Improved chemotherapy modeling with RAG-based immune deficient mice.

We have previously characterized an acute myeloid leukemia (AML) chemotherapy model for SCID-based immune deficient mice (NSG and NSGS), consisting of 5 days of cytarabine (AraC) and 3 days of anthracycline (doxorubicin), to simulate the standard 7+3 chemotherapy regimen many AML patients receive. While this model remains tractable, there are several limitations, presumably due to the constitutional Pkrdcscid (SCID, severe combined immune deficiency) mutation which affects DNA repair in all tissues of the mouse. These include the inability to combine preconditioning with subsequent chemotherapy, the inability to repeat chemotherapy cycles, and the increased sensitivity of the host hematopoietic cells to genotoxic stress. Here we attempt to address these drawbacks through the use of alternative strains with RAG-based immune deficiency (NRG and NRGS). We find that RAG-based mice tolerate a busulfan preconditioning regimen in combination with either AML or 4-drug acute lymphoid leukemia (ALL) chemotherapy, expanding the number of samples that can be studied. RAG-based mice also tolerate multiple cycles of therapy, thereby allowing for more aggressive, realistic modeling. Furthermore, standard AML therapy in RAG mice was 3.8-fold more specific for AML cells, relative to SCID mice, demonstrating an improved therapeutic window for genotoxic agents. We conclude that RAG-based mice should be the new standard for preclinical evaluation of therapeutic strategies involving genotoxic agents.

Introduction

Currently, NOD/SCID IL2Rγ-/- (NSG) mice are the most commonly used strain for engraftment of both normal and malignant human hematopoietic tissues. These mice represent a dramatic improvement over older strains for engraftment of normal HSCs [1] as well as AML and ALL cell lines and patient samples [2]. NSG mice with transgenic expression of human SCF/GM-CSF/IL-3 cytokines (NSGS) further improved AML engraftment efficiency, latency, and levels [3, 4]. Similarly, NRGS mice (NOD/RAG IL2Rγ-/-(NRG) mice harboring the same SCF/GM-CSF/IL-3 transgene) also exhibited improved engraftment of patient AML samples when compared to NRG [5].

We previously characterized the therapeutic response of AML samples to combined Ara-C and doxorubicin in NSGS mice [6]. Importantly, this model revealed differential response of patient samples to a 5+3 regimen; de novo samples showed delayed disease progression while relapse/refractory samples were resistant. This is consistent with the finding of excellent concordance between the response of a large, diverse group of patient derived xenograft (PDX) models to patient outcome using a variety of therapies [7].

While several groups have successfully employed SCID-based immune deficient mice for studies involving PDX response to standard chemotherapies [8-14], there are limitations for doses, frequency, and prior conditioning. These shortcomings are presumably related to the Prkdc mutation, which is responsible for defects in DNA repair [15] and extreme radio-sensitivity [16]. For unknown reasons, these issues are even more pronounced in IL-2ry mice [1]. In contrast, NRG mice tolerate much higher doses of radiation [17] yet retain the ability to engraft human HSCs and give rise to human blood cell levels and subpopulations that are very similar to NSG mice [18]. It is important to recognize that SCID mutation has functional consequences for every cell in SCID mice, while RAG knockout should only affect differentiation and maturation of lymphocytes. Concerns about SCID-related toxicity are not limited to the hematopoietic compartment for PDX models. For example, it is well established that anthracyclines, which are a common agent in leukemia therapy, have significant toxic effects on cardiac tissues which could be exacerbated in the presence of a SCID mutation [19].

One limitation with previous SCID chemotherapy models was the inability to administer repeated cycles of chemotherapy. Current guidelines for adult and pediatric AML call for two induction cycles, followed by additional intensification/consolidation cycles [20, 21]. Repeated cycles in PDX models may allow for more realistic modeling of response and improved efficacy. Another limitation with the SCID-based model is the inability to give chemotherapy after prior conditioning with either gamma irradiation or busulfan injection. Such conditioning is required for reliably robust engraftment of some PDX samples.

In our previous study, we were careful to examine the effects of chemotherapy on both AML and non-malignant host BM cells [6]. We showed increased sensitivity of AML cells to chemotherapy, particularly with doxorubicin. Ara-C had only minimal selective effects on AML, but increased treatment toxicity. However, these experiments were done in SCID mice, which are likely artificially sensitive to DNA damage-inducing chemotherapy. This sensitivity may artificially lower the relative AML response readout. The maximum tolerable doses of chemotherapies are also likely artificially low and sub-optimal for therapeutic effect.

Recent PDX ALL therapy models in NSG mice utilized a 3-drug induction regimen with vincristine, dexamethasone and L-asparaginase (VXL). This approach has been successfully used along with bioluminescent imaging [22] and combined with Bcl inhibitors [23, 24]. A 4-drug induction protocol (VXL+daunorubicin) optimized for T-ALL engrafted NOD/SCID resulted in 2 of 4 PDX developing signs of resistance [25]. We are unaware of a 4-drug induction protocol for NSG mice. One likely pitfall is increased sensitivity of NSG to anthracyclines [1].

Here, we determined the sensitivity of RAG-based mice (NRG and NRGS) to standard AML and 4-drug ALL induction chemotherapy. RAG-based mice tolerated significantly higher doses of daunorubicin/Ara-C, repeated cycles of therapy as well as combination with busulfan conditioning. Interestingly, we also uncovered a differential activity of doxorubicin and daunorubicin in RAG mice that highlights the importance of full characterization of therapeutics in the various immune deficient models. Finally, we showed that RAG-based host BM cells are more resistant to DA therapy, resulting in an approximate 3.8-fold increase in therapeutic window relative to SCID-based mice. These experiments illustrate the degree to which the choice of host strain may affect results with genotoxic therapies in PDX systems.

