兽研尊敬的编辑：
我们感谢贵刊近期对我们之前发表的手稿的回应，并本着科学探究的精神，借此机会回应作者的评论和批评。首先，我们欢迎作者对这一主题及所引用研究文献的深入参与。对于我们的荟萃分析部分中任何遗漏或错误记录的数据，我们深表遗憾。
关于荟萃分析数据收集中的错误，我们感谢这位作者对细节的关注并发现这些错误。事实上，我们已经修订了补充表1，以纳入Zarea、Abdullah和Shaw等人研究中跳蚤的更正数据[1–3]。
对于使用Jensen引物的研究，我们承认更正意见，即Shaw等人[3]确实进行了洗涤，并且在从猫身上采集的跳蚤的44份DNA提取物中，有16份在聚合酶链反应（PCR）后进行了测序，发现在16S rRNA基因193 bp片段上与M. haemominutum匹配。我们在此附上更新后的表1，反映Shaw研究的错误分类。然而，我们认为这一更正并不改变Jensen引物可能扩增了螺原体属DNA的结论，洗涤并不一定降低其流行率。Shaw研究使用了洗涤这一事实进一步支持了我们的假设，因为即使跳蚤经过洗涤，其中约40%（44/90）的样本使用Jensen引物仍产生阳性结果。
Table 1. Primer set, flea pooling, and flea washing are associated with the detection of hemotropic Mycoplasma spp
| Primer | Washed | Percentage of infected fleas ± 95% CI (# positive/total) |
|---|---|---|
| Myco_Hf_F/Myco_Hf_F.1 and Myco_Hf_R [ 4 ] | No | 45.7 ± 1.9% (148/324) a |
| Myco_Hf_F/Myco_Hf_F.1 and Myco_Hf_R [ 4 ] | Yes | 17.1 ± 4.2% (44/258) b |
| Other primers | No | 1.5 ± 0.3% (11/734) c |
| Other primers | Yes | 1.4 ± 0.1% (3/215) c |
对于使用非Jensen引物的研究，更正后的数字并未显著改变我们的解读。在Zarea等人的研究[1]中，比例应如何计算略不清楚，因为从猫身上采集的跳蚤中有两只犬栉首蚤按我们的纳入标准不应被纳入，但作者未说明所有七只嗜血支原体阳性跳蚤是否均为猫栉首蚤。保守估计为7/91，即流行率<8%。同样，对于Abdullah等人的研究[2]，根据我们的纳入标准，我们应仅纳入从猫采集的跳蚤池结果，共包括210个池。然而，根据他们的结果：“针对嗜血支原体种属的特异性qPCR发现，仅3份跳蚤样本对M. haemofelis或M. haemocanis呈阳性，均发现于猫栉首蚤指名亚种跳蚤中，其中2份采集自猫，1份无宿主物种记录。”因此，该研究的分子也應修正为仅包括已知采集自猫的两个阳性池。使用2/210（1.0%）代替报告的3/467（0.6%）在统计上对结果的解读没有有意义的影响。使用各种种属特异性（非Jensen）引物更新从猫栉首蚤扩增嗜血支原体DNA的总体比例为14/949（1.5%），我们认为这仍然表明使用这些种属特异性引物通过定量（q）PCR从跳蚤中扩增嗜血支原体仍然极为罕见。
关于合并的作用，研究内池大小不一致确实使荟萃分析复杂化。由于大多数研究未制备一致大小的跳蚤池，也未报告所有阳性和阴性池中的跳蚤数量，我们的荟萃分析仅旨在捕获任何关于合并的报告。我们敦促未来的研究人员纳入元数据或分析不同大小跳蚤池中嗜血支原体检测的情况。Willi等人[5]和Abdullah等人[2]研究中关于池大小的遗留问题，以及不同读者的不同解读，凸显了清晰全面报告方法学的重要性，包括报告此类研究的池大小。此外，池大小对检测跳蚤携带或传播的嗜血支原体或其他细菌的影响尚不清楚，再次强调了未来研究应重点关注确定最佳条件（包括池大小），以便通过PCR适当检测跳蚤中的细菌DNA。
