Gene editing of SAMHD1 in macrophage-like cells reveals complex relationships between SAMHD1 phospho-regulation, HIV-1 restriction, and cellular dNTP levels

ABSTRACT Sterile α motif and HD domain-containing protein 1 (SAMHD1) is a dNTP triphosphate triphosphohydrolase (dNTPase) and a potent restriction factor for immunodeficiency virus 1 (HIV-1), active in myeloid and resting CD4+ T cells. The anti-viral activity of SAMHD1 is regulated by dephosphorylation of the residue T592. However, the impact of T592 phosphorylation on dNTPase activity is still under debate. Whether additional cellular functions of SAMHD1 impact anti-viral restriction is not completely understood. We report BLaER1 cells as a novel human macrophage HIV-1 infection model combined with CRISPR/Cas9 knock-in (KI) introducing specific mutations into the SAMHD1 locus to study mutations in a physiological context. Transdifferentiated BLaER1 cells harbor active dephosphorylated SAMHD1 that blocks HIV-1 reporter virus infection. As expected, homozygous T592E mutation, but not T592A, relieved a block to HIV-1 reverse transcription. Co-delivery of VLP-Vpx to SAMHD1 T592E KI mutant cells did not further enhance HIV-1 infection indicating the absence of additional SAMHD1-mediated anti-viral activity independent of T592 dephosphorylation. T592E KI cells retained dNTP levels similar to WT cells indicating uncoupling of anti-viral and dNTPase activity of SAMHD1. The integrity of the catalytic site in SAMHD1 was critical for anti-viral activity, yet a poor correlation between HIV-1 restriction and global cellular dNTP levels was observed in cells harboring catalytic core mutations. Together, we emphasize the complexity of the relationship between HIV-1 restriction, SAMHD1 enzymatic function, and T592 phospho-regulation and provide novel tools for investigation in an endogenous and physiological context. IMPORTANCE We introduce BLaER1 cells as an alternative myeloid cell model in combination with CRISPR/Cas9-mediated gene editing to study the influence of sterile α motif and HD domain-containing protein 1 (SAMHD1) T592 phosphorylation on anti-viral restriction and the control of cellular dNTP levels in an endogenous, physiologically relevant context. A proper understanding of the mechanism of the anti-viral function of SAMHD1 will provide attractive strategies aiming at selectively manipulating SAMHD1 without affecting other cellular functions. Even more, our toolkit may inspire further genetic analysis and investigation of restriction factors inhibiting retroviruses and their cellular function and regulation, leading to a deeper understanding of intrinsic anti-viral immunity.

Only SAMHD1 dephosphorylated at residue T592 is active against HIV-1 (19-21).SAMHD1 is phosphorylated in cycling cells by cyclin dependent kinases CDK1 and CDK2 in complex with Cyclin A2 in S and G 2 /Mphase (19,21).At mitotic exit, SAMHD1 is rapidly dephosphorylated at residue T592 due to the action of the PP2A-B55α phosphatase complex (8).While the effect of SAMHD1 T592 phosphorylation on HIV-1 restriction is consistently demonstrated, the consequence for its dNTPase activity is still under debate.Biochemical approaches to measure the effect of SAMHD1 T592 phosphorylation and phosphomimetic mutants on SAMHD1 tetramer formation and dNTPase activity have not been able to reveal a functional relationship (12,(20)(21)(22)(23).Still, cell cycle dependent SAMHD1 phosphorylation, loss of HIV-1 restriction and increased dNTP levels in S-and G 2 /M phases in synchronized HeLa cells show a clear timely correlation (8).In contrast, mutagenic analysis of T592 site in myeloid cells challenges a causative link.Phosphoabla tive T592A or T592V, but not phosphomimetic T592E or T592D mutants, were able to inhibit HIV-1 replication, when overexpressed in PMA-activated U937 cells (12,20,21,24).Conversely, not only phosphoablative but also phosphomimetic SAMHD1 T592 mutants efficiently limited the cellular dNTP pool (14,20,21).This obvious discrepancy might be due to biological reasons [for details, refer to discussion and review in reference (1)].However, technical limitations might also be the cause for this problem.
Genetic studies of SAMHD1 phospho-mutants in myeloid cells are currently limited to PMA-activated macrophage-like THP-1 or U937 cells.As treatment with PMA can activate non-physiological intracellular pathways (25,26), alternative myeloid models are needed, which ideally would be both genetically amendable and based on physiological myeloid differentiation pathways.
So far, anti-viral restriction has been tested with mutant constructs of SAMHD1 using lenti-or retroviral transduction.In this case, an exogenous promotor mediates overex pression of SAMHD1.The use of CRISPR/Cas9 allowed us to modify SAMHD1 within the native genetic environment and to analyze the impact of selected mutations on anti-viral restriction in a physiological context, avoiding potential unwanted effects of mutant protein overexpression.