Materials and methods

Mice

NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) and NOD.Cg-Rag1tm1MomIl2rgtm1Wjl/SzJ (NRG) mice were obtained from Jackson Laboratories. Generation of NOD.Cg-PrkdcscidIl2rgtm1WjlTg(CMV-IL3,CSF2,KITLG)1Eav/MloySzJ (NSGS) [3] and NOD.Cg-Rag1tm1MomIl2rgtm1WjlTg(CMV-IL3,CSF2,KITLG)1Eav/J (NRGS) [26] have been previously described. All strains were housed and bred in a pathogen-free facility at Cincinnati Children’s Hospital in accordance with an IACUC protocol. Veternary Services of Cincinnati Children’s Hosptial provided hands on and classroom training concerning proper animal handling for all research staff. Mice (both males and females, aged 8–12 weeks) subjected to chemotherapy protocols were monitored twice daily for signs of toxicity. Mice showing poor mobility, labored breathing, or cumulative weight loss of 30% of their initial body weight were immediately euthanized. These humane endpoints discriminate mice with lethal toxicities from those showing less severe, transient signs of illness from chemotherapy exposure (scruffy appearance and slight hunched posture). Chemotherapy exposed mice were provided moistened food to allow easier feeding and aid hydration. Leukemic mice often rapidly develop hind limb paralysis when tumor burden is high therefore mice with signs of hind limb weakness were also euthanized. Additionally, BM and PB samples were periodically taken from leukemic mice in order to ascertain the level of leukemic burden and to better predict the onset of illness. Bone marrow aspirates were taken from live mice under general anesthesia with isoflurane as previously described in detail [27]. Mice received buprenorphine hydrochloride injections to minimize pain and discomfort before the procedure and after, as necessary. Death was not used as an endpoint for any experiment, however, occasionally mice were found dead, presumably due to rapid progression and onset of disease symptoms and/or toxicities during the overnight hours. This was limited to fewer than 5% of mice involved in our studies. All leukemic and chemotherapy protocols and humane endpoints were reviewed and approved by the Cincinnati Children’s Hospital IACUC prior to study initiation.

Cells

The MA9.3RAS cell line was generated by sequential retroviral expression of MLL/AF9 and NRasG12D cDNAs into umbilical cord blood (UCB) CD34+ cells, as described previously [3, 28], and was maintained in IMDM/20%FBS. 2X105 cells (i.v. injection) were used to engraft mice for experiments. Upon sacrifice due to AML, control spleen preparations were frozen for later use in the experiments designed to determine AML/BM toxicity. 8-9X105 cells were i.v. injected into non-conditioned mice for these experiments. The AE46T cell line was originally established by sequential retroviral transduction of UCB CD34+ cells with cDNAs encoding AML/ETO and hTERT [29, 30] and was maintained in IMDM/20%FBS supplemented with 10ng/mL SCF, TPO, FLT3-L, IL-3, and IL-6. 1X106 cultured cells were injected i.v. to induce AML.

Patient samples were obtained from patients at Cincinnati Children’s Hospital Medical Center following informed written consent of parents/guardians and assent of patients over 11 years old. Residual diagnostic specimens were used according to a study protocol (#2008–0021) approved by the Cincinnati Children’s Hospital Institutional Review Board (Office for Research Compliance and Regulatory Affairs). Additionally, we used a pre-existing PDX model (frozen spleen from secondary engrafted mice) which was previously generated from cells from a deidentified sample (DFAM-64519-V2, PRoXe.org [31]). Initial primary specimens were incubated with OKT3 antibody to eliminate the potential for xenogeneic GVHD [26]. Following successful engraftment, BM and spleen preparations from primary mice were viably frozen for future experiments. 1–3 X106 thawed cells were injected i.v. for the PDX experiments described in this study. All cell transplants in this study were done by i.v. injection. A table with patient sample and PDX model information is included as supplementary material (S1 Table).

Chemotherapy

A single i.p. dose of 30mg/kg busulfan was used as a preconditioning regimen in some experiments as described previously [26, 32]. For AML therapy, mice received 1.2mg/kg daunorubicin (D) and 50mg/kg cytarabine (A, Ara-C) by i.v. injection for 3 consecutive days beginning 2–3 weeks after busulfan conditioning and/or cell engraftment (low dose DA therapy). DA was repeated for some mice. Alternatively, a higher dose of 3.0mg/kg daunorubicin and 75mg/kg Ara-C was used (high dose DA therapy). For some experiments, doxorubicin was substituted for daunorubicin. For B-ALL 4-drug induction therapy, we used a 4-week schedule of vincristine (V, 0.5mg/kg, i.p., each Monday), dexamethasone (X, 15mg/kg, i.p., each day Monday-Friday), pegaspargase (P, 1200kU/kg, i.p., 1st and 3rd Monday), and daunorubicin (D, 2.5mg/kg, i.v., each Monday). This treatment is abbreviated “VXPD”. When optimizing chemotherapy doses, mice were monitored for at least 6 weeks after exposure in an attempt to detect longer-term toxicities.

PB and BM analysis

Tail bleeds were analyzed on a Hemavet9500 (Drew Scientific). Engraftment was determined from flow cytometry of PB and BM preparations using a FACSCantoII instrument (BD) with analysis by FlowJo software. Our standard flow panel consists of antibodies to block mouse and human Fc IgG receptors (Miltenyi Biotech) as well as mCD45-APC/Cy7(BD), CD45-FITC (BD), CD3-PE/Cy7 (BD), CD19-VioBlue (Miltenyi Biotech), CD13-PE (BD), CD33-PE (BD), CD34-APC (BD), and CD56-v510 (BD). Leukemia percentage was determined by calculating the number of cells with positive staining for CD33 and CD45 (AML) or CD19 and CD45 (ALL) as a fraction of viable cells.

Statistics

Statistics were calculated with Prism 7 software. The Mann-Whitney U test was used to compare 2 groups. 2-way ANOVA was used to compare groups with repeated measurements. Log-rank analysis was used to compare survival curves. Linear regression analysis was performed to compare trendlines.

Results

RAG mice tolerate higher doses of AML chemotherapy

We began our comparison of SCID and RAG-based mice by searching for the maximum tolerated dose for combined daunorubicin and Ara-C intravenous infusions over three consecutive days. Our initial chemotherapy model utilized 5 days of Ara-C, however, we found that Ara-C alone produced very little leukemia cell specific killing benefit while adding measurable normal BM toxicity [6]. Therefore, we eliminated the final 2 days of Ara-C exposure. We also switched the anthracycline, replacing doxorubicin with daunorubicin in order to better mimic pediatric AML therapy protocols. NSGS (SCID) mice experienced lethal toxicities at all doses higher than 1.2 mg/kg daunorubicin and 50mg/kg Ara-C (Fig 1A) in line with our previous findings. NRGS (RAG) mice survived a 50% higher dose but succumbed to a double dose of 2.4mg/kg daunorubicin and 100mg/kg Ara-C (Fig 1B). NRGS mice were resistant to increasing daunorubicin to 3.0mg/kg if Ara-C remained at 75mg/kg, implying that Ara-C may contribute more substantially to off-target toxicities. These results establish that RAG mice tolerate substantially higher chemotherapy doses.

Fig 1

RAG knockout mice tolerate higher AML therapy than SCID-based mice.