尽管Willi等人[5]在整篇手稿中提到了混合池的影响，但遗憾的是，他们并未报告每个混合池中的蚤数量。该研究报告了来自17只猫的73只蚤的结果，混合池为1-5只蚤（根据Willi等人[5]表1），方法部分指出“部分蜱和蚤样本被混合用于DNA提取（表1）；混合池由同一物种的节肢动物组成，且这些节肢动物采集自同一只动物。”在结果部分，作者指出“从动物采集的三个蜱和两个蚤通过实时PCR检测嗜血支原体呈阳性；所有阳性样本均从单个节肢动物中提取，”而在讨论部分，作者提出了以下作为嗜血支原体低感染率估计值的潜在解释：“本研究中较低的样本感染率可能是因为大多数蚤是单独分析而非混合池分析。”因此，无法在荟萃分析中将蚤按混合池或非混合池进行区分。同样，在Abdullah等人的研究[2]中，虽然方法部分报告蚤进行了混合，但该研究报告的至少一半“混合池”为单个蚤。除此之外，该研究未提供关于混合池大小的进一步信息，也未进一步研究每个混合池中单个蚤与多个蚤的影响。无论如何，所注意到的混合池解释差异并不改变以下结论：使用物种特异性引物时，嗜血支原体感染的蚤极少：Willi等人[5]中为两个单个蚤，Abdullah等人[2]中200多个混合池中有两个混合池（可能但不确定每个混合池由一只蚤组成）。
在本研究的实验部分，我们重新检测了Assarakorn等人[6]先前发表的一项研究的样本，该研究报告使用Jensen引物在蚤中嗜血支原体DNA感染率很高。我们承认，重新分析中感染率显著较低的一个潜在解释可能是样本储存问题，尽管大量研究表明DNA在冷冻储存数年后仍保持稳定[7]。然而，混合池方法的问题可能与此无关。原始Assarakorn研究中混合池方法和结果的报告，与许多参考文献中的报告一样，有些不够具体，容易引起误解，从而导致我们手稿中的误报[8]。根据Assarakorn等人研究的主要研究者（M. Lappin）和我们手稿的共同作者，在原始Assarakorn研究中，蚤按猫进行混合，许多猫每个混合池只有一只蚤，但每个混合池最多五只蚤。没有关于每个混合池蚤数量或仅采集到单只蚤的猫数量的进一步信息。然而，在原始Assarakorn等人研究中，有些猫有超过5只蚤，这些猫创建了多个混合池，但仅检测了1个混合池，这解释了Assarakorn等人检测的50个混合池（来自50只猫）与我们手稿中检测的67个混合池（来自50只猫）之间的数值差异。在我们的重新分析中，所有67个原始DNA提取物的等分试样均在科罗拉多州立大学使用Jensen引物和北卡罗来纳州立大学使用Manvell引物重新检测。由于所有猫的所有混合池均被检测（来自50只猫的67个混合池），结果仅可在50个原始混合池之间直接比较。在先前未检测的混合池中，7/17（41%）使用Jensen引物呈阳性，而先前检测的混合池中为18/50（36%）。结果在此表（表2）中更明确地呈现。
Table 2. Flea pools from Assarakorn et al. [ 6 ] study, with results shown by the original result and retesting in Moore et al. [ 8 ]
| Original Assarakorn et al. [ 6 ] result | Moore et al. [ 8 ] retesting result | |||
|---|---|---|---|---|
| Mhf | Spiroplasma | Other bacteria | Negative | |
| Mhf or Mt | 0 | 0 | 0 | 2 |
| Mhm | 0 | 13 | 1 | 1 |
| Negative | 0 | 2 | 2 | 29 |
| Not tested | 2 | 1 | 4 | 10 |
关于Manvell和Jensen引物的分组，在Manvell等人研究[9]中，最初有大量阳性PCR结果在最终分析中被重新分类，因为测序显示这些阳性实际上为螺旋体属，而非嗜血支原体。在荟萃分析中，这些引物组被归为一组，因为它们共享一个反向引物，且正向引物存在单碱基对偏移，该偏移在猫嗜血支原体和猫栉首蚤的螺旋体属之间同源。
最后，关于我们实验研究中的洗涤部分，我们承认由于蚤中嗜血支原体存在情况未知，无法了解洗涤的效果。我们在原始发表的讨论部分提到了这一点。然而，这一点本可以更加强调。遗憾的是，没有已知感染嗜血支原体的蚤可用于研究洗涤蚤在支原体检测中的作用。这些重要实验仍有待完成。