Here, we use CRISPR/Cas9-mediated knock-in (KI) in combination with transdifferen tiated macrophage-like BLaER1 cells as a tool to study the impact of SAMHD1 T592 phosphorylation on HIV-1 restriction and dNTP pools in myeloid cells.Transdifferenti ated macrophage-like BLaER1 cells expressed SAMHD1, which was dephosphorylated at residue T592.Concomitantly, transdifferentiated BLaER1 cells restricted HIV-1 replica tion in a SAMHD1 dependent manner.Introduction of SAMHD1 homozygous T592E mutations via CRISPR/Cas9 KI led to loss of HIV-1 restriction, while SAMHD1 T592A mutants maintained their anti-viral activity.Interestingly, HIV-1 infection was not further enhanced by SAMHD1 T592E mutant depletion suggesting the absence of an additional anti-viral activity of SAMHD1 that is independent of T592 dephosphorylation.This highlights the T592 phospho-site as the critical residue for anti-viral activity.Remarkably, neither endogenous SAMHD1 T592E nor T592A mutants had an impact on cellular dNTP levels in transdifferentiated BLaER1 cells, indicating that the regulation of anti-viral and dNTPase activity of SAMHD1 can be uncoupled.However, mutagenic analysis of the catalytic residues H210, D218, and D311 highlights the importance of the integrity of the catalytic site for anti-viral restriction.Yet, we observed again a lack of correla tion between cellular dNTP levels and HIV-1 restriction potential, indicating that the relationship between SAMHD1 function, HIV-1 restriction, and T592 phospho-regulation is complex.Also, we emphasize that regulation of dNTP levels is neither sufficient nor necessary for SAMHD1-dependent HIV-1 restriction in macrophage-like BLaER1 cells.

SAMHD1 is dephosphorylated at residue T592 in macrophage-like BLaER1 cells
Myeloid models to study HIV-1 restriction by mutagenesis are very limited.Transdifferen tiated BLaER1 cells are a novel myeloid cell model, which has successfully been used to study innate immune signaling in macrophage-like cells (27,28).The native, B-line age-derived BLaER1 cells undergo macrophage transdifferentiation by induction of the myeloid transcription factor C/EBPα.Transdifferentiated BLaER1 cells have been shown to closely resemble human macrophages with respect to mRNA expression, cell cycle arrest, and immune functions (28,29).In order to test whether these cells can serve as a model to study SAMHD1-mediated anti-viral restriction, we analyzed SAMHD1 expres sion in transdifferentiated BLaER1 cells.Flow cytometry analysis of transdifferentiated BLaER1 cells showed loss of B cell marker CD19 and acquisition of surface expression of the macrophage marker CD11b (Fig. 1A), as demonstrated earlier (28).Transdifferentia tion of BLaER1 cells, using an adopted protocol, was highly reproducible and yielded 89.3 ± 8.8% (n = 33) of CD19 − CD11b + cells in viable BLaER1 cells expressing GFP (Fig. 1B).In addition, transdifferentiated BLaER1 cells expressed monocyte-derived macrophage and dendritic cell markers CD14, CD163, CD206, and CD11c (Fig. 1C) highlighting the myeloid phenotype of these cells and validating previous results based on mRNA expression (28,30,31).Interestingly, transdifferentiated BLaER1 cells displayed HIV-1 entry receptor CD4 expression and high surface level expression of both co-receptors CXCR4 and CCR5 indicating that they may be amenable to infection with CCR5-and CXCR4-tropic HIV-1 (Fig. 1C).Transdifferentiated BLaER1 cells expressed levels of SAMHD1 comparable to cycling THP-1 cells (Fig. 1D), whereas native BLaER1 cells showed no SAMHD1 expression.As T592 phosphorylation in SAMHD1 is the major regulator of anti-viral restriction (8,19,21), we analyzed the phosphorylation status in transdifferentiated BLaER1 cells.Relative SAMHD1 T592 phosphorylation was 31-fold lower in transdifferentiated BLaER1 compared to cycling THP-1 cells (0.032 ± 0.013 relative SAMHD1 pT592 normalized to cycling THP-1, n = 6) and in fact was barely detectable by immunoblotting even after long exposure times (Fig. 1D).Absence of SAMHD1 pT592 correlated well with the reported G 1 /G 0 cell cycle arrest in transdifferentiated BLaER1 cells (8,28), as well as with low Cyclin A2 expression (Fig. 1D), which in complex with CDK1 and CDK2 is known to mediate T592 phosphorylation (19).Thus, macrophage-like transdifferentiated BLaER1 cells expressed SAMHD1 dephosphorylated at residue T592, suggesting it to be anti-virally active.