A) Naïve male NSGS mice (n = 3–4, 25.6 +/- 1.1g average body weight) with SCID immune deficiency or B) NRGS male mice (n = 3–4, 27.8 +/- 2.3g) with RAG knockout immune deficiency were challenged with various doses (doses in mg/kg) of combined daunorubicin (D) and Ara-C (A) injection for 3 consecutive days. Survival was monitored to determine maximum tolerable doses. Mice were sacrificed when they reached humane endpoints as described in the methods. (C-H) Mice (n = 4–5 per group) were conditioned with busulfan 3 weeks before exposure to 1.2mg/kg daunorubicin and 50mg/kg Ara-C (BU/DA). Survival (C), relative body weight (D), PB WBCs (E), RBCs (F), hemoglobin (G), and platelets (H) were monitored for responses to chemotherapy. For C-H, 8–12 week old female mice were used with starting weights of 25.4 +/- 3.0g (NRGS) and 24.4 +/- 1.9g (NSGS). Asterisks indicate p<0.05. For C-H, asterisks indicate significant differences between the SCID DA and RAG DA groups. CNTL = PBS controls, DA = combined daunorubicin/Ara-C, WBCs = white blood cells, RBCs = red blood cells.

Next, we exposed NSGS and NRGS mice to sub-lethal busulfan doses 3 weeks prior to chemotherapy to mimic an approach requiring preconditioning for successful engraftment of leukemia. Preconditioning is required for reliable engraftment of many samples and can significantly speed up disease latency for most. SCID and RAG strains received 1.2mg/kg daunorubicin and 50mg/kg Ara-C for 3 consecutive days. Previously we showed that NSGS mice cannot tolerate a similar 5+3 doxorubicin/Ara-C protocol after either irradiation or busulfan conditioning [6]. Consistent with those findings, the NSGS busulfan+DA group experienced lethal toxicities several days after chemotherapy while similarly-treated NRGS mice survived for the duration of the 5-week post chemotherapy observation period (Fig 1C). The NSGS busulfan+DA group experienced more profound weight loss and failure to recover WBC, RBC, and PLT counts while these parameters returned to baseline levels in NRGS (Fig 1D–1H).

Optimization of DA therapy in NRGS PDX mice

Next, we sought to test the efficacy of combined daunorubicin and AraC in leukemic NRGS mice, with and without prior busulfan conditioning. For this, we used a paired set of de novo and relapse PDX samples from the same patient. Busulfan conditioning was used to aid engraftment of the de novo sample but not the relapse sample. The lower dose of 1.2mg/kg daunorubicin and 50mg/kg AraC (Low Dose, LD) was used because that was the dose successfully tested with busulfan conditioning in Fig 1. Marrow aspirates taken after therapy showed significantly decreased AML levels in the mice harboring the de novo sample, but not in those engrafted with the relapse sample (Fig 2A). However, this effect did not translate into increased survival in the de novo group (Fig 2B). Similarly, DA treatment did not affect survival of the mice with the relapse sample either (Fig 2C).

Fig 2

Lack of efficacy with DA in AML-engrafted NRGS.

A) NRGS mice were engrafted with PDX samples generated from a paired de novo/relapse AML case. Busulfan was used to pre-condition mice for de novo engraftment, but not for mice receiving the relapse sample. Mice were treated with 1.2mg/mL daunorubicin and 50mg/kg Ara-C at 3 weeks and BM aspirates were analyzed at day 25. Survival of the mice engrafted with the B) de novo and C) relapse PDX samples was monitored. Asterisk indicates p<0.05 by Mann-Whitney U test. CNTL = PBS controls, DA = combined daunorubicin/Ara-C. Mice were randomly assigned to treatment or control groups.

One possibility for the disconnect between initial treatment response and survival time is that the treatment damages both normal and leukemic cells which then compete to repopulate the bone marrow. If the AML is not sufficiently repressed, then the remaining cells may expand rapidly after therapy and effectively eliminate the gap in AML burden between the treated and control cohorts. Another possibility is that daunorubicin is not as effective as doxorubicin in PDX models. We tested both anthracyclines in two separate approaches to address this lack of efficacy.

First, we tested whether multiple cycles of LD chemotherapy would be tolerated in NRGS mice and improve survival. We engrafted NRGS mice with MA9.3Ras cells and initiated chemotherapy at day 10. After a 1-week break, some mice received a second round of chemotherapy. Others went on to receive a third round according to the same schedule. This schedule of repeated cycles more closely resembles typical patient therapy, which calls for additional therapy in MRD+ or high-risk cases. NRGS mice tolerated additional chemotherapy cycles and survived longer with each successive round of therapy. Consistent with our previous results, a single cycle of 1.2mg/kg daunorubicin and 50mg/kg Ara-C did not show efficacy (Fig 3A). However, when doxorubicin was substituted for daunorubicin, a statistically significant extension of latency was observed with a single cycle which was also further improved by additional cycles (Fig 3B).

Fig 3

Comparison of doxorubicin to daunorubicin in AML PDX models.

A) Survival of NRGS mice engrafted with the MA9.3Ras cell line and treated with 0, 1, 2, or 3 cycles of 1.2mg/kg daunorubicin and 50mg/kg Ara-C (0 DA, 1 DA, 2 DA, 3 DA) or B) 1.5mg/kg doxorubicin and 50mg/kg Ara-C. C) A de novo PDX was engrafted into NRGS mice and treatment began 3 weeks after busulfan conditioning and engraftment using HD DA using either 3.0mg/kg doxorubicin or daunorubicin. AML burden was determined from BM aspirates at day 27. The red points indicate mice with undetectable disease and are plotted as 0.001 in order to include them in the log based plot. D) The mice in C were followed for survival. Log rank tests were used for A,B,D. Mann-Whitney U tests were used to determine significance for C. Asterisks indicate p<0.05. CNTL = PBS controls. Mice in A-D were randomly assigned to treatment or control groups.

Secondly, since we found that RAG mice could tolerate higher chemotherapy doses, we treated NRGS mice engrafted with a chemo naïve PDX sample with 3.0mg/kg daunorubicin or doxorubicin and 75mg/kg AraC (High Dose, HD). NRGS mice tolerated this higher chemotherapy dose 21 days after busulfan conditioning. Mice treated with either HD daunorubicin or doxorubicin (at the same dose) exhibited similar AML burden after therapy (Fig 3C). Approximately half of the mice in each group had AML at less than 0.1% by flow, a clinical cut-off for MRD status. However, most mice did relapse, although survival time was significantly extended (Fig 3D). Notably, doxorubicin resulted in a greater extension of lifespan compared to daunorubicin. In fact, 2 of the 11 HD doxorubicin treated mice had no detectable disease at the end of the experiment.