总体而言，关于蚤传播血支原体的实验证据很少，这表明需要更稳健的实验。事实上，我们研究中引用的血支原体实验感染数量很少，但考虑到在实验环境中使用猫蚤和猫的难度，对于如此劳动密集的动物研究而言，并不算过少[10]。这些结果虽然有限，但在发表时是唯一已发表的利用蚤进行的实验传播研究。显然，血支原体的蚤传播仍然是一个需要进一步研究的问题。在这方面，自我们的手稿发表以来，另一项实验研究已经发表[11]。该研究发现对蚤传播的支持较少，而对直接传播的支持较多。具体而言，在共享同一围栏的感染蚤的猫之间观察到了血支原体传播（1/6感染血猫支原体，3/6感染血最小支原体），但在仅允许蚤移动而不允许猫直接接触的单独围栏中的猫之间未观察到传播[11]。相比之下，在该研究中，Bartonella clarridgeiae在同一围栏中所有感染蚤的猫之间传播，并且也传播到了相邻围栏中允许蚤移动的八只猫中的两只。这证明了猫蚤在B. clarridgieae传播中具有更明确的作用，该属已知主要通过节肢动物媒介传播，与嗜血支原体属相比。
地理因素在解释猫和蚤中血支原体感染流行率方面的作用也存在争议。从我们研究中纳入的出版物来看，与Safwat引用的瑞士研究[12]相比，日本[13]和意大利[14]分别呈现出更明确的气候类型范围，从亚寒带到湿润亚热带以及从苔原到夏季炎热的地中海气候[15]。在Willi等人的瑞士研究[12]中，该国南部和西部的猫血支原体感染流行率高于其他地区。虽然瑞士西部是与北部相当的温带海洋性气候，但瑞士南部因阿尔卑斯山而属于亚寒带到苔原气候。因此，Willi等人的研究结果并未将猫中较高的血支原体流行率与温度和湿度较高的地区联系起来，也就未与预计猫蚤更丰富的地区联系起来。此外，意大利和日本的调查在同一篇[14]或后续相关[16]手稿中包含了蚤传Bartonella spp.的监测，从而能够更有效地比较血支原体和蚤传Bartonella spp.。
我们完全同意作者的观点，即有许多重要因素可能影响感染流行率和/或蚤中血支原体DNA的检测，并希望未来的研究不仅关注洗涤和混合的作用，还应关注物种和菌株差异，以及常规PCR、qPCR和其他分子检测方法之间的分析差异。事实上，不断有报告确认使用特异性不足的16S rRNA引物对蚤微生物组成分产生脱靶扩增，突显了PCR测序和稳健的基本局部比对搜索工具（BLAST）分析或比对的重要性[17]。
总之，我们的目的并非忽视使用Jensen引物记录到的高血支原体流行率，而是提醒这些引物在开发和验证时可能并不像最初认为的那样具有血支原体特异性。此外，我们承认蚤可能是有效的媒介；然而，地理调查和最新证据更支持嗜血支原体属在猫之间的直接传播。我们非常同意这一结论，即需要更多研究来阐明蚤作为这些缺乏细胞壁的细菌的媒介的作用，并阐明这种重要猫病原体自然传播的主要途径。
补充信息
[内容请查看原文]
致谢
不适用。
作者贡献
EL和CM完成了原始手稿的起草准备。所有作者审阅并编辑了最终手稿草稿。EL和EB监督了该项目。
资助
本项目的一部分是在CM获得北卡罗来纳州立大学美国国立卫生研究院分子生物技术培训项目资助（奖项编号1T32GM133366）期间完成的。
数据和材料的可用性
用于荟萃分析的数据可在更新的补充数据文件中获取。
声明
伦理批准
本信函中描述的工作不需要伦理批准。
发表同意
所有作者同意发表本手稿。
脚注
[内容请查看原文]
参考文献
[内容请查看原文]
关联数据
数据可用性声明
用于荟萃分析的数据可在更新的补充数据文件中获取。
查看英文原文▾
Dear Editors,
We appreciate the recent response to our previously published manuscript in your journal, and in the spirit of scientific inquiry, would like to take this opportunity to respond to the author’s comments and critiques. Firstly, we welcome the author’s deep engagement with this subject and the body of referenced work. We regret any omitted or incorrectly recorded data from the meta-analysis component of our study.