SAMHD1 restricts HIV-1 in macrophage-like BLaER1 cells
To define the restrictive capacity of SAMHD1 in the context of transdifferentiated BLaER1 cells, we infected the cells with a single-cycle HIV-1 luciferase reporter virus (HIV-1-luc), in presence or absence of virus-like particles containing Vpx (VLP-Vpx).VLP-Vpx treatment led to efficient degradation of SAMHD1 (0.013 ± 0.007 relative SAMHD1 expression normalized to no VLP-Vpx, n = 3) (Fig. 2A) and increased HIV-1-luc infection.Linear regression revealed a significant (P = 0.0125, n = 3, unpaired t-test) increase over a wide range of MOIs (Fig. 2B).To validate this further, we generated SAMHD1 knock-out (KO) BLaER1 cells using CRISPR/Cas9 ribonucleoprotein (RNP).Three independent SAMHD1 KO BLaER1 single cell clones were analyzed in detail and showed bi-allelic InDels at the intended target site (Fig. 2C) leading to a frameshift, the introduction of premature stop codons and therefore, absence of SAMHD1 expression in transdifferentiated BLaER1 cells (Fig. 2D).While SAMHD1 KO did not affect BLaER1 transdifferentiation (Fig. 2E), it strongly increased HIV-1-luc infection at 24 hpi, as compared to wild-type (WT) cells.Significance of differences in the slopes of linear regressions are suggesting SAMHD1 to be a major restriction factor in these cells over a wide range of MOIs (P < 0.0001 for Clone #1, 2, and 3, n = 7, one-way ANOVA) (Fig. 2F).In order to rule out a potential confounding effect of a minor CD11b − native-like population, we developed a flow cytometry workflow combining the use of a single-cycle HIV-1 mCherry (HIV-1-mCherry) reporter virus together with staining for viable CD11b + cells.Thereby, we could specifically analyze infection in transdifferentiated CD11b + macrophage-like BLaER1 cells.HIV-1-mCherry infection, as measured by %mCherry + cells in CD11b + viable GFP + transdifferentiated BLaER1 cells, was strongly increased upon SAMHD1 KO at 24 hpi (Clone #1 P = 0.3633, #2 P = 0.0360, #3 P = 0.0013, n = 5, Kruskal-Wallis test) (Fig. 2G and H).This indicates Representative immunoblot analysis of SAMHD1, Cyclin B1, and Cyclin A2 expression in native (N) and transdifferentiated (td) BLaER1 cells, as well as cycling THP-1 cells.GAPDH serves as a loading control.Mean signal of SAMHD1 T592 phosphorylation (pT592) relative to total SAMHD1 expression in transdifferentiated BLaER1 cells was normalized to cycling THP-1 (n = 6).that SAMHD1 is a major anti-lentiviral restriction factor in macrophage-like transdifferen tiated BLaER1 cells.We therefore conclude that transdifferentiated BLaER1 cells are an excellent model to study SAMHD1-mediated HIV-1 restriction.

A pipeline to generate mutants of SAMHD1 by CRISPR/Cas9-mediated knock-in
So far, mutagenic analysis of SAMHD1 has been limited to model systems in which SAMHD1 is overexpressed by transient transfection or retroviral transduction.Overex pression of SAMHD1, especially in the context of phosphomimetic T592E or phosphoa blative T592A mutation and their effect on viral restriction and intracellular dNTP levels, might affect functional readouts due to non-physiological expression levels, abnormal genomic context, and altered post-translational regulation (1).To overcome this challenge, we decided to introduce SAMHD1 point mutations directly into the SAMHD1 gene locus by CRISPR/Cas9 KI.Therefore, we developed a pipeline based on the introduction of RNPs and single-stranded DNA correction templates by electro poration, followed by an allelespecific PCR (KASP-genotyping assay screening) and rigorous validation by Sanger sequencing and quantitative genomic PCR to exclude large genomic deletions (qgPCR) (32) (Fig. 3A).We identified single cell clones, displaying homozygous introduction of T592A and T592E mutations into the SAMHD1 locus of BLaER1 cells (Fig. 3B).Quantification of allele numbers of SAMHD1 exon 16 revealed that the majority of homozygous single cell T592A and T592E KI clones still contained two alleles of SAMHD1 exon 16 (Fig. 3C).However, we could identify 1 out of 8 clones analyzed (Clone X), which showed loss of one allele in qgPCR, indicative of pseudohomozygosity (32).In total, we were able to generate and validate two homozygous T592A, as well as three homozygous T592E BLaER1 KI mutants, corresponding to a homozygous KI frequency of ~1% and highlighting the necessity of KASP-screening to reduce the number of KI candidates (Fig. 3D).Expression of SAMHD1 mutants in transdifferentiated T592A or T592E KI BLaER1 single cell clones was at similar level compared to WT protein in the respective parental cell line (Fig. 3E).SAMHD1 KI had no negative impact on BLaER1 transdifferentiation (Fig. 3F).In summary, using our pipeline we introduced homozygous T592A and T592E mutations into the endogenous SAMHD1 locus of BLaER1 cells without affecting SAMHD1 expression or BLaER1 transdifferentia tion into macrophage-like cells.