Use of HD daunorubicin/AraC in de novo and relapse PDXs

We tested the optimized HD chemotherapy treatment protocol in our paired de novo / relapse PDX set. Mice engrafted with the de novo sample responded to therapy with a significantly longer latency while the relapse-engrafted mice showed no response to therapy (Fig 4A). In addition, the HD chemotherapy but not the LD protocol extended the lifespan of busulfan conditioned NRGS mice engrafted with a second chemotherapy-naïve sample (Fig 4B).

Fig 4

HD chemotherapy for de novo and relapse AML PDXs.

A) Survival of NRGS mice engrafted with a matched de novo (DN)–relapsed/refractory (R/R) patient sample were treated with the higher dose of 3.0mg/kg daunorubicin and 75 mg/kg Ara-C. B) Survival of NRGS mice engrafted with a PDX sample from a second de novo case and treated with 1.2mg/kg daunorubicin and 50mg/kg AraC (DA-LD) or a higher dose (DA-HD) as in A. C) NRGS mice engrafted with the AE46T cell line were monitored for AML response to HD DA treatment. D) A relapse adult sample was subjected to two rounds of HD DA chemotherapy. BM AML burden and E) survival are shown. Asterisks indicate p<0.05 by log rank test (panel A, B, E), or 2-way ANOVA (panel C, D). Comparisons are treated versus controls. CNTL = PBS controls. Mice were randomly assigned to treatment or control groups.

We also tested the HD chemotherapy response of mice engrafted with the chemotherapy-naïve AE46T cell line which was derived from UCB CD34+ cells with retroviral directed expression of RUNX1/RUNX1T1 (AML1/ETO) and TERT [29]. HD chemotherapy was initiated at day 46, after engraftment was confirmed in the busulfan preconditioned recipients, resulting in delayed progression of leukemia (Fig 4C). We attempted to further delay leukemia by re-treatment at day 110, however the treated mice experienced significant toxicities and the experiment was ended. We repeated this approach with conditioned mice engrafted with a refractory adult MDS/AML sample with the first round of therapy at day 25 followed more closely by a second round of HD DA 2 weeks later. This timing resulted in mice with low tumor burden after therapy and increased lifespan (Fig 4D and 4E). Importantly, treated mice had similar AML in the BM at sacrifice as controls (CNTL, 64.0 +/- 5.9% vs DA, 72.0 +/- 12.1%), suggesting these mice succumbed to leukemia rather than treatment-related toxicities. Together, these results suggest that treatment toxicities increase in severity as tumor burden increases in PDX models.

RAG mice better tolerate an ALL 4-drug induction protocol

To explore the suitability of RAG-based mice for B-ALL modeling, we examined the durability of NSG and NRG mice to a 4-drug induction protocol for high risk B-lymphoid leukemia. To test for tolerance, we initially exposed non-conditioned, non-leukemic RAG and SCID based mice to vincristine, dexamethasone, pegaspargase, and daunorubicin. SCID-based NSG mice experienced a more dramatic weight loss relative to RAG-based NRG mice, but both strains recovered from the 4-week treatment (Fig 5A). However, when busulfan conditioning was included 3 weeks prior to chemotherapy, half of the SCID-based mice experienced lethal toxicities several weeks post exposure (Fig 5B and 5C). There were no obvious or consistent statistically significant alterations in hematopoietic parameters as measured by CBC analysis, indicating that this effect was unlikely to be related to excessive BM damage or failure and points to non-hematopoietic toxicity (Fig 5D–5G).

Fig 5

Modeling 4-drug induction for high risk B-ALL.

A) Weights of NRG (RAG, n = 10) and NSG (SCID, n = 6) mice were treated with a 4-week course of VXPD. B) Weights of mice conditioned with busulfan 3 weeks prior to VXPD (n = 6 per group). C) Survival of mice in A and B. Mice were sacrificed when they reached humane endpoints as described in the methods. D) WBCs, E) RBCs F) hemoglobin, and G) platelets were monitored before, during, and after VXPD. H) PDX samples from a pediatric B-ALL sample were engrafted into NRG mice after busulfan conditioning. BM aspirates were analyzed by flow cytometry to monitor engraftment. I) Survival and J) B-ALL levels at time of sacrifice were determined for these mice. Asterisks indicate p<0.05 by Mann-Whitney U test (panels A-B, D-H, J) or log rank (panel I). CNTL = PBS controls.

To test the efficacy of 4-drug ALL induction, a chemotherapy-naïve B-ALL was engrafted into busulfan conditioned NRG mice. The 4-week treatment started once B-ALL was detectable in the PB. Serial BM aspirations revealed a dramatic decrease in ALL burden in treated mice relative to controls (Fig 5H) which resulted in a significant latency shift (Fig 5I). Importantly, treated mice showed the same level of ALL as control mice at the time of sacrifice, indicating that the mice did not experience treatment-related toxicities (Fig 5J). These data demonstrate the utility of NRG mice in the modeling of high risk 4-drug ALL induction therapy.

RAG mice offer a better therapeutic window for genotoxic agents

To this point, we have established that SCID mice have lower tolerance for chemotherapy regimens. We assume that this is at least partly due to the Prkdc mutation being consequential in all cells. DNA repair should be compromised in SCID mice and therefore we would expect higher rates of apoptosis in response to DNA damaging agents. This problem should be avoided in RAG mice, because RAG knockout should specifically affect lymphocyte development and play no direct role in DNA damage response. The MA9.3Ras cell line causes fatal AML in both NSGS and NRGS with very similar kinetics (Fig 6A), making this model suitable for testing this hypothesis, and for quantifing any SCID/RAG differences. To examine initial response to chemotherapy, we subjected non-conditioned mice engrafted with MA9.3Ras cells to 3 consecutive days of DA exposure and sacrificed them 3 days later. For this experiment we used the SCID MTD of 1.2mg.kg daunorubicin and 50mg/kg AraC (LD). In NSGS DA treated mice, BM cellularity was reduced to 34.7% of controls (25.4 X106 vs 8.8 X106 WBCs/femur) while a somewhat smaller decrease was observed in NRGS (31.4 X106 to 13.7 X106 cells, 43.7%) (Fig 6B). NRGS DA mice had significantly more surviving BM cells than NSGS DA mice (p = 0.0073), however, increased NRGS BM cellularity was also noted in control mice. This finding might be at least partially explained by the overall larger size of age-matched male NRGS compared to NSGS mice (32.0+/-2.3g vs 28.5+/-2.0g, p = 8E-06, Fig 6C). Linear regression analysis of age/weight data confirmed that the NRGS mice used in these experiments were larger than their SCID-based counterparts. Separate analysis of NRG and NSG showed similar significant differences in both age-matched males and females (not shown). Absolute AML cells per femur was decreased in both strains in response to chemotherapy. However, this decrease was more dramatic in NRGS mice, where DA-treated mice contained on average only 0.34%+/-0.26% of control levels while NSGS DA-treated mice retained 0.96%+/-0.73% (p = 0.027, Fig 6D). By taking the percent decrease of normal mouse BM and human AML cell numbers in response to DA together, we calculated the relative AML specific toxicity in both strains. For NRGS mice, DA treatment resulted in a 49.5-fold decline in absolute AML number compared to normal mouse BM cells while NSGS mice experienced only a 13.1-fold difference, suggesting a larger therapeutic window (a 3.8 fold difference between strains) for cytotoxic chemotherapy in NRGS mice (Fig 6E).