In regard to the mistakes in our data collection for the meta-analysis, we thank this author for their attention to detail and for finding these errors. Indeed, we have revised the supplementary table 1 to include the corrected figures for fleas in the Zarea, Abdullah, and Shaw et al. studies [1–3].
For the study utilizing Jensen primers, we acknowledge the correction that Shaw et al. [3] did utilize washing, and they followed their polymerase chain reaction (PCR) with sequencing in 16 of the 44 DNA extracts from fleas removed from cats, finding a match with M. haemominutum over a 193 bp fragment of the 16S rRNA gene. We have included here an updated table 1 reflecting the miscategorization of the Shaw study. However, in our opinion, this correction does not change the conclusion that the Jensen primers could have been amplifying Spiroplasma spp. DNA, the prevalence of which was not necessarily reduced by washing. The fact that the Shaw study used washing lends further support to our hypothesis, in that even though the fleas were washed, ~40% of them (44/90) yielded positive results from the Jensen primers.
Table 1. Primer set, flea pooling, and flea washing are associated with the detection of hemotropic Mycoplasma spp
| Primer | Washed | Percentage of infected fleas ± 95% CI (# positive/total) |
|---|---|---|
| Myco_Hf_F/Myco_Hf_F.1 and Myco_Hf_R [ 4 ] | No | 45.7 ± 1.9% (148/324) a |
| Myco_Hf_F/Myco_Hf_F.1 and Myco_Hf_R [ 4 ] | Yes | 17.1 ± 4.2% (44/258) b |
| Other primers | No | 1.5 ± 0.3% (11/734) c |
| Other primers | Yes | 1.4 ± 0.1% (3/215) c |
For the studies using non-Jensen primers, the corrected numbers do not significantly change our interpretation. In the Zarea et al. study [1], it is slightly unclear how the proportion should be calculated, since there were two C. canis fleas removed from cats that would not have been included on the basis of our inclusion criteria, but the authors do not specify whether all seven hemotropic mycoplasma-positive fleas were C. felis. Conservatively, we can estimate this as 7/91, a prevalence of < 8%. Similarly, for the Abdullah et al. study [2], based on our inclusion criteria, we should have included only the results from the flea pools collected from cats, which included 210 pools. However, according to their results: “The species-specific qPCRs for haemoplasma species found that only 3 flea samples were positive for M. haemofelis or M. haemocanis, all of which were found in C. f. felis fleas, 2 of which were collected from cats, and one had no record of the host species.” Therefore, the numerator for this study should also be revised to include only the two positive pools known to be collected from cats. Using 2/210 (1.0%) instead of the reported 3/467 (0.6%) does not make a statistically meaningful difference in the interpretation of these results. Updating the overall proportion of hemoplasma DNA amplified from C. felis using the various species-specific (non-Jensen) primers gives us 14/949 (1.5%), which, in our opinion, still suggests that amplification of hemotropic mycoplasma from fleas when using these species-specific primers via quantitative (q)PCR remains exceedingly rare.
When it comes to the role of pooling, it is true that inconsistent pool sizes within studies complicate meta-analysis. As most studies do not make flea pools of a consistent size and do not report the number of fleas in all positive and negative pools, our meta-analysis only aimed to capture any report of pooling. We urge future researchers to include metadata or analysis of the detection of hemoplasmas in flea pools of varying sizes. The questions remaining on pool size from both the Willi et al. [5] and Abdullah et al. [2] studies, and the different interpretations of different readers, highlight the importance of clear and comprehensive methods reporting, including reporting of pool sizes for these types of studies. Additionally, the impact of pool size on the detection of hemoplasma or other bacteria harbored or transmitted by fleas is unknown, again stressing the importance of focusing future research on determining optimal conditions, including pool size, for appropriate detection of bacterial DNA in fleas via PCR.