Homozygous SAMHD1 T592E mutation increases HIV-1 infection in transdif ferentiated BLaER1 cells
We infected several clones of transdifferentiated homozygous SAMHD1 phosphoablative T592A and phosphomimetic T592E KI BLaER1 cell mutants with HIV-1-mCherry reporter virus and measured the fold change of %mCherry + cells in CD19 + viable GFP + cells relative to infection in WT cells.In all three clones, homozygous SAMHD1 T592E mutation significantly increased HIV-1-mCherry infection up to 31-fold (T592E/T592E Clone #1 P = 0.0017, #2 and #3 P <0.0001, n = 3, one-way ANOVA) (Fig. 4A and B).In contrast, SAMHD1 T592A KI mutants completely retained their restrictive potential in transdiffer entiated BLaER1 cell clones and behaved similar to WT BLaER1 cells upon challenge with HIV-1-mCherry (Fig. 4A and B).Using CRISPR/Cas9 KI, we were able to validate the loss of HIV-1 restriction in SAMHD1 phosphomimetic T592E mutants in macrophage-like cells.In this model, mutants of SAMHD1 are analyzed in the native genomic context and show physiological expression levels, confirming the role of T592 phosphorylation in the regulation of the anti-viral activity of SAMHD1.

SAMHD1 T592E knock-in relieves a block to HIV-1 reverse transcription and HIV-1 infection, which is not further enhanced by SAMHD1 T592E mutant depletion
Previous results indicate that SAMHD1 inhibits HIV-1 replication at the step of reverse transcription (3,7,8).To better understand the effect of endogenous SAMHD1 T592E mutation on HIV-1 replication, we measured the abundancy of HIV-1-mCherry reverse transcription products in CD19-depleted transdifferentiated BLaER1 cells.While late reverse transcription product copy numbers in WT BLaER1 cells stayed low until 24 hpi, both SAMHD1 KO and SAMHD1 T592E KI mutant BLaER1 cells showed a strong increase in HIV-1 late reverse transcription product copy numbers starting from 9 hpi (Fig. 6A).This indicates that endogenous mutation of SAMHD1 T592E in transdifferenti ated BLaER1 cells relieves a block to HIV-1 replication which is situated at or before the step of reverse transcription.Several restriction mechanisms have been proposed for SAMHD1, notably dNTP degradation, RNAse activity, and nucleic acid binding (1, 10, 24, 34-39).To understand if additional SAMHD1-mediated, but T592 dephosphorylation independent anti-lentiviral mechanisms could influence HIV-1 replication in transdiffer entiated BLaER1 cells, we tested whether SAMHD1 mutant depletion using VLP-Vpx would further enhance HIV-1 replication.As anticipated, VLP-Vpx-mediated SAMHD1 depletion (Fig. 6B) in WT cells significantly increased the percentage of HIV-1-mCherry infected CD11b + macrophage-like BLaER1 cells, while there was no effect on SAMHD1 KO cells, when compared to empty VLP-treated control cells (Fig. 6C; WT P < 0.0001, KO Clone #2 P = 0.9615, n = 3, one-way ANOVA).In contrast, depletion of endogenous T592A mutant SAMHD1 significantly increased HIV-1 infection rates (Clone #1 and #2 P < 0.0001, n = 3, one-way ANOVA).VLP-Vpx treatment in SAMHD1 T592E KI mutant clones, however, had no effect on HIV-1 infection rates (Fig. 6C; Clone #2 P = 0.2654, #3 P = 0.1788, n = 3, one-way ANOVA).Thus, our data indicate that no T592 dephosphorylation independent anti-lentiviral mechanism is exerted by SAMHD1 in macrophage-like cells.

The integrity of the catalytic site in SAMHD1 is critical for anti-viral activity; however, antiviral activity correlates poorly with global cellular dNTP levels
Previous results indicate the loss of SAMHD1's anti-viral potential in mutants of key catalytic residues, as shown by overexpression of mutants such as combined H206A_D207A or the D311A mutant (2,(11)(12)(13).However, overexpression might introduce substantial experimental bias; therefore, we decided to validate the require ment of the catalytic dNTPase pocket by CRISPR/Cas9 knock-in.To do so, we modified the residues, H210, D218, and D311 in BLaER1 cells.These residues have been suggested to be directly or indirectly implicated in the triphosphohydrolase activity of SAMHD1 (40).The introduction of homozygous mutations was verified by Sanger sequencing and qgPCR (Fig. S1).The mutant proteins, with alanine substitution of the respective residues, were expressed in macrophage-like BLaER1 cells (Fig. 7A).While H210A showed similar or slightly higher expression compared to WT protein, D218A and D311A expressions were slightly reduced.Importantly, we did not see increased SAMHD1 pT592 in any of the mutant clones, indicating phospho-regulation of all three endogenous catalytic site mutants similar to WT (Fig. 7A).As expected, macrophage-like BLaER1 cells harboring SAMHD1 D311A mutation displayed a significant increase in cellular dNTP levels (dATP #1-4 P < 0.0001; dCTP #1 P = 0.0079, #2 P = 0.0003, #3 P = 0.1271, #4 P = 0.0169; dGTP #1-4 P < 0.0001; dTTP #1-4 P < 0.0001, n = 3, one-way ANOVA) (Fig. 7B through E), for dATP, dGTP, and dTTP even above the levels measured in SAMHD1 KO BLaER1 cell clone #2, indicating a complete loss of SAMHD1 dNTPase activity in D311A mutant protein.