Fig 6

Determination of therapeutic window in NSGS and NRGS mice.

A) Latency of MA9.3RAS leukemia induction in NSGS and NRGS mice. B) Day 14 BM cellularity of mice engrafted with MA9.3RAS cells. Mice were treated with 3 days of PBS (CNTL) or DA (1.2mg/kg daunorubicin and 50mg/kg Ara-C) beginning on day 9. Mice were randomly assigned to treatment or control groups. C) Body Weight plotted against age of mice used in panel B. D) Absolute MA9.3RAS cell number in mice presented in panel B. E) The toxicity ratio was calculated for each DA treated mouse. Log rank test was used for panel A. Asterisks indicate p<0.05 by Mann-Whitney U test (panel B, D, and E) or linear regression analysis (panel C).

Discussion

We show that RAG-based mice tolerate a busulfan preconditioning regimen in combination with leukemia chemotherapy, expanding the number of samples that can be studied in vivo. RAG-based mice also tolerate multiple cycles of therapy, thereby allowing for more aggressive, realistic modeling. Furthermore, standard AML therapy in RAG mice was 3.8-fold more specific for AML cells, relative to SCID mice, demonstrating an improved therapeutic window for genotoxic agents. We conclude that RAG-based mice represent the new standard for preclinical evaluation of therapeutic strategies involving genotoxic agents.

We did not cure any mice using either our daunorubicin HD or repeated cycle protocols, even in a PDX model of a de novo patient that achieved a MRD(-) remission clinically. This suggests that while we have improved modeling of standard therapy, the models are still not optimized. One difficulty is likely the relative lack of benefit from Ara-C in these models. Ideally, Ara-C would be delivered as a continuous slow infusion by implanted osmotic pumps over the course of a week rather than as several bolus injections. Alternatively, liposomal formulations could also improve efficacy [33]. Supportive care to combat treatment toxicities is also largely absent in PDX models and this may limit the successful implementation of any protocol, particularly if disease burden is high. However, further optimization may not be the preferred approach, since a perfected model would likely make it more difficult to realize a PDX benefit from additional novel therapies.

The ability to tolerate higher chemotherapy doses also suggests that additional cytotoxic agents could be added to the DA backbone. For example, pediatric AML patients are commonly treated with etoposide in addition to DA (so-called ADE induction therapy). The RAG background should allow for expanded chemotherapy modeling in mice. Additionally, targeting anti-apoptotic Bcl-2 familiy proteins along with standard therapy is of great interest, however these inhibitors have been shown to sensitize to anthracyclines [34-36] posing a major challenge for in vivo experiments with SCID mice. The SCID defect would also cause sensitivities to other therapies that induce double strand DNA breaks. As a result, experiments that couple proton therapy to chemotherapy or other sensitizers [37] could be more easily done with RAG-based mice.

The finding of a 3.8-fold improvement of the therapeutic window with RAG relative to SCID mice is similar to the 2–3 fold difference in sensitivity reported between SCID and BALB/c fibroblasts exposed to bleomycin or gamma irradiation, both of which induce double strand DNA breaks [16]. Anthracyclines such as daunorubicin and doxorubicin inhibit the ability of Topoisomerase II to reseal double strand DNA breaks. Given that the SCID mutation renders cells defective in double strand break repair, this likely provides the rationale for increased sensitivity of SCID mice to both the DA and VXPD induction protocols and a worse therapeutic window.

Interestingly, we have found that SCID and RAG mice have similar sensitivities to busulfan conditioning but react differently upon additional genotoxic stress. Busulfan works by induction of intra-strand crosslinks and mono-alkylation of DNA [38], so one might predict repair to be independent of the Pkrdc mutation. However, prior exposure to busulfan further increased the sensitivity of SCID mice to ds-DNA break inducing agents as evidenced by the failure of conditioned NSGS mice to tolerate DA therapy at doses tolerated by naïve NSGS mice. This has clear implications for studies that combine anthracyclines with other DNA damage inducing agents, even if the mechanisms of action are distinct. RAG-based immune-deficient mice should be used for chemotherapy modeling that requires conditioning of mice prior to engraftment.

Although the SCID-associated genotoxic sensitivities are especially severe when ds-DNA break-inducing agents are used, it should be appreciated that some agents that do not damage DNA may increase sensitivity to anthracyclines. For example, the BCL inhibitor venetoclax enhances the effects of ionizing radiation [39]. Experimental MDM2 inhibitors are potent activators of p53 which could be expected to sensitize cells to anthracycline therapy [40, 41]. Furthermore, some compounds may produce unexpected toxicities in the same way, as has been described recently for abemaciclib [42]. These activities are likely to increase non-specific toxicities of standard chemotherapy in SCID-based mouse models, effectively limiting detection and measurement of a pre-clinical therapeutic window.

In the current study we substituted doxorubicin with daunorubicin in order to more closely follow accepted clinical protocols. Surprisingly, we found less efficacy than expected with daunorubicin and a marked improvement of response to doxorubicin over daunorubicin in our PDX models in head to head experiments. This could simply reflect a difference in human and mouse metabolism of the drugs. On the other hand, it could indicate a true difference in the efficacy of these anthracyclines. The optimal dose for individual anthracyclines is different for each drug and there is active study and debate about the relative efficacy between the members of the class. Non-hematopoietic toxicities are an important clinical consideration that must be balanced against the anti-leukemic effect. A retrospective study of childhood cancer survivors demonstrated that daunorubicin resulted in approximately half the risk of future cardiac failure relative to doxorubicin [43]. Doxorubicin has also been found to be associated with more complications due to infections than daunorubicin when given to ALL patients during delayed intensification [44]. Another study with retrospective analysis of a large group of patients found that in children over 3 years of age, doxorubicin was associated with significantly higher rates of induction related mortality, but fewer induction failures than were observed with daunorubicin [45].

