Figure 3D ). A similar but less pronounced loss of POSH cKO cells was observed within the Teff/Tem CD4 T cell compartment, along with a significant reduction in all CD25⁺ POSH cKO CD4 T cells ( Supplementary Figure 1 ). Collectively, these data suggest that the few GFP^- cells that did not delete POSH in these mice have a distinct advantage upon activation, providing strong support for the role of POSH in the differentiation, proliferation, and/or survival of activated CD8 T cells.

The reduced number of POSH cKO CD8 T cells in the effector memory pool in POSH^fl/fl CD4-Cre mice led to the hypotheses that, upon stimulation, POSH cKO CD8 T cells are either unable to differentiate into effector or effector memory cells or are unable to survive following activation. To test this, we stimulated polyclonal CD8 T cells with α-CD3 and α-CD28 in the presence of IL-2 for 4 days and assessed their activation state (CD25) and differentiation state (CD44, CD62L) each day. There were no significant differences in the frequency or importantly, the number of CD8 T cells between the POSH^fl/fl CD4-Cre and control cultures after stimulation (data not shown). However, the percentage of CD25⁺ GFP⁺ POSH cKO CD8 T cells decreased each day, from 90% on day 1 to only 20% on day 4 ( Figures 4A, B ). In contrast, the GFP^- POSH WT CD8 T cells increased from 10% on day 1 to 80% on day 4 ( Figures 4A, B ).

Next, we examined whether the change in the frequency of POSH cKO and WT cells was due to defects in proliferation or apoptosis. Using a viability stain, we observed no significant difference in cell death between POSH cKO and WT cells ( Figure 4C ). However, significantly fewer POSH cKO cells expressed Ki-67, a marker of proliferation, compared to WT CD8 T cells ( Figure 4D ), indicating a proliferation defect in POSH cKO cells. These findings are consistent with previous results obtained using our POSH inhibitor (7).

Given the phenotype of POSH cKO cells in the periphery, we also assessed the frequency of CD8 Tcm precursors (CD44⁺ CD62L⁺) and CD8 Teff (CD44⁺ CD62L^-) cells in vitro. Notably, the frequency of POSH cKO Teff cells was significantly reduced at day 4, while POSH WT Teff cells continued to increase and comprised the majority of the CD8 T cell population ( Figure 4E ). Although the frequency of POSH cKO Tcm cells decreased slightly over time, the remaining cells at day 4 more often exhibited the Tcm phenotype ( Figure 4F ). Due to the decrease in POSH cKO Teff cells, we next evaluated their effector functions. There was a slight but statistically significant decrease in killing capacity, likely not to be physiologically or biologically relevant ( Supplementary Figure 2 ). Additionally, fewer POSH cKO cells expressed Granzyme B, but there was no difference in the frequency of IFN-γ-expressing POSH cKO cells ( Supplementary Figure 3 ). Thus, while POSH cKO Teff cells were capable of acquiring effector functions and being recruited into the effector pool early, they exhibited reduced proliferation and Teff differentiation in vitro compared to WT CD8 T cells.

Given the observed defects during in vitro differentiation ( Figure 4 ) and the lack of activated POSH cKO CD8 T cells (Teff/Tem) in the periphery of POSH^fl/fl CD4-Cre mice ( Figure 3 ), we hypothesized that these results were linked to defects in JNK, NF-κB, and/or Akt signaling. To examine these potential defects, we purified and stimulated GFP⁺ POSH cKO CD8 T cells from POSH^fl/fl CD4-Cre mice and analyzed signaling components by Western blot. We found that phosphorylation of both JNK1 and JNK2 was decreased in CD8 T cells compared to controls ( Figure 5A ). Additionally, defects in the NF-κB pathway were evident, as phosphorylation of IκBα and p65 were also markedly reduced ( Figure 5B ). Finally, phosphorylation of Akt at Serine 473 was significantly diminished upon stimulation ( Figure 5C ).

Thus, all three major pathways associated with POSH function are reduced in POSH cKO CD8 T cells. These findings differ subtly from those observed with POSH SH3.3 inhibition, where no defect in NF-κB or Akt signaling was detected. This highlights the importance of the remaining protein-binding domains within the scaffold and provides insight into their essential roles in POSH function in CD8 T cells.