The increase of dNTP levels in homozygous D210A and D218A mutant cells was much less pronounced and only consistently significant in D210A clone #2 and D218A clone #1 for dATP, dGTP, and dTTP (dATP D210A #2 P = 0.0025, D218A #1 P = 0.0724; dGTP D210A #2 P < 0.0001, D218A #1 P = 0.0003; dTTP D210A #2 P = 0.0052, D218A #1 P = 0.0457, n = 3, one-way ANOVA) with levels similar to SAMHD1 KO (Fig. 7B and C).Importantly, several cell clones, namely D210A #3 and 4, as well as D218A #2, revealed dNTP levels similar to or only marginally increased, when compared to WT or T592A mutant cells indicating that D210A and D218A mutations do not lead to complete loss of SAMHD1 dNTPase activity or alternatively, loss of control of dNTP levels could be compensated.Next, we infected transdifferentiated D210A, D218A, and D311A mutant cells with HIV-1-mCherry reporter virus at three different MOIs and quantified infected mCherry + cells in CD11b + BLaER1 cells after 24 h (Fig. 8).CRISPR/Cas9 D210A and D311A KI mutant cells showed a significant (P < 0.0001 for all clones, n = 3, one-way ANOVA) increase in infection level, which was consistent over a range of MOIs from 0.1 to 1 (Fig. 8B).Surprisingly, however, the loss of restrictive potential in the D218A mutant cell clones was strikingly lower and only significant for Clone #2 (#1 P = 0.1975, #2 P < 0.0001, n = 3, one-way ANOVA).When compared to SAMHD1 KO, HIV-1-mCherry infection at MOI1 was significantly lower in D218A (P < 0.0001, n = 3, one-way ANOVA), but similar in D210A and D311A mutants.Depletion of SAMHD1 D210A and D311A mutant protein by VLP-Vpx treatment did not further increase HIV-1-mCherry infection at MOI0.1 (Fig. 8C and D).In contrast, VLP-Vpx-mediated depletion of SAMHD1 D218A mutants allowed higher infection levels, which after depletion were similar to SAMHD1 KO cells (Fig. 8D).In summary, using CRISPR/Cas9 KI mutants of the SAMHD1 catalytic dNTP pocket, we were able to show that the residues H210 and D311 are critical for anti-viral restriction.
However, this seems not to be the case for D218, highlighting that the individual residues of the catalytic pocket might be differentially involved in HIV-1 restriction.Importantly, lack of correlation between cellular dNTP levels and HIV-1 restriction potential in D210A and D218A mutants, together with loss of HIV-1 restriction in dNTP low T592E mutants, indicates that maintenance of low global dNTP levels is a poor predictor of SAMHD1 anti-viral activity.

DISCUSSION
SAMHD1 is a major cellular dNTPase and a potent HIV-1 restriction factor (2-6, 10, 39).However, whether SAMHD1 dNTPase activity is meditating its anti-viral activity and how this is regulated by T592 phosphorylation is still a matter of debate (1,14).
To tackle this question with a novel toolkit, we used transdifferentiated BLaER1 cells as an alternative and versatile myeloid model to study HIV-1 infection in macrophagelike cells.Transdifferentiated BLaER1 cells closely resemble human macrophages and have successfully been used to study innate immune signaling (27,28).In contrast to PMA-activated THP-1 or U937 cells, BLaER1 cell transdifferentiation relies on the activa tion of the fusion protein of the macrophage transcription factor C/EBPα and the estradiol receptor leading to the induction of a myeloid cell program (28,29).Transdiffer entiated BLaER1 cells show surface expression of classical monocyte-derived macro phage and dendritic cell markers CD14, CD163, CD206, and CD11c, validating previous transcriptional data (28), as well as HIV-1 CD4, CCR5, and CXCR4 (co-) receptor expression.This cell system is an interesting physiological model for HIV-1 infection in macrophagelike cells, as we were able to show that SAMHD1 is completely dephosphorylated at residue T592 in transdifferentiated BLaER1 cells and serves as a major restriction factor for HIV-1.