Similarly, we have previously used L-asparaginase for PDX ALL induction therapy but switched to pegaspargase in order to update our models to more closely follow practices in pediatric oncology. A large multi-center trial of childhood de novo ALL found that results and toxicities from biweekly pegaspargase were very similar to those observed with weekly intra-muscular injection of native L-asparaginase given after initial induction induced remission [46]. Similarly, a comparison in a relatively low number of adult high-risk ALL patients found no difference in clinical outcomes [47]. In a cohort of relapsed pediatric ALL patients, while pegaspargase demonstrated a prolonged half-life, there was an observed trend towards lower asparagine clearance in the CNS [48]. It remains to be seen whether this substitution has any effects in the PDX setting.

Recently, we have shown that humanized NSGS mice have improved hematopoietic function over humanized NSG mice [49]. Moving forward, it will be important to test immune function in NRGS mice as well, since these mice could be better hosts to build immune therapy models with, particularly if exposure to genotoxic agents is planned. The ability to busulfan condition prior to chemotherapy will be an important advantage for building better models of therapy. For example, conditioning is required for engraftment and humanization with UCB. UCB-transplanted mice might allow for the evaluation of therapies in the context of human immune cells [50].

Summary of PDX models.

Age, sex, and stage of disease of the source material used to build each PDX model is listed. The genomic alterations present in the PDX models as well as a description of the latencies in mouse strains with and without busulfan conditioning is also listed.

(XLSX)

Click here for additional data file.

10 Oct 2019

PONE-D-19-20568

Improved chemotherapy modeling with RAG-based immune deficient mice

PLOS ONE

Dear Mr. WUNDERLICH,

Thank you for submitting your manuscript to PLOS ONE. Apologies for delay - it was difficult to find timely reviewers with appropriate expertise. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Nov 24 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Daniel Thomas, MD

Academic Editor

PLOS ONE

Journal Requirements:

1. Thank you for inclouding your funding statement; "This work was funded in part by an NIH/NCI R50 award (#CA21140, MW). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

Please provide an amended statement that declares *all* the funding or sources of support (whether external or internal to your organization) received during this study, as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now.  Please also include the statement “There was no additional external funding received for this study.” in your updated Funding Statement.

Additional Editor Comments (if provided):

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: An extremely valuable piece of work at a time when many academic institutions and industry are expanding their patient-derived xenograft models. The superior robustness of NRG and NRGS mice to tolerate high doses, repeated doses and busulfan pre-conditioning has been suspected by many in the field but never definitively compared with the same patient samples/ cell lines and conditions. The improved efficacy of doxorubicin compared to daunorubicin for AML is also intriguing and correlates with increased toxicity observed in humans. The calculation of AML vs normal murine cell ratio comparing NSG and RAG mice is an informative measurement; if only similar studies were performed by other groups.

English and figure presentation is excellent.

Minor points

1] It is not clear as you read how old the mice are after weaning, in general, or specifically, at the time of engraftment and conditioning. Sex and weight of mice in Figure 1 are also not specified.

2] A small table of patient details for the de novo and relapse pairs would be useful for other groups to compare their own rates of engraftment.

3] I assume all samples were tail vein engrafted but BM sampling was with live aspiration - this was not completely clear in methods.

4] The lack of busulfan conditioning in the relapsed sample Fig 2A makes it difficult to assess the lack of reponse to DA. Any other data to show what happens to relapsed sample with bu conditioning? Any data to show complete lack of engraftment of de novo without any conditioning?

5] Fig 2 CNTL not obvious to every reader - please define in figure legend. Busulfan not stated as given in the Figure legend - this is important variable.

6] A similar indice comparing engrafted AML response to chemo to engrafted normal human cord blood derived hematopoiesis response to chemo would be interesting to extrapolate and compare with clinical trials.

Reviewer #2: The manuscript follows, broadly, a similar approach to earlier publications by these authors, and establishes the utility of RAG-based immunodeficient mice as a model fro AML PDX. The work presented in this manuscript is sound, and well performed, and will be a useful resource for others working in the field. I would judge that there a number of groups that have perhaps already reached similar conclusions as those resented in this manuscript. Nevertheless, this is a paper that warrants publication, as it presents a solid characterisation of the model and response to standard chemotherapeutic regimens. I do not have any major changes to suggest. I would argue that there needs to be some attention paid to the presentation of the figures. In many instances, the acronyms used in the figures are not described in the figure legends. For example in Figure 1 BU/DA is not explained (I accept that one can guess easily enough) or there is somewhat eclectic title to a figure panel such as 6 C. However, beyond these very minor suggestions, i judge the manuscript will find a readership in those interested in AML PDX models largely as it is written.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Daniel Thomas

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

25 Oct 2019

We thank the reviewers for their time and careful consideration of our manuscript. Please find the responses to individual points below. Changes to the manuscript are indicated using the track change feature in Word.

Reviewer #1: An extremely valuable piece of work at a time when many academic institutions and industry are expanding their patient-derived xenograft models. The superior robustness of NRG and NRGS mice to tolerate high doses, repeated doses and busulfan pre-conditioning has been suspected by many in the field but never definitively compared with the same patient samples/ cell lines and conditions. The improved efficacy of doxorubicin compared to daunorubicin for AML is also intriguing and correlates with increased toxicity observed in humans. The calculation of AML vs normal murine cell ratio comparing NSG and RAG mice is an informative measurement; if only similar studies were performed by other groups.

Thank you for the positive review of our work.

English and figure presentation is excellent.

Minor points

1] It is not clear as you read how old the mice are after weaning, in general, or specifically, at the time of engraftment and conditioning. Sex and weight of mice in Figure 1 are also not specified.

We have included additional information regarding age and sex in the methods and the specific information for figure 1 in the legend.

2] A small table of patient details for the de novo and relapse pairs would be useful for other groups to compare their own rates of engraftment.

We have included a small table (Table S1) with patient sample and PDX information.

3] I assume all samples were tail vein engrafted but BM sampling was with live aspiration - this was not completely clear in methods.

Yes, the reviewer is correct. We have made this clearer in the methods. See line 132 for aspirate information and 162 for i.v. injection.