So far, our data suggest that POSH plays an important role in the differentiation and proliferation of CD8 T effector subsets in vitro. To further assess the role of POSH in CD8 T cell differentiation, we next tested the ability of POSH cKO CD8 T cells to differentiate in response to viral infection in vivo. To do this, we adoptively transferred OT1 POSH^fl/fl (OT1 control) or OT1 POSH^fl/fl CD4-Cre T cells into congenically marked host mice, infected them with VSV-OVA, and tracked antigen-specific donor CD8 T cells in the blood over the course of the immune response.

Interestingly, POSH^fl/fl CD4-Cre T cells showed a slight delay in their ability to expand in response to VSV-OVA infection compared to controls, as the peak response shifted from day 5 to day 7 ( Figure 6A ). However, the donor OT1 POSH^fl/fl CD4-Cre CD8 T cells were significantly reduced by days 15 and 28 post-infection ( Figure 6A ). Correspondingly, we observed a decrease in the frequency of POSH cKO cells and an increase in WT POSH-sufficient donor cells on days 15 and 28 post-infection in mice transferred with OT1 POSH^fl/fl CD4-Cre CD8 T cells ( Figure 6B ). This suggests that POSH cKO CD8 T cells are less capable of survival.

The balance between Bcl-2 (pro-survival) and Bim (pro-apoptotic) molecules is critical for the survival of both effector and effector memory precursors (33–36). Therefore, we monitored their levels throughout the in vivo response and found that OT1 POSH cKO cells expressed significantly lower levels of Bcl-2 and higher levels of Bim on days 5 and 7 compared to OT1 controls ( Figures 6C, D ). This Bim/Bcl-2 phenotype is associated with cell death and likely results from defects in NF-κB (10) and Akt signaling ( Figure 5 ). Supporting this, at the endpoint of the experiment, we found very few OT1 POSH cKO donor T cells in the spleen ( Figure 6E ), and nearly none in the lymph nodes ( Figure 6F ) or kidneys ( Figure 6G ). Taken together, in addition to the apparent defects in survival, these findings also indicate that the loss of POSH cKO CD8 T effector/effector memory cells ( Figure 3 ) was likely not due to a lack of CD4 T cell help.

To corroborate these findings and to determine whether POSH availability during early activation could rescue differentiation, we performed the same adoptive transfer experiment using OT1 POSH^fl/fl GzmB-Cre mice. These mice were generated in the same manner as described above, and POSH knockout was confirmed by day 3 post-stimulation ( Supplementary Figure 4 , data not shown). As before, OT1 Control or OT1 POSH^fl/fl GzmB-Cre CD8 T cells were transferred into congenically marked hosts followed by VSV-OVA infection. Similar to what we observed with OT1 POSH^fl/fl CD4-Cre CD8 T cells, the peak response of OT1 POSH^fl/fl GzmB-Cre donor CD8 T cells shifted from day 5 to day 7 ( Supplementary Figure 4A ). However, by day 28, the OT1 POSH cKO donor CD8 T cells were again lost from circulation, similar to the POSH cKO cells in OT1 POSH^fl/fl CD4-Cre mice ( Figure 6 , Supplementary Figure 4A ). Additionally, POSH cKO cells failed to outcompete WT POSH cells in mice adoptively transferred with OT1 POSH^fl/fl GzmB-Cre cells ( Supplementary Figure 4B ). Similarly, there was a significant decrease in Bcl-2 expression and a notable increase in Bim within OT1 POSH cKO donor CD8 T cells ( Supplementary Figures 4C, D ). The OT1 POSH cKO CD8 T cells were also not detected in the spleen, lymph nodes, or kidneys of the recipient mice ( Supplementary Figures 4E–G ). Thus, POSH cKO CD8 T cells, regardless of when POSH expression is lost, are unable to survive into the memory phase following VSV infection.