To test mutants of SAMHD1 for anti-viral restriction activity in a physiological genetic and relevant cellular context, we combined transdifferentiated BLaER1 cells, as a novel myeloid model, with CRISPR/Cas9-mediated knock-in.We developed a gene editing strategy using CRISPR/Cas9 RNP and ssDNA oligo introduction by nucleofection, KASP screening of single cell clones and rigorous validation by sequencing and qgPCR (Fig. 3A).By introducing point mutations that correspond to phosphoablative T592A and phosphomimetic T592E mutations into the genomic SAMHD1 locus, we were able to genetically uncouple SAMHD1-mediated anti-viral restriction and cellular dNTPase activity.As expected, we showed that endogenous T592 phospho-mimetic mutants lose their anti-viral activity against HIV-1 in macrophage-like BLaER1 cells, while phosphoablative KI mutants maintained their anti-viral potential (Fig. 4).Furthermore, using endogenous mutants of T592, we highlighted that T592 phospho-regulation indeed affects anti-lentiviral activity of SAMHD1 at the step of reverse transcription (Fig. 6A).Thus, for the first time, we both validated and expanded our knowledge on the pheno typic consequences of SAMHD1 T592E phosphomimetic mutation, and hence the effect of T592 phosphorylation, for its anti-viral restriction activity in a myeloid model not based on overexpression.
Overexpression of SAMHD1 mutants in U937 background using retroviral transduc tion has several technical limitations.Even though PMA-activated U937 cells are often considered not to express SAMHD1, they actually can express small amounts of endoge nous SAMHD1, which can be further enhanced upon interferon treatment (41).Presence of endogenous WT SAMHD1 might affect the function of overexpressed mutant SAMHD1, especially if heterotetramers are formed.In addition, strong exogenous, often viral promotors drive SAMHD1 expression here, leading to high expression levels and non-physiological phosphorylation ratios, i.e., hyperphosphorylation (data not shown).In mutants generated by CRISPR/Cas9 KI, modified SAMHD1 is expressed in the physiologi cal genomic context from the endogenous promotor and thus under normal transcrip tional regulation.In BLaER1 cells, SAMHD1 expression is very low in native cells, but strongly induced upon transdifferentiation (Fig. 1D).This is also the case for SAMHD1 mutants.The use of CRISPR/Cas9 KIs avoids potential effects of constitutively expressed SAMHD1 mutants on cycling BLaER1 cells.
In macrophage-like WT BLaER1 cells, we measured dNTP levels and concentrations, which were similar or slightly lower than those found in resting T cells (Table 1) (33).After depletion of CD19 + incompletely transdifferentiated cells from bulk preparations of transdifferentiated BLaER1 cells, we were able to further reduce the levels of all dNTPs (Table 1) (33).Considering HIV-1 reverse transcriptase K m and K d values measured in vitro, this indicates that the dNTP concentrations found in transdifferentiated BLaER1 cells are sufficiently low to restrict or delay HIV-1 RT (33,(42)(43)(44).Concomitantly, SAMHD1 KO increased cellular dNTP concentrations in transdifferenti ated BLaER1 cells up to fourfold (Fig. 5A), which is reminiscent of the five to eightfold increase upon T cell activation (33).In stark contrast, however, neither endogenous SAMHD1 T592E nor T592A mutation increased cellular dNTP concentrations in transdif ferentiated BLaER1 cells (Fig. 5A).This indicates that the loss of restriction observed in endogenous T592E mutants is probably not caused by increased dNTP levels or reduced SAMHD1 dNTPase activity in transdifferentiated BLaER1 cells.This lack of correlation between cellular dNTP levels and HIV-1 restriction was also observed in a study published during revision, highlighting the possibility that T592 phosphorylation might indeed impact SAMHD1 tetramer stability (45).However, using our endogenous approach, we exclude artifacts of overexpression and the use of U937 cells as a model and thereby improve the physiological value of the conclusions.
In addition, we demonstrated that SAMHD1 T592E and T592A mutations had no consistent effect on dNTP pool composition in macrophage-like BLaER1 cells (Fig. 5B), ruling out an effect of the phosphomimetic mutation on SAMHD1 dNTPase substrate preferences and thus dNTP ratios, which was proposed earlier (46).More specifically, endogenous SAMHD1 T592E mutations did not increase cellular dCTP concentration (Fig. 5A).Taken together, mutagenic analysis of SAMHD1 residue T592 indicates that SAMHD1 dNTPase activity or substrate preference in transdifferentiated BLaER1 cells is not regulated by phosphorylation at this specific residue.Consequently, loss of HIV-1 restriction in SAMHD1 T592E mutants cannot be attributed to changes in SAMHD1 dNTPase activity.In addition, by combining SAMHD1 T592E KI mutants with VLP-Vpxmediated depletion of SAMHD1 in trans, we did not find an anti-lentiviral activity of SAMHD1 independent of SAMHD1 T592 dephosphorylation (Fig. 6B and C).
To address the relationship between SAMHD1 enzymatic function and its anti-lenti viral activity further, we introduced KI mutations for specific residues in the catalytic dNTPase pocket of SAMHD1 (Fig. 7A) (40).Loss of restrictive potential of H210A and D311A mutations in an endogenous context indicate that the integrity of the catalytic site is indeed required for HIV-1 restriction (Fig. 8A and B).While the loss of restriction with the endogenous H210A and D311A mutations is in concordance with previous data employing overexpression models (2,(11)(12)(13), two alternative hypotheses can be drawn from this observation: either the enzymatic SAMHD1 dNTPase activity per se is required for the inhibition of HIV-1 replication, or an (enzymatic) activity other than the canonical dNTPase activity, but still involving these residues is at play.