4] The lack of busulfan conditioning in the relapsed sample Fig 2A makes it difficult to assess the lack of reponse to DA. Any other data to show what happens to relapsed sample with bu conditioning? Any data to show complete lack of engraftment of de novo without any conditioning?

Busulfan conditioning speeds up engraftment of almost every AML PDX. As it is a chemotherapy, busulfan conditioning is somewhat toxic, but the effects are transient, and mice fully recover after a couple of weeks, before chemotherapy (DA) exposure. This is why our protocol is to wait 3 weeks before chemotherapy treatment. By this time, prior busulfan exposure should not be a factor in the response to DA. With aggressive samples such as Pt#55 relapse, busulfan conditioning would likely cause the latency to fall before that three-week timepoint. We have seen this in previous work with cell lines that have similar latencies. Therefore, using busulfan for such an aggressive sample is not feasible. Even though the relapse sample was engrafted into non-conditioned mice, the aspirate data in figure 2A shows that engraftment is actually better than the presentation sample with busulfan pre-conditioning. We have seen previously that some samples do not engraft without conditioning. More frequently, conditioning allows for a more robust early engraftment and shorter latency of disease. This is an important tool which can decrease time and money spent on lengthy in vivo experiments involving samples with latencies that would otherwise be 100 days or longer.

Our study only sought to determine conditions under which pre-conditioning could be combined with subsequent chemotherapy. This was important to us because pre-conditioning with irradiation or busulfan has well known benefits and is a common practice in PDX modeling. However, we did not design experiments to specifically demonstrate the benefits of conditioning as part of this study.

5] Fig 2 CNTL not obvious to every reader - please define in figure legend. Busulfan not stated as given in the Figure legend - this is important variable.

We have defined CNTL in the legend. We have also included a statement regarding busulfan conditioning in the legend to make this clear.

6] A similar indice comparing engrafted AML response to chemo to engrafted normal human cord blood derived hematopoiesis response to chemo would be interesting to extrapolate and compare with clinical trials.

We agree that this could be an interesting and potentially useful system to also measure the effects on normal human cells in RAG based xenografts. Perhaps it would be even more appealing to use the assay to test the activities of a novel experimental agent, in order to show pre-clinical specificity or potential for a therapeutic window.

However, this approach would require significant effort to establish and optimize as several technical considerations make this experiment very difficult to perform and interpret. First, significant engraftment of UCB CD34+ requires either very high cell numbers or busulfan conditioning. We could evaluate effects after reconstitution of the BM has stabilized and after busulfan effects on the BM are repaired, ~8-10 weeks later, but the engraftment levels and cell type subpopulations within a cohort of UCB CD34+ transplanted mice are quite variable at this time. This makes determination of total normal cells before and after therapy (averaging control / treated groups) very imprecise. This is made worse by the fact that human cell engraftment affects the remaining murine BM cells, including a progressive BM failure in NSGS and NRGS mice (Wunderlich, JCI Insight, 2016, https://insight.jci.org/articles/view/88181), thereby altering total cellularity. Also, human cells do not fully recapitulate normal hematopoiesis in mice, so a human BM graft would likely have altered sensitivities to chemo as well, especially given the lack of human RBC production by human cells. Another variable that would need to be considered and tested is the rate of repopulation of the BM by the UCB graft and residual mouse hematopoiesis. The timing of sacrifice and quantification after chemo exposure will greatly affect the results if human and mouse cells repopulate at different rates.

Alternatively, if we engrafted UCB CD34+ cells and treated early with chemotherapy, it is unclear how many CD34+ cells would be required for this approach. We transplanted 800-900k MA9.3Ras cells for the AML experiment and using similar numbers of CD34+ would require an entire UCB unit for each 1 or 2 mice, making this approach cost-prohibitive. Busulfan conditioning would also be required and we would be initiating chemo before our tested window of 3 weeks. In short, there are many logistical issues to be considered and weighed to reach reliable conclusions with such an approach.

Reviewer #2: The manuscript follows, broadly, a similar approach to earlier publications by these authors, and establishes the utility of RAG-based immunodeficient mice as a model fro AML PDX. The work presented in this manuscript is sound, and well performed, and will be a useful resource for others working in the field. I would judge that there a number of groups that have perhaps already reached similar conclusions as those resented in this manuscript. Nevertheless, this is a paper that warrants publication, as it presents a solid characterisation of the model and response to standard chemotherapeutic regimens. I do not have any major changes to suggest. I would argue that there needs to be some attention paid to the presentation of the figures. In many instances, the acronyms used in the figures are not described in the figure legends. For example in Figure 1 BU/DA is not explained (I accept that one can guess easily enough) or there is somewhat eclectic title to a figure panel such as 6 C. However, beyond these very minor suggestions, i judge the manuscript will find a readership in those interested in AML PDX models largely as it is written.

We thank the reviewer for the positive assessment of our work. We agree that while others may have reached similar conclusions, we hope that publishing our findings will help the vast majority of researchers who are not as experienced in the intricacies of PDX modeling.

We have defined acronyms in the figure legends for most figures. Additionally, we have changed the figure 6C title to “Body Weight Comparison”.

Reviewer#3

Review for Wunderlich et al.

PONE-D-19-20568

“Improved chemotherapy modeling with RAG-based immune deficient mice”

Overall recommendation: accept with revision

To the Editors:

This manuscript describes a new NRG-based mouse model of leukaemia that is able to tolerate a more realistic human AML chemotherapy regimen thus providing a superior in vivo model than the NSG-based models currently used. Additionally, the inability of current NSG/NSGS mouse models to tolerate pre-chemotherapy busulfan conditioning is also overcome increasing the number of PDX engraftments able to be modelled. The manuscript is well written, the data sound and the study would be of interest to the readers of the PLOS ONE. However, the manuscript would benefit from some minor alterations/clarifications which I have in the comments to authors below.

To the Authors:

Overall comments to authors

This study describes a new NRG-based mouse model of leukaemia that is able to better tolerate treatment regimens currently used in the clinic that the more widely used NSG models. The manuscript is well written and the data and conclusions sound. However, some additional clarifications and amendments are required before I can support publication. Please address my questions and comments as per below.

Thank you for the positive feedback.

1. It’s not always mentioned with what the control mice have been treated (busulfan alone, busulfan + PBS?). Can this please be interrogated throughout?

We have now defined CNTL in each legend.

2. Fig 2A, 3C, 4D, 4F, 5H, 5J: How was %AML determined, what % positivity of which markers from the flow panel? Is %AML the % of human markers relative to mCD45 or absolute % of human markers?

Also, which marker/s was used to determine B-ALL cells in the blood (line 339)?