Despite referencing the influence of pooling throughout the manuscript, Willi et al. [5] unfortunately do not report the number of fleas per pool. That study reports results of 73 fleas from 17 cats, with pools of 1–5 fleas (per Willi et al. [5] Table 1), and the methods state “Some tick and flea samples were pooled for DNA extraction (Table 1); pools consisted of the arthropods of one species and those collected from one animal.” In the results, the authors state “Three ticks and two fleas collected from animals tested positive with real-time PCR for hemotropic mycoplasmas; all positive samples were extracted from individual arthropods,” while in the discussion the authors offer the following as a potential explanation for a low prevalence estimate of hemoplasmas: “The lower sample prevalence in the present study could be explained by the fact that most fleas were analyzed individually and not in pools.” Therefore, it was not possible to separate fleas by pooled or unpooled for meta-analysis. Similarly, in the Abdullah et al. study [2], while the methods reported that fleas were pooled, at least half of the “pools” as reported by the study were single fleas. No further information was available on pool size otherwise from this study, and the effect of single versus multiple fleas per pool was not investigated further. Regardless, noted differences in interpretation of pooling do not change the conclusion that with the use of species-specific primers, there were very few hemoplasma-infected fleas: two individual fleas in Willi et al. [5] and two pools (likely but not definitively consisting of one flea each) out of more than 200 pools in Abdullah et al. [2].
In the experimental component of our study, we retested samples from a previously published study by Assarakorn et al. [6] that reported a high prevalence of hemoplasma DNA in fleas using Jensen primers. We acknowledge that one potential explanation for the significantly lower prevalence in the reanalysis could be due to sample storage, though there is a substantial body of research suggesting that DNA remains stable when stored frozen for years [7]. However, the issues with the pooling methodology are likely not related. The reporting of pooling methodology and results in the original Assarakorn study were, like in many of the referenced publications, somewhat nonspecific, leaving it open to misunderstanding and therefore misreporting in our manuscript [8]. According to the primary investigator of the Assarakorn et al. study (M. Lappin) and a co-author of our manuscript, in the original Assarakorn study, fleas were pooled by cat, with many cats having only one flea per pool, but up to a maximum of five fleas per pool. There is no further information available on the number of fleas per pool or the number of cats with only a single individual flea collected. However, in the original Assarakorn et al. study, there were some cats with more than 5 fleas, and those cats had multiple pools created, but only 1 pool was tested, which explains the numerical discrepancy between the 50 pools (from 50 cats) tested in Assarakorn et al. and the 67 pools (from 50 cats) tested in our manuscript. For our reanalysis, aliquots from those original DNA extracts—all 67—were retested at both Colorado State University using the Jensen primers and North Carolina State University using the Manvell primers. Because all pools from all cats were tested (67 pools from 50 cats), the results can be directly compared between only the 50 original pools. Of the previously untested pools, 7/17 (41%) were positive using the Jensen primers, compared with 18/50 (36%) of the previously tested pools. The results are presented more explicitly here in the table (Table 2).
Table 2. Flea pools from Assarakorn et al. [ 6 ] study, with results shown by the original result and retesting in Moore et al. [ 8 ]
| Original Assarakorn et al. [ 6 ] result | Moore et al. [ 8 ] retesting result | |||
|---|---|---|---|---|
| Mhf | Spiroplasma | Other bacteria | Negative | |
| Mhf or Mt | 0 | 0 | 0 | 2 |
| Mhm | 0 | 13 | 1 | 1 |
| Negative | 0 | 2 | 2 | 29 |
| Not tested | 2 | 1 | 4 | 10 |
In regard to the grouping of Manvell and Jensen primers, in the Manvell et al. study [9], there were initially a large number of positive PCR results that were recategorized in the final analysis due to sequencing showing that these positives were actually Spiroplasma spp., not hemoplasma. In meta-analysis, these primer sets were grouped, as they share a reverse primer and display a single base pair shift in the forward primer that is homologous among feline hemoplasmas and Spiroplasma spp. of C. felis.
Finally, when it comes to the washing component of our experimental study, we acknowledge that due to unknown hemoplasma presence in the fleas, it is impossible to know the effect of washing. We mentioned this in the Discussion section of our original publication. However, this point could have been emphasized more strenuously. Unfortunately, fleas with known hemoplasma infection were not available to study the role of washing fleas in mycoplasma detection. These important experiments remain to be done.
Overall, the experimental evidence for hemoplasma transmission by fleas is minimal, indicating that more robust experiments are needed. Indeed, the hemoplasma experimental infection numbers cited in our study are small, but not excessively small for such a labor-intensive animal study, considering the difficulty of working with C. felis and cats in an experimental setting [10]. The results, while limited, at the time of publication, were the only published experimental transmission study using fleas. Clearly, flea transmission of hemoplasmas remains a subject requiring additional investigation. In that regard, another experimental study has been published [11] since our manuscript was published. That study found less support for flea transmission and more support for direct transmission. Specifically, hemoplasma transmission was observed between flea-infested cats sharing the same enclosure (1/6 infected with M. haemofelis and 3/6 with M. haemominutum), but no transmission was observed between cats in separate enclosures that only permitted flea movement, but not direct cat contact [11]. In comparison, in that study, Bartonella clarridgeiae was transmitted among all of the flea-infested cats in the same enclosure, as well as to two of the eight cats in the adjacent enclosures that allowed for flea movement. This demonstrated a clearer role of C. felis in transmission of B. clarridgieae, a genus known for predominantly arthropod vector-borne transmission, compared with hemotropic Mycoplasma spp.
The role of geography in interpreting the prevalence of hemoplasma infection in cats and fleas is also contentious. From the publications included in our study, when compared with the Switzerland study cited by Safwat [12], Japan [13] and Italy [14] present a clearer range of climate types, from subarctic to humid subtropical and tundra to hot-summer Mediterranean climate, respectively [15]. In the Swiss study by Willi et al. [12], there was increased prevalence of hemoplasma infection in cats from the south and west compared with other parts of the country. While the west of Switzerland is a temperate oceanic climate comparable to the north, the south of Switzerland is a subarctic to tundra climate due to the Alps. The findings by Willi et al. do not therefore associate higher hemoplasma prevalence in cats with areas of increased temperature and humidity, and therefore areas predicted to have more abundant C. felis. In addition, the Italy and Japan surveys included flea-borne Bartonella spp. surveillance in the same [14] or subsequent related [16] manuscripts, allowing for more effective comparisons between hemoplasma and flea-borne Bartonella spp.
We agree wholeheartedly with the author that there are many important factors that may influence infection prevalence and/or detection of hemoplasma DNA in fleas, and hope that future studies will focus not only on the role of washing and pooling, but also on species and strain differences, as well as analytical differences between conventional PCR, qPCR, and other molecular detection methods. Indeed, reports continue to acknowledge the off-target amplification of flea microbiome constituents utilizing 16S rRNA primers with inadequate specificity, highlighting the importance of PCR sequencing and robust Basic Local Alignment Search Tool (BLAST) analysis or alignment [17].
In conclusion, it was not our purpose to disregard the high prevalences of hemoplasma documented using Jensen primers, but to caution that those primers may not be as hemoplasma-specific as originally thought at the time of development and validation. In addition, we acknowledge that fleas may be a competent vector; however, geographic surveys and recent evidence are more supportive of direct transmission of hemotropic Mycoplasma species among cats. We very much concur with the conclusion that additional studies are needed to clarify the role of fleas as vectors for these cell-wall-deficient bacteria and elucidate the major route(s) of natural transmission for this important feline pathogen.
Supplementary Information
[内容请查看原文]
Acknowledgement
Not Applicable.
Author contributions
EL and CM performed original manuscript draft preparation. All authors reviewed and edited the final manuscript draft. EL and EB supervised the project.
Funding
A portion of this project was completed while CM was supported by the North Carolina State University’s Molecular Biotechnology Training Program of the National Institutes of Health under award number 1T32GM133366.
Availability of data and materials
The data utilized for meta-analysis are available in the updated supplementary data file.
Declarations
Ethical approval
Ethical approval was not needed for the work described in this letter.
Consent for publication
All authors consent to the publication of this manuscript.
Footnotes
[内容请查看原文]
References
[内容请查看原文]
Associated Data
Data Availability Statement
The data utilized for meta-analysis are available in the updated supplementary data file.