In favor of the first and, so far, most often discussed hypothesis speaks data from further studies showing that addition of exogenous dNs and a concomitant increase in cellular dNTP levels can leverage SAMHD1-mediated block to HIV-1 restriction (5,11).Also, in concordance with previous results, the D311A mutation increases cellular dNTPs to levels equal to or higher than those found in SAMHD1 KO macrophage-like BLaER1 cells (Fig. 7B through E), correlating HIV-1 restriction to SAMHD1 dNTPase activity.In addition, no D311 residue independent HIV-1 restriction potential was observed (Fig. 8C  and D).
In contrast, however, data on H210A and D218A mutants seem to be at odds with this hypothesis.First, while all SAMHD1 H210A mutant clones showed complete loss of HIV-1 restriction potential (Fig. 8B and D), enhancement of dNTP levels was less pronounced (Fig. 7B through E).In particular, clones #3 and #4 showed global dNTP levels comparable to WT or T592A mutant clones, indicating that the H210 residue is not essential for dNTP degradation and that loss of dNTPase activity can only be partial in H210A mutant cells.Even more striking is the phenotype of the D218A mutation.SAMHD1 D218 was recently shown to be involved in the triphosphohydrolase reaction (40).D218A mutation only leads to partial loss of dNTPase activity in vitro (40), which is in concordance with our data in cellulo, showing a clone-dependent significant increase of cellular dNTP levels, similar to what was found in H210A mutants (Fig. 7B through E).Importantly however, at least one of the D218A mutant cells maintained their potential to restrict HIV-1 replication (Fig. 8B and D).Thus, in the case of both H210A and D218A mutant cells, cellular dNTP levels, and presumably SAMHD1 dNTPase activity, do not entirely correlate with their anti-lentiviral capacity.
A possible explanation is that a still-to-be defined function, other than dNTP triphosphohydrolase, which also depends on the residues of the catalytic pocket, is required for HIV-1 restriction and that this function is directly or indirectly modulated by T592 phosphorylation.Ribonuclease (RNase) activity was proposed as a mechanism responsible for HIV-1 restriction (24,34,35).While initial reports of SAMHD1 associated RNase activity were strongly questioned by the community because of the reversible nature of SAMHD1 restriction and due to potential copurification of unknown RNase(s) (37,(47)(48)(49), recent evidence links SAMHD1 RNase activity to the inflammatory pheno type in Aicardi-Goutières syndrome (50).Catalytic residues, such as H206, D207, D311, and H167 were proposed to be critical for RNase activity (24,35,50), while reports on the role of T592 phospho-regulation are contradictory (24,50).The contribution of the proposed SAMHD1 RNase activity to HIV-1 restriction thus merits further investigation.
Nucleic acid binding and exo-/endonuclease recruitment could be an alternative anti-lentiviral restriction mechanism.SAMHD1 is participating in the resolution of stalled replication forks and homologous recombination by recruitment of endo-/exonuclease MRE11 and CtIP (16,18).Interestingly, SAMHD1 T592 phosphorylation is required for DNA end resection and resolution of stalled replication forks (16).However, crucial residues of the catalytic dNTPase pocket, such as H206, D207, and K312 of SAMHD1, seem dispensable for this process.Thus, endo-/exonuclease recruitment is unlikely to contribute to the phenotypes we observed in BLaER1 cells (16,18).
SAMHD1 was shown to bind to single-stranded RNA and DNA, as well as RNA or DNA with complex secondary structures in vitro (35,37).Whether nucleic acid bind ing contributes to the recruitment of endo-/exonucleases in cells or anti-viral activity remains an open question.Short phosphorothioate oligonucleotides can bind to the allosteric sites of SAMHD1, promoting formation of a distinct SAMHD1 tetramer (51).Mutations that abolished phosphorothioate oligonucleotide binding also reduced HIV-1 restriction potential.How SAMHD1 nucleic acid binding activity is regulated is currently not clear.SAMHD1 phosphomimetic T592E mutant had no influence on ssRNA and ssDNA binding in vitro (37).Intriguingly, ssDNA or ssRNA binding to SAMHD1's dimerdimer interface can inhibit the formation of catalytically active SAMHD1 tetramer and thus interferes with dNTPase activity (52).Even though the SAMHD1-nucleic acid binding interface seems to be distinct from the catalytic pocket (37,51), a deeper investigation is needed to understand this phenomenon and its implications for HIV-1 restriction in cells.
Another possible explanation could be that dNTPase activity per se is required for HIV-1 restriction, but global cellular dNTP levels do not correlate with this anti-viral activity.Sub-cellular dNTP levels, i.e., in the nucleus or at sites of HIV replication, such as the nuclear pore, nuclear speckles, or the capsid shell (53,54) could be modulated by the active site mutants, such as H210A and D311A.In reverse, T592 phosphorylation could regulate the sub-cellular distribution of SAMHD1.T592 phosphorylation, however, has no influence on cytoplasmic vs. nuclear localization of SAMHD1 (13).In addition, mutation of the n-terminal nuclear-localization signal (NLS) does not affect HIV-1 restriction ( 55).Yet, it is possible that T592 dephosphorylation alters SAMHD1 distribution, i.e., allows the recruitment of SAMHD1 to the capsid shell or into the sub-nuclear compartments in which HIV-1 reverse transcription occurs and thereby locally modulates the availability of dNTPs.