Given the loss of T effector cells observed in vitro and the reduction of memory cells in vivo ( Figures 4 , 6 ), we assessed the activation and differentiation status of antigen-specific donor CD8 T cells in response to VSV-OVA infection. We found that, at both day 5 and day 10 post-infection, there was a significant reduction in the frequency of CD44⁺ CD25⁺ POSH cKO CD8 T cells compared to CD44⁺ CD25^- POSH cKO cells ( Figures 7A–C ). This finding aligns with the observed defects in JNK and NF-κB signaling ( Figure 5 ).

Importantly, CD25 and IL-2 are essential for SLEC differentiation via the Akt/mTOR pathway, and loss of these signals has been shown to increase MPEC differentiation (12, 14, 15, 37–39). Based on the loss of CD25 expression and the significant reduction in Akt phosphorylation in POSH cKO cells, we hypothesized that POSH cKO CD8 T cells would exhibit decreased SLEC and increased MPEC differentiation. This is supported by our data showing that, by day 10 post-infection, POSH cKO CD8 T cells within the SLEC population were significantly reduced, with these cells primarily differentiating into MPECs ( Figures 7D–F ). Collectively, the lack of CD25 expression and defects in Akt signaling ( Figure 5 ) suggest that impaired CD25 signaling through Akt contributed to defective SLEC differentiation and survival. These findings highlight a role for POSH in coordinating JNK, NF-κB, and/or Akt signals that support proper SLEC differentiation.

Finally, due to the impaired Akt signaling in POSH cKO cells and the decreased SLEC differentiation, we examined downstream activity of the Akt/mTOR pathway, which is known to be essential for T-bet expression and SLEC differentiation (12, 14, 15). Given the reduced SLEC and increased MPEC differentiation observed, we hypothesized that POSH cKO CD8 T cells would exhibit decreased T-bet expression and p-S6, a downstream marker of mTOR activity, along with increased Eomes expression.

We found that, 24 hours post-stimulation in vitro, POSH cKO CD8 T cells had significantly lower levels of p-S6 compared to control CD8 T cells ( Figures 8A, B ). There was also a significant decrease in T-bet⁺ POSH cKO CD8 T cells ( Figures 8C, D ), confirming their impaired differentiation into T effector cells. Interestingly, Eomes⁺ POSH cKO CD8 T cells were also reduced in POSH^fl/fl CD4-Cre cultures ( Figures 8E, F ). While unexpected, the reduced Eomes expression—likely mediated by the observed NF-κB defects ( Figure 5 )—may explain the poor survival of the early-formed MPEC population during the immune response (2, 10). Together, these results suggest that POSH is critical for coordinating JNK, NF-κB and Akt/mTOR signaling pathways that promote SLEC and MPEC differentiation, proliferation, and survival.

POSH^fl/fl mice were generated by the University of Missouri Transgenic Mouse facility from Sh3rf1^{tm1a(EUCOMM)Hmgu} KO first allele (reporter-tagged insertion with conditional potential) from the International Mouse Phenotype constortium (IMPC). Transgene positive mice were crossed with FLPo mice (Jackson Labs) to delete FRT flanked luciferase/neo cassette and backcrossed >9 generations to create POSH^fl/fl mice ( Figure 1A ).

POSH^fl/fl mice were then crossed with Rosa26-STOP-eGFP, CD4-Cre mice, GzmB-Cre and OT-I mice (all strains: Jackson Labs) and are kept homozygous for POSH^fl/fl and Rosa26-STOP-eGFP alleles, and hemizygous for Gzmb-Cre, CD4-Cre and OT-I alleles. Mice were maintained in our animal facilities at the University of Missouri. Animal procedures were in accordance with Intuitional Animal Care and Use Committee regulations.

Spleens from OT1 POSH^fl/fl, OT1 POSH^fl/fl CD4-Cre, and OT1 POSH^fl/fl GzmB-Cre were harvested and processed into single cell suspensions. For POSH^fl/fl and POSH^fl/fl CD4-Cre, cells were stained with α-CD8 and CD8+GFP- or CD8+GFP+ cells were sorted via Cytek Aurora CS. For POSH^fl/fl GzmB-Cre, cells were stimulated with 20 nM OVA peptide, treated with 1x murine recombinant IL-2, and harvested on day 3 post stimulation. Cells were stained with α-CD8 and CD8+GFP- and CD8+GFP+ cells were sorted via Cytek Aurora CS.

1 million cells were lysed and RNA was harvested according to manufacturer instructions (PureLink RNA Mini Kit ThermoFisher 12183018A). RT PCR was performed using SuperScript IV VILO Master Mix (ThermoFisher 11756050) according to manufacturer instructions. Finally, a PCR on the cDNA to identify the presence or absence of exon seven was run using the following primers: FWD – GTG ACC CCA CCC CCT AGC REV - CCT GTG ACG GGC GCC ACA.

Cells were lysed using RIPA buffer containing leupeptin, aprotinin, PMSF, Na₃VO₄, and NaF for 30 minutes on ice. Cells were spun at 13,000 rpm for 30 minutes followed by the addition of 4x Laemmli Sample buffer (Bio-Rad 1610747) containing 2-mercaptoethanol (Sigma 7522). Lysates from 2 million cells per well were run on a 10% SDS-PAGE gel, transferred to a membrane, and blocked in 5% Bovine Serum Albumin (Fisher BP-1600-100) in TBST for 1 hour. Membranes were probed with primary antibody diluted in 5% BSA in TBST at 4C overnight. Membranes were washed then probed with secondary antibodies in 5% BSA in TBST at room temperature for 1 hour. Membranes were imaged on the Bio-Rad ChemiDoc MP Imaging System and images were analyzed using ImageJ.

Tissues harvested were processed into single-cell suspensions by smashing the tissue between two pieces of mesh using a 3 mL syringe. Cells were washed in 1xPBS then resuspended in 1x ACK red blood cell lysis buffer for 3 minutes followed by an additional wash. Cells were subsequently blocked in Fc blocker for 15 minutes at room temperature and washed in FACS buffer. Cells were then stained extracellularly with a cocktail of antibodies prepared in FACS buffer and incubated in the dark on ice for 30–40 minutes. For thymic cells, the CCR7 antibody was added separately and incubated at 37C for 30–40 minutes. Cells were then washed twice and resuspended in FACS buffer for flow cytometry. Cell counts were obtained using 123count eBeads (Invitrogen 01-1234-42). For intracellular staining, the BD Pharmagen Transcription Factor set was utilized following manufactures instructions. All flow cytometry was performed on the Cytek Aurora.

For phospho-S6 analysis, cells were stained extracellularly as described above. Cells were then fixed in 2% PFA at room temperature for 20 minutes then permeabilized in 90% ice cold methanol for 20 minutes at –20C. Cells were then stained with phospho-S6 antibody diluted in 0.5% BSA in 1xPBS for 30 minutes in the dark at room temperature. Cells were washed twice followed by analysis by flow cytometry.

Spleens from 6-12-week-old mice were harvested and processed into single-cell suspensions in complete RPMI 1640 media (1mM sodium pyruvate, 1x MEM non-essential amino acids solution, 100units/mL penicillin, 100units/mL streptomycin, 2mM L-glutamine, 0.02mM 2-mercaptoethanol, and 10% heat-inactivated FBS) at a concentration of 5 million/mL. Polyclonal cells were stimulated with 1 ug/mL α-CD3 and 1 ug/mL α-CD28 and incubated at 37C for 4 days. OT1 cells were stimulated with 20 nM OVA peptide and incubated at 37C for 4 days. At 24 hours, murine recombinant IL-2 (50 U/mL) was added to the cell culture. A portion of cells were taken each day for phenotyping analysis by flow cytometry as described above.

For cytokine production, cells were stimulated as described above. On day 2 post stimulation, cells were restimulated with 10 ug/mL PMA and 100 ug/mL ionomycin in the presence of Golgi-Plug (BD Biosciences) for 5 hours. Cells were then harvested and stained as described above.

Splenocytes from 6–12-week-old OT1 POSHfl/fl, OT1 POSHfl/fl CD4-Cre, and OT1 POSHfl/fl GzmB-Cre (CD45.2) mice were harvested and processed into single cells suspensions in 1x PBS+1%FBS. 100,000 CD8 T cells were adoptively transferred via retro orbital injection into B6 (CD45.1) mice. The next day, recipient mice were infected with 2 million PFU of VSV-OVA via tail vein injection. At indicated time points, tail bleeds were performed to collect blood for phenotyping by flow cytometry as described above. At the endpoint of the experiment, blood, spleen, lymph nodes, and kidney were harvested and processed into single-cell suspensions. Cells were then stained on ice for 30–40 minutes, washed, and resuspended in FACS buffer. All flow cytometry was performed on the Cytek Aurora.

EG7-OVA cells were cultured in complete RPMI 1640 media (1mM sodium pyruvate, 1x MEM non-essential amino acids solution, 100units/mL penicillin, 100units/mL streptomycin, 2mM L-glutamine, 0.02mM 2-mercaptoethanol, and 10% heat-inactivated FBS). Cells were maintained in an incubator at 37C and 5% CO₂ and cell media was changed every 2 days.

Splenic CD8 T cells were harvested from 6–12-week-old mice and stimulated as described above. On day 3 post stimulation, 6 million EG7-OVA target cells were harvested and stained with Tag-It Violet Proliferation and Cell Tracking Dye (Biolegend) according to manufacture instructions. 6 million CD8 T cells were harvested and co-cultured with stained EG7-OVA target cells at indicated effector:target ratios (E:T) for 5 hours. Cells were then washed, then stained with 7-AAD in the dark on ice for 30 minutes. Target cell death was analyzed by flow cytometry performed on the Cytek Aurora.

p values were calculated with paired two-tailed T test, one-way ANOVA with Tukey post-hoc test, or two-way ANOVA with Tukey multiple comparisons as denoted in figure legends. All analyses were performed with Prism 9 software (GraphPad).

The 123count eBeads (01–1234–42), BSA (BP1600-100), PureLink RNA Mini Kit (12183018A), SuperScript IV VILO Master Mix (11756050), and Live/Dead™ Aqua (L34957) were purchased from ThermoFisher.

CD4 BV785 (RM4-5, 100552), CD4 PE Fire 640 (GK1.5, 100482), CD4 PerCP-Cy5.5 (RM4-5, 100540), CD8 BV421 (53-6.7, 100756), CD8 BV785 (53-6.7, 100750), TCRβ BV605 (H57-597, 109241), CD69 APC-Cy7 (H1.2F3, 104526), CD25 APC-Cy7 (PC61, 102025), CD62L PE-Cy7 (MEL-14, 104418), CD44 BV605 (IM7, 103047), CD62L PE-Cy7 (MEL-14, 104418), CD25 APC-Cy7 (PC61, 102025), CD69 AF700 (H1.2F3, 104539), CD127 BV421 (A7R34, 135027), CD45.2 BV711 (104, 109847), KLRG1 PE/Dazzle 594 (2F1, 138424), Ki67 BV650 (11F6, 151215), Streptavidin BV650 (405232), and Tag-It Violet Proliferation and Cell Tracking Dye (425101) were purchased from BioLegend.

CD8 PerCP-Cy5.5 (53-6.7, 551162), CCR7 APC (4B12, 17-1971-81), MHC1 PE (AF6-88.5, 553570), CD69 PE-Cy7 (H1.2F3, 552879), CD8 AF700 (53-6.7, 557959), CD8 PerCP-Cy5.5 (53-6.7, 551162), CD45.2 AF700 (104, 56–0454–82), and S6 (pS235/pS236) (560433) were purchased from BD Biosciences.

α-POSH (YY5, sc-100815) and α-p-JNK Thr183/Tyr185 (sc-6254) were purchased from Santa Cruz Biotechnology.

α-p-p65 (S536) (3033), α-p65 (6956), α-p-Akt S473 (4060), α-Akt (2920), α-JNK (9252), α-p-IκBα S32 (2859), and α-IκBα (4814) were purchased from Cell Signaling.

Bio Rad.

α-actin (12004163), α-mouse 700 (12004158), and α-rabbit 520 (12005869) were purchased from Bio-Rad.

OVA peptide (SIINFEKL) was purchased from Biosynth.