CD33 was used to label the AMLs and CD19 was used for ALLs. The models used here also all had human CD45 expression. We present the percentage of positive cells within the total viable cell fraction. A sentence has been added to methods to relay this point.

3. The methods say "Additionally, BM and PB samples were periodically taken from leukemic mice in order to ascertain the level of leukemic burden and to better predict the onset of illness.” Was there a threshold level of engraftment that mice had to reach prior to treatment initiation and if so what was it?

No, we did not establish a threshold for engraftment. Our PDX models were all tested and characterized prior to the study and showed reliable engraftment without failure. The above quoted sentence is in reference to measures taken that aid in our efforts to minimize discomfort and needless suffering of the mice.

Line 288 states: Engraftment was confirmed in the busulfan preconditioned recipients before HD chemotherapy at day 46. However, in Fig 3AB it's not clear whether MA9.3Ras engraftment was confirmed prior to day 10 treatment initiation? Similarly, in Fig 3C,D it's also not clear whether PDX engraftment was confirmed prior to day 7 treatment initiation. Whether confirmation of engraftment prior to treatment initiation occurred (or not) needs to be explicitly stated throughout.

Figure legends 2,3,4, and 6 have been updated to indicate random assignment of mice to treatment groups.

4. Can the number of days relating to the survival curve comparisons please be stated in the text with p-values included throughout? It's a little hard to estimate from the graphs eg: single cycle comparison in Fig 3B appears to be 50 vs 53 days which doesn't seem like it would be a “significant extension of latency”.

The asterisks indicate statistically significant differences only (p<0.05 by log rank). The main findings are that the latency shifts further to the right with each successive round of therapy and that overall, doxo has a more robust effect than dauno in these PDX models. We have inserted “statistically” into the text to avoid suggesting that a modest extension represents a significant (in terms of scale) biological effect.

5. Fig 3C:There are 9 dots in the CNTL group, not 8 and 2x dots are on the dotted zero line so shouldn’t the numbers quoted be 2/9 <0.1%? The data may be better represented with a split y-axis.

We have re-analyzed and re-plotted these results with log scale in order to better show the data points with low disease burden. Upon this review, one of the low-level control points represented a mouse that was sacrificed after this BM check with an apparent mouse cell leukemia/lymphoma soon after the aspirate data was taken and was therefore censored from the experiment. The 2 red points in the doxo/arac column had undetectable disease. The legend has been edited to reflect this.

6. Line 271, 272: “AML burden …. was equally suppressed…” But the authors don't know if the AML level was suppressed in treated groups as there was no initial determination of %AML, only at day 27. Similarly, line 273 “less than 0.1% AML by flow” – the low level of assessable AML may be due to lack of initial engraftment rather than response to treatment.

These points may not be relevant once point 3 is addressed though.

We have rephrased this sentence to “Mice treated with either HD daunorubicin or doxorubicin (at the same dose) exhibited similar AML burden after therapy (Fig 3C).” This change avoids assuming that the levels were equivalent at the start of treatment (the mice were randomly assigned to groups).

No control mice exhibit engraftment of less than 0.1% at this timepoint, therefore we attribute the lowered level of engraftment in the treated cohorts to the activity of the drugs.

7. For Fig 4A shouldn't the Log rank and not the M-W test (as indicated in the legend) have been applied?

Yes, thank you for your attention to detail. This has been corrected in the legend.

Also, does the asterisk in 4A denote significance of de novo DN treatment group compared with all other groups, or just the DN CNTL group? Please specify in the legend. Similarly for Fig 4B, what comparisons have been made and which are significant?

Only controls vs treated were analyzed. A clarification was added to the legend.

8. Fig 5: The error bars overlap for 12/17 of the data points denoted with asterisks, how can there be significant?

We are simply reporting the result of the Mann-Whitney calculation done in Prism software. We do not make any claims that the individual CBC points represent meaningful differences. In fact, in the text we offer the interpretation that “There were no obvious alterations in hematopoietic parameters measured by CBC analysis, indicating that this effect was unlikely to be related to BM failure (Fig 5D-G).” We have edited this sentence to reinforce that idea. We say this because for each of the parameters (WBC, RBCs, Hg, Plt) there are only sporadic timepoints with p <0.05 and many of these are contradictory (such as PLT sometimes higher, sometimes lower in SCID relative to RAG, Fig 5G). The weights are likely to be more meaningful. For those measurements, the SCID line tracks continuously below the RAG line with recurrent timepoints of p<0.05. Also, extreme weight loss by some SCID mice was a key factor leading to the necessary sacrifice of mice in Fig 5B.

9. Lines 354-357: If MA9.3Ras engrafted mice were treated and all animals sacrificed 3 days later, how was the survival curve plotted in Fig 6A obtained? Please clarify or reword these sentences.

Figure 6A are results from a separate, initial experiment comparing MA9.3Ras cells in NSGS/NRGS. We agree that these sentences were a bit out of order. We have made edits to this paragraph to attempt to improve the flow of information.

Minor Comments

10. Spelling error, line 73 “agenet”

Please also define that UBC is (I presume) umbilical cord blood, line 286

Line 359: ‘M’ is not standard nomenclature for cell numbers, please use x10^6. Please alter on Fig 6B as well

Thank you, these corrections have been made.

11. Can more details for the cell lines used in the study please be given for readers who may not be familiar: MA9.3RAS, is this an AML cell line and from where was it obtained? The AE46T cell line is not mentioned in the methods section either

Our lab has generated both of these cell lines through retroviral transduction of UCB-CD34+ cell cultures. We have added references along with some information in the methods section and in the newly added Table S1.

12. Fig 1 legend: the description of D-H are a bit out of order. C=survival, not E. Please check the parameters on the graph and assigned figure letters in the legend

Thank you. This has been corrected.

13. Line 368: what is the p-value of the DA-treated vs control mice absolute cell number comparison?

The Mann-Whitney p value is 0.0073. We have added this to the text, line 369.

14. Fig 6B: what does the “m” in mBM represent?

We have changed Fig 6B to read as “mouse BM Cell Death” to make this clear.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

7 Nov 2019

Improved chemotherapy modeling with RAG-based immune deficient mice

PONE-D-19-20568R1

Dear Dr. WUNDERLICH,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Daniel Thomas, MD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

All points have been adequately addressed.

Reviewers' comments:

12 Nov 2019

PONE-D-19-20568R1

Improved chemotherapy modeling with RAG-based immune deficient mice

Dear Dr. WUNDERLICH:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Daniel Thomas

Academic Editor

PLOS ONE