SAMHD1 regulation is certainly more complex than commonly assumed.In addi tion to multiple potential phosphorylation sites, SAMHD1 is modified by acetylation, SUMOylation, ubiquitination, and O-GlcNAcylation (20,21,(56)(57)(58)(59)(60)(61)(62)(63) and harbors redoxactive cysteines (64,65).Recently, SAMHD1 SUMOylation at residue K595 was shown to be required for HIV-1 restriction in PMA differentiated U937 cells.Overexpression of SAMHD1 mutants that abrogate restriction and SUMOylation at residue K595 did not show increased dATP levels, phenocopying the phosphomimetic T592 mutants of SAMHD1 (66).It will be interesting to investigate in more detail how T592 phosphoryla tion and K595 SUMOylation are integrated and it will be crucial to validate SAMHD1 (co-) regulation via diverse proposed post-translational modifications in physiological settings.Post-translational regulation of SAMHD1 might not only be achieved by the direct modification of single residues, but also by interaction partners, that could modulate or mediate SAMHD1 anti-viral activity.
A better understanding of SAMHD1 regulation in relevant HIV-1 target cells will improve our understanding of how SAMHD1 inhibits HIV-1 replication and which conditions determine SAMHD1's anti-viral capacity.

Cell lines
Human 293T/17 (ATCC No.: CRL-11268) cells were cultured in DMEM (Sigma-Aldrich) supplemented with 10% fetal calf serum (FCS; Sigma-Aldrich) and 2 mM L-glutamine (Sigma-Aldrich) at 37°C and 5% CO 2 .Human BLaER1 cells (a kind gift of Thomas Graf ) (28) were grown in RPMI (Sigma-Aldrich) supplemented with 10% FCS and 2 mM L-glutamine at 37°C and 5% CO 2 .For transdifferentiation, 1 × 10 6 BLaER1 cells per well of a six-well tissue culture plate were treated with 10 ng/mL human recombinant M-CSF and IL-3 (PeproTech) and 100 nM β-estradiol (Sigma-Aldrich) for 7 days.Half of the cell culture supernatant was replaced with medium containing cytokines and β-estradiol at days 2 and 6.All cell lines were free of mycoplasma contamination, as tested by PCR Myco plasma Test Kit II (PanReac AppliChem).

FIG 3 A
FIG 3 A pipeline to generate mutants of SAMHD1 by CRISPR/Cas9-mediated knock-in.(A) Schematic representation of CRISPR/Cas9 meditated knock-in (KI) to generate mutants of SAMHD1 in BLaER1 cells.Cas9 ribonucleoprotein (RNP) together with ssDNA correction template was introduced into BLaER1 cells via nucleofection.Single cell clones generated by limiting dilution were screened using KASP assay and validated by Sanger sequencing and quantitative genomic PCR (qgPCR) (32).(B) Representative sections of Sanger sequencing traces obtained from genomic SAMHD1 exon 16, highlighting successful bi-allelic single base exchange at the base triplet corresponding to amino acid position T592 in BLaER1 SAMHD1 KI T592A and T592E mutant single cell clones.No further mismatches were detected up-or downstream of shown section in the amplified region.Two independent sequencing runs were performed.Homozygous T592E mutants were additionally confirmed by allelespecific sequencing after TA-cloning.(C) Quantitative genomic PCR for SAMHD1 exon 16 against reference gene TERT was performed and 2 −Δct value obtained from SAMHD1 KI clones normalized to WT in order to obtain the allele number.As a control, half of the WT (WT 1/2) DNA was inoculated and Δct of SAMHD1 was calculated against ct of TERT, which was obtained in the WT with normal DNA amount.Error bars indicate standard deviation of technical triplicates in a representative experiment (n = 2).(D) Number of single cell clones obtained from CRISPR/Cas9 RNP and ssDNA correction oligo-treated BLaER1 cells and number of clones scoring positive in KASP assay, as well as homozygous (Mut/Mut) mutants identified by Sanger sequencing and confirmed by qgPCR are shown.(E) Transdifferentiated SAMHD1 KI BLaER1 cells were analyzed by immunoblot for SAMHD1 expression and compared to WT cells.GAPDH was used as a loading control (representative for n = 3).(F) Percentage of CD19 − CD11b + cells in viable GFP + transdifferentiated WT and SAMHD1 KI cells was quantified by flow cytometry.Error bars indicate standard deviation of biological replicates (n = 4).

TABLE 1
dNTP concentrations in transdifferentiated and CD19 depleted BLaER1 cells