Response to Comment on “The Common R71H-G230A-R293Q Human TMEM173 Is a Null Allele”

In their “Letter to the Editor,” Sivick et al. (1) disputed our results and conclusion. They claimed human R71H-G230A-R293Q (HAQ) STING is “fully functional.” We stand by our results and conclusion. Below are our responses to their comments.

Human PBMC studies

In contrast to our human PBMC (hPBMC) studies, Sivick et al. (1) claimed HAQ/HAQ cells act “in a matter indistinguishable from [wild type] WT/WT donor cells.” A previous genomewide association study identified a single-nucleotide polymorphism (rs13181561, ~4.2kb to the TMEM173 gene) associated with decreased IFN-α production in Europeans (Fig. 1A) (2). Upon vaccinia virus stimulation, hPBMC from the GG genotype of rs13181561 (a total 28 individuals) have a more than 10-fold decrease in IFN-α production in comparison to AA (272 individuals) or AG genotype (192 individuals) (p value: 2.71E-14) (2). We analyzed the relationship between rs13181561 and HAQ in Europeans (Fig. 1B). All HAQ/HAQ individuals have the GG genotype whereas all R232/R232 individuals are AA or AG of rs13181561 (Fig. 1B). The median IFN-α value of the difference between the unstimulated and the stimulated samples is 5.6 for the GG genotype in comparison to 88.4 for the AA genotype and 51.5 for the AG genotype of rs13181561 (2). This large scale hPBMC analysis agrees with our conclusion that the human HAQ allele is defective.

FIGURE 1.

HAQ is a null TMEM173 allele. (A) The panel illustrates the single-nucleotide polymorphism rs13181561 near the human TMEM173 gene, ~4.2kb downstream of the TMEM173 gene. (B) An analysis of 1000 Genomes Project showing, in Europeans, all HAQ/HAQ individuals have the GG genotype of rs13181561, whereas all R232/R232 individuals have the AA (243 individuals) or AG genotype (8 individuals) of rs13181561. (C) STING expression was determined by Western blot in the human B cells as before (3). Ethnic origins were indicated. R232-AA: homozygous R232 with the AA genotype of rs13181561; R232-GG: homozygous R232 with GG of rs13181516; HAQ-GG: homozygous HAQ with the GG of rs13181561; AQ-GG, homozygous AQ with the GG of rs13181561; AQ-AA, homozygous AQ with the AA of rs13181561; Q293-AA, homozygous Q293 with the AA of rs13181561 (n > 3). (D) TMEM173 mRNA was measured by quantitative real-time PCR in the indicated cells (n = 3). (E and F) Cells from indicated samples were activated with RpRp-ssCDA (5 μg/ml) for five hours in culture. Nuclear fractions were isolated and probed as before (2) (E) and IFN-β was determined in the supernatant by ELISA (F)(n = 3). Graph presents means ± SE from three independent experiments. *p < 0.05.

The small scale hPBMC study from Sivick et al. (1) had great variability and is inconclusive. Noticeably, there is no information regarding the ethnic background of their hPBMC samples. Our hPBMC study, showing a significant difference, was done in German R232/R232 and HAQ/HAQ individuals (3). If the hPBMC samples in the study by Sivick et al. (1) were from different ethnic groups, this could contribute to the high variability in their study.

Decreased STING expression in HAQ/HAQ individuals

Sivick et al. (1) claimed “STING is expressed similarly in these cells at the RNA and protein level.” Their intracellular stain lacked a positive control (e.g., R232/R232 cells), and western blots lacked a negative control (e.g., STING^−/− THP-1 cells). Critically, the low STING expression in human HAQ/HAQ cells is due to R71H-G230A-R293Q changes and decreased TMEM173 gene transcripts caused by a mutation in linkage disequilibrium with the human HAQ allele (3). We showed that despite the same TMEM173 genotype (G230A-R293Q, AQ), the AQ cells with the GG allele of rs13181561 (AQ-GG) have much lower STING expression than the AQ cells with the AA allele of rs13181561 (AQ-AA) (Fig. 1C). The AQ-GG cells also have decreased TMEM173 transcript (Fig. 1D) and did not respond to RpRp-ssCDA stimulation whereas the AQ-AA cells do (Fig. 1E, 1F). Thus, AQ-GG is also a null TMEM173 allele. Noticeably, Sivick et al. (1) found an HAQ/AQ individual (donor no. 8), along with two HAQ/HAQ individuals, have low STING expression in CD4⁺ T cells, monocytes and dendritic cells when compared against the rest of their samples, supporting our conclusion that HAQ has decreased STING expression.

Human B cell lines

Our human B cells studies were well-controlled, comparing HAQ/HAQ B cells to the R232/R232 B cells (3). We obtained these lines from National Human Genome Research Institute Repository and used them at a low passage. To our knowledge, our study is the first to use these 1000 Genomes Project–derived lymphoblastoid lines to study STING function. Our data showed these low passage lymphoblastoid cell lines express STING, respond to cyclic dinucleotide (CDN), activate IRF3 and produce inducible IFN-β (3), and they are thus suitable for CDN-STING study.

The HAQ mouse

To our knowledge, our knock-in mouse expressing a mouse equivalent of the HAQ allele (mHAQ) mouse is the only mouse model so far for any human STING variant. Our mHAQ mouse recapitulates the main feature of human HAQ (decreased STING expression and lack of response to CDN), thus indicating it is a valid mouse model for the human HAQ. We will share our mouse with the scientific community (with a material transfer agreement). As for the amino acid labels in Supplemental Figure 3, they indicated the location within an equivalent human HAQ. To avoid confusion, we have changed these labels from A230 (human STING) to A229 (mouse STING) and Q293 (human STING) to Q292 (mouse STING) in a Correction (4) to this figure. We have also provided new sequencing results for the A229 and Q292 changes of the mHAQ mouse in response to criticism from Sivick et al. (1) regarding our mHAQ mouse. Please see the accompanying Correction (4) to our original article (3).

Regarding previous HAQ publications

As we stated in the Discussion section of our original article (3), the only previous HAQ hPBMC study was done using one HAQ/HAQ and one R232/R232 individual with no information on ethnic background (5). Other HAQ studies used HEK293T cells stably expressing human HAQ cDNA (6–9). The STING level is much higher in these overexpressed cells than the endogenous level (3). Nevertheless, two such studies saw ~50% reduction of CDN responses in HAQ expressing HEK 293T cells when compared with R232 expressing HEK 293T cells (7, 8). Other mentioned studies used the THP-1 cells (10, 11) and did not compare HAQ with R232.

THP-1 cells

The THP-1 line was derived from a Japanese line in 1980 (12). It is not surprising the THP-1 cells are positive for HAQ because ~63% of the East Asians have at least one copy of the HAQ allele (3). However, results solely based on a cell line should be interpreted very carefully (13). We speculated that, like the HEK293T cells overexpressing HAQ, the THP-1 cells might have high STING expression that making them respond to CDN.

In summary, we stand by our conclusion that the human HAQ STING is a null allele. This conclusion is supported by data from well-controlled primary human cells (hPBMC), low-passage human B cell lines and a mouse model of HAQ.

Abbreviations used in this article:

CDN

cyclic dinucleotide

HAQ

R71H-G230A-R293Q

hPBMC

human PBMC

mHAQ

knock-in mouse expressing a mouse equivalent of the HAQ allele

WT

